Adverse Events Lead to Drug Discontinuation More Commonly among Patients Who Receive Nafcillin than among Those Who Receive Oxacillin

Abstract

Nafcillin and oxacillin are used interchangeably in clinical practice, yet few studies have evaluated the safety of these two agents. Our objective was to compare the differential tolerabilities of nafcillin and oxacillin among hospitalized patients. We conducted a retrospective cohort study of all patients who received 12 g/day of nafcillin or oxacillin for at least 24 h. Two hundred twenty-four patients were included. Baseline characteristics and comorbidities were similar among patients receiving nafcillin (n = 160) and those receiving oxacillin (n = 64). Hypokalemia, defined as a potassium level of ≤3.3 mmol/liter or ≤2.9 mmol/liter or as a ≥0.5-mmol/liter decrease from the baseline level, occurred more frequently among patients who received nafcillin (51%, 20%, and 56%, respectively) than among those who received oxacillin (17%, 3%, and 34%, respectively; P < 0.0001, P = 0.0008, and P = 0.005, respectively). By multivariate logistic regression analysis, receipt of nafcillin was an independent predictor of severe hypokalemia (odds ratio [OR] = 6.74; 95% confidence interval [CI], 1.46 to 31.2; P = 0.02). Rates of hepatotoxicity did not differ between groups; however, acute kidney injury occurred more commonly with nafcillin than with oxacillin (18% versus 6%; P = 0.03). Overall, 18% of patients who received nafcillin discontinued therapy prematurely due to adverse events, compared to 2% of patients who received oxacillin (P = 0.0004). Nafcillin treatment is associated with higher rates of adverse events and treatment discontinuation than oxacillin among hospitalized adult patients. These findings have important implications for patients in both inpatient and outpatient settings, particularly patients who require long-term therapy and cannot be monitored routinely. Future randomized controlled studies evaluating the efficacy, costs, and tolerability of nafcillin versus oxacillin are warranted.

INTRODUCTION

Infections due to Staphylococcus aureus are a major public health concern and contribute to significant patient morbidity and mortality (1–3). Common infections include bacteremia, bone and joint infections, pneumonia, and infective endocarditis, which require long-term treatment. Vancomycin is recommended as empirical therapy for S. aureus given its reliable activity against methicillin-resistant S. aureus (MRSA) isolates (4). For methicillin-susceptible S. aureus (MSSA) isolates, however, nafcillin and oxacillin are preferred agents in the United States (5, 6). Indeed, treatment with nafcillin or oxacillin is associated with improved patient outcomes compared to those with vancomycin for MSSA infections (7–12). As such, both agents are commonly prescribed in inpatient and outpatient settings. Given their comparable acquisition costs, antimicrobial spectra, efficacies, and pharmacokinetics, nafcillin and oxacillin are used interchangeably (5, 6). Notably, however, few studies have directly compared the tolerabilities of the two agents in adult patients (13, 14).

A need for further discrimination of the safety of nafcillin and that of oxacillin has been motivated in recent years by the emergence of outpatient parenteral antimicrobial therapy (OPAT) programs. Patients receiving OPAT are monitored less frequently than hospitalized patients, necessitating the selection of the safest antimicrobial regimen prior to hospital discharge. In devising our institutional OPAT guidelines, we sought to further examine the tolerability of nafcillin given anecdotal reports of hypokalemia at our center. We hypothesized that oxacillin was better tolerated than nafcillin among hospitalized patients and may be more ideally suited for patients being discharged with OPAT. The objective of this study was to evaluate the differential tolerabilities of nafcillin and oxacillin administered in the hospital.

MATERIALS AND METHODS

We performed a retrospective cohort study of adult inpatients treated with 12 g per day of nafcillin or oxacillin at the University of Pittsburgh Medical Center (UPMC) from 1 April 2012 to 30 June 2013. Nafcillin or oxacillin therapy was selected based on physician preference; both agents were available in the hospital formulary throughout the study period. All patients who received treatment for at least 24 h were included in the analysis. We excluded patients with a baseline potassium level of ≤3.3 mmol/liter, those who received an antistaphylococcal penicillin within the preceding 6 months, those with baseline hepatic dysfunction (liver transaminase levels 3 times the upper limit of normal or documented cirrhosis), and those who required hemodialysis. Patients were monitored for 30 days after hospital discharge or discontinuation of therapy. Patient demographics, clinical characteristics, laboratory data, and outcomes were extracted from electronic medical records. The study was approved by the University of Pittsburgh institutional review board.

