The relationship between healthcare workers’ attachment styles and patient outcomes: a systematic review

The relationship between healthcare workers’ attachment styles and patient outcomes: a... Abstract Purpose To examine whether attachment styles of healthcare workers influence the quality of their relationships with patients, or impact patients’ health outcomes. Data source Literature database searches on the CINAHL, Cochrane Library, Embase, MEDLINE and PsyCinfo, and hand searching of reference lists of the retained articles. Study selection Original empirical studies reporting an analysis of the relationship of interest were selected for review. Data extraction Estimates of association between healthcare workers’ attachment style and patients’ health outcomes were extracted. Results of data synthesis Results from 13 studies were mixed in terms of which attachment styles related to patients’ perceptions of care or health outcomes, and the evidence overall was of poor quality and methodologically heterogeneous. However, there is limited evidence that secure attachment styles of healthcare workers have little or a negative effect on patients’ health outcomes or perceptions in the short term but in the long term have a more positive effect. Conversely, insecure styles tend to have a positive effect in the short term but little or a negative effect on long-term relationships. Studies which used self-report attachment measurements tended to report stronger associations with patients’ outcome measurements than studies using the interviewer rated Adult Attachment Interview. Conclusion It is unclear whether or not there is a relationship between attachment style of health workers and patients’ health outcomes. Further research using consistent data collection tools, especially in relation to the attachment measurement construct selected, and analysis methods across studies is required to draw recommendations for clinical practice. attachment style, psychological bonding, healthcare, outcome assessment, professional patient relationship Introduction The quality of the interpersonal relationships between healthcare workers and patients is an important process of care variable and may influence patients’ outcomes since it may influence patients’ compliance with treatment. Previous studies have demonstrated that, among a wide variety of influential factors, attachment styles are important antecedents of interpersonal relationship quality [1]. Building on Bowlby’s attachment theory that conceptualizes attachment styles as secure or insecure [2–4], Bartholomew and Horowitz [5] proposed an attachment style model comprising intersecting dimensions of positive and negative perceptions of the self and others. They proposed four attachment styles: secure, dismissing, pre-occupied and fearful. People with secure attachment have positive regard for both themselves and others. In contrast, individuals with a dismissing attachment style have a positive self-image but a negative image of others. Those with a pre-occupied attachment style have a negative self-image but are positive towards others. People with a fearful attachment style perceive both themselves and others as unworthy and unlikable. These theories suggest that attachment styles of both the healthcare provider and their patients can influence the quality of clinician–patient relationships. However, research on the association between patient attachment styles and the patient–clinician relationship has been criticized for not having clinical applications [6]. A more productive approach might be to consider the attachment styles of healthcare workers as a potential prerequisite of the quality of patient care. A recent systematic review investigating the impact of therapist attachment style on psychotherapeutic alliance and outcomes, found diverse findings across studies but also some evidence that therapist attachment style is a predictor of the therapeutic alliance and patients’ health outcomes [7]. The present review builds on this finding by investigating the relationship between the attachment style of a wider group of healthcare professionals and the quality of patient care. To assess adult attachment styles in research, a variety of tools have been developed. Some of these involve data collection by interviewer-rating assessment and others through self-report. Amongst the first group is the Adult Attachment Interview (AAI) [8], which focuses on how interviewees coherently organize their speech with respect to childhood experiences with significant caregivers irrespective of whether they as adults perceive such loving relationships as positive or negative [9]. In contrast, self-report measures ask respondents to report how they think and feel about current close relationships [10]. This review investigates whether there is an association between healthcare workers’ attachment styles, measured with the AAI or self-rating tools, and patients’ health outcomes or their perception of care provided by their healthcare workers. Methods Literature search An electronic literature search was performed in February 2017 on the following databases: CINAHL, Cochrane Library, Embase (1974-), MEDLINE (1946-) and PsyCinfo (1806-). The search strategy is summarized in Table 1. Table 1 Database literature searching strategy Keywords Population facet Measurement facet doctor psychiatrist nurse practitioner clinician therapist psychologist counsellor health worker healthcare worker health professional healthcare professional attachment Adult Attachment Interview Adult Attachment Scale Attachment Style Questionnaire Adult Attachment Style Measure Experience in Close Relationship Scale Self-Reliance Inventory Strategy Keywords were truncated where applicable and MeSH terms were also used including all subheadings where available. Keywords and terms were combined with the Boolean operator OR within each facet. These combinations of the two facets were further combined with the Boolean operator AND. Studies were limited to those using humans. Keywords Population facet Measurement facet doctor psychiatrist nurse practitioner clinician therapist psychologist counsellor health worker healthcare worker health professional healthcare professional attachment Adult Attachment Interview Adult Attachment Scale Attachment Style Questionnaire Adult Attachment Style Measure Experience in Close Relationship Scale Self-Reliance Inventory Strategy Keywords were truncated where applicable and MeSH terms were also used including all subheadings where available. Keywords and terms were combined with the Boolean operator OR within each facet. These combinations of the two facets were further combined with the Boolean operator AND. Studies were limited to those using humans. Table 1 Database literature searching strategy Keywords Population facet Measurement facet doctor psychiatrist nurse practitioner clinician therapist psychologist counsellor health worker healthcare worker health professional healthcare professional attachment Adult Attachment Interview Adult Attachment Scale Attachment Style Questionnaire Adult Attachment Style Measure Experience in Close Relationship Scale Self-Reliance Inventory Strategy Keywords were truncated where applicable and MeSH terms were also used including all subheadings where available. Keywords and terms were combined with the Boolean operator OR within each facet. These combinations of the two facets were further combined with the Boolean operator AND. Studies were limited to those using humans. Keywords Population facet Measurement facet doctor psychiatrist nurse practitioner clinician therapist psychologist counsellor health worker healthcare worker health professional healthcare professional attachment Adult Attachment Interview Adult Attachment Scale Attachment Style Questionnaire Adult Attachment Style Measure Experience in Close Relationship Scale Self-Reliance Inventory Strategy Keywords were truncated where applicable and MeSH terms were also used including all subheadings where available. Keywords and terms were combined with the Boolean operator OR within each facet. These combinations of the two facets were further combined with the Boolean operator AND. Studies were limited to those using humans. The titles and abstracts of all the papers identified were screened to ensure they met the following inclusion criteria: (a) Published in English or Japanese language. (b) Report the findings of empirical research studies conducted within a healthcare setting examining attachment styles of any healthcare workers using any measurement tool based on Bowlby’s attachment theory. (c) Report patients’ somatic and/or psychological health response to care using a measure/indicator. To be included in the review, healthcare workers had to be defined by the study authors as people whose paid employment involved promotion, protection, prevention, curation or improvement of the health of the population [11] in settings where the primary purpose is the treatment and prevention of disease and illness. In addition to the electronic literature searches, hand searching was conducted by scanning reference lists of retained articles, relevant review papers and bibliographies. Full texts of any potentially relevant studies were sought, and if they could not be obtained, information on the study was sought from the corresponding author by email. Retrieved full-text papers were excluded if they met any of the following criteria: (d) Studies using a simulated intervention and (e) Studies in which more than 50% of the healthcare providers were pre-registration students. These studies were excluded because students are always supervised by senior professionals and their ways of relating patients are likely to be influenced by the supervisor, so the effects of their own attachment styles might be attenuated. Quality assessment We used a standardized quality assessment checklist, the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies by the National Heart, Lung and Blood Institute [12], and modified the phrases to make them applicable to this review (Table 2). Table 2 Quality assessment checklist: a modification of Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies (NHLBI 2014) Item Yes C/D, N/A or N/R* No 1 Was the research question/objective clearly stated? 2 Was the study population (healthcare workers and patients/clients) clearly specified and defined? 3 Was the participation rate of eligible persons at least 50%? 4 Were all the subjects selected or recruited from the same or similar populations (including the same time period)? Were inclusion/exclusion criteria pre-specified and applied uniformly to all participants? 5 Was a sample size justification provided? 6 For the analyses, was the independent variable (attachment representation of healthcare workers) measured prior to the outcome (response of patients/clients) being measured? 7 Was the timeframe of interpersonal relationship between healthcare workers and patients/clients sufficient so that one could reasonably expect to see an association between the independent variable (attachment representation of healthcare workers) and the outcome (response of patients/clients) if it existed? 8 Did the study examine different status/levels of the attachment representation of healthcare workers as related to the responses of patients/clients (e.g. categories of attachment representation, or attachment representation measured as continuous variable)? 9 Was the attachment measure clearly defined, valid/reliable (validity/reliability in the present study was sufficiently reported), and implemented consistently across all healthcare workers? 10 Was the intervention, if any, standardized and assessed/checked? 11 Were the measures of response of patients/clients clearly defined, valid/reliable (validity/reliability in the present study was sufficiently reported), and implemented consistently across all patients/clients? 12 Were the outcome (response of patients/clients) assessors blinded/independent to the outcome status (attachment representation of healthcare workers)? 13 Was loss to follow-up after baseline 20% or less? 14 Were key potential confounding variables measured and adjusted statistically for their impact on the relationship between the independent variable (attachment representation of healthcare workers) and the outcome (response of patients/clients)? Item Yes C/D, N/A or N/R* No 1 Was the research question/objective clearly stated? 2 Was the study population (healthcare workers and patients/clients) clearly specified and defined? 3 Was the participation rate of eligible persons at least 50%? 4 Were all the subjects selected or recruited from the same or similar populations (including the same time period)? Were inclusion/exclusion criteria pre-specified and applied uniformly to all participants? 5 Was a sample size justification provided? 6 For the analyses, was the independent variable (attachment representation of healthcare workers) measured prior to the outcome (response of patients/clients) being measured? 7 Was the timeframe of interpersonal relationship between healthcare workers and patients/clients sufficient so that one could reasonably expect to see an association between the independent variable (attachment representation of healthcare workers) and the outcome (response of patients/clients) if it existed? 8 Did the study examine different status/levels of the attachment representation of healthcare workers as related to the responses of patients/clients (e.g. categories of attachment representation, or attachment representation measured as continuous variable)? 9 Was the attachment measure clearly defined, valid/reliable (validity/reliability in the present study was sufficiently reported), and implemented consistently across all healthcare workers? 10 Was the intervention, if any, standardized and assessed/checked? 11 Were the measures of response of patients/clients clearly defined, valid/reliable (validity/reliability in the present study was sufficiently reported), and implemented consistently across all patients/clients? 12 Were the outcome (response of patients/clients) assessors blinded/independent to the outcome status (attachment representation of healthcare workers)? 13 Was loss to follow-up after baseline 20% or less? 14 Were key potential confounding variables measured and adjusted statistically for their impact on the relationship between the independent variable (attachment representation of healthcare workers) and the outcome (response of patients/clients)? Overall quality: good fair poor Overall quality: good fair poor Note. *C/D = cannot determine; N/A = not applicable; N/R = not reported. Table 2 Quality assessment checklist: a modification of Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies (NHLBI 2014) Item Yes C/D, N/A or N/R* No 1 Was the research question/objective clearly stated? 2 Was the study population (healthcare workers and patients/clients) clearly specified and defined? 3 Was the participation rate of eligible persons at least 50%? 4 Were all the subjects selected or recruited from the same or similar populations (including the same time period)? Were inclusion/exclusion criteria pre-specified and applied uniformly to all participants? 5 Was a sample size justification provided? 6 For the analyses, was the independent variable (attachment representation of healthcare workers) measured prior to the outcome (response of patients/clients) being measured? 7 Was the timeframe of interpersonal relationship between healthcare workers and patients/clients sufficient so that one could reasonably expect to see an association between the independent variable (attachment representation of healthcare workers) and the outcome (response of patients/clients) if it existed? 8 Did the study examine different status/levels of the attachment representation of healthcare workers as related to the responses of patients/clients (e.g. categories of attachment representation, or attachment representation measured as continuous variable)? 9 Was the attachment measure clearly defined, valid/reliable (validity/reliability in the present study was sufficiently reported), and implemented consistently across all healthcare workers? 10 Was the intervention, if any, standardized and assessed/checked? 11 Were the measures of response of patients/clients clearly defined, valid/reliable (validity/reliability in the present study was sufficiently reported), and implemented consistently across all patients/clients? 12 Were the outcome (response of patients/clients) assessors blinded/independent to the outcome status (attachment representation of healthcare workers)? 13 Was loss to follow-up after baseline 20% or less? 14 Were key potential confounding variables measured and adjusted statistically for their impact on the relationship between the independent variable (attachment representation of healthcare workers) and the outcome (response of patients/clients)? Item Yes C/D, N/A or N/R* No 1 Was the research question/objective clearly stated? 2 Was the study population (healthcare workers and patients/clients) clearly specified and defined? 3 Was the participation rate of eligible persons at least 50%? 4 Were all the subjects selected or recruited from the same or similar populations (including the same time period)? Were inclusion/exclusion criteria pre-specified and applied uniformly to all participants? 5 Was a sample size justification provided? 6 For the analyses, was the independent variable (attachment representation of healthcare workers) measured prior to the outcome (response of patients/clients) being measured? 7 Was the timeframe of interpersonal relationship between healthcare workers and patients/clients sufficient so that one could reasonably expect to see an association between the independent variable (attachment representation of healthcare workers) and the outcome (response of patients/clients) if it existed? 8 Did the study examine different status/levels of the attachment representation of healthcare workers as related to the responses of patients/clients (e.g. categories of attachment representation, or attachment representation measured as continuous variable)? 9 Was the attachment measure clearly defined, valid/reliable (validity/reliability in the present study was sufficiently reported), and implemented consistently across all healthcare workers? 10 Was the intervention, if any, standardized and assessed/checked? 11 Were the measures of response of patients/clients clearly defined, valid/reliable (validity/reliability in the present study was sufficiently reported), and implemented consistently across all patients/clients? 12 Were the outcome (response of patients/clients) assessors blinded/independent to the outcome status (attachment representation of healthcare workers)? 13 Was loss to follow-up after baseline 20% or less? 14 Were key potential confounding variables measured and adjusted statistically for their impact on the relationship between the independent variable (attachment representation of healthcare workers) and the outcome (response of patients/clients)? Overall quality: good fair poor Overall quality: good fair poor Note. *C/D = cannot determine; N/A = not applicable; N/R = not reported. Data extraction and analysis A pre-determined data extraction sheet was used, comprising items as listed in Tables 3 and 4. The data needed for assessing the quality of studies (Table 2) were also extracted. Table 3 Characteristics, findings and appraisals of studies using Adult Attachment Interview Design and setting Population and sampling Attrition or response rate Relationship/ intervention duration Workers’ attachment measure [measure point] Patients’ outcome measure [measure point] Statistical analysis and findings Quality assessment Relationship quality Health outcome Petrowski et al. (2013)  Repeated measure design  Germany  A university hospital inpatient setting 22 therapists  15 psychologists  7 physicians Sampling not clear Not reported 12 sessions (average 62.5 days, SD = 28.45) AAI [pre-therapy] GSI CATS [post-therapy] Hierarchical linear model  AAI(secure–insecure) not associated with  CATS (secure) β not reported (P > 0.05)  AAI(dismissing–preoccupied) not  associated with CATS(preoccupied/merger) β = −0.88 (P = 0.06)  AAI(dismissing–preoccupied) associated  with CATS (avoidant/fearful) β = 0.94 (P = 0.03)* Pearson correlation AAI (secure–insecure) not correlated with GSI r = −0.06 (P = 0.07) AAI(dismissing–preoccupied) not correlated with GSI r = −0.12 (P = 0.51) Fair [7–2] 429 patients Sampling not clear (randomly assigned to therapists) Sibrava (2009)  Repeated measure  design 4 therapists Sampling not clear Not reported 14 Weekly sessions (1.5–2 h) AAI [pre-therapy] WAI-36 [end of 2, 5, 10, 14 sessions] Hierarchical linear model  AAI associated with WAI [at session 2  AAI(secure) showed lower WAI than  AAI(insecure), at session 14  AAI(secure) showed higher WAI than  AAI(insecure)] β = 0.62 (P < 0.05)* Poor [6–2] Small sample size in therapists  Not clear 69 clients Sampling not clear Petrowski et al. (2011)  Repeated measure design  Germany  A university hospital inpatient setting 19 therapists  13 psychologists  6 physicians  Sampling not clear Not reported Not clear AAI [pre-therapy] HAQ [post-therapy] Hierarchical regression  AAI(secure–insecure) not associated with   HAQ (statistics not reported)  AAI (dismissing–preoccupied) associated   with HAQ (relationship) β = −0.65  (P = 0.02)*   with HAQ(outcome) β = −0.65 (P <  0.01)*  In all other 7 models, AAI (secure– insecure) or AAI (dismissing–preoccupied) not associated with HAQ (relationship) or HAQ (outcome) (statistics not reported) Poor [5–1] Repeated data of therapists in analysis 59 patients  Sampling not clear  (randomly assigned to therapists) Schauenburg et al. (2010)  Repeated measure design  Germany  2 psychotherapy hospitals, inpatient settings 31 therapists  27 physicians  4 psychologists  Sampling not clear Not reported Mean treatment duration 12.01 weeks (SD = 5.4) AAI [pre-therapy] GSI IIP IS [pre-, post-therapy] HAQ [post-therapy] Hierarchical linear model AAI(secure–insecure) not associated with  HAQ β = 0.07 (P = 0.53) AAI(dismissing–preoccupied) not associated with HAQ β = 0.10 (P = 0.66) Hierarchical linear model AAI(secure–insecure) not associated with GSI  β = 0.07  IIP  β = 0.07  IS  β = −0.22 (all P > 0.05) AAI(dismissing–preoccupied) not associated with GSI  β = −0.01 IIP  β = −0.06 IS  β = −0.14 (all P > 0.05) Poor [5–1] 1381 patients assigned to therapists in consecutive order  Sampling not clear Tyrrel et al. (1999)  Cross-sectional study  USA  A community-based setting 21 case managers  Sampling not clea 54 clients  Sampling