Skip to main content
PMC home page
The Milbank Quarterly logoLink to The Milbank Quarterly
. 2002 Dec;80(4):709–755. doi: 10.1111/1468-0009.00029

Voices Unheard: Barriers to Expressing Dissatisfaction to Health Plans

Mark Schlesinger 1, Shannon Mitchell 1, Brian Elbel 1
PMCID: PMC2690133  PMID: 12532645

As American medicine and health policy have incorporated market perspectives, the emphasis on patients’ ability to act as effective consumers has grown. Empowering consumers is expected to achieve two goals (Andreason 1988). First, informed consumers can protect themselves from inadequate or inequitable treatment. Second, their actions can provide incentives for health plans to improve the quality of care for all enrollees. To produce these effects, consumers must be able to assess their health care experiences accurately and take appropriate action when they believe the plan has not performed satisfactorily. In recent years, state and federal policies to facilitate medical consumerism have been designed to expand consumer options, improve their decision-making abilities, and provide supportive services (Pollitz, Dallek, and Tapay 1998; Rodwin 2001). During the 1990s, many private employers pursued similar policies for their workers (Maxwell et al. 1998).

Since Hirschman's seminal work in 1970, scholars have recognized that consumers can respond to unsatisfactory treatment in one of two ways: by “exiting,” or switching to another provider of services, or through “voice,” by expressing their dissatisfaction with their current provider. In the market for health plans, exit is often quite limited. Employers are more likely to select a plan for their workers rather than allowing employees to choose among several (Marquis and Long 1999; Maxwell, Temin, and Watts 2001; Rice et al. 2002). In public programs such as Medicare or Medicaid, a choice among health plans is rarely available outside large metropolitan areas (Gold and McCoy 2002). Even when individuals do have choices, switching plans may be so burdensome that enrollees stay in their plan despite a history of unsatisfactory treatment. This is particularly true when the enrollees have chronic health problems and changing plans would disrupt the continuity of care with their physicians (Schlesinger, Druss, and Thomas 1999).

Recognizing these constraints on exit, policymakers have tried to bolster consumers’ ability to voice their dissatisfaction directly to their health plan (Dallek and Pollitz 2000; Miller 1998; Rodwin 2001). These interventions, drawing on grievance arrangements previously used in hospitals and other health care facilities (Charters 1993; Krajewski and Bell 1992), have been incorporated into contracts with managed care plans enrolling Medicare and Medicaid beneficiaries (Annas 1997; Perkins, Olson, and Rivera 1998) and have been made a part of many states’ efforts to regulate managed care (Miller 1997; Noble and Brennan 1999). Regulations have a variety of forms, such as making third-party mediators available to assess appeals, requiring plans to notify their members about the scope and use of grievance procedures, or collecting from plans data on the number and nature of their enrollees’ grievances.

In view of this recognition of the importance of voice, we need to understand better the factors that discourage consumers from expressing their dissatisfaction effectively. A substantial literature in consumer sciences (virtually ignored in health policy circles) has documented those attributes of medical care that hinder the expression of grievances (Andreason 1984; Brown and Swartz 1984; Kolodinsky 1993; Kucukarslan and Pathak 1994; Singh 1989; Solnick and Hemenway 1992). These include the complexity of the care-giving process and the difficulties of pursuing complaints in the face of a debilitating illness (Day 1983). These barriers are likely to be worse for certain disadvantaged groups. The introduction of managed care may either facilitate or hinder voice, depending on how enrollees perceive and interact with their health plans (Rosenthal and Schlesinger 2002). But these implications of managed care have been little studied.

This article is the first comprehensive assessment of voicing in response to problems with health plans. A small body of research has examined the determinants of complaints for specific HMOs (Gresenz et al. 2002; Hemenway and Killen 1989; Solnick and Hemenway 1992). But these studies are largely atheoretical and not representative of either consumers or the health plans in which they are typically enrolled. In this article, we develop a theoretical framework and specific hypotheses regarding the characteristics of problems, patients, and settings that might inhibit effective voicing, drawing on the broader literature on consumer complaint behavior. Using a nationally representative survey of working-aged consumers, conducted at the end of 1999, we test these hypotheses and assess the impact of state regulations intended to enhance voice.

Specifically, we identify the factors related to consumers’ willingness and ability to voice their grievances to their health plan, as well as to pursue alternative forms of voicing. We also identify the factors associated with effectiveness of voice, as measured by consumers’ judgments about whether the problem in question had been satisfactorily resolved. We show that while voicing is a far more common response to problems than exit is, it is limited by a number of factors. Our findings also suggest that while past regulatory interventions have done little to increase the frequency of voice, they have made grievances more effective, at least in those jurisdictions in which citizens are aware of the laws. We conclude by discussing the implications for public policy, especially those that might improve the performance or fairness of grievance mechanisms.

The Behavioral Foundations of Consumer Grievances

The foundational studies of consumer voice established that although Americans have considerable experience with unsatisfactory goods and services (roughly 15 to 25 percent reporting an unsatisfactory purchase in the previous year), usually they do not complain, even when they suffered substantial losses of time or money (Andreason 1988). For the average service, consumers complained about slightly more than half of all serious problems. Fewer than half of these complaints led to a satisfactory resolution of the problem. Although much of this research was on services other than medical care or focused on medical care settings in other countries (most commonly the National Health Service [NHS] in the United Kingdom; see Klein 1973), we can extrapolate from these findings to generate hypotheses about how voicing to health plans might emerge in the United States.

The Behavioral and Contextual Determinants of Consumer Voice

The broader literature on consumer sciences has documented the circumstances under which dissatisfaction translates into expressed complaints. These can be grouped into three categories: (1) characteristics of the problem, (2) characteristics of the consumer, and (3) characteristics of the setting in which the problem occurred.

Characteristics of the Problem

Because consumers need to understand (or at least believe that they have some modicum of understanding) the problem before they can feel empowered to express their grievance, more complicated problems may inhibit voice.1 In addition to complexity, the literature cites three other factors:

  • The Severity of the Problem: All other things being equal, problems that have more serious consequences for the patient are more likely to induce a complaint (Kolodinsky 1992; Kucukarslan and Pathak 1994; Warland, Herrmann, and Moore 1984). For example, in their comprehensive study of consumer behavior, researchers from Center for the Study of Responsive Law found that for an array of services, 54.6 percent of all “strong problems” produced a grievance, compared with 38.6 percent of “weak problems” (Best 1981, 122).

  • Attributable Responsibility: Numerous studies have documented that consumers are more likely to complain to those whom they feel are responsible for the problem (Andreason 1988; Brown and Swartz 1984; Coyle 1999). One reason that complex problems limit voice is that it is more difficult to assign responsibility. However, recent research has established that consumers distinguish between responsibility and blame. That is, they may avoid complaining to parties that they see as blameworthy because they do not expect culpable parties to respond in a constructive manner (Rosenthal and Schlesinger 2002).

  • Persistent Problems: Given the costs and potential risks of expressing a grievance, consumers are likely to remain quiet when they judge a problem to be an isolated mistake or a transitory error (Kucukarslan and Pathak 1994). Conversely, when problems recur, they are more likely to stimulate a complaint (Miller and Sarat 1980–81).

Characteristics of the Patient

Because complaining is less likely when consumers cannot understand their situation, their willingness to complain and their ability to complain effectively may require a minimal threshold of cognitive capacity. Although this hypothesis has not been tested directly, there is considerable indirect but suggestive evidence. Studies have found a positive correlation between the propensity to voice and perceptions of self-efficacy (Kritzer, Vidmar, and Bogart 1991; Singh 1989) and a negative relationship with feelings of powerlessness (Mulcahy and Tritter 1998). Earlier we observed that illness, and particularly debilitating medical conditions, might sap time and energy that otherwise would be used to express dissatisfaction. In this case, studies have documented a direct linkage between illness and voice, demonstrating that consumers with more serious chronic health conditions are less likely to complain, even about problems that they view as blameworthy (Rosenthal and Schlesinger 2002). The literature also identified another five factors associated with an enhanced or reduced propensity to voice, some in rather counterintuitive ways.

Characteristics of the Setting

Three features of the setting in which consumers make decisions affect their propensity to voice. First, studies consistently show that complaining is more common when the dissatisfied consumers have extensive networks of family and friends (Coyle 1999; May and Stengel 1990; Rosenthal and Schlesinger 2002; Warland, Herrmann, and Moore 1984). These networks help the aggrieved party make sense of the situation. In addition, family and friends often act as advocates, particularly for consumers who are debilitated by poor health (Lim et al. 1998). In Britain's National Health Service, for instance, about half of all complaints are filed by a family member acting as an advocate for the patient (Webb 1995, 35).

Second, the possibility of exit may alter the propensity for voice. In his original conceptualization (1970), Hirschman assumed that exit typically trumped voice, so that constraints on exit were necessary in order for voice to be significant response to dissatisfaction.2 He later acknowledged a more complex and dynamic interaction between the two responses to dissatisfaction (Hirschman 1980). (See also Singh 1989 for a discussion of the dynamics of voicing.) For instance, when consumers worry about the potential for retaliation, the possibility of exit may actually increase their willingness to complain; that is, if necessary, they can “get away” if their complaints trigger a threatening response. Empirical research has found few consistent relationships between exit and voice in either employment (Borloff and Lewin 1997; Farrell 1983) or health care settings (Kucukarslan and Pathak 1994; Rosenthal and Schlesinger 2002). Some studies have documented a modest overlap between the two behaviors, suggesting that at least under some circumstances the two responses are complementary (Annendale and Hunt 1998; Best 1981; Feick 1987; Solnick and Hemenway 1992).

Finally, the extent of voicing is likely to depend on the availability of other actors who might become involved in the dispute. Third parties may play several different roles in the grievance process. First, they can provide an alternative to voicing directly to the culpable party, which is useful for consumers wishing to avoid confrontations (Kucukarslan and Pathak 1994). Third parties also can act as advocates, making the expression of dissatisfaction more effective (Charters 1993; Krajewski and Bell 1992; May and Stengel 1990). In medical care, this advocacy role can be assumed by physicians, employers, labor unions, or other agencies representing the enrollees’ interests (Miller 1998). Third, engaging additional parties can transform the complaint process into a more “public” event, engaging social norms that force the culpable party to acknowledge the harm that he or she has inflicted on the consumer (Bearden and Teel 1983; Rosenthal and Schlesinger 2002).

For third parties to assist in the complaint process, they must be made aware that a problem exists. But many aggrieved consumers keep their concerns to themselves. Empirical research suggests that complaints are seldom made to third parties, occurring in about 15 to 20 percent of all problematic consumer experiences (Best 1981; Feick 1987; Kritzer, Vidmar, and Bogart 1991; Kucukarslan and Pathak 1994).

