Original Articles: Original Articles

Beta-Blockers, Trimethoprim-Sulfamethoxazole, and the Risk of Hyperkalemia Requiring Hospitalization in the Elderly

A Nested Case-Control Study

Weir, Matthew A.^*,†; Juurlink, David N.^‡,§,‖; Gomes, Tara^‡; Mamdani, Muhammad^{‡,§,‖,¶}; Hackam, Daniel G.^†,**; Jain, Arsh K.^*,†; Garg, Amit X.^*,†,‡

Author Information

^*Division of Nephrology, Department of Medicine, and

^†Department of Epidemiology and Biostatistics, University of Western Ontario, London, Canada;

^‡The Institute for Clinical Evaluative Sciences, Toronto, Ontario, Canada;

^§Department of Medicine, University of Toronto, Toronto, Ontario, Canada;

^‖Department of Health Policy, Management, and Evaluation, University of Toronto, Toronto, Ontario, Canada;

^¶Applied Health Research Center, Li Ka Shing Knowledge Institute of St Michael's Hospital, Toronto, Ontario, Canada; and

^**Division of Clinical Pharmacology, Department of Medicine, University of Western Ontario, London, Ontario, Canada

Correspondence: Dr. Amit X. Garg, University of Western Ontario, Room ELL-101, 800 Commissioner's Road East, London, Ontario, Canada, N6A 5W9. Phone: 519-685-5802; Fax: 519-685-8269; E-mail: [email protected]

Received March 3, 2010

Accepted May 3, 2010

Clinical Journal of the American Society of Nephrology 5(9):p 1544-1551, September 2010. | DOI: 10.2215/CJN.01970310

Free

Abstract

Adverse drug reactions among the elderly are common. It is estimated that 50 serious adverse drug reactions occur for every 1000 patient-years (¹). Drug-induced hyperkalemia is of particular concern because of its association with commonly used medications and its significant potential for harm, including sudden death (²^–⁵). Common cardiovascular agents such as angiotensin converting enzyme inhibitors, angiotensin receptor blockers, and potassium-sparing diuretics all impart a risk of hyperkalemia (²^,⁶^–¹⁰). Many studies have suggested an augmented risk of hyperkalemia associated with simultaneous use of two or more of these agents (¹¹^–¹⁵). However, there may also be significant risks associated with other combinations of common, hyperkalemia-inducing medications such as beta adrenergic receptor blockers (β-blockers) and trimethoprim-sulfamethoxazole antibiotics (TMP-SMX).

β-Blocker prescriptions have steadily increased over the last 10 years (¹⁶). Their role in hyperkalemia through inhibition of cellular adrenergic receptor-dependent potassium translocation has been extensively studied (¹⁷^–¹⁹). TMP-SMX is also in common use, representing 30% of all antibiotics prescribed for urinary tract infections (²⁰^,²¹). Trimethoprim is structurally related to the potassium-sparing diuretic amiloride and has been shown to block sodium channels in the distal nephron thereby limiting the electrochemical gradient driving potassium elimination (²²^–²⁷).

Given their popularity, β-blockers and TMP-SMX antibiotics are frequently co-administered. We conducted a nested case-control study using health administrative data to explore the risk of hyperkalemia conferred by this combination of drugs. We hypothesized that concurrent use of β-blockers and TMP-SMX would pose a substantially greater risk of hyperkalemia requiring hospital admission than would the use of TMP-SMX alone.

Materials and Methods

Design

Using health administrative data from July 1, 1994 to March 31, 2008, we established two cohorts of outpatient residents of Ontario, age 66 years and older. One cohort included only those with evidence of continuous β-blocker use, and the other included only those without evidence of any β-blocker use. Within these cohorts, we conducted separate nested case-control studies comparing the risk of hyperkalemia posed by TMP-SMX to that of amoxicillin. To assess interaction between β-blockers and TMP-SMX, we compared the estimates of risk from the two cohorts. We acquired and analyzed exposure, outcome, and covariate data according to a predefined protocol. The study was approved by the institutional review board at Sunnybrook Health Sciences Center, Toronto, Canada. The reporting of this study follows the STROBE statement guidelines (²⁸).

