Corsonello A, Pedone C, Corica F, Mussi C, Carbonin P, Antonelli Incalzi R, Gruppo Italiano di Farmacovigilanza nell’Anziano (GIFA) Investigators. Concealed Renal Insufficiency and Adverse Drug Reactions in Elderly Hospitalized Patients. Arch Intern Med. 2005;165(7):790–795. doi:10.1001/archinte.165.7.790
Adverse drug reactions (ADRs) are common causes of in-hospital complications for elderly people. The purpose of the present study is to verify whether concealed renal insufficiency, that is, reduction of the estimated glomerular filtration rate (GFR) in people with normal serum creatinine levels, is a risk factor for ADRs in elderly hospitalized patients.
We used data on 11 687 hospitalized patients enrolled in the Gruppo Italiano di Farmacovigilanza nell’Anziano study. The outcomes of the study were any ADR, ADR to hydrosoluble drugs, and ADR to any other drug during the hospital stay. We compared 3 groups: normal renal function (normal serum creatinine levels and normal estimated GFRs), concealed (normal serum creatinine levels and reduced estimated GFRs), or overt (increased creatinine levels and reduced estimated GFRs) renal insufficiency. The relationship between renal function and ADR was evaluated using contingency tables and multiple regression analysis including potential confounders.
Concealed renal insufficiency was detected in 1631 (13.9%) patients and was frequently associated with male sex and poor nutritional status. Hydrosoluble drugs were responsible for 301 of the 941 recorded ADRs. After adjusting for potential confounders, both concealed (odds ratio [OR], 1.61; 95% confidence interval [CI], 1.15-1.25) and overt (OR, 2.02; 95% CI, 1.54-2.65) renal failure were associated with ADR to hydrosoluble drugs, but not with ADR to other drugs (OR, 0.83 [95% CI, 0.65-1.08], and OR, 1.01 [95%CI, 0.83-1.23], respectively).
Older hospitalized patients frequently have impaired renal function despite normal serum creatinine levels and are exposed to an increased risk of ADRs to hydrosoluble drugs.
Adverse drug reactions (ADRs) are important and common causes of hospitalization as well as of in-hospital complications for elderly people.1- 4 Age has been shown to be an independent risk factor for ADRs in most but not all the studies, while comorbidity and polypharmacy have been reported to be the main risk factors for ADRs.3- 8
Aging is associated with decreased renal and liver reserve and with the risk of delayed renal and hepatic clearance of drugs.9 The available research suggests that this conclusion is uniformly true for hydrosoluble drugs, which are cleared by the kidney, and for most lipophilic drugs.10 Lipophilic drugs that undergo low extraction with phase II (conjugation) hepatic metabolism are as efficiently cleared in the elderly as in young adults.10 Therefore, loss of renal function has a uniformly foreseeable effect on the risk of ADRs, whereas the same is not true for impaired liver function.
Renal function can be readily estimated by the serum creatinine level, but this estimate is hardly reliable in the elderly because of the frequent loss of muscle mass secondary to age itself and aging-related conditions.11 Some of these diseases, such as type 2 diabetes mellitus or hypertension, are also a cause of renal dysfunction,12 and elderly patients with these diseases are more likely to have concealed renal insufficiency, that is, renal insufficiency despite a normal serum creatinine level. The present study has been designed to verify whether concealed renal insufficiency can be considered a risk factor for ADRs in a geriatric population.
The present study uses data from a large collaborative observational study group, the Gruppo Italiano di Farmacovigilanza nell’Anziano (GIFA), based in community and university hospitals located throughout Italy, that periodically surveys drug consumption, occurrence of ADRs, and quality of hospital care. We used data on patients consecutively admitted to the participating centers during 4 months of surveys carried out in 1993, 1995, 1997, and 1998. Methods of the GIFA have been previously described.4,13 Briefly, after obtaining a written informed consent, a study physician with specific training completed a questionnaire for each patient at admission to hospital and updated it daily. Data recorded included sociodemographic characteristics, medical variables, complete blood count, and neuropsychological and physical function variables.
Overall, 17 186 patients were enrolled in the survey period. We excluded patients who died during hospital stay (n = 878), those for whom the information needed for the estimation of the GFR was not available (n = 4472), and those admitted to wards other than geriatrics or internal medicine (n = 149). After exclusion of these 5499 patients, the final sample comprised 11 687 patients.
