Observational Bias during Nutrition Surveillance: Results of a Mixed Longitudinal and Cross-Sectional Data Collection System in Northern Nigeria

Abstract

BackgroundThe Sahel is subject to seasonal hungry periods with increasing rates of malnutrition. In Northern Nigeria, there is no surveillance system and surveys are rare. The objectives were to analyse possible observational bias in a sentinel surveillance system using repeated mixed longitudinal/cross-sectional data and estimate the extent of seasonal variation.MethodsThirty clusters were randomly selected using probability proportional to size (PPS) sampling from Kazaure Local Government Area, Jigawa State. In each cluster, all the children aged 6–59 months within 20 randomly selected households had their mid-upper arm circumference measured and were tested for oedema. The surveys were repeated every 2 or 4 weeks. At each survey round, three of the clusters were randomly selected to be replaced by three new clusters chosen at random by PPS. The seasonal variation of acute malnutrition was assessed using cyclical regression. The effect of repeated visits to the same cluster was examined using general linear mixed effects models adjusted for the seasonal change.ResultsThere was a significant seasonal fluctuation of Global Acute Malnutrition (GAM) with a peak in October. With each repeat survey of a cluster, the prevalence of GAM decreased by 1.6% (95% CI: 0.4 to 2.7; p = 0.012) relative to the prevalence observed during the previous visit after adjusting for seasonal change.ConclusionsNorthern Nigeria has a seasonal variation in the prevalence of acute malnutrition. Repeated surveys in the same cluster-village, even if different children are selected, lead to a progressive improvement of the nutritional status of that village. Sentinel site surveillance of nutritional status is prone to observational bias, with the sentinel site progressively deviating from that of the community it is presumed to represent.

Are free anti-tuberculosis drugs enough? An empirical study from three cities in China

“…Prefectural level’s city, for instance Zhengjiang City, can also be divided into district, and further into street. To sample adapt to China’s administration unit, Probability-Proportional-to-Size Sampling (PPS) was adopted [ 39 , 40 ]. Using selected key indicators, the required sample size per city was calculated to be 264 cases.…”

Methods of nutrition surveillance in low-income countries

“…The main disadvantage is the unknown level of bias which is likely to vary on a case-by-case basis depending on the environmental context and exact methods applied. With respect to bias, further evidence has since been published of how, during the process of surveillance, the nutritional situation in the selected sites may become progressively different from the rest of community that they were chosen to represent [ 60 , 61 ]. This is due to the inputs of the survey teams who may provide education, advice and counselling; treating illness or referring malnourished children to a treatment programme; and providing employment and spending money within the community [ 62 ].…”

“…The disadvantage of such longitudinal data collection is that it is labour intensive and expensive to repeatedly find and measure the same individuals, something that is usually done only in cohort studies in wealthy countries. Children can be lost to follow-up, as in the system using data from sentinel clinics in Malawi [ 97 , 98 ], and there may be bias introduced by the Hawthorne effect of repeatedly studying the same children, as noted in Northern Nigeria [ 60 ]. Furthermore, if children are lost to follow-up and are replaced, the sample becomes a mixture of longitudinal and cross-sectional samples, which makes the analysis more statistically complex and there may not be enough statistical power to detect changes.…”

Nutrition surveillance using a small open cohort: experience from Burkina Faso

“…It is probable that in the field, information provided to families by our staff and referral of malnourished children have improved the nutritional status of the children within these sentinel villages. Likewise, the observational bias reported in a recent study piloted in Northern Nigeria by Médecins sans Frontières (MSF) was a progressive deviation of the nutritional status of the sentinel site from that of the wider community it is presumed to represent [10]. In the MSF study, the prevalence of GAM decreased by 1.6 % (95 % CI 0.4–2.7 %; p = 0.012) relative to the prevalence observed during the previous visit.…”

“…Second, an observational effect that acts to reduce prevalence over time as the selected sites tend to be progressively positively affected by the inputs of the survey teams (e.g. giving education, advice and counselling, referral of cases for treatment, and treating illness) [10]. It is not clear, however, that a statistically representative sample, as might be used in a population survey, is an essential attribute of a surveillance system.…”