Physicians in Retainer (“Concierge”) Practice

Abstract

Background

Retainer practices represent a new model of care whereby physicians charge an up-front fee for services that may not be covered by health insurance. The characteristics of these practices are largely unknown.

Design, Setting, and Participants

We conducted a cross-sectional mail survey of 144 retainer physicians (58% response rate) and a national random sample of 463 nonretainer physicians (50% response rate) to compare retainer and nonretainer practices. Outcomes of interest included physician demographics, size and case-mix of patient panel, services offered and, for retainer practices, characteristics of practice development.

Results

Retainer physicians have much smaller patient panels (mean 898 vs 2303 patients, P<.0001) than their nonretainer counterparts, and care for fewer African-American (mean 7% vs 16%, P<.002), Hispanic (4% vs 14%, P<.001), or Medicaid (5% vs 15%, P<.001) patients. Physicians in retainer practices are more likely to offer accompanied specialist visits (30% vs 1%), house calls (63% vs 26%), 24-hour direct physician access (91% vs 40%), and several other services (all P values <.05). Most retainer physicians (85%) converted from nonretainer practices but kept few of their former patients (mean 12%). Most retainer physicians (84%) provide charity care and many continue to see some patients (mean 17%) who do not pay retainer fees.

Conclusions

Despite differences between retainer and nonretainer practices, there is also substantial overlap in services provided. These findings, in conjunction with the scope of patient discontinuity when physicians transition to retainer practice, suggest that ethical and legal debates about the standing of these practices will endure.

Retainer practices, also known as “concierge,”“luxury,” or “boutique” practices, have been argued to represent the best and worst of the U.S. health care system. These practices first arose in the mid-1990s and are distinguished by a supplemental fee that patients pay their physicians for enhanced access and certain services (e.g., accompanied visits to specialists) that might not be offered in traditional primary care practices. Largely because they can offer longer and more frequent visits, retainer practices have been heralded for enabling greater patient-centered care and enhancing both patient and physician satisfaction.^1–4

On the other hand, concerns about these practices have arisen around a number of issues. Ethical concerns, articulated in statements by the American Medical Association Council on Ethical and Judicial Affairs ² and elsewhere,^3,5,6 include a possible decrease in physicians' provision of charity care, the risk of patient abandonment as physicians convert from traditional to retainer practice, and exacerbation of existing health care inequities based on rationing by ability to pay. Retainer practices have also been challenged on legal grounds that retainer fees constitute duplicate billing if physicians also accept insurance payments for clinical services provided to the same patients.⁷ Finally, some have argued that these practices may offer inappropriate services and contribute to health care overuse.⁵

Despite intense debate about retainer practices, no systematic studies of the physicians or patients who join them have been published. This leaves many open questions. How much do the services offered by retainer and nonretainer physicians overlap? How many patients are displaced when their physician converts to a retainer practice? How frequently do retainer physicians provide charity care or see patients who do not pay the retainer fee? The answers to these questions are important, both to inform debates over the ethical standing of this new model of health care delivery and to help judge their policy implications. To investigate these areas, we conducted a national survey of retainer and nonretainer physicians.

METHODS

Participants and Protocol

To identify retainer physicians, we performed a comprehensive search using a general Internet search engine (Google^®), a specialized business search engine (bizjournals.com^®), and a search of print media (LexisNexis^®). We conducted these searches from April to September 2003 using the terms ([boutique, concierge, OR retainer] AND [physician, medical care, medicine, or practice]). We supplemented our findings by using public phone directories. We then used snowball sampling of this initial sample of 91 physicians to identify 61 additional retainer physicians. Each practice was called to ascertain new retainer physicians, until every physician's office had been contacted and no new names were given. For comparison, we selected a random sample of 1,200 nonretainer physicians from the American Medical Association Masterfile of all licensed physicians in the U.S. For this study, we limited our comparison group to primary care physicians (general internists, family practitioners, and general practitioners) and internal medicine specialists because most retainer physicians are in these fields. We excluded physicians who were retired, in training, or without a forwarding address.

