Abstract
Introduction
Timely access to primary care is essential for Medicare beneficiaries. Amid growing workforce shortages and consolidation, little is known about whether and how organizational and market-level factors affect access.
Methods
We conducted a simulated-patient study of 444 primary care clinics in Chicago, Los Angeles, New York, and Portland to measure acceptance of new Medicare patients and wait times to the earliest appointment.
Results
Overall, 77.5% of clinics accepted new Medicare patients, ranging from 96.9% in Los Angeles to just 35.0% in Portland. Among accepting clinics, median wait times for a physician varied from 8 days in New York City to 61 days in Portland. In adjusted analyses, each additional practice site was associated with a 1.5–percentage-point lower probability of accepting new Medicare patients (P < 0.001), and hospital or health system–affiliated practices had waits about 15 days longer than independent practices, with prolonged delays concentrated in Portland.
Conclusions
Findings highlight the importance of local organizational structure and market context in shaping access, with implications for workforce planning and access monitoring.
Keywords: primary care, access to care, Medicare, wait times, appointment availability, secret shopper
Introduction
Access to primary care is central to a well-functioning, equitable health system,1 but nearly one in four adults in the U.S. lacks a usual source of care. Policymakers have warned of a critical shortage of primary care clinicians,1-3 exacerbated by patient demand rebounding after the COVID-19 pandemic and clinicians exiting the workforce.4,5 Simultaneously, the structure of primary care delivery has changed dramatically, with independent practices increasingly being absorbed into large hospital systems and corporate entities.6 These trends may alter how primary care is organized, who gets seen, and how quickly.
Understanding the factors that shape primary care access is especially important for older adult populations. Medicare covers over 60 million beneficiaries, roughly one-fifth of the U.S population, yet evidence on how accessible primary care is for this group remains surprisingly limited.7 Recent data have documented rising access challenges,8,9 with limited access to a consistent primary care provider associated with poorer chronic disease management, delayed diagnosis, avoidable emergency department use, and preventable hospitalizations among beneficiaries.10
Evidence gaps persist in understanding primary care access for Medicare beneficiaries amid an evolving practice landscape.6,10 For example, national metrics of provider supply or appointment availability often obscure substantial variation at the local level, where workforce shortages, market consolidation, and organizational structure interact in ways that shape access. Moreover, Medicare beneficiaries differ meaningfully from younger, commercially insured adults in their care-seeking patterns, frequency of visits, and vulnerability to disruptions in access. To inform policy responses tailored to primary care needs, more detailed, practice-level information can help to understand how real-world access differs across markets and organizational settings.
Simulated patient (aka “secret shopper,” or audit) studies provide a valuable way to measure access to care by replicating the process of someone seeking a new patient appointment.11-16 By capturing pragmatic, real-world data, these studies capture appointment availability more accurately than administrative supply metrics or provider directory listings, which often overstate access.17,18 Prior secret shopper studies have focused on Medicaid or commercially insured populations and have reported results at broad geographic levels. Frequently cited in the gray literature, one proprietary audit conducted by the consulting firm AMN Healthcare (formerly known as Merritt-Hawkins), is based on an average of 16 calls per specialty per city. This study estimated that new patient appointment wait times averaged 31 days in metropolitan areas in the U.S., with wide variation by city and specialty.19
In this study, we add to the existing literature via a multi-city simulated patient study of primary care access for new Medicare patients, measuring two dimensions of access: (1) panel availability (ie, open to new Medicare patients) and (2) wait time across clinics in multiple metropolitan markets. We then examined the association of these outcomes with practice-level structural and organizational features including ownership, size, and specialty composition and city-level market characteristics (eg, market consolidation).
Methods
Study design and data sources
This cross-sectional simulated patient study assessed access to primary care for Medicare beneficiaries in four metropolitan areas: Chicago, Los Angeles, New York City, and Portland, Oregon. Practices were sampled from the 2023 IQVIA OneKey database, which we used to characterize organizational features and clinician composition.20 City selection incorporated metrics on practice size distribution, ownership mix, and specialty composition from IQVIA OneKey; hospital market concentration from the Health Care Cost Institute; social and demographic context from the Agency for Healthcare Research and Quality's Social Determinants of Health Database; Medicare Advantage penetration from the Centers for Medicare & Medicaid Services; and prior findings from the Merrit Hawkins survey. These sources together allowed identification of markets that differ by both market structure (eg, degree of consolidation, prevalence of large systems) and population characteristics (see Appendix A1a for more details) (To access the appendix, click on the Details tab of the article online).
