Cohort profile: the potentially preventable burden of community-acquired pneumonia in South African adults in the era of widespread PCV13 immunisation and antiretroviral therapy roll-out, before and during the COVID-19 pandemic – the multicentre, multimethod PotPrev Study

Abstract

Purpose

In the setting of an established childhood pneumococcal vaccination programme with immediate initiation and treatment of antiretroviral therapy (ART) for people living with HIV (PLWH), the risk of adult pneumococcal community-acquired pneumonia (CAP) is not recently described. We aimed to investigate CAP incidence, recurrence, mortality, risk factors and microbiology before and during the COVID-19 pandemic.

Participants

Adults aged ≥18 years were enrolled in three South African provinces from March 2019 to October 2021, with a brief halt during the initial COVID-19 lockdown. The first group, PdCAP, a surveillance cohort, had their data abstracted to estimate the population incidence of physician-diagnosed CAP by counting incident CAP patients presenting to emergency rooms (ER) and mapping them to catchment areas linked to census data. From those admitted to wards from ERs, a prospective cohort (HospCAP) was enrolled and followed up to 1 year after discharge. Microbiology testing was performed and data were abstracted and collected for economic assessments. A third group (StART) of PLWH without respiratory illness at enrolment, attending primary healthcare clinics to initiate or reinitiate ART, was prospectively enrolled and followed. HospCAP and StART participants (totalling 2950 participants) were followed for at least 1 year and assessed for CAP episodes, hospitalisations and mortality.

Findings to date

Surveillance identified 6546 patients attending ERs with physician-diagnosed CAP; 61/6546 (0.9%) died in the ER. We prospectively enrolled 2000 hospitalised patients with CAP of whom 1079/2000 (54.0%) were PLWH. Overall, 271/2000 (13.6%) hospitalised CAP patients died during their first admission and 298/2000 (14.9%) died during follow-up. Among StART cohort, 18/950 (1.9%) died during follow-up.

Future plans

Planned analyses include incidence estimates of pneumococcal serotype-specific adult CAP and its recurrence, using Urinary Antigen Detection assay results to model the burden of pneumococcal CAP better and health economics analyses.

Correspondence to Dr Firdaus Nabeemeeah; [email protected]

STRENGTHS AND LIMITATIONS OF THIS STUDY

• We used WHO-based guidelines for defining catchment and enumeration areas helping to provide more accurate CAP prevalence estimates, particularly in low-resource settings.

• The study followed up a relatively large, well-characterised population of 2950 adults with CAP from both rural and urban areas across four different public sector hospitals.

• The prospective cohort design included multiple follow-ups, both via phone and in-person, to thoroughly track participant recovery, symptom resolution, complications and any new symptoms or diseases.

• A comprehensive range of clinical, laboratory and economic data was collected to analyse episodes and recurrences of pneumonia.

• The COVID-19 pandemic may have led to an underestimation of pneumococcal pneumonia cases, due to COVID-19 restrictions and the strain on hospital processes during waves, which likely affected documentation and triage practices.

Introduction

Streptococcus pneumoniae continues to predominate as the aetiological agent for community-acquired pneumonia (CAP) and is a leading infectious cause of admissions and deaths globally, particularly in low-income and middle-income countries (LMICs).^{1 2} Despite some provincial variation, high national infant 13-valent pneumococcal conjugate vaccine (PCV13) coverage has been achieved in South Africa,³ providing indirect protection against invasive pneumococcal disease (IPD) to unvaccinated groups, including people living with HIV (PLWH).^{4 5}

With an estimated 8.45 million PLWH in 2022,⁶ South Africa faces an extreme burden of HIV and associated opportunistic infections. PLWH are at 13–19 fold higher risk than seronegative individuals for severe respiratory illness⁷ and have a 24-fold increased risk of IPD and CAP, and for poor outcomes.^{8 9} Since the 2016 implementation of universal test-and-treat strategy with same-day antiretroviral therapy (ART) initiation, there has been a marked increase in PLWH receiving ART,^{10 11} reducing opportunistic infections including pneumonia.

Available data on the burden of vaccine-preventable CAP, vaccine effectiveness and cost-effectiveness of PCV, influenza and other vaccinations relevant to South Africa, predate paediatric PCV immunisation and test-and-treat ART initiation. Without current data on the potential benefit and cost-effectiveness of PCVs, investments in vaccines for adults in high HIV prevalence settings are unlikely.¹²

The overall objective of this study was to estimate the population-level adult annual incidence of CAP requiring hospital attendance in this LMIC setting, stratified by HIV serostatus, at sites with differing HIV prevalence and heterogenous childhood PCV13 coverage and to characterise risk factors and outcomes for CAP in these populations, to inform on the potential benefit of adult vaccinations administered in South Africa, especially to high-risk groups. We anticipated that incidence of CAP and mortality risk would be higher in facilities with higher HIV prevalence but that ART use and viral suppression status would attenuate this risk when adjusted for other risk factors.

Cohort description

Setting

Three hospitals (Chris Hani Baragwanath Academic Hospital, Klerksdorp/Tshepong Hospital Complex and Polokwane/Mankweng Hospital (PMH)), with university-affiliated Departments of Internal Medicine overseeing general adult medical wards and two close primary healthcare clinics (PHCs) were study sites, their provinces have varying HIV prevalence, tuberculosis (TB) prevalence, paediatric immunisation coverage and urbanisation rates (table 1).