not clear Not reported Mean dyad relationship 31 months (SD = 17, range=7–69) AAI WAI-36 QLI GAF BDI Psychiatric hospitalization days in previous 1 year Correlation AAI(dismissing–preoccupied) not correlated with WAI r = 0.07 (P not reported) Hierarchical regression AAI(dismissing–preoccupied) not associated with WAI β = 0.04 (P not reported) Correlation  AAI(dismissing-preoccupied) not correlated with  QLI(global) r = 0.02  QLI(relation-q) r = -0.08  QLI(relation-s) r = 0.10  GAF r = 0.23  BDI r = 0.10  Hospitalization r = 0.06 (all p not reported) Poor [5–2] Repeated data of therapists in analysis Hierarchical regression  AAI(dismissing-preoccupied) not associated with  QLI(global) β = −0.03  QLI(relation-q) β = −0.17  QLI(relation-s) β = 0.02  GAF β = 0.20  BDI β = 0.10  Hospitalization β = 0.01 (all p not reported) Design and setting Population and sampling Attrition or response rate Relationship/ intervention duration Workers’ attachment measure [measure point] Patients’ outcome measure [measure point] Statistical analysis and findings Quality assessment Relationship quality Health outcome Petrowski et al. (2013)  Repeated measure design  Germany  A university hospital inpatient setting 22 therapists  15 psychologists  7 physicians Sampling not clear Not reported 12 sessions (average 62.5 days, SD = 28.45) AAI [pre-therapy] GSI CATS [post-therapy] Hierarchical linear model  AAI(secure–insecure) not associated with  CATS (secure) β not reported (P > 0.05)  AAI(dismissing–preoccupied) not  associated with CATS(preoccupied/merger) β = −0.88 (P = 0.06)  AAI(dismissing–preoccupied) associated  with CATS (avoidant/fearful) β = 0.94 (P = 0.03)* Pearson correlation AAI (secure–insecure) not correlated with GSI r = −0.06 (P = 0.07) AAI(dismissing–preoccupied) not correlated with GSI r = −0.12 (P = 0.51) Fair [7–2] 429 patients Sampling not clear (randomly assigned to therapists) Sibrava (2009)  Repeated measure  design 4 therapists Sampling not clear Not reported 14 Weekly sessions (1.5–2 h) AAI [pre-therapy] WAI-36 [end of 2, 5, 10, 14 sessions] Hierarchical linear model  AAI associated with WAI [at session 2  AAI(secure) showed lower WAI than  AAI(insecure), at session 14  AAI(secure) showed higher WAI than  AAI(insecure)] β = 0.62 (P < 0.05)* Poor [6–2] Small sample size in therapists  Not clear 69 clients Sampling not clear Petrowski et al. (2011)  Repeated measure design  Germany  A university hospital inpatient setting 19 therapists  13 psychologists  6 physicians  Sampling not clear Not reported Not clear AAI [pre-therapy] HAQ [post-therapy] Hierarchical regression  AAI(secure–insecure) not associated with   HAQ (statistics not reported)  AAI (dismissing–preoccupied) associated   with HAQ (relationship) β = −0.65  (P = 0.02)*   with HAQ(outcome) β = −0.65 (P <  0.01)*  In all other 7 models, AAI (secure– insecure) or AAI (dismissing–preoccupied) not associated with HAQ (relationship) or HAQ (outcome) (statistics not reported) Poor [5–1] Repeated data of therapists in analysis 59 patients  Sampling not clear  (randomly assigned to therapists) Schauenburg et al. (2010)  Repeated measure design  Germany  2 psychotherapy hospitals, inpatient settings 31 therapists  27 physicians  4 psychologists  Sampling not clear Not reported Mean treatment duration 12.01 weeks (SD = 5.4) AAI [pre-therapy] GSI IIP IS [pre-, post-therapy] HAQ [post-therapy] Hierarchical linear model AAI(secure–insecure) not associated with  HAQ β = 0.07 (P = 0.53) AAI(dismissing–preoccupied) not associated with HAQ β = 0.10 (P = 0.66) Hierarchical linear model AAI(secure–insecure) not associated with GSI  β = 0.07  IIP  β = 0.07  IS  β = −0.22 (all P > 0.05) AAI(dismissing–preoccupied) not associated with GSI  β = −0.01 IIP  β = −0.06 IS  β = −0.14 (all P > 0.05) Poor [5–1] 1381 patients assigned to therapists in consecutive order  Sampling not clear Tyrrel et al. (1999)  Cross-sectional study  USA  A community-based setting 21 case managers  Sampling not clea 54 clients  Sampling not clear Not reported Mean dyad relationship 31 months (SD = 17, range=7–69) AAI WAI-36 QLI GAF BDI Psychiatric hospitalization days in previous 1 year Correlation AAI(dismissing–preoccupied) not correlated with WAI r = 0.07 (P not reported) Hierarchical regression AAI(dismissing–preoccupied) not associated with WAI β = 0.04 (P not reported) Correlation  AAI(dismissing-preoccupied) not correlated with  QLI(global) r = 0.02  QLI(relation-q) r = -0.08  QLI(relation-s) r = 0.10  GAF r = 0.23  BDI r = 0.10  Hospitalization r = 0.06 (all p not reported) Poor [5–2] Repeated data of therapists in analysis Hierarchical regression  AAI(dismissing-preoccupied) not associated with  QLI(global) β = −0.03  QLI(relation-q) β = −0.17  QLI(relation-s) β = 0.02  GAF β = 0.20  BDI β = 0.10  Hospitalization β = 0.01 (all p not reported) Note. Healthcare workers’ attachment style measure: AAI = Adult Attachment Interview (category: secure, dismissing, preoccupied, unresolved) (score: secure–insecure, dismissing–preoccupied). Relationship quality measure: CATS = Client Attachment to Therapist Scale (subscale: secure, avoidant/fearful, preoccupied/merger); WAI-36 = Working Alliance Inventory 36-Item Version; HAQ = Helping Alliance Questionnaire (subscales: relationship satisfaction, outcome satisfaction). Health outcome measure: GSI = Global Severity Index of the Symptom Checklist; IIP = Inventory of Interpersonal Problems; IS = Impairment Score, caregiver-rating; QLI = Quality of Life Interview (subscale: global life satisfaction, relationship quality, relationship satisfaction); GAF = Global Assessment of Functioning; BDI = 13-item Beck Depression Inventory. Findings: *Statistically significant. Quality Assessment: [the number of ‘Yes’ out of 14 items in total—the number of ‘Yes’ among items 7, 9, 11 and 12] (see Table 1). Table 3 Characteristics, findings and appraisals of studies using Adult Attachment Interview Design and setting Population and sampling Attrition or response rate Relationship/ intervention duration Workers’ attachment measure [measure point] Patients’ outcome measure [measure point] Statistical analysis and findings Quality assessment Relationship quality Health outcome Petrowski et al. (2013)  Repeated measure design  Germany  A university hospital inpatient setting 22 therapists  15 psychologists  7 physicians Sampling not clear Not reported 12 sessions (average 62.5 days, SD = 28.45) AAI [pre-therapy] GSI CATS [post-therapy] Hierarchical linear model  AAI(secure–insecure) not associated with  CATS (secure) β not reported (P > 0.05)  AAI(dismissing–preoccupied) not  associated with CATS(preoccupied/merger) β = −0.88 (P = 0.06)  AAI(dismissing–preoccupied) associated  with CATS (avoidant/fearful) β = 0.94 (P = 0.03)* Pearson correlation AAI (secure–insecure) not correlated with GSI r = −0.06 (P = 0.07) AAI(dismissing–preoccupied) not correlated with GSI r = −0.12 (P = 0.51) Fair [7–2] 429 patients Sampling not clear (randomly assigned to therapists) Sibrava (2009)  Repeated measure  design 4 therapists Sampling not clear Not reported 14 Weekly sessions (1.5–2 h) AAI [pre-therapy] WAI-36 [end of 2, 5, 10, 14 sessions] Hierarchical linear model  AAI associated with WAI [at session 2  AAI(secure) showed lower WAI than  AAI(insecure), at session 14  AAI(secure) showed higher WAI than  AAI(insecure)] β = 0.62 (P < 0.05)* Poor [6–2] Small sample size in therapists  Not clear 69 clients Sampling not clear Petrowski et al. (2011)  Repeated measure design  Germany  A university hospital inpatient setting 19 therapists  13 psychologists  6 physicians  Sampling not clear Not reported Not clear AAI [pre-therapy] HAQ [post-therapy] Hierarchical regression  AAI(secure–insecure) not associated with   HAQ (statistics not reported)  AAI (dismissing–preoccupied) associated   with HAQ (relationship) β = −0.65  (P = 0.02)*   with HAQ(outcome) β = −0.65 (P <  0.01)*  In all other 7 models, AAI (secure– insecure) or AAI (dismissing–preoccupied) not associated with HAQ (relationship) or HAQ (outcome) (statistics not reported) Poor [5–1] Repeated data of therapists in analysis 59 patients  Sampling not clear  (randomly assigned to therapists) Schauenburg et al. (2010)  Repeated measure design  Germany  2 psychotherapy hospitals, inpatient settings 31 therapists  27 physicians  4 psychologists  Sampling not clear Not reported Mean treatment duration 12.01 weeks (SD = 5.4) AAI [pre-therapy] GSI IIP IS [pre-, post-therapy] HAQ [post-therapy] Hierarchical linear model AAI(secure–insecure) not associated with  HAQ β = 0.07 (P = 0.53) AAI(dismissing–preoccupied) not associated with HAQ β = 0.10 (P = 0.66) Hierarchical linear model AAI(secure–insecure) not associated with GSI  β = 0.07  IIP  β = 0.07  IS  β = −0.22 (all P > 0.05) AAI(dismissing–preoccupied) not associated with GSI  β = −0.01 IIP  β = −0.06 IS  β = −0.14 (all P > 0.05) Poor [5–1] 1381 patients assigned to therapists in consecutive order  Sampling not clear Tyrrel et al. (1999)  Cross-sectional study  USA  A community-based setting 21 case managers  Sampling not clea 54 clients  Sampling not clear Not reported Mean dyad relationship 31 months (SD = 17, range=7–69) AAI WAI-36 QLI GAF BDI Psychiatric hospitalization days in previous 1 year Correlation AAI(dismissing–preoccupied) not correlated with WAI r = 0.07 (P not reported) Hierarchical regression AAI(dismissing–preoccupied) not associated with WAI β = 0.04 (P not reported) Correlation  AAI(dismissing-preoccupied) not correlated with  QLI(global) r = 0.02  QLI(relation-q) r = -0.08  QLI(relation-s) r = 0.10  GAF r = 0.23  BDI r = 0.10  Hospitalization r = 0.06 (all p not reported) Poor [5–2] Repeated data of therapists in analysis Hierarchical regression  AAI(dismissing-preoccupied) not associated with  QLI(global) β = −0.03  QLI(relation-q) β = −0.17  QLI(relation-s) β = 0.02  GAF β = 0.20  BDI β = 0.10  Hospitalization β = 0.01 (all p not reported) Design and setting Population and sampling Attrition or response rate Relationship/ intervention duration Workers’ attachment measure [measure point] Patients’ outcome measure [measure point] Statistical analysis and findings Quality assessment Relationship quality Health outcome Petrowski et al. (2013)  Repeated measure design  Germany  A university hospital inpatient setting 22 therapists  15 psychologists  7 physicians Sampling not clear Not reported 12 sessions (average 62.5 days, SD = 28.45) AAI [pre-therapy] GSI CATS [post-therapy] Hierarchical linear model  AAI(secure–insecure) not associated with  CATS (secure) β not reported (P > 0.05)  AAI(dismissing–preoccupied) not  associated with CATS(preoccupied/merger) β = −0.88 (P = 0.06)  AAI(dismissing–preoccupied) associated  with CATS (avoidant/fearful) β = 0.94 (P = 0.03)* Pearson correlation AAI (secure–insecure) not correlated with GSI r = −0.06 (P = 0.07) AAI(dismissing–preoccupied) not correlated with GSI r = −0.12 (P = 0.51) Fair [7–2] 429 patients Sampling not clear (randomly assigned to therapists) Sibrava (2009)  Repeated measure  design 4 therapists Sampling not clear Not reported 14 Weekly sessions (1.5–2 h) AAI [pre-therapy] WAI-36 [end of 2, 5, 10, 14 sessions] Hierarchical linear model  AAI associated with WAI [at session 2  AAI(secure) showed lower WAI than  AAI(insecure), at session 14  AAI(secure) showed higher WAI than  AAI(insecure)] β = 0.62 (P < 0.05)* Poor [6–2] Small sample size in therapists  Not clear 69 clients Sampling not clear Petrowski et al. (2011)  Repeated measure design  Germany  A university hospital inpatient setting 19 therapists  13 psychologists  6 physicians  Sampling not clear Not reported Not clear AAI [pre-therapy] HAQ [post-therapy] Hierarchical regression  AAI(secure–insecure) not associated with   HAQ (statistics not reported)  AAI (dismissing–preoccupied) associated   with HAQ (relationship) β = −0.65  (P = 0.02)*   with HAQ(outcome) β = −0.65 (P <  0.01)*  In all other 7 models, AAI (secure– insecure) or AAI (dismissing–preoccupied) not associated with HAQ (relationship) or HAQ (outcome) (statistics not reported) Poor [5–1] Repeated data of therapists in analysis 59 patients  Sampling not clear  (randomly assigned to therapists) Schauenburg et al. (2010)  Repeated measure design  Germany  2 psychotherapy hospitals, inpatient settings 31 therapists  27 physicians  4 psychologists  Sampling not clear Not reported Mean treatment duration 12.01 weeks (SD = 5.4) AAI [pre-therapy] GSI IIP IS [pre-, post-therapy] HAQ [post-therapy] Hierarchical linear model AAI(secure–insecure) not associated with  HAQ β = 0.07 (P = 0.53) AAI(dismissing–preoccupied) not associated with HAQ β = 0.10 (P = 0.66) Hierarchical linear model AAI(secure–insecure) not associated with GSI  β = 0.07  IIP  β = 0.07  IS  β = −0.22 (all P > 0.05) AAI(dismissing–preoccupied) not associated with GSI  β = −0.01 IIP  β = −0.06 IS  β = −0.14 (all P > 0.05) Poor [5–1] 1381 patients assigned to therapists in consecutive order  Sampling not clear Tyrrel et al. (1999)  Cross-sectional study  USA  A community-based setting 21 case managers  Sampling not clea 54 clients  Sampling not clear Not reported Mean dyad relationship 31 months (SD = 17, range=7–69) AAI WAI-36 QLI GAF BDI Psychiatric hospitalization days in previous 1 year Correlation AAI(dismissing–preoccupied) not correlated with WAI r = 0.07 (P not reported) Hierarchical regression AAI(dismissing–preoccupied) not associated with WAI β = 0.04 (P not reported) Correlation  AAI(dismissing-preoccupied) not correlated with  QLI(global) r = 0.02  QLI(relation-q) r = -0.08  QLI(relation-s) r = 0.10  GAF r = 0.23  BDI r = 0.10  Hospitalization r = 0.06 (all p not reported) Poor [5–2] Repeated data of therapists in analysis Hierarchical regression  AAI(dismissing-preoccupied) not associated with  QLI(global) β = −0.03  QLI(relation-q) β = −0.17  QLI(relation-s) β = 0.02  GAF β = 0.20  BDI β = 0.10  Hospitalization β = 0.01 (all p not reported) Note. Healthcare workers’ attachment style measure: AAI = Adult Attachment Interview (category: secure, dismissing, preoccupied, unresolved) (score: secure–insecure, dismissing–preoccupied). Relationship quality measure: CATS = Client Attachment to Therapist Scale (subscale: secure, avoidant/fearful, preoccupied/merger); WAI-36 = Working Alliance Inventory 36-Item Version; HAQ = Helping Alliance Questionnaire (subscales: relationship satisfaction, outcome satisfaction). Health outcome measure: GSI = Global Severity Index of the Symptom Checklist; IIP = Inventory of Interpersonal Problems; IS = Impairment Score, caregiver-rating; QLI = Quality of Life Interview (subscale: global life satisfaction, relationship quality, relationship satisfaction); GAF = Global Assessment of Functioning; BDI = 13-item Beck Depression Inventory. Findings: *Statistically significant. Quality Assessment: [the number of ‘Yes’ out of 14 items in total—the number of ‘Yes’ among items 7, 9, 11 and 12] (see Table 1). Table 4 Characteristics, findings and appraisals of studies using self-report adult attachment measurements Design and setting Population and sampling Attrition or response rate Relationship/intervention duration Workers’ attachment measure [measure point] Patients’ outcome measure [measure point] Statistical analysis and findings Quality assessment Relationship quality Health outcome Bruck et al. (2006)  Interrupted time series study  USA  A medical centre outpatient service 26 therapists psychologists psychiatrists social workers Convenience sampling 26 patients Recruited through advertisement 20 patients dropped out From 46 (43%) 30 sessions RSQ [pre-therapy] WAI-12 SEQ [6th session] TC GAS SCL IIP [post-therapy] Pearson correlation  RSQ not associated with WAI, SEQ or TC  (statistics not reported) Pearson correlation  RSQ(fearful) associated with   GAS r = −0.41 (P = 0.05)*  RSQ(preoccupied) associated with   GAS r = −0.35 (P = 0.05)*   SCL r = −0.30 (P = 0.05)*  RSQ(dismissing) associated with   IIP r = −0.30 (P < 0.05)*  All other correlations not significant (statistics not reported) Fair [7–2] Bucci et al. (2015)  Cross-sectional study  UK  Primary care psychology services 30 therapists psychologists practitioners counsellors other workers Convenience sampling 30 clients Recruited by therapists (next client at least 3 sessions) 12 clients out of 42 not answered (29%) At least 3 sessions RQ WAI-36 PHQ GAD Spearman correlation  RQ(secure) rho=−0.13 (P = 0.49)  RQ(fearful) rho=−0.17 (P = 0.38)  RQ(preoccupied) rho=−0.08 (P = 0.67)  RQ(dismissing) rho = 0.06 (P = 0.74)  All not associated with WAI Spearman correlation, clients split low/high (by median of PHQ + GAD) symptom  RQ(fearful) associated with WAI in high  symptom clients rho=−0.63 (P = 0.02)*  All other correlations not significant (statistics not reported) Fair [5–2] Kafetsios et al. (2016)  Cross-sectional study  Greece  Private and public system 40 doctors (various specialities) Sampling not clear 160 patients Sampling not clear 100% (No one refused) Average contact time = 34 months 27days RQ(converted into avoidant and anxious) PSQ Multilevel model  RQ(avoidant) not associated with PSQ  β = −0.006 (P > 0.10)  RQ(anxious) not associated with PSQ  β = −0.009 (P > 0.10) Fair [7–2] Dunkle and Friedlander (1996)  Cross-sectional survey 73 therapists Convenience sampling 73 clients Selected by therapists 29% (Data of 73 dyads/252 packets mailed) 3 to 5 sessions AAS WAI-12 Correlation  AAS(close) correlated with   WAI(total) r = 0.39 WAI(task) r = 0.34   WAI(bond) r = 0.41 WAI(goal) r = 0.32 (all P < 0.01)*  AAS(depend) correlated with   WAI(total) r = 0.42 WAI(task) r = 0.37   WAI(bond) r = 0.44 WAI(goal) r = 0.36 (all P < 0.01)*  AAS(anxiety) not correlated with   WAI(total) r = −0.16 WAI(task) r = −0.09   WAI(bond) r = −0.22 WAI(goal) r = −0.12 (all P > 0.05) Regression  AAS(close) associated with WAI(bond) β = 0.38 (P = 0.01)*  AAS(depend) not associated with WAI(bond)  β = 0.31 (P = 0.09)  AAS(anxiety) not associated with WAI(bond)  β = 0.16 (P = 0.37) Fair [6–3]  USA  15 university counselling centres, 6 clinics Wiseman and Tishby (2014)  Repeated measure design  Israel  A university counselling centre 27 therapists psychologists social workers Convenience sampling 54 clients Recruited at reception 13 clients dropped out from 67 (19%) 32 sessions (50-min session, once a week) ECR [pre-therapy] OQ-45 [pre, 5th, 15th, 23th, 28th session] CATS [5th, 15th, 28th session] Hierarchical mixed model  ECR(anxiety) associated with CATS at 5th  β = −0.23 (P = 0.05)*  All other association not significant  [ECR(anxiety)(avoidance) and CATS at 5th, 15th, 28th session] (statistics not reported)   Hierarchical linear model  ECR(anxiety) β = 0.12 (P > 0.05)  ECR(avoidance) β = −0.07  (P > 0.05) not associated with OQ change Good [9–4] Fuertes et al. (2007)  Cross-sectional survey  USA  Private settings, university counselling centres, hospitals, clinics 59 therapists Random sampling from American Counselling Association 59 clients selected by therapists 20% (Data of 59 dyads/300 therapists selected) At least 5 sessions (median = 16, n=30) ECR WAI-12 CATS RRI EUS COM Correlation  ECR(avoidance) associated with CATS(secure) r = −0.27 (P < 0.05)*  All other correlations not significant  [ECR(anxiety)(avoidance) and WAI, RRI,  CATS(secure)(avoidant/fearful)(preoccupied/merger), EUS] r = −0.44 to 0.05 (P not reported) Correlation  ECR(avoidance) correlated  with COM r = −0.45 (P < 0.001)*  ECR(anxiety) correlatedwith COM  r = −0.52 (P < 0.001)* Hierarchical regression  ECR(anxiety) associated  with COM β = −0.41 (P<0.002)* Fair [7–3] Wongpakaran and Wongpakaran (2012)  Cross-sectional study  Thailand  A psychiatric hospital outpatient setting 13 therapists psychiatrists residents Sampling not clear 121 patients Convenience sampling Not clear Not clear (each session 5 min to more than 1 h) ECR (secure or preoccupied) WAI-12 PDQ t-Test  WAI not differed by ECR (statistics not reported) MANOVA  ECR not associated with WAI  (statistics not reported) t-Test  PDQ(interpersonal difficulties) differed by   ECR(secure) 14.12 (SD = 4.20)   ECR(preoccupied) 12.74 (SD = 4.82) (P = 0.026)* ANOVA  ECR(secure) associated with   worse PDQ(anxiety) β = 8.95  (P = 0.001)*   worse PDQ(depression)  β = 8.25 (P = 0.002)*  ECR(secure) not associated  with PDQ (interpersonal difficulty) β = 2.02 (P=0.291) Poor [6–1] Very short relationship duration ECR cut-off not clear Berry and Greenwood (2016)  Cross-sectional study  UK  Community mental health service 33 care co-ordinators  24 nurses  5 occupational therapists  3 social workers  1 clinical psychologists Not reported 3 or more months PAM WAI-12 Correlation  PAM(anxious) not correlated  with WAI(total) r = −0.06 (P > 0.05)  PAM(avoidant) not correlated  with WAI(total) r = 0.04 (P > 0.05) Poor [6–2] Repeated data of therapists in analysis Sampling not clear 61 patients with psychosis Sampling not clear Design and setting Population and sampling Attrition or response rate Relationship/intervention duration Workers’ attachment measure [measure point] Patients’ outcome measure [measure point] Statistical analysis and findings Quality assessment Relationship quality Health outcome Bruck et al. (2006)  Interrupted time series study  USA  A medical centre outpatient service 26 therapists psychologists psychiatrists social workers Convenience sampling 26 patients Recruited through advertisement 20 patients dropped out From 46 (43%) 30 sessions RSQ [pre-therapy] WAI-12 SEQ [6th session] TC GAS SCL IIP [post-therapy] Pearson correlation  RSQ not associated with WAI, SEQ or TC  (statistics not reported) Pearson correlation  RSQ(fearful) associated with   GAS r = −0.41 (P = 0.05)*  RSQ(preoccupied) associated with   GAS r = −0.35 (P = 0.05)*   SCL r = −0.30 (P = 0.05)*  RSQ(dismissing) associated with   IIP r = −0.30 (P < 0.05)*  All other correlations not significant (statistics not reported) Fair [7–2] Bucci et al. (2015)  Cross-sectional study  UK  Primary care psychology services 30 therapists psychologists practitioners counsellors other workers Convenience sampling 30 clients Recruited by therapists (next client at least 3 sessions) 12 clients out of 42 not answered (29%) At least 3 sessions RQ WAI-36 PHQ GAD Spearman correlation  RQ(secure) rho=−0.13 (P = 0.49)  RQ(fearful) rho=−0.17 (P = 0.38)  RQ(preoccupied) rho=−0.08 (P = 0.67)  RQ(dismissing) rho = 0.06 (P = 0.74)  All not associated with WAI Spearman correlation, clients split low/high (by median of PHQ + GAD) symptom  RQ(fearful) associated with WAI in high  symptom clients rho=−0.63 (P = 0.02)*  All other correlations not significant (statistics not reported) Fair [5–2] Kafetsios et al. (2016)  Cross-sectional study  Greece  Private and public system 40 doctors (various specialities) Sampling not clear 160 patients