Physician Manipulation of Reimbursement Rules for Patients:  Between a Rock and a Hard Place FREE

Physicians' decisions about what services to offer their patients affect almost 80% of all health care expenditures and have an enormous influence on health care quality.^1- 2 To control cost and quality and ensure adherence to their contracts, health care delivery organizations and payers frequently review physician recommendations and pay for services (eg, diagnostic tests, drugs, or hospitalization for treatment) only in predefined circumstances. This utilization review may occur prospectively, in the form of preauthorization; concurrently, requiring immediate approval over the telephone; or retrospectively, with payment decisions being made after services have been delivered. Specific utilization review criteria are rarely spelled out in advance; what physicians agree to when they sign health plan contracts varies greatly and may both alter and constrain physicians' decisions regarding their patients' use of services.

Although utilization review may be intended to improve quality and save money, many physicians and patients dislike conforming to its rules. Utilization review is often viewed as overly time consuming, a hassle, and an undue questioning of professional authority.^3- 4 Appealing adverse coverage decisions may seem inconvenient, unproductive, or even risky for physicians. In some cases, physicians may feel trapped between professional obligations to advocate for their patients and conflicting contractual obligations to follow coverage rules. Furthermore, it has been suggested that some insurers are "gaming" patients and physicians—tricking them into paying for covered services by routinely denying coverage but then approving services that are subsequently appealed, knowing that time and other constraints will prevent some appeals.^4- 6 To retaliate, physicians may be tempted to turn the tables.

Physicians could manipulate reimbursement rules to help their patients obtain coverage for care that the physicians perceive to be necessary, for example, through ambiguous documentation or by exaggerating the severity of patients' conditions.^7- 9 This sort of deception of third-party payers has been called "gaming the system" for patients⁵ and it is not a new idea, nor is it limited to managed care. Fee-for-service payment often provides a greater financial incentive for physicians to manipulate reimbursement rules since they directly profit from increased services. Yet recent data suggest that physicians in regions with higher managed care penetration are more likely to condone manipulating reimbursement rules.⁹ Some may believe that it is necessary to manipulate reimbursement rules in the course of patient advocacy today—even in the face of well-publicized crackdowns on fraud and abuse.^10- 12 These physicians may see manipulation of reimbursement rules as an indirect, or covert, form of patient advocacy and even a professional obligation.

Previous studies have posed hypothetical coverage dilemmas to physicians and offered deception of insurers as potential solutions.^9,13- 14 These studies demonstrate that many physicians believe there are scenarios in which manipulating reimbursement rules would be justifiable in theory. Anecdotes suggest that using such tactics is not merely hypothetical but occurs in practice.¹⁵ This study sought to determine how many physicians are manipulating reimbursement rules to get coverage for their patients' needed care, and what factors might affect how frequently physicians circumvent utilization review in practice.

We performed a self-administered mailed survey of physicians in 1998, using a random sample of 1350 physicians from the American Medical Association (AMA) Masterfile of all physicians practicing in the United States. Physicians who were deceased, retired, or no longer seeing patients; who had moved with no forwarding address; or who were in training were excluded, bringing the final sample to 1124 eligible physicians.

Because of the sensitive legal nature of the survey questions, an independent survey research firm, the National Opinion Research Center at the University of Chicago, Chicago, Ill, conducted all mailings and telephone calls, and all potentially identifying information for respondents, including city of residence, was removed from the data set prior to its release to the research team. Respondents were informed in the cover letter that the information would remain strictly confidential. It is not possible to track data back to identifiable individual respondents or health plans using the hard copies of the surveys. The study protocol was approved by an independent institutional review board (Western IRB, Olympia, Wash).

Questionnaires were mailed with a modest financial incentive enclosed (average value, $10). Those who did not respond to the initial mailing of the survey received up to 3 follow-up mailings and 2 telephone calls. In the second call, physicians were given the opportunity to respond by telephone interview.

