A typology of preferences for participation in healthcare decision making

Abstract

Classifying patients as “active” or “passive” with regard to healthcare decision making is misleading, since patients have different desires for different components of the decision-making process. Distinguishing patients’ desired roles is an essential step towards promoting care that respects and responds to individual patients’ preferences. We included items on the 2004 Wisconsin Longitudinal Study mail survey measuring preferences for four components of the decision-making process: physician knowledge of patient medical history, physician disclosure of treatment choices, discussion of treatment choices, and selection of treatment choice. We characterized preference types for 5199 older adults using cluster analysis. Ninety-six percent of respondents are represented by four preference types, all of which prefer maximal information exchange with physicians. Fifty-seven percent of respondents wanted to retain personal control over important medical decisions (“autonomists”). Among the autonomists, 81% preferred to discuss treatment choices with their physician. Thirty-nine percent of respondents wanted their physician to make important medical decisions (“delegators”). Among the delegators, 41% preferred to discuss treatment choices. Female gender, higher educational attainment, better self-rated health, fewer prescription medications, and having a shorter duration at a usual place of care predicted a significantly higher probability of the most active involvement in discussing and selecting treatment choices. The overwhelming majority of older adults want to be given treatment options and have their physician know everything about their medical history; however, there are substantial differences in how they want to be involved in discussing and selecting treatments.

Introduction

Successful physician–patient communication about decision making plays a critical role in healthcare delivery, yet there are conflicting messages about changing roles for physicians and patients. A mandatory role for patients as autonomous decision makers has recently been recommended, since patients alone weather the consequences of their medical care (McNutt, 2004). However, others advocate that patients be allowed to participate to the extent they desire, completely delegating decisions to their physicians if they so choose (Whitney, McGuire, & McCullough, 2004). The model of patient participation where both physician and patient communicate information and values and make decisions together—called shared decision making—has been embraced in literature (Charles, Gafni, & Whelan, 1999; Emanuel & Emanuel, 1992; Quill & Brody, 1996), and the Institute of Medicine has recommended including it in medical school curricula as a mechanism to improve care (Institute of Medicine, 2004). Yet there has been little discussion resolving the potential conflict between promoting shared decision making as ideal and accommodating individual patient preferences for participation (or lack of participation) in decision making.

There is evidence that not all components of the shared decision making model are equally desired by patients. Early work in this area identified two distinct components of patient participation: information exchange and decision making (Ende, Kazis, Ash, & Moskowitz, 1989; Ong, de Haes, Hoos, & Lammes, 1995; Strull, Lo, & Charles, 1984). Later work further refined the components to three: information exchange, deliberation, and decisional control (Charles et al., 1999). It is well established that patients want to receive information from their physicians almost universally (Deber, Kraetschmer, & Irvine, 1996; Ende et al., 1989; Nease & Brooks, 1995), but less is known about information transmission, i.e., patient preferences for providing information to their physicians. Preferences for deliberation (i.e., formal discussion about options) and preferences for decisional control (i.e., power over the final selection of treatment) are known to differ substantially among patients (Robinson & Thomson, 2001), although it is unclear whether patients who prefer discussing treatment options also prefer making the final decision about treatment. Where one study suggests that patients do not want to be involved in problem-solving tasks while retaining control over decisions (Deber et al., 1996), others advise patient involvement in both stages (Quill & Brody, 1996). Often the concept of deliberation as distinct from decisional control is ignored entirely.

Consequently, there is little information available to assist clinicians in understanding and distinguishing the different roles patients prefer in the decision-making process. Distinguishing these roles is an essential step towards promoting patient-centered care, care that respects and responds to individual patients’ preferences (Gerteis et al., 1993). Physicians have been classified according to their attitudes about paternalism, patient autonomy, and deliberation (Falkum & Forde, 2001), but patients have not been similarly characterized according to their preferences for participation in multiple, distinct stages of healthcare decision making.

To address this gap, we categorize a population-based cohort of older adults into types based on preferences for participation in each of four components of healthcare decision making (physician knowledge of patient, physician disclosure of treatment choices, discussion of treatment choices, and selection of treatment choice). We then describe the preference types with respect to key sociodemographic and health variables. We use data from the Wisconsin Longitudinal Study (WLS), which for 47 years has followed over 10,000 men and women who graduated from Wisconsin high schools in 1957. Respondents have completed extensive telephone and mail surveys with high sample retention and represent a wide variety of health states.