Sampling not clear 100% (No one refused) Average contact time = 34 months 27days RQ(converted into avoidant and anxious) PSQ Multilevel model  RQ(avoidant) not associated with PSQ  β = −0.006 (P > 0.10)  RQ(anxious) not associated with PSQ  β = −0.009 (P > 0.10) Fair [7–2] Dunkle and Friedlander (1996)  Cross-sectional survey 73 therapists Convenience sampling 73 clients Selected by therapists 29% (Data of 73 dyads/252 packets mailed) 3 to 5 sessions AAS WAI-12 Correlation  AAS(close) correlated with   WAI(total) r = 0.39 WAI(task) r = 0.34   WAI(bond) r = 0.41 WAI(goal) r = 0.32 (all P < 0.01)*  AAS(depend) correlated with   WAI(total) r = 0.42 WAI(task) r = 0.37   WAI(bond) r = 0.44 WAI(goal) r = 0.36 (all P < 0.01)*  AAS(anxiety) not correlated with   WAI(total) r = −0.16 WAI(task) r = −0.09   WAI(bond) r = −0.22 WAI(goal) r = −0.12 (all P > 0.05) Regression  AAS(close) associated with WAI(bond) β = 0.38 (P = 0.01)*  AAS(depend) not associated with WAI(bond)  β = 0.31 (P = 0.09)  AAS(anxiety) not associated with WAI(bond)  β = 0.16 (P = 0.37) Fair [6–3]  USA  15 university counselling centres, 6 clinics Wiseman and Tishby (2014)  Repeated measure design  Israel  A university counselling centre 27 therapists psychologists social workers Convenience sampling 54 clients Recruited at reception 13 clients dropped out from 67 (19%) 32 sessions (50-min session, once a week) ECR [pre-therapy] OQ-45 [pre, 5th, 15th, 23th, 28th session] CATS [5th, 15th, 28th session] Hierarchical mixed model  ECR(anxiety) associated with CATS at 5th  β = −0.23 (P = 0.05)*  All other association not significant  [ECR(anxiety)(avoidance) and CATS at 5th, 15th, 28th session] (statistics not reported)   Hierarchical linear model  ECR(anxiety) β = 0.12 (P > 0.05)  ECR(avoidance) β = −0.07  (P > 0.05) not associated with OQ change Good [9–4] Fuertes et al. (2007)  Cross-sectional survey  USA  Private settings, university counselling centres, hospitals, clinics 59 therapists Random sampling from American Counselling Association 59 clients selected by therapists 20% (Data of 59 dyads/300 therapists selected) At least 5 sessions (median = 16, n=30) ECR WAI-12 CATS RRI EUS COM Correlation  ECR(avoidance) associated with CATS(secure) r = −0.27 (P < 0.05)*  All other correlations not significant  [ECR(anxiety)(avoidance) and WAI, RRI,  CATS(secure)(avoidant/fearful)(preoccupied/merger), EUS] r = −0.44 to 0.05 (P not reported) Correlation  ECR(avoidance) correlated  with COM r = −0.45 (P < 0.001)*  ECR(anxiety) correlatedwith COM  r = −0.52 (P < 0.001)* Hierarchical regression  ECR(anxiety) associated  with COM β = −0.41 (P<0.002)* Fair [7–3] Wongpakaran and Wongpakaran (2012)  Cross-sectional study  Thailand  A psychiatric hospital outpatient setting 13 therapists psychiatrists residents Sampling not clear 121 patients Convenience sampling Not clear Not clear (each session 5 min to more than 1 h) ECR (secure or preoccupied) WAI-12 PDQ t-Test  WAI not differed by ECR (statistics not reported) MANOVA  ECR not associated with WAI  (statistics not reported) t-Test  PDQ(interpersonal difficulties) differed by   ECR(secure) 14.12 (SD = 4.20)   ECR(preoccupied) 12.74 (SD = 4.82) (P = 0.026)* ANOVA  ECR(secure) associated with   worse PDQ(anxiety) β = 8.95  (P = 0.001)*   worse PDQ(depression)  β = 8.25 (P = 0.002)*  ECR(secure) not associated  with PDQ (interpersonal difficulty) β = 2.02 (P=0.291) Poor [6–1] Very short relationship duration ECR cut-off not clear Berry and Greenwood (2016)  Cross-sectional study  UK  Community mental health service 33 care co-ordinators  24 nurses  5 occupational therapists  3 social workers  1 clinical psychologists Not reported 3 or more months PAM WAI-12 Correlation  PAM(anxious) not correlated  with WAI(total) r = −0.06 (P > 0.05)  PAM(avoidant) not correlated  with WAI(total) r = 0.04 (P > 0.05) Poor [6–2] Repeated data of therapists in analysis Sampling not clear 61 patients with psychosis Sampling not clear Note. Healthcare workers’ attachment style measure: RSQ = Relationship Scale Questionnaire (subscale: secure, dismissing, preoccupied, fearful); RQ = Relationship Questionnaire (subscale: secure, dismissing, preoccupied, fearful); AAS = Adult Attachment Scale (subscale: depend, anxiety, close); ECR = Revised Experience of Close Relationship Questionnaire (subscale: anxiety, avoidance); PAM = Psychosis Attachment Measure (subscale: anxious, avoidant). Relationship quality measure: WAI-12 = Working Alliance Inventory 12-Item Version (subscales: bond, task, goal); SEQ = Session Evaluation Questionnaire; TC = Target Complaints; PSQ = Patient Satisfaction Questionnaire; WAI-36 = Working Alliance Inventory 36-Item Version; CATS = Client Attachment to Therapist Scale (subscale: secure, avoidant/fearful, preoccupied/merger); RRI = Real Relationship Inventory; EUS = Empathic Understanding Scale. Health outcome measure: GAS = Global Assessment Scale; SCL = Symptom Checklist 90 Revised; IIP = Inventory of Interpersonal Problems; PHQ = Patient Health Questionnaire 9-Item Version (depression scale); GAD = General Anxiety Disorder Assessment; OQ = Outcome Questionnaire (symptom distress, interpersonal problem and social role measure); COM = Counselling Outcome Measure; PDQ = Psychological Distress Questionnaire (subscales: anxiety, depression, interpersonal difficulties). Findings: *Statistically significant. Quality Assessment: [the number of ‘Yes’ out of 14 items in total—the number of ‘Yes’ among items 7, 9, 11 and 12] (see Table 1). Table 4 Characteristics, findings and appraisals of studies using self-report adult attachment measurements Design and setting Population and sampling Attrition or response rate Relationship/intervention duration Workers’ attachment measure [measure point] Patients’ outcome measure [measure point] Statistical analysis and findings Quality assessment Relationship quality Health outcome Bruck et al. (2006)  Interrupted time series study  USA  A medical centre outpatient service 26 therapists psychologists psychiatrists social workers Convenience sampling 26 patients Recruited through advertisement 20 patients dropped out From 46 (43%) 30 sessions RSQ [pre-therapy] WAI-12 SEQ [6th session] TC GAS SCL IIP [post-therapy] Pearson correlation  RSQ not associated with WAI, SEQ or TC  (statistics not reported) Pearson correlation  RSQ(fearful) associated with   GAS r = −0.41 (P = 0.05)*  RSQ(preoccupied) associated with   GAS r = −0.35 (P = 0.05)*   SCL r = −0.30 (P = 0.05)*  RSQ(dismissing) associated with   IIP r = −0.30 (P < 0.05)*  All other correlations not significant (statistics not reported) Fair [7–2] Bucci et al. (2015)  Cross-sectional study  UK  Primary care psychology services 30 therapists psychologists practitioners counsellors other workers Convenience sampling 30 clients Recruited by therapists (next client at least 3 sessions) 12 clients out of 42 not answered (29%) At least 3 sessions RQ WAI-36 PHQ GAD Spearman correlation  RQ(secure) rho=−0.13 (P = 0.49)  RQ(fearful) rho=−0.17 (P = 0.38)  RQ(preoccupied) rho=−0.08 (P = 0.67)  RQ(dismissing) rho = 0.06 (P = 0.74)  All not associated with WAI Spearman correlation, clients split low/high (by median of PHQ + GAD) symptom  RQ(fearful) associated with WAI in high  symptom clients rho=−0.63 (P = 0.02)*  All other correlations not significant (statistics not reported) Fair [5–2] Kafetsios et al. (2016)  Cross-sectional study  Greece  Private and public system 40 doctors (various specialities) Sampling not clear 160 patients Sampling not clear 100% (No one refused) Average contact time = 34 months 27days RQ(converted into avoidant and anxious) PSQ Multilevel model  RQ(avoidant) not associated with PSQ  β = −0.006 (P > 0.10)  RQ(anxious) not associated with PSQ  β = −0.009 (P > 0.10) Fair [7–2] Dunkle and Friedlander (1996)  Cross-sectional survey 73 therapists Convenience sampling 73 clients Selected by therapists 29% (Data of 73 dyads/252 packets mailed) 3 to 5 sessions AAS WAI-12 Correlation  AAS(close) correlated with   WAI(total) r = 0.39 WAI(task) r = 0.34   WAI(bond) r = 0.41 WAI(goal) r = 0.32 (all P < 0.01)*  AAS(depend) correlated with   WAI(total) r = 0.42 WAI(task) r = 0.37   WAI(bond) r = 0.44 WAI(goal) r = 0.36 (all P < 0.01)*  AAS(anxiety) not correlated with   WAI(total) r = −0.16 WAI(task) r = −0.09   WAI(bond) r = −0.22 WAI(goal) r = −0.12 (all P > 0.05) Regression  AAS(close) associated with WAI(bond) β = 0.38 (P = 0.01)*  AAS(depend) not associated with WAI(bond)  β = 0.31 (P = 0.09)  AAS(anxiety) not associated with WAI(bond)  β = 0.16 (P = 0.37) Fair [6–3]  USA  15 university counselling centres, 6 clinics Wiseman and Tishby (2014)  Repeated measure design  Israel  A university counselling centre 27 therapists psychologists social workers Convenience sampling 54 clients Recruited at reception 13 clients dropped out from 67 (19%) 32 sessions (50-min session, once a week) ECR [pre-therapy] OQ-45 [pre, 5th, 15th, 23th, 28th session] CATS [5th, 15th, 28th session] Hierarchical mixed model  ECR(anxiety) associated with CATS at 5th  β = −0.23 (P = 0.05)*  All other association not significant  [ECR(anxiety)(avoidance) and CATS at 5th, 15th, 28th session] (statistics not reported)   Hierarchical linear model  ECR(anxiety) β = 0.12 (P > 0.05)  ECR(avoidance) β = −0.07  (P > 0.05) not associated with OQ change Good [9–4] Fuertes et al. (2007)  Cross-sectional survey  USA  Private settings, university counselling centres, hospitals, clinics 59 therapists Random sampling from American Counselling Association 59 clients selected by therapists 20% (Data of 59 dyads/300 therapists selected) At least 5 sessions (median = 16, n=30) ECR WAI-12 CATS RRI EUS COM Correlation  ECR(avoidance) associated with CATS(secure) r = −0.27 (P < 0.05)*  All other correlations not significant  [ECR(anxiety)(avoidance) and WAI, RRI,  CATS(secure)(avoidant/fearful)(preoccupied/merger), EUS] r = −0.44 to 0.05 (P not reported) Correlation  ECR(avoidance) correlated  with COM r = −0.45 (P < 0.001)*  ECR(anxiety) correlatedwith COM  r = −0.52 (P < 0.001)* Hierarchical regression  ECR(anxiety) associated  with COM β = −0.41 (P<0.002)* Fair [7–3] Wongpakaran and Wongpakaran (2012)  Cross-sectional study  Thailand  A psychiatric hospital outpatient setting 13 therapists psychiatrists residents Sampling not clear 121 patients Convenience sampling Not clear Not clear (each session 5 min to more than 1 h) ECR (secure or preoccupied) WAI-12 PDQ t-Test  WAI not differed by ECR (statistics not reported) MANOVA  ECR not associated with WAI  (statistics not reported) t-Test  PDQ(interpersonal difficulties) differed by   ECR(secure) 14.12 (SD = 4.20)   ECR(preoccupied) 12.74 (SD = 4.82) (P = 0.026)* ANOVA  ECR(secure) associated with   worse PDQ(anxiety) β = 8.95  (P = 0.001)*   worse PDQ(depression)  β = 8.25 (P = 0.002)*  ECR(secure) not associated  with PDQ (interpersonal difficulty) β = 2.02 (P=0.291) Poor [6–1] Very short relationship duration ECR cut-off not clear Berry and Greenwood (2016)  Cross-sectional study  UK  Community mental health service 33 care co-ordinators  24 nurses  5 occupational therapists  3 social workers  1 clinical psychologists Not reported 3 or more months PAM WAI-12 Correlation  PAM(anxious) not correlated  with WAI(total) r = −0.06 (P > 0.05)  PAM(avoidant) not correlated  with WAI(total) r = 0.04 (P > 0.05) Poor [6–2] Repeated data of therapists in analysis Sampling not clear 61 patients with psychosis Sampling not clear Design and setting Population and sampling Attrition or response rate Relationship/intervention duration Workers’ attachment measure [measure point] Patients’ outcome measure [measure point] Statistical analysis and findings Quality assessment Relationship quality Health outcome Bruck et al. (2006)  Interrupted time series study  USA  A medical centre outpatient service 26 therapists psychologists psychiatrists social workers Convenience sampling 26 patients Recruited through advertisement 20 patients dropped out From 46 (43%) 30 sessions RSQ [pre-therapy] WAI-12 SEQ [6th session] TC GAS SCL IIP [post-therapy] Pearson correlation  RSQ not associated with WAI, SEQ or TC  (statistics not reported) Pearson correlation  RSQ(fearful) associated with   GAS r = −0.41 (P = 0.05)*  RSQ(preoccupied) associated with   GAS r = −0.35 (P = 0.05)*   SCL r = −0.30 (P = 0.05)*  RSQ(dismissing) associated with   IIP r = −0.30 (P < 0.05)*  All other correlations not significant (statistics not reported) Fair [7–2] Bucci et al. (2015)  Cross-sectional study  UK  Primary care psychology services 30 therapists psychologists practitioners counsellors other workers Convenience sampling 30 clients Recruited by therapists (next client at least 3 sessions) 12 clients out of 42 not answered (29%) At least 3 sessions RQ WAI-36 PHQ GAD Spearman correlation  RQ(secure) rho=−0.13 (P = 0.49)  RQ(fearful) rho=−0.17 (P = 0.38)  RQ(preoccupied) rho=−0.08 (P = 0.67)  RQ(dismissing) rho = 0.06 (P = 0.74)  All not associated with WAI Spearman correlation, clients split low/high (by median of PHQ + GAD) symptom  RQ(fearful) associated with WAI in high  symptom clients rho=−0.63 (P = 0.02)*  All other correlations not significant (statistics not reported) Fair [5–2] Kafetsios et al. (2016)  Cross-sectional study  Greece  Private and public system 40 doctors (various specialities) Sampling not clear 160 patients Sampling not clear 100% (No one refused) Average contact time = 34 months 27days RQ(converted into avoidant and anxious) PSQ Multilevel model  RQ(avoidant) not associated with PSQ  β = −0.006 (P > 0.10)  RQ(anxious) not associated with PSQ  β = −0.009 (P > 0.10) Fair [7–2] Dunkle and Friedlander (1996)  Cross-sectional survey 73 therapists Convenience sampling 73 clients Selected by therapists 29% (Data of 73 dyads/252 packets mailed) 3 to 5 sessions AAS WAI-12 Correlation  AAS(close) correlated with   WAI(total) r = 0.39 WAI(task) r = 0.34   WAI(bond) r = 0.41 WAI(goal) r = 0.32 (all P < 0.01)*  AAS(depend) correlated with   WAI(total) r = 0.42 WAI(task) r = 0.37   WAI(bond) r = 0.44 WAI(goal) r = 0.36 (all P < 0.01)*  AAS(anxiety) not correlated with   WAI(total) r = −0.16 WAI(task) r = −0.09   WAI(bond) r = −0.22 WAI(goal) r = −0.12 (all P > 0.05) Regression  AAS(close) associated with WAI(bond) β = 0.38 (P = 0.01)*  AAS(depend) not associated with WAI(bond)  β = 0.31 (P = 0.09)  AAS(anxiety) not associated with WAI(bond)  β = 0.16 (P = 0.37) Fair [6–3]  USA  15 university counselling centres, 6 clinics Wiseman and Tishby (2014)  Repeated measure design  Israel  A university counselling centre 27 therapists psychologists social workers Convenience sampling 54 clients Recruited at reception 13 clients dropped out from 67 (19%) 32 sessions (50-min session, once a week) ECR [pre-therapy] OQ-45 [pre, 5th, 15th, 23th, 28th session] CATS [5th, 15th, 28th session] Hierarchical mixed model  ECR(anxiety) associated with CATS at 5th  β = −0.23 (P = 0.05)*  All other association not significant  [ECR(anxiety)(avoidance) and CATS at 5th, 15th, 28th session] (statistics not reported)   Hierarchical linear model  ECR(anxiety) β = 0.12 (P > 0.05)  ECR(avoidance) β = −0.07  (P > 0.05) not associated with OQ change Good [9–4] Fuertes et al. (2007)  Cross-sectional survey  USA  Private settings, university counselling centres, hospitals, clinics 59 therapists Random sampling from American Counselling Association 59 clients selected by therapists 20% (Data of 59 dyads/300 therapists selected) At least 5 sessions (median = 16, n=30) ECR WAI-12 CATS RRI EUS COM Correlation  ECR(avoidance) associated with CATS(secure) r = −0.27 (P < 0.05)*  All other correlations not significant  [ECR(anxiety)(avoidance) and WAI, RRI,  CATS(secure)(avoidant/fearful)(preoccupied/merger), EUS] r = −0.44 to 0.05 (P not reported) Correlation  ECR(avoidance) correlated  with COM r = −0.45 (P < 0.001)*  ECR(anxiety) correlatedwith COM  r = −0.52 (P < 0.001)* Hierarchical regression  ECR(anxiety) associated  with COM β = −0.41 (P<0.002)* Fair [7–3] Wongpakaran and Wongpakaran (2012)  Cross-sectional study  Thailand  A psychiatric hospital outpatient setting 13 therapists psychiatrists residents Sampling not clear 121 patients Convenience sampling Not clear Not clear (each session 5 min to more than 1 h) ECR (secure or preoccupied) WAI-12 PDQ t-Test  WAI not differed by ECR (statistics not reported) MANOVA  ECR not associated with WAI  (statistics not reported) t-Test  PDQ(interpersonal difficulties) differed by   ECR(secure) 14.12 (SD = 4.20)   ECR(preoccupied) 12.74 (SD = 4.82) (P = 0.026)* ANOVA  ECR(secure) associated with   worse PDQ(anxiety) β = 8.95  (P = 0.001)*   worse PDQ(depression)  β = 8.25 (P = 0.002)*  ECR(secure) not associated  with PDQ (interpersonal difficulty) β = 2.02 (P=0.291) Poor [6–1] Very short relationship duration ECR cut-off not clear Berry and Greenwood (2016)  Cross-sectional study  UK  Community mental health service 33 care co-ordinators  24 nurses  5 occupational therapists  3 social workers  1 clinical psychologists Not reported 3 or more months PAM WAI-12 Correlation  PAM(anxious) not correlated  with WAI(total) r = −0.06 (P > 0.05)  PAM(avoidant) not correlated  with WAI(total) r = 0.04 (P > 0.05) Poor [6–2] Repeated data of therapists in analysis Sampling not clear 61 patients with psychosis Sampling not clear Note. Healthcare workers’ attachment style measure: RSQ = Relationship Scale Questionnaire (subscale: secure, dismissing, preoccupied, fearful); RQ = Relationship Questionnaire (subscale: secure, dismissing, preoccupied, fearful); AAS = Adult Attachment Scale (subscale: depend, anxiety, close); ECR = Revised Experience of Close Relationship Questionnaire (subscale: anxiety, avoidance); PAM = Psychosis Attachment Measure (subscale: anxious, avoidant). Relationship quality measure: WAI-12 = Working Alliance Inventory 12-Item Version (subscales: bond, task, goal); SEQ = Session Evaluation Questionnaire; TC = Target Complaints; PSQ = Patient Satisfaction Questionnaire; WAI-36 = Working Alliance Inventory 36-Item Version; CATS = Client Attachment to Therapist Scale (subscale: secure, avoidant/fearful, preoccupied/merger); RRI = Real Relationship Inventory; EUS = Empathic Understanding Scale. Health outcome measure: GAS = Global Assessment Scale; SCL = Symptom Checklist 90 Revised; IIP = Inventory of Interpersonal Problems; PHQ = Patient Health Questionnaire 9-Item Version (depression scale); GAD = General Anxiety Disorder Assessment; OQ = Outcome Questionnaire (symptom distress, interpersonal problem and social role measure); COM = Counselling Outcome Measure; PDQ = Psychological Distress Questionnaire (subscales: anxiety, depression, interpersonal difficulties). Findings: *Statistically significant. Quality Assessment: [the number of ‘Yes’ out of 14 items in total—the number of ‘Yes’ among items 7, 9, 11 and 12] (see Table 1). Several studies analysed healthcare workers’ attachment measurements combined with other variables and such combinations differed between studies. Therefore, in this review, only estimates for which an attachment measurement was input as a categorical variable in comparative analyses or as a simple independent variable (not an interaction term) in correlational analyses were considered, which enabled us to summarize findings of interest reported across the studies reviewed. Meta-analytical synthesis was not attempted because of methodological heterogeneity across studies. A qualitative sensitivity analysis was additionally conducted (see below). Review of titles, abstracts and full-manuscripts, data extraction and quality assessment were conducted by the first author (C.M.). The results of this analysis were checked by the second author (I.J.N.) who also responded to any queries. Discrepancies were resolved through discussion. Results Overview of included studies Figure 1 summarizes the progress of papers through different stages of the review. A total of 13 papers [13–25] met criteria for inclusion, and were included in the final analysis (Tables 3 and 4). No randomized controlled trials or cohort studies were found. In total, 12 studies were conducted in mental healthcare settings, and one study in a general somatic health setting. Twelve studies used correlational analysis for examining the association of interest. One study used comparative analysis of t-test, MANOVA and ANOVA. Figure 1 View largeDownload slide Flow diagram of literature search. Figure 1 View largeDownload slide Flow diagram of literature search. Attachment style measurements Tables 3 and 4 summarize characteristics and findings of the five studies which gathered data using the AAI and other studies that used self-report attachment measurements. Eight studies used self-report methods, including the Experience of Close Relationship Questionnaire (ECR) [26], the Relationship Questionnaire (RQ) [5], the Relationship Scale Questionnaire (RSQ) [27], the Adult Attachment Scale (AAS) [28] and the Psychosis Attachment Measure (PAM) [29]. Patients’ perception and health outcome measurements A variety of tools were used to measure patients’ perceptions following their care, which were of two types: measures of relationship quality with healthcare workers; and measures of patients’ health outcomes. Thus, findings were reviewed for these two outcomes. The most widely used instrument of relationship quality was the 12-item Working Alliance Inventory (WAI-12) [30]. Other widely used tools were the 36-item Working Alliance Inventory (WAI-36) [31], the Client Attachment to Therapist Scale (CATS) [32] and the Helping Alliance Questionnaire (HAQ) [33]. Five other scales [34–38] were used, but each scale was not used in more than one study. Widely used tools to examine patients’ health outcomes included the Global Severity Index of the Symptom Check List (GSI) [39] and the Inventory of Interpersonal Problems (IIP) [40]. A variety of nine other measurements were used to collect outcome data [41–49]. Quality assessment This review was not concerned with evaluating the effectiveness of an intervention, thus standardization of the intervention was not a key issue. However, we recognize that the form of intervention could influence patients’ perceptions and health outcomes, and that standardizing interventions would control for this bias to some extent. Therefore, we considered interrupted time series studies as more robust than repeated measure study designs, and cross-sectional designs as the lowest level of evidence. Tables 3 and 4 summarize studies in this hierarchy of study designs grouped by use of similar attachment measurements. A frequent limitation of the studies was their small sample size, especially of healthcare workers. Only two studies used a sample of more than 50 of healthcare workers (mean = 30.6, SD = 18.4). Only