The outcome of the study was the occurrence of ADRs during the hospital stay. Adverse drug reactions were classified according to the World Health Organization.14 The attending physician judged if the patient had an event, and the strength of the association between the event and the suspected drug was assessed by the algorithm of Naranjo et al.15 Analytical variables for the occurrence of any ADR, ADR to hydrosoluble drugs, and ADR to any other drug during the hospital stay were created.16
The GFR was computed using the Modification of Diet and Renal Disease (MDRD) Study formula17:
170 × [Serum Creatinine]–0.999 × [Age]–0.176 × [Serum Urea Nitrogen]–0.170 × [Serum Albumin]0.318.
For women, the result was multiplied by 0.762.
We grouped the participants into 3 categories according to their creatinine levels and estimated GFRs. We considered as normal a creatinine level up to 1.2 mg/dL (106 μmol/L)18 and an estimated GFR of 60 mL/min for a body surface area of 1.73 m2.19 We considered 3 conditions: normal renal function (serum creatinine levels within the normal range and normal estimated GFRs), concealed renal insufficiency (serum creatinine levels within the normal range but reduced estimated GFRs), or overt renal insufficiency (creatinine levels over the normal range and reduced estimated GFRs).
Variables specifically considered in this study were age, sex, type of ward, and length of hospital stay. Nutritional status was assessed by measuring body mass index (BMI) (calculated as weight in kilograms divided by the square of height in meters) and serum albumin levels. To investigate the effect of malnutrition on emerging associations, we considered BMI values lower than 20 and serum albumin levels lower than 3.5 g/dL as potential confounders. Functional capabilities were rated by activities of daily living (ADL) scale. We categorized the functional status as independent (no need of assistance in any ADL), or dependent (needs assistance in at least 1 ADL). Cognitive status was assessed by the Hodkinson Abbreviated Mental Test, and we considered patients with a score of 7 or less as having cognitive impairment.20 The number of clinical problems was calculated and used as an index of comorbidity. Drugs were coded by Anatomical and Therapeutical Classification,21 and the number of drugs was also calculated. Procedures conformed to guidelines provided by the Catholic University of Rome Ethical Committee.
We used contingency tables to compare the demographic and clinical characteristics of the 3 groups. The association between ADRs and the variables of interest was evaluated by calculating the odds ratios (ORs) and the corresponding 95% confidence intervals (CIs), adjusting for age and sex. We then built a multivariable logistic regression model to obtain a deconfounded estimate of the association between concealed or overt renal insufficiency and ADRs during hospital stay, adjusting for age, sex, and variables proven to be associated with the outcomes in preliminary analysis. The presence of collinearity between variables entered in the final regression model was excluded by assessing the effect of individual variables on the standard error of regression coefficients, as reported by Hosmer and Lemeshow.22 All statistical analyses were performed using SPSS software (version 10.0; SPSS Inc, Chicago, Ill).
The main characteristics of patients grouped according to their renal function are reported in Table 1. Compared with patients having normal renal function, those with concealed or overt renal insufficiency were older and more frequently male, had a poorer physical and mental status, had more active diagnoses and prescribed drugs, and were more frequently hypoalbuminemic and admitted to a geriatric ward. Prevalence of congestive heart failure and hypertension was lower and that of malnutrition higher in the concealed than in the overt renal insufficiency group. Male sex was also significantly associated with concealed renal insufficiency.
Patients excluded from the analysis because of missing estimated GFR were characterized by older age (age ≥80: 36.2% vs 30.4%; P<.001), higher percentage of women (53.1% vs 47.9%; P<.001), and shorter length of hospital stay (>14 days: 23.2% vs 34.3%: P<.001). The 2 groups had a comparable prevalence of both functional (14.9% and 13.7%, respectively) and cognitive impairment (25.4% and 25.5%, respectively).
We found 941 patients (8.0% of all patients considered in the analysis) with any ADR during hospital stay. Older age and female sex were more frequent in patients with any ADR, as were hypoalbuminemia, comorbidity, polypharmacy, and a length of stay longer than 14 days. Physical and mental performance was similar in the 2 groups, as was the fraction of patients admitted to a geriatric ward. The prevalence of overt, but not concealed, renal insufficiency was significantly higher in patients with any ADR (Table 2).
We detected 301 ADRs to hydrosoluble drugs and 640 ADRs to any other drug. The hydrosoluble drugs most commonly responsible for ADRs were diuretics (17.9%), digitalis (16.3%), angiotensin-converting enzyme inhibitors (15.3%), and hypoglycemic agents (13.9%). The inherent 4 most common ADRs were hypoglycemia caused by hypoglycemic drugs, hypotension caused by angiotensin-converting enzyme inhibitors or diuretics, bradycardia caused by digitalis, and hypokalemia caused by diuretics. Among all the remaining drugs, nitrates (13.7%), calcium channel blockers (9.1%), antithrombotics (8.0%), and psycholeptics (7.6%) were the most frequent cause of ADRs, which were mainly gastrointestinal signs or symptoms caused by antithrombotics, headache and hypotension caused by nitrates or calcium channel blockers, and sleepiness caused by psycholeptics.
