Automated Adverse Drug Event Monitoring in the VHA
John F. Hurdle MD PhD
Salt Lake City
Salt Lake City, UT
Funding Period: January 2000 - June 2002

BACKGROUND/RATIONALE:
Considerable effort has been devoted to the study of adverse drug events (ADEs), redoubled in the wake of the now-famous Institute of Medicine report on the nature of error in health care. Investigators have shown that inpatient ADEs constitute an important source of patient morbidity and mortality, contribute substantially to health care costs, and are preventable in many cases. However, the general frequency and character of inpatient ADEs is not at all clear from the literature: in just the past 10 years the reported incidence by admission varies 10-fold, from 2-3 percent to over 32 percent. The United States sees over 30,000,000 admissions annually; it is vital to understand the genesis and preventability of inpatient ADEs.

OBJECTIVE(S):
The original aims of the study were to mount a comprehensive ADE surveillance of inpatient ADEs and then to use the resulting portfolio of cases as a gold standard in evaluating a new ADE monitor added to the clinical patient record system (CPRS). However, we detected ADEs at much higher rates than predicted by the literature. So we emphasized ADE classification and analysis, undertaking several substudies designed to capitalize on the wealth of the new information. Expanded by a six-month extension, the objectives now include reviews of non-CPRS ADE signals and of pre/post admission ADES; and a reassessment of clinical responses to ADEs.

METHODS:
This is a prospective, experimental design where inpatients to our tertiary-care facility (single site) over a 20 week period were selected randomly (inclusion:exclusion ratio 2:3), providing a sample of 937 admissions. Study participants were followed by a clinical pharmacist for their entire admission, and a subset (10%) were interviewed twice by a non-pharmacist blinded to the case review. Pharmacists followed daily all electronic progress notes, laboratory values, and orders, looking for signals indicating an ADE. Suspected ADEs were characterized by drug(s) involved, event syndrome, signal type, patient characteristics, team response to ADE, medical severity, hospital resources consumed, drug-event causality, pharmacologic type, and detailed error analysis. All suspected ADEs were verified by a multidisciplinary panel (MDs, pharmacists, nurse). In the study extension, clinical pharmacists retrospectively reviewed the same admits for evidence of pre/post admission ADEs, and also reviewed nursing narratives not otherwise available in CPRS for evidence of ADEs.

FINDINGS/RESULTS:
We verified 521 non-trivial ADEs (4% mild, 94% moderate, 2% severe) for an incidence density of 7.6/100 patient days. This translates to one or more ADEs in 27 percent of admissions. This is seven times higher than rates typically reported (by incidence density). This result underscores two important points: 1) even with strong preventative measure in place, ADEs are far more common than the literature suggests; and 2) a sensitive, standards-based surveillance methodology is well-suited to highlighting the true extent of inpatient ADEs. The CPRS showed no documentation of the events by the care team in 40 percent of cases; 26 percent of events were judged clearly due to error: failing to prophylax against (21%) or to monitor for (21%) common side effects, or improper dosing (18%) being the most frequent errors. The commonly involved drugs were: narcotic analgesics (27%), cardiovascular-renal/non-diuretics (18%),cardiovascular-renal/diuretics (17%), and replacements-regulators of electrolytes/water (10%). Age and polypharmacy were shown to correlate significantly with ADEs. For example, we found that patients with a daily average of 12 or more dispensed doses (considering all drugs) were 2.8 times (95% CI 1.9 to 4.2) more likely to have an ADE than those with less than 12.

IMPACT:
A remarkable finding that requires a thorough re-evaluation of medication ordering and administration across the VHA is that, in spite of having in place ten highly-regarded, literature-based anti-ADE preventative measures (both computerized and clinical), very high rates of ADEs persist. The analysis of the clinical context of the ADEs we analyzed will lead to improved ADE interventions and concomitant reductions in veteran morbidity and mortality.

PUBLICATIONS:

Journal Articles

1. Nebeker JR, Hoffman JM, Weir CR, Bennett CL, Hurdle JF. High rates of adverse drug events in a highly computerized hospital. Archives of Internal Medicine. 2005; 165(10): 1111-6.
2. Weir C, Hoffman J, Nebeker JR, Hurdle JF. Nurse's role in tracking adverse drug events: the impact of provider order entry. Nursing Administration Quarterly. 2005; 29(1): 39-44.
3. Hurdle JF. Can the electronic medical record improve geriatric care? Geriatric Times. 2004; 5(2): 25-26.
4. Nebeker JR, Hurdle JF, Bair BD. Future history: medical informatics in geriatrics. The Journals of Gerontology Series A, Biological Sciences and Medical Sciences. 2003; 58(9): M820-5.
5. Weir CR, Hurdle JF, Felgar MA, Hoffman JM, Roth B, Nebeker JR. Direct text entry in electronic progress notes. An evaluation of input errors. Methods of Information in Medicine. 2003; 42(1): 61-7.

DRA: Health Services and Systems, Substance Abuse, Addictive Disorders
DRE: Prevention, Quality of Care, Technology Development and Assessment
Keywords: Adverse events, Decision support, Safety
MeSH Terms: none