Studies generating estimates of deaths due to medical error*
| Study | Study features | Deaths contributed or caused by medical error | Limitations relevant to estimating deaths due to medical error Inclusion in estimate of deaths due to medical error |
|---|---|---|---|
| Office of the Inspector General6 (used by Makary and Daniel4 and by James9) | 838 hospitalisations involving 780 Medicare beneficiaries randomly selected from the total of 999 645 Medicare beneficiaries discharged from acute care hospitals during October 2008 | 12 | No stated method for judging the causal relationship between preventable adverse event and eventual death. |
| Office of the Inspector General11 (used by James9) | 278 Medicare hospitalisations of at least 24 hours randomly sampled from the 2549 total discharges from 24 hospitals in two counties during 1 week in August 2008. | 3 | Very small number of deaths. Study population does not include some patient populations at lower risk of death (eg, younger, non-Medicare patients), but extrapolation includes these patients as if they had same risk of death. No assessment of preventability for adverse events. Unclear method for judging the causal relationship between the adverse event and death. |
| Landrigan et al7 (used by Makary and Daniel4 and by James9) | 2341 admissions to 10 North Carolina hospitals from 2002 to 2007 | 14 | No stated method for judging the causal relationship between preventable adverse event and eventual death. |
| Classen et al8 (used by Makary and Daniel4 and by James9) | 795 admissions randomly selected from hospitalisations of at least 24 hours from October 2004 in three hospitals | 9 | No distinction between preventable and non-preventable adverse events. No stated method for judging the causal relationship between adverse event and eventual death. |
| HealthGrades5 (used by Makary and Daniel4) | Hospital-acquired conditions identified using Patient Safety Indicators and resulting prevalence combined with estimates of the mortality attributable to these conditions | 389 576 | Estimate depends on questionable values for the attributable mortality associated with hospital-acquired conditions. This number of preventable deaths represents over half of all inpatient deaths in the USA in a year.17 |
| Leapfrog group10† | Similar to above | 206 021 | Estimate depends on questionable values for the attributable mortality associated with hospital-acquired conditions. This number of preventable deaths represents about 30% of all inpatient deaths in the USA in a year.17 |
*As discussed in the text, none of these studies directly assessed the prevalence of preventable deaths. The several studies12 ,13–15 that did focus on identifying preventable deaths all reported under 5% of all hospital deaths as preventable, leading to a 10-fold lower estimate of deaths due to medical error.
†This estimate was not included by either Makary and Daniel4 or by James.9 It has been included because it uses similar methodology to the estimate from HealthGrades but does so with some refinements and more contemporary data.