The Etiology of Pneumonia in HIV-infected Zambian Children

Supplemental Digital Content is available in the text.

I n the past 10 years, significant declines in new pediatric HIV infections and childhood HIV-related deaths have occurred, largely due to widespread access to highly efficacious prevention of mother to child transmission (PMTCT) regimens, as well as highly active antiretroviral therapy for children. 1 Despite these efforts, in 2016, there were an estimated 2.1 million children (<15 years) living with HIV and 160,000 new infections, the overwhelming majority (90%) living in Sub-Saharan Africa where the epidemic is most prominent. 1 The leading cause of morbidity and mortality in these children is pneumonia. [2][3][4] HIV-infected children are more likely to be hospitalized from pneumonia than any other illness, 5,6 are more likely to fail antimicrobial treatment 7,8 and are more likely to suffer worse outcomes (including death) from pneumonia than HIV-uninfected children. 7,9,10 In addition to common causative pathogens found in childhood community-acquired pneumonia [CAP; Streptococcus pneumoniae, Haemophilus influenzae, respiratory syncytial virus (RSV)], HIV-infected children were also found to be susceptible to opportunistic pathogens such as Pneumocystis jirovecii (Pj), cytomegalovirus (CMV) and Mycobacterium tuberculosis (MTB). 2,4,11,12 Because of these factors, the 2014 World Health Organization (WHO) recommendations for health facility treatment were updated for empiric antimicrobial treatment of CAP in HIVinfected children. 2,13 Unfortunately, the majority of studies used to support these recommendations were conducted before widely

Location
The Zambia PERCH study site was located at the University Teaching Hospital (UTH) in the densely populated capital, Lusaka (population 1.7 million 17 ). While Zambia is considered a lowermiddle-income country (per capita income $4300), at the time of the study (November 2011 to October 2013), approximately 74% of the country's population was living in extreme poverty (<$2/d). 18 UTH provides free health services to the most impoverished in Lusaka. As the primary academic and tertiary healthcare facility and the main country-wide referral center, UTH has a 425-bed inpatient pediatric ward with dedicated Malnutrition and Intensive Care Units. Access to mechanical ventilation was limited and rarely used. Obtaining radiographs required taking children to the Radiology Department at some distance from the pediatric wards and at the caregivers' expense, therefore, were not routinely performed (for nonstudy patients). Oxygen, however, was routinely available. The majority of children presenting for pneumonia care at UTH are referred from outlying Lusaka clinics after receiving 1 dose of antibiotics.
Hib conjugate vaccine was introduced in 2004 with 81% estimated 3-dose coverage before the study. 19 PCV10 was only universally introduced in July 2013 during the final 3 months of study enrollment.
In 2013, HIV prevalence in Lusaka among women of childbearing age was 19.4%. 20 Antenatal PMTCT care was nearly universal (91%) in Zambia, leading to a decline in vertical HIV transmission from 24% in 2009 to 12% by 2012. 21 By 2013, an estimated 54% of the 150,000 Zambian HIV-infected children were accessing antiretroviral therapy (ART). 22 While pediatric HIV seroprevalence rates at UTH were unavailable at the time of PERCH, a 2007 study found 29.2% HIV antibody positive rate among over 11,500 children tested in the pediatric ward. 23 At the time of PERCH, the standard of care for HIV-infected children <2 years was to initiate ART regardless of CD4 count or clinical staging. 24 For 2-to 5-year-old children, ART was initiated if the CD4 count was ≤750 cells/mm 3 , CD4 percentage <25% or clinical concern for advanced disease based on WHO staging. 24 The most common first-line ART regimen consisted of lopinavir/ ritonavir plus 2 nucleoside reverse transcriptase inhibitors. CTM prophylaxis beginning at 4-6 weeks of age was universally recommended and available for all HIV-infected and exposed children.

