Old and new cluster designs in emergency field surveys: in search of a one-fits-all solution

Methodological issues of retrospective surveys for measuring mortality of highly clustered diseases: case study of the 2014-16 Ebola outbreak in Bo District, Sierra Leone

BACKGROUND

There is a lack of empirical data on design effects (DEFF) for mortality rate for highly clustered data such as with Ebola virus disease (EVD), along with a lack of documentation of methodological limitations and operational utility of mortality estimated from cluster-sampled studies when the DEFF is high.

OBJECTIVES

The objectives of this paper are to report EVD mortality rate and DEFF estimates, and discuss the methodological limitations of cluster surveys when data are highly clustered such as during an EVD outbreak.

METHODS

We analysed the outputs of two independent population-based surveys conducted at the end of the 2014-2016 EVD outbreak in Bo District, Sierra Leone, in urban and rural areas. In each area, 35 clusters of 14 households were selected with probability proportional to population size. We collected information on morbidity, mortality and changes in household composition during the recall period (May 2014 to April 2015). Rates were calculated for all-cause, all-age, under-5 and EVD-specific mortality, respectively, by areas and overall. Crude and adjusted mortality rates were estimated using Poisson regression, accounting for the surveys sample weights and the clustered design.

RESULTS

Overall 980 households and 6,522 individuals participated in both surveys. A total of 64 deaths were reported, of which 20 were attributed to EVD. The crude and EVD-specific mortality rates were 0.35/10,000 person-days (95%CI: 0.23-0.52) and 0.12/10,000 person-days (95%CI: 0.05-0.32), respectively. The DEFF for EVD mortality was 5.53, and for non-EVD mortality, it was 1.53. DEFF for EVD-specific mortality was 6.18 in the rural area and 0.58 in the urban area. DEFF for non-EVD-specific mortality was 1.87 in the rural area and 0.44 in the urban area.

CONCLUSION

Our findings demonstrate a high degree of clustering; this contributed to imprecise mortality estimates, which have limited utility when assessing the impact of disease. We provide DEFF estimates that can inform future cluster surveys and discuss design improvements to mitigate the limitations of surveys for highly clustered data.

WASH and Health in Sindhupalchowk District of Nepal after the Gorkha Earthquake

An earthquake of magnitude 7.8 MW and 6.8 MW struck Nepal on 25 April and 12 May, 2015, respectively, which caused massive damage. In such crises, understanding the water, sanitation, and hygiene (WASH) situation is of paramount importance. Therefore, we aimed to assess the WASH situation and its impact on health, particularly in the Sindhupalchowk district. A questionnaire survey and microbial analysis of water samples were conducted. Descriptive statistics and parametric and non-parametric statistical tests were employed. The results revealed that 97.1% of water samples from the source during the pre-monsoon season and 98.5% during the monsoon season had fecal contamination. Similarly, 92.8% of water samples during the pre-monsoon season and 96.7% during the monsoon season at point of use (PoU) had fecal contamination. Furthermore, water consumption was comparatively less during the pre-monsoon season. The increase in water consumption improved hygiene behavior and lowered the prevalence of waterborne diseases. Similarly, less water consumption affected water handling behavior; for example, the cleaning interval of storage vessels was less frequent. An increase in cleaning interval resulted in fecal contamination of water at PoU. The findings of this study can be useful in the review of existing WASH policy and plans and integration with the disaster management plan for disaster risk reduction.

Risk Factors for Obesity and Overfat among Primary School Children in Mashonaland West Province, Zimbabwe

