The first round of the study has been initiated in Kerman in 2009 to recruit 5900 subjects aged between 15 to 75 years old.
The details of the recruited sample and the key findings are illustrated in . To maximize the sample size to be efficient for all the NCD and their risk factors, we have calculated the sample size for a prevalence of 50%, and consider the precision as the level of 5%. The initial calculation for the sample size was 384.16 cases. We did not apply the Finite Population Correction. However, we have corrected the sample size based on eight strata (15–24, 25–34, 35–54, 55–74 by sex), the design effect of 1.5 and the response rate 78% where lead to 5900 as the final sample size ().
Standardized estimates for the clinical and para-clinical findings in KERCADR study round one, (N=5900), Kerman, Iran, 2009–11
the sample size calculation formula
The project has been funded by Kerman University of Medical sciences with the grant number 88/110.
The sampling method was a one stage cluster sampling. In the first stage, 250 postal codes (called seeds) were selected randomly among an updated roster of residential addresses in provincial post office. The data collection team firstly mapped the seeds and then contact them one by one. After briefing the household’s member, all the eligible members (15–75 years old) have been listed in the Kish household coversheet and recruited to the study.
The written informed consent was signed by all of the participants after ensuring their well-understanding of the harm and benefits of the participation in the survey. It should be noted that the study protocol and procedures were reviewed and approved by the Research Review Board of the Kerman University of Medical Sciences (Ethic code 88–110KA).
In the case of any household being absent for twice, the other neighborhood households from the right direction of the seed were approached systematically and with the same method, eligible people were asked to participate in the study. The recruitment was continued to reach 24 subjects in each cluster. All recruited people were given an appointment card having the date, time and the place of attending collaborating clinic for blood sampling and face-to-face interview. They were asked to be fasted for 12–14 hours before the appointment time in the morning and bring their medicines with themselves.
In , the age distribution of the recruited sample is compared with the target population in Kerman (National consensus of population size for 2006). Base on the sampling methods (non-proportionate to size), the younger age groups were under-sampled and the older groups have been over-sampled. Although this will bring more precision to age-stratum specific estimates, the total combined estimates have to be always standardized based on the real age distribution of the target population.
The age distribution in the sample and target population, (N=5900) Kerman, Iran, 2009–11
The standardized estimates (based on age and sex subgroups) for the key measures of clinical and para-clinical findings were presented separately for men and women in . The standardized estimates are lower than the crude estimates. For sake of simplicity, we did not present crude estimates (as they have been confounded estimates). The reason for this confounding is the over-sampling of the older age-groups in the survey and the direct standardization has removed this sort of confounding.
For further analysis, we have considered household as the primary sampling unit (cluster). Proportions of categorical and ordinal variables reported by sex group and then compared using Chi square test. The means and 95% confidence interval (CI) of numerical variables were reported by sex group and compared using Post-Survey Estimation Student T-test. Association between disorders and risk factors were calculated using Poisson regression model. All statistical analyses were conducted under survey data analysis by STATA v.11. P-Value less than 0.05 considered as significant level.