Several models of decision making have been described. In the paternalistic model, physicians perform information management, assess options, and make treatment decisions for patients (presumably in patients’ best interests) without consideration of patient preferences (Katz, 1984). This model assumes that there are objective criteria to determine the best course of action, that physicians choose this course of action and that patients assent to this course with gratitude (Emanuel & Emanuel, 1992). It is also assumed that patients who prefer not to make the final decision also do not want to participate more than minimally in information exchange or discussing options. Conversely, in the informed or consumer model, physicians provide all relevant information to their patients, and patients alone assess their options and make the final decision. The physician in this model serves as a technical expert to provide information and facilitate decisions made by a fully autonomous patient (Emanuel & Emanuel, 1992). Again assumptions are made about the congruence of preferences for information, deliberation, and decisional control, namely, that patients who want to make the final decision do not want to first discuss all the options with physicians. The shared model merges these models into one where patients and physicians participate equally in all stages of decision making (Charles et al., 1999). This model assumes that patients want equal participation with their physicians at every stage. Variations of this model include the interpretive and deliberative models, which differ in their treatment of patient values (Emanuel & Emanuel, 1992). However, none of these models accommodates situations where patients want to participate to different degrees in different stages.

This paper categorizes older adults into types based on preferences for participation in healthcare decision making using the Charles, Gafni and Whelan (1997) and Charles et al. (1999) framework, which describes decision making in three analytic stages: (1) information exchange (subcategorized into flow, direction, type, and amount), (2) deliberation, and (3) control over the final decision. Although the shared decision-making model Charles et al. outlined has been extensively cited (Thomson ISI, 2003), an explicit test of the model showed that in 62 consultations regarding prescription drugs, there was little evidence that both parties were involved or shared information, let alone shared deliberation or decisional control (Stevenson, Barry, Britten, Barber, & Bradley, 2000). Though not explicitly based on the Charles et al. framework, when Norwegian physicians were profiled with respect to their attitudes toward paternalism, patient autonomy, and deliberation, only 15% were pure paternalists, 19% fit under the informed model (autonomists), and 12% could be called shared (deliberationists) (Falkum & Forde, 2001). The approach to categorizing decision making as purely paternalistic, informed, or shared is clearly limited in its ability to accurately reflect actual practice, as Charles et al. acknowledge (1999). To address this, we develop a typology that allows the components of the decision-making process to vary within patients, e.g., a strong preference for discussing treatment options does not necessarily indicate a strong preference for retaining control over the decision. By allowing patients to vary in their preferences for each of the dimensions, our typology may better reflect the complex nature of patient preferences for participation in healthcare visits.

No single validated instrument exists to measure preferences for participation in all components of the decision-making process as conceptualized by Charles et al. (1999), though two validated instruments capture fundamental parts. The Autonomy Preference Index (API) includes two scales to distinguish between information seeking and decision making (Ende et al., 1989), and the Health Opinions Survey (HOS) includes two scales to distinguish between preferences for information and behavioral involvement (Krantz, Baum, & Wideman, 1980). A comparison of these two instruments identified differences in how they framed questions, specifically that the API focuses on what patients prefer the physician to do and the HOS asks what the patient usually does (Nease & Brooks, 1995). Because we were interested in preferences rather than respondent-reported behaviors, we favored the decisional control items in the API. However, the API was limited for our purposes because information exchange is measured only in the direction of doctor to patient and there are no explicit items about deliberation. Similarly the HOS does not measure information exchange from patient to physician, but the information subscale does include items about being given choices, which is the basis for the discussing options during deliberation. We operationalized the components of patient participation in decision making by including items from these validated instruments when possible, but we developed our own item to assess preference for physician knowledge of patient. The other frequently used method to elicit the role patients prefer is illustrated response cards (Degner & Sloan, 1992), but this option was not feasible given constraints of the WLS mail survey. Additionally, the illustrated response cards would not distinguish preferences for physician knowledge of patient or discussion of treatment choices.

We expected that respondents would have high preferences for information about treatment choices from their physicians but would vary in their preferences for the other components of decision making. Since female gender, (Kaplan, Gandek, Greenfield, Rogers, & Ware, 1995) higher education (Kaplan et al., 1995; Thompson, Pitts, & Schwankovsky, 1993), and better health (Degner & Sloan, 1992; Ende et al., 1989) have been associated with preferences for increased participation in healthcare decision making, we expect to find similar relationships in the WLS population as well (Kaplan et al., 1995; Thompson et al., 1993). We also expect to see preferences vary according to interactions with the healthcare system or specific providers. There is some evidence that chronic illness visits are positively associated with increased shared decision making (Gotler et al., 2000). We expect that respondents with greater illness burden or more prescription medications will prefer more active participation.