The Particular Challenges of Voicing in Medical Care

This previous research suggests that a consumer's voice is the product of three sequential cognitive processes: assessment, attribution, and prediction. Assessment requires that consumers must judge whether their experiences have fallen short of reasonable expectations. Attribution involves determining the parties who were responsible for the problem. Finally, they must predict whether expressing their dissatisfaction will be worth their time and effort. In medical settings, circumstances influencing all three stages of consumer cognition often deter the expression of dissatisfaction.

Assessment of Experiences

Studies have documented that patients do not recognize many problems (Brennan, Sox, and Burstin 1996). Furthermore, even when consumers sense that something is wrong, they may hesitate to complain about services they poorly understand, fearing that their claims will be dismissed as foolish or unrealistic (Annandale and Hunt 1998). Patients may feel empowered only “to voice dissatisfaction about standards they could be expected to know something about, such as hotel services” (Mulcahy and Tritter 1998, 835).

Attribution of Responsibility

Attributions of responsibility in medical settings can also be more challenging than they are for other goods and services (Coyle 1999; Rodwin 2001). It is often difficult for even the most knowledgeable patients to determine whether their recovery (or lack thereof) had anything to do with the treatment that they received. And if they consider their treatment to be inadequate, it is difficult for patients to determine whether the blame rests with their physician, their insurer, the employer who purchased their insurance, or even themselves (Rosenthal and Schlesinger 2002). The fragmented patterns of authority in the American health care system make these judgments even more difficult than they are in countries with more unified systems of health care financing (Emanuel 1999).

Predicted Outcomes from Voicing

The costs of voicing in medical settings often seem to exceed the potential benefits. The most seriously ill consumers are those most likely to encounter problems with their health care or health insurance (Druss et al. 2000). These patients are most debilitated by their medical conditions, inhibiting their ability to complain effectively (Mulcahy and Tritter 1998; Rosenthal and Schlesinger 2002). Even those who are capable of expressing their dissatisfaction may doubt that their physician (Annandale and Hunt 1998; Hak 1994; Kaye and MacManus 1990) or health plan (Mechanic and Schlesinger 1996) would respond constructively to their complaints.

Taken together, these factors significantly inhibit voice in medical settings (Kolodinsky 1993). Even in Britain's simpler National Health Service (NHS), a system that since the early 1990s has explicitly promoted grievance rights through its Patient's Charter, many patients report being confused about how to present complaints (Webb 1995). The most extensive study of consumer voice in the NHS found that only 38 percent of problematic episodes led to any expressed dissatisfaction. The researchers concluded that “complaining is an atypical reaction to dissatisfaction and is rarely considered as an option by aggrieved patients … it reflects a perception of complaints as an expensive distraction from everyday problems and concerns” (Mulcahy and Tritter 1998, 828).

Although representative data from the United States are somewhat dated, they paint a similar picture. Voicing in response to serious problems is significantly less common for medical care (38.3%) than for all services (54.6%) or for other commonly used services, such as car repair (63%) (Best 1981). Some studies have found as few as 11 percent of American patients complain about the problems that they experience (Andreason 1988). Effective voice also appears more difficult in health settings. Of those who complained about problems with any kind of service, 43.9 percent reported a successful resolution of the dispute (Best 1981). But for health care, problems that had produced a complaint were satisfactorily resolved only about a third of the time.

Voicing and the Influence of Managed Care

The introduction of managed care may alter the expression of consumer grievances by affecting any of the preceding three stages. The limited research on this topic suggests that this influence depends on the nature of the health plan, its interaction with its enrollees, and the regulatory environment. Consequently, not only is the net impact of managed care difficult to predict, but it also is likely to vary across plans and geographic locales.

Assessment of Experience

The spread of managed care has been accompanied by expanded information for consumers, in the form of report cards and other standardized assessments (General Accounting Office 1994), which should enhance consumers’ ability to assess their personal experiences. Research suggests, however, that although consumers are able to make sense of report card measures related to costs or administrative concerns, they have a much harder time even understanding measures related to quality, let alone applying those measures to real-world events (Lubalin and Harris-Kojetin 1999). Consequently, these improvements in assessment are likely to be modest, particularly when consumers are forced to evaluate care in the context of a complex health care plan (Rodwin 2001).

Attributions of Responsibility

Conventionally, health care consumers have attributed responsibility for health outcomes to their physicians (and other health care professionals) and hold their insurance plans responsible for issues related to costs and paperwork (Rosenthal and Schlesinger 2002; Zaslavsky et al. 2000). Under these circumstances, we would expect aggrieved patients to complain to their health plan largely about administrative matters. Over the past decade, however, Americans have become more aware of the ways in which managed care plans can directly influence the quality of their medical care (Jacobs and Shapiro 1999). The implications for voice are ambiguous. On the one hand, enrollees having problems with their doctor could turn to their health plan for help. This would induce more grievances regarding the quality of medical care. On the other hand, once consumers recognize that plans can directly affect the quality of their treatment, they may suspect that the plan caused the problem. With multiple parties (the physicians, the plans, pharmaceutical companies) potentially held responsible for medical outcomes, it becomes much more difficult to determine whom to trust and whom to enlist as an ally in any effort to resolve a grievance. And if they are sufficiently bewildered, consumers will be unable to form any coherent attributions (Weiner 1995).

Although it is unclear whether managed care will increase or decrease consumers’ propensity to hold plans accountable for quality, it is apparent that the magnitude of these effects will depend on the type of health plan in which consumers are enrolled. Some forms of managed care are highly visible: all their physicians practice in the same office buildings, and they regularly contact their enrollees about their benefits, health promotion programs, or other administrative matters. Other forms of managed care, in contrast, are virtually transparent to consumers (Schlesinger 1997). Independent practice associations (IPAs) bring together physicians through contractual arrangements, with no shared office arrangements. They shape treatment through incentives or utilization reviews that affect their practice but are never observed by patients (Mechanic and Schlesinger 1996). The more visible the plan is, the more dramatically its role may alter consumer behavior with respect to voice.

Predicted Outcomes from Voicing

The introduction of managed care is likely to reduce the costs of consumer voice in several ways. First, enrollees may face fewer psychological barriers when complaining to a health plan than to a physician. Dealing with a plan representative can be less threatening psychologically, because there are smaller gaps in social status between patients and administrators than between patients and physicians. Unlike with doctors, administrators do not (at least in most circumstances) literally hold peoples’ lives in their hands. To anger an administrator with a complaint thus may seem less risky than angering one's physician (Andreason 1984; Annendale and Hunt 1998).

Second, for decades policymakers have attempted to use the infrastructure of managed care plans to establish grievance mechanisms that are visible and accessible to patients (Gresenz et al. 2002). Federally qualified health plans (which were once eligible for subsidies under the HMO Act of 1973) were required to have formal grievance mechanisms. Once these subsidies were phased out, many states stepped in to require that HMOs have grievance procedures (although these state requirements do not apply to ERISA-exempt plans) (Families USA 1998). In response to evidence that many enrollees had little understanding of this grievance process (Tapay, Feder, and Dallek 1998), some states required health plans to notify their enrollees about their grievance rights. The number of states mandating such notification grew from 17 in 1995 to 41 in 1999.

Although enrollment in managed care plans should therefore reduce the costs of complaining to one's health plan, the implications for the expected benefits of voice are less clear. On the positive side, a number of states now require that disputes between plans and their members be adjudicated through third-party mediation. States with these requirements have grown from three in 1990 to six in 1995 and to 30 by 1999. By providing a forum for an impartial hearing, these third-party reviews ought to increase the perceived odds of winning an appeal and thus stimulate more grievances. But in nearly all these states, enrollees must first exhaust their opportunities for appeals within the health plan before engaging external mediators (Dallek and Pollitz 2000; Tapay, Feder, and Dallek 1998). This may weaken the impact of these regulations.

On the negative side, over the past ten years managed care plans have increasingly become targets of public mistrust (Mechanic 1998). Indeed, the states’ requirements for external mediation were a response to several well-publicized cases in which the grievances processes established by health plans appeared to have been biased in favor of the plan (Pollitz, Dallek, and Tapay 1998).3 This “backlash” against managed care has left many Americans skeptical of the motives of those who run the plans (Blendon et al. 1998; Jacobs and Shapiro 1999). The more that enrollees doubt that their complaints will receive a favorable response, the less likely they will be to voice their dissatisfaction. Since external mediation typically requires enrollees first to exhaust their rights of appeal within the plan, these perceptions may also reduce the extent to which grievances reach the purview of these third-party mediators.

Implications for Voice to Health Plans: Some Testable Hypotheses

Our review of the literature on consumer behavior, its application to medical care, and our extrapolation to settings involving managed care plans all suggest factors that may promote or discourage the expression of dissatisfaction to health plans. In our later statistical analyses, we measure and control for each of these factors. Some sets of factors, however, stand out as potentially being most relevant to our understanding of consumer voice or as being most important for designing policy to enhance voice. We view these as our five core hypotheses to be tested in the empirical models. The first four come from past research on the predictors of consumer complaints.

Hypothesis 1: Compounded Vulnerability

Past studies have demonstrated that certain groups of enrollees, including racial minorities (Phillips, Mayer, and Aday 2000) and people with serious health conditions (Druss et al. 2000), tend to be less satisfied with their care in managed health plans. Research on complaint behavior suggests that these groups are also less likely than other enrollees to voice their dissatisfaction, making it more difficult for health plans to respond to their problems.

Hypothesis 2: The Role of External Support

Given the challenges and costs of voice, particularly in health settings in which the sickest patients have the most problems, the propensity and effectiveness of voice to health plans are enhanced by access to a supportive third party, such as a family member, employer, physician, or third-party mediator.

Hypothesis 3: The Complex Consequences of Severe Medical Problems

Given the costs of voice, minor problems are not likely to trigger a complaint to one's health plan. But severe care-related problems may also discourage complaint, either because the patient is too sick (and his family too busy with his care) to take the time and effort to express dissatisfaction or because the problem is too complicated to attribute blame precisely.

Hypothesis 4: Perceptions of Health Plans

Our review of the potential implications of managed care practices shows that their likely impact depends on the enrollees’ perception of their health plan. The more easily the enrollees can understand the plan, the more likely they are to regard it positively; and the more influence that they believe the plan has over the delivery of medical care, the more often enrollees will express their grievance to the plan.

There is much less empirical evidence about the effectiveness of voice than on the determinants of voice, and little of this evidence applies to medical settings (for exceptions, see Best 1981; Kolodinsky 1993). Some of the factors that have been shown to affect the frequency of complaining may also be expected to influence the effectiveness of voice. Consumers who are less apt to voice their concerns can be expected to pursue their complaints less aggressively because they feel less empowered or entitled to make their claims (Rodwin 2001). But in other cases, factors that may increase the propensity to voice may decrease its effectiveness.

One example involves the economic incentives facing managed care plans. These incentives may encourage plans to respond to some sorts of complaints more effectively than others. Because payments to plans are adjusted only imperfectly (and often not at all) to the health status of enrollees, plans can expect to lose money on enrollees who have complex and chronic health needs (Kronick and de Beyer 1999; Schlesinger 1986). It is thus in the plans’ financial interests not to respond to complaints from enrollees with these conditions and instead to encourage them to switch to another plan (Schlesinger 1997). This leads to our fifth core hypothesis.

