Skip to main content
JAMA Network logoLink to JAMA Network
. 2017 Jun 5;177(6):838–845. doi: 10.1001/jamainternmed.2017.0410

Association of Primary Care Practice Location and Ownership With the Provision of Low-Value Care in the United States

John N Mafi 1,2,, Christina C Wee 3, Roger B Davis 3, Bruce E Landon 3,4
PMCID: PMC5540052  NIHMSID: NIHMS903722  PMID: 28395013

Key Points

Question

What is the influence of practice location and ownership on the provision of low-value care?

Findings

In this nationally representative sample of primary care visits, hospital-based outpatient practices used more low-value computed tomography and magnetic resonance imaging, radiographs, and specialty referrals for common conditions than community-based office practices, particularly during hospital-based visits with someone other than the patient’s primary care provider. Hospital-owned community-based practices made more specialty referrals than physician-owned community-based practices but were otherwise similar.

Meaning

Hospital-based practices provided more low-value care than community-based practices, and hospital-owned community-based practices made more specialty referrals than physician-owned community-based practices. These findings raise concerns about the provision of low-value care at hospital-associated primary care practices.

Abstract

Importance

Hospital-employed physicians provide primary care within the hospital or within community-based office practices. Yet, little is understood regarding the influence of hospital location and ownership on the delivery of low-value care.

Objective

To assess the association of hospital location and hospital ownership with the provision of low-value health services.

Design, Setting, and Participants

This study compared low-value service use after primary care visits at hospital-based outpatient practices from January 1, 1997, to December 31, 2011, vs community-based office practices and at hospital-owned vs physician-owned community-based office practices from January 1, 1997, to December 31, 2013. Logistic regression models adjusted for patient and health care professional characteristics and year, and weighted results were used to reflect population estimates. Results were also stratified by symptom acuity and whether a generalist physician (eg, general internist or family practitioner) was the patient’s primary care provider. This study used nationally representative data from the National Ambulatory Medical Care Survey (January 1, 1997, to December 31, 2013) and the National Hospital Ambulatory Medical Care Survey (January 1, 1997, to December 31, 2011) on outpatient visits to generalist physicians. Participants were patients seen with 3 common primary care conditions, namely, upper respiratory tract infection, back pain, and headache.

Main Outcomes and Measures

The use of antibiotics (for upper respiratory tract infection), computed tomography or magnetic resonance imaging (for back pain and headache), radiographs (for upper respiratory tract infection and back pain), and specialty referrals (for all 3 conditions).

Results

This study identified 31 162 visits for upper respiratory tract infection, back pain, and headache, representing an estimated 739 million US primary care visits from 1997 to 2013. Compared with visits with community-based physicians, patients in visits to hospital-based physicians were younger (mean age, 44.5 vs 49.1 years; P < .001) and less frequently saw their primary care provider (52.7% vs 81.9%, P < .001). Although antibiotic use was similar in both settings, hospital-based visits had more orders for computed tomography and magnetic resonance imaging (8.3% vs 6.3%, P = .01), radiographs (12.8% vs 9.9%, P < .001), and specialty referrals (19.0% vs 7.6%, P < .001) than community-based visits. Multivariable adjustment and symptom acuity stratification revealed similar findings. Visits with a generalist other than the patient’s primary care provider were associated with greater provision of low-value care but mainly within hospital-based settings. Practice patterns were similar among hospital-owned vs physician-owned community-based practices with the exception of specialty referrals, which were more frequent in hospital-owned community-based practices.

Conclusions and Relevance

Visits to US hospital-based practices are associated with greater use of low-value computed tomography and magnetic resonance imaging, radiographs, and specialty referrals than visits to community-based practices, and visits to hospital-owned community-based practices had more specialty referrals than visits to physician-owned community-based practices. These findings raise concerns about the provision of low-value care at hospital-associated primary care practices.


This study of a nationally representative sample of primary care visits assesses the association of hospital location and hospital ownership with the provision of low-value health services.

Introduction

Low-value care is defined as patient care that provides minimal average benefit in specific clinical scenarios. Despite decades of widely publicized studies and national clinical guidelines, our health care system still provides large amounts of low-value care, such as advanced diagnostic imaging for low back pain or headache or antibiotics for upper respiratory tract infection (URTI). Evidence also suggests that many of these unnecessary services are increasing over time. Eliminating low-value care could not only substantially lower health care costs but also reduce preventable patient harm, such as exposure to adverse reactions from unnecessary antibiotics or excessive irradiation from diagnostic imaging.

Little is known about the correlates of low-value care in the United States, and such an understanding is essential in determining the most effective interventions to curb the use of such services. Previous work suggests that variation in health care delivery is in part influenced by structural factors, such as practice location or practice ownership. For example, hospital-based outpatient practices (outpatient practices located on hospital campuses) typically have more readily accessible technologies, such as nearby advanced imaging scanners and specialty services compared with community-based office practices. In contrast, community-based office practices that are owned by a hospital may have more salaried physicians with less incentive to overuse health services, compared with physician-owned community-based practices, where compensation may be more related to the volume of services provided. Few studies have evaluated the extent to which practice location and ownership characteristics influence the provision of low-value ambulatory care.

