Acute kidney injury associated with trimethoprim/sulfamethoxazole

Abstract

Objectives

Trimethoprim/sulfamethoxazole effectively treats community-acquired soft tissue infections and urinary tract infections, both of which occur in patients with risk factors for renal impairment. We systematically studied the adverse renal effects of trimethoprim/sulfamethoxazole in a middle-aged veteran population.

Methods

We reviewed complete electronic records for all patients who, during a 3 year period, had received ≥6 days of treatment with trimethoprim/sulfamethoxazole and for whom a baseline and follow-up determination of serum creatinine and blood urea nitrogen (BUN) were available.

Results

Of 573 patients who met inclusion criteria, 64 (11.2%) had increases in both serum creatinine and BUN that met predetermined criteria for acute kidney injury (AKI): in 33 (5.8%), AKI was judged likely due to trimethoprim/sulfamethoxazole; in 28 (4.9%), possibly due to trimethoprim/sulfamethoxazole; and in 3 (0.52%), unrelated to trimethoprim/sulfamethoxazole. Five additional patients (0.9%) had elevations only in serum creatinine. In nearly all cases likely due to trimethoprim/sulfamethoxazole, AKI resolved promptly after discontinuation of therapy, but one patient required dialysis. Pyuria appeared in only 2 of 37 patients who had urinalyses; eosinophiluria was not observed. In a multivariate model, patients with hypertension and diabetes mellitus had increased risk for renal insufficiency, especially if these conditions were considered poorly controlled.

Conclusions

In a middle-aged male inpatient population treated for a minimum of 6 days, AKI is much more common with trimethoprim/sulfamethoxazole therapy than previously reported. Intrinsic renal impairment rather than interstitial nephritis or competition for creatinine clearance appears responsible for the great majority of cases, and neither an effect of dose nor duration was detected in a univariate analysis. Impairment is transient if therapy is discontinued.

nephrotoxicity, soft tissue infections, urinary tract infections

Introduction

The emergence of methicillin-resistant Staphylococcus aureus (MRSA) as a prominent cause of community-acquired infection of skin and soft tissues in the USA has increased reliance on treatment with trimethoprim/sulfamethoxazole, because nearly all MRSA isolates are susceptible to this combination.¹ Trimethoprim/sulfamethoxazole also remains the recommended treatment for uncomplicated urinary tract infection in the USA.² This drug combination has been identified as a cause of renal toxicity.^3–14 Unfortunately, many older adults who are at risk of skin and soft tissue or urinary infections and who therefore might be candidates for therapy with trimethoprim/sulfamethoxazole have comorbid conditions such as diabetes mellitus, hypertension or vascular disease that may compromise renal function. For these reasons, it is important to define the adverse effects of this drug combination on renal function.

Although case series have described elevations in serum creatinine^3–8 and case reports have cited interstitial nephritis,^9–11 hyperkalemia^12–14 and acute tubular necrosis¹⁵ with trimethoprim/sulfamethoxazole therapy, there has been no systematic, population-based study to determine the incidence, extent and severity of acute kidney injury (AKI) associated with this therapy. Post-marketing surveys of adverse experiences associated with trimethoprim/sulfamethoxazole and individual case reports^5,15–22 suggest that AKI is a very uncommon side effect. We now describe the incidence, extent and severity of AKI associated with trimethoprim/sulfamethoxazole in a largely middle-aged veteran population.

Methods

Patients

The Michael E. DeBakey Veterans Affairs Medical Center (MEDVAMC) provides medical care for ∼120 000 persons. MEDVAMC has fully automated electronic medical records that document every encounter between a patient and the healthcare system. The mean age of patients enrolled at MEDVAMC is 65 years. For the purposes of this study, we reviewed the records of consecutive inpatients for whom any dose of trimethoprim/sulfamethoxazole was prescribed during a 3 year period. Patients who received less than a 6 day course of treatment, those who did not have serum creatinine and blood urea nitrogen (BUN) determinations both within 7 days before starting and within 3 days after completing therapy, patients on haemodialysis and patients who began their antibiotic course as an outpatient were excluded. We focused on patients who were begun on treatment while they were hospitalized, because that maximized our chances of finding measurements of serum creatinine and BUN, but we did not exclude those who completed the course of trimethoprim/sulfamethoxazole as outpatients. This study was approved by the Institutional Review Board, Baylor College of Medicine and the Research and Development Committee, MEDVAMC.