Hypothesis 5: Chronic Illness and the Effectiveness of Voice

Enrollees who have multiple problems, more chronic health conditions, and persistent health problems have a harder time complaining effectively, in part because health plans can be expected to respond less constructively to their complaints.

Next we describe the methods we used to test these hypotheses and provide the first assessment of consumer voice for a nationally representative sample of working-aged Americans who are covered by health insurance.

Methods

We begin by describing the survey used to test our hypotheses and then discuss the dependent variables, the explanatory variables, and the specifications of our regression models.

Data Sources

The data we analyzed came from the Kaiser Family Foundation's National Survey on Consumer Experiences with Health Plans. This survey (conducted by telephone in both English and Spanish) used a nationally representative, randomly selected sample of 2,500 adults aged 18 to 64 who had health insurance other than Medicare. The survey was conducted in late 1999. It achieved an overall response rate of 53 percent (producing a sampling error of two percentage points for the entire sample). The respondents were asked a series of 14 questions to identify any problem that they may have had with their health plan in the previous 12 months. Just over half (1,278) of the respondents had experienced at least one problem that they attributed in part to their health plan. This is slightly higher than the annual prevalence of problems (42%) reported in an earlier survey of enrollees in managed care plans (Schauffler et al. 2001).

Dependent Variables

We examined three different dependent variables: two measures of voice and one measure of the effectiveness of voice. The primary voice variable identified whether the respondent had complained to his or her health insurance plan. The complaint could have been either oral or written. This information was collected from all respondents who reported that they had had at least one problem with their health plan but they were asked follow-up questions about only the most recent problem. (Our analyses therefore focused on this most recent problem, although we controlled for the total number of problems and the nature of the other problems reported.) Among those who had had a problem, 55 percent voiced their dissatisfaction to their plan.

To explore other avenues for voice, we also looked at whether the respondents had lodged a complaint with their employer concerning the problem with their health plan. Since most individuals had employer-based coverage, employee-benefit administrators represented a potential alternative venue for voicing.4 Among those who had experienced a problem with their health plan, 13 percent had expressed concerns to their employer.

The second stage in the analysis was designed to assess the effectiveness of voice. Here we determined whether the respondents felt that the problem with the health plan had been “resolved to [their] satisfaction” (measured as a dichotomous variable). To assess the effectiveness of voice, we included our measure of voicing as an explanatory variable in the model on problem resolution. If voice does matter, we should observe a positive and statistically significant relationship between the expression of dissatisfaction and the probability that the problem was resolved.

Explanatory Variables

Consistent with the preceding discussion, the independent variables in the regressions fell into three different categories: characteristics of the problem, characteristics of the enrollee, and characteristics of the setting.

Characteristics of the Problem

The first group of independent variables characterized the nature of the problems reported by the respondents. Three variables captured different aspects of severity: (1) whether the most recent problem delayed their medical care for more than two weeks, (2) whether the problem in question resulted in a somewhat or very serious decline in health, and (3) whether the respondent incurred $200 or more in out-of-pocket spending. The percentages of cases passing these three thresholds of severity were 4.3, 4.8, and 8.4 percent, respectively.

The second set of problem characteristics pertained to the nature of the problem and the corresponding domain of responsibility. Past research suggests that consumers classify health problems as being in the domains of health care and health administration (Rosenthal and Schlesinger 2002; Zaslavsky et al. 2000), holding the health plans more responsible for administration. We would thus expect more voice to plans for administrative problems. We therefore grouped the problems into administrative-related or care-related categories. Respondents were asked a series of questions to identify “problems that you may have had with your main health plan” in the previous 12 months. Those related to care included (1) delays in receiving care or treatment, (2) the plan's denial of care or treatment, (3) the plan's refusal to cover a particular treatment needed by the enrollee, (4) difficulty getting an appointment with a doctor, (5) difficulty getting referrals to see a medical specialist, (6) a forced change of doctors or a doctor's being dropped from the plan, and (7) the inability to get a specific medication. Problems related to administration included (1) misunderstandings over which services the plan covered, (2) difficulty selecting a doctor in the plan, (3) insensitivity or lack of help by the administrators or other plan staff, (4) impediments to language or communication with plan staff, (5) issues with billing or payment for services, and (6) difficulty getting someone on the phone to answer questions.

Both care and administrative indices were internally coherent, with alpha coefficients of 0.70 and 0.69, respectively. Nonetheless, it is important to recognize the ambiguity in these groupings of problems. For example, misunderstandings about coverage could have had implications for obtaining care, and so forth. We therefore experimented with modified categorizations, but these did not alter the results reported later.

We next constructed dichotomous variables to identify those respondents who experienced problems associated only with care, problems that involved both care and administration, or problems of an unspecified nature (administrative problems were the omitted comparison group). Among those who had a problem, 22 percent had care-only problems; 28 percent, administrative-only problems; 49 percent, problems of both types; and 3 percent, unspecified problems.

Because past consumer research suggests that voice (and effective voice) is discouraged when consumers do not understand their experiences, we also tried to construct a measure of the enrollees’ ability to understand their problem. On this survey, respondents were asked whether their (most recent) problem was related to “paperwork” or “medical treatment.” We created a dichotomous variable identifying those that could not so classify their reported problem (giving new meaning to the term dummy variable).

The propensity to complain is positively related to the perceived probability that the culpable party will respond constructively. Among health plans, we expect that coverage issues (as identified by respondents) will be resolved less often to the enrollee's satisfaction. In these cases, the plan has a plausible legal and ethical rational for not responding to the complaint, since coverage provisions are typically specified by contractual provisions between the plan and the purchaser (Daniels and Sabin 1998). If enrollees follow this same logic, they also may be less inclined to complain about such issues. The respondents reported that 18 percent of the problems were related to coverage.

The final problem-related characteristic was a count of the total number of problems that the respondent reported having with their health plan in the last 12 months. When enrollees report multiple problems, they are likely to experience a continuing conflict with the health plan. Because past research indicates that persistent problems stimulate voice, we expected that those consumers with many problems would be more willing to express their dissatisfaction. As we hypothesized earlier, however, health plans may be less likely to respond constructively both because repeated problems are most common for enrollees with complex, costly illnesses and because enrollees with this track record may be labeled as chronic complainers.

Characteristics of the Enrollee

Several individual characteristics can affect voice and the response to voicing. Because voicing is such a complex task, we constructed a variable to control for the respondents’ cognitive capacity. This was measured by the number of complex tasks that were viewed as stressful, from a list including “doing your taxes,” “dealing with your auto mechanic or the place that services your vehicle,” and “dealing with your auto insurance company.” (Respondents were asked to report on a four-point scale ranging from a lot of stress to no stress at all.) Seventy-one percent reported that none of these tasks was stressful; 23 percent reported stress for one task; 5 percent for two; and 0.5 percent for all three. Although the validity of this measure has not been assessed, it provides a plausible proxy for a difficult-to-measure characteristic of the respondents.

To the extent that ill health debilitates enrollees, we expected that it would also inhibit their voice. Although the effect on voicing should hold for all forms of illness, we expected that problem resolution in particular would be reduced for chronic illnesses because plans have a financial incentive to disenroll such enrollees. We therefore used two measures of the respondents’ health status. The first of these was a count of the chronic conditions the respondents reported (they were asked if they had ever been diagnosed with heart disease [including high blood pressure], stroke, emotional or mental illness, cancer, or diabetes). In addition, self-reported health status was included, measured in the standard five-level format (“poor” to “excellent”).

Past research suggests that voice is positively related to education but negatively related to income. Effective voice should be positively related to both aspects of socioeconomic status. Education was measured as an ordinal variable. In most past studies of complaint behavior, educational attainment measured both social status and cognitive capacity. In this analysis, education can be interpreted as social status, since we had another measure that captured cognitive capacity. With education as a measure of status, we viewed income (also measured as an ordinal variable) as a measure of the availability of resources that could support voicing.

Other demographic factors have been linked to voicing. Age was measured as a continuous variable, and sex, as a dummy variable (with male as the excluded reference group). Race was coded as a series of dichotomous variables (with white as the omitted group), with black, Asian, and those of mixed or unknown race included as explanatory variables. We measured Hispanic ethnicity as a separate dichotomous variable. The distributions of these sociodemographic variables are presented in table 1. Again, we experimented with different specifications for these variables (e.g., categorical as opposed to continuous, multiple dichotomous variables as opposed to categorical variables). Since these did not alter the results, they are not reported here.

TABLE 1.

Descriptive Statistics: Study Sample versus United States (Persons Aged 18 to 64)

Study Sample (%) United States (%)
Geographic Regiona
New England 6 6
Mid Atlantic 16 11
South Atlantic 18 19
East North Central 8 20
East South Central 19 6
West North Central 4 8
West South Central 1 10
Mountain 6 10
Pacific 12 10
Demographic Characteristicsb
Age
18–30 23 22
31–49 51 46
50–64 26 32
Race
White 78 71
Black 9 13
Hispanic 8 12
Asian 2 2
Income
<$10,000 4 31
$10,000–$19,999 7 12
$20,000–$29,999 13 19
$30,000–$50,000 26 23
$50,000–$75,000 19 6
$75,000–$99,999 10 4
$100,000+ 9 3
Education
<High School 9 14
High School 56 60
Bachelor's Degree 19 18
Post Graduate Degree 11 8
Health Status
Excellent 29 36
Very Good 38 31
Good 23 24
Fair 8 7
Poor 2 2
Open in a new tab
a

Kaiser Family Foundation State Health Facts Online.

b

Bureau of Labor Statistics, Current Population Survey 1997.

Two other characteristics of enrollees may affect voicing to plans, although neither has been studied previously in this context. The length of time the respondent has been enrolled in the health plan can affect voice, but the direction of the association is uncertain. Voice may be suppressed if long membership leads to socializing enrollees into the plan's norms and practices. Conversely, voice may be enhanced if prolonged enrollment leads to better understanding of the plan's administrative procedures (Ross, Wheaton, and Duff 1981). The survey contained little information about the duration of enrollment. We could distinguish only between those who had been enrolled for at least one year and more recent enrollees: 14 percent of respondents had been enrolled for less than one year.

The final characteristic pertained to the respondents’ general attitude toward consumer activism. Past research suggests that voicing is positively related to consumer activism. We assessed the respondents’ general propensity to complain according to their assessment of whether it had proved to be generally “worth their time and effort to complain to someone” when having a problem with a major purchase. Two-thirds of the respondents indicated that it was generally worth their time and effort to complain. We expected those who so responded to be more likely to express their dissatisfaction with their health plan.

