Selection and Exclusion of Primary Care Physicians by Managed Care Organizations FREE

MANAGED care has fundamentally altered the balance of power between physicians and third-party payers. One of the elements of managed care most responsible for this transformation is selective contracting.¹ Patients enrolled in managed care plans must choose from a limited panel of physicians offered by the plan. Selective contracting, thus, places in the hands of managed care plans considerable power over the economic survival of practicing physicians, since contracts are a necessary condition for access to patients in these plans. As managed care comes to increasingly dominate the health insurance market, physicians excluded from managed care networks face the prospect of becoming physicians without patients.

Little is known about the problems physicians may be encountering in gaining access to managed care networks. Among the concerns that have been expressed is that contracting decisions may be discriminatory based on the characteristics of physicians or their patients. A recent National Public Radio story featured minority physicians who suspected that they had been denied managed care contracts because of their race.² Similar stories have been written about older physicians who believe they have been excluded from managed care because of their age.³ Managed care networks may develop around existing informal physician networks based on traditional referral patterns and social groupings and may perpetuate segregated arrangements. Physicians who have cared for many low-income patients or patients with costly chronic illnesses may also be considered unattractive by managed care plans that seek to market to more affluent populations and to avoid adverse risk. Many managed care plans are believed to use "economic credentialing" in their contracting decisions, whereby physicians with costly practice profiles are excluded from the plan.⁴

Representatives of the managed care industry have defended selective contracting as a means to lower costs and improve patient care. They argue that plans consider quality of care to be of paramount importance in contracting decisions, as demonstrated by policies in some plans that require qualifications such as board certification for participation in the physician network.

Concerns about selective contracting have sparked a great deal of legislative activity to regulate this facet of managed care. The American Medical Association (AMA) has been a proponent of "any willing provider" legislation that would force managed care plans to offer contracts to any physician meeting basic standards for participation.⁵ Consumer and physician groups have rallied behind "patient bill-of-rights" measures that would prohibit summary termination or denial of physician contracts and force plans to demonstrate "just cause" for excluding a physician.⁶

We performed a study in California to investigate more systematically how commonly primary care physicians experienced denials or terminations of managed care contracts, to evaluate the characteristics of physicians and their practices that might be associated with denials and terminations, and to determine the impact of denials and terminations on primary care physicians involvement with managed care. California has one of the highest degrees of managed care market penetration of any state and features an array of well-developed physician networks. These conditions make California a useful "laboratory" for examining how the health system's uncontrolled experiment in managed care may be affecting primary care physicians.

We conducted a mail survey of a sample of physicians providing primary care in the 13 largest urban counties of California, which constitute 79% of the state's primary care physicians and 78% of the state's population. Physicians who were listed in the 1995 AMA Masterfile as providing direct patient care, not in training, not employed by the federal government, and with a primary specialty of general internal medicine, family practice, pediatrics, or obstetrics-gynecology were sampled. Within each county, a stratified random sample was taken that included 5 categories of physician race or ethnicity: African American, Asian, Latino, white, and race not specified. A sample of 1600 from the eligible population of 13404 physicians was selected with a minimum of 100 and a maximum of 200 physicians selected in each county. Nonwhite physicians were oversampled throughout to give adequate numbers in each of the 5 race or ethnicity strata. Of the 1600 physicians, 205 (13%) were African American, 328 (20%) Asian, 288 (18%) Latino, 546 (34%) white, and 233 (15%) race not specified compared with 327 (3%), 2238 (17%), 476 (4%), 6282 (48%), and 3782 (29%) for the corresponding categories in the AMA list of primary care physicians.

The self-administered questionnaire included items about physicians' demographics, their training, their practice size, the racial and insurance composition of their patient population, whether physicians had a managed care contract through an independent practice association (IPA) and/or directly with a health maintenance organization (HMO), and the percentage of capitated patients in their practice. (In California, many office-based physicians participate in IPAs, which serve as organizational structures for managing contracts with HMOs.) Physicians were also asked whether they had ever been denied or terminated from a contract with an IPA or HMO. Physicians who had been denied or terminated from IPA and HMO contracts were asked the reasons they were told and what they believed to be the reasons for that action. Physicians could provide more than 1 reason.

