Using focus groups to design systems science models that promote oral health equity

Using focus groups to design systems science models that promote oral health equity Background: While the US population overall has experienced improvements in oral health over the past 60 years, oral diseases remain among the most common chronic conditions across the life course. Further, lack of access to oral health care contributes to profound and enduring oral health inequities worldwide. Vulnerable and underserved populations who commonly lack access to oral health care include racial/ethnic minority older adults living in urban environments. The aim of this study was to use a systematic approach to explicate cause and effect relationships in creating a causal map, a type of concept map in which the links between nodes represent causality or influence. Methods: To improve our mental models of the real world and devise strategies to promote oral health equity, methods including system dynamics, agent-based modeling, geographic information science, and social network simulation have been leveraged by the research team. The practice of systems science modeling is situated amidst an ongoing modeling process of observing the real world, formulating mental models of how it works, setting decision rules to guide behavior, and from these heuristics, making decisions that in turn affect the state of the real world. Qualitative data were obtained from focus groups conducted with community-dwelling older adults who self-identify as African American, Dominican, or Puerto Rican to elicit their lived experiences in accessing oral health care in their northern Manhattan neighborhoods. Results: The findings of this study support the multi-dimensional and multi-level perspective of access to oral health care and affirm a theorized discrepancy in fit between available dental providers and patients. The lack of information about oral health at the community level may be compromising the use and quality of oral health care among racial/ethnic minority older adults. Conclusions: Well-informed community members may fill critical roles in oral health promotion, as they are viewed as highly credible sources of information and recommendations for dental providers. The next phase of this research will involve incorporating the knowledge gained from this study into simulation models that will be used to explore alternative paths toward improving oral health and health care for racial/ethnic minority older adults. Keywords: Oral public health, Dental public health, Oral health equity, Systems science, Agent-based modeling, Qualitative analysis, Focus group analysis, Racial/ethnic minorities, Older adults, Community-based oral health care * Correspondence: men6@nyu.edu Department of Epidemiology and Health Promotion, New York University College of Dentistry, 433 First Avenue, Room 726, New York, NY 10010, USA Department of Sociomedical Sciences, Columbia University Mailman School of Public Health, 722 West 168th Street, New York, NY 10032, USA Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Kum et al. BMC Oral Health (2018) 18:99 Page 2 of 11 Background Several conceptual models for studying access to Since the end of World War II and the subsequent health services have been proposed. For instance, a increase in living standards and community water framework devised by Penchansky and Thomas defines fluoridation [1], the US population has benefitted access as the fit between service providers and service from substantial reductions in tooth decay and eden- users [12]. There are 5 dimensions for which fit is tulism (complete tooth loss). The declining preva- compared: (1) accessibility, i.e., the locations of service lence of edentulism in the older adult population providers versus the locations of service users; (2) avail- means there is greater retention of natural teeth ability, i.e., the volume and type (supply) of services ver- requiring oral health care as people age [2, 3]. Pre- sus the volume and types of needs; (3) affordability, i.e., vention and treatment efforts, technological advance- the cost of services versus the ability of service users to ments in dentistry, and dental insurance have also pay for them; (4) accommodation, i.e., the approach to contributed to improved oral health [4]. Yet despite provision by service providers versus the perceptions of these documented improvements in oral health over appropriateness by service users; and (5) acceptability, time, the major oral diseases—dental caries and peri- i.e., service users’ reactions to and expectations of ser- odontal diseases—are still among the most prevalent vice providers versus service providers’ reactions to and chronic conditions in the US population across the expectations of service users [12]. Discrepancies in fit in- life span [5]. fluence use of services, satisfaction of service users, and There remain critical unmet oral health needs among practices of service providers. children and adolescents, young and working age adults, Powell offers an alternative taxonomic definition of ac- and older adults. While the public views oral health as a cess to health services that considers social accessibility priority, oral health concerns are inadequately addressed to be comprised of 3 dimensions: affordability, accom- by public policies, especially for older adults [6]. For in- modation, and acceptability [13]. In the present analyses, stance, because the Medicare program in the United the question asked was: “Why do older adults go to a States for persons aged 65 years and older and disabled dental provider?” Responses were used to determine as- adults does not cover routine dental care, many older pects of dental providers and oral health care settings adults are unable to afford the necessary preventive and that influence older adults’ decisions about where they restorative treatments they need [7]. Racial/ethnic mi- seek oral health care, i.e., social accessibility. nority older adults are at higher risk for edentulism than the white majority population [8]. They have also been Methods found to be less likely to report dental visits in the past Qualitative methods year, perhaps due to language barriers, especially among Valuable information for understanding and addressing the foreign-born [9]. In addition to experiencing poorer problems is often held in qualitative forms of data, such clinical measures of oral health, racial/ethnic minority as in mental models and written texts [14]. Similarly, the older adults also report worse self-rated oral health than humanistic approach emphasizes the lived experiences their white counterparts [10]. and personal histories of individuals [15]. Narratives To effectively address oral health and health care in- may provide meaning and context to an individual’s ex- equities requires reforming public policies for how oral periences with illness and recovery [16]. Interpreting health services are financed and delivered [6]. There are narratives may reveal characteristics associated with an multiple identified factors operating at different scales individual’s health status that cannot otherwise be de- that contribute to individual and population oral health tected, such as hope, despair, fear, guilt, and grief. Narra- inequities [11]. Therefore, strategies are needed to sup- tives may point to new hypotheses and stimulate more port and advance oral health promotion and treatment patient-centered research [17]. initiatives, along with oral health policy changes at local, state, and national levels. These strategies include Focus group approach and participants for the study place-based programs that apply principles of geographic Focus groups were conducted with a sample of 194 ra- targeting or directed population approaches to promote cial/ethnic minority men and women aged 50 years and oral health equity. A primary objective of these strategies older living in northern Manhattan who participated in is to identify gaps in dental services and population oral one of 24 focus group sessions about improving oral health needs. Toward advancing research designed to health for older adults [18]. The investigators of the promote oral health equity, this paper presents an ap- study selected focus groups over individual interviews proach to explicating and incorporating the perceptions because group discussions may facilitate greater disclos- and knowledge of African American, Dominican, and ure by participants through reciprocity, i.e., disclosure by Puerto Rican older adults related to oral health care in one participant may prompt greater disclosure by others their neighborhoods. [19]. Further, focus groups allow participants to respond Kum et al. BMC Oral Health (2018) 18:99 Page 3 of 11 to and elaborate on topics raised by fellow participants, / Inwood (home to large numbers of Dominicans), thus facilitating discussion of a greater breadth of topics and East Harlem (home to large numbers of Puerto [19]. Finally, focus groups may be less fatiguing than in- Ricans). These 3 neighborhoods have historically been dividual interviews, which may be particularly important considered as racial/ethnic enclaves, with large num- when interviewing older adults [20]. bers of recent immigrants and many residents qualify- Focus group participants had to meet the following ing for Medicaid and other forms of public assistance. criteria: (1) aged 50 years or older; (2) attended a senior Further details of the recruitment and screening pro- center or other community locale where older adults cedures are available elsewhere [18]. gather in northern Manhattan; (3) speak fluent English The study design of 24 focus groups was selected a or Spanish; and (4) self-identify as African American, priori in order to obtain multiple groups of each Dominican, or Puerto Rican. The demographic charac- demographic segment, thereby allowing conclusions teristics of the focus group participants overall and by about each demographic segment to be based on gender are presented in Table 1. multiple focus group discussions rather than on a sin- gle focus group discussion. Consistent with standard focus group techniques [21], thegroupswereseg- Recruitment procedure, sampling strategy, and context mented based on important characteristics that may Field recruitment staff visited senior centers in northern influence the issues discussed or the ability of the Manhattan and directly approached older adults to ex- members to build rapport. A total of 24 focus groups plain the study, screen them for eligibility, and solicit were conducted, including 12 groups of men and 12 participation in the focus groups. Senior centers were se- groups of women. Within each gender set, 4 groups lected rather than places where older adults receive den- were conducted with African Americans, 4 groups tal care in order to obtain a sample of individuals who were conducted with Dominicans, and 4 groups were did not necessarily have access to, or seek, dental care. conducted with Puerto Ricans. Within each gender / To ensure geographic and demographic representa- ethnic / racial set, half of the groups were conducted tion of northern Manhattan, approximately equal with participants who had visited a dentist in the past numbers of participants were recruited from senior year and half were conducted with participants who centers in each of three northern Manhattan neigh- had not visited a dentist in the past year. Ten groups borhoods: Central / West Harlem (home to large were conducted in English (including two groups with numbers of African Americans), Washington Heights Table 1 Characteristics of participants in focus groups for the total sample and by gender, New York, NY, 2013–2015 Participants and Focus Groups Total Sample Women Men Participants N = 194 n = 104 n =90 Focus groups N =24 n = 12 n =12 Characteristics % (n) % (n) % (n) Age group in years 50–59 14.4% (28) 16.3% (17) 12.2% (11) 60–69 34.0% (66) 32.7% (34) 35.6% (32) 70–79 36.1% (70) 34.6% (36) 37.8% (34) 80–89 11.9% (23) 11.5% (12) 12.2% (11) 90+ 3.6% (7) 4.8% (5) 2.2% (2) Race/ethnicity Dominican 35.6% (69) 33.7% (35) 37.8% (34) Puerto Rican 27.3% (53) 