Participants
PERCH methods have been published elsewhere. 15,25 Unless otherwise noted, cases for this analysis were HIV-infected hospitalized children between the ages of 1 and 59 months living in the Lusaka catchment area presenting with signs and symptoms of WHO-defined severe or very severe pneumonia (2005 definition), including cough and/or difficulty in breathing, plus danger signs (central cyanosis, difficulty breast-feeding/drinking, vomiting everything, multiple or prolonged convulsions, lethargy/unconsciousness or head nodding) defined as "very severe pneumonia," or lower chest wall indrawing in the absence of danger signs defined as "severe pneumonia." 26 Cases were enrolled on weekdays from 0730 to 1800 hours due to weekend constraints for sample processing and limited weekend and nighttime staffing. Nighttime admissions were eligible for study enrollment the following morning.
HIV-infected controls, recruited from Pediatric HIV clinics in the Lusaka area without evidence of pneumonia, were age-frequency matched to HIV-infected cases using 4 strata: 1-5, 6-11, 12-23 and 24-59 months. A few HIV-infected controls were (incidentally) enrolled during routine community control recruitment. 16

Clinical Procedures
Cases were examined at admission and 24 and 48 hours postadmission. Cases that survived to discharge were seen 30 days after discharge to ascertain vital status. Chest radiographs (CXRs) were performed at admission and classified as normal, consolidation, other infiltrate, consolidation and other infiltrate or uninterpretable based on WHO methods. 27,28 Clinical assessments of controls were completed at the time of enrollment.

Specimen Collection and Laboratory Methods
Specimen collection and laboratory methods were highly standardized across study sites. [29][30][31][32][33] From all participants, we collected nasopharyngeal/oropharyngeal (NP/OP) swabs for polymerase chain reaction (PCR) for respiratory pathogens using a 33-pathogen multiplex quantitative PCR (FTD Resp-33; Fast-track Diagnostics, Sliema, Malta) and culture (plus serotyping) for pneumococcus, blood for pneumococcal PCR and serum for antibiotic activity testing. 31 From cases, we also collected blood for bacterial culture and induced sputum for MTB culture. For four pathogens with similar prevalence in cases and controls, positivity was defined using quantitative PCR density thresholds; including S. pneumoniae (≥2.2 log10 copies/mL) from whole blood 34 and S. pneumoniae (≥6.9 log10 copies/mL), 35 H. influenzae (≥5.9 log10 copies/mL), 36 CMV (≥4.9 log10 copies/mL) and Pj (≥4 log10 copies/mL), 36 NP/ OP (CMV threshold analysis available from authors). Maternal HIV status was obtained from the infant perinatal card or if the mother indicated she was HIV-infected. The child's blood was tested by PCR if <18 months or HIV antibody if ≥18 months as per Zambian guidelines.

Statistical Analysis
Odds ratios (OR) and 95% confidence intervals (CI) of pathogens detected on NP/OP PCR in cases compared with controls were calculated using logistic regression adjusted for age in months and presence of all other pathogens detected on NP/OP PCR to account for associations between pathogens. Logistic regression adjusted for age in months was used to compare clinical characteristics by case-control status and, among cases, by vital status. Results were stratified by HIV infection and exposure status.
The percent of pneumonia due to each pathogen was estimated using the PERCH Integrated Analysis (PIA) method, which is described in detail elsewhere (see reference 39, Appendix Section III B). [37][38][39] In brief, the PIA is a Bayesian nested partially latent class analysis that integrates the results for each case from blood culture, NP/OP PCR, whole-blood PCR for pneumococcus and induced sputum culture for MTB. The PIA also integrates test results from controls to account for imperfect test specificity of NP/ OP PCR and whole-blood PCR. Blood culture results (excluding contaminants) and MTB results were assumed to be 100% specific (ie, the etiology for a case was attributed 100% to the pathogen that S52 | www.pidj.com was detected in their blood by culture). The model assumes that each child's pneumonia was caused by a single pathogen.
The PIA accounts for imperfect sensitivity of each test/pathogen measurement by using a priori estimates of their sensitivity (ie, estimates regarding the plausibility range of sensitivity which varied by laboratory test method and pathogen) (Supplemental Digital Content 1, http://links.lww.com/INF/D818). Sensitivity of blood culture was reduced if blood volume was low (<1.5 mL) or if antibiotics were administered before specimen collection. Sensitivity of NP/OP PCR for S. pneumoniae and H. influenzae was reduced if antibiotics were administered before specimen collection.
As a Bayesian analysis, both the list of pathogens and their starting "prior" etiologic fraction values were specified a priori, which favored no pathogen over another (ie, "uniform"). The pathogens selected for inclusion in the analysis included any noncontaminant bacteria detected by culture in blood at any of the 9 PERCH sites, regardless of whether it was observed at the Zambia site specifically, MTB, and all of the multiplex quantitative PCR pathogens except those considered invalid because of poor assay specificity (Klebsiella pneumoniae 40 and Moraxella catarrhalis). A category called "Pathogens Not Otherwise Specified" was also included to estimate the fraction of pneumonia caused by pathogens not tested for or not observed. A child negative for all pathogens would still be assigned an etiology, which would be either one of the explicitly estimated pathogens (implying a "false negative," accounting for imperfect sensitivity of certain measurements) or "Pathogens Not Otherwise Specified." All analyses were adjusted for age (<1 vs. ≥1 year) to account for differences in pathogen prevalences by this factor; analyses stratified by pneumonia severity could not adjust for age due to small sample size. For results stratified by case clinical data (eg, to CXR+, very severe, etc), the test results from all controls were used. However, for analyses stratified by age, only data from controls representative of that age group were used.
The PIA estimated both the individual and population-level etiology probability distributions, each summing to 100% across pathogens where each pathogen has a probability ranging from 0% to 100%. The population-level etiologic fraction estimate for each pathogen was approximately the average of the individual case probabilities and was provided with a 95% credible interval (95% CrI), the Bayesian analog of the confidence interval. Statistical analyses were conducted using SAS 9.3 (SAS Institute, Cary, NC), R Statistical Software 3.3.1 (The R Development Core Team, Vienna, Austria), Bayesian inference software JAGS 4.2.0 (http:// mcmc-jags.sourceforge.net/) and BAKER, the R package used to perform the PIA (https://github.com/zhenkewu/baker).

