Minding the gap and overlap: a literature review of fragmentation of primary care for chronic dialysis patients

Background Care coordination is a challenge for patients with kidney disease, who often see multiple providers to manage their associated complex chronic conditions. Much of the focus has been on primary care physician (PCP) and nephrologist collaboration in the early stages of chronic kidney disease, but less is known about the co-management of the patients in the end-stage of renal disease. We conducted a systematic review and synthesis of empirical studies on primary care services for dialysis patients. Methods Systematic literature search of MEDLINE/PubMED, CINAHL, and EmBase databases for studies, published until August 2015. Inclusion criteria included publications in English, empirical studies involving human subjects (e.g., patients, physicians), conducted in US and Canadian study settings that evaluated primary care services in the dialysis patient population. Results Fourteen articles examined three major themes of primary care services for dialysis patients: perceived roles of providers, estimated time in providing primary care, and the extent of dialysis patients’ use of primary care services. There was general agreement among providers that PCPs should be involved but time, appropriate roles, and miscommunication are potential barriers to good primary care for dialysis patients. Although many dialysis patients report having a PCP, the majority rely on primary care from their nephrologists. Studies using administrative data found lower rates of preventive care services than found in studies relying on provider or patient self-report. Discussion The extant literature revealed gaps and opportunities to optimize primary care services for dialysis patients, foreshadowing the challenges and promise of Accountable Care / End-Stage Seamless Care Organizations and care coordination programs currently underway in the United States to improve clinical and logistical complexities of care for this commonly overlooked population. Studies linking the relationship between providers and patients’ receipt of primary care to outcomes will serve as important comparisons to the nascent care models for ESRD patients, whose value is yet to be determined.


Background
Overall and disease-specific management and care coordination is a challenge for patients with complex and multiple chronic conditions (MCC), who typically see multiple providers to manage one or more co-morbid conditions. This problem is amplified for patients with kidney disease, whose condition is associated and co-morbid with highly prevalent conditions such as hypertension, diabetes, and heart disease. Much of the literature and intervention efforts have focused on primary care and nephrology physician collaborative care in the early stages of chronic kidney disease (CKD), with an emphasis on primary care physician (PCP) intervention to prevent, detect, treat, and slow the progression of kidney disease. Studies have generally found lack of clarity on the roles and responsibilities of clinical management among providers [1], poor communication and coordination between primary care physicians and nephrologists [2] and suboptimal general and CKD-related care in patients [3,4].
However, less is known about primary care physician (PCP) and nephrologist co-management of patients in the end-stage of renal disease. Specifically, most patients with advanced kidney failure undergo chronic dialysis treatments and are clinically managed by clinical staff at dialysis facilities on a frequent basis (e.g., thrice weekly for in-center hemodialysis or monthly for home-based dialysis). Due in part to the rigidity of the dialysis treatment schedule, management of co-morbid illness and their complications by providers outside of the dialysis unit is challenging. Patients commonly receive supervision of non-renal health needs during their dialysis treatment visits due to convenience and familiarity. As a result, primary care management may default to renal providers. Compared to early-stage CKD care, this observed focusof nephrologist involvement in treating primary care needs of patients with end-stage renal disease (ESRD)raises important questions about the role of the PCP, traditionally considered the patient's medical home for continuous comprehensive care and the "quarterback" responsible for assessing, balancing, and coordinating the care of patients' multiple competing conditions [5,6]. Who is responsible for and actively managing the primary care needs of these dialysis patients? Discrepant expectations and subsequent provision of primary care may exacerbate care fragmentation in this highly complex and vulnerable patient population, increasing the potential for unnecessary duplication of care or adverse outcomes. To date, PCP and nephrology roles in the actual provision of primary care services for chronic dialysis patients is not well-defined. We conducted a systematic review of primary care service provision for dialysis patients, in order to assess 1) patient and provider perceptions of PCPs and nephrologists roles; 2) the extent to which PCPs and nephrologists deliver primary care services to chronic dialysis patients; 3) reported barriers to patients' receipt or physicians' delivery of primary care services; and 4) the measures used to assess provider provision and primary care outcomes in dialysis patients.
Understanding the provision of primary care among nephrologists and PCPs, fragmentation of care, and outcomes in dialysis patients is an emerging and important concern [7]. The anticipated growth and surveillance of patients with CKD [8] and the diminishing supply of practicing nephrologists [9,10] suggests that nephrologists may be unable to continue or increase their provision of primary care services to dialysis patients, and it remains to be seen if nephrologists' provision of primary care services is adequate or appropriate. In addition, recent efforts to organize chronic disease management, such as multidisciplinary care teams and the patient-centered medical home, commonly exclude patients with ESRD who may also benefit from models designed to mitigate care fragmentation and improve coordination. Elucidating the types of services and providers managing dialysis patients' primary care needs and their effectiveness will illuminate the gaps in our understanding of primary care and nephrologist collaboration to guide further research and intervention to improve care for this overlooked, high healthcare utilizing, co-morbid dialysis patient population.