In this context, we used nationally representative data on ambulatory visits to generalist physicians (eg, general internists or family practitioners) to determine whether ambulatory practice location and ownership are associated with the provision of low-value health services for 3 common conditions (URTI, back pain, and headache) seen in the primary care setting.

Methods

We used the most recently available data from the National Ambulatory Medical Care Survey (NAMCS) and the National Hospital Ambulatory Medical Care Survey (NHAMCS) on ambulatory visits to physicians from January 1, 1997, to December 31, 2013, for the NAMCS and from January 1, 1997, to December 31, 2011, for the NHAMCS. The NAMCS is composed of probability samples of outpatient visits to nonfederal office-based physician practices, while the NHAMCS consists of data on visits to nonfederal hospital outpatient departments and emergency departments. Designed in parallel, the NAMCS and NHAMCS share common design, variables, and patient visit weights. They are also both structured to represent ambulatory visits to physicians across the United States. The University of California, Los Angeles Institutional Review Board determined that this study was exempt from review and from participant informed consent.

Data Collection Procedures

The NAMCS and NHAMCS use standardized surveys completed soon after a visit. Both include chief symptom information, symptom duration and context, and 2 other nonprimary reasons for the visit. The NAMCS and NHAMCS also contain up to 3 International Classification of Diseases, Ninth Revision, Clinical Modification diagnostic codes, and both include patient, health care professional, payer, and practice setting characteristics as well as diagnostic tests, treatments, and medications listed during the visit. From 2005 to 2013, the NAMCS and NHAMCS also collected data on 14 common comorbidities (eg, congestive heart failure, chronic obstructive pulmonary disease, and cancer). Missing data are rare (eg, answers to the survey question, “Have you or anyone in this practice seen this patient before?” are missing 2.9% of the time), and multiple imputation techniques are used in these cases.

Eligible Visits

We selected outpatient visits for common conditions seen in the primary care setting, which we defined as patient visits to an internist, general practitioner, or family practitioner physician in the NAMCS and as patient visits to physicians in general medical clinics in the NHAMCS (we also included a small number of visits to nurse practitioners and physician assistants). Based on prior research, we chose the following 3 episodic conditions that are frequently associated with low-value care: (1) URTI (such as acute bronchitis and pharyngitis), (2) back pain, and (3) headache. These conditions are commonly treated by generalists or in general medical clinics and are based on prior research.

We excluded visits with clinical red flags noted on the symptoms or diagnoses addressed at the visit that would indicate more serious or complex presentations. As described previously, such visits include symptoms or diagnoses of cancer for patients seen with back pain or symptoms or diagnoses of chronic obstructive pulmonary disease for patients seen with URTI.

Practice Location and Ownership of Community Practices

Hospital-based practices were represented by the NHAMCS outpatient department data, whereas community-based practices were derived from the NAMCS data, using the private solo or group practice setting variable. We excluded other settings included in the NAMCS, such as health maintenance organizations, federally qualified health centers, mental health centers, non–federal government clinics, family planning clinics, freestanding clinics and urgent centers, faculty practice plans, and other settings because they fell outside of the scope of our primary care–focused study or had insufficient numbers of visits.

The NAMCS also provided information on the owners of the community practice, specifically, whether the practice setting is owned by a physician group or owned by a hospital. We defined hospital-owned practices as community-based practices owned by an academic medical center or other hospital. Physician-owned practices were defined as community-based practices owned by a physician or physician group.

Primary Outcome Measures

Our primary outcomes were broadly accepted measures of low-value care that were also endorsed by the Choosing Wisely campaign for the 3 common conditions of uncomplicated URTI, back pain, and headache. We considered low-value care as the use of antibiotics for patients with URTI, the use of computed tomography (CT) or magnetic resonance imaging (MRI) for patients with back pain or headache, and the use of radiographs for patients with URTI or back pain. As a secondary outcome, we evaluated referrals to another physician for all 3 conditions because they typically should be managed in primary care settings, and specialty referrals could be indicative of further unnecessary downstream use (the referral variable was added to the survey instruments starting in 1999).

Potential Confounders

Our analyses adjusted for potential confounders. They included age (measured as a continuous variable in years), sex, race/ethnicity (defined as a single 4-level variable of white, black, Hispanic, and other), modified Charlson comorbidity disease count, insurance status, urban vs rural location, geographic region (1 of 4 quadrants in the United States), and year.