A confidential survey and cover letter was mailed to all physicians during the fall of 2003. A second wave of surveys was sent approximately 2 weeks after the first, followed by a postcard reminder after another 2 weeks. The third survey wave, sent in November 2003, included a $2 cash incentive, and a fourth wave, mailed in January 2004, included a laser pointer (cash value $2.50). The Institutional Review Board of the University of Chicago approved the study protocol.

Survey

We designed a survey instrument to explore patient (demographics and case-mix), physician (age, sex, years in practice, amount of charity care performed), and practice (size of patient panel, number of patients seen per day, services offered, and frequency of use) characteristics of retainer and nonretainer physicians. We defined retainer practices as practices in which “patients pay an up-front fee to join the practice and receive in return special services and amenities.” We were particularly interested in the use of 6 services (accompanied specialist visits, house calls, 24-hour physician access, same-day appointments, coordinated hospital care, and private waiting rooms) that news reports indicated retainer practices offered.⁸ Retainer physicians were also asked about their length of time in retainer practice, whether their retainer practice was de novo or a conversion from a nonretainer practice and, if the latter, the proportion of their former patients who joined. Finally, for converted practices, we assessed physicians' involvement in helping patients who did not join the retainer practice to find a new physician. Survey items were pretested among both retainer and nonretainer physicians to examine psychometric performance and to ensure construct validity.

Analysis

First, we used descriptive statistics to examine the distribution of each survey variable. Second, we stratified our sample into retainer and nonretainer physicians. We used t tests for continuous variables and Pearson χ²test for categorical variables to explore the bivariate associations between practice type and each variable of interest. Because our total sample of 152 retainer physicians were clustered into 70 practices, we considered the potential impact of clustering on our analyses by performing logistic regression with a random-effects model. Third, we examined items assessing patient case-mix (e.g., proportion of patients on Medicaid) as both ordinal and continuous variables. To derive population means for case-mix variables with ordinal response frames (<5%, 5% to 15%, 16% to 25%, 26% to 50%, and >50%), we used the midpoint of each response option to impute a score for each respondent. In order to evaluate whether differences between retainer and nonretainer services and case-mix might be caused by population-level differences in the areas where these 2 types of practices are located, we compared our baseline models with those that adjusted for the average zip code level income derived from census data of the area where the physician practice was located. Overall, this adjustment did not alter either the direction or magnitude of the differences that we describe.

RESULTS

Respondent Characteristics

Of 1,200 nonretainer physicians initially selected from the Masterfile, 668 were not in primary care, 36 were unreachable by mail, and 33 were ineligible because of retiree or trainee status. Of the remaining 463, 231 (50%) returned completed surveys. Of the 152 retainer physicians, 8 were ineligible because of missing or incomplete addresses. Of the remaining 144 physicians, 83 (58%) returned completed surveys. There was no association between response time to the survey and a global measure of support for retainer practices (r=−.04, P=.56), suggesting the absence of response wave bias. Compared with respondents, nonrespondents did not differ significantly based on age (50 vs 49 years, P=.10), sex (75% vs 72% male, P=.55), or specialty (34% vs 32% internal medicine specialists, P=.77).

Retainer and Nonretainer Physician Demographics

There was a wide geographic distribution of retainer and nonretainer physicians in our sample, with prominent clustering of retainer physicians in large urban areas on the east and west coasts. Most retainer practices were relatively new.Table 1) describes the basic characteristics of respondents stratified by type of practice. Retainer physicians did not differ from nonretainer physicians with regard to mean age (48 vs 49 years, P=.65), sex (73% male vs 72% male, P=.88), or years in practice (16.7 vs 18.4, P=.21). Most retainer physicians practice general internal medicine (62%) or family practice (28%), with fewer in internal medicine subspecialties (8%) or other fields (1%). On a monthly basis, there was a nonstatistically significant trend towards retainer physicians providing more hours of charity care compared to nonretainer physicians (mean of 9.07 hours vs 7.48 hours/month, P=.24).

