RSS Feeds
Once-Daily Aminoglycoside Dosing in Immunocompetent Adults
A Meta-Analysis
- Rosemarie Hatala, MD, MSc;
- Tuan Dinh, MSc, RPh; and
- Deborah J. Cook, MD, MSc
- From McMaster University, Hamilton, Ontario, Canada. Acknowledgments: The authors thank L. Griffith for statistical analysis and J.M. Aguado, D. Gilbert, P. Gonzalez, A. Herchuelz, E. Hippe, W. Kern, R. Maller, J. Prins, E. ter Braak, R.W. Vreede, and P.J. de Vries for their generous responses to our inquiries. Requests for Reprints: Rosemarie Hatala, MD, Department of Medicine, Henderson General Hospital, 711 Concession Street, Hamilton, Ontario L8V 1C3, Canada. Current Author Addresses: Dr. Hatala: Department of Medicine, Henderson General Hospital, 711 Concession Street, Hamilton, Ontario L8V 1C3, Canada.
Abstract
Objective: To compare the efficacy, nephrotoxicity, and ototoxicity of once-daily aminoglycoside dosing with those of standard aminoglycoside regimens in immunocompetent adults.
Data Sources: A structured MEDLINE search from 1966 to April 1995 using the keywords aminoglycosides, drug administration schedule, and adult; bibliographic searching of review articles, position papers, and references of the selected articles; contact with primary authors of selected articles to obtain information not in the published reports and lists of potentially relevant articles.
Study Selection: Randomized, controlled trials that 1) compared an intravenous once-daily aminoglycoside regimen with a standard aminoglycoside regimen in infected immunocompetent adults and 2) examined efficacy, mortality, or toxicity.
Data Extraction: For each selected study, two independent reviewers assessed methodologic quality and abstracted data. The heterogeneity of individual study risk ratios was assessed and data were pooled using a random-effects model.
Results: Forty-two studies were reviewed for possible inclusion. Thirteen independent studies met the selection criteria, and their results were pooled. The trials had a mean methodologic quality score of 0.69 (range, 0.50 to 0.91). Heterogeneity exists among the individual risk ratios for clinical cure (P = 0.07); significant heterogeneity does not exist for the other outcomes. For the pooled efficacy outcomes, the risk ratio for bacteriologic cure is 1.02 (95% CI, 0.99 to 1.05), and the risk ratio for mortality is 0.91 (CI, 0.63 to 1.31). For the pooled toxicity outcomes, the risk ratio for nephrotoxicity is 0.87 (CI, 0.60 to 1.26), and the risk ratio for ototoxicity is 0.67 (CI, 0.35 to 1.28).
Conclusions: Standard and once-daily aminoglycoside dosing regimens are equivalent with regard to bacteriologic cure, and once-daily dosing shows a trend toward reduced mortality and toxicity. However, additional studies are needed for more precise estimates of mortality and toxicity risk ratios. The equivalency of the dosing regimens, the ease of administration, reduced nursing time, and reduced variability in the timing of drug administration that are associated with once-daily dosing may mean that the once-daily regimen is clinically advantageous.
Aminoglycoside antibiotics remain an integral part of modern therapy for severe bacterial infections. Recent interest in the aminoglycosides has emphasized their unique synergistic role in combination with β-lactams against resistant gram-negative strains and in bacteremia caused by Enterococci species [1-5].
Aminoglycoside effectiveness depends on achieving adequate serum levels relative to the minimum concentration needed to inhibit bacterial growth [6, 7]. However, a minimal trough serum concentration may be required to avoid the risks of ototoxicity and nephrotoxicity [8, 9]. Conventional dosing strategies have used either population-based nomograms or labor-intensive services to achieve optimal drug levels [10, 11]. Alternatively, preliminary human and animal studies [12] support the use of a single, once-daily dose of an aminoglycoside. This simplified regimen is designed to both capitalize on aminoglycoside concentration-dependent bacterial killing and postantibiotic effect and reduce the risk for toxicity caused by excessive drug accumulation [13, 14].