Ethical Considerations
The study protocol was approved by the Institutional Review Boards at Boston University, the Johns Hopkins Bloomberg School of Public Health in the United States, and by the ERES Converge Ethical Review Committee in Zambia. Parents or guardians of participants provided written informed consent.
Only 36.9% (27.3% for children <1 year) of case caregivers reported knowing that their child was HIV-infected (Table 1). Pediatric HIV clinic attendance in the past 3 months was low in cases (18.5%) compared with controls (47.1%; P < 0.0001), as was being on ART (13.6% vs. 41.2%) and receiving CTM prophylaxis (35.0% vs. 70.6%; both P < 0.0001); only 2.6% of cases <1 year were receiving ART. Restricting to those children whose caregivers knew their children's HIV status, these characteristics were similar between cases and controls (Supplemental Digital Content 2, http:// links.lww.com/INF/D819). CD4 count data were largely unavailable and viral load testing was not routine during PERCH.

DISCUSSION
We present here updated clinical and etiologic findings among HIV-infected children in Zambia with CAP who are seen in a typical large urban Sub-Saharan African setting, characterized by high HIV prevalence and limited access to quality healthcare. Because PERCH enrollment in Zambia occurred during a period of greater routine childhood vaccine coverage (although limited access to PCV10), nearly universal access to PMTCT regimens, and increased access to pediatric HIV care (including CTM prophylaxis), we believed that our findings would differ from several foundational analyses on causes and outcomes of pneumonia in HIVinfected children conducted from 1990s to 2010. 2,4,7,10,11,41 However, we found that they are largely similar. Among HIV-infected children hospitalized with severe or very severe pneumonia, common bacterial pathogens, as well as Pj, remained a frequent cause of CXR+ pneumonia; relatively low ART and CTM coverage existed; etiology differed between HIV-infected and uninfected cases and HEU etiology was intermediate; malnutrition was common, and the mortality rate was high.
Six of the top 10 etiologies (S. pneumoniae, H. influenzae, Enterobacteriaceae, and S. aureus, Pj and MTB) are potentially treatable with available antibiotics and antituberculosis medications. While these same 6 organisms were also among the top 10 in our HIV-uninfected analysis, 16 their cumulative proportion was almost double (75.2% vs. 36.7%) in HIV-infected cases. We recognize the challenges in settings such as UTH in Zambia for conducting routine etiologic and antimicrobial resistance analyses, but periodic analyses, as well as updates to evidence-based treatment guidelines, 42  Prematurity defined as <37 wk gestational age or maternal report of premature. d HIV characteristics that were missing were assumed to be negative. e Moderate or severe malnutrition defined as less than −2 SD weight-for-age Z scores. f Defined as serum bioassay positive (cases and controls), antibiotics administered at the referral facility or antibiotic administration before whole-blood specimen collection at the study facility (cases only). g Very severe pneumonia defined as cough or difficulty breathing, and at least one of the following: central cyanosis, difficulty breast-feeding/drinking, vomiting everything, convulsions, lethargy, unconsciousness or head nodding. h Severe anemia defined as hemoglobin < 7.5 g/dL. i Leukocytosis count defined as >15 × 10 9 cells/L for children 1-11 mo and >13 × 10 9 cells/L for children 12-59 mo. j Hypoxemia defined as oxygen saturation <90% or on supplemental oxygen if a room air oxygen saturation reading was not available. k Lethargic or unresponsive (responds to voice or pain, unresponsive or pharmacologically sedated). l Duration of illness defined as duration (in days) of cough, wheeze, fever or difficulty breathing, whichever is longest. m Restricted to those children discharged alive. CRP indicates C-reactive protein; CXR, chest radiograph; DTP, diphtheria-tetanus-pertussis vaccine; HAART, highly active antiretroviral therapy; HIV, human immunodeficiency virus; IQR, interquartile range.
www.pidj.com | S55 therapy and/or validating current treatment guidelines. The preponderance of bacterial causes (50.6%) may have contributed to the higher in-hospital fatality rate in the HIV-infected cases, compared with HEU (21.6%) and HUU (11.1%) cases at the PERCH Zambia site.
