Non-adherence is common and contributes to adverse health outcomes, reduced quality of life, and increased healthcare expenditure. The objective of this study was to assess the diagnostic validity to estimate the prevalence of non-adherence in patients with mild cognitive impairment (MCI) and dementia using two self-reported methods (SRMs) that are useful and easy in clinical practice, considering the pill count as a reference method (RM).
The cohort study was nested in a multicenter randomized controlled trial NCT03325699. A total of 387 patients from 8 health centers were selected using a non-probabilistic consecutive sampling method. Inclusion criteria were as follows: a score of 20–28 points on the Mini-Mental State Examination (MMSE); older than 55 years; taking prescribed medication; and are in charge of their own medication use. Participants were followed up for 18 months after the baseline visit, i.e., 6, 12, and 18 months. Variables related with treatment adherences were measured in all visits. The variables included age, sex, treatment, comorbidities, and the MMSE test. Adherences included pill counts and Morisky–Green test (MGT) and Batalla test (BT) as SRMs. Statistical analysis included descriptive analysis and 95% confidence intervals (CIs). The diagnostic validity included the following: 1) open comparison statistical association between SRMs and RMs and 2) hierarchy comparison: the RM as the best method to assess non-adherence, kappa value (k), sensitivity (S), specificity (Sp), and likelihood ratio (PPV/PPN).
A total of 387 patients were recruited with an average age of 73.29 years (95% CI, 72.54–74.04), of which 59.5% were female. Comorbidities were 54.4% HTA, 35.9% osteoarticular pathology, and 24.5% DM. The MMSE mean score was 25.57 (95% CI, 25.34–25.8). The treatment adherence for the RM oscillates between 22.5% in the baseline and 26.3%, 14.8%, and 17.9% in the follow-up visits. For SRMs, the treatment adherence oscillates between 43.5% in the baseline and 32.4%, 21.9%, and 20.3% in the follow-up visits. The kappa value was statistically significant in all the comparison in all visits with a score between 0.16 and 035. Regarding the diagnostic validity, for the MGT, the sensibility oscillated between 0.4 and 0.58, and the specificity oscillated between 0.68 and 0.87; for the BT, the sensibility oscillated between 0.4 and 0.7, and the specificity oscillated between 0.66 and 0.9; and when both tests were used together, the sensibility oscillated between 0.22 and 0.4, and the specificity oscillated between 0.85 and 0.96.
SRMs classify non-adherent subjects correctly. They are very easy to use and yield quick results in clinical practice, so SRMs would be used for the non-adherence diagnosis in patients with MCI and mild dementia.