Table 1.

Characteristics of Retainer and Nonretainer Physicians and Practices

	Retainer Physicians (N=82)	Nonretainer Physicians (N=231)	P Value
Physician characteristics
Age, mean (median) (y)	48 (48)	49 (48)	.65
Male, n(%)	73	72	.88
Years in practice, mean (median)	16.7 (17.5)	18.4 (18)	.21
Specialty, n(%)			<.0001
General internal medicine	62	27
Internal medicine subspecialty	8	28
Family practice	28	40
Other	1	5
Patient panel, mean (median)	898 (400)	2303 (2000)	<.0001
Patients seen per day, mean (median)	11.6 (10)	22 (20)	<.0001
Hours of monthly charity care, mean (median)	9.07 (8)	7.48 (4)	.24
Practice characteristics
Services offered (%)
Accompanied specialist visits	30	0.5	.03
House calls	63	26	.008
24-h physician access	91	40	.02
Same-day appointments	96	83	.007
Coordinated hospital care	86	59	.045
Private waiting room	31	3	<.0001
Services utilized, mean % patients/3 mo *
Accompanied specialist visits	1.4	0.3	.35
House calls	3.0	0.3	<.0001
24-h physician access	15.0	10.7	.48
Same-day appointments	21.1	8.6	<.0001
Coordinated hospital care	4.5	3.2	.30
Private waiting room	25.7	0.3	.24

^*Analyses limited to physicians offering service, % calculated based on average panel size.

Practice Demographics and Services Offered

Compared to an average nonretainer physician, retainer physicians report having smaller patient panels (mean size 898 vs 2303 patients, P <.0001) and report seeing about half as many patients per day (12 vs 22, P <.0001;Table 1). Retainer physicians are also significantly more likely to offer each of the 6 special services examined (all P <.05), with the largest relative differences between retainer and nonretainer physicians in the use of accompanied specialist visits (30% vs 1%, P <.03) and private waiting rooms (31% of retainer vs 3% of nonretainer physicians, P <.0001). However, a considerable proportion of nonretainer physicians reported providing 24-hour access (40%), same-day appointments (83%), and coordinated hospital care (59%). The proportion of patients utilizing each service also varied based on physician type. After taking into account that retainer patient panels were approximately one third the size of their nonretainer counterparts, most services were utilized by a larger share of the patients of retainer than nonretainer physicians. For example, among physicians providing same-day appointments, nonretainer physicians reported an average of 8.6% of their patients having used this service over the previous 3 months, while retainer physicians reported an average of 21.1% of their patients having used the service over the same time period (P <.0001).

CHARACTERISTICS OF PATIENTS IN RETAINER AND NONRETAINER PRACTICES

Retainer physicians, as expected, reported caring for few patients on Medicaid compared to nonretainer physicians (Table 2) In addition, minority patients were also under-represented in most of these practices, with the majority of retainer physicians reporting patient panels with 0% to 5% African-American and Hispanic patients. Retainer physicians also reported caring for significantly fewer patients with diabetes (mean 17% vs 24% of patients, P=.008), and there were nonstatistically significant trends, suggesting they may care for slightly fewer patients with coronary artery disease (21% vs 25%, P=.40) and hypertension (30% vs 35%, P=.26).

Table 2.

Characteristics of Retainer and Nonretainer Patients

	Retainer Physicians (N=82)	Nonretainer Physicians (N=231)	P Value
Mean * proportion of …
Medicaid patients	5.0	15.3	<.001
African-American patients	6.8	15.5	<.002
Hispanic patients	4.4	13.5	<.001
Patients age 65 or greater	39.1	37.3	.64
Patients with diabetes	16.7	23.9	.008
Patients with CAD	21.0	24.8	.40
Patients with hypertension	30.3	34.7	.26

^*Mean calculated by imputing a midpoint answer for each respondent (e.g., 0% to 5% recoded as 2.5%, 6% to 15% recoded as 10.5%, etc.).