Secret shopper data collection
Between November 2024 and April 2025, four trained callers used a standardized script and protocol to contact primary care clinics during local business hours (9 Am–4 Pm), up to three attempts on different days. Before calling, practice name, address, phone number, and adult primary-care status were verified through web searches. Callers used unique online phone numbers and recorded call date, time, and outcome after each attempt. Among clinics accepting Medicare, callers asked front-desk staff whether the practice was accepting new Medicare patients and the earliest available appointment date for a physician (MD/DOs) or advanced practice provider (APP; nurse practitioners and physician assistants).
From a sampling frame of 1094 clinics across four cities, 788 clinics were sampled for contact. Clinics were excluded if they had nonworking numbers, permanent closures, or duplicates (n = 56); inability to reach staff after 3 attempts on different days (n = 128); ineligibility based on clinic type (concierge or membership models), subspecialty clinics, or restricted populations (eg, HIV, homebound, or employer-specific clinics) (n = 101); inability to provide information (n = 13), and not accepting Medicare (n = 46). See Appendix A1b for further detail on study procedures, A1c for the call script, and A1d for the Sample Flow Diagram (To access the appendix, click on the Details tab of the article online).
Measures
The primary outcomes were (1) whether the practice reported accepting new Medicare patients (open panel), and (2) wait time, defined as the number of days from the call date to the earliest available new patient appointment. Practice characteristics included organization type (independent, hospital or health system–affiliated, other corporate medical practice, and safety net), multispecialty status, organizational scale (number of practice sites), clinician count, and percent physicians among clinicians (scaled to 10-percentage-point increments). Additional detail is provided in Appendix A1e (To access the appendix, click on the Details tab of the article online).
Statistical analysis
We summarized outcomes by city and practice characteristics. For open panel status, we estimated linear probability models with city fixed effects and HC3 robust standard errors and reported marginal effects in percentage points (PP). For wait time, we estimated generalized linear models with a log link and negative binomial family with city fixed effects and robust standard errors; average marginal effects are reported in days.
As sensitivity analyses, we re-estimated open-panel models using logistic regression and wait time models using Poisson regression; rationale for the primary specifications and overdispersion testing is reported in Appendix A1f (To access the appendix, click on the Details tab of the article online).We also conducted city-stratified and city-interaction models, along with additional robustness checks as described in Appendix A1f (To access the appendix, click on the Details tab of the article online).
Analyses were conducted in Stata MP 18.5 and R, version 4.5.1 (R Core Team, 2023).
Limitations
Several limitations warrant consideration. First, the study included only four urban markets and may not generalize to other metropolitan areas or rural settings. Although the cities vary in size, geography, and market structure, our findings are hypothesis-generating and do not isolate the effects of specific market characteristics. Additionally, wait times reflect routine access to primary care; more acute scenarios may be scheduled more quickly.21
Second, despite multiple call attempts, approximately 17% of potentially eligible clinics could not be reached. This nonresponse may introduce bias if unreachable clinics systemically differ by city, organization type, or other unobserved characteristics. For example, overloaded practices may be less likely to answer the phone or accept new patients. At the same time, our call outcomes reflect real-world barriers patients encounter when trying to schedule care.
Third, each metropolitan area was assigned a single caller, making caller identity collinear with city. Although callers all used a single standardized script and uniformly disclosed Medicare coverage, unmeasured caller-specific factors could contribute to between-city differences, and city comparisons should be interpreted cautiously.
Fourth, our analyses are observational and reflect associations rather than causal effects. Consolidation and organizational scale are not exogenous; large health systems may preferentially acquire practices in markets with supply constraints or growth opportunities, which could independently influence appointment availability and wait times.
Fifth, we excluded direct primary care and concierge practices because appointment availability could not be determined without membership. The prevalence of these excluded models varied across markets approximately 17% of out-of-sample clinics in New York and Portland and <2% in Chicago and Los Angeles highlighting differences in market composition. These models may offer alternative access pathways for some Medicare beneficiaries.
Finally, although we observed significant variation across cities, particularly in Portland, the specific drivers of this variation remain incompletely understood. Broader national sampling is needed to disentangle the role of local contexts in shaping primary care access.
