63 Patient safety outcomes and their association with the level of digitization in Dutch hospitals DISCUSSION We did not find a significant relation between the EMRAM-score and the number of patients with adverse events (AE), preventable AEs, AE caused by medication, the number of readmissions (RA) and the length of stay (LOS). Those are remarkable findings, and this is not what we expected. This may be caused by the low level of digitalization of Dutch hospitals. No EMRAM stage 7 hospitals are present yet in the Netherlands. Hospitals struggle with the introduction of advanced functionalities like closed loop medication administration, evidence-based decision support and the introduction of clinical pathways in a broad sense. These characteristics are a prerequisite for stage 7. As many hospitals in the Netherlands are investing in information technology on a large scale by replacing their EMR systems it is expected this may change in the nearby future. Another factor is the quality of the collected EMRAM data. The actual use of EMR software can only be measured reliably by measuring the actual quality of stored data and that is not done. In primary care such tools are available.19 It cannot be excluded that unknown or non-measured factors affect both the dependent and the independent variables. Through stratification and the use of multivariate techniques, these effects are addressed for the known distorting factors. Further research in the future may provide clarity. Remarkable is the high median number of AE in the hospitals that did not join the EMRAM program (Table 4). Because earlier studies indicated that smaller hospitals did not join the EMRAM study 20 and at the same time these studies indicated that smaller hospitals in the Netherlands have a lower EMRAM score. This may lead to the conclusion that smaller hospitals can win by implementing more sophisticated EMR capabilities. Limitations of the study and suggestions for further research There are limitations to our study. First, although we achieved a 77% response rate, the hospitals that did not respond to the HIMSS survey were somewhat different from those that did respond. Small hospitals and hospitals located in the northern part of the Netherlands were underrepresented in the study. Given that non-responding hospitals were more likely to have characteristics associated with lower levels of adoption of electronic health records (especially size), residual bias may be present in estimating the associations. Another remark has to be made about the time frame in whichmeasurement were taken in both studies. The Nivel databasemeasurements were taken in the period 2011/2012, while the EMRAMmeasurements were done in the period 2012/2014 (with only 2 out of 17 in 2014). We assume that the effect of implementing 4
RkJQdWJsaXNoZXIy MjY0ODMw