The Epidemiology of Pulmonary Hypertension in Sub‐Saharan Africa—A Systematic Report

ABSTRACT

Noncommunicable diseases (NCDs) are on the rise in sub‐Saharan Africa (SSA), with cardiovascular NCD accounting for most deaths. Pulmonary hypertension (PH) is a highly morbid NCD; however, contemporary prevalence data are lacking. We performed a systematic literature search of all echocardiographic or right heart catheterization studies performed in SSA. We grouped studies based on disease populations and performed random‐effects meta‐analyses, meta‐regression, and extrapolation analyses. We included n = 105 reports, comprising 23,384 individuals. Most investigated cohorts included sickle cell disease (SCD; n = 24), human immunodeficiency virus (HIV; n = 19), rheumatic heart disease (RHD; n = 18), hypertensive heart failure (HHF; n = 7), and other heart failure (HF) syndromes (n = 18), while others, including chronic obstructive pulmonary disease (COPD), pulmonary tuberculosis (TB), chronic kidney disease (CKD) and schistosomiasis were underreported and prevalence of PH in patients with specific PH‐predisposing diseases from other global regions was used. Estimated PH prevalence was strongly dependent on tricuspid regurgitation velocity (TRV) threshold, referral reason for echocardiography (HIV), and disease severity (RHD, HHF, HF). When taking these factors into account, extrapolation suggests that up to 1.2% may had “likely” PH TRV > 3.4 m/s) and 1.9% “at risk” PH (TRV > 2.8 m/s) in SSA in 2021, with the largest contributor being RHD (31%), followed by other HF (25%), COPD (18%), CKD (14%), HIV (5%), SCD (4%), schistosomiasis (2%), and TB (1%). PH has a unique epidemiological profile in SSA and based on our extrapolation methods PH prevalence is substantial but varies by disease population, disease severity, and specificity of diagnostic criteria. Population health initiatives focusing on PH prevention, detection, and management in SSA are warranted.

Trial Registration: PROSPERO: CRD42023478199.

1. Background

Until recently, infectious diseases were the single most important cause of disability adjusted life years (DALYS) and deaths in sub‐Saharan Africa (SSA) [1]. However, recent trends show that because of enhanced screening, transmission breaks, and treatment initiatives for human immunodeficiency virus (HIV) and pulmonary tuberculosis (TB), noncommunicable diseases (NCDs) are on the rise [2]. Cardiovascular diseases are the largest contributor to the NCD burden in SSA, accounting for up to 38% of NCD deaths in 2021 [2]. In fact, ischemic heart disease and chronic obstructive pulmonary disease (COPD) were amongst the four leading causes of death (next to COVID‐19 and stroke) in low‐ to middle‐income countries (LMICs) in 2021 [3].

Pulmonary hypertension (PH) is a heterogenous condition [4], defined by an elevated mean pressure in the pulmonary arteries, which is diagnosed with right heart catheterization (RHC) [5]. In turn, echocardiography presents the most important noninvasive tool to suspect PH, is quantitative, and used for clinical prognostication [6]. PH development is associated with a wide range of diseases endemic to SSA, such as HIV and TB, but also rheumatic heart disease (RHD) and sickle cell disease (SCD). When present, PH is independently associated with decreased quality of life, exercise capacity, and shortened lifespan [7, 8, 9, 10, 11]. Importantly, PH is an important marker of disease severity and identifies patients at increased risk who warrant careful evaluation and optimization of the underlying condition. For example, adequate treatment of underlying chronic kidney disease (CKD) can decrease PAP, which is associated with improved longevity [12].

Most available data on PH and its prevalence originate from high‐income countries, although these only represent ~25% of the world's population. The scope and prevalence of PH in SSA, however, remains widely unknown. In this study, we aimed to report on the prevalence of PH in SSA based on comprehensive systematic literature searches, meta‐analyses, and data extrapolation methods. Findings from this study establish a scope of PH prevalence in SSA that is essential to understand the burden of PH‐associated risk in SSA. Epidemiology data are positioned to help aim direction of public health programs, specifically tailored to needs and considerations unique to SSA, with the goal to reduce PH‐associated risk, improve quality of life and outcome in at‐risk populations living in SSA.

2. Methods

2.1. Sub‐Saharan Africa

SSA is a diverse global region comprising 49 countries south of the Sahara Desert. The region is home to ~1.19 billion people in 2023, characterized by a relatively young population with a median age of 19 years [13].