Pj (24.9%) was the most common pathogen attributed as the cause of pneumonia among HIV-infected CXR+ cases (and almost nonexistent among HUU cases at our Zambia site), consistent with prior pneumonia studies and systematic pneumonia etiology reviews among HIV-infected children, and similar to the findings in the South Africa site (22.5%). 2,4,10,11,44 These findings were also consistent with 2 postmortem studies conducted 15 years apart in Zambia among children dying from reported respiratory causes at UTH. 41,45 In the Bates study, 45 conducted at the same time as PERCH, the smaller proportion of histopathologic evidence of Pj (9%) among HIV-infected deaths is likely attributable to an older median age of cases (19 vs. 6 months in PERCH), given the known higher risk during infancy; Pj was nearly absent (0.2%) among PERCH cases over 1 year of age. Due to small sample size, we were unable to determine whether Pj, and lack of CTM prophylaxis, were risk factors for in-hospital death in the multivariate analysis (data not shown). Regardless, Pj continues to play a large etiologic role in HIV-infected pneumonia cases in Zambia, stressing the importance of appropriate CTM prophylaxis and empiric high-dose CTM therapy.
S. pneumoniae was the second leading cause for pneumonia among Zambian HIV-infected CXR+ cases (19.8%), consistent with prior pneumonia studies 4,7 as well as studies reporting increased risk of invasive pneumococcus disease and bacteremia among HIV-infected children. 46,47 S. pneumoniae was much more common among severe versus very severe pneumonia cases [22.8% (95% CrI: 12.2-41.5) vs. 4.3% (95% CrI: 0.0-29.4)]. All serotypes identified by blood culture were covered by PCV10, introduced in Zambia at the end of PERCH enrollment, leading to cautious optimism that PCV10's introduction may lead to fewer pneumonia cases (and subsequent deaths) in Zambia's HIV-infected children. Post-PCV10 surveillance in Zambia, particularly for outcomes and etiologic causes of pneumonia in HIV-infected children, will be important.
MTB was estimated to cause 4.5% of CXR+ pneumonia among hospitalized HIV-infected cases in Zambia, 10.1% among HEU (10.1%) and 14.7% among HUU cases. Although these estimates are based on few MTB-positive cases (n = 6), a large systematic review of MTB prevalence among childhood pneumonia cases in similar settings supports MTB as a common cause, 12 as do 2 Zambian postmortem studies, 41,45 where MTB was common among acute pneumonia cases. MTB continues to be under-appreciated as a cause of acute pneumonia in children in settings similar to Zambia, and efforts for improving real-time diagnostics, and thus initiating treatment for MTB in children, should be made to improve childhood pneumonia and outcomes.  Despite overall improvements in access to care for HIV-exposed and HIV-infected children such as scaled-up pediatric HIV services, nearly universal access to CTM prophylaxis these beginning at 6 weeks of age, and improved early HIV-1 DNA PCR testing, only 37% of caregivers of HIV-infected cases were aware of their child's HIV status. Furthermore, both the low number of HIV-infected cases on ART (13.6% overall, 2.6% infants) and on CTM prophylaxis (31.2% of infants) at enrollment indicates an access to healthcare failure. When limiting to HIV-infected children whose caregivers were aware of their status (Supplemental Digital Table 2, http://links.lww.com/INF/D819), these numbers for accessing HIV preventative and definitive care remain surprisingly low, both for cases and controls. Reasons for such limited awareness and access to pediatric HIV care go beyond the scope of this analysis, but this underscores the need for earlier identification of HIV infection status and better caregiver education to ensure that appropriate life-saving chemoprophylaxis and ART can be given.
With 39.8% of cases dying in hospital, outcomes for HIVinfected severe and very severe pneumonia cases at UTH in Lusaka were very poor and much worse than for HEU and HUU children at our site. While similar studies in the region among HIV-infected children show high case fatality rates, notably in Malawi 10 and South Africa, 7,11 our in-hospital fatality rate remains alarmingly high. Some likely causes are well-established risk factors for poor outcomes, such as disease severity, younger age and danger signs ( Table 2). The larger proportion of bacterial, Pj and MTB causes for pneumonia also likely played a role, particularly with the lack of a standardized approach for care of HIV-infected children with pneumonia among the clinical staff at UTH. Poor immunologic status, high HIV-1 viral load and advanced WHO HIV clinical staging likely influenced outcomes (see below), but these data were unavailable in PERCH. Lack of mechanical ventilation and early resuscitative efforts in an ICU setting likely also contributed.
