The impact of vancomycin trough concentrations on outcomes in non-deep seated infections: a retrospective cohort study

Vancomycin was discovered over 60 years ago; however, it was not until the early 1980s that its clinical use sharply increased in response to a rise in the worldwide prevalence of methicillin-resistant Staphylococcus aureus (MRSA) [1]. Unfortunately, controversy regarding the optimal dosing and pharmacokinetic/pharmacodynamic targets continues to plague the use of vancomycin [2]. Current guidelines recommend targeting vancomycin serum trough concentrations of 15–20 mg/L for complicated infections (e.g. endocarditis, osteomyelitis, pneumonia, meningitis, etc.), and suggest maintaining a trough of > 10 mg/L for all patients (including those with non-deep seated infections with a planned duration of ≤14 days) to avoid development of resistance to vancomycin (Quality and Grade of Evidence for each recommendation in the 2009 IDSA guidelines was IIIB) [3–5].

The emergence of vancomycin resistance has not been observed in patients treated with short term vancomycin (≤14 days) for the management of non-deep seated infections (e.g. bacteremia when endocarditis or other non-cardiac seeding [e.g. bone or brain] has been ruled out; uncomplicated skin and soft tissue infections; and urinary tract infections in the absence of anatomic abnormalities, renal stones, or renal abscess).

Clinical evidence to support the need to maintain vancomycin serum trough concentrations above 10 mg/L for management of non-deep seated infections is lacking and the practice of targeting higher trough concentrations has obvious downsides. This practice increases the complexity of the dosing and monitoring of vancomycin, results in additional blood draws from the patient, increases the workload for the clinical team, and may therefore, increase the risk of medical error and harm to the patient. We hypothesize that there is no difference in clinical or microbiological outcomes associated with vancomycin trough concentrations of ≤10 mg/L versus > 10 mg/L when vancomycin is used in the setting of non-deep seated infections for a period of ≤14 days.

The primary objective of this study was to evaluate patients with non-deep seated infections who were treated with vancomycin for a period of ≤14 days to determine whether there were differences in clinical outcome with serum trough concentrations of vancomycin of ≤10 mg/L versus > 10 mg/L. The secondary objectives of this study were to identify factors which may affect clinical outcome, evaluate safety outcomes (kidney injury), assess workload based on dose adjustment(s), and identify changes in microbial resistance during the vancomycin treatment course.

A total of 2098 patients on vancomycin during the study period were identified from the SPIRIT database query for screening, and 103 (5%) met inclusion criteria for the study (Low trough cohort: n = 48, High trough cohort: n = 55) (Fig. 1). Of the 1995 excluded patients, the most common reason for exclusion was the presence of a deep-seated infection (1134 patients). The two groups were well-balanced in terms of baseline demographic, clinical and microbiological factors (Table 1).

	Low Trough (N = 48)	High Trough (N = 55)	Odds Ratio	95% Confidence Interval	p-value
Gender (Male)	27 (56%)	28 (51%)	1.24	0.57–2.70	0.69
Age on Admission (Years), mean ± SD (Range)	59 ± 19 (21–91)	67 ± 21 (19–96)			0.06
Hospital Location at Vancomycin Initiation (Ward)	38 (79%)	44 (80%)	0.95	0.36–2.48	> 0.99
Length of Stay at the time of Vancomycin Initiation (Days),mediana(Range)	2(0–70)	2(0–187)			0.89
Any Comorbidityb	19 (40%)	25 (45%)	0.79	0.36–1.72	0.56
• Congestive Heart Failure	3 (6%)	8 (15%)	0.39	0.098–1.57	0.21
• Chronic Obstructive Pulmonary Disease	2 (4%)	7 (13%)	0.30	0.06–1.51	0.17
• Diabetes Mellitus	7 (15%)	10 (18%)	0.77	0.27–2.21	0.79
• Immunosuppression due to disease or drugc	8 (17%)	6 (11%)	1.63	0.52–5.10	0.57
APACHE IId, mean ± SD (ICU patients)	17 ± 6 (11–25)	23 ± 11 (6–41)			0.32
Critically Ill Ward Patients (Pitt Bacteremia Score ≥ 4)	1 (2%)	4 (7%)	0.27	0.03–2.52	0.37
Baseline Creatinine (mmol/L)	77 ± 32 (23–185)	83 ± 45 (18–254)			0.42
Use of Concomitant Nephrotoxinse	35 (73%)	36 (65%)	1.42	0.61–3.31	0.52
• # of Concomitant Nephrotoxinsa (median)	1 (0–3)	1 (0–4)			0.49
Use of Concomitant Antibioticsf	21 (44%)	17 (31%)	1.74	0.78–3.90	0.22
• Same Indication as Vancomycing	18 (86%)	14 (82%)	1.29	0.22–7.37	> 0.99
Included Infections
• All SSTIs	31 (65%)	32 (58%)	1.31	0.59–2.91	0.55
o Cellulitis	25 (52%)	22 (40%)	1.63	0.75–3.57	0.24
o Wound/Surgical Site Infection	6 (12%)	10 (18%)	0.64	0.21–1.92	0.59
• All UTIs	11 (23%)	7 (13%)	2.04	0.72–5.77	0.20
o MRSA UTI	2 (4%)	1 (2%)	2.35	0.21–26.75	0.60
o Enteroccocal UTI	9 (19%)	6 (11%)	1.88	0.62–5.75	0.28
• All Bacteremias	4 (8%)	10 (18%)	0.41	0.12–1.40	0.16
o CNST Bacteremiah	3 (6%)	8 (15%)	0.39	0.10–1.57	0.21
• Any Positive CNST in Blood	5 (10%)	14 (25%)	0.34	0.11–1.03	0.07
Microbiology
• Patients With Positive Non-Screening Cultures for Resistant Gram positive isolatesi	10 (21%)	22 (40%)	0.39	0.16–0.95	0.054
o Patients with MRSA Clinical Culture	2 (4%)	5 (9%)	0.43	0.08–2.35	0.44
o Patients with CNST Clinical Culture	8 (17%)	17 (31%)	0.45	0.17–1.16	0.11
• MRSA-colonized Patients	2 (4%)	8 (15%)	0.26	0.05–1.27	0.10
• VRE-colonized Patients	1 (2%)	0	3.50	0.14–88.15	0.47

