Treatment of Mycoplasma Pneumonia: A Systematic Review Eric Biondi, Russell McCulloh, Brian Alverson, Andrew Klein, Angela Dixon and Shawn Ralston Pediatrics 2014;133;1081; originally published online May 26, 2014; DOI: 10.1542/peds.2013-3729 The online version of this article, along with updated information and services, is located on the World Wide Web at: http://pediatrics.aappublications.org/content/133/6/1081.full.html PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly publication, it has been published continuously since 1948. PEDIATRICS is owned, published, and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk Grove Village, Illinois, 60007. Copyright © 2014 by the American Academy of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275. Downloaded from pediatrics.aappublications.org by guest on September 9, 2014 REVIEW ARTICLE Treatment of Mycoplasma Pneumonia: A Systematic Review AUTHORS: Eric Biondi, MD,a Russell McCulloh, MD,b Brian Alverson, MD,c Andrew Klein, BS,a Angela Dixon, BSN, MLS, AHIP,a and Shawn Ralston, MDd aDepartment of Pediatrics, University of Rochester, Rochester, New York; bDepartment of Pediatrics, Children’s Mercy Hospitals & Clinics, Kansas City, Missouri; cDepartment of Pediatrics, Hasbro Children’s Hospital, Providence, Rhode Island; and dDepartment of Pediatrics, Children’s Hospital at Dartmouth– Hitchcock, Hanover, New Hampshire KEY WORDS pediatric, pneumonia, macrolide, azithromycin, atypical pneumonia, mycoplasma ABBREVIATIONS CA-LRTI—community-acquired lower respiratory tract infection CAP—community-acquired pneumonia CI—95% conﬁdence interval IDSA—Infectious Diseases Society of America RCT—randomized controlled trial URTI—upper respiratory tract infection Dr Biondi conceptualized and designed the review, reviewed articles, ran the meta-analysis, and drafted the original manuscript; Dr McCulloh conceptualized and designed the review, reviewed articles, helped to run the meta-analysis, and drafted the original manuscript; Dr Alverson participated in the design, helped interpret the included studies, and critically reviewed the manuscript; Mr Klein participated in study design, tabulated articles, and reviewed the manuscript; Ms Dixon participated in study design, helped perform the literature review, and critically reviewed portions of the manuscript; Dr Ralston participated in study design, supervised Drs Biondi and McCulloh in study interpretation and translated 1 study, and participated in the meta-analysis; and all authors approved the ﬁnal manuscript as submitted. www.pediatrics.org/cgi/doi/10.1542/peds.2013-3729 doi:10.1542/peds.2013-3729 Accepted for publication Feb 21, 2014 Address correspondence to Russell McCulloh, MD, Pediatric Infectious Diseases, Children’s Mercy Hospitals & Clinics, 2401 Gillham Rd, Kansas City, MO 64108. E-mail: [email protected] abstract BACKGROUND AND OBJECTIVE: Children with community-acquired lower respiratory tract infection (CA-LRTI) commonly receive antibiotics for Mycoplasma pneumoniae. The objective was to evaluate the effect of treating M. pneumoniae in children with CA-LRTI. METHODS: PubMed, Cochrane Central Register of Controlled Trials, and bibliography review. A search was conducted by using Medical Subject Headings terms related to CA-LRTI and M. pneumoniae and was not restricted by language. Eligible studies included randomized controlled trials (RCTs) and observational studies of children #17 years old with conﬁrmed M. pneumoniae and a diagnosis of CA-LRTI; each must have also compared treatment regimens with and without spectrum of activity against M. pneumoniae. Data extraction and quality assessment were completed independently by multiple reviewers before arriving at a consensus. Data were pooled using a random effects model. RESULTS: Sixteen articles detailing 17 studies were included. The most commonly selected primary outcome was symptomatic improvement. Nine studies examined M. pneumoniae treatment in CA-LRTI secondary to M. pneumoniae, and 5 RCTs met criteria for meta-analysis. The suggested pooled risk difference of 0.12 (95% conﬁdence interval, 20.04 to 0.20) favoring treatment was not signiﬁcantly different and demonstrated signiﬁcant heterogeneity. Limitations included substantial bias and subjective outcomes within the individual studies, difﬁculty interpreting testing modalities, and the inability to correct for mixed infections or timing of intervention. CONCLUSIONS: We identiﬁed insufﬁcient evidence to support or refute treatment of M. pneumoniae in CA-LRTI. These data highlight the need for well-designed, prospective RCTs assessing the effect of treating M. pneumoniae in CA-LRTI. Pediatrics 2014;133:1081–1090 PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275). Copyright © 2014 by the American Academy of Pediatrics FINANCIAL DISCLOSURE: The authors have indicated they have no ﬁnancial relationships relevant to this article to disclose. FUNDING: No external funding. POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no conﬂicts of interest to disclose. COMPANION PAPER: A companion to this article can be found on page 1124, and online at www.pediatrics.org/cgi/doi/10.1542/peds. 2014-0871. PEDIATRICS Volume 133, Number 6, June 2014 Downloaded from pediatrics.aappublications.org by guest on September 9, 2014 1081 Community-acquired pneumonia (CAP) accounts for .150 000 pediatric hospitalizations each year in the United States, and there are great disparities in patterns of care and outcomes.1 Speciﬁcally, wide variations in antibiotic prescribing practices exist among physicians, often because the causative agent is not identiﬁed.2 For this reason, recent evidence-based pediatric CAP practice guidelines published by the Infectious Diseases Society of America (IDSA) recommend that children hospitalized with CAP be tested