Associated childhood obesity risk factors are not well established in developing countries such as Zimbabwe and this information is essential for tailored intervention development. This study aimed to identify prominent risk factors for overweight/obese and overfat/obese among primary school children of Mashonaland West Province in Zimbabwe. A school-based cross-sectional study was conducted using multi-stage random cluster sampling approach (30 × 30). Bivariate and multivariable logistic regression was employed and identified the risk factors for overweight/obese and overfat/obese. A total of 974 participants were enrolled in the study. Prominent significant risk factors of overweight/obese after multivariable adjustment were higher socio-economic households; parental diabetes status; and living in Makonde, Zvimba, Sanyati or Mhondoro-Ngezi district as opposed to Hurungwe district. Risk factors for overfat/obese that remained statically significant were children in urban areas (aOR = 3.19, 95% CI: 2.18-4.66, p = 0.000), being one child in a household, and parents who have diabetes mellitus. Living in Makonde, Sanyati, and Zvimba district remained associated with overfat/obese compared to Hurungwe district. This study has identified prominent proximal determinants of overweight/obese and overfat/obese among primary school children in Zimbabwe, to better assist policy guidance. Aggressive education on good nutrition activities should be tailored and targeted to most affected urban areas within high-risk districts.

Parameters associated with design effect of child anthropometry indicators in small-scale field surveys

Background

Cluster surveys provide rapid but representative estimates of key nutrition indicators in humanitarian crises. For these surveys, an accurate estimate of the design effect is critical to calculate a sample size that achieves adequate precision with the minimum number of sampling units. This paper describes the variability in design effect for three key nutrition indicators measured in small-scale surveys and models the association of design effect with parameters hypothesized to explain this variability.

Methods

380 small-scale surveys from 28 countries conducted between 2006 and 2013 were analyzed. We calculated prevalence and design effect of wasting, underweight, and stunting for each survey as well as standard deviations of the underlying continuous Z-score distribution. Mean cluster size, survey location and year were recorded. To describe design effects, median and interquartile ranges were examined. Generalized linear regression models were run to identify potential predictors of design effect.

Results

Median design effect was under 2.00 for all three indicators; for wasting, the median was 1.35, the lowest among the indicators. Multivariable linear regression models suggest significant, positive associations of design effect and mean cluster size for all three indicators, and with prevalence of wasting and underweight, but not stunting. Standard deviation was positively associated with design effect for wasting but negatively associated for stunting. Survey region was significant in all three models.

Conclusions

This study supports the current field survey guidance recommending the use of 1.5 as a benchmark for design effect of wasting, but suggests this value may not be large enough for surveys with a primary objective of measuring stunting or underweight. The strong relationship between design effect and region in the models underscores the continued need to consider country- and locality-specific estimates when designing surveys. These models also provide empirical evidence of a positive relationship between design effect and both mean cluster size and prevalence, and introduces standard deviation of the underlying continuous variable (Z-scores) as a previously unexplored factor significantly associated with design effect. The magnitude and directionality of this association differed by indicator, underscoring the need for further investigation into the relationship between standard deviation and design effect.

Methods of nutrition surveillance in low-income countries

BACKGROUND

In 1974 a joint FAO/UNICEF/WHO Expert Committee met to develop methods for nutrition surveillance. There has been much interest and activity in this topic since then, however there is a lack of guidance for practitioners and confusion exists around the terminology of nutrition surveillance. In this paper we propose a classification of data collection activities, consider the technical issues for each category, and examine the potential applications and challenges related to information and communication technology.

ANALYSIS

There are three major approaches used to collect primary data for nutrition surveillance: repeated cross-sectional surveys; community-based sentinel monitoring; and the collection of data in schools. There are three major sources of secondary data for surveillance: from feeding centres, health facilities, and community-based data collection, including mass screening for malnutrition in children. Surveillance systems involving repeated surveys are suitable for monitoring and comparing national trends and for planning and policy development. To plan at a local level, surveys at district level or in programme implementation areas are ideal, but given the usually high cost of primary data collection, data obtained from health systems are more appropriate provided they are interpreted with caution and with contextual information. For early warning, data from health systems and sentinel site assessments may be valuable, if consistent in their methods of collection and any systematic bias is deemed to be steady. For evaluation purposes, surveillance systems can only give plausible evidence of whether a programme is effective. However the implementation of programmes can be monitored as long as data are collected on process indicators such as access to, and use of, services. Surveillance systems also have an important role to provide information that can be used for advocacy and for promoting accountability for actions or lack of actions, including service delivery.

CONCLUSION

This paper identifies issues that affect the collection of nutrition surveillance data, and proposes definitions of terms to differentiate between diverse sources of data of variable accuracy and validity. Increased interest in nutrition globally has resulted in high level commitments to reduce and prevent undernutrition. This review helps to address the need for accurate and regular data to convert these commitments into practice.

