Mycoplasma pneumoniae detection in children with respiratory tract infections and influence on management - a retrospective cohort study in Switzerland.

AIM: To evaluate the frequency of Mycoplasma pneumoniae in nasopharyngeal specimens from children with respiratory tract infections (RTIs) and to detail clinical characteristics and management.
METHODS: The study was designed as a retrospective cohort study. All children with RTI and nucleic acid amplification testing from nasopharyngeal specimens were analysed. Clinical data were extracted from electronic health records for all M. pneumoniae-positive cases. Stored samples of cases and a random selection of matched controls were retested using a M. pneumoniae-specific nucleic acid amplification test.
RESULTS: Of 4460 children, 70 (1.6%) were positive for M. pneumoniae with a median age of 6.4 (IQR: 2.7-9.7). M. pneumoniae was the only organism identified in 50/64 (78%) cases. Macrolide treatment was prescribed in 52/65 (80%); prescription was empirical in 29/52 (56%) and targeted in 23/52 (44%) with no difference regarding patient age, oxygen requirement or duration of hospitalisation.
CONCLUSION: The prevalence of M. pneumoniae in nasopharyngeal specimens of children with RTI was low. The detection of M. pneumoniae influenced antibiotic prescriptions, but the benefit of early empirical versus targeted treatment remains unclear. ©2019 Foundation Acta Paediatrica. Published by John Wiley & Sons Ltd.

Community‐acquired pneumonia

Chest x‐ray

Mycoplasma pneumoniae

Polymerase chain reaction

Respiratory syncytial virus

Respiratory tract infection

This Swiss study investigated the prevalence of M. pneumoniae in 4460 children with respiratory tract infection (RTI).

In children with RTI, in whom M. pneumoniae was detected by nucleic acid amplification testing, the majority had no other organism detected suggesting causal relationship.

The detection of M. pneumoniae resulted in prescription of a macrolide antibiotic in the majority of cases, and empirical versus delayed macrolide treatment did not result in improved short‐term outcome.

Introduction

Mycoplasma pneumoniae respiratory tract infections (RTIs) are common, particularly in children and young adults between 5 and 20 years of age 1. M. pneumoniae causes upper and lower RTI and has been shown to be the most common bacterial pathogen for community‐acquired pneumonia together with Streptococcus pneumoniae 2, 3. M. pneumoniae‐associated RTIs are most often mild and self‐limiting in nature, and severe courses of disease are not common 4, 5. In the past, the detection of pathogen‐specific IgM and IgG antibodies in serum has been the standard diagnostic test for M. pneumoniae RTI. However, elevated M. pneumoniae‐specific serum antibodies have been shown to persist for several months after infection and are also common in asymptomatic children, thus limiting their use for detection of an acute infection 6. Molecular methods for the diagnosis of M. pneumoniae have therefore emerged and are now being used in clinical practice. However, the relationship between the detection of M. pneumoniae by nucleic acid amplification testing from a nasopharyngeal specimen and disease has been questioned in the light of evidence that it is also commonly detected among asymptomatic children 6. The aim of this study was to investigate the frequency of M. pneumoniae in routine nasopharyngeal specimens from children with a RTI and to detail clinical characteristics and management of M. pneumoniae‐positive patients.

Methods

Study setting and participants

Eligible for inclusion were children with an acute RTI and undergoing routine multiplex nucleic acid amplification testing from nasopharyngeal specimens, who presented to the emergency department of the University of Basel Children’s Hospital, Basel, Switzerland, between June 2010 and December 2014. They were identified using laboratory records of the Division of Infection Diagnostics, Department of Biomedicine (Haus Petersplatz) at the University of Basel.

There were no further inclusion or exclusion criteria. Electronic health records were used to extract the following data: age at presentation, gender, date of presentation, admission status, clinical examination findings, results of multiplex nucleic acid amplification test from nasopharyngeal specimens, laboratory inflammatory markers, results of chest radiography as reported by senior radiologist, oxygen requirement, antibiotic treatment and duration of admission. Data from patients without a complete information from  electronic health records were included in the analysis and specified where applicable. The study was approved by the ethics committee of the University of Basel (EKNZ 2015‐277).