Randomized trials of once-daily aminoglycoside dosing in adults have yielded conflicting results about efficacy and toxicity [15-31]; many of these studies have had insufficient power to produce a definitive answer. Recent editorials [13, 32] have highlighted selected samples of this available literature on once-daily dosing, adding to uncertainty in this area. Comprehensive overviews examining the once-daily regimen in adults have not used quantitative methods to aggregate study results [13, 14, 33, 34]. Therefore, to address the inadequate power of the published trials and to synthesize their results, we did a meta-analysis of the randomized, clinical trials that compared once-daily aminoglycoside regimens with standard aminoglycoside dosing regimens in immunocompetent adults. Both the efficacy and the toxicity of the regimens were evaluated.
Methods
Data Sources
To identify relevant studies, two independent reviewers did MEDLINE searches from 1966 to April 1995 using the following keywords: aminoglycosides (exploded), drug administration schedule, and adult. Selected infectious disease journals published from November 1994 to April 1995 were searched manually. Additional articles were identified from review articles, position papers, and references of the selected articles. Any article identified as relevant by either reviewer was retrieved.
Study Selection
To meet the study selection criteria, a study had to 1) be a randomized, controlled trial; 2) compare an intravenous once-daily aminoglycoside regimen with a standard aminoglycoside regimen; 3) comprise infected, immunocompetent adult patients [exclusion of surgical prophylaxis]; 4) use any of the following outcome measures: bacteriologic or clinical cure, mortality, nephrotoxicity, or ototoxicity; and 5) have a patient sample in which fewer than 50% of the patients had lower urinary tract infections (optimal bactericidal levels of aminoglycoside in patients with lower urinary tract infections are achieved regardless of the dosing regimen because the entire daily dose is excreted unchanged into the bladder). All potentially relevant articles identified by the search were examined by two independent reviewers for adherence to selection criteria. Agreement statistics (κ) for study selection were calculated, and disagreement was resolved by consensus.
Methodologic Quality
Articles meeting the inclusion criteria were examined independently by two reviewers for methodologic quality, which was scored using a standardized, weighted quality grading system with a maximum possible score of 12. The following characteristics were assessed: randomization of patients, blinding of patients and observers, completeness of follow-up, explicitly stated inclusion and exclusion criteria, cointervention with other antibiotic agents, treatment compliance, defined outcome measures, and intention-to-treat analysis. The achieved score was divided by the maximum score [12] to obtain a methodologic quality grading between 0 and 1.0. We calculated κ statistics, and disagreement was resolved by consensus.
When information pertaining to our methodologic quality criteria or outcome assessments was missing from the published report, we contacted the primary author of the selected article by mail to obtain the missing information. Requested information included the specific method of randomization, whether the outcome assessors were blinded to patient intervention, the specific numbers of patients for each outcome assessment, and nephrotoxicity and ototoxicity data analyzed by the standard definitions used in our meta-analysis. The authors were also asked to list any potentially relevant articles that were not included in our reference list.
Data Extraction
We used data conforming to the following definitions for the outcome assessments.
1. Bacteriologic cure: any study definition.
2. Clinical cure: any study definition.
3. Mortality: all-cause mortality.
4. Nephrotoxicity: an increase in serum creatinine concentration of at least 35 to 45 µmol/L (0.5 mg/dL) during the study period (the standard definition of nephrotoxicity) [35].
5. Ototoxicity: a 15-dB change in hearing at any frequency, as assessed by audiometry.
Additional information about outcome measures that was not explicitly provided in the published report was sought through correspondence with the primary investigators. The outcome data were extracted by two independent reviewers, and disagreement was resolved by consensus.
Analysis
For study selection and methodologic quality assessment, both raw agreement and κ statistics were calculated. The κ statistic measures the agreement between observers beyond that expected by chance [36].
When duplicate information was reported in preliminary abstracts and subsequent articles (15 and 17, 20 and 28, 29 and 31), the analyzed data were taken from the most complete data set available [17, 20, 31].
When a meta-analysis is being conducted, it is important to consider the variation between individual study risk ratios for each outcome when deciding whether it is logical to combine the study results into a pooled risk ratio. This variation can be quantitated statistically, in an assessment of heterogeneity, to measure the extent to which it represents measurement of the same effect size (with variation due to chance) rather than true differences in individual results. If heterogeneity exists between individual study risk ratios, it may reflect clinically important differences in the patient populations, intervention effects, or research methods of the individual trials [37].
