Abstract
Objective
To compare physician‐level versus practice‐level primary care continuity and their association with expenditure and acute care utilization among Medicare beneficiaries and evaluate whether continuity of outpatient primary care at either/both physician or/and practice level could be useful quality measures.
Data Source
Medicare Fee‐For‐Service claims data for community dwelling beneficiaries without end‐stage renal disease who were attributed to a national random sample of primary care practices billing Medicare (2011–2017).
Study Design
Retrospective secondary data analysis at per Medicare beneficiary per year level. We used multivariable linear regression with practice‐level fixed effects to estimate continuity of care score at physician versus practice level and their associations with outcomes.
Data Collection/Extraction Method
We calculated clinician‐ and practice‐level Bice–Boxerman continuity of care index scores, ranging from 0 to 1, using primary care outpatient claims. Medicare expenditures, hospital admissions, emergency department (ED) visits, and readmissions were obtained from the Medicare Beneficiary Summary File: Cost and Utilization Segment. Ambulatory care sensitive conditions (ACSC) were defined using diagnosis codes on inpatient claims.
Principal Findings
We studied 2,359,400 beneficiaries who sought care from 13,926 physicians. Every 0.1 increase in physician continuity score was associated with a $151 reduction in expenditure per beneficiary per year (p < 0.01), and every 0.1 increase in practice continuity score was associated with $282 decrease (p < 0.01) per beneficiary per year. Both physician‐ and practice‐level continuity were associated with lower Medicare expenditures among small, medium, and large practices. Both physician‐ and practice‐level continuity were associated with lower probabilities of hospitalization, ED visit, admissions for ACSC, and readmission.
Conclusions
Primary care continuity of care could serve as a potent value‐based care quality metric. Physician‐level continuity is a unique value center that cannot be supplanted by practice‐level continuity.
Keywords: continuity of care, population health, primary care, value‐based payment
What is known on this topic
Continuity of care is one of the core functions of primary care, which is motivated by the holistic patient‐centered approach to improve human health.
Continuity of primary care has the potential to serve as a simpler and more effective measure to build trust with patients and communities and proactively manage population health.
The measurement of continuity of primary care is nuanced because of the complex structure of primary care delivery organizations and the communities they serve.
What this study adds
We conducted a rigorous analysis of physician‐ and practice‐level continuity of care measures and their association with outcomes using Medicare claims data.
We found continuity of primary care at both physician and practice levels are associated with lower Medicare cost and acute care utilization.
Continuity of primary care could serve as a potent value‐based care metric. Continuity of care with physician cannot be substituted by practice‐level continuity.
1. INTRODUCTION
Primary care is the cornerstone of the U.S. health care system, with nearly 230,000 primary care physicians (PCPs) providing more than 500 million visits per year addressing preventive, acute, and chronic care. 1 , 2 , 3 The positive impact of primary care on community and population health is well established, due in part to the core function of providing continuous, relationship‐based care. 4 Specifically, volumes of literature show that greater continuity of care is associated with better performance on each of the elements of the Quintuple Aim of health care, including better clinical outcomes, less acute care utilization (including hospitalizations and emergency department [ED] visits), decreased spending (Medicare, private insurance, out‐of‐pocket), better patient experience, greater physician wellness, and improved equity in access and utilization of health care services. 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16
Despite the importance of continuity, federal value‐based payment and accountability programs have yet to use continuity of care measures for assessing clinician quality. Because of its broad scope and substantial scale, more clinical quality measures have been implemented in primary care than in any other specialty. However, most of the existing measures are disease‐specific or process measures that do not capture the core primary care functions, which are motivated by the holistic patient‐centered approach to improve human health. The diagnoses, treatment, and management of major chronic diseases such as diabetes and chronic kidney disease are long‐term processes involving multiple clinical encounters and relationships with patients, many of whom suffer from multiple complex comorbidities. 17 , 18 Therefore, quality payment mechanisms based on isolated clinical procedure measures do not necessarily lead to better population health. 4 , 19 , 20
Continuity of care has the potential to serve as a simpler and more effective measure that might capture PCP and practice efforts to build trust with patients and communities and to proactively manage population health. It could also serve to counter trends toward paring down patient time and overfilling clinician schedules, which may result in short‐term financial benefit but may be detrimental to patient outcomes overall, including through compromising continuity. 21 The measurement of continuity of care is nuanced because of the complex structure of primary care delivery organizations and the communities they serve. For the sake of policy design, the absence of continuity of care among quality metrics may be due, in part, to gaps in rigorous evidence‐based analysis of viable potential physician‐ and practice‐level measures and their association with outcomes.
Prior work has shown that patients with high continuity of care measured across all physician visit types have lower probability of hospitalization and lower spending. 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 One study has shown these benefits to be associated with continuity of care measured among primary care visits specifically (representing a metric over which a patient's PCP may have greater control). 11 No studies to our knowledge have simultaneously estimated whether primary care‐focused continuity of care for individual PCPs and/or practices is associated with better population health outcomes; nor have they compared the relative strength of association of physician‐ versus practice‐level continuity. Although there is evidence that primary care quality and community engagement may vary by clinic size, it is unknown how the association between continuity and outcomes varies by practice size. 22 , 23 Continuity of care at the practice level is likely higher for medium or large practices than for solo or small practices With more staff members and a greater scope of services, large practices may be able to maintain patient engagement with the organization more easily. However, it is not known whether practice‐level continuity of care could substitute for the continuity of care with a single physician within the practice. 24 , 25
In this study, we evaluated the effectiveness of physician‐ and practice‐level primary care continuity measures by estimating both simultaneously using a nationally representative sample of primary care practices. In order to investigate the role of primary care clinic size in these associations, we conducted analysis among the entire study sample, the solo practice sample, the small to medium practice sample (2–12 physicians), and the large practice sample (13 or more physicians) separately.
2. METHODS
2.1. Study design
We conducted a secondary data analysis using claims of Medicare beneficiaries who received services from a random sample of U.S. primary care practices between 2011 and 2017. The unit of analysis was per beneficiary per year. We calculated a primary care‐focused continuity of care score for each physician as well as each primary care practice per year. We then examined the associations between these two scores and annual Medicare expenditures and acute health care utilization per beneficiary. To enable within‐practice comparisons of individual physicians and obtain unbiased estimates of the practice level score, we used practice fixed effects to control for unobserved confounders that correlate to both care delivery patterns and outcomes. We also included a secondary analysis using continuity scores from the previous year's data to formulate a lagged continuity of care model.
2.2. Source of data
We used Medicare hospital care and outpatient clinical care claims for all beneficiaries attributed to a stratified random sample of 4265 primary care practices nationwide. The sample of practices was generated in 2011 from among all primary care practices submitting Medicare claims. We then merged the claims data with two additional data sources from Centers for Medicare and Medicaid Services (CMS): the Medicare Beneficiary Summary File (MBSF) Cost and Utilization Segment that contains the summary of annual Medicare expenditures and service utilizations, and the Medicare Data on Provider Practice and Specialty (MD‐PPAS) that contains characteristics of PCPs and their practices.