Analysis of nasopharyngeal specimens

All nasopharyngeal specimens were routinely taken by experienced clinical staff and analysed using a commercially available multiplex nucleic acid amplification test (Respifinder®, Pathofinder, Maastricht, Netherlands) 7, 8. The multiplex nucleic acid amplification test used detects M. pneumoniae, Bordetella pertussis, Chlamydophila (synonym: Chlamydia) pneumoniae, adenovirus, bocavirus, coronavirus (229E, HKU1, OC43, and NL63), human metapneumovirus (hMPV), influenzavirus (A, B and A(H1NA)pdm09), rhinovirus/enterovirus, parainfluenza virus 1, 2, 3, 4, and respiratory syncytial virus (RSV) A and B. Routinely stored nasopharyngeal specimens were retested with an independent M. pneumoniae‐specific nucleic acid amplification test based on Nilsson et al.9 using the primers GGCAGTCAACAAACCACGTATG, GGTGGTTGATGCGGTCAAA and the TaqMan probe CCCACCCGAACCGAAGCGG. This was done in all children where M. pneumoniae was detected in the multiplex nucleic acid amplification test and in 1:2‐matched children with negative initial M. pneumoniae detection. Matching criteria were age (±24 months), gender and season of presentation. Season was defined as autumn/winter (22 Sep–19 March) and spring/summer (20 March–21 Sept). The analysis of all samples was done at the Division of Infection Diagnostics in the Department of Biomedicine (Haus Petersplatz) at the University of Basel.

Statistical analysis

Statistical analysis for numerical data was performed using Mann–Whitney U‐test or Kruskal–Wallis test; for categorical data, Pearson’s chi‐square test was used. A p‐value < 0.05 was considered statistically significant.

Results

Characteristics of all children with nasopharyngeal specimen results

Over the 4.5‐year study period, 4460 nasopharyngeal specimens were obtained from children (median age 1.3 years) with acute RTI undergoing routine multiplex nucleic acid amplification testing from nasopharyngeal specimens. Of those, 70 (1.6%, 95% CI: 1.2–1.9) were positive for M. pneumoniae with a median age of 6.4 (IQR 2.7–9.7) years. The annual proportion for M. pneumoniae‐positive samples ranged between 0.7 and 2.5%. Figure 1 shows age‐stratified numbers and proportions of M. pneumoniae‐positive versus negative cases in the study period.

Figure 1

Age‐stratified analysis of Mycoplasma pneumoniae cases between 2010 and 2014.

For the reanalysis of nasopharyngeal specimens, 135 matched controls were identified. Complete medical records including all variables of interest were available in 63 (90%) of 70 M. pneumoniae‐positive cases and in 124 (92%) of 135 controls. The most frequent discharge diagnoses of M. pneumoniae‐positive cases were pneumonia 46/65 (71%), bronchitis/bronchiolitis 9/65 (14%) and upper RTI 4/65 (6%).

In controls, the discharge diagnoses were more diverse: upper RTI 42/125 (33%), pneumonia 32/125 (26%) and bronchitis/bronchiolitis 30/125 (24%) were the most frequently recorded. Further baseline characteristics including clinical characteristics of both groups are summarised in Table 1.

Table 1

Characteristics of cases with nasopharyngeal specimen positive for Mycoplasma pneumoniae and matched M. pneumoniae‐negative controls

	Cases		Matched controls
	n*	Result	n*	Result
Median (IQR) age in years	70	6.4 (2.7–9.7)	133	6.1 (4.5–9.2)
Number (%) male gender	70	42 (60)	133	81 (61)
Number (%) sampled in winter season	70	46 (66)	133	87 (65)
Number (%) other pathogen identified †	64	14 (22)	124	80 (64)
Number (%) admission to hospital	65	48 (74)	125	85 (68)
Median (IQR) CRP (mg/L)	52	17.5 (8.3–42.5)	85	42.5 (4.1–32)
Number (%) x‐ray performed	65	53 (81)	126	64 (51)
Number (%) oxygen requirement	64	18 (28)	125	44 (35)
Pneumonia as diagnosis (%)	65	46 (71)	125	32 (26)
Number (%) any antibiotic treatment	65	55 (85)	125	52 (42)
Number (%) macrolide prescribed	65	52 (80)	125	19 (15)
Number (%) nonmacrolide antibiotic prescribed	65	4 (6)	125	33 (26)
Number (%) >1 antibiotic prescribed	65	17 (26)	125	6 (5)

Information available.

Matching criteria.

CRP = C‐reactive protein; IQR = interquartile range.