one study reported a power calculation. Another limitation of the studies was that they reiterated healthcare workers’ attachment measurement in analysis in which the nested nature of patient data on healthcare worker data was not accounted for; this statistical problem was identified in three studies. In addition, the duration of clinical interpersonal relationship was relatively short in most studies. Finally, the selection of tools tended to be based on reliability/validity evaluated in previous research, and most studies (n = 9) did not sufficiently test the reliability/validity tools selected in their study setting. In using the quality assessment checklist, we regarded Items 7, 9, 11 and 12 of the checklist to be most important in assessing the quality of studies for the purpose of this review. However, we did not simply use the total number of ‘yes’ ratings among these four items, or among whole the set of items, to rate studies as good, fair or poor; rather we considered risk of selection bias, information bias, measurement bias or confounding. In addition, if the statistical analysis was assessed as flawed, then the study was rated as poor. One study was rated as good and six each as fair or poor. Findings Relationship quality In the studies which used the AAI for measuring healthcare workers’ attachment style (Table 3), one study found that at an early stage of treatment, patients seen by securely attached therapists rated their working alliance as poorer than those patients of insecurely attached therapists. However, approximately 3 months later the secure attachment group rated their working alliance more positively than the insecure attachment group [14]. On the other hand, statistically significant results indicated that the more healthcare workers were dismissing, the more their patients’ attachment to them were avoidant or fearful [13] and patients were less satisfied with the therapeutic relationship [15]. Within the studies using self-report (Table 4), two studies found that healthcare workers’ secure attachment was not associated with patients’ perception of working alliance [18, 19]. However, in one study, healthcare workers’ close (secure) attachment style was found to be associated with patients’ positive perception of their working alliance [21]. Statistically significant results showed that if healthcare workers were more avoidant [23] or anxious [22] patients believed their attachment to healthcare workers to be less secure. One study showed a positive relationship between dependence of healthcare workers and patients’ perception of the quality of their working alliance [21]. Patients’ health outcome Three studies using the AAI [13, 16, 17] explored the association between healthcare workers’ attachment style and patients’ health conditions, and all analyses produced statistically insignificant results (Table 3). As for studies using self-report attachment measurements (Table 4), four studies examined the association between attachment and patients’ health outcomes [18, 22–24]. Statistically significant relationships were found between insecure attachment style (fearful, pre-occupied and dismissing) and anxiety, somatic symptoms or interpersonal problems [18] and between insecure attachment style (anxiety and avoidant) and counselling outcome. In contrast, one study showed that patients treated by securely attached healthcare workers had worse interpersonal problems than those cared for by insecurely attached workers, and that secure attachment style of healthcare workers was associated with higher levels of anxiety and depression in patients [24]. Sensitivity analysis Of the 14 publications that could not be retrieved (Figure 1), abstracts were available for 12. One was found to report findings from the same study [50] as one of the studies included in the review [21], thus, it was excluded. Six out of 11 abstracts reported findings of interest [51–56]. As with the studies included in the review, study abstracts also reported mixed results. However, studies which investigated early stages of therapeutic relationships tended to show that healthcare workers’ attachment styles were not associated with patients’ outcomes [51, 53, 56] and that secure attachment styles of healthcare workers were associated with poorer patients’ outcomes [51]. One study indicated an association between healthcare workers’ insecure attachment styles and poorer patient ratings of alliance [56]. In addition, two studies reported a positive influence of securely attached workers on patients’ response/outcomes compared to that of insecurely attached workers [54, 55]. These findings are largely consistent with the conclusion of this review. Discussion Comparison between interviewer-report and self-report attachment measurements Studies using self-report measures appear to be more likely to detect a relationship between attachment styles and relationship quality or patients’ health outcomes than studies using the AAI and these relationships reported are generally statistically stronger (Tables 3 and 4). This difference might be explained by the different attachment constructs as described earlier. Self-report attachment measures study an entity which is similar to therapeutic working alliance. These findings support previous studies which conclude that the AAI and self-report attachment measures measure different aspects of attachment and are not interchangeable [57, 58]. Synthesis of overall findings This review shows that there is variation across studies regarding the association between attachment style of healthcare workers and patients’ outcome, and that associations, if any, are weak to moderate. Hence it is difficult to draw firm conclusions. This finding supports the conclusion of a recent systematic review on the role of psychotherapist attachment style in therapeutic alliance and outcome [7]. However, close inspection of the associations reported by studies included in this review shows that associations between secure attachment style and positive patients’ perception of relationship quality or health outcomes tend to be positive in direction, whereas associations between insecure dimensions of attachment style (dismissing, pre-occupied, fearful, avoidant, anxious and dependent) and patients’ perceptions of relationship quality or health outcomes tend to be negative in direction, irrespective of whether the AAI or self-report was used. Findings from the two studies that investigated change in patients’ response over time [14, 22] suggest that secure attachment styles of healthcare workers tend to have little or a negative effect on patients’ health outcomes or perceptions in the short term, but in the long term tend to have a more positive effect. In contrast, insecure styles tend to have a positive effect in the short term but little or a negative effect in long-term relationships. These findings are in line with Bartholomew and Horowitz’ model [5] of the self and others in relation to attachment style discussed earlier. Healthcare workers with positive self-perceptions (namely, secure or dismissing attachment styles) might appear superior and overbearing to patients at an early stage of the relationship, which may in turn contribute to a negative effect on patients’ health conditions or perceptions. In contrast, those healthcare workers with negative images of self (namely dependent, pre-occupied, fearful or anxious attachment styles) are likely to be valued by patients at the beginning of therapeutic contact because of the tendency of these health workers to invest highly in establishing a rapport with patients. However, healthcare workers hidden negative model of either themselves or others may present to patients in long-term relationships as unsure, inconsistent or distant attitudes, which might reduce patients’ feeling of safety and support. Practical implications It is now common practice for higher education institutions to screen applicants to healthcare programmes for empathy and positive attitudes towards patients which are considered prerequisites of high quality healthcare, although valid and reliable screening tools are in short supply. We are not aware of attempts to screen for attachment style when recruiting psychotherapy practitioners. Most theorists of adult attachment claim that attachment patterns in adulthood consist of generalized thoughts, feeling and expectations regulating the way that an individual engages in close relationships [59]. Attachment styles may have potential as valid predictors of the quality of interpersonal relationships between professional healthcare providers and patients, although more research would be needed to establish this. Knowledge of their attachment style may assist healthcare workers to enhance the quality of their care. In addition, healthcare workers may be able to modify attachment styles of patients in positive ways [60]. If patients who experienced rejecting and neglectful treatment in their childhood receive, as adults, responsive and supportive caregiving from healthcare workers, it seems possible that their internalized working model of relationships could be modified with resulting benefits for their health and wellbeing. Limitations This review retrieved a large number of papers, however, few papers were included in the analysis. Most of these studies were conducted in Western countries, which may induce cultural bias. The literature search sought only published or grey literature, which may have led to some selection bias. The healthcare workers and patients assessed were predominantly from mental healthcare settings. However, patients in physical care settings are usually treated by multiple healthcare workers and relationships between patients and professional carers may be relatively superficial. Kafetsios et al.’s study [20] suggests that in such cases, the influence of attachment style of care providers on patients’ outcomes and perceptions are likely to be attenuated and therefore difficult to detect. This review included only uncontrolled before–after studies and cross-sectional studies. We acknowledge the limitation of these research designs in evaluating cause and effect relationships. Future research One of the main limitations of studies conducted hitherto is use of a variety of measures and so constructs of attachment style and health outcomes, which make it difficult to compare the findings of different studies and so draw definitive conclusions. Surveying larger samples of the same number of healthcare workers and patients is also desirable because this would reduce the bias of repeated variable data. Future studies could also investigate the effect of attachment styles of healthcare workers on patients’ response over time rather than over short periods or at a single time point. Conclusion The results of this review are mixed, and there is currently no robust, statistically significant evidence of any relationship between attachment style of healthcare workers and patients’ health outcomes. It is uncertain whether this is due to the poor quality of the available evidence, or because there is no relationship, and so this question may not be a priority for future research. References 1 Shaver PR , Mikulincer M . An attachment-theory framework for conceptualizing interpersonal behavior. In: Horowitz LM , Strack S (eds) . Handbook of Interpersonal Psychology: Theory, Research, Assessment, and Therapeutic Interventions . New Jersey : John Wiley & Sons Inc , 2012 : 17 – 35 . Google Scholar CrossRef Search ADS 2 Bowlby J . Attachment and Loss Volume 1: Attachment . London : Pimlico , 1997 . 3 Bowlby J . Attachment and Loss Volume 2: Separation, Anger and Anxiety . London : Pimlico , 1998 . 4 Bowlby J . Attachment and Loss Volume 3: Loss, Sadness and Depression . London : Pimlico , 1998 . 5 Bartholomew K , Horowitz LM . Attachment styles among young adults: a test of four category model . J Pers Soc Psychol 1991 ; 61 : 226 – 44 . Google Scholar CrossRef Search ADS PubMed 6 Silberschatz G . Commentary: what have we learned about how the alliance develops over the course of therapy? Psychother Theory Res Pract Train 2009 ; 46 : 295 – 7 . Google Scholar CrossRef Search ADS 7 Degnan A , Seymour-Hyde A , Harris A et al. . The role of therapist attachment in alliance and outcome: a systematic literature review . Clin Physiol Psychother 2016 ; 23 : 47 – 65 . Google Scholar CrossRef Search ADS 8 George C , Kaplan N , Main M . Adult Attachment Interview . University of California , 1985 . 9 Hesse E . The Adult Attachment Interview: historical and current perspectives. In: Cassidy J , Shaver PR (eds) . Handbook of Attachment: Theory, Research, and Clinical Applications . New York : Guilford Press , 1999 : 395 – 433 . 10 Cassidy J , Shaver PR . Handbook of Attachment: Theory, Research and Clinical Applications . New York : Guilford Press , 1999 . 11 Dal Poz MR , Kinfu Y , Dräger S et al. . Counting Health Workers: Definitions, Data, Methods and Global Results. World Health Organization, 2007 . http://www.who.int/hrh/documents/counting_health_workers.pdf (September 2016, date last accessed). 12 National Heart, Lung and Blood Institute . Quality Assessment Tool for Observational Cohort and Cross-sectional Studies. U.S. Department of Health & Human Services, 2014 . http://www.nhlbi.nih.gov/guidelines/cvd_adult/tools/cohort.htm (September 2016, date last accessed). 13 Petrowski K , Pokorny D , Nowacki K et al. . The therapist’s attachment representation and the patient’s attachment to the therapist . Psychother Res 2013 ; 23 : 25 – 34 . Google Scholar CrossRef Search ADS PubMed 14 Sibrava NJ . Therapist and client attachment and the therapeutic alliance. Dissertation. The Graduate School College of the Liberal Arts, the Pennsylvania State University 2009 . 15 Petrowski K , Nowacki K , Pokorny D et al. . Matching the patient to the therapist: the roles of the attachment status and the helping alliance . J Nerv Ment Dis 2011 ; 199 : 839 – 44 . Google Scholar CrossRef Search ADS PubMed 16 Schauenburg H , Buchheim A , Beckh K et al. . The influence of psychodynamically oriented therapists’ attachment representations on outcome and alliance in inpatient psychotherapy . Psychother Res 2010 ; 20 : 193 – 202 . Google Scholar CrossRef Search ADS PubMed 17 Tyrrell CL , Dozier M , Teague GB et al. . Effective treatment relationships for persons with serious psychiatric disorders: the importance of attachment states of mind . J Consult Clin Psychol 1999 ; 67 : 725 – 33 . Google Scholar CrossRef Search ADS PubMed 18 Bruck E , Winston A , Aderholt S et al. . Predictive validity of patient and therapist attachment and introject styles . Am J Psychother 2006 ; 60 : 393 – 406 . Google Scholar CrossRef Search ADS PubMed 19 Bucci S , Seymour-Hyde A , Harris A et al. . Client and therapist attachment styles and working alliance . Clin Psychol Psychother 2015 ; 23 : 155 – 65 . Google Scholar CrossRef Search ADS PubMed 20 Kafetsios K , Hantzara K , Anagnostopoulos F et al. . Doctors’ attachment orientations, emotion regulation strategies, and patient satisfaction: a multilevel analysis . Health Commun 2016 ; 31 : 772 – 7 . Google Scholar CrossRef Search ADS PubMed 21 Dunkle JH , Friedlander ML . Contribution of therapist experience and personal characteristics to the working alliance . J Counseling Psychol 1996 ; 43 : 456 – 60 . Google Scholar CrossRef Search ADS 22 Wiseman H , Tishby O . Client attachment, attachment to the therapist and client-therapist attachment match: how do they relate to change in psychodynamic psychotherapy? Psychother Res 2014 ; 24 : 392 – 406 . Google Scholar CrossRef Search ADS PubMed 23 Fuertes JN , Mislowack A , Brown S et al. . Correlates of the real relationship in psychotherapy: a study of dyads . Psychother Res 2007 ; 17 : 423 – 30 . Google Scholar CrossRef Search ADS 24 Wongpakaran T , Wongpakaran N . How the interpersonal and attachment styles of therapists impact upon the therapeutic alliance and therapeutic outcomes . J Med Assoc Thai 2012 ; 95 : 1583 – 92 . Google Scholar PubMed 25 Berry C , Greenwood K . The relevance of professionals’ attachment style, expectations and job attitudes for therapeutic relationships with young people who experience psychosis . European Psychiatry 2016 ; 34 : 1 – 8 . Google Scholar CrossRef Search ADS PubMed 26 Brennan KA , Clark CL , Shaver PR . Self-report measurement of adult attachment: an integrative overview. In: Simpson JA , Rholes WS (eds) . Attachment Theory and Close Relationships . New York : Guilford Press , 1998 : 46 – 76 . 27 Griffin DW , Bartholomew K . The metaphysics of measurement: the case of adult attachment. In: Bartholomew K , Perlman D (eds) . Advances in Personal Relationships: Attachment Process in Adulthood 5 . London : Jessica Kingsley Publishers , 1994 : 17 – 52 . 28 Collins NL , Read SJ . Adult attachment, working models, and relationship quality in dating couples . J Pers Soc Psychol 1990 ; 58 : 644 – 63 . Google Scholar CrossRef Search ADS PubMed 29 Berry K , Wearden A , Barrowclough C et al. . Attachment styles, interpersonal relationships and psychotic phenomena in a non-clinical student sample . Pers Individ Dif 2006 ; 41 : 707 – 18 . Google Scholar CrossRef Search ADS 30 Tracey TJ , Kokotovic AM . Factor structure of the Working Alliance Inventory . Psychol Assessment 1989 ; 1 : 207 – 10 . Google Scholar CrossRef Search ADS 31 Horvath AO . Research on the alliance. In: Horvath A , Greenberg L (eds) . The Working Alliance: Theory, Research and Practice . New York : Wiley , 1994 . 32 Mallinckrodt B , Gantt DL , Goble HM . Attachment patterns in the psychotherapy relationship: development of the client attachment to therapist scale . J Counsel Psychol 1995 ; 42 : 307 – 17 . Google Scholar CrossRef Search ADS 33 Bassler M , Potratz B , Krauthauser H . The ‘Helping Alliance Questionnaire’ (HAQ) by Luborsky (Der ‘Helping Alliance Questionnaire’ HAQ von Luborsky) . Psychotherapeut 1995 ; 40 : 23 – 32 . 34 Stiles WB . Measurement of the impact of psychotherapy sessions . J Consult Clin Psychol 1994 ; 2 : 176 – 85 . 35 Battle CC , Imber SD , Hoehn-Saric R et al. . Target complaints as criteria of improvement . Am J Psychother 1996 ; 20 : 184 – 92 . Google Scholar CrossRef Search ADS 36 Grogan S , Conner M , Norman P et al. . Validation of a questionnaire measuring patient satisfaction with general practitioner services . Qual Health Care 2000 ; 9 : 210 – 5 . Google Scholar CrossRef Search ADS PubMed 37 Kelley FA , Gelso CJ , Fuertes JN et al. . The real relationship inventory: development and psychometric investigation of the client form . Psychotherapy 2010 ; 47 : 540 – 53 . Google Scholar CrossRef Search ADS PubMed 38 Barrett-Lennard GT . Dimensions of therapists response as causal factors in therapeutic change . Psychol Monog 1962 ; 76 : 1 – 36 . Google Scholar CrossRef Search ADS 39 Franke G . SCL-R-90. The Symptom-Checklist by Derogatis . Weinheim : Beltz , 1995 . 40 Horowitz L , Strauss B , Kordy H . The Inventory of Interpersonal Problems IIP (Das Inventar zur Erfassung interpersoneller Probleme—IIP) . Weinheim, Germany : Beltz , 2000 . 41 Schepank H . The impairment score (Der Beeintraechtigungssschwere-Score). In: Goettingen . Germany : Beltz , 1995 . 42 Lehman AF . A quality life of interview for the chronically mentally ill . Eval Program Plann 1988 ; 11 : 51 – 62 . Google Scholar CrossRef Search ADS 43 American Psychiatric Association . Diagnostic and Statistical Manual of Mental Disorders , 4th edn . Washington, DC : Author , 1994 . PubMed PubMed 44 Beck AT , Ward CH , Mendelson M et al. . An inventory for measuring depression . Arch Gen Psychiatry 1961 ; 4 : 53 – 63 . Google Scholar CrossRef Search ADS 45 Endicott J , Spitzer RL , Fleiss JL et al. . The Global Assessment Scale: a procedure for measuring overall severity of psychiatric disturbance . Arch Gen Psychiatry 1976 ; 33 : 766 – 71 . Google Scholar CrossRef Search ADS PubMed 46 Derogatis LR . SCL-90-R: Administration, Scoring and Procedures Manual II . Towson, MD : Clinical Psychiatric Research , 1983 . 47 Lambert M , Burlingame GM , Umphress V et al. . The reliability and validity of the Outcome Questionnaire . Clin Psychol Psychother 1996 ; 3 : 249 – 58 . Google Scholar CrossRef Search ADS 48 Gelso CJ , Johnson DH . Explorations in Time-limited Psychotherapy and Psychotherapy . New York : Teachers College Press, Columbia University , 1983 . 49 Wongpakaran T , Wongpakaran N . Development and Validation of Psychological Distress Questionnaire . Chiang Mai : Department of Psychiatry, Chiang Mai University, Thailand , 2011 . 50 Dunkle JH . The contribution of therapists’ personal and professional characteristics to the strength of the therapeutic alliance . Diss Abstr Int 1996 ; 56 : 5763 . 51 Biscoglio RL . Patient and therapist personality, therapeutic alliance, and overall outcome in Brief Relational Therapy . Diss Abstr Int 2005 ; 65 : 6642 . 52 Britton PC . The alliance and adult attachment . Diss Abstr Int 2006 ; 67 : 531 . 53 Frehling R . The relationship of therapist and client: attachment styles to measures of the working alliance . Diss Abstr Int 2005 ; 66 : 2818 . 54 Mahoney KSB . Effects of client satisfaction and attachment styles in the counseling dyad on therapy continuation . Diss Abstr Int 2008 ; 68 : 6972 . 55 Niederhauser RK . Adult attachment styles in primary care clinics for the medically underserved . Diss Abstr Int 2001 ; 61 : 6716 . 56 Sack A . The influence of therapist attachment style on the strength of the therapeutic alliance . Diss Abstr Int 1997 ; 57 : 5343 . 57 Roisman GI , Holland A , Fortuna K et al. . The Adult Attachment Interview and self-reports of attachment style: an empirical rapprochement . J Pers Soc Psychol 2007 ; 92 : 678 – 97 . Google Scholar CrossRef Search ADS PubMed 58 Kirchmann H , Mestel R , Schreiber-Willnow K et al. . Associations among attachment characteristics, patients’ assessment of therapeutic factors, and treatment outcome following inpatient psychodynamic group psychotherapy . Psychother Res 2009 ; 19 : 234 – 48 . Google Scholar CrossRef Search ADS PubMed 59 Rholes WS , Simpson JA . Adult Attachment: Theory, Research, and Clinical Implications . New York : Guilford Publications , 2004 . 60 Bowlby J . A Secure Base: Parent-Child Attachments and Healthy Human Development . New York : Basic Books , 2008 . © The Author(s) 2018. Published by Oxford University Press in association with the International Society for Quality in Health Care. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png International Journal for Quality in Health Care Oxford University Press

The relationship between healthcare workers’ attachment styles and patient outcomes: a systematic review

Loading next page...
