Research CMAJ

Research CMAJ Trimethoprim–sulfamethoxazole and risk of sudden death among patients taking spironolactone Tony Antoniou PhD, Simon Hollands MSc, Erin M. Macdonald MSc, Tara Gomes MHSc, Muhammad M. Mamdani PharmD MPH, David N. Juurlink MD PhD; for the Canadian Drug Safety and Effectiveness Research Network Competing interests: Abstract Muhammad Mamdani has Background: Trimethoprim–sulfamethoxazole Results: Of the 11 968 patients who died of served on advisory boards increases the risk of hyperkalemia when used sudden death while receiving spironolactone, for AstraZeneca, Bristol- with spironolactone. We examined whether we identified 328 whose death occurred Myers Squibb, Eli Lilly, GlaxoSmithKline, this drug combination is associated with an within 14 days after antibiotic exposure. Hoffman-La Roche, increased risk of sudden death, a consequence Compared with amoxicillin, trimethoprim– Novartis, Novo Nordisk of severe hyperkalemia. sulfamethoxazole was associated with a more and Pfizer. No other than twofold increase in the risk of sudden Methods: We conducted a population-based competing interests were death (adjusted OR 2.46, 95% confidence in- declared. nested case–control study involving Ontario resi- terval [CI] 1.55–3.90). Ciprofloxacin (adjusted dents aged 66 years or older who received spi- This article has been peer OR 1.55, 95% CI 1.02–2.38) and nitrofuran- ronolactone between Apr. 1, 1994, and Dec. 31, reviewed. toin (adjusted OR 1.70, 95% CI 1.03–2.79) 2011. Within this group, we identified cases as pa- were also associated with an increased risk Correspondence to: tients who died of sudden death within 14 days Tony Antoniou, of sudden death, although the risk with ni- after receiving a prescription for trimethoprim– [email protected] trofurantoin was not apparent in a sensitivity sulfamethoxazole or one of the other study anti- analysis. CMAJ 2015. DOI:10.1503 biotics (amoxicillin, ciprofloxacin, norfloxacin or /cmaj.140816 nitrofurantoin). For each case, we identified up to Interpretation: The antibiotic trimethoprim– 4 controls matched by age and sex. We deter- sulfamethoxazole was associated with an mined the odds ratio (OR) for the association be- increased risk of sudden death among older tween sudden death and exposure to each antibi- patients taking spironolactone. When clinically otic relative to amoxicillin, adjusted for predictors appropriate, alternative antibiotics should be of sudden death using a disease risk index. considered in these patients. he use of spironolactone increased con- azole, trimethoprim is most often used for the siderably following publication of the treatment of urinary tract infections. More than Randomized Aldactone Evaluation Study, 20 million prescriptions are written for the com- T 8 which showed that the drug improved morbidity bination each year in the United States. and mortality in carefully selected patients with We have previously shown that the use of severe systolic heart failure.1,2 Although spirono- trimethoprim–sulfamethoxazole in patients re- lactone is generally well tolerated, hyperkalemia ceiving spironolactone increased the risk of is a potentially life-threatening adverse effect of hospital admission with hyperkalemia more the drug in clinical practice.3–5 Strategies for miti- than 12-fold relative to amoxicillin.9 However, gating the risk of serious hyperkalemia include we did not examine whether the drug interac- cautious dosing of spironolactone, close monitor- tion was associated with an increased risk of ing of electrolyte levels and avoidance of other sudden cardiac death, a predictable conse- drugs that cause hyperkalemia. quence of severe hyperkalemia.10,11 This is im- The widely used antibiotic trimethoprim has portant because sudden death in patients taking pharmacologic similarities to the potassium-spar- spironolactone may erroneously be attributed to ing diuretic amiloride. It reduces urinary potas- intrinsic heart disease. sium excretion by about 40% and can increase Because treatment with trimethoprim– the risk of life-threatening hyperkalemia in sus- sulfamethoxazole can precipitate life-threatening ceptible individuals, including those taking spi- hyperkalemia in patients receiving spironolac- ronolactone.6,7 In combination with sulfamethox- tone, we conducted a study to determine whether E138 CMAJ, March 3, 2015, 187(4) ©2015 8872147 Canada Inc. or its licensors Research this drug interaction would be associated with an period 100 days from the date of the last pre- increased risk of sudden death. scription to identify outcomes that might have precipitated cessation of therapy. We used pre- Methods scription intervals of 100 days to define continuous use of spironolactone, because this is the Setting maximum prescription