Guideline-Concordant Therapy for Community-Acquired Pneumonia in the Hospitalized Population: A Systematic Review and Meta-analysis

Abstract Background A commonly used guideline for community-acquired pneumonia (CAP) is the joint American Thoracic Society and Infectious Diseases Society of America practice guideline. We aimed to investigate the effect of guideline-concordant therapy in the treatment of CAP. Methods We systematically searched MEDLINE, Embase, CENTRAL, Web of Science, and Scopus from 2007 to December 2023. We screened citations, extracted data, and assessed risk of bias in duplicate. Primary outcomes were mortality rates, intensive care unit (ICU) admission, and length of stay. Secondary outcomes were guideline adherence, readmission, clinical cure rate, and adverse complications. We performed random-effect meta-analysis to estimate the overall effect size and assessed heterogeneity using the I2 statistics. Results We included 17 observational studies and 82 240 patients, of which 10 studies were comparative and pooled in meta-analysis. Overall guideline adherence rate was 65.2%. Guideline-concordant therapy was associated with a statistically significant reduction in 30-day mortality rate (crude odds ratio [OR], 0.49 [95% confidence interval .34–.70; I2 = 60%]; adjusted OR, 0.49 [.37–.65; I2 = 52%]) and in-hospital mortality rate (crude OR, 0.63 [.43–.92]; I2 = 61%). Due to significant heterogeneity, we could not assess the effect of guideline-concordant therapy on length of stay, ICU admission, readmission, clinical cure rate, and adverse complications. Conclusions In hospitalized patients with CAP, guideline-concordant therapy was associated with a significant reduction in mortality rate compared with nonconcordant therapy; however, there was limited evidence to support guideline-concordant therapy for other clinical outcomes. Future studies are needed to assess the clinical efficacy and safety of current guideline recommendations.

Community-acquired pneumonia (CAP) is one of the leading causes of hospitalization and death affecting all age groups globally [1].Empiric antimicrobial therapy is the mainstay of treatment of the majority of patients with pneumonia.To prevent antimicrobial misuse, resistance, and complications, an individualized risk-benefit analysis and evidence-based selection of antibiotic regimen is needed.To address this, several practice guidelines for CAP have been published [2][3][4][5].
One of the commonly referenced CAP guidelines is the American Thoracic Society (ATS) and Infectious Diseases Society of America (IDSA) guideline, first published in 2007 and updated in 2019 [3,4].The updated guideline reaffirms many prior recommendations including a β-lactam plus a macrolide therapy or a respiratory fluoroquinolone monotherapy for nonsevere inpatient CAP management and a β-lactam plus a macrolide or fluoroquinolone dual therapy for severe inpatient CAP management, noting subtle differences in the indications for empiric coverage for methicillin-resistant Staphylococcus aureus (MRSA) and Pseudomonas between the guidelines (Table 1).
Several studies have associated the use of guidelineconcordant CAP therapy with improved mortality rates among patients with pneumonia [6][7][8][9].However, criticism has been raised about potential overtreatment and misuse of empiric broad-spectrum therapy based on MRSA and Pseudomonas risk factors alone, with discordance in the recommended therapy and local epidemiology and antimicrobial resistance patterns in different areas of the world [10].To date, there has been no comprehensive review of the clinical outcomes following ATS/IDSA guideline-concordant CAP therapy to our knowledge.Therefore, the current study aimed to Guideline-Concordant Therapy for Community-Acquired Pneumonia • OFID • 1 Open Forum Infectious Diseases M A J O R A R T I C L E systematically review the literature and meta-analyze all available studies to provide an evidence-based appraisal of the following clinical questions pertaining to the ATS/IDSA guideline-concordant care for adults hospitalized with CAP: (1) What are the outcomes of CAP treated with ATS/IDSA guideline-concordant therapy, including mortality rate, length of stay (LOS), and intensive care unit (ICU) admission?(2) What is the prevalence of ATS/IDSA guideline-concordant CAP treatment?And (3) What are the readmission and clinical cure rates and adverse complications with guidelineconcordant versus nonconcordant therapy?

Protocol Registration
This systematic review was registered in the Open Science Framework Registries (https://doi.org/10.17605/OSF.IO/ X8VFA) and followed the statement of Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) reporting guidelines (Supplementary Figure 1) [11].