Results
Descriptive market and sample characteristics
Of 444 primary care clinics in the final sample, 111 were in Chicago, 97 in Los Angeles, 136 in New York City, and 100 in Portland (Appendix A2a) (To access the appendix, click on the Details tab of the article online). Practice composition differed across markets (Appendix A2a) (To access the appendix, click on the Details tab of the article online). Mean practice sites were higher in New York (9.2) and Portland (14.1) than in Chicago (3.5) and Los Angeles (4.2). Portland also had a greater share of health system–affiliated clinics (57.0%) and single-specialty practices (86.0%), whereas safety-net clinics were more common in Chicago (33.3%) and Los Angeles (36.1%) than in New York (11.8%) and Portland (3.0%). City-level demographic and market context measures also varied (Appendix A2b) (To access the appendix, click on the Details tab of the article online). Notably, Portland had the highest primary care physician supply (116.3 per 100 000), the most concentrated hospital market (HHI = 2580), greater consolidation in primary care delivery, and the highest Medicare Advantage penetration (67.1%). Additional market and sample characteristics are reported in Appendix A2c (To access the appendix, click on the Details tab of the article online).
Acceptance of new Medicare patients
Nearly eight in ten clinics (77.5%) accepted new Medicare patients (Table 1). Descriptively, acceptance varied across cities: 96.9% of clinics in Los Angeles, 92.8% in Chicago, and 82.3% in New York City had open Medicare panels, compared with 35.0% in Portland (χ2(3) = 141.16, P < 0.001). Hospital or health system affiliated practices were less likely to accept new Medicare patients (57.4%) than safety-net clinics, independent, and other corporate practices (χ2(3) = 49.86, P < 0.001). Acceptance declined with organizational scale: clinics in large multi-site systems (>10 locations) had lower acceptance (53.9%) than single-site and mid-sized organizations (86.0% and 82.9%, respectively) (χ2(2) = 43.93, P < 0.001). Smaller practices (<5 providers) were more likely to accept new Medicare patients and had shorter wait times than larger practices (χ2(2) = 25.60, P < 0.001). Appendix A2d illustrates the overlap between organizational scale, ownership, and open-panel status across markets (To access the appendix, click on the Details tab of the article online).
Table 1.
Patient appointments available by practice location and characteristics, 2025.
| All practices: N = 444 | Open panel: N = 345 | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Practice characteristics | Accepting new medicare patients (n, %) |
P-Value | >1 provider accepting new patients (n, %) |
MD/DO provider available (n, %) |
MD/DO wait time (days), mean | MD/DO wait time (days), median (IQR) | APC available (n, %) |
APC wait time (days), mean | APC wait time (days), median, (IQR) | |
| Overall | 344 (77.5%) | 205 (59.6%) | 296 (86.1%) | 29 | 14 (4-39) | 77 (22.4%) | 17 | 8 (2-25) | ||
| Practice sites | 43.93 | <0.001 | ||||||||
| 1 site | 172 (86.0%) | 83 (48.3%) | 155 (90.12%) | 20 | 8 (3-26) | 34 (19.8%) | 15 | 4 (2-15) | ||
| 2-9 sites | 116 (82.9%) | 88 (75.9%) | 97 (83.6%) | 32 | 20 (7-42) | 33 (28.5%) | 15 | 11 (2-25) | ||
| >10 sites | 56 (53.9%) | 34 (60.7%) | 44 (78.6%) | 58 | 41 (16-64) | 10 (17.9%) | 26 | 25 (9-38) | ||
| Practice size | 25.60 | <0.001 | ||||||||
| < 5 Providers | 212 (85.5%) | 111 (52.4%) | 188 (88.3%) | 21 | 11 (3-25) | 49 (23.1%) | 14 | 7 (2-16) | ||
| 5-10 Providers | 80 (62.5%) | 60 (75.0%) | 69 (86.2%) | 38 | 22 (7-52) | 22 (27.5%) | 23 | 14 (3-38) | ||
| > 10 Providers | 52 (76.5%) | 34 (65.4%) | 40 (76.9%) | 54 | 27 (10-70) | 6 (11.5%) | 12 | 9 (4-15) | ||
| Percent physicians vs APCs | 4.69 | 0.030 | ||||||||
| ≥50th percentile (≥75%) | 171 (73.4%) | 92 (53.8%) | 155 (90.6%) | 33 | 12 (4-41) | 16 (9.4%) | 14 | 8 (2-25) | ||