To construct the sample for analysis, we started with the pool of 65,443 practices (defined by unique Tax Identification Number [TIN]) from the 50 states of the United States, District of Columbia, and Puerto Rico that billed to Medicare in 2011. We then limited the practices to those in which the majority of physicians (50% or higher) were PCPs, including family medicine, internal medicine, general practice, and geriatric medicine, resulting in a pool of 50,076 TINs. We then drew a national random sample from this pool, stratified by rurality at the level of zip code and practice size. We oversampled larger practices with more than 12 physicians and practices in rural areas. The final random sample included 4265 practices with unique TINs. Because of clinic closures and mergers in the past decade, 26 , 27 , 28 , 29 our sample of unique TINs decreased year over year between 2011 and 2017 with an average 6.1% attrition rate annually. While TIN is often used to identify primary care practices in empirical research, 30 , 31 , 32 we acknowledge the limitation of TIN as an identifier because, while multiple practices within a larger health system, for example, Kaiser Permanente, may use a single TIN to bill for their services, physicians in other practices may bill under individual TINs.
2.3. Sample for analysis and beneficiary attribution
We analyzed inpatient and outpatient physician service claims of beneficiaries who were ever seen for any service at one of the 4256 practices identified by TIN between 2011 and 2017. The unit of analysis was per beneficiary per year. We included beneficiaries who did not have end‐stage renal disease (ESRD) as their reason for enrollment, were community‐dwelling, were continuously enrolled in FFS Medicare for a given year, did not die during the year, and had at least two evaluation and management service visits within a calendar year. The reason to exclude ESRD patients is that they are referred and managed by nephrologists. Primary care physicians diagnose and manage chronic kidney disease patients but refer the ESRD patients to nephrologists. We exclude the ESRD patients to focus on the primary care patient population. The reason to exclude decedents is that the beneficiaries die across the years from January to December. Therefore, the time frame to calculate continuity of care is not unanimous across the sample. To ensure the unbiasedness of the continuity of care score calculation, we excluded all the decedents. We attributed each beneficiary to the unique physician and practice that they visited the most frequently, and randomly assigned the beneficiaries to physicians or practices to resolve ties in the maximum number of visits. See Appendix A for further details.
2.4. Measures
2.4.1. Continuity of care
The physician‐level continuity measure was constructed in two steps. In the first step, we calculated a primary care‐focused Bice–Boxerman Continuity of Care (BB‐COC) Index (Bice–Boxerman Index) per beneficiary per year. 33 This index reflects the relative share of each beneficiary's outpatient primary care visits during the year that were billed by distinct physicians; the index ranges from 0 (each visit involved a different physician) to 1 (all visits to a single physician). 34 We chose BB‐COC Index because of the National Quality Forum's endorsement of the BB‐COC as a quality measure for care of children with complex needs in addition to its high correlation with other commonly used measures of continuity. 11 In the second step, we calculated the weighted average of the BB‐COC among all beneficiaries (i.e., weighting patient continuity scores by the total number of visits to each unique physician) who were attributed to a given physician defined by NPI per year as the continuity of care score of the physician. 11
Similarly, the primary care‐focused practice‐level continuity of care score was constructed by first calculating the Bice–Boxerman Index as the relative share of each beneficiary's outpatient primary care visits provided by a distinct practice TIN during the year. We then calculated the weighted average of the Bice–Boxerman Index across all the beneficiaries attributed to a given practice per year. 11
2.4.2. Outcomes
The major dependent variables included total Medicare expenditures (total of all covered services including inpatient care, outpatient clinical care, physician services, and skilled nursing facility, adjusted to 2020 dollar value based on the Consumer Price Index of Medical Care 35 ), probability of any inpatient admission, probability of any ED visit per beneficiary per year, and probability of 30‐day hospital readmission.
We also examined hospital admission for ambulatory care‐sensitive conditions (ACSC admission), defined using the primary diagnosis on the inpatient claims. We create a dummy variable to indicate whether each beneficiary had at least one hospital stay that met the criteria of Preventable Quality Indicator Overall Component published by the Agency for Healthcare Research and Quality, including hospital admission for diabetes complications (short‐term and long‐term), chronic obstructive pulmonary disease, hypertension, heart failure, and community‐acquired pneumonia. 36
2.4.3. Covariates
We used two groups of variables to control for beneficiary health status: risk adjustment score and pre‐existing chronic diseases. We calculated the Centers for Medicare and Medicaid Services Hierarchical Condition Category risk adjustment score (HCC) based on each beneficiary's age, gender, Medicare eligibility status (new vs. continuous), Medicaid enrollment, and ICD codes for primary diagnoses on inpatient and outpatient care claims. 37 We identified pre‐existing chronic diseases from the “ever” flag in the MBSF: Chronic Condition Segment, including diabetes, acute myocardial infarction, Alzheimer's isdease, congestive heart failure, stroke, chronic kidney disease, and cancer. Physician characteristics, including age, gender, TIN of the practice they are associated with, and the total number of physicians within their practice, were obtained from MD‐PPAS.
2.5. Analytical approach
We used regression models to estimate annual health care spending and utilization per year as a function of physician continuity of care score and practice continuity of care score controlling for patient age, race and ethnicity, number of months enrolled in Medicaid, Medicare Part D participation and other prescription drugs benefits, HCC score, and pre‐existing conditions, as well as physician age and gender.
We used ordinary least square (OLS) regression to estimate expenditures and a linear probability model to estimate the probabilities of hospital readmission, ED visits, ACSC hospital admission, and readmission. We estimated the regression model among the entire study sample, the solo practice sample, the small to medium practice sample (2–12 physicians), and the large practice sample (13 or more physicians) separately.
Acknowledging unobserved supply‐ and demand‐side community‐level characteristics that may confound the relationship between continuity and outcomes, we used TIN‐level fixed and random effects to control for time‐invariant practice‐level confounders. The Hausman test rejected the null hypothesis that the coefficients of random effects were consistent with fixed effect (p < 0.05) so we reported fixed‐effect estimates to draw conclusions. Standard errors were nested at the practice level. For those beneficiaries for whom there were 2 years of Medicare data available, we conducted a robustness check using continuity of care from the previous year as a primary independent variable and estimating a fixed‐effects model at the practice level.
All analyses were conducted on the Virtual Research Data Center (VRDC) platform of Research Data Assistance Center of CMS using Stata Version 16.1. This study was approved by the American Academy of Family Physicians Institutional Review Board.
2.6. Sensitivity analysis
Although we included beneficiaries with at least two primary care visits in our main analysis to maximize generalizability of the sample, we repeated our analysis among a subsample of beneficiaries with four or more visits per year, as prior studies have done, to optimize stability of the measure. 38 We found slight differences in the magnitude of the regression coefficient but the sign of the coefficients and the conclusions of the analysis did not change. The results of the sensitivity analysis are presented in the Appendix A.
3. RESULTS
3.1. Descriptive statistics of the beneficiaries
Our final sample included 6,503,512 observations of 2,359,400 unique beneficiaries who sought care from 13,926 unique physicians from 2011 to 2017. The average annual Medicare expenditures of all covered services per beneficiary was $13,408 among the sample population, and the rate of hospital admissions was 187.51 per 1000 beneficiaries (Table 1). The sample included 7.1% non‐Hispanic Black beneficiaries, 2.5% non‐Hispanic Asian and Pacific Islander beneficiaries, and 4.4% Hispanic beneficiaries. 39
TABLE 1.
