Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Adverse drug reactions

Beijer & Blaey [2]
Wiffen et al [3]
UK burden

Adverse drug events or reactions (for convenience, ADR) have important consequences for individuals and organisations ( Bandolier 28 ) . In-hospital ADRs have been ranked as the fourth to sixth most common cause of death in the USA, based on a meta-analysis of only North American studies [1]. We now have two more systematic reviews of the world literature [2,3] which confirm the size of the problem and provide some insights about what to look for.

Beijer & Blaey [2]

This Dutch study looked for studies relating hospital admission to adverse drug reaction. Excluding papers about drug and alcohol abuse, and drug-related problems during hospital stay, they were left with 68 studies. They used a WHO definition of ADR that excluded therapeutic failure, intentional and accidental poisoning, and drug abuse.


There were 6,000 ADR-related admissions in a total of 124,000, giving an overall rate of 4.9%, but with large inter-study variation. Almost all of the variation was found in small studies (especially those with just a few hundred patients), with more consistent results in larger studies. The variation could be related just to size, or to special circumstances examined in particular studies.

Subgroup analysis looked at elderly versus non-elderly, (over 65 was the usual criterion). Here the ADR admission rate was 16.6%, compared with 4.1% in younger people. Analysis by condition or medicine was not possible, but year of publication and type of hospital made no difference.

The authors attempt to calculate what the cost was to the Dutch health care system in 1998, and came up with the range of Euro186 million to Euro430 million a year. This is based on estimates that a significant proportion of ADR-related admissions in the elderly are preventable.

Wiffen et al [3]

The Oxford study (available as a PDF on the Bandolier Internet site) used a similar if wider search strategy looking for:

Primarily the ADR rate was the key outcome, rather than admissions from ADR, though the ADR rate could be the proportion of patients or admissions. Admission because of an ADR and ADR whilst in hospital were also examined separately. Several prespecified sensitivity analyses were defined.


There were 69 unique studies with evaluable data on 412,000 patients, with an overall ADR rate of 6.7%. Of the 69 studies, 54 were prospective with an ADR rate of 5.5% (193,000 subjects) and 15 were retrospective with an ADR rate of 7.7% (220, 000 subjects). Larger studies tended to have lower ADR rates than small studies (Figure 1).

Figure 1: ADR rate in studies of different sizes. Vertical line is overall average of 6.7%

Neither geographical setting nor publication before or after 1985 made much difference, except when clinical setting was added (Table 1). Studies in general medicine after 1985 had a lower ADR rate than those before (3% versus 9%), and those in older people had a higher rate in post-1985 studies (20% versus 4%). Adverse drug reactions were also examined by specialist clinical setting (cancer, emergency departments, for instance) and with specific classes of medicines, and by gender and culture. ADR rates in inpatients and admissions with ADRs were similar.

Table 1: ADR rates with specialty and age of study


Pre/Post 1985

No of subjects

ADR rate % (95% CI)

General medicine Pre 60401 8.5 (8.2-8.7)
General medicine Post 243803 2.9 (2.8-3.0)
Geriatric Pre 11212 4.3 (3.9-4.7)
Geriatric Post 3488 20 (19-21)
Paediatric Pre 469 4.2 (2.4-6.0)
Paediatric Post 837 3.1 (1.9-4.3)

Several studies examined the interaction between age and number of medicines taken. Older age, and increasing numbers of medicines, especially in women, were associated with ADR rates of between 20% and 50%.

UK burden

Calculating from the number of accident and emergency visits and inpatient days, the rates of ADR likely from UK, European and US studies, and average stays, the estimate for the burden on the UK NHS was equivalent to about 15-20 400-bed hospitals. This would consume about 4% of available bed-days and cost about £380 million.


Both studies also examine the evidence on methods, like computer monitoring or decision aids that can help reduce ADR rates ( Bandolier 73 ) . Estimates in both studies were that medication errors can be substantially reduced by using computer systems. When lack of capacity is a major constraint on delivery of health care, avoiding ADRs would, together with reducing hospital-acquired infections, be a major contribution to making things better, and would probably be cost saving if introduced.


  1. Lazarou J et al. Incidence of Adverse Drug Reactions in Hospitalized Patients: A Meta-analysis of Prospective Studies. JAMA 1998; 279(15):1200-05.
  2. HJ Beijer, CJ de Blaey. Hospitalisations caused by adverse drug reactions (ADR): a meta-analysis of observational studies. Pharmacy World & Science 2002 24: 46-54.
  3. PJ Wiffen et al. Adverse drug reactions in hospital patients: a systematic review of the prospective and retrospective studies. Bandolier Extra, June 2002, www/
previous or next story in this issue