Setting

Ontario is the most populous Canadian province, with approximately 12 million residents in the year 2008, of whom 1.6 million were older than 65 years (²⁹). All residents received universal access to hospital and physician services, and elderly residents received coverage for prescription medications. Coverage for medical services and medications from a single provincial payer provided a comprehensive set of health administrative data.

Sources of Data

We identified prescription drug use using the Ontario Drug Benefits (ODB) database. The ODB program provides residents of Ontario 65 years of age or older with coverage for most prescription medications. We identified all hospitalizations attributed to hyperkalemia based on ninth and tenth editions of the International Classification of Disease (ICD-9 and ICD-10) codes recorded in the Canadian Institute for Health Information Discharge Abstract Database (CIHI-DAD).We obtained covariate information from multiple databases including the ODB database, CIHI-DAD, and the Ontario Health Insurance Plan (OHIP) database. The OHIP database contains claims information on inpatient, outpatient, and laboratory services rendered to residents of Ontario. We obtained demographic data from the Registered Persons Database (RPD), which contains demographic information on all Ontarians ever issued a health card. Census data from Statistics Canada was linked to postal codes and used to determine neighborhood income quintile. These databases have been validated in previous epidemiologic studies (³⁰^–³⁴). Data were complete except for 5% of income estimates that we imputed as the mean.

Participants

Cohorts.

Drug coverage for each individual prescription in ODB is maximally 100 days duration; therefore, we established the cohort of continuous β-blocker users by including only those with repeated β-blocker prescriptions that were no more than 100 days apart. We established a cohort of nonusers by including only those without a β-blocker prescription in the ODB database.

Cases.

Within the cohorts, we identified hospitalizations for hyperkalemia within 14 days after filling a prescription for one of the five study antibiotics: TMP-SMX, ciprofloxacin, norfloxacin, nitrofurantoin, or amoxicillin. These antibiotics are typically prescribed for urinary tract infections, which is one of the most common reasons for outpatient antibiotic therapy (³⁵). We chose these antibiotics to help mitigate indication bias by restricting the range of possible infections underlying the prescriptions. We chose amoxicillin as the reference drug because it is one of the most commonly prescribed antibiotics and its use is not associated with hyperkalemia. We defined cases as hospitalizations attributed to hyperkalemia at the time of admission (ICD-9 or ICD-10 diagnosis codes 276.7 and E87.5, respectively). The date of hospital admission served the index date for all analyses. Among individuals with multiple episodes of hyperkalemia, we considered only the first hospitalization.

Controls.

For each case, we randomly selected up to four patients who had filled a prescription for a study antibiotic but who had not been hospitalized for hyperkalemia. Potential controls had the same index date as their corresponding case patient and were included only if a study antibiotic prescription was detected in the preceding 14 days. Controls were also matched to cases on age at the index date (±1 year), sex, history of chronic kidney disease, and history of diabetes mellitus.

In both cases and controls, we excluded patients with prescriptions for multiple study antibiotics or a single nonstudy antibiotic within the 14 days before the index date. We also excluded those with evidence of end-stage renal disease or renal transplantation in the 3 years preceding the index date. All patients were at least 66 years of age at the time of analysis to ensure the availability of at least 1 year of drug use records preceding the index date.

Statistical Methods

We assessed differences between the baseline characteristics of case and control groups and those who received various types of antibiotics using standardized differences (³⁶^,³⁷). We examined odds ratios (OR) for hyperkalemia comparing five groups of patients prescribed different study antibiotics. We used conditional logistic regression to estimate the OR and their 95% confidence intervals (CI). We adjusted for characteristics that may have predisposed to hyperkalemia in the multivariable model using all available data in the 3 years preceding the index date, including the Charlson comorbidity score (0, 1, ≥2) (³⁸^;³⁹), the number of distinct prescriptions in the preceding year (≤5, 6 to 10, 11 to 15, 16 to 20, 21 to 25, ≥26) (⁴⁰), socioeconomic status (quintiles 1 to 2, quintiles 3 to 5), congestive heart failure, coronary artery disease, and previous episodes of hyperkalemia. We adjusted for medications used within the 120-day interval before the index date that may have influenced serum potassium concentrations: Nonpotassium-sparing diuretics, potassium-sparing diuretics (⁷^,⁸), potassium supplements, nonsteroidal anti-inflammatory drugs, angiotensin converting enzyme inhibitors (⁹), and angiotensin receptor blockers (⁶). We assessed for a statistical interaction by comparing the OR from the two cohorts using the technique of Altman and Bland, the OR from the first cohort reflected the risk of hyperkalemia among users of both β-blockers and TMP-SMX, and the OR from the second cohort reflected the same risk among users of TMP-SMX alone (⁴¹). Two-tailed P values of < 0.05 were interpreted as statistically significant. We conducted all analyses using SAS 9.1.3 software (SAS Institute, Carey, NC).

