Frequency and nature of adverse drug reactions in elderly in-patients of two Indian medical college hospitalsA Harugeri1, G Parthasarathi1, M Ramesh1, S Guido2, H Basavanagowdappa3
1 Department of Pharmacy Practice, J.S.S. College of Pharmacy and J.S.S. Medical College Hospital, J.S.S. University, Mysore, India
2 Department of Pharmacology, St. John's Medical College Hospital, Bangalore, India
3 Department of Medicine, J.S.S. Medical College Hospital, J.S.S. University, Mysore, Karnataka, India
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0022-3859.85201
Source of Support: None, Conflict of Interest: None
Background : Adverse drug reactions (ADRs) are a major public health problem in the elderly. Although the Indian elderly represent 12.8% of the entire global elderly population, data on prevalence and predictors of ADRs in elderly Indians is extremely limited. Aim : To determine the prevalence, severity, preventability, length of hospital stays, and risk factors for ADRs in hospitalized Indian elderly. Setting : Medicine wards of two tertiary care teaching hospitals. Design : Prospective study was conducted between July 2007 and December 2009. Materials and Methods : In-patients of either sex and aged ≥60 years were included and monitored for ADRs throughout their hospital stay. Severity (Hartwig et al. scale), preventability (Shumock and Thornton criteria) and increased length of stay (considering underlying disease, ADR, and discussion with clinicians) were assessed. Statistical Analysis : Bivariate analysis and subsequently multivariate logistic regression were used to determine the risk factors for developing ADRs. Results : Over the study period, among the 920 patients monitored, 296 patients (32.2%) experienced 419 ADRs. Among all ADRs, 48.4% (203) were preventable. Majority of ADRs [226 (53.9%)] were moderate in severity. Therapeutic classes of drugs frequently associated with ADRs were the drugs used in diabetes [76 (18.1%)] and antibacterials for systemic use [54 (12.9%)]. ADRs increased the hospital stay in 5.9% (54) of patients. Female gender [Odds Ratio: 1.52, 95% Confidence Interval:1.04-2.22, P=0.03] was observed as the influential risk factor for ADRs. Conclusion : One third of hospitalized elderly experienced ADRs. Interventions focused at preventable ADRs should be developed and implemented to reduce their implications.
Keywords: Adverse drug reaction, aged, geriatrics, India, pharmacovigilance, risk
Adverse drug reactions (ADRs) are regarded as top of the pyramid containing all drug-related problems (DRPs) and are a major public health concern. ,, ADRs cause 3−7% of hospital admissions, ,,, and account for 5−9% of hospital in-patient costs. , More than half of the ADRs are considered to be preventable. , ADRs in the elderly are common in various settings. ,,,, Multiple drug use, and changes in pharmacokinetics and pharmacodynamics predispose the elderly to ADRs. , Risk factors reported to be associated with ADRs in elderly include pathology of ageing, female gender, multiple co-morbidities, multiple drug therapy, length of hospital stay, and use of drugs that are inappropriate for the elderly. ,,,,
Indian elderly (93 million) constitute 12.8% of the global elderly population and is expected to rise to 100 million by 2013. , Prevalence of drug use increases with ageing. Though elderly account for 15% of the population in Western countries, they consume one-third of total volume of prescription medications.  In India, management of an ADR is estimated to cost US $ 15 to 150. ,,
With the increasing elderly population, geriatrics is emerging as a clinical specialty in India.  Although ADRs represent an important DRP in elderly, in India, data on prevalence and predictors of ADRs in elderly is limited. We report the results of our study, which explores the impact of ADRs in hospitalized Indian elderly in terms of prevalence, severity, preventability, length of stay, and risk factors for ADRs.
The study was conducted between July 2007 and December 2009 in the medicine wards of two tertiary care teaching hospitals in southern India. The study protocol was assessed and approved by respective Institutional Ethics Review Boards. Both study hospitals are 1200 bed representative tertiary care teaching hospitals. Our estimate of the prevalence of ADRs excludes patients treated in intensive care units, intensive therapy units, and private medicine wards as our focus was on ADRs occurring in medicine wards that represent most tertiary care teaching hospitals in India.
The clinical pharmacists included patients of either sex aged ≥60 years after obtaining informed consent. Patients were recruited over blocks of 12 weeks from each of the hospital's six medical units. Patients treated in the out-patient departments, admitted on Sundays, and discharged within a day after admission were excluded. Eligible patients were identified using the ward census registers (Monday to Saturday) and monitored daily by a clinical pharmacist.
Suspected ADRs were detected by daily review of patients' treatment charts, clinicians' and nurses' notes by a clinical pharmacist. Clinicians were encouraged to refer patients with suspected ADRs to the clinical pharmacist to record the ADRs in the enrolled patients. World Health Organization (WHO) definition of an ADR was adopted.  Objective findings of ADRs (e.g., biochemical investigation results) were identified from laboratory reports, while subjective markers of ADRs (e.g., abdominal pain) were identified through review of clinicians' and nurses' notes, and also discussion with the patient and clinicians. All patients who developed an ADR were monitored throughout their hospital stay and data were updated to ensure complete data collection.
