TY - JOUR
AU - Wald,, Heidi
AB - Abstract Background and Objectives Clinical decision support systems (CDSS) hold promise to influence clinician behavior at the point of care in nursing homes (NHs) and improving care delivery. However, the success of these interventions depends on their fit with workflow. The purpose of this study was to characterize workflow in NHs and identify implications of workflow for the design and implementation of CDSS in NHs. Research Design and Methods We conducted a descriptive study at 2 NHs in a metropolitan area of the Mountain West Region of the United States. We characterized clinical workflow in NHs, conducting 18 observation sessions and interviewing 15 staff members. A multilevel work model guided our data collection and framework method guided data analysis. Results The qualitative analysis revealed specific aspects of multilevel workflow in NHs: (a) individual, (b) work group/unit, (c) organization, and (d) industry levels. Data analysis also revealed several additional themes regarding workflow in NHs: centrality of ongoing relationships of staff members with the residents to care delivery in NHs, resident-centeredness of care, absence of memory aids, and impact of staff members’ preferences on work activities. We also identified workflow-related differences between the two settings. Discussion and Implications Results of this study provide a rich understanding of the characteristics of workflow in NHs at multiple levels. The design of CDSS in NHs should be informed by factors at multiple levels as well as the emergent processes and contextual factors. This understanding can allow for incorporating workflow considerations into CDSS design and implementation. Multilevel work, Clinical decision support, Information technology, Institutional/residential care, Organizational and institutional issues Clinical decision support systems (CDSS; mobile or desktop based) hold promise for influencing clinician behavior at the point of care in nursing homes (NHs) and improving care delivery (Alexander, Pasupathy, Steege, Strecker, & Carley, 2014; Handler et al., 2013; McMaughan et al., 2016). Antibiotic stewardship programs that support judicious, evidence-based use of antibiotics in NHs, can benefit from CDSS, because of the reliance of these programs on providers integrating detailed pharmacy data with clinical determinations that are subject to cognitive bias (Lim, Kong, & Stuart, 2014). However, the success of these interventions depends on the usability of the decision support applications (Jones et al., 2017; Sittig et al., 2008), including their clinical fit within the NH setting (Qian & Yu, 2014; Thorpe-Jamison, Culley, Perera, & Handler, 2013). Understanding workflow in NHs is essential to design and implement an effective CDSS in this clinical environment. Workflow can be defined as a sequence of activities by single or multiple roles in providing care (Ozkaynak & Brennan, 2012). This definition highlights three building blocks of workflow: activities, roles, and temporal relationships (e.g., sequence). Recent work has underscored the importance of context in examining workflow (Ozkaynak, Valdez, Holden, & Weiss, 2018). Workflow studies have been conducted in various clinical settings to inform the design and implementation of systematic organizational interventions (including CDSS) and to improve quality and safety of care (Ozkaynak, Unertl, Johnson, Brixey, & Haque, 2016). However, there are few studies that specifically focus on workflow in NHs (Alexander et al., 2014; Qian & Yu, 2013; Stokoe et al., 2016; Tariq, Georgiou, Raban, Baysari, & Westbrook, 2016). Within this limited scope of research, these studies focus on a specific task (or a select group of tasks) such as pressure ulcer-risk assessment (Alexander et al., 2014), prescribing (Tariq et al., 2016), medication management (Qian & Yu, 2013), and transfer to emergency departments (Stokoe et al., 2016). McCloskey, Donovan, Stewart, and Donovan (2015) examined how registered nurses, licensed practical nurses, and resident aides utilize their time in NHs. Although they identified no-value added activities, their design did not allow an understanding of how these activities are temporally organized or the possible relationship between these activities and structure and outcomes. There is a need for studies that look at workflows in NHs in a more holistic manner (i.e., a study with broader scope looking at various workflow elements simultaneously) to better understand the unique dynamics in NHs. As shown in other clinical settings, without sufficient workflow information, organizational interventions can lead to serious unintended consequences (Ozkaynak et al., 2014; Ozkaynak & Brennan, 2013). NH care entails distinct workflows (Morrill, Caffrey, Jump, Dosa, & LaPlante, 2016; Walker, McGeer, Simor, Armstrong-Evans, & Loeb, 2000) that include accommodation for the numerous daily living activities of residents and asynchronous communication between team members. This asynchrony often occurs because, unlike hospital settings, providers are not constantly present in facilities. The result is that nurses have an enhanced role in clinical decision making in response to clinical changes in residents (Lim, Kwong, et al., 2014). The purpose of this study was to characterize workflow in NHs and develop workflow guidelines to support the design and implementation of CDSS in NHs. We used a multilevel model (Holden & Karsh, 2009) to study workflow. The Holden and Karsh model conceptualizes work at four levels: clinicians, work group/unit, organization, and industry. This model suggests that health information technology (IT) should fit to work at each level. The utility of technology depends on its fit. We collected data in all the levels suggested by the model and used the concept of fit to inform the design and implementation of the CDSS. Multilevel models have been successful in capturing workflow comprehensively and delineating individual, unit, organizational activities, and their relationships (Ozkaynak et al., 2016). Organizations are naturally comprised of interacting levels such as departments, groups, and individuals (Kozlowski & Klein, 2000). Studying these levels separately provides insight into micro and macro subsystems of the examined organizations. Additionally, studying the interaction of these levels are essential to examining bottom-up emergent processes (Costa et al., 2013) and top-down mechanisms, that express the influence of higher level contextual factors on lower levels of the examined organizations (Costa et al., 2013). In the case of NHs, multilevel frameworks can provide insights about important work-related factors that affect the safety and quality of delivery of care (Siegel, Bakerjian, & Zysberg, 2017). These factors should be taken into account when designing health information technologies such as CDSS. Research Design and Methods We conducted a descriptive workflow study, using both direct field observations and semi-structured interviews. Qualitative methods were utilized because of the exploratory nature of the study. The study setting was two, for-profit NHs in a metropolitan area of the Mountain West Region of the United States. Both NHs had more than 100 beds. One facility (NH-1) had a Medicare Nursing Home Compare 5-star rating; the other (NH-2) had a Medicare Nursing Home Compare 4-star rating. Both had a skilled nursing facility unit (SNF), although the majority