CAD, coronary artery disease.

Transition to Retainer Practice

Table 3 describes some of the factors associated with transition to a retainer practice. The mean length of time in retainer practice was 17 months (median 12 months), and 85% of retainer physicians converted from a nonretainer to a retainer practice. On average, physicians who converted to a retainer practice retained only 12% of their former patients (interquartile range 5% to 15%). Among all retainer physicians, approximately 17% of their patients did not pay the retainer fee. There was no statistically significant correlation between length of time in retainer practice and proportion of patients not paying the retainer fee (r=−.20, P=.15), suggesting that nonpayment of the retainer fee by some patients is a stable phenomenon unrelated to practice transition. Many physicians (42%) reported a high level of involvement with facilitating the transfer of care of patients not joining the practice, and two thirds of physicians had given their former patients at least 3 months to find a new physician.

Table 3.

Characteristics of Retainer Practices (n=82 Physicians in 56 Practices)

	Retainer Physicians
Characteristics of new and converted practices
Months in retainer practice, mean (median)	17 (12)
Practice development (%)
Started new practice	15
Converted or moved to retainer practice	85
Characteristics of converted practices
Proportion of patients in practice not paying retainer fee, % mean (median)	17 (10)
Level of involvement in facilitating transfer of patients not joining practice (%)
Very involved	42
Somewhat involved	11
A little involved	4
Not at all involved	2
Proportion of prior patient population joining retainer practice, % mean (median)	12 (10)
Amount of time patients given to find new physician (%)
<31 d	2
31 to 90 d	36
>90 d	63

DISCUSSION

This study represents one of the first systematic nationwide studies of retainer physicians. Currently, there is no comprehensive listing of retainer physicians in the U.S., and our sampling strategy revealed only a modest group of retainer physicians around the country. However, although we identified fewer than 200 of these physicians, ongoing debates and an evaluation by the Government Accountability Office demonstrate that the ethical and legal questions surrounding these practices persist. We sampled nonretainer primary care physicians as well, to provide comparisons between the 2 groups. In sum, we found significant differences in case-mix and services offered based on the physicians' practice type. Nonetheless, some physicians in traditional practices offer many of the same “special” services offered by retainer physicians. We also found that most retainer practices are urban, new, represent conversions from previously nonretainer practices, and that retainer physicians carry over relatively few patients from their former, nonretainer practices. These findings have a number of implications that help to inform ongoing debates about retainer medicine.

Patient Case-Mix and Access to Care

As expected, access to care for patients enrolled in retainer practices is enhanced compared to patients in nonretainer practices. We found that virtually all retainer physicians offer 24-hour access and same-day appointments, and house calls are very common. But critics of retainer practices have argued that these practices might attract wealthier and healthier patients (the “worried well”) rather than sick patients with complex illnesses, who tend to be less wealthy but who might benefit most from the additional attention retainer practices can offer.⁵ Similar claims have been made against “boutique” hospitals.⁹ Our data are limited to physicians' estimates of their patients' demographic and illness characteristics and therefore do not allow for examination of case-mix severity in detail. Nevertheless, we found that retainer physicians have smaller proportions of patients with diabetes, and perhaps other chronic diseases, than do their nonretainer counterparts and they care for fewer African-American and Hispanic patients. Given that minorities are already underserved and at risk for worse health outcomes,¹⁰ our findings suggest that retainer practices could contribute to tiering of health care and to disparities in health care according to race as well as wealth. However, the effects of these findings are likely to be small given the current prevalence of these practices and the magnitude of differences that we describe. Though there has been increasing momentum for retainer practice over the past decade, these efforts remains tiny relative to the more than 600,000 practicing physicians in the U.S. Our study cannot definitively address whether the case-mix of retainer practices is causally driven by their retainer versus nonretainer status. An alternative explanation is that these practices tend to emerge in high-income areas whether there are fewer African-American patients or Medicare beneficiaries. However, we did not find that adjustment for the population level income of practice location significantly changed our findings.