We also specifically evaluated whether practice location was confounded by symptom acuity, patient comorbidities, and whether the generalist was the patient’s primary care provider (PCP) (in this study, provider refers to physicians, nurse practitioners, and physician assistants). For all chief symptom items, the NAMCS and NHAMCS symptom duration or context variable divides into the following 5 categories: (1) new-onset symptoms (<3 months’ duration), (2) acute on chronic flare-up (acute exacerbation of a chronic condition), (3) chronic routine, (4) routine or preventive care, and (5) preoperative or postoperative visit. We defined acute visits by combining the categories of new-onset symptoms with acute on chronic flare-up. For additional case-mix assessment beyond the Charlson comorbidity disease count, we also evaluated the 14 comorbidities (eg, congestive heart failure, chronic obstructive pulmonary disease, and cancer, only available from 2005 onward) directly collected by the survey (collected from the problem list or medical record) from 2005 to 2011.

Statistical Analysis

Our primary analyses focused on examining the following 2 associations: (1) the association between practice location (hospital based vs community based) and low-value service use and (2) the association between hospital ownership of community-based practices and low-value service use. We used the combined data from the NAMCS and NHAMCS through 2011 to examine the first association and used only the NAMCS data through 2013 to test the second association.

We pooled all years in each outcome category and used bivariable logistic regression models to compare low-value service use between hospital-based practices vs all community-based practices (NHAMCS vs NAMCS, January 1, 1997, to December 31, 2011) and between hospital-owned community-based practices vs physician-owned community-based practices (NAMCS vs NAMCS, January 1, 1997 to December 31, 2013). In our multivariable-adjusted analyses, we adjusted for the previously mentioned demographic, clinical, and health care professional variables. We also controlled for year.

Because our first study question might be particularly prone to patient selection bias (eg, patients seen at hospital-based practices may be more complex than those seen at community-based practices), we performed 4 prespecified sensitivity analyses in which we evaluated the following important potential effect modifiers: (1) stratification of results by whether the patient saw his or her usual PCP or another generalist (and interacting the PCP variable with the practice location variable), (2) stratification by acute vs nonacute presentations, (3) comparison of the proportion of patients who had any of the 14 comorbidities between hospital-based vs community-based visits (and a subanalysis of adding patient comorbidity counts to the models from 2005 to 2011), and (4) evaluation of trends in use over time among hospital-based practices vs community-based practices.

In 2007, the NHAMCS survey changed the question, “Are you the patient’s primary care physician/provider?” to “Is this clinic the patient’s primary care provider?” This alteration could affect the PCP vs non-PCP analysis. Therefore, we performed an additional sensitivity analysis evaluating this stratification before and after the survey wording change (eTable 1 and eTable 2 in the Supplement), which revealed consistent results. In addition, to assess whether the PCP variable also was reflecting continuity of care, we performed a sensitivity analysis of only established clinic patient visits by using a second variable that asked, “Have you or anyone in this practice seen this patient before?” The rationale is that established patients seeing their PCP would better reflect continuity of care.

We performed all of our analyses with a software program (SAS-callable SUDAAN, version 11.0; RTI International) using SUDAAN subpopulation procedures. These functions use data from the entire NAMCS and NHAMCS sample to account for the complex multistage survey design and population weighting to produce national estimates.

Results

This study identified 31 162 visits for URTI (n = 11 049), back pain (n = 14 821), and headache (n = 5695), representing an estimated 739 million US primary care visits from 1997 to 2013 (during a small percentage of visits, more than one of the study conditions was addressed). Compared with community-based visits, patients seen at hospital-based practices generally were younger (mean age, 44.5 vs 49.1 years; P < .001), and the hospital-based health care professionals were less likely to be the patient’s usual PCP (52.7% vs 81.9%, P < .001) (Table 1). Patients seen at hospital-owned and physician-owned community practices were more similar demographically, although physician-owned community-based practices had a greater proportion of visits of patients of nonwhite race/ethnicity than hospital-owned community-based practices.

Table 1. Characteristics of All Community-Based Office Practices vs Hospital-Based Practices and of Physician-Owned vs Hospital-Owned Community-Based Office Practicesa.

Variable 1997-2011 1997-2013
All Community-Based Office Practices
(n = 9783) [Reference]
Hospital-Based Practices
(n = 16 306)
P Value Physician-Owned Community-Based Office Practices
(n = 10 395)
Hospital-Owned Community-Based Office Practices
(n = 1038)
P Value
Age, mean, y 49.1 44.5 <.001 49.7 49.6 .88
Female, % 62.7 65.0 .01 62.5 62.8 .85
Race/ethnicity, %
White 77.8 72.5 .001 76.5 85.4 <.001
Black 10.2 15.1 10.7 8.7
Hispanic 7.6 9.2 8.2 3.2
Other 4.4 3.1 4.6 2.7
Charlson comorbidity disease count, mean 0.048 0.043 .39 0.049 0.041 .27
Acute symptoms, % 72.2 77.8 .001 71.2 77.1 .02
Insurance status, % .68
Private 59.2 45.2 <.001 58.1 60.3 .08
Medicare or Medicaid 26.4 33.3 27.7 28.7
Other 14.4 21.6 14.2 10.9
Clinician identifies as PCP, % 81.9 52.7 <.001 82.7 82.1 .79
Urban location, % 81.7 74.9 .05 82.4 79.4 .40
Geographic region, %
Northeast 20.3 17.6 .01 21.3 16.5 <.001
Midwest 24.1 34.7 21.4 38.0
South 37.6 32.8 38.3 37.3
West 18.1 14.9 19.1 8.2
Problem list items, %b
URTI 36.0 38.1 .20 35.3 37.8 .31
Back pain 47.1 45.0 .14 48.4 45.9 .23
Headache 18.2 18.1 .94 17.7 17.2 .79