We examined the yellow pages of the 1996 California telephone book, the 1996 CD-ROM California Medical Board listing of licensed physicians, and the directory of board-certified specialty physicians to determine eligibility of physicians not responding to the survey mailings. Physicians not responding after 2 mailings were contacted by telephone to verify location and eligibility. Physicians still not responding after 4 mailings were contacted by telephone and office visits to encourage participation. Physicians who we identified as having retired, moved out of state or died or who indicated a main practice specialty other than primary care were considered ineligible and removed from the sample.

We limited the analysis to physicians reporting their main practice setting as office-based practice. Based on their self-reports of the percentage of uninsured, Medicaid, and nonwhite patients in their practices, we compared physicians in the top quartile of each of these categories with those in the lower 3 quartiles. The top quartiles for percentage uninsured, Medicaid, and nonwhite patients were greater than 4%, 20%, and 75%, respectively. We analyzed separate variables describing whether a physician had been denied an IPA contract, terminated from an IPA contract, denied a direct HMO contract, or terminated from a direct HMO contract. In addition, we created a combined dependent variable of denial or termination from an IPA or a direct HMO contract. Similarly, we developed a combined dependent variable describing whether a physician had a contract with at least 1 IPA or an HMO. Statistical analyses were performed using χ² for bivariate comparisons of categorical data and linear regression or logistic regression models for multivariate models with continuous and dichotomous outcomes, respectively. In multivariate analyses, we found an interaction between physician age and sex for several of the outcome variables. We, therefore, combined age and sex predictor variables to create 4 age-sex categories. Analyses generalizing all physicians at the county level or the state level (all 13 counties) were weighted by the inverse of the product of the sampling fraction and the participation rate to account for oversampling of nonwhite physicians and differences in response rates.

Of the original sample of 1600 physicians, 1336 were ultimately determined to be eligible. We obtained a 71% completion rate (n=947); 5% (62) refused, and 24% (327) did not respond. There were no significant differences between the age, sex, race or ethnicity, or specialty of respondents and nonrespondents. The 15% of sampled physicians who did not specify race in the AMA Masterfile were found to be disproportionately white (67%) or Asian (28%) based on survey responses.

Of the 520 office-based primary care physicians, 24% reported that they had been denied or terminated from a contract with an IPA or HMO (Table 1). Weighting our results of office-based physicians to adjust for the stratified sampling and differences in response rates by county resulted in an estimate that 22% of primary care physicians in the 13 sampled counties had experienced either a denial or termination from an IPA or direct HMO contract. In general, it was more common for physicians to report that they had been denied contracts with IPAs or HMOs than it was for them to report terminations.

In multiple logistic regression models that included all of the predictor variables in Table 1, solo practice was the strongest predictor of being denied or terminated from an IPA or direct HMO contract. Nonwhite physicians did not experience a higher percentage of denial or terminations than white physicians. Obstetrician-gynecologists were the primary care specialists with the highest percentage of denials and terminations. There was a trend toward physicians with higher proportions of uninsured patients to experience a contract denial or termination (odds ratio [OR], 1.4; 95% confidence interval [CI], 0.9-2.3).

Ninety-eight physicians, including 60 obstetrician-gynecologists, reported that they had not tried to obtain either an IPA or a direct HMO contract as a primary care physician. When we limited our analysis to the remaining 422 physicians, the rates of denials or terminations increased from 1% to 5% for all of the characteristics listed in Table 1. For example, the overall total rate of denials or terminations increased from 24% to 26%. After excluding physicians who had not tried to obtain a primary care contract, obstetrician-gynecologists continued to have the highest percentage of denials and terminations among primary care specialists, but the adjusted OR decreased from 3.6 to 2.5 (95% CI, 1.05-5.92) compared with pediatricians. The only other meaningful change in adjusted OR was for physicians whose practices included more than 4% of uninsured patients (OR=1.8; 95% CI, 1.1-3.2; P=.03).