27.9% (29) 26.7% (24) African American 37.1% (72) 38.5% (40) 35.6% (32) Last dental visit Within past year 54.1% (105) 52.9% (55) 55.6% (50) 1–3 years ago 27.3% (53) 31.7% (33) 22.2% (20) > 3 years ago 18.6% (36) 15.4% (16) 22.2% (20) Primary language English 42.3% (82) 46.2% (48) 37.8% (34) Spanish 48.5% (94) 45.2% (47) 52.2% (47) Both 9.3% (18) 8.7% (9) 10.0% (9) Women and men did not differ significantly on any of the characteristics listed above, in accordance with the sampling strategy Kum et al. BMC Oral Health (2018) 18:99 Page 4 of 11 Puerto Ricans who preferred to speak English) and 14 specify the direction of cause and effect. A polarity is groups were conducted in Spanish. An average of 8 older assigned to each relationship indicating whether the ef- adults participated in each of the 24 focus groups. fect (variable Y) increases or decreases relative to the cause (variable X) ceteris paribus, i.e., holding all else Systems science methods equal. Paina and Peters propose viewing health systems A protocol proposed by Kim and Andersen for cod- through the lens of complex adaptive systems to offer ing qualitative text data and developing causal maps insights and inform planning, implementing, monitor- from purposive text data was adapted here to analyze ing, and evaluating more effective, equitable, sustainable, decisions on where older adults seek oral health care and context-relevant approaches to population health [14]. The initial stage involves critically evaluating the needs and demands [22]. For the study of complex text to extract data segments and cause-and-effect re- phenomena, systems science provides a variety of lationships. This corresponds to open coding, i.e., the research methods that are complementary to qualitative summarizing and division of text that captures spe- approaches [23]. These include system dynamics, cific phenomena or experiences that may be used as agent-based modeling, social network analysis, and geo- codes. As codes are grouped, dominant patterns or graphic information science [24–28]. themes emerge and are observed and subsequently Among the findings of a recent systematic review is coded. The next stage involves the grouping together that soft systems modeling techniques are likely to be of similar data segments to arrive at more generalized the most useful addition to public health [29]. This is cause-and-effect relationships and implicit structures because the methodological positioning and subsequent in order to construct a composite causal map. This metaphors in systems science, such as feedback, accu- step corresponds to axial coding, i.e., the organization mulation, and endogenous behavior, provide a way to of relationships among categories of codes to establish conceptualize complex and politically sensitive problems relationships between categories. Implicit structures, and policies of the health system, and in the process, fa- i.e., intermediate variables, guide the merging of cat- cilitate knowledge transfer among researchers, practi- egories of codes. tioners, and policymakers [29]. On the other hand, there A topic guide consisting of open-ended questions was also needs to be greater accountability in hard systems developed by the research team and used by the focus modeling, especially in terms of explicating the proce- group moderators to facilitate the sharing of personal dures used to build computer models. This would allow experiences among participants (see Additional file 1). for other researchers and practitioners to contribute to The initial categories for organizing the qualitative data the modeling process and assess whether or not the were based upon the topic guide; additional categories right questions are being addressed. More realistic, were included during subsequent readings of the tran- data-intensive computer models may lack transparency scripts. Focus groups with content related to categories and flexibility, becoming too complicated and difficult to of interest were indicated and recorded. An iterative comprehend to be of practical use [30, 31]. process was used to extract excerpts of transcript text, A causal map is a visual artifact used in the field of which are referred to as data segments. system dynamics, a scientific approach that involves repre- Data segments are typically a single response from a senting, testing, and modifying assumptions about dynam- participant, i.e., a portion of text indicated in the focus ically complex problems attributed to feedback loops and group by “P:” indicating “Participant:” in the transcripts. delays [32]. Causal mapping is a method for articulating a For certain data segments, it was important to capture plausible explanation for these dynamically complex prob- the dialogue. Hence, a forward slash (/) is used to indi- lems that is used to identify feedback loops. Causal maps cate multiple responses. Cause-and-effect relationships are integral elements in a preliminary blueprint for com- are then identified from data segments. puter models, and in this research, are considered to be Table 2 presents the topics by type (either finding den- boundary objects [33]. Black defines a boundary object as tal providers or going to dental providers) identified “a representation—perhaps a diagram, sketch, sparse text, from the 24 focus groups. or prototype—that helps individuals collaborate effectively Topics refer to the pattern or theme that was deter- across some boundary, often a difference in knowledge, mined by examining a grouping of similar data segments training, or objective.” ([34], 76, p). and were identified at the time of extraction of data segments. Labels were then assigned to types and topics Integrating qualitative and systems science methods in to facilitate the filtering of the data segments in a the data analysis spreadsheet. Next, each data segment was assigned a Importantly, a causal map depicts cause-and-effect rela- data segment Type, Topic, and identification number tionships, where arrows between variables are used to (SegmentID). When assigning SegmentIDs, replicate Kum et al. BMC Oral Health (2018) 18:99 Page 5 of 11 Table 2 Decisions on where to go for oral health care organized by types and topics Types (in shaded rows) and Topics Label Finding dental providers finding Information from insurance company insurance Third-party resources (1-800-Dentist, 311, Internet) third-party Referral from another health provider referral Recommendations from friends, family, community recommendation Going to dental providers going Recommendations from friends, family, community recommendation Dental care environment environment Training and credentials (title) of oral health providers credentials Character of oral health providers character Shared language or culture and understanding of community understanding (Dis)Trust in oral health provider trust Quality of dental care quality Dental treatment options treatment Dental need (urgent care) need Appointments appointments Insurance coverage and cost of care cost entries were removed that had been introduced errone- effect (cmEffect). For example, Fig. 1c presents the ously during data extraction. cmCause and cmEffect to which SegmentID 56 contrib- uted. Data segments in the same topic group were com- Results pared to derive a term that was representative of the Figure 1 provides an illustration of the process by which relationships expressed in that subset of data segments. a data segment and corresponding focus group segmen- Deliberate attempts were made to use variables that tation information is recorded and organized. were already assigned; new variables were typically gen- Certain data segments yielded more than a single erated to clarify relationships. Each of the generalized re- cause-and-effect relationship; thus, there are 287 rela- lationships was then assembled in Vensim software [35] tionships from 240 data segments. The data segment to provide a visual overview of the existing relationships. shown in Fig. 1a is SegmentID 56, which was extracted Certain of the generalized relationships were collapsed from focus group 5, comprised of Dominican men who during the process of assembly and intermediate struc- had received a dental visit in the past year (translated tures were added. The order in which the variables were into English from Spanish). This data segment was used assigned a cmID, indicating the generalized cause and to explicate 2 cause-and-effect relationships: the first for effect relationships, followed the descending order of finding dental providers and the second for going to topics presented in Table 1. One of the generalized rela- dental providers. This is an especially interesting data tionships in a topic group would often provide a link to segment because it represents both the view of the another topic group. To complete the example, Fig. 1d speaker and the view of his friend. presents an illustration of how the generalized A data segment cause (dsCause) and a data segment cause-and-effect relationships are summarized and effect (dsEffect) were then identified for each data tabulated. segment. Attempts were made to extract terms from The behavior (noted simply as Behavior) of the each data segment to express dsCause and dsEffect. To cmCause and cmEffect was then assessed. The Behavior continue with the illustration, Fig. 1b presents the was articulated to check the logic of each cause and ef- cause-and-effect relationships identified for SegmentID fect relationship and assign a polarity to that relation- 56. For data segments where it was difficult to maintain ship. The polarity of the relationship, that is, whether complete data integrity, terms derived from the inter- the behavior of the cause and effect changes in the same pretation of the first author for the data segment direction (positive) or in the opposite direction (nega- were used. tive), is recorded in Relationship Type. For a positive re- Similar expressions of cause and effect relationships, lationship type, an increase (decrease) of the cause leads aided by the filtering of labels, were then generalized to an increase (decrease) of the effect; for a negative re- into a causal map cause (cmCause) and a causal map lationship type, an increase (decrease) of the cause leads Kum et al. BMC Oral Health (2018) 18:99 Page 6 of 11 Fig. 1 Process by which a data segment is recorded and organized. An illustration of the process by which an extracted data segment is recorded and organized. Specifically, data segment 56 (SegmentID 56) originates from focus group 5 conducted with Dominican men who received oral health care in the past year (translated into English from Spanish). The 4 panels correspond to the following steps: (a) Data segment is identified and extracted; (b) Cause and effect relationships are explicated; (c) Similar expressions of cause and effect relationships contribute to a generalized cause-and-effect relationship; and (d) Generalized cause-and-effect relationships are summarized and tabulated to a decrease (increase) of the effect. A causal map ID and data segments. Instead of attempting to capture all (cmID) was assigned after all of the 37 generalized of the attendant details, data segments were aggregated cause-and-effect relationships were established. To to highlight the major constructs and the potential complete the illustration, a summary and tabulation of mechanisms that connect these constructs. Both the task the generalized cause-and-effect relationships used to of manually extracting and organizing the text and the systematically construct the causal map is presented in task of generalizing the selected text were time intensive. Fig. 1d, with the contribution of SegmentID 56 Yet it was only through multiple and iterative readings highlighted in yellow. of the text that the analyses were effectively framed. This Modifications to the Kim and Andersen protocol [14] was especially important in generating the causal map, for constructing a causal map were motivated by the since justifying the underlying logic of relationships re- need to be able to efficiently compare different topics quired considerable abstraction. Kum et al. BMC Oral Health (2018) 18:99 Page 7 of 11 Finally, a composite causal map representing the experi- The first column in Table 3 lists the 19 variables in the ence of the patient was systematically constructed to causal map that are included in any of 12 feedback loops explicate the generalized relationships involved in the involved in finding or going to a dental provider. decision-making process of dental provider choice (Fig. 2). All 12 feedback loops include the focal variable dental These considerations correspond to the access dimen- visits. Three feedback loops are associated with finding sions of acceptability, accommodation, and affordability, dentists – knowledge about dental provider – and 9 i.e., social accessibility to oral health care. Two separate feedback loops are associated with going to dentists – but related issues are manifest: the first involves finding motivation for dental visit. Variables that are not dental providers, knowledge about dental providers, and included in any feedback loops are listed in the last the places where dental providers practice; and the sec- column of Table 2 as preceding variable(s), i.e., a cause ond involves going to and returning to dental providers, of the associated variable in the feedback loop. more specifically, the circumstances at dental practices Below illustrative data segments are presented and dis- that motivate dental visits. cussed, including those related to the example in Fig. 1 The focal variable of the causal map is dental visits (SegmentID 56). Participant responses for how they (highlighted in orange), an indicator of utilization of or found dental providers included: resources from insur- realized access to the oral health care system. The two ance plans; phone calls to third-party services; Internet issues of finding and going to dentists are represented, searches; referrals from health care providers; and rec- respectively, as knowledge about dental provider and ommendations from relatives and friends. For example, motivation for dental visit (also highlighted in orange an African American man explained how he was re- in the causal map). ferred by a dental provider to another dental provider Fig. 2 Causal map derived from focus group data. A composite causal map of decisions on where to go for oral health care based on information extracted from focus groups with African American, Dominican, and Puerto Rican older adults. The solid arrows indicate a positive effect (same direction), whereas the dashed arrows indicate a negative effect (opposite direction) Kum et al. BMC Oral Health (2018) 18:99 Page 8 of 11 Table 3 Feedback loops of decisions on where to go for oral health care for specialized care: “So, finally you got a lot of dentists practices towards oral health care. For example, an who are really at another level. I’ve been to dentist who African American man shared the following remarks: have been like, “Listen, I can’t do this, but your union is “She’ll [wife] find a dentist. [pause]. She’ll notice where going to pay.” Mr. [name] is a good dentist for, what do I’m taking all the pill because it don’t matter to me. I just you call it? The root canal!” (focus group 17: African went to bed. I won’t notice [pause] things like that. But American men with a dental visit in the past year) she will. She’s good that way.” (focus group 17: African Family members and friends were considered to be both American men with a dental visit in the past year) Older important sources of recommendations for finding den- adults who lack the support of family and friends may tal providers and influential in going to dental visits. neglect their health and encounter difficulties in obtain- The role of referrals and recommendations in finding ing resources to address their health needs, such as dentists are captured in feedback loops 1, 2, and 3; the information about oral health care options in their influence of communication about dental experiences neighborhoods and transportation assistance. are captured in feedback loops 4, 5, and 6. These feed- According to participants, information and opinions back loops indicate the potential of using social net- about dental providers and care settings often circulate works to deliver knowledge and change attitudes and through word of mouth among community members Kum et al. BMC Oral Health (2018) 18:99 Page 9 of 11 that may affect their reputations. The following quote inability to control for exposure times, selection bias due from a Dominican man emphasizes the impact of com- to loss to follow-up, and underrepresentation of racial/ munication in decision-making for selecting dentists: ethnic minorities [38]. Agent-based modeling has been proposed as a way forward since it facilitates representa- “There is something important that I want to explain tion of multiple scales and heterogeneity [39, 40]. More- here, the vast majority of us Hispanics look for our over, agent-based models are theory-based and doctors through references, this is very important. In data-driven, and allow for the testing of different plaus- other words if a doctor is good, the neighbor will say… ible mechanisms [41]. look, so and so is a wonderful dentist. Then we start A computer model may be used to communicate and looking for references. And it’s through the references learn about the impact of life events and social relation- that we start communicating with one another and we ships on oral health. The proposed model design con- even make an appointment to go see that doctor. tributes to an existing portfolio of models that have been Then, if the doctor is bad and did not complete the informed by quantitative and spatial data, as well as the work well, for this and that reason. It’s always like experiences and expertise of research team members that, the neighbor will communicate with the second [33, 42–46]. Here particular insights from older adults’ person and that spreads the word. In other words, a experiences that are reflected in focus groups transcripts job well done will be well received by the community, are included, along with supplementary knowledge about but people will also know about bad work.” (focus the older adult population in Manhattan and the Bronx. group 15: Dominican men with a dental visit in the The model is designed to simulate: (1) the life course, past year, translated into English from Spanish) i.e., life stages and life events; (2) oral health status, oral health care seeking orientation, and oral health care use; Several participants were frightened about going to the and (3) multiple social relationships as dynamic social dentist either because of fear and the associated pain of networks based on person agent attributes and geo- dental procedures or the fear of contracting diseases at graphic proximity. the dental office. Recommendations from trusted family Below are key insights gleaned from the focus groups members and friends may minimize such fears because that are considered important in the model design. First, dental providers and oral health care settings have the inability to pay for oral health care is a significant already been vetted (feedback loop 6). A Dominican barrier to accessing services. Second, family members woman confidently stated: “I say, ‘You are afraid? But let and friends are both important sources of recommenda- me tell you, I have a dentist you will love. Here is the tions for finding dental providers and influential in mo- phone number. Here is the phone number. Go to him tivating dental visits. Third, participants believe there is and you will remember me.’” (focus group 14: Dominican a lack of information about oral health in the commu- women without a dental visit in the past year, translated nity and they would like more information about oral into English from Spanish). health and health care. The design includes 2 active agent classes: person Discussion (older adult, dental provider) and place. Place is further In systems science, both the problems and their specified into home, work, third place, e.g., senior center, solutions are understood as being generated from within public library, religious institution, and dental clinic [46]. the system. Social disparities in oral health result from Characteristics of all place agents will include a unique factors at multiple scales [11]. With regular dental identification number and an indication of status (either hygiene and professional care, adverse oral health out- open or closed). An additional characteristic of dental comes such as tooth loss may often be prevented. Oral clinics is the types of insurance accepted. Further, the health behaviors and practices are transferred through identification numbers of older adults who visited each social relationships throughout the life course, which dental clinic and the dates of these visits as well as re- may either increase or decrease social disparities in oral minder messages to older adults regarding scheduled ap- health. For example, children often adopt the behaviors pointments will need to be tracked. An additional and practices of their parents. Social relationships characteristic of third places is whether health outreach among older adults provide mechanisms for the ex- events are held at each location, and if so, the identifica- change of resources, such as information that is critical tion number of participants at outreach events, and the in decision-making. participants who needed a referral to a dental clinic affil- While the life course approach is useful in understand- iated with the outreach events. ing oral health inequities, there are empirical challenges According to the focus group transcript analysis, avail- in testing the proposed theoretical models [36, 37]. ability (the supply of dental providers) and accessibility Problems with using retrospective studies include the (the means of traveling to oral health care) were not Kum et al. BMC Oral Health (2018) 18:99 Page 10 of 11 perceived to be significant barriers to oral health care. fill critical roles in oral health promotion, as they are Rather, the inability to pay for oral health care, poor re- viewed as highly credible sources of information and lationships with dental providers, and lack of informa- recommendations to dental providers. Disseminating tion on oral health and health care were the major up-to-date information at frequented sites to older challenges. Therefore, the model design includes 2 types adults and the community at large about the importance of interventions: (1) social and behavioral interventions; of oral health, proper dental hygiene practices, and local and (2) policy interventions. The first set of modeled in- oral health care options remain public health priorities. terventions would involve community-based outreach education and delivery at places in the neighborhood Additional file that provide information, and use different social rela- tionships to direct information and influence changes in Additional file 1: Focused Group Interview Topic Guide. A topic guide consisting of open-ended questions that was developed by the research oral health care-seeking orientation and oral health care team and used by the focus group moderators to facilitate the sharing of status. The second set of proposed interventions would personal experiences among participants regarding reasons why people involve health insurance coverage, specifically, the im- may or may not visit a dentist. (PDF 57 kb) pact of expanding and ensuring dental insurance cover- age throughout the life span, and reducing restrictions Abbreviations cmCause: Causal map cause; cmEffect: Causal map effect; cmID: Causal map to preventive oral health care. identification number; dsCause: Data segment cause; dsEffect: Data segment effect; SegmentID: Data segment identification number Conclusions The findings of this study support the multi-dimensional Acknowledgments The authors thank the participants and recruitment staff whose engagement and multi-level perspective of access to oral health care in this qualitative study made the reporting of the results possible. and affirm a theorized discrepancy in fit between avail- able providers and