of beds were for long-term care (LTC). Both NHs were staffed with certified nursing assistants (CNAs) and nurses (Table 1). Nurses were licensed practical nurse (LPNs) or registered nurses (RNs). NH-1 employed 27 CNAs and 18 RNs/LPNs per day whereas NH-2 had 26 CNAs and 11 RNs/LPNs per 24-hr period. Of the staff members considered “active” by human resources, there were 21 LPNs, 15 RNs, and 53 CNAs in NH-1. In NH-2, of the staff members considered “active” by human resources, there are 21 LPNs, 17 RNs, and 93 CNAs. “Active” means that the staff member was PRN, full time, or part-time. In NH-1, there were three sections: long-term care, skilled care, and memory care and nurses and CNAs worked within only one section throughout a shift. NH-2 has five patient floors. One floor was strictly skilled care and four floors were long-term care. There was no designated memory care section in NH-2. This study was approved by the Colorado Multiple Institutional Review Board. Table 1. Study Participants (n = 31) Characteristics NH-1 NH-2 Total number of staff 15 RNs 17 RNs 21 LPNs 21 LPNs 53 CNAs 93 CNAs Nursing position of observation participants 3 RNs 3 RNs 3 LPNs 3 LPNs 3 CNAs 3 CNAs Nursing position of interview participants 2 RNs 5 LPNs 2 LPNs 3 CNAs 3 CNAs Average years in current role (participants) 7.7 (range 2–30) 8.0 (range 0.1–21) Average years in current facility (participants) 3.5 (range 0.3–8) 4.4 (range 0.1–16) Characteristics NH-1 NH-2 Total number of staff 15 RNs 17 RNs 21 LPNs 21 LPNs 53 CNAs 93 CNAs Nursing position of observation participants 3 RNs 3 RNs 3 LPNs 3 LPNs 3 CNAs 3 CNAs Nursing position of interview participants 2 RNs 5 LPNs 2 LPNs 3 CNAs 3 CNAs Average years in current role (participants) 7.7 (range 2–30) 8.0 (range 0.1–21) Average years in current facility (participants) 3.5 (range 0.3–8) 4.4 (range 0.1–16) Note: CNAs = certified nursing assistants; LPNs = licensed practical nurses; RNs = registered nurses. Open in new tab Table 1. Study Participants (n = 31) Characteristics NH-1 NH-2 Total number of staff 15 RNs 17 RNs 21 LPNs 21 LPNs 53 CNAs 93 CNAs Nursing position of observation participants 3 RNs 3 RNs 3 LPNs 3 LPNs 3 CNAs 3 CNAs Nursing position of interview participants 2 RNs 5 LPNs 2 LPNs 3 CNAs 3 CNAs Average years in current role (participants) 7.7 (range 2–30) 8.0 (range 0.1–21) Average years in current facility (participants) 3.5 (range 0.3–8) 4.4 (range 0.1–16) Characteristics NH-1 NH-2 Total number of staff 15 RNs 17 RNs 21 LPNs 21 LPNs 53 CNAs 93 CNAs Nursing position of observation participants 3 RNs 3 RNs 3 LPNs 3 LPNs 3 CNAs 3 CNAs Nursing position of interview participants 2 RNs 5 LPNs 2 LPNs 3 CNAs 3 CNAs Average years in current role (participants) 7.7 (range 2–30) 8.0 (range 0.1–21) Average years in current facility (participants) 3.5 (range 0.3–8) 4.4 (range 0.1–16) Note: CNAs = certified nursing assistants; LPNs = licensed practical nurses; RNs = registered nurses. Open in new tab Observations and interviews were conducted with a total of 31 staff members. One RN in NH-1 and one CNA in NH-2 participated in both the observation and interview. The data collection team consisted of a PhD-trained industrial engineer with workflow expertise, a PhD-trained informatics researcher and information system designer, a public health PhD student and a researcher with social work training. All team members had prior experience working together on CDSS projects and had qualitative research experience and training. Clinical workflow in the NHs was characterized from 18 observations (9 at each NH) and 15 interview sessions (7 in NH-1 and 8 in NH-2) from November 2016 through February 2017. Data collection was completed first in NH-1 followed by NH-2. Both observations and interviews were guided by a multilevel work model (Holden & Karsh, 2009) and workflow definition (Ozkaynak & Brennan, 2012). Multilevel theory suggests several hierarchical perspectives to understand work system, specifically individual, work group/unit, organization, and industry. The model implies that the fits between technology and individual, work group/unit, organization and industry aspects of workflow together, determine the appropriate use of a technology such as CDSS. Therefore, both observations and interviews aimed to capture these four aspects of workflow. Data collection was designed to capture various aspects of workflow at multiple levels, temporality, and important dimensions of context. Observation Sessions Observation notes captured: (a) contextual information about the time, place, and direct-care staff involved in the observation, (b) objective and subjective descriptions of events and conversations, and (c) information about staff members’ task, activities, conversations, and situational context. We sought to characterize routines and variations in tasks, individuals, information, relationships, tools, and technologies, as well as contextual factors shaping NH workflow (Ozkaynak et al., 2016, 2018). During the observation period, participants were asked to clarify information about the activities performed. These short-duration, purposive field interviews, were distinguishable from the detailed semi-structured interviews, in that they were conducted in the immediate context of the activities being performed. Observation sessions lasted 4 hr during different shifts; morning, n = 10; mid-day, n = 5; and evening, n = 3. Two members of the research team (M. Ozkaynak, B. Reeder, C. Drake, or P. Ferrarone), conducted each observation with M. Ozkaynak always one of the observers. Observers debriefed each other after the first two observation sessions, to establish consistency. Observation sessions also included reviewing documents and forms such as facility calendars, staff task management documents, data collection forms on residents’ needs and preferences, posted flyers and information sheets, manuals, and reference tools. Interviews Interviews with NH staff lasted on average 38.8 min (range 25.8–51.5 min). Each interview was conducted by two researchers (M. Ozkaynak, C. Drake, or P. Ferrarone). The interview guide was developed and tested in a previous design study of nursing work and adapted for the current research (Reeder, Ozkaynak, Makic, & Sousa, 2017). Researchers asked all the questions from the interview guide (Supplementary Appendix 1) of each participant, with follow-up questions as necessary. Interview questions enabled capture of daily routines, common and rare tasks, variety in activities, various factors that lead to variability, obstacles and facilitators of work, organizational and inter-organizational context and urinary tract infection (UTI) diagnosis. We examined UTI diagnosis for two reasons: (a) The parent project of this study was development of a mobile CDSS specific to UTI diagnosis; and (b) it provided a rich case study to gain insights for the workflow. Data collection (both observations and semi-structured interviews) ceased when the investigative team agreed there was redundancy in collected data and, thus, data saturation had been reached (Bowen, 2008). Analysis Qualitative analysis was based on a theory-driven approach (Boyatzis, 1998) and Gale and colleagues’ framework method of analysis (Gale, Heath, Cameron, Rashid, & Redwood, 2013). This method of analysis is particularly appropriate for multidisciplinary health research teams and fruitful in generating themes by making comparisons within and between cases. Each observer digitally