Table 1

Characteristics of study sites used to conduct the PotPrev Study, South Africa, 2019–2022

	PMH	CHBAH	KTHC
Province	Limpopo	Gauteng	North West
Municipality	Capricorn District	City of Johannesburg	Matlosana Municipality
City/town	Polokwane	Soweto	Klerksdorp
Population size	438 486	1.7 million	1 372 355
Type of hospital	Joint regional and some tertiary functions	Tertiary/academic specialist services	Joint regional and some tertiary functions
Region serviced	Semi-rural/rural	Urban	Semi-urban
Urbanisation	29% urban, 71% rural	100% urban	92% urban, 8% rural
Beds per hospital	500	3200	890
Average daily adult admissions	10	100	40
HIV prevalence (2022) by province*	11.9%	15.0%	16.5%
General population ART coverage (2019/2020) by province†	77%	58.9%	59.9%
TB burden (2018)‡	Low	Low	High
Overall Paediatric immunisation (2019–2020 by province)§	70.5%	81.4%	75.2%
Study clinics	Evelyn Lekganyane Clinic Buite Clinic	Lillian Ngoyi Clinic Mofolo Clinic	Grace Mokhomo Clinic Jouberton CHC

*The Sixth South African National HIV Prevalence, Incidence, Behaviour and Communication Survey 2022: https://sahivsoc.org/Files/SABSSMVI-SUMMARY-SHEET-2023.pdf.

†District Health Barometer: https://www.hst.org.za/publications/District/Health/Barometers/DHB/2019-/20Section/A./chapter/2/-/Infectious/disease/control.pdf.

‡The First National TB Prevalence Survey South Africa 2018, Short Report: https://www.nicd.ac.za/wp-content/uploads/2021/02/TB-Prevalence-surveyreport_A4_SA_TPS-Short_Feb-2021.pdf.

§Expanded Programme on Immunisation national coverage survey report https://www.health.gov.za/wp-content/uploads/2022/03/National-EPI-CoverageSurvey_Final-full-report-Dec-2020.pdf.

ART, antiretroviral therapy; CHBAH, Chris Hani Baragwanath Academic Hospital; CHC, community health centre; KTHC, Klerksdorp-Tshepong Hospital Complex; PMH, Polokwane Mankweng Hospital; TB, tuberculosis.

Study objectives, population and enrolment criteria

Three groups of adults aged ≥18 years were reviewed to estimate CAP incidence and characterise risk factors and outcomes for CAP in this population, particularly among PLWH (table 2, figure 1). Eligible participants were enrolled from March 2019 to January 2022. Inclusion and exclusion criteria are shown in table 3. Study-specific definitions of CAP and other key terms are provided in online supplemental table 1.

Table 2

Study primary and secondary objectives and public health importance, Potprev Study, 2019–2022

Substudy	Type of objective	Objective	Public health importance
PdCAP	Primary objective	To establish population incidence of physician-diagnosed CAP in adults attending three study hospitals.	Gather recent evidence of CAP in high HIV, LMIC setting.
HospCAP	Primary objectives	1.1. To estimate the incidence of confirmed, hospitalised CAP 1.2. To estimate the incidence of recurrent CAP among a prospective cohort of hospitalised adults with confirmed CAP.	Gather recent evidence for hospitalised CAP and CAP associated mortality in high HIV, LMIC setting.
	Secondary objectives	2.1. To estimate population incidence of confirmed, hospitalised CAP with PCV13 Streptococcus pneumoniae detected by UAD and/or culture. 2.2. To determine the incidence of CAP patients with S. pneumoniae identified by BinaxNOW, and/or UAD 2.3. To estimate the incidence of recurrent cases of confirmed CAP. 2.4. To estimate the proportion of patients with PCV13 S. pneumoniae detected by UAD assay among those with recurrent confirmed CAP. 2.5. To determine the proportion of recurrent confirmed CAP with S. pneumoniae identified by BinaxNOW and/or UAD. 2.6. To determine the S. pneumoniae serotype distribution from UAD among initial and recurrent confirmed CAP cases. 2.7. To explore the impact of nasopharyngeal and oropharyngeal bacteria and viruses on subsequent morbidity. 2.8. To determine all-cause mortality and if possible cause-specific mortality. 2.9. To define risk factors for recurrence and for mortality assessing in particular the impact of HIV serostatus, age, smoking, among others. 2.10. To determine the proportion of hospitalised patients who are identified by LytA PCR and determine its diagnostic value compared with UAD or BinaxNOW 2.11. To determine if cough sound analysis can be used to understand and predict respiratory illness such as pneumonia or TB.
StART	Primary objectives	1.1. To measure the incidence of confirmed CAP among newly diagnosed PLWH or PLWH non-adherent to ART for ≥3 months 1.2. To identify risk factors associated with CAP among PLWH	Identify the risk factors associated with CAP in high-risk populations.
	Secondary objectives	2.1. To determine the S. pneumoniae serotype distribution from UAD among PLWH
	Exploratory objectives	3.1. To estimate pneumococcal carriage rate among PLWH 3.2. To estimate the influenza incidence rate and SARS-CoV-2 as determined by PCR swab among PLWH

ART, antiretroviral therapy; CAP, community acquired pneumonia; LMIC, low-income and middle-income country; PCV13, 13-valent pneumococcal conjugate vaccine; PLWH, people living with HIV; TB, tuberculosis; UAD, urinary antigen detection.

Table 3

Participant eligibility criteria by cohort for the Potprev Study in South Africa, 2019–2022

	PdCAP Adults attending hospital with presumptive CAP	HospCAP Adults hospitalised (from within PdCAP cohort) with confirmed CAP	StART Outpatient PLWH who were initiating or reinitiating ART without respiratory symptoms
Inclusion	≥18 years old
	Attend hospital or attached outpatient clinics	Willing to consent to the study, or in the event of acute incapacity due to illness, next of kin ≥21 years old willing to consent until participant has recovered sufficiently to consent personally or withdraw consent to data use.
	Presumptive CAP defined as cases listed in routine attendance documentation with a clinical diagnosis or differential of pneumonia*	Admitted to hospital (including overnight observation) where treating physician clinically suspects CAP in presence of 2 or more of the following symptoms and signs: Fever (oral >38°C or tympanic >38.5°C) within prior 24 hours; Hypothermia (<35.5°C) within prior 24 hours Pleuritic chest pain. Cough. Sputum production. Dyspnoea. Tachypnoea (respiratory rate >20/min). Abnormal auscultatory findings: rales or clinical evidence of pulmonary consolidation (dullness on percussion, bronchial breath sounds or aegophony).	Hard copy of laboratory test confirming HIV seropositivity for newly diagnosed patient OR Unsuppressed HIV viral load OR Known PLWH who has been non-adherent to ART for at least 3 months
			ART naïve OR Initiated ART for the first time in the past three weeks OR Restarted ART after a period of≥three months of non-adherence
		Resides in the relevant catchment area	Willing to start/reinitiate ART
		Positive chest X-ray†
Exclusion		Previous enrolment to HospCAP	Symptoms suggestive of acute (≤7 days) respiratory illness, including any duration of cough in the presence of fever (≥38°C)
		Prior or current enrolment in any study assessing novel vaccines or novel treatment/s for pneumonia or other bacterial illness.	Any history of hospital admission (including overnight stay) in the month prior.
		Unable to provide a contact telephone number or traceable address.	Unable to provide contact telephone number or traceable address
		Unable to complete the study visit requirements.	Planning to move from study area in the next 6 months

