Short-term versus usual-term antibiotic treatment for uncomplicated Staphylococcus aureus bacteremia: a systematic review and meta-analysis

Abstract

Introduction:

Uncomplicated Staphylococcus aureus bacteremia remains a leading cause of morbidity and mortality in hospitalized patients. Current guidelines recommend a minimum of 14 days of treatment.

Objective:

To evaluate the efficacy and safety of short versus usual antibiotic therapy in adults with uncomplicated S. aureus bacteremia (SAB).

Methods:

We developed a search strategy to identify systematic review and meta-analysis of non-randomized studies (NRS), comparing short versus usual or long antibiotic regimens for uncomplicated SAB in MEDLINE, Embase, and the Cochrane Register up to June 2023. The risk of bias was assessed using the ROBINS I tool. The meta-analysis was performed using Review Manager software with a random effect model.

Results:

Six NRS with a total of 1700 patients were included. No significant differences were found when comparing short versus prolonged antibiotic therapy as defined by the authors for 90-day mortality [odds ratio (OR): 1.09; 95% confidence interval (CI): 0.82–1.46, p: 0.55; I² = 0%] or 90-day recurrence or relapse of bacteremia [OR: 0.72; 95% CI: 0.31–1.68, p: 0.45; I² = 26%]. Sensitivity analysis showed similar results when comparing a predefined duration of <14 days versus ⩾14 days and when excluding the only study with a high risk of bias.

Conclusion:

Shorter-duration regimens could be considered as an alternative option for uncomplicated SAB in low-risk cases. However, based on a small number of studies with significant methodological limitations and risk of bias, the benefits and harms of shorter regimens should be analyzed with caution. Randomized clinical trials are needed to determine the best approach regarding the optimal duration of therapy.

Plain language summary

Comparing short and regular antibiotic treatment duration, for a type of blood infection caused by S. aureus

We investigated the optimal duration of antibiotic treatment for adults with a specific type of blood infection (uncomplicated Staphylococcus aureus), a condition with a significant global impact on mortality and costs. After a thorough search, only six trials involving 1700 patients were identified. We therefore decided to perform a meta-analysis (a type of statistical analysis). The results showed that the duration of antibiotics, whether short or long (less or more than 14 days), did not significantly affect mortality or recurrence of infection within 90 days. Consequently, we suggested that shorter courses of antibiotics might be appropriate for less severe cases. However, we emphasized caution because of the limitations of the studies. We recommended further research with improved methods to determine the optimal approach to treating this type of infection.

Introduction

Bacteremia due to Staphylococcus aureus (S. aureus) is one of the three leading causes of healthcare-associated bacteremia worldwide.^1,2 Its presence is associated with mortality of up to 10.4% at 7 days and 27% at 30 days. ³ It is also estimated that each episode of bacteremia due to methicillin-sensitive S. aureus (MSSA) can cost approximately 10,000 euros, and up to twice that for methicillin-resistant S. aureus (MRSA). ⁴

Reducing the duration of antibiotic treatment for bacterial infections is being evaluated in various clinical settings, due to its potential to decrease adverse events, costs, and hospitalization duration while also maintaining comparable effectiveness regarding mortality, relapse, and infection complications.^5–8 For example, in community-acquired pneumonia with clinical stability, one study demonstrated non-inferiority of a 3-day versus 8-day regimen. ⁹ Similar findings were reported in afebrile males with urinary tract infections, comparing 7 versus 14 days. ¹⁰ Recently published clinical trials have evaluated the efficacy of shortened antibiotic regimens for uncomplicated Gram-negative bacteremia, showing non-inferiority in efficacy-related outcomes and a lower incidence of adverse effects compared to longer antibiotic regimens.^5,7,11–13

Current guidelines recommend a minimum of 14 days of antibiotic treatment for uncomplicated S. aureus bacteremia (SAB) ¹⁴ ; however, evidence supporting this recommendation is scarce and mainly based on small observational studies^15,16 with several methodological limitations. New evidence on relevant clinical and microbiological outcomes of shorter antibiotic regimens is emerging and may change this recommendation.

This systematic review evaluates whether shortened antibiotic treatment in adult patients with uncomplicated SAB differs from standard therapy (14 days or longer) in terms of efficacy and safety-related outcomes.

Methods

Protocol

The protocol was registered in the International Prospective Registry of Systematic Reviews: CRD42023440714.