Extending cluster lot quality assurance sampling designs for surveillance programs

Lot quality assurance sampling (LQAS) has a long history of applications in industrial quality control. LQAS is frequently used for rapid surveillance in global health settings, with areas classified as poor or acceptable performance on the basis of the binary classification of an indicator. Historically, LQAS surveys have relied on simple random samples from the population; however, implementing two-stage cluster designs for surveillance sampling is often more cost-effective than simple random sampling. By applying survey sampling results to the binary classification procedure, we develop a simple and flexible nonparametric procedure to incorporate clustering effects into the LQAS sample design to appropriately inflate the sample size, accommodating finite numbers of clusters in the population when relevant. We use this framework to then discuss principled selection of survey design parameters in longitudinal surveillance programs. We apply this framework to design surveys to detect rises in malnutrition prevalence in nutrition surveillance programs in Kenya and South Sudan, accounting for clustering within villages. By combining historical information with data from previous surveys, we design surveys to detect spikes in the childhood malnutrition rate.

Need for a gender-sensitive human security framework: results of a quantitative study of human security and sexual violence in Djohong District, Cameroon

Background

Human security shifts traditional concepts of security from interstate conflict and the absence of war to the security of the individual. Broad definitions of human security include livelihoods and food security, health, psychosocial well-being, enjoyment of civil and political rights and freedom from oppression, and personal safety, in addition to absence of conflict.

Methods

In March 2010, we undertook a population-based health and livelihood study of female refugees from conflict-affected Central African Republic living in Djohong District, Cameroon and their female counterparts within the Cameroonian host community. Embedded within the survey instrument were indicators of human security derived from the Leaning-Arie model that defined three domains of psychosocial stability suggesting individuals and communities are most stable when their core attachments to home, community and the future are intact.

Results

While the female refugee human security outcomes describe a population successfully assimilated and thriving in their new environments based on these three domains, the ability of human security indicators to predict the presence or absence of lifetime and six-month sexual violence was inadequate. Using receiver operating characteristic (ROC) analysis, the study demonstrates that common human security indicators do not uncover either lifetime or recent prevalence of sexual violence.

Conclusions

These data suggest that current gender-blind approaches of describing human security are missing serious threats to the safety of one half of the population and that efforts to develop robust human security indicators should include those that specifically measure violence against women.

New tools for evaluating LQAS survey designs

Lot Quality Assurance Sampling (LQAS) surveys have become increasingly popular in global health care applications. Incorporating Bayesian ideas into LQAS survey design, such as using reasonable prior beliefs about the distribution of an indicator, can improve the selection of design parameters and decision rules. In this paper, a joint frequentist and Bayesian framework is proposed for evaluating LQAS classification accuracy and informing survey design parameters. Simple software tools are provided for calculating the positive and negative predictive value of a design with respect to an underlying coverage distribution and the selected design parameters. These tools are illustrated using a data example from two consecutive LQAS surveys measuring Oral Rehydration Solution (ORS) preparation. Using the survey tools, the dependence of classification accuracy on benchmark selection and the width of the ‘grey region’ are clarified in the context of ORS preparation across seven supervision areas. Following the completion of an LQAS survey, estimation of the distribution of coverage across areas facilitates quantifying classification accuracy and can help guide intervention decisions.

Sexual violence among host and refugee population in Djohong District, Eastern Cameroon

The following is a population-based survey of the Central African Republic (CAR) female refugee population displaced to rural Djohong District of Eastern Cameroon and associated female Cameroonian host population to characterise the prevalence and circumstances of sexual violence. A population-based, multistage, random cluster survey of 600 female heads of household was conducted during March 2010. Women heads of household were asked about demographics, household economy and assets, level of education and sexual violence experienced by the respondent only. The respondents were asked to describe the circumstances of their recent assault. The lifetime prevalence of sexual violence among Djohong district female heads of household is 35.2% (95% CI 28.7-42.2). Among heads of household who reported a lifetime incident of sexual violence, 64.0% (95% CI 54.3-72.5) suffered sexual violence perpetrated by their husband or partner. Among the host population, 3.9% (95% CI 1.4-10.5) reported sexual violence by armed groups compared to 39.0% (95% CI 25.6-54.2) of female refugee heads of household. Women who knew how to add and subtract were less likely to report sexual violence during their lifetime (OR 0.16, 95% CI 0.08-0.34). Sexual violence is common among refugees and host population in Eastern Cameroon. Most often, perpetrators are partners/husbands or armed groups.