Hypokalemia and severe hypokalemia were defined as serum potassium levels of ≤3.3 and 2.9 mmol/liter, respectively (15). In separate analyses, we considered hypokalemia to be a ≥0.5-mmol/liter decrease from the baseline level. Acute kidney injury (AKI) was classified according to consensus guidelines as a 1.5-fold increase in serum creatinine levels from the baseline (16, 17). Acute hepatotoxicity was defined as total bilirubin, aspartate aminotransferase (AST), or alanine aminotransferase (ALT) levels higher than three times the upper limit of normal (>4.5 mg/dl, >123 IU/liter, and >162 IU/liter, respectively).

Continuous and categorical variables were compared with Mann-Whitney U and χ² (or Fisher's exact) tests, respectively. Using variables with a P value of <0.20 upon univariate analysis, we applied multivariable logistic regression methods to identify independent predictors of hypokalemia using a backward-selection procedure. Two-tailed P values of <0.05 were considered statistically significant.

RESULTS

Two hundred twenty-four patients were included; 64% (143/224) were male, and the median age was 56 years (range, 19 to 90 years). The most common indications for treatment were bacteremia (44%; 98/224), bone and joint infections (22%; 50/224), and endocarditis (21%; 46/224). Fifty percent (49/98) of bacteremia cases were complicated by secondary sites of infection. MSSA was the targeted pathogen in 91% (203/224) of cases. One hundred sixty and 64 patients received treatment with nafcillin and oxacillin, respectively. Charlson comorbidity scores, baseline serum creatinine levels, and receipt of concomitant medications were similar between groups (Table 1). Median baseline potassium levels were 4 and 4.1 mmol/liter among patients who received nafcillin and oxacillin, respectively. Patients were in the hospital for a median of 4 days prior to treatment initiation in both groups; however, nafcillin therapy was continued for a median of 1.5 days longer than oxacillin (P = 0.046) during the index admission.

TABLE 1.

Patient demographics, clinical characteristics, and concomitant medications

Parameter	Value for treatment group		P value
	Oxacillin (n = 64)	Nafcillin (n = 160)
No. (%) of male patients	40 (63)	103 (64)	0.88
Median age (yr) (range)	61 (22–88)	54 (19–90)	0.02
Median Charlson comorbidity score (range)	3 (0–10)	3 (0–10)	0.053
No. (%) of immunosuppressed patientsa	8 (13)	19 (12)	1.000
No. (%) of patients with indication for treatment of:
Bacteremiab	38 (59)	60 (38)	0.004
Bone and jointc	15 (23)	35 (22)	0.86
Endocarditis	7 (11)	39 (24)	0.03
Pneumonia	2 (3)	9 (6)	0.73
Skin/soft tissue infection	2 (3)	6 (4)	1.00
Ventriculitis	0 (0)	8 (5)	0.11
Intra-abdominal infection	0 (0)	3 (2)	0.56
No. (%) of patients with infection due to MSSA	57 (89)	146 (91)	0.62
Median baseline potassium level (mmol/liter) (range)	4.1 (3.4–5.2)	4 (3.4–5.2)	0.03
Median baseline serum creatinine level (mg/dl) (range)	0.8 (0.3–4.7)	0.9 (0.2–4.1)	0.16
Median no. of days in hospital prior to start of therapy (range)	4 (0–17)	4 (0–105)	0.04
No. (%) of patients taking concomitant medication
ACE inhibitord	14 (22)	33 (21)	0.86
ARBd	6 (9)	2 (1)	0.008
Aldosterone antagonist	3 (5)	6 (4)	0.72
Loop diuretic	13 (20)	47 (29)	0.18
Thiazide diuretic	7 (11)	10 (6)	0.27
Trimethoprim-sulfamethoxazole	1 (2)	6 (4)	0.68
Calcineurin inhibitor/sirolimus	3 (5)	7 (4)	1.00
Median inpatient treatment duration (days) (range)	4 (1–40)	5.5 (1–48)	0.046

^aImmunosuppressed was defined as history of solid-organ or hematopoietic stem cell transplantation, human immunodeficiency virus infection, chronic steroid use, neutropenia, or other conditions requiring immunosuppressive medications.

^bBacteremia was complicated by a secondary site of infection in 42% and 55% of patients receiving oxacillin and nafcillin, respectively (P = 0.48).

^cIncludes discitis, osteomyelitis, and septic arthritis.

^dACE, angiotensin-converting enzyme; ARB, angiotensin II receptor blocker.