Methods
This literature review is guided by conceptual frameworks for care provided to individuals with MCC [11] and interprofessional and co-managed care [12]. Together, these frameworks illustrate the complexity of care for patients receiving maintenance dialysis and is a valuable lens to frame our understanding of how their specialized and general healthcare needs are met. Briefly, the National Quality Forum's model for MCC care recognizes the various ways in which patient preferences, care settings and providers, and types of healthcare services interact to impact health outcomes (e.g., care coordination, prevention of disease, cost). The interaction between these domains of care is dynamic, as patients' healthcare needs evolve over time [11]. Our review of the empirical literature focuses on the domains regarding dialysis patients' providers, their use, and measures used to assess their primary care. We also incorporate perspectives from Retchin's collaborative model of interprofessional and co-managed care, which posits that interaction and coordination between MCC provider types is influenced by temporality (e.g., concurrent vs. sequential care), urgency, and delineation of authority in patient care [12]. To this end, our review also considers the evidence on perceptions regarding provider roles and skills in primary care delivery.

Definitions and eligibility criteria for relevant literature
In this review, variables of interest pertain to the provision of primary care services among patients undergoing chronic dialysis treatment. For the purpose of this structured literature review, we conducted a broad search of the two main terms of 'dialysis' and' primary care' , due to the anticipated paucity of published studies in this area. We defined dialysis patients as those with ESRD and undergoing chronic dialysis treatment (inclusive of all modalities, in-center or home-based dialysis). Primary care is more loosely defined. We adopted the Institute of Medicine's definition of primary care's scope of services [13], which includes general health maintenance; prevention (e.g., immunizations) and early detection (e.g., screening for cancer, depression); counseling of patients (e.g., diet, nutrition, tobacco cessation); risk assessment; management of acute care and chronic care (e.g., diabetes, hypertension, heart disease), care coordination and referrals. As follows, providers of primary care services include physicians represented in internal medicine, family medicine, geriatric medicine, general medicine, and nephrology, as well as non-physician providers such as physician assistant, nurse practitioner, and registered nurse. Last, we considered several settings where provision of primary care services take place, including physician offices and clinics of internal, family, and geriatric medicine, nephrology clinics, and dialysis centers.
From the published literature, we applied several eligibility criteria for inclusion in critical synthesis of the literature. Publications in English, empirical studies involving human subjects (e.g., patients, physicians), conducted in US and Canadian study settings and published until August 2015 were included for review. Editorials, letters, and literature review articles, and duplicate citations and publications that did not evaluate primary care services in the dialysis patient population were excluded.

Literature search and selection criteria
Literature included for synthesis were identified using 3 search strategies. The primary approach involved searching MEDLINE/PubMED, CINAHL, and Embase electronic databases for articles related to primary care services for dialysis patients. This search strategy employed numerous Medical Subject Headings (MeSH) and keywords related to primary care and dialysis patients (Fig. 1). Eligibility criteria were first applied to titles and abstracts and then full text review of potential references. To identify qualifying articles and ensure saturation of search results, the reference lists of eligible articles and literature reviews (that were not included in the literature synthesis) were examined. We applied similar eligibility criteria to these references through this backward search method. Inclusion and exclusion criteria were applied by four reviewers and dual coding of all citations found 98.7% agreement in title and abstract reviews and 90.2% in full text reviews. Disagreements were resolved by discussion and consensus among all four reviewers.

