Contextualizing Evidence for Action on Diabetes - Population Survey

Objective: To provide rigorous epidemiological data on diabetes risk and morbidity in Health District 17D06 in Quito, Ecuador; and Eloy Alfaro District, located in the coastal region of Esmeraldas.

Design: The investigators will undertake two cross-sectional population surveys, with representative samples of the selected health districts, to estimate the prevalence of diabetes, impaired glucose tolerance, and NCD risk factors.

Sample design: Participants will be selected using multi-stage cluster sampling. We will use parishes or 'micro-areas' as primary Enumeration Areas (EA) for which official GIS shape files and population estimates are available. The EAs will be randomly selected with a probability proportional to its population density. In urban areas, after having sampled EAs, buildings or homes will be identified by randomly generated GPS points. The investigators will then sample one person at random in each household.

In rural areas, the investigators will randomly select villages from within the samples EAS, and then undertake the simple random sampling of individuals by using an existing census already developed by voluntary health promotors, and digitised by a local NGO. This census will be updated in the 3 months preceding sample selection to ensure accuracy.

Sample size: The sample size is proposed assuming the prevalence of diabetes is no higher than 10%, to provide an estimate with an absolute precision of +/- 3%, with a 95% confidence level. The investigators assume a design effect of 1.5 (recommended in STEPS guideline for complex designs in absence of locally available alternatives) and will further increased the sample size to allow for up to 20% refusal or loss due to individuals not attending their appointment for physical and biological measurements.

Survey procedures: The questionnaire will be designed according to the WHO STEPS NCD risk factor survey forms , and some survey-specific questions. Detailed demographic information will also be included following the WHO STEPS core questions and extended questions (highest level of education, ethnicity, marital status, employment status and household income). Cultural adaptations to the questions will be made as required.

The questions will be used to compile a purpose designed electronic survey form loaded onto tablets purchased for the research. The survey tool will be pre-piloted before the survey, after training the data collection team.

Thirty trained interviewers will establish survey eligibility, solicit informed consent, and carry out the survey in the participant's home. Composition of the survey team will be gender balanced, and include staff who represent local cultural, ethnic and religious groups. The survey will be carried out in Spanish but some interviewers will also be fluent in Quechua, Chachi and any other relevant indigenous language to translate questions verbally if required. Data collection in rural areas will engage the volunteer health promotors. The surveyors will arrange an appointment for physical and biological measurements together with the participant considering geographical convenience and availability. Individuals with a fasting blood glucose ≥ 126 mg/dl and/or a 2 hour plasma glucose ≥ 200 mg/dl, or blood pressure over 140/90 mmHg will be referred to their local health facility for care. All participants will receive the results of their physical measurements and blood analyses and will receive a health promotion booklet about diabetes, cardiovascular risk, and the importance of healthy diet and physical activity.

Physical measurements and laboratory: Weight, height, waist circumference will be measured and we will include limited biochemical measurements (fasting blood glucose, cholesterol and creatinine) analysed at regional reference laboratories. Participants will also undergo an Oral Glucose Tolerance Test (OGTT). Survey participants will be invited to a well-lit hygienic location after at least 8 hours fasting. Appointments will be first thing in the morning and participants will be instructed to eat or drink nothing but water from 10pm the night before. Instructions for fasting will be given on the appointment card given to the participants the survey day and telephone reminders will be sent when possible. All participants will receive breakfast after the OGTT test. Diabetic patients on medication will be asked to bring their medication and take it immediately after their breakfast after blood draw. They will not be asked to complete an OGTT. Similarly, pregnant women are eligible to participate in the study but will not undergo an OGTT, only fasting blood glucose. In order to ensure absence of pregnancy, the investigators will provide pregnancy tests to women in fertile age who are unable to confirm the absence of pregnancy at the household interview. Steps will be taken to minimise variability (at study zone level) in the pre-analytical treatment of samples. Blood extraction, storage and transportation will follow standard infection control measures.

Data management and analysis: Stickers with Quick Response (QR) codes that can be scanned by the android devices at each step will be used to ensure linkage of data collected in participant's home, their physical and biological measurements and laboratory results. The investigators will use and modify the WHO STEPS forms where possible and import them to Kobo toolbox free open source software (or ODK collect if necessary). Stata/SE (StataCorp, Texas, U.S.A.) Version 12.1. will be used for statistical analysis. The prevalence of impaired glucose tolerance, diabetes, and the other NCD risk factors will be calculated with 95% confidence intervals. Outcomes will be disaggregated by sex, age group, ethnicity and socioeconomic position. Geographical variation in survey outcomes will also be described. Steps will be taken to prevent missing data, but some level is unavoidable and we will incorporate methods analysing missing data or data from uncertain sources when necessary.