| <50th percentile (<75%) | 173 (82.0%) | 113 (65.3%) | 141 (81.5%) | 25 | 15 (5-34) | 61 (35.3%) | 17 | 8 (2-25) | ||
| Practice type | 3.30 | 0.069 | ||||||||
| Single-specialty | 236 (75.2%) | 132 (55.9%) | 209 (88.6%) | 27 | 11 (3-32) | 57 (24.2%) | 13 | 6 (2-15) | ||
| Multi-specialty | 108 (83.1%) | 73 (67.6%) | 87 (80.6%) | 35 | 22 (7-54) | 20 (18.5%) | 26 | 24 (8-40) | ||
| Organization type | 49.86 | <0.001 | ||||||||
| Safety Net | 84 (92.3%) | 58 (69.1%) | 68 (81.0%) | 28 | 22 (10-44) | 32 (38.1%) | 21 | 17 (7-35) | ||
| Hospital or Health System | 74 (57.4%) | 54 (73.0%) | 63 (85.1%) | 60 | 28 (10-73) | 5 (6.8%) | 25 | 25 (15-35) | ||
| Corporate medical practice | 33 (97.1%) | 24 (72.7%) | 28 (84.9%) | 13 | 10 (3-20) | 15 (45.5%) | 5 | 3 (1-11) | ||
| Independent practice | 153 (80.5%) | 69 (45.1%) | 141 (89.8%) | 20 | 8 (3-26) | 25 (16.3%) | 17 | 3 (2-14) | ||
| City | 141.16 | <0.001 | ||||||||
| Chicago | 103 (92.8%) | 71 (68.9%) | 86 (83.5%) | 27 | 11 (4-41) | 23 (22.3%) | 18 | 11 (3-35) | ||
| Los Angeles | 94 (96.9%) | 60 (63.8%) | 85 (90.4%) | 22 | 14 (5-26) | 28 (29.8%) | 12 | 7 (2-15) | ||
| New York | 112 (82.3%) | 52 (46.4%) | 98 (87.5%) | 20 | 8 (3-26) | 14 (12.5%) | 12 | 3 (2-25) | ||
| Portland Metro | 35 (35.0%) | 22 (62.9%) | 27 (77.1%) | 90 | 61 (28-130) | 12 (34.3%) | 30 | 19 (3-42) | ||
Source: Authors' analysis of primary data collection linked to IQVIA OneKey for practice characteristics.
Secret shopper calls were conducted from November 2024 to April 2025. Practice size was calculated from IQVIA 2023 data, based on number of affiliated providers (physicians and advanced practice clinicians). Organizational types are mutually exclusive. MD/DO and APC provider availability does not total to 100% because some practices could not access specific schedules without patient registration.
After adjustment, the lower acceptance observed among health system-affiliated clinics were attenuated and reversed, indicating that practice size rather than ownership alone explained much of the difference (Table 2). Each additional practice site was associated with a 1.5-percentage-point lower probability of an open panel (CI −2.0 to −1.1; P < 0.001), and hospital and health system practices were 17.3 pp more likely than independent practices to accept new Medicare patients (CI 5.2 to 29.4; P = 0.005). Total clinician count was not associated with Medicare patient acceptance. Results were qualitatively similar using logistic regression (Appendix A3a) (To access the appendix, click on the Details tab of the article online).
Table 2.
Association between practice characteristics and acceptance of new medicare patients.
| Difference (pp) | 95% CI | P-value | |
|---|---|---|---|
| Characteristics | |||
| Practice Sites (per 1 additional practice) | −1.5 | (−2.0, −1.1) | <0.001 |
| Provider Count (per 1 additional provider) | 0.1 | (−0.2, 0.4) | 0.475 |
| Percent Physicians (vs APCs) | −1.2 | (−2.1, −0.3) | 0.011 |
| Practice Type | |||
| Multi-specialty | ref | — | — |
| Single-specialty | −0.1 | (−8.2, 6.1) | 0.777 |
| Organization Type | |||
| Independent practice | ref | — | — |
| Safety Net | 6.6 | (−2.1, 15.3) | 0.136 |
| Hospital or Health System | 17.3 | (5.2, 29.4) | 0.005 |
| Other corporate medical practice | 12.1 | (3.7, 20.6) | 0.005 |
| City | |||
| Chicago | ref | — | — |
| Los Angeles | 7.0 | (0.4, 13.6) | 0.034 |
| New York | −0.1 | (−8.4, 08.2) | 0.987 |
| Portland Metro | −43.8 | (−55.9, −31.8) | <0.001 |
Source: Authors' analysis of primary data collection linked to IQVIA OneKey for practice characteristics.
Open panel is a binary outcome indicating whether a practice accepts new Medicare patients. Probability estimates are displayed in PP to represent percentage-point differences in the probability of accepting new Medicare patients. A linear probability model with city fixed effects and HC3 robust standard errors was used (n = 444). Baseline mean acceptance is 77.5% (see Table 1).