Forty-two percent (93/224) of patients developed hypokalemia (potassium level of ≤3.3 mmol/liter) during their hospitalization (Table 2); rates were higher for patients who received nafcillin (51%; 82/160) than for those who received oxacillin (17%; 11/64) (P < 0.0001). The median times to hypokalemia were 3 and 4 days among patients who received nafcillin and oxacillin, respectively. Rates of severe hypokalemia (potassium level of ≤2.9 mmol/liter) and an acute drop in potassium levels (≥0.5-mmol/liter decrease from the baseline level) were also higher in the nafcillin group (20% [32/160] and 56% [90/160], respectively) than in the oxacillin group (3% [2/64] [P = 0.0008] and 34% [22/64] [P = 0.005], respectively). By multivariate logistic regression analysis (Tables 3 and 4), receipt of nafcillin was an independent predictor of both severe hypokalemia (odds ratio [OR] = 6.74; 95% confidence interval [CI], 1.45 to 31.2; P = 0.02) and an acute drop in potassium levels (OR = 3.82; 95% CI, 1.81 to 8.08; P < 0.0001). Hypokalemia was more common within the first 7 days among patients who received nafcillin, as rates, defined as potassium levels of ≤3.3 or ≤2.9 mmol/liter or as a ≥0.5-mmol/liter decrease from the baseline level, were 44% (71/160), 15% (24/160), and 51% (82/160), respectively. The corresponding rates in the oxacillin group were 14% (9/64; P < 0.0001), 0% (0/64; P = 0.0002), and 28% (18/64; P = 0.002), respectively.

TABLE 2.

Patient adverse events and outcomes

Parametera	Value for treatment group		P value
	Oxacillin (n = 64)	Nafcillin (n = 160)
No. (%) of patients with hypokalemia (K level of ≤3.3 mmol/liter)	11 (17)	82 (51)	<0.0001
Median time to hypokalemia (days) (range)	4 (1–14)	3 (1–15)	0.21
No. (%) of patients with severe hypokalemia (K level of ≤2.9 mmol/liter)	2 (3)	32 (20)	0.0008
Median time to severe hypokalemia (days) (range)	16.5 (14–19)	5 (1–27)	0.06
No. (%) of patients with ≥0.5-mmol/liter decrease in K level from baseline	22 (34)	90 (56)	0.005
Median time to decrease in K level from baseline (days) (range)	2 (1–17)	2 (1–15)	0.14
No. (%) of patients with any liver function test result ≥3× ULN	1 (2)	7 (4)	0.45
No. (%) of patients with bilirubin levels of ≥4.5 mg/dl	2 (3)	7 (4)	1.000
No. (%) of patients with serum creatinine levels ≥1.5× baseline	4 (6)	29 (18)	0.03
No. (%) of patients with discontinuation due to toxicity	1 (2)	29 (18)	0.0004
Acute kidney injury	0	15	0.007
Hypokalemia	0	12	0.02
Leukopenia	1	1	0.49
Hypernatremia	0	1	1.00
No. (%) of patients with discontinuation of therapy due to hypersensitivity	4 (6)	5 (3)	0.28
Median no. of days in hospital after start of therapy (range)	7.5 (1–85)	11 (1–84)	0.02
No. (%) of patients with readmission within 30 days of discharge	18 (28)	35 (22)	0.38
No. (%) of patients who died within 30 days of discharge	4 (6)	15 (9)	0.60

^aULN, upper limit of normal; K, potassium.

TABLE 3.

Risk factors for severe hypokalemia^a

Parameter	Value for group		Univariate P value	Multivariate P value	Odds ratio (95% CI)
	Severe hypokalemia (n = 34)	No hypokalemia (n = 190)
Median age (yr) (range)	57 (27–85)	55 (19–90)	0.42
Median Charlson comorbidity score (range)	4 (0–10)	3 (0–10)	0.12	0.18
No. (%) of male patients	20 (59)	123 (65)	0.56
Median baseline potassium level (mmol/liter) (range)	3.9 (3.4–4.6)	4.0 (3.4–5.2)	0.06	0.02	0.22 (0.06–0.77)
Median baseline serum creatinine level (mg/dl) (range)	1.0 (0.2–4.1)	0.9 (0.3–4.7)	0.13	0.28
No. (%) of patients taking concomitant medication
ACE inhibitorb	8 (34)	39 (21)	0.65
ARBb	0 (0)	8 (4)	0.61
Aldosterone antagonist	2 (6)	7 (4)	0.63
Loop diuretic	16 (47)	44 (23)	0.006	0.07
Thiazide diuretic	4 (12)	13 (7)	0.30
Trimethoprim-sulfamethoxazole	1 (3)	6 (3)	1.00
Calcineurin inhibitor/sirolimus	0 (0)	10 (5)	0.37
Median treatment duration (days) (range)	12.5 (2–48)	4.5 (1–46)	<0.0001	<0.0001	1.08c (1.04–1.13)
No. (%) of patients who received nafcillin	32 (94)	128 (67)	0.001	0.02	6.74 (1.46–31.2)

^aDefined as a K level of ≤2.9 mmol/liter.

