Human bocavirus, coronavirus, and polyomavirus detected among patients hospitalised with severe acute respiratory illness in South Africa, 2012 to 2013

Abstract Aim To investigate the prevalence of human bocavirus (hBoV), human coronaviruses (hCoV), and human polyomaviruses (hPyV) among patients with severe acute respiratory illness (SARI), in South Africa. Methods The study included 680 South African patients randomly selected in age‐defined categories from hospitalised patients enrolled through SARI surveillance during 2012 to 2013. A multiplex reverse transcription real‐time polymerase chain reaction assay was used to detect hBoV; hCoV‐OC43, hCoV‐229E, hCoV‐NL63, and hCoV‐HKU1; and Washington University hPyV (hPyV‐WU) and Karolinska Insitute hPyV (hPyV‐KI), in respiratory tract specimens collected from patients with SARI. All respiratory specimens from patients enrolled through SARI surveillance were also routinely tested by multiplex reverse transcription real‐time polymerase chain reaction for adenovirus; enterovirus; human metapneumovirus; parainfluenza virus types 1, 2, and 3; respiratory syncytial virus; rhinovirus; influenza A, and influenza B. Results Human bocavirus, hCoV‐229E, and hPyV‐WU were detected in 3.7% (25/680), 4.1% (28/680), and 4.1% (28/680) of respiratory specimens, respectively. All other viruses were detected in <2% of specimens. Rhinovirus was the most common coinfecting virus (21.4%‐60.7%), followed by adenovirus (21.4%‐39.3%), and respiratory syncytial virus (10.7%‐24.0%). Testing for the additional viruses (hBoV, hCoV, and hPyV) decreased the number of specimens that initially tested negative by 2.9% (20/680). Conclusion Inclusion of laboratory tests for hBoV, hCoV‐229E, and hPyV‐WU in differential testing algorithms for surveillance and diagnostics for suspected cases of respiratory illness of unknown cause may improve our understanding of the etiology of SARI, especially in a country like South Africa with a high number of immune compromised persons.

Conclusion: Inclusion of laboratory tests for hBoV, hCoV-229E, and hPyV-WU in differential testing algorithms for surveillance and diagnostics for suspected cases of respiratory illness of unknown cause may improve our understanding of the etiology of SARI, especially in a country like South Africa with a high number of immune compromised persons.

| Determination of HIV status
For consenting patients, HIV results were obtained from medical records available at the time of enrollment, or testing of dried-blood spots were performed at the National Institute for Communicable Diseases, Johannesburg, South Africa. Human immunodeficiency virus testing included polymerase chain reaction (PCR)for children aged <18 months and enzyme-linked immunosorbent assay for patients aged ≥18 months.

| Study specimens, sample size, and laboratory procedures
Nasopharyngeal aspirates were collected from children aged <5 years, and combined nasopharyngeal and oropharyngeal swabs were collected from persons aged ≥5 years. Specimens were placed in viral (Highveld Biological, Johannesburg, South Africa) or universal (Copan, Murrieta, California, USA) transport medium and stored at 2 to 8°C.
Within 72 hours of collection, the specimens were transported to the National Institute for Communicable Diseases (Johannesburg, South Africa) for testing. Respiratory specimens were stored at −70°C.
After sorting by collection date, a subset of respiratory specimens was randomly selected in age-categories (<1, 1-4, 5-24, 25-44, 45-65, and > 65 years of age), for inclusion in this study. From a total of 5784 specimens collected during the surveillance period, every eighth specimen was then systematically selected for inclusion in this study to obtain a study sample of 680 (11.8%).

| Statistical analysis
Prevalence of hBoV, hCoV, and hPyV viruses were compared between age groups, HIV-infected and HIV-uninfected individuals, and individuals with and without non-HIV underlying medical conditions (predominantly, asthma, heart disease, pregnancy, diabetes, seizure disorders, and malnutrition) using the chi-square test or Fisher exact test. Stata® version 14 (StataCorp, College Station, Texas, USA) was used to implement the analysis.

| Study population
Among the patients included in this study, children aged <1 year accounted for the largest proportion of cases (46.8%; 318/680). Among individuals with available information, the HIV prevalence was 30.5% (204/668), and the prevalence of non-HIV underlying medical conditions was 7.5% (51/678). CD4 + count information was only available for a       limited number of individuals; therefore, these results could not be utilized. Age distribution and HIV prevalence of cases included in and excluded from this study were statistically similar (Table 1).

| DISCUSSION
The detection rate of hBoV, hCoV, and hPyV-WU was <4.2%. Children Although we detected hBoV in 3.7% (25/680) of specimens, we did not test these cases for hBoV mRNA or viral DNA load to assess the proportion of cases with active infection as described by Xu et al. 13 In a similar study conducted in South Africa during 2006 to 2007, 610 hospitalised children aged <5 years that required medical attention or hospitalization were tested for several respiratory viruses including those tested in our study 14 . Among these children, hBoV, hCoV-229E, and hPyV-WU were detected in 6.1% (37/610), 0.3% (2/610), and 3.4% (21/610) of specimens, respectively. In our study, a higher prevalence of hCoV-229E was reported (Table 2). This may be due to the different case definitions used in the 2 studies. Findings similar to ours were reported in a study conducted in the Gulf region in children <5 years of age. In the latter study, the reported detection rates for these viruses were 4.9% (14/331) hBoV and hCoV-229E, and 3.5% (10/331) hPyV-WU. 5 Globally, hBoV and hPyV-WU are predominantly observed in children aged <4 years with upper and lower respiratory tract infection, which was also observed in our study. 15 A recent South African study found that among children aged <1 year, hBoV (but not hCoV) detection was significantly associated with severe illness. 16 Human polyomaviruses was not investigated in the latter study. Therefore, inclusion of laboratory tests for hBoV, hCoV-229E, and hPyV-WU in differential testing algorithms for surveillance and diagnostics for suspected cases of respiratory illness of unknown cause may improve our understanding of the etiology of SARI, especially in a country like South Africa with a high number of immune compromised persons.
This study has limitations that warrant discussion. First, only patients with SARI were included in this study. As no control participant group was evaluated, the association and/or fraction of hBoV, hCoV, and hPyV detections that may contribute to severe respiratory illness could not be investigated. Second, the sample size was too small. Sero-prevalence data for South Africa is lacking; therefore, using 50% expected prevalence to obtain the biggest sample size for a 95% confidence interval, a 5% desired absolute precision, and a design effect of 1, plus inflation by 25% for spoilage/quality of samples, the annual expected sample for the study should have been 480 (960 for the 2 years). Lastly, the study duration was too short, as a 4-to 5-year period is required to assess seasonality and, therefore, the seasonality of HBoV, hCoV, and hPyV was not determined.
In conclusion, while low, the prevalence of hBoV, hCoV-229E, and hPyV-WU were similar to or higher than some other important viruses commonly tested for such as HMPV, PIV types 1 to 3, INFA, and INFB.
Including tests for hBoV, hCoV, and hPyV viruses in differential diag-