Abstract
Purpose: Continuity of care between patients and physicians is a defining element of primary care and a pillar of the Patient Centered Medical Home (PCMH) program. We aimed to investigate the level of short- and long-term continuity within a network of Federally Qualified Health Centers (FQHCs) and the relationship of continuity to PCMH recognition.
Methods: This multi-method study utilized Electronic Health Record data to investigate patient continuity, and survey data to investigate PCMH history. The study population included patients with at least 2 visits between 2008 and 2023 to one of 18 FQHCs. Continuity was measured by calculating the number of primary care providers (PCPs) seen by the patient and the usual provider of care index (UPC Index [the number of visits with the most frequent PCP/total visits]).
Results: Our population consisted of 1,323,547 patients and 19,768,516 encounters. The mean (SD) number of PCPs per patient over one year was 2.01 (1.1). For patients who had visits spanning at least 5 years, the mean was 7.2 (4.7). The mean one-year UPC was .72 (.25) and 5+ year UPC was .47 (.21). No meaningful association was found between continuity measures and PCMH recognition.
Conclusions: These findings show, on average, high numbers of PCPs and poor continuity with a single “usual provider of care” for each patient’s care over time at FQHCs. Leveraging performance measures, such as PCMH recognition, to incentivize continuity may be inadequate. Different approaches should be considered to preserve the long-term continuity at the heart of primary care.
- Community Health Centers
- Community Medicine
- Continuity of Care
- Electronic Health Records
- Occupational Burnout
- Patient-Centered Care
- Patient Centered Medical Home
- Physicians
- Primary Health Care
Introduction
Relationships matter in primary care. A patient’s relationship with their primary care physician is an instrumental asset and is associated with improved health outcomes,1–8 increased patient satisfaction,7,9 and lower utilization7,10,11 and costs.12,13 Indeed, relational continuity14 between physicians and care teams and the patients they serve is one of the core defining elements of primary care.15 There are numerous, well-validated, patient level measures to assess the degree to which a patient’s care is concentrated among a set of providers.16–19 These measures have rarely been applied to Federally Qualified Health Center (FQHC) settings.20 This is particularly concerning given evidence that continuity is more highly valued by vulnerable patients.21 There is also evidence that strong relationships with patients and continuity have been shown to be impactful in efforts to reduce and prevent physician burnout,22 a significant concern in the resource-constrained environment of FQHCs.
As the patient-centered medical home (PCMH) model was being conceptualized, relational continuity was a major pillar.23 PCMH is a National Committee for Quality Assurance (NCQA) evaluation program that provides recognition for organizations who adopt a model of care that facilitates partnership between patients, physicians, and their families. Core concepts of the model include team-based care, knowing your patients, continuity, care management, performance measurement, and quality improvement. A key principle is that “each patient has an ongoing relationship with a personal physician trained to provide first contact, continuous and comprehensive care.”23 A core (required) PCMH measure is to monitor the percentage of patient visits with their PCP/team during the calendar year (Patient Centered Access and Continuity 11: Patients Visits with Clinician/Team).24
Past studies have shown mixed associations of the PCMH model with quality and costs.25–30 While some studies found improvement on quality indicators26,27 and lower hospital utilization26 for PCMH recognized sites, other studies did not25,31 and gaps in knowledge remain.32 Few studies examined the patient and provider experience, and those that did had mixed results.33 One study found lower levels of staff burnout for PCMH model practices but the sample size was small.34 Few studies of the PCMH examined continuity,35,36 and none did so in a safety net setting.
Our previous research has explored some unintended consequences of performance metrics, such as high rates of repetition and inaccuracy for depression and anxiety screenings linked to performance metrics.37,38 These findings may indicate a focus on meeting the measure in place of the ultimate goal of enhancing care, a concept known as goal displacement or metric fixation.39,40 The result is an increase in provider and care team burden to comply with these measures without the intended patient benefit. Similarly, we hypothesize that existing performance metrics for continuity (specifically, those used for PCMH recognition) may be missing the mark and imposing an opportunity cost on organizations by spending limited administrative time and energy reporting on a measure. The act of measuring short-term continuity may not be the right lever to improve and ensure continuity. To investigate this hypothesis, we used Electronic Health Record (EHR) data both to examine and quantify the levels of short- and long-term relational continuity in FQHCs and to analyze the relationship between continuity and PCMH recognition.