Organisms identified

In 50 (78%) of 64 cases, M. pneumoniae was the only organism detected in nasopharyngeal specimens. Concomitant organisms were identified in 22%; 13 (20%) had two and one (1.5%) had three organisms detected. In M. pneumoniae‐negative patients, 80/124 (64%) had one organism detected, with rhinovirus/enterovirus being the most frequent organism, and 7 (6%) had two organisms detected. Of the stored nasopharyngeal specimen samples, 64/70 (91%) of the M. pneumoniae‐positive cases and 128/135 (95%) of the M. pneumoniae‐negative cases were available for retesting using a M. pneumoniae‐specific nucleic acid amplification test. Of the samples with sufficient RNA/DNA, 55 (92%) were confirmed positive and none of those initially M. pneumoniae‐negative tested positive (Table 2). Therefore, the M. pneumoniae‐specific nucleic acid amplification test of stored samples showed a sensitivity of 91.7% and a specificity of 100%.

Table 2

Results for retesting of M. pneumoniae‐specific nucleic acid amplification test

	Cases (n = 70)	Controls (n = 135)
Number (%) available for retesting	64 (91)	128 (95)
Number PCR/DNA empty	4	2
Number (%) positive	55 (92)	0 (0)
Number (%) negative	5 (8)	126 (100)

Antibiotic treatment of cases and controls

A total of 52/65 (80%) cases were treated with a macrolide, which was clarithromycin in 50 (96%) (Table 1). While 34 (38%) cases received a macrolide only, 18 (15%) were also prescribed concomitant treatment with a beta‐lactam antibiotic. In 5 (10%) cases, macrolide treatment was initiated before presentation to the hospital, in 24 (46%), it was started empirically at presentation and in 23 (44%) targeted; that is, it was initiated after the result of M. pneumoniae detection had become available. Children receiving empiric compared to targeted macrolide treatment were statistically not different for age (p = 0.21), oxygen requirement (p = 0.78) or duration of hospitalisation (p = 0.56). Thirteen cases did not receive an antibiotic treatment. In 11/13 (85%) cases, clinical improvement and/or discharge from the hospital prior to the result being available was documented. In 2/13 (14%), the reason for withholding macrolide treatment is unknown. In controls, 19/125 (15%) were treated with a macrolide (Table 1). Of those, 13 (10%) received a macrolide only and 6 (5%) a combination treatment with a beta‐lactam antibiotic. In addition, 33/125 (26%) of controls received a beta‐lactam antibiotic only.

Chest radiography

A chest x‐ray (CXR) was performed in 53 (81%) of 65 cases. Of those, 31 (58%) were classified as pneumonia/infiltrate, 18 (34%) as bronchitis/bronchial wall thickening, 3 (6%) as normal and for one the report was missing. In 126 children with negative M. pneumoniae detection, 64 (51%) had a CXR performed. Of those, 32 (50%) were classified as pneumonia/infiltrate, 17 (27%) as bronchitis/bronchial wall thickening, 14 (22%) as normal and one (2%) as an effusion. A CXR was more commonly performed in cases compared to controls (p < 0.001), and they were more commonly abnormal in cases (p = 0.013).

Discussion

We investigated the prevalence of M. pneumoniae in nasopharyngeal specimens from children with RTI. Of 4460 samples, only 70 (1.6%) were positive for M. pneumoniae. This is in line with findings from two previous studies in England/Wales and Germany in which M. pneumoniae was detected by PCR in 0.5–4% of patients with RTI 10, 11. These studies were performed before the occurrence of the most recent M. pneumoniae epidemics in Europe in 2010–2012 and other recent studies describing much higher rates of M. pneumoniae detection in children with RTI 6, 12, 13, 14, 15, 16, 17. For example, a study in the Netherlands including children aged 3 months to 16 years from July 2008 to November 2011 showed that 16% of children with RTI (mean age 2.7 years) and 21% of asymptomatic children (mean age 5.6 years) were positive for M. pneumoniae by PCR 6. Importantly, the prevalence of M. pneumoniae in children with RTI was lower in the years 2008/2009 compared to 2010/2011, with 4–7% and 21–24%, respectively, suggesting a considerable year‐to‐year variation likely influenced by the presence of local or national M. pneumoniae epidemics. The lower prevalence of M. pneumoniae in our study might be explained by an absence of an M. pneumoniae epidemic between 2010 and 2014 in our region. Such a diverse regional prevalence of M. pneumoniae seems possible as other European regions or countries in the same time interval also did not report an increasing M. pneumoniae prevalence 18. Unfortunately, Switzerland has no national surveillance of M. pneumoniae‐related RTI and it remains therefore difficult to confer our data to the whole country.