for Mycoplasma pneumoniae.3 M. pneumoniae is a common cause of CAP and other community-acquired lower respiratory tract infections (CA-LRTIs), particularly in school-age children and adolescents, but there are large gaps in our understanding of this disease.3,4 Prevalence estimates vary from 10% to 40% in pediatric CA-LRTI,5–9 and few studies address treatment recommendations.4,10 Additionally, the IDSA guidelines target the use of macrolides in CA-LRTI as an area needing additional research.3 First-line treatment of children hospitalized with CAP currently includes a b-lactam to treat common causative bacterial agents such as Streptococcus pneumoniae and a macrolide to provide additionally treatment of atypical pathogens such as M. pneumoniae.3 Studies of antibiotic use in the pre– IDSA guidelines era showed marked variability in prescription practices,2 probably because the efﬁcacy of macrolides in the treatment of M. pneumoniae remains unclear and treatment recommendations, even in major textbooks, are variable.4,10 The available evidence is also conﬂicting. One small trial suggests that b-lactam use in children with CAP is more costeffective than macrolides,11 several randomized controlled trials (RCTs) suggest no difference in efﬁcacy,12,13 and 2 recent pediatric cohort studies suggest that b-lactam–macrolide com1082 bination therapy decreased length of stay by 20% to 30% over b-lactam monotherapy.2,4 Data on antibiotic treatment of pediatric CA-LRTI caused by M. pneumoniae are generally considered inconclusive,3,4,10,14 yet macrolides remain the most commonly overprescribed antibiotic at pediatric clinics in the United States: .6 million annual doses for respiratory symptoms without a clear indication.15 A Cochrane review on the use of antibiotics to treat CA-LRTI secondary to M. pneumoniae in children found insufﬁcient evidence to draw any speciﬁc conclusions about the efﬁcacy of antibiotics for the condition.10 However, this review omitted at least 3 RCTs (roughly one-third of the total RCTs) for which there appeared to be applicable data on the topic,12,16,17 and data from observational studies were omitted. The objective of our study was to provide a more comprehensive review of all available published literature on the use of antibiotics in children to treat CA-LRTI secondary to M. pneumoniae. METHODS Search Design This was a systematic review of all observational and randomized trials comparing antibiotics with spectrum of activity for M. pneumoniae (eg, macrolide, tetracycline, or quinolone class) with placebo or antibiotics from any other class without spectrum of activity against M. pneumoniae in children ,18 years of age with CA-LRTI. The review was conducted in concordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement. Outcome Types The primary outcome was clinical improvement or cure at follow-up. Clinical improvement or cure could include resolution of fever; resolution or improvement in symptoms such as cough, congestion, shortness of breath, fatigue, or chest pain; or improvement or cure as deﬁned by the authors of the individual studies. Literature Search With the assistance of a librarian (A.D.), we performed a comprehensive search of PubMed (January 1966–August 2012) with the Medical Subject Headings terms and keywords (Table 1). After duplicates were removed, we used inclusion and exclusion criteria to select studies based on their title and abstract to include in our review. Additional studies were identiﬁed through manual search of the bibliographies of qualifying studies. Before manuscript submission, a second search was performed (August 2012–September 2013) to ensure that the review was as up-todate as possible and any new studies were included in the analysis. Study Selection Studies were considered eligible for inclusion in the review if they met the TABLE 1 Medical Subject Headings Used for Primary Literature Search Mycoplasma pneumoniae pneumonia, mycoplasma respiratory tract infections pneumonia recurrent respiratory tract infectiona respiratory infectiona RRTI community-acquired infections lower respiratory tract infectiona lower respiratory infectiona a pediatrica infant child child, preschool adolescent Represents open-ended term. BIONDI et al Downloaded from pediatrics.aappublications.org by guest on September 9, 2014 antibiotica anti-bacterial agents macrolidea roxithromycin erythromycin telithromycin macrolides azithromycin clarithromycin REVIEW ARTICLE following criteria: randomized or observational studies, included children #17 years of age, included $10 children who were diagnosed with CA-LRTI secondary to laboratory-conﬁrmed M. pneumoniae, data were provided for at least 1 outcome measuring improvement, and the study compared any antibiotic with spectrum of activity against M. pneumoniae with either placebo or another antibiotic without activity against M. pneumoniae. Studies that examined children with chronic respiratory illnesses such as cystic ﬁbrosis, bronchiectasis, bronchopulmonary dysplasia, congenital heart disease, or immunodeﬁciency were excluded, as were those that examined children with nosocomial and congenital infections. Studies were considered eligible for meta-analysis if, in addition to meeting criteria for inclusion in the qualitative review, information was either provided or could be extrapolated regarding the total number and outcomes of patients with M. pneumoniae within each treatment group. A subgroup analysis of the initial meta-analysis was performed that included only the RCTs. Identiﬁcation of Trials and Data Extraction The principal investigators (E.B. and R.M.) independently screened each citation identiﬁed through the initial search strategy as deﬁnitely, possibly, or clearly not meeting inclusion criteria. Full-text articles of all studies definitely or