2.2. Systematic Review

This is a systematic literature review, meta‐analysis, and extrapolation. This study was conducted and reported following the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) checklist, the Meta‐analysis of Observational Studies in Epidemiology (MOOSE) reporting guidelines, and the Guidelines for Accurate and Transparent Health Estimates Reporting (GATHER). The study protocol was registered with PROSPERO (CRD42023478199).

2.3. Search Strategy and Eligibility Criteria

We conducted a comprehensive systematic literature search in PubMed, Embase, Web of Science, and LILACS with no language or age restrictions for articles published between 1945 and September 1, 2023, in SSA. Detailed information on specific search strings are given in Online Supplement S1. Briefly, we performed three independent searches for (1) “pulmonary hypertension,” (2) “echocardiogra*,” and (3) “right heart catheter*” or “pulmonary cath*.” Each search string was used in combination with 15 predefined countries in SSA, balanced for income level and population size, resulting in 45 independent searches. We included primary quantitative studies that provided (1) a clear PH definition and PH prevalence, (2) a systolic pulmonary arterial pressure (sPAP) or tricuspid regurgitation velocity (TRV) by echocardiography, or (3) a mean (or median) pulmonary arterial pressure (mPAP) by RHC with corresponding standard deviation (or interquartile range). We excluded studies with unclear PH definitions, studies using billing or ICD codes to define PH, secondary studies (including scoping, rapid, and systematic reviews), studies conducted outside the SSA region, studies that only included PH populations or where PH served as exclusion criteria.

2.4. Study Selection, Data Extraction, and Study Quality Assessment

The search outputs were exported into EndNote to remove duplicates and uploaded to Covidence for title and abstract screening, full‐text review, and data extraction, which was done in duplicate by at least two independent researchers (K. Z., V. N., A. S., E. W. R.). Data extraction was performed using a standardized extraction form; a full list of all variables is shown in Table S1. Any disagreements were adjudicated through group consensus opinion. All data for this study were obtained from the published literature, and ethical approval for this study was waived. Critical appraisal of the included studies was conducted independently and in duplicate to determine the risk of bias using the tool by Munn et al. [14] for epidemiological studies. An overall risk of bias was assigned to each study and is summarized in Table S2. Publication bias was assessed using Funnel plots and Egger's test and was not significant (p > 0.5) for all meta‐analyses.

2.5. Definitions

Based on European Society of Cardiology (ESC)/European Respiratory Society (ERS) 2022 Guidelines [6] as well as proceedings from the World Symposium of PH 2024 [15], we used three different criteria for assessing PH prevalence. For echocardiography studies, peak TRV > 2.8 m/s [6] or sPAP > 35 mmHg was used to group patients with at least moderate PH probability into “at risk” PH. SPAP was estimated from TRV in addition to right atrial pressure. A TRV > 3.4 m/s [6] was used to group patients with high PH probability into “likely” PH. For RHC studies, if available, mPAP > 20 mmHg was used to confirm PH diagnosis [6]. The diagnoses of the underlying analyzed cohorts were based on the respective guidelines and are summarized in Table S3.

2.6. Data Extrapolation

To contextualize the number of patients affected by PH in SSA in total, we used the published prevalence of these diseases from the literature, as presented in Table S4. We weighed our extrapolation based on the referral scenarios for primary extra‐cardiopulmonary diseases, such as SCD and HIV, into referred for evaluation of cardiopulmonary symptoms versus asymptomatic echocardiographic screening studies. This covariate was determined a priori. For cardiac disease populations, including, for example, RHD and HF, we weighted our extrapolation based on the severity of RHD into mild versus severe forms, as defined by indication for surgical intervention in the case of the latter. For hypertensive heart failure (HHF), we used left ventricular ejection fraction (LVEF) to distinguish between systolic and diastolic heart failure. These categories for cardiac subgroups were determined based on results from the meta‐regression analysis, where these factors turned out to significantly influence PH prevalence (Table S7).

There are paucity of data using echocardiography or RHC to evaluate PAP in SSA. For example, there were only n = 1 and n = 2 studies investigating TB and kidney failure (hemodialysis cohorts only) and none that analyzed COPD or schistosomiasis. For these highly prevalent diseases, PH prevalence estimates were used from published literature derived from other global regions (Tables S4 and S5) in order to provide an overall estimation of possible affected persons by PH in SSA. Echocardiographic TRV > 2.8 m/s was used to assess “at risk” PH and RHC confirmed prevalence data for confirmed cases to increase specificity over echocardiography, as RHC was largely available from these regions.