Search strategy

We developed a search strategy to identify randomized clinical trials (RCTs) and non-randomized studies (NRS), comparing short versus usual or long antibiotic regimens for S. aureus uncomplicated bacteremia in MEDLINE, Embase, and the Cochrane Central Register of Controlled Trials (CENTRAL), from inception to June 2023. It included controlled vocabulary and free-text terms using field labels, Boolean, and proximity operators adapted for each search engine, without language restrictions. Search strategies are available in Supplemental Table 1. In addition, we searched for clinical trial registries in the International Clinical Trials Registry Platform of the World Health Organization; and we hand-searched a reference list of the selected studies.

Studies selection

Two authors (CD-B and SGP) independently reviewed the identified studies, initially by title and abstract and later by full text. Disagreements in the selection process were resolved by consensus or by involving a third reviewer (OMMV). The selection process is presented in the PRISMA (Preferred reporting items for systematic review and meta-analysis) flowchart (Figure 1).

Figure 1.

PRISMA flowchart.

*NRS, non-randomized studies.

**Only the abstract was found.

Studies were included if they considered hospitalized patients older than 18 years with uncomplicated SAB confirmed by at least one positive blood culture. Studies were excluded if they reported on polymicrobial bacteremia, sources of bacteremia requiring prolonged treatment, evidence of metastatic sites of infection, and prolonged bacteriemia.

Studies had to compare a short course with a longer or usual duration of antibiotic treatment, as defined by the author, versus the standard of care for the treatment of uncomplicated SAB at the time of the study.

In addition, studies had to report at least one of the following outcomes: in-hospital mortality from any cause, outpatient mortality from any cause at 30/90 days, recurrence of bacteremia, distant relapse (defined as evidence of metastatic sites of infection after completion of antibiotic treatment before 90 days), or adverse events related to antibiotic therapy. Bacteremia recurrence was determined by the presence of a new bloodstream infection of S. aureus after the initial 90-day proposed antibiotic regimen completion. If genetic testing was available, it had to confirm that the microorganism was identical to the initial one. Otherwise, it was classified as a reinfection after the 90-day threshold. Adverse events associated with antibiotics for MRSA and MSSA included acute kidney injury (as defined by KDIGO criteria), hypersensitivity reactions, infusion-related reactions, gastrointestinal symptoms, rash, phlebitis, chills, and less frequently neutropenia (<500/mm³).

Risk of bias assessment

We planned to assess the risk of bias for RCT using the ROB 2 tool ¹⁷ and for NRS using the ROBINS I tool ¹⁸ proposed by the Cochrane Collaboration. This assessment was performed by two investigators (CD-B and SGP) in parallel and independently, with discrepancies resolved by a third investigator (OMMV).

The NRS were categorized as having low, moderate, high, or critical risk of bias according to seven domains: bias due to confounding, bias in the selection of participants for the study, bias in the classification of interventions, bias due to deviations from the intended interventions, bias due to missing data, and bias due to the measurement of outcomes. ¹⁷

Data extraction

The following information was extracted for each study: author, country in which the study was conducted, study date, baseline characteristics of the population (including sex, age, prevalence of diabetes, intravenous drug use, immunosuppression, the presence of a catheter, hemodialysis, and patients with recent surgery), intervention and comparators, percentage of MRSA, most commonly used antibiotic, and the frequency of outcomes. If necessary, the corresponding authors were contacted to obtain missing data; if not available, the study was discarded.

Analysis

Meta-analysis was carried out using Review Manager software 5.4 version. ¹⁹ A random effects model was used, as proposed by Valentine for meta-analyses including NRS because it takes into account the unexplained heterogeneity of results associated with different sources of bias between studies. ²⁰ For dichotomous outcomes, results are presented as odds ratio (OR) with their 95% confidence intervals (CIs), displayed in forest plot figures.