patients. The presence of resources Funding does not directly translate into use of services by racial/ The authors were supported in the research, analysis, and writing of this paper by the National Institute for Dental and Craniofacial Research (NIDCR) ethnic minority older adults. Despite the relatively high and the Office of Behavioral and Social Sciences Research (OBSSR) of the US volume of dental providers and the range of transporta- National Institutes of Health (NIH) for the project titled, “Integrating Social tion options, focus group participants did not believe and Systems Science Approaches to Promote Oral Health Equity” (grant R01- DE023072) and by the National Center for Advancing Translational Sciences that their oral health needs were being adequately ad- (NCATS) of the NIH for the project titled, “Primary Care Screening by Dental dressed, whether or not they had recently visited a Hygienists at Chairside: Developing and Evaluating an Electronic Tool” (grant dentist. UL1TR000038). The funding agencies had no role in the design of the study and collection, analysis, and interpretation of data or in the writing of the The systematic approach to explicating cause and ef- manuscript. fect relationships from focus group transcripts intro- duced here may prove transferable to other research Availability of data and materials contexts. The product of this approach, a causal map, De-identified raw data and materials described in the manuscript are freely available from the corresponding author on reasonable request. provides a visual representation of major factors and re- lationships involved in the decision-making process. Authors’ contributions From both epistemological and ontological stand- SSK led the writing of this paper and conducted the analyses as part of her dissertation research. MEN closely edited the paper, provided interpretation, points, however, system dynamics involves more than finalized the tables and figures, and contributed oral health and public the mechanics of creating a causal map. Rather, there is health expertise. SSM conceived of and supervised the analyses, and a philosophical understanding that in order to solve contributed systems science expertise. All authors contributed to the writing and editing of this paper and approved it for publication. large, complex problems, it is important and effective to consider the needs of others [47]. The ability to incorp- Ethics approval and consent to participate orate qualitative data into a causal map allows direct in- This research has been performed in accordance with the Declaration of clusion of the views of underrepresented populations Helsinki. The following Institutional Review Boards reviewed and approved all study procedures: Columbia University Medical Center Institutional Review into the hypothesized cause and effect mechanisms Board [protocol AAAL4104(M01Y04)] and NYU School of Medicine explicated. Institutional Review Board (protocol i12-02947_CR4). All Health Insurance The lack of information about oral health may be Portability and Accountability Act safeguards were followed. All participants signed written consent forms. compromising the use and quality of oral health care among racial/ethnic minority older adults. This finding Competing interests is consistent with key informant views that senior center The authors declare that they have no competing interests. attendees did not regard oral health concerns with the same degree of immediacy as high blood pressure (indi- Publisher’sNote cative of hypertension) or high blood sugar (indicative of Springer Nature remains neutral with regard to jurisdictional claims in diabetes) [48]. Well-informed community members may published maps and institutional affiliations. Kum et al. BMC Oral Health (2018) 18:99 Page 11 of 11 Author details 23. Ip EH, Rahmandad H, Shoham DA, et al. Reconciling statistical and systems science Department of Geography, The State University of New York at Buffalo, 115 approaches to public health. Health Educ Behav. 2013;40(1 suppl):123S–31S. Wilkeson Quad, Ellicott Complex, Buffalo, NY 14261-0055, USA. Department 24. Metcalf SS, Northridge ME, Lamster IB. A systems perspective for dental of Population Health, New York University School of Medicine, 227 East 30th health in older adults. Am J Public Health. 2011;101(10):1820–3. Street, 8th Floor, New York, NY 10016, USA. Department of Epidemiology 25. McLeroy K. Thinking of systems. Am J Public Health. 2006;96(3):402. and Health Promotion, New York University College of Dentistry, 433 First 26. Mabry PL, Marcus SE, Clark PI, Leischow SJ, Méndez D. Systems science: a Avenue, Room 726, New York, NY 10010, USA. Department of Sociomedical revolution in public health policy research. Am J Public Health. 2010;100(7): Sciences, Columbia University Mailman School of Public Health, 722 West 1161–3. 168th Street, New York, NY 10032, USA. 27. Luke DA, Stamatakis KA. Systems science methods in public health. Annu Rev Public Health. 2012;33:357–76. Received: 10 April 2017 Accepted: 22 May 2018 28. Torrens PM. Geography and computational social science. GeoJournal. 2010; 75(2):133–48. 29. Carey G, Malbon E, Carey N, Joyce A, Crammond B, Carey A. Systems science and systems thinking for public health: a systematic review of the field. BMJ Open. 2015;5(12):e009002. References 30. Ness RB, Koopman JS, Roberts MS. Causal system modeling in chronic 1. Centers for Disease Control and Prevention (CDC). Achievements in public disease epidemiology: a proposal. Ann Epidemiol. 2007;17(7):564–8. health, 1900–1999. Fluoridation of drinking water to prevent dental caries. 31. O’Sullivan D, Evans T, Manson S, Metcalf S, Ligmann-Zielinska A, Bone C. MMWR Morb Mortal Wkly Rep. 1999;48(41):933–40. Strategic directions for agent-based modeling: avoiding the YAAWN 2. Centers for Disease Control and Prevention (CDC). Public health and aging: syndrome. J Land Use Sci. 2016;11(2):177–87. retention of natural teeth among older adults–United States, 2002. MMWR 32. Sterman J. Business dynamics: systems thinking and modeling for a Morb Mortal Wkly Rep. 2003;52(50):1226–9. complex world. Boston: Irwin/McGraw-Hill; 2000. 3. Slade GD, Akinkugbe AA, Sanders AE. Projections of U.S. edentulism 33. Northridge ME, Metcalf SS. Enhancing implementation science by applying prevalence following 5 decades of decline. J Dent Res. 2014;93(10):959–65. best principles of systems science. Health Res Policy Syst. 2016;14:74. 4. Kiyak HA, Reichmuth M. Barriers to and enablers of older adults’ use of 34. Black LJ. When visuals are boundary objects in system dynamics work. Syst dental services. J Dent Educ. 2005;69(9):975–86. Dyn Rev. 2013;29(2):70–86. 5. Benjamin RM. Oral health: the silent epidemic. Public Health Rep. 2010; 35. Ventana Systems, Inc. Vensim. Harvard; 2015. 125(2):158–9. 36. Ben-Shlomo Y, Kuh D. A life course approach to chronic disease 6. Fisher-Owens SA, Barker JC, Adams S, et al. Giving policy some teeth: epidemiology: conceptual models, empirical challenges and interdisciplinary routes to reducing disparities in oral health. Health Aff (Millwood). 2008; perspectives. Int J Epidemiol. 2002;31(2):285–93. 27(2):404–12. 37. Nicolau B, Thomson WM, Steele JG, Allison PJ. Life-course epidemiology: 7. Greenblatt AP, Estrada I, Schrimshaw EW, Metcalf SS, Kunzel C, Northridge concepts and theoretical models and its relevance to chronic oral ME. Acceptability of chairside screening for racial/ethnic minority older conditions. Community Dent Oral Epidemiol. 2007;35(4):241–9. adults: a qualitative study. JDR Clin Trans Res. 2017;2(4):343–52. 38. Shavers VL. Measurement of socioeconomic status in health disparities 8. Wu B, Liang J, Plassman BL, Remle RC, Luo X. Edentulism trends among research. J Natl Med Assoc. 2007;99(9):1013–23. middle-aged and older adults in the United States: comparison of five 39. Diez Roux AV. Complex systems thinking and current impasses in health racial/ethnic groups. Community Dent Oral Epidemiol. 2012;40(2):145–53. disparities research. Am J Public Health. 2011;101(9):1627–34. 9. Shelley D, Russell S, Parikh NS, Fahs M. Ethnic disparities in self-reported oral 40. Noble J, Silverman E, Bijak J, et al. Linked lives: the utility of an agent-based health status and access to care among older adults in NYC. J Urban Health. approach to modeling partnership and household formation in the context 2011;88(4):651–62. of social care. In: Proceedings of the 2012 winter simulation conference; 10. Huang DL, Park M. Socioeconomic and racial/ethnic oral health disparities 2012. p. 1–12. among US older adults: oral health quality of life and dentition. J Public 41. Salgado M, Marchione E, Gilbert N. Analysing differential school Health Dent. 2015;75(2):85–92. effectiveness through multilevel and agent-based modelling. J Artif Soc Soc 11. Northridge ME, Ue FV, Borrell LN, et al. Tooth loss and dental caries in Simul. 2014;17(4):3. community-dwelling older adults in northern Manhattan. Gerodontology. 42. Metcalf SS, Northridge ME, Widener MJ, Chakraborty B, Marshall SE, Lamster 2012;29(2):e464–73. IB. Modeling social dimensions of oral health among older adults in urban 12. Penchansky R, Thomas JW. The concept of access: definition and environments. Health Educ Behav. 2013;40(1 suppl):63S–73S. relationship to consumer satisfaction. Med Care. 1981;19(2):127–40. 43. Wang H, Northridge ME, Kunzel C, et al. Modeling social capital as dynamic 13. Powell M. On the outside looking in: medical geography, medical networks to promote access to oral healthcare. Soc Cult Behav Model. 2016; geographers and access to health care. Health Place. 1995;1(1):41–50. 9708:117. 14. Kim H, Andersen DF. Building confidence in causal maps generated from 44. Chakraborty B, Widener MJ, Mirzaei Salehabadi S, et al. Estimating peer purposive text data: mapping transcripts of the Federal Reserve. Syst Dyn density effects on oral health for community-based older adults. BMC Oral Rev. 2012;28(4):311–28. Health. 2017;17(1):166. 15. Phillips J, Ajrouch KJ, Hillcoat-Nallétamby S. Key concepts in social 45. Zhang Q, Northridge ME, Jin Z, Metcalf SS. Modeling accessibility of gerontology. Los Angeles: SAGE; 2010. screening and treatment facilities for older adults using transportation 16. Sweeney K. Complexity in primary care: understanding its value. Oxford; networks. Appl Geogr. 2018;93:64–75. Seattle: Radcliffe Pub.; 2006. 46. Northridge ME, Kum SS, Chakraborty B, et al. Third places for health 17. Greenhalgh T, Hurwitz B. Why study narrative? BMJ. 1999;318(7175):48–50. promotion with older adults: using the consolidated framework for 18. Northridge ME, Shedlin M, Schrimshaw EW, et al. Recruitment of racial/ implementation research to enhance program implementation and ethnic minority older adults through community sites for focus group evaluation. J Urban Health. 2016;93(5):851–70. discussions. BMC Public Health. 2017;17(1):563. 47. Meadows DH. Thinking in Systems: A Primer. White River Junction: Chelsea 19. Morgan DL. Focus group interviewing. In: Gubrium JF, Holstein JA, editors. Green Publishing Co.; 2008. Handbook of interview research: context and method. Thousand Oaks: 48. Marshall S, Schrimshaw EW, Metcalf SS, et al. Evidence from ElderSmile for Sage; 2002. p. 141–59. diabetes and hypertension screening in oral health programs. J Calif Dent 20. Wenger GC. Interviewing older people. In: Gubrium JF, Holstein JA, editors. Assoc. 2015;43(7):379–87. Handbook of interview research: context and method. Thousand Oaks: Sage; 2002. p. 259–78. 21. Krueger RA, Casey MA. Focus groups: a practical guide for applied research. 4th ed. Thousand Oaks: Sage; 2009. 22. Paina L, Peters DH. Understanding pathways for scaling up health services through the lens of complex adaptive systems. Health Policy Plan. 2012; 27(5):365–73. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png BMC Oral Health Springer Journals

Using focus groups to design systems science models that promote oral health equity

Free
11 pages
Loading next page...
 
/lp/springer_journal/using-focus-groups-to-design-systems-science-models-that-promote-oral-6ilq3SdMno
Publisher
BioMed Central
Copyright
Copyright © 2018 by The Author(s).