possibly meeting inclusion criteria were obtained regardless of primary language of the publication. Reviewers (E.B. and R.M.) independently reviewed each full-text article and then reached consensus before data abstraction. For full-text articles meeting study criteria, reviewers (E.B. and R.M.) extracted data for each study independently and then discussed their ﬁndings to form a consensus. Discrepancies were resolved through discussion with the collaborating reviewers (B.A. and S.R.). When necessary, the principal reviewers attempted to contact study authors to verify methods and extracted data, although no responses were received. Validity and Quality Assessment Studies were assessed and graded for level of evidence based on criteria published by the Oxford Center for Evidence-Based Medicine.18 An 18-item study quality assessment tool (Supplemental Information) was developed based on recommendations published by the Agency for Healthcare Research and Quality.19 Items on the quality assessment tool fell within 6 domains: subject selection, study performance bias, detection bias, subject attrition, reporting bias, and ﬁnancial conﬂict of interest. Items were marked as factor present, not present, or unclear based on review of the manuscript. Each study was given a total quality score, and for this purpose, labels of “not present” or “unclear” were grouped together. Studies were considered to be unblinded when it was not speciﬁcally stated that participants or investigators were blinded. Three reviewers (E.B., R.M., and B.A.) performed individual assessments on all included studies independently and then discussed each until all 3 reviewers were in agreement. Statistical Analyses Because much of the information relevant to this review was extrapolated from data presented in the included studies, the review authors attempted to calculate statistical signiﬁcance wherever possible using x 2 or Fisher’s exact tests. In cases for which M. pneumoniae and Chlamydia pneumoniae were grouped together, a “worstcase scenario” was performed by the review authors that assumed all C. pneumoniae were in the “improved on treatment” group.20 These were then subtracted from the total. This allowed an interpretation of the results that would ensure that, if a treatment effect was identiﬁed, it was not secondary to C. pneumoniae. For comparison, the analysis was also performed assuming that C. pneumoniae and M. pneumoniae responded equally to treatment. Treatment effects for dichotomous outcomes were calculated as risk difference using random effects modeling. Study heterogeneity was assumed for a P ,.10 and I2 $25%. If the primary outcome (clinical improvement) was assessed in separate methods and separate time periods, each set of outcomes was included in the review and meta-analysis. Meta-analysis was performed by using Comprehensive Meta-Analysis Version 2 (Biostat, Englewood, NJ). RESULTS The initial database search identiﬁed 4667 articles. One other citation was identiﬁed via bibliography review.4 Most articles were excluded based on a preliminary screen of the abstracts, leaving 40 full-text articles to be assessed for eligibility. Before submission, a second search was performed (August 2012–September 2013) to ensure that no recent articles were missed in the review. This search revealed 275 new articles, 1 of which was assessed for full-text eligibility but did not meet study criteria.2 Of the 4943 total citations reviewed, 16 met criteria for inclusion in the qualitative synthesis.5,12,13,16,17,20–30 One RCT20 used different outcome metrics at 2 different times and, for the purposes of this review, was considered 2 separate RCTs. Therefore, 17 studies were included. Four citations were non-English and were translated by the review authors (2 in Spanish,23,30 1 in French,21 and 1 in German29). Eight studies provided PEDIATRICS Volume 133, Number 6, June 2014 Downloaded from pediatrics.aappublications.org by guest on September 9, 2014 1083 data20,22,27,28 or enabled the review authors (E.B. and R.M.) to extract data13,16,23 speciﬁcally comparing treatment with no treatment of M. pneumoniae in children diagnosed with M. pneumoniae and were included in the quantitative analysis; 5 were RCTs (Fig 1).13,16,20,23 There were 4294 patients enrolled in the 17 included studies (Table 2). Of these, an aggregate 2648 patients could be used to compare an agent treating M. pneumoniae with an agent that did not treat M. pneumoniae in CA-LRTI. This number excludes comparator arms in which antibiotics of the same class or spectrum were compared and patients in whom therapy could not be determined. One randomized trial provided data on a total of 155 patients but compared macrolide therapy with either a b-lactam or another macrolide and did not provide separate statistics.30 Ten RCTs12,13,17,20,23,24,26,29,30 and 3 cohort studies5,22,25 compared a macrolide with a nonmacrolide in the treatment of CA-LRTI. One RCT16 compared levoﬂoxacin with a b-lactam, 1 retrospective cohort study21 compared M. pneumoniae spectrum with non–M. pneumoniae spectrum antibiotics, and 2 case–control studies27,28 compared macrolide treatment in patients with macrolide-sensitive and macrolideresistant M. pneumoniae. Esposito et al 20 included upper respiratory infections (URTIs) (199/352, 57%) in addition to CA-LRTI (153/352, 43%) but did not provide a separate analysis, and therefore all are included in this review, as was done previously.10 One study did not report the total number of pediatric patients.17 Studies used various combinations of methods for M. pneumoniae detection; 5 used culture of nasopharyngeal specimens,17,23,24,27,28 15 used serology,5,12,13,16,17,20–26,28–30 and 6 used polymerase chain reaction assays of upper airway specimens.12,20,23,25,26,28 No trials were placebo