Summary statistics of study sample: per beneficiary per year a
| Full sample N = 6,503,512 | Solo practice N = 1,774,438 | Small/medium practice (2–12 physicians) N = 2,024,843 | Large practice (≥13 physicians) N = 2,704,231 | |
|---|---|---|---|---|
| Health care cost and utilization | ||||
| Annual Medicare expenditures b | $13,408 (24,537) c | $13,990(25,498) | $13,579 (24,661) | $12,889 (23,778) |
| Hospital admission per 1000 beneficiaries | 187.51 | 184.12 | 187.83 | 187.18 |
| Health outcomes | ||||
| ED visits per 1000 beneficiaries | 329.32 | 321.90 | 330.81 | 333.30 |
| ACSC hospital admission per 1000 beneficiaries | 39.84 | 40.23 | 40.63 | 38.61 |
| 30‐day readmission per 1000 admissions | 151.81 | 154.41 | 153.31 | 149.64 |
| Demographics | ||||
| Age | 76.21 | 76.32 | 76.27 | 76.23 |
| Male | 40.63% | 41.34% | 41.12% | 39.92% |
| Race and ethnicity | ||||
| % White | 85.41 | 79.15 | 83.52 | 88.01 |
| % Black | 7.12 | 7.62 | 6.97 | 7.32 |
| % Asian and Pacific Islander | 2.53 | 5.41 | 3.42 | 1.35 |
| % Hispanic | 4.41 | 7.04 | 5.40 | 2.94 |
| % Other | 0.62 | 0.92 | 0.81 | 0.52 |
| Supplemental insurance | ||||
| Months enrolled in Medicaid | 1.53 | 1.81 | 1.52 | 0.91 |
| Months enrolled in Medicare Part D | 7.67 | 7.72 | 7.73 | 7.65 |
| Months enrolled in other drug benefits | 1.32 | 1.43 | 1.51 | 1.52 |
| Comorbidities | ||||
| HCC score | 0.65 (0.55) b | 0.68(0.58) | 0.66 (0.57) | 0.64 (0.54) |
| %Cancer (breast, prostate, lung) | 16.31 | 16.21 | 16.33 | 16.33 |
| %AMI | 5.12 | 5.03 | 5.01 | 5.13 |
| %Stroke | 15.43 | 15.91 | 15.87 | 14.82 |
| %CHF | 26.52 | 29.86 | 28.21 | 24.09 |
| %Diabetes | 40.32 | 45.57 | 42.50 | 37.18 |
| %Alzheimer | 4.71 | 5.32 | 5.01 | 4.32 |
| %COPD | 27.02 | 29.41 | 28.51 | 25.02 |
All the means and standard deviation reported are adjusted by sampling weight.
All the expenditures are inflated to 2020 dollar value based on the Consumer Price Index of Medical Services.
Standard deviation in parentheses.
3.2. Descriptive statistics of physicians and practices
The average continuity of care score per physician per year was 0.72, the continuity of care per practice per year was 0.86 (Table 2), and the distribution of the practice‐level continuity score was more skewed to the right than the distribution of the physician‐level continuity score (Figure 1).
TABLE 2.
Characteristics of physicians and practices
| Overall N = 54,971 | Solo practice N = 12,797 | Small or medium practice N = 14,286 | Large practice N = 27,558 | |
|---|---|---|---|---|
| Physician continuity score | 0.72 (0.21) a | 0.85(0.11) | 0.79 (0.15) a | 0.65 (0.24) a |
| Practice continuity score b | 0.86 (0.11) a | N/A | 0.87 (0.10) a | 0.83 (0.14) a |
| Physician age | 52.49 (11.31) a | 57.68(10.12) a | 52.14(10.78) a | 50.30 (11.31) a |
| % Male physicians | 66.81 | 77.02 | 70.31 | 60.43 |
| Physician‐patient panel | 324.72 (274.99) a | 310.34(265.20) a | 389.66 (309.36) a | 296.98 (254.53) a |
| Census track region | ||||
| % Midwest | 22.06 | 15.22 | 22.80 | 31.18 |
| % Northeast | 14.23 | 20.30 | 15.81 | 11.12 |
| % South | 48.00 | 38.88 | 40.41 | 43.78 |
| % West | 15.49 | 22.74 | 19.89 | 13.67 |
| % Puerto Rico | 1.10 | 2.81 | 1.01 | 2.24 |
| Census track rurality | ||||
| % Urban | 77.88 | 82.89 | 77.49 | 75.74 |
| % Urban rural combination | 7.61 | 0.38 | 1.15 | 15.42 |
| % Rural | 14.51 | 16.72 | 21.36 | 9.84 |
Standard deviation in parentheses.
Report of continuity score is the Bice–Boxerman score at per practice per year level.
FIGURE 1.
Distribution of continuity of care scores per physician versus per practice [Color figure can be viewed at wileyonlinelibrary.com]
Physicians and practice characteristics differed by size (Table 2). Solo practice physicians had the highest average physician continuity score at 0.85, compared with 0.79 among small or medium practice physicians and 0.65 among large practice physicians. The practice continuity scores of small and medium practices were slightly higher than those of large practices (0.87 vs. 0.84). Solo practice physicians were about 5 years older on average (57.7) than the small or medium practice physicians (52.1) and more than 7 years older than the large practice physicians (50.3). Solo practice physicians were mostly (77.0%) male, and in large practices, 60.4% of physicians were male. Large practices were most likely to serve beneficiaries from urban or urban/rural combination census tracts (75.7% and 15.4%), while small to medium practices were most likely serve beneficiaries exclusively in rural census tracts (21.4%).
3.3. Association between continuity of care, Medicare expenditures, and acute care utilization
Beneficiaries attributed to physicians or practices with higher continuity scores had lower Medicare expenditures and lower rates of hospital admissions, ED visits, ACSC admissions, and readmissions per member per year (Table 3).
TABLE 3.
Impact of 0.1 increase in continuity of care on Medicare expenditures and acute care utilization a
| Overall | Solo practice | Small or medium practice | Large practice | |
|---|---|---|---|---|
| Annual Medicare expenditures | ||||
| Physician continuity | −$151** | −$238** | −$217** | −$129** |
| Practice continuity | −$282** | N/A | −$136* | −$288** |
| Hospital admission per 1000 beneficiaries | ||||
| Physician continuity | −20** | −32** | −23** | −18** |
| Practice continuity | −24** | N/A | −13* | −11 |
| ED visits per 1000 beneficiaries | ||||
| Physician continuity | −52** | −45** | −55** | −51** |
| Practice continuity | −41** | N/A | −44** | −35** |
| ACSC admission per 1000 beneficiaries | ||||
| Physician continuity | −5** | −9** | −11** | −5** |
| Practice continuity | −4** | N/A | 1 | −7** |
| Readmission per 1000 beneficiaries | ||||
| Physician continuity | −6** | −8* | −14** | |
| Practice continuity | −11** | N/A | −31** | −43** |
Full results of regression analysis are presented in the Appendix A.