 
/lp/ou_press/the-relationship-between-healthcare-workers-attachment-styles-and-yjzRREL7aF
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press in association with the International Society for Quality in Health Care. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
1353-4505
eISSN
1464-3677
D.O.I.
10.1093/intqhc/mzy034
Publisher site
See Article on Publisher Site

Abstract

Abstract Purpose To examine whether attachment styles of healthcare workers influence the quality of their relationships with patients, or impact patients’ health outcomes. Data source Literature database searches on the CINAHL, Cochrane Library, Embase, MEDLINE and PsyCinfo, and hand searching of reference lists of the retained articles. Study selection Original empirical studies reporting an analysis of the relationship of interest were selected for review. Data extraction Estimates of association between healthcare workers’ attachment style and patients’ health outcomes were extracted. Results of data synthesis Results from 13 studies were mixed in terms of which attachment styles related to patients’ perceptions of care or health outcomes, and the evidence overall was of poor quality and methodologically heterogeneous. However, there is limited evidence that secure attachment styles of healthcare workers have little or a negative effect on patients’ health outcomes or perceptions in the short term but in the long term have a more positive effect. Conversely, insecure styles tend to have a positive effect in the short term but little or a negative effect on long-term relationships. Studies which used self-report attachment measurements tended to report stronger associations with patients’ outcome measurements than studies using the interviewer rated Adult Attachment Interview. Conclusion It is unclear whether or not there is a relationship between attachment style of health workers and patients’ health outcomes. Further research using consistent data collection tools, especially in relation to the attachment measurement construct selected, and analysis methods across studies is required to draw recommendations for clinical practice. attachment style, psychological bonding, healthcare, outcome assessment, professional patient relationship Introduction The quality of the interpersonal relationships between healthcare workers and patients is an important process of care variable and may influence patients’ outcomes since it may influence patients’ compliance with treatment. Previous studies have demonstrated that, among a wide variety of influential factors, attachment styles are important antecedents of interpersonal relationship quality [1]. Building on Bowlby’s attachment theory that conceptualizes attachment styles as secure or insecure [2–4], Bartholomew and Horowitz [5] proposed an attachment style model comprising intersecting dimensions of positive and negative perceptions of the self and others. They proposed four attachment styles: secure, dismissing, pre-occupied and fearful. People with secure attachment have positive regard for both themselves and others. In contrast, individuals with a dismissing attachment style have a positive self-image but a negative image of others. Those with a pre-occupied attachment style have a negative self-image but are positive towards others. People with a fearful attachment style perceive both themselves and others as unworthy and unlikable. These theories suggest that attachment styles of both the healthcare provider and their patients can influence the quality of clinician–patient relationships. However, research on the association between patient attachment styles and the patient–clinician relationship has been criticized for not having clinical applications [6]. A more productive approach might be to consider the attachment styles of healthcare workers as a potential prerequisite of the quality of patient care. A recent systematic review investigating the impact of therapist attachment style on psychotherapeutic alliance and outcomes, found diverse findings across studies but also some evidence that therapist attachment style is a predictor of the therapeutic alliance and patients’ health outcomes [7]. The present review builds on this finding by investigating the relationship between the attachment style of a wider group of healthcare professionals and the quality of patient care. To assess adult attachment styles in research, a variety of tools have been developed. Some of these involve data collection by interviewer-rating assessment and others through self-report. Amongst the first group is the Adult Attachment Interview (AAI) [8], which focuses on how interviewees coherently organize their speech with respect to childhood experiences with significant caregivers irrespective of whether they as adults perceive such loving relationships as positive or negative [9]. In contrast, self-report measures ask respondents to report how they think and feel about current close relationships [10]. This review investigates whether there is an association between healthcare workers’ attachment styles, measured with the AAI or self-rating tools, and patients’ health outcomes or their perception of care provided by their healthcare workers. Methods Literature search An electronic literature search was performed in February 2017 on the following databases: CINAHL, Cochrane Library, Embase (1974-), MEDLINE (1946-) and PsyCinfo (1806-). The search strategy is summarized in Table 1. Table 1 Database literature searching strategy Keywords Population facet Measurement facet doctor psychiatrist nurse practitioner clinician therapist psychologist counsellor health worker healthcare worker health professional healthcare professional attachment Adult Attachment Interview Adult Attachment Scale Attachment Style Questionnaire Adult Attachment Style Measure Experience in Close Relationship Scale Self-Reliance Inventory Strategy Keywords were truncated where applicable and MeSH terms were also used including all subheadings where available. Keywords and terms were combined with the Boolean operator OR within each facet. These combinations of the two facets were further combined with the Boolean operator AND. Studies were limited to those using humans. Keywords Population facet Measurement facet doctor psychiatrist nurse practitioner clinician therapist psychologist counsellor health worker healthcare worker health professional healthcare professional attachment Adult Attachment Interview Adult Attachment Scale Attachment Style Questionnaire Adult Attachment Style Measure Experience in Close Relationship Scale Self-Reliance Inventory Strategy Keywords were truncated where applicable and MeSH terms were also used including all subheadings where available. Keywords and terms were combined with the Boolean operator OR within each facet. These combinations of the two facets were further combined with the Boolean operator AND. Studies were limited to those using humans. Table 1 Database literature searching strategy Keywords Population facet Measurement facet doctor psychiatrist nurse practitioner clinician therapist psychologist counsellor health worker healthcare worker health professional healthcare professional attachment Adult Attachment Interview Adult Attachment Scale Attachment Style Questionnaire Adult Attachment Style Measure Experience in Close Relationship Scale Self-Reliance Inventory Strategy Keywords were truncated where applicable and MeSH terms were also used including all subheadings where available. Keywords and terms were combined with the Boolean operator OR within each facet. These combinations of the two facets were further combined with the Boolean operator AND. Studies were limited to those using humans. Keywords Population facet Measurement facet doctor psychiatrist nurse practitioner clinician therapist psychologist counsellor health worker healthcare worker health professional healthcare professional attachment Adult Attachment Interview Adult Attachment Scale Attachment Style Questionnaire Adult Attachment Style Measure Experience in Close Relationship Scale Self-Reliance Inventory Strategy Keywords were truncated where applicable and MeSH terms were also used including all subheadings where available. Keywords and terms were combined with the Boolean operator OR within each facet. These combinations of the two facets were further combined with the Boolean operator AND. Studies were limited to those using humans. The titles and abstracts of all the papers identified were screened to ensure they met the following inclusion criteria: (a) Published in English or Japanese language. (b) Report the findings of empirical research studies conducted within a healthcare setting examining attachment styles of any healthcare workers using any measurement tool based on Bowlby’s attachment theory. (c) Report patients’ somatic and/or psychological health response to care using a measure/indicator. To be included in the review, healthcare workers had to be defined by the study authors as people whose paid employment involved promotion, protection, prevention, curation or improvement of the health of the population [11] in settings where the primary purpose is the treatment and prevention of disease and illness. In addition to the electronic literature searches, hand searching was conducted by scanning reference lists of retained articles, relevant review papers and bibliographies. Full texts of any potentially relevant studies were sought, and if they could not be obtained, information on the study was sought from the corresponding author by email. Retrieved full-text papers were excluded if they met any of the following criteria: (d) Studies using a simulated intervention and (e) Studies in which more than 50% of the healthcare providers were pre-registration students. These studies were excluded because students are always supervised by senior professionals and their ways of relating patients are likely to be influenced by the supervisor, so the effects of their own attachment styles might be attenuated. Quality assessment We used a standardized quality assessment checklist, the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies by the National Heart, Lung and Blood Institute [12], and modified the phrases to make them applicable to this review (Table 2). Table 2 Quality assessment checklist: a modification of Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies (NHLBI 2014) Item Yes C/D, N/A or N/R* No 1 Was the research question/objective clearly stated? 2 Was the study population (healthcare workers and patients/clients) clearly specified and defined? 3 Was the participation rate of eligible persons at least 50%? 4 Were all the subjects selected or recruited from the same or similar populations (including the same time period)? Were inclusion/exclusion criteria pre-specified and applied uniformly to all participants? 5 Was a sample size justification provided? 6 For the analyses, was the independent variable (attachment representation of healthcare workers) measured prior to the outcome (response of patients/clients) being measured? 7 Was the timeframe of interpersonal relationship between healthcare workers and patients/clients sufficient so that one could reasonably expect to see an association between the independent variable (attachment representation of healthcare workers) and the outcome (response of patients/clients) if it existed? 8 Did the study examine different status/levels of the attachment representation of healthcare workers as related to the responses of patients/clients (e.g. categories of attachment representation, or attachment representation measured as continuous variable)? 9 Was the attachment measure clearly defined, valid/reliable (validity/reliability in the present study was sufficiently reported), and implemented consistently across all healthcare workers? 10 Was the intervention, if any, standardized and assessed/checked? 11 Were the measures of response of patients/clients clearly defined, valid/reliable (validity/reliability in the present study was sufficiently reported), and implemented consistently across all patients/clients? 12 Were the outcome (response of patients/clients) assessors blinded/independent to the outcome status (attachment representation of healthcare workers)? 13 Was loss to follow-up after baseline 20% or less? 14 Were key potential confounding variables measured and adjusted statistically for their impact on the relationship between the independent variable (attachment representation of healthcare workers) and the outcome (response of patients/clients)? Item Yes C/D, N/A or N/R* No 1 Was the research question/objective clearly stated? 2 Was the study population (healthcare workers and patients/clients) clearly specified and defined? 3 Was the participation rate of eligible persons at least 50%? 4 Were all the subjects selected or recruited from the same or similar populations (including the same time period)? Were inclusion/exclusion criteria pre-specified and applied uniformly to all participants? 5 Was a sample size justification provided? 6 For the analyses, was the independent variable (attachment representation of healthcare workers) measured prior to the outcome (response of patients/clients) being measured? 7 Was the timeframe of interpersonal relationship between healthcare workers and patients/clients sufficient so that one could reasonably expect to see an association between the independent variable (attachment representation of healthcare workers) and the outcome (response of patients/clients) if it existed? 8 Did the study examine different status/levels of the attachment representation of healthcare workers as related to the responses of patients/clients (e.g. categories of attachment representation, or attachment representation measured as continuous variable)? 9 Was the attachment measure clearly defined, valid/reliable (validity/reliability in the present study was sufficiently reported), and implemented consistently across all healthcare workers? 10 Was the intervention, if any, standardized and assessed/checked? 11 Were the measures of response of patients/clients clearly defined, valid/reliable (validity/reliability in the present study was sufficiently reported), and implemented consistently across all patients/clients? 12 Were the outcome (response of patients/clients) assessors blinded/independent to the outcome status (attachment representation of healthcare workers)? 13 Was loss to follow-up after baseline 20% or less? 14 Were key potential confounding variables measured and adjusted statistically for their impact on the relationship between the independent variable (attachment representation of healthcare workers) and the outcome (response of patients/clients)? Overall quality: good fair poor Overall quality: good fair poor Note. *C/D = cannot determine; N/A = not applicable; N/R = not reported. Table 2 Quality assessment checklist: a modification of Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies (NHLBI 2014) Item Yes C/D, N/A or N/R* No 1 Was the research question/objective clearly stated? 2 Was the study population (healthcare workers and patients/clients) clearly specified and defined? 3 Was the participation rate of eligible persons at least 50%? 4 Were all the subjects selected or recruited from the same or similar populations (including the same time period)? Were inclusion/exclusion criteria pre-specified and applied uniformly to all participants? 5 Was a sample size justification provided? 6 For the analyses, was the independent variable (attachment representation of healthcare workers) measured prior to the outcome (response of patients/clients) being measured? 7 Was the timeframe of interpersonal relationship between healthcare workers and patients/clients sufficient so that one could reasonably expect to see an association between the independent variable (attachment representation of healthcare workers) and the outcome (response of patients/clients) if it existed? 8 Did the study examine different status/levels of the attachment representation of healthcare workers as related to the responses of patients/clients (e.g. categories of attachment representation, or attachment representation measured as continuous variable)? 9 Was the attachment measure clearly defined, valid/reliable (validity/reliability in the present study was sufficiently reported), and implemented consistently across all healthcare workers? 10 Was the intervention, if any, standardized and assessed/checked? 11 Were the measures of response of patients/clients clearly defined, valid/reliable (validity/reliability in the present study was sufficiently reported), and implemented consistently across all patients/clients? 12 Were the outcome (response of patients/clients) assessors blinded/independent to the outcome status (attachment representation of healthcare workers)? 13 Was loss to follow-up after baseline 20% or less? 14 Were key potential confounding variables measured and adjusted statistically for their impact on the relationship between the independent variable (attachment representation of healthcare workers) and the outcome (response of patients/clients)? Item Yes C/D, N/A or N/R* No 1 Was the research question/objective clearly stated? 2 Was the study population (healthcare workers and patients/clients) clearly specified and defined? 3 Was the participation rate of eligible persons at least 50%? 4 Were all the subjects selected or recruited from the same or similar populations (including the same time period)? Were inclusion/exclusion criteria pre-specified and applied uniformly to all participants? 5 Was a sample size justification provided? 6 For the analyses, was the independent variable (attachment representation of healthcare workers) measured prior to the outcome (response of patients/clients) being measured? 7 Was the timeframe of interpersonal relationship between healthcare workers and patients/clients sufficient so that one could reasonably expect to see an association between the independent variable (attachment representation of healthcare workers) and the outcome (response of patients/clients) if it existed? 8 Did the study examine different status/levels of the attachment representation of healthcare workers as related to the responses of patients/clients (e.g. categories of attachment representation, or attachment representation measured as continuous variable)? 9 Was the attachment measure clearly defined, valid/reliable (validity/reliability in the present study was sufficiently reported), and implemented consistently across all healthcare workers? 10 Was the intervention, if any, standardized and assessed/checked? 11 Were the measures of response of patients/clients clearly defined, valid/reliable (validity/reliability in the present study was sufficiently reported), and implemented consistently across all patients/clients? 12 Were the outcome (response of patients/clients) assessors blinded/independent to the outcome status (attachment representation of healthcare workers)? 13 Was loss to follow-up after baseline 20% or less? 14 Were key potential confounding variables measured and adjusted statistically for their impact on the relationship between the independent variable (attachment representation of healthcare workers) and the outcome (response of patients/clients)? Overall quality: good fair poor Overall quality: good fair poor Note. *C/D = cannot determine; N/A = not applicable; N/R = not reported. Data extraction and analysis A pre-determined data extraction sheet was used, comprising items as listed in Tables 3 and 4. The data needed for assessing the quality of studies (Table 2) were also extracted. Table 3 Characteristics, findings and appraisals of studies using Adult Attachment Interview Design and setting Population and sampling Attrition or response rate Relationship/ intervention duration Workers’ attachment measure [measure point] Patients’ outcome measure [measure point] Statistical analysis and findings Quality assessment Relationship quality Health outcome Petrowski et al. (2013)  Repeated measure design  Germany  A university hospital inpatient setting 22 therapists  15 psychologists  7 physicians Sampling not clear Not reported 12 sessions (average 62.5 days, SD = 28.45) AAI [pre-therapy] GSI CATS [post-therapy] Hierarchical linear model  AAI(secure–insecure) not associated with  CATS (secure) β not reported (P > 0.05)  AAI(dismissing–preoccupied) not  associated with CATS(preoccupied/merger) β = −0.88 (P = 0.06)  AAI(dismissing–preoccupied) associated  with CATS (avoidant/fearful) β = 0.94 (P = 0.03)* Pearson correlation AAI (secure–insecure) not correlated with GSI r = −0.06 (P = 0.07) AAI(dismissing–preoccupied) not correlated with GSI r = −0.12 (P = 0.51) Fair [7–2] 429 patients Sampling not clear (randomly assigned to therapists) Sibrava (2009)  Repeated measure  design 4 therapists Sampling not clear Not reported 14 Weekly sessions (1.5–2 h) AAI [pre-therapy] WAI-36 [end of 2, 5, 10, 14 sessions] Hierarchical linear model  AAI associated with WAI [at session 2  AAI(secure) showed lower WAI than  AAI(insecure), at session 14  AAI(secure) showed higher WAI than  AAI(insecure)] β = 0.62 (P < 0.05)* Poor [6–2] Small sample size in therapists  Not clear 69 clients Sampling not clear Petrowski et al. (2011)  Repeated measure design  Germany  A university hospital inpatient setting 19 therapists  13 psychologists  6 physicians  Sampling not clear Not reported Not clear AAI [pre-therapy] HAQ [post-therapy] Hierarchical regression  AAI(secure–insecure) not associated with   HAQ (statistics not reported)  AAI (dismissing–preoccupied) associated   with HAQ (relationship) β = −0.65  (P = 0.02)*   with HAQ(outcome) β = −0.65 (P <  0.01)*  In all other 7 models, AAI (secure– insecure) or AAI (dismissing–preoccupied) not associated with HAQ (relationship) or HAQ (outcome) (statistics not reported) Poor [5–1] Repeated data of therapists in analysis 59 patients  Sampling not clear  (randomly assigned to therapists) Schauenburg et al. (2010)  Repeated measure design  Germany  2 psychotherapy hospitals, inpatient settings 31 therapists  27 physicians  4 psychologists  Sampling not clear Not reported Mean treatment duration 12.01 weeks (SD = 5.4) AAI [pre-therapy] GSI IIP IS [pre-, post-therapy] HAQ [post-therapy] Hierarchical linear model AAI(secure–insecure) not associated with  HAQ β = 0.07 (P = 0.53) AAI(dismissing–preoccupied) not associated with HAQ β = 0.10 (P = 0.66) Hierarchical linear model AAI(secure–insecure) not associated with GSI  β = 0.07  IIP  β = 0.07  IS  β = −0.22 (all P > 0.05) AAI(dismissing–preoccupied) not associated with GSI  β = −0.01 IIP  β = −0.06 IS  β = −0.14 (all P > 0.05) Poor [5–1] 1381 patients assigned to therapists in consecutive