Erratum: Mediation and moderation effects of health system structure and process on the quality of mental health services in Ghana- structural equation modelling (PLoS ONE (2020) 15: 5 (e0233351) DOI: 10.1371/journal.pone.0233351)

There are errors in Fig 1 and Fig 2 regarding the covary between observed variable "G"(under latent variable Structure) and observed variable "M"(under latent variable Process). In Fig 2, the covary for the modification of the model is performed between observed variables under the same latent variable. Goodness of Fit: RMSEA = 0.05; 90% CI, lower bound = 0.050; upper bound = 0.063; pclose = 0.062 Probability RMSEA < = 0.05; Comparative fit index (CFI) = 0.940; Tucker-Lewis index (TLI) = 0.917; Coefficient of determination = 0.88. Please see the correct Fig 1 and Fig 2 here. These errors also occur in the Confirmatory factor analysis results subsection of the Results and in Table 4. The correct paragraph is: As shown in Table 3, the correlation matrix demonstrates sufficient convergent and discriminant correlation between the VSS-54 construct and WHODAS construct. Fig 1, graphically describes the hypothesized SEM. Results from the CFA showed that the hypothesized model had a good fit with the Residual Mean Square of Approximation (RMSEA) = 0.05; 90% CI, lower bound = 0.050; upper bound = 0.063; pclose = 0.062; Comparative fit index (CFI) = 0.940; Tucker-Lewis index (TLI) = 0.917; Coefficient of determination = 0.88. Details of the standardized coefficient estimates for CFA is represented in Fig 2 and Table 4. The path model result shows that the health system structure was significantly related to the process and outcome (β = 0.47; p<0.001). Similarly, the health systems structure construct was mediated by the process construct in its relationship with the outcome (β = 0.346; p<0.001). Please see the correct Table 4 here. (Figure Presented) (Table Presented).