transcribed his or her own observations from handwritten notes. Both sets of notes from each observation session were reviewed by the first author, who sought clarification on any differences in observations. Audio recordings of interviews were transcribed verbatim by a professional research assistant. The research team familiarized themselves with the data as a whole by reading notes and transcriptions, and through discussion during regular research team meetings as well as ad hoc meetings of subgroups from the research team. The first author attended all meetings. Six primary codes (individual, tasks, organization, tools and technology, environment, and temporality) were identified a priori by synthesizing the human factors and workflow literature (Carayon et al., 2006; Ozkaynak et al., 2016) to develop a theory-based codebook (Boyatzis, 1998). These predetermined codes were consistent with the multilevel model guiding the study. Two sets of observation notes and one interview transcript were coded by three coders (M. Ozkaynak, C. Drake, B. Reeder) to refine the codebook and understand its application. Initial coding results were reconciled to identify subcodes of the six primary codes, producing a combined total of 27 codes and subcodes in the final codebook (Supplementary Appendix 2). Three researchers (M. Ozkaynak, C. Drake, P. Ferrarone) analyzed the remaining observation notes and interview transcripts using the codebook and ATLAS.ti 6 (Scientific Software Development GmbH, Berlin, Germany). Led by one coder (MO), with input from the research team. Emergent themes were identified from the coded data. Final results were member-checked for accuracy through follow-up with five nurses and two CNAs in May 2017. Results Analysis characterized themes of multilevel workflow in NHs. Themes included the impact of individual, work group/unit, organization, and industry factors on workflow. Results are described in detail below. Individual This level of work is concerned with characteristics of individual staff members and their activities. We identified a wide variety of individual tasks (goals) and activities (steps taken to achieve the goal) conducted in NHs. The majority of the RN’s time was for medication management. In contrast, CNAs spent the majority of their time assisting with activities of daily living (ADLs) such as bathing, transferring from bed to chair, etc. Although both CNAs and RNs may use the assistance of a second person (e.g., ambulating a frail resident), CNAs more often worked in pairs. Between these two routine activities (medications and ADL), both RN and CNA roles perform latent (i.e., underlying) activities. These include: establishing baseline information (e.g., health status and daily routines) for new residents; maintaining baseline information for current residents; time management; task prioritizing; supporting the resident’s sense of independence; and handling interruptions. [When I start my day] ... “I make a list of everything that I need to do, then I prioritize. The medication is usually first…” (Nurse from NH-2) Priorities. Priorities. You have to look for priorities. If you’re very busy. ‘Ok, now I need to change your bed. Look, it’s wet.’ This person is sick. He had an accident. This is a priority. I can do your bed later. The next person can do your bed. I can do your bed later. You have to be able to prioritize (CNA from NH-1) [for a dementia patient] You know if that’s their baseline, or if they’re just confused. You need to know your patient. (Nurse from NH-2) These latent activities, occurring on daily basis, were support activities that impacted the effectiveness and efficiency of the main activities (medication management and assisting with ADL’s). For example, the first two excerpts above highlight the importance of time management (i.e., prioritizing) to complete all needed responsibilities on a given day. In the third excerpt, the pre-established baseline is decisive for a nurse to determine the next course of action for the patient. Latent activities may be necessary to accomplish the main activities within a given time and with available resources. Observation notes highlighted that the computer literacy of NH staff was highly variable. Some staff members were very comfortable using institutionalized electronic health records on desktop computers as well as on touch screens mounted on wall. These staff members were able to use a wide variety of functions in a rapid manner. However, some staff members used only essential functions and were much slower in completing entries or searching for information. These individual characteristics of users, affect the use of information systems that are essential for delivering care. Work group/unit This level of work is concerned with characteristics of unit-based activities in NHs. Six main unit-based activities were identified in both NHs: (1) pain management; (2) assessment and detecting change in condition; (3) treatment (e.g., wound care or treating productive cough); (4) admission to NH; (5) discharge to home; and (6) incident (e.g., falls) follow-up. These activities may involve RNs and CNAs, both performing multiple and overlapping roles. Although these activities can be considered routine, staff shortages can adversely impact work routines, resulting in hectic work shifts, missing assessments or treatments, etc. Each of these six activities were observed to be present multiple times. Assessing and detecting a change in condition is a critical function of care delivery in NHs, and performed in a unique way, compared with inpatient and outpatient care settings. There are several mechanisms that serve to detect “change of condition” in NHs. Typically, CNAs detect the first signs of a change in a resident’s condition. They then must communicate their observation to the shift RN, who is the central processor of the information. However, other individuals such as family members, nurse practitioners, physical therapists, or residents may communicate indicators of change. Once the nurse obtains this preliminary information, s/he assesses the resident. After the assessment, the nurse decides on the next step, which can be initiating a treatment, contacting a provider, or continuing to monitor. Nurses are required to document any change in a resident’s condition. Although we described the subactivities of detecting change of condition in a sequence, these subactivities can occur simultaneously, or due to interruptions, there might be long intervals among these subactivities. There was someone who did not get a good night’s sleep or someone was in more pain or somebody had a change of condition or something. So we get a report from [the clinician of the previous shift]. (CNA from NH-1) Other mechanisms for detecting and assessing change include reporting between shifts (as mentioned in the excerpts above) and opportunistic (ad hoc sporadically) assessments by staff members. Two main activities during shift change are counting controlled substances and discussing residents who had changes in medical conditions, behavioral changes or were involved in an incident. Opportunistic assessments may occur when a staff member is in a resident’s room for a nonassessment reason. For example, they may observe an event while serving food, administering medication, or be present outside the resident’s room (such as on the hallway). These assessments are facilitated by being alert to their surroundings and knowing