Search Strategy
We performed a systematic search in MEDLINE (Ovid), Embase (Ovid), Cochrane Central Register of Controlled Trials (CENTRAL), Web of Science, and Scopus using a predefined search strategy that consisted of a Boolean combination of medical subject headings and keywords to identify studies on guideline-concordant therapies for CAP (filtered by English language and adult population, defined as age ≥18 years) (Supplementary Figure 2).We manually screened the bibliographies of relevant articles to identify potentially eligible studies that were not captured in the initial database search.We limited the search results to articles indexed from 2007 (ie, when the first joint CAP ATS/IDSA guideline was published) through 9 December 2023.

Study Eligibility Criteria
Included studies were (1) quantitative or epidemiological peerreviewed primary articles; (2) cross-sectional, observational cohort, longitudinal, case-controlled, quasi-experimental, and randomized controlled trials; (3) on adults (ie, ≥ 18 years of age) with diagnosed CAP requiring hospitalization; (4) concordant with the ATS/IDSA 2007 or 2019 CAP guidelines for the study intervention, as determined by study authors (ie, use of guideline recommended antimicrobials according to disease severity and risk factors); and (5) reporting on outcome measures of interest as listed below.We excluded studies that were (1) primarily focused on the pediatric population, (2) on hospitalacquired or ventilator-associated pneumonia, or (3) commentaries, case studies, reviews, editorial, letters to the editor, or opinion articles.

Study Selection
We used COVIDence (Veritas Health Information) as the primary screening tool.Citations were imported from each database and duplicates were removed.Two reviewers (C. S. and M. C.) independently screened the included articles in duplicate, and any discrepancies were resolved by consensus.A 2-step screening process was conducted.First, we performed a title-abstract screening based on the exclusion criteria and removed all articles that were deemed to be irrelevant for this review.Then we performed a full-text screening of the remaining articles and included articles that satisfied all the inclusion criteria.In both steps, we performed a calibration on the first 20 articles to ensure satisfactory interrater reliability (>80%).a Doxycycline can be used as an alternative to the macrolide.
b Doxycycline is an alternative to the macrolide and fluoroquinolone.

Data Extraction
We used a standardized data collection table to extract relevant data on study characteristics (ie, study identifier, design, time frame, country, and guideline edition), patient population (ie, population size, age, and sex), CAP outcome measures including adjusted variables for overall effect estimates when available.Two reviewers (C. S. and M. C.) independently extracted data in duplicate, and any discrepancies were resolved by consensus.This study did not include factors necessitating patient consent.

Outcomes
Primary outcomes of this study included the mortality rate (ie, in-hospital and 30-day rates), hospital LOS, and ICU admission rate.Secondary outcomes included clinical cure rate, readmission rate, any complications related to guideline-concordant and nonconcordant therapies, and overall adherence rate of guideline-concordant therapy in treating CAP.

Risk of Bias Assessment and Certainty of Evidence Assessment
We assessed the risk of bias of individual studies using the Risk Of Bias In Nonrandomized Studies-of Exposure (ROBINS-E) tool [12].Two reviewers (C. S. and M. C.) independently appraised the risk of bias of included studies and any discrepancies were resolved by consensus.We assessed the overall evidence level of outcomes using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) tool [13].
The overall evidence was summarized as very low, low, moderate, or high.

Statistical Analysis
Due to anticipated heterogeneity between studies, we performed meta-analyses using a random-effect model when ≥2 studies reported on the same outcome measures to calculate a pooled odds ratio (OR) and 95% confidence interval (CI) values.We pooled the data by computing ORs when the number of events were reported; otherwise, we pooled the logarithm of the ORs using an inverse variance approach.We performed separate meta-analyses for studies that reported adjusted ORs (aORs) as opposed to crude ORs.For continuous variables (ie, LOS), we converted individual effect sizes into standardized mean differences and corresponding 95% CIs to meta-analyze.We planned to perform subgroup analyses of the primary outcomes according to the different ATS/IDSA guideline editions, disease severity (eg, CURB-65 and the pneumonia severity index), and general ward versus ICU destination.However, due to a lack of studies reporting outcomes using the 2019 ATS/ IDSA guideline and disease severity, only a subgroup analysis of ward versus ICU destination was performed.
We assessed the heterogeneity among included studies using I 2 statistics with the thresholds of 25%, 50%, and 75% representing low, medium, and high heterogeneity respectively [14].When there was considerable heterogeneity (ie, I 2 ≥ 50%), we performed a sensitivity analysis by comparing studylevel characteristics and repeating the analysis using a subset of the data that minimized variations between the included studies (ie, high-quality vs low-quality studies, prospective cohort vs retrospective cohort, and leave-one-out analysis).If there remained persistently significant heterogeneity, we performed narrative synthesis instead to report the outcomes.
To assess publication bias, we first visually examined funnel plots, and if >10 studies were included in the analysis, we used the Egger test of regression and Begg and Mazumdar's rank correlation test to determine the significance of funnel plot asymmetry [15][16][17].We considered differences to be statistically significant at P ≤ .05.All statistical analyses and metaanalyses were performed using the Cochrane Collaboration's RevMan Web and R software (version 4.3.2).

