Can Patient Safety Incident Reports Be Used to Compare Hospital Safety? Results from a Quantitative Analysis of the English National Reporting and Learning System Data

BACKGROUND: The National Reporting and Learning System (NRLS) collects reports about patient safety incidents in England. Government regulators use NRLS data to assess the safety of hospitals. This study aims to examine whether annual hospital incident reporting rates can be used as a surrogate indicator of individual hospital safety. Secondly assesses which hospital characteristics are correlated with high incident reporting rates and whether a high reporting hospital is safer than those lower reporting hospitals. Finally, it assesses which health-care professionals report more incidents of patient harm, which report more near miss incidents and what hospital factors encourage reporting. These findings may suggest methods for increasing the utility of reporting systems. METHODS: This study used a mix methods approach for assessing NRLS data. The data were investigated using Pareto analysis and regression models to establish which patients are most vulnerable to reported harm. Hospital factors were correlated with institutional reporting rates over one year to examine what factors influenced reporting. Staff survey findings regarding hospital safety culture were correlated with reported rates of incidents causing harm; no harm and death to understand what barriers influence error disclosure. FINDINGS: 5,879,954 incident reports were collected from acute hospitals over the decade. 70.3% of incidents produced no harm to the patient and 0.9% were judged by the reporter to have caused severe harm or death. Obstetrics and Gynaecology reported the most no harm events [OR 1.61(95%CI: 1.12 to 2.27), p<0.01] and pharmacy was the hospital location where most near-misses were captured [OR 3.03(95%CI: 2.04 to 4.55), p<0.01]. Clinicians were significantly more likely to report death than other staff [OR 3.04(95%CI: 2.43 to 3.80) p<0.01]. A higher ratio of clinicians to beds correlated with reduced rate of harm reported [RR = -1.78(95%Cl: -3.33 to -0.23), p = 0.03]. Litigation claims per bed were significantly negatively associated with incident reports. Patient satisfaction and mortality outcomes were not significantly associated with reporting rates. Staff survey responses revealed that keeping reports confidential, keeping staff informed about incidents and giving feedback on safety initiatives increased reporting rates [r = 0.26 (p<0.01), r = 0.17 (p = 0.04), r = 0.23 (p = 0.01), r = 0.20 (p = 0.02)]. CONCLUSION: The NRLS is the largest patient safety reporting system in the world. This study did not demonstrate many hospital characteristics to significantly influence overall reporting rate. There were no association between size of hospital, number of staff, mortality outcomes or patient satisfaction outcomes and incident reporting rate. The study did show that hospitals where staff reported more incidents had reduced litigation claims and when clinician staffing is increased fewer incidents reporting patient harm are reported, whilst near misses remain the same. Certain specialties report more near misses than others, and doctors report more harm incidents than near misses. Staff survey results showed that open environments and reduced fear of punitive response increases incident reporting. We suggest that reporting rates should not be used to assess hospital safety. Different healthcare professionals focus on different types of safety incidents and focusing on these areas whilst creating a responsive, confidential learning environment will increase staff engagement with error disclosure.