In our meta-analysis, heterogeneity of individual study risk ratios was assessed by using a chi-square test for homogeneity and by visual inspection. A pooled risk ratio, using a random-effects model [38], was calculated for each outcome in which homogeneity of the individual study risk ratios was seen. For outcomes with significant heterogeneity among risk ratios, the results were not pooled.
From each pooled risk ratio, the number needed to treat was calculated to show the clinical consequences of treatment. The number needed to treat is the number of patients who need to receive once-daily therapy to prevent one complication or to derive one beneficial outcome compared with standard therapy [39]. This number was calculated for each outcome by dividing the absolute risk reduction obtained through once-daily dosing into 1 [39].
Before data analysis, four hypotheses were identified as possible explanations for the differences among the risk ratios of individual trials. We hypothesized that individual study results may have varied as a result of different methodologic quality, different aminoglycosides administered, different classes of antibiotics responsible for cointervention and different primary sources of infection. These four hypotheses were tested in sensitivity analyses whenever heterogeneity was seen among individual risk ratios for a given outcome.
Results
Trial Characteristics
We reviewed 42 studies for possible inclusion [12, 15-31, 40-63] and excluded 25 [12, 40-63]. Reasons for exclusion were nonrandomized trial design [59, 63], nonintravenous route of drug administration [54-56], nonadult population [57, 58], lack of documented infection [40-44], lower urinary tract infection in more than 50% of patients [61], prophylactic administration [62], lack of a control group receiving a twice-daily or three-times-daily regimen [12, 45-49], and lack of outcome criteria [60]. Four studies were review articles [50-53]. One study was reported as an abstract [30] and adequate data extraction was not possible.
For study selection, the raw agreement between the two reviewers was 83%; the κ statistic was 0.66, indicating substantial agreement between reviewers [36].
The characteristics of the individual trials, as well as their combinable outcomes (meeting the prespecified definitions), are given in Table 1. For the single study [19] that compared patients receiving two different doses of a once-daily dosing regimen with a control group, each intervention arm was compared separately with the control group (Table 1). Differences in patient sample sizes for efficacy and toxicity risk ratio estimates were attributable to different studies that were pooled for each outcome and to different sample sizes for each outcome within individual studies. Within-study sample size variability was a result of different treatment periods for the intra-study assessments of efficacy and toxicity [20, 23, 25], patient exclusion from efficacy analysis when diagnostic confirmation by culture could not be obtained [16], and different definitions of bacteriologic and clinical cure [18, 21, 23, 25].
The trials had a mean methodologic quality score of 0.69 (range, 0.50 to 0.91). The scoring of individual trials on each item within the methodologic quality score is shown in Table 2. The κ statistic for ascertainment of the methodologic quality score was 0.62.
Efficacy Outcomes
For bacteriologic cure, the pooled risk ratio is 1.02 (95% CI, 0.99 to 1.05) (Figure 1), indicating that the standard and once-daily aminoglycoside dosing regimens are equivalent.
For clinical cure (Figure 2), both visual inspection and statistical analysis showed that individual trial results were heterogeneous. Four separate sensitivity analyses, wherein individual studies were categorized according to 1) methodologic quality, 2) the specific aminoglycoside used, 3) the cointervention antibiotic used, and 4) the site of infection, showed persistent heterogeneity. Thus, the individual study results were not pooled to obtain a common risk ratio.
Examination of mortality shows a pooled mortality risk ratio of 0.91 (CI, 0.63 to 1.31), favoring once-daily dosing and corresponding to a relative risk reduction of 0.09, or 9%. Estimating a baseline mortality rate of 10% from the crude event rates, 111 patients would need to be treated with once-daily dosing instead of standard dosing to prevent one death.
Toxicity Outcomes
Examination of toxicity shows a pooled risk ratio for nephrotoxicity of 0.87 (CI, 0.60 to 1.26), corresponding to a relative risk reduction of 13% with the once-daily regimen (Figure 3). Given a 10% baseline risk for nephrotoxicity, 77 patients would need to be treated with once-daily instead of standard dosing to prevent nephrotoxicity in 1 patient.
For ototoxicity, the combined risk ratio is 0.67 (CI, 0.35 to 1.28) with once-daily dosing, corresponding to a relative risk reduction of 33%. With a 5% baseline risk for ototoxicity, the associated number needed to treat is 61.
