In the news, (BMJ, May 3) . The article explains that the 1999 Institute of Medicine (IOM report is limited and outdated. The results were contested by Leape in 1993, a chief investigator of the 1984 Harvard study which published an article arguing that the study’s estimate was too low, contending that 78% rather than 51% of the 180,000 iatrogenic deaths were preventable. Similarly, the Inspector General of the USDHHS Office examining the health records of hospital inpatients in 2008, reported that 180,000 deaths due to medical error a year among Medicare beneficiaries alone, a rate of 1.13%. If this rate is applied to all registered US hospital admissions in 2013, it translates to over 400,000 deaths a year, more than four times the IOM estimate. The ICD-10 coding system has limited ability to capture most types of medical error. At best, there are only a few codes where the role of error can be inferred, such as the code for anticoagulation causing adverse effects and the code for overdose. When a medical error results in death, both the physiologic cause of the death and the related problem with deliver of care should be captured. Data should be shared nationally and internationally to improve safety, much in the way scientists share data about disease. Click here to read full-text.