Percentage of adverse drug reactions by drug class among depressed (n = 1363) and nondepressed (n = 1771) patients. NSAIDs indicates nonsteroidal anti-inflammatory drugs; ASA, acetylsalicylic acid; and ACE, angiotensin-converting enzyme.
Odds ratios of adverse drug reactions according to Geriatric Depression Scale (GDS) scores (GDS score 0-1 is the reference group). The analysis was adjusted for age; sex; living alone; activities of daily living disability; cognitive impairment; education; alcohol use; Charlson Comorbidity Index; congestive heart failure; diabetes mellitus; cerebrovascular disease; chronic obstructive pulmonary disease; neoplasm; number of drugs taken during the hospital stay; use of angiotensin-converting enzyme inhibitors, diuretics, digoxin, nitrates, benzodiazepines, corticosteroids, insulin, and antidepressants; and length of hospital stay. P = .002 for linear trend.
Onder G, Penninx BWJH, Landi F, Atkinson H, Cesari M, Bernabei R, Pahor M, . Depression and Adverse Drug Reactions Among Hospitalized Older Adults. Arch Intern Med. 2003;163(3):301-305. doi:10.1001/archinte.163.3.301
Copyright 2003 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2003
Depression is a common disorder among hospitalized older adults, and it has been associated with adverse outcomes during hospital stays, including increased risk of morbidity and mortality and reduced recovery rates from illness and disability. The aim of this study was to assess whether depression may be a risk factor for adverse drug reactions (ADRs) among hospitalized older adults.
A total of 3134 older patients admitted to 23 hospitals throughout Italy between May 1 and June 30 and September 1 and October 31, 1998, entered the study. Patients with a short-form Geriatric Depression Scale score of 5 or greater were considered depressed. Adverse drug reactions observed during hospital stays and classified as definite or probable on the basis of the Naranjo algorithm were considered for this study.
During the hospital stays, 192 ADRs were identified in 183 patients (5.8% of the sample). Cardiovascular and arrhythmic complications (20.3% of all ADRs) were the most frequent ADRs, followed by gastrointestinal (18.8%), dermatologic and allergic (12.5%), hemorrhagic (11.5%), and electrolyte (9.9%) disturbances. Adverse drug reactions were recorded in 101 (7.4%) of 1363 depressed patients and in 82 (4.6%) of 1771 nondepressed patients (P = .001). After adjusting for potential confounders, depression was associated with a significantly higher rate of ADRs (odds ratio, 1.58; 95% confidence interval, 1.14-2.20; P = .006). This effect seemed more pronounced in women (odds ratio, 1.85; 95% confidence interval, 1.16-2.95) than in men (odds ratio, 1.38; 95% confidence interval, 0.85-2.34).
In older hospitalized patients, depression seems to be associated with a greater occurrence of ADRs.
DEPRESSION IS a common disorder among hospitalized older adults: an estimated 10% to 20% of older inpatients experience major depression, and another 20% to 30% have minor depression.1 These conditions have been associated with adverse outcomes during hospitalization, including increased risk of morbidity and mortality, reduced recovery rates from illness and disability, and worse functional and health status after discharge.2- 4
Depression is also associated with increased use of psychotropic and nonpsychotropic drugs, inappropriate drug prescription, and reduced pharmacologic compliance.1,5,6 These factors, in turn, can potentially increase the risk of pharmacologic adverse events in individuals with depression.
In Western countries, adverse drug reactions (ADRs) result in 3% to 5% of all hospital admissions, account for 5% to 10% of in-hospital costs, and substantially increase morbidity and mortality.7- 9 Age, sex, comorbidity, multiple-drug regimens, alcohol intake, and cognitive function have been shown to be independently associated with ADRs in previous studies.7,10- 12 Despite concerns that ADRs represent a significant medical problem, it seems that the relationship between conditions such as depression and the occurrence of ADRs has not yet been studied, to our knowledge.
We assessed whether depression may represent a risk factor for ADRs among hospitalized older adults. To address this potential association, we used data from the Gruppo Italiano di Farmacoepidemiologia nell'Anziano, a study specifically designed to collect data about ADRs among in-hospital patients in Italy.
The Gruppo Italiano di Farmacoepidemiologia nell'Anziano study involved community- and university-based hospitals throughout Italy. During the study, information was periodically collected about patient drug use, occurrence of ADRs, and quality of hospital care.7,11,12
All patients admitted to 81 geriatric and internal medicine wards participating in the study were enrolled and followed until discharge. The study periods were May 1 to June 30 and September 1 to December 31, 1988; May 15 to June 15, 1991; and May 1 to June 30 and September 1 to October 31 in 1993, 1995, 1997, and 1998. The study was approved by the ethical committee of the Catholic University in Rome.
For each participant, a questionnaire was completed at hospital admission, and it was updated daily by a study physician who received specific training. Data recorded included sociodemographic characteristics, indicators of physical function and cognitive status, clinical diagnoses at admission and at discharge, and medication use before and during the hospital stay as well as medications prescribed at discharge.