Additional Analyses

Severity of Hospital Admissions.

For case patients, we determined the median duration of the hospital admission, whether there was an intensive care unit admission, and whether the patient died during the hospital admission.

Dose Response.

To identify a dose-response relationship, we repeated the primary analysis with the TMP-SMX exposure stratified into single-strength (400 mg/80 mg) and double-strength (800 mg/160 mg) dose categories. We expected a higher risk of hyperkalemia with a larger dose of TMP-SMX.

Time from Antibiotic Prescription.

We altered the time from antibiotic prescription and index date first to 7 days and then 21 days rather than 14 days as used in the primary analysis.

β-1 Receptor Selectivity.

We repeated the primary analysis and test for interaction differentiating between β-1-selective and nonselective β-blockers. Given the role of the β-2 adrenergic receptor in translocation of potassium, we expected to find a stronger relationship with hyperkalemia when TMP-SMX was used in the setting of nonselective β-blockers (¹⁷^,⁴²^,⁴³).

Serum Potassium Testing.

For the purposes of billing, the performing of serum potassium tests but not the test results are recorded in Ontario administrative databases. We examined the number of outpatient serum potassium tests performed between the dates of antibiotic prescription and the index date.

Absolute Risk of Hyperkalemia.

To obtain a crude estimate of the absolute risk of hospital admission for hyperkalemia, we ascertained the number of cohort patients who received at least one TMP-SMX prescription and the number of those who required admission for hyperkalemia within the following 21 days.

Absolute Risk of Death.

To obtain a crude estimate of the absolute risk of death, we ascertained the number of cohort patients who received at least one TMP-SMX prescription and the number of those who died within the following 21 days.

Results

Primary Analysis

During the 14-year accrual period, we identified a cohort of 299,749 β-blocker users. Of these, 31,186 filled at least one prescription for TMP-SMX, and 195 were admitted to hospital for hyperkalemia. We excluded six cases that could not be matched to a control. The final sample comprised 189 cases of hyperkalemia and 641 controls.

Characteristics according to hyperkalemia and antibiotic use are shown in Table 1 and Table 2, respectively. Cases had slightly lower incomes and greater comorbidity (Table 1). The distribution of baseline characteristics was relatively consistent across the five types of antibiotics (Table 2). As shown in Table 3, among patients receiving β-blockers, those receiving concomitant TMP-SMX were five times more likely to be hospitalized for hyperkalemia than patients using amoxicillin (adjusted OR, 5.1; 95% CI, 2.8 to 9.4). No significant associations were observed between the other study antibiotics and admissions for hyperkalemia.

T1-3 — Table 1:
Characteristics of β-blocker users admitted to hospital with hyperkalemia and matched controlsa

T2-3 — Table 2:
Characteristics of β-blocker users by antibiotic usea

T3-3 — Table 3:
Association between hospitalization for hyperkalemia and antibiotic use among β-blocker users

The cohort of non-β-blocker users had a distribution of baseline characteristics similar to that of the β-blocker cohort (tables available upon request). Among non-β-blocker users, we identified 1349 cases patients and 5378 controls (Table 4). The OR for hyperkalemia requiring hospital admission comparing TMP-SMX to amoxicillin was 5.8 (95% CI, 4.7 to 7.3). This was not significantly different from the risk associated with TMP-SMX use among β-blocker users (P value for interaction 0.65). This indicated the risk of hyperkalemia was attributable to TMP-SMX alone, rather than its combination with β-blockers.

T4-3 — Table 4:
Association between hospitalization for hyperkalemia and antibiotic use among non-β-blocker users