order  Sampling not clear Tyrrel et al. (1999)  Cross-sectional study  USA  A community-based setting 21 case managers  Sampling not clea 54 clients  Sampling not clear Not reported Mean dyad relationship 31 months (SD = 17, range=7–69) AAI WAI-36 QLI GAF BDI Psychiatric hospitalization days in previous 1 year Correlation AAI(dismissing–preoccupied) not correlated with WAI r = 0.07 (P not reported) Hierarchical regression AAI(dismissing–preoccupied) not associated with WAI β = 0.04 (P not reported) Correlation  AAI(dismissing-preoccupied) not correlated with  QLI(global) r = 0.02  QLI(relation-q) r = -0.08  QLI(relation-s) r = 0.10  GAF r = 0.23  BDI r = 0.10  Hospitalization r = 0.06 (all p not reported) Poor [5–2] Repeated data of therapists in analysis Hierarchical regression  AAI(dismissing-preoccupied) not associated with  QLI(global) β = −0.03  QLI(relation-q) β = −0.17  QLI(relation-s) β = 0.02  GAF β = 0.20  BDI β = 0.10  Hospitalization β = 0.01 (all p not reported) Design and setting Population and sampling Attrition or response rate Relationship/ intervention duration Workers’ attachment measure [measure point] Patients’ outcome measure [measure point] Statistical analysis and findings Quality assessment Relationship quality Health outcome Petrowski et al. (2013)  Repeated measure design  Germany  A university hospital inpatient setting 22 therapists  15 psychologists  7 physicians Sampling not clear Not reported 12 sessions (average 62.5 days, SD = 28.45) AAI [pre-therapy] GSI CATS [post-therapy] Hierarchical linear model  AAI(secure–insecure) not associated with  CATS (secure) β not reported (P > 0.05)  AAI(dismissing–preoccupied) not  associated with CATS(preoccupied/merger) β = −0.88 (P = 0.06)  AAI(dismissing–preoccupied) associated  with CATS (avoidant/fearful) β = 0.94 (P = 0.03)* Pearson correlation AAI (secure–insecure) not correlated with GSI r = −0.06 (P = 0.07) AAI(dismissing–preoccupied) not correlated with GSI r = −0.12 (P = 0.51) Fair [7–2] 429 patients Sampling not clear (randomly assigned to therapists) Sibrava (2009)  Repeated measure  design 4 therapists Sampling not clear Not reported 14 Weekly sessions (1.5–2 h) AAI [pre-therapy] WAI-36 [end of 2, 5, 10, 14 sessions] Hierarchical linear model  AAI associated with WAI [at session 2  AAI(secure) showed lower WAI than  AAI(insecure), at session 14  AAI(secure) showed higher WAI than  AAI(insecure)] β = 0.62 (P < 0.05)* Poor [6–2] Small sample size in therapists  Not clear 69 clients Sampling not clear Petrowski et al. (2011)  Repeated measure design  Germany  A university hospital inpatient setting 19 therapists  13 psychologists  6 physicians  Sampling not clear Not reported Not clear AAI [pre-therapy] HAQ [post-therapy] Hierarchical regression  AAI(secure–insecure) not associated with   HAQ (statistics not reported)  AAI (dismissing–preoccupied) associated   with HAQ (relationship) β = −0.65  (P = 0.02)*   with HAQ(outcome) β = −0.65 (P <  0.01)*  In all other 7 models, AAI (secure– insecure) or AAI (dismissing–preoccupied) not associated with HAQ (relationship) or HAQ (outcome) (statistics not reported) Poor [5–1] Repeated data of therapists in analysis 59 patients  Sampling not clear  (randomly assigned to therapists) Schauenburg et al. (2010)  Repeated measure design  Germany  2 psychotherapy hospitals, inpatient settings 31 therapists  27 physicians  4 psychologists  Sampling not clear Not reported Mean treatment duration 12.01 weeks (SD = 5.4) AAI [pre-therapy] GSI IIP IS [pre-, post-therapy] HAQ [post-therapy] Hierarchical linear model AAI(secure–insecure) not associated with  HAQ β = 0.07 (P = 0.53) AAI(dismissing–preoccupied) not associated with HAQ β = 0.10 (P = 0.66) Hierarchical linear model AAI(secure–insecure) not associated with GSI  β = 0.07  IIP  β = 0.07  IS  β = −0.22 (all P > 0.05) AAI(dismissing–preoccupied) not associated with GSI  β = −0.01 IIP  β = −0.06 IS  β = −0.14 (all P > 0.05) Poor [5–1] 1381 patients assigned to therapists in consecutive order  Sampling not clear Tyrrel et al. (1999)  Cross-sectional study  USA  A community-based setting 21 case managers  Sampling not clea 54 clients  Sampling not clear Not reported Mean dyad relationship 31 months (SD = 17, range=7–69) AAI WAI-36 QLI GAF BDI Psychiatric hospitalization days in previous 1 year Correlation AAI(dismissing–preoccupied) not correlated with WAI r = 0.07 (P not reported) Hierarchical regression AAI(dismissing–preoccupied) not associated with WAI β = 0.04 (P not reported) Correlation  AAI(dismissing-preoccupied) not correlated with  QLI(global) r = 0.02  QLI(relation-q) r = -0.08  QLI(relation-s) r = 0.10  GAF r = 0.23  BDI r = 0.10  Hospitalization r = 0.06 (all p not reported) Poor [5–2] Repeated data of therapists in analysis Hierarchical regression  AAI(dismissing-preoccupied) not associated with  QLI(global) β = −0.03  QLI(relation-q) β = −0.17  QLI(relation-s) β = 0.02  GAF β = 0.20  BDI β = 0.10  Hospitalization β = 0.01 (all p not reported) Note. Healthcare workers’ attachment style measure: AAI = Adult Attachment Interview (category: secure, dismissing, preoccupied, unresolved) (score: secure–insecure, dismissing–preoccupied). Relationship quality measure: CATS = Client Attachment to Therapist Scale (subscale: secure, avoidant/fearful, preoccupied/merger); WAI-36 = Working Alliance Inventory 36-Item Version; HAQ = Helping Alliance Questionnaire (subscales: relationship satisfaction, outcome satisfaction). Health outcome measure: GSI = Global Severity Index of the Symptom Checklist; IIP = Inventory of Interpersonal Problems; IS = Impairment Score, caregiver-rating; QLI = Quality of Life Interview (subscale: global life satisfaction, relationship quality, relationship satisfaction); GAF = Global Assessment of Functioning; BDI = 13-item Beck Depression Inventory. Findings: *Statistically significant. Quality Assessment: [the number of ‘Yes’ out of 14 items in total—the number of ‘Yes’ among items 7, 9, 11 and 12] (see Table 1). Table 3 Characteristics, findings and appraisals of studies using Adult Attachment Interview Design and setting Population and sampling Attrition or response rate Relationship/ intervention duration Workers’ attachment measure [measure point] Patients’ outcome measure [measure point] Statistical analysis and findings Quality assessment Relationship quality Health outcome Petrowski et al. (2013)  Repeated measure design  Germany  A university hospital inpatient setting 22 therapists  15 psychologists  7 physicians Sampling not clear Not reported 12 sessions (average 62.5 days, SD = 28.45) AAI [pre-therapy] GSI CATS [post-therapy] Hierarchical linear model  AAI(secure–insecure) not associated with  CATS (secure) β not reported (P > 0.05)  AAI(dismissing–preoccupied) not  associated with CATS(preoccupied/merger) β = −0.88 (P = 0.06)  AAI(dismissing–preoccupied) associated  with CATS (avoidant/fearful) β = 0.94 (P = 0.03)* Pearson correlation AAI (secure–insecure) not correlated with GSI r = −0.06 (P = 0.07) AAI(dismissing–preoccupied) not correlated with GSI r = −0.12 (P = 0.51) Fair [7–2] 429 patients Sampling not clear (randomly assigned to therapists) Sibrava (2009)  Repeated measure  design 4 therapists Sampling not clear Not reported 14 Weekly sessions (1.5–2 h) AAI [pre-therapy] WAI-36 [end of 2, 5, 10, 14 sessions] Hierarchical linear model  AAI associated with WAI [at session 2  AAI(secure) showed lower WAI than  AAI(insecure), at session 14  AAI(secure) showed higher WAI than  AAI(insecure)] β = 0.62 (P < 0.05)* Poor [6–2] Small sample size in therapists  Not clear 69 clients Sampling not clear Petrowski et al. (2011)  Repeated measure design  Germany  A university hospital inpatient setting 19 therapists  13 psychologists  6 physicians  Sampling not clear Not reported Not clear AAI [pre-therapy] HAQ [post-therapy] Hierarchical regression  AAI(secure–insecure) not associated with   HAQ (statistics not reported)  AAI (dismissing–preoccupied) associated   with HAQ (relationship) β = −0.65  (P = 0.02)*   with HAQ(outcome) β = −0.65 (P <  0.01)*  In all other 7 models, AAI (secure– insecure) or AAI (dismissing–preoccupied) not associated with HAQ (relationship) or HAQ (outcome) (statistics not reported) Poor [5–1] Repeated data of therapists in analysis 59 patients  Sampling not clear  (randomly assigned to therapists) Schauenburg et al. (2010)  Repeated measure design  Germany  2 psychotherapy hospitals, inpatient settings 31 therapists  27 physicians  4 psychologists  Sampling not clear Not reported Mean treatment duration 12.01 weeks (SD = 5.4) AAI [pre-therapy] GSI IIP IS [pre-, post-therapy] HAQ [post-therapy] Hierarchical linear model AAI(secure–insecure) not associated with  HAQ β = 0.07 (P = 0.53) AAI(dismissing–preoccupied) not associated with HAQ β = 0.10 (P = 0.66) Hierarchical linear model AAI(secure–insecure) not associated with GSI  β = 0.07  IIP  β = 0.07  IS  β = −0.22 (all P > 0.05) AAI(dismissing–preoccupied) not associated with GSI  β = −0.01 IIP  β = −0.06 IS  β = −0.14 (all P > 0.05) Poor [5–1] 1381 patients assigned to therapists in consecutive order  Sampling not clear Tyrrel et al. (1999)  Cross-sectional study  USA  A community-based setting 21 case managers  Sampling not clea 54 clients  Sampling not clear Not reported Mean dyad relationship 31 months (SD = 17, range=7–69) AAI WAI-36 QLI GAF BDI Psychiatric hospitalization days in previous 1 year Correlation AAI(dismissing–preoccupied) not correlated with WAI r = 0.07 (P not reported) Hierarchical regression AAI(dismissing–preoccupied) not associated with WAI β = 0.04 (P not reported) Correlation  AAI(dismissing-preoccupied) not correlated with  QLI(global) r = 0.02  QLI(relation-q) r = -0.08  QLI(relation-s) r = 0.10  GAF r = 0.23  BDI r = 0.10  Hospitalization r = 0.06 (all p not reported) Poor [5–2] Repeated data of therapists in analysis Hierarchical regression  AAI(dismissing-preoccupied) not associated with  QLI(global) β = −0.03  QLI(relation-q) β = −0.17  QLI(relation-s) β = 0.02  GAF β = 0.20  BDI β = 0.10  Hospitalization β = 0.01 (all p not reported) Design and setting Population and sampling Attrition or response rate Relationship/ intervention duration Workers’ attachment measure [measure point] Patients’ outcome measure [measure point] Statistical analysis and findings Quality assessment Relationship quality Health outcome Petrowski et al. (2013)  Repeated measure design  Germany  A university hospital inpatient setting 22 therapists  15 psychologists  7 physicians Sampling not clear Not reported 12 sessions (average 62.5 days, SD = 28.45) AAI [pre-therapy] GSI CATS [post-therapy] Hierarchical linear model  AAI(secure–insecure) not associated with  CATS (secure) β not reported (P > 0.05)  AAI(dismissing–preoccupied) not  associated with CATS(preoccupied/merger) β = −0.88 (P = 0.06)  AAI(dismissing–preoccupied) associated  with CATS (avoidant/fearful) β = 0.94 (P = 0.03)* Pearson correlation AAI (secure–insecure) not correlated with GSI r = −0.06 (P = 0.07) AAI(dismissing–preoccupied) not correlated with GSI r = −0.12 (P = 0.51) Fair [7–2] 429 patients Sampling not clear (randomly assigned to therapists) Sibrava (2009)  Repeated measure  design 4 therapists Sampling not clear Not reported 14 Weekly sessions (1.5–2 h) AAI [pre-therapy] WAI-36 [end of 2, 5, 10, 14 sessions] Hierarchical linear model  AAI associated with WAI [at session 2  AAI(secure) showed lower WAI than  AAI(insecure), at session 14  AAI(secure) showed higher WAI than  AAI(insecure)] β = 0.62 (P < 0.05)* Poor [6–2] Small sample size in therapists  Not clear 69 clients Sampling not clear Petrowski et al. (2011)  Repeated measure design  Germany  A university hospital inpatient setting 19 therapists  13 psychologists  6 physicians  Sampling not clear Not reported Not clear AAI [pre-therapy] HAQ [post-therapy] Hierarchical regression  AAI(secure–insecure) not associated with   HAQ (statistics not reported)  AAI (dismissing–preoccupied) associated   with HAQ (relationship) β = −0.65  (P = 0.02)*   with HAQ(outcome) β = −0.65 (P <  0.01)*  In all other 7 models, AAI (secure– insecure) or AAI (dismissing–preoccupied) not associated with HAQ (relationship) or HAQ (outcome) (statistics not reported) Poor [5–1] Repeated data of therapists in analysis 59 patients  Sampling not clear  (randomly assigned to therapists) Schauenburg et al. (2010)  Repeated measure design  Germany  2 psychotherapy hospitals, inpatient settings 31 therapists  27 physicians  4 psychologists  Sampling not clear Not reported Mean treatment duration 12.01 weeks (SD = 5.4) AAI [pre-therapy] GSI IIP IS [pre-, post-therapy] HAQ [post-therapy] Hierarchical linear model AAI(secure–insecure) not associated with  HAQ β = 0.07 (P = 0.53) AAI(dismissing–preoccupied) not associated with HAQ β = 0.10 (P = 0.66) Hierarchical linear model AAI(secure–insecure) not associated with GSI  β = 0.07  IIP  β = 0.07  IS  β = −0.22 (all P > 0.05) AAI(dismissing–preoccupied) not associated with GSI  β = −0.01 IIP  β = −0.06 IS  β = −0.14 (all P > 0.05) Poor [5–1] 1381 patients assigned to therapists in consecutive order  Sampling not clear Tyrrel et al. (1999)  Cross-sectional study  USA  A community-based setting 21 case managers  Sampling not clea 54 clients  Sampling not clear Not reported Mean dyad relationship 31 months (SD = 17, range=7–69) AAI WAI-36 QLI GAF BDI Psychiatric hospitalization days in previous 1 year Correlation AAI(dismissing–preoccupied) not correlated with WAI r = 0.07 (P not reported) Hierarchical regression AAI(dismissing–preoccupied) not associated with WAI β = 0.04 (P not reported) Correlation  AAI(dismissing-preoccupied) not correlated with  QLI(global) r = 0.02  QLI(relation-q) r = -0.08  QLI(relation-s) r = 0.10  GAF r = 0.23  BDI r = 0.10  Hospitalization r = 0.06 (all p not reported) Poor [5–2] Repeated data of therapists in analysis Hierarchical regression  AAI(dismissing-preoccupied) not associated with  QLI(global) β = −0.03  QLI(relation-q) β = −0.17  QLI(relation-s) β = 0.02  GAF β = 0.20  BDI β = 0.10  Hospitalization β = 0.01 (all p not reported) Note. Healthcare workers’ attachment style measure: AAI = Adult Attachment Interview (category: secure, dismissing, preoccupied, unresolved) (score: secure–insecure, dismissing–preoccupied). Relationship quality measure: CATS = Client Attachment to Therapist Scale (subscale: secure, avoidant/fearful, preoccupied/merger); WAI-36 = Working Alliance Inventory 36-Item Version; HAQ = Helping Alliance Questionnaire (subscales: relationship satisfaction, outcome satisfaction). Health outcome measure: GSI = Global Severity Index of the Symptom Checklist; IIP = Inventory of Interpersonal Problems; IS = Impairment Score, caregiver-rating; QLI = Quality of Life Interview (subscale: global life satisfaction, relationship quality, relationship satisfaction); GAF = Global Assessment of Functioning; BDI = 13-item Beck Depression Inventory. Findings: *Statistically significant. Quality Assessment: [the number of ‘Yes’ out of 14 items in total—the number of ‘Yes’ among items 7, 9, 11 and 12] (see Table 1). Table 4 Characteristics, findings and appraisals of studies using self-report adult attachment measurements Design and setting Population and sampling Attrition or response rate Relationship/intervention duration Workers’ attachment measure [measure point] Patients’ outcome measure [measure point] Statistical analysis and findings Quality assessment Relationship quality Health outcome Bruck et al. (2006)  Interrupted time series study  USA  A medical centre outpatient service 26 therapists psychologists psychiatrists social workers Convenience sampling 26 patients Recruited through advertisement 20 patients dropped out From 46 (43%) 30 sessions RSQ [pre-therapy] WAI-12 SEQ [6th session] TC GAS SCL IIP [post-therapy] Pearson correlation  RSQ not associated with WAI, SEQ or TC  (statistics not reported) Pearson correlation  RSQ(fearful) associated with   GAS r = −0.41 (P = 0.05)*  RSQ(preoccupied) associated with   GAS r = −0.35 (P = 0.05)*   SCL r = −0.30 (P = 0.05)*  RSQ(dismissing) associated with   IIP r = −0.30 (P < 0.05)*  All other correlations not significant (statistics not reported) Fair [7–2] Bucci et al. (2015)  Cross-sectional study  UK  Primary care psychology services 30 therapists psychologists practitioners counsellors other workers Convenience sampling 30 clients Recruited by therapists (next client at least 3 sessions) 12 clients out of 42 not answered (29%) At least 3 sessions RQ WAI-36 PHQ GAD Spearman correlation  RQ(secure) rho=−0.13 (P = 0.49)  RQ(fearful) rho=−0.17 (P = 0.38)  RQ(preoccupied) rho=−0.08 (P = 0.67)  RQ(dismissing) rho = 0.06 (P = 0.74)  All not associated with WAI Spearman correlation, clients split low/high (by median of PHQ + GAD) symptom  RQ(fearful) associated with WAI in high  symptom clients rho=−0.63 (P = 0.02)*  All other correlations not significant (statistics not reported) Fair [5–2] Kafetsios et al. (2016)  Cross-sectional study  Greece  Private and public system 40 doctors (various specialities) Sampling not clear 160 patients Sampling not clear 100% (No one refused) Average contact time = 34 months 27days RQ(converted into avoidant and anxious) PSQ Multilevel model  RQ(avoidant) not associated with PSQ  β = −0.006 (P > 0.10)  RQ(anxious) not associated with PSQ  β = −0.009 (P > 0.10) Fair [7–2] Dunkle and Friedlander (1996)  Cross-sectional survey 73 therapists Convenience sampling 73 clients Selected by therapists 29% (Data of 73 dyads/252 packets mailed) 3 to 5 sessions AAS WAI-12 Correlation  AAS(close) correlated with   WAI(total) r = 0.39 WAI(task) r = 0.34   WAI(bond) r = 0.41 WAI(goal) r = 0.32 (all P < 0.01)*  AAS(depend) correlated with   WAI(total) r = 0.42 WAI(task) r = 0.37   WAI(bond) r = 0.44 WAI(goal) r = 0.36 (all P < 0.01)*  AAS(anxiety) not correlated with   WAI(total) r = −0.16 WAI(task) r = −0.09   WAI(bond) r = −0.22 WAI(goal) r = −0.12 (all P > 0.05) Regression  AAS(close) associated with WAI(bond) β = 0.38 (P = 0.01)*  AAS(depend) not associated with WAI(bond)  β = 0.31 (P = 0.09)  AAS(anxiety) not associated with WAI(bond)  β = 0.16 (P = 0.37) Fair [6–3]  USA  15 university counselling centres, 6 clinics Wiseman and Tishby (2014)  Repeated measure design  Israel  A university counselling centre 27 therapists psychologists social workers Convenience sampling 54 clients Recruited at reception 13 clients dropped out from 67 (19%) 32 sessions (50-min session, once a week) ECR [pre-therapy] OQ-45 [pre, 5th, 15th, 23th, 28th session] CATS [5th, 15th, 28th session] Hierarchical mixed model  ECR(anxiety) associated with CATS at 5th  β = −0.23 (P = 0.05)*  All other association not significant  [ECR(anxiety)(avoidance) and CATS at 5th, 15th, 28th session] (statistics not reported)   Hierarchical linear model  ECR(anxiety) β = 0.12 (P > 0.05)  ECR(avoidance) β = −0.07  (P > 0.05) not associated with OQ change Good [9–4] Fuertes et al. (2007)  Cross-sectional survey  USA  Private settings, university counselling centres, hospitals, clinics 59 therapists Random sampling from American Counselling Association 59 clients selected by therapists 20% (Data of 59 dyads/300 therapists selected) At least 5 sessions (median = 16, n=30) ECR WAI-12 CATS RRI EUS COM Correlation  ECR(avoidance) associated with CATS(secure) r = −0.27 (P < 0.05)*  All other correlations not significant  [ECR(anxiety)(avoidance) and WAI, RRI,  CATS(secure)(avoidant/fearful)(preoccupied/merger), EUS] r = −0.44 to 0.05 (P not reported) Correlation  ECR(avoidance) correlated  with COM r = −0.45 (P < 0.001)*  ECR(anxiety) correlatedwith COM  r = −0.52 (P < 0.001)* Hierarchical regression  ECR(anxiety) associated  with COM β = −0.41 (P<0.002)* Fair [7–3] Wongpakaran and Wongpakaran (2012)  Cross-sectional study  Thailand  A psychiatric hospital outpatient setting 13 therapists psychiatrists residents Sampling not clear 121 patients Convenience sampling Not clear Not clear (each session 5 min to more than 1 h) ECR (secure or preoccupied) WAI-12 PDQ t-Test  WAI not differed by ECR (statistics not reported) MANOVA  ECR not associated with WAI  (statistics not reported) t-Test  PDQ(interpersonal difficulties) differed by   ECR(secure) 14.12 (SD = 4.20)   ECR(preoccupied) 12.74 (SD = 4.82) (P = 0.026)* ANOVA  ECR(secure) associated with   worse PDQ(anxiety) β = 8.95  (P = 0.001)*   worse PDQ(depression)  β = 8.25 (P = 0.002)*  ECR(secure) not associated  with PDQ (interpersonal difficulty) β = 2.02 (P=0.291) Poor [6–1] Very short relationship duration ECR cut-off not clear Berry and Greenwood (2016)  Cross-sectional study  UK  Community mental health service 33 care co-ordinators  24 nurses  5 occupational therapists  3 social workers  1 clinical psychologists Not reported 3 or more months PAM WAI-12 Correlation  PAM(anxious) not correlated  with WAI(total) r = −0.06 (P > 0.05)  PAM(avoidant) not correlated  with WAI(total) r = 0.04 (P > 0.05) Poor [6–2] Repeated data of therapists in analysis Sampling not clear 61 patients with psychosis Sampling not clear Design and setting Population and sampling Attrition or response rate Relationship/intervention duration Workers’ attachment measure [measure point] Patients’ outcome measure [measure point] Statistical analysis and findings Quality assessment Relationship quality Health outcome Bruck et al. (2006)  Interrupted time series study  USA  A medical centre outpatient service 26 therapists psychologists psychiatrists social workers Convenience sampling 26 patients Recruited through advertisement 20 patients dropped out From 46 (43%) 30 sessions RSQ [pre-therapy] WAI-12 SEQ [6th session] TC GAS SCL IIP [post-therapy] Pearson correlation  RSQ not associated with WAI, SEQ or TC  (statistics not reported) Pearson correlation  RSQ(fearful) associated with   GAS r = −0.41 (P = 0.05)*  RSQ(preoccupied) associated with   GAS r = −0.35 (P = 0.05)*   SCL r = −0.30 (P = 0.05)*  RSQ(dismissing) associated with   IIP r = −0.30 (P < 0.05)*  All other correlations not significant (statistics not reported) Fair [7–2] Bucci et al. (2015)  Cross-sectional study  UK  Primary care psychology services 30 therapists psychologists practitioners counsellors other workers Convenience sampling 30 clients Recruited by therapists (next client at least 3 sessions) 12 clients out of 42 not answered (29%) At least 3 sessions RQ WAI-36 PHQ GAD Spearman correlation  RQ(secure) rho=−0.13 (P = 0.49)  RQ(fearful) rho=−0.17 (P = 0.38)  RQ(preoccupied) rho=−0.08 (P = 0.67)  RQ(dismissing) rho = 0.06 (P = 0.74)  All not associated with WAI Spearman correlation, clients split low/high (by median of PHQ + GAD) symptom  RQ(fearful) associated with WAI in high  symptom clients rho=−0.63 (P = 0.02)*  All other correlations not significant (statistics not reported) Fair [5–2] Kafetsios et al. (2016)  