Preliminary analysis showed that ADRs to both hydrosoluble and nonhydrosoluble drugs were associated with hypoalbuminemia, comorbidity, polypharmacy, and length of stay, while older age was associated with ADRs with hydrosoluble drugs, and female sex with ADRs to nonhydrosoluble drugs. Both concealed and overt renal insufficiency were significantly more prevalent in patients with ADRs to hydrosoluble drugs (Table 3).
In the summary regression models, the occurrence of any ADR during hospital stay was significantly associated to comorbidity, polypharmacy, length of hospital stay, and overt renal insufficiency, while male sex was inversely associated to this outcome. Independent correlates of ADRs to hydrosoluble drugs were comorbidity, polypharmacy, length of hospital stay, and the presence of concealed or overt renal insufficiency. Finally, predictors of ADRs to nonhydrosoluble drugs were comorbidity, polypharmacy, and length of hospital stay (Table 4).
The main finding from this study is that concealed renal insufficiency is highly prevalent in elderly hospitalized patients and qualifies as an independent risk factor for ADRs to hydrosoluble but not to the remaining drugs. Impaired clearance of hydrosoluble drugs secondary to defective renal function likely explains this finding. The lack of association between overt renal insufficiency and the risk of ADRs to liposoluble drugs confirms that the positive association between concealed renal insufficiency and ADRs to hydrosoluble drugs was mediated by a pharmacokinetic mechanism rather than by a generic condition of frailty for which concealed renal insufficiency seems to be a marker.
The most important implication of our results is that assessing renal function by estimating GFR may help to avoid drug-related events by adjusting the dose regimens and reducing the number of prescribed drugs in patients with concealed or overt renal dysfunction. It is important to note that our study underestimates the impact of ADRs in elderly patients because we only consider events occurring during hospital stay. Actually, ADRs occur more commonly in home-dwelling patients and are an important cause of hospital admission3,23; therefore, paying more attention to concealed renal dysfunction in the outpatient setting may have an impact far greater than reducing ADRs in hospitalized patients. Furthermore, the estimated cost of drug-related morbidity and mortality in the ambulatory setting in the United States ranges from $30.1 to $136.8 billion, and the largest amount of this total cost is associated with drug-related hospitalizations.8 Thus, using estimated GFR to identify patients at risk may result in relevant savings with a negligible added cost.
Patients having concealed renal insufficiency were older, more cognitively impaired, and physically dependent compared with those having normal renal function. Furthermore, they were characterized by a higher prevalence of caloric-protein malnutrition, as expressed by serum albumin level, and a greater number of used drugs. On the other hand, patients with overt and concealed renal insufficiency had fairly comparable characteristics except for a greater percentage of women and prevalence of malnutrition among the latter. This finding suggests that a lower fat-free mass in older women24 may have contributed to conceal renal insufficiency by blunting the rise of creatinine levels.
The finding of a greater prevalence of hypoalbuminemia in patients with concealed renal insufficiency was expected because a strong association between hypoalbuminemia and low GFR has been recently reported.25 This finding emphasizes the need for a comprehensive approach to nutritional status in older patients instead of one based only on anthropometric indexes. Furthermore, chronic conditions predisposing to proteinuria (such as hypertension, diabetes, and heart failure) are highly prevalent in our study, and hypoalbuminemia is an independent predictor of the progression of diabetic nephropathy to end-stage renal disease in patients with type 2 diabetes mellitus.26 Thus, our findings add to the knowledge of its role as a correlate of early renal dysfunction in elderly patients. Finally, even if albumin is one of the parameters considered in the MDRD formula, adjusting for hypoalbuminemia did not modify the strength of the association between concealed renal insufficiency and ADRs to hydrosoluble drugs in the multivariable summary regression model.