Dependent Variable. Questionnaire items were developed from the literature,³ from a 1996 pilot survey of 134 physicians attending a professional meeting,⁷ and from a focus group of practicing physicians. Our dependent variable was whether the physician had manipulated reimbursement rules "to help patients secure coverage for needed treatments or services" in the last year. This was operationalized as the frequency of using 3 general strategies: exaggerating the severity of the patient's condition, changing the patient's official (billing) diagnosis, and/or reporting signs or symptoms that the patient did not have. These 3 items were each reported on 5-point Likert scales ("never" to "very often"). Descriptive statistics showed that all 3 of these variables had skewed distributions, and, using confirmatory factor analysis, we determined that the 3 items clustered together and the resulting scale had acceptable internal consistency reliability (Cronbach α = .68). Therefore, we felt it was appropriate to use these 3 variables to create a dichotomized summary variable for a multivariate model. Based on their distributions, we made an a priori decision (ie, before examining results of bivariate and multivariate analyses) to dichotomize each of these variables such that "never" and "rarely" were in 1 group and "sometimes," "often," and "very often" were in the other. The 3 items were then aggregated into a single dichotomous variable such that those who answered "never" or "rarely" to all 3 items were in 1 group and all other physicians were in the other.

Independent Variables. We hypothesized that a number of personal and practice characteristics might affect a physician's propensity to manipulate reimbursement rules. We examined characteristics such as number of years in practice; specialty; country of origin, employment type; average number of patients seen each week; number of managed care contracts; proportion of income from capitated, fee-for-service, or salary payments; and total proportion of income at risk for the costs of patient care.

We assessed physicians' attitudes using a series of items that asked for agreement or disagreement on a range of statements, using 5-point Likert scale responses ("strongly agree" to "strongly disagree"). Each of these items was collapsed for analysis into 3 categories: those who agreed (including "agree" or "strongly agree"), those who disagreed (including "disagree" or "strongly disagree"), and those who were not sure.

One item asked whether physicians agreed that "today it is necessary to game the system to provide high-quality care" (italics in the original). Three statements were designed to measure whether the respondent believed that "gaming the system" was a legitimate part of patient advocacy: (1) "In general, it is ethical to game the system for your patient's benefit"; (2) "It is a physician's responsibility to advocate for his/her patient's needs even if it involves exaggerating the severity of the patient's condition to third parties"; and (3) "Assuming your intent is to further a patient's interest, exaggerating the severity of a patient's condition to an insurer is okay." These items were intercorrelated (all correlation coefficients >0.5) so that only the results for item 1 are reported. Other items addressed numerous potentially relevant attitudes and experiences, including "All things considered, I am satisfied with my medical practice"; "I feel financially secure"; "Worry about prosecution for fraud prevents me from exaggerating patients' conditions to third parties"; and "I have enough time with each patient to get everything done that I need to."

Three statements addressed physicians' perceptions of the inconvenience involved in the utilization review process: (1) "Appealing an adverse utilization decision usually takes an excessive amount of my own time"; (2) "I find insurance company intrusion into medical decision making extremely annoying"; and (3) "Appealing adverse utilization decisions is a tremendous hassle." We used these items to form a 3-to-15-point scale confirmed by factor analysis (Cronbach α = .68), which we call the Hassle of Utilization Review scale. Physicians scoring higher on this scale felt more inconvenienced by utilization review. Finally, physicians were asked how often in the last year patients had requested that the physician "in some way deceive their third-party payer to help them secure coverage for a service."

Bivariate relationships between the dichotomous summary gaming variable and each independent variable were assessed using t tests for comparisons that used continuous variables and χ² tests for those that used ordered variables. Variables with bivariate relationships to the dependent variables of P≤.20 were candidates for the multivariate model. Multivariate logistic regression models were estimated using forward and backward selection methods. Identical independent variables were significant using both criteria, so a straight-entry model was created using those variables, and those results are presented. To obtain adjusted odds ratios (ORs) in these models, continuous variables were dichotomized at reasonable cut points based on their distributions and clinical importance. For categorical variables with "agree," "disagree," and "not sure" categories, the "agree" and "not sure" categories were each compared with the "disagree" category. One categorical variable ("How often have your patients requested that you in some way deceive their third-party payer to help them secure coverage for a service?") had 5 response options, from "never" to "very often." This variable was dichotomized by collapsing the "never" and "rarely" categories and the "sometimes," "often," and "very often" categories. All analyses used the SPSS statistical program (SPSS Inc, Chicago, Ill).