The most commonly stated reason provided by IPAs or HMOs to physicians for denying a contract with an IPA or HMO was that there were too many physicians in the respondent's specialty (Table 2). By and large, physicians who were denied contracts with IPAs and HMOs believed the reasons that they were given for the decision. However, more than 30% of the physicians who were denied a contract with an IPA or HMO and more than 40% of the physicians who were terminated from a contract with an IPA or HMO received no reason for the decision. Of the physicians denied an IPA contract, 12 (20%) believed that the denial was due to racial discrimination and 15 (25%) believed it was related to not being in the necessary social circles. All of these physicians were nonwhite.

While denials and terminations were relatively common among primary care physicians, they did not preclude a high percentage of physicians from participating in managed care. Ninety percent of responding primary care physicians had either an IPA contract, a direct HMO contract, or both (Table 3). In the weighted analysis, 87% of primary care physicians in the 13 study counties had at least 1 IPA or direct HMO contract. Being denied or terminated from an IPA or HMO contract was not significantly associated with currently having at least 1 such contract. Ninety-three percent of physicians who reported a denial or termination from an IPA or direct HMO contract had at least 1 IPA or direct HMO contract at the time of the survey compared with 89% of physicians who did not report a denial or termination (P=.26). Similarly, the rate of terminations was not associated with the number of managed care contracts. Overall, a mean of 41% of the patients in the physicians' practices were insured under capitated HMO plans. In the weighted analysis, a mean of 37% of the patients were insured by capitated HMO plans. Physicians who were denied or terminated from an IPA or HMO contract had significantly fewer patients in their practice covered under capitated HMO contracts (35% vs 43%; P<.01).

In a multiple regression analysis predicting participation in either an IPA or direct HMO contract, practice size was again the strongest predictor of being involved with managed care. Consistent with the analysis of denials and terminations, the larger the practice group, the more likely the primary care physician was to be involved in managed care and the higher the percentage of capitated HMO patients in a physician's practice. In general, primary care physicians from underrepresented minority groups were as likely or more likely than white physicians to be involved in managed care. Primary care physicians who were in the highest quartile for providing care to the uninsured were significantly less likely to be in managed care and had significantly fewer capitated HMO patients compared with physicians who had lower percentages of uninsured patients. Similarly, providers who were in the highest quartile for providing care to nonwhite patients also had significantly fewer capitated HMO patients than did primary care physicians with lower percentages of nonwhite patients. Although lack of board certification was not associated with denials and terminations, it did predict less involvement with managed care.

Our study is the first that we are aware of to systematically evaluate physician exclusion from managed care in a representative sample of primary care physicians. We found that 22% of primary care physicians in California's large urban counties report that they had been denied or terminated from contracts with IPAs and HMOs. Given the demand for primary care physicians in managed care plans, this perhaps represents a larger number than might have been expected. We found that denials from both IPAs and direct HMO contracts were more common than terminations. This finding is consistent with statements by managed care administrators that they are more selective about whom they initially choose as participants than they are about whom they choose to retain.⁷ We also found that the patterns of denials and terminations were similar for IPAs and direct HMO contracts; this suggests that greater physician control in IPAs over decisions about physician contracting does not necessarily lead to different patterns of denials and terminations compared with those made by HMO managers.

While denials and terminations are commonplace among primary care physicians, we found that they did not completely preclude most primary care physicians from participating in managed care through either IPA or direct HMO contracts. Only 10% of primary care physicians had no IPA or HMO contract. While primary care physicians may not have been successful in obtaining every managed care contract that they seek, for the most part, they were able to participate in at least 1 managed care plan. Physicians who did experience a denial or termination, however, had a smaller proportion of their patients covered under capitated HMO contracts, indicating that denials and terminations may result in lower enrollment of HMO patients in these physicians' practices.

The patterns of denials and terminations we observed suggest that managed care plans are distinguishing among whom they choose to contract. However, we did not find that these practices are systematically discriminatory on the basis of a physician's race or ethnicity. In general, nonwhite physicians were no more likely to be denied or terminated from IPA or direct HMO contracts than were white physicians. Furthermore, nonwhite primary care physicians participated to the same or greater degree in IPAs and direct HMO contracts as white providers.