Abbreviations: PCP, primary care provider; URTI, upper respiratory tract infection.

a

Hospital-based outpatient practice data are only available from 1997 to 2011.

b

Symptom group proportions may not sum to 100% because a small number of patients fit into more than 1 category.

Practice Location

Antibiotic use was similar in visits to both locations. However, visits to hospital-based practices had higher use of CT or MRI (8.3% vs 6.3%, P = .01), radiographs (12.8% vs 9.9%, P < .001), and specialty referrals (19.0% vs 7.6%, P < .001) compared with community-based visits (Table 2). Multivariable adjustment revealed findings similar to the unadjusted results.

Table 2. Unadjusted and Adjusted Use Rates by Stratified Practice Setting, 1997-2011.

Variable Unadjusteda Adjustedb
Community-Based Practice
(n = 9783) [Reference]
Hospital-Based Practices
(n = 16 306)
P Value Community-Based Practice
(n = 9783) [Reference]
Hospital-Based Practices (n = 16 306) aOR (95% CI)
Antibiotics for URTI 1684/3389 (49.1) 2898/6023 (52.1) .16 49.4 49.2 0.99 (0.82-1.21)
CT or MRI for back pain or headachec 404/6443 (6.3) 726/10 346 (8.3) .01 5.8 8.2 1.44 (1.13-1.85)
Radiographs for back painc 775/8123 (9.9) 1689/13 451 (12.8) <.001 9.5 12.9 1.41 (1.16-1.71)
Specialty referrals for all 3 conditionsc 740/9783 (7.6) 3128/16 306 (19.0) <.001 7.1 17.3 2.74 (2.23-3.36)

Abbreviations: aOR, adjusted odds ratio; CT, computed tomography; MRI, magnetic resonance imaging; URTI, upper respiratory tract infection.

a

Presented as unadjusted sample numerator divided by sample denominator (percentage). Percentages may not equal the value of the numerator divided by the denominator because they are population weighted.

b

Presented as multivariable-adjusted use rates and aOR. Models are adjusted for age, sex, race/ethnicity, insurance status, symptom acuity, comorbidities, geographic region, urban location, and whether the generalist was the patient’s usual primary care provider.

c

Comparisons with P < .05.

Practice Location–Stratified and Sensitivity Analyses

Regarding our first study question of comparing low-value service use at hospital-based vs all community-based practices, we evaluated whether important effect modifiers could explain differences in practice patterns. Stratification by PCP vs non-PCP visits revealed that non-PCP visits were associated with higher use of imaging and specialty referrals but primarily within hospital-based practices (Table 3). For instance, at hospital-based practice visits with non-PCPs, radiographs were ordered 13.7% of the time vs 8.5% of the time among community-based non-PCP visits (adjusted odds ratio, 1.70; 95% CI, 1.21-2.39; P = .08 for interaction), and specialty referrals were ordered 26.1% of the time vs 5.5% of the time among community-based non-PCP visits (adjusted odds ratio, 6.10; 95% CI, 4.42-8.42; P < .001 for interaction). In contrast, stratification by symptom acuity revealed similar findings in both subgroups (Table 4).

Table 3. Multivariable-Adjusted PCP vs Non-PCP Visit Use Rates by Practice Setting, 1997-2011a.

Variable Community-Based Practice [Reference] Hospital-Based Practice aOR (95% CI) P Value for Interaction
PCP, No. 7913 7957 NA NA
Non-PCP, No. 1870 8349 NA NA
Antibiotics, %
PCP 49.0 45.1 0.85 (0.70-1.04) .06b
Non-PCP 50.2 55.3 1.23 (0.92-1.64)
CT or MRI, %
PCP 6.2 7.2 1.18 (0.89-1.58) .07c
Non-PCP 4.1 8.8 2.27 (1.49-3.44)
Radiographs, %
PCP 9.7 11.4 1.20 (0.98-1.48) .08c
Non-PCP 8.5 13.7 1.70 (1.21-2.39)
Specialty referrals, %
PCP 7.4 9.3 1.29 (1.05-1.60) <.001c
Non-PCP 5.5 26.1 6.10 (4.42-8.42)

Abbreviations: aOR, adjusted odds ratio; CT, computed tomography; MRI, magnetic resonance imaging; NA, not applicable; PCP, primary care provider.

a

Models are adjusted for age, sex, race/ethnicity, insurance status, symptom acuity, comorbidities, geographic region, and urban location.

b

For interaction between PCP vs non-PCP and hospital-based practice vs community-based practice setting.

c

For interaction between PCP vs non-PCP and hospital-based outpatient practice vs community-based office practice setting.