Methods
This multi-method study utilized EHR data to investigate patient continuity, and data from surveys with FQHC administrators to investigate organizational characteristics associated with continuity. All patient data were deidentified to protect confidentiality. This study was granted an exemption from review by the Chicago Department of Public Health Institutional Review Board because it evaluated only previously collected, nonidentifiable data.
EHR Metadata
EHR data were collected from 18 health centers across 10 states within a network of FQHCs between June 1, 2008, and May 31, 2023. Variables collected from the EHR include: date of encounter, encounter provider, encounter provider type, provider specialty, responsible provider (designated primary care provider) and service location.
The study population included patients with at least 2 visits during the study period with a PCP. Our definition of a PCP included the following provider types: family physicians, general practitioners, nurse practitioners, internists, pediatricians, and physician assistants. Patients in hospice or long-term care facilities were excluded. As a measure of FQHC size, we calculated annual visit volume for each organization.
FQHC Administrator Survey Data
The survey instrument included 5 items capturing information about each organization’s history with PCMH recognition. The survey gathered information about whether none, some, or all an organization’s locations of care were granted PCHM recognition as well as the date of initial recognition. The survey was developed de novo and is available on request of the authors. It was emailed to administrative contacts at 18 health centers on November 1, 2023 with follow up e-mails sent 70 days later. Survey results are available in Appendix 1.
Statistical Analysis
Descriptive statistics, including frequency counts, percentages, means, and standard deviations (SDs) were calculated for all quantitative variables. A version of the Usual Provider of Care Index (UPC) developed by Breslau et al.16 was calculated as a continuous outcome variable. We selected this measure because the PCMH program requires that organizations monitor the proportion of a patient's visits with his or her provider and because it is highly correlated with other widely used measures such as the Bice-Boxerman Continuity of Care Index17 and the Herfindahl Index.41,42 The UPC Index is often calculated as the number of visits with the patient’s designated PCP over the patient’s total number of visits during the time period. However, we had limitations and concerns about the integrity of the PCP designations in our data set. Specifically, we only had access to the most recently designated PCP for a given patient with no information on when the designation was made. Further, there is great variation in how the designations are made across health centers. Finally, there is a significant number of patients with no designated PCP or an indication that the PCP is to be assigned (19.2%). Thus, we calculated the UPC Index as the number of visits with the patient’s most frequently seen PCP over the patient’s total number of primary care visits during the time period. A short-term UPC Index (1-year UPC Index) was calculated for all patients over a 1-year period from June 1, 2022–May 31, 2023. A longer-term UPC Index (5+ year UPC Index) was calculated for all patients that had at least 2 visits 5 or more years apart.
A linear mixed model was built to analyze the impact of PCMH status on UPC Index with clinic volume as a covariate. In the analysis, organizations reporting that some or all locations of care received PCMH recognition were grouped together to create a binary measure of PCMH status. PCMH recognition status was treated as a fixed effect; a random intercept was created to account for unobserved heterogeneity among health centers that are not explicitly accounted for in the model as fixed effects. Bivariate correlations were used to assess the strength of the association between PCMH recognition and variables of interest (UPC Index and the total number of PCPs seen over the last year). All analyses were conducted using IBM SPSS Version 29 for Windows and R version 3.6.0 for Windows, with a 2-sided significance threshold of P < .05.
Results
The study population included 1,323,547 patients seen by 3,435 providers (including 708 Advanced Practice Providers (APPs) and 2,727 physicians) at 19,768,516 encounters at 18 participating health centers. We excluded dental encounters (4.6%, 907,425/19,768,516), asynchronous patient communication such as secure messages and phone notes (5.5%, 1,082,346/19,768,516), and further limited our analysis to primary care encounters (54%, 10,741,773/19,768,516).