Cross-sectional study  Greece  Private and public system 40 doctors (various specialities) Sampling not clear 160 patients Sampling not clear 100% (No one refused) Average contact time = 34 months 27days RQ(converted into avoidant and anxious) PSQ Multilevel model  RQ(avoidant) not associated with PSQ  β = −0.006 (P > 0.10)  RQ(anxious) not associated with PSQ  β = −0.009 (P > 0.10) Fair [7–2] Dunkle and Friedlander (1996)  Cross-sectional survey 73 therapists Convenience sampling 73 clients Selected by therapists 29% (Data of 73 dyads/252 packets mailed) 3 to 5 sessions AAS WAI-12 Correlation  AAS(close) correlated with   WAI(total) r = 0.39 WAI(task) r = 0.34   WAI(bond) r = 0.41 WAI(goal) r = 0.32 (all P < 0.01)*  AAS(depend) correlated with   WAI(total) r = 0.42 WAI(task) r = 0.37   WAI(bond) r = 0.44 WAI(goal) r = 0.36 (all P < 0.01)*  AAS(anxiety) not correlated with   WAI(total) r = −0.16 WAI(task) r = −0.09   WAI(bond) r = −0.22 WAI(goal) r = −0.12 (all P > 0.05) Regression  AAS(close) associated with WAI(bond) β = 0.38 (P = 0.01)*  AAS(depend) not associated with WAI(bond)  β = 0.31 (P = 0.09)  AAS(anxiety) not associated with WAI(bond)  β = 0.16 (P = 0.37) Fair [6–3]  USA  15 university counselling centres, 6 clinics Wiseman and Tishby (2014)  Repeated measure design  Israel  A university counselling centre 27 therapists psychologists social workers Convenience sampling 54 clients Recruited at reception 13 clients dropped out from 67 (19%) 32 sessions (50-min session, once a week) ECR [pre-therapy] OQ-45 [pre, 5th, 15th, 23th, 28th session] CATS [5th, 15th, 28th session] Hierarchical mixed model  ECR(anxiety) associated with CATS at 5th  β = −0.23 (P = 0.05)*  All other association not significant  [ECR(anxiety)(avoidance) and CATS at 5th, 15th, 28th session] (statistics not reported)   Hierarchical linear model  ECR(anxiety) β = 0.12 (P > 0.05)  ECR(avoidance) β = −0.07  (P > 0.05) not associated with OQ change Good [9–4] Fuertes et al. (2007)  Cross-sectional survey  USA  Private settings, university counselling centres, hospitals, clinics 59 therapists Random sampling from American Counselling Association 59 clients selected by therapists 20% (Data of 59 dyads/300 therapists selected) At least 5 sessions (median = 16, n=30) ECR WAI-12 CATS RRI EUS COM Correlation  ECR(avoidance) associated with CATS(secure) r = −0.27 (P < 0.05)*  All other correlations not significant  [ECR(anxiety)(avoidance) and WAI, RRI,  CATS(secure)(avoidant/fearful)(preoccupied/merger), EUS] r = −0.44 to 0.05 (P not reported) Correlation  ECR(avoidance) correlated  with COM r = −0.45 (P < 0.001)*  ECR(anxiety) correlatedwith COM  r = −0.52 (P < 0.001)* Hierarchical regression  ECR(anxiety) associated  with COM β = −0.41 (P<0.002)* Fair [7–3] Wongpakaran and Wongpakaran (2012)  Cross-sectional study  Thailand  A psychiatric hospital outpatient setting 13 therapists psychiatrists residents Sampling not clear 121 patients Convenience sampling Not clear Not clear (each session 5 min to more than 1 h) ECR (secure or preoccupied) WAI-12 PDQ t-Test  WAI not differed by ECR (statistics not reported) MANOVA  ECR not associated with WAI  (statistics not reported) t-Test  PDQ(interpersonal difficulties) differed by   ECR(secure) 14.12 (SD = 4.20)   ECR(preoccupied) 12.74 (SD = 4.82) (P = 0.026)* ANOVA  ECR(secure) associated with   worse PDQ(anxiety) β = 8.95  (P = 0.001)*   worse PDQ(depression)  β = 8.25 (P = 0.002)*  ECR(secure) not associated  with PDQ (interpersonal difficulty) β = 2.02 (P=0.291) Poor [6–1] Very short relationship duration ECR cut-off not clear Berry and Greenwood (2016)  Cross-sectional study  UK  Community mental health service 33 care co-ordinators  24 nurses  5 occupational therapists  3 social workers  1 clinical psychologists Not reported 3 or more months PAM WAI-12 Correlation  PAM(anxious) not correlated  with WAI(total) r = −0.06 (P > 0.05)  PAM(avoidant) not correlated  with WAI(total) r = 0.04 (P > 0.05) Poor [6–2] Repeated data of therapists in analysis Sampling not clear 61 patients with psychosis Sampling not clear Note. Healthcare workers’ attachment style measure: RSQ = Relationship Scale Questionnaire (subscale: secure, dismissing, preoccupied, fearful); RQ = Relationship Questionnaire (subscale: secure, dismissing, preoccupied, fearful); AAS = Adult Attachment Scale (subscale: depend, anxiety, close); ECR = Revised Experience of Close Relationship Questionnaire (subscale: anxiety, avoidance); PAM = Psychosis Attachment Measure (subscale: anxious, avoidant). Relationship quality measure: WAI-12 = Working Alliance Inventory 12-Item Version (subscales: bond, task, goal); SEQ = Session Evaluation Questionnaire; TC = Target Complaints; PSQ = Patient Satisfaction Questionnaire; WAI-36 = Working Alliance Inventory 36-Item Version; CATS = Client Attachment to Therapist Scale (subscale: secure, avoidant/fearful, preoccupied/merger); RRI = Real Relationship Inventory; EUS = Empathic Understanding Scale. Health outcome measure: GAS = Global Assessment Scale; SCL = Symptom Checklist 90 Revised; IIP = Inventory of Interpersonal Problems; PHQ = Patient Health Questionnaire 9-Item Version (depression scale); GAD = General Anxiety Disorder Assessment; OQ = Outcome Questionnaire (symptom distress, interpersonal problem and social role measure); COM = Counselling Outcome Measure; PDQ = Psychological Distress Questionnaire (subscales: anxiety, depression, interpersonal difficulties). Findings: *Statistically significant. Quality Assessment: [the number of ‘Yes’ out of 14 items in total—the number of ‘Yes’ among items 7, 9, 11 and 12] (see Table 1). Table 4 Characteristics, findings and appraisals of studies using self-report adult attachment measurements Design and setting Population and sampling Attrition or response rate Relationship/intervention duration Workers’ attachment measure [measure point] Patients’ outcome measure [measure point] Statistical analysis and findings Quality assessment Relationship quality Health outcome Bruck et al. (2006)  Interrupted time series study  USA  A medical centre outpatient service 26 therapists psychologists psychiatrists social workers Convenience sampling 26 patients Recruited through advertisement 20 patients dropped out From 46 (43%) 30 sessions RSQ [pre-therapy] WAI-12 SEQ [6th session] TC GAS SCL IIP [post-therapy] Pearson correlation  RSQ not associated with WAI, SEQ or TC  (statistics not reported) Pearson correlation  RSQ(fearful) associated with   GAS r = −0.41 (P = 0.05)*  RSQ(preoccupied) associated with   GAS r = −0.35 (P = 0.05)*   SCL r = −0.30 (P = 0.05)*  RSQ(dismissing) associated with   IIP r = −0.30 (P < 0.05)*  All other correlations not significant (statistics not reported) Fair [7–2] Bucci et al. (2015)  Cross-sectional study  UK  Primary care psychology services 30 therapists psychologists practitioners counsellors other workers Convenience sampling 30 clients Recruited by therapists (next client at least 3 sessions) 12 clients out of 42 not answered (29%) At least 3 sessions RQ WAI-36 PHQ GAD Spearman correlation  RQ(secure) rho=−0.13 (P = 0.49)  RQ(fearful) rho=−0.17 (P = 0.38)  RQ(preoccupied) rho=−0.08 (P = 0.67)  RQ(dismissing) rho = 0.06 (P = 0.74)  All not associated with WAI Spearman correlation, clients split low/high (by median of PHQ + GAD) symptom  RQ(fearful) associated with WAI in high  symptom clients rho=−0.63 (P = 0.02)*  All other correlations not significant (statistics not reported) Fair [5–2] Kafetsios et al. (2016)  Cross-sectional study  Greece  Private and public system 40 doctors (various specialities) Sampling not clear 160 patients Sampling not clear 100% (No one refused) Average contact time = 34 months 27days RQ(converted into avoidant and anxious) PSQ Multilevel model  RQ(avoidant) not associated with PSQ  β = −0.006 (P > 0.10)  RQ(anxious) not associated with PSQ  β = −0.009 (P > 0.10) Fair [7–2] Dunkle and Friedlander (1996)  Cross-sectional survey 73 therapists Convenience sampling 73 clients Selected by therapists 29% (Data of 73 dyads/252 packets mailed) 3 to 5 sessions AAS WAI-12 Correlation  AAS(close) correlated with   WAI(total) r = 0.39 WAI(task) r = 0.34   WAI(bond) r = 0.41 WAI(goal) r = 0.32 (all P < 0.01)*  AAS(depend) correlated with   WAI(total) r = 0.42 WAI(task) r = 0.37   WAI(bond) r = 0.44 WAI(goal) r = 0.36 (all P < 0.01)*  AAS(anxiety) not correlated with   WAI(total) r = −0.16 WAI(task) r = −0.09   WAI(bond) r = −0.22 WAI(goal) r = −0.12 (all P > 0.05) Regression  AAS(close) associated with WAI(bond) β = 0.38 (P = 0.01)*  AAS(depend) not associated with WAI(bond)  β = 0.31 (P = 0.09)  AAS(anxiety) not associated with WAI(bond)  β = 0.16 (P = 0.37) Fair [6–3]  USA  15 university counselling centres, 6 clinics Wiseman and Tishby (2014)  Repeated measure design  Israel  A university counselling centre 27 therapists psychologists social workers Convenience sampling 54 clients Recruited at reception 13 clients dropped out from 67 (19%) 32 sessions (50-min session, once a week) ECR [pre-therapy] OQ-45 [pre, 5th, 15th, 23th, 28th session] CATS [5th, 15th, 28th session] Hierarchical mixed model  ECR(anxiety) associated with CATS at 5th  β = −0.23 (P = 0.05)*  All other association not significant  [ECR(anxiety)(avoidance) and CATS at 5th, 15th, 28th session] (statistics not reported)   Hierarchical linear model  ECR(anxiety) β = 0.12 (P > 0.05)  ECR(avoidance) β = −0.07  (P > 0.05) not associated with OQ change Good [9–4] Fuertes et al. (2007)  Cross-sectional survey  USA  Private settings, university counselling centres, hospitals, clinics 59 therapists Random sampling from American Counselling Association 59 clients selected by therapists 20% (Data of 59 dyads/300 therapists selected) At least 5 sessions (median = 16, n=30) ECR WAI-12 CATS RRI EUS COM Correlation  ECR(avoidance) associated with CATS(secure) r = −0.27 (P < 0.05)*  All other correlations not significant  [ECR(anxiety)(avoidance) and WAI, RRI,  CATS(secure)(avoidant/fearful)(preoccupied/merger), EUS] r = −0.44 to 0.05 (P not reported) Correlation  ECR(avoidance) correlated  with COM r = −0.45 (P < 0.001)*  ECR(anxiety) correlatedwith COM  r = −0.52 (P < 0.001)* Hierarchical regression  ECR(anxiety) associated  with COM β = −0.41 (P<0.002)* Fair [7–3] Wongpakaran and Wongpakaran (2012)  Cross-sectional study  Thailand  A psychiatric hospital outpatient setting 13 therapists psychiatrists residents Sampling not clear 121 patients Convenience sampling Not clear Not clear (each session 5 min to more than 1 h) ECR (secure or preoccupied) WAI-12 PDQ t-Test  WAI not differed by ECR (statistics not reported) MANOVA  ECR not associated with WAI  (statistics not reported) t-Test  PDQ(interpersonal difficulties) differed by   ECR(secure) 14.12 (SD = 4.20)   ECR(preoccupied) 12.74 (SD = 4.82) (P = 0.026)* ANOVA  ECR(secure) associated with   worse PDQ(anxiety) β = 8.95  (P = 0.001)*   worse PDQ(depression)  β = 8.25 (P = 0.002)*  ECR(secure) not associated  with PDQ (interpersonal difficulty) β = 2.02 (P=0.291) Poor [6–1] Very short relationship duration ECR cut-off not clear Berry and Greenwood (2016)  Cross-sectional study  UK  Community mental health service 33 care co-ordinators  24 nurses  5 occupational therapists  3 social workers  1 clinical psychologists Not reported 3 or more months PAM WAI-12 Correlation  PAM(anxious) not correlated  with WAI(total) r = −0.06 (P > 0.05)  PAM(avoidant) not correlated  with WAI(total) r = 0.04 (P > 0.05) Poor [6–2] Repeated data of therapists in analysis Sampling not clear 61 patients with psychosis Sampling not clear Design and setting Population and sampling Attrition or response rate Relationship/intervention duration Workers’ attachment measure [measure point] Patients’ outcome measure [measure point] Statistical analysis and findings Quality assessment Relationship quality Health outcome Bruck et al. (2006)  Interrupted time series study  USA  A medical centre outpatient service 26 therapists psychologists psychiatrists social workers Convenience sampling 26 patients Recruited through advertisement 20 patients dropped out From 46 (43%) 30 sessions RSQ [pre-therapy] WAI-12 SEQ [6th session] TC GAS SCL IIP [post-therapy] Pearson correlation  RSQ not associated with WAI, SEQ or TC  (statistics not reported) Pearson correlation  RSQ(fearful) associated with   GAS r = −0.41 (P = 0.05)*  RSQ(preoccupied) associated with   GAS r = −0.35 (P = 0.05)*   SCL r = −0.30 (P = 0.05)*  RSQ(dismissing) associated with   IIP r = −0.30 (P < 0.05)*  All other correlations not significant (statistics not reported) Fair [7–2] Bucci et al. (2015)  Cross-sectional study  UK  Primary care psychology services 30 therapists psychologists practitioners counsellors other workers Convenience sampling 30 clients Recruited by therapists (next client at least 3 sessions) 12 clients out of 42 not answered (29%) At least 3 sessions RQ WAI-36 PHQ GAD Spearman correlation  RQ(secure) rho=−0.13 (P = 0.49)  RQ(fearful) rho=−0.17 (P = 0.38)  RQ(preoccupied) rho=−0.08 (P = 0.67)  RQ(dismissing) rho = 0.06 (P = 0.74)  All not associated with WAI Spearman correlation, clients split low/high (by median of PHQ + GAD) symptom  RQ(fearful) associated with WAI in high  symptom clients rho=−0.63 (P = 0.02)*  All other correlations not significant (statistics not reported) Fair [5–2] Kafetsios et al. (2016)  Cross-sectional study  Greece  Private and public system 40 doctors (various specialities) Sampling not clear 160 patients Sampling not clear 100% (No one refused) Average contact time = 34 months 27days RQ(converted into avoidant and anxious) PSQ Multilevel model  RQ(avoidant) not associated with PSQ  β = −0.006 (P > 0.10)  RQ(anxious) not associated with PSQ  β = −0.009 (P > 0.10) Fair [7–2] Dunkle and Friedlander (1996)  Cross-sectional survey 73 therapists Convenience sampling 73 clients Selected by therapists 29% (Data of 73 dyads/252 packets mailed) 3 to 5 sessions AAS WAI-12 Correlation  AAS(close) correlated with   WAI(total) r = 0.39 WAI(task) r = 0.34   WAI(bond) r = 0.41 WAI(goal) r = 0.32 (all P < 0.01)*  AAS(depend) correlated with   WAI(total) r = 0.42 WAI(task) r = 0.37   WAI(bond) r = 0.44 WAI(goal) r = 0.36 (all P < 0.01)*  AAS(anxiety) not correlated with   WAI(total) r = −0.16 WAI(task) r = −0.09   WAI(bond) r = −0.22 WAI(goal) r = −0.12 (all P > 0.05) Regression  AAS(close) associated with WAI(bond) β = 0.38 (P = 0.01)*  AAS(depend) not associated with WAI(bond)  β = 0.31 (P = 0.09)  AAS(anxiety) not associated with WAI(bond)  β = 0.16 (P = 0.37) Fair [6–3]  USA  15 university counselling centres, 6 clinics Wiseman and Tishby (2014)  Repeated measure design  Israel  A university counselling centre 27 therapists psychologists social workers Convenience sampling 54 clients Recruited at reception 13 clients dropped out from 67 (19%) 32 sessions (50-min session, once a week) ECR [pre-therapy] OQ-45 [pre, 5th, 15th, 23th, 28th session] CATS [5th, 15th, 28th session] Hierarchical mixed model  ECR(anxiety) associated with CATS at 5th  β = −0.23 (P = 0.05)*  All other association not significant  [ECR(anxiety)(avoidance) and CATS at 5th, 15th, 28th session] (statistics not reported)   Hierarchical linear model  ECR(anxiety) β = 0.12 (P > 0.05)  ECR(avoidance) β = −0.07  (P > 0.05) not associated with OQ change Good [9–4] Fuertes et al. (2007)  Cross-sectional survey  USA  Private settings, university counselling centres, hospitals, clinics 59 therapists Random sampling from American Counselling Association 59 clients selected by therapists 20% (Data of 59 dyads/300 therapists selected) At least 5 sessions (median = 16, n=30) ECR WAI-12 CATS RRI EUS COM Correlation  ECR(avoidance) associated with CATS(secure) r = −0.27 (P < 0.05)*  All other correlations not significant  [ECR(anxiety)(avoidance) and WAI, RRI,  CATS(secure)(avoidant/fearful)(preoccupied/merger), EUS] r = −0.44 to 0.05 (P not reported) Correlation  ECR(avoidance) correlated  with COM r = −0.45 (P < 0.001)*  ECR(anxiety) correlatedwith COM  r = −0.52 (P < 0.001)* Hierarchical regression  ECR(anxiety) associated  with COM β = −0.41 (P<0.002)* Fair [7–3] Wongpakaran and Wongpakaran (2012)  Cross-sectional study  Thailand  A psychiatric hospital outpatient setting 13 therapists psychiatrists residents Sampling not clear 121 patients Convenience sampling Not clear Not clear (each session 5 min to more than 1 h) ECR (secure or preoccupied) WAI-12 PDQ t-Test  WAI not differed by ECR (statistics not reported) MANOVA  ECR not associated with WAI  (statistics not reported) t-Test  PDQ(interpersonal difficulties) differed by   ECR(secure) 14.12 (SD = 4.20)   ECR(preoccupied) 12.74 (SD = 4.82) (P = 0.026)* ANOVA  ECR(secure) associated with   worse PDQ(anxiety) β = 8.95  (P = 0.001)*   worse PDQ(depression)  β = 8.25 (P = 0.002)*  ECR(secure) not associated  with PDQ (interpersonal difficulty) β = 2.02 (P=0.291) Poor [6–1] Very short relationship duration ECR cut-off not clear Berry and Greenwood (2016)  Cross-sectional study  UK  Community mental health service 33 care co-ordinators  24 nurses  5 occupational therapists  3 social workers  1 clinical psychologists Not reported 3 or more months PAM WAI-12 Correlation  PAM(anxious) not correlated  with WAI(total) r = −0.06 (P > 0.05)  PAM(avoidant) not correlated  with WAI(total) r = 0.04 (P > 0.05) Poor [6–2] Repeated data of therapists in analysis Sampling not clear 61 patients with psychosis Sampling not clear Note. Healthcare workers’ attachment style measure: RSQ = Relationship Scale Questionnaire (subscale: secure, dismissing, preoccupied, fearful); RQ = Relationship Questionnaire (subscale: secure, dismissing, preoccupied, fearful); AAS = Adult Attachment Scale (subscale: depend, anxiety, close); ECR = Revised Experience of Close Relationship Questionnaire (subscale: anxiety, avoidance); PAM = Psychosis Attachment Measure (subscale: anxious, avoidant). Relationship quality measure: WAI-12 = Working Alliance Inventory 12-Item Version (subscales: bond, task, goal); SEQ = Session Evaluation Questionnaire; TC = Target Complaints; PSQ = Patient Satisfaction Questionnaire; WAI-36 = Working Alliance Inventory 36-Item Version; CATS = Client Attachment to Therapist Scale (subscale: secure, avoidant/fearful, preoccupied/merger); RRI = Real Relationship Inventory; EUS = Empathic Understanding Scale. Health outcome measure: GAS = Global Assessment Scale; SCL = Symptom Checklist 90 Revised; IIP = Inventory of Interpersonal Problems; PHQ = Patient Health Questionnaire 9-Item Version (depression scale); GAD = General Anxiety Disorder Assessment; OQ = Outcome Questionnaire (symptom distress, interpersonal problem and social role measure); COM = Counselling Outcome Measure; PDQ = Psychological Distress Questionnaire (subscales: anxiety, depression, interpersonal difficulties). Findings: *Statistically significant. Quality Assessment: [the number of ‘Yes’ out of 14 items in total—the number of ‘Yes’ among items 7, 9, 11 and 12] (see Table 1). Several studies analysed healthcare workers’ attachment measurements combined with other variables and such combinations differed between studies. Therefore, in this review, only estimates for which an attachment measurement was input as a categorical variable in comparative analyses or as a simple independent variable (not an interaction term) in correlational analyses were considered, which enabled us to summarize findings of interest reported across the studies reviewed. Meta-analytical synthesis was not attempted because of methodological heterogeneity across studies. A qualitative sensitivity analysis was additionally conducted (see below). Review of titles, abstracts and full-manuscripts, data extraction and quality assessment were conducted by the first author (C.M.). The results of this analysis were checked by the second author (I.J.N.) who also responded to any queries. Discrepancies were resolved through discussion. Results Overview of included studies Figure 1 summarizes the progress of papers through different stages of the review. A total of 13 papers [13–25] met criteria for inclusion, and were included in the final analysis (Tables 3 and 4). No randomized controlled trials or cohort studies were found. In total, 12 studies were conducted in mental healthcare settings, and one study in a general somatic health setting. Twelve studies used correlational analysis for examining the association of interest. One study used comparative analysis of t-test, MANOVA and ANOVA. Figure 1 View largeDownload slide Flow diagram of literature search. Figure 1 View largeDownload slide Flow diagram of literature search. Attachment style measurements Tables 3 and 4 summarize characteristics and findings of the five studies which gathered data using the AAI and other studies that used self-report attachment measurements. Eight studies used self-report methods, including the Experience of Close Relationship Questionnaire (ECR) [26], the Relationship Questionnaire (RQ) [5], the Relationship Scale Questionnaire (RSQ) [27], the Adult Attachment Scale (AAS) [28] and the Psychosis Attachment Measure (PAM) [29]. Patients’ perception and health outcome measurements A variety of tools were used to measure patients’ perceptions following their care, which were of two types: measures of relationship quality with healthcare workers; and measures of patients’ health outcomes. Thus, findings were reviewed for these two outcomes. The most widely used instrument of relationship quality was the 12-item Working Alliance Inventory (WAI-12) [30]. Other widely used tools were the 36-item Working Alliance Inventory (WAI-36) [31], the Client Attachment to Therapist Scale (CATS) [32] and the Helping Alliance Questionnaire (HAQ) [33]. Five other scales [34–38] were used, but each scale was not used in more than one study. Widely used tools to examine patients’ health outcomes included the Global Severity Index of the Symptom Check List (GSI) [39] and the Inventory of Interpersonal Problems (IIP) [40]. A variety of nine other measurements were used to collect outcome data [41–49]. Quality assessment This review was not concerned with evaluating the effectiveness of an intervention, thus standardization of the intervention was not a key issue. However, we recognize that the form of intervention could influence patients’ perceptions and health outcomes, and that standardizing interventions would control for this bias to some extent. Therefore, we considered interrupted time series studies as more robust than repeated measure study designs, and cross-sectional designs as the lowest level of evidence. Tables 3 and 4 summarize studies in this hierarchy of study designs grouped by use of similar attachment measurements. A frequent limitation of the studies was their small sample size, especially of healthcare workers. Only two