Characteristics of Setting

Based on previous research, we hypothesized that voicing complaints to health plans would be shaped by four features of the setting: (1) the availability of social support, (2) the opportunities for exit, (3) the perceived characteristics of the plan, and (4) the regulatory environment. Regarding social support, respondents who had experienced a problem with their health plan were asked whether they had discussed the matter with their family and friends; if they indicated that they had, they were asked how useful it had been to do so (responses were on a four-point scale ranging from “not at all useful” to “extremely useful”). The social support measure was included in the model as an ordinal variable.

Hirschman (1970, 1980) hypothesized that exit would reduce the propensity to voice. We captured two types of exit in this study, using data indicating whether the respondents had changed their health plans (5% had done so) or their provider (6% had done so) in response to their most recent problem with their health plan.5

The propensity to voice grievances to a health plan may depend on the enrollees’ understanding of the plan as well as the expected response from the plan. The more readily that respondents could identify the managed care practices that shaped their medical care, the more we expected them to hold the plan responsible for problems with their care. The survey included questions on four features that could make plans visible to enrollees: (1) copayments for out-of-network services, (2) the requirement that enrollees choose a specific primary care provider, (3) gatekeeper provisions to manage referrals to specialists, and (4) prior authorization before seeing an out-of-network provider. These were combined into an index of plan visibility that ranged from zero to four visible features. The index was internally coherent (α= 0.70). Forty-six percent reported two or fewer features, and 32 percent reported that their health plan contained all four practices.

Because past research suggests that voice is more likely when complainants expect a favorable response, enrollees may express more grievances if they perceive that their health plan is administered with their interests in mind. To gauge this perception, we used a series of three questions that asked whether the plan was generally “friendly and helpful,” “easy to get a hold of,” and “fair to health plan members like yourself.” An index was constructed measuring the number of positive attributions for these three questions. It was highly coherent (α= 0.86). About half of all respondents reported a favorable impression on all three measures in the index.

As we mentioned earlier, external regulations may also shape the perception that the plan will respond favorably. We assessed the three regulations most salient to voicing and to the plan's response to complaints: (1) mandated external mediation or independent external review, (2) the requirement that health plans disclose their internal grievance and complaint procedure to both enrollees and potential enrollees, and (3) state initiatives that establish an ombudsman program or comparable forms of support services. Regulations (identified by the researchers in a companion study) were assigned to the respondents based on their state of residence. We counted only those regulations that existed as of 1998 to allow for sufficient time for implementation and for consumers to learn about the requirements.6

These three regulations could affect behavior in two ways. First, they could alter the performance of plans directly, since they are under greater oversight from state officials (a sort of Hawthorne effect). Second, regulations could affect the behavior of consumers by changing the perceived costs and benefits of voicing their dissatisfaction. This latter influence could occur only if enrollees were aware of the regulations. Unfortunately, we were able to assess an enrollee's awareness for only the requirement for external mediation. Survey respondents were asked: “As far as you know, in your state do you have the right to appeal to the state or to an independent medical expert if your health plan refuses coverage for medical services you think you need?” Forty-two percent of the respondents reported that their state had such a regulation. Combining the survey data with information on state regulations, we constructed an interaction term measuring both the existence of a requirement for external review and the respondents’ recognition of the regulation. Four categories were thus possible: (1) the regulation existed and the respondent knew about it (25% of all respondents), (2) the regulation existed and the respondent did not know about it (30%), (3) the regulation did not exist and the respondent believed that it did (18%), and (4) the regulation did not exist and the respondent knew that it did not exist (27%). (This fourth category served as the omitted comparison group in the regression models.) These findings suggest that many respondents are poorly informed about managed care regulations.

Distinctive Factors Shaping the Effectiveness of Voice

Many factors that promote (or inhibit) voice can also be expected to make voice more (or less) effective. However, some factors seem unlikely to carry over in this manner. Although the understanding of a health plan by its enrollees should encourage voice, there is no reason to expect it to make voice more effective. Enrollees who are likely to complain in other settings will not necessarily complain more effectively about medical care, given its unique attributes.

Other variables may be endogenous and thus not appropriate predictors for a model of the effectiveness of voice. Plans may be perceived as friendly to their enrollees because they are more responsive to complaints, not the other way around. Given the wording of the measure regarding family support, respondents might have reported that talking to their family about the problem had been useful because the problem was subsequently resolved, whether or not the family support actually helped in the resolution. Because of the ambiguities in interpreting these two variables, we omitted them from the model predicting effectiveness of voice.

Conversely, the efficacy of voice to the health plan may be enhanced by other factors not included in the voicing regression. Contacting third parties is a prime example—we would expect their advocacy to make consumer voice more effective. We explore the influence of two potential advocates: employers (contacted by 16% of respondents with a plan-related problem) and physicians (contacted by 28% of respondents with problems).

Statistical Methods

We used logistic regression to identity predictors of voice and the effectiveness of voice. In order to make the results more readily interpretable, we report them in terms of odds ratios for the independent variables.

Assessing the impact of voice on the resolution of plan-related problems required somewhat more sophisticated statistical methods. Simply including the expression of voice as an explanatory variable in the regression on problem resolution might lead to a biased estimate. That is, the failure to resolve a problem in the first instance might actually induce subsequent grievances, creating a spurious negative relationship between voice and problem resolution. This statistical artifact would bias downward the measured relationship between voice and problem resolution. To avoid this bias, we estimated a two-stage regression model. The voice regression represents the first stage. We then used the predicted value of voice as an explanatory variable in the second stage regression predicting problem resolution. The coefficient on predicted voice provided an unbiased estimate of the relationship between voice and problem resolution.

To assess the impact of regulations on the efficacy of voice, we also used a more sophisticated technique. We estimated two separate regressions for problem resolution, one for respondents who had voiced their grievances and the other for respondents who had not. This allowed us to explore whether the regulations intended to enhance voice might also benefit enrollees who did not file grievances. All our analyses used SAS software.

Results

Among those who experienced a problem with their health care or health benefits (n= 1,278), 55 percent voiced a complaint to the plan, and 13 percent expressed their dissatisfaction to their employer. Although voicing was therefore not uncommon, a number of people who experienced serious problems failed to express their concerns to their health plan (table 2). Even for the most serious problems, fewer than 15 percent of the respondents filed a formal grievance with the plan. Fewer than half of those who voiced their concerns had their problem satisfactorily resolved. To better understand the origins of these outcomes, first we explore the determinants of voice and effective voice through several stages of regression models. We conclude by examining other factors related to resolving problems with health plans.

TABLE 2.

Prevalence of Voicing by Seriousness of Problem

Frequency of Voicing: Percentage of Those with Serious Problem Who
Measures ofSeriousness of Problem Contacted Plan (%) Filed Formal Appeal/Grievance (%)
Problem Caused Serious Decline in Health (n= 122) 49.2 16.4
Problem Significantly Delayed Treatment (n= 108) 50.0 7.4
Problem Caused Large Out-of-Pocket Costs (n= 108) 62.2 13.4
More than Three Problems in the Past Year (n= 434) 61.8 11.0
Open in a new tab

Factors Associated with a Greater Propensity to Voice

The results from the regression identifying correlates of voicing are presented in table 3.

TABLE 3.

Determinants of Voice

Voice to the Plan Voice to the Employer
odds ratio p-value odds ratio p-value
Characteristics of the Problem
Problem Severity: Delays 0.778 0.2697 0.766 0.411
Problem Severity: Health Decline 0.677 0.095 1.067 0.833
Problem Severity: Cost 1.388 0.071 1.514 0.067
Problem Attribution: Related to Care 0.343 <.0001 0.653 0.163
Problem Attribution: Related to Care and Administration 0.638 0.017 1.297 0.316
Problem Is Difficult to Understand 0.219 0.000 0.170 0.087
Problem Is Less Discretionary (Coverage) 0.694 0.029 1.103 0.671
Problem Repeated over Time 1.224 <.0001 1.024 0.632
Characteristics of the Enrollee
Enrollee Is Challenged by Complexity 0.873 0.170 1.122 0.392
Self-Reported Health Status 0.981 0.786 1.166 0.124
Number of Chronic Conditions 0.909 0.389 1.403 0.019
Socioeconomic Status: Education 1.096 0.289 1.042 0.733
Socioeconomic Status: Income 1.058 0.233 1.067 0.323
Gender: Female 1.173 0.218 0.710 0.056
Age 1.001 0.838 0.998 0.830
Race/Ethnicity: Black 0.568 0.005 0.700 0.243
Race/Ethnicity: Hispanic 0.601 0.247 1.929 0.233
Race/Ethnicity: Asian 0.540 0.103 0.734 0.590
Race/Ethnicity: Other 1.230 0.819 2.975 0.242
Consumer Activism 0.975 0.724 1.078 0.451
Experience with Health Plan 0.950 0.778 0.727 0.177
Characteristics of the Setting
Social Support 0.963 0.563 1.353 0.000
Exit: Plan 0.700 0.238 0.911 0.815
Exit: Physician 0.815 0.437 1.177 0.624
Other Avenues for Voice: Physician 0.504 <.0001 1.109 0.586
Health Plan Visibility 0.959 0.390 1.131 0.082
Health Plan Benevolence 1.455 <.0001 0.925 0.331
Third-Party Mediation: Enrollee Knows about Regulation 0.932 0.696 0.896 0.676
Third-Party Mediation: Enrollee Doesn't Know about Regulation 0.890 0.548 1.303 0.317
Third-Party Mediation: Enrollee Mistakenly Thinks Regulation Exists 0.747 0.085 1.169 0.512
Regulation: Requirement of Grievance Notification to Enrollees 0.865 0.258 1.019 0.915
Regulation: Established Ombudsprogram 1.170 0.317 0.746 0.196
Open in a new tab

Voicing to the Health Plan

Overall, the problem's characteristics displayed the most consistent association with voicing to the health plan. Some aspects of the context also were associated with voice, but less consistently. In contrast, the respondents’ characteristics (with the exception of race) generally did not predict voice to the health plan.

The problem's characteristics were strongly related to voicing to the health plan. For example, if the problem was one in a series of problems, the probability of voice increased substantially. But not all the characteristics of the problem worked in the hypothesized direction. Our measures of problem severity illustrate this. Problems that substantially delayed treatment did not appear to induce complaints. Problems that added substantially to enrollees’ out-of-pocket costs were associated with more frequent voice. By contrast, enrollees whose health suffered because of the problem were only two-thirds as likely to voice to the plan as were those whose health did not suffer.

Other features of the problem were related to the propensity to voice dissatisfaction in the predicted manner. Problems attributed purely to care received were much less likely to produce a grievance to the health plan than those that were purely administrative (OR = 0.34). This pattern persisted even if the respondent had problems with both care and administration; these problems were also less frequently reported than those related to administration alone. Complex problems (those that the enrollee was unable to classify as related to either care or paperwork) were only one-fifth as likely to result in voicing to the health plan, compared with other problems. Problems that involved coverage decisions by the health plan were also less likely to result in consumer voice (OR = 0.69).

