Patients’ perceptions and treatment effectiveness

Abstract

An extensive literature relating patients’ expectations to treatment outcomes has not addressed the determinants of these expectations. We argue that treatment history is part of a reference point that influences the patients’ expectations of how effective further treatment might be, thus influencing whether to proceed with additional treatment or not. We hypothesize that those patients with unsuccessful prior treatments have diminished expected improvement from subsequent treatments. Prospect Theory (PT) provides a theoretical foundation for reference frame effects, and the model is tested with data on patients diagnosed with Idiopathic Intracranial Hypertension (IIH). Our results support the reference frame hypothesis.

Notes

¹ The data include both male and female patients. For convenience, only the female pronoun is used in the remainder of this article.

² For summaries of the earlier literature, see Jones (1982), Ditto and Hilton (1990), Jones (1990).

³ See Appendix A for a detailed explanation of this disorder.

⁴ A revised version titled ‘cumulative prospect theory’ (Tversky and Kahneman, 1992; Wakker and Tversky, 1993) applies to uncertain and risky prospects with multiple outcomes. Under the extended theory decision weights are applied to cumulative, as opposed to individual, probabilities. However, the extension gives the same results as the original theory for all two-outcome prospects and mixed three-outcome prospects. This article focuses on three-outcome prospects, hence the earlier version of PT (Kahneman and Tversky, 1979) is sufficient for our purposes.

⁵ Even though the model is conditioned on the assumption that v(0) = 0, it is a trivial extension to extend the results to the nonrestricted case. The math is available upon request from the authors. Also, as we mentioned above, v(0) is most likely less than zero, otherwise treatment would not be sought. This assumption essentially allows us to group v(0) and v(w) and analyse the probability of feeling better relative to not feeling better.

⁶ In our application to IIH our ordering assumes that the acetazolamide was ineffective from its initial use. If acetazolamide was at first effective and improving a patient's condition, but over time lost its effectiveness, the ordering of anticipated probabilities may be opposite to what we discuss here.

⁷ The registry is co-sponsored by the Intracranial Hypertension Research Foundation (IHRF) of Vancouver, Washington and the Casey Eye Institute at the Oregon Health and Science University (OHSU). http://www.ihrfoundation.org/news/registry.asp.

⁸ We are analysing the effects that an individual's reference point has on her perceived effectiveness of future treatment. Her reference point is defined as the patient's pre-treatment status and the successfulness of her earlier treatments. As our dependent variable is relative to her pre-treatment condition, including the severity of the condition (which is, in any case, almost unmeasurable because of the nature of the disease) in the model would create confounding issues, such as endogeneity bias.

⁹ See the discussion of the dependent variable in the Section ‘Data’.

¹⁰ See Appendix B for a detailed discussion of the MRE, including the sufficient condition that must be met for consistent estimates.

¹¹ We thank Jayson Abrevaya for providing the MRE code used in our analysis.

¹² Comparability could also be obtained by transforming the coefficients to marginal effects; however, the knowledge of the misclassification probabilities is required for consistent estimates of the MRE marginal effects (Hausman et al., 1998).

¹³ Had the MRE reversed the signs of the logistic coefficient estimates, this could have been an indication that the sufficient condition for consistent MRE estimates was not met (Abrevaya and Hausman, 1999). As the coefficient estimates from both procedures are similar misspecification bias is probably not a problem.

¹⁴ This is possible because all of these variables are binary and share the same omitted category.

¹⁵ The long-term success rate among all dieters is only 31%. For females the success rate is only 27%. (Kruger et al., 2006).

¹⁶ See Rosenman et al. (2009) for a detailed discussion on the economic costs of IIH.

¹⁷ As is mentioned in Appendix A, although we are not able to control for the treating physician, a treatment algorithm for IIH does exist, implying that treatment paths should not vary a great deal from one physician to the next.

¹⁸ All of the patients in that study were on acetazolamide at one time, and it is unclear whether they continued to take the drug after surgery.