Strengthening the SQUEAC & SLEAC methods

20/01/2016 by CMN
In mid-2014, the Coverage Monitoring Network commissioned an independent, participatory review of the three (3) most commonly used coverage methodologies in the humanitarian nutrition sector: Semi-Quantitative Evaluation of Access and Coverage (SQUEAC), Simplified LQAS Evaluation of Access and Coverage (SLEAC) and Simple Spatial Survey Method (S3M). The objective of this review was to increase understanding, ownership and uptake of the different methodologies. Overall, Epicentre, the independent body who conducted the review, received a total of 73 questions, comments or queries from a variety of organisations including the Centres for Disease Control and Prevention, Concern Worldwide, the World Food Program, UNICEF, the International Rescue Committee as well as independents experts and participants from unspecified affiliations. The review highlighted that “a number of comments received […] do not raise substantial concerns but rather represent solvable misunderstandings that require clarification from the methods developers.” Since the report’s publication, the CMN has been working with the key partners to strengthen the coverage assessment methodologies. The work has focussed on the following topics:
  1. Active Adaptive Case Finding
  2. Errors in LQAS at Stage II of SQUEAC assessments
  3. The application of Bayesian Beta-Binomial conjugate analysis
  4. Single Coverage Estimate
Also, a new tool to assist survey implementers has been produced. Active Adaptive Case Finding The within-community case-finding sampling procedure most often used in SQUEAC small-area surveys, SQUEAC likelihood surveys, SLEAC surveys, and CSAS / S3M surveys is known as active and adaptive case-finding (AACF). In order to strengthen this approach the following detailed guidance has been developed:
  1. An active and adaptive case-finding procedure: Data gathering & testing.
  2. Investigating the performance of a case-finding procedure using capture-recapture studies
Read the full article here: developing an active and adaptive case finding
Furthermore, a review of historical data on the case-finding sensitivity of active and adaptive case-finding procedures for severe acute malnutrition was carried out. A large majority of the studies reviewed here show active and adaptive case-finding procedures achieving excellent levels of case-finding sensitivity for SAM cases.
Read the full article here:  historical data on active and adaptive case finding Errors in LQAS at Stage II SQUEAC Concerns regarding error when using the small sample simplified LQAS method to test SQUEAC hypotheses were raised in the review. Further methodological guidance is given to the LQAS approach in SQUEAC hypothesis setting in this document.The SQUEAC coverage assessment uses hypothesis testing by simplified LQAS to confirm (or deny) the findings made from routine program data and qualitative data. Most SQUEAC assessments will test a spatial hypotheses regarding the homogeneity (evenness) or heterogeneity (unevenness) of coverage. Although spatial hypotheses should always be tested, further guidance is provided on how hypotheses can be adapted to test other issues.
The application of Bayesian Beta-Binomial conjugate analysis This analysis focussed on survey implementers’ capacity to correctly apply Bayesian statistics. The analysis was based on data from the CMN’s global database, which sought to respond to 3 main questions;
  1. Are survey supervisors able to translate their prior mode into a properly specified probability density?
  2. Do SQUEAC stage III surveys yield estimates of coverage with an adequate precision (i.e. adequate 95% credible intervals on estimates)?
  3. Do survey supervisors properly specify priors with accurate modes and / or appropriate levels of uncertainty in order to avoid prior-likelihood conflicts?
The analyses presented here clearly demonstrates that the concerns relating to the perceived difficulties of the Bayesian Beta-Binomial conjugate analysis are unfounded. Mid-level UNO, NGO, and MoH staff are able to conduct the required analyses.
Read the full article here:  Bayesian analysis
Single Coverage Estimate In order to address the question of the preferred coverage indicator (either point or period coverage), research was carried out to create a unique estimator: the single coverage estimator. The single coverage estimator was made available in 2015, and is now the recommended coverage indicator for all survey implementers. Retrospective analysis was carried out to understand what the general effect of using the new single coverage estimate is. This analysis will be published soon. SQUEAC & SLEAC Tool In order to simplify calculations for both methods, a tool has been developed, which automatically calculated numerators and denominators, as well as classifying coverage. This toll has several new functionalities: Numbers: This tab calculates numerators and denominators for Single Coverage.. Single SLEAC: This tab automatically classifies coverage based on the findings from a SLEAC assessment. Multiple SLEACs: Here you are able to paste survey results in to the tool in order to derive coverage classifications over multiple health districts. The calculator also produces an overall coverage estimation, as well as testing for heterogeneity of results. Barriers Plot: For a standardised Barriers Plot, use this tab to produce a clear Pareto chart which can be saved in an image format, ready to be pasted in to a report. Capture-Recapture: For those who would like to test the sensitivity of case-finding methods, this tab compares case finding methods against each other.
As an integral par of the methodological update toward the use of the Single Coverage Estimator, the data plotted by the CMN on its website maps and graphs will be updated.