Epidemiology and Molecular Identification and Characterization of Mycoplasma pneumoniae, South Africa, 2012–2015

During 2012–2015, we tested respiratory specimens from patients with severe respiratory illness (SRI), patients with influenza-like illness (ILI), and controls in South Africa by real-time PCR for Mycoplasma pneumoniae, followed by culture and molecular characterization of positive samples. M. pneumoniae prevalence was 1.6% among SRI patients, 0.7% among ILI patients, and 0.2% among controls (p<0.001). Age <5 years (adjusted odd ratio 7.1; 95% CI 1.7–28.7) and HIV infection (adjusted odds ratio 23.8; 95% CI 4.1–138.2) among M. pneumonia–positive persons were associated with severe disease. The detection rate attributable to illness was 93.9% (95% CI 74.4%–98.5%) in SRI patients and 80.7% (95% CI 16.7%–95.6%) in ILI patients. The hospitalization rate was 28 cases/100,000 population. We observed the macrolide-susceptible M. pneumoniae genotype in all cases and found P1 types 1, 2, and a type 2 variant with multilocus variable number tandem repeat types 3/6/6/2, 3/5/6/2, and 4/5/7/2.


Study Design
We enrolled patients and asymptomatic persons during June 2012-May 2015 as part of 2 surveillance programs (1 for severe respiratory illness [SRI] and 1 for influenza-like illness [ILI]). SRI surveillance was conducted at 2 sentinel sites, Edendale Hospital in KwaZulu Natal Province and Klerksdorp-Tshepong Hospital Complex in North West Province. Patients enrolled in SRI surveillance were those hospitalized with clinical signs and symptoms of lower respiratory tract infection (LRTI), regardless of symptom duration. We included children 2 days to <3 months old who had physician-diagnosed sepsis or acute LRTI, children 3 months to <5 years old with physician-diagnosed LRTI, and persons >5 years old who met the World Health Organization case definition for LRTI (sudden onset of fever [temperature >38°C] or reported fever, cough or sore throat, and shortness of breath or difficulty breathing [14]).
ILI patients were outpatients who were seen at 2 primary health care clinics serving the 2 SRI sentinel sites. Patients were considered to have ILI if they had an acute fever of >38°C or a self-reported fever within the last 7 days and either a cough or sore throat. Asymptomatic persons included those who were seen at the same primary health care clinics and had no history of respiratory illness, diarrheal illness, or fever in the preceding 14 days. For asymptomatic persons, we aimed to enroll 1 HIV-infected and 1 HIV-uninfected person weekly in each clinic within the following age categories: 0-1, 2-4, 5-14, 15-54, and >55 years.
We obtained demographic and clinical information from all enrollees by using a standardized questionnaire. We reviewed hospital records of SRI patients to assess disease progression and outcome.

Specimen Collection
We collected combined nasopharyngeal and oropharyngeal swabs from >5 year-old persons and nasopharyngeal aspirates from <5 year-old persons (nasopharyngeal specimens) and placed the specimens in universal transport medium (Copan Italia, Brescia, Italy). We collected induced or expectorated sputum from SRI patients only. HIV status was determined as part of standard care or by using anonymized-linked dried blood spot testing for consenting enrollees (PCR for children <18 months old and ELISA for persons >18 months old [15]). We tested nasopharyngeal specimens for 10 respiratory viruses (influenza types A and B, adenovirus, enterovirus, rhinovirus, human metapneumovirus, respiratory syncytial virus, and parainfluenza virus types 1-3) by using an in-house multiplex real-time reverse transcription PCR (16).

Detection of M. pneumoniae
We extracted DNA from 200 µL of nasopharyngeal specimen and digested sputum by using the MagNA Pure 96 instrument (Roche Diagnostics, Mannheim, Germany) with the DNA and Viral NA SV kit (Roche Diagnostics). We performed an in-house multiplex real-time PCR for the detection of M. pneumoniae, Chlamydia (Chlamydophila) pneumoniae, and Legionella spp., with human ribonuclease P gene serving as an internal control, as previously described (17). A positive M. pneumoniae patient was defined as a patient having a positive PCR result with a cycle threshold value <45 for M. pneumoniae on the nasopharyngeal specimen, sputum specimen, or both.

Culture and Molecular Characterization
We detected 82 cases of PCR-positive M. pneumoniae during study periods 1 (June 2012-May 2013) and 2 (June 2013-May 2014) and performed culture and molecular characterization retrospectively on 77 (94%) samples. Culture and further characterization could not be performed for 5 cases because of insufficient specimens.
We inoculated M. pneumoniae-positive specimens in SP4 medium (Thermo Fisher Scientific, Waltham, Massachusetts, USA) and incubated them at 37°C in 5% CO 2 for up to 10 days. Growth was indicated by a color change from red to orange, without turbidity. We performed macrolide susceptibility analysis by using real-time PCR followed by high-resolution melt-curve (HRM) analysis by means of the Rotor-Gene Q6000 system (QIAGEN, Hilden, Germany), according to previously described methods (18).
We performed P1 genotyping by using real-time PCR targeting the 1900-bp region of the P1 gene, followed by HRM analysis using the Rotor-Gene Q6000 system according to previously described methods (12). We also performed MLVA typing on the same specimens by using 5 variable-number tandem-repeat loci (Mpn1, Mpn13, Mpn14, Mpn15, and Mpn16), as described by Dégrange et al. (13). However, for analysis, we used the 4-loci nomenclature as described by Sun et al. (19) because of the instability of the Mpn1 locus (20).