However, we recognize that in these circumstances, the CI around a summary meta-analytic estimate represents only the identifiable statistical variation and does not fully reflect the uncertainty due to the unknown direction and magnitude of bias in each study. Therefore, we choose not only to perform the meta-analysis but also to include a description of possible confounding to allow the reader to make their assessment of heterogeneity. ²⁰

Clinical heterogeneity was assessed by analyzing the variability between studies due to differences in the characteristics of the participants, the interventions, the comparators, and the way the outcomes were measured. Statistical heterogeneity was assessed by visual inspection of the forest plot and the χ², I², and TAU ² statistical tests. A publication bias analysis was planned in case there were at least 10 studies evaluating a certain outcome, but it was not executed as this criterion was not fulfilled. ²¹ We planned a subgroup analysis differentiating short versus long according to the definition of the studies, and according to our definition of < or ⩾14 days. Likewise, a sensitivity analysis was performed excluding those with a high risk of bias.

Results

A total of 3146 individual studies were identified, of which 6 non-RCTs^22–27 with a total of 1700 patients were finally included. No RCTs met the inclusion criteria. Of the selected studies, four were retrospective cohorts^22,25–27 conducted in Switzerland, Denmark, the Netherlands, and the United States. Another study was a prospective cohort ²³ from South Korea, and finally one was a post hoc analysis ²⁴ from the United Kingdom. Zeylemaker et al. ²⁵ included three different cohorts with different moments of data collection, assessment of comorbidities, and sample size. A table with the inclusion and exclusion criteria for the individual studies has been addressed in Supplemental Table 2. The selection process is shown in the PRISMA diagram ²⁸ (Figure 1).

The characteristics of the studies and the patients included in each study are shown in Table 1. The median age was between 50 and 70 years, and the proportion of male patients was more than 60%. The proportion of patients with diabetes ranged from 12% to 30%, and the proportion of patients with immunosuppression was variable, ranging from 5% to 40% related to different definitions in each study. Hemodialysis and surgery were present in <10% of patients in most studies.

Table 1.

Included studies and patients’ characteristics.

Author, year, country	Type of study	Uncomplicated SAB definition	Short group a	Usual group a	Assessed outcomes	AB Most Used Name, %	N	Age, years, median IQR	Male%	MRSA%	DM%	IDU%	IS%	HD%	CX %	CVC %
Abbas, 2019, Switzerland	RC	Absence of: EC, IP. Fever <3 days ,SAB duration < 2 days.	⩽14	>14	Mortality 90 days, Relapse 90 days.	MRSA Vanc, 48.6 MSSA Flux 24.3	185	ST 74.3 (51.2–83.4) UT 70.4 (53.5–81.2)	ST 66.7 UT 66.6	ST 32 UT 15	ST 13.3 UT 13.9	ST 2.7 UT 5.4	ST 44.0 b UT 42.4 b	ST 9.3 UT 7.3	NR	ST 20 UT 12
Thorlacius-Ussing, 2020, Denmark d	RC	Absence of: EC, IP, OM, MG, AR, SP, PNEU. Positive BC <48 h AT.	CH1 8 (7–10) CH2 9 (8–10) CH3 8 (7–10)	CH1 14 (13–15) CH2 14 (13–15) CH3 13 (12–15)	Mortality 30 days , Mortality 90 days , Relapse 90 days.	NR	CH1 645 CH2 219 CH3 141	CH1 ST 71(58–81) UT 71(59–83) CH2 ST 53(49–77) UT 63(52–73) CH3 ST 62 (52–75) UT 70 (55–80)	CH1 ST 63 UT 64 CH2 ST 68 UT 66 CH3 ST 63 UT 57	CH1 ST 63 UT 64 CH2 ST 68 UT 66 CH3 ST 63 UT 57	CH1 ST 25 UT 24 CH2 ST 14 UT 20 CH3 NR	CH1 ST 4 UT 3 CH2 ST 5 UT 3 CH3 ST 2 UT 3	CH1 c ST 5 UT 6 CH2 c ST 19 UT 17 CH3 NR	CH1 ST 2 UT 3 CH2 NR CH3 ST 12 UT 12	CH1 ST 2 UT 2 CH2 ST 3 UT 4 CH3 ST 0 UT 1	CH1 ST 30 UT 21 CH2 ST 32 UT 26 CH3 ST 22 UT 14
Zeylemaker, 2001, Netherlands	RC	Absence of EC, OM, MA, SS.	⩽14	>14	Favorable outcome e Mortality due to complications ,Mortality due to underlying disease.	NR	44	57 (20–92)	47	NR	12.2	NR	30.6 f	NR	NR	100
Evans, 2022, United Kingdom	Post hoc	Absence of: EC, IP.	<10	10–18	Mortality 28 days	MRSA Vanc 77.2 MSSA Flux 78.0	291	ST 67(45–77) UT 66(51–76)	ST 60.9 UT 68.3	ST 16 UT 12.2	ST 25.4 UT 21.7	NR	NR	ST 9.6 UT 11.7	ST 8.6 UT 12.2	ST 12.2 UT 13.9
Chong, 2013, South Korea	PC	Absence of EC, MA ,Fever <3 days ,SAB duration <2–4 days.	<14	⩾14	Relapse 90 days ,Mortality 90 days	MRSA Vanc 85 MSSA Cfz 48.3	111	ST 63(51–73) UT 59(49–67)	ST 57.9 UT 61.6	ST 47.4 UT 47.9	ST 23.7 UT 24.7	NR	ST 18.4 f UT 16.4 f	ST 5.3 UT 13.7	NR	ST 34.2 UT 53.4
Taupin, 2020, United States	RC	Absence of EC, AR, OM, MA, PNEU, IP, Fever <3 days ,SAB duration <3 days.	<14	⩾14	Complicated bacteremia , Relapse and 90-day mortality.	NR	64	ST 60(51–70) GU 64(50–74)	ST 48 UT 53	ST 19 UT 37	ST 29 UT 40	ST 0 UT 2	ST 14 g UT 44 g	ST 0 UT 12	NR	ST 10 UT 47