The 720 physicians (response rate, 64%) who completed the questionnaire did not differ from nonrespondents on any available demographic variables. Respondents represented a range of incomes, types of practice, and specialties, as shown in Table 1. Most (80%) were men, had been in practice more than 10 years (mean, 17 years), and were in active clinical practice (seeing a mean of 86 patients per week). In all but 1 respect, the demographic frequencies in Table 1 are also similar to available national statistics.¹⁶ For unexplained reasons, our overall random sample contained a larger-than-usual proportion of foreign-born physicians (nearly 40%) compared with nationally (27% in the AMA Masterfile; Tom Pasco, written communication, July 21, 1999). However, foreign-born physicians were not more likely to complete the questionnaire and there were no differences in our dependent variable or its components between foreign-born and US-born physicians.

Figure 1 shows physicians' reported use of 3 ways to manipulate reimbursement rules, using the exact wording of the survey items. Exaggerating the severity of patients' conditions to help them avoid early hospital discharge was the most commonly used tactic, with 28% of respondents reporting that they had done this sometimes or more often in the last year. Almost as many respondents (24%) reported having changed patients' official (billing) diagnoses, while 10% had recorded signs or symptoms that patients did not actually have to help them secure coverage for needed care.

Altogether, 275 physicians (39%; Figure 1) reported that they had sometimes, often, or very often used at least 1 of the 3 tactics for manipulating reimbursement rules, and 424 (61%) had rarely or never used any of the tactics in the last year. While we dichotomized respondents in this way to create our dependent variable for subsequent analyses, it may be of interest that 193 respondents (28%) reported never using any of the 3 tactics, while only 3 physicians (0.4%) reported using all 3 tactics often or very often and 49 (7%) reported using at least 1 manipulative tactic often or very often.

We also asked each physician who reported using any of these 3 tactics in the last year whether "you find yourself using deception of third-party payers to obtain needed benefits for patients" more or less often now than 5 years ago. Fifty-four percent responded "more often," while 12% reported doing so less often and the rest said it was unchanged.

As shown in Table 2, 28.5% of physicians agreed with the statement, "Today it is necessary to game the system to provide high quality care" and 15.3% agreed with the statement, "In general it is ethical to game the system for your patients' benefit." These 2 items were strongly associated (χ²₁₆ = 187.5; P<.001); among those who believed that gaming the system is necessary, 42.9% also believed it was ethical, while among those who did not think gaming is necessary, only 4.2% believed it was ethical. Given this significant association, only the first item was used for subsequent statistical analyses, yet it is of interest that even among those who believed that gaming the system is necessary, the majority (57.2%) did not believe that it was ethical.

The Hassle of Utilization Review scale had a mean score of 12.8 (median, 13; range, 3-15). The middle value for this scale was 9, so a mean of 12.8 indicates that, on average, physicians fell between "agree" and "strongly agree" on these items. Roughly 58% of physicians were satisfied with their practice, 47% felt financially secure, 57% reported that worry about prosecution for fraud might prevent them from exaggerating patients' conditions to third-party payers, and 44% believed they had enough time with each patient to get everything done that they needed to. Finally, more than 1 in 3 physicians (37%) reported that their patients sometimes, often, or very often requested that they deceive third-party payers to help them obtain coverage for services in the last year.

Physicians who reported manipulating reimbursement rules were almost twice as likely to report patient requests to deceive third-party payers as those who did not (50.5% vs 27.2%; P<.001; Table 3). Physicians who agreed with the statement, "Today it is necessary to game the system to provide high-quality care" reported manipulating reimbursement systems more often than those who did not agree with the statement (64.3% vs 35.7%; P<.001). The mean score on the Hassle of Utilization Review scale was also significantly higher for those who reported manipulating reimbursement rules compared with those who did not (13.2 vs 12.4; P<.001). Perceptions of time pressure during office visits were positively correlated with physicians' decisions to manipulate reimbursement rules; those who reported using these tactics were less likely to agree with the statement that they have enough time with patients to get everything done that they needed to (35.8% vs 48.8%; P<.001). Physicians who manipulate reimbursement rules also reported a somewhat higher percentage of patients receiving Medicaid (20.6% vs 15.3%; P = .002), were less satisfied with their medical practice (51.6% vs 63.0%; P = .003), and felt less financially secure (40.4% vs 52.6%; P = .002) compared with physicians who reported rarely or never using these tactics. Those who reported manipulating reimbursement rules were less likely to agree that worry about prosecution for fraud would prevent them from exaggerating patients' conditions to third parties (42.3% vs 57.7%; P = .04). Other variables, including physician demographics (specialty, sex, rural practice, physician country of origin, and years of practice); primary payment by fee for service, capitation, or salary; physician income; and number of managed care contracts, were not associated with manipulating reimbursement rules for patients (P>.20 for all).

