Development and external validation of a logistic regression derived algorithm to estimate a 12-month post open defecation free slippage risk

Abstract

Appropriate open defecation free (ODF) sustainability interventions are key to mobilising communities to consume sanitation and hygiene products and services that enhance quality of life and result in embedded behavioural change. This study aims to develop a logistic regression derived risk algorithm to estimate the risk of the loss of ODF status over a 12-month period, and to externally validate the model using an independent data set. ODF status loss occurs when one or more toilet adequacy parameters is no longer present for one or more toilets in a community. Data collected in the Zambia district health information software for water sanitation and hygiene management was utilised in this study. Datasets for the Chungu and Chabula chiefdoms were selected for this study. The data was collected from the date of attainment of ODF status (October 2016) for a period of 12 months until September 2017. The Chungu chiefdom data set was utilised as the development data set whilst the Chabula chiefdom data set was utilised as the validation data set. Data was assumed to be missing at random and the complete case analysis approach was used. The events per variables were satisfactory for both the development and validation data sets. Multivariable regression with a backwards selection procedure was used to decide candidate predictor variables with p values less than 0.05 meriting inclusion. To correct for optimism, the study compared amount of heuristic shrinkage by comparing the model’s apparent C-statistic to the C-statistic computed by non-parametric bootstrap resampling. In the resulting model, an increase in the covariates ‘months after ODF attainment’, ‘village population’ and ‘latrine built after CLTS’, were all associated with a higher probability of ODF status loss. Conversely, an increase in the covariate ‘presence of a handwashing station with soap’, was associated with reduced probability of ODF status loss. The predictive performance of the model was improved by the heuristic shrinkage factor of 0.988. The external validation test confirmed good prediction performance with an area of 0.85 under the receiver operating characteristic curve and no significant lack of fit (Hosmer-Lemeshow test: p = 0.246). The results of this study must be interpreted with caution in context where ODF definitions, cultural and other factors are different from those described in the study.