450,000; these are likely underestimates given the higher rates of another study (Jha et al., 1998). There is one estimate of the extra costs of inpatient care for a preventable ADE—$5,857 (Bates et al., 1997); this figure excludes health care costs outside the hospital and is derived from 1993 cost data. Assuming conservatively an annual incidence of 400,000 preventable ADEs, each incurring extra hospital costs of $5,857, gives a total cost of $2.3 billion (1993 dollars) or $3.5 billion8 (2006 dollars).
Long-term care—Gurwitz and colleagues (2005) projected an annual incidence of 800,000 preventable ADEs—again likely an underestimate given the higher ADE rates of earlier studies (Gerety et al., 1993; Cooper, 1999). However, there is no estimate of the health care costs for this group of preventable ADEs.
Ambulatory care—The best estimate derives from a study (Field et al., 2005) of the costs of ADEs in older adults, which estimated the annual cost of preventable ADEs for all Medicare enrollees aged 65 and older. The cost per preventable ADE was $1,983, and the national annual costs were estimated to be $887 million in 2000. These figures include the costs of inpatient stays (62 percent of the total cost), emergency department visits (6 percent), outpatient care and physician fees (28 percent), and prescribed medicines (4 percent). The national estimate is almost certainly conservative because the detection approach used did not include direct patient contact, which identifies many more ADEs than other approaches.
In addition to the likelihood of being underestimates, the above estimates have some important omissions. First, the costs of some highly common medication errors, such as drug use without a medically valid indication and failure to receive drugs that should have been prescribed, were excluded from the Medicare study of ambulatory ADEs (Field et al., 2005). Moreover, the costs of morbidity and mortality arising from the lack of adherence to the drug regimen were not assessed. Second, all the cost studies omitted other important costs—lost earnings, the costs of not being able to carry out household duties (lost household production), and compensation for pain and suffering. Third, few data are available for any setting regarding the costs of medication errors that do not result in harm. While no injury is involved, these errors often create extra work, and the costs involved may be substantial. For example, one estimate suggested that a 700-bed hospital has 300,000 medication errors per year, each of which creates approximately 20 minutes of extra work for providers—mainly nurses and pharmacists (Bates et al., 1995a). Near-misses may also cost