In the multivariate model, a physician's belief that gaming the system is necessary to provide high-quality care remained the strongest predictor of manipulating reimbursement rules. As shown in Table 4, those who agreed with this statement were more likely to report having manipulated reimbursement rules compared with those who disagreed or were not sure (OR, 3.67; 95% confidence interval [CI], 2.54-5.29). Having patients who requested that physicians deceive third-party payers was also strongly associated with manipulating reimbursement rules, with those who received such requests being more likely to use these tactics (OR, 2.44; 95% CI, 1.72-3.45). Similarly, physicians who reported not having enough time with each patient had significantly higher odds of reporting manipulating reimbursement rules (OR, 1.69; 95% CI, 1.21-2.37), as did those with more than 25% of their patients covered by Medicaid (OR, 1.60; 95% CI, 1.08-2.38). Two negative findings from the multivariate model also merit mention. First, the physician's level of financial security was no longer related to physicians' manipulation of reimbursement rules (P = .11). Second, physicians who reported worrying about prosecution for fraud were not less likely to report manipulating reimbursement rules (P = .34).

In this study of a nationally representative sample of physicians, we sought to determine how frequently physicians manipulate reimbursement rules to help their patients secure coverage for care that the physicians perceive to be necessary. Unlike previous studies, which have examined whether physicians might condone gaming the system under hypothetical circumstances,^9,13,17 we were interested in estimating how often physicians actually use deceptive tactics in the current health care system and whether certain factors might be increasing this frequency. We found that 3 different ways of manipulating reimbursement rules were all relatively uncommon but far from rare; between 10% and 27% of physicians reported using each tactic "sometimes" or more often in the last year to help patients obtain coverage. Added together, while the majority of physicians (61%) reported rarely or never exaggerating patients' severity of illness, changing patients' billing diagnoses, or reporting signs or symptoms that patients did not have in the last year, the rest (39%) reported using 1 or more of these tactics at least sometimes. In a multivariate model, the belief that gaming the system was necessary to get patients needed care, more requests from patients to deceive payers, believing that visit lengths were too short to get everything done that they needed to do, and having more than 25% of one's patients covered by Medicaid were independently and significantly associated with having manipulated reimbursement rules in the last year.

Physicians might manipulate reimbursement rules to benefit patients, themselves, or both. Since the financial and health-related interests of patients and physicians can coincide, it is impossible to completely separate these motivations. However, 2 aspects of our findings suggest that financial self-interest is not the sole motivation for most physicians who manipulate reimbursement rules. First, manipulation of reimbursement rules was most common in the situation in which an individual physician could not possibly provide free or reduced-cost care (ie, hospitalization). Physicians may reserve gaming the system for situations in which free care cannot be offered. Second, to our surprise, we found no association between manipulation of reimbursement rules and any of the financial markers we examined, such as proportion of income at risk, principal type of reimbursement (fee for service vs capitation or salary), or recent practice-related income losses. Although the physician's perceived level of financial security was significant at the bivariate level, it was not significant in the multivariate model. Physicians with potentially greater financial stakes in manipulating reimbursement rules did not report doing so more often.