Literature search

Using Medline and Google Scholar, we searched existing English-language literature from inception, including post-marketing studies, for references to renal insufficiency, renal failure, acute kidney injury, azotemia or acute tubular necrosis and trimethoprim or sulfamethoxazole singly or together. We also searched for reports of adverse effects of trimethoprim/sulfamethoxazole.

Data recorded

Demographic data that were recorded included age, race, gender, the presence and control of diabetes mellitus and hypertension, the presence of chronic kidney disease, and HIV infection. We noted the indication for dose and duration of trimethoprim/sulfamethoxazole therapy. Serum creatinine and BUN within 7 days before starting (‘baseline’) and within 3 days after completing therapy (‘end of treatment’) were recorded. For patients who had received multiple courses of trimethoprim/sulfamethoxazole, only the first course was included in this study.

Definitions

Well-controlled diabetes mellitus was defined as a haemoglobin A1c of <8% and well-controlled hypertension as three consecutive blood pressure measurements of ≤140/90 prior to admission.^23,24 Estimated glomerular filtration rate (EGFR) was calculated using the Modification of Diet in Renal Disease (MDRD) equation, and chronic kidney disease was staged according to National Kidney Foundation guidelines.²⁵ According to the Acute Kidney Injury Network (AKIN) criteria,²⁶ AKI for patients who did not have chronic kidney disease was defined as an absolute increase in the serum creatinine concentration of ≥0.3 mg/dL (26.4 μM) from baseline, a percentage increase in the serum creatinine concentration of ≥50% or oliguria (≤0.5 mL urine output per kilogram of body weight per hour for >6 h). For patients with chronic kidney disease, AKI was defined as an increase in serum creatinine of ≥50%.²⁷ Because trimethoprim/sulfamethoxazole has been associated with decreased tubular secretion of creatinine without compromise of glomerular filtration,^3,19 patients with an initially normal renal function also needed to have an increase in BUN of 10 mg/dL, and those with chronic kidney disease needed to exhibit an increase in BUN of ≥50% in order to be classified in the AKI group.

Kidney injury

For patients who developed AKI during trimethoprim/sulfamethoxazole treatment, we recorded the presence of white blood cells (WBCs), red blood cells (RBCs), protein, eosinophils, granular casts and crystals in urine at the time AKI was documented. We also recorded the number of days the patient had received trimethoprim/sulfamethoxazole prior to documentation of AKI and, in patients for whom the serum creatinine returned to baseline within 1month, the time to return to baseline after discontinuation of trimethoprim/sulfamethoxazole. The fully electronic medical record was reviewed for alternative explanations for AKI, including the presence and stability of conditions such as volume depletion, heart failure or cirrhosis, that might diminish renal perfusion, as well as the concurrent administration of diuretics or other potentially nephrotoxic medications. Based on these data, AKI was classified as follows according to the likelihood that it was attributable to trimethoprim/sulfamethoxazole. If patients had stable comorbid disease states and either were on no medications that might elevate creatinine or BUN, or were continued on such medications unchanged after discontinuation of this drug combination as the creatinine and BUN returned to baseline, AKI was attributed to trimethoprim/sulfamethoxazole. Patients in whom an unstable underlying disease state was present, a diuretic agent or a potentially nephrotoxic medication was recently added and/or the dosage was increased prior to the appearance of AKI were classified as having AKI possibly due to trimethoprim/sulfamethoxazole. Patients in whom another condition, such as hypotension, or a drug, such as amphotericin, was likely to cause AKI during the time they were taking trimethoprim/sulfamethoxazole were classified as having AKI unlikely to be caused by trimethoprim/sulfamethoxazole. Records of AKI patients were also studied for the presence of new onset rash or fever.

