Physician-Patient Relationship and Medication Compliance: A Primary Care Investigation

Kerse et al. present their original and important work on patient- provider relationships and medication adherence. Because the patient-provider relationship is so potentially complex and difficult to model, the investigators are to be commended for their thoughtful selection of multiple measures, which also allowed them to identify the relationship feature that best accounts for subsequent medication adherence. The sample was large and representative, and followup attrition was fairly low. The implication of their findings is that interventions to enhance short-term medication adherence might target patients who report that their provider disagrees or does not understand the patient’s perspective on the nature of the presenting problem.

And yet, before the findings can guide intervention design, it will be important clarify the meaning of concordance… Did patients actively disagree with their provider, or vise-versa, or both? Does a patient’s “feeling misunderstood” correlate with actual within-encounter behavior patterns? Who is responsible for the misunderstanding: the patient, the provider, the healthcare system, or some other entity? Could feeling misunderstood actually be a proxy for generalized distress? Or perhaps correlate with drug- seeking, secondary gain, or some other inappropriate use of services? Which of these many factors are operating, and which are appropriate for intervention?

Another issue is that readers should bear in mind that 4-day adherence to an diverse set of acute regimens has very different determinants than long term adherence to a chronic disease regimen. Adherence even varies widely between chronic diseases (due to differences in illness severity and illness meanings) and regimens for a given chronic disease (due to differences in regimen-specific barriers like side effects). One of the best-known adherence predictors is symptom severity, which was beyond the study focus. It is even possible that the discordance was about the severity of presenting symptoms.

Finally, I wonder whether the prediction model might vary whether the actual patient was an independent adult that was seeking his or her own care, versus a child or non-autonomous adult.

James E. Aikens, Ph.D.

Competing interests:   None declared