Mycoplasma Pneumoniae Testing and Treatment Among Children With Community-Acquired Pneumonia

We conducted a retrospective cohort study using the Pediatric Health Information System (PHIS) database. PHIS includes administrative and billing data from 41 children’s hospitals affiliated with the Children’s Hospital Association (Lenexa, KS). Given the use of deidentified data, this study was determined to be non–human subjects research by our medical center’s institutional review board.

We included children 3 months to 18 years old hospitalized with CAP between 2012 and 2018. We defined CAP by International Classification of Diseases, Ninth Revision and International Classification of Diseases, 10th Revision discharge diagnosis codes (International Classification of Diseases, Ninth Revision: 480–483, 485–487; International Classification of Diseases, 10th Revision: J12.x, J13, J14, J15.x, J16.x, J18.0–J18.1, J18.8–J18.9, 510.0, 510.9, 511.0–511.1, 511.9, J86.x, R09.1, J90, J91.8).⁸  We excluded children who were transferred from a different hospital because testing or medications prescribed at the original hospital would not be available in the PHIS data set. We excluded children with diagnosis codes consistent with complex chronic conditions because CAP testing and management might be different in these children.⁹ 

At the hospital level, we examined the proportion of patients receiving Mp testing and macrolide therapy during hospitalization. At the patient level, the primary exposure was a billed Mp diagnostic test, including serology or polymerase chain reaction. Patient characteristics included demographics, payer, asthma hospitalization within the past 12 months, diagnostic testing (ie, chest radiography, computed tomography, ultrasound, viral testing), and use proxies for illness severity (ie, blood gas, blood products, chest tube). Macrolide prescription on hospital day 0 to 1 served as a proxy for empirical prescription because some Mp test results would likely not be available during this time. Patient outcomes included ICU admission, length of stay (LOS), and hospital readmission within 14 days of index discharge.¹⁰  We also examined the proportion of patients receiving Mp testing and macrolide therapy over time.

Continuous variables were described by using median, range, and interquartile range (IQR) and compared across groups by using the Wilcoxon rank test. Categorical variables were compared by using the χ² test. At the hospital level, Spearman’s correlation coefficient was performed to evaluate the correlation between Mp testing and macrolide prescription. Trends in the proportion of patients receiving Mp testing and macrolide therapy over time were assessed by using the Cochran-Armitage trend test. P < .05 was determined as statistically significant. All analyses were performed by using SAS statistical software (version 9.4; SAS Institute, Inc, Cary, NC).

Among 103 977 children hospitalized with CAP, 17.3% underwent Mp testing and 31.1% received macrolides (Table 1). At the hospital level, the median proportion of patients undergoing Mp testing was 7.1% (range: 0.4%–78.7%) and the median proportion of patients receiving macrolides was 29.1% (range: 15.8%–54.7%). Of the patients receiving macrolides, 97.5% received azithromycin, 1.9% received erythromycin, and 0.6% received clarithromycin. There was no correlation between Mp testing and macrolide prescription at the hospital level (R² = 0.05; P = .11) (Fig 1A). Mp testing increased from 15.8% in 2012 to 18.6% in 2018 (P < .001), and macrolide prescription decreased from 40.9% in 2012 to 20.6% in 2018 (P < .001) (Fig 1B). There was no significant change in the proportion of patients requiring ICU admission during the study period (1.04% vs 1.25%; P = .23).

Patients tested for Mp had overall increased markers of illness severity and were more likely to receive imaging and viral pathogen testing when compared with patients who were not tested (Table 1). Patients who were tested were also more likely to have ICU admission and longer median LOS when compared with patients without Mp testing.

Nearly one-third of hospitalized children with CAP received macrolide antibiotic therapy. We found hospital-level variability in Mp testing and macrolide prescribing; however, there was no correlation between testing and prescribing practices. The proportion of patients receiving macrolides (31%) exceeded the previously reported Mp prevalence of 8% to 14%,¹  although macrolide prescription steadily decreased during our study period.

Systematic reviews of children with Mp pneumonia reveal insufficient evidence to support clinical benefit from macrolide therapy.^3,7  In one study, 77% of children who were managed without antibiotics demonstrated clinical resolution of illness.¹¹  Our findings corroborate previous findings of potential overuse of macrolide therapy in children with CAP.¹² 

Patients who received Mp testing had higher markers of illness severity, which suggests that clinicians tested for Mp in ill-appearing children or that clinicians performed more overall diagnostic testing (ie, viral testing and imaging studies) in general in these children. Previous studies have indicated that Mp pneumonia is often a milder disease than CAP caused by other pathogens.^6,13  One approach to reduce macrolide overuse could involve testing children with mild disease and withholding empirical antibiotics until test results are received. However, we did not find a statistical correlation between testing and treatment; therefore, it is unclear in which direction, if any, increased or targeted testing would change antibiotic prescription practices. Furthermore, examination of testing and treatment practices over time revealed a small increase in testing accompanied by a substantial decrease in macrolide prescription between 2012 and 2018. These encouraging temporal trends suggest that clinicians have identified opportunities to reduce the overprescription of macrolide antibiotics without indiscriminately overtesting for Mp among children with CAP.

This study has several limitations. First, we did not have detailed clinical covariates (eg, previous receipt of antibiotics, physical examination findings) that may have been associated with testing, disease severity, treatment, or outcomes. Second, PHIS does not provide reliable information on test type (ie, polymerase chain reaction or serology), test results, and the exact timing of Mp test result availability. Therefore, some misclassification likely exists. Lastly, the results may not be generalizable to patients admitted to non–children’s hospitals.

Macrolide prescribing among children hospitalized with CAP exceeded currently reported estimates of Mp prevalence. Testing and treatment of Mp were not correlated at the hospital level, and temporal trends revealed substantial decreases in macrolide prescription without a sizeable increase in Mp testing during the study period. It remains unclear if clinicians use Mp testing to drive decisions regarding macrolide prescription. Reducing macrolide overuse in CAP is of critical importance given the emergence of macrolide-resistant Mp strains.¹⁴  Novel diagnostics, when used in combination with clinical features, reveal promise for improving reliable identification of Mp pneumonia and, as such, may allow for a coordinated clinical approach that further reduces overprescription of macrolide antibiotics among children with CAP.¹⁵