Appointments offered and wait times
Among clinics accepting new Medicare patients (n = 344), 86.1% offered appointments with physicians and 22.4% with APCs (Table 1). Median physician wait times were 14 days overall, ranging from 8 days (IQR, 3-26) in New York to 61 days (IQR, 28-130) in Portland. Median APC wait times ranged from 7 days (IQR, 2-15) in Los Angeles to 19 days (IQR, 3-42) in Portland. Adjusted mean predicted waits (Figure 1) were 20.5 days in Chicago (IQR, 17-31), 20.6 days in Los Angeles (IQR, 14-26), 15.6 days (IQR, 11-26) in New York City, and 57.7 days (IQR, 48-114) in Portland, indicating substantial between-city variation after accounting for ownership, size, and staffing. Findings were similar using Poisson models (Appendix A3b).
Figure 1.
Adjusted mean wait times for new patient appointments by metropolitan area, 2025. Source: Authors' analysis of primary data collection. Figure displays adjusted mean wait times (in days) by city, estimated from a generalized linear model with a log link and negative binomial family. Estimates reflect predicted wait times adjusted for practice characteristics, organization type, and number of providers. Error bars represent 95% confidence intervals, calculated using robust standard errors.
Regression models (Table 3) identified several contributors to longer times. Single-specialty practices had shorter waits than multispecialty practices (−11.3 days; CI = −21.5, −1.2; P = 0.029). Relative to independent practices, hospital or health system practices had longer predicted waits (+15.5 days; CI = 0.5, 30.5, P = 0.043), while other corporate practices had shorter waits (−7.9 days; CI = −15.3, −0.5; P = 0.038). Safety net clinics did not differ significantly from independent practices (+5.4 days; CI = −3.4, 14.1; P = 0.227). The number of practices, total clinician count per practice, and physician share were not associated with meaningful differences in wait time.
Table 3.
Association between practice characteristics and wait time for new patient appointments.
| AME (Days) | 95% CI | P-value | |
|---|---|---|---|
| Characteristics | |||
| Practice Sites (per 1 additional practice) | 0.3 | (−0.1, 0.8) | 0.154 |
| Provider Count (per 1 additional provider) | 0.1 | (−0.1, 0.4) | 0.331 |
| Percent Physicians (vs APCs) | 1.3 | (0.7, 26.2) | 0.049 |
| Practice Type | |||
| Multi-specialty | ref | — | — |
| Single-specialty | −11.3 | (−21.5, −1.2) | 0.029 |
| Organization Type | |||
| Independent practice | ref | — | — |
| Safety Net | 5.4 | (−3.4, 14.1) | 0.227 |
| Hospital or Health System | 15.5 | (0.5, 30.5) | 0.043 |
| Other corporate medical practice | −7.9 | (−15.3, −0.5) | 0.038 |
| City | |||
| Chicago | ref | — | — |
| Los Angeles | −0.2 | (−8.0, 8.3) | 0.971 |
| New York | −5.6 | (−13.6, 2.4) | 0.169 |
| Portland Metro | 42.4 | (19.0, 65.8) | <0.001 |
Source: Authors' analysis of primary data collection linked to IQVIA OneKey for practice characteristics.
Wait time is measured in days. Estimates represent average marginal effects (AMEs) from a generalized linear model with a log link and negative binomial family, using Huber–White robust standard errors. Coefficients indicate the average change in predicted wait time (in days) associated with each characteristic, holding other covariates constant; city fixed effects are included. N = 323. Baseline mean wait time is 26 days.
Sensitivity analyses
City-stratified and city-by-covariate interaction models (Appendix A4a-d) showed that associations varied across markets, suggesting that wide confidence intervals in pooled models reflect heterogeneity rather than insufficient power (To access the appendix, click on the Details tab of the article online). For example, in New York, single-specialty practices had shorter waits than multispecialty practices (−16.0 days; CI −28.6, −3.4; P = 0.013), while in Los Angeles other corporate medical practices had shorter waits than independent practices. In Portland, each additional practice site was strongly associated with lower Medicare acceptance (−2.3 pp; CI = −3.1, −1.4; P < 0.001), and safety-net and hospital–affiliated clinics were more likely than independent practices to accept new Medicare patients.
Discussion
This multi-market secret shopper study provides novel post-pandemic evidence on how primary care access for Medicare beneficiaries varies across organizational and geographic contexts. Across 444 clinics sampled in four metropolitan markets, nearly 80% accepted new Medicare patients, but access varied substantially from near universal acceptance in Los Angeles to 35.0% in Portland. Among clinics with open panels, predicted wait times varied from just over 2 weeks in New York to nearly 2 months in Portland, even after adjusting for practice ownership, size, and staffing models. These findings align with a recent proprietary survey of commercial insurance appointment availability,19 but extend the literature by focusing on Medicare and by linking observed access to organizational characteristics.