2.7. Data Analysis

2.7.1. Meta Analysis Summary

We performed separate meta‐analyses for each PH‐associated disease using random‐effects models. The observed proportions were transformed using the Freeman–Tukey double arcsine method [16, 17]. Subsequently, a back‐transformation was computed to convert the transformed proportions back to the original proportion scale. The summary effect size was pooled using inverse‐variance weighting, where variance was defined as the sum of within‐study variance and the between‐study variance [18]. Between‐study heterogeneity variance (${\tau }^2$) was estimated using the restricted maximum likelihood estimator [19], and its confidence interval was obtained via the Q‐profile method. A p < 0.05 was considered statistically significant. Where possible, univariate meta‐regression analyses were conducted to assess whether study‐level characteristics (e.g., average age, percent male, study design, region, study endpoint) explained heterogeneity in prevalence estimates. Results of the meta‐regression as well as effect size are given in Table S7. A 95% confidence interval for the pooled prevalence was calculated assuming a normal approximation. Detailed information on data harmonization, meta‐regression analysis, data imputation, and data extrapolation is given in Online Supplement S1.

3. Results

This systematic review yielded N = 105 reports from n = 94 studies with a total of 23,384 subjects (Figure 1). Baseline characteristics for the included studies summarized by underlying disease are shown in Table 1. Individual study reports as well as risk of bias assessment are available in Table S6.

Figure 1.

PRSIMA flowchart. PAP, pulmonary arterial pressure; PH, pulmonary hypertension; WHO, World Health Organization.

Table 1.

Baseline characteristics of included studies grouped based on disease cohort.

	HIV	Sickle cell disease	Rheumatic heart disease	Hypertensive heart failure	Congenital heart disease	HFrEF other	Right heart failure
Reports, N	16	24	18	7	4	12	2
Individuals, N	3389	4379	3181	1173	451	2368	818
Age, years	39 ± 16	18 ± 8	32 ± 14	54 ± 14	3 ± 10	56 ± 15	54 ± 16
Sex, %male	40 ± 12	52 ± 7	17 ± 12	53 ± 16 (n = 4 missing)	40 ± 10	54 ± 30	42 ± 5
Publication year, range (median)	2005–2023 (2019)	2007–2023 (2018)	1991–2023 (2011)	2008–2019 (2012)	2015–2022 (2021)	2008–2023 (2018)	2008, 2011
Study endpoint, PH vs. no PH	4/12	15/9	2/16		3/1	5/7	0/2
Clinical indication/research only	4/12	3/19	—	—	—	—	—
Design
Prospective/retrospective/NA	16/0/0	22/1/0	7/11/0	7/0/0	3/1/0	10/2	2/0/0
Random/not random	2/14	0/24	0/18	0/7	0/4	2/10	0/2
Hospital/community based/NA	15/1/0	21/2/1	18/0	7/0	4/0	12/0	2/0/0
Single/multicenter	15/1	17/7	15/3	7/0	1/3	10/2	2/0
Geography of SSA West/East/South/Central/Overlap	2/8/6/0/0	20/1/0/0/3	2/3/13/0/0	6/0/1/0/0	3/1/0/0/0	6/3/3/0/0	0/0/2/0/0
Income level Low/lower‐middle/upper‐middle	5/8/3	0/24/0	3/2/13	0/6/1	1/3/0	1/8/3	0/0/2
Ejection fraction, %	—	—	—	40 ± 16	—	37 ± 14	55 ± 17
Hemoglobin, g/dL	—	8.5 ± 0.7	—	—	—	—	—
Disease severity, mild/severe	—	—	6/12	—	—	—	—
Stable/crisis	—	23/1	—	—	—	—	—
Risk of bias, low/middle/high	11/2/3	17/4/3	13/1/4	7/0/0	4/0/0	12/0/0	2/0/0

Abbreviations: HFrEF, heart failure with reduced ejection fraction; PH, pulmonary hypertension; SSA, sub‐Saharan Africa.