Market Segmentation and the Legality of Retainer Practices

Some policy experts have speculated that the health care marketplace will become increasingly segmented based on ability to pay.^11,12 Retainer practices offer one example of such segmentation. We found, for example, that retainer physicians were more likely to offer each of 6 “special services” and that in many cases their patients utilize these services more often. But most of these services (e.g., house calls) are also available in at least some traditional practices, even though nonretainer physicians may not prominently advertise the provision of these services. The fact that nonretainer physicians occasionally offer what might be called a “retainer level of care,” combined with the fact that some practices are using retainer fees to subsidize the costs of health care for nonretainer patients,¹³ poses a dilemma. Both private and governmental insurance plans are trying to determine whether physicians may charge retainer fees to their enrolled patients.⁷ Contractually, the acceptability of such retainer fees might depend upon the fees not being used to pay for otherwise routinely offered “covered services.” Our results underscore the heterogeneity of the services offered by both retainer and nonretainer physicians, and the difficulty of distinguishing “special services” from routinely covered care.² It is also noteworthy that these findings did not change after adjusting for the average income of the population in the zip code area where the physician practice was located, suggesting that differences in services offered may not merely reflect differences in the incomes of the populations living where these 2 types of practices are located.

Continuity of Care and the Transition to Retainer Practice

We found considerable patient discontinuity among physicians converting to retainer practice. For example, the average reported panel size of nonretainer physicians was about 2,300 patients. We calculate that, on average, conversion to retainer practice entails transferring care for approximately 2,025 patients, adding approximately 560 new patients, and continuing to see approximately 140 patients who do not pay the retainer fee. Of note, many physicians reported active involvement in transitioning patients to other practitioners, which is consistent with professional guidelines ² and may minimize harms potentially associated with discontinuity of care.¹⁴ In addition, most retainer practices are in urban areas that are not as affected by physician shortages as more rural settings, where the harms of discontinuity might be exacerbated by difficulties finding a new physician. Nevertheless, the ethics of discontinuing care for only a certain subset of one's patients (as distinct from discontinuing care for all patients, as when a physician moves or retires) has not been fully explored. Our data do not allow for us to examine the types of patients who discontinue seeing physicians converting to retainer practice, but they do suggest this issue will continue to pose a challenge.

Study Limitations

This study has several limitations. First, as was noted, there is no comprehensive listing of all retainer physicians in the U.S. and our sample may not be representative of the universe of retainer practices. Second, any survey based on self-report is potentially subject to estimation errors and socially desirable response bias. Survey results can also be subject to nonresponse bias, though we did not detect response wave bias, nonrespondents were similar to respondents on measurable variables, and our response rates are similar to those of other rigorously conducted physician mail surveys.¹⁵ Third, we did not explore several other domains of interest regarding these practices, including magnitude of retainer fees, coordination of care within other benefits patients have, or additional services offered beyond the 6 services that we focused on. Finally, retainer practices remain dynamic and heterogeneous within a fluid health care environment. As we conducted the survey, there were media reports of physicians charging their patients additional fees for various purposes, such as to offset increasing malpractice insurance premiums, but without promise of new or additional services in return.¹⁶ This suggests that retainer practices as we defined them might be one manifestation of broader efforts by physicians to recapture income, time, and other resources lost in an evolving health care system.^17,18

CONCLUSION

There are considerable differences between physicians in retainer versus nonretainer practices, most notably in the average patient panel size, but also in a variety of measures of access to care and other services. Despite these differences, some nonretainer physicians provide many of the “special services” that tend to be associated with retainer practice. Such findings, along with information on the displacement of patients in retainer practice conversion, suggest that ethical and legal debates about the standing of these practices are likely to endure. Although we explored retainer practices from the perspective of physicians, we cannot assess the experiences of patients in retainer practices, or of those who choose not to follow their physicians into such practices. Such information would be an invaluable complement to our findings.