Statistical Analysis
We used the χ 2 or Fisher exact test for comparison of categorical variables. We used unconditional logistic regression to estimate the attributable fraction (AF) of M. pneumoniae-associated hospitalization and outpatient consultation by comparing the M. pneumoniae detection rate among SRI or ILI patients to that of controls. The AF was estimated from the odds ratio (OR) obtained from the regression models.
Among SRI patients, we estimated the AF for patients positive on nasopharyngeal specimens only as well as for patients positive on both nasopharyngeal and sputum specimens. We adjusted all estimates for age, HIV status, underlying medical conditions other than HIV infection, and co-infections with the 10 respiratory viruses investigated in this study.
In addition, we used unconditional logistic regression to assess factors associated with M. pneumoniae-associated SRI hospitalization by comparing the characteristics of M. pneumoniae-positive SRI patients with those of M. pneumoniae-positive ILI patients. For the multivariable model, we assessed all variables that were significant at p<0.2 on univariate analysis and dropped nonsignificant factors (p>0.05) with manual backward elimination. We assessed pairwise interactions by inclusion of product terms for all variables remaining in the final multivariable additive model. We performed the analysis by using

Calculation of Rates of M. pneumoniae-Associated SRI Hospitalization
We estimated the overall and age-specific rates of M. pneumoniae-associated SRI hospitalizations (per 100,000 population) by using the number of SRI hospitalizations and adjusting for nonenrollment (e.g., refusals to participate and no enrollment on weekends [21]) and healthcareseeking behavior during 2013-2014. For all calculations, we assumed that the M. pneumoniae detection rate among persons tested and not tested was the same within age groups. We obtained age-and year-specific population denominators from projections of the 2011 census data (21), and we obtained age-and year-specific HIV prevalence in the study population from the projections of the Thembisa model (22).
We calculated 95% CIs for all estimated rates by using bootstrap resampling of all parameters included in the estimation over 1,000 replications. The upper and lower limits of the 95% CI were the 2.5th and 97.5th percentile of the estimated values from the bootstrapped datasets, respectively.
Among persons for whom age was known, children <5 years old accounted for 35

AF of M. pneumoniae-Associated Hospitalization and Outpatient Consultation
The AF of M. pneumoniae detection to illness using nasopharyngeal specimens only for patients with ILI was 80.7% (95% CI 16.7%-95.6%) and for patients with SRI was 90.1% (95% CI 58.3%-97.7%). The AF of M. pneumoniae detection to illness for patients with SRI using both nasopharyngeal and sputum specimens was 93.9% (95% CI 74.4%-98.5%).  Table 2).

