Study design
In this study, a total of 2426 people from the baseline (second stage of the first phase) of a population-based prospective cohort study, the Bushehr Elderly Health (BEH) program (18, 19) were included in the analysis. The purpose of the BEH study, whose methodology has been described elsewhere (18, 19), is to investigate the incidence of non-communicable diseases and associated risk factors among people 60 years and older. The participants were selected through multi-stage stratified cluster random sampling in Bushehr, southern Iran (18). The first stage of the first phase of the BEH program was implemented from March 2013 to October 2014. Prevalence of cardiovascular risk factors was investigated among 3000 men and women (participation rate = 90.2%) in this stage (18). The second stage of the first phase of the study was conducted 2.5 years later on 2772 eligible persons from the first stage. The prevalence of musculoskeletal and cognitive diseases and their risk factors was investigated in this stage. 78 subjects eligible to participate in the second stage of the first phase, due to death, and 268 subjects due to loss to follow-up, did not attend at this stage (12.5%) (19). Both stages of the first phase of the study are the baseline phase of the BEH program, and each stage has been focused on measuring the prevalence of the specific groups of non-communicable diseases and their risk factors (18, 19). Follow-up of the non-communicable disease incidence in the enrolled subjects will be done in the next phases of the BEH program.
Measurements Outcomes
In this study, disability was measured by two questionnaires of Basic Activity of Daily Living (BADL) using Barthel scale(3), and Instrumental Activity of Daily Living (IADL) using Lawton (2) scale, through face to face interview with the participants by the trained questioners. The validity and reliability of these questionnaires were previously assessed in Iran and were at acceptable levels (20, 21). The BADL questionnaire has 10 items including eating, bathing, urine control, toilet use, moving from bed to chair and vice versa, dressing, self-cleaning, stool control, climbing stairs, and ability to move on a flat surface. The IADL questionnaire has 8 items including the ability to use the phone, cooking meals, washing clothes, taking medication, shopping, housekeeping, transportation, and financial ability. These two questionnaires assess the degree of dependency of older adults. In the BADL questionnaire, the subjects with total scores of <95 on the Barthel scale were considered as a dependent, and those with scores of 95 to 100 were considered as independent. The subjects with total scores of 0 to 7 of the IADL questionnaire were also defined as dependent and those with a score of 8 were defined as an independent.
Cardio-metabolic and socio-demographic risk factors
In this study, we considered cardio-metabolic and socio-demographic risk factors as independent variables. Cardio-metabolic risk factors refer to risk factors that increase the chance of experiencing cardiovascular events, such as age, sex, obesity, hypertension, dyslipidemia (high LDL cholesterol, high triglycerides, and low HDL cholesterol), dysglycemia, smoking, abdominal obesity, lack of consumption of fruits and vegetables, and sedentary lifestyle.
Socio-demographic characteristics including sex, age, marital status, and education level, as well as information on cardio-metabolic risk factors including history of diabetes mellitus and hypertension, smoking, physical activity, and daily intake of calories, were collected using the standardized questionnaires. Diabetes mellitus was defined as current fasting blood sugar ≥ 126 mg/dL, or HbA1c ≥ 6.5, or subject’s self-reporting of diabetes mellitus based on a doctor's diagnosis, or current use of anti-diabetic drugs. Hypertension was defined as current systolic blood pressure ≥ 140 mmHg or diastolic blood pressure ≥ 90 mmHg, or subject’s self-reporting of hypertension based on a doctor's diagnosis, or current use of anti-hypertension drugs. Smoking refers to current or past use of cigarettes or hookahs or pipes. Hookahs is a single- or multi-stemmed instrument for vaporizing and smoking flavored tobacco, whose vapor or smoke is passed through a water basin-often glass-based-before inhalation. Physical Activity was measured using Aadahl et al. physical activity questionnaire (22). Daily intake of calories (Kcal) was assessed by a standardized 24-hour dietary recall questionnaire.
Anthropometric measurements including height, weight, and waist circumference as well as laboratory measurements including LDL, HDL, total cholesterol, and triglyceride were performed under standard conditions with calibrated instruments. Anthropometric measurements were taken with shoes removed and the participants wearing light clothing. Height and weight were measured with a fixed stadiometer and a digital scale according to the standard protocol. A flexible, circumference measuring tape is used to measure the waist (WC). WC should be measured at a point midway between the iliac crest and the lowest rib in a standing position (19). High WC was defined as waist circumference ˃102 cm in males and ˃88cm in females.
An overnight fasting venous blood sample was obtained for every participant for biochemical measurements. A total of 25 cc of whole blood is collected by a trained nurse. Fasting blood sugar was measured by enzymes (glucose oxidase) colorimetric method using a commercial kit (Pars Azmun, Karaj, Iran). HbA1C was measured by boranate affinity method using a CERA-STAT system (CERAGEM MEDISYS, chungcheongnam-do, Korea). Total cholesterol was measured by enzymatic (cholesterol oxidase phenol aminoantipyrine (CHOD-PAP)) colorimetric method using a commercial kit (Pars Azmun) (19). High total cholesterol was considered as total cholesterol ≥ 200 mg/dL. HDL cholesterol was measured by enzymatic (cholesterol esterase and cholesterol oxidase (CHE & CHO)) colorimetric method using a commercial kit (Pars Azmun) (19). Low HDL cholesterol was defined as high density lipoprotein cholesterol <40 mg/dL in males, and <50 mg/dL in females. LDL cholesterol was measured by enzymatic (CHE & CHO) colorimetric method using a commercial kit (Pars Azmun) (19). High LDL cholesterol was considered as low-density lipoprotein cholesterol ≥ 110 mg/dL. Triglyceride was measured by enzymatic (glycerol-3- phosphate oxidase phenol aminoantipyrine (GPO-PAP)) colorimetric method using a commercial kit (Pars Azmun) (19). High serum triglyceride was defined as serum triglyceride ≥ 150 mg/dL.
Statistical analysis:
In the descriptive analysis, we used the mean (standard deviation), and number (percent) for the continuous and categorical variables respectively. To investigate the association between the risk factors and dependency in IADL and BADL, at first, the directed acyclic graph (DAG) was depicted based on the literature review, considering the activities of daily living as the outcome and the cardio-metabolic and socio-demographic risk factors as the explanatory variables (Figure 1). We used the DAG to help us choose the proper covariates (confounders, not intermediate or collider variables) to enter into the multiple models. The logistic regression model via Hosmer and Lemeshow suggested strategy (23) was used to investigate the association between BADL and IADL and cardio-metabolic and socio-demographic risk factors based on the causal graph. The risk factors that their effect should be controlled based on the DAG, as well as the P-value of their association with the outcome was ≤ 0.25 in the bivariate analysis, were entered into the multiple logistic model. Then the included risk factors were removed from the model one by one when they lost their significance while checking via a likelihood ratio test. Afterward, the statistical significance of the plausible interaction terms between the remaining risk factors in the model was assessed. We chose the interaction terms based on the literature review and expert opinion. No interaction term was statistically significant. Eventually, the goodness of fit of the final model to the data was checked out using the Hosmer-Lemeshow test. Stata version 15.1 was used for statistical analysis.