*Pneumonia and associated keywords including lower respiratory tract infection, tuberculosis etc included.

†Exceptions: PCP and COVID-19 where X-ray may be reported as normal.

ART, antiretroviral therapy; CAP, community-acquired pneumonia; PCP, Pneumocystis pneumonia; PLWH, people living with HIV.

Figure 1. Study flow disposition diagram for the Potprev Study in South Africa, 2019-2022. *Enrolment is not possible due to study logistical difficulty—inadequate equipment, staff or otherwise. †Prescreening processes excluded those with previous enrolment to HospCAP, enrolment in novel vaccine or treatment studies, attending clinics for symptomatic illness or with a recent admission. § <=50 copies/ml. CAP, community-acquired pneumonia; ED, emergency department; ER, emergency room; TB, tuberculosis; MHCU, mental health care user.

Methods

PdCAP: surveillance of physician-diagnosed CAP among adults presenting to emergency rooms

To estimate the population incidence of CAP in adults attending the three public sector hospitals

Study staff retrospectively conducted daily counts of all patients with a clinical diagnosis CAP as diagnosed by a physician and documented in routine records, counting patients who presented to adult emergency rooms (ERs), general medical wards, medical admission wards and intensive care units. Patient records and attendance logs were searched for any of the following keywords: “hypoxic pneumonia”, “lower respiratory tract infection”, “pneumocystis pneumonia”, “tuberculosis” or “bacterial pneumonia”, and “CAP” included in differential diagnoses, or conditions assessed as exacerbated by underlying CAP (eg, heart failure or acute asthma). Patients with a recent history of physical trauma to the chest, chemical pneumonitis or suspected nosocomial pneumonia evidenced by a recent hospital admission were excluded (figure 1, table 3). All consecutive CAP patients irrespective of disease severity and outcome of the emergency visit (admitted, discharged or deceased) were counted. Study staff abstracted age, gender, residential address, symptoms and their duration, and vital signs from hospital files. HIV serostatus, CXR findings and antibiotics prescribed were captured.

Sample size and planned data analyses

There were no recent estimates for time-bound population incidence rates of PdCAP from the study hospitals’ catchment areas on which to estimate sample sizes. Additionally, as this was a surveillance study, all eligible participants were included, and no sample size estimates were conducted.

Data analyses for all three cohorts (substudies) will be conducted according to a stasticical analysis plan. For the PdCAP cohort, our numerator was planned as all PdCAP cases over 1 year to account for seasonal variation (every patient entering the facility with a preliminary diagnosis of CAP was counted), but the surveillance period was extended to capture the impact of the COVID-19 pandemic. We based our determination of annual incidence on WHO directives.¹³ First, we defined the catchment area for each hospital by geolocating the street name of participants’ addresses at the time of seeking care using Google Earth to identify the central point of an enumeration area (EA) (online supplemental figure 1), which are the smallest geographical unit in South Africa for which census data is reported, consisting of 100–250 households.^{14 15} EAs where at least 80% of physician-diagnosed CAP cases resided were identified to define the hospital catchment area and thereby the denominator for incidence determination. Second, the most recent mid-term population size and age estimates were abstracted. The overall population incidence of CAP will thus be estimated using the number of patients with PdCAP, divided by the total population of the hospital catchment area as defined above. Subgroup incidence by gender, age and HIV serostatus will be estimated after adjusting for those using self or insurance-funded healthcare.¹⁶

HospCAP: prospective cohort of patients hospitalised for CAP

To estimate incidence of confirmed, hospitalised CAP and recurrent CAP and the proportion of potentially vaccine-preventable recurrence (CAP due to S. pneumoniae or influenza virus)

Staff approached enrolled PdCAP participants to obtain written informed consent. Adults admitted ≤48 hours previously with at least two CAP symptoms/signs, living in areas close to the hospital, with no history of recent hospitalisation in the 30 days prior to current admission and who had a suggestive CXR (excepting Pneumocystis jiroveccii pneumonia and COVID-19 in whom CXR may be reported normal) were eligible (figure 1, table 3).^{17 18} A signed proxy consent was obtained from a family member in those with acute delirium and a personally signed consent was obtained on recovery if they agreed to remain in the study.

At entry, and at recurrent hospitalisations for CAP during follow-up, we assessed the proportion of patients with pneumococcal VT serotypes (S. pneumoniae serotypes covered by PCV13) in whole blood lytA real-time PCR (lytA PCR)¹⁹ or a urine specimen subjected to Abbott BinaxNOW Streptococcus pneumoniae Antigen Card Test Kit (Abbott, Maine, USA) and pneumococcal serotype differentiation using the serotype-specific Urinary Antigen Detection (ssUAD) assays conducted at the Pfizer Vaccine Research and Development facility (Pearl River, New York, USA) (online supplemental table 2).

