Modelling Needs for Mental Healthcare from Epidemiological Surveys with Validation Using Sociodemographic Census Data
Viviane Kovess-Masfety*, Anders Boyd
Identifiers and Pagination:Year: 2015
First Page: 186
Last Page: 194
Publisher Id: CPEMH-11-186
Article History:Received Date: 3/4/2015
Revision Received Date: 6/7/2015
Acceptance Date: 8/8/2015
Electronic publication date: 31/12/2015
Collection year: 2015
open-access license: This is an open access articles licensed under the terms of the Creative Commons Attribution-Non-Commercial 4.0 International Public License (CC BY-NC 4.0) (https://creativecommons.org/licenses/by-nc/4.0/legalcode), which permits unrestricted, non-commercial use, distribution and reproduction in any medium, provided that the work is properly cited.
Purpose: To develop and validate a prediction model for mental health needs (MHN) and psychiatric needs (PN) using specific social indicators, obtainable from census data, within low-density departments (LDD) and high- density departments (HDD). Methods: In a population-based study of 20,404 participants from 22 departments in France, mental health needs were defined into three categories (no needs, MHN, and PN) using the Composite International Diagnosis Interview Short-Form, Sheehan disability scale, and presence of depressive and alcohol disorders. Within HDD (n=9) and LDD (n=13) departments, two separate logistic regression models, using MHN or PN as an endpoint, were fitted using available sociodemographic data. Model validation was performed using 2007 census data. Overall accuracy was evaluated using average residuals (AR) calculated within density stratum.
Results: In LDD and HDD respectively, 26.6% and 28.7% of persons had MHN and 9.8% and 11.3% had PN. In LDD, housing type, age, employment, living alone, housing support, and household size predicted MHN and PN. In HDD, housing type, living alone, household size, living in a marriage/partnership, and duration of dwelling habitation predicted MHN and PN. Predictions were more accurate in HDD, in which the AR was 30% lower for MHN and 40% lower for PN. Predictions were less accurate when using census data, yet they were consistently better in HDD.
Conclusions: Sociodemographic indicators from either survey or census data may be useful in predicting MHN and PN in high-density settings. The ideal territorial size still needs to be evaluated when planning psychiatric and mental health resources.