controlled, and a number of studies received pharmaceutical company funding.12,13,16,20,24,25 When we evaluated for overall quality among all studies included, detection bias, which included items such as validity and reliability of study outcomes, was the most frequently iden- tiﬁed bias category (12/17 studies, 71%) (Table 3). The most common cause of study performance bias was unblinded participants or observers. More than half of the RCTs (6/11, 55%) reported a potential conﬂict of interest, which in all cases was pharmaceutical sponsorship with use of that company’s drug in the study. Spectrum-Speciﬁc Treatment in Pediatric CA-LRTI Secondary to M. pneumoniae Nine studies5,13,16,20,22,23,27,28 provided enough detail to allow a comparison of M. pneumoniae spectrum and nonspectrum treatment in pediatric patients with CA-LRTI secondary to M. pneumoniae and included a total of 723 such patients (Table 4). In terms of overall clinical improvement, 4 of 5 RCTs found no clinical beneﬁt.13,16,20,23 One retrospective cohort study identiﬁed a signiﬁcant improvement in duration of fever in patients treated with macrolides, although clinically there was a large overlap in fever duration range (3.0–6.8 vs 3.4–7.9 days; P = .04), and there was not a signiﬁcant FIGURE 1 Flowchart for included studies. a One article20 used 2 different outcome metrics at 2 different time periods and was therefore treated as 2 separate studies. 1084 BIONDI et al Downloaded from pediatrics.aappublications.org by guest on September 9, 2014 Retro, multisite cohort/5 y RCT/2 y Pro, cohort/2 y Multisite, RCT/1.5 y Multisite, RCT/2 y RCT/2 y Multisite, RCT/3 y Pro, multisite cohort/1 y RCT/3 y RCT/2 y Multisite, RCT/2 y Retro, cohort/2 y Retro, case– control/4 y Garo et al (1988) Gomez Campdera et al (1996) Gendrel et al (1997) Harris et al (1998) Sáez-Llorens et al (1998) Wubbel et al (1999) Ferwerda et al (2001) Principi et al (2001) Esposito et al (2005), study A Esposito et al (2005), study B Bradley et al (2007) Lu et al (2008) Matsubara et al (2009) RCT/2 y RCT/1 y Ruhrmann et al (1982) Kogan et al (2003) RCT/1.5 y Design/Length Sterner et al (1967) Study PEDIATRICS Volume 133, Number 6, June 2014 Downloaded from pediatrics.aappublications.org by guest on September 9, 2014 N = 139 (8 mo–12 y) with MP CA-RTI N = 94 (0–14 y) with MP CAP treated with macrolide N = 352 (1–14 y) with recurrent RTI N = 352 (1–14 y) with recurrent RTI N = 738 (6 mo–16 y) with CAP N = 613 (2–14 y) with CAP N = 110 (30 d–14 y) with CAP N = 118 (3 mo–12 y) with CAP N = 174 (6 mo–16 y) with CAP N = 335 (6 mo–15 y) with CAP N = 104 (1.5–13 y) with CAP N = 456 (6 mo–16 y) with CAP N = 155 (6 mo–16 y) with CAP N = 120 (6 mo–14 yr) with CAP N = 182 (mean age = 29 y) with MP N = 180 (unknown ages) with CAP Patients TABLE 2 Study Characteristics and Quality, in Chronological Order Macrolide sensitive MP/ resistant MP Macrolide/no macrolide AZM and symptomatic/ symptomatic AZM and symptomatic/ symptomatic Levoﬂoxacin/b-lactam (,5 yr); macrolide ($5 yr) Classic CAP: AZM/amoxicillin Macrolide/no spectrum AZM/co-amoxiclav AZM/co-amoxiclav (,5 yr); EES ($5 yr) AZM/co-amoxiclav (,5 yr); EES ($5 yr) AZM/co-amoxiclav (,5 yr); EES ($5 yr) Macrolide/b-lactam AZM/co-amoxiclav (,5 y); EES (.5 y) 1. Excellent or good clinical response 2. Mean fever duration 3. Cough duration Duration of fever Improved or cured at 1–3 d and cured at 10–17 d #2 RTIs at 6 mo Clinical response: Fever duration CXR .75% improved Clinical success at 4–6 wk 1. Cured or improved at 10–13 d 2. Cured or improved at 25–30 d 3. Adverse events Cured or improved at 4–6 wk 1. Cured 3 d after treatment 2. Adverse events 1. Cured or improved at 15–19 d 2. Cured or improved at 4–6 wk 3. Adverse events Cured or improved after 3 d of treatment Fever at 2–18 d Cured or improved at 3, 10, or 30 d Fever duration Fever duration EES/amoxicillin MP spectrum/no spectrum Failure = daily fevers for 5 d or fever $10 d Relevant Outcomes EES/cephaloridine Intervention/Comparator 76/76 (AZM) vs 88/114 (comparator) with MP or CP had clinical success (P , .001). 53/71(AZM) vs 61/109 (comparator) with MP or CP #2 RTIs (P = .01).a Success was 409/441(levoﬂoxacin) vs 138/147 (comparator) at 1–3 d and 439/503 vs 145/170 at 10–17 d. Mean fever = 4.90 6 1.89 d (macrolide) vs 5.63 6 2.22 d (no macrolide) (P = .04). 43/47 (sensitive) vs 5/22 (resistant) had positive clinical response (P , .01). Mean fever = 1.5 vs 4.0 d (P , .01) and mean cough = 1.5 vs 4.0 d (P , .01). 2b 8/87 (EES) vs 17/93 (cephaloridine) failures.a Mean fever = 2.7 vs 3.8 d. In those with MP, 0/13 vs 3/9 failures.a Does not separate adults and children. Mean fever = 2.6 6 4.1 d (EES) vs 2.4 6 1.9 d (amoxicillin). Mean fever = 3 d (spectrum) vs 7 d (no spectrum). 84% presented with atypical pneumonia. Does not separate adults and children. At 3, 10, and 30 d, respectively: 78/82 (AZM) vs 66/73 (comparator),a 80/82 vs 68/73,a and 80/82 vs 69/73.a No difference in duration of fever or cough. 15 total with MP. In children with MP: 11/11 (macrolide) vs 2/32 (b-lactam) became afebrile (P , .001).a In ,5 y group: clinical success in 114/125(AZM) vs 59/63 (co-amoxiclav) at 15–19 d,a and 97/114 vs 41/48 at 4–6 wk.1 35/310 vs 45/146 had adverse events (P , .001).a 96/97 (AZM) vs 114/116 (co-amoxiclav) cured or improved. In cases of MP, 9/9 and 5/5 cured or improved.a 68/69 (AZM) vs 75/78 (comparator) cured. 10/69 (AZM) vs 33/49 (co-amoxiclav) had adverse events (P , .001).a Success in 50/55 (AZM) vs 46/53 (co-amoxiclav) at 10–13 d and 46/51 vs 43/50 at 25–30 d. 33/59 vs 41/58 adverse events.a In subjects with MP or CP, success in 106/109 (macrolide) vs 67/82 (comparator) (P , .001).a Mean fever = 1.7 d (AZM) vs 2.0 d (amoxicillin). By day 14 all 47 CXRs had improved by .75%. 