Data collection and analysis
Key information in each article was extracted by 2 reviewers (VW, RCG). Data included overall study design, study setting and participants, and key measures. Key variables of interest included the types of providers, primary care services, and outcomes that were assessed or reported and authors' results.

Literature search
Keyword searches in electronic databases resulted in 13,841 citation hits, of which 10 met inclusion criteria. The large number of excluded citations and papers were largely due to the broad search terminology regarding primary care and prevention. Another 4 relevant articles were identified through backward search methods, resulting in a total of 14 articles regarding primary care services for dialysis patients (Fig. 2).

Study designs of reviewed literature
Characteristics of the studies addressing primary care services for dialysis patients are described in Table 1. The overwhelming majority of articles were conducted in the US (n = 13, 93%) and one study was conducted in Canada.
No studies assessed relationships between primary care services and clinical outcomes. Moreover, no clinical trials were identified addressing our key questions.
Across studies included in this review, the dialysis patient was the most common unit of observation (n = 11 studies), followed by nephrologists (n = 4), and PCPs (n = 3). Interestingly, however, there was less variation among the types of outcomes assessed from the patient perspective, compared to the wider array of provideroriented outcomes examined. Key measures about or reported by patients included whether patients had or used a PCP [14,17,19,20,23], the extent to which patients relied on primary care from nephrologists or non-nephrologists [19,23] and the receipt of specific types of primary care services [17][18][19][20][21][22][24][25][26][27]. In contrast, measures referencing or reported by providers assessed the estimated time devoted to and capacity for providing primary care to dialysis patients [14][15][16][17]20], self-reported frequency of delivering specific types of primary care services [14-17, 19, 20, 25], nephrologists' confidence in their ability to provide primary care [14,17] and nephrologist and PCP opinions of who should provide primary care services [15,17].