^bACE, angiotensin-converting enzyme; ARB, angiotensin II receptor blocker.

^cPer day of therapy.

TABLE 4.

Risk factors for hypokalemia^a

Parameter	Value for group		Univariate P value	Multivariate P value	Odds ratio (95% CI)
	Hypokalemia (n = 112)	No hypokalemia (n = 112)
Median age (yr) (range)	59 (19–88)	53 (19–90)	0.003	0.02	1.03 (1.01–1.06)
Median Charlson comorbidity score (range)	3 (0–10)	2 (0–10)	0.004	0.32
No. (%) of male patients	73 (65)	70 (63)	0.68
Median baseline potassium level (mmol/liter) (range)	4.1 (3.4–5.2)	3.8 (3.4–5.2)	<0.0001	<0.0001	5.86 (2.53–13.6)
Median baseline serum creatinine level (mg/dl) (range)	0.8 (0.3–4.7)	1 (0.2–4.1)	0.0004	0.37
No. (%) of patients taking concomitant medication
ACE inhibitorb	27 (24)	20 (18)	0.32		2.76 (1.31–5.81)
ARBb	4 (4)	4 (4)	1.00
Aldosterone antagonist	6 (5)	3 (3)	0.49
Loop diuretic	43 (38)	17 (15)	<0.0001	0.008
Thiazide diuretic	11 (10)	6 (5)	0.31
Trimethoprim-sulfamethoxazole	3 (3)	4 (4)	1.00
Calcineurin inhibitor/sirolimus	7 (6)	3 (3)	0.33
Median treatment duration (days) (range)	6 (1–48)	4.5 (1–46)	0.004	0.03	1.04c (1.01–1.08)
No. (%) of patients who received nafcillin	90 (80)	22 (20)	0.005	<0.0001	3.82 (1.81–8.08)

^aDefined as a ≥0.5-mmol/liter decrease from baseline levels.

^bACE, angiotensin-converting enzyme; ARB, angiotensin II receptor blocker.

^cPer day of therapy.

AKI occurred in 18% (29/160) of patients who received nafcillin, compared to 6% (4/64) of patients who received oxacillin (P = 0.03). Concomitant nephrotoxins (aminoglycosides, angiotensin-converting enzyme inhibitors, calcineurin inhibitors, ganciclovir, intravenous contrast dye, loop diuretics, nonsteroidal anti-inflammatory drugs [NSAIDs], vancomycin, and vasopressors) were present in 86% (25/29) and 100% (4/4) of patients who developed AKI on nafcillin and oxacillin, respectively (P = 1.00). Rates of abnormal liver function test results were comparable between groups (Table 2). Adverse events led to early treatment discontinuation in 18% (29/160) and 2% (1/64) of patients who received nafcillin and oxacillin, respectively (P = 0.0004). Discontinuation ascribed to AKI or hypokalemia was identified solely among patients who received nafcillin (P = 0.007 and P = 0.02, respectively) (Table 2). The median time to nafcillin discontinuation secondary to AKI was 4 days (range, 2 to 14 days). Fifty-eight percent (7/12) of patients who required nafcillin treatment discontinuation due to hypokalemia were switched to oxacillin, and only 1 patient (14%) experienced hypokalemia thereafter.

Rates of death and hospital readmission within 30 days of discharge were similar for patients treated with nafcillin and those treated with oxacillin (Table 2). Patients who received nafcillin were hospitalized for a median of 3.5 days longer than patients treated with oxacillin (P = 0.02).

DISCUSSION

This is the first study to compare the tolerabilities of nafcillin and oxacillin among adult patients in the hospital setting. As a result, we have identified several new findings that have important implications for clinical practice. First, we affirmed our hypothesis that rates of hypokalemia are significantly higher among patients receiving nafcillin than among those receiving oxacillin. Second, we found that AKI occurs more frequently among hospitalized patients who receive nafcillin. Finally, we demonstrated that patients who received nafcillin were more likely to discontinue therapy due to adverse events than were patients who received oxacillin. These data corroborate results of previous studies evaluating the tolerability of nafcillin or oxacillin among outpatients (18, 19).