Discussion
This meta-analysis of standard compared with once-daily aminoglycoside dosing regimens in immunocompetent adults shows that once-daily dosing is equivalent to conventional dosing in terms of bacteriologic cure and may be associated with reduced nephrotoxicity, ototoxicity, and mortality. The efficacy of the once-daily regimen may result from the higher peak serum drug concentrations it produces, which lead to enhanced tissue penetration and a longer postantibiotic effect [64]. Moreover, a higher peak concentration may be beneficial for patients with an increased volume of distribution or clearance for aminoglycosides [65].
In contrast, the trials evaluating the efficacy of these regimens in producing clinical cure show visual and statistical heterogeneity of individual study risk ratios, implying differences in the estimates of clinical cure in individual trials (Figure 2). Preplanned sensitivity analyses evaluating the extent to which the treatment effect varies with the type of aminoglycoside, type of concomitant antibiotic, and primary source of infection could not explain this heterogeneity. Cautious post hoc examination of these data suggests three alternative hypotheses. First, the heterogeneity may reside mainly in the CIs of the two studies [16, 22], which do not overlap [66]. One of these trials had a markedly different patient population, with a preponderance of intra-abdominal infections treated nonsurgically [22]. Second, the effect of comorbid illness was not examined as a potential source of heterogeneity. Finally, broad inclusion criteria for our meta-analysis resulted in substantial heterogeneity of the definitions of clinical cure in each study. This may have translated into the variable risk ratio estimates that were obtained.
The final efficacy analysis was an examination of mortality rates with either regimen. The pooled risk ratio for mortality suggests equivalent risks for death with either therapy, if not a trend toward reduced mortality with the once-daily regimen. An explanation for this potential benefit with the once-daily regimen is that underdosing may occur with standard regimens in critically ill patients who require higher individual doses to compensate for an increased volume of distribution [65, 67]. Despite the inclusion of approximately 600 patients in each treatment group, the pooled risk ratio does not have the power to clearly delineate a mortality benefit of either treatment regimen.
Considering the toxicity outcomes, the nephrotoxicity pooled risk ratio indicates a nonsignificant trend toward reduced nephrotoxicity with once-daily dosing (Figure 3). In addition, all studies that had increases in the serum creatinine level of 45 µmol/L (the standard definition of nephrotoxicity [35]) or less were combined to maximize the number of patients in the analysis; this may have led to a conservative toxicity estimate by overestimating the incidence of nephrotoxicity as defined conventionally. The net effect of this approach depends on any true differences between the regimens that exist at the upper (increase of 45 µmol/L) but not the lower definition of nephrotoxicity. Surrogate markers such as renal β2-microglobulin or N-acetyl-β-D-glucosaminidase levels were not examined because they were rarely used to define nephrotoxicity in the selected trials and their predictive utility for nephrotoxicity has not been established [68]. The potential relative risk reduction for nephrotoxicity with once-daily dosing may be explained by the rationale, from animal and human studies, that once-daily dosing results in less renal accumulation of the aminoglycoside through complete brush border membrane phospholipid saturation of cells lining the proximal tubule [14, 69].
As does the outcome of nephrotoxicity, the pooled risk ratio for ototoxicity indicates a nonsignificant trend toward reduced ototoxicity with the once-daily regimen. Evidence from animal studies is conflicting as to whether ototoxicity is related to dosing regimen or total administered dose [33]. In human studies, irreversible vestibular toxicity leading to vertigo and nystagmus may correlate with the duration of aminoglycoside exposure in both periand endolymph fluid [70]. It has also been suggested [71] that once-daily administration may not pose a higher risk for ototoxicity if peak serum concentrations of gentamicin do not exceed 12 µg/mL [71]. Our pooled relative risk reduction of 33% with once-daily dosing is consistent with potentially reduced ototoxicity, despite the higher peak serum concentrations attained with the once-daily regimen, but it is not statistically significant.