Participants were interviewed to obtain a detailed past medical, vaccination, smoking, substance use history and all demographic data, medical history, examination, investigations, health economics and management were abstracted. This included routine investigations done (complete blood count, urea, creatinine, C reactive protein); if not collected, study staff drew the specimen and the results were shared with treating doctors. Additional blood specimens included whole blood for PCR, CD4 count and HIV viral load (VL). Sputum was collected from all participants for TB testing using Xpert MTB/RIF Ultra (Cepheid, Sunnyvale, California, USA) and mycobacterial growth inhibitor tube automated liquid culture (Becton Dickinson, Franklin Lakes, New Jersey, USA). Two urine specimens were collected from those able to provide; one for two point-of-care tests: cotinine assay (Abon, Hangzhou, China) for all participants, and the DetermineTB LAM Ag test kit (Abbott Laboratories, Abbott Park, Illinois, USA) for PLWH; the second specimen was buffered, stored and shipped frozen to the Pfizer laboratory for BINAXNow and the ssUAD testing. Carbon monoxide breathalyser testing (CareFusion, Basingstoke, UK) was performed to quantify smoking. Nasopharyngeal (NP) and oropharyngeal (OP) swabs were collected and combined per participant in a single collection tube containing universal transport medium, and tested together for the detection of pneumococcus by lytA PCR (all study investigations in online supplemental table 2). Pneumococcus-positive samples then underwent serotyping by PCR. When possible, staff also recorded participants’ cough and voice while repeating a generic simple sentence in English for future analysis.

Follow-up of participants was three pronged: telephonic via a central call centre database; in-person every 4 months; and unscheduled visits when required in the case of illness, readmission or if scheduled visits were missed. Telephonic follow-up was planned weekly for the first 6 months after discharge and in the influenza season (May–October in South Africa),²⁰ otherwise fortnightly to symptom screen for new-onset respiratory illness, document new admissions and to record coughs and the study generic sentence. If new symptoms, readmission or death were reported, clinical staff were alerted. Ill participants were directed to an unscheduled visit for care or an emergency department. Recurrent CAP was a new episode of CAP, occurring ≥30 days after a prior CAP diagnosis, with either a history or clinical record of improvement from the initial episode.²¹

In-person visits included focused physical examination and NP/OP swabs were collected (stored in repository if no respiratory symptoms). In suspected pneumonia, study entry investigations were repeated and the participant was referred to routine care for treatment. Some health economics data were also collected related to the initial admission. At the final study visit, all those who were HIV seronegative at admission were offered voluntary counselling and testing by the study team if they had not tested again since enrolment.

Outcomes, clinical information and duration of participants’ admissions were abstracted from hospital records. In the event of death, this was supplemented with results from the National Health Laboratory database and by interview of family members. A thorough determination of cause of death (CoD) was completed by two investigators after access to treating physicians’ clinical notes, participant laboratory and histopathology results, radiological tests and clinical records to ascertain the most likely CoDs.²² If there was any ambiguity in the determination of CoD, a third senior internist’s advice was sought.

Sample size and planned data analyses

The sample size calculation was based on the incidence of recurrent CAP per 100 person-years. At the time of planning up the study, we assumed the incidence of recurrent CAP was between 51.8 and 55.5 per 100 person-years (initially planning 1700 HospCAP patients, which were increased following an interim analysis presenting incidence data). A range of power values between 80% and 90% and precision estimates between ±3.7% and ±5.2% were used to estimate possible sample size values. Adjusting for at least 20% loss to follow-up, a precision of ±3.7%, 85% power and an incidence rate of 55.5 per 100 person-years, at least 2000 participants would be enrolled in HospCAP. The number of CAP and recurrent CAP cases will be reported as events per 100 person-years and stratified by HIV or vaccine-preventable CAP. Characterisation of risk factors for recurrent CAP and mortality from CAP will be by initial χ² testing by study outcome (survived vs died) and by HIV status (negative, virally suppressed and virally unsuppressed). Kaplan-Meier plot and log-rank tests will compare probability of death at particular intervals since admission and Cox-proportional hazard regression to determine time-dependent risk factors associated with mortality.

StART: prospective cohort of adult PLWH without respiratory illness at entry, initiating or reinitiating ART from PHC clinics

To measure the incidence of confirmed CAP and identify risk factors associated with CAP among PLWH who are newly diagnosed or reinitaiting ART

At two primary clinics per site, we enrolled and followed up PLWH without concurrent respiratory illness (based on symptoms history and investigations) who had recently (≤3 weeks) initiated ART (newly diagnosed or reinitiating after ≥3 months non-adherence). Although initially excluding PLWH with viral suppression, owing to universal test-and-treat programmes and rapid viral suppression with dolutegravir, introduced in December 2019, many were already virally suppressed at enrolment and this criterion was removed. Only those presenting asymptomatic to the clinic for ART initiation or reinitiation were screened (figure 1, table 3).

After obtaining written informed consent, through participant interview and medical record abstraction, the following data were recorded: demographic, and medical history data, CD4 count and HIV VL. All participants provided a sputum specimen that was tested for TB in the same manner as in the HospCAP cohort. An NP/OP swab was sent for pneumococcal lytA PCR, respiratory syncytial virus (RSV), SARS-CoV-2 and influenza testing. Those who subsequently tested positive for either TB or SARS-CoV-2 were excluded from follow-up and analyses and referred to routine care. The remaining eligible participants were followed up for episodes of CAP as per the schedule of the HospCAP cohort.

Sample size and planned data analyses

The sample size for StART was revised from 1200 to 1010, based on an interim assessment of the incidence of CAP which was lower than originally anticipated (posited that the incidence rate of CAP would be 50% lower than that of HospCAP). Following adjustment for an incidence rate as low as 5 per 100 person-years, a range of precision estimates were applied (±2.5%, ±3.5%, ±4.0% and ±4.5%). At least 1000 participants would need to be enrolled assuming an incidence rate of 5 per 100 person-years, a precision of±2.5%, 87% power and a 20% adjustment of loss to follow-up. An additional 62 participants replaced those excluded retrospectively with asymptomatic laboratory-confirmed TB or SARS-CoV-2. The number of CAP cases and person-years of follow-up will be ascertained to estimate incidence. Risk factors for CAP and mortality will be investigated by preliminary χ² testing, followed by Cox-proportional hazards models to define time-dependent risk factors.