Not surprisingly, physicians who reported manipulating reimbursement rules were much more likely to believe that gaming the system is necessary to provide high-quality care today. Of perhaps more importance is that physicians who believed that gaming the system for patients is necessary were also significantly more likely to believe that it is ethically acceptable, indicating some degree of ethical adaptation to perceived necessity. However, even among physicians who reported that gaming is necessary, the majority still did not believe it is an ethical practice. This accords with the strongly worded official declarations of professional societies, wherein manipulating reimbursement rules is considered an ineffective and socially irresponsible way to advocate for patients' interests.^18- 20 These and other ethical arguments against gaming reimbursement systems are well developed.^{5,8,18- 19,21- 22} Yet, although the AMA and others have consistently opposed the manipulation of reimbursement rules, regardless of intent, and though most physicians agree with this position, some nevertheless appear to view this advice as untenable given their immediate patient care obligations. Moreover, in the face of the current crackdown on fraud and abuse,^11- 12 it is interesting that worry about prosecution for fraud did not seem to inhibit physicians from manipulating reimbursement rules to obtain coverage for needed care. Thus, neither ethical pronouncements nor further fraud and abuse enforcement are likely to be very effective in deterring manipulation of reimbursement rules among physicians who see it as the only feasible way to provide high-quality care.

Many physicians who reported manipulating reimbursement rules also reported receiving patient requests to deceive third-party payers. This suggests that both physicians and patients may not view current utilization review processes as legitimate or responsive to their needs. However, physicians could also be manipulating reimbursement rules in response to patient requests for another reason. Some might be deceiving insurers rather than confronting patients who present with unreasonable demands. Manipulating reimbursement rules may be easier and less time-consuming than attempting to explain to patients why a requested intervention is marginally beneficial or unnecessary, or exploring whether it could be paid for out-of-pocket by the patient.

Increasing time pressures have been linked to poorer health outcomes²³ and possibly increased malpractice claims.²⁴ We found that feeling pressed for time is also associated with increases in manipulating reimbursement systems among physicians. This could reflect the time required to pursue regular advocacy routes, such as filing formal appeals. When time pressures make using regular advocacy routes inconvenient or impossible, physicians may be tempted to use covert advocacy tactics, such as manipulating reimbursement rules, to help their patients obtain needed care. We do not have information on the specific health plans in which manipulating reimbursement rules occurred and whether regular appeals processes were readily available, but manipulating reimbursement rules might be perceived to be the most rapid and convenient short-term way to obtain services in some situations.

It is not clear from these data why having more than 25% of one's patients covered by Medicaid is associated with a higher propensity to report manipulating reimbursement rules. Medicaid reimbursement rates are often lower than those of commercial payers, so accepting more Medicaid patients may put financial stresses on physicians that increase the perceived need to game the system to achieve target incomes. Some individuals are also more willing to mislead government entities than smaller enterprises.²⁵ Medicaid plans may also carry out more stringent or more intrusive utilization review, or patients covered by Medicaid simply may need more services and, hence, present more potential opportunities for physicians to manipulate reimbursement rules.

Although many physicians report manipulating reimbursement rules to help their patients obtain coverage for care that the physicians perceive to be necessary, the majority report doing so rarely or never. From these data, we cannot tell whether this is because they did not have to or because they chose other avenues to obtain the care, or whether their patients did not receive needed care. The last scenario would be particularly worrisome for patients and their physicians, in part because current case law suggests that physicians are legally accountable for providing needed care and that they do not derive legal immunity by agreeing to accept health plan coverage rules.^26- 27 That is, compliance with health plan rules does not provide protection against malpractice litigation and losses stemming from failure to provide an acceptable standard or care. Of course, failure to adhere to health plan coverage rules and manipulating these rules, even to benefit patients, also has potential legal repercussions.¹⁰

This study has several important strengths. We surveyed a random sample of physicians nationwide, including all specialties, and achieved a suitable response rate despite the controversial topic. For the first time that we are aware of, physicians were asked to report the frequency of actually manipulating reimbursement rules in practice. In previous studies, researchers have used second- or even third-party hypothetical scenarios to explore these behaviors.^9,13- 14 While using hypothetical scenarios may be less threatening to respondents, it cannot address key issues in contention, such as whether these scenarios arise in practice and whether physicians really act as they say they might. Another strength is that we asked only about manipulating reimbursement rules to help patients obtain coverage for care that the physicians perceived to be necessary. While we cannot ascertain whether the physicians' judgments of medical necessity were accurate, we intentionally avoided issues of large-scale fraud that the Health Care Financing Administration and others have primarily targeted, such as billing for services not rendered, kickbacks, and self-referral.¹⁰ Most physicians oppose such activities, so they pose no real ethical dilemmas. Since we wished to address the frequency of physicians facing true ethical dilemmas due to reimbursement policies, the situations we addressed were intentionally ethically challenging, with no clear "right" answer.