Only a few characteristics of the settings were significantly related to the expression of consumer dissatisfaction to the health plan. When enrollees viewed the health plan as being on their side (interacting with them in a friendly and respectful manner), they were significantly more likely to voice their complaints. But enrollees who discussed their problem with their physician (a potential advocate on their behalf) were only one-half as likely to voice to the health plan themselves (OR = 0.50). Other contextual factors did not predict consumer voice to the health plan. Most strikingly, none of the state regulatory initiatives intended to facilitate voice appeared to increase the prevalence of grievances, even among enrollees who were aware of the regulations.

The only significant enrollee characteristic associated with voicing to the health plan was the enrollee's race. Compared with whites, black and Asian enrollees were approximately one-half as likely (OR = 0.57 and OR = 0.54, respectively) to voice their dissatisfaction to their health plan, although the result for Asians was only marginally significant. Enrollees’ cognitive capacity, health status, socioeconomic status, gender, and age were not related to the prevalence of complaints to the plan.

Alternative Avenues for Voice

These results suggest that voice to the plan is inhibited by a number of conditions: complex problems, problems seen as outside the purview of the health plan, problems with severe health consequences, enrollees from minority groups, and health plans that have less friendly relations with their enrollees. Do alternative avenues for expressing grievances provide recourse for those unable or unwilling to complain to their health plan?

Overall, predictors of voice to the employer did differ somewhat from predictors of voicing to the health plan. For example, some of the problem's characteristics that displayed a strong significant relationship with voicing to the health plan were unrelated or only marginally related to voicing to the employer. However, among problem characteristics that predicted voice to the employer, the magnitudes and direction of effect were similar to voicing to the health plan. For instance, experiencing a financial impact because of the problem was positively related to voicing to both one's employer (OR = 1.51) and one's health plan (OR = 1.38). More complex problems continued to deter voice, with enrollees facing such problems being only one-fifth as likely to complain to their employer as those with less complex problems.

Some of the enrollees’ characteristics that predicted voice to the employer differed from those associated with voice to the health plan. A diagnosis involving several chronic diseases was associated with an odds ratio 1.4 times larger in voicing to the employer. Men were more likely to voice to their employer than women were. Other characteristics generally did not predict the likelihood of voicing to the employer.

Different characteristics of the context in which voice occurs predicted voicing to the employer. The presence of social support increased the likelihood that enrollees would complain to their employer (OR = 1.3), although this was not associated with complaining to the plan. The visibility of the health plan's characteristics also increased the likelihood of complaining to the employer, although the relationship was marginally significant. Regulations designed to enhance voice were not related to expressing dissatisfaction to employers.

Effectiveness of Voice

We explore the efficacy of voicing in several different regression models. In the first set of models (table 4), we determine whether voicing to the plan increases the probability that a problem will be successfully resolved. In the second set of models (table 5), we compare the features associated with problem resolution among enrollees who voiced their complains with the factors affecting those who had not voiced.

TABLE 4.

Problem Resolution

With Actual Voice
 With Predicted Values
odds ratio p-value odds ratio p-value
Characteristics of the Problem
Problem Severity: Delays 1.463 0.124 1.471 0.118
Problem Severity: Health Decline 0.320 <.0001 0.322 <.0001
Problem Severity: Cost 0.337 <.0001 0.340 <.0001
Problem Attribution: Related to Care 0.974 0.894 1.380 0.210
Problem Attribution: Related to Care and Administration 0.680 0.046 0.777 0.214
Problem Is Difficult to Understand 0.327 0.002 0.483 0.078
Problem Is Less Discretionary (Coverage) 0.315 <.0001 0.337 <.0001
Problem Repeated over Time 0.783 <.0001 0.749 <.0001
Characteristics of the Enrollee
Enrollee Is Challenged by Complexity 1.010 0.920 1.045 0.670
Self-Reported Health Status 1.170 0.034 1.167 0.037
Number of Chronic Conditions 1.214 0.097 1.248 0.060
Socioeconomic Status: Education 1.039 0.664 1.000 0.996
Socioeconomic Status: Income 0.962 0.425 0.942 0.226
Gender: Female 1.165 0.251 1.092 0.518
Age 1.002 0.768 1.001 0.824
Race/Ethnicity: Black 1.299 0.214 1.589 0.046
Race/Ethnicity: Hispanic 1.128 0.791 1.337 0.526
Race/Ethnicity: Asian 0.943 0.888 1.087 0.844
Race/Ethnicity: Other 4.580 0.134 4.003 0.175
Experience with Health Plan 1.315 0.147 1.288 0.181
Characteristics of the Context
Exit: Plan 1.302 0.417 1.454 0.254
Exit: Physician 1.322 0.317 1.393 0.235
Voiced to Plan 1.255 0.101 5.315 0.020
Other Avenues for Voice: Employer 1.089 0.669 1.120 0.570
Other Avenues for Voice: Physician 1.092 0.543 1.338 0.097
Third-Party Mediation: Enrollee Knows about Regulation 1.649 0.007 1.651 0.007
Third-Party Mediation: Enrollee Doesn't Know about Regulation 1.237 0.283 1.255 0.252
Third-Party Mediation: Enrollee Mistakenly Thinks Regulation Exists 1.028 0.876 1.115 0.541
Regulation: Requirement of Grievance Notification to Enrollees 1.108 0.435 1.139 0.327
Regulation: Established Ombudsprogram 0.896 0.501 0.859 0.354
Open in a new tab

TABLE 5.

Problem Resolution: Voicers Compared with Those Who Did Not Voice

Voicers
 Non-Voicers
odds ratio p-value odds ratio p-value
Characteristics of the Problem
Problem Severity: Delays 1.67 0.1551 1.41 0.3341
Problem Severity: Health Decline 0.415 0.0349 0.255 0.0016
Problem Severity: Cost 0.443 0.0019 0.19 <.0001
Problem Attribution: Related to Care 1.093 0.7616 0.882 0.6619
Problem Attribution: Related to Care and Administration 0.676 0.115 0.624 0.1585
Problem Is Difficult to Understand 0.405 0.2917 0.276 0.0028
Problem Is Less Discretionary (Coverage) 0.295 <.0001 0.315 <.0001
Problem Repeated over Time 0.776 <.0001 0.792 0.0035
Characteristics of the Enrollee
Enrollee Is Challenged by Complexity 0.946 0.6978 1.101 0.5428
Self-Reported Health Status (Excellent Scored High) 1.232 0.0447 1.113 0.3432
Number of Chronic Conditions 1.046 0.7895 1.424 0.04
Socioeconomic Status: Education 0.905 0.4155 1.273 0.0715
Socioeconomic Status: Income 0.974 0.7022 0.952 0.4957
Gender: Female 1.061 0.7442 1.272 0.2539
Age 1.008 0.3641 0.993 0.418
Race/Ethnicity: Black 0.922 0.7955 1.527 0.1636
Race/Ethnicity: Hispanic 0.72 0.6248 1.502 0.5342
Race/Ethnicity: Asian 0.645 0.4845 1.217 0.7309
Race/Ethnicity: Other 6.768 0.1441 1.784 0.7505
Experience with Health Plan 1.555 0.1049 1.082 0.7855
Characteristics of the Setting
Exit: Plan 0.84 0.7107 1.672 0.295
Exit: Physician 0.886 0.7571 2.08 0.0937
Other Avenues for Voice: Employer 1.205 0.4595 1.178 0.6514
Other Avenues for Voice: Physician 0.683 0.0668 1.514 0.0508
Third-Party Mediation: Enrollee Knows about Regulation 1.989 0.0063 1.373 0.2777
Third-Party Mediation: Enrollee Doesn't Know about Regulation 1.204 0.49 1.224 0.5175
Third-Party Mediation: Enrollee Mistakenly Thinks Regulation Exists 1.35 0.2104 0.72 0.2236
Regulation: Requirement of Grievance Notification to Enrollees 1.129 0.5022 1.072 0.734
Regulation: Established Ombudsprogram 1.046 0.8337 0.713 0.208
Open in a new tab

Does Voice to Plan Enhance Problem Resolution?

Table 4 presents the results from two different specifications of the model. The first uses reported voice to plan as the explanatory variable, and the second uses the predicted values of voice derived from the regression in table 3. As these results show, voicing to the plan initially appears not to be related to problem resolution. But the voice coefficient in the first model reflects the downward bias described in the methods section. When predicted voice is substituted, expressing dissatisfaction to one's health plan is very strongly associated with problem resolution (OR = 5.3). Similarly, complaining to one's physician is not related to problem resolution initially, but positive and marginally significant association with resolution is found using the predicted voice specification.

These models also reveal that while regulations intended to encourage voice do not increase its frequency, they do increase the chance that problems will be successfully resolved. Those who were aware that they resided in a state mandating third-party mediation of disputes were nearly twice as likely to report successful problem resolution (OR = 1.7) (table 4). However, simply residing in states with these mandates was not associated with more effective voice. Only the combination of supportive regulations and knowledge of the regulation increased the likelihood of effective voice.

Problem Resolution with and without Voicing

Comparing problem resolution for those who voiced and those who did not further illustrates how voice shapes the process of problem resolution (table 5). Although problem complexity was not related to resolution among those who did voice, among those who did not voice, problem complexity was associated with significantly reduced problem resolution. Enrollees’ health status was associated with better problem resolution, but in different ways for each of the two subsets of respondents. Among those who voiced, enrollees with better self-reported health reported greater problem resolution. Among those who did not voice, those with a greater number of chronic conditions reported greater problem resolution.

The effect of setting on problem resolution also differed between voicers and nonvoicers. Complaining to a physician was negatively associated with problem resolution among those who did voice to the plan but was positively associated with problem resolution among those who did not voice. This indicates that the role of physicians in the process of problem resolution differs for the two groups. More specifically, for enrollees who are unwilling or unable to express their grievances directly, turning to their physician as an advocate seems to be an effective strategy. But a physician's advocacy does not appear to have the same positive effects when enrollees are also voicing their own concerns to the plan. That is, the two forms of voice may interfere with each other. (Although these results may also be an artifact of our inability to isolate the time sequence. Enrollees whose physicians were unsuccessful as advocates may subsequently act on their own to rectify the situation.)

Regulation also has a different effect on problem resolution for voicers and nonvoicers. Knowing about their state's mandated third-party mediation was associated with a greater likelihood of successful problem resolution for those who voiced (OR = 2.0) but had no effect on those who did not express their dissatisfaction to the plan.

Other Predictors of Satisfactory Problem Resolution

Although we are focusing here on the correlates and consequences of voicing, we shall also note some of the other correlates of problem resolution (table 4). Characteristics of the problem were most noteworthy in this respect. Problems that seriously delayed care were more often associated with a satisfactory resolution. This finding contrasts with other measures of problem severity. Those who experienced a severe decline in health and those who experienced a serious financial impact because of the problem were only one-third as likely to have their problem satisfactorily resolved. Problems involving care or treatment were less likely to be resolved satisfactorily compared with problems regarding purely administrative matters (OR = 0.69). Problems that were too complex to categorize were also much less likely to be resolved to the enrollee's satisfaction (OR = 0.32).

