The EMPAC Tool: Development and Validation of a Context‐Specific Questionnaire for Measuring Patient‐Centered Surgical Care in Sub‐Saharan Africa

BACKGROUND: Despite the potential benefits of patient-centered care (PCC), to our knowledge, there is no research on measuring patient-centered surgical care (PCSC) particularly in sub-Saharan Africa (SSA). This study aimed to develop and test the psychometric properties of a scale for measuring PCSC in hospitals offering surgical care in SSA. METHODS: A 41-item interviewer-administered questionnaire was generated and tested by surveying 422 surgical patients in four hospitals in Malawi, chosen as a case study. After exploratory factor (EFA) and confirmatory factor analysis (CFA), standard model fit indices and qualitative judgment were used to determine the composition of the final questionnaire. Convergent and discriminant validity, composite reliability (CR) and intraclass correlation coefficient (ICC) were determined. All factor analyses were conducted using Mplus software. RESULTS: The participant average age was 46 (18-84) years, 53% (223 of 422) were male and 69% (290 of 422) were elective patients. Only 18 items (out of 41) had inter-item correlations ≥ 0.3. At CFA, 9 items (out of 18) were dropped due to low correlations and poor model fit. This resulted in a 9-item scale with 3 subscales: empowerment, patient-provider relationship, and access to surgical care (EMPAC). The instrument had good psychometric properties: RMSEA = 0.075, SRMR = 0.077, CFI = 0.958, TLI = 0.937, CR = 0.81, and ICC = 0.61. Patients with a poor satisfaction score also scored low on the PCSC scale signifying criterion validity. CONCLUSION: A novel parsimonious scale for measuring PCSC, a first of its kind for SSA, was developed using standard instrument development guidelines. The scale has shown evidence for reliability and construct validity paving the way for further validation among surgical patients in similar settings. The test-retest reliability, which was conducted on a sample of only 57 participants from four different study sites, was moderate to low (ICC: 0.61) pointing to the need for further evaluation using a larger sample size.