Study design
This study is a cross-sectional survey. Based on the estimated sample size of the survey, a simple random sampling method was used to interview and investigate 34 community day care centers under the jurisdiction of Fuzhou City, Fujian Province, and 30–40 cases of elderly individuals were randomly selected from the health records of each day care center in the community. If the following situations occurred, re-sampling was conducted: (1) the selected individuals refuse to participate; (2) the archival information is old or false; or (3) the selected individuals have died and therefore, the corresponding individuals cannot be contacted.
Study subjects
Individuals aged55-75 years; those with permanent residence in communities of Fuzhou; those having normal activities of daily living and self-care or basic self-care ability; and those able to understand and cooperate, voluntarily participate, and sign the informed consent form were included in the analysis.
The exclusion criteria were as follows: (1) GDS-15 > 8 points[6], or a history of depression; (2) presence of brain tumors, Parkinson’s disease, or other unstable internal medical diseases that can affect brain function or the assessment of cognitive function; (3) an acute disease history within the past 3 months; (4) current diagnosis of active epilepsy; (5) secondary disturbance of sleeping-waking rhythms caused by somatic diseases or mental disorders; and (6) rejection or poor cooperation with the research.
MCI diagnostic criteria were as follows[7]: (1) cognitive impairment reported by patients or informant, or detected by experienced clinicians; (2) objective evidence of impairments in one or more cognitive fields (from cognitive testing); (3) slight impairments in complex instrumental activities of daily living, but maintenance of the ability to independently lead daily life; and (4) No current diagnosis of dementia.
Sample size estimation
According to the literature, the prevalence rate of cognitive impairment in older adults staying at home and in nursing institutions for the aged in Fuzhou is 30.17%[8]. The test level α was set at 0.05, and the permissible error δ was 3%. The sample size according to the calculation formula was about 931 subjects. Considering the possibility that 15% of survey data are incomplete, the survey sample size was accordingly adjusted to be approximately 1,071.
The calculation formula for sample size is
Evaluation item
(1) General survey
Demographic data: age, gender, ethnicity, years of education, marital status, current state of residence, status of work, etc.
Behavior and lifestyle: diet structure, sports activities, social participation, etc.
Health status: This included self-reported smoking and drinking status, past medical history, and medication use and height, weight, BMI, and blood pressure as measured by the assessors.
The cardiovascular risk scoring was conducted according to the “Chinese ICVD 10-year Morbidity Risk Assessment Form” recommended in Chinese Cardiovascular Disease Prevention Guidelines[9]. Based on the levels of seven risk factors of subjects (age, gender, systolic blood pressure, body mass index, hypercholesterolemia, smoking status, and diabetes status), the total scores for cardiovascular risks were calculated.
(2) Cognitive function assessment
MoCA scale of Fuzhou version: The MoCA scale of Fuzhou version that has passed the reliability and validity test[10] was used to assess the subjects’ cognitive function in a face-to-face interview, involving eight cognitive fields: executive function, visual space structure, memory, attention, speech fluency, abstract ability, calculation ability, and orientation ability. The total score for the scale is 30 points; ≥24 points, and 19–24 points if the years of education are ≤6 years, indicate normal cognition, and 14-19 points indicate MCI[11].
AD8 Dementia Screening Scale: The AD8 Dementia Screening Scale was compiled by the University of Washington in 2005, which involves a total of eight items[12]. In its Chinese version, ≥2 points is considered as the cut-off value for cognitive dysfunction, with the sensitivity of 85.7% and the specificity of 77.6%[13]. The Chinese version of AD8 quicker and is convenient for the elderly to understand and self-assess. Accordingly, it is widely applied in non-specialist medical institutions such as communities and general medicine.
(3) Instrumental activity of daily living
The scale is compiled by Lawton et al., with good reliability and validity[14]. The scale involves eight items such as phone use, shopping, food cooking, housekeeping, clothes washing, transportation, drug administration, and financial management. The total score ranges between 0 and 23. The higher the score, the more complete the ability of daily living. A score that is 2 standard deviations less than the norm indicates that the ability of daily living is severely impaired[15].
(4) Subjective sleep quality assessment
The subjects completed the assessment of the Pittsburgh Sleep Quality Index (PSQI) through self-assessment[16]. Its Chinese version has passed the reliability and validity test[17]. The PSQI scale is composed of 19 self-assessment items, constituting seven dimensions such as sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbance, hypnotics, and daytime dysfunction. Each dimension is scored from 0 to 3, with a total score of 21 points. The higher the points, the worse the sleep quality. The result of 0-5 points indicates good sleep quality; points ≥ 6 indicate sleep disturbance.