Subject
Dentistry; Dentistry; Oral and Maxillofacial Surgery
eISSN
1472-6831
D.O.I.
10.1186/s12903-018-0560-0
Publisher site
See Article on Publisher Site

Abstract

Background: While the US population overall has experienced improvements in oral health over the past 60 years, oral diseases remain among the most common chronic conditions across the life course. Further, lack of access to oral health care contributes to profound and enduring oral health inequities worldwide. Vulnerable and underserved populations who commonly lack access to oral health care include racial/ethnic minority older adults living in urban environments. The aim of this study was to use a systematic approach to explicate cause and effect relationships in creating a causal map, a type of concept map in which the links between nodes represent causality or influence. Methods: To improve our mental models of the real world and devise strategies to promote oral health equity, methods including system dynamics, agent-based modeling, geographic information science, and social network simulation have been leveraged by the research team. The practice of systems science modeling is situated amidst an ongoing modeling process of observing the real world, formulating mental models of how it works, setting decision rules to guide behavior, and from these heuristics, making decisions that in turn affect the state of the real world. Qualitative data were obtained from focus groups conducted with community-dwelling older adults who self-identify as African American, Dominican, or Puerto Rican to elicit their lived experiences in accessing oral health care in their northern Manhattan neighborhoods. Results: The findings of this study support the multi-dimensional and multi-level perspective of access to oral health care and affirm a theorized discrepancy in fit between available dental providers and patients. The lack of information about oral health at the community level may be compromising the use and quality of oral health care among racial/ethnic minority older adults. Conclusions: Well-informed community members may fill critical roles in oral health promotion, as they are viewed as highly credible sources of information and recommendations for dental providers. The next phase of this research will involve incorporating the knowledge gained from this study into simulation models that will be used to explore alternative paths toward improving oral health and health care for racial/ethnic minority older adults. Keywords: Oral public health, Dental public health, Oral health equity, Systems science, Agent-based modeling, Qualitative analysis, Focus group analysis, Racial/ethnic minorities, Older adults, Community-based oral health care * Correspondence: men6@nyu.edu Department of Epidemiology and Health Promotion, New York University College of Dentistry, 433 First Avenue, Room 726, New York, NY 10010, USA Department of Sociomedical Sciences, Columbia University Mailman School of Public Health, 722 West 168th Street, New York, NY 10032, USA Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Kum et al. BMC Oral Health (2018) 18:99 Page 2 of 11 Background Several conceptual models for studying access to Since the end of World War II and the subsequent health services have been proposed. For instance, a increase in living standards and community water framework devised by Penchansky and Thomas defines fluoridation [1], the US population has benefitted access as the fit between service providers and service from substantial reductions in tooth decay and eden- users [12]. There are 5 dimensions for which fit is tulism (complete tooth loss). The declining preva- compared: (1) accessibility, i.e., the locations of service lence of edentulism in the older adult population providers versus the locations of service users; (2) avail- means there is greater retention of natural teeth ability, i.e., the volume and type (supply) of services ver- requiring oral health care as people age [2, 3]. Pre- sus the volume and types of needs; (3) affordability, i.e., vention and treatment efforts, technological advance- the cost of services versus the ability of service users to ments in dentistry, and dental insurance have also pay for them; (4) accommodation, i.e., the approach to contributed to improved oral health [4]. Yet despite provision by service providers versus the perceptions of these documented improvements in oral health over appropriateness by service users; and (5) acceptability, time, the major oral diseases—dental caries and peri- i.e., service users’ reactions to and expectations of ser- odontal diseases—are still among the most prevalent vice providers versus service providers’ reactions to and chronic conditions in the US population across the expectations of service users [12]. Discrepancies in fit in- life span [5]. fluence use of services, satisfaction of service users, and There remain critical unmet oral health needs among practices of service providers. children and adolescents, young and working age adults, Powell offers an alternative taxonomic definition of ac- and older adults. While the public views oral health as a cess to health services that considers social accessibility priority, oral health concerns are inadequately addressed to be comprised of 3 dimensions: affordability, accom- by public policies, especially for older adults [6]. For in- modation, and acceptability [13]. In the present analyses, stance, because the Medicare program in the United the question asked was: “Why do older adults go to a States for persons aged 65 years and older and disabled dental provider?” Responses were used to determine as- adults does not cover routine dental care, many older pects of dental providers and oral health care settings adults are unable to afford the necessary preventive and that influence older adults’ decisions about where they restorative treatments they need [7]. Racial/ethnic mi- seek oral health care, i.e., social accessibility. nority older adults are at higher risk for edentulism than the white majority population [8]. They have also been Methods found to be less likely to report dental visits in the past Qualitative methods year, perhaps due to language barriers, especially among Valuable information for understanding and addressing the foreign-born [9]. In addition to experiencing poorer problems is often held in qualitative forms of data, such clinical measures of oral health, racial/ethnic minority as in mental models and written texts [14]. Similarly, the older adults also report worse self-rated oral health than humanistic approach emphasizes the lived experiences their white counterparts [10]. and personal histories of individuals [15]. Narratives To effectively address oral health and health care in- may provide meaning and context to an individual’s ex- equities requires reforming public policies for how oral periences with illness and recovery [16]. Interpreting health services are financed and delivered [6]. There are narratives may reveal characteristics associated with an multiple identified factors operating at different scales individual’s health status that cannot otherwise be de- that contribute to individual and population oral health tected, such as hope, despair, fear, guilt, and grief. Narra- inequities [11]. Therefore, strategies are needed to sup- tives may point to new hypotheses and stimulate more port and advance oral health promotion and treatment patient-centered research [17]. initiatives, along with oral health policy changes at local, state, and national levels. These strategies include Focus group approach and participants for the study place-based programs that apply principles of geographic Focus groups were conducted with a sample of 194 ra- targeting or directed population approaches to promote cial/ethnic minority men and women aged 50 years and oral health equity. A primary objective of these strategies older living in northern Manhattan who participated in is to identify gaps in dental services and population oral one of 24 focus group sessions about improving oral health needs. Toward advancing research designed to health for older adults [18]. The investigators of the promote oral health equity, this paper presents an ap- study selected focus groups over individual interviews proach to explicating and incorporating the perceptions because group discussions may facilitate greater disclos- and knowledge of African American, Dominican, and ure by participants through reciprocity, i.e., disclosure by Puerto Rican older adults related to oral health care in one participant may prompt greater disclosure by others their neighborhoods. [19]. Further, focus groups allow participants to respond Kum et al. BMC Oral Health (2018) 18:99 Page 3 of 11 to and elaborate on topics raised by fellow participants, / Inwood (home to large numbers of Dominicans), thus facilitating discussion of a greater breadth of topics and East Harlem (home to large numbers of Puerto [19]. Finally, focus groups may be less fatiguing than in- Ricans). These 3 neighborhoods have historically been dividual interviews, which may be particularly important considered as racial/ethnic enclaves, with large num- when interviewing older adults [20]. bers of recent immigrants and many residents qualify- Focus group participants had to meet the following ing for Medicaid and other forms of public assistance. criteria: (1) aged 50 years or older; (2) attended a senior Further details of the recruitment and screening pro- center or other community locale where older adults cedures are available elsewhere [18]. gather in northern Manhattan; (3) speak fluent English The study design of 24 focus groups was selected a or Spanish; and (4) self-identify as African American, priori in order to obtain multiple groups of each Dominican, or Puerto Rican. The demographic charac- demographic segment, thereby allowing conclusions teristics of the focus group participants overall and by about each demographic segment to be based on gender are presented in Table 1. multiple focus group discussions rather than on a sin- gle focus group discussion. Consistent with standard focus group techniques [21], thegroupswereseg- Recruitment procedure, sampling strategy, and context mented based on important characteristics that may Field recruitment staff visited senior centers in northern influence the issues discussed or the ability of the Manhattan and directly approached older adults to ex- members to build rapport. A total of 24 focus groups plain the study, screen them for eligibility, and solicit were conducted, including 12 groups of men and 12 participation in the focus groups. Senior centers were se- groups of women. Within each gender set, 4 groups lected rather than places where older adults receive den- were conducted with African Americans, 4 groups tal care in order to obtain a sample of individuals who were conducted with Dominicans, and 4 groups were did not necessarily have access to, or seek, dental care. conducted with Puerto Ricans. Within each gender / To ensure geographic and demographic representa- ethnic / racial set, half of the groups were conducted tion of northern Manhattan, approximately equal with participants who had visited a dentist in the past numbers of participants were recruited from senior year and half were conducted with participants who centers in each of three northern Manhattan neigh- had not visited a dentist in the past year. Ten groups borhoods: Central / West Harlem (home to large were conducted in English (including two groups with numbers of African Americans), Washington Heights Table 1 Characteristics of participants in focus groups for the total sample and by gender, New York, NY, 2013–2015 Participants and Focus Groups Total Sample Women Men Participants N = 194 n = 104 n =90 Focus groups N =24 n = 12 n =12 Characteristics % (n) % (n) % (n) Age group in years 50–59 14.4% (28) 16.3% (17) 12.2% (11) 60–69 34.0% (66) 32.7% (34) 35.6% (32) 70–79 36.1% (70) 34.6% (36) 37.8% (34) 80–89 11.9% (23) 11.5% (12) 12.2% (11) 90+ 3.6% (7) 4.8% (5) 2.2% (2) Race/ethnicity Dominican 