3.1. Prevalence Estimates for PH Associated With HIV

The search yielded n = 16 reports from n = 3389 people living with HIV with a mean age of 39 ± 16 years. In all studies, PH was assessed using echocardiography. Baseline characteristics are shown in Table 1. Estimated prevalence for “at risk” PH was found in 4.8% (95% CI: 1.9%–8.7%) of the patients and “likely” PH in 0.3% (95% CI: 0.03%–0.7%; Table 2). Estimated PH prevalence varied significantly based on whether patients were referred for echocardiographic evaluation due to cardiopulmonary symptoms (TRV > 2.8 m/s: 16.6% [95% CI: 8.6%–26.6%]) compared to screening in asymptomatic individuals (TRV > 2.8 m/s: 2.3% [95% CI: 20.7%–4.6%], p < 0.001). These prevalence estimates were drastically decreased when using the TRV > 3.4 m/s threshold to 2.1% (95% CI: 0.2%–5.2%) and 0.3% (95% CI: 0.03%–0.07%), respectively. Based on a published report that implemented a survey in people living with HIV in an HIV outpatient clinic, 84% of all patients did not report any cardiopulmonary symptoms, while 16% had dyspnea or chest discomfort of any cause [20]. Based on WHO estimates, there were 26 million people (95% CI: 23.6–28.8) living with HIV in SSA in 2023 [21]. Using this as a referent and weighting our calculations based on the percentage of the respective cohorts that did or not did not present with symptoms and their respective PH prevalence, our extrapolation suggests that there might be up to 1.2 million people (95% CI: 0.5–2.4) with “at risk” PH‐HIV and 0.16 million people (95% CI: 0.05–0.43) with “likely” PH‐HIV in SSA in 2023 (Table 2 and Figure 2).

Table 2.

PH prevalence estimates based on TRV > 2.8 m/s and TRV > 3.4 m/s based on the systematic literature search and meta‐analysis for HIV, SCD, and RHD.

Cohort	TRV	Estimated PH prevalence (95% CI)	Estimated PH prevalence (95% CI)	Prevalence in SSA, million (95% UI)	PH‐associated prevalence (95% UI)
		Asymptomatic screening	Cardiopulmonary symptoms
HIV	> 2.8 m/s	2.3 (0.7–4.6)	16.6 (8.6–26.6)	26.0 (23.6–28.8)	1,223,000 (429,870–2,015,380)
	> 3.4 m/s	0.3 (0.03–0.7)	2.1 (0.2–5.2)	26.0 (23.6–28.8)	155,220 (55,690–432,600)
SCD	> 2.8 m/s	14.1 (9.8–19.0)	12.8 (1.5–31.9)	5.7 (4.8–6.6)	812,420 (415,230–1,209,610)
	> 3.4 m/s	0.4 (0.0–2.1)	7.6 (2.2–15.4)	5.7 (4.8–6.6)	162,160 (19,360–463,620)
		Mild disease	Surgical indication
RHD	> 2.8 m/s	25.7 (13.8–39.8)	78.2 (69.9–85.4)	15.5 (11.8–19.6)	9,178,630 (6,661,030–11,696,230)
	> 3.4 m/s	20.2 (15.0–28.7)	61.3 (45.6–75.8)	15.5 (11.8–19.6)	7,212,220 (4,695,230–9,729,220)

Abbreviations: HIV, human immunodeficiency virus; RHD, rheumatic heart disease; SCD, sickle cell disease; TRV, tricuspid regurgitation velocity.

Figure 2.

Prevalence estimates for PH associated with HIV. *Selection bias and other limitations associated with extrapolating prevalence estimates from targeted cohorts to the broader population may substantially increase the uncertainty surrounding these estimates.

3.2. Prevalence Estimates for PH Associated With SCD

We identified n = 24 reports including n = 4397 subjects with SCD in our systematic review. The mean age of the cohorts was 18 ± 8 years (Table 1). Across all studies, the estimated “at risk” PH prevalence was 13.9% (95% CI: 9.8–18.5). For SCD, we did not identify a significant difference in PH prevalence estimates in terms of echocardiographic referral. Based on a meta‐analytic result of two cross‐sectional studies originating from SSA, 33% of outpatient SCD patients reported any cardiopulmonary symptoms [22, 23]. The PH prevalence for patients presenting with cardiopulmonary symptoms was 14.1% (95% CI: 9.7–19.0; Table 2) compared to patients who volunteered to undergo asymptomatic screening, which was 12.8% (95% CI: 11.5–31.9; p = 0.819). The prevalence of “likely” PH was only 0.8% (95% CI: 0–2.9; based on 10 reports). The all‐age prevalence of SCD in SSA in 2021 was estimated to be 5.7 million (95% CI: 3.8–6.6) [24]. Using this as a referent and weighing our calculations for echocardiographic referral based on symptoms, we extrapolated that there might be 0.776 million (95% UI: 0.273–1.605) with “at risk” PH‐SCD and 0.162 million (95% UI: 0.019–0.463) with “likely” PH‐SCD in SSA in 2021 (Table 2 and Figure 3).