Consensus guidelines recommend nafcillin or oxacillin as a first-line treatment for MSSA infections without a preference between the agents (5, 6). Thus, it is striking to identify a higher rate of treatment discontinuation associated with nafcillin. Early discontinuation was attributed to disproportionately high rates of AKI and hypokalemia. Indeed, renal tubular damage and interstitial nephritis are well-described limiting factors of the antistaphylococcal penicillins (20) and have been reported following the administration of nafcillin (21). It is important to note, however, that our study was not designed to differentiate hypersensitivity-mediated interstitial nephritis, and thus, we characterized AKI more broadly. In doing so, we found that 15% of hospitalized patients developed AKI overall. Among these patients, 52% and 0% who received nafcillin and oxacillin, respectively, required drug discontinuation (P = 0.03). Rates of nafcillin-associated AKI reported here are higher than those among patients receiving OPAT (18). Although we cannot definitively associate nafcillin therapy with the type or rate of AKI, our findings suggest that future investigations are needed.

Previous studies evaluating the tolerability of antistaphylococcal penicillins have not included electrolyte imbalances as an endpoint (18, 19). This is noteworthy given that hypokalemia was the most common adverse event identified in our study. Drug-induced hypokalemia is related to nonabsorbable ion effects in the distal tubule and/or intracellular redistribution due to volume depletion (22–25). The available literature suggests that this occurs rarely after instituting therapy with penicillin-type antimicrobials (22, 25). With this in mind, we analyzed and interpreted our data with extreme caution. After controlling for potentially confounding factors, we determined that hypokalemia was independently associated with nafcillin therapy and resulted in treatment discontinuation in 8% of patients (compared to 0% of patients who received oxacillin; P = 0.02). To limit sampling bias, we further defined hypokalemia using several clinically relevant definitions, all of which provided the same conclusion. In addition, we found that the baseline potassium level is a key determinant of hypokalemia. Patients with a higher baseline potassium level were less likely to develop severe hypokalemia (Table 3) but more likely to have an acute drop in the potassium level (Table 4). Moreover, the risk for hypokalemia was potentiated by each additional day of nafcillin therapy and receipt of concomitant loop diuretics.

If our findings are validated in randomized controlled studies, the implications for clinical practice are significant. Nafcillin and oxacillin are widely prescribed antimicrobials with similar pharmacokinetics, costs, and efficacies. Recognition of improved tolerability may motivate the selection of one agent over the other, particularly among patients receiving extended treatment courses. On the other hand, the semisynthetic penicillins may be replaced altogether for patients who cannot be routinely monitored. Accumulating evidence suggests that cefazolin may be a better-tolerated alternative to both nafcillin and oxacillin (18, 19, 26, 27). These findings will need to be balanced with practical considerations like cost and administration of the agents. To this end, another important observation in our study was the longer duration of hospitalization (3.5 days) for patients who received nafcillin. It is unclear if higher rates of adverse events, more frequent laboratory monitoring, or the need for electrolyte replacement may have contributed to prolonged hospitalizations; however, diagnosis and treatment of adverse events are time- and cost-ineffective. As such, our study is likely to have direct implications for OPAT and antimicrobial stewardship programs that must regulate costs while improving patient outcomes.

Like all retrospective studies, ours is not without limitations. Of importance is the potential for undetected differences in baseline characteristics of patients who received nafcillin or oxacillin. While demographics were similar between groups and were controlled for in multivariable models, indication bias can be excluded only through a randomized controlled trial. In our study, the risk for indication bias is mitigated by the lack of a formulary preference between the agents, which are used interchangeably. Notwithstanding, however, a randomized approach would allow for a comparative-effectiveness study of nafcillin and oxacillin. Based on our preliminary findings, outcomes of interest would include the rate and severity of adverse events, frequency of treatment discontinuation, and potential cost benefit of oxacillin. It is also important to note that our patient population may not be representative of others. The UPMC is a tertiary-care referral center that houses highly dynamic, critically ill patients. Rates of adverse events reported here should not be directly compared to those among patients receiving OPAT (18, 19). Furthermore, we cannot definitively comment on differences in the long-term tolerabilities of nafcillin and oxacillin. This is particularly important for detecting differences in hepatotoxicity, which may have a delayed onset (13). Finally, our findings should be regarded as specific for nafcillin and oxacillin and not be extrapolated to other antistaphylococcal penicillins like dicloxacillin and flucloxacillin, which are commonly used outside the United States. Nevertheless, we conclude that nafcillin is less well tolerated than oxacillin among hospitalized patients. Adverse events led to higher rates of early treatment discontinuation among patients who received nafcillin. These pilot data should be validated by future studies aimed to determine the optimal treatment strategies for patients with MSSA infections who require long-term therapy.