On average, frailty (expressed by reduced physical and cognitive performance and comorbidity) was a clear-cut correlate of concealed renal insufficiency and, thus, should be considered a clue to detect this harmful condition. Age did not emerge as a risk factor for ADRs in this and in some of previous studies on hospitalized patients.4,5 The different result of a Norwegian study might depend on the outcome used, which was fatal ADRs instead of ADRs, making the prognostic role of age evident.6 Any difference in the prevalence of patients with diabetes is a further source of heterogeneity among studies because type 2 diabetes mellitus is strictly associated with renal insufficiency even in the absence of albuminuria and abnormal serum creatinine levels.27
Polypharmacy was the main risk factor for ADRs to both hydrosoluble and nonhydrosoluble drugs. This finding is in agreement with previous studies and further stresses the need to prescribe the least necessary number of drugs to the elderly.9 The crude number of drugs prescribed provides a convenient, yet far from optimal, measure of the risk of ADRs. The dose might be a determinant of the risk: drugs cleared by the kidney and most of those cleared by the liver should be prescribed at a dose 30% lower than the adult dose.28 In this perspective, the number of drugs taken is a generic indicator of risk because the greater the number of drugs used, the higher the possibility of overdosing. Pharmacokinetic drug-drug interaction is a further mechanism likely explaining the link between ADRs and polypharmacy irrespective of the patient’s age.29 Thus, the impact of polypharmacy on the risk of ADRs might be greater than that observed in both the present and previous studies, provided that more reliable methods of categorizing polypharmacy were available.
Both a length of stay longer than 14 days and a number of diagnoses greater than 4 qualified as risk factors for any ADR and for ADR to both hydrosoluble and nonhydrosoluble drugs. While both of these variables are obviously related to polypharmacy, the multivariable analysis suggests that they really played an independent role as risk factors. Furthermore, the same finding has been previously reported.1,4,6 Thus, it can be argued that the cumulative interaction between patient and hospital, as mediated by the length of stay and the burden of diseases and inherent care, is the main risk factor for ADRs.
Women have been reported to be more exposed to the risk of ADRs.3,30 In the present study, this was true for all ADRs and not for ADRs to hydrosoluble or lipophilic drugs, but this difference likely reflects the smaller sizes of the 2 subgroups. Alternatively, women might be more prone to ADRs to lipophilic drugs because both phase I and phase II liver metabolic processes are slower in women, with only few exceptions to this general rule.31,32
The present results should be interpreted with some caution because we did not measure directly the GFR. However, the MDRD formula has been proved to predict the GFR with a high degree of accuracy, while the creatinine clearance estimated by the Cockroft-Gault formula overestimated the measured GFR by 16% in the MDRD study.17
We separated hydrosoluble drugs from all the other drugs: the latter category includes a complex mix of drugs with a wide spectrum of lipophilia. The liver and kidney are therefore variously involved in the kinetics of a consistent proportion of the drugs in this group. Accordingly, the present results provide a reliable estimate of the relationship between concealed renal insufficiency and ADRs to hydrosoluble drugs, but they do not rule out some causal role of concealed renal insufficiency also toward ADRs to selected nonhydrosoluble drugs.
We did not consider the severity of the ADRs in the present study, and this limitation may partly explain the different impact of age on ADRs with respect to other studies.6 Finally, we tried to interpret the relationship between concealed renal insufficiency and ADRs on purely pharmacokinetic bases. Theoretically, defective renal function can also affect the pharmacodynamics of selected drugs, such as by decreasing receptor sensitivity through its positive effect on serum drug levels.
The present study shows that a high proportion of older hospitalized patients have impaired renal function despite a normal serum creatinine level and are exposed to a risk of ADRs to water-soluble drugs, which is comparable to that observed in patients with overt renal insufficiency. Given that ADRs represent a dramatic burden for older populations, any attempt should be made to prevent them. This objective can be pursued not only by limiting the number of prescribed drugs but also by using the minimal effective doses. Recognizing patients with impaired renal function is an obvious condition for optimal dosing. Thus, GFR should be indirectly obtained in older people before prescribing or reevaluating drug therapy. This procedure seems highly recommendable even in the absence of known risk factors for renal insufficiency such as diabetes. Further research is needed to verify whether such a strategy can really decrease the incidence of ADRs in the elderly.
Correspondence: Andrea Corsonello, MD, Via D. Frugiuele, 39, I-87100 Cosenza, Italy (firstname.lastname@example.org).
Accepted for Publication: August 20, 2004.
Financial Disclosure: None.
Funding/Support: The GIFA is partially supported by a grant 94000402 from the Italian National Research Council, Rome, Italy.
Additional Information: The GIFA is a research group of the Italian Society of Gerontology and Geriatrics (SIGG)–Fondazione Italiana per la Ricerca sull’Invecchiamento (FIRI-ONLUS).
Group Information: A complete list of the GIFA investigators has been published previously (Eur J Epidemiol. 1999;15:893-901).