Prior work measuring appointment wait times includes national benchmarking reports based on telephone scheduling inquiries and simulated patient and administrative-data studies, which similarly find substantial variation in wait times across settings and geographic areas.22,23 Our findings are consistent with this literature. For example, AMN Healthcare's benchmarking report finds longer wait times in certain metropolitan areas, including Portland (45 days for family medicine), compared with areas with relatively shorter wait times, like Los Angeles (13 days) and New York (4 days). Although the report is not Medicare-specific and reflects a different sampling frame and set of specialties,19 these directionally consistent findings of market-level differences highlight the need for future research to understand causes. Simulated patient studies and VA-based analyses likewise report variable delays in securing outpatient appointments, using different populations and scheduling pathways than in our study.22,23 In one study examining differences in wait times for veterans, mean wait times ranged from 25 to 52 days in community primary care settings, with shorter wait times in the VA system.23
Our findings add to this literature by demonstrating that access is shaped not just by workforce supply, but also by how care is organized. Despite all four cities in our study having relatively high primary care physician supply,24 we observed striking differences in new patient access. Prior research in a single metropolitan area found no association between primary care appointment availability and local physician supply.13 While the reasons for variable access to care remain poorly understood, local market structures, workforce distribution, and practice ownership may be as important as aggregate supply alone. Importantly, our findings pertain to urban areas, and may underestimate different barriers in rural settings, where workforce shortages are more acute.24
A growing body of work has examined within-market differences in access to care.12,16,25,26 For example, Saloner and colleagues16 found that FQHCs had greater appointment availability but longer wait times than private practices, varying significantly by local contexts.16 Additional secret shopper studies have shown that even within a given city, access may vary based on organization type, practice size, multispecialty status, and participation in insurance programs.11,12,14-16 In this analysis, clinics within large, multi-site systems were associated with lower acceptance rates but not with longer waits. These effects persisted after adjusting for clinician counts, indicating that organizational scale and structure, and not just staffing, are important to understanding access constraints. One interpretation is that panel closure could be a gatekeeping strategy to manage patient inflow or maintain operational efficiency. However, centralized decisions about new patient acceptance may have broad effects on local access, even when individual clinics have appointment capacity.
Other factors could also be at play. For example, Medicare Advantage plans rely on network contracting and integrated care management, and practices in high-penetration markets may face stronger incentives to prioritize patients within aligned systems or contracted networks. As a result, access pathways for traditional Medicare beneficiaries may become more uneven even when overall clinician supply is adequate. Our study was not designed to evaluate the effects of MA penetration, but such trends underscore the importance of monitoring access for traditional Medicare beneficiaries as the program's market share evolves.
Finally, even after accounting for practice-level ownership and structure, we observed persistent differences across cities. In particular, Portland stood out with both markedly lower Medicare acceptance and substantially longer wait times, the latter of which has been confirmed in other audits.19 While the sample is not nationally representative, this finding suggests that market-level dynamics, including local competition, health system dominance, or insurance mix, may be critical. Notably, Portland did not have lower primary care physician supply per capita than the other markets, suggesting that clinician supply alone is unlikely to explain its longer waits. Portland's market structure was also more concentrated compared with other cities, and prior literature has raised concerns that concentrated markets may be less responsive to pressures to reduce wait times or expand access.27,28
Of note, the observed association between organizational scale and access should be interpreted cautiously. Large health systems may expand in markets with existing capacity constraints or strong demand, meaning that consolidation may both respond to and reinforce local access pressures. These findings warrant further investigation into how market power and delivery system configuration influence access for Medicare beneficiaries.
Our findings highlight the need for updated access monitoring and accountability tools. Workforce expansion alone may not resolve access barriers if organization and structure of delivery systems determine how that capacity is deployed. Additional levers may be needed, including more detailed, market-specific access metrics that can help inform both federal and state timely access policies. For example, granular time and distance standards apply to Medicare Advantage, varying by specialty and county type, but currently do not apply to traditional Medicare.29 Routine monitoring of appointment availability, including by payer type, could help identify markets with limited responsiveness to the needs of publicly-insured patients.30 Policymakers might also consider incentivizing transparent reporting of new patient appointment availability, incorporating access measures into payment or contracting decisions, and reassessing the role of local market contexts in shaping primary care access.
Conclusion
Access to primary care for Medicare beneficiaries varies substantially across and within cities. Our findings suggest that practice organization and market structure may play important roles in shaping new patient appointment availability. Strengthening primary care access for Medicare beneficiaries will therefore require continued investments in the primary care workforce, but also deeper attention to how care is organized and how local markets function.