35.6% (69) 33.7% (35) 37.8% (34) Puerto Rican 27.3% (53) 27.9% (29) 26.7% (24) African American 37.1% (72) 38.5% (40) 35.6% (32) Last dental visit Within past year 54.1% (105) 52.9% (55) 55.6% (50) 1–3 years ago 27.3% (53) 31.7% (33) 22.2% (20) > 3 years ago 18.6% (36) 15.4% (16) 22.2% (20) Primary language English 42.3% (82) 46.2% (48) 37.8% (34) Spanish 48.5% (94) 45.2% (47) 52.2% (47) Both 9.3% (18) 8.7% (9) 10.0% (9) Women and men did not differ significantly on any of the characteristics listed above, in accordance with the sampling strategy Kum et al. BMC Oral Health (2018) 18:99 Page 4 of 11 Puerto Ricans who preferred to speak English) and 14 specify the direction of cause and effect. A polarity is groups were conducted in Spanish. An average of 8 older assigned to each relationship indicating whether the ef- adults participated in each of the 24 focus groups. fect (variable Y) increases or decreases relative to the cause (variable X) ceteris paribus, i.e., holding all else Systems science methods equal. Paina and Peters propose viewing health systems A protocol proposed by Kim and Andersen for cod- through the lens of complex adaptive systems to offer ing qualitative text data and developing causal maps insights and inform planning, implementing, monitor- from purposive text data was adapted here to analyze ing, and evaluating more effective, equitable, sustainable, decisions on where older adults seek oral health care and context-relevant approaches to population health [14]. The initial stage involves critically evaluating the needs and demands [22]. For the study of complex text to extract data segments and cause-and-effect re- phenomena, systems science provides a variety of lationships. This corresponds to open coding, i.e., the research methods that are complementary to qualitative summarizing and division of text that captures spe- approaches [23]. These include system dynamics, cific phenomena or experiences that may be used as agent-based modeling, social network analysis, and geo- codes. As codes are grouped, dominant patterns or graphic information science [24–28]. themes emerge and are observed and subsequently Among the findings of a recent systematic review is coded. The next stage involves the grouping together that soft systems modeling techniques are likely to be of similar data segments to arrive at more generalized the most useful addition to public health [29]. This is cause-and-effect relationships and implicit structures because the methodological positioning and subsequent in order to construct a composite causal map. This metaphors in systems science, such as feedback, accu- step corresponds to axial coding, i.e., the organization mulation, and endogenous behavior, provide a way to of relationships among categories of codes to establish conceptualize complex and politically sensitive problems relationships between categories. Implicit structures, and policies of the health system, and in the process, fa- i.e., intermediate variables, guide the merging of cat- cilitate knowledge transfer among researchers, practi- egories of codes. tioners, and policymakers [29]. On the other hand, there A topic guide consisting of open-ended questions was also needs to be greater accountability in hard systems developed by the research team and used by the focus modeling, especially in terms of explicating the proce- group moderators to facilitate the sharing of personal dures used to build computer models. This would allow experiences among participants (see Additional file 1). for other researchers and practitioners to contribute to The initial categories for organizing the qualitative data the modeling process and assess whether or not the were based upon the topic guide; additional categories right questions are being addressed. More realistic, were included during subsequent readings of the tran- data-intensive computer models may lack transparency scripts. Focus groups with content related to categories and flexibility, becoming too complicated and difficult to of interest were indicated and recorded. An iterative comprehend to be of practical use [30, 31]. process was used to extract excerpts of transcript text, A causal map is a visual artifact used in the field of which are referred to as data segments. system dynamics, a scientific approach that involves repre- Data segments are typically a single response from a senting, testing, and modifying assumptions about dynam- participant, i.e., a portion of text indicated in the focus ically complex problems attributed to feedback loops and group by “P:” indicating “Participant:” in the transcripts. delays [32]. Causal mapping is a method for articulating a For certain data segments, it was important to capture plausible explanation for these dynamically complex prob- the dialogue. Hence, a forward slash (/) is used to indi- lems that is used to identify feedback loops. Causal maps cate multiple responses. Cause-and-effect relationships are integral elements in a preliminary blueprint for com- are then identified from data segments. puter models, and in this research, are considered to be Table 2 presents the topics by type (either finding den- boundary objects [33]. Black defines a boundary object as tal providers or going to dental providers) identified “a representation—perhaps a diagram, sketch, sparse text, from the 24 focus groups. or prototype—that helps individuals collaborate effectively Topics refer to the pattern or theme that was deter- across some boundary, often a difference in knowledge, mined by examining a grouping of similar data segments training, or objective.” ([34], 76, p). and were identified at the time of extraction of data segments. Labels were then assigned to types and topics Integrating qualitative and systems science methods in to facilitate the filtering of the data segments in a the data analysis spreadsheet. Next, each data segment was assigned a Importantly, a causal map depicts cause-and-effect rela- data segment Type, Topic, and identification number tionships, where arrows between variables are used to (SegmentID). When assigning SegmentIDs, replicate Kum et al. BMC Oral Health (2018) 18:99 Page 5 of 11 Table 2 Decisions on where to go for oral health care organized by types and topics Types (in shaded rows) and Topics Label Finding dental providers finding Information from insurance company insurance Third-party resources (1-800-Dentist, 311, Internet) third-party Referral from another health provider referral Recommendations from friends, family, community recommendation Going to dental providers going Recommendations from friends, family, community recommendation Dental care environment environment Training and credentials (title) of oral health providers credentials Character of oral health providers character Shared language or culture and understanding of community understanding (Dis)Trust in oral health provider trust Quality of dental care quality Dental treatment options treatment Dental need (urgent care) need Appointments appointments Insurance coverage and cost of care cost entries were removed that had been introduced errone- effect (cmEffect). For example, Fig. 1c presents the ously during data extraction. cmCause and cmEffect to which SegmentID 56 contrib- uted. Data segments in the same topic group were com- Results pared to derive a term that was representative of the Figure 1 provides an illustration of the process by which relationships expressed in that subset of data segments. a data segment and corresponding focus group segmen- Deliberate attempts were made to use variables that tation information is recorded and organized. were already assigned; new variables were typically gen- Certain data segments yielded more than a single erated to clarify relationships. Each of the generalized re- cause-and-effect relationship; thus, there are 287 rela- lationships was then assembled in Vensim software [35] tionships from 240 data segments. The data segment to provide a visual overview of the existing relationships. shown in Fig. 1a is SegmentID 56, which was extracted Certain of the generalized relationships were collapsed from focus group 5, comprised of Dominican men who during the process of assembly and intermediate struc- had received a dental visit in the past year (translated tures were added. The order in which the variables were into English from Spanish). This data segment was used assigned a cmID, indicating the generalized cause and to explicate 2 cause-and-effect relationships: the first for effect relationships, followed the descending order of finding dental providers and the second for going to topics presented in Table 1. One of the generalized rela- dental providers. This is an especially interesting data tionships in a topic group would often provide a link to segment because it represents both the view of the another topic group. To complete the example, Fig. 1d speaker and the view of his friend. presents an illustration of how the generalized A data segment cause (dsCause) and a data segment cause-and-effect relationships are summarized and effect (dsEffect) were then identified for each data tabulated. segment. Attempts were made to extract terms from The behavior (noted simply as Behavior) of the each data segment to express dsCause and dsEffect. To cmCause and cmEffect was then assessed. The Behavior continue with the illustration, Fig. 1b presents the was articulated to check the logic of each cause and ef- cause-and-effect relationships identified for SegmentID fect relationship and assign a polarity to that relation- 56. For data segments where it was difficult to maintain ship. The polarity of the relationship, that is, whether complete data integrity, terms derived from the inter- the behavior of the cause and effect changes in the same pretation of the first author for the data segment direction (positive) or in the opposite direction (nega- were used. tive), is recorded in Relationship Type. For a positive re- Similar expressions of cause and effect relationships, lationship type, an increase (decrease) of the cause leads aided by the filtering of labels, were then generalized to an increase (decrease) of the effect; for a negative re- into a causal map cause (cmCause) and a causal map lationship type, an increase (decrease) of the cause leads Kum et al. BMC Oral Health (2018) 18:99 Page 6 of 11 Fig. 1 Process by which a data segment is recorded and organized. An illustration of the process by which an extracted data segment is recorded and organized. Specifically, data segment 56 (SegmentID 56) originates from focus group 5 conducted with Dominican men who received oral health care in the past year (translated into English from Spanish). The 4 panels correspond to the following steps: (a) Data segment is identified and extracted; (b) Cause and effect relationships are explicated; (c) Similar expressions of cause and effect relationships contribute to a generalized cause-and-effect relationship; and (d) Generalized cause-and-effect relationships are summarized and tabulated to a decrease (increase) of the effect. A causal map ID and data segments. Instead of attempting to capture all (cmID) was assigned after all of the 37 generalized of the attendant details, data segments were aggregated cause-and-effect relationships were established. To to highlight the major constructs and the potential complete the illustration, a summary and tabulation of mechanisms that connect these constructs. Both the task the generalized cause-and-effect relationships used to of manually extracting and organizing the text and the systematically construct the causal map is presented in task of generalizing the selected text were time intensive. Fig. 1d, with the contribution of SegmentID 56 Yet it was only through multiple and iterative readings highlighted in yellow. of the text that the analyses were effectively framed. This Modifications to the Kim and Andersen protocol [14] was especially important in generating the causal map, for constructing a causal map were motivated by the since justifying the underlying logic of