2.1 Study Area
The study was conducted in the two governmental Hospitals, Fiche and Chancho ART clinics. Fiche is located 112km north of Addis Ababa latitude of 9°48′N and longitude of 38°44′E and an elevation between 2,738 and 2,782 meters above sea level. Chancho is located about 40km north of Addis Ababa, with a latitude of 9° 18' 29.1240'' N and longitude 38° 45' 11.2320'' E.
2.2 Source of Population
The target population was obtained from the two hospitals attending treatment between 1st September 2016 and 30th August 2018. The data were extracted from the patient chart and identification card. Both contain epidemiological, laboratory, and clinical information of all Diabetic patients receiving treatments in follow-up.
2.3 Study Design
A retrospective study was conducted to obtain secondary data among diabetic patients attending treatment from September 2016 to August 2018 in the hospitals.
2.4 Eligibility Criteria
All Type I and Type II diabetic patients and placed under insulin and metformin attending follow-up in the study area. Patients of all ages who are the follow-up were included in the study. Patients who were both diabetic and hypertensive were included in the study. Patients who were out of the study period were excluded from the study.
2.5 Operational Definitions
Body mass index: Calculated from the patients’ weight (in kilograms) divided by their height (in meters) squared.
Diabetics Mellitus: According to the American Diabetes Association should be 5.0–7.2 mmol/l (90–130 mg/dL) before meals and less than 180 mg/dL (10 mmol/L) after meals.
Insulin: is a hormone produced naturally in humans and animals in the beta cells of the pancreas.
Mg⁄dL: It stands for milligram per decilitre. It is a measurement that indicates the amount of a particular substance (such as glucose) in a specific amount of blood.
2.6 Sampling Technique
The sampling technique, the recorded identification card of patients, was filtered first from their cards according to their entry to the follow-up. Then, patients were clarified using inclusion criteria, given a code for the remaining records, and selected each recorded card for the study using simple random sampling to get real, relevant, and detailed information from diabetic patients who follow up at hospitals.
2.7 Sample Size Determination
The sample size was calculated using single population proportion formula, taking into account the following premise: 95% confidence level, 5% margin of error, and progression of fasting blood glucose level was 5% (Gebermariam et al. 2020), design effect=0.05, and non-response rate=10%.
2.8 Data Collection Tools and Procedures
Medical records were reviewed by using a checklist. Both the inpatient and outpatients’ records were reviewed during the study. The data was collected from the patient’s identification card and charts.
2.9 Data quality assurance
The data collectors were familiar with medical cards review. The three-day training was given for data collectors and supervisors about the study's objectives and the data collection process. Strict supervision was assumed to meanwhile, any doubt in the checklist was clarified.
2.10 Study Variables
The dependent variable was fasting blood glucose level for three years. The independent variables were age, BMI, systolic blood pressure, and diastolic blood pressure were continuous variables, whereas sex, functional status, residence, regimen, education level, and marital status were categorical variables (Table 1).
Table 1: Independent variable with their categories
Variables
|
Categories (if any)
|
Sex
|
0=male, 2=Female
|
Baseline Age (in years)
|
|
Marital status
|
0=Married, 1=Others (single, widowed or Divorced)
|
Body Mass Index Kg/m2
|
|
Observation time (in month)
|
|
Residence
|
1= Urban, 2= Rural
|
Education Level
|
1= Illiterate, 2= Primary, 3=Secondary, 4=Above Secondary
|
Functional status
|
1= Working, 2=Bedridden
|
Regime
|
1= Insulin Agent 2=Oral Agent
|
Systolic Blood Pressure (MmHg)
|
|
Diastolic Blood Pressure (MmHg)
Clinical Diagnosis
|
1=Type I 2=Type 2
|
2.11 Statistical Data Analysis
2.11.1 Exploratory Data Analysis
To determine the evolution and balances of the data, the individual and mean profiles to time were plotted. The mean, the variance, and the correlation structures were also explored through graphical techniques. A random-effects model was chosen to define a covariance model in parallel to the fixed-effects model. After deciding the fixed effects, the study selected a set of random effects in the model.
2.11.2 Linear Mixed Model (LMM)
The study considered the changes in the fasting blood glucose (FBG) levels in diabetes patients on treatment throughout the study. Longitudinal data can be analyzed using various methods, but the approach employed in this study was to fit the linear mixed-effects (LME) model. This approach is pliant, considers the natural heterogeneity within the population, and can effectively handle drop-out and missing data. A general LME model can be written as (Laird and Ware 1982).
Statistical inference for a linear mixed model is typically based on the maximum likelihood method or the restricted maximum likelihood method (Laird and Ware 1982; Lindstrom and Bates 1988). The longitudinal study analysis using SAS version 9.4 was used for statistical analysis and graphics, and a 5% level of significance was used in statistical tests.