Additional analyses

Detection of pneumococcal pneumonia using the ssUAD and BinaxNOW

Three distinct groups of well-characterised participants provided urine samples for analysis using the BinaxNOW S. pneumoniae urinary antigen test and the ssUAD assay to detect pneumococcal serotypes. These three groups included HospCAP participants as well as two groups of participants without respiratory illness who served as controls (1) 400 PLWH from the StART cohort and (2) 400 confirmed HIV seronegative participants (at the time of consenting) recruited at the same clinics as the StART cohort. These 800 participants met additional eligibility criteria to be a control (without pulmonary illness or severe chronic disease and no recent pneumococcal vaccination) and provided a urine specimen. The results of ssUAD assays are for research and surveillance purposes only and were not used to influence treatment for participants (online supplemental table 3).

Health economics

We will estimate cost-effectiveness of PCV using health economics data collected from participants about their initial admission, and the estimated break-even price of a vaccine from the health provider perspective under varying estimates of vaccine effectiveness. These estimates will assume vaccination is administered either at discharge from hospital for adults with HospCAP or, for outpatient PLWH, at the point of initiating ART. Costing will be from both a societal and healthcare perspective. We will estimate the cost-effectiveness over a range of likely PCV effectiveness estimates; when administered to PLWH and HIV seronegative adults at the time of discharge from hospital for an episode of CAP, or in those initiating or reinitiating ART. The model will be used to define incremental cost-effectiveness ratios that include thresholds of disease incidence and vaccine effectiveness that are cost-effective for PLWH and adults discharged from hospital after a CAP episode. Sensitivity analyses and triangulation will be used to assess the robustness of estimates and conclusions, and their applicability in contexts beyond the study sites.

Impact of COVID-19 pandemic on the study

The study was initiated in March 2019. In March 2020, South Africa registered its first case of SARS-CoV-2, followed by 2 years of government-imposed restrictions on movement. From 27 March 2020 to 1 June 2020, access to, and seeking of non-COVID-19 medical care was limited,²³ thus participant recruitment and follow-up for this observational study were halted as per local ethics committee requirements. In this period, the protocol was amended to include SARS-CoV-2 as an aetiological agent for CAP, and the study team was trained on COVID-19, including SARS-CoV-2 testing and personal protective equipment use. Carbon monoxide breathalyser tests were stopped. Recruitment restarted on 1 June 2020, when the aetiology of CAP was dominated by SARS-CoV-2. There were no classic influenza or RSV seasons in 2020–2021. SARS-CoV-2-exposed or symptomatic staff were either quarantined or isolated and unable to work. Hard copy medical records were difficult to trace, especially for patients who died or were transferred to other hospitals and planned visit schedules were disrupted because neither staff nor participants were able to attend clinics. Follow-up visits were originally scheduled for 2 years after discharge but for those recruited after August 2021, follow-up was truncated to 1 year due to study interruptions and unplanned costs due to COVID-19 (online supplemental figure 2). A detailed questionnaire assessing the presence of symptoms suggestive of post-acute COVID-19 syndrome was added at the final visit.

Patient and public involvement

PHRU research sites have formalised collaboration with leaders and members of the local communities through a Community Advisory Board (CAB). Updates on study recruitment and findings were shared during community research days. Future dissemination to the public and patients will be in consultation with the CAB and at presentations to the patients and staff. Informed consent was obtained using the preferred languages of participants.

Findings to date

A total of 6546 patients with CAP were counted between March 2019 and January 2022 within PdCAP. Data for this cohort are still being reviewed and analyses will be published separately.

Preliminary findings for the two prospective cohorts are detailed below.

HospCAP: prospective hospitalised CAP

We enrolled 2000 participants; 1551/1556 (99.7%) enrolled during the COVID-19 era were tested for SARS-CoV-2 and 700/1551 (45.1%) were SARS-CoV-2 positive at admission. Females comprised 1014/2000 (50.7%) of the cohort and the median age at admission was 46 years (IQR 36–58); 1079/2000 (53.95%) were PLWH with a median CD4 count of 109 cells/mm³ (IQR: 33–276) (table 4). In PLWH with VL data, 288/993 (29.0%) were virally suppressed (≤50 copies/mL). Overall, 569 out of 2000 participants (28.5%) died, with a median time to death of 24 days (IQR: 7–161 days). Among the 569 deaths, 81 (14.2%) were PLWH who were virally suppressed, and 235 (41.3%) were HIV seronegative.

Table 4

Baseline characteristics of admitted adults with confirmed CAP (HospCAP) in the Potprev Study, 2019–2022