There were several study limitations particular to the Zambia PERCH site. First, 35% (N = 36) of cases were excluded from the primary etiology analysis due to missing or uninterpretable CXRs. Most missing CXRs were due to staff shortages causing delays in which sicker cases died before obtaining a CXR, or children were deemed too sick to attempt imaging. Second, nearly all children received antibiotics before referral to UTH, likely diminishing bacterial detection by culture and NP/OP PCR; thus, the bacterial fraction is likely underestimated. Third, with poor (~50%) posthospitalization follow-up, we likely underestimated mortality. Fourth, lacking immunologic (CD4 counts and percentages), viral load and HIV clinical staging status, we cannot assess these cofactors' influence on outcomes and etiology. Fifth, autopsies were not performed, an important though challenging process for informing etiology of fatal cases; however, in the contemporary Bates autopsy study, 45 evidence of bronchopneumonia (47%), pulmonary Line represents the 95% credible interval. The size of the symbol is scaled on the basis of the ratio of the estimated etiologic fraction to its standard error. Of 2 identical etiologic fraction estimates, the estimate associated with a larger symbol is more informed by the data than the priors. Bacterial summary excludes Mtb. Pathogens that were estimated at the subspecies level but grouped to the species level for display include Parainfluenza virus type 1, 2, 3 and 4; S. pneumoniae PCV10 and S. pneumoniae non-PCV10 types; H. influenzae type b and H. influenzae non b; and influenza A, B and C. Etiologic fraction estimates, including subspecies and serotype disaggregation (eg, PCV10 type and non-PCV10 type), are given in Supplemental Digital Content  www.pidj.com | S57 mycobacteria (12%) and Pj (9%) were found in HIV-infected children, closely matching our findings.
There are also limitations inherent to pneumonia etiology studies and case-control studies for pneumonia. Although PERCH relied on multiple samples (blood, NP/OP swabs, induced sputum) to estimate etiology, they were taken from locations outside the site of infection (the lung), a common problem for etiology studies. Despite using specimens obtained from outside the lung, some specimen results do have diagnostic potential, either based on standard clinical practice or association with case status, such as NP/OP PCR results for pertussis and RSV or induced sputum for tuberculosis. In addition, for case-control pneumonia etiology studies in children, bacteria which are commonly carried in the nasal passages of children likely lead to an overall underestimation of these bacteria as causative agents. Unfortunately, there is no easy method for distinguishing chronic carriage from newly acquired exposure or infection; however, we have mitigated some of the influence by using quantitative PCR and applied density thresholds that were associated with detection in the blood to improve the value of the data. Finally, the PIA model assumes each case's pneumonia episode is caused by a single pathogen and it does not attempt to identify or quantify pathogen combinations. While we acknowledge that copathogen causes will result in an underestimate of any single cause, exploratory analyses do not support a large underestimate (data not shown).
These results provide a clinical and microbiologic view of HIV-infected children with severe CAP in a typical Sub-Saharan African setting. These results should be viewed concomitantly and contrasted with the HIV-uninfected results from Zambia 16 and HIVinfected results from South Africa 44 in this issue. In aggregate, HIVinfected cases in Zambia had high in-hospital mortality, with common bacterial pathogens, Pj and MTB comprising a large proportion of the etiology. Until improvements in early HIV detection, appropriate CTM prophylaxis and early ART occur, outcomes among HIVinfected children with pneumonia in Zambia will likely remain poor.