Statistically significant, p < 0.1.
Statistically significant, p < 0.05.
Spending: (Table 4) Each 0.1 increase in physician continuity score was associated with $151 lower annual expenditure (p < 0.01), while each 0.1 increase in practice continuity was associated with $282 lower annual expenditures (p < 0.01). Among solo practices, where the physician continuity score was the same as the practice continuity score, a 0.1 increase in continuity was associated with a $238 cost reduction (p = 0.01). Among small to medium practices, each 0.1 increase in physician continuity of care was associated with an average $217 reduction, and each 0.1 increase in practice continuity of care was associated with an average reduction of $136. Among the large practices, every 0.1 increase in physician continuity score was associated with $129 reduction in Medicare spending, and every 0.1 increase in practice continuity score was associated with $288 reduction in Medicare spending. In the lagged continuity of care model, each 0.1 increase in continuity of care in the previous year was significantly associated (p < 0.05) with a reduction of either $104 (physician‐level continuity metric) or $88 (practice‐level continuity metric).
TABLE 4.
Impact of 0.1 increase in continuity of care in previous year on Medicare expenditures and acute care utilization a
| Overall | Solo practice | Small or medium practice | Large practice | |
|---|---|---|---|---|
| Annual Medicare expenditures | ||||
| Physician continuity | −$104** | −127 | −$140** | −$119** |
| Practice continuity | −$88** | N/A | −$30 | −$57 |
| Hospital admission per 1000 beneficiaries | ||||
| Physician continuity | −19** | 9 | −15** | −25** |
| Practice continuity | −10* | N/A | −6 | −11 |
| ED visits per 1000 beneficiaries | ||||
| Physician continuity | −44** | −73** | −37** | −46** |
| Practice continuity | −1 | N/A | −10 | −4 |
| ACSC admission per 1000 beneficiaries | ||||
| Physician continuity | −3** | −4 | −1 | −6** |
| Practice continuity | −4 | N/A | −10* | 3 |
| Readmission per 1000 beneficiaries | ||||
| Physician continuity | −4** | −1 | −7** | − 4** |
| Practice continuity | −1 | N/A | −4 | − 5 |
All dollar values are 2020 value.
Statistically significant, p < 0.1.
Statistically significant, p < 0.05.
Utilization: Both physician and practice continuity of care scores were significantly associated with a lower probability of hospitalization when studied among the full sample and the sample subsets by practice size, with the exception of practice‐level continuity among large practices (coefficient = −11, p = 0.21). Every 0.1 increase in the physician continuity score was associated with a 1.8–3.2 reduction in hospital admissions per 1000 beneficiaries.
Continuity of care was also significantly associated with lower probability of ED admission, ACSC admission, and readmission. There was a larger association between physician continuity of care and ED visits than for practice continuity and ED visits for the overall sample (−52 vs. −41, p < 0.05). We also stratified by practice size and found that, while both physician level and practice level continuity of care scores were associated with lower ACSC admissions overall (coefficients at −5, p < 0.05 and −4, p < 0.05) and among large practices (coefficients at −7 and −5), practice‐level continuity was not associated with ACSC admission among small and medium‐sized practices. There was a larger association between practice continuity and readmission probability than for physician continuity and readmission probability, both across all practices and when stratifying by practice size. Among solo practices, physician continuity of care was not associated with readmission probability. In our lagged continuity of care model, a 0.1 increase in previous year's physician continuity score was associated with 19 fewer hospital admissions per 1000 beneficiaries, 44 fewer ED visits per 1000 beneficiaries, 3 fewer ACSC admissions per 1000 beneficiaries, and 4 fewer readmissions per 1000 beneficiaries (all p < 0.05). In the same model, this increase in past year's practice‐level continuity was not associated with any outcomes at the p < 0.05 level.
4. DISCUSSION
This study found that continuity of care at both the physician and practice levels was associated with lower Medicare expenditures and lower probability of acute care utilization. The association of continuity of care with the studied outcomes varied by practice size. Physician‐level continuity had a larger association than practice‐level continuity on spending among solo, small, and medium practices, while practice‐level continuity had a stronger association among the large practices. Notably, practice‐level continuity had a larger association than physician‐level continuity on hospital readmission rates regardless of practice size.
Our findings suggest that the implication of using continuity measures may vary by practice size. For solo and small to medium‐sized practices, individual physician practice patterns are the most influential factor of the practice. These practices usually face limited resources for care coordination and population health management, especially for complex and frail patients. 40 Financial incentives that reward continuity of care not only have the potential to help these practices maintain their connections with the community but also support their investment in and promotion of care coordination with specialists and hospitals. For large practices, which likely had more resources and greater administrative capacity, the results suggest that it is valuable to explore the pathways to improve physician‐level continuity within the practice for better outcomes.
Our findings suggest a complementary role for physician‐ and practice‐level continuity and may assuage some potential concerns about comparing physician‐level continuity among practices of different sizes. Intuitively, these two measures capture two dimensions of continuity of care that could vary greatly, especially in large practices. At the same time, smaller or solo physician practices naturally have higher physician‐level continuity of care, due to the practice scale and paucity of other practice physicians, such that implementing a physician‐level continuity of care measure could risk favoring small or solo practices. However, we find that physician‐level continuity is associated with better outcomes even when controlling for practice‐level continuity, including within the subsample of the large practices. This finding was also robust to analysis utilizing a lagged continuity of care score from the previous year, strengthening the potential impact of our findings. Our incorporation of both physician‐ and practice‐level continuity into our models helps distill the impact of physician behaviors within the practice, providing potentially useful guidance in formulating future continuity of care reporting measures that do not unduly favor small or solo practices.
We focused on continuity within outpatient primary care visits because this approach could help simplify measurement construction for documentation and data processing. 38 Such an approach follows the outpatient pediatric continuity of care measure that was endorsed by the National Quality Forum recently. 41 Although prior studies have measured continuity across visits with all specialists in all clinical settings, 34 we believe that future research that expands the scope of the study to include all the services provided by PCPs in both inpatient and outpatient care settings will further enrich the literature on the value of care continuity in improving population health, particularly for complex and frail patients.
4.1. Limitation
Because of limitations of the claims data, we could not observe the role of primary care team members such as nurses despite their importance to care continuity and coordination. We are aware that several community‐level characteristics that influence continuity of care at both the physician and practice levels, including certain measures of beneficiary health care utilization, expenditures, and outcomes, are not observed in our sample. These unobserved characteristics may include both supply‐side factors, such as practice infrastructure and capacity for care coordination (e.g., electronic health records [EHR] adoption and data sharing), team‐based care capacity, and workforce professionalism, as well as demand‐side factors, such as social determinants of health of the practicing community (e.g., transportation and food insecurity). 16 , 42 , 43 , 44 , 45 , 46 Studies that use EHR or clinical data will help fill this gap. In addition, although we used practice‐level fixed effects to obtain the most consistent estimates of the relationship between continuity of care measures and outcomes, this is a cross‐sectional observational study that is not designed to infer causality. Future research could include a more comprehensive sample with sophisticated stratification, including patients' mortality and disability for more consistent estimates.