	Overall N=2000	CHBAH N=914	KTHC N=865	PMH N=221
Median age, years (IQR)	46 (36–58)	45 (36–58)	48 (35–59)	45 (34–56)
PLWH	41 (34–50)	41 (34–49)	41 (34–53)	39 (32–48)
Female (%)	1015 (50.8)	476 (52.1)	423 (48.9)	116 (52.5)
Enrolled in COVID-19 era* (%)	1557 (77.9)	753 (82.4)	583 (67.4)	221 (100.0)
PLWH (%)	1079 (54.0)	521 (57.0)	460 (53.2)	98 (44.3)
Self-reported on ART (%)	653 (60.5)	302 (57.9)	284 (61.7)	67 (69.1)
Median CD4, cells/uL (IQR)	109 (33–276)	91(23–234)	142(47–318)	75 (17–198)
Median VL, cps/mL (IQR)	n=993; 71 244 (920–4500)	n=510; 84 719 (920–5300)	n=392; 36 921 (423–2800)	n=91; 206 000 (6530–5300)
≤50 cps/mL (%)†	288 (29.0)	136 (26.7)	126 (32.1)	26 (28.6)
51–400 cps/mL (%)	109 (11.0)	62 (12.2)	44 (11.2)	03 (3.3)
Current smoker, self-report (%)	391 (19.6)	159 (17.3)	208 (24.0)	24 (10.9)
Urine cotinine positive (%)	467/1631 (28.6)	202/806 (25.1)	242/618 (39.2)	23/207 (11.1)
Alcohol, median units/week (IQR), male/female	6.0 (4–12)/4.0 (3–6)	6.0 (3–9)/3.0 (3–6)	10.0 (6–15)/6.0 (4–9)	4.5 (3–6)/4.0 (3–5)
Cardiovascular disease (%)	514 (25.7)	202 (22.1)	258 (29.8)	54 (24.4)
Diabetes (%)	204 (10.2)	85 (9.3)	78 (9.0)	41 (18.6)
Chronic pulmonary disease (%)	109 (5.5)	44 (4.8)	55 (6.4)	10 (4.5)
TB diagnosis at admission (%)	483 (24.2)	211 (23.1)	236 (27.3)	36 (16.4)
ssUAD positive (%)	268/1716 (15.6)	107/793 (13.5)	139/724 (19.2)	22/199 (11.1)
CURB-65 high risk (%)	43 (2.2)	20 (2.2)	15 (1.7)	8 (3.6)
Median hospital stay, days	8 (5–12)	8 (5–13)	7 (4–11)	8 (5–11)
Outcome of admission
Discharged (%)	1617 (80.9)	733 (80.1)	709 (82.0)	175 (79.6)
Died during admission (%)	271 (13.6)	135 (14.8)	102 (11.8)	34 (15.5)
Median follow-up, days (IQR)	533 (33–704)	511 (16–691)	591 (79–726)	478 (55–603)
Lost to follow-up (%)	161 (8.1)	50 (5.5)	96 (11.1)	15 (6.8)

*Enrolled after 20 March 2020.

†Considered virally suppressed.

ART, antiretroviral therapy; CAP, community-acquired pneumonia; CHBAH, Chris Hani Baragwanath Academic Hospital; CHC, community health centre; KTHC, Klerksdorp-Tshepong Hospital Complex; PLWH, people living with HIV; PMH, Polokwane Mankweng Hospital; ssUAD, serotype-specific Urinary Antigen Detection; VL, viral load.

Of those who survived their first admission, 98/1729 (5.8%) had a subsequent readmission; of those 44/98 (44.9%) were readmitted for recurrent pneumonia during their follow-up period. Median time of actual follow-up during that year was 533 days.

StART: adult patients initiating or reinitiating ART without respiratory illness at baseline

We enrolled 1072 participants and included in the analysis 950 who met eligibility criteria and completed at least 1 year of follow-up. At baseline, we excluded from analysis 26 participants with asymptomatic pulmonary TB and 29 with SARS-CoV-2 infection. Overall, 151/950 (15.9%) were virally suppressed. 24/950 (2.5%) of the StART participants died during follow-up, 3/950 (0.3%) were admitted for CAP and 135/950 (14.2%) were lost to follow-up (table 5).

Table 5

Baseline characteristics of outpatient participants without respiratory illness recruited to StART Cohort

	Overall N=950	CHBAH N=377	KTHC N=267	PMC N=306
Median age (IQR) years	35 [28–42]	35 [29–42]	35 [27–43]	34 [27–43]
Female (%)	597 (62.8)	217 (57.6)	168 (62.9)	212 (69.3)
Median CD4, cells/uL (IQR)	294 (143–481)	244 (124–450)	343 (156–516)	306 (152–477)
Median VL, cps/mL (IQR)	7558 (280–57 880)	23 284 (1784–98 071)	2833 (144–27 678)	3344 (232–34 656)
VL≤50 cps/mL (%)	151 (15.9)	54 (14.3)	49 (18.4)	48 (15.7)
Current/previous smoker, self-report (%)	297 (31.3)	98 (26.0)	135 (50.6)	64 (20.9)
Previous TB (%)	48 (5.1)	16 (4.2)	30 (11.2)	2 (0.7)
Asthma (%)	2 (0.2)	2 (0.5)	0 (0.0)	0 (0.0)
Diabetes mellitus (%)	2 (0.2)	1 (0.3)	0 (0.0)	1 (0.3)
Cardiovascular disease (%)	41 (4.3)	11 (2.9)	21 (7.9)	9 (2.9)
Epilepsy	1 (0.1)	0 (0.0)	1 (0.4)	0 (0.0)
Admission for CAP	3 (0.3)	0 (0.0)	1 (0.4)	2 (0.7)
Completed follow-up ≥1 year (%)	726 (76.4)	292 (77.5)	199 (74.5)	235 (76.8)
Lost to follow-up (%)	135 (14.2)	83 (22.0)	28 (10.5)	24 (7.8)
Death (%)	24 (2.5)	7 (1.9)	6 (2.2)	11 (3.6)
Median length of follow-up, days (IQR)	494 (364–691)	482 (361–682)	520 (360–698)	492 (369–630)

ART, antiretroviral therapy; CHBAH, Chris Hani Baragwanath Academic Hospital; CHC, Community Health Centre; KTHC, Klerksdorp-Tshepong Hospital Complex; PLWH, people living with HIV; TB, tuberculosis; VL, viral load.

Detection of pneumococcal pneumonia using the ssUAD and BinaxNOW

In total, 1794/2000 (89.7%) HospCAP participants provided urine samples for ssUAD testing, including 955 PLWH and 839 HIV seronegative participants. Of 98 patients who were readmitted, 30/98 (30.6%) provided a urine specimen. The leading reasons for not providing urine were dehydration, use of diapers and concurrent diarrhoea. Of the StART cohort, 588/1072 (54.9%) provided urine specimens for ssUAD at baseline (online supplemental table 3).