Discussion and Conclusion

Our analyses reveal that enrollees in health plans who have problems express their grievances inconsistently. These problems are resolved to the enrollee's satisfaction even less often. Of course, not all perceived problems represent bad care, and not all expressions of dissatisfaction merit a response by the health plan. But if the enrollees fail to express their concerns, the plan's administrators may not even recognize that there is a problem, let alone respond appropriately. Our findings also suggest that state regulators cannot rely on expressed grievances to monitor the plan's performance. Only a few problems lead to formal grievances that would be recognized by state agencies, even when the enrollee perceives the consequences to have been quite deleterious (table 2).

Returning to our five core hypotheses, our findings are strongly consistent with some of these predictions, have mixed implications for others, and directly conflict with one hypothesis. We uncovered strong evidence that the propensity and effectiveness of voice are enhanced by third parties, including family, physicians, and external mediators. We also found that the propensity to express complaints is affected by enrollees’ perceptions of health plans, whether they seemed to put the enrollees’ welfare first and whether they could be held responsible for the problem in question. (Problems related to medical care led to grievances far less frequently than did those associated with administrative matters.) We found mixed support for the hypotheses of compounded vulnerability and problem severity. Minority enrollees appeared less likely to express their grievances (although no less likely to have them resolved satisfactorily). But the severity of illness did not appear to inhibit voicing, although it did make voice less effective. The severity of the consequences produced by the problem had mixed effects, with some promoting voice and others deterring it.

We found no support for the prediction that enrollees with chronic conditions would have more difficulty resolving their grievances. To the contrary, these enrollees were more successful than otherwise comparable consumers without a chronic illness. It is possible that plans have less incentive to disenroll these individuals than we had expected (or less opportunity, given the limited plan choice for most of those with employer-based insurance). Alternatively, enrollees with chronic conditions may become sufficiently knowledgeable and skilled at negotiating with their health plan that these advantages offset any financial disincentives for the plan (Thorne 1993).

These findings should, of course, be interpreted in light of our study's methodological limitations. The respondents subjectively assessed their problems with their health plan. Consumers, therefore, may have blamed plans for some outcomes that were beyond their control. Conversely, enrollees may have overlooked other damaging episodes, not recognizing that they were problems. We also could not accurately determine the characteristics of health plans in which respondents were enrolled. The features of the health plans studied here were reported by respondents and may, therefore, have been measured imperfectly (Cunningham, Denk, and Sinclair 2001). The impact of state regulations could be only imperfectly assessed, since we had measures of enrollees’ awareness for only one regulation and no measures of regulatory enforcement for any of the interventions in question. Each of these relationships merits additional research, as does an extension of our analyses to other populations, such as the elderly, who were not included in the Kaiser survey.

Some of our independent variables related to the characteristics of the enrollees or the setting were not ideal for measuring the relevant concepts. For example, cognitive sophistication was not associated with voice in our models, but this may have reflected an inadequate proxy measure rather than a true null finding. These relationships should therefore be viewed as preliminary, in need of additional verification.

Finally, we could not measure with the Kaiser Foundation data the relationships among consumers’ various responses to problems. Because the survey did not identify when particular responses occurred, we could not determine whether exit came before or after voice, or the timing of voicing to physicians, employers, and health plans. Because this last factor could bias our findings with respect to voice, we employed a two-stage regression model to construct an unbiased estimate of the impact of complaining to the plan on the resolution of problems. Our findings suggest that (1) the estimated effects from a one-stage regression are seriously biased downward and (2) when appropriately measured, voice has a dramatic positive effect on the probability that problems will be resolved to the enrollee's satisfaction.

It is this last finding that makes the unevenness of voice so important. We now consider the implications of our findings in greater detail. We highlight three sets of results: (1) those relevant to understanding consumer behaviors in health care, (2) those important to public policy, and (3) those illustrating promising directions for additional research.

Consumer Behavior in Medical Care

Three sets of results are interesting in this respect. First, although chronic medical conditions did not inhibit effective voice, our findings provide substantial evidence that problems involving medical care are less adequately protected by grievance mechanisms than are those related to administrative concerns. Problems related to health care, as opposed to plan administration, were less likely to trigger a complaint. Problems involving combinations of care and administration were less likely to be resolved. When the problem itself seriously hurt health, voice was less common and less effective. Finally, those who reported themselves to be in poor health were also less able to have their problems resolved satisfactorily. Taken together, these findings suggest that voice mechanisms function least well for enrollees who encounter problems with their medical care (as opposed to their benefit paperwork). Since grievance mechanisms have been promoted by state and federal regulation precisely because policymakers were concerned that managed care plans were inappropriately interfering with clinical practice, this finding is ironic and not particularly reassuring.

Second, several results suggest that voicing can be difficult, especially for certain types of problems. Problems that crossed the boundaries between care and administration were less frequently resolved to enrollees’ satisfaction. Problems that were hard for respondents to classify (we could not determine whether the problems themselves were more complex or the respondents were simply less capable) were associated with less frequent expression of complaints to both the health plan and employers. This last finding is important, since as table 3 reveals, these hard-to-classify problems were least often resolved for those who remained silent about their concerns. As suggested earlier, third-party advocates can overcome these barriers to voice to some extent. But even with their involvement, disparities remain in consumers’ ability to respond to problems.

Third, minority enrollees were only about half as likely to express complaints to the plan, compared with otherwise similar non-Hispanic whites. Interestingly, this was not because they viewed voice as less effective. Indeed, African American respondents were more likely to have their problems resolved. There is certainly no evidence suggesting that minorities felt their health plan would be less responsive to their concerns. So why, then, was their voicing inhibited? Perhaps previous episodes from outside the health care system conditioned minority respondents to suppress their dissatisfaction (LaVeist, Nickerson, and Bowie 2000). Or their expectations for care may have been lower based on a history of past difficulties obtaining access to needed treatment (Collins et al. 2002).

Policy Implications

Several sets of findings also should be important to policymakers. First and foremost, our study is the first empirical assessment of state regulations intended to promote consumer voice or enhance its effectiveness. Our findings paint a rather mixed picture. On the positive side, mandating third-party mediation of disputes does appear to increase the efficacy of voicing to the plan. Respondents who lived in states with these regulations (and were aware of the requirements) were twice as likely to have their problems resolved as were other enrollees who had expressed their grievances to the health plan (table 5). But a substantial number of the residents of these states were not aware that the regulations were in place. This ignorance completely undermined the efficacy of the law. And even those who were aware of the law were no more likely to have voiced their concerns in the first place. Other regulatory interventions (required notification about grievance processes; mandated reporting of grievances to state agencies) appeared to have no positive effects on voicing.

We could not determine from our data the origins of this checkered track record. The fact that many plans are exempt from state regulations because of ERISA may be a part of the explanation. Learning curves may be another reason. The states enacted a large number of managed care regulations over a relatively brief period of time (Noble and Brennan 1999), and these new tasks may simply have exceeded their administrative capacities. The impact of these regulations may grow as the states learn to administer their provisions more effectively. Since many of these regulations were adopted in the late 1990s, their impact may also increase as more residents in the state learn about their existence. Consequently, the relatively negative findings reported here should be viewed as a conservative estimate of the longer-term effects of voice-promoting regulations.

A second finding has particular policy relevance, albeit in a rather perverse way. Plans that established friendly relationships with their enrollees were significantly more likely to receive complaints, even while controlling for the severity and frequency of problems. This suggests that plans that foster a positive, caring relationship with their enrollees are penalized by a greater number of grievances. Conversely, plans that are seen as unfriendly and unhelpful discourage complaints. Ironically, the more extensively state regulators assess a plan's performance in terms of frequency of complaints, the stronger the incentive is for plan administrators to create an image of aloofness and unresponsiveness to their members.

A third set of policy implications relate to physicians’ advocacy for their patients. Although both the American Medical Association and scholars who study professionalism recognize the need for doctors to advocate for their patients’ interests (Council on Ethical and Judicial Affairs 1995; Mechanic 2000), the norms of appropriate advocacy are not well defined. Although some consumers with problems with their health plan discussed them with their physician, many more did not. In some cases, that silence may have reflected a perception that their physician could do nothing to help. In other instances, it may have resulted from the belief that their physician would do nothing, indeed, may have been the culpable party. Whatever the factors that inhibit consumer voice to physicians, it is clear that if we expect health care professionals to be advocates, not only must they accept this role, but consumers must be made aware that they have potential allies who can help deal with their health plan. To our knowledge, there has been no concerted effort to date to do this.

Fourth, if American health policy is to rely on voice as a means of improving quality and a source of consumer protection, it is clear from our findings that policymakers need to develop more aggressive and comprehensive approaches for facilitating voice. Existing interventions have yielded inconsistent results. Many Americans remain ill-informed about their state's supportive policies and programs. Many others live in states that have not yet enacted these protections. But even if they were enacted in all states, conventional policy interventions remain far too passive to enhance consumer voice dramatically. Clearly, if one expects even the most rudimentary regulatory interventions to simulate voice more effectively, there must be widespread consumer education.

Education alone is not sufficient, however. Existing regulatory interventions to promote voice remain woefully inadequate. Worst is the requirement that enrollees first exhaust their internal appeals within the plan. This almost guarantees that the plans with the worst reputations and the least responsive administrations will remain untouched by pressures from consumers’ voice, precisely because their enrollees have no reason to expect them to respond constructively to a grievance or appeal. No existing policies have the capacity to offset the disparities in voice that have been documented here.

We can and should expect more of consumer voice and so should promote it more effectively. Looking to the experience of other countries, such as New Zealand (Paterson 2001) and the United Kingdom (Annendale and Hunt 1998; Klein 1973; Webb 1995), would provide a good start. Those countries have struggled for a long time to develop a meaningful “right to complain.” Their failures, as much as their successes, can help guide American policymakers.

Our final set of policy implications relates to means other than voice for resolving problems related to health care and health insurance. Although our focus in this article has been on consumers’ expression of grievances, it is also noteworthy that a number of problems are resolved even for those who do not voice to the plan at all. Forty-five percent of all problems are not voiced to the plan, and of these, 43 percent were resolved to the enrollee's satisfaction. The results reported in table 5 shed some light on the processes involved. Those who had contacted their physician about the problem were 50 percent more likely to have their problem resolved, compared with those who did not express their concerns to their doctor. In addition, exit proved to be an important source of consumer protection for this group (though not for those who voiced). Both switching plans and physicians were associated with higher levels of problem resolution, although only the latter was statistically significant (and then only at a 10 percent confidence level).

Directions for Future Research

Several findings point to the need for additional research on this topic. We found, for example, that enrollees reporting a larger number of chronic illnesses were more likely to have their problems resolved to their satisfaction (controlling for their current health status). Although this finding might be interpreted as consumers learning to deal more effectively with managed care as they cope with a number of different medical conditions, this positive relationship was true only for those who did not voice their complaints (table 5). What other causal mechanisms might come into play remains unclear.