3b 2b 2b 2b 2b 2b 2b 1b 2b 2b 1b 4 2b 4 2b Level of Evidence Relevant Results REVIEW ARTICLE 1085 Differences not signiﬁcant or not calculable unless otherwise stated. AZM, azithromycin; CP, Chlamydia pneumoniae; CXR, chest x-ray; EES, erythromycin ethylsuccinate; MP, Mycoplasma pneumoniae; pro, prospective; retro, retrospective; RTI, respiratory tract infection. a Signiﬁcance calculated by the review authors. 3b Kawai et al (2012) Retro, case– control/5 y N = 29 (1–15 y) with MP CAP treated with macrolide Macrolide sensitive MP/resistant MP 1. Afebrile 2 d after starting therapy 2. MP DNA load at 2 d 8/8 (sensitive) vs 6/21 (resistant) became afebrile (P , .001).a Mean DNA load decreased signiﬁcantly. Level of Evidence Relevant Results Relevant Outcomes Intervention/Comparator Patients Design/Length Study TABLE 2 Continued 1086 difference in other signs or symptoms, white blood cell count, or C-reactive protein levels.5 One retrospective case–control study found an overall clinical beneﬁt and 2 other studies identiﬁed a decrease in fever duration, although these were rated as being among the lowest quality of evidence in this review.22,27,28 Esposito et al2o (study A) was the only RCT to identify a clinical beneﬁt, but the patient population included children with URTI. Additionally, the data were categorized only as “atypical” bacteria and did not separate C. pneumoniae from M. pneumoniae. Twenty-seven patients had acute C. pneumoniae infection only, and the “worstcase scenario” excludes all patients with C. pneumoniae after assuming each improved on macrolide therapy. Two meta-analyses were attempted using risk difference as the treatment effect. The ﬁrst included all studies listed in Table 4 that used dichotomous variables.13,16,20,22,23,27,28 However, there was a large degree of heterogeneity (P , .001) in addition to publication bias and largely variable treatment effects. A second analysis used only the RCTs13,16,20,23 and demonstrated a pooled risk difference of 0.12 (95% conﬁdence interval [CI], 20.04 to 0.20) (Fig 2). This risk difference represents the absolute change in risk attributable to treatment with a macrolide. In our case, the risk in the treated group minus the risk in the control group was 12% (95% CI, 24% to 20%). This ﬁnding suggests that 12% of children treated with a macrolide will have more rapid clinical improvement, corresponding to a number needed to treat of 8.33, but the conﬁdence interval overlapping 0% negates statistical signiﬁcance. There remained signiﬁcant heterogeneity between the studies (P = .02). The funnel plot revealed potential for publication bias against small studies that show a treatment effect (Fig 3). To provide comparison with the “worst-case scenario,” the same meta-analysis, assuming C. pneumoniae and M. pneumoniae responded equally to treatment, demonstrated a pooled risk difference of 0.12 (95% CI, 0.01 to 0.22). DISCUSSION The majority of studies included in our systematic review did not show a signiﬁcant clinical beneﬁt of M. pneumoniae spectrum therapy in CA-LRTI. Of the 9 studies that speciﬁcally examined the issue of M. pneumoniae treatment in children with CA-LRTI secondary to M. pneumoniae, almost all the prospective studies showed no clinical beneﬁt. The remaining studies generally suggest a statistical, but not necessarily clinically relevant, decrease in fever duration, and most of these are rated as low- or lowest-quality evidence. Our meta-analysis of RCTs suggests a small treatment beneﬁt in patients with CA-LRTI secondary to M. pneumoniae. However, the pooled effect favoring treatment was driven primarily by the result of a single RCT that included children with URTIs (57% of enrolled patients). Four of the 5 studies included in the meta-analysis did not show a beneﬁt of M. pneumoniae spectrum therapy, and the summary statistic is not statistically signiﬁcant. The associated funnel plot demonstrates a paucity of small studies showing a treatment effect, suggesting either publication bias against such studies, poor methodological design, or artifact heterogeneity among smaller studies. The question of M. pneumoniae spectrum (speciﬁcally macrolide) use in CAP, or other CA-LRTIs, is commonly encountered in pediatric practice. Two large, retrospective cohort studies based on administrative data sets have indirectly addressed the issue of M. pneumoniae treatment by comparing differences in length of hospital stay among children receiving b-lactam BIONDI et al Downloaded from pediatrics.aappublications.org by guest on September 9, 2014 REVIEW ARTICLE TABLE 3 Bias Assessment for Individual Studies Study Selection Performance Detection Attrition Reporting Conﬂict of Interest Overall Sterner et al (1967) Ruhrmann et al (1982) Garo et al (1988) Gomez Campdera et al (1996) Gendrel et al (1997) Sáez-Llorens et al (1998) Harris et al (1998) Wubbel et al (1999) Ferwerda et al (2001) Principi et al (2001) Kogan et al (2003) Esposito et al (2005), study A Esposito et al (2005), study B Bradley et al (2007) Lu et al (2008) Matsubara et al (2009) Kawai et al (2012) x — — — — x — — — — — — 4/15 5/15 x x x x x x — x x — — — 8/14 7/17 x x x — x x — x x — — — 8/14 6/15 — x — — x — — x — x x — — x — x x x 4/15 10/15 3/15 — x — — x x x x x x x — x — — x — x 7/14 8/15 6/15 x x x x — x 8/15 — — — x — — x — — — — x — — — x — — 5/15 2/14 2/15 — x x — x — 6/15 —, substantial bias not identiﬁed within the domain; x, substantial bias identiﬁed within the domain. TABLE 4 Studies Comparing Spectrum With Nonspectrum Treatment of Children With Acute Respiratory Infection With M. Pneumoniae Grouped by Outcome Terma Clinical Improvement at #5 d Study Sáez-Llorens et al (1998) Matsubara et al (2009) Kawai et al (2012) Study Lu et al (2008) Time Frame 3d ,5 d 2d Outcome Overall Fever Fever Time Frame N/A Fever Treatment (n) Comparator (n) Improved Total