Further to this, younger age has been shown in several studies to be associated with lower detection rates of M. pneumoniae 11, 19, 20 . In the Dutch study mentioned above, the prevalence of M. pneumoniae in nasopharyngeal specimens in children under five years of age was 15% compared to 23% in older children. A study from Italy including children admitted to hospital with RTI with a mean age of 5.2 years showed a M. pneumoniae prevalence of 23% 12. Similarly, in a study in Chile including children with community‐acquired pneumonia at a mean age of 4.9 years, 26.4% were positive for M. pneumoniae by PCR 13. A recent study in the United States showed only 3% M. pneumoniae prevalence in children below five years of age with community‐acquired pneumonia, compared to 17 and 24% in children aged five to nine years and 10–17 years, respectively 21. In our study, the median age of children included was 1.3 years and therefore lower than in all previous published studies. This has likely contributed to the low prevalence of M. pneumoniae in our setting as younger age is associated with lower M. pneumoniae detection rates.

The use of macrolides prior to testing might have also influenced the results, as effective antibiotic treatment may shorten the period of M. pneumoniae DNA persistence. In our study, 15% of the negative controls were already treated with a macrolide before a nasopharyngeal specimen was obtained. However, it is not known how long DNA might persist in children on effective treatment and as it is also impossible to reconstruct the duration of antimicrobial treatment these children received prior to testing, the use of macrolides might have had an influence.

As asymptomatic carriage rates as high as 56% have been reported, we had expected higher M. pneumoniae rates in our study, including in cases with a likely other cause of RTI such as influenza or RSV 6, 22. However, the proportion of other pathogens identified was low in the group of M. pneumoniae‐positive children. This strongly suggests that the detection of M. pneumoniae in the nasopharyngeal specimen as a single organism increases the likelihood of causality for the RTI. This notion is supported by another study reporting that over 40% of symptomatic children with M. pneumoniae detected had no other pathogen identified in their nasopharyngeal specimen 6. There are limited data comparing sensitivity of upper versus lower respiratory tract samples for M. pneumoniae. However, the available evidence indicates that in most cases, both specimens are positive. Thus, the detection of M. pneumoniae in the nasopharyngeal specimen of a symptomatic child with lower RTI strongly implicates this pathogen as the cause of the pneumonia 23, 24. This interpretation is also in line with evidence showing the narrowing of the microbiota present in nasopharyngeal specimen from children with RTI 25. Further to this, CXR was more frequently performed and abnormal in the group of children with M. pneumoniae detected in nasopharyngeal specimens compared to those with negative results. This suggests an association of pneumonia in those with positive results.

The detection of M. pneumoniae led to a change of the empiric antibiotic treatment in almost half of the cases in the present study. However, data on the efficacy of macrolide treatment are controversial. Two recent systematic reviews concluded that there is insufficient evidence to draw conclusions about the efficacy of macrolides in the treatment of M. pneumoniae lower RTIs in children, despite the fact that there is considerable evidence on anti‐inflammatory properties of these agents 26, 27, 28. Interestingly, in our study, no short‐term treatment benefit such as reduced oxygen requirement or shorter hospital stay was found when comparing patients with empiric administration of a macrolide and those with prescription 48 hours later after M. pneumoniae infection had been confirmed by nucleic acid amplification testing. This is consistent with another recently published study, which failed to show a significant difference in duration of admission in children with M. pneumoniae‐positive community‐acquired pneumonia treated with a macrolide compared to no treatment 21.

One potential limitation of our study is that in our emergency department, multiplex nucleic acid amplification testing from NPS is not a standard diagnostic tool for children not requiring admission. This may lead to a bias of testing children with clinically more severe RTI presentations, thereby rendering our findings not applicable to children with mild RTI. A further limitation is the potential lower sensitivity of multiplex PCR assays. However, retesting of stored samples with an independent M. pneumoniae‐specific nucleic acid amplification testing did not suggest lower sensitivity in line with recent results 29. However, sensitivity may also be influenced by the storage itself, which cannot be excluded in the current study setting. The study was done in a limited region and time frame, and this may limit generalisability of the study results. Further to this, results and comparisons of results from the group with negative M. pneumoniae testing require cautious interpretation as they represent only a small and potentially nonrepresentative group of M. pneumoniae‐negative patients.

Conclusions

Multiplex nucleic acid amplification testing is a sensitive screening tool, and the detection of M. pneumoniae influenced antibiotic prescription. Empiric treatment did not result in shorter admission duration. In view of the uncertainties about efficacy of antibiotic treatment and increasing antibiotic resistance worldwide, the significance of M. pneumoniae detection in children with RTI needs to be further investigated.

Conflict of interest

JAB declares that her husband is employed by Novartis. All other authors have nothing to declare.

Funding

This study did not receive any external funding.