4.2. Implication
This study affirms and builds on previous studies suggesting that continuity of care may be associated with lower spending and acute care utilization. 47 , 48 , 49 In addition, results reveal novel approaches to measuring and comparing primary care‐specific continuity at the practice level and at the physician level. 11
Measuring primary care‐focused continuity of care may present several advantages for practice physicians. These claims‐based measures may require less clerical work relative to existing measures and serve to reward what PCPs already value: namely longitudinal, trusting relationships with their patients. 50 , 51 Existing volume‐based incentives in the primary care setting emphasize patient access to care, leading to fuller physician schedules and, therefore, might work against continuity by overscheduling clinician time and making it hard for a given patient to see his or her own doctor. 15 , 52 , 53 Our results suggest that using continuity of care may be a valuable avenue toward improvement in population health and mitigation of system costs.
Contrary to anxiety over continuity and patient access becoming competing priorities, the recent rise of digital technology has the potential to serve both while improving the overall value of care. One recent study found direct online scheduling to be associated with higher continuity of care at the physician level; practice‐ and system‐level innovations serving both convenience and continuity will be essential to future improvement in access and outcomes. 54 Given the expanded use of telemedicine and digital care during and post the COVID−19 pandemic, future work should examine a broader definition of continuity that includes office visits as well as virtual care to inform value‐based payment reform.
ACKNOWLEDGEMENT
None.
APPENDIX A. TECHNICAL APPENDIX: TECHNICAL NOTE ABOUT BENEFICIARY ATTRIBUTION
The beneficiary who visited multiple physicians was attributed to the physician that he or she visited the most. If there was a tie between physicians for the greatest number of beneficiary visits to a single physician per year, we randomly picked one of the physicians.
The same principle applies to the beneficiaries who visited multiple practices. We attributed them to the practice they visited the most in a year, and unless there was a tie between most visited practices, we randomly picked one TIN number.
Applying the above attribution methodology produced two groups of beneficiaries: those attributed to the physicians within the 4256 TINS selected to be part of our 5% national random sample and those attributed to physicians outside of the random sample. We limited our sample of analysis to the first group. However, not all of the 4256 TINs qualified to the be attributed TINs; only 3856 TINs met the attribution rule to form the final sample of analysis.
In addition, the attributed practice (defined by TIN) is not necessarily the practice of the attributed physicians. For example, patient Jane Doe had five visits in 2011, two with Dr. Smith within practice A, and three with three different physicians (Drs. Johnson, William, and Brown) within Practice B. Our attribution rule assigns her to Dr. Smith of Practice A to evaluate the impact of physician quality but to practice B to evaluate the impact of practice quality. Jane Does account for 3% of the beneficiaries seeking care from the 3856 TINs, and we excluded them in our analysis (Tables A1, A2, A3, A4, A5, A6).
TABLE A1.
Regression results of the impact of continuity of care on annual Medicare expenditure
| Whole sample | Solo practice | Small practice | Large practice | |
|---|---|---|---|---|
| Continuity of care with physician | −1512.43** | −2382.52** | −2171.82** | −1291.16** |
| Continuity of care with practice | −2820.78** | N/A | −1361.36** | −2887.67** |
| Demographics | ||||
| Age | −292.29** | −442.86** | −246.95** | −201.00** |
| Age ^2/100 | −11.61 | 80.55** | −42.22* | −68.71** |
| Male | −544.29** | −572.18** | −562.21** | −550.02** |
| Race: Black | −119.67** | −166.01** | −135.46* | −114.03** |
| Race: Asian and Pacific Islander | −2372.09** | −2222.10** | −2500.02** | −2526.84** |
| Race: Hispanics | −830.95** | −1028.34** | −746.16** | −733.47** |
| Race: Others | −1100.99** | −1264.65** | −1076.54** | −968.84** |
| Supplemental insurances | ||||
| Part D coverage months | 232.19** | 218.43** | 232.26** | 234.05** |
| Other Rx benefit coverage months | −4.53 | 6.22 | 3.70 | −0.70 |
| Medicaid coverage months | 224.46** | 230.60** | 213.95** | 228.74** |
| Existing chronic diseases | ||||
| HCC score | 12,572.80** | 12,909.86** | 12,722.28** | 12,537.88** |
| Cancer | 3789.91** | 3907.19** | 3764.03** | 3703.56** |
| AMI | 6766.77** | 7547.66** | 6746.82** | 6218.95** |
| Stroke | 5512.70** | 5887.15** | 5517.62** | 5228.02** |
| CHF | 8665.64** | 8241.87** | 8512.43** | 9073.53** |
| COPD | 4789.72** | 4838.44** | 4826.79** | 4678.36** |
| Physician characteristics | ||||
| Physician age | −6.78** | 409.55** | −10.85** | −17.85** |
| Male physician | −43.88 | −1521.70** | 22.09 | 25.65 |
Statistically significant, p < 0.1.
Statistically significant, p < 0.05.
TABLE A2.
Regression results of the impact of continuity of care on hospital admission
| Whole sample | Solo practice | Small practice | Large practice | |
|---|---|---|---|---|
| Continuity of care with physician | −0.0198** | −0.0321** | −0.0226** | −0.0185** |
| Continuity of care with practice | −0.0244** | 0.0000 | −0.0128 | −0.0110 |
| Demographics | ||||
| Age | −0.0145** | −0.0166** | −0.0142** | −0.0131** |
| Age ^2/100 | 0.0084** | 0.0097** | 0.0082** | 0.0075** |
| Male | −0.0102** | −0.0109** | −0.0102** | −0.0101** |
| Race: Black | −0.0128** | −0.0135** | −0.0135** | −0.0128** |
| Race: Asian and Pacific Islander | −0.0321** | −0.0320** | −0.0314** | −0.0330** |
| Race: Hispanics | −0.0106** | −0.0107** | −0.0108** | −0.0108** |
| Race: Others | −0.0203** | −0.0200** | −0.0210** | −0.0210** |
| Supplemental insurances | ||||
| Part D coverage months | 0.0005** | 0.0005** | 0.0005** | 0.0006** |
| Other Rx benefit coverage months | −0.0002** | −0.0001 | −0.0003** | −0.0001 |
| Medicaid coverage months | −0.0003** | −0.0006** | −0.0003** | −0.0001 |
| Existing chronic diseases | ||||
| HCC score | 0.1435** | 0.1406** | 0.1449** | 0.1477** |
| Cancer | 0.0318** | 0.0324** | 0.0312** | 0.0316** |
| AMI | 0.1292** | 0.1356** | 0.1291** | 0.1242** |
| Stroke | 0.0967** | 0.0964** | 0.0967** | 0.0966** |
| CHF | 0.1239** | 0.1132** | 0.1218** | 0.1329** |
| COPD | 0.0816** | 0.0797** | 0.0831** | 0.0813** |
| Physician characteristics | ||||
| Physician age | −0.0002** | 0.0031** | −0.0003** | −0.0002** |
| Male physician | 0.0027** | −0.0143 | 0.0022** | 0.0033** |
Statistically significant, *p < 0.1.
Statistically significant, p < 0.05.
TABLE A3.