Table 4. Multivariable-Adjusted Acute vs Nonacute Visit Use Rates by Practice Setting, 1997-2011a.

Variable Community-Based Practice [Reference] Hospital-Based Practice aOR (95% CI)
Acute, No. 7101 12 428 NA
Nonacute, No. 2682 3878 NA
Antibiotics, %
Acute 51.5 51.4 0.99 (0.82-1.21)
Nonacute 34.2 27.0 0.71 (0.41-1.24)
CT or MRI, %
Acute 6.9 8.6 1.27 (1.02-1.58)
Nonacute 3.8 7.4 2.03 (1.25-3.31)
Radiographs, %
Acute 10.5 13.1 1.28 (1.05-1.57)
Nonacute 6.4 12.5 2.09 (1.46-3.00)
Specialty referrals, %
Acute 7.1 17.9 2.85 (2.29-3.54)
Nonacute 6.9 13.4 2.10 (1.49-2.96)

Abbreviations: aOR, adjusted odds ratio; CT, computed tomography; MRI, magnetic resonance imaging; NA, not applicable.

a

Models are adjusted for age, sex, race/ethnicity, insurance status, comorbidities, geographic region, urban location, and whether the generalist was the patient’s usual primary care provider.

In our sensitivity analysis of only established clinic patients, we found results similar to our main findings among non-PCP and PCP visits, supporting the notion that differences in low-value care were associated with discontinuity of care within the hospital setting (eTable 3 in the Supplement). In our comorbidity sensitivity analysis, which was restricted to 2005 onward when these data were available (eTable 4 in the Supplement), we found that patients seen at community-based practices had more comorbidities than those seen at hospital-based practices (mean, 1.25 vs 1.09 comorbidities per patient, respectively; P = .03). When we added comorbidity counts to our adjusted models, results remained unchanged. Finally, use trends revealed that hospital-based practices consistently provided more low-value CT or MRI, radiographs, and specialty referrals than community-based practices over time (eTable 5 in the Supplement).

Community Practice Ownership

We found no significant differences between visits to hospital-owned community-based practices and physician-owned community-based practices in terms of the use of antibiotics, CT and MRI, or radiographs. However, we observed that visits to hospital-owned community-based practices were associated with more specialty referrals, although the magnitude of the odds ratio was much smaller than was seen for hospital-based practices. These results are summarized in Table 5.

Table 5. Visit Use Rates in the Community-Based Office Practice Setting by Ownership Arrangement, 1997-2013.

Variable Unadjusteda Adjustedb
Physician-Owned Community-Based Practice [Reference]
(n = 10 395)
Hospital-Owned Community-Based Practice
(n = 1038)
P Value Physician-Owned Community-Based Practice [Reference]
(n = 10 395)
Hospital-Owned Community-Based Practice
(n = 1038)
aOR (95% CI)
Antibiotics for URTI 1704/3488 (48.3) 194/360 (54.3) .23 48.3 53.7 1.24 (0.84-1.84)
CT or MRI for back pain or headache 420/6968 (6.2) 51/681 (6.1) .98 5.8 5.3 0.92 (0.58-1.45)
Radiographs for back pain 801/8656 (9.6) 97/872 (10.2) .60 9.2 9.5 1.04 (0.79-1.37)
Specialty referrals for all 3 conditions 831/10 395 (7.9) 112/1038 (9.8) .07 7.2 9.8 1.40 (1.09-1.82)

Abbreviations: aOR, adjusted odds ratio; CT, computed tomography; MRI, magnetic resonance imaging; URTI, upper respiratory tract infection.

a

Presented as unadjusted numerator divided by denominator (percentage). Percentages may not equal the value of the numerator divided by the denominator because they are population weighted.

b

Presented as multivariable-adjusted use rates and aOR. Models are adjusted for age, sex, race/ethnicity, insurance status, symptom acuity, comorbidities, geographic region, urban location, and whether the generalist was the patient’s usual primary care provider.

Discussion

In this large and nationally representative sample of ambulatory care, we find that visits to hospital-based practices are associated with higher use of 3 forms of low-value health care services compared with visits to community-based practices, specifically, greater use of low-value CT and MRI and radiographs and more low-value referrals to specialists. Moreover, in hospital-based practices, low-value care was associated with visits in which the health care professional was not the patient’s PCP. These findings suggest that practice location has a larger role than ownership or incentive factors in low-value care delivery and that the association of these structural attributes also appears to be related to visits to physicians who are not the patient’s PCP. We also found that hospital-owned community-based practices were associated with more low-value specialty referrals than physician-owned community-based practices, although the magnitude of the difference was much smaller than was seen in hospital-based practices. Overall, these findings raise concerns about the provision of low-value care at hospital-associated primary care practices.