Comparator group

Because of the high prevalence of underlying comorbid conditions that might contribute to AKI in our population, and in order to provide reassurance that any observed increase in creatinine and BUN in our study population was due to trimethoprim/sulfamethoxazole rather than to infection in a susceptible host, we also studied patients who received a course of clindamycin for a skin or soft tissue infection. Medical records of consecutive patients who received clindamycin during the same 3year period were searched to identify patients who were prescribed 600 mg by mouth three to four times daily for skin or soft tissue infection and met the same criteria as set out above for the availability of baseline and post-therapy determinations of serum creatinine and BUN.

Statistical analysis

Descriptive analysis, including mean and standard deviation, was done for the continuous variables—age, baseline and post-treatment creatinine and BUN, and treatment duration. Means are presented (±SD). Student's t-test was used to compare the continuous variables in the AKI group with those in the non-AKI group. Additional bivariate characteristics of the AKI group, including the proportion of patients with a treatment duration >10 days, age ≥60 years, doses of more than two trimethoprim/sulfamethoxazole 800/160 tablets per day, indication for treatment, gender and race, were compared with individuals in the non-AKI group using the χ² test and Fisher's exact test when appropriate. In addition, the frequency of comorbid conditions, including the proportion of patients with diabetes mellitus, hypertension, chronic kidney disease and HIV infection, were compared in the AKI and non-AKI groups. Following bivariate analyses of categorical data, a multivariate model was hypothesized based on biologically plausible covariates and included variables having a P value <0.25 by univariate logistic regression.

Statistical significance was defined by a P value <0.05, and parameter estimates were calculated using maximum likelihood methods with SAS statistical software (version 9.2; SAS Institute, Cary, NC, USA).

Results

Patients and demographic considerations

During the 3 years under study, trimethoprim/sulfamethoxazole was prescribed for 1662 inpatients. Of these, 573 met inclusion criteria; the conditions for which trimethoprim/sulfamethoxazole was prescribed are shown in Table 1. Of the 1089 that did not meet inclusion criteria, 495 received less than a 6 day course, 581 did not have serum creatinine and BUN measurements at the required times, and 13 were on haemodialysis before treatment began (Figure 1).

Table 1.

Characteristics of patients with and without AKI

Category	AKI patients, N = 64	Non-AKI patients, N = 509	P value
>2 tablets/daya	4 (6.3)	24 (4.7)	0.59
>10 days of treatment	23 (35.9)	174 (34.2)	0.78
Age >60 years	36 (56.3)	263 (51.7)	0.49
Race
white	32 (50.0)	295 (58.0)	0.23
black	23 (35.9)	149 (29.3)	0.31
Hispanic	4 (6.3)	33 (6.5)	1.0
other/undocumented	5 (7.8)	32 (6.3)	0.59
Gender
male	64 (100)	493 (96.9)	0.24
female	0 (0)	16 (3.1)	—
Diabetes mellitus	40 (62.5)	169 (33.2)	<0.001
well controlled	26 (40.6)	131 (25.7)	0.01
poorly controlled	14 (21.9)	38 (7.5)	<0.001
Hypertension	49 (76.6)	256 (50.3)	<0.001
Hypertension and well-controlled diabetes	20 (31.2)	88 (17.3)	0.007
Hypertension and poorly controlled diabetes	11 (17.2)	27 (5.3)	<0.001
Chronic kidney disease	22 (34.4)	104 (20.4)	0.01
stage 1	1 (4.5)	0 (0)	—
stage 2	8 (36.4)	47 (45.2)	0.45
stage 3	13 (59.1)	54 (51.9)	0.54
stage 4	0 (0)	2 (1.9)	—
stage 5	0 (0)	1 (1.0)	—
HIV	6 (9.4)	58 (11.4)	0.63
CD4 >200 cells/mm3	2 (3.1)	15 (2.9)	1
CD4 <200 cells/mm3	4 (6.3)	43 (8.4)	0.81
Indication
bacteraemia	0 (0)	9 (1.8)	0.28
osteomyelitis	3 (4.7)	6 (1.2)	0.07
PCP prophylaxis	4 (6.3)	57 (11.2)	0.23
PCP pneumonia	2 (3.1)	13 (2.6)	0.79
pneumonia	4 (6.3)	33 (6.5)	0.94
skin and soft tissue	28 (43.8)	183 (36.0)	0.22
urinary tract infection	16 (25)	139 (27.3)	0.70
miscellaneous/undocumented	7 (10.9)	69 (13.6)	0.56