Among the most interesting, but most mysterious, findings is the estimated empirical relationship between exit and voice. The relationship between exiting from the plan or physician and the successful resolution of the problem reported in table 4 appears to be a positive (complementary) relationship of exit and voice, albeit one that is not statistically significant. But the disaggregated results reported in table 5 reveal that this positive relationship between exit and problem resolution existed only for those who did not voice their grievances. The direction of causality cannot be discerned from our present data. We need data that can track consumer responses longitudinally, so as to determine whether lack of effective voice was stimulating exit, or the other way around.

Given the constraints on exiting the plan imposed by purchasers and inherent in the need for continuity of care when sick, voice is likely to remain the most common response to unsatisfactory experiences with health plans. The unevenness of voicing reported here and the systematic biases in expressing grievances that are revealed in our findings clearly demonstrate the need to make voice more effective and more equitable. Our findings suggest that public policy can encourage these consumer behaviors. However, only by paying more attention to the implementation of managed care regulations can policymakers ensure that the potential benefits of consumer voice are realized and that all health care consumers can benefit from this response to health plans’ failures.

Acknowledgments

We are grateful that the Kaiser Family Foundation was willing to share with us the survey data used in this article. Our analyses and presentation also benefited from the comments of the Milbank Quarterly’s three reviewers. This project was supported in part from grants by the Robert Wood Johnson Foundation and Atlantic Philanthropic Services.

Endnotes

1

Consumers don't need to feel that they completely understand a problem before they can complain. Indeed, grievances are often motivated wholly or partly by a desire for better understanding, to obtain some sort of explanation from the culpable parties (Rosenthal and Schlesinger 2002). In their study of complaints in the NHS, Mulcahy and Tritter found that 55 percent of the complaints to doctors and 42 percent of the complaints to administrators were motivated primarily by a quest for more information (1998, 836).

2

Hirschman recognized (1980) that his original specification presumed that exit was relatively costless, an assumption that he later retracted. Although he concluded that “exit can imply considerable cost in purely economic market situations even in the absence of loyalty,” he continued to assert that the costs of exit would be experienced largely in employment arrangements or contracts between firms, suggesting that “such costs are least in evidence in the case of consumer products”(439). This assertion ignores the crucial differences between products and services, and the high value that consumers often place on maintaining a continuity of relationships with service providers (Schlesinger, Druss, and Thomas 1999).

3

For example, the California Supreme Court determined that Kaiser Permanente had failed to establish an impartial appeals process to resolve disputes. The case involved evidence that Kaiser established a self-administered appeals process that delayed resolution for an average of 674 days instead of the promised 60 or fewer days and that it took an average of almost two and a half years for a case to reach a hearing (Hilzenrath 1998).

4

Two other possibilities were not studied. The first was complaints lodged with state agencies. This was simply too uncommon (n) to study in much detail. The second was complaining to physicians. We did not treat this as a form of voice because it was difficult to distinguish from conventional symptom reporting. We did, however, control for comments to physicians in assessing the effectiveness of voice to health plans.

5

Given the question format, we cannot discern the order in which the respondent exited the plan and/or voiced a complaint. Consequently, a positive coefficient on the exit variable may not indicate that opportunities for exit actually encouraged voice, since it is possible that the respondents exited only after an unsuccessful experience with voice. It would have been more useful to have measures of opportunities for exit, whether or not these opportunities were pursued.

6

We also considered whether the number of years that the law had been in effect influenced any of the dependent variables. We found no significant duration effects for any of the regulations.