Supplementary Material
Acknowledgments
Dr. J.M.Z. reports funding from NIH, AHRQ, Commonwealth Fund, and NIHCM Foundation on work unrelated to this study. Dr. M.L.B. reports funding from NIH unrelated to this work. No other potential conflicts of interest were reported.
Contributor Information
Tamara Beetham, Center for Gerontology and Healthcare Research, Brown University School of Public Health, Providence, RI 02912, United States.
Trisha Marsh, Division of General Internal Medicine, Oregon Health & Science University, Portland, OR 97239, United States.
Michael L Barnett, Center for Gerontology and Healthcare Research, Brown University School of Public Health, Providence, RI 02912, United States.
Ruby M Aaron, Division of General Internal Medicine, Oregon Health & Science University, Portland, OR 97239, United States.
Emmanuel Greenberg, Department of Medicine, University of California San Francisco, San Francisco, CA 94143, United States.
Alexandra Do, Division of General Internal Medicine, Oregon Health & Science University, Portland, OR 97239, United States.
Jane M Zhu, Center for Gerontology and Healthcare Research, Brown University School of Public Health, Providence, RI 02912, United States.
Supplementary material
Supplementary material is available at Health Affairs Scholar online.
Funding
None declared.
Notes
- 1. Jabbarpour Y, Petterson S, Jetty A, Byun H. The health of US primary care: a baseline scorecard tracking support for high-quality primary care. The Milbank Memorial Fund and The Physicians Foundation. Updated February 2023. Accessed October 25, 2025. Report No. https://www.milbank.org/publications/health-of-us-primary-care-a-baseline-scorecard/iii-access-the-percentage-of-adults-reporting-they-do-not-have-a-usual-source-of-care-is-increasing/
- 2. NEWS: Sanders and Marshall Announce Bipartisan Legislation on Primary Care | The U.S. Senate Committee on Health, Education, Labor & Pensions . 2023. Accessed November 3, 2025. https://www.help.senate.gov/dem/newsroom/press/news-sanders-and-marshall-announce-bipartisan-legislation-on-primary-care
- 3. HPC Research Spotlights Challenges in Primary Care Delivery and Opportunities to Stabilize Provider Workforce | Massachusetts Health Policy Commission . Accessed February 20, 2026. https://masshpc.gov/news/press-release/hpc-research-spotlights-challenges-primary-care-delivery-and-opportunities
- 4. Neprash HT, Chernew ME. Trends in physician exit from fee-for-service Medicare. JAMA Health Forum. 2025;6(7):e252267. 10.1001/jamahealthforum.2025.2267 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Rotenstein LS, He Z, Dziura J, et al. Trends in and predictors of physician attrition from clinical practice across specialties. Ann Intern Med. 2025;178(12):1698–1708. 10.7326/ANNALS-25-00564 [DOI] [PubMed] [Google Scholar]
- 6. Singh Y, Radhakrishnan N, Adler L, Whaley C. Growth of private equity and hospital consolidation in primary care and price implications. JAMA Health Forum. 2025;6(1):e244935. 10.1001/jamahealthforum.2024.4935 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Centers for Medicare & Medicaid Services . Medicare and Medicaid by the numbers. Centers for Medicare & Medicaid Services. Updated July 2025. Accessed December 15, 2025. Report No. https://www.cms.gov/files/document/medicare-and-medicaid-numbers.pdf
- 8. Biennial Health Insurance Survey 2022 | Commonwealth Fund . Accessed November 21, 2025. https://www.commonwealthfund.org/publications/issue-briefs/2022/sep/state-us-health-insurance-2022-biennial-survey
- 9. US Census Bureau . Census.gov. Health Insurance Coverage in the United States: 2023. Accessed November 21, 2025. https://www.census.gov/library/publications/2024/demo/p60-284.html
- 10. McCauley L, Phillips RL, Meisnere M, Robinson SK. Implementing High-Quality Primary Care: Rebuilding the Foundation of Health Care. National Academies Press; 2021. [Google Scholar]
- 11. Polsky D, Richards M, Basseyn S, et al. Appointment availability after increases in Medicaid payments for primary care. N Engl J Med. 2015;372(6):537–545. 10.1056/NEJMsa1413299 [DOI] [PubMed] [Google Scholar]
- 12. Rhodes KV, Kenney GM, Friedman AB, et al. Primary care access for new patients on the eve of health care reform. JAMA Intern Med. 2014;174(6):861–869. 10.1001/jamainternmed.2014.20 [DOI] [PubMed] [Google Scholar]