Figure 3.

Prevalence estimates for PH associated with SCD. *Selection bias and other limitations associated with extrapolating prevalence estimates from targeted cohorts to the broader population may substantially increase the uncertainty surrounding these estimates.

3.3. Prevalence Estimates for PH Associated With RHD

In total, we included n = 18 studies comprising n = 3181 patients with a mean age of 32 ± 14 years (Table 1). Echocardiography was used to diagnose PH in n = 16 studies (89%) and RHC in n = 2 (11%). Overall, across the echocardiographic studies the estimated prevalence of “at risk” PH was 59.9% (95% CI: 43.3%–73.7%), which was 3‐fold greater in severe (=indication for surgical repair) versus mild RHD (78.2% [95% CI: 69.6%–85.4%] vs. 25.7% [95% CI: 13.8%–39.8%]; p < 0.001; Table S7). Based on published data from the 2021 Global Burden of Disease Study, RHD affected 15.5 (95% uncertainty interval [UI]: 11.8–19.6) million in SSA [25]. In a National RHD database from Uganda, 64% of all RHD met criteria for index operation at timepoint of RHD diagnosis [26], which we termed “severe” RHD. Using the total RHD prevalence as a referent and weighing our calculations by the proportion of people who have severe versus mild RHD, our extrapolation suggests that 9.2 (95% UI: 6.6–11.7) million people may be affected by at least “at risk” PH‐RHD and 7.2 (95% UI: 4.7–9.7) by “likely” PH‐RHD in SSA in 2021 (Table 2 and Figure 4).

Figure 4.

Prevalence estimates for PH associated with RHD. *Selection bias and other limitations associated with extrapolating prevalence estimates from targeted cohorts to the broader population may substantially increase the uncertainty surrounding these estimates.

3.4. Prevalence Estimates for PH Associated With All‐Cause Heart Failure

Other subgroups of heart disease that were identified included HHF (n = 7 reports, n = 1173 patients), isolated right heart failure (RHF; n = 2, n = 818), congenital heart disease (CHD; n = 4, n = 451), and a group of patients suffering from heart failure with reduced ejection fraction (HFrEF) syndromes of mixed etiology (ischemic, myocardial fibrosis, dilated, and peripartum cardiomyopathy; n = 12, n = 2368). Cohort characteristics are given in Table 1. PH prevalence estimates for “at risk” and “likely” PH are provided in Table 3. Based on the 2019 Global Burden of Disease Study, the prevalence of all‐cause HF was 7.6 (95% UI: 6.2–9.3) million people in SSA in 2019 [27]. The distribution of underlying causes of heart failure varies significantly by age group. Based on published reports, at the median of the SSA population (19 years [13]), ~1% were attributed to RHF, ~3% to HHF, ~19% to CHD, ~24% to HF associated with RHD (HF‐RHD), and ~53% to other HFrEF syndromes [28]. At a median age of 55 years old, the distribution changes dramatically to ~33% due to HHF, ~14% due to RHF, ~9% due to HF‐RHD, and ~44% to other HFrEF syndromes [28, 29]. Weighing our calculation on these distributions, data extrapolation suggested that 5.7 (95% UI: 4.2–7.6) million people may have “at risk” PH‐HF and 4.5 (95% UI: 2.6–7.0) million people may have “likely” PH‐HF in SSA in 2019 (Figure 5). Of note, decreased LVEF was associated with increased PH prevalence HFrEF syndromes (p = 0.024) and HHF (p = 0.059; Figure 6A,B and Table S7).

Table 3.

PH prevalence estimates based on TRV > 2.8 m/s and TRV > 3.4 m/s based on the systematic literature search and meta‐analysis for heart failure subgroups.

Cohort	Estimated PH prevalence, % (95% CI)	Estimated PH prevalence, % (95% CI)
	TRV > 2.8 m/s	TRV > 3.4 m/s
Hypertensive heart failure	49.6 (28.1–71.2)	26.3 (9.8–47.1)
Congenital heart disease	72.9 (50.8–90.3)	62.3 (19.5–95.8)
Right heart failure	75.7 (58.7–89.4)	66.3 (62.7–69.8)
HFrEF other	82.8 (79.9–85.5)	58.8 (38.8–77.4)

Abbreviations: HFrEF, heart failure with reduced ejection fraction; TRV, tricuspid regurgitation velocity.

Figure 5.

Prevalence estimates for PH associated with all‐cause heart disease. *Selection bias and other limitations associated with extrapolating prevalence estimates from targeted cohorts to the broader population may substantially increase the uncertainty surrounding these estimates.