Data are shown as the number of patients (% of total patients in that column).

^aTablet = 800 mg of sulfamethoxazole and 160 mg of trimethoprim.

Figure 1.

Flow chart showing the process by which the authors ascertained the patients under study.

AKI and the role of trimethoprim/sulfamethoxazole

Of the 573 patients who met the inclusion criteria, 64 (11.2%) had increases in both serum creatinine and BUN that met predetermined criteria for AKI. Five (0.9%) had only an elevation in serum creatinine and 504 (88.0%) had no change in creatinine or BUN that met AKI criteria. Thirty-three of the 64 (5.8%) patients who developed AKI had no recognized contributing factors, and their AKI was classified as likely to be caused by trimethoprim/sulfamethoxazole (Table 2). Twenty-eight (4.9%) were found to have one or more alternative explanations for AKI and were classified as having AKI possibly caused by trimethoprim/sulfamethoxazole. Of these, 8 were taking a new diuretic, an increased dosage of an existing diuretic or another potentially nephrotoxic medication, 3 had a comorbid condition such as uncontrolled congestive heart failure, 3 had pre-renal causes of AKI, 4 had received intravenous contrast agents in the few days prior to developing AKI and 10 had more than one alternative explanation for AKI. Three patients had AKI that was regarded as unlikely to be due to trimethoprim/sulfamethoxazole: in one case rhabdomyolysis preceded trimethoprim/sulfamethoxazole therapy, and in two cases amphotericin was given simultaneously with trimethoprim/sulfamethoxazole. No patient with AKI had a concurrent fever or rash.

Table 2.

Classification of AKI based on likelihood of causation by trimethoprim/sulfamethoxazole

Classification of AKI	Number of patients
Likely to be caused by trimethoprim/sulfamethoxazole	33
Possibly caused by trimethoprim/sulfamethoxazole	28
medications	8
comorbidities	3
intravenous contrast	4
pre-renal	3
post-renal	0
multiple causes	10
Unlikely to be caused by trimethoprim/sulfamethoxazole	3

Degree of renal impairment

Post-treatment creatinine levels in the AKI and non-AKI group were 2.1 ± 0.8 and 1.1 ± 0.5 mg/dL, respectively (P < 0.001), and BUN levels in the AKI and non-AKI groups were 33.7 ± 14.0 and 17.5 ± 11.3 mg/dL, respectively (P = 0.016). The greatest increase observed in serum creatinine was from 1.0 to 6.5 mg/dL, and this was in the single patient who required dialysis.

Time to onset of AKI

AKI was first documented as early as the third day of therapy and as late as 2 days after completion of therapy. Of the nine cases in which AKI developed on day 3, five were likely, three were possibly and one was unlikely to be attributable to trimethoprim/sulfamethoxazole. The mean time to documentation was 7.9 days (median 6.5 days). In 56 of 64 patients (87.5%) who were judged to have AKI, serum creatinine and BUN were recorded within 1 month after the discontinuation of trimethoprim/sulfamethoxazole; in these, renal function had returned to baseline in 52 (92.9%), and this return was detected in a mean of 9.1 ± 9.4 days after discontinuation of the medication (median 5.5 days).

