Community Health Department, Christian Medical College, Vellore, Tamil Nadu, India.
Sree Chitra Tirunal Institute for Medical Sciences and Technology, Trivandrum, Kerala, India; Global Institute for Public Health, Ananthapuri Hospitals and Research Institute, Trivandrum, Kerala, India.
Public Health. 2019 Mar;168:102-106. doi: 10.1016/j.puhe.2018.12.019. Epub 2019 Feb 6.
Most World Health Organisation (WHO) STEPS surveys use cluster sampling to assess the prevalence of risk factors for non-communicable diseases (NCDs) for which design effects need to be estimated using intracluster correlation (ICCs) coefficients, for sample size calculation. Although there are many reports of risk factor surveys reported from developing countries, there are very few reports of ICCs for risk factors for NCDs, which can inform planning the appropriate sample size needed for such surveys. This study reports the ICCs for NCD risk factors, obtained from a WHO STEPS survey conducted in Vellore district, in the state of Tamil Nadu, South India.
Cross-sectional study.
A cross-sectional study was carried out in 48 urban clusters (wards) and nine rural clusters (villages) between 2011 and 2012, using the WHO STEPS methodology for assessing behavioural, anthropometric, physical and biochemical risk factors. The ICC estimates for various risk factors were obtained using loneway and xtmelogit commands using STATA to study clustering of risk factors.
The number of respondents was 6196 adults aged 30-64 years. The median ICC of cardiovascular risk factors in the urban area was 0.046, while it was 0.064 in the rural area. Clustering was higher for behavioural risk factors such as physical activity (ICC: 0.179 rural, 0.049 urban) and fruit and vegetable intake (ICC: 0.105 rural, 0.091 urban) as compared with physical risk factors (ICCs for hypertension: 0.044 rural, 0.006 urban; body mass index: 0.046 rural, 0.041 urban) and biochemical outcomes such as fasting plasma glucose (ICC: 0.017 rural, 0.027 urban).
This study provides estimates of ICCs for cardiovascular risk factors from Vellore, South India, as such data have not been reported from WHO STEPS surveys in India or neighbouring countries. Such estimates of ICCs if reported from various WHO STEPS being carried out across the country can contribute to better planning of epidemiological surveys. Clustering of behavioural risk factors at village/ward level as seen in this study points to the need for community-based interventions for health promotion, as spatial clustering influences behaviour, which in turn affects chronic disease outcomes.
世界卫生组织（WHO）的大多数 STEPS 调查使用聚类抽样来评估非传染性疾病（NCD）风险因素的流行率，为此需要使用组内相关系数（ICC）来估计设计效果，以进行样本量计算。尽管有许多来自发展中国家的风险因素调查报告，但关于 NCD 风险因素的 ICC 报告却很少，这可以为规划此类调查所需的适当样本量提供信息。本研究报告了来自印度南部泰米尔纳德邦维洛尔区进行的 WHO STEPS 调查中获得的 NCD 风险因素的 ICC。
横断面研究。
2011 年至 2012 年期间，采用 WHO STEPS 方法评估行为、人体测量、身体和生化风险因素，在 48 个城市集群（区）和 9 个农村集群（村）中进行了一项横断面研究。使用 STATA 中的 loneway 和 xtmelogit 命令获得各种风险因素的 ICC 估计值，以研究风险因素的聚类情况。
共有 6196 名 30-64 岁的成年人接受了调查。城市地区心血管风险因素的中位数 ICC 为 0.046，而农村地区为 0.064。与身体风险因素（高血压的 ICC：0.044 农村，0.006 城市；体重指数：0.046 农村，0.041 城市）和生化结果（空腹血糖的 ICC：0.017 农村，0.027 城市）相比，行为风险因素（如身体活动和水果和蔬菜摄入量）的聚类程度更高（农村的 ICC：0.179，城市的 ICC：0.049；农村的 ICC：0.105，城市的 ICC：0.091）。
本研究提供了来自印度南部维洛尔的心血管风险因素的 ICC 估计值，因为在印度或邻国的 WHO STEPS 调查中没有报告此类数据。如果从全国范围内进行的各种 WHO STEPS 调查中报告此类 ICC 估计值，可以为更好地规划流行病学调查做出贡献。本研究中观察到的村级/区级行为风险因素聚类表明，需要进行基于社区的健康促进干预，因为空间聚类会影响行为，进而影响慢性病结果。