- 13. Grande D, Zuo JX, Venkat R, et al. Differences in primary care appointment availability and wait times by neighborhood characteristics: a mystery shopper study. J Gen Intern Med. 2018;33(9):1441–1443. 10.1007/s11606-018-4407-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Tipirneni R, Rhodes KV, Hayward RA, et al. Primary care appointment availability and nonphysician providers one year after Medicaid expansion | AJMC. 2025. Accessed November 20, 2025. https://www.ajmc.com/view/primary-care-appointment-availability-and-nonphysician-providers-one-year-after-medicaid-expansion
- 15. Saloner B, Polsky D, Friedman A, Rhodes K. Primary care appointment availability and preventive care utilization: evidence from an audit study. Med Care Res Rev. 2015;72(2):149–167. 10.1177/1077558715569541 [DOI] [PubMed] [Google Scholar]
- 16. Saloner B, Kenney GM, Polsky D, Rhodes K, Wissoker D, Zuckerman S. The availability of new patient appointments for primary care at federally qualified health centers: findings from an audit study. The Urban Institute Health Policy Center. Updated April 7, 2014. Accessed January 6, 2026. https://www.urban.org/research/publication/availability-new-patient-appointments-primary-care-federally-qualified-health-centers-findings-audit-study
- 17. Zhu JM, Charlesworth CJ, Polsky D, McConnell KJ. Phantom networks: discrepancies between reported and realized mental health care access in Oregon Medicaid. Health Aff (Millwood). 2022;41(7):1013–1022. 10.1377/hlthaff.2022.00052 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Beetham T, Saloner B, Wakeman SE, Gaye M, Barnett ML. Access to office-based buprenorphine treatment in areas with high rates of opioid-related mortality: an audit study. Ann Intern Med. 2019;171(1):1–9. 10.7326/M18-3457 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Survey of Physician Appointment Wait Times . AMN healthcare. Accessed October 25, 2025. Report No. https://www.amnhealthcare.com/amn-insights/physician/whitepapers/2025-survey-of-physician-appointment-wait-times/
- 20. IQVIA . OneKey reference data. 2021. Accessed November 10, 2025. https://www.iqvia.com/locations/united-states/solutions/life-sciences/information-solutions/onekey-reference-data
- 21. Candon M, Rhodes K, Polsky D. Acuity-based scheduling in primary care. Med Care. 2018;56(10):818–821. 10.1097/MLR.0000000000000960 [DOI] [PubMed] [Google Scholar]
- 22. Wisniewski JM, Walker B. Association of simulated patient race/ethnicity with scheduling of primary care appointments. JAMA Netw Open. 2020;3(1):e1920010. 10.1001/jamanetworkopen.2019.20010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Feyman Y, Asfaw DA, Griffith KN. Geographic variation in appointment wait times for US military veterans. JAMA Netw Open. 2022;5(8):e2228783. 10.1001/jamanetworkopen.2022.28783 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Health Resources and Services and Administration . Designated health professional shortage areas statistics [Quarterly Summary]. Updated June 2022. Accessed August 24, 2022. Report No. https://data.hrsa.gov/Default/GenerateHPSAQuarterlyReport
- 25. McConnell KJ, Watson K, Choo E, Zhu JM. Geographical variations in emergency department visits for mental health conditions for medicaid beneficiaries. Health Aff (Millwood). 2023;42(2):172–181. 10.1377/hlthaff.2022.00796 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Gotlieb EG, Rhodes KV, Candon MK. Disparities in primary care wait times in medicaid versus commercial insurance. J Am Board Fam Med. 2021;34(3):571–578. 10.3122/jabfm.2021.03.200496 [DOI] [PubMed] [Google Scholar]
- 27. Access, Quality, And Financial Performance Of Rural Hospitals Following Health System Affiliation | Health Affairs . Accessed February 20, 2026. https://www.healthaffairs.org/doi/10.1377/hlthaff.2019.00918
- 28. Desai SM, Padmanabhan P, Chen AZ, Lewis A, Glied SA. Hospital concentration and low-income populations: evidence from New York State Medicaid. J Health Econ. 2023;90:102770. 10.1016/j.jhealeco.2023.102770 [DOI] [PubMed] [Google Scholar]
- 29. Centers for Medicare & Medicaid Services . Contract year 2025 medicare advantage and part D final rule (CMS-4205-F) | CMS. 2024. Accessed December 15, 2025. https://www.cms.gov/newsroom/fact-sheets/contract-year-2025-medicare-advantage-and-part-d-final-rule-cms-4205-f
- 30. Magellan Healthcare . Provider focus—New medicare advantage access-to-care standards. Accessed November 20, 2025. https://www.magellanproviderfocus.com/issues/fall-2023/features/new-medicare-advantage-access-to-care-standards.aspx
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.