Bayes-LQAS: classifying the prevalence of global acute malnutrition

Lot Quality Assurance Sampling (LQAS) applications in health have generally relied on frequentist interpretations for statistical validity. Yet health professionals often seek statements about the probability distribution of unknown parameters to answer questions of interest. The frequentist paradigm does not pretend to yield such information, although a Bayesian formulation might. This is the source of an error made in a recent paper published in this journal. Many applications lend themselves to a Bayesian treatment, and would benefit from such considerations in their design. We discuss Bayes-LQAS (B-LQAS), which allows for incorporation of prior information into the LQAS classification procedure, and thus shows how to correct the aforementioned error. Further, we pay special attention to the formulation of Bayes Operating Characteristic Curves and the use of prior information to improve survey designs. As a motivating example, we discuss the classification of Global Acute Malnutrition prevalence and draw parallels between the Bayes and classical classifications schemes. We also illustrate the impact of informative and non-informative priors on the survey design. Results indicate that using a Bayesian approach allows the incorporation of expert information and/or historical data and is thus potentially a valuable tool for making accurate and precise classifications.

Association between Mycobacterium Avium Subsp. Paratuberculosis Infection among Offspring and their Dams in Nondomestic Ruminant Species Housed in a Zoo

The objective of the present study was to determine whether offspring of dams infected with Mycobacterium avium subsp. paratuberculosis (Map) have an increased risk for Map infection. Antemortem and postmortem disease surveillance data were used to identify positive and test-negative ruminants born at the Zoological Society of San Diego (ZSSD) between 1991 and 2007 and to estimate cumulative lifetime incidence. A matched case-control study, nested within the population, was conducted and conditional logistic regression analyses were used to quantify the association between infection status of offspring and their dams. Cases (infected ruminants, n = 47) were matched to controls (test-negative ruminants, n = 152) by species, birth date, birth enclosure, and follow-up time to control for confounding factors. The overall cumulative lifetime incidence of infection was estimated at 2.2%, but it decreased over time and varied by species. There was a significant association between infection status of offspring and their dams (odds ratio [OR] = 6.8, P < 0.01), which is consistent with studies in domestic livestock species. The association was stronger for animals whose dam was diagnosed within 2 years of their birth (OR = 9.0, P < 0.01) than for animals whose dam was diagnosed more than 2 years after their birth (OR = 6.0, P < 0.01) compared to animals with test-negative dams. For positive animals born to a positive dam, 85.3% of the Map infections were attributable to having a positive dam. For the entire population of ZSSD ruminants, 36.8% of the cases were attributable to having a positive dam. Findings will help guide future management of Map infection in zoo ruminant populations.

Interpreting results of cluster surveys in emergency settings: is the LQAS test the best option?

Cluster surveys are commonly used in humanitarian emergencies to measure health and nutrition indicators. Deitchler et al. have proposed to use Lot Quality Assurance Sampling (LQAS) hypothesis testing in cluster surveys to classify the prevalence of global acute malnutrition as exceeding or not exceeding the pre-established thresholds. Field practitioners and decision-makers must clearly understand the meaning and implications of using this test in interpreting survey results to make programmatic decisions. We demonstrate that the LQAS test–as proposed by Deitchler et al. – is prone to producing false-positive results and thus is likely to suggest interventions in situations where interventions may not be needed. As an alternative, to provide more useful information for decision-making, we suggest reporting the probability of an indicator's exceeding the threshold as a direct measure of "risk". Such probability can be easily determined in field settings by using a simple spreadsheet calculator. The "risk" of exceeding the threshold can then be considered in the context of other aggravating and protective factors to make informed programmatic decisions.