A Simple Algorithm to Predict Post Traumatic Stress (PTS) Symptom Prevalence and Local Distribution
The Post-Terremoto survey (EPT for its Spanish acronym) contains unusual longitudinal data about the same persons before and after a major disaster. It comprehends nationally representative data of a household survey that was gathered a few months before the 2010 Chilean earthquake and tsunami streaked. This was complemented by post-disaster follow up information, from a representative subsample of the original households. In the follow up, persons were requested to respond the Davidson’s trauma battery to evaluate PTS symptoms, leading to 25600 valid PTS scores available for the analyses. We complemented EPT data with information about the strength of the earthquake and the tsunami, the history of replicas, and the death rate.
The objective of the project is to derive an algorithm to predict PTS symptom prevalence and its local distribution, using the database described above. Several model specifications for the mean and centiles of the distribution of PTS symptoms (at the municipal level), together with PTS disorder prevalence are estimated via linear and quantile regressions. Models vary in the set of covariates included.
Preliminary results show that it is possible to devise simple algorithms to predict PTS prevalence and distribution, even in a setting where data is scarce. Such settings are highly probable at the immediate aftermath of a large-scale disaster. This rough but quick estimate could be of use for emergency managers that must decide where to assign the scarce mental health personnel that is available at the aftermath of a major natural disaster.