Causality of suspected ADRs was established using WHO scale  and Naranjo's algorithm.  Preventability and severity of ADRs were assessed by using the modified Schumock and Thornton criteria  and modified Hartwig, et al. scale,  respectively. Causality, preventability, and severity of ADRs were assessed independently by a clinical pharmacist and a pharmacologist/senior academic clinical pharmacist. Discrepancy in scoring between assessors, if any was addressed by discussion. Those ADRs at least with 'possible' causal association with the suspected drug were considered for further analyses.
The length of stay for each patient was calculated by considering the day of admission and discharge. Attribution of increased length of hospital stay due to ADR was made in consultation with the clinicians who were treating the patients by considering the clinical features of the underlying disease and the event. If suspected ADR was the only reason for patient's extended hospital stay, the additional length of hospital stay was attributed to the reaction.
The overall prevalence of ADRs during hospital stay was determined by the ratio of total number of patients who experienced ADRs and total number of patients included in the study. Drugs received and ADRs experienced by the study patients during their hospital stay were recorded and coded using WHO Anatomical Therapeutic and Chemical classification  and WHO-Adverse Reaction Terminologies (WHO-ART),  respectively.
Patients who did and did not experience an ADR were compared with Pearson Chi-Square tests for categorical variables and by Mann-Whitney U test for continuous variables. Risk factors for ADRs were determined at a P value of <0.05 by investigating the effects of age, gender, number of diseases, number of drugs prescribed, number of doses used, length of hospital stay, and history of medication allergy in the bivariate analysis. Multivariate logistic regression was used to evaluate the influence of these risk factors on development of ADRs. All statistical calculations were performed using Statistical Package for Social Science (SPSS) Version 17.0. A P value of <0.05 was considered statistically significant.
Over the study period, 4815 patients were admitted to medicine wards in study settings (58.9% male). A total of 920 patients were monitored for ADRs (59.1% male). Sex distribution of patient included in the study was similar to patient not included in the study (n=2294; 58.9% male, 41.1% female; c2 =0.02, df=1, P=0.464). Median age of the study population was 66 years (range, 60 to 108 years), while the median age of patients not included in the study was 67 years (range, 60 to 101 years) (U=1412020, P=0.377). The median duration of hospital stay was 7 days in both groups of patients included in the study (range, 1 to 43 days) and not included in the study (range, 1 to 48 days) (U=1322668, P=0.202). Characteristics of the study population and prevalence of ADRs are listed in [Table 1].
Women experienced more ADRs (n=135, 35.9%) than men (n=161, 29.6%) (χ2= 4.1, df=1, P=0.026). The median age was higher in the non ADR group (67 years, range, 60 to 95 years) compared to ADR group (65 years, range, 60 to 108 years) (U=64999, P=0.054). Median length of hospital stay in patients who experienced ADRs was 7 days (range, 1 to 43 days) compared to 6 days (range, 1 to 20 days) in non-ADR group (U=34224, P=0.02).
Number of diseases was higher in the ADR group (median 3, range, 1 to 6) than non-ADR group of patients (median 2, range, 1 to 9) (U=61552, P=0.008). During the hospital stay, number of medications prescribed was significantly higher in the ADR group (median 10, range, 3 to 22) (U=79980, P<0.001) compared to non-ADR (median 9, range, 1 to 21). Similarly, number of doses used was significantly higher in ADR group than non-ADR group (median 77, range, 8 to 360 vs. median 64, range, 4 to 369) (U=79670, P<0.001).
Of the 426 ADRs identified, only 419 ADRs were considered for the analyses as remaining seven ADRs were found to be unlikely in their causal relationship to suspected drugs. A total of 296 (32.2%) patients experienced 419 ADRs. The drug classes most commonly implicated in ADRs and the frequency of their use in the study population are shown in [Table 2]. Organ system most affected was metabolic and nutritional disorders [139 (33.2%)] [Table 3]. Overall, there was 80% agreement between the assessors in the causality, preventability, and severity assessments. Results of causality, severity, and preventability of identified ADRs are shown in [Table 4].
The reactions of type A and type B accounted for 92.1% (n=386) and 7.9% (n=33) of ADRs, respectively. In majority of cases [n=312 (74.5%)], patients experienced reactions to medications that were prescribed after hospital admission. Whilst hospital admission, 71 (16.9%) ADRs were observed to prior medications. Thirty six (3.9%) patients were hospitalized due to ADRs. In 54 (5.9%) patients, ADRs directly led to increased length of hospital stay (median 2 days, range, 1 to 5 days). Of the 6487 bed days of study patients, ADRs accounted for 145 (2.2%) bed days. [0.2 bed days/patient (145/920)].
Bivariate analysis identified number of diseases (≥4), number of drugs used during the hospital stay (≥10), length of hospital stay (≥10 days), and number of doses used (≥200) as the risk factors for developing ADRs [Table 1]. However, female gender was the only influential risk factor identified in the logistic regression model [Table 5].