Interstitial nephritis

Thirty-seven patients had a urinalysis in the week following the development of AKI (time to urinalysis, median of 2 days after AKI and 10 days after initiation of trimethoprim/sulfamethoxazole). Of these, 20 had <5 WBCs or RBCs (Table 3). Twelve patients (32.4%) had ≥5 WBCs per high-power field (HPF), and 11 (29.7%) had ≥5 RBCs per HPF at the end of the treatment period. Of the 12 patients with pyuria, 5 had had elevated urine WBCs prior to treatment with trimethoprim/sulfamethoxazole and 5 had no baseline urinalysis, leaving only 2 patients in whom pyuria appeared in association with the drug combination. One of these two patients also had peripheral eosinophilia and was thought to have interstitial nephritis. Of the 11 patients with microscopic haematuria, 4 had RBCs in their urine prior to treatment and 5 had no baseline urinalysis. No RBC casts, WBC casts, coarse granular casts or renal epithelial cells were found in any of the 37 patients. Urine eosinophils were specifically sought in 18 patients and detected in none (0%). Crystalluria was documented in four patients (10.8%): one patient had many uric acid crystals, one had rare uric acid crystals and two had rare calcium oxalate crystals, but no sulpha crystals were reported in any sample.

Table 3.

Urinalysis findings after documentation of AKI

Study findings	Number of patients
≥5 WBCs/HPF	12 (32.4)
≥5 RBCs/HPF	11 (29.7)
Urine pH ≥6	14 (37.8)
Urine crystals	4 (10.8)
Calcium oxalate	2 (5.4)
Uric acid	2 (5.4)
Sulpha	0 (0)
RBC casts	0 (0)
WBC casts	0 (0)
Coarse granular casts	0 (0)
Renal epithelial cells	0 (0)
Urine eosinophilsa	0 (0)

Data are shown as the number (%) of abnormal findings in 37 patients for whom a urinalysis was obtained.

^aUrine eosinophils were obtained in 18 patients.

Relation to age, duration of treatment and pre-morbid conditions

The mean ages of patients in the AKI and non-AKI groups were 63.8 ± 10.3 and 62.4 ± 12.8, respectively (P = 0.035). The duration of treatment was greater in the non-AKI group than in the AKI group (11.3 ± 8.8 versus 10.8 ± 5.8 days, respectively, P < 0.01). Baseline creatinine and BUN were significantly higher in AKI patients than in non-AKI patients (1.2 ± 0.4 and 18.3 ± 9.8 mg/dL, respectively, compared with 1.1 ± 0.5 and 17.6 ± 13.4 mg/dL, respectively, P < 0.01 for each comparison).

Although the incidence of AKI appeared to be lower in patients aged 50–59 years than in patients ≥60 years, when we used a regression equation to analyse the tendency to develop AKI by decade of age there was no significant association of AKI with age (R = 0.24). Of the 64 patients who developed AKI during trimethoprim/sulfamethoxazole therapy, 40 (62.5%) were diabetic compared with 169 of 509 (33.2%) of those who did not have AKI (Table 1, P < 0.001). Patients with poorly controlled diabetes were three times more likely than well-controlled diabetics to have AKI (P < 0.001). Twenty patients (31.2%) who developed AKI had both hypertension and well-controlled diabetes, compared with 88 (17.3%) who did not (P = 0.007). Eleven patients (17.2%) in the AKI group had both hypertension and poorly controlled diabetes, compared with 27 (5.3%) patients who did not develop AKI (P < 0.001). Patients with hypertension without regard to control (P < 0.001) and those with chronic kidney disease (P = 0.01) were also significantly more likely to have AKI (Table 1).

Relation to infection for which trimethoprim/sulfamethoxazole was given

Two-hundred-and-eleven patients were treated with trimethoprim/sulfamethoxazole for skin or soft tissue infection, of whom 28 (13.3%) developed AKI. This compared with 16 of 155 (10.3%) for whom trimethoprim/sulfamethoxazole was prescribed for a urinary tract infection, indicating that infections of the urinary tract were unlikely to account for the observed renal insufficiency.