Improved Total 9 43 8 9 47 8 5 5 6 5 22 21 Pb ..99 ,.001 ,.001 Days of Feverc Days of Feverc 4.9 6 1.89 5.63 6 2.22 .04 Comparator (n) Pb Clinical Improvement at .5 d Study Gendrel et al (1997) Harris et al (1998) Esposito et al (2005), study Ad Esposito et al (2005), study Bd Bradley et al (2007) Time Frame 2–18 d 15–19 d 1m 6m 10–17 d Outcome Treatment (n) Improved Total Improved Total Fever Overall Overall 9 21 49 9 21 49 2 9 88 32 9 114 ,.001 ..99 ,.001 Recurrent 26 45 61 109 .86 Clinical 59 66 15 18 .44 a Numbers often extrapolated by the review authors. Includes P values calculated by the review authors. c Results are mean 6 SD. d Worst-case scenario with all 27 C. pneumoniae patients placed in the treatment improved group and then excluded. b monotherapy and children receiving combination therapy with a macrolide.2,31 These studies found a shorter length of stay among children receiving combination therapy. However, testing for M. pneumoniae was not recorded in either database and therefore could not be assessed. Macrolides are among the most frequently prescribed classes of antibiotics for respiratory tract infections in children, particularly in conditions for which antibiotics are not clearly indicated.15 Additionally, national guidelines intended to reduce practice variability in the management of CAP in children recommend consideration of macrolide use but state that this is based on limited data.3 Our review found insufﬁcient evidence to support or refute antibiotic use in CALRTI secondary to M. pneumoniae. This suggests that macrolide use in CA-LRTI could represent an area of opportunity for reducing antibiotic use, which could have a signiﬁcant impact on health care costs and the development of antimicrobial resistance.32,33 Our review provides novel information in comparison with the previous review on the topic by including more enrolled patients (4294 vs 1912) and more studies (17 vs 7).10 Although the previous review reports insufﬁcient data to draw conclusions, we suggest that the available literature does not currently support treatment of CA-LRTI secondary to M. pneumoniae. We examined observational studies, many of which were assessed as having the same level of evidence as the RCTs. In the absence of high-quality RCTs, experts suggest it is reasonable to include cohort and case–control studies in a systematic review, particularly when the quality of evidence is equivalent.34 Additionally, we identiﬁed 3 randomized trials that had not been previously examined, 1 of which was assessed as having the highest level of evidence of any included study.12,16,17 Although our review failed to identify clear therapeutic efﬁcacy of macrolides in pediatric CA-LRTI due to M. pneumoniae, there are data in support of antibiotic treatment. Military studies from the 1960s, performed on recruits in PEDIATRICS Volume 133, Number 6, June 2014 Downloaded from pediatrics.aappublications.org by guest on September 9, 2014 1087 FIGURE 2 Meta-analysis of randomized trials comparing spectrum and nonspectrum treatment of M. pneumoniae in children. Studies were weighted by size and degree of heterogeneity. The overall effect (diamond) represents the pooled risk differences of the 5 studies. The conﬁdence interval crosses 0.00, suggesting a lack of statistical signiﬁcance. basic training, found that recruits with atypical pneumonia more often had serologic evidence of M. pneumoniae infection (identiﬁed as Eaton’s agent at the time) than asymptomatic controls.35,36 Additionally, in a randomized, double-blind, placebo-controlled study among 300 recruits with atypical pneumonia due to M. pneumoniae, treatment with a tetracycline was found to decrease the duration of a number of clinical signs and symptoms faster than placebo.37 These seminal studies do not necessarily represent the typical clinical situation encountered by pediatricians today, given their young adult, homogenous population, nor can they account for mixed infections or modernized diagnostic testing. Nonetheless, these studies describe principles of study design and conduct in this area that can serve as a template for pediatric investigations. One of the strengths of meta-analysis is in the process of evaluating the available literature for heterogeneity. It is often difﬁcult to ﬁnd a middle ground between throwing out all summary estimates and inappropriately combining studies, particularly in pediatrics, where the paucity of high-quality RCTs is acute. We provided a summary statistic for 1 broad study question, but we noted signiﬁcant study heterogeneity that we could not resolve with subgroup analysis. Therefore, our interpretation of the entirety of the statistical testing associated with our meta-analysis is that despite multiple FIGURE 3 Funnel plot of standard errors by risk difference for all studies in the second meta-analysis. Publication bias is evident against small studies that show a treatment effect. 1088 BIONDI et al Downloaded from pediatrics.aappublications.org by guest on September 9, 2014 REVIEW ARTICLE studies, a clear treatment effect has not emerged. Furthermore, the study heterogeneity lends credence to the idea that a large RCT is necessary to provide guidance to pediatricians on this clinical question. There are several limitations to our review, most related to the quality of the available data on the topic. First, most studies contained substantial bias or ﬁnancial conﬂict of interest. We attempted to account for this through indepth quality assessment and reporting to allow a critical interpretation of study results. In particular, detection bias was commonly identiﬁed in the studies comparing spectrum with nonspectrum therapy in patients diagnosed with M. pneumoniae. This bias most often represented lack of blinding of investigators or participants or lack