Improving Health Systems Responsiveness among Chronic Care Centers in Selected Tier Three Hospitals in Kenya: Predictors and Effect of Health Workers Training

Abstract

This study assessed health system responsiveness, predictors (accountability mechanisms, access, structural, valuations, organizational culture, justice perceptions) and role of training among patients with diabetes mellitus and hypertension at three primary hospitals in Kenya: Kimilili (Bungoma county), Uasin Gishu (Uasin Gishu county) and Gatundu (Kiambu county). It used a quasi-experimental design with baseline survey, a training intervention on responsiveness and client interaction skills for health workers, and an end-line survey. A sample of 323 respondents was selected from a sampling frame of 853 using Fisher's formula. The end-line survey included 258 participants. Data were collected using an interviewer-assisted structured questionnaire, with responses rated on a five-point Likert scale and then dichotomized into favorable and unfavorable categories using a demarcation threshold formula. Responsiveness levels increased from 63.7% to 67.4%, while proportion of favorable responsiveness rose from 38.3% to 52.7%. The odds of favorable responsiveness nearly doubled from 0.620 to 1.114 between baseline and end line. Comparatively, Kimilili hospital exhibited higher levels and distribution of responsiveness. Communication scored highest with 72.7% at the end line, while Choice performed lowest with 53.3% at baseline. Dignity was most valued domain at 23.1%, while social support was least at 3.9%. A paired samples t-test indicated significant (P<0.05) positive deviations across all responsiveness domains and four predictors: valuations, accountability, structural factors, and organizational culture. Chi-square test indicated the socio -demographic factors: facility (p=0.001), medical condition (p=0.001), religion (p=0.033), marital status (p=0.001), and occupation (p=0.001) significantly impacted responsiveness at baseline, while marital status (p=0.012) and occupation (p=0.039) remained significant at end line. Following an iterative binary logistic regression, the final predictive model for responsiveness was based on end line survey findings and indicates structural (p=0.010, OR=2.171), accountability (p=0.001, OR=2.730), organizational culture (p=0.009, OR=2.267), and justice perceptions (p=0.001, OR=2.909) were significant predictors. After intervention, the model improved significantly; explained variation increased from 15.7% to 32.8%, the logit improved from 68.5% to 85.1%, and the correct classification of responsiveness categories rose from 66.9% to 70.5%. Including two significant sociodemographic factors: marital status and occupation; into the predictive model raised the explained variation to 40.4%, reflecting the context. The -2log likelihood ratio indicated justice perceptions significantly moderated the association between responsiveness and all other predictors. Qualitative analysis underscored the importance of respect for persons domains, with varied perceptions across contexts. Challenges included low client voice, insufficient managerial support, inadequate accountability mechanisms, corruption, and limited supplies. In conclusion, responsiveness, although generally low, improved with training. Lower socioeconomic groups received more responsive care, suggesting potential for lowered expectations. The study suggests hospital managers prioritize an integrated, patient-centered approach, conduct regular client-provider feedback sessions, and implement audits for continuous improvement. Training institutions and hospital management should integrate attitude and cultural competence training across all levels to enhance sensitivity to cultural differences in healthcare. Furthermore, the Ministry of Health and county health departments should enhance accountability, improve structural factors, foster positive organizational culture, and ensure fairness in justice perceptions to enhance patient experiences and health outcomes.