Search Results
The electronic search of the aforementioned databases retrieved 9748 citations, of which 5263 remained after duplicates were removed (Figure 1).After screening for title-abstract, a total of 5212 articles were excluded.No additional articles were identified from the manual bibliography screen.The full texts of the remaining 51 articles were assessed for inclusion.In total, 17 articles met the inclusion criteria for this review [6-9, 18-30], of which 10 were comparative and fulfilled criteria for meta-analysis [6-8, 20, 21, 24, 25, 27, 29, 30].There was high interrater reliability (κ > 95%) in all steps of the screening process.

Study Characteristics
The characteristics of the included studies are collated (Table 2).Studies were universally observational in nature (15 retrospective cohort, 1 prospective cohort, and 1 casecontrol study), with a total patient population of 82 240.Fourteen studies followed the 2007 and 3 the 2019 ATS/IDSA guideline.Of these, 5 studies assessed the impact of guidelineconcordant therapy in an ICU setting.Only 1 study assessed the clinical impact of guideline-concordant therapy in patient populations at high risk for MRSA [23].Studies were conducted mainly in the United States (n = 5), Canada (n = 2), Japan (n = 2), South Korea (n = 1), Taiwan (n = 1), New Zealand (n = 1), Saudi Arabia (n = 1), and Spain (n = 1).Three studies were conducted in Europe.

Risk of Bias and Publication Bias
The risk of bias assessment for the included studies is reported in Supplementary Figure 3. Overall, 5 studies were deemed low risk, 10 studies were deemed to have some concerns for risk of bias, and 2 studies were deemed to have high risk of bias.We did not detect strong evidence of publication bias based on funnel plot symmetry (Supplementary Figure 4).Because there were <10 individual studies in each meta-analysis, we could not perform the Egger test of regression or Begg and Mazumdar's rank correlation test to quantitatively assess publication bias.
In a sensitivity analysis excluding low-quality studies, there was persistently significant association between guidelineconcordant therapy and reduced 30-day and in-hospital mortality rates (Supplementary Figure 6).A similarly strong association was found in a leave-one-out analysis for 30-day mortality rate.However, when the study by McCabe et al [6] was excluded in a leave-one-out analysis for in-hospital mortality rate, there was no longer a significant association, although the general trend remained the same (OR, 0.56 [95% CI, .31-1.02];P = .06;I 2 = 61%).

ICU Admission
Three studies reported outcomes on ICU admission when patients were treated with guideline-concordant versus nonconcordant therapy.Individually, Grenier et al [8] and Sims et al [30] reported statistically significant reductions in ICU admission when patients were treated with guideline-concordant therapy (OR, 0.17 [95% CI, .10-.29] and 0.45 [.27-.75], respectively) while Cilloniz et al [21] found no significant association (1.42 [.96-2.08]).When the outcomes from the 3 studies were combined, there was no significant association between the receipt of guideline-concordant therapy and ICU admission, barring significant heterogeneity between the studies (I 2 = 95%) (Supplementary Figure 6).Insignificant effect size and high heterogeneity persisted even after leave-one-out sensitivity analysis was performed (Supplementary Figure 7).