Regression results of the impact of continuity of care on the probability of ED visit
| Whole sample | Solo practice | Small practice | Large practice | |
|---|---|---|---|---|
| Continuity of care with physician | −0.0519** | −0.0454** | −0.0552** | −0.0513** |
| Continuity of care with practice | −0.0410** | 0.0000** | −0.0442** | −0.0351** |
| Demographics | ||||
| Age | −0.0299** | −0.0314** | −0.0297** | −0.0286** |
| Age ^2/100 | 0.0204** | 0.0211** | 0.0203** | 0.0196** |
| Male | −0.0301** | −0.0318** | −0.0299** | −0.0295** |
| Race: Black | 0.0427** | 0.0360** | 0.0425** | 0.0447** |
| Race: Asian and Pacific Islander | −0.0552** | −0.0546** | −0.0539** | −0.0583** |
| Race: Hispanics | 0.0142** | 0.0125** | 0.0149** | 0.0143** |
| Race: Others | −0.0206** | −0.0272** | −0.0191** | −0.0162** |
| Supplemental insurances | ||||
| Part D coverage months | 0.0000 | −0.0001 | −0.0002** | 0.0000 |
| Other Rx benefit coverage months | −0.0016** | −0.0010** | −0.0016** | −0.0016** |
| Medicaid coverage months | 0.0049** | 0.0041** | 0.0047** | 0.0056** |
| Existing chronic diseases | ||||
| HCC score | 0.1140** | 0.1137** | 0.1141** | 0.1189** |
| Cancer | 0.0212** | 0.0217** | 0.0213** | 0.0205** |
| Ami | 0.1221** | 0.1261** | 0.1232** | 0.1177** |
| Stroke | 0.1403** | 0.1381** | 0.1411** | 0.1409** |
| CHF | 0.1295** | 0.1209** | 0.1266** | 0.1377** |
| COPD | 0.1022** | 0.0977** | 0.1004** | 0.1059** |
| Physician characteristics | ||||
| Physician age | −0.0003** | 0.0069** | −0.0004** | −0.0004** |
| Male physician | 0.0015** | −0.0286** | 0.0013 | 0.0026** |
Statistically significant, *p < 0.1.
Statistically significant, p < 0.05.
TABLE A4.
Regression results of the impact of continuity of care on the probability of ACSC admission
| Whole sample | Solo practice | Small practice | Large practice | |
|---|---|---|---|---|
| Continuity of care with physician | −0.0053** | −0.0086** | −0.0111** | −0.0052** |
| Continuity of care with practice | −0.0051** | 0.0000 | 0.0040 | −0.0069* |
| Demographics | ||||
| Age | −0.0109** | −0.0115** | −0.0112** | −0.0103** |
| Age ^2/100 | 0.0068** | 0.0071** | 0.0069** | 0.0064** |
| Male | −0.0042** | −0.0041** | −0.0045** | −0.0040** |
| Race: Black | 0.0039** | 0.0029** | 0.0036** | 0.0042** |
| Race: Asian and Pacific Islander | −0.0043** | −0.0040** | −0.0052** | −0.0037** |
| Race: Hispanics | 0.0013** | 0.0016** | 0.0019** | 0.0004 |
| Race: Others | −0.0003 | −0.0009 | −0.0005 | 0.0007 |
| Supplemental insurances | ||||
| Part D coverage months | 0.0000 | 0.0000 | 0.0000 | 0.0000 |
| Other Rx benefit coverage months | −0.0001** | −0.0002** | −0.0001** | 0.0000 |
| Medicaid coverage months | 0.0006** | 0.0003** | 0.0007** | 0.0008** |
| Existing chronic diseases | ||||
| HCC score | 0.0459** | 0.0465** | 0.0456** | 0.0461** |
| Cancer | 0.0021** | 0.0025** | 0.0020** | 0.0020** |
| Ami | 0.0282** | 0.0323** | 0.0289** | 0.0247** |
| Stroke | 0.0132** | 0.0144** | 0.0145** | 0.0113** |
| CHF | 0.0604** | 0.0534** | 0.0600** | 0.0658** |
| COPD | 0.0466** | 0.0454** | 0.0473** | 0.0466** |
| Physician characteristics | ||||
| Physician age | −0.0002** | −0.0007** | −0.0002** | −0.0001** |
| Male physician | 0.0019** | 0.0072 | 0.0019** | 0.0016** |
Statistically significant, p < 0.1;
Statistically significant, p < 0.05.
TABLE A5.
Regression results of the impact of continuity of care on the probability of readmission
| Whole sample | Solo practice | Small practice | Large practice | |
|---|---|---|---|---|
| Continuity of care with physician | −0.0059** | −0.0051 | −0.0085** | −0.0059** |
| Continuity of care with practice | −0.0108** | 0.0000** | −0.0084** | −0.0138** |
| Demographics | ||||
| Age | −0.0019** | −0.0025** | −0.0022** | −0.0011** |
| Age ^2/100 | 0.0005** | 0.0009** | 0.0006** | 0.0000 |
| Male | −0.0026** | −0.0025** | −0.0030** | −0.0024** |
| Race: Black | −0.0005* | 0.0003 | −0.0015** | −0.0006 |
| Race: Asian and Pacific Islander | −0.0038** | −0.0044** | −0.0035** | −0.0031** |
| Race: Hispanics | −0.0010** | −0.0008 | −0.0018** | −0.0006 |
| Race: Others | −0.0016** | −0.0009 | −0.0017 | −0.0024 |
| Supplemental insurances | ||||
| Part D coverage months | 0.0000** | 0.0000 | 0.0000 | 0.0000 |
| Other Rx benefit coverage months | −0.0002** | −0.0002** | −0.0002** | −0.0002** |
| Medicaid coverage months | −0.0002** | −0.0004** | −0.0002** | −0.0002** |
| Existing chronic diseases | ||||
| HCC score | 0.0525** | 0.0529** | 0.0527** | 0.0531** |
| Cancer | 0.0063** | 0.0063** | 0.0065** | 0.0062** |
| Ami | 0.0403** | 0.0426** | 0.0407** | 0.0379** |
| Stroke | 0.0210** | 0.0208** | 0.0219** | 0.0204** |
| CHF | 0.0348** | 0.0310** | 0.0344** | 0.0379** |
| COPD | 0.0196** | 0.0189** | 0.0202** | 0.0194** |
| Physician characteristics | ||||
| Physician age | 0.0000** | 0.0014** | −0.0001** | −0.0001** |
| Male physician | 0.0006** | −0.0012 | 0.0009** | 0.0007** |
Statistically significant, p < 0.1;
Statistically significant, p < 0.05.
TABLE A6.