studies used a sample of more than 50 of healthcare workers (mean = 30.6, SD = 18.4). Only one study reported a power calculation. Another limitation of the studies was that they reiterated healthcare workers’ attachment measurement in analysis in which the nested nature of patient data on healthcare worker data was not accounted for; this statistical problem was identified in three studies. In addition, the duration of clinical interpersonal relationship was relatively short in most studies. Finally, the selection of tools tended to be based on reliability/validity evaluated in previous research, and most studies (n = 9) did not sufficiently test the reliability/validity tools selected in their study setting. In using the quality assessment checklist, we regarded Items 7, 9, 11 and 12 of the checklist to be most important in assessing the quality of studies for the purpose of this review. However, we did not simply use the total number of ‘yes’ ratings among these four items, or among whole the set of items, to rate studies as good, fair or poor; rather we considered risk of selection bias, information bias, measurement bias or confounding. In addition, if the statistical analysis was assessed as flawed, then the study was rated as poor. One study was rated as good and six each as fair or poor. Findings Relationship quality In the studies which used the AAI for measuring healthcare workers’ attachment style (Table 3), one study found that at an early stage of treatment, patients seen by securely attached therapists rated their working alliance as poorer than those patients of insecurely attached therapists. However, approximately 3 months later the secure attachment group rated their working alliance more positively than the insecure attachment group [14]. On the other hand, statistically significant results indicated that the more healthcare workers were dismissing, the more their patients’ attachment to them were avoidant or fearful [13] and patients were less satisfied with the therapeutic relationship [15]. Within the studies using self-report (Table 4), two studies found that healthcare workers’ secure attachment was not associated with patients’ perception of working alliance [18, 19]. However, in one study, healthcare workers’ close (secure) attachment style was found to be associated with patients’ positive perception of their working alliance [21]. Statistically significant results showed that if healthcare workers were more avoidant [23] or anxious [22] patients believed their attachment to healthcare workers to be less secure. One study showed a positive relationship between dependence of healthcare workers and patients’ perception of the quality of their working alliance [21]. Patients’ health outcome Three studies using the AAI [13, 16, 17] explored the association between healthcare workers’ attachment style and patients’ health conditions, and all analyses produced statistically insignificant results (Table 3). As for studies using self-report attachment measurements (Table 4), four studies examined the association between attachment and patients’ health outcomes [18, 22–24]. Statistically significant relationships were found between insecure attachment style (fearful, pre-occupied and dismissing) and anxiety, somatic symptoms or interpersonal problems [18] and between insecure attachment style (anxiety and avoidant) and counselling outcome. In contrast, one study showed that patients treated by securely attached healthcare workers had worse interpersonal problems than those cared for by insecurely attached workers, and that secure attachment style of healthcare workers was associated with higher levels of anxiety and depression in patients [24]. Sensitivity analysis Of the 14 publications that could not be retrieved (Figure 1), abstracts were available for 12. One was found to report findings from the same study [50] as one of the studies included in the review [21], thus, it was excluded. Six out of 11 abstracts reported findings of interest [51–56]. As with the studies included in the review, study abstracts also reported mixed results. However, studies which investigated early stages of therapeutic relationships tended to show that healthcare workers’ attachment styles were not associated with patients’ outcomes [51, 53, 56] and that secure attachment styles of healthcare workers were associated with poorer patients’ outcomes [51]. One study indicated an association between healthcare workers’ insecure attachment styles and poorer patient ratings of alliance [56]. In addition, two studies reported a positive influence of securely attached workers on patients’ response/outcomes compared to that of insecurely attached workers [54, 55]. These findings are largely consistent with the conclusion of this review. Discussion Comparison between interviewer-report and self-report attachment measurements Studies using self-report measures appear to be more likely to detect a relationship between attachment styles and relationship quality or patients’ health outcomes than studies using the AAI and these relationships reported are generally statistically stronger (Tables 3 and 4). This difference might be explained by the different attachment constructs as described earlier. Self-report attachment measures study an entity which is similar to therapeutic working alliance. These findings support previous studies which conclude that the AAI and self-report attachment measures measure different aspects of attachment and are not interchangeable [57, 58]. Synthesis of overall findings This review shows that there is variation across studies regarding the association between attachment style of healthcare workers and patients’ outcome, and that associations, if any, are weak to moderate. Hence it is difficult to draw firm conclusions. This finding supports the conclusion of a recent systematic review on the role of psychotherapist attachment style in therapeutic alliance and outcome [7]. However, close inspection of the associations reported by studies included in this review shows that associations between secure attachment style and positive patients’ perception of relationship quality or health outcomes tend to be positive in direction, whereas associations between insecure dimensions of attachment style (dismissing, pre-occupied, fearful, avoidant, anxious and dependent) and patients’ perceptions of relationship quality or health outcomes tend to be negative in direction, irrespective of whether the AAI or self-report was used. Findings from the two studies that investigated change in patients’ response over time [14, 22] suggest that secure attachment styles of healthcare workers tend to have little or a negative effect on patients’ health outcomes or perceptions in the short term, but in the long term tend to have a more positive effect. In contrast, insecure styles tend to have a positive effect in the short term but little or a negative effect in long-term relationships. These findings are in line with Bartholomew and Horowitz’ model [5] of the self and others in relation to attachment style discussed earlier. Healthcare workers with positive self-perceptions (namely, secure or dismissing attachment styles) might appear superior and overbearing to patients at an early stage of the relationship, which may in turn contribute to a negative effect on patients’ health conditions or perceptions. In contrast, those healthcare workers with negative images of self (namely dependent, pre-occupied, fearful or anxious attachment styles) are likely to be valued by patients at the beginning of therapeutic contact because of the tendency of these health workers to invest highly in establishing a rapport with patients. However, healthcare workers hidden negative model of either themselves or others may present to patients in long-term relationships as unsure, inconsistent or distant attitudes, which might reduce patients’ feeling of safety and support. Practical implications It is now common practice for higher education institutions to screen applicants to healthcare programmes for empathy and positive attitudes towards patients which are considered prerequisites of high quality healthcare, although valid and reliable screening tools are in short supply. We are not aware of attempts to screen for attachment style when recruiting psychotherapy practitioners. Most theorists of adult attachment claim that attachment patterns in adulthood consist of generalized thoughts, feeling and expectations regulating the way that an individual engages in close relationships [59]. Attachment styles may have potential as valid predictors of the quality of interpersonal relationships between professional healthcare providers and patients, although more research would be needed to establish this. Knowledge of their attachment style may assist healthcare workers to enhance the quality of their care. In addition, healthcare workers may be able to modify attachment styles of patients in positive ways [60]. If patients who experienced rejecting and neglectful treatment in their childhood receive, as adults, responsive and supportive caregiving from healthcare workers, it seems possible that their internalized working model of relationships could be modified with resulting benefits for their health and wellbeing. Limitations This review retrieved a large number of papers, however, few papers were included in the analysis. Most of these studies were conducted in Western countries, which may induce cultural bias. The literature search sought only published or grey literature, which may have led to some selection bias. The healthcare workers and patients assessed were predominantly from mental healthcare settings. However, patients in physical care settings are usually treated by multiple healthcare workers and relationships between patients and professional carers may be relatively superficial. Kafetsios et al.’s study [20] suggests that in such cases, the influence of attachment style of care providers on patients’ outcomes and perceptions are likely to be attenuated and therefore difficult to detect. This review included only uncontrolled before–after studies and cross-sectional studies. We acknowledge the limitation of these research designs in evaluating cause and effect relationships. Future research One of the main limitations of studies conducted hitherto is use of a variety of measures and so constructs of attachment style and health outcomes, which make it difficult to compare the findings of different studies and so draw definitive conclusions. Surveying larger samples of the same number of healthcare workers and patients is also desirable because this would reduce the bias of repeated variable data. Future studies could also investigate the effect of attachment styles of healthcare workers on patients’ response over time rather than over short periods or at a single time point. Conclusion The results of this review are mixed, and there is currently no robust, statistically significant evidence of any relationship between attachment style of healthcare workers and patients’ health outcomes. It is uncertain whether this is due to the poor quality of the available evidence, or because there is no relationship, and so this question may not be a priority for future research. References 1 Shaver PR , Mikulincer M . An attachment-theory framework for conceptualizing interpersonal behavior. In: Horowitz LM , Strack S (eds) . Handbook of Interpersonal Psychology: Theory, Research, Assessment, and Therapeutic Interventions . New Jersey : John Wiley & Sons Inc , 2012 : 17 – 35 . Google Scholar CrossRef Search ADS 2 Bowlby J . Attachment and Loss Volume 1: Attachment . London : Pimlico , 1997 . 3 Bowlby J . Attachment and Loss Volume 2: Separation, Anger and Anxiety . London : Pimlico , 1998 . 4 Bowlby J . Attachment and Loss Volume 3: Loss, Sadness and Depression . London : Pimlico , 1998 . 5 Bartholomew K , Horowitz LM . Attachment styles among young adults: a test of four category model . J Pers Soc Psychol 1991 ; 61 : 226 – 44 . Google Scholar CrossRef Search ADS PubMed 6 Silberschatz G . Commentary: what have we learned about how the alliance develops over the course of therapy? Psychother Theory Res Pract Train 2009 ; 46 : 295 – 7 . Google Scholar CrossRef Search ADS 7 Degnan A , Seymour-Hyde A , Harris A et al. . The role of therapist attachment in alliance and outcome: a systematic literature review . Clin Physiol Psychother 2016 ; 23 : 47 – 65 . Google Scholar CrossRef Search ADS 8 George C , Kaplan N , Main M . Adult Attachment Interview . University of California , 1985 . 9 Hesse E . The Adult Attachment Interview: historical and current perspectives. In: Cassidy J , Shaver PR (eds) . Handbook of Attachment: Theory, Research, and Clinical Applications . New York : Guilford Press , 1999 : 395 – 433 . 10 Cassidy J , Shaver PR . Handbook of Attachment: Theory, Research and Clinical Applications . New York : Guilford Press , 1999 . 11 Dal Poz MR , Kinfu Y , Dräger S et al. . Counting Health Workers: Definitions, Data, Methods and Global Results. World Health Organization, 2007 . http://www.who.int/hrh/documents/counting_health_workers.pdf (September 2016, date last accessed). 12 National Heart, Lung and Blood Institute . Quality Assessment Tool for Observational Cohort and Cross-sectional Studies. U.S. Department of Health & Human Services, 2014 . http://www.nhlbi.nih.gov/guidelines/cvd_adult/tools/cohort.htm (September 2016, date last accessed). 13 Petrowski K , Pokorny D , Nowacki K et al. . The therapist’s attachment representation and the patient’s attachment to the therapist . Psychother Res 2013 ; 23 : 25 – 34 . Google Scholar CrossRef Search ADS PubMed 14 Sibrava NJ . Therapist and client attachment and the therapeutic alliance. Dissertation. The Graduate School College of the Liberal Arts, the Pennsylvania State University 2009 . 15 Petrowski K , Nowacki K , Pokorny D et al. . Matching the patient to the therapist: the roles of the attachment status and the helping alliance . J Nerv Ment Dis 2011 ; 199 : 839 – 44 . Google Scholar CrossRef Search ADS PubMed 16 Schauenburg H , Buchheim A , Beckh K et al. . The influence of psychodynamically oriented therapists’ attachment representations on outcome and alliance in inpatient psychotherapy . Psychother Res 2010 ; 20 : 193 – 202 . Google Scholar CrossRef Search ADS PubMed 17 Tyrrell CL , Dozier M , Teague GB et al. . Effective treatment relationships for persons with serious psychiatric disorders: the importance of attachment states of mind . J Consult Clin Psychol 1999 ; 67 : 725 – 33 . Google Scholar CrossRef Search ADS PubMed 18 Bruck E , Winston A , Aderholt S et al. . Predictive validity of patient and therapist attachment and introject styles . Am J Psychother 2006 ; 60 : 393 – 406 . Google Scholar CrossRef Search ADS PubMed 19 Bucci S , Seymour-Hyde A , Harris A et al. . Client and therapist attachment styles and working alliance . Clin Psychol Psychother 2015 ; 23 : 155 – 65 . Google Scholar CrossRef Search ADS PubMed 20 Kafetsios K , Hantzara K , Anagnostopoulos F et al. . Doctors’ attachment orientations, emotion regulation strategies, and patient satisfaction: a multilevel analysis . Health Commun 2016 ; 31 : 772 – 7 . Google Scholar CrossRef Search ADS PubMed 21 Dunkle JH , Friedlander ML . Contribution of therapist experience and personal characteristics to the working alliance . J Counseling Psychol 1996 ; 43 : 456 – 60 . Google Scholar CrossRef Search ADS 22 Wiseman H , Tishby O . Client attachment, attachment to the therapist and client-therapist attachment match: how do they relate to change in psychodynamic psychotherapy? Psychother Res 2014 ; 24 : 392 – 406 . Google Scholar CrossRef Search ADS PubMed 23 Fuertes JN , Mislowack A , Brown S et al. . Correlates of the real relationship in psychotherapy: a study of dyads . Psychother Res 2007 ; 17 : 423 – 30 . Google Scholar CrossRef Search ADS 24 Wongpakaran T , Wongpakaran N . How the interpersonal and attachment styles of therapists impact upon the therapeutic alliance and therapeutic outcomes . J Med Assoc Thai 2012 ; 95 : 1583 – 92 . Google Scholar PubMed 25 Berry C , Greenwood K . The relevance of professionals’ attachment style, expectations and job attitudes for therapeutic relationships with young people who experience psychosis . European Psychiatry 2016 ; 34 : 1 – 8 . Google Scholar CrossRef Search ADS PubMed 26 Brennan KA , Clark CL , Shaver PR . Self-report measurement of adult attachment: an integrative overview. In: Simpson JA , Rholes WS (eds) . Attachment Theory and Close Relationships . New York : Guilford Press , 1998 : 46 – 76 . 27 Griffin DW , Bartholomew K . The metaphysics of measurement: the case of adult attachment. In: Bartholomew K , Perlman D (eds) . Advances in Personal Relationships: Attachment Process in Adulthood 5 . London : Jessica Kingsley Publishers , 1994 : 17 – 52 . 28 Collins NL , Read SJ . Adult attachment, working models, and relationship quality in dating couples . J Pers Soc Psychol 1990 ; 58 : 644 – 63 . Google Scholar CrossRef Search ADS PubMed 29 Berry K , Wearden A , Barrowclough C et al. . Attachment styles, interpersonal relationships and psychotic phenomena in a non-clinical student sample . Pers Individ Dif 2006 ; 41 : 707 – 18 . Google Scholar CrossRef Search ADS 30 Tracey TJ , Kokotovic AM . Factor structure of the Working Alliance Inventory . Psychol Assessment 1989 ; 1 : 207 – 10 . Google Scholar CrossRef Search ADS 31 Horvath AO . Research on the alliance. In: Horvath A , Greenberg L (eds) . The Working Alliance: Theory, Research and Practice . New York : Wiley , 1994 . 32 Mallinckrodt B , Gantt DL , Goble HM . Attachment patterns in the psychotherapy relationship: development of the client attachment to therapist scale . J Counsel Psychol 1995 ; 42 : 307 – 17 . Google Scholar CrossRef Search ADS 33 Bassler M , Potratz B , Krauthauser H . The ‘Helping Alliance Questionnaire’ (HAQ) by Luborsky (Der ‘Helping Alliance Questionnaire’ HAQ von Luborsky) . Psychotherapeut 1995 ; 40 : 23 – 32 . 34 Stiles WB . Measurement of the impact of psychotherapy sessions . J Consult Clin Psychol 1994 ; 2 : 176 – 85 . 35 Battle CC , Imber SD , Hoehn-Saric R et al. . Target complaints as criteria of improvement . Am J Psychother 1996 ; 20 : 184 – 92 . Google Scholar CrossRef Search ADS 36 Grogan S , Conner M , Norman P et al. . Validation of a questionnaire measuring patient satisfaction with general practitioner services . Qual Health Care 2000 ; 9 : 210 – 5 . Google Scholar CrossRef Search ADS PubMed 37 Kelley FA , Gelso CJ , Fuertes JN et al. . The real relationship inventory: development and psychometric investigation of the client form . Psychotherapy 2010 ; 47 : 540 – 53 . Google Scholar CrossRef Search ADS PubMed 38 Barrett-Lennard GT . Dimensions of therapists response as causal factors in therapeutic change . Psychol Monog 1962 ; 76 : 1 – 36 . Google Scholar CrossRef Search ADS 39 Franke G . SCL-R-90. The Symptom-Checklist by Derogatis . Weinheim : Beltz , 1995 . 40 Horowitz L , Strauss B , Kordy H . The Inventory of Interpersonal Problems IIP (Das Inventar zur Erfassung interpersoneller Probleme—IIP) . Weinheim, Germany : Beltz , 2000 . 41 Schepank H . The impairment score (Der Beeintraechtigungssschwere-Score). In: Goettingen . Germany : Beltz , 1995 . 42 Lehman AF . A quality life of interview for the chronically mentally ill . Eval Program Plann 1988 ; 11 : 51 – 62 . Google Scholar CrossRef Search ADS 43 American Psychiatric Association . Diagnostic and Statistical Manual of Mental Disorders , 4th edn . Washington, DC : Author , 1994 . PubMed PubMed 44 Beck AT , Ward CH , Mendelson M et al. . An inventory for measuring depression . Arch Gen Psychiatry 1961 ; 4 : 53 – 63 . Google Scholar CrossRef Search ADS 45 Endicott J , Spitzer RL , Fleiss JL et al. . The Global Assessment Scale: a procedure for measuring overall severity of psychiatric disturbance . Arch Gen Psychiatry 1976 ; 33 : 766 – 71 . Google Scholar CrossRef Search ADS PubMed 46 Derogatis LR . SCL-90-R: Administration, Scoring and Procedures Manual II . Towson, MD : Clinical Psychiatric Research , 1983 . 47 Lambert M , Burlingame GM , Umphress V et al. . The reliability and validity of the Outcome Questionnaire . Clin Psychol Psychother 1996 ; 3 : 249 – 58 . Google Scholar CrossRef Search ADS 48 Gelso CJ , Johnson DH . Explorations in Time-limited Psychotherapy and Psychotherapy . New York : Teachers College Press, Columbia University , 1983 . 49 Wongpakaran T , Wongpakaran N . Development and Validation of Psychological Distress Questionnaire . Chiang Mai : Department of Psychiatry, Chiang Mai University, Thailand , 2011 . 50 Dunkle JH . The contribution of therapists’ personal and professional characteristics to the strength of the therapeutic alliance . Diss Abstr Int 1996 ; 56 : 5763 . 51 Biscoglio RL . Patient and therapist personality, therapeutic alliance, and overall outcome in Brief Relational Therapy . Diss Abstr Int 2005 ; 65 : 6642 . 52 Britton PC . The alliance and adult attachment . Diss Abstr Int 2006 ; 67 : 531 . 53 Frehling R . The relationship of therapist and client: attachment styles to measures of the working alliance . Diss Abstr Int 2005 ; 66 : 2818 . 54 Mahoney KSB . Effects of client satisfaction and attachment styles in the counseling dyad on therapy continuation . Diss Abstr Int 2008 ; 68 : 6972 . 55 Niederhauser RK . Adult attachment styles in primary care clinics for the medically underserved . Diss Abstr Int 2001 ; 61 : 6716 . 56 Sack A . The influence of therapist attachment style on the strength of the therapeutic alliance . Diss Abstr Int 1997 ; 57 : 5343 . 57 Roisman GI , Holland A , Fortuna K et al. . The Adult Attachment Interview and self-reports of attachment style: an empirical rapprochement . J Pers Soc Psychol 2007 ; 92 : 678 – 97 . Google Scholar CrossRef Search ADS PubMed 58 Kirchmann H , Mestel R , Schreiber-Willnow K et al. . Associations among attachment characteristics, patients’ assessment of therapeutic factors, and treatment outcome following inpatient psychodynamic group psychotherapy . Psychother Res 2009 ; 19 : 234 – 48 . Google Scholar CrossRef Search ADS PubMed 59 Rholes WS , Simpson JA . Adult Attachment: Theory, Research, and Clinical Implications . New York : Guilford Publications , 2004 . 60 Bowlby J . A Secure Base: Parent-Child Attachments and Healthy Human Development . New York : Basic Books , 2008 . © The Author(s) 2018. Published by Oxford University Press in association with the International Society for Quality in Health Care. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

International Journal for Quality in Health CareOxford University Press

Published: Mar 21, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off