^aThe results are displayed as days.

^bImmunosuppression includes any of the following: HIV (CD4 200/mm³), an end-stage renal disease requiring dialysis, insulin-dependent DM, active malignancy, cytotoxic chemotherapy 6 months prior, prednisone therapy 10 mg/day, child C cirrhosis, neutropenia 500/mm³, hematopoietic stem cell transplantation, solid organ transplantation.

^cImmunosuppression was defined as the use of immunosuppressive medication.

^dThis study has three cohorts which were named CH1, CH2, and CH3, respectively.

^eDefined as no complications and no death during the 1-year follow-up period.

^fImmunosuppression had no specific definition.

^gDefined as transplant recipients, active cancer, corticosteroid use, or use of other immunomodulatory agents.

AB, antibiotic; AR, arthritis; AT, after treatment initiation; BC, blood culture; Cfz, Cefazolin; CVC, central venous catheter; CX, surgery; DM, diabetes mellitus; EC, endocarditis; Flux, flucloxacillin; HD, hemodialysis; IDU, intravenous drug user; IP, implanted prosthetics; IQR, interquartile range; IS, immunosuppression; MA, metastatic abscesses; MG, meningitis; N, sample size; NR, unreported; OM, osteomyelitis; PC, prospective cohort; PNEU, pneumonia; RC, retrospective cohort; MRSA, methicillin-resistant Staphylococcus aureus; MSSA, methicillin-sensitive S. aureus; SAB, S. aureus bacteremia; SP, spondylodiscitis; SS, severe sepsis; ST, shortened treatment group; UT, treatment group as usual; Vanc, vancomycin.

The duration of the shortened treatment group was <14 days in most studies, but two counted as 14 days in this group^25,27 and one counted as short <10 days. ²⁴ The same study classified medium duration as 10–18 days and long duration as >18 days. The most commonly used antibiotic in both groups was vancomycin for MRSA and Flux for MSSA, except in one study ²³ where cefazolin was the most commonly used treatment.

Risk of bias

Some studies^22–25 did not evaluate confounding factors like patients’ clinical status or comorbidities. In addition, there were clinical and statistically significant differences in basal characteristics between groups, which could have resulted in selection bias. In the study by Taupin et al., ²² there were no differences in variables such as age, sex, sensitivity to oxacillin, and community origin, but there were differences in terms of immunosuppression and the presence of valvular heart disease, which were greater in the long-course group. In the study by Chong et al., ²³ no differences were found in both comparisons regarding age, sex, community origin, antibiotic treatment, MRSA isolation, and comorbidities such as diabetes mellitus (DM), immunosuppression, hemodialysis dependence, liver cirrhosis, prosthetic device, among others. The only difference was found in peripheral catheter-related infection, which was greater in the short-course arm. In the study by Evans et al., ²⁴ baseline characteristics were broadly similar between groups using standardized mean differences (SMD) to quantify disparities. The only variables for which the SMD was greater than 10% were estimated glomerular filtration rate (better renal function in patients in the short-course group) and source of infection (more often skin and soft tissue or uncertain in the short-course group). In the study by Zeylemaker et al., ²⁵ comorbidities such as DM, renal failure, and neoplasm were greater in the long-course group.