the baseline of the resident. Organization Organizational rules (formal and informal) and policies are relevant to this level of work and present in both NHs. These rules are related to documentation, pain medication, contacting providers, incident reporting, admission criteria, and medication management. Additionally, resources and institutionalized technologies are components of organizational level work. Our analyses revealed that NH-1 is more centralized. For example, if a nurse identified a change in the status of a resident in NH-1, they were required to call the Director of Nursing before reaching out to a provider. However, nurses at NH-2 could contact outside physicians directly. In both NHs, a provider is more likely to be present in the SNF unit of the NH. Organizational rules also ensure that activities occur in temporal “zones,” which are consistent with the rhythms of daily living of the residents (Figure 1). The structure of the nursing shifts in both settings are consistent with these temporal zones. These time zones are determined by daily routines of residents and activities by staff members. In the breakfast zone, nurses pass morning medications and perform assessments. CNAs help residents with getting up and getting ready for breakfast. Staff members also help with distributing and supporting residents having breakfast. In the lunch zone, nurses pass mid-day medications and CNAs help residents with various daily living activities such as lunch and snacks. CNAs may also assist residents with recreational and social activities during the breakfast and lunch zone. In the dinner zone, nurses administer evening medications, and CNAs prepare residents for sleep. The night shift includes caring for the residents who don’t sleep at night and fulfilling nonurgent documentation and cleaning responsibilities. However, there are zone-independent activities as well, such as bathing and opportunistic assessments. The timing of bathing is determined by the preference of the residents. Each zone has its own pace and responsibilities; however written policies may not accurately reflect the work performed at each time zone. For instance, the nursing policy at one NH required resident assessment on each of the three shifts. However, observations revealed little routine assessment occurred outside of the morning period. Figure 1. Open in new tabDownload slide Temporal zones in nursing homes. Figure 1. Open in new tabDownload slide Temporal zones in nursing homes. In both NHs, staff members work in low-resource environments, probably due to efforts to decrease the cost of care. For example, cases were observed where staff members spent time looking for functioning, calibrated, equipment for vital sign measurement (because there were only a few), or nurses were required to relocate to another unit (s) to find a specific type of wound dressing. The scarcity of functional material and equipment (e.g., nonworking phones, paper-based records) decreased the productivity of staff members and increased their workload. Industry This level of work is related to interorganizational activities such as resident admission to the facility and transferring a resident from the facility to a hospital. Admission (as well as transfer/discharge) was mentioned by nurses as cumbersome and time-consuming, thereby affecting the service they provided to other residents. State laws are relevant since they dictate policies regarding care delivery. Also, accreditation requirements can dictate policy and procedures. Interviews revealed that assessment frequency depends on what state laws are required for admission or discharge assessment for reimbursement and how payment incentives are designed. State policies also impact priorities. For example, scheduled medications should be administered within 1 hr before or after their scheduled times. Both NHs work with outside pharmacies, imaging, and lab facilities. Seamless communication between NHs and these entities are essential for a smooth care delivery to residents. Other Emerging Workflow-Related Themes Several other important themes that encompass multiple levels emerged from the data. These themes did not fit within a single level described above. The first is the centrality of ongoing relationships of staff members with the residents in NHs. This relationship was important for establishing the resident’s baseline status and identifying a change in the status of the resident. Ongoing and established relationships save time for clinicians because they bolster familiarity with behavior and activity patterns of residents and their needs for care. For example, knowing a resident may become incontinent upon awakening, the nurse can proactively anticipate their need for the bathroom, as opposed to an emergent or unexpected situation where incontinence requires unplanned time spent bathing, changing clothing, and linens. Also, documentation of routine assessments is facilitated by established baseline information about residents, allowing clinicians to make assumptions and document efficiently. A second emergent theme is the resident-centeredness of care. Needs and preferences of residents are the primary drivers of care in NHs. Both health needs and resident preferences shape the timing and other characteristics of care. A resident’s preferences are collected on admission and throughout their stay. For example, some patients like to “sleep-in” later than other residents, thereby requiring their breakfast be held/kept warm until they arise. A third theme is the absence of memory aids. Staff members rely heavily on their memory. For example, reporting during end of shift hand-offs may be based on recall rather than notes, due to limited documentation. Lastly, staff member’s preferences also affect work activities. For example, medication drawers may be moved in closer proximity to residents, thereby saving the time required returning to the nurses’ station for each person’s medications. Our analysis also revealed the range of time to provide care related to the complexity of the residents. Not all patients take the same amount of time. As reported by a CNA, “half of the residents are 80% of workload.” Contributing factors to patient complexity include adherence, cooperative behavior, level of cognition, level of independence, and medical needs. Complexity can be a permanent characteristic of a resident because of a chronic (physical or mental) condition. Complexity can also be temporary in the case of an acute condition or incident. During the study, we noted a high level of variability in routines of clinicians and individualized care of residents. There was variability at the task level between staff members, but limited variation within an individual task for a given staff member. For example, nurses have different assessment routines. However, a nurse assesses all of his/her patients in a similar way, after considering the special needs of the patient. The pace of the work may differ on a daily basis; shortages in staffing (which happen occasionally), incidents, and unusual numbers of admissions and discharges affect pace as well. Various elements potentially impacting resident safety were observed. These issues spanned several key aspects of NH workflow, such as environment, activities, rules/policies, and individuals. Some safety issues were related to shortcuts or workarounds in medication administration. Some examples of safety issues included; not watching