Our sample contains more than 70% of all office-based African American physicians practicing primary care in the largest urban counties in California, except in Los Angeles where the sample includes about a quarter of African American physicians. Given the broad interest in the California model of managed care⁸ and the opportunity to study a high concentration of nonwhite physicians, we believe our data provide an important opportunity to reveal evolving trends. However, our sample cannot determine with certainty that nationwide these physicians are not being discriminated against in managed care. In a survey of almost 300 African American physicians attending the annual conference of the National Medical Association, 92% reported that they believed that African American physicians were more likely than white physicians to be terminated from managed care plans.⁹ In our study, 20% of primary care physicians who were denied a contract with an IPA believed it was due to racial or ethnic discrimination; none of these physicians were white.

We did not find any evidence to suggest that managed care organizations were discriminating against physicians on the basis of their gender or older age. Older physicians may potentially be exposed to a longer period of opportunity for denial or termination and might therefore be expected to report more of these events than younger physicians. This could produce confounding between physician age and exposure to contracting decisions. While it might have been preferable to have measured the rate of denials and terminations within a fixed period, we suspect that this would not have dramatically altered our results since selective managed care contracting is a relatively recent phenomenon.

Board certification was not associated with the experience of a denial or termination. However, board-certified physicians were twice as likely to have 1 or more managed care contracts. Because board certification is an explicit criterion for a contract with many IPAs and HMOs,¹⁰ many of those without board certification may not even try to get these contracts and thereby not experience a denial. Our survey did not determine how aggressively physicians tried to obtain managed care contracts.

The strongest predictor of denials and terminations from contracts, as well as actual participation in managed care, was the size of the physician's practice. The larger the practice, the less likely the experience of a denial or termination from a contract. Physicians in large group practices may report lower rates of denials and terminations because in some cases they may be less aware of contracting decisions that occur for a group rather than solo practice. However, solo practitioners are less likely than physicians in group practice to participate in managed care. Physicians in solo practice were the least likely to be in either an IPA or to have a direct HMO contract, and they had significantly fewer HMO patients in their practice. The growing horizontal integration among physicians may be a necessary condition for obtaining managed care contracts. Between 1983 and 1994, the number of physicians in solo practice dropped from 41% to 29%.¹¹

Among primary care specialties, pediatricians were least likely to be denied or terminated and most likely to have either an IPA or HMO contract. Obstetrician-gynecologists, on the other hand, reported the highest amounts of denials and terminations among primary care physicians, but they also had the highest participation in managed care after pediatricians. This reflects obstetrician-gynecologists' ability to be denied or terminated as primary care physicians but still obtain managed care contracts as specialists. We cannot contrast obstetrician-gynecologists' experiences as specialists with other nonprimary care specialties because we did not include other specialists in our sample. The often described oversupply of specialist physicians and the recent reports of the decrease in demand for their services^12- 13 make us believe that rates of denials and terminations might be even greater among specialists than they were for primary care physicians.

The IPA and HMO administrators may have provided a different perception of the rates of and reasons for denials and terminations than did physicians. However, in most cases, physicians reported that the reasons they believed that denials and terminations occurred were the same as the ones they were told by managed care organizations.

Perhaps our most worrisome finding is that primary care physicians who provide a disproportionate amount of care to the uninsured and to nonwhite patients are significantly less likely to have managed care patients. The significant association of more frequent experience of denial from IPA and direct HMO contracts among physicians trying to obtain contracts with higher percentages of uninsured patients in their practices suggests that the lower participation in managed care may not be by these physicians' choice. Selective contracting for managed care may be occurring at the expense of physicians who provide a greater share of charity care. Rather than rewarding these physicians for their socially responsible deeds, the health care marketplace seems to be excluding them. If the penalty for caring for the uninsured and nonwhite patients is exclusion from managed care organizations, then more physicians may decide that they cannot afford to care for the underserved.

In conclusion, our study shows that managed care contracting is in fact being conducted with some selectivity. It is relatively common for primary care physicians in California to experience a denial or termination of a managed care contract. Although the experience of a denial or termination does not completely exclude physicians from participating in managed care, it may limit the numbers of managed care patients in a physician's practice. Our study failed to document that managed care organizations are disproportionately excluding minority or older physicians. However, managed care organizations may be reluctant to fully include physicians who care for substantial proportions of uninsured patients. Office-based physicians who care for underserved populations may find themselves at a disadvantage as the competition for managed care contracts and patients intensifies in the future.