Few prior studies directly compare the quality and efficiency of care between hospital-based practices and community-based practices. A comprehensive analysis using Medicare data from the 1980s found no meaningful differences in quality and costs of care when comparing the 2 settings. Another national analysis using the NAMCS and NHAMCS data from the 1990s found that clinicians at hospital-based practices ordered more imaging, minor surgical procedures, and specialty referrals compared with community-based clinicians, which is generally consistent with our findings. However, the data from the latter study are almost 20 years old, and the use measures were not calibrated to determine the quality and efficiency of care because the analysis did not exclude important guideline-based clinical red flags or account for symptom acuity. Moreover, neither of these prior studies specifically compared differences in the use of low-value care services. Therefore, our work adds recent data to the literature and specifically highlights the greater use of costly and low-value services at hospital-based practices compared with community-based practices.

We found that community-based practices that are owned by hospitals use similar amounts of low-value antibiotics, CT and MRI, and radiographs but made more specialty referrals compared with physician-owned community-based practices. It is possible that hospital ownership of a practice leads to easier access to specialists or a broader network of specialty services than physician ownership. Prior studies have determined that more integrated health care systems deliver more costly care, without any benefit in quality or efficiency of care (and by some measures worse quality of care). However, at least in part, these results might be related to higher reimbursement for hospital-owned practices, which we did not assess in this study. Some commentators have raised concerns that such arrangements could result in higher prices and higher rates of overuse. Nevertheless, our study findings suggest that hospital location is more important than hospital ownership in terms of the delivery of low-value care, although hospital-owned community-based practices also more frequently used specialty referrals.

Higher use of low-value services within hospital-based practices may be related to greater availability of diagnostic imaging and specialty services within hospitals. This notion is supported by the fact that our results revealed differences primarily with the use of CT and MRI, radiographs, and specialty referrals but not antibiotics, which would not be influenced by hospital location. We also found that the health care professional was identified as the patient’s PCP in only half of the visits in hospital-based practices vs in more than 80% of visits to community-based generalists. These non-PCP visits are associated with greater use of low-value services but primarily within hospital-based settings. Such findings suggest that the combination of hospital-based settings and discontinuity of care may be related to greater use of low-value health services. Prior studies found that greater continuity of care with a PCP is associated with higher-value care in terms of increased use of preventive care, reduced costs, and lower rates of emergency department visits, hospitalizations, and all-cause mortality.

Limitations

Our study has some limitations. The measures used herein do not represent all low-value care, and we cannot definitively conclude that our outcomes are low value in every instance because we lacked the detailed clinical data required to identify the rare instances when the service might have been of higher value. However, the measures we studied have been widely identified as being low value in most circumstances, and we used rigorous methods to exclude cases with any symptoms or diagnoses of clinical red flags, which were created from evidence-based guidelines and prior research. Moreover, our analysis excluded patients with the most important markers of complexity relevant to our outcomes, such as any diagnosis or symptom of trauma or neurological deficit among patients with back pain. We also cannot eliminate the possibility of selection bias because patients seen at hospital-based practices may be more complex or may have different preferences than patients seen at community-based practices. However, neither adjustment nor stratification by symptom acuity altered our findings. In addition, our analysis revealed that community-based practice patients were older than hospital-based patients, and our comorbidity sensitivity analysis (using medical record–verified comorbidities) found that community-based patients had more comorbidities than hospital-based patients. Because the NAMCS and NHAMCS share common survey designs, patient weights, and variables (including clinical diagnoses, imaging studies, specialty referrals, etc), measurement error would be unlikely to differ by practice setting. Finally, we could not control for physician age, and it is possible that younger physicians were in part driving the greater use of low-value services within hospital-based practices.

Conclusions

In this nationally representative sample, we found that hospital-based practices were associated with more low-value care compared with community-based practices and that low-value care is associated with hospital-based practice visits in which the health care professional was not the patient's usual PCP. In contrast, hospital-owned practices were not associated with greater use of some low-value services compared with physician-owned practices in the community-based setting, but they used more specialty referrals than physician-owned community-based practices. These findings suggest that practice location plays a greater role than practice ownership in the delivery of low-value care and that low-value care may be related to greater discontinuity of care within hospital-based settings. Ultimately, these results raise general concerns about the provision of low-value care at hospital-associated primary care practices. Because almost one-third of health spending is considered potentially wasteful, our findings have important implications for policymakers, health care practice leaders, and clinicians, who have an interest in providing the highest-quality care at the lowest per capita cost.