a resident taking their medication, borrowing a medication from one resident for another, excess reliance on memory while documenting and reporting, dispensing medication for multiple residents at one time, not taking a finger stick glucose test before a meal. Differences were noted in the physical layout of the two NHs, and the physical layout, in turn, affected workflow (Table 2). NH-1 was a one-story building with long hallways in a spoke-and-hub configuration. Each unit was located on a different hallway. NH-2 on the other hand was a tall building with each of the five floors constituting a unit. At NH-1, there was one central dining area for residents (plus one exclusively for dementia residents). At NH-2, there is one dining area on each floor. The NH-2 dining areas are next to nursing stations which allow nurses to observe residents while dining. In contrast, at NH-1, a nurse (RN or LPN) must be temporarily reassigned from his/her unit, to supervise residents while dining. Table 2. Summary of Findings in Two Nursing Homes (NH-1 and NH-2) Description NH-1 NH-2 Majority of nurse time used for medication management ✓ ✓ Majority of certified nurse assistant time spent assisting with activities of daily living ✓ ✓ Existence of the latent activities ✓ ✓ Existence of the following unit level activities: (a) pain management; (b) assessment and detecting change in condition; (c) treatment; (d) admission; (e) discharge; and (f) incident follow-up ✓ ✓ Centralized More Less Existence of temporal zones ✓ ✓ Prolonged relationships between staff and residents ✓ ✓ Resident-centered care ✓ ✓ Absence of memory aids ✓ ✓ Staff member’s preferences affect work activities ✓ ✓ Physical layout One story building with long hallways in a spoke and hub configuration Multiple floors Medication administration records Electronic Paper Proximity of management Nearby Distant Interdependence between shifts Lower Higher High variability in complexity of patients ✓ ✓ High variability at the task level between staff members and limited variation within individual’s routines ✓ ✓ Description NH-1 NH-2 Majority of nurse time used for medication management ✓ ✓ Majority of certified nurse assistant time spent assisting with activities of daily living ✓ ✓ Existence of the latent activities ✓ ✓ Existence of the following unit level activities: (a) pain management; (b) assessment and detecting change in condition; (c) treatment; (d) admission; (e) discharge; and (f) incident follow-up ✓ ✓ Centralized More Less Existence of temporal zones ✓ ✓ Prolonged relationships between staff and residents ✓ ✓ Resident-centered care ✓ ✓ Absence of memory aids ✓ ✓ Staff member’s preferences affect work activities ✓ ✓ Physical layout One story building with long hallways in a spoke and hub configuration Multiple floors Medication administration records Electronic Paper Proximity of management Nearby Distant Interdependence between shifts Lower Higher High variability in complexity of patients ✓ ✓ High variability at the task level between staff members and limited variation within individual’s routines ✓ ✓ Open in new tab Table 2. Summary of Findings in Two Nursing Homes (NH-1 and NH-2) Description NH-1 NH-2 Majority of nurse time used for medication management ✓ ✓ Majority of certified nurse assistant time spent assisting with activities of daily living ✓ ✓ Existence of the latent activities ✓ ✓ Existence of the following unit level activities: (a) pain management; (b) assessment and detecting change in condition; (c) treatment; (d) admission; (e) discharge; and (f) incident follow-up ✓ ✓ Centralized More Less Existence of temporal zones ✓ ✓ Prolonged relationships between staff and residents ✓ ✓ Resident-centered care ✓ ✓ Absence of memory aids ✓ ✓ Staff member’s preferences affect work activities ✓ ✓ Physical layout One story building with long hallways in a spoke and hub configuration Multiple floors Medication administration records Electronic Paper Proximity of management Nearby Distant Interdependence between shifts Lower Higher High variability in complexity of patients ✓ ✓ High variability at the task level between staff members and limited variation within individual’s routines ✓ ✓ Description NH-1 NH-2 Majority of nurse time used for medication management ✓ ✓ Majority of certified nurse assistant time spent assisting with activities of daily living ✓ ✓ Existence of the latent activities ✓ ✓ Existence of the following unit level activities: (a) pain management; (b) assessment and detecting change in condition; (c) treatment; (d) admission; (e) discharge; and (f) incident follow-up ✓ ✓ Centralized More Less Existence of temporal zones ✓ ✓ Prolonged relationships between staff and residents ✓ ✓ Resident-centered care ✓ ✓ Absence of memory aids ✓ ✓ Staff member’s preferences affect work activities ✓ ✓ Physical layout One story building with long hallways in a spoke and hub configuration Multiple floors Medication administration records Electronic Paper Proximity of management Nearby Distant Interdependence between shifts Lower Higher High variability in complexity of patients ✓ ✓ High variability at the task level between staff members and limited variation within individual’s routines ✓ ✓ Open in new tab Medication administration records are also in different formats. At NH-1, records were kept in electronic format, while NH-2 records were kept on paper. The physical location of management differed. At NH-1, unit managers were closer to units and were observed to spend more time on the units. In NH-2 unit, managers worked more remotely, spending most of their time on the first floor, where the administrators’ offices were located. Interdependence between shifts was limited in NH-1 whereas in NH-2, the workload of a nurse depended on what a staff member accomplished on the previous shift. In NH-1, job descriptions for each shift were more clear and delineated, while in NH-2, the incoming nurse assumed responsibility for what the outgoing nurse was doing. Interaction Among Levels Multilevel models highlight the interaction among levels through two mechanisms. We identified both bottom-up emerging processes, and top-down mechanisms that express the influence of higher level contextual factors on lower levels. Flexibility is an example for the emerging process. Despite the highly regulated setting, there was a great deal of flexibility inherent in how staff members performed their work. To a certain degree, variability was introduced by meeting residents’ wishes and by accommodating staff needs for efficiency. In addition, many work processes were under the control of the staff. For instance, documentation was often performed at the end of the shift rather than as work occurred. Safety orientation was also an emerging process. In interviews and observations, resident safety was a prime concern and a driver of work across several domains. However, the pace of the work and resident preferences could supersede safety precautions, for example, not watching the resident take their medication. Both observations and interviews revealed how contextual factors such as organizational rules, state-level policies and physical and technological infrastructure shape individual and work group/unit level activities. All the activities listed above are affected by context in a way that context can speed up the activities or slow them down for the safety, quality and cost-related goals. Discussion and Implications Examining workflow helps identify