Perceptions of provider roles and skills in primary care delivery
Four cross-sectional surveys examined patients and providers' perceptions of PCP and nephrologists' roles and capacity in the provision of primary care for dialysis patients (Table 2). Across these studies, most nephrologists' were confident in their abilities serving as the primary care provider for dialysis patients [14,17] or managing and treating symptoms, despite their perceived limited  training and comfort in managing aspects of symptom management such as pain and depression [15]. In comparing nephrologists' and family physicians' perspectives, there was general agreement that primary care should not be exclusively provided by nephrologists [15,17]. Yet, both nephrologists and family practitioners reported similar lack of confidence in family practitioners' knowledge, training, and available time to provide quality primary care to dialysis patients [17]. Further, there was incongruence in provider perceptions related to trust, where nephrologists encouraged their patients to maintain relationships with their PCPs, but family practitioners were uncertain about nephrologists' encouragement of PCP involvement [17].
From the patient perspective, overall satisfaction with medical care did not differ significantly by provider type [25]. In contrast to physicians' perceptions, more patients believed family practitioners had the training (69%) and time (69%) to address their non-dialysis related symptoms compared to nephrologists (training: 46% and time: 37%) [17].
Time, use, reliance and primary care service delivery Physician estimates of delivery of primary care was determined via self-report, whereas receipt of primary care services was ascertained from patient self-report or administrative data collection (Table 3). Four studies assessed physician time dedicated to provision of primary care, all of which were reported by nephrologists. Most nephrologists reported spending at least a portion of their time delivering primary care, with varying reports of the amount of actual clinical time dedicated for primary care [14][15][16][17]. This variation may reflect secular trends. For example, only 8% of nephrologists in 1981 reported spending 75% or more of their time on general internal medicine [16] compared to nephrologists in 1993 who reported spending an average 38% of their practice time on primary care [14] and 85% nephrologists in 2010 who reported "moderate" to "a lot" of their time managing symptoms in dialysis patients [15].
Dialysis patient use of PCPs was examined in several studies, from nephrologist, non-nephrologist provider, and patient perspectives [14,16,17,19,20,23]. There was wide variation in patients either having or relying on non-nephrologist PCPs (ranges of 20-87% of patients) [14,17,19,20,23]. Nephrologists and patients reported generally similar extents (20-35%) of dialysis patients having a PCP [14,19,20,23]. Patient-reported reliance on PCPs reflected these general trends, where   81% of patients relied on nephrologists for annual exams and minor illness [14] and reliance on PCPs diminished with dialysis vintage [19,23]. General trends in provision and receipt of primary care differed by type of service. Overall, studies reported similar rates of primary care services delivered by nephrologist and PCPs. Within broad categories of primary care services, we found several discernible patterns. Nephrologists reported providing higher rates of consultative referrals [14] than those reported by patients [19,20]. For counseling and preventive care services, nephrologists, PCPs, and patient reported generally consistent rates of preventive care (e.g., annual physicals, counseling, screening) delivered by both provider types [14,17,19,20].
There were, however, some inconsistencies in reporting by data source. Patients reported lower rates of receipt of mammogram and cervical cancer screening than reportedly delivered by providers [17]. In contrast to the relatively high rates of patient-or provider-reported prevention services [14,17,19,20], studies using administrative claims and medical chart data [21,22,24,26,27] found dialysis patients received suboptimal rates of preventive care. Rates of documented immunizations and vaccinations varied from 47% to 85% [21,22,26,27]. In a cohort of dialysis patients enrolled in Medicare and Medicaid, Winkelmayer and colleagues (2002) found high rates of diabetic eye exams (76%) but low rates of hemoglobin A1c testing (11%) and cancer screening (21-27%).
Most of the studies examining physicians' involvement in acute illness, symptom and disease management found similar rates of physician provision and patient receipt of care. For example, treatment of acute minor illness by nephrologists was reported by 72-85% of nephrologists and 91% of patients while 91% of PCPs also reported managing acute care illnesses for dialysis patients [14,17,20]. Similarly, high rates of nephrologist provision of disease management for diabetes, heart and gastrointestinal disease was found in patient-and nephrologistreported surveys [14,19,20]; and nephrologist and PCPdriven medication management for hyperlipidemia, diabetes, heart disease, and gastrointestinal disease reported by these physician types [17]. Notable exceptions to these trends were found in patients reporting overwhelmingly high to exclusive use of PCPs for new problems and mental health symptoms, yet lower rates of PCP use for follow-up care [15,17,18].

Discussion
In this review, we found a limited evidence base to inform our understanding of primary care services for chronic dialysis patients. From our systematic search of the healthcare literature, we found 14 studies spanning approximately three decades of research. Overall, we identified three themes related to primary care of dialysis patients: perceived roles of providers, estimated time in providing primary care, and the extent of dialysis patients' use of primary care services (e.g., referrals, prevention, acute illness and disease management).
Among the findings in this literature, we found overall general agreement among providers that PCPs should be involved in primary care of dialysis patients. Despite this agreement, a potential barrier to realizing this may be the adequacy of time for PCPs to provide good primary care to dialysis patients and subsequent incongruent For certain types of preventive care (e.g., cancer screening), physician referrals and direct delivery of preventive service are differentiated, appearing in separate outcome categories, where indicated expectation and coordination between provider specialties. At the same time, the logistical barriers to accessing care and time burden of dialysis care for patients are a significant factor, which may also attribute to the majority of dialysis patients relying on primary care services from their nephrologists, despite their report of greater confidence in primary care delivered by nonnephrologist PCPs. These findings generally apply to all dialysis patients, as found in the few studies that explicitly assessed differences between in-center hemodialysis and home-based peritoneal dialysis patients. Although more PD than HD patients report having family practitioner [19], PD patients were less likely to see them over time [19] and have similarly low vaccination rates as HD patients [21]. With increasing use of home-based dialysis [8], future research should explore differences in patient care perceptions and utilization by treatment modality. Across studies reported from a variety of patient and provider perspectives, we found varying degrees of consistencies of primary care service delivery and use with important implications on patient outcomes. On one hand, consistent reporting of delivery and receipt of primary care services may reflect accurate reporting across survey respondent types. On the other hand, high provision of care reported by nephrologists and PCPs may also suggest duplicative care. The pooled findings related to overall delivery and receipt of preventive care suggest potential gaps in care, where studies of administrative data found lower rates of preventive care services than found in studies relying on provider or patient selfreport. The mechanisms underlying these inconsistent findings is unclear: this may be due to recall bias in selfreported data and potentially incomplete documentation of care in administrative data, but may also highlight discordant expectations and communication between patients, nephrologists, and PCPs that lead to gaps in care. The difficulty in communication between providers [17] is also exacerbated by a lack of streamlined electronic health records between health systems and dialysis units, resulting in a figurative chasm of information exchange. Further, the included studies were limited in their ability to assess the contribution of patient non-adherence to provider recommendations for preventive care, which may reflect patient behavior as a barrier to receipt of primary care services rather than inappropriate, duplicative, or missing care patterns by providers.
Although providers generally agreed that PCPs should be involved in primary care of dialysis patients, the optimal extent of this involvement is unclear. For example, many patients report having a non-nephrologist PCP, but simultaneously report reliance on nephrologists for referrals, counseling, preventive care, symptom and disease management; services that traditionally fall under the expertise of a PCP in the non-dialysis setting. There are no formal guidelines on the provision of primary care for individuals with ESRD, and the clinical benefit and cost-effectiveness of traditional health maintenance in this high-risk population remains controversial [28,29]. While recent literature and consensus statements support reductions in cost-ineffective use of cancer screening in dialysis patients [30][31][32], the extent to which PCPs and nephrology providers are aware and/or agree with these recommendations is unclear.