Sensitivity analysis about the impact of 0.1 increase in Bice–Boxerman score on outcomes, among the beneficiaries with four or more primary care visits per year
| Overall N = 3,634,706 | Solo practice N = 1,124,495 | Small or medium practice N = 1,146,894 | Large practice N = 1,146,894 | |
|---|---|---|---|---|
| Annual Medicare expenditures | ||||
| Physician continuity | −$238** | −$290** | −$306** | −$199** |
| Practice continuity | −$241** | N/A | −$93* | −$236** |
| Hospital admission per 1000 beneficiaries | ||||
| Physician continuity | −26** | −34** | −35** | −23** |
| Practice continuity | −19** | N/A | −11* | −1 |
| ED visits per 1000 beneficiaries | ||||
| Physician continuity | −58** | −50** | −60** | −56** |
| Practice continuity | −39** | N/A | −43** | −27** |
| ACSC admission per 1000 beneficiaries | ||||
| Physician continuity | −6** | −9** | −11** | −7** |
| Practice continuity | −5 | N/A | −1 | −8* |
| Readmission per 1000 beneficiaries | ||||
| Physician continuity | −9** | −5 | −13* | −8** |
| Practice continuity | −7** | N/A | −10** | −3* |
Statistically significant, p < 0.1;
Statistically significant, p < 0.05.
Yang Z, Ganguli I, Davis C, et al. Physician‐ versus practice‐level primary care continuity and association with outcomes in Medicare beneficiaries. Health Serv Res. 2022;57(4):914‐929. doi: 10.1111/1475-6773.13999
REFERENCES
- 1. Phillips RL, McCauley LA, Koller CF. Implementing high‐quality primary care: a report from the National Academies of sciences, engineering, and medicine. JAMA. 2021;325(24):2437. doi: 10.1001/jama.2021.7430 [DOI] [PubMed] [Google Scholar]
- 2. Brian A, Bazemore A, George J, et al. Primary Care in the United States: A Chartbook of Facts and Statistics. Robert Braham Center, American Board of Family Medicine, IBM Watson Health. 2021.
- 3. Green LA, Fryer G, Yawn B, Lanier D, Dovey S. The ecology of medical care revisited. N Engl J Med. 2001;344(26):2021‐2025. [DOI] [PubMed] [Google Scholar]
- 4. National Academies of Sciences, Engineering, and Medicine . Implementing High‐Quality Primary Care: Rebuilding the Foundation of Health Care. The National Academies Press; 2021. [PubMed] [Google Scholar]
- 5. White Paper . Redifining Primary Care for the 21st Century. https://www.ahrq.gov/ncepcr/tools/workforce-financing/white-paper.html
- 6. Ionescu‐Ittu R, McCusker J, Ciampi A, et al. Continuity of primary care and emergency department utilization among elderly people. Can Med Assoc J. 2007;177(11):1362‐1368. doi: 10.1503/cmaj.061615 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Kao Y‐H, Tseng T‐S, Ng Y‐Y, Wu S‐C. Association between continuity of care and emergency department visits and hospitalization in senior adults with asthma‐COPD overlap. Health Policy. 2019;123(2):222‐228. doi: 10.1016/j.healthpol.2018.11.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Marshall EG, Clarke B, Burge F, Varatharasan N, Archibald G, Andrew MK. Improving continuity of care reduces emergency department visits by long‐term care residents. J Am Board Fam Med. 2016;29(2):201‐208. doi: 10.3122/jabfm.2016.12.150309 [DOI] [PubMed] [Google Scholar]
- 9. Katz DA, McCoy KD, Vaughan‐Sarrazin MS. Does greater continuity of veterans administration primary care reduce emergency department visits and hospitalization in older veterans? J Am Geriatr Soc. 2015;63(12):2510‐2518. doi: 10.1111/jgs.13841 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Knight JC, Dowden JJ, Worrall GJ, Gadag VG, Murphy MM. Does higher continuity of family physician care reduce hospitalizations in elderly people with diabetes? Popul Health Manag. 2009;12(2):81‐86. doi: 10.1089/pop.2008.0020 [DOI] [PubMed] [Google Scholar]
- 11. Bazemore A, Petterson S, Peterson LE, Bruno R, Chung Y, Phillips RL. Higher primary care physician continuity is associated with lower costs and hospitalizations. Ann Fam Med. 2018;16(6):492‐497. doi: 10.1370/afm.2308 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Amjad H, Carmichael D, Austin AM, Chang C‐H, Bynum JPW. Continuity of care and health care utilization in older adults with dementia in fee‐for‐service Medicare. JAMA Intern Med. 2016;176(9):1371‐1378. doi: 10.1001/jamainternmed.2016.3553 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Chen C‐C, Cheng S‐H. Better continuity of care reduces costs for diabetic patients. AJMC. 2011;17(6). https://www.ajmc.com/view/ajmc_11jun_chenchen_420to27. Accessed February 28, 2021. [PubMed] [Google Scholar]
- 14. Hussey PS, Schneider EC, Rudin RS, Fox DS, Lai J, Pollack CE. Continuity and the costs of Care for Chronic Disease. JAMA Intern Med. 2014;174(5):742‐748. doi: 10.1001/jamainternmed.2014.245 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Goodwin JS. Continuity of care matters in all health care settings. JAMA Netw Open. 2021;4(3):e213842. doi: 10.1001/jamanetworkopen.2021.3842 [DOI] [PubMed] [Google Scholar]
- 16. Baker R, Freeman GK, Haggerty JL, Bankart MJ, Nockels KH. Primary medical care continuity and patient mortality: a systematic review. Br J Gen Pract. 2020;70(698):e600‐e611. doi: 10.3399/bjgp20X712289 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Shrivastav M, Gibson W, Shrivastav R, et al. Type 2 diabetes Management in Primary Care: the role of retrospective, professional continuous glucose monitoring. Diabetes Spectr. 2018;31(3):279‐287. doi: 10.2337/ds17-0024 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Chen Y‐C, Weng S‐F, Hsu Y‐J, Wei C‐J, Chiu C. Continuity of care: evaluating a multidisciplinary care model for people with early CKD via a nationwide population‐based longitudinal study. BMJ Open. 2020;10(12):e041149. doi: 10.1136/bmjopen-2020-041149 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Adler R, Hamdan S, Scanlon C, Altman W. Quality measures: how to get them right. American Academy of Family Physicians. 2018. https://www.aafp.org/fpm/2018/0700/fpm20180700p23.pdf [PubMed]
- 20. Shuemaker JC, Phillips RL, Newton WP. Clinical quality measures in a post‐pandemic world: measuring what matters in family medicine (ABFM). Ann Fam Med. 2020;18(4):380‐382. doi: 10.1370/afm.2564 [DOI] [Google Scholar]
- 21. Phillips RL, Bazemore AM, Peterson LE. Effectiveness over efficiency: underestimating the primary care physician shortage. Med Care. 2014;52(2):97‐98. doi: 10.1097/MLR.0000000000000047 [DOI] [PubMed] [Google Scholar]
- 22. Liaw WR, Jetty A, Petterson SM, Peterson LE, Bazemore AW. Solo and small practices: a vital, diverse part of primary care. Ann Fam Med. 2016;14(1):8‐15. doi: 10.1370/afm.1839 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Isaacs SL, Jellinek PS, Ray WL. The independent physician — going. N Engl J Med. 2009;360(7):655‐657. doi: 10.1056/NEJMp0808076 [DOI] [PubMed] [Google Scholar]
- 24. Johnston KJ, Mittler J, Hockenberry JM. Patient social risk factors and continuity of care for Medicare beneficiaries. Health Serv Res. 2020;55(3):445‐456. doi: 10.1111/1475-6773.13272 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Colla C, Yang W, Mainor AJ, et al. Organizational integration, practice capabilities, and outcomes in clinically complex medicare beneficiaries. Health Serv Res. 2020;55(S3):1085‐1097. doi: 10.1111/1475-6773.13580 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. 163 Rural Hospital Closures: January 2005 – Present (121 since 2010). Cecil F. Sheps Center for Health Services Research. https://www.shepscenter.unc.edu/programs-projects/rural-health/rural-hospital-closures/. Accessed January 24, 2020.