of a reliable outcome measure (ie, clinical improvement). This quality issue could potentially skew study results toward or away from the null hypothesis. Second, we were unable to extract data on outcomes for all studies, reducing the number of articles that could be included in the meta-analysis and potentially causing publication bias. This limitation should be balanced against the fact that we abstracted quantitative data from more studies than previous meta-analyses on this subject. CONCLUSIONS Fourth, M. pneumoniae testing modalities have changed dramatically over time, with variable testing characteristics and lack of overall consistency. This limits the uniformity of the patient population included in our review. Furthermore, coinfection with a pathogen in addition to M. pneumoniae is common but rarely identiﬁed via clinical testing, making it difﬁcult to draw speciﬁc conclusions about therapeutic effects.9,38 Our systematic review provides insufﬁcient evidence to support conclusions about the efﬁcacy of macrolide treatment of CA-LRTI due to M. pneumoniae in children. Fever duration may be decreased, but the clinical impact of this effect is unclear, and we identiﬁed few high-quality studies to support it. Our review represents the most comprehensive evaluation of treating M. pneumoniae in children with CA-LRTI to date. The ﬁndings from this review highlight the need for high-quality, prospective studies to assess the impact of antibiotic therapy to treat CALRTI caused by M. pneumoniae in children. These studies should speciﬁcally address the potential for confounding mixed infections, timing of intervention relative to symptom onset, and testing modalities that include a combination of serology and polymerase chain reaction assays. Finally, no study included in our analysis took into account the duration of illness before the start of antibiotic therapy. It is possible that the timing of intervention relative to the start of symptoms is related to the outcome. ACKNOWLEDGMENTS The authors thank Dr Swati Murthy for her help with data collection and Dr Jill Halterman for her review of the ﬁnal manuscript. 4. Shah SS, Test M, Shefﬂer-Collins S, Weiss AK, Hall M. Macrolide therapy and outcomes in a multicenter cohort of children hospitalized with Mycoplasma pneumoniae pneumonia. J Hosp Med. 2012;7(4):311– 317 5. Lu YJ, Chen TH, Lin LH, Shen CM, Huang CH. Macrolide use shortens fever duration in Mycoplasma pneumoniae infection in children: a 2-year experience. J Microbiol Immunol Infect. 2008;41(4):307–310 6. Deﬁlippi A, Silvestri M, Tacchella A, et al. Epidemiology and clinical features of Mycoplasma pneumoniae infection in children. Respir Med. 2008;102(12):1762–1768 7. Sakurai N, Nagayama Y, Honda A, Makuta M, Yamamoto K, Kojima S. Mycoplasma pneumoniae and other pathogens in the aetiology of lower respiratory tract infections among Japanese children. J Infect. 1988;16 (3):253–261 Waites KB. New concepts of Mycoplasma pneumoniae infections in children. Pediatr Pulmonol. 2003;36(4):267–278 Korppi M, Heiskanen-Kosma T, Kleemola M. Incidence of community-acquired pneumonia in children caused by Mycoplasma pneumoniae: serological results of a prospective, population-based study in primary health care. Respirology. 2004;9(1):109–114 Mulholland S, Gavranich JB, Gillies MB, Chang AB. Antibiotics for communityacquired lower respiratory tract infections secondary to Mycoplasma pneumoniae in children. Cochrane Database Syst Rev. 2012;9:CD004875 Aurangzeb B, Hameed A. Comparative efﬁcacy of amoxicillin, cefuroxime and Third, our assessment of treatment effect was limited primarily to subjective outcome measures such as symptomatic improvement or resolution. This limitation is inevitable given the primary endpoints used in the included studies, but our bias assessment speciﬁcally addresses the subjective nature of the data. REFERENCES 1. Pati S, Lorch SA, Lee GE, Shefﬂer-Collins S, Shah SS. Health insurance and length of stay for children hospitalized with communityacquired pneumonia. J Hosp Med. 2012;7 (4):304–310 2. Ambroggio L, Taylor JA, Tabb LP, Newschaffer CJ, Evans AA, Shah SS. Comparative effectiveness of empiric b-lactam monotherapy and b-lactam–macrolide combination therapy in children hospitalized with community-acquired pneumonia. J Pediatr. 2012;161(6):1097–1103 3. Bradley JS, Byington CL, Shah SS, et al. The management of community-acquired pneumonia in infants and children older than 3 months of age: clinical practice guidelines by the Pediatric Infectious Diseases Society and the Infectious Diseases Society of America. Clin Infect Dis. 2011;53(7):e25–76 8. 9. 10. 11. PEDIATRICS Volume 133, Number 6, June 2014 Downloaded from pediatrics.aappublications.org by guest on September 9, 2014 1089 12. 13. 14. 15. 16. 17. 18. 19. 20. clarithromycin in the treatment of community-acquired pneumonia in children. J Coll Physicians Surg Pak. 2003;13 (12):704–707 Ferwerda A, Moll HA, Hop WC, et al. Efﬁcacy, safety and tolerability of 3 day azithromycin versus 10 day co-amoxiclav in the treatment of children with acute lower respiratory tract infections. J Antimicrob Chemother. 2001;47(4):441–446 Harris JA, Kolokathis A, Campbell M, Cassell GH, Hammerschlag MR. Safety and efﬁcacy of azithromycin in the treatment of communityacquired pneumonia in children. Pediatr Infect Dis J. 1998;17(10):865–871 Toltzis P. Determining the best regimen for children hospitalized with CAP. J Pediatr. 2012;161(6):A1–A2 Hersh AL, Shapiro DJ, Pavia AT, Shah SS. Antibiotic prescribing in ambulatory pediatrics in the United States. Pediatrics. 