Patient safety without borders: measuring the global burden of adverse events.

To cite: Adhikari NKJ. BMJ Qual Saf 2013;22:798–801. In high-income countries, prominent cases and seminal epidemiologic data 3 have focused attention on the importance of safe medical care for hospitalised patients. Since this initial work, reports from high-income countries have painted a remarkably consistent picture, with around 9% of hospital admissions complicated by an adverse event (ie, an injury due to medical care), of which around 44% may be preventable. In this issue of BMJ Quality & Safety, Jha et al make a substantial contribution to the patient safety literature by estimating the burden of adverse events among hospitalised patients worldwide. They incorporate a systematic search of published data and recent multicountry observational studies commissioned by WHO, construct statistical models to pool these data, and meticulously report the underlying assumptions, methods and results. The main findings are that approximately 43 million adverse events occur each year around the globe and cause a staggering 23 million associated disability-adjusted life years (DALYs, the sum of years of life lost and years lost to disability). Importantly, two-thirds of these occur in low-income and middle-income countries (LMICs). This level of DALYs places adverse events ahead of maternal disorders (16 million) and behind cirrhosis (31 million), although as the authors point out, DALYs due to adverse events are not new but comprise some of the total global burden of 2.5 billion DALYs. The widely cited estimate from the US Institute of Medicine of 44 000– 98 000 preventable deaths annually due to medical care made medical error the eighth leading cause of death in the USA. Similarly, the analysis by Jha et al suggests that preventable adverse events represent a leading cause of morbidity and mortality worldwide. WHERE DO THE NUMBERS COME FROM? Jha et al considered DALYs from seven in-hospital adverse events: adverse drug events, nosocomial pneumonia, catheterrelated blood stream infections, catheterrelated urinary tract infections, venous thromboembolism, falls, and decubitus ulcers. Although their estimates of burden are subject to biases, several would lead to underestimation of the number of adverse events. The most notable such bias is the exclusion (due to limitations of available data) of clinically important and common adverse events in LMICs related to peripartum care, counterfeit drugs, unsafe injection practices, blood use, surgery and outpatient care. Inclusion of these events would only increase the estimated global toll of adverse events. In contrast, one bias that would overestimate the number of DALYs in LMICs is the assumption of equal life expectancy as in high-income countries, but this methodology is consistent across the Global Burden of Disease research programme. Accepting the findings at face value, one limitation is that the highlighted deficiencies in inpatient medical care may be refractory to intervention. For example, preventability has not been considered. Since not all adverse events are preventable, and since case reviewers generally only agree modestly on the extent of preventability, the opportunity to eliminate adverse events and subsequent DALYs is less than implied in the summary estimates. A more fundamental issue is that of causality. For example, Jha et al use disability weights for endocarditis to calculate DALYs for catheter-related blood stream infection. The authors turned to analogous conditions because data on disabilities caused by in-hospital adverse events are lacking. However, disability in a patient admitted with endocarditis is much easier EDITORIAL