Model 3: Factors associated with individual HH compliance change between study periods In multivariable analysis, a higher nurse‐to‐patient ratio was positively associated, and professional category ‘medical doctor/ students’, higher activity index, and higher individual baseline HH compliance were negatively associated with individual changes in HH compliance between study periods (Table 2). The multivariable regression model explained 43% of the variance in individual changes in HH compliance between study periods (marginal R2), 22% due to the independent variables (fixed effects) and 21% due to nested clustering on the HCWs and hospital level (random effects). Model 4 Although lower baseline compliance overall was associated with a larger increase in HH compliance, it can be seen in Fig. 1 that Improving HCWs with low baseline HH compliance remained lower in intervention compared to Improving HCWs who were already high during baseline. Figure 1 shows different HCW compliance patterns and changes across the study ICUs. In some centres the variability in HH compliance among individual HCWs increased from baseline to intervention – resulting from higher variability in compliance between HCWs (Additional file 1: Fig. S1) and/or between sessions of individual HCWs (Additional file 1: Fig. S3). DISCUSSION The goal of this study was to understand the determinants and dynamics of individual HH compliance in response to an intervention. The intervention was a success, HH compliance increased significantly and importantly, overall and in each ICU. Individual compliance change was positively correlated with the overall change per ICU. However, we found large inter‐ICU and inter‐individual differences in the observed HH compliance and their dynamics. In some ICUs the overall result of the intervention was produced by almost exclusively improving HCWs, while in other ICUs the contribution of improving HCWs was to some extent offset by worsening or non‐changing HCWs. We also found that individual compliance change was positively associated with a higher nurse‐to‐ patient ratio, and negatively
RkJQdWJsaXNoZXIy MTk4NDMw