where in the stream of interactions, deficits may occur at any level of work in NHs. For example, when trying to find the root cause of an issue, workflow patterns would indicate the problem may stem from the individual, group, organizational, or industry levels. Understanding the relationship between these levels can occur more easily with workflow analysis. For example, in a fictional case, organizational policy may dictate that a nurse respond to a call light within 10 min, but on the way to attend to the patient, a patient is observed to fall in the hallway. The nurse must prioritize the urgent nature of the fall and “trust” the call light was for a less urgent nature. Examining workflows can reveal these interactions and inform interventions to improve care delivery in NHs. Work in NHs is impacted by the long-term relationship between staff members and residents. Knowing baseline information of a resident and their preferences was the driver of a staff member’s decision making. For NHs with a high staff turnover, this bonding could impact residents emotionally when staff left (Castle, Engberg, & Men, 2007; Lerner, Johantgen, Trinkoff, Storr, & Han, 2014; Roberts & Bowers, 2015; Secrest, Iorio, & Martz, 2005; Thomas, Mor, Tyler, & Hyer, 2013). In addition, relevant to clinical decision making, turnover could impact the time requirements for the staff member to assess the resident and their preferences. This qualitative study revealed various aspects of workflow in NHs. Future studies can be designed to quantitatively characterize a specific aspect of work, highlighted in this study. For example, tools such as WOMBAT (Ballermann, Shaw, Mayes, Gibney, & Westbrook, 2011) can be used to quantitatively characterize duration and sequence of activities by nurses and CNAs to better understand pain management and examine different factors that affect pain outcomes. The value of workflow studies has been repeatedly shown to improve care and inform informatics interventions. We highlighted an important uniqueness of workflow, being sensitive to daily routines of residents. Methodological challenges for future quantitative studies can be achieved by using multiple data sources such as sensors and radio frequency identification technologies (Kim et al., 2010; Ku, Wang, Su, Liu, & Hwang, 2011). This study provided a rich understanding of the characteristics of NH work that can inform both the design and implementation of CDSSs. A misfit between these systems and workflow at any level in NHs can lead to disruptions that cause suboptimal staffing and patient outcomes. NHs have a variety of policies regarding assessment, communication, testing, and prescribing. Thus, CDSS should be consistent with organizational clinical guidelines. This may require generalizable recommendations or recommendations that are customizable. For example, each facility may have a different hydration protocol. If CDSS makes a recommendation to consider hydration, it will need to be consistent with that protocol. Our work in two NHs revealed a marked difference in organizational structure (e.g., physical structure, policies, hierarchy, etc.). The role of organizational structure in NHs on outcomes merits further exploration. Implications for CDSS Design The design of CDSS in NHs crosses all levels as reported in the results. At the individual level, computer literacy of NH staff is highly variable. Thus, in this setting, the application must be user-friendly with an intuitive interface and clear, concise instructions. Further, there are most likely limited resources to support the user in the NH setting, thus the application must be robust, including the downloading function. The variable degree of computer literacy of the NH staff will impact adoption of CDSS with implications for training of staff. At the work group/unit level, resident assessment was a team activity that occurred both at scheduled intervals and as needed. The use of a mobile application such as a CDSS may be helpful, given the temporal and ad hoc nature of the assessment work. The teamwork aspect to assessment means that CNAs, while not the primary target of the application, may be asked to interact with it. Field testing may be helpful in determining how a CNA might work with the CDSS. From organizational level perspective, in a NH where nursing units are proximate to one another, CDSS might be used as a collaborative tool. In a NH where nursing units are isolated, the CDSS may provide needed feedback. Nurse manager oversight will also impact how the CDSS is used. If the nurse manager is involved in all clinical decision making, then s/he can require application use with all changes in condition, and the CDSS serve as collaborative tool. If the nurses are not involving the nurse manager in clinical decision making, the nurse will be using the application independently. Finally, the use of CDSS may differ between LTC units and SNF units. The LTC unit staff are more attuned to subtle/nonspecific changes in residents because of extended interactions. CDSS may be helpful in avoiding premature reactions, when noting these subtle changes. In contrast, the SNF staff have fewer opportunities to develop an understanding of a resident’s current baseline state that may be changeable. CDSS may be helpful in reminding staff of what they need to assess before proceeding with a resident’s assessment. NHs can also benefit from technologies documentation at the point of care, using mobile devices that synchronize to the NH electronic health records. Such technologies will allow clinicians to rely less on their memory, and result in more timely, accurate, and complete documentation. Our results highlighted that staff members work in low-resource environments. This finding confirms previous studies (Harrington, Olney, Carrillo, & Kang, 2012; Ouslander et al., 2010; Seymour, Kumar, & Froggatt, 2011; Zhang, Punnett, Gore, & CPH-NEW Research Team, 2014) and should be taken into account when planning for future design requirements in NHs. Any CDSS in NH should be compatible with industry level variables (e.g., interorganizational activities, state-level regulations, and scientific developments). State level regulations and scientific developments directly affect the algorithms. Given the frequency of interorganizational activities with a variety of institutions, portability of the data, and information in the CDSS is important. Therefore, CDSS can import data from other settings for more data-oriented approaches and information produced by CDSS can be transferred to other settings that can use such information for their decision making. Some of the activities we described here were previously described (McCloskey et al., 2015). An important additional contribution of this study would be in providing a holistic perspective that highlights the multilevel aspect of workflow in NHs in a temporal context. Moreover, we examined workflow for the purpose of health IT implementation. As a result, we were able to make recommendations for workflow informed health IT in NHs. We argue that the value of using multilevel model in developing CDSS in NHs is two-fold. First, the multilevel model provided an easy to follow (in data collection and analysis) categorization of work elements. This categorization helped inclusive and holistic consideration of all elements. Second, multilevel model highlighted emerging processes and contextual factors. Emergent processes in NHs are mostly due