- 27. Phillips RL. Preserving primary care robustness despite increasing health system integration. Fam Med. 2017;49(8):591‐593. [PubMed] [Google Scholar]
- 28. Fulton BD. Health care market concentration trends in the United States: evidence and policy responses. Health Aff. 2017;36(9):1530‐1538. doi: 10.1377/hlthaff.2017.0556 [DOI] [PubMed] [Google Scholar]
- 29. Furukawa MF, Machta RM, Barrett KA, et al. Landscape of health systems in the United States. Med Care Res Rev. 2020;77(4):357‐366. doi: 10.1177/1077558718823130 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Ganguli I, Souza J, McWilliams JM, Mehrotra A. Practices caring for the underserved are less likely to adopt Medicare's annual wellness visit. Health Aff. 2018;37(2):283‐291. doi: 10.1377/hlthaff.2017.1130 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. McWilliams JM, Chernew ME, Dalton JB, Landon BE. Outpatient care patterns and organizational accountability in Medicare. JAMA Intern Med. 2014;174(6):938‐945. doi: 10.1001/jamainternmed.2014.1073 [DOI] [PubMed] [Google Scholar]
- 32. Roberts ET, Mehrotra A, McWilliams JM. High‐price and low‐price physician practices do not differ significantly on care quality or efficiency. Health Aff Proj Hope. 2017;36(5):855‐864. doi: 10.1377/hlthaff.2016.1266 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Bice TW, Boxerman SB. A quantitative measure of continuity of care. Med Care. 1977;15(4):347‐349. [DOI] [PubMed] [Google Scholar]
- 34. Pollack CE, Hussey PS, Rudin RS, Fox DS, Lai J, Schneider EC. Measuring care continuity: a comparison of claims‐based methods. Med Care. 2016;54(5):e30‐e34. doi: 10.1097/MLR.0000000000000018 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Consumer Price Index. https://www.bls.gov/cpi/
- 36. Agency for Healthcare Research and Quality Prevention Quality Indicators Overview. 2020. https://www.qualityindicators.ahrq.gov/Modules/PQI_TechSpec_ICD10_v2020.aspx
- 37. Risk Adjustment. https://www.cms.gov/Medicare/Health-Plans/MedicareAdvtgSpecRateStats/Risk-Adjustors
- 38. Arthur KC, Mangione‐Smith R, Burkhart Q, et al. Quality of care for children with medical complexity: an analysis of continuity of care as a potential quality indicator. Acad Pediatr. 2018;18(6):669‐676. doi: 10.1016/j.acap.2018.04.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Race/Ethnicity of Medicare Population. 2019. https://www.kff.org/medicare/state-indicator/medicare-beneficiaries-by-raceethnicity/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22:%22asc%22%7D
- 40. Rich E, Lipson D, Libersky J, Parchman M. Coordinating care for adults with complex care needs in the patient‐centered medical home: challenges and solutions. 2012. https://pcmh.ahrq.gov/sites/default/files/attachments/coordinating-care-for-adults-with-complex-care-needs-white-paper.pdf [DOI] [PMC free article] [PubMed]
- 41. Pediatric Performance Measures 2017: Final Technical Report. National Quality Forum. 2017.
- 42. Bazemore AW, Cottrell EK, Gold R, et al. “Community vital signs”: incorporating geocoded social determinants into electronic records to promote patient and population health. J Am Med Inform Assoc. 2016;23(2):407‐412. doi: 10.1093/jamia/ocv088 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Fraze TK, Brewster AL, Lewis VA, Beidler LB, Murray GF, Colla CH. Prevalence of screening for food insecurity, housing instability, utility needs, transportation needs, and interpersonal violence by US physician practices and hospitals. JAMA Netw Open. 2019;2(9):e1911514. doi: 10.1001/jamanetworkopen.2019.11514 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Lichkus J, Liaw WR, Phillips RL. Utilizing PHATE: a population health–mapping tool to identify areas of food insecurity. Ann Fam Med. 2019;17(4):372. doi: 10.1370/afm.2387 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Altschuler J, Margolius D, Bodenheimer T, Grumbach K. Estimating a reasonable patient panel size for primary care physicians with team‐based task delegation. Ann Fam Med. 2012;10(5):396‐400. doi: 10.1370/afm.1400 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Syed ST, Gerber BS, Sharp LK. Traveling towards disease: transportation barriers to health care access. J Community Health. 2013;38(5):976‐993. doi: 10.1007/s10900-013-9681-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. Tammes P, Morris R, Murphy M, Salisbury C. Is continuity of primary care declining in England? Practice‐level longitudinal study, 2012–2017. Br J Gen Pract. 2020;71(707):e432‐e440. doi: 10.3399/BJGP.2020.0935 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48. Tammes P, Purdy S, Salisbury C, MacKichan F, Lasserson D, Morris RW. Continuity of primary care and emergency hospital admissions among older patients in England. Ann Fam Med. 2017;15(6):515‐522. doi: 10.1370/afm.2136 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49. Pereira Gray DJ, Sidaway‐Lee K, White E, Thorne A, Evans PH. Continuity of care with doctors—a matter of life and death? A systematic review of continuity of care and mortality. BMJ Open. 2018;8(6):e021161. doi: 10.1136/bmjopen-2017-021161 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50. Kocher B, Rajkumar R. Setting the stage for the next 10 years of health care payment innovation. JAMA. 2021;326(10):905‐906. doi: 10.1001/jama.2021.13153 [DOI] [PubMed] [Google Scholar]
- 51. Nowak DA, Sheikhan NY, Naidu SC, Kuluski K, Upshur REG. Why does continuity of care with family doctors matter? Can Fam Physician. 2021;67(9):679‐688. doi: 10.46747/cfp.6709679 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52. Goodwin JS, Zhou J, Kuo Y‐F, Nattinger AB. Work schedules of hospitalists and continuity of care during hospital stays of Texas Medicare patients: an observational study. J Gen Intern Med. 2019;34(4):489‐491. doi: 10.1007/s11606-018-4777-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53. Farid M, Tsugawa Y, Jena AB. Assessment of care handoffs among hospitalist physicians and 30‐day mortality in hospitalized Medicare beneficiaries. JAMA Netw Open. 2021;4(3):e213040. doi: 10.1001/jamanetworkopen.2021.3040 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54. Ganguli I, Orav EJ, Lupo C, Metlay JP, Sequist TD. Patient and visit characteristics associated with use of direct scheduling in primary care practices. JAMA Netw Open. 2020;3(8):e209637. doi: 10.1001/jamanetworkopen.2020.9637 [DOI] [PMC free article] [PubMed] [Google Scholar]