relationships re- need to be able to efficiently compare different topics quired considerable abstraction. Kum et al. BMC Oral Health (2018) 18:99 Page 7 of 11 Finally, a composite causal map representing the experi- The first column in Table 3 lists the 19 variables in the ence of the patient was systematically constructed to causal map that are included in any of 12 feedback loops explicate the generalized relationships involved in the involved in finding or going to a dental provider. decision-making process of dental provider choice (Fig. 2). All 12 feedback loops include the focal variable dental These considerations correspond to the access dimen- visits. Three feedback loops are associated with finding sions of acceptability, accommodation, and affordability, dentists – knowledge about dental provider – and 9 i.e., social accessibility to oral health care. Two separate feedback loops are associated with going to dentists – but related issues are manifest: the first involves finding motivation for dental visit. Variables that are not dental providers, knowledge about dental providers, and included in any feedback loops are listed in the last the places where dental providers practice; and the sec- column of Table 2 as preceding variable(s), i.e., a cause ond involves going to and returning to dental providers, of the associated variable in the feedback loop. more specifically, the circumstances at dental practices Below illustrative data segments are presented and dis- that motivate dental visits. cussed, including those related to the example in Fig. 1 The focal variable of the causal map is dental visits (SegmentID 56). Participant responses for how they (highlighted in orange), an indicator of utilization of or found dental providers included: resources from insur- realized access to the oral health care system. The two ance plans; phone calls to third-party services; Internet issues of finding and going to dentists are represented, searches; referrals from health care providers; and rec- respectively, as knowledge about dental provider and ommendations from relatives and friends. For example, motivation for dental visit (also highlighted in orange an African American man explained how he was re- in the causal map). ferred by a dental provider to another dental provider Fig. 2 Causal map derived from focus group data. A composite causal map of decisions on where to go for oral health care based on information extracted from focus groups with African American, Dominican, and Puerto Rican older adults. The solid arrows indicate a positive effect (same direction), whereas the dashed arrows indicate a negative effect (opposite direction) Kum et al. BMC Oral Health (2018) 18:99 Page 8 of 11 Table 3 Feedback loops of decisions on where to go for oral health care for specialized care: “So, finally you got a lot of dentists practices towards oral health care. For example, an who are really at another level. I’ve been to dentist who African American man shared the following remarks: have been like, “Listen, I can’t do this, but your union is “She’ll [wife] find a dentist. [pause]. She’ll notice where going to pay.” Mr. [name] is a good dentist for, what do I’m taking all the pill because it don’t matter to me. I just you call it? The root canal!” (focus group 17: African went to bed. I won’t notice [pause] things like that. But American men with a dental visit in the past year) she will. She’s good that way.” (focus group 17: African Family members and friends were considered to be both American men with a dental visit in the past year) Older important sources of recommendations for finding den- adults who lack the support of family and friends may tal providers and influential in going to dental visits. neglect their health and encounter difficulties in obtain- The role of referrals and recommendations in finding ing resources to address their health needs, such as dentists are captured in feedback loops 1, 2, and 3; the information about oral health care options in their influence of communication about dental experiences neighborhoods and transportation assistance. are captured in feedback loops 4, 5, and 6. These feed- According to participants, information and opinions back loops indicate the potential of using social net- about dental providers and care settings often circulate works to deliver knowledge and change attitudes and through word of mouth among community members Kum et al. BMC Oral Health (2018) 18:99 Page 9 of 11 that may affect their reputations. The following quote inability to control for exposure times, selection bias due from a Dominican man emphasizes the impact of com- to loss to follow-up, and underrepresentation of racial/ munication in decision-making for selecting dentists: ethnic minorities [38]. Agent-based modeling has been proposed as a way forward since it facilitates representa- “There is something important that I want to explain tion of multiple scales and heterogeneity [39, 40]. More- here, the vast majority of us Hispanics look for our over, agent-based models are theory-based and doctors through references, this is very important. In data-driven, and allow for the testing of different plaus- other words if a doctor is good, the neighbor will say… ible mechanisms [41]. look, so and so is a wonderful dentist. Then we start A computer model may be used to communicate and looking for references. And it’s through the references learn about the impact of life events and social relation- that we start communicating with one another and we ships on oral health. The proposed model design con- even make an appointment to go see that doctor. tributes to an existing portfolio of models that have been Then, if the doctor is bad and did not complete the informed by quantitative and spatial data, as well as the work well, for this and that reason. It’s always like experiences and expertise of research team members that, the neighbor will communicate with the second [33, 42–46]. Here particular insights from older adults’ person and that spreads the word. In other words, a experiences that are reflected in focus groups transcripts job well done will be well received by the community, are included, along with supplementary knowledge about but people will also know about bad work.” (focus the older adult population in Manhattan and the Bronx. group 15: Dominican men with a dental visit in the The model is designed to simulate: (1) the life course, past year, translated into English from Spanish) i.e., life stages and life events; (2) oral health status, oral health care seeking orientation, and oral health care use; Several participants were frightened about going to the and (3) multiple social relationships as dynamic social dentist either because of fear and the associated pain of networks based on person agent attributes and geo- dental procedures or the fear of contracting diseases at graphic proximity. the dental office. Recommendations from trusted family Below are key insights gleaned from the focus groups members and friends may minimize such fears because that are considered important in the model design. First, dental providers and oral health care settings have the inability to pay for oral health care is a significant already been vetted (feedback loop 6). A Dominican barrier to accessing services. Second, family members woman confidently stated: “I say, ‘You are afraid? But let and friends are both important sources of recommenda- me tell you, I have a dentist you will love. Here is the tions for finding dental providers and influential in mo- phone number. Here is the phone number. Go to him tivating dental visits. Third, participants believe there is and you will remember me.’” (focus group 14: Dominican a lack of information about oral health in the commu- women without a dental visit in the past year, translated nity and they would like more information about oral into English from Spanish). health and health care. The design includes 2 active agent classes: person Discussion (older adult, dental provider) and place. Place is further In systems science, both the problems and their specified into home, work, third place, e.g., senior center, solutions are understood as being generated from within public library, religious institution, and dental clinic [46]. the system. Social disparities in oral health result from Characteristics of all place agents will include a unique factors at multiple scales [11]. With regular dental identification number and an indication of status (either hygiene and professional care, adverse oral health out- open or closed). An additional characteristic of dental comes such as tooth loss may often be prevented. Oral clinics is the types of insurance accepted. Further, the health behaviors and practices are transferred through identification numbers of older adults who visited each social relationships throughout the life course, which dental clinic and the dates of these visits as well as re- may either increase or decrease social disparities in oral minder messages to older adults regarding scheduled ap- health. For example, children often adopt the behaviors pointments will need to be tracked. An additional and practices of their parents. Social relationships characteristic of third places is whether health outreach among older adults provide mechanisms for the ex- events are held at each location, and if so, the identifica- change of resources, such as information that is critical tion number of participants at outreach events, and the in decision-making. participants who needed a referral to a dental clinic affil- While the life course approach is useful in understand- iated with the outreach events. ing oral health inequities, there are empirical challenges According to the focus group transcript analysis, avail- in testing the proposed theoretical models [36, 37]. ability (the supply of dental providers) and accessibility Problems with using retrospective studies include the (the means of traveling to oral health care) were not Kum et al. BMC Oral Health (2018) 18:99 Page 10 of 11 perceived to be significant barriers to oral health care. fill critical roles in oral health promotion, as they are Rather, the inability to pay for oral health care, poor re- viewed as highly credible sources of information and lationships with dental providers, and lack of informa- recommendations to dental providers. Disseminating tion on oral health and health care were the major up-to-date information at frequented sites to older challenges. Therefore, the model design includes 2 types adults and the community at large about the importance of interventions: (1) social and behavioral interventions; of oral health, proper dental hygiene practices, and local and (2) policy interventions. The first set of modeled in- oral health care options remain public health priorities. terventions would involve community-based outreach education and delivery at places in the neighborhood Additional file that provide information, and use different social rela- tionships to direct information and influence changes in Additional file 1: Focused Group Interview Topic Guide. A topic guide consisting of open-ended questions that was developed by the research oral health care-seeking orientation and oral health care team and used by the focus group moderators to facilitate the sharing of status. The second set of proposed interventions would personal experiences among participants regarding reasons why people involve health insurance coverage, specifically, the im- may or may not visit a dentist. (PDF 57 kb) pact of expanding and ensuring dental insurance cover- age throughout the life span, and reducing restrictions Abbreviations cmCause: Causal map cause; cmEffect: Causal map effect; cmID: Causal map to preventive oral health care. identification number; dsCause: Data segment cause; dsEffect: Data segment effect; SegmentID: Data segment identification number Conclusions The findings of this study support the multi-dimensional Acknowledgments The authors thank the participants and recruitment staff whose engagement and multi-level