to NHs being daily living settings of residents and the centrality of relationship between residents and staff members. The design of the CDSS should take these emergent processes such as flexibility into account. CDSS should not impose constraints that will limit the staff’s flexibility. Similarly, contextual factors (i.e., impact of higher level variables on lower level variables) should be considered because these contextual factors (e.g., working in low resource environments) heavily shape the individual and team activities. CDSS must, therefore, be congruent with these individual and team activities. Limitations This study has limitations related to study site selection, participant enrollment, and methods utilized. First, the convenience sample of two NHs may limit transferability of results even though staff within NHs have similar work requirements based on regulatory requirements and many common resident medical conditions. Factors that may introduce variability are differences in the layout of the work environment, site-level policy, resident population, geography, staffing patterns, and for-profit/nonprofit status. However, a strength of this study that allowed us to identify commonalities between the two sites is that the two NHs served similar resident populations in the same metropolitan area with similar staffing patterns, under the same corporate ownership structure, while offering variations in layout and site-level policy. Second, we enrolled a convenience sample of volunteer participants at both study sites and the views of those who opted not to participate may diverge from those who volunteered. Third, the qualitative methods utilized in this study required actions to mitigate bias and establish reliability of results. Observation notes are subject to observers’ backgrounds and perspectives. As described in the Research Design and Methods section, all observations were conducted and documented independently and simultaneously by two observers. Individual observers independently transcribed their own notes and sets of notes were reviewed for differences by the first author. Interviews also carry potential for bias and similar steps were taken to establish reliability of results. All interviews were conducted by two interviewers, all transcripts were reviewed by the research team, and all transcripts were analyzed by multiple coders. Funding This work was supported by the Centers for Disease Control and Prevention (Contract 200-2016-92277). Acknowledgments We thank Ms Suzanne Lareau for editorial support. Conflict of Interest None reported. References Alexander , G. L. , Pasupathy , K. S. , Steege , L. M. , Strecker , E. B. , & Carley , K. M . ( 2014 ). Multi-disciplinary communication networks for skin risk assessment in nursing homes with high IT sophistication . International Journal of Medical Informatics , 83 , 581 – 591 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Ballermann , M. A. , Shaw , N. T. , Mayes , D. C. , Gibney , R. T. , & Westbrook , J. I . ( 2011 ). Validation of the Work Observation Method By Activity Timing (WOMBAT) method of conducting time-motion observations in critical care settings: An observational study . BMC Medical Informatics and Decision Making , 11 , 32 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Bowen , G. A . ( 2008 ). Naturalistic inquiry and the saturation concept: A research note . Qualitative Research , 8 , 137 – 152 . doi: Google Scholar Crossref Search ADS WorldCat Crossref Boyatzis , R. E . ( 1998 ). Transforming qualitative information: Thematic analysis and code development . Thousand Oaks, CA : Sage Publications . Google Preview WorldCat COPAC Carayon , P. , Schoofs Hundt , A. , Karsh , B. T. , Gurses , A. P. , Alvarado , C. J. , Smith , M. , & Flatley Brennan , P . ( 2006 ). Work system design for patient safety: The SEIPS model . Quality & Safety in Health Care , 15 ( Suppl. 1 ), i50 – i58 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Castle , N. G. , Engberg , J. , & Men , A . ( 2007 ). Nursing home staff turnover: Impact on nursing home compare quality measures . The Gerontologist , 47 , 650 – 661 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Costa , P. L. , Graça , A. M. , Marques-Quinteiro , P. , Santos , C. M. , Caetano , A. , & Passos , A. M . ( 2013 ). Multilevel research in the field of organizational behavior: An empirical look at 10 years of theory and research . SAGE Open , 3 . doi: WorldCat Crossref Gale , N. K. , Heath , G. , Cameron , E. , Rashid , S. , & Redwood , S . ( 2013 ). Using the framework method for the analysis of qualitative data in multi-disciplinary health research . BMC Medical Research Methodology , 13 , 117 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Handler , S. M. , Boyce , R. D. , Ligons , F. M. , Perera , S. , Nace , D. A. , & Hochheiser , H . ( 2013 ). Use and perceived benefits of mobile devices by physicians in preventing adverse drug events in the nursing home . Journal of the American Medical Directors Association , 14 , 906 – 910 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Harrington , C. , Olney , B. , Carrillo , H. , & Kang , T . ( 2012 ). Nurse staffing and deficiencies in the largest for-profit nursing home chains and chains owned by private equity companies . Health Services Research , 47 ( 1 Pt 1 ), 106 – 128 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Holden , R. J. , & Karsh , B.-T . ( 2009 ). A theoretical model of health information technology usage behaviour with implications for patient safety . Behaviour & Information Technology , 28 , 21 – 38 . doi: Google Scholar Crossref Search ADS WorldCat Crossref Jones , W. , Drake , C. , Mack , D. , Reeder , B. , Trautner , B. , & Wald , H . ( 2017 ). Developing mobile clinical decision support for nursing home staff assessment of urinary tract infection using goal-directed design . Applied Clinical Informatics , 8 , 632 – 650 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Kim , J. Y. , Lee , H. J. , Byeon , N. S. , Kim , H. C. , Ha , K. S. , & Chung , C. Y . ( 2010 ). Development and impact of radio-frequency identification-based workflow management in health promotion center: Using interrupted time-series analysis . IEEE Transactions on Information Technology in Biomedicine , 14 , 935 – 940 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Kozlowski , S. W. J. , & Klein , K. J . ( 2000 ). A multi-level approach to theory and research in organizations: Contextual, temporal, and emergent processes . In K. J. Klein & S. W. J. Kozlowski (Eds.), Multilevel theory, research, and methods in organizations . San Francisco, CA : Jossey-Bass . Google Preview WorldCat COPAC Ku , H. L. , Wang , P. C. , Su , M. C. , Liu , C. C. , & Hwang , W. Y . ( 2011 ). Application of radio-frequency identification in perioperative care . AORN Journal , 94 , 158 – 169 ; quiz 170. doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Lerner , N. B. , Johantgen , M. , Trinkoff , A. M. , Storr , C. L. , & Han , K . ( 2014 ). Are nursing home survey deficiencies higher in facilities with greater staff turnover . Journal of the American Medical Directors Association , 15 , 102 – 107 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Lim , C. J. , Kong , D. C. , & Stuart , R. L . ( 2014 ). Reducing inappropriate antibiotic prescribing in the residential care setting: Current perspectives . Clinical Interventions in Aging , 9 , 165 – 177 . doi: Google Scholar PubMed WorldCat Crossref