•Supervision is regularly cited as a key factor to maintain community health worker (CHW) motivation and performance.
•Considered one of the most important determinants of a functioning CHW programme, supervision is also one of the weakest and most difficult components to implement consistently (Lehmann et al, 2007).
•While most CHW programmes stress the importance of consistent supervision for CHWs, previous studies suggest that regular interaction with a supervisor on its own is insufficient to positively impact CHW performance (Frimpong et al. 2011). This suggests that supportive and regular supervision are two separate components of supervisory support.
•Given that whether an external factor (i.e. supervision) diminishes or enhances motivation depends on how it is perceived by the individual, the development of a subjective measure of supervision represents an important contribution to the field of CHW research.
•The purpose of this research was to develop, validate, and make freely available a cross-culturally valid instrument to measure supervisory support, from the perspective of the community health worker.
•Phase I was an exploratory phase whereby the latent structure of a 20-item PSSS was explored with a sample of 327 Sierra Leonean CHWs.
•The initial model tested a 20-item, three-factor solution measuring: Supportive Supervision (10 items), Negative Experiences of Supervision (7 items) and Regular Supervision (3 items).
•Phase II sought to validate the results of Phase I across a sample of 741 CHWs in four different sub-Saharan African (Ethiopia, Kenya, Mozambique, Malawi) and two Asian contexts (Bangladesh, Indonesia), see map. Approximately 100 CHWs were recruited from each country.
•In both phases, confirmatory factor analysis was conducted using MPlus (Version 6.0). A multi-mode fit indices approach was employed whereby a model was considered to exhibit good fit if the chi-squared to df ratio was less than 3:1, if the RMSEA was 0.06 or lower, the SRMR was less than 0.06, and the CFI and TLI were greater than 0.90.
•Model comparison was performed to assess the factorial validity of the scale comparing: a unidimensional model, a two-factor structure, and the hypothesised three-factor structure
•The temporal stability of the factor structure was assessed across three different timepoints.
•Removing misperforming items during Phase I, we reduced the scale to a 11-item, three factor model producing good model fit results (X2=48.63, df=41, p<0.05; RMSEA=.024 (CI 90%=[.000-.047]); SRMR=.049; CFI=.98; TLI=.97). This model was subsequently tested in Phase II:
•During Phase II, the three-factor model was found to produce the best fit-statistics (X2=88.25, df=41, p<0.05; RMSEA=.044 (CI 90%=[.031-.056]); SRMR=.024; CFI=.95; TLI=.94). This finding was consistent across all three time points.
•The final three-factor PSSS includes five items measuring supportive supervision (i.e. My supervisor takes into consideration my views and ideas), three items measuring regular supervision (i.e. My supervisor meets with me regularly to discuss problems and solutions) and three items measuring negative experiences of supervision (i.e. My supervisor makes me uncomfortable or embarrassed by discussing my work in front of others).
•Simple, quick to administer and currently freely available in nine languages, the validated PSSS can contribute towards a more accurate understanding of the CHW’s perspective of supervision, as a critical determinant of successful CHW programmes across a wide range of contexts.
•The development of the PSSS scale, as described here, provides a parsimonious, valid instrument that could be used across health programmes to measure, and subsequently monitor, perceived approaches to supervision from the perspective of the CHW.