References

  1. Allsop J. Two Sides to Every Story: Complaints and Doctors’ Perspectives in Disputes about Medical Care in a General Practice Setting. Law and Policy. 1994;16(2):149–83. [Google Scholar]
  2. Andreason AR. Consumer Satisfaction in Loose Monopolies: The Case of Medical Care. Journal of Public Policy and Marketing. 1984;2:122–35. [Google Scholar]
  3. Andreason AR. Consumer Complaints and Redress: What We Know and What We Don't Know. In: Maynes ES, editor. The Frontiers of Research in the Consumer Interest. Columbia, Mo.: American Council on Consumer Interests; 1988. pp. 675–722. [Google Scholar]
  4. Annandale E, Hunt K. Accounts of Disagreements with Doctors. Social Science and Medicine. 1998;46(1):19–29. doi: 10.1016/s0277-9536(97)00149-4. [DOI] [PubMed] [Google Scholar]
  5. Annas GJ. Patients’ Rights in Managed Care—Exit, Voice and Choice. New England Journal of Medicine. 1997;337(3):210–5. doi: 10.1056/NEJM199707173370323. [DOI] [PubMed] [Google Scholar]
  6. Bearden WO, Teel JE. Selected Determinants of Consumer Satisfaction and Complaint Reports. Journal of Marketing Research. 1983;20(1):21–8. [Google Scholar]
  7. Best A. When Consumers Complain. New York: Columbia University Press; 1981. [Google Scholar]
  8. Blendon RJ, Brodie M, Benson JM, Altman DE, Levitt L, Hoff T, Hugick L. Understanding the Managed Care Backlash. Health Affairs. 1998;17(4):80–94. doi: 10.1377/hlthaff.17.4.80. [DOI] [PubMed] [Google Scholar]
  9. Boroff KE, Lewin D. Loyalty, Voice and Intent to Exit a Union Firm: A Conceptual and Empirical Analysis. Industrial and Labor Relations Review. 1997;51(1):50–62. [Google Scholar]
  10. Brennan T, Sox C, Burstin H. Relation between Negligent Adverse Events and the Outcomes of Medical Malpractice Litigation. New England Journal of Medicine. 1996;335(26):1963–7. doi: 10.1056/NEJM199612263352606. [DOI] [PubMed] [Google Scholar]
  11. Brown SW, Swartz TA. Consumer Medical Complaint Behavior: Determinants and Alternatives to Malpractice Litigation. Journal of Public Policy and Marketing. 1984;3:85–98. [Google Scholar]
  12. Charters MA. The Patient Representative Role and Sources of Power. Hospital & Health Services Administration. 1993;38(3):429–42. [PubMed] [Google Scholar]
  13. Collins KS, Hughes DL, Doty MM, Ives BL, Edwards JN, Tenney K. Diverse Communities, Common Concerns: Assessing Health Care Quality for Minority Americans. New York: Commonwealth Fund; 2002. [Google Scholar]
  14. Cornwell J. Hard Earned Lives: Accounts of Health and Illness from East London. London: Tavistock; 1984. [Google Scholar]
  15. Council on Ethical and Judicial Affairs, American Medical Association. Ethical Issues in Managed Care. Journal of the American Medical Association. 1995;273(4):330–5. [Google Scholar]
  16. Coyle J. Exploring the Meaning of “Dissatisfaction” with Health Care: The Importance of “Personal Identity Threat.”. Sociology of Health and Illness. 1999;21(1):95–124. [Google Scholar]
  17. Cunningham PJ, Denk C, Sinclair M. Do Consumers Know How Their Health Plan Works? Health Affairs. 2001;20(2):159–66. doi: 10.1377/hlthaff.20.2.159. [DOI] [PubMed] [Google Scholar]
  18. Dallek G, Pollitz K. External Review of Health Plan Decisions: An Update. Menlo Park, Calif.: Kaiser Family Foundation; 2000. [Google Scholar]
  19. Daniels N, Sabin J. The Ethics of Accountability in Managed Care Reform. Health Affairs. 1998;17(5):50–64. doi: 10.1377/hlthaff.17.5.50. [DOI] [PubMed] [Google Scholar]
  20. Day RL. Proceedings of the Fourteenth Annual Conference of the Association for Consumer Research. Chicago: 1983. Modeling Choices among Alternative Responses to Dissatisfaction; pp. 496–9. October 20-23. [Google Scholar]
  21. Druss B, Schlesinger M, Thomas T, Allen H. Chronic Illness and Plan Satisfaction under Managed Care. Health Affairs. 2000;19(1):203–9. doi: 10.1377/hlthaff.19.1.203. [DOI] [PubMed] [Google Scholar]
  22. Emanuel E. Choice and Representation in Health Care. Medical Care Research and Review. 1999;56(suppl. 1):113–40. doi: 10.1177/1077558799056001S07. [DOI] [PubMed] [Google Scholar]
  23. Families USA. Hit and Miss: State Managed Care Laws. Washington, D.C: 1998. [Google Scholar]
  24. Farrell D. Exit, Voice, Loyalty and Neglect as Responses to Job Dissatisfaction: A Multidimensional Scaling Study. Academy of Management Journal. 1983;26:596–607. [Google Scholar]
  25. Feick L. Latent Class Models for the Analysis of Behavioral Hierarchies. Journal of Marketing Research. 1987;(May):174–86. [Google Scholar]
  26. General Accounting Office. Health Care Reform: Report Cards Are Useful but Significant Issues Need to Be Addressed. Washington, D.C: 1994. GAO/HEHS-94-219. [Google Scholar]
  27. Gilly MC. Post-complaint Processes: From Organizational Response to Repurchase Behavior. Journal of Consumer Affairs. 1987;21(2):293–313. [Google Scholar]
  28. Gold M, McCoy J. Monitoring Medicare + Choice: Fast Facts. Washington, D.C.: Mathematica Policy Research; 2002. Choice Continues to Erode in 2002. no. 7. [Google Scholar]
  29. Granbois D, Summers JO, Frazier GL. Correlates of Consumer Expectation and Complaining Behavior. In: Day RL, editor. Consumer Satisfaction, Dissatisfaction and Complaining Behavior. Bloomington: Indiana University Press; 1977. pp. 18–25. [Google Scholar]
  30. Gresenz C, Studdert D, Campbell N, Henster D. Patients in Conflict with Managed Care: A Profile of Appeals in Two HMOs. Health Affairs. 2002;21(4):189–96. doi: 10.1377/hlthaff.21.4.189. [DOI] [PubMed] [Google Scholar]
  31. Hak T. The Interactional Form of Professional Dominance. Sociology of Health and Illness. 1994;16:469–88. [Google Scholar]
  32. Hemenway D, Killen A. Complainers and Non-Complainers. Journal of Ambulatory Care Management. 1989;12(3):19–27. doi: 10.1097/00004479-198908000-00005. [DOI] [PubMed] [Google Scholar]
  33. Hilzenrath DS. The Battle for Patient Rights; Final Say: One Patient's Ordeal. Washington Post. 1998;(June 24):A1. [Google Scholar]
  34. Hirschman AO. Exit, Voice and Loyalty: Responses to Declines in Firms, Organizations and States. Cambridge, Mass.: Harvard University Press; 1970. [Google Scholar]
  35. Hirschman AO. Exit, Voice, and Loyalty: Further Reflections and a survey of Recent Contributions. Milbank Memorial Fund Quarterly/Health and Society. 1980;58(3):430–53. [PubMed] [Google Scholar]
  36. Jacobs LR, Shapiro RY. The American Public's Pragmatic Liberalism Meets Its Philosophical Conservatism. Journal of Health Politics, Policy and Law. 1999;24(5):1021–31. doi: 10.1215/03616878-24-5-1021. [DOI] [PubMed] [Google Scholar]
  37. Kaye C, MacManus T. Understanding Complaints. Health Service Journal. 1990;100:1254–5. [PubMed] [Google Scholar]
  38. Klein R. Complaints against Doctors: A Study in Professional Accountability. London: Charles Knight; 1973. [Google Scholar]
  39. Kolodinsky J. A System for Estimating Complaints, Complaint Resolution and Subsequent Purchases of Professional and Personal Services. Journal of Consumer Satisfaction, Dissatisfaction and Complaining Behavior. 1992;5:36–44. [Google Scholar]
  40. Kolodinsky J. Complaints, Redress, and Subsequent Purchases of Medical Services by Dissatisfied Consumer. Journal of Consumer Policy. 1993;16:193–214. [Google Scholar]
  41. Krajewski TF, Bell C. A System of Patients’ Rights Advocacy in State Psychiatric Inpatient Facilities in Maryland. Hospitals and Community Psychiatry. 1992;43(2):127–31. doi: 10.1176/ps.43.2.127. [DOI] [PubMed] [Google Scholar]
  42. Kritzer HM, Vidmar N, Bogart WA. To Confront or Not to Confront: Measuring Claiming Rates in Discrimination Grievances. Law and Society Review. 1991;25(4):875–87. [Google Scholar]
  43. Kronick R, de Beyer J, editors. Medicare HMOs: Making Them Work for the Chronically Ill. Chicago: Health Administration Press; 1999. [Google Scholar]
  44. Kucukarslan S, Pathak DS. The Pharmaceutical Market's Complaining Process: An Application of Hirschman's Theory to a Public Policy Problem. Journal of Consumer Satisfaction, Dissatisfaction and Complaining Behavior. 1994;7:87–95. [Google Scholar]
  45. LaVeist TA, Nickerson KJ, Bowie JV. Attitudes about Racism, Medical Mistrust and Satisfaction with Care among African American and White Cardiac Care Patients. Medical Care Research and Review. 2000;57(suppl. 1):146–61. doi: 10.1177/1077558700057001S07. [DOI] [PubMed] [Google Scholar]
  46. Lim MC, Tan CB, Goh LG, Ling SL. Why Do Patients Complain? A Primary Health Care Study. Singapore Medical Journal. 1998;39(9):390–5. [PubMed] [Google Scholar]
  47. Lubalin J, Harris-Kojetin L. What Do Consumers Want and Need to Know in Making Health Care Choices? Medical Care Research and Review. 1999;56(suppl. 1):67–102. doi: 10.1177/1077558799056001S04. [DOI] [PubMed] [Google Scholar]
  48. Marquis MS, Long S. Trends in Managed Care and Managed Competition, 1993–97. Health Affairs. 1999;18(6):75–88. doi: 10.1377/hlthaff.18.6.75. [DOI] [PubMed] [Google Scholar]
  49. Maxwell J, Briscoe F, Davidson S, Eisen L, Robbins M, Temin P, Young C. Managed Competition in Practice: “Value Purchasing” by Fourteen Employers. Health Affairs. 1998;17(3):216–26. doi: 10.1377/hlthaff.17.3.216. [DOI] [PubMed] [Google Scholar]
  50. Maxwell J, Temin P, Watts C. Corporate Health Care Purchasing among Fortune 500 Firms. Health Affairs. 2001;20(3):181–8. doi: 10.1377/hlthaff.20.3.181. [DOI] [PubMed] [Google Scholar]
  51. May ML, Stengel DB. Who Sues Their Doctors? How Patients Handle Medical Grievances. Law and Society Review. 1990;24(1):105–20. [Google Scholar]
  52. Mechanic D. Managed Care as a Target of Distrust. Journal of the American Medical Association. 1998;277(22):1810–2. [PubMed] [Google Scholar]
  53. Mechanic D. Managed Care and the Imperative for a New Professional Ethic. Health Affairs. 2000;19(5):100–11. doi: 10.1377/hlthaff.19.5.100. [DOI] [PubMed] [Google Scholar]
  54. Mechanic D, Schlesinger M. The Impact of Managed Care on Patients’ Trust in Medical Care and Their Physicians. Journal of the American Medical Association. 1996;275(21):1693–7. [PubMed] [Google Scholar]
  55. Miller RE, Sarat A. Grievances, Claims and Disputes: Assessing the Adversarial Culture. Law and Society Review. 1980–81;13:3–4. 525–66. [Google Scholar]
  56. Miller TE. Managed Care Regulation: In the Laboratory of the States. Journal of the American Medical Association. 1997;278(13):1102–9. doi: 10.1001/jama.278.13.1102. [DOI] [PubMed] [Google Scholar]
  57. Miller TE. Center Stage on the Patient Protection Agenda: Grievance and Appeal Rights. Journal of Law, Medicine and Ethics. 1998;26:89–99. doi: 10.1111/j.1748-720x.1998.tb01665.x. [DOI] [PubMed] [Google Scholar]
  58. Mulcahy L, Tritter JQ. Pathways, Pyramids and Icebergs? Mapping Links between Dissatisfaction and Complaints. Sociology of Health and Illness. 1998;20(6):825–47. [Google Scholar]
  59. Nader L, editor. No Access to Law: Alternatives to the American Judicial System. London: Academic Press; 1980. [Google Scholar]
  60. Noble A, Brennan T. The Stages of Managed Care Regulation: Developing Better Rules. Journal of Health Politics, Policy and Law. 1999;24(6):1275–935. doi: 10.1215/03616878-24-6-1275. [DOI] [PubMed] [Google Scholar]
  61. Paterson R. The Patients’ Complaints System in New Zealand. Health Affairs. 2001;21(3):70–8. doi: 10.1377/hlthaff.21.3.70. [DOI] [PubMed] [Google Scholar]
  62. Perkins J, Olson K, Rivera L. Ombudsprograms and Member Advocates: Consumer-Oriented Approaches to Problem-Solving in Medicaid Managed Care. Los Angeles: National Health Law Program; 1998. [Google Scholar]
  63. Phillips KA, Mayer ML, Aday LA. Barriers to Care among Racial/Ethnic Groups under Managed Care. Health Affairs. 2000;19(4):65–75. doi: 10.1377/hlthaff.19.4.65. [DOI] [PubMed] [Google Scholar]
  64. Pollitz K, Dallek G, Tapay N. External Review of Health Plan Decisions: An Overview of Key Program Features in the States and Medicare. Menlo Park, Calif.: Kaiser Family Foundation; 1998. [Google Scholar]
  65. Rice T, Gabel J, Levitt L, Hawkins S. Workers and Their Health Care: Free to Choose? Health Affairs. 2002;21(1):182–7. doi: 10.1377/hlthaff.21.1.182. [DOI] [PubMed] [Google Scholar]
  66. Richins ML. Consumer Perceptions of Costs and Benefits Associated with Complaining. In: Hunt HK, Day RL, editors. Refining Concepts and Measures of Consumer Satisfaction and Complaining Behavior: Papers from the Fourth Annual Conference on Consumer Satisfaction, Dissatisfaction and Complaining Behavior. Bloomington: Indiana University Press; 1979. pp. 50–3. [Google Scholar]
  67. Rodwin M. Consumer Voice and Representation in Managed Healthcare. Journal of Health Law. 2001;(Spring):223–76. [Google Scholar]
  68. Rosenthal M, Schlesinger M. Not Afraid to Blame: The Neglected Role of Blame Attribution in Medical Consumerism and Some Implications for Health Policy. Milbank Quarterly. 2002;80(1):41–95. doi: 10.1111/1468-0009.00003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Ross C, Wheaton B, Duff R. Client Satisfaction and the Organization of Medical Practice: Why Time Counts. Journal of Health and Social Behavior. 1981;22(3):243–55. [PubMed] [Google Scholar]
  70. Schauffler H, McMenamin S, Cubanski J, Hanley H. Differences in the Kinds of Problems Consumers Report in Staff/Group HMOs, IPA/Network HMOs and PPOs in California. Medical Care. 2001;39(1):15–25. doi: 10.1097/00005650-200101000-00004. [DOI] [PubMed] [Google Scholar]
  71. Schlesinger M. On the Limits of Expanding Health Care Reform: Chronic Care in Prepaid Settings. Milbank Quarterly. 1986;64(2):189–215. [PubMed] [Google Scholar]
  72. Schlesinger M. Countervailing Agency: A Strategy of Principaled Regulation under Managed Competition. Milbank Quarterly. 1997;75(1):35–87. doi: 10.1111/1468-0009.00044. [DOI] [PMC free article] [PubMed] [Google Scholar]
  73. Schlesinger M, Druss B, Thomas T. No Exit? The Effects of Health Status on Dissatisfaction and Disenrollment from Health Plans. Health Services Research. 1999;34(2):547–76. [PMC free article] [PubMed] [Google Scholar]
  74. Schwartz LR, Overton DT. Emergency Department Complaints: A One-Year Analysis. Annals of Emergency Medicine. 1987;16(8):857–61. doi: 10.1016/s0196-0644(87)80522-x. [DOI] [PubMed] [Google Scholar]
  75. Singh J. Determinants of Consumers’ Decisions to Seek Third Party Redress: An Empirical Study of Dissatisfied Patients. Journal of Consumer Affairs. 1989;23(2):329–63. [Google Scholar]
  76. Solnick SJ, Hemenway D. Complaints and Disenrollment at a Health Maintenance Organization. Journal of Consumer Affairs. 1992;26(1):90–104. [Google Scholar]
  77. Tapay N, Feder J, Dallek G. Protection for Consumers in Managed Care Plans: A Comparison of Medicare, Medicaid and the Private Insurance Market. Menlo Park, Calif.: Kaiser Family Foundation; 1998. [Google Scholar]
  78. Thorne SE. Negotiating Health Care: The Social Context of Chronic Illness. Newbury Park, Calif.: Sage; 1993. [Google Scholar]
  79. Warland RH, Herrmann RO, Moore DE. Consumer Complaining and Community Involvement: An Exploration of Their Theoretical and Empirical Linkages. Journal of Consumer Affairs. 1984;18(1):64–78. [Google Scholar]
  80. Webb B. A Study of Complaints by Patients of Different Age Groups in an NHS Trust. Nursing Standard. 1995;9(42):34–7. doi: 10.7748/ns.9.42.34.s36. [DOI] [PubMed] [Google Scholar]
  81. Weiner B. Judgments of Responsibility. New York: Guildford Press; 1995. [Google Scholar]
  82. Zaslavsky AM, Beaulieu ND, Landon BE, Cleary PD. Dimensions in Consumer-Assessed Quality of Medicare Managed-Care Health Plans. Medical Care. 2000;38(2):162–74. doi: 10.1097/00005650-200002000-00006. [DOI] [PubMed] [Google Scholar]

Articles from The Milbank Quarterly are provided here courtesy of Milbank Memorial Fund

RESOURCES