Multivariate logistic regression model

A multivariate model was created to determine the best predictors of AKI in our study population when comparing AKI with non-AKI patients, given the presence of other covariates (see Table 1). In our final model, only patients with diabetes mellitus (P < 0.01) and hypertension (P = 0.01) were significant predictors for AKI when other variables were controlled for.

Comparator group

During the 3 years under study, 2360 patients received clindamycin; of these, 84 met our criteria for drug dosage, indication and availability of serum creatinine and BUN before and after therapy. There were no statistically significant differences in age or demographics when these patients were compared with 211 patients who received trimethoprim/sulfamethoxazole for skin or soft tissue infection (data not shown). Of these 84 patients, none (0%) developed AKI, and 3 had an increase in serum creatinine without a corresponding increase in BUN (P < 0.001).

Discussion

To our knowledge, the present study is the first systematic evaluation of alterations in renal function during therapy with trimethoprim/sulfamethoxazole in hospitalized patients. In a largely middle-aged male, hospitalized veteran population, 11.2% of patients developed AKI during or immediately following trimethoprim/sulfamethoxazole therapy.

Three earlier post-marketing studies reported an extremely low rate of nephrotoxicity (<0.01%).^18–20 In two, nurses used self-coding sheets to record information on consecutive hospitalized patients on trimethoprim/sulfamethoxazole therapy, but the studies did not appear to have included routine monitoring of renal function. The third evaluated outpatients treated with trimethoprim/sulfamethoxazole and recorded side effects only in those patients who sought evaluation from a healthcare provider during or shortly after therapy; renal function tests were not routinely performed. In contrast, in a closely monitored small group of subjects, Trollfors et al.⁵ showed that glomerular filtration rate decreased in 4 of 25 patients (16%), 2 of whom had received more than the recommended dosage.

In slightly more than half of our cases, no other factors contributing to AKI were identified, and the trimethoprim/sulfamethoxazole was felt to be responsible. Other possible contributing factors were identified in the remainder of cases, including exacerbation of a pre-morbid condition, volume depletion, diuresis, injection of contrast dyes and administration of potentially nephrotoxic drugs. In these cases, the role of trimethoprim/sulfamethoxazole was uncertain, and we regarded their AKI as possibly due to trimethoprim/sulfamethoxazole. Patients who had other conditions that are highly associated with AKI, such as rhabdomyolysis or amphotericin therapy were regarded as having AKI unrelated to trimethoprim/sulfamethoxazole.

In a univariate analysis, patients with diabetes mellitus were at increased risk, and those with poorly controlled diabetes were three times more likely than those with well-controlled diabetes to develop AKI; hypertension without regard to control and chronic kidney disease were also significant risk factors. In a multivariate analysis, only hypertension and diabetes were significantly associated with AKI. The rate at which AKI developed was greater among patients who were treated for skin or soft tissue infection than those treated for urinary tract infection. Of note, no AKI was documented in a comparator group of 84 patients who received oral clindamycin for skin or soft tissue infection.

Our results did not show a relationship between the dose of trimethoprim/sulfamethoxazole and the likelihood of developing AKI, but most patients in our series received 800 mg sulfamethoxazole and 160 mg trimethoprim (one ‘double strength’ tablet) twice daily. Higher doses were prescribed for patients with Pneumocystis jirovecii (PCP) pneumonia, and this group, nearly all of whom had AIDS, was younger than non-HIV-infected patients and had lower rates of diabetes and hypertension. The duration of treatment was actually slightly shorter among patients with AKI than those who did not have this complication, perhaps reflecting early discontinuation of the medication because of deteriorating renal function.

