TRADING OFF BENEFIT AND RISK OF SIDE EFFECTS FOR PERSONALIZED MEDICINE? A CONJOINT ANALYSIS STUDY

Purpose:  

   Discrete choice experiments in medicine usually focus on the patient’s perspective, however, the method can also be used to examine how a clinician trades off risks and benefits when evaluating a treatment.  In an era of personalised medicine, clinicians may be more inclined to recommend treatments to which they believe their patients are more responsive.

Method:   

   We developed twenty-six vignettes showing an individual patient’s change score on the positive and negative syndrome scale of schizophrenia (PANSS), a widely used symptom report scale in psychiatry, after either of two possible treatments for major psychosis.  Each treatment bore a given risk of side effects in terms of the probability of a ten kilogram weight gain and was associated with specified acute hospital treatment costs.  Psychiatrists were also informed of whether or not the patient possessed a “hyper-responsiveness” genotype (for one of the alternative treatments).  The clinicians (n=53) were asked to determine if they believed the patients had responded to treatment and how confident they were in their judgement.  After considering the risks of side effects and costs of each treatment, clinicians decided which of the two treatments they would recommend.

Result:   

   We applied multi-nomial logit models to investigate the way that trade-offs between risks and benefits are considered by the psychiatrist when prescribing a treatment.  Our results showed that increases in both the risk of weight gain and number of days in hospital associated with the treatment decreases the probability of prescribing that treatment (OR=.92;.91:p=<.001;<.001 respectively).  Clinicians were more likely to prescribe a treatment when they believed the patient change score (on the PANSS scale) was large enough for them to be classified as a “responder”, but less likely to prescribe a treatment when the patient possessed the hyper-responsiveness genotype.

Conclusion:

   When recommending a treatment, psychiatrists seem to take appropriate account of the risk of treatment side effects and of their individual judgement of whether the patient is a "responder". The major issue in the recommendation of treatment was judgement of response, the threshold for which was not affected by genotype.