In our short-term analysis, the mean (SD) number of PCPs per patient over 1-year follow-up was 2.01 (1.1). In our long-term analysis including patients that had visits over at least a 5-year span, the mean number of PCPs was 7.2 (4.7) (Figure 1). The mean 1-year UPC Index was 0.72 (0.25) and 5+ year UPC was 0.47 (0.21). See Table 1 for a summary of continuity measure results for all organizations and for those with and without PCMH recognition.
Frequency distribution of number of Primary Care Providers (PCPs) per patient over 5+ years.
Continuity Measures Results
No meaningful differences were found between PCMH and non-PCMH groups as both had a mean 5+ year UPC Index of 0.47 (t = −0.082, P = .94) (Figure 2). There were small differences in mean 1-year UPC Index by PCMH recognition status, as the mean for the PCMH group was 0.71 whereas the mean UPC Index for the non-PCMH group was 0.76 (t = 25.6, P < .01). There was a weak, positive association between PCMH recognition and the number of PCPs seen over the last year (r = 0.037, P < .01); there was a similarly weak but negative association between PCMH recognition and the number of PCPs over 5+ years (r = −0.012, P < .01) (Figure 3). Both correlations fall below the 0.10 threshold indicating the strength of these associations is negligible. These results do not suggest a meaningful relationship between continuity and PCMH recognition. No significant association between health center visit volume and patient continuity were found (t = −2.0, P = .05).
Box plots of 5+ year UPC for organizations with and without PCMH recognition. Abbreviations: UPC, Usual Provider of Care Index; PCMH, Patient Centered Medical Home.
Box plots of number of PCPs per patient over 5+ years for organizations with and without PCMH recognition. Abbreviations: PCP, Primary Care Providers; PCMH, Patient Centered Medical Home.
Discussion
Continuity of care is a defining element of primary care. Various structural factors such as the consolidation from small physician practices to larger practice settings, increased physician turnover, growing number of physicians choosing to work part-time, and increasing pressures on physicians and APPs to meet productivity standards, however, threaten the ability of a patient to consistently see their designated PCP for primary care visits.
In our data set of over 15 years of patient encounters, the average patient who had been a patient for at least 5 years saw 7.2 PCPs and only saw their “usual provider of care” (provider seen most frequently) 47% of the time. On the extreme end, 1,802 (0.5%) of patients had seen 25 or more different PCPs, and 1 patient saw 83 PCPs.
Prior studies showed a UPC Index ranging from 0.4 to 0.84. These studies varied between 6 months to 3 year time intervals and also varied in the minimum number of visits required for inclusion in the data set. Several studies used a UPC Index of greater than or equal to 0.75 as a benchmark of good continuity. While the mean 1-year UPC Index in the current analysis (0.72) was slightly below the 0.75 threshold, our mean observed 5-year UPC Index was 0.47, indicating that long-term continuity was substantially lower than the desired standard. Despite this benchmark data in the literature, it is still unclear what the ideal UPC and number of PCPs per patient should be in a setting where continuity is optimized. If a practice is able to optimize continuity without impairing access, ideal UPC index would approximate 1.0 and ideal number of PCPs per patient would be closer to 1 or 2, rather than the 7.2 we observed.
The patient-centered medical home movement recognizes the importance of continuity between patients and their primary care clinician and team. However, the NCQA’s PCMH measure for continuity (AC 11: Patient Visits with Clinician/Team) did not seem to be a sufficient lever to ensure continuity in the practices we studied. Our data show no meaningful association between PCMH status and both short-term and long-term continuity (as measured by number of PCPs and UPC Index). We do not believe the very weak associations that we found in some cases to be practically meaningful. There could be numerous contributing factors to the lack of association. The NCQA PCMH continuity measure is a relatively short-term (calendar year) measure that does not necessarily align with the goal of long-term continuity. For example, if a patient’s PCP changes every year but the practice updates the PCP of record, perfect continuity of care is achieved according to the measure, yet true long-term relational continuity is not.