Discussion

Despite childhood vaccination and widespread ART uptake, our data suggest that the burden of CAP in South Africa, and particularly among PLWH remains high. Those PLWH admitted with CAP often have lower CD4 counts than PLWH initiating or reinitiating ART although whether this effect is due to acute illness requires further elucisation.²⁴ ART test-and-treat campaigns have, however, clearly, resulted in rapid viral suppression among adherents. S. pneumoniae as the causative agent for CAP among hospitalised patients remains substantial and considering the high mortality noted in our cohort raises the possibility that adult vaccination strategies may warrant further consideration.

Strengths

We successfully completed this study which started prior to COVID-19 and continued, after a brief interruption, during the remaining surges of the COVID-19 pandemic at three sites with varying HIV and COVID-19 burdens. A strength of the study is that all eligible patients were enrolled. Contemporaneous estimates of the adult annual incidence of hospital-attended, physician-diagnosed CAP, stratified by HIV serostatus, age and COVID-19 epochs will be generated. The ssUAD testing will provide valuable information on vaccine-preventable pneumococcal serotypes causing CAP in South African adults. Although initially designed to report pneumococcal CAP, detailed diagnostic data were collected describing prevalence and coinfection with respiratory pathogens including Mycobacterium tuberculosis, Pneumocystis jirovecii, Cryptococcus neoformans and SARS-CoV-2. Moreover, this study will provide updated data on morbidity and mortality and costs of the care of critically ill PLWH, stratified by their VL.

Limitations

Some CAP cases may have been missed if abbreviations or synonyms possibly relating to CAP used by medical doctors were not identified by study staff on reviewing medical records. We excluded patients from HospCAP (but not from PdCAP) who had severe CAP, who were intubated or too ill to consent if the family was unable to provide initial consent. Due to the movement restrictions and other challenges caused by the COVID-19 pandemic, the true incidence of CAP and its recurrence will be underestimated, especially as patient care was rationed with fatal consequences.²⁵ Telephonic follow-up was lower than expected as participant call responses fell, likely as a result of fatigue from high call numbers, as well as high turnover in cell phone numbers. This study was based in only three of nine provinces in South Africa and its generalisability to other settings may be limited.

We thank the participants for consenting to study procedures and follow-up. Staff members who worked on this project. Soweto: Khomotso Joy Zitha, Lingiwe Mnune, Nomathemba Mofokeng, Nokuthula Mbadaliga, Tshililo Mafunisa, Vallentia Mathapelo Mofokeng, Thabo Gerald Els, Ntombifuthi Luruli, Nonhlanhla Phume, Xoli Menziwa, Halalisile Cebekhulu, Selina Mashabane, Dumisani Mthembu. PHRU Klerksdorp: Andisiwe Sizani, Mbusi Ngema, Joseph Lechalaba,Tshegofatso Setlhare, Nozililo Mthembu,Mary Modisadife. Siza Khanyile, Obusitse Seipone. PHRU Polokwane: Kegaugetswe Motsomi, Thabo Phahlamohlaka, Omphemetse Masisi, Moorane Mogale, Anastasia Mohale, Alex Mahlare, Puledi Mashala NICD for their support regarding the logistics, SOPs and testing of all the NP/OP swabs. Provincial Departments of Health of Gauteng, North West and Limpopo, study hospital and district managers and hospital staff for hosting the study and providing participant care. Mr Hamza Cajee for accounting support and Ms Charity Leeuw for administrative and logistic support.

Data availability statement

Data available on reasonable request. Deidentified datasets used and/or analysed for the current study and biological specimens in the study repository are available from the corresponding author on reasonable request. Access will require approval from the funder, fully executed data and/or material transfer agreements and at a minimum, notification to the local Human Research Ethics Committee of the planned analyses.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

The Human Research Ethics Committee (Medical HREC), of the University of the Witwatersrand (reference no 171009) and provincial and facility Research Ethics Committees approved the protocol. Participants followed prospectively, or donating specimens, provided written informed consent and signed separate consents for storage of specimens. We specifically requested and were granted, a waiver by HREC for consenting patients in the ERs from whom we abstracted routinely collected de-identified data for PdCAP and with whom we had no direct contact.

Footnote

Contributors Conception of the work: NM, KO, EV, JS and SG. Designed the study and wrote the protocol: NM, KO, EV and JN with intellectual contributions from BDG, BN-T, AvG and CC. Provided project oversight: FN, NM, MF and JS. Responsible for clinical oversight: FN, NM, MW, EV, PM, NS and TM. Statistical analytical plan design: NM, KO with ED and EB. Oversaw laboratory testing and serotyping: FM. Data management: PA and FS. Data analysis and review: KO, KH, MM, LMM and PA. Drafting, revision and final approval of the manuscript: all authors. NM is the guarantor.

Funding This work was supported by Pfizer, grant number WI230621, under a collaborative research agreement with the Perinatal HIV Research Unit, a division of the Wits Health Consortium, which was the study sponsor.

Map disclaimer The inclusion of any map (including the depiction of any boundaries therein), or of any geographic or locational reference, does not imply the expression of any opinion whatsoever on the part of BMJ concerning the legal status of any country, territory, jurisdiction or area or of its authorities. Any such expression remains solely that of the relevant source and is not endorsed by BMJ. Maps are provided without any warranty of any kind, either express or implied.

Competing interests JS, BG, SG, EB, EMD, MF, BNT are employed by Pfizer and may own Pfizer stock or stock options. JN was a Pfizer employee and is currently a GSK employee. All other authors report no competing interests.

Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.

Provenance and peer review Not commissioned; externally peer reviewed.

Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

References

¹ Gessner BD. Severe Acute Respiratory Illness in Sub-Saharan Africa. J Infect Dis 2015; 212: 843–4. doi:10.1093/infdis/jiv104

² Steel HC, Cockeran R, Anderson R, et al. Overview of community-acquired pneumonia and the role of inflammatory mechanisms in the immunopathogenesis of severe pneumococcal disease. Mediators Inflamm 2013; 2013: 490346. doi:10.1155/2013/490346

³ Makamba-Mutevedzi P, Madhi S, Burnett R. Republic of South Africa expanded programme on immunisation (EPI) national coverage survey report. 2020. Available: https://www.health.gov.za/wp-content/uploads/2022/03/National-EPI-Coverage-Survey_Final-full-report-Dec-2020.pdf

⁴ Madhi SA, Nunes MC. The potential impact of pneumococcal conjugate vaccine in Africa: Considerations and early lessons learned from the South African experience. Hum Vaccin Immunother 2016; 12: 314–25. doi:10.1080/21645515.2015.1084450

⁵ Weinberger DM, Pitzer VE, Regev-Yochay G, et al. Association Between the Decline in Pneumococcal Disease in Unimmunized Adults and Vaccine-Derived Protection Against Colonization in Toddlers and Preschool-Aged Children. Am J Epidemiol 2019; 188: 160–8. doi:10.1093/aje/kwy219

⁶ Statssa. Mid-year population estimates. 2022. Available: https://www.statssa.gov.za/publications/P0302/P03022022.pdf

⁷ Cohen C, Walaza S, Moyes J, et al. Epidemiology of severe acute respiratory illness (SARI) among adults and children aged ≥5 years in a high HIV-prevalence setting, 2009-2012. PLoS One 2015; 10: e0117716. doi:10.1371/journal.pone.0117716

⁸ Harboe ZB, Larsen MV, Ladelund S, et al. Incidence and risk factors for invasive pneumococcal disease in HIV-infected and non-HIV-infected individuals before and after the introduction of combination antiretroviral therapy: persistent high risk among HIV-infected injecting drug users. Clin Infect Dis 2014; 59: 1168–76. doi:10.1093/cid/ciu558

⁹ Søgaard OS, Lohse N, Gerstoft J, et al. Mortality after hospitalization for pneumonia among individuals with HIV, 1995-2008: a Danish cohort study. PLoS One 2009; 4: e7022. doi:10.1371/journal.pone.0007022

¹⁰ Onoya D, Sineke T, Hendrickson C, et al. Impact of the test and treat policy on delays in antiretroviral therapy initiation among adult HIV positive patients from six clinics in Johannesburg, South Africa: results from a prospective cohort study. BMJ Open 2020; 10: e030228. doi:10.1136/bmjopen-2019-030228

¹¹ Rosen S, Maskew M, Fox MP, et al. Initiating Antiretroviral Therapy for HIV at a Patient’s First Clinic Visit: The RapIT Randomized Controlled Trial. PLoS Med 2016; 13: e1002015. doi:10.1371/journal.pmed.1002015

¹² Blasi F, Akova M, Bonanni P, et al. Community-acquired pneumonia in adults: Highlighting missed opportunities for vaccination. Eur J Intern Med 2017; 37: 13–8. doi:10.1016/j.ejim.2016.09.024

¹³ World Health O. A manual for estimating disease burden associated with seasonal influenza. World Health Organization; 2015.

¹⁴ Qader S, Lefebvre V, Tatem A, et al. n.d. Semi-automatic mapping of pre-census enumeration areas and population sampling frames. Hum Soc Sci Commun 8. doi:10.1057/s41599-020-00670-0

¹⁵ Statssa. Community survey 2007: methodology, processes and highlights of key results. 2007. Available: https://www.statssa.gov.za/publications/Report-03-01-20/Report-03-01-20.pdf

¹⁶ Statssa. General household survey. 2022. Available: https://www.statssa.gov.za/publications/P0318/P03182022.pdf

¹⁷ Stephanie S, Shum T, Cleveland H, et al. Determinants of Chest X-Ray Sensitivity for COVID- 19: A Multi-Institutional Study in the United States. Radiol Cardiothorac Imaging 2020; 2: e200337. doi:10.1148/ryct.2020200337

¹⁸ Wills NK, Adriaanse M, Erasmus S, et al. Chest X-ray Features of HIV-Associated Pneumocystis Pneumonia (PCP) in Adults: A Systematic Review and Meta-analysis. Open Forum Infect Dis 2024; 11: ofae146. doi:10.1093/ofid/ofae146

¹⁹ Carvalho M da G, Tondella ML, McCaustland K, et al. Evaluation and improvement of real-time PCR assays targeting lytA, ply, and psaA genes for detection of pneumococcal DNA. J Clin Microbiol 2007; 45: 2460–6. doi:10.1128/JCM.02498-06

²⁰ McAnerney JM, Cohen C, Moyes J, et al. Twenty-five years of outpatient influenza surveillance in South Africa, 1984-2008. J Infect Dis 2012; 206 Suppl 1: S153–8. doi:10.1093/infdis/jis575

²¹ Garcia-Vidal C, Carratalà J, Fernández-Sabé N, et al. Aetiology of, and risk factors for, recurrent community-acquired pneumonia. Clin Microbiol Infect 2009; 15: 1033–8. doi:10.1111/j.1469-0691.2009.02918.x

²² WHO. International classification of diseases, eleventh revision (ICD-11) reference guide. 2021. Available: https://icdcdn.who.int/icd11referenceguide/en/html/index.html?sfvrsn=9ec05f86_1#icd11-reference-guide

²³ Burger R, Day C, Deghaye N, et al. Examining the unintended consequences of the COVID-19 pandemic on public sectorhealth facility visits: the first 150 days. 2020. Available: https://cramsurvey.org/wp-content/uploads/2020/12/16.-Examining-the-unintended-consequences-of-the-COVID-19-pandemic-on-public-sector-health-facility-visits-The-first-150-days-2.pdf

²⁴ Battistini Garcia S, Guzman N. Acquired immune deficiency syndrome CD4+ count. Treasure Island, FL: StatPearls Publishing, 2024. Available: https://www.ncbi.nlm.nih.gov/books/NBK513289

²⁵ Sabet N, Omar T, Milovanovic M, et al. Undiagnosed Pulmonary Tuberculosis (TB) and Coronavirus Disease 2019 (COVID-19) in Adults Dying at Home in a High-TB-Burden Setting, Before and During Pandemic COVID-19: An Autopsy Study. Clin Infect Dis 2023; 77: 453–9. doi:10.1093/cid/ciad212