In summary, this meta-analysis of once-daily compared with standard aminoglycoside dosing regimens shows that the two regimens have equivalent efficacy and that the once-daily regimen may potentially reduce toxicity and mortality when used in combination with other antibiotics. The wide CIs around the risk ratios, which include 1.0, show that additional studies are needed to obtain more precise estimates of the mortality and toxicity risk ratios. However, the pooled risk ratios for each efficacy and toxicity outcome indicate that once-daily dosing appears to be no more harmful than and at least as efficacious as standard dosing regimens. Individual randomized trials of similar drug regimens may differ in their estimates of treatment effect for many reasons, including variability in the patient populations, interventions, or study methods; an example of such variations is the different aminoglycosides included in this overview. Although these aminoglycosides differ in their chemical structure and their in vitro efficacy against certain bacterial strains, their pharmacokinetics and rates of nephrotoxicity are similar [72, 73]. In addition, the in vitro differences among the individual aminoglycosides do not translate into significant clinical differences because significant variation among the individual study risk ratios does not exist for any outcome except clinical cure and no pooled risk ratio estimate varied significantly when examined by sensitivity analysis. Thus, it is methodologically sound to use meta-analytic techniques to combine studies that used different specific aminoglycosides.
Rigorous systematic reviews provide a summary and a critical appraisal of the current literature on a given topic. Although methodologic quality grading scales have been used widely to assess the quality of individual trials in meta-analyses [74], until recently there has been little empiric evidence for the validity of the scale or of any individual item in the scale [75]. Concealment of randomization and double-blinding are two methodologic trial features associated with biased estimates of treatment effects [76]. Table 2 shows that although the mean methodologic quality of the trials in our meta-analysis was fair (0.69), these trials did not have adequate, consistent concealment or double-blinding. Consequently, the pooled risk ratios may overestimate the true treatment effect.
One limitation of this meta-analysis is a potential susceptibility to publication bias. Although a comprehensive literature search was done using both MEDLINE and bibliographic sources, we searched for unpublished research by communicating with experts in the field and retrieved no additional data. If significant, unpublished, negative results of trials of once-daily dosing exist, then our results may reflect an overestimation of the pooled risk ratio [77]. In addition, most but not all primary authors responded to our inquiries about methodologic quality criteria and missing data; thus, some trials are more accurately represented than others.
A final limitation of our meta-analysis pertains to the generalizability of the pooled risk ratios. Because all of the included trials administered aminoglycosides in combination with other antibiotics, and because these antibiotics were equally distributed between both treatment regimens, the results of our meta-analysis are only applicable to combination regimens involving aminoglycosides; single-agent aminoglycoside therapy has not been assessed. Given that aminoglycosides are only effective against gram-negative bacteria and that broader-spectrum antibiotic therapy is generally initiated before specific bacteriologic culture reports are available to clinicians, it would be challenging to conduct a prospective, randomized trial with sufficient power to definitively assess the efficacy and toxicity of different single-agent aminoglycoside regimens.
Whenever two therapies appear to be equivalent, as in our meta-analysis, other clinically relevant benefits must be examined. For aminoglycoside administration, the potential equivalence of efficacy and toxicity with the two regimens confers important clinical advantages on the once-daily regimens. First, once-daily drug administration is more convenient for patients and nursing staff. It is associated with less variability in the timing of drug administration [13]. In addition, the reduced nursing time involved in once-daily drug administration may result in cost-minimization [78]. Even with the limited evidence currently available, once-daily dosing regimens have already been instituted as standard therapy at many centers [79, 80].
Aside from additional studies to improve the precision of the toxicity estimates, further directions for research on once-daily dosing regimens remain. Methodologically rigorous studies of tobramycin administered once daily are lacking. The nephrotoxicity of once-daily regimens in patients with preexisting renal failure has not been evaluated. Additionally, optimal drug monitoring remains unclear. Finally, certain serious infections commonly treated with aminoglycosides, such as endocarditis, have not been evaluated for their response to a once-daily dosing regimen. Future research on this clinically relevant topic may lead to the judicious use of once-daily aminoglycosides in defined adult populations.
Dr. Cook: Department of Medicine, St. Joseph's Hospital, 50 Charlton Avenue East, Hamilton, Ontario L8N 4A6, Canada.
Mr. Dinh: Department of Pharmacy, Hamilton General Hospital, 237 Barton Street East, Hamilton, Ontario, Canada.
- Copyright ©2004 by the American College of Physicians
References
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
Related Clinical Content