So why did even short-term continuity not correlate with PCMH recognition status? We believe this finding may result from a focus on measurement without the corresponding improvement effort needed to improve performance on the measure. As Don Berwick said, “we’ve put the cart before the horse—measurement is the cart and improvement is the horse.” Simply measuring continuity may not be enough to overcome structural barriers to continuity, such as physician turnover and lack of time due to productivity pressures. Rates of burnout and turnover among physicians is at an all-time high, and a patient cannot see their PCP, of course, if they are no longer there. In addition, if a PCP has a full template of 15-minute visits, it is often not possible for patients to get in to see their PCP when a need arises. The PCMH recognition strategy, with its focus on dozens of EHR metrics like “patient visits with clinician/team”24 puts the emphasis on measurement (the cart) and potentially distracts the limited attention and bandwidth that administrative leaders have from improving the deeper structural problems (the horse). To fix the structural problems plaguing primary care (high turnover; burnout; insufficient team size and stability; limited time), however, would require significant transformation from the typical care model currently practiced in FQHCs.
Our study has several limitations. First, the study was limited to a network of FQHCs that share a common EHR. Results may not be generalizable to other types of organizations. However, our sample does include 18 diverse health centers in 10 states. Our analysis does not capture visits outside the patient’s designated FQHC, such as urgent care visits. If we were able to include this data, the rates of continuity would be even lower than we found. The FQHC patient population, however, seeks care at urgent care centers less frequently than the commercially insured population69 so we speculate that the accuracy of PCP continuity data are better in our dataset than if working with commercially insured population and not having urgent care visit data at patient level. Another limitation is that data constraints prevented us from studying visits with designated PCP, which is the numerator used in the NCQA continuity measure (rather than most frequented PCP). We believe, however, that the UPC Index used in this analysis and the “Patient visits with clinician/team” NCQA measure24 to be highly correlated as the designated PCP is often a frequently seen provider. Our analysis of the relationship between long-term continuity and PCMH recognition is limited by the fact that our survey captured only current PCMH status. Our attempt to gather historic PCMH status information was incomplete and thus not included in the analysis. Further, only 4 of the 18 health centers fell into the non-PCMH recognition category. Despite the small number of organizations, this category included 32,344 patients and 99,137 visits. In addition, it is possible that some nurse practitioners or physician assistants in our data set were psychiatric providers as opposed to primary care. We were unable to differentiate these provider types in our data set but believe it represents a very small proportion of included providers given FQHCs’ focus on primary care. Finally, it is possible we may have gotten slightly different results if we had used different continuity measures (ie, Bice-Boxerman17). However, this limitation is mitigated by our use of a common, validated metric.
Conclusion
Continuity of care between patients and their primary care clinician is a cornerstone of primary care. The findings in our study show high numbers of primary care clinicians involved in each individual patient’s care over time. Patient centered medical home recognition is not associated with improved continuity, as measured by UPC index and number of PCPs per patient. The strategy of using process measures like “patient visits with clinician” feels like an intuitive step in the right direction but simply measuring continuity is not enough to overcome structural barriers to continuity, such as physician turnover, insufficient use of teams, and limited time. The solution to preserve continuity (and the heart of primary care) requires, instead, foundational care model transformation.
Acknowledgments
Ta-Yun Yang, MS, AllianceChicago, created the data set for analysis.
Appendix
Appendix 1. Organizational Survey Results
Notes
This article was externally peer reviewed.
Funding: This manuscript was prepared with the financial support of the American Medical Association’s Practice Transformation Initiative. The funder had no role in the design or conduct of the study, collection, management, analysis, or interpretation of the data.
Conflict of interest: The authors report no conflicts of interest. One of our authors, Dr. Christine Sinsky, was employed by the American Medical Association, but the opinions expressed in this article are those of the authors and should not be interpreted as American Medical Association policy.
To see this article online, please go to: http://jabfm.org/content/38/3/490.full.
- Received for publication October 2, 2024.
- Revision received December 12, 2024.
- Accepted for publication January 13, 2025.
