The purpose of this study was to expand on the patient preference literature for individuals with psoriasis by examining preferences for levels within psoriasis treatment attributes. We demonstrated not only the utility of receiving a preferred treatment, but also the disutility associated with receiving a nonpreferred treatment. Moreover, we identified several correlates of patient preferences for treatment attributes that were most and least preferred.
We observed that the magnitude of the mean preference scores for least preferred attribute levels was consistently higher than the magnitude of the mean scores for the most preferred attribute levels. Our results suggest that patients are not only concerned about receiving preferred treatments, but perhaps are more concerned that treatments associated with strong disutility not be recommended.27
Evidence from the psychology literature reveals a negativity bias, specifically that many are affected to a greater extent by bad things happening to them rather than good events. As a result, individuals may be more motivated to avoid bad things than to pursue good things.15
Moreover, previous work suggests that the saliency of bad events or negative emotions lasts longer than good events or positive emotions.28
Taken together, these prior findings may explain the magnitude of the negative utility that participants in our study assigned to the least preferred treatment attribute levels. Based on our results, it is possible that failure to acknowledge negative preferences in treatment decision-making, ie, recommending treatments associated with strong disutility, may affect patient satisfaction and adherence with recommended treatment to a greater degree than recommendation of treatments with attributes that are only weakly preferred.
For example, the strongly negative utilities identified through the conjoint analysis in our study suggest that participants were more concerned about improvement of their skin condition than about the reversibility or the severity of treatment side effects. Studies exploring the significance of treatment side effects and treatment benefit in chronic disease management have reported mixed results.11
In some studies, patients valued the risk and the severity of treatment side effects over treatment benefit,11
while in others patients were reported to be willing to accept the risks of treatment if benefits were perceived to be high.12
These contradictory findings may have resulted from the way in which the severity of side effects was explicitly defined or presented (eg, “risk of liver damage or skin cancer” versus “side effects involving more than the skin”). In cases where the severity of side effects was defined, treatment side effects became the most important determinant of patient treatment preferences.11
When eliciting patient preferences, explicitly defining treatment attributes has both advantages and disadvantages. For example, use of explicit language (eg, “risk of liver damage or skin cancer”) may result in dominant preferences in which respondents become unwilling to trade between attributes.32
However, using explicit language has the advantage of being more realistic and makes the choice task less abstract.32
Therefore, patients’ preferences, may be influenced not only by the processes of care (eg, treatment frequency) or benefits (eg, side effects), but also by the way the information is presented to patients.33
Further, our findings showed that sociodemographic, treatment, and behavioral aspects of individual patients appear to be important in determining patient preferences for treatment attributes. In multivariate analysis, we found that younger patients and women preferred treatments with greater likelihood of benefit, while older patients appeared more concerned with treatment side effects. A study by Gelfand et al reported that younger patients and women with psoriasis had greater impairment in quality of life as compared with men and older patients, potentially explaining why younger patients and women prefer treatments with greater likelihood of benefit.34
We also found that patients with higher educational attainment least preferred prolonged inpatient treatment, possibly because greater education correlates with acquiring a position with higher earning potential, greater responsibility,35
and therefore busier schedules that would limit the ability to devote blocks of time to treatment.
Findings from previous studies have shown that smoking is significantly associated with increased psoriasis severity, and heavy and long-term smokers are more likely to have severe psoriasis,36
potentially explaining why patients still smoking in our sample appeared to be more concerned with the probability of treatment benefit and reduction in psoriasis plaques. The correlations we observed between patient characteristics, treatment history, behavioral factors, and their treatment preferences () raise the possibility that patient “profiles” or patient preference subgroups may exist. Acknowledging these profiles, if present, may be useful as clinicians develop treatment recommendations to optimize adherence.
To our knowledge, this is the first study to examine patients’ treatment preferences in terms of features they most and least prefer. However, there are studies that have examined preferences for treatment attributes. Similar to a study reporting that time to achieve moderate improvement was rated higher than time to relapse,13
our results suggest strong patient preferences for treatment with high (100%) probability of benefit as compared with a long (6–8-month) duration of benefit (or time to relapse). These findings may indicate that patients are more concerned with the onset than the duration of treatment benefit. Further, in our previous study, location of treatment was the most important treatment attribute.12
From our current results, we may interpret this prior finding as arising from patients’ strong disutility for inpatient treatment associated with hospital stays of 3 weeks. Strong preferences for the location of treatment, in particular, suggest that patients are concerned and have opinions about the experience of treatment beyond its ultimate outcomes. This conclusion is also supported by other studies that have used conjoint analysis in the health care setting to assess the value of process versus outcomes of treatments.12
The findings presented here justify further research. When examining patient preferences, future preference elicitation studies should consider both preferences for most and least preferred treatment options or attributes. A focus solely on aspects of treatment that a patient prefers may provide incomplete information on forces that influence treatment choice and adherence.
Strengths and limitations
We used choice-based conjoint analysis to measure patient preferences for psoriasis treatment attributes, a method considered to represent best the way people make everyday choices and which is easy to use and efficient in assessing patient preferences.18
Moreover, the hierarchical Bayes method we used to estimate individual-level preference scores is said to improve the reliability and predictive validity of individual preference models because the method borrows information from other respondents in the sample to stabilize estimates for each individual.26
However, it is important to note that, in preference elicitation, simplifications are made to make the exercises feasible. In conjoint analysis we assume, for example, that the utility functions attributable to respondents are additive and preferences for attribute levels are independent. However, preferences for attribute levels may not be independent in all cases.18
Further, although it is a strength of our study that we accounted for individual characteristics in our multivariable models, there might be individual differences even within these subgroups that are not captured by the rather broad categories we used. Therefore, treatment recommendations may need to be individualized to an even greater extent than suggested by our reference to patient profiles or preference subgroups. It may also be considered a limitation of this study that treatment attributes and attribute levels were specified without input from the patients. However, during the pilot phase of the study, feedback from the patients was used to refine and improve the comprehensiveness and clarity of the identified features of psoriasis treatment. Finally, our study sample included only patients with moderate-to-severe psoriasis, and the treatments they were offered may have differed from those offered to patients with mild psoriasis. Therefore, the generalizability of the study sample is limited.