Baseline Profiles of Drug Prescriptions Prior to Diagnosis of Mild Cognitive Impairment (MCI) Obtained by Latent Class Analysis (LCA), and Assessment of Their Association with Conversion to Dementia

• Published in: Healthcare (Basel) Vol. 11; no. 15; p. 2219
• Main Authors: Gómez-Gómez, Carmen, Moya-Molina, Miguel Ángel, Tey-Aguilera, Manuel Jesús, Flores-Azofra, Jorge, González-Caballero, Juan Luis
• Format: Journal Article
• Language: English
• Published: Switzerland MDPI AG 07-08-2023; MDPI
• Subjects: Activities of daily living; Alzheimer's disease; Antidepressants; Antihypertensives; Cognition & reasoning; Cognitive ability; Dementia; Family physicians; Latent class analysis; Medical diagnosis; Medical records; mild cognitive impairment; modifiable risk factors; Polypharmacy; Prescription drugs; progression to dementia; Variables; mild cognitive impairment; latent class analysis; modifiable risk factors; progression to dementia; polypharmacy
• ISSN: 2227-9032; 2227-9032
• DOI: 10.3390/healthcare11152219

Polypharmacy has been linked to cognitive decline. However, interventions targeting modifiable risk factors, some of which are targets of the most commonly used drugs, could reduce the prevalence of dementia. Our aim was to determine the drug prescription regimen at baseline, prior to the diagnosis of mild cognitive impairment (MCI), and its possible association with progression to dementia. Data were collected from the electronic medical records of 342 MCI outpatients diagnosed during 2006-2017 at their first neurology consultation. We followed the classical three-step method of statistical analysis, starting with a Latent Class Analysis (LCA) to discover subgroups of drug prescription probability. Half of the patients were under polypharmacy (≥5 drugs), 17.5% had no recorded medication, 33.3% progressed to dementia (94.7% in ≤5 years), and 84.1% of them to Alzheimer's disease (AD). According to the LCA and based on 20 therapeutic indicators obtained from 240 substances and regrouped according the Anatomical Therapeutic Chemical Classification, we identified a four-profile model: (1) low (35.7% of patients); (2) mixed (28.7%); (3) cardio-metabolic (19.3%); and (4) psychotropic (16.4%). The binomial regression logistic model showed that profiles 2 and 3 (and 4 for AD), with a higher drug prescription conditioned probability against classic risk factors, were protective than profile 1 (OR = 0.421,
= 0.004; OR = 0.278,
= 0.000; OR = 0.457,
= 0.040, respectively), despite polypharmacy being significant in profiles 2 and 3 (mean > 7 drugs) vs. profile 1 (1.4 ± 1.6) (
= 0.000). Patients in the latter group were not significantly older, although being aged 65-79 years old quadrupled (OR = 4.217,
= 000) and being >79 tripled (OR = 2.945,
= 0.010) the conversion risk compared to patients <65 years old. According to the proposed analytical model, profiling the heterogeneous association of risk factors, which were taken prior to diagnosis, could be explored as an indicator of prior care and a predictor of conversion to dementia.