Rates of M. pneumoniae SRI Hospitalization
The mean annual rate of hospitalization for M. pneumoniae patients during 2013-2014 was 27.9 cases/100,000 population (95% CI 18.9-37.4) ( The prevalence of M. pneumoniae varies depending on whether a study was performed during an endemic or epidemic year, the laboratory detection method used, or the study participants (3). During 2010-2012, an epidemic of M. pneumoniae occurred in Denmark, England, Wales, Sweden, Finland, and Germany, with detection rates ranging from 12% to 17% (23)(24)(25)(26). In France, detection rates of M. pneumoniae ranged from 2% to 10% during a 5-year period among outpatients with an acute respiratory illness (5). However, higher detection rates of 27%-30% among children with community-acquired pneumonia have been reported in the United States and Finland and up to 60% among hospitalized adults with pneumonia in Japan (3,4,27). Jain et al. reported that, among hospitalized children in the United States, M. pneumoniae was the most common bacterial cause of community-acquired pneumonia, accounting for 8% of cases (28), and among hospitalized adults in the United States, M. pneumoniae was identified in ≈2% of cases (29). The prevalence differences in our study compared with other studies might be attributable to a difference in enrollment criteria, the age group of participants, and HIV prevalence among the participants.
Despite having low detection rates, M. pneumoniae was significantly associated with illness. The fraction of illness attributable to M. pneumoniae in patients testing positive was 80.7% in ILI patients, 90.1% in SRI patients with M. pneumoniae detected on nasopharyngeal specimens only, and 93.9% in SRI patients with M. pneumoniae detected on both nasopharyngeal and sputum specimens. These results suggest that M. pneumoniae can be considered a likely pathogen when detected in patients with ILI or SRI, regardless of specimen type.
We did not observe a distinct seasonal pattern of M. pneumoniae. Several more years of surveillance of M. pneumoniae is essential to elucidate seasonality in our setting. However, a significant difference was noted in the detection rate over the 3 study periods. Layani-Milon et al. reported that, during a 5-year period (1993)(1994)(1995)(1996)(1997), rates of M. pneumoniae disease varied monthly and yearly and M. pneumoniae occurred in epidemic cycles (5). Furthermore, in a serologic study in Johannesburg, South Africa, during  †Only variables found to be statistically significant (p<0.2) in univariate analysis were assessed in the multivariable model. ‡Underlying conditions defined as previously diagnosed chronic conditions including asthma, chronic lung diseases, cirrhosis or liver failure, chronic renal failure, heart failure, valvular heart disease, coronary heart disease, diabetes, burns, kwashiorkor or marasmus, nephrotic syndrome, spinal cord injury, seizure disorder, emphysema, and cancer, or history of immunosuppressive therapy or splenectomy. §Viruses tested were influenza types A and B, adenovirus, enterovirus, rhinovirus, human metapneumovirus, respiratory syncytial virus, and parainfluenza virus types 1-3.  (30).
We found that young age (<5 years) and HIV infection among M. pneumoniae-positive persons were independently associated with severe disease. HIV association with M. pneumoniae disease was reported in a study conducted in India during 2004-2007 (31). The incidence rate of M. pneumoniae in South Africa was 28 cases/100,000 population, with the highest incidence occurring in children <5 years old at a rate of 87 cases/100,000 population. We observed a greater disease prevalence among HIV-infected patients than HIV-uninfected patients. Other studies have reported incidence rates ranging from 180 to 1,290 cases/100,000 population/year (3,5,32). Studies have shown variability in detection rates among different age groups, especially in M. pneumoniae-endemic areas, where M. pneumoniae has occurred predominantly among children <5 years old (33,34).
A lack of consensus exists regarding the preferred specimen type for the identification of M. pneumoniae (35,36). We observed a significantly higher detection rate of M. pneumoniae in sputum compared to nasopharyngeal specimens, similar to results reported by Dorigo-Zetsma et al. (37) and Räty et al. (38). Although we detected a higher rate in sputum, nasopharyngeal specimens remain the preferred specimen type for surveillance because collecting a nasopharyngeal specimen is less invasive. In addition, a positive result on a nasopharyngeal specimen is a good indicator of disease as indicated by the AF.
During the M. pneumoniae epidemic that occurred in Europe during 2010-2012, P1 type 1 was dominant (24,39). In our population, P1 types 1 and 2 were circulating at equal frequencies. Likewise, in the United States, during an 8-year period (2006-2013) both P1 types were co-circulating (40). In China, during 2009-2011, P1 type 1, type 2, and variants of type 2 were identified; however, a higher frequency of type 1 compared with the other P1 types was observed (41). Continued surveillance is important to identify longer-term trends in M. pneumoniae strain prevalence in South Africa.
Macrolide resistance of 17% was documented in M. pneumoniae in Japan during 2000-2003 (43), with even higher rates of up to 90% reported in China (44). In Germany, 1.2% and 3% of M. pneumoniae found in respiratory tract specimens were resistant to macrolides during 2003-2008 and 1991-2009, respectively (45). In a US study, macrolide resistance was reported for ≈3% of M. pneumoniae cases in patients hospitalized with community-acquired pneumonia (46). However, other studies have reported macrolide resistance of 10%-13% in sporadic and outbreak specimens in the United States (40,47). Resistance in Europe and the United States remains low relative to Asia, possibly because of the restricted availability of antimicrobial drugs. We did not identify macrolide resistance among the isolates in our study, and therefore macrolide treatment is probably effective against M. pneumoniae in our setting. However, excessive use of macrolides should be discouraged, given that in Japan inappropriate use of macrolides was shown to increase the likelihood of the organism developing mutations in the 23S rRNA gene (11). Therefore, identification of the etiologic cause of infection and its appropriate treatment are essential. In South Africa, first-line treatment for community-acquired pneumonia is penicillin (48). In severely ill persons or those in whom atypical pneumonia is suspected, macrolides are administered. A limitation of our study is that treatment data for patients with M. pneumoniae were limited.
We performed molecular characterization for samples collected during periods 1 and 2 of our study. Most of the positive specimens were obtained during these 2 periods, and results from these periods can be inferred for period 3 because no intervention was implemented. We obtained a low yield of isolates and were unable to determine the macrolide susceptibility trait and strain type for a proportion of specimens, most likely because of a low bacterial load in the specimen, which might have affected the ability to detect resistance particularly if a low prevalence of macrolide-resistant M. pneumoniae strains exists in South Africa.
We have shown that, although the M. pneumoniae detection rate was low, M. pneumoniae detection is probably associated with illness, underscoring the need for testing, especially among patients at higher risk for severe disease.
Such testing would result in an earlier diagnosis and improved management. Our study provides baseline data that can be used for future surveillance programs to better understand M. pneumoniae epidemiology in South Africa.