ACEI Angiotensin converting enzyme inhibitor, AIDS Acquired immune deficiency syndrome, APACHE Acute Physiology and Chronic Health Evaluation, ARB Angiotensin II receptor blocker, ASA Acetylsalicylic acid, HCTZ Hydrochlorothiazide, CNST Coagulase-negative Staphylococci, HIV Human immunodeficiency virus, ICU Intensive Care Unit, NSAID Nonsteroidal anti-inflammatory drug, PPI Proton pump inhibitor, SD Standard deviation, SSTI Skin and soft tissue infection, TNF-α Tumor necrosis factor alpha, UTI Urinary tract infection, VRE Vancomycin-resistant Enterococci

^aFor non-normally distributed data, the median with range was reported

^bPatients may have had more than 1 comorbidity, thus totals for specific comorbidities sum to a value greater than the number of patients with any comorbidity

^cDisease: HIV/AIDS, asplenia, hematological malignancies, transplantation. Drug: Corticosteroids (prednisone > 5 mg/day, chemotherapy, TNF-α inhibitors, transplant medications)

^dArterial blood gases were not available for all ICU patients and APACHE II scoring could not be completed for these patients; reported values are based on 6 patients in low trough cohort and 7 patients in high trough cohort

^eNSAIDs, ACEIs, ARBs, cyclosporine, tacrolimus, acyclovir, aminoglycosides, amphotericin, colistin, indinavir, adefovir, cidofovir, tenofovir, chemotherapy (e.g. carmustine, semustine, cisplatin, methotrexate, mitomycin), foscarnet, contrast dye, zoledronate, loop diuretics, HCTZ, triamterene, hydralazine, interferons, PPIs, sulfonamides, lithium, aristocholic acid, acetaminophen at > 1 g/day for > 2 years, ASA at > 1 g/day for > 2 years)

^fConcomitant antibiotic defined as: ≥48 h overlap with vancomycin and administered for ≥48 h

^gDenominator for percentage calculations is the number of patients (n) on concomitant antibiotics (i.e. low trough n = 18; high trough n = 14)

^hDefined as ≥2 positive blood cultures on the same day

ⁱMethicillin Resistant Staphylococcus aureus, Coagulase-negative Staphylococci, or Vancomycin Resistant Enterococci cultured from 3 days prior to initiation or during vancomycin course of therapy and includes a single positive culture for CNST

The average final steady state vancomycin troughs were 7.38 ± 1.95 mg/L and 16.56 ± 6.56 mg/L in the low and high trough groups, respectively, and the median duration of therapy in both groups was 7 ± 3 days. Although the median total daily dose between groups was similar (2 g per day in both groups), patients in the high trough group were more likely to receive 3 or more grams of vancomycin per day (p = 0.04) and had a significant increase from initial to final vancomycin troughs (p = 0.0009). Patients in the high trough group had more vancomycin dosing changes (p = 0.02), which in turn translated to a three-fold higher investment of health care worker time related to vancomycin therapy per patient in the high trough group (p = 0.02) (Table 2). Renal function outcomes were similar between the two groups, with no significant difference in change in serum creatinine (p = 0.13) and number of patients at risk for kidney injury (p = 0.12). However, it is noteworthy that while no patient in the low trough group was at risk for kidney injury, 4 patients in the high trough group were at risk for kidney injury (Table 2).