Acute interstitial nephritis is said to be the usual mechanism of renal toxicity associated with trimethoprim/sulfamethoxazole,²⁸ but in our study only one case of acute interstitial nephritis was documented. Although only 37 subjects had urinalyses, haematuria and/or pyuria were newly detected in only 4 patients. In 18 cases, eosinophiluria was sought, and was found in none. No patients had new-onset rash or fever. Trimethoprim/sulfamethoxazole may also form crystals in the urine of volume-depleted patients, resulting in AKI, and crystalluria may be seen in 0.4%–49% of these patients.²⁹ Such crystals were not detected in our study. Thus, based on all these data, an adverse effect on intrinsic renal function appears to be a better explanation for the observed renal insufficiency. Several case reports have demonstrated acute tubular necrosis (ATN) on renal biopsy following trimethoprim/sulfamethoxazole therapy.^15,16 Acute tubular necrosis appeared not to be a significant contributor to AKI in our study, as no coarse granular casts or renal epithelial cells were seen on examination of the urine sediment in any case. Trimethoprim/sulfamethoxazole may interfere with renal excretion of creatinine;^3–8 we observed five patients in whom the serum creatinine increased without a corresponding increase in BUN, and these elevations would be explained by this interference. However, our study shows that the BUN increases concomitantly with creatinine in nearly every case, suggesting that renal insufficiency is a more common cause of increasing creatinine in patients taking trimethoprim/sulfamethoxazole than competition with creatinine clearance.

There are several limitations to the present study. First is its retrospective nature. We tried to reduce the impact of this limitation by including only those patients who had baseline determinations of serum BUN and creatinine. There may be a selection bias toward detecting renal insufficiency, since patients whose renal function was never open to question, either due to a lack of risk factors at baseline or the failure to develop symptoms of azotemia, may not have had a baseline BUN and creatinine and would not have met inclusion criteria. Further, restricting our inclusion criteria to patients who were hospitalized at the time trimethoprim/sulfamethoxazole was begun may have led to a falsely elevated incidence of AKI, as hospitalized patients are more likely than outpatients to have comorbidities that predispose them to AKI; however, a clindamycin comparator group, in which patients had a similar incidence of comorbidities as the study group, did not have any cases of AKI. In the classification of AKI based on the likelihood that it was attributable to trimethoprim/sulfamethoxazole, we used clinical judgment, but any misclassification of these groups may affect study findings. Although we extensively reviewed the medical record to determine risk factors for AKI, it is possible that there were confounders, which we did not recognize. In addition, a urinalysis was done during the trimethoprim/sulfamethoxazole treatment period in only 57.8% of AKI patients, so the incidence of acute interstitial nephritis, crystal-induced nephropathy and ATN could be underreported.

We chose to include patients treated with clindamycin for skin and soft tissue infections as the comparator group because skin and soft tissue infections were the most common indication for trimethoprim/sulfamethoxazole in this study; a population treated for the same infection is most likely to have a similar incidence of comorbid conditions. We specifically chose clindamycin because in the USA clinicians often choose between trimethoprim/sulfamethoxazole and clindamycin for the treatment of skin and soft tissue infections, and, further, clindamycin has not been identified as a cause of renal insufficiency.

Because AKI occurred in our study in patients who had no specific new risk factors for renal impairment, trimethoprim/sulfamethoxazole should be considered an independent risk factor for AKI. AKI developed as early as day 3, but on average occurred on day 8 of therapy in our study; it thus seems reasonable to check creatinine and BUN after the first week of trimethoprim/sulfamethoxazole therapy. If creatinine and BUN rise acutely in a patient on trimethoprim/sulfamethoxazole, it is appropriate to discontinue the drug and investigate other potential causes of AKI. Few cases in our study were symptomatic, and only one patient required dialysis, which likely explains the low incidence of renal side effects reported in previous population-based studies.^19,20 However, AKI is an independent risk factor for mortality in hospitalized patients and therefore renal function should be monitored in patients at risk.^26,30 Since the deterioration of renal function observed in our patients appeared to be transient, if renal function does not return to baseline within 1 month, the source of the AKI is less likely to be trimethoprim/sulfamethoxazole.

Funding

This study was supported by a grant from the Infectious Diseases Society of America (to T. N. F.).

Transparency declarations

None to declare.