Certainty of Evidence Assessment
There were serious or very serious concerns with 2 GRADE domains (risk of bias [ROB], inconsistency) across the 4 studies that reported 30-day mortality rates (Supplementary Figure 8).The overall certainty of the evidence was very low for guidelineconcordant therapy on reducing 30-day mortality rates among hospitalized patients with CAP.There were further serious or very serious concerns with 3 GRADE domains (ROB, inconsistency, imprecision) across the 5 studies that reported inhospital mortality rates.The overall certainty of evidence was similarly very low for guideline-concordant therapy on reducing in-hospital mortality rates among hospitalized patients with CAP.Finally, there were serious or very serious concerns with 4 GRADE domains (ROB, inconsistency, indirectness, imprecision) across 3 and 4 studies that reported ICU admission and hospital LOS, respectively.The overall certainty of evidence was very low for both outcomes.

DISCUSSION
In this systematic review of 17 observational studies and metaanalysis of 10 comparative studies enrolling 59 621 patients hospitalized for CAP, we found a statistically significant reduction in 30-day mortality rate for non-ICU patients and inhospital mortality rate for both non-ICU and ICU patients when they were treated with ATS/IDSA guideline-concordant therapy compared with nonconcordant therapy.However, there were serious concerns with ≥2 GRADE domains, resulting in a very low certainty of evidence for guideline-concordant therapy for all 4 primary outcomes.Moreover, due to significant between-study heterogeneity and limited available evidence, no definitive comparative conclusion could be made for hospital LOS, ICU admission, readmission and clinical cure rates, and adverse complications between guidelineconcordant and nonconcordant therapies.
Although the pooled OR is suggestive of the overall mortality benefit of guideline-concordant therapy, administration of empirical anti-MRSA therapy in high-risk patients as recommended by the ATS/IDSA CAP guidelines was counterintuitively associated with an increased 30-day mortality risk [23].The observation that guideline-concordant broad-spectrum antibiotic treatment is associated with higher mortality rates and adverse complications has been also reported in other patient populations, such as those with healthcare-associated pneumonia and community-onset sepsis [31][32][33].Notably, since publication of the ATS/IDSA pneumonia guidelines [3,4] and the Surviving Sepsis Campaign guidelines [34], the use of broadspectrum antibiotics (ie, vancomycin and piperacillintazobactam) has increased, yet the rates of isolation of multidrug-resistant pathogens, including MRSA and Pseudomonas, and overall clinical outcomes of pneumonia have remained unchanged [25,35,36].
In this context, the clinical validity of the ATS/IDSA CAP guidelines' recommendation of initiating empirical broadspectrum antibiotics in patients at high risk for MRSA and Pseudomonas is elusive.Previous studies have shown that pathogen-directed antibiotic treatment in patients with CAP is feasible and noninferior to empirical broad-spectrum antibiotics [37,38].Further research is needed to assess the clinical benefits and harms of empirical broad-spectrum antibiotics in high-risk patient populations to identify effective antibiotic decision-making strategies for patients admitted with CAP and subsequent postempirical treatment de-escalation.
Our findings did not identify convincing evidence supporting guideline-concordant therapy for other clinical measures, including hospital LOS, ICU admission, readmission and clinical cure rates, and adverse complications.While this highlights a paucity in available data pertaining to clinical outcomes following guideline-concordant therapy, we also noted considerable heterogeneity between the few studies that did report such outcomes.One possible explanation of the observed heterogeneity between the studies may be in part due to different local prevalence of bacterial pathogens responsible for CAP.For instance, Peto et al [39] reported in their systematic review of 48 studies that Streptococcus pneumoniae was found in much higher numbers of hospitalized patients with CAP in Europe (25.9%)compared with Asia (13.3%).Conversely, gram-negative enteric bacteria and S aureus were more commonly implicated in hospitalized CAP cases in Asia (9.0% and 4.0%, respectively) than in Europe (2.7% and 1.4%, respectively).

Figure 1 .
Figure 1.PRISMA flow diagram of the study screening process.Abbreviations: ATS, American Thoracic Society; CAP, community-acquired pneumonia; IDSA, Infectious Diseases Society of America.

Table 1 . Comparison of Recommendations in the 2007 and 2019 American Thoracic Society/Infectious Diseases Society of America CAP Guidelines
Abbreviations: ATS, American Thoracic Society; ICU, intensive care unit; IDSA, Infectious Diseases Society of America; MRSA, methicillin-resistant Staphylococcus aureus.