This is the largest prospective study of ADRs in Indian hospitalized elderly. About one third of study population experienced ADRs. Of the identified ADRs, 48.4% were preventable. Increased length of hospital stay due to ADRs accounted for 0.2 bed days/patient. Female gender was observed as influential risk factor for developing ADRs. The national cost of ADRs-related hospital stay in elderly was estimated to be US $ 540000.
The cut off age (60 years) for the elderly in India and other developing countries is justified.  To our knowledge, only one study on ADRs in elderly from India is published.  Compared to our study, the study by S. Malhotra, et al. included fewer patients (n=578) at one hospital, evaluated ADRs in emergency setting, and did not assess preventability and severity of ADRs.  Prevalence of ADR-related hospital admissions observed in our study (5.9%) and the study by S. Malhotra, et al. (6.7%) suggests that ADR-related hospital admissions are common both in medicine and emergency wards. Hypoglycemia to drugs used in diabetes was observed as most frequent ADR both in our study (15.7%) and the study by S. Malhotra, et al. (30.8%),  thus suggesting the need to focus on monitoring elderly receiving antihyperglycemic drugs to reduce the medication related harm. Prevalence of ADRs in medical and surgical wards may differ due to differences in drug use patterns.
High degree of confidence was ensured in causality assessment by using two tools and two assessors. Although, inter and intra-rater agreements vary in causality assessment of ADRs,  in our study the disagreement between two assessors was minimal. Accuracy of ADR assessment was ensured through data collection by daily follow-up and active surveillance method as adopted in other studies. ,, Therapeutic class of drugs that were most frequently implicated in ADRs were in concordance with other studies. ,,, This may be partly due to their frequent use. This suggests efforts to prevent ADRs in hospitalized Indian elderly should focus on frequently implicated medications as about half of the ADRs were probably or definitely preventable consistent with the literature (30−70%). ,,,
The ADRs like hypoglycemia, hypokalaemia, hyponatremia, hyperglycemia, and hyperkalaemia contributed to metabolic and nutritional disorders observed as the most commonly affected organ system due to ADRs. The severity of ADRs was either level 3 or below in 77.1% of ADRs that require interventions including cessation of drug and/or specific/symptomatic treatment. For example, use of pyridoxine for peripheral neuropathy due to isoniazid, or dextrose for insulin-induced hypoglycemia. The remaining ADRs (22.9%) were serious as defined by International Conference on Hormonization.  However, no mortality due to ADRs was observed.
Overall, there are 818311 beds in Indian hospitals.  Our study found 2.2% bed days are due to ADRs. Therefore it is estimated that approximately 18000 bed days in a given time are due to ADRs in elderly. This suggests ADR-related increased hospital stay in elderly lead to 100% occupancy of fifteen 1200-bed hospitals in India. We believe that the assessment of ADR-related extended hospital stay was robust as all possible causes other than ADR were eliminated. Considering the average cost of treatment in patients presenting with ADRs as US $30 per day,  total cost of hospital stay due to ADRs is estimated to be US $4350 (INR 200100), that is US $80.5 per patient (108.7% of per capita per year expenditure on health).  The national cost estimate for 18000 bed days of ADRs related hospital stay in elderly is likely to be US $540000 (INR 24.84 million).
The national cost estimate of ADR-related hospital stay in our study is a crude estimate. An accurate estimation of cost of treating ADRs is difficult as US $30 per day considered in the calculation does not include the indirect costs like loss of wages, transportation cost of the patient and attendants/caretakers, and food expenses during hospital stay. Regardless, the estimated cost is very high for Indians considering the percentage of out of pocket expenditure on health (91.4%). 
Female gender, multiple diseases (≥4), longer hospital stay (≥10 days), number of drugs (≥10), and doses (≥200) identified as risk factors for developing ADRs are consistent with literature. ,,, The substantial association of ADRs for females is interesting and merits further study. With standard drug dosing, it may be possible that lighter weight females are at increased risk of ADRs because of relatively higher doses per patient.  Nevertheless, further large cohort studies are required to confirm our findings of risk factors for developing ADRs.
There is a need for research to develop and implement the strategies to prevent/minimize ADRs and to increase the safety in hospitalized Indian elderly. Intervention strategies like appropriate drug use,  focusing on drugs commonly implicated in ADRs,  drug monitoring supported by information technology,  pharmacist participation in ward rounds,  linking pharmacy and laboratory data,  and use of simple and practical method to identify patients who are at increased risk for ADRs  can be considered to prevent and reduce the harm associated with ADRs.
We could not review all the elderly patients admitted to medicine wards due to limited human resource and due to unavailability of electronic medical records at the study settings. The variables tested to determine the risk factors were not adjusted for potential cofounders. The results cannot be extrapolated or generalized as the study did not include the patients of other wards, and the characteristics of patients from other wards may be different than that observed in this study.
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5]