Some cases of complicated SAB could potentially have been misclassified as low-risk SAB in one study. ²⁶ Two studies^22,25 had a risk of bias due to missing data, where 7.8% and 12.3% of the participants, respectively, were excluded because of lack of available data. Finally, the risk of bias associated with the selection of the reported result was observed in one study, where 11 patients were excluded from analysis due to mortality during antibiotic therapy, with some having serious comorbidities. ²³

One study was assessed as low risk, ²⁷ four studies as moderate risk,^22–24,26 and one as serious risk of bias. ²⁵ The results of the risk of bias assessment are presented in Figure 2.

Figure 2.

Risk of bias assessment.

Mortality at 90 days

Five trials evaluated this outcome in a total of 1409 patients. An initial meta-analysis was performed using the authors’ definition of abbreviated versus standard therapy (Figure 3) and showed no statistical difference (OR: 1.09, 95% CI, 0.82–1.46, p: 0.55; I² = 0%). Subsequently, a subgroup analysis was performed defining shortened therapy as <14 days versus ⩾ 14 days, for a total of 3 studies with 1039 patients, also showing no statistical difference (OR: 0.84, 95% CI: 0.59–1.22, p = 0.36; I² = 0%). Finally, a sensitivity analysis excluding those with a high risk of bias was consistent with the results above (OR: 1.08, 95% CI: 0.80–1.45, p: 0.62; I² = 0%).

Figure 3.

Forest plot.

30-Day mortality

Only one study ²³ evaluated 30-day mortality in a total of 291 patients. Compared with the extended regimens, the shortened course antibiotic had higher mortality (OR: 7.82, 95% CI: 2.10–29.17) but the intermediate course (OR: 1.37, 95% CI: 0.38–4.97) did not reach statistical significance.

Relapse 90 days

This outcome was reported in five of the selected studies, with a total of 864 patients. The initial meta-analysis using the authors’ definition of shortened versus standard therapy (Table 1), yielded an OR of 0.72 (95% CI: 0.31–1.68, p: 0.45; I² = 26%). Subgroup analysis with a predefined length of therapy <14 days versus ⩾14 days, with three studies and a total of 820 patients found OR 1.03 (95% CI: 0.24–4.41, p: 0.97; I² = 45%). The sensitivity analysis without the high-risk study showed a similar result (OR: 1.03, 95% CI: 0.24–4.41, p: 0.97; I² = 45%).

Adverse events

Only one study reported associated adverse events. ²² In the shortened treatment group (<14 days), 5 adverse events occurred in 21 patients, 4 of which were due to Clostridium difficile infection, and 1 was not specified, while in the usual or long therapy group (⩾14 days), 7 events were documented in 43 patients, 1 Clostridium difficile infection and 6 not specified, with no statistically significant differences.

Discussion

The objective of this systematic review was to assess whether shortened versus standard antibiotic treatment differs in terms of mortality, recurrence of bacteremia, or occurrence of adverse events in adults with uncomplicated S. aureus bacteremia. In the six NRS included, we did not find significant differences in the odds of 90-day mortality or 90-day relapse or recurrence of bacteremia with the administration of shorter versus longer regimens. Only one study ²² reported adverse events at follow-up, with no significant differences between the two comparisons. No serious adverse events were reported in the included study.

To our knowledge, this is the first systematic review assessing the safety and efficacy of shortened versus extended antibiotic therapy in adults with uncomplicated S. aureus bacteremia. Regarding 30- and 90-day mortality, our results support the findings of a retrospective cohort study ²⁹ in which there was no difference in survival rate between patients with S. aureus bacteremia treated for more than 14 days compared to those receiving shorter antibiotic courses: That study was not included in our review because it did not differentiate between complicated and not complicated bacteremia but the results were similar. Similarly, in terms of relapse of bacteremia, our findings are in line with evidence from a previous study ³⁰ in which treatment for less than 14 days among patients with a short duration of bacteremia (less than 3 days) did not show an increase in relapse rate of bacteremia where also uncomplicated bacteremia was not well differentiated. It is worth noting that Abbas et al.’s ²⁷ study also evaluated mortality rates by comparing short versus usual duration antibiotic treatments for complicated S. aureus bacteremia. A mortality reduction was observed in the antibiotic treatment regimen lasting >14 days (hazard ratio: 0.32, 95% CI: 0.16–0.64). This finding implies that shorter therapy time may only be suitable for uncomplicated bacteremia.