interval allowed by the We conducted a population-based, nested case– provincial drug benefit program. control study involving patients aged 66 years or Within the cohort of continuous users of older living in the province of Ontario who re- spironolactone, we defined cases as patients ceived spironolactone between Apr. 1, 1994, and who died of sudden death within 14 days after Dec. 31, 2011. These individuals have universal receiving a prescription for one of trimethoprim– access to physician services and hospital care sulfamethoxazole, ciprofloxacin, norfloxacin, and have provincial prescription drug coverage. nitrofurantoin or amoxicillin (excluding amoxicillin–clavulinic acid). The date of death served as Data sources the index date for all analyses. We excluded pa- We identified prescription drug records using the tients who received prescriptions for more than Ontario Drug Benefit Database, which contains one of the antibiotics of interest, or any other anti- comprehensive records of prescription drugs dis- biotic in the 30 days before the index date. We also pensed to residents in the province aged 65 years excluded patients admitted to hospital within or older. We obtained hospital admission data 30 days before the index date, individuals with a from the Canadian Institute for Health Informa- history of HIV infection or organ transplantation, tion’s Discharge Abstract Database, which con- and individuals who filled prescriptions for oral tains detailed clinical information on all hospital corticosteroids or immunomodulating drugs in the admissions in Ontario. We used the Ontario 100 days before the index date, to avoid the con- Health Insurance Plan database to identify claims founding effects of recent serious illness and im- for physician services, and we used validated dis- mune compromise. We identified sudden death us- ease registries to define the presence of diabetes ing a previously validated case definition, which mellitus, hypertension, HIV infection and conges- has a positive predictive value of 86.8%.19 tive heart failure.12–15 We obtained basic demo- From within the cohort of patients receiving graphic data from the Registered Persons Data- spironolactone, we selected up to 4 controls for base, a registry of all residents of the province each case matched by age at the index date who are eligible for health insurance. We deter- (within 1 yr) and sex. Controls were required to mined emergency department visits using the Na- be alive at the index date, and to have received tional Ambulatory Care Reporting System. We one of the study antibiotics within 14 days be- ascertained sudden deaths from the Ontario Office fore the index date. We excluded controls who of the Registrar General’s database, which con- received prescriptions for any other antibiotic in tains the cause of death reported on individual the 30 days before the index date. Consequently, death certificates. These datasets were linked with all cases and controls were older patients receiv- the use of unique, encoded identifiers and an- ing spironolactone who did not receive any anti- alyzed at the Institute for Clinical Evaluative Sci- biotic in the 30 days before the index date other ences, and are routinely used to study the conse- than one of the study antibiotics. When fewer quences of drug interactions.9,16–18 than 4 controls were available for each case, we included only those controls and maintained the Study population matching process. We excluded cases that could We used the Ontario Drug Benefit Database to not be matched to at least 1 control. identify patients who were prescribed spironolactone between Apr. 1, 1994, and Dec. 31, 2011. Statistical analysis For each patient, we defined a period of continu- We used standardized differences to compare ous spironolactone use beginning with the first baseline demographic and clinical characteristics prescription following their 66th birthday. We of cases and controls. Standardized differences excluded the first year of eligibility for prescrip- of less than 0.1 indicate good balance between tion drug coverage (age 65) to avoid incomplete the cases and controls for a given covariate.20 medication records. Observation ended with the We used conditional logistic regression to esti- first occurrence of death, the end of the study pe- mate the odds ratio (OR) and 95% confidence riod or the cessation of spironolactone treatment, intervals (CIs) for the association between sudden which was defined as a lapse of more than 100 death and receipt of a prescription for trime- days between successive prescriptions. In the event thoprim–sulfamethoxazole, with patients receiving of such a lapse, we extended the observation amoxicillin as the reference group. We selected CMAJ, March 3, 2015,

Load more