There are several relevant study limitations. We could not independently confirm reported acts of manipulating reimbursement rules for patients. If there were a systematic reporting bias, however, we believe that it was in the direction of underreporting. Although we tried to frame the survey questions in neutral tones, the words gaming and deception were both used in some items and these words have negative connotations. Combining this socially desirable response bias with the perceived legal risk involved in admitting to using these tactics, it is likely that respondents underreported how often they manipulate reimbursement systems. A second limitation, with a similar underreporting bias, is that we only asked about 3 general ways in which physicians might manipulate reimbursement rules to help patients obtain care. It is likely that there are other ways of manipulating reimbursement rules that we did not address.

There are no simple solutions to the problem of physicians manipulating reimbursement rules for their patients. The root causes of this problem are tensions that are structural, largely unavoidable, and likely to become increasingly intense. These tensions reflect an underlying uncertainty as to whether health care is best viewed in a market-based contractual model or a profession-based fiduciary model, when both models have ethical and legal strengths. A contractual model of health care, wherein patients and physicians are closely held to their written agreements with health plans and insurers, supports the compelling ethical principle of contractual justice.²⁸ And American society is leaning strongly toward a market-oriented emphasis on value for money and free choice in health care. Patients with good information and options should "get what they pay for." On the other hand, physicians traditionally act as their patients' agents (fiduciaries) and strive to provide an equally high standard of care to all, regardless of what individual patients can afford to pay.²⁹ It is well-known that many patients are not "good" health care consumers,^30- 31 and can unwittingly agree to or be assigned to contracts that may not fully serve their medical needs. Bending the rules for patients to provide a high-quality standard of care may be seen as an act of beneficence, or mercy, which is also ethically compelling.³² While insurance policies are contracts between patients and insurers, our findings suggest that some physicians believe that strictly enforcing these contracts is contrary to their professional role as patients' agents and caregivers.

Physicians are thus caught in 2 sets of conflicting demands. Legally, physicians are contractually bound to adhere to reimbursement policies yet are also liable for failure to deliver an equally high standard of care to all patients, regardless of ability to pay. Ethically, physicians are caught between the compelling principles of social justice on the one hand and beneficence, or mercy, on the other. We can document the effects of these tensions and perhaps find ways to alleviate them, but the question no empirical study can answer is, "When, if ever, is it ‘good' for a contract to be broken in a surreptitious act of mercy?"

Without answering this question, our findings still have several practical implications. As pressures to control health care costs increase, it is likely that manipulating reimbursement systems will increase in parallel. Efforts to more tightly control utilization will likely increase physicians' perceived need to manipulate reimbursement rules to provide high-quality care and will increase patients' requests for the services being controlled. Further attempts to reduce physician reimbursement will put pressures on physicians to increase patient volumes, necessarily reducing the time physicians and patients spend with each other. Our findings suggest that such conditions will be associated with more covert manipulation of reimbursement rules, and that enforcement of fraud and abuse statutes will not alleviate this. On the other hand, possible approaches to minimize the perceived need for physicians to manipulate reimbursement rules might include providing adequate time for patient-physician discussions and involving physicians and patients in benefits determinations and utilization review processes. In our view, making utilization review procedures more user-friendly might encourage patients and physicians to use legitimate advocacy channels. Physicians and patients who are positively disposed to a utilization review system and who view its processes as legitimate, responsive, and nonpunitive might be less tempted to manipulate it covertly for their own benefit. But attaining this goal must involve patients and physicians becoming committed to the social systems through which coverage decisions are made and overcoming their apparently high levels of cynicism regarding the effectiveness and fairness of these systems today. Finally, as a marker for systemic problems, evolving systems of care delivery could be individually assessed for their effects on physicians' perceived need to covertly manipulate reimbursement rules to provide high-quality care. Health plans in which the use of these tactics is common should carefully review their rules and procedures and work with physicians to reduce the perceived need for covert advocacy.