(5) Determination of TCM constitution
The data of the surveyed subjects were collected according to the 33 items in the standard questionnaire of “TCM Service Record Sheet for the Elderly” issued by the State Administration of Traditional Chinese Medicine in 2013, and the results of constitution determination were analyzed. The types of TCM constitution included biased constitution (Qi deficiency, Yang deficiency, Yin deficiency, phlegm, damp-heat, blood stasis, Qi stagnation, and special constitution) and gentle constitution. Criteria for the identification of biased constitution were: “Yes” if the cumulative score of all items ≥11; “Tendency” if the cumulative score of all items =9-10; “No” if the cumulative score of all items ≤ 8. Criteria for the identification of gentle constitution: “Yes” if the cumulative score of all items ≥17 and the final score of each of the other eight constitutions < 8; “Roughly yes” if the cumulative score of all items ≥17 and the final score of each of the other eight constitutions <10.
(6) Evaluation of the simplified geriatric depression scale
The geriatric depression scale has been simplified by Burke et al. [18], and the Cronbach’s α coefficient for internal consistency in the Chinese version of the scale is 0.82[19]. The scale involves 15 items, among which items 1, 5, 7, 11, and 13 have negative scoring, and the remaining 10 items have positive scoring. A score of 0 or 1 point is given to each item, with the maximum score of 15 points. The higher the score, the more obvious the tendency of depression. If GDS-15 score is higher than 8 points, it indicates the existence of depressive symptoms.
Ethical approval and clinical trial registration
This study strictly follows “The Helsinki Declaration” by the World Medical Association and has passed the ethical review by the ethics committee of the Rehabilitation Hospital affiliated to Fujian University of Traditional Chinese Medicine (2019KY-002-02) and the Second Affiliated Hospital of Fujian University of Traditional Chinese Medicine (SPHFJP-K2019001-1). The research project is registered in the Chinese Clinical Trial Registry (http://www.chictr.org.cn/index.aspx,ChiCTR2000039411).
Statistical analysis
The SPSS 24.0 software was used for data processing and analysis. Independent sample t test was used for the intergroup comparison of measurement data conforming to the normal distribution; the Mann-Whitney U test was performed for the intergroup comparison of the measurement data not conforming to normal distribution; the χ2 test was used for the intergroup comparison of disordered classification data; chi-square trend test and Wilcoxon test were used for the intergroup comparison of ranked data. Inspection level was set at α=0.05. The binary logistical stepwise regression model was used to screen the influencing factors, and the OR value for relative risk of a single factor and the corresponding 95% confidence interval were obtained. The dependent variables were MCI events (binary classification), and the exposure variables were age, gender, years of education, BMI, ICVD, marital status, status of work, exercise, smoking, PSQI score, various sleep dimensions, and TCM constitution.
Based on the selected potential influencing factors, multi-factor logistical regression model was constructed:
(1) Model 1: a crude model only with the exposure as independent variables;
(2) Model 2: Based on of Model 1, the general demographic data such as age and the years of education were adjusted;
(3) Model 3: A Fully adjusted model. Through the Directed Acyclic Graph (DAG) theory[20], the causal relationship network was explored to determine the independent variables that fit the model, with the focus on exploring the association of subjective sleep characteristics and Qi-stagnation with the MCI outcome (Figure 1). After the construction of different models, the Akaike information criterion (AIC) and Area Under the Curve (AUC) were calculated, and the AIC and AUC of different models were compared. The smaller the values of AIC and AUC, the better performance of the model.
The Delta method was employed to analyze the additive interaction between sleep disturbance and Qi-stagnation, and the obtained regression coefficient β for the independent factors and interactive items was substituted into the Excel table compiled by Andersson[21] to calculate the three important indices for the interaction strength of the additive model: relative excess risk due to interaction (RERI), attributable proportion due to interaction (AP), and Synergy index (S) as well as their 95% CI. When the 95% CI of RERI and AP does not contain 0, or the 95% CI of S does not contain 1, it is considered that an additive model interaction synergy exists.
Missing data processing: If the exposure factors (PSQI scale, TCM constitution scale) in the research hypothesis was missing without being evaluated, they were removed. For other missing data, the multiple imputation method was used to perform statistical analysis on the filled data.