Background, objectives and setting
Despite global activity over the past 15 years to improve patient safety, the measurement of adverse events (AEs) remains challenging.
We aimed to obtain definitive longitudinal data on harm across NHS Wales and to compare the performance of the Global Trigger Tool (GTT) with the two-stage retrospective review process, using our findings to consolidate an approach to the ongoing surveillance of harm in Wales.
Eleven of the 13 major Welsh NHS hospitals.
The two-stage retrospective review methodology was used to quantify harm across NHS Wales. In total, 4536 inpatient episodes were screened for AEs by research nurses. Records that were highly suggestive of AEs were further assessed by physicians. NHS-led teams undertook GTT reviews on the same case notes.
At least one AE was determined in 10.3% of episodes of care [95% confidence interval (CI) 9.4% to 11.2%] and 51.5% were preventable (95% CI 46.9% to 56.1%). The percentage of patients identified with AEs using the GTT methodology was lower, at 9.0% (95% CI 8.82% to 9.18%). Differences in AEs were evident across study sites. Methods were developed to profile the risk of AEs in individual organisations by producing signatures of harm for each NHS site. Analysis indicated that neither the GTT nor the two-stage process was a candidate tool for routine surveillance, and a hybrid one-stage tool (Harm2), based on phase 1 findings, was developed for ongoing AE monitoring. Using the Harm2 tool, AEs were identified in 371 out of 3352 randomly selected discharge reviews (11.3%, 95% CI 10.2% to 12.4%), and 59.6% (95% CI 55.3 to 63.9) of these were preventable. In a cohort of randomly selected deceased patient reviews, at least one AE was determined in 315 out of 1018 admissions (30.1%, 95% CI 28.1% to 33.8%), and 61.7% (95% CI 57.5% to 65.9%) of these were preventable. Factors associated with AEs in the randomly selected discharge reviews included having peripheral vascular disease [odds ratio (OR) 2.52], hemiplegia (OR 2.27) or dementia (OR 2.27). No association with chronic disease was identified in the deceased episodes of care.
The dependence on our health service partners in identifying notes to be reviewed, along with the small sample examined each month, limits the generalisability of these findings and rates were not standardised for hospital and size and level of services provided. We cannot rule out the possibility that the rates we report may be underestimated.
The extent of harm detected across NHS Wales using both the two-stage retrospective review process and the new Harm2 tool conforms to the findings in the literature, but this is the first longitudinal study using these methods. With training and using a structured review process, non-physician reviewers can undertake case note review efficiently and effectively, and the rates of AEs and of the preventability and the breakdown of problems in care conform to those reported in studies in which physicians undertake these classifications. Whether the patient died or was discharged alive significantly influences the rate and composition of AEs. The Harm2 tool performed with moderate reliability in the determination of AEs.
Future large-scale studies should attempt to specify types of AEs, such as hospital-acquired infections and surgical complications, to enable the surveillance of the specific types of harm as well as the overall level of AEs. In the longer term, we need to automate harm surveillance and set measures of harm alongside measures of the beneficial effects of health care.
The National Institute for Health Research Health Services and Delivery Research programme.