Lim , C. J. , Kwong , M. W. , Stuart , R. L. , Buising , K. L. , Friedman , N. D. , Bennett , N. J. ,… Kong , D. C . ( 2014 ). Antibiotic prescribing practice in residential aged care facilities—Health care providers’ perspectives . The Medical Journal of Australia , 201 , 98 – 102 . doi: Google Scholar PubMed WorldCat Crossref McCloskey , R. , Donovan , C. , Stewart , C. , & Donovan , A . ( 2015 ). How registered nurses, licensed practical nurses and resident aides spend time in nursing homes: An observational study . International Journal of Nursing Studies , 52 , 1475 – 1483 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref McMaughan , D. K. , Nwaiwu , O. , Zhao , H. , Frentzel , E. , Mehr , D. , Imanpour , S. ,… Phillips , C. D . ( 2016 ). Impact of a decision-making aid for suspected urinary tract infections on antibiotic overuse in nursing homes . BMC Geriatrics , 16 , 81 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Morrill , H. J. , Caffrey , A. R. , Jump , R. L. , Dosa , D. , & LaPlante , K. L . ( 2016 ). Antimicrobial stewardship in long-term care facilities: A call to action . Journal of the American Medical Directors Association , 17 , 183.e1 – 183.e16 . doi: Google Scholar Crossref Search ADS WorldCat Crossref Ouslander , J. G. , Lamb , G. , Perloe , M. , Givens , J. H. , Kluge , L. , Rutland , T. ,… Saliba , D . ( 2010 ). Potentially avoidable hospitalizations of nursing home residents: Frequency, causes, and costs: [see editorial comments by Drs. Jean F. Wyman and William R. Hazzard, pp 760–761] . Journal of American Geriatrics Society , 58 , 627 – 635 . doi: Google Scholar Crossref Search ADS WorldCat Crossref Ozkaynak , M. , & Brennan , P. F . ( 2012 ). Characterizing patient care in hospital emergency departments . Health Systems , 1 , 104 – 117 . doi: Google Scholar Crossref Search ADS WorldCat Crossref Ozkaynak , M. , & Brennan , P. F . ( 2013 ). Revisiting sociotechnical systems in a case of unreported use of health information exchange system in three hospital emergency departments . Journal of Evaluation in Clinical Practice , 19 , 370 – 373 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Ozkaynak , M. , Johnson , S. , Shimada , S. , Petrakis , B. A. , Tulu , B. , Archambeault , C. ,… Woods , S . ( 2014 ). Examining the multi-level fit between work and technology in a secure messaging implementation . AMIA. Annual Symposium Proceedings. AMIA Symposium , 2014 , 954 – 962 . Google Scholar PubMed WorldCat Ozkaynak , M. , Unertl , K. M. , Johnson , S. A. , Brixey , J. J. , & Haque , S. N . ( 2016 ). Clinical workflow analysis, process redesign, and quality improvement . In J. Finnel & B. Dixon (Eds.), Clinical informatics study guide (pp. 135 – 161 ). New York : Springer International Publishing . doi:10.1007/978-3-319-22753-5_7 Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Ozkaynak , M. , Valdez , R. S. , Holden , R. J. , & Weiss , J . ( 2018 ). Infinicare framework for an integrated understanding of health-related activities in clinical and daily-living contexts . Health Systems , 7 , 66 – 78 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Qian , S. , & Yu , P . ( 2013 ). Model medication management process in Australian nursing homes using business process modeling . Studies in Health Technology and Informatics , 192 , 1061 . doi:10.3233/978-1-61499-289-9-1061 Google Scholar PubMed WorldCat Qian , S. , & Yu , P . ( 2014 ). Fitting clinical workflow: The case for wound care in a residential aged care home . Studies in Health Technology and Informatics , 204 , 130 – 136 . doi:10.3233/978-1-61499-427-5-130 Google Scholar PubMed WorldCat Reeder , B. , Ozkaynak , M. , Makic , M. B. , & Sousa , K. H . ( 2017 ). Examining nurses’ use of electronic health records to document symptoms in acute care settings . Paper presented at the American Medical Informatics Association Annual Symposium , Washington, DC . Roberts , T. , & Bowers , B . ( 2015 ). How nursing home residents develop relationships with peers and staff: A grounded theory study . International Journal of Nursing Studies , 52 , 57 – 67 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Secrest , J. , Iorio , D. H. , & Martz , W . ( 2005 ). The meaning of work for nursing assistants who stay in long-term care . Journal of Clinical Nursing , 14 , 90 – 97 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Seymour , J. E. , Kumar , A. , & Froggatt , K . ( 2011 ). Do nursing homes for older people have the support they need to provide end-of-life care? A mixed methods enquiry in England . Palliative Medicine , 25 , 125 – 138 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Siegel , E. O. , Bakerjian , D. , & Zysberg , L . ( 2017 ). Quality improvement in nursing homes: Alignment among leaders across the organizational chart . Gerontologist , 58, e281–e290. doi: WorldCat Crossref Sittig , D. F. , Wright , A. , Osheroff , J. A. , Middleton , B. , Teich , J. M. , Ash , J. S. ,… Bates , D. W . ( 2008 ). Grand challenges in clinical decision support . Journal of Biomedical Informatics , 41 , 387 – 392 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Stokoe , A. , Hullick , C. , Higgins , I. , Hewitt , J. , Armitage , D. , & O’Dea , I . ( 2016 ). Caring for acutely unwell older residents in residential aged-care facilities: Perspectives of staff and general practitioners . Australasian Journal on Ageing , 35 , 127 – 132 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Tariq , A. , Georgiou , A. , Raban , M. , Baysari , M. T. , & Westbrook , J . ( 2016 ). Underlying risk factors for prescribing errors in long-term aged care: A qualitative study . BMJ Quality & Safety , 25 , 704 – 715 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Thomas , K. S. , Mor , V. , Tyler , D. A. , & Hyer , K . ( 2013 ). The relationships among licensed nurse turnover, retention, and rehospitalization of nursing home residents . The Gerontologist , 53 , 211 – 221 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Thorpe-Jamison , P. T. , Culley , C. M. , Perera , S. , & Handler , S. M . ( 2013 ). Evaluating the impact of computer-generated rounding reports on physician workflow in the nursing home: A feasibility time-motion study . Journal of the American Medical Directors Association , 14 , 358 – 362 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref Walker , S. , McGeer , A. , Simor , A. E. , Armstrong-Evans , M. , & Loeb , M . ( 2000 ). Why are antibiotics prescribed for asymptomatic bacteriuria in institutionalized elderly people? A qualitative study of physicians’ and nurses’ perceptions . Canadian Medical Association Journal = Journal De l’Association medicale canadienne , 163 , 273 – 277 . Google Scholar PubMed WorldCat Zhang , Y. , Punnett , L. , & Gore , R .; CPH-NEW Research Team . ( 2014 ). Relationships among employees’ working conditions, mental health, and intention to leave in nursing homes . Journal of Applied Gerontology , 33 , 6 – 23 . doi: Google Scholar Crossref Search ADS PubMed WorldCat Crossref © The Author(s) 2018. Published by Oxford University Press on behalf of The Gerontological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Characterizing Workflow to Inform Clinical Decision Support Systems in Nursing Homes
JO - The Gerontologist
DO - 10.1093/geront/gny100
DA - 2019-11-16
UR - https://www.deepdyve.com/lp/oxford-university-press/characterizing-workflow-to-inform-clinical-decision-support-systems-in-2eAhf5DRX3
SP - 1024
VL - 59
IS - 6
DP - DeepDyve
ER -