This retrospective study evaluated whether differences in clinical cure exist when targeting trough concentrations ≤10 mg/L versus > 10 mg/L in patients with non-deep seated infections treated with vancomycin for ≤14 days; and, to the best of our knowledge, is the first study to do so. Although limited by a small sample size, the study enrollment reflects the entire population of patients with non-deep seated infections treated with vancomycin at our hospital between March 12th, 2010 and December 31st, 2015, inclusive. Strict eligibility criteria for study entry were used and the cohorts were well balanced for all baseline characteristics. We did not measure any difference in clinical cure with higher vancomycin trough concentrations and there was no signal for selection of resistant Gram positive bacteria in patients with a non-deep seated infection receiving ≤14 days of vancomycin. However, there was a three-fold increase in healthcare personnel workload, which introduces unnecessary complexity (increased pharmacist, pharmacy technician, nursing and physician time involvement, additional blood work, and risk of medication error with more frequent dose adjustments).

In current infectious diseases guidelines, the rationale for maintaining trough concentrations above 10 mg/L is to prevent the development of resistance; emergence of vancomycin resistance in patients treated with ≤14 days of vancomycin has never been reported in the literature [3–5]. We did not observe the emergence of any resistance to vancomycin in our study, and no patients with low vancomycin levels (< 10 mg/L) required re-admission for recurrent infection; this is a reassuring finding, since one reason for recurrent infection may be the emergence of vancomycin resistance. Vancomycin resistance in Staphylococcus aureus with low vancomycin levels for ≤14 days has only been observed in vitro with purposeful selection; and clinically in patients who received prolonged vancomycin exposure (6–18 weeks) [2, 13–23]. No association between vancomycin trough concentrations and emergence of vancomycin resistance has been observed for any other bacteria (e.g. CNST, Enterococci, other Gram positive organisms) for which vancomycin may be indicated.

As a single-center, retrospective analysis, our study has several limitations. To maximize our sample size, we screened all potential patients since the inception of our Antimicrobial Stewardship database until study closing. Despite our efforts, our final sample size was only sufficient to detect a difference of 25% points between the two groups for our primary outcome; we set a difference of 10% between the two groups as being clinically important to identify in our sample size calculation, and observed a difference of less than 1% (non-rounded clinical cure rates in low versus high trough cohorts 87.5% vs. 87.3%) between the cohorts. To be adequately powered for a difference in treatment failure of 1% or 0.5%, a sample size of 17,370 or 68,850 patients per cohort, respectively, would be necessary. Since a difference of 1% in treatment failure would not be considered clinically important in non-deep seated bacterial infections, this point illustrates the futility for the need to increase the sample size of our study and minimizes any argument that insufficient sample size may have biased against seeing a difference in clinical outcomes between the low and high trough cohorts. To further address the risk of a type II error, we conducted a MCS of 1 million iterations, using a binomial distribution, with the study observed probability of cure (Low trough = 0.88, High trough = 0.87) and found only a 3% probability that clinical cure in the high trough cohort could theoretically be ≥10% better than the low trough cohort. This result further supports the validity of the study findings that clinical cure was not significantly different between cohorts. Patients that were treated with vancomycin for less than 48 h were excluded to minimize the risk of including patients in whom infection was not considered by the physician. Although this exclusion may introduce selection bias due to early mortality, early death attributable to infection would not be anticipated with non-deep seated infections.

Due to the retrospective nature of this study, unidentified confounding factors may exist, and we made assumptions in the definition of steady state levels, as well as the definition of clinical cure and failure. We were unable to calculate the APACHE II score for all ICU patients, as not all patients had arterial blood gases available; we also extrapolated the use of the Pitt Bacteremia score to all hospital ward patients, as there are no other validated measures of severity of illness in this patient population. However, it is reassuring that both groups were well-balanced for all identified demographic, clinical and microbiological factors, including the number of patients with concomitant antibiotics, as well as the number of patients with concomitant antibiotics for the same indication as vancomycin. Only 5% of patients on vancomycin were eligible for study inclusion, primarily because most patients had deep-seated infections. Therefore, the non-deep seated infection patient population may constitute a small portion of those patients placed on vancomycin, but nevertheless, avoidance of unnecessary increases in vancomycin dosing to provide higher trough concentrations in these patients is an important quality improvement initiative to potentially reduce medication errors associated with frequent dose changes.

To our knowledge, this is the first study to evaluate clinical and microbiological outcomes associated with trough concentrations in patients with non-deep seated infections treated with short course vancomycin. Our results show an increase in workload with no corresponding clinical benefit, and no signal for increased risk of resistance with vancomycin trough concentrations ≤10 mg/L for short term therapy (≤ 14 days) in non-deep seated infections.