Adverse drug reactions in elderly patients as contributing factor for hospital admission: cross sectional studyBMJ 1997; 315 doi: https://doi.org/10.1136/bmj.315.7115.1057 (Published 25 October 1997) Cite this as: BMJ 1997;315:1057
- Cyndie K Mannesse, consultant in internal and geriatric medicinea,
- Frans H M Derkx, consultant in clinical pharmacology and internal medicineb,
- Maria A J de Ridder, statisticianc,
- Tischa J M van der Cammen, consultant in internal and geriatric medicinea,
- Arie J Man in't Veld, professor in internal medicineb
- a Department of Internal Medicine I, Division of Geriatric Medicine, Room D442, University Hospital Rotterdam Dijkzigt, Dr Molewaterplein 40, 3015 GD Rotterdam, Netherlands
- b Department of Internal Medicine I, University Hospital Rotterdam Dijkzigt
- c Institute for Epidemiology and Biostatistics, Erasmus University Rotterdam, Rotterdam
- Correspondence to: Dr Mannesse
- Accepted 25 September 1997
Adverse reactions to drugs in elderly patients are a major clinical problem.1 Patients' perceptions about the presence of drug reactions may contribute to their earlier detection,2 but no studies have related patients' complaints about adverse reactions to the objective presence of reactions, and no study has been performed in elderly patients.
Subjects, methods, and results
All admissions of patients aged 70 and over to our medical wards from 1 February to 1 May 1994 were analysed. We excluded patients transferred from other wards and those readmitted within a month. Patients gave their informed consent. Data were collected on drug treatment, and patients were asked whether they thought any of their medical problems were caused by their drugs. The presence and severity of an adverse reaction was assessed from the history, examination, and laboratory tests, using the methods of Kramer et al3 and Hutchinson et al4 and the British National Formulary.5 A reaction was defined as severe if it was potentially life threatening or led to admission. Differences in patient characteristics (1) and the use of specific drugs in patients with adverse reactions and those without were tested with χ2 or Fisher's test for dichotomous variables and Student's t test or the Mann-Whitney test for continuous variables. Variables with a significant influence on univariate analysis (P0.05) were used in a backward selection (using P<0.05 as criterion) to build a multiple logistic regression model. Results are expressed as two tailed P values.
A total of 128 patients were identified (24% of all admissions), but 22 were excluded because of transfer, readmission, refusal, or death. The 106 patients had a mean age of 78 (70-91); 60 were women. Of the total, 102 were receiving drug treatment (mean 5.9 drugs); 44 patients admitted (42%, 95% confidence interval 32% to 51%) had had one or more adverse reactions and 25 had had severe reactions (24%, 16% to 32%). None of the 15 deaths during the study were directly related to an adverse drug reaction.
Comparison of patients with and without an adverse reaction causing admission was feasible in 12 patients with gastrointestinal bleeding or haematuria: 5 were taking an oral anticoagulant, which was used by only 9 of the 94 patients who were not admitted for bleeding (P=0.009). In multivariate analysis a fall before admission (odds ratio 51.3 (3.2 to 834.3), P=0.006), the presence of gastrointestinal bleeding or haematuria (19.8 (4.4 to 88.6), P<0.001), and the use of three or more drugs (14.5 (1.2 to 175.3), P=0.04) were significant factors in patients with severe adverse reactions (1). Severe reactions in 5 patients admitted after falling included orthostatic hypotension (2 patients were taking diuretics), drug intoxication (1 patient taking carbamazepine and 1, phenobarbitone), and anaemia caused by gastrointestinal bleeding (1 taking a non-steroidal anti-inflammatory drug).
The 102 patients having drug treatment were asked whether they had complaints caused by their drugs. A correct opinion about the presence of adverse drug reactions was found in 73 of the 93 patients (79%, 70% to 87%) who could answer the question: 28 correctly recognised a reaction, while 45 who did not complain of an adverse reaction correctly affirmed its absence—a sensitivity of 0.70 and a specificity of 0.85. The negative predictive value of the question was 0.79 and the positive predictive value 0.78. Eighteen of the 25 patients with a severe reaction could not, however, implicate the drug in their current illness; these severe reactions were unrecognised by 14 of the 57 patients who did not complain of adverse reactions (24.6%) and in 4 of the 9 patients who could not answer the question.
Overall, 1 in 6 elderly patients admitted to a general ward experienced adverse drug reactions; severe reactions occurred in 24%. This study confirms the view that drug reactions are related to the use of more drugs rather than the presence of more diagnoses. Other significant factors in severe reactions were a fall before admission and the presence of gastrointestinal bleeding or haematuria. Mild drug reactions were well recognised by these patients, but not severe ones.
Funding: The Erasmus Centre for Research on Aging gave financial support for the statistical analysis.
Conflict of interest: None.