perspective of access to oral health care in this qualitative study made the reporting of the results possible. and affirm a theorized discrepancy in fit between avail- able providers and patients. The presence of resources Funding does not directly translate into use of services by racial/ The authors were supported in the research, analysis, and writing of this paper by the National Institute for Dental and Craniofacial Research (NIDCR) ethnic minority older adults. Despite the relatively high and the Office of Behavioral and Social Sciences Research (OBSSR) of the US volume of dental providers and the range of transporta- National Institutes of Health (NIH) for the project titled, “Integrating Social tion options, focus group participants did not believe and Systems Science Approaches to Promote Oral Health Equity” (grant R01- DE023072) and by the National Center for Advancing Translational Sciences that their oral health needs were being adequately ad- (NCATS) of the NIH for the project titled, “Primary Care Screening by Dental dressed, whether or not they had recently visited a Hygienists at Chairside: Developing and Evaluating an Electronic Tool” (grant dentist. UL1TR000038). The funding agencies had no role in the design of the study and collection, analysis, and interpretation of data or in the writing of the The systematic approach to explicating cause and ef- manuscript. fect relationships from focus group transcripts intro- duced here may prove transferable to other research Availability of data and materials contexts. The product of this approach, a causal map, De-identified raw data and materials described in the manuscript are freely available from the corresponding author on reasonable request. provides a visual representation of major factors and re- lationships involved in the decision-making process. Authors’ contributions From both epistemological and ontological stand- SSK led the writing of this paper and conducted the analyses as part of her dissertation research. MEN closely edited the paper, provided interpretation, points, however, system dynamics involves more than finalized the tables and figures, and contributed oral health and public the mechanics of creating a causal map. Rather, there is health expertise. SSM conceived of and supervised the analyses, and a philosophical understanding that in order to solve contributed systems science expertise. All authors contributed to the writing and editing of this paper and approved it for publication. large, complex problems, it is important and effective to consider the needs of others [47]. The ability to incorp- Ethics approval and consent to participate orate qualitative data into a causal map allows direct in- This research has been performed in accordance with the Declaration of clusion of the views of underrepresented populations Helsinki. The following Institutional Review Boards reviewed and approved all study procedures: Columbia University Medical Center Institutional Review into the hypothesized cause and effect mechanisms Board [protocol AAAL4104(M01Y04)] and NYU School of Medicine explicated. Institutional Review Board (protocol i12-02947_CR4). All Health Insurance The lack of information about oral health may be Portability and Accountability Act safeguards were followed. All participants signed written consent forms. compromising the use and quality of oral health care among racial/ethnic minority older adults. This finding Competing interests is consistent with key informant views that senior center The authors declare that they have no competing interests. attendees did not regard oral health concerns with the same degree of immediacy as high blood pressure (indi- Publisher’sNote cative of hypertension) or high blood sugar (indicative of Springer Nature remains neutral with regard to jurisdictional claims in diabetes) [48]. Well-informed community members may published maps and institutional affiliations. Kum et al. BMC Oral Health (2018) 18:99 Page 11 of 11 Author details 23. Ip EH, Rahmandad H, Shoham DA, et al. Reconciling statistical and systems science Department of Geography, The State University of New York at Buffalo, 115 approaches to public health. Health Educ Behav. 2013;40(1 suppl):123S–31S. Wilkeson Quad, Ellicott Complex, Buffalo, NY 14261-0055, USA. Department 24. Metcalf SS, Northridge ME, Lamster IB. A systems perspective for dental of Population Health, New York University School of Medicine, 227 East 30th health in older adults. Am J Public Health. 2011;101(10):1820–3. Street, 8th Floor, New York, NY 10016, USA. Department of Epidemiology 25. McLeroy K. Thinking of systems. Am J Public Health. 2006;96(3):402. and Health Promotion, New York University College of Dentistry, 433 First 26. Mabry PL, Marcus SE, Clark PI, Leischow SJ, Méndez D. Systems science: a Avenue, Room 726, New York, NY 10010, USA. Department of Sociomedical revolution in public health policy research. Am J Public Health. 2010;100(7): Sciences, Columbia University Mailman School of Public Health, 722 West 1161–3. 168th Street, New York, NY 10032, USA. 27. Luke DA, Stamatakis KA. Systems science methods in public health. Annu Rev Public Health. 2012;33:357–76. Received: 10 April 2017 Accepted: 22 May 2018 28. Torrens PM. Geography and computational social science. GeoJournal. 2010; 75(2):133–48. 29. Carey G, Malbon E, Carey N, Joyce A, Crammond B, Carey A. Systems science and systems thinking for public health: a systematic review of the field. BMJ Open. 2015;5(12):e009002. References 30. Ness RB, Koopman JS, Roberts MS. Causal system modeling in chronic 1. Centers for Disease Control and Prevention (CDC). Achievements in public disease epidemiology: a proposal. Ann Epidemiol. 2007;17(7):564–8. health, 1900–1999. Fluoridation of drinking water to prevent dental caries. 31. O’Sullivan D, Evans T, Manson S, Metcalf S, Ligmann-Zielinska A, Bone C. MMWR Morb Mortal Wkly Rep. 1999;48(41):933–40. Strategic directions for agent-based modeling: avoiding the YAAWN 2. Centers for Disease Control and Prevention (CDC). Public health and aging: syndrome. J Land Use Sci. 2016;11(2):177–87. retention of natural teeth among older adults–United States, 2002. MMWR 32. Sterman J. Business dynamics: systems thinking and modeling for a Morb Mortal Wkly Rep. 2003;52(50):1226–9. complex world. Boston: Irwin/McGraw-Hill; 2000. 3. Slade GD, Akinkugbe AA, Sanders AE. Projections of U.S. edentulism 33. Northridge ME, Metcalf SS. Enhancing implementation science by applying prevalence following 5 decades of decline. J Dent Res. 2014;93(10):959–65. best principles of systems science. Health Res Policy Syst. 2016;14:74. 4. Kiyak HA, Reichmuth M. Barriers to and enablers of older adults’ use of 34. Black LJ. When visuals are boundary objects in system dynamics work. Syst dental services. J Dent Educ. 2005;69(9):975–86. Dyn Rev. 2013;29(2):70–86. 5. Benjamin RM. Oral health: the silent epidemic. Public Health Rep. 2010; 35. Ventana Systems, Inc. Vensim. Harvard; 2015. 125(2):158–9. 36. Ben-Shlomo Y, Kuh D. A life course approach to chronic disease 6. Fisher-Owens SA, Barker JC, Adams S, et al. Giving policy some teeth: epidemiology: conceptual models, empirical challenges and interdisciplinary routes to reducing disparities in oral health. Health Aff (Millwood). 2008; perspectives. Int J Epidemiol. 2002;31(2):285–93. 27(2):404–12. 37. Nicolau B, Thomson WM, Steele JG, Allison PJ. Life-course epidemiology: 7. Greenblatt AP, Estrada I, Schrimshaw EW, Metcalf SS, Kunzel C, Northridge concepts and theoretical models and its relevance to chronic oral ME. Acceptability of chairside screening for racial/ethnic minority older conditions. Community Dent Oral Epidemiol. 2007;35(4):241–9. adults: a qualitative study. JDR Clin Trans Res. 2017;2(4):343–52. 38. Shavers VL. Measurement of socioeconomic status in health disparities 8. Wu B, Liang J, Plassman BL, Remle RC, Luo X. Edentulism trends among research. J Natl Med Assoc. 2007;99(9):1013–23. middle-aged and older adults in the United States: comparison of five 39. Diez Roux AV. Complex systems thinking and current impasses in health racial/ethnic groups. Community Dent Oral Epidemiol. 2012;40(2):145–53. disparities research. Am J Public Health. 2011;101(9):1627–34. 9. Shelley D, Russell S, Parikh NS, Fahs M. Ethnic disparities in self-reported oral 40. Noble J, Silverman E, Bijak J, et al. Linked lives: the utility of an agent-based health status and access to care among older adults in NYC. J Urban Health. approach to modeling partnership and household formation in the context 2011;88(4):651–62. of social care. In: Proceedings of the 2012 winter simulation conference; 10. Huang DL, Park M. Socioeconomic and racial/ethnic oral health disparities 2012. p. 1–12. among US older adults: oral health quality of life and dentition. J Public 41. Salgado M, Marchione E, Gilbert N. Analysing differential school Health Dent. 2015;75(2):85–92. effectiveness through multilevel and agent-based modelling. J Artif Soc Soc 11. Northridge ME, Ue FV, Borrell LN, et al. Tooth loss and dental caries in Simul. 2014;17(4):3. community-dwelling older adults in northern Manhattan. Gerodontology. 42. Metcalf SS, Northridge ME, Widener MJ, Chakraborty B, Marshall SE, Lamster 2012;29(2):e464–73. IB. Modeling social dimensions of oral health among older adults in urban 12. Penchansky R, Thomas JW. The concept of access: definition and environments. Health Educ Behav. 2013;40(1 suppl):63S–73S. relationship to consumer satisfaction. Med Care. 1981;19(2):127–40. 43. Wang H, Northridge ME, Kunzel C, et al. Modeling social capital as dynamic 13. Powell M. On the outside looking in: medical geography, medical networks to promote access to oral healthcare. Soc Cult Behav Model. 2016; geographers and access to health care. Health Place. 1995;1(1):41–50. 9708:117. 14. Kim H, Andersen DF. Building confidence in causal maps generated from 44. Chakraborty B, Widener MJ, Mirzaei Salehabadi S, et al. Estimating peer purposive text data: mapping transcripts of the Federal Reserve. Syst Dyn density effects on oral health for community-based older adults. BMC Oral Rev. 2012;28(4):311–28. Health. 2017;17(1):166. 15. Phillips J, Ajrouch KJ, Hillcoat-Nallétamby S. Key concepts in social 45. Zhang Q, Northridge ME, Jin Z, Metcalf SS. Modeling accessibility of gerontology. Los Angeles: SAGE; 2010. screening and treatment facilities for older adults using transportation 16. Sweeney K. Complexity in primary care: understanding its value. Oxford; networks. Appl Geogr. 2018;93:64–75. Seattle: Radcliffe Pub.; 2006. 46. Northridge ME, Kum SS, Chakraborty B, et al. Third places for health 17. Greenhalgh T, Hurwitz B. Why study narrative? BMJ. 1999;318(7175):48–50. promotion with older adults: using the consolidated framework for 18. Northridge ME, Shedlin M, Schrimshaw EW, et al. Recruitment of racial/ implementation research to enhance program implementation and ethnic minority older adults through community sites for focus group evaluation. J Urban Health. 2016;93(5):851–70. discussions. BMC Public Health. 2017;17(1):563. 47. Meadows DH. Thinking in Systems: A Primer. White River Junction: Chelsea 19. Morgan DL. Focus group interviewing. In: Gubrium JF, Holstein JA, editors. Green Publishing Co.; 2008. Handbook of interview research: context and method. Thousand Oaks: 48. Marshall S, Schrimshaw EW, Metcalf SS, et al. Evidence from ElderSmile for Sage; 2002. p. 141–59. diabetes and hypertension screening in oral health programs. J Calif Dent 20. Wenger GC. Interviewing older people. In: Gubrium JF, Holstein JA, editors. Assoc. 2015;43(7):379–87. Handbook of interview research: context and method. Thousand Oaks: Sage; 2002. p. 259–78. 21. Krueger RA, Casey MA. Focus groups: a practical guide for applied research. 4th ed. Thousand Oaks: Sage; 2009. 22. Paina L, Peters DH. Understanding pathways for scaling up health services through the lens of complex adaptive systems. Health Policy Plan. 2012; 27(5):365–73.

Journal

BMC Oral HealthSpringer Journals

Published: Jun 4, 2018

References

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