Eligibility and enrollment of participants
Participants diagnosed with mild-to-moderate AD according to the 2011 the National Institute of Neurological Disorders and Stroke–Alzheimer Disease and Related Disorders (NINCDS-ADRDA) criteria were enrolled in this prospective study from Guangzhou First People’s Hospital from June 2014 to June 2018. The participants were evaluated and screened using the Mini-Mental State Examination (MMSE) score and Global Deterioration Scale (GDS). Patients with the MMSE score ranging from 11 to 26 points with the graduated primary school education level or above, patients with the MMSE score ranging from 11 to 22 points with primary school education level, and patients with GDS scales between mild and moderate degree were included in the study. Patients with severe and extremely severe dementia or with dementia caused by vascular diseases, depression, and other diseases were excluded. Patients under the primary school education level were excluded due to the need for the ADAS-Cog examination. All patients should have stable and well-communicated caregivers. Fifty-eight potentially eligible patients signed inform consent form. Twenty-seven patients were excluded because of lacking on collaboration in pre-experiment, such as caregivers did not have enough time to monitor medication intake for patients, or patients could not finish all scales in any reasons. Finally, thirty-one participants finished the procedures.
Follow-up visit procedures
Follow-up visits were conducted for 10 months after enrollment, and then scale evaluations were implemented in the 1st, 2nd, 4th, and 7th months and at the end of the follow-up period (10th month). After the second follow-up, two patients dropped out of the study due to the changes in their health conditions, and one patient quit before the last follow-up for personal reasons.
Improving the medication adherence
Physicians, clinical research nurses, and clinical pharmacists participated in the study as researchers. The principal investigator clarified the importance of medication adherence to participating physicians, patients, and caregivers before the study. The physicians encouraged patients and their caregivers to maintain adherence during every visit. The clinical research nurses handed out medication record cards to caregivers. The clinical pharmacists created and maintained medication files for each patient independently, calculated the remaining pills before taking, distributed the medications of next phase and recorded the medication adherence by counting the pills. The caregivers supervised the patients in taking the drugs, recorded the card after taking medicine, and recorded the reasons for poor adherence, such as missing by mistake or barely forgetting to take medicine. White pills made from starch were used in the study.
Due to the frequent intake of medicine (two times per day) and pills (three pills per day), the medication adherence was directly measured by the pill counting method and rechecked using the medication record cards during the whole study period. The specific calculation method for the medication adherence rate was as follows: (Number of drugs dispensed minus the number of remaining drugs)/(Number of drugs prescribed per day× Number of days between two visits). For the treatment of overdose (i.e., the medication compliance rate exceeding 100%), the adherence rate started with 100%; the excess part was considered poor adherence, which was subtracted from 100% (if the actual measurement adherence was 125%, it was converted as follows: 100 – (125–100) % = 75%).
Implementation of scales and data collection
The data on name, age, sex, education level, relation between caregivers and patients, AD disease duration, and Apolipoprotein e 4 (APOE4) gene types of participants were collected after signing informed consent. Details on the MMSE score and GDS level of participants were collected after enrollment immediately. The same neurologist implemented the screening scales (MMSE and GDS) and all experimental scales (ADAS-Cog, ADL, and NPI) except the CIBIC-Plus. Another doctor independently completed the evaluation of the CIBIC-Plus scale. The scores of ADAS-Cog, ADL, NPI, and CIBIC-Plus scales were recorded at every visit. Because CIBIC-Plus is scoring on a 7-point Likert scale ranging from 1 (markedly improved) to 7 (markedly worse), we set no change as 0. The level of 3, 2, and 1 was set as markedly improved, moderately improved, and slightly improved, respectively. Also, the level of − 3, − 2, and − 1 was set as markedly deteriorated, moderately deteriorated, and slightly deteriorated, respectively. The NPI scale was divided into two parts: the total scores of the frequency and severity (frequency multiplied by severity) and the scores of caregivers’ own distress, being tagged as NPI score and NPI boring.
Data were analyzed using SPSS 23.0 software (SPSS Inc., IL, USA), and OriginPro 8.0 SR2 (OriginLab Corp., MA, USA) was used for creating graphics. The scores and medication adherence data of all scales were consistent with the homogeneity of variance but not with normal distribution. According to clinical experience (population characteristics, disease incidence, and medication habits of Chinese patients), the distribution of patients' cognitive function, daily behavioral function, and overall function followed the normal distribution; medication adherence rate, the mental and behavioral abnormalities followed the skewed distribution. CIBIC-plus scores were orderly graded into seven levels and considered as multiple sets of non-qualitative variables. Because continuous variables also can be considered as interval variables, the relation between the medication adherence rate and CIBIC-Plus scores was evaluated using the two-sample correlation analysis Kendall test, which was more efficient for ranked data. The correlation analyses for scales and adherence were performed using two-samples Spearman correlation analysis. The non-normal distribution data cannot undergo partial correlation analysis, and therefore scale scores and medication adherence rate were transformed before analysis being carrying out. An ascending order was found for ADL scores and medication adherence rate, while a descending order was observed for the remaining scales (the higher the scale score, the higher the functional deterioration). Our previous study showed that some objective factors (sex and disease course) related to patients significantly or nearly significantly correlated with medication adherence rate, so sex and disease course of patients were included as control factors in the partial correlation analysis after rank-sum transformation. Spearman correlation analysis was conducted between the 12 items of ADAS- cog scores and medication adherence separately. A significance level of p < 0.05 was assumed as statistically significant.