Supplement.

eTable 1. Multivariable Adjusted* PCP vs non-PCP Visit Utilization Rates by Practice Setting Before the Survey Wording Change, 1997-2006 (%)

eTable 2. Multivariable Adjusted* PCP vs non-PCP Visit Utilization Rates by Practice Setting After the Survey Wording Change, 2007-2011 (%)

eTable 3. Multivariable Adjusted* PCP vs non-PCP Visit Utilization Rates by Practice Setting Among Established Patients Only, 1997-2011 (%)

eTable 4. Comorbidities* Stratified by Practice Location, 2005-2011 (%)

eTable 5. Unadjusted Utilization Time-Trends* Among Hospital-Based Practices vs Community-Based Practices, 1997-2011

References

  • 1.Mafi JN, Wee CC, Davis RB, Landon BE. Comparing use of low-value health care services among U.S. advanced practice clinicians and physicians. Ann Intern Med. 2016;165(4):237-244. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Fisher ES, Wennberg DE, Stukel TA, Gottlieb DJ, Lucas FL, Pinder EL. The implications of regional variations in Medicare spending, part 1: the content, quality, and accessibility of care. Ann Intern Med. 2003;138(4):273-287. [DOI] [PubMed] [Google Scholar]
  • 3.Barnett ML, Linder JA. Antibiotic prescribing for adults with acute bronchitis in the United States, 1996-2010. JAMA. 2014;311(19):2020-2022. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Schwartz AL, Landon BE, Elshaug AG, Chernew ME, McWilliams JM. Measuring low-value care in Medicare. JAMA Intern Med. 2014;174(7):1067-1076. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Colla CH, Morden NE, Sequist TD, Schpero WL, Rosenthal MB. Choosing Wisely: prevalence and correlates of low-value health care services in the United States. J Gen Intern Med. 2015;30(2):221-228. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Mafi JN, McCarthy EP, Davis RB, Landon BE. Worsening trends in the management and treatment of back pain. JAMA Intern Med. 2013;173(17):1573-1581. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Mafi JN, Edwards ST, Pedersen NP, Davis RB, McCarthy EP, Landon BE. Trends in the ambulatory management of headache: analysis of NAMCS and NHAMCS data 1999-2010. J Gen Intern Med. 2015;30(5):548-555. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Welch HG, Schwartz L, Woloshin S. Overdiagnosed: Making People Sick in the Pursuit of Health. Boston, Massachusetts: Beacon Press; 2011. [Google Scholar]
  • 9.Tyler PD, Larochelle MR, Mafi JN. Access to prescription opioids—primum non nocere: a teachable moment. JAMA Intern Med. 2016;176(9):1251-1252. [DOI] [PubMed] [Google Scholar]
  • 10.Donabedian A. The quality of care: how can it be assessed? JAMA. 1988;260(12):1743-1748. [DOI] [PubMed] [Google Scholar]
  • 11.Neprash HT, Chernew ME, Hicks AL, Gibson T, McWilliams JM. Association of financial integration between physicians and hospitals with commercial health care prices. JAMA Intern Med. 2015;175(12):1932-1939. [DOI] [PubMed] [Google Scholar]
  • 12.Kocher R, Sahni NR. Hospitals’ race to employ physicians: the logic behind a money-losing proposition. N Engl J Med. 2011;364(19):1790-1793. [DOI] [PubMed] [Google Scholar]
  • 13.Isaacs SL, Jellinek PS, Ray WL. The independent physician: going, going.... N Engl J Med. 2009;360(7):655-657. [DOI] [PubMed] [Google Scholar]
  • 14.Baker LC, Atlas SW, Afendulis CC. Expanded use of imaging technology and the challenge of measuring value. Health Aff (Millwood). 2008;27(6):1467-1478. [DOI] [PubMed] [Google Scholar]
  • 15.Forrest CB, Whelan EM. Primary care safety-net delivery sites in the United States: a comparison of community health centers, hospital outpatient departments, and physicians’ offices. JAMA. 2000;284(16):2077-2083. [DOI] [PubMed] [Google Scholar]
  • 16.McCaig LF, McLemore T. Plan and operation of the National Hospital Ambulatory Medical Survey, series 1: programs and collection procedures. Vital Health Stat 1. 1994;(34):1-78. [PubMed] [Google Scholar]
  • 17.National Center for Health Statistics. Ambulatory health care data. https://www.cdc.gov/nchs/ahcd/ahcd_questionnaires.htm. Updated October 31, 2016. Accessed January 24, 2017.
  • 18.Barnett ML, Linder JA. Antibiotic prescribing to adults with sore throat in the United States, 1997-2010. JAMA Intern Med. 2014;174(1):138-140. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Berenson RA, Ginsburg PB, May JH. Hospital-physicians relations: cooperation, competition, or separation? Health Aff (Millwood). 2007;26(1):w31-w43. [DOI] [PubMed] [Google Scholar]
  • 20.Cassel CK, Guest JA. Choosing Wisely: helping physicians and patients make smart decisions about their care. JAMA. 2012;307(17):1801-1802. [DOI] [PubMed] [Google Scholar]
