An oncologist discusses published PRO data from an RCT (see Scenario 1) with a patient with metastatic gastric cancer. Based on the PRO data, the patient prefers regimen B (the triplet of cytotoxic agents) over regimen A, because QoL scores of this group are higher at follow-up. Whether or not the patient would undergo the same response shift as the study respondents is not certain. Not knowing about the response shift causing the higher HRQoL scores means that the patient’s decision is not fully informed. Consequently, the patient may be overtreated, resulting in unnecessary side-effects and lower health state at follow-up than regimen A would have yielded.
The example is problematic from the perspective of nonmaleficence. At the moment of the decision, no harm is done yet. However, the overtreatment that may be the consequence, leading to a worse health state, equals ‘doing harm’. In addition, the principle of autonomy is at stake as well, since the decision is not fully informed. Whereas possible differences between study groups and the individual patient - such as gender, age, and possibly lifestyle - are ideally taken into consideration, influences of response shifts and dispositions are less well-known and rarely discussed in SDM. However, the patient is still included in the decision making and informed about options, expected benefits and risks. Therefore, this may be considered only a minor violation of the autonomy principle, especially as it is not possible to tease out all health changes from response shift and disposition in PRO data.