For the present analysis, we used data collected in 1998, because the depression survey (the short form [15 items] of the Geriatric Depression Scale [GDS]) was administered only during this period.13 This instrument, which was administered to hospitalized study patients in stable health condition, has proved to be reliable for detecting depression among inpatients.14 It has also been validated in the Italian population.15 Patients with GDS scores of 5 or greater were considered to be depressed, based on previous observations among in-hospital patients.14
Cognitive performance was assessed using the Hodkinson Abbreviated Mental Test.16 Based on a previous observation in an Italian population, a score of less than 7 defined cognitive impairment.17
Drugs were coded according to Anatomical Therapeutic and Chemical codes.18 Discharge diagnoses were coded according to International Classification of Diseases, Ninth Revision, Clinical Modification, codes.19 Comorbidity was quantified using the Charlson Comorbidity Index by adding scores assigned to specific discharge diagnoses, as illustrated in the original publication.20
An ADR was considered to be any noxious, unintended, and undesired effect of a drug, excluding therapeutic failures, intentional and accidental poisoning, and abuse.21 A study physician investigated each ADR detected during hospital stays by gathering information from patients, nurses, and attending physicians and by reviewing medical charts and records. For each suspected ADR, the study physician coded the clinical description, severity, and outcome of the ADR and collected detailed information about the drug(s) that potentially caused the ADRs. Causality between drug use and ADR was assessed using scores on the Naranjo algorithm.22 Adverse drug reactions were classified as definite (score, 9-12), probable (score, 5-8), possible (score, 1-4), or doubtful (score, ≤0). Only definite and probable ADRs observed during hospital stays were used in this study.
Differences between depressed and nondepressed patients in categorical variables were tested using the Fisher exact test. Differences between continuous variables were assessed using analysis of variance comparisons for normally distributed variables; alternatively, the Kruskal-Wallis test was adopted. To establish whether depression represented a risk factor for experiencing any ADR, a logistic regression model was performed in the 3134 patients participating in the 1998 survey who had valid GDS data. To explore a potential trend between depressive symptom severity and ADRs, an additional logistic regression model was conducted using categorization of the GDS score: 0 to 1 (n = 839), 2 to 4 (n = 932), 5 to 7 (n = 664), 8 to 10 (n = 464), and 11 to 15 (n = 235). Variables considered for adjustment were those associated with depression at P≤.05 in the univariate analyses.
To assess whether depression was associated with ADRs independent of comorbidity level, we performed additional stratified analyses across 3 groups, classified on the basis of the Charlson Comorbidity Index score: 0 to 1 (n
= 1745), 2 to 3 (n = 973), and 4 or higher (n = 416). We also assessed the association between depression and (1) ADRs that reflect subjective symptoms (headache, abdominal pain, nausea, etc) and (2) ADRs that reflect objective signs, laboratory tests, or diagnostic procedures (skin rashes, hemorrhagic complications, electrolyte disturbances, etc). All analyses were performed using statistical software (SPSS for Windows, version 10.0; SPSS Inc, Chicago, Ill).
A total of 3134 patients were enrolled in the 1998 portion of the study, during which GDS data were collected. Mean ± SD patient age was 72.0 ± 14.1 years, and 45.6% of the study population was female. Of the total enrolled sample, 1363 (43.5%) experienced depression during hospitalization (GDS score, ≥5). The mean ± SD GDS score was 4.5 ± 3.7. Other characteristics of the study population are summarized in Table 1. Compared with nondepressed patients, those with depression were older, were more likely to be female, had a higher prevalence of cognitive impairment and disability, and had a more severe Charlson Comorbidity Index. Depression was associated with a significantly higher prevalence of congestive heart failure, diabetes mellitus, cerebrovascular disease, chronic obstructive pulmonary disease, and neoplasms. Patients with depression used more drugs during hospital stays. In particular, they were more likely to use cardiovascular medications, antidiabetic drugs, corticosteroids, and neurotropic drugs.
During the hospital stays, a total of 192 probable or definite ADRs were diagnosed in 183 patients (5.8% of the sample). An ADR was recorded in 101 (7.4%) of the 1363 patients with depression and in 82 (4.6%) of the 1771 patients without depression (P = .001) (Table 2). In the unadjusted model, depression was associated with a 65% increased risk of developing ADRs (odds ratio [OR], 1.65; 95% confidence interval [CI], 1.22-2.23).
After adjusting for potential confounders, this association was still present (OR, 1.58; 95% CI, 1.14-2.20). The adjusted association between depression and risk of developing ADRs was more pronounced in women (OR, 1.85; 95% CI, 1.16-2.95) than in men (OR, 1.38; 95% CI, 0.85-2.34) (Table 2), although the interaction was not significant (P = .41). Associations between depression and ADRs were similar for ADRs reflecting subjective symptoms (n = 60; OR, 1.59, 95% CI, 0.90-2.81) and those reflecting objective signs or measures (n = 123; OR, 1.56; 95% CI, 1.05-2.31).