2011; 128(6):1053–1061 Bradley JS, Arguedas A, Blumer JL, Sáez-Llorens X, Melkote R, Noel GJ. Comparative study of levoﬂoxacin in the treatment of children with community-acquired pneumonia. Pediatr Infect Dis J. 2007;26(10):868–878 Sterner G, Fransén H, Tunevall G, Svedmyr A. A comparative trial of cephaloridine and erythromycin in hospitalized pneumonic patients. Postgrad Med J. 1967;43(Suppl 43):43–, 53–54 Oxford Centre for Evidence-Based Medicine. The Oxford levels of evidence 2. 2013. Available at: www.cebm.net/index.aspx? o=5653. Accessed August 29, 2013 Viswanathan M, Ansari MT, Berkman ND, et al. Assessing the Risk of Bias of Individual Studies in Systematic Reviews of Health Care Interventions. Rockville, MD: Methods Guide for Effectiveness and Comparative Effectiveness Reviews; 2008 Esposito S, Bosis S, Faelli N, et al. Role of atypical bacteria and azithromycin therapy for children with recurrent respiratory 1090 21. 22. 23. 24. 25. 26. 27. 28. tract infections. Pediatr Infect Dis J. 2005;24 (5):438–444 Garo B, Garré M, Quiot JJ, et al. Mycoplasma pneumoniae infections. A multicenter retrospective study of 182 cases [in French]. Presse Med. 1988;17(29):1475– 1478 Gendrel D, Raymond J, Moulin F, et al Etiology and response to antibiotic therapy of community-acquired pneumonia in French children. Eur J Clin Microbiol Infect Dis. 1997;16(5):388–391 Sáez-Llorens X, Castaño E, Wubbel L, et al. Importance of Mycoplasma pneumoniae and Chlamydia pneumoniae in children with community-acquired pneumonia [in Spanish]. Rev Med Panama. 1998;23(2):27– 33 Wubbel L, Muniz L, Ahmed A, et al. Etiology and treatment of community-acquired pneumonia in ambulatory children. Pediatr Infect Dis J. 1999;18(2):98–104 Principi N, Esposito S, Blasi F, Allegra L; Mowgli Study Group. Role of Mycoplasma pneumoniae and Chlamydia pneumoniae in children with community-acquired lower respiratory tract infections. Clin Infect Dis. 2001;32(9):1281–1289 Kogan R, Martínez MA, Rubilar L, et al. Comparative randomized trial of azithromycin versus erythromycin and amoxicillin for treatment of community-acquired pneumonia in children. Pediatr Pulmonol. 2003;35(2):91–98 Matsubara K, Morozumi M, Okada T, et al A comparative clinical study of macrolidesensitive and macrolide-resistant Mycoplasma pneumoniae infections in pediatric patients. J Infect Chemother. 2009;15(6): 380–383. Kawai Y, Miyashita N, Yamaguchi T, et al. Clinical efﬁcacy of macrolide antibiotics against genetically determined macrolideresistant Mycoplasma pneumoniae pneumonia in paediatric patients. Respirology. 2012;17(2):354–362 29. Ruhrmann H, Blenk H. Erythromycin versus amoxicillin for the treatment of pneumonia in children (author’s transl) [in German]. Infection. 1982;10(Suppl 2):S86–S91 30. Gomez Campdera JA, Navarro Gomez ML, Hernandez Sampelayo T, Merello Godino C, Sanchez Sanchez C. Azithromycin in the treatment of ambulatory pneumonia in children. Acta Pediatr Esp. 1996;54(8):554– 562 31. Leyenaar JK, Shieh MS, Lagu T, Pekow PS, Lindenauer PK. Comparative effectiveness of ceftriaxone in combination with a macrolide compared with ceftriaxone alone for pediatric patients hospitalized with community acquired pneumonia [published online ahead of print]. Pediatr Infect Dis J. 2013 32. Carbon C, Isturiz R. Narrow versus broad spectrum antibacterials: factors in the selection of pneumococcal resistance to betalactams. Drugs. 2002;62(9):1289–1294 33. Davey P, Brown E, Charani E, et al. Interventions to improve antibiotic prescribing practices for hospital inpatients. Cochrane Database Syst Rev. 2013;4:CD003543 34. Walker E, Hernandez AV, Kattan MW. Metaanalysis: its strengths and limitations. Cleve Clin J Med. 2008;75(6):431–439 35. Forsyth BR, Bloom HH, Johnson KM, Chanock RM. Etiology of primary atypical pneumonia in a military population. JAMA. 1965;191: 364–368 36. Chanock RM, Mufson MA, Bloom HH, James WD, Fox HH, Kingston JR. Eaton agent pneumonia. JAMA. 1961;175:213–220 37. Kingston JR, Chanock RM, Mufson MA, et al. Eaton agent pneumonia. JAMA. 1961;176:118– 123 38. Michelow IC, Olsen K, Lozano J, Duffy LB, McCracken GH, Hardy RD. Diagnostic utility and clinical signiﬁcance of naso- and oropharyngeal samples used in a PCR assay to diagnose Mycoplasma pneumoniae infection in children with community-acquired pneumonia. J Clin Microbiol. 2004;42(7):3339–3341 BIONDI et al Downloaded from pediatrics.aappublications.org by guest on September 9, 2014 Treatment of Mycoplasma Pneumonia: A Systematic Review Eric Biondi, Russell McCulloh, Brian Alverson, Andrew Klein, Angela Dixon and Shawn Ralston Pediatrics 2014;133;1081; originally published online May 26, 2014; DOI: 10.1542/peds.2013-3729 Updated Information & Services including high resolution figures, can be found at: http://pediatrics.aappublications.org/content/133/6/1081.full. html Supplementary Material Supplementary material can be found at: http://pediatrics.aappublications.org/content/suppl/2014/05/2 0/peds.2013-3729.DCSupplemental.html References This article cites 36 articles, 6 of which can be accessed free at: http://pediatrics.aappublications.org/content/133/6/1081.full. html#ref-list-1 Citations This article has been cited by 2 HighWire-hosted articles: http://pediatrics.aappublications.org/content/133/6/1081.full. html#related-urls Permissions & Licensing Information about reproducing this article in parts (figures, tables) or in its entirety can be found online at: http://pediatrics.aappublications.org/site/misc/Permissions.xh tml Reprints Information about ordering reprints can be found online: http://pediatrics.aappublications.org/site/misc/reprints.xhtml PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly publication, it has been published continuously since 1948. PEDIATRICS is owned, published, and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk Grove Village, Illinois, 60007. Copyright © 2014 by the American Academy of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275. Downloaded from pediatrics.aappublications.org by guest on September 9, 2014