We could identify a moderate statistical heterogeneity among the studies for the outcome of 90-day relapse of bacteremia (I² = 45%), with non-significant results (p: 0.97). We believe that these findings may be explained in part by the differences in the proportion of included patients with MRSA, DM status, and hemodialysis requirement in shortened compared to prolonged antibiotic therapy groups among the summarized studies.

An important finding of our review was the differences in the definitions used to classify shortened or prolonged duration of antibiotic therapy between studies. While the definition of standard treatment was more homogeneous among the included studies (14 days or longer) in all included studies except the study by Evans et al. ²⁴ (defined as 10–18 days), in the case of a shortened regimen, it was as short as 7–14 days of antibiotic treatment, with one study ²⁶ being 8–9 days. The fact that there is no uniform and universally accepted definition of what is considered a short or prolonged regimen underscores the need to standardize these definitions in further clinical trials to reduce the heterogeneity of the interventions and comparators evaluated and to draw more precise conclusions on this topic. There is a novel proposal for the classification and management of patients with S. aureus bacteremia to systematically identify those patients at high risk of developing infectious complications and treat them accordingly rather than classifying patients as having complicated or uncomplicated bacteremia. ³¹ It consists of four steps: an initial risk stratification for the presence of metastatic infection, a diagnostic work-up in search of these infections based on the initial risk assessment, a final clinical diagnosis, and finally a general direction for treatment that can be individualized according to favorable clinical features.

The results of this systematic literature review provide interesting data on the prescription of shorter antibiotic regimens compared with standard therapy in adults with uncomplicated S. aureus bacteremia. However, there are some considerations regarding our findings: First, only one study was assessed as overall low risk of bias. To mitigate this concern, a sensitivity analysis was performed including only low and moderate risk of bias studies, with similar findings. Second, the number of total events was limited, presented in 77 out of 864 patients for relapse of bacteremia at 90 days and in 266 out of 1409 patients for 90-day mortality. This could explain a potential insufficient precision of the analyzed outcomes, which implies the need to generate studies with a greater sample size to corroborate our results. Third, among the included studies, we could identify little representation of patients with chronic and debilitating diseases such as DM, chronic renal failure, and immunosuppressive states, which may contribute to the prognosis of bacteremia, It is important to mention that in some studies,^22–25 there were differences in basal clinical characteristics (DM, immunosuppression, renal failure) that could have contributed to selection bias. The difficulties in the lack of appropriate description of these variables could be explained by the retrospective nature of the included studies.

A limitation was that relatively few studies were available to be included in this meta-analysis, and some outcomes could only include a subset of these, which may limit our conclusions. Future studies are needed to overcome this limitation. Currently, there is an ongoing randomized, non-blinded interventional study (NCT03514446) that aims to evaluate if 7 days of antibiotic treatment in patients with uncomplicated S. aureus bacteremia is non-inferior to 14 days of treatment, which results may contribute to determining the best time duration of antibiotic therapy in these population. Nowadays, there is an urgent need for well-designed, controlled clinical trials with a robust sample size, evaluating clinical and microbiologic outcomes associated with S. aureus uncomplicated bacteremia to elucidate the best course of action regarding the optimal duration of therapy.

Conclusion

In summary, considering that no differences were found in the outcomes of 90-day mortality, 90-day relapse or recurrence of bacteremia, or the frequency of adverse events according to the length of therapy comparing shorter regimens with prolonged antibiotic therapy, shorter duration regimens could be considered as an alternative option in individual low-risk cases. Some of the included studies had important limitations regarding bias due to confounding factors and showed an imbalance in basal clinical characteristics such as DM, immunosuppression, and renal failure. This population remains underrepresented in the studies, which limits the generalizability of the results in these conditions. Because of the risk of bias, imprecision for several of the outcomes analyzed, and the few studies taken into account in this study, the benefits and harms of using shorter treatment regimens for uncomplicated S. aureus bacteremia should be analyzed with caution and on an individual basis.