Table 3 provides the frequency of ADRs by type in depressed and nondepressed patients. Cardiovascular and arrhythmic complications (20.3% of all ADRs) were the most frequent ADRs, followed by gastrointestinal (18.8%), dermatologic and allergic (12.5%), hemorrhagic (11.5%), and electrolyte disturbances (9.9%). Except for neurologic and neuropsychiatric ADRs, which were significantly more common among depressed patients (P = .001), no significant differences were found for other types of ADRs between the 2 groups.
Figure 1 shows the drug classes that contributed most frequently to ADRs in the study sample. The most common culprit drugs were diuretics, antibiotics, and angiotensin-converting enzyme inhibitors among depressed patients, and digoxin, nitrates, and anticoagulants among nondepressed patients.
Figure 2 summarizes the ORs for ADRs across different groups, according to GDS scores. The risk of developing ADRs progressively and significantly increased as GDS score increments increased (signifying more severe depression) (P = .002 for linear trend).
An ADR was observed in 7.3% of depressed and 4.3% of nondepressed participants scoring 0 to 1 in the Charlson Comorbidity Index (P = .007), in 7.3% of depressed and 5.0% of nondepressed participants scoring 2 to 3 (P = .12), and in 7.9% of depressed and 5.4% of nondepressed participants scoring 4 or more (P = .31). After adjusting for potential confounders, depression was associated with a similar increased risk of ADRs in all 3 comorbidity groups (Charlson Comorbidity Index score, 0-1: OR of ADRs for depression, 1.54; score, 2-3: OR, 1.60; score, ≥4: OR, 1.94). Results of these stratified analyses indicate that the risk of ADRs associated with depression is independent of the level of comorbidity because ORs were similar across groups.
On the basis of our findings in hospitalized older patients, depression seems to be associated with an increased risk of developing ADRs. This association increases with severity of depression.
The prevalence of depression in our sample is higher than that observed among patients in the community, but it is similar to that found in other studies1- 4 conducted in the hospital setting. The correlation between depression and hospitalization has been documented by other researchers,23 who reported that depressive symptoms frequently represent a reaction to severe disability and discomforts that are associated with medical illness.
The higher risk of ADRs that we observed among depressed patients may be due to a variety of factors. First, depressed patients can amplify somatic symptoms, leading to a higher reported rate of ADRs.24 In this context, it has been suggested that emotional distress can lead to increased attention directed toward one's body, with a consequent decrease in the threshold of any noxious somatic sensation.25 However, this hypothesis is not supported by our results, given that in our study population, associations between depression and either subjective or objective ADRs were similar.
Second, it has been hypothesized that psychological distress can activate neurally regulated biological processes. This can result in diminished ability to combat pathologic processes, thus favoring the onset of negative outcomes such as ADRs. This phenomenon, described by Engel26 as the "giving-up–given-up complex," could explain the increased risk of adverse outcomes observed in depressed patients. According to this hypothesis, depression adversely affects cardiac, gastrointestinal, endocrine, neurologic, and immune processes by increasing sympathetic tone and decreasing vagal tone.27- 29
A third possible explanation for our findings is that depressive symptoms may occur as a consequence of ADRs and the high comorbidity associated with ADRs. Because depression data for our study were collected on inpatients with stable health conditions, we are unable to evaluate whether a temporal relationship existed between the onset of ADRs and subsequent development of depression.
The present study has several strengths. First, the relationship between depression and ADRs was studied using a dedicated database. Second, the hospital was an ideal setting to evaluate this association because pharmacologic noncompliance, which can play an important role in the onset of ADRs among depressed patients, is reduced. Finally, to describe the causal relationship between ADRs and drug exposure, we used an algorithm that is associated with 85% interobserver agreement.22
An important limitation of this study relates to generalizability of the results. Our findings, which are based on an elderly hospitalized population, cannot be extrapolated to younger individuals living in the community.
In conclusion, depression seems to be associated with a higher rate of developing ADRs. This finding may be relevant to physicians prescribing medications, who may want to monitor patients with depressive symptoms more closely. Future studies on ADRs in the older population should consider multiple complex aspects of aging, including depression.
Corresponding author: Graziano Onder, MD, Centro Medicina dell'Invecchiamento, Università Cattolica del Sacro Cuore, Policlinico A. Gemelli, L.go Francesco Vito 1, 00168 Rome, Italy (e-mail: email@example.com).
Accepted for publication June 13, 2002.
This study was partially supported by grant 94000402 from the National Research Council, Rome, Italy; by Neopharmed (Gruppo Italiano di Farmacoepidemiologia nell'Anziano), Rome; and by grant 5P60 AG 10484-07 from the US National Institutes of Health, Claude Pepper Older Americans Independence Center, Bethesda, Md (Drs Onder, Penninx, and Pahor).