Validation of an Announced Telephone Pill Count Compared to a Home-Visit Pill Count in People With Type 2 Diabetes or Cardiovascular Disease

We aimed to assess the validity of an announced telephone pill count in people with type 2 diabetes or cardiovascular disease by comparing this method to a home-visit pill count. We also assessed whether a second telephone pill count improved accuracy. People aged ≥35 years using oral type 2 diabetes or cardiovascular disease medication were included. Thirty-four participants completed a telephone pill count followed by a home-visit pill count, and a subsample of this population (n = 11) completed a second telephone pill count. Scatterplots were used for a visual representation of the number of pills counted with both methods, intraclass correlation coefficients for agreement, and Bland-Altman plots for absolute differences and outliers. A total of 203 pill counts were conducted. The study population consisted of 53% men, with a mean age of 69.6 (±9.2) years and an average of 6.1 (±2.8) medication prescriptions per participant. Scatterplots showed that pills counted with both methods were mostly scattered around the y = x equation. Agreement between the first telephone pill count and home-visit pill count was high, with intraclass correlation coefficients of 0.96 (medication count level) and 0.98 (individual level). No learning effects were observed in the subsample (n = 11), the intraclass correlation coefficient for the first telephone pill count was 0.88 versus 0.89 for the second telephone pill count. Bland-Altman plots indicated high agreement between the two methods. An announced telephone pill count is considered a valid alternative for a home-visit pill count in people with type 2 diabetes or cardiovascular disease. A single pill count appears sufficient.