Data from: Mental health ecosystem of Gipuzkoa (2015) for Bayesian network modelling
Data files
Mar 29, 2022 version files 247.86 KB
-
Data_Gipuzkoa_2015_model.xlsx
16.49 KB
-
README_Gipuzkoa_2015_model.pdf
231.37 KB
Abstract
This dataset include data from Mental Health network of Gipuzkoa (Spain). It is included information on resources (inputs) and outcomes (outputs) of care, which are described in the manuscript: “Almeda, N., Garcia-Alonso, C. R., Gutierrez-Colosia, M. R., Salinas-Perez, J. A., Iruin-Sanz, A., & Salvador-Carulla, L. (2022). Modelling the balance of care: Impact of an evidence-informed policy on a mental health ecosystem. PLoS ONE, 17(1 January), 1–16. https://doi.org/10.1371/journal.pone.0261621”. This manuscript has been published in Plos One journal.
This research focused on developing a formal causal model based on Bayesian network prototypes which were designed by formalizing expert knowledge (by using Expertbased Cooperative Analysis) and resulting in Direct Acyclic Graphs. The best Bayesian networks and their corresponding regression models were used to estimate the statistical ranges or confidence intervals for the dependent variable (potential effect, consequence, or output) given the independent variable values. These ranges, adjusted to delimited statistical distributions (triangular, trapezoidal and gamma), were managed by a Monte Carlo simulation engine for intervention assessment. A computer-based Decision Support System (DSS) was used to assess the status of ecosystem performance: RTE, statistical stability and entropy.
Main results of the analyses pointed out that by combining causal reasoning and statistical methods, decision makers can obtain a deep view of both pre-implementing and post-implementing situations. Knowing the causal levers, it is possible to act directly to the causes in order to potentially produce de appropriate results considering the uncertainty: to provide a more balanced and integrated MH care provision in the community. In this particular case, an improvement in the outpatient workforce increases both ecosystem performance (RTE) and stability and slightly decreases entropy.
The Mental Health Network of Gipuzkoa manages its community and rehabilitation MH services (774,700 residents). These services include outpatient and core health day care facilities closely related to both hospital services (inpatient) and the social care network for severe cases. The Mental Health Network of Gipuzkoa is divided into 13 catchment areas, all of which have a mental health centre, considered as a reference, with a distinct orientation to outpatient care (some of them also include health day care facilities).
Gipuzkoa has a population of 640,635 adults older than 17 years of age in 2015. It is one of the three historic territories of Basque Country autonomous community in Spain. The Department of Health in each historic territory has total governance capacity and centralizes healthcare management and provision. The MH ecosystem of Gipuzkoa is structured in 13 catchment areas, which are considered the decision-making units (DMUs) for policy assessment. Each catchment area of Gipuzkoa corresponds to a community MH centre. A single acute MH hospital unit provides care to all the DMUs. The 13 catchment areas are: Alto Deba-Arrasate, Amara, Andoain, Azpeitia, Beasain, Eguia, Eibar, Irun, Ondarreta, Renteria, Tolosa, Zarautz and Zumarraga.
This dataset content 49 variables, where combined generated 85 variables which were used to describe MH care provision and to assess ecosystem performance by using RTE scores. Seventy-five variables were identified as resources used by the DMUs to provide MH care, which were classified as inputs in Group C: quality of care. Ten variables were classified as outcomes (outputs) obtained by using the inputs, classified in Group B: service utilization. However, only a few of these variables are considered by experts for designing potential causal relationships between outpatient and residential MH care.
The readme file contains an explanation of each variable in the dataset, including its variable name, units and explanation. This dataset has been used for assessing relative technical efficiency, stability, entropy indicators and modelling.