  • 21.Peek ME, Tang H, Alexander GC, Chin MH. National prevalence of lifestyle counseling or referral among African-Americans and whites with diabetes. J Gen Intern Med. 2008;23(11):1858-1864. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.National Center for Health Statistics. Using ultimate cluster models with NAMCS and NHAMCS public use files. https://www.cdc.gov/nchs/data/ahcd/ultimatecluster.pdf. Published 2004. Accessed January 24, 2017.
  • 23.Starfield B, Powe NR, Weiner JR, et al. . Costs vs quality in different types of primary care settings. JAMA. 1994;272(24):1903-1908. [PubMed] [Google Scholar]
  • 24.Casalino LP, Pesko MF, Ryan AM, et al. . Small primary care physician practices have low rates of preventable hospital admissions. Health Aff (Millwood). 2014;33(9):1680-1688. [DOI] [PubMed] [Google Scholar]
  • 25.McWilliams JM, Chernew ME, Zaslavsky AM, Hamed P, Landon BE. Delivery system integration and health care spending and quality for Medicare beneficiaries. JAMA Intern Med. 2013;173(15):1447-1456. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Gill JM, Mainous AG III, Nsereko M. The effect of continuity of care on emergency department use. Arch Fam Med. 2000;9(4):333-338. [DOI] [PubMed] [Google Scholar]
  • 27.Dreiher J, Comaneshter DS, Rosenbluth Y, Battat E, Bitterman H, Cohen AD. The association between continuity of care in the community and health outcomes: a population-based study. Isr J Health Policy Res. 2012;1(1):21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Saultz JW, Lochner J. Interpersonal continuity of care and care outcomes: a critical review. Ann Fam Med. 2005;3(2):159-166. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Amjad H, Carmichael D, Austin AM, Chang CH, Bynum JP. 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] [PMC free article] [PubMed] [Google Scholar]
  • 30.Wasson JH, Sauvigne AE, Mogielnicki RP, et al. . Continuity of outpatient medical care in elderly men: a randomized trial. JAMA. 1984;252(17):2413-2417. [PubMed] [Google Scholar]
  • 31.National Committee for Quality Assurance. HEDIS 2014 technical specifications for physician measurement. http://www.ncqa.org/Portals/0/HEDISQM/HEDIS2014/HEDIS_2014%20_List_of_Physician_Measures.pdf. Published 2014. Accessed January 24, 2017.
  • 32.Chou R, Qaseem A, Snow V, et al. ; Clinical Efficacy Assessment Subcommittee of the American College of Physicians; American College of Physicians; American Pain Society Low Back Pain Guidelines Panel . Diagnosis and treatment of low back pain: a joint clinical practice guideline from the American College of Physicians and the American Pain Society. Ann Intern Med. 2007;147(7):478-491. [DOI] [PubMed] [Google Scholar]
  • 33.Gonzales R, Bartlett JG, Besser RE, et al. ; American Academy of Family Physicians; American College of Physicians–American Society of Internal Medicine; Centers for Disease Control and Prevention; Infectious Diseases Society of America . Principles of appropriate antibiotic use for treatment of uncomplicated acute bronchitis: background. Ann Intern Med. 2001;134(6):521-529. [DOI] [PubMed] [Google Scholar]
  • 34.Morey SS. Headache Consortium releases guidelines for use of CT or MRI in migraine work-up. Am Fam Physician. 2000;62(7):1699-1701. [PubMed] [Google Scholar]
  • 35.Sistrom C, McKay NL, Weilburg JB, Atlas SJ, Ferris TG. Determinants of diagnostic imaging utilization in primary care. Am J Manag Care. 2012;18(4):e135-e144. [PubMed] [Google Scholar]
  • 36.Mehrotra A, Reid RO, Adams JL, Friedberg MW, McGlynn EA, Hussey PS. Physicians with the least experience have higher cost profiles than do physicians with the most experience. Health Aff (Millwood). 2012;31(11):2453-2463. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement.

eTable 1. Multivariable Adjusted* PCP vs non-PCP Visit Utilization Rates by Practice Setting Before the Survey Wording Change, 1997-2006 (%)

eTable 2. Multivariable Adjusted* PCP vs non-PCP Visit Utilization Rates by Practice Setting After the Survey Wording Change, 2007-2011 (%)

eTable 3. Multivariable Adjusted* PCP vs non-PCP Visit Utilization Rates by Practice Setting Among Established Patients Only, 1997-2011 (%)

eTable 4. Comorbidities* Stratified by Practice Location, 2005-2011 (%)

eTable 5. Unadjusted Utilization Time-Trends* Among Hospital-Based Practices vs Community-Based Practices, 1997-2011


Articles from JAMA Internal Medicine are provided here courtesy of American Medical Association

RESOURCES