Conclusion
Altogether, these findings highlight opportunities for improvement and have important implications on the amount and quality of primary care that dialysis patients receive and on healthcare policy, as depicted in Fig. 3. This figure illustrates the Swiss Cheese Model of accident causation, used in engineering safety and healthcare and described here in the context of provider care and coordination of dialysis patients. Although dialysis patients likely see more providers than depicted here, our example presents a simple model of just PCPs and nephrologists. Each slice represents care delivered by a provider, where the holes reflect unprovided care. Patient care is completed neglected when holes between the slices of cheese align. For patient care, there is potential overlap in care that may be unnecessary and duplicative, placing time burdens on patients and their families and providers and financial burden on the healthcare system. In addition, the pooled, complementary findings suggest the possibility of a distinction in roles and responsibilities for some primary care services that result in non-duplicative, singular care. Importantly, this evidence base suggests the greater likelihood of gaps in care, leaving missed opportunities for clinical intervention and increasing the risk of adverse events, exacerbation of symptoms and disease, and increased healthcare utilization and costs. While this is an emerging area of concern, it is useful to consider its importance and the findings from the extant literature in the context of changes in the healthcare delivery system for ESRD patients. Aided by improving survival rates, the US is experiencing growing ESRD prevalence and a shrinking nephrology and primary care workforce. Recent system-level interventions, such as Medicare's managed care program demonstration for ESRD patients and the Comprehensive ESRD Care model aim to reorganize ESRD patient care through the collaboration of dialysis clinics, nephrologists and other US providers. This latter effort has resulted in Medicarefunded ESRD Seamless Care Organizations (ESCO) to improve care coordination and outcomes for Medicare beneficiaries. Thus, the research synthesized here foreshadows the challenges and opportunities for ESCOs to retool structures and processes of care across provider types to improve clinical and logistical complexity of care for patients on chronic dialysis. Future research will want to examine the extent to which singular care is coordinated and communicated across patients, nephrologists and PCPs, as well as ways to optimize such multidisciplinary care for dialysis patients and improve outcomes. The extant literature presented here reveal gaps and opportunities for future research to augment and improve the evidence base. First, there is a paucity of claims-based research. The differential findings of studies based on administrative data versus individual self-reported observational studies, suggest that the truth is probably somewhere in between. Well-designed mixed methods research can examine the impacts of structures and processes of care coordination in a more nuanced way than quantitatively or qualitatively-based studies alone. Second, the studies examined in this review examined a limited set of outcomes in dialysis patients' primary care. More studies linking the relationship between providers and patients' use and receipt of primary care to outcomes (e.g., physical function, morbidity, hospitalization, mortality) will serve as important comparisons to the nascent ESCOs, whose value is yet to be determined.