Figure 6.

The association between estimated PH prevalence and left ventricular ejection fraction in (A) hypertensive heart disease and (B) heart failure with reduced ejection fraction syndromes. PH, pulmonary hypertension.

3.5. Prevalence Estimates for PH Associated With COPD, Pulmonary Tuberculosis, Schistosomiasis, and Chronic Kidney Disease

Besides HIV, SCD, RHD, and heart failure, other infectious and NCDs play a crucial role when attempting to estimate the total prevalence of PH in SSA. These include COPD, TB, schistosomiasis, and CKD. Due to paucity of published data of these diseases in the SSA region (n = 0 reports, n = 1, n = 0 and n = 2, respectively; Table S6), we extracted data based on previously published data from other regions in the world. We preferred meta‐analytic reports if available, such as is the case for COPD [30], TB [31], and CKD [32]. For schistosomiasis, no meta‐analytic reports were available, and we performed our own meta‐analysis from published reports. We used a TRV threshold > 2.8 m/s for “at risk,” used RHC using a mPAP > 20 mmHg (the available gold standard) to confirm PH, if available and a TRV > 3.4 m/s for “likely” PH alternatively. Detailed description and analyses are available in Online Supplement S1. The overall PH prevalence estimates for “at risk” and “likely” PH for each group as well as the total extrapolated PH cases for SSA are given in Table 4. We did not include PH associated with chronic liver disease, due to the low prevalence of PH in liver disease, even in referral populations with clinically significant portal hypertension [33].

Table 4.

PH prevalence estimates in chronic obstructive pulmonary disease, pulmonary tuberculosis, schistosomiasis, and chronic kidney disease in SSA. Data is extrapolated based on previously published, meta‐analytic reports.

Cohort	PH probability	Estimated PH prevalence (95% CI)	Prevalence in SSA, million (95% UI)	PH‐associated prevalence (95% UI)
COPD	“at risk”	39.2% (34.0–44.4)	10.3 (9.7–10.9) [34]	4,038,000 (3,493,000–4,667,000)
	“likely”	15.4% (11.6–19.7)		1,586,000 (1,212,000–2,077,000)
TB	“at risk”	9.4% (6.3–13.0)	2.6 (2.3–2.9)a	239,700 (164,500–349,300)
	“likely”	5.9% (3.9–9.2)		151,700 (95,500–241,100)
CKD	“at risk”	33.0% (28.0–42.0)	9.5 (7.7–11.7) [35]	3,128,400 (2,322,420–4,214,080)
	“likely”	22.2% (17.3–28.3)		2,100,770 (1,518,490–2,906,320)
Schistosomiasis,			112.6 (82.1–148.8) [38]
Hepatosplenic form [36, 37]	“at risk”	18.5% (10.8–29.5)	2.9 (1.6–5.2)	537,400 (241,500–1,196,000)
	“likely”	7.7% (3.3–16.7)		203,300 (65,000–636,100)

Note: Data are extrapolated based on previously published, meta‐analytic reports. We used a TRV threshold > 2.8 m/s for “at risk,” used RHC using a mPAP > 20 mmHg (the available gold standard) to confirm PH, if available and a TRV > 3.4 m/s for “likely” PH alternatively. PH prevalence estimates for COPD based on Zhang et al. [30], TB based on van Heerden et al. [31], CKD based on Bolignano et al. [32], and schistosomiasis based on Conceição et al. [36, 37].

Abbreviations: CI, confidence interval; UI, uncertainty interval.

^aAnnual incident infections based on the WHO Tuberculosis Report 2023.

3.6. Total PH Prevalence in SSA

Akin to prospective deep phenotyping studies on PH that identify overlapping phenotypes between patients [39], it is difficult to determine the totality of PH cases in SSA as the same individual with PH may harbor multiple risk factors (e.g., HIV and SCD). Nevertheless, using our methods that included heavy extrapolation, we determined that of all PH cases estimated in all‐cause heart failure, 24% were attributed to RHD. When accounting for the overlap of counted PH cases between RHD and all‐cause heart failure (excluding other potential phenotypic overlap combinations), we estimate that there might be up to ~22.7 million “at risk” and ~14.3 million “likely” PH cases in SSA in 2021, which may result in an overall estimate of ~1.9% “at risk” for PH and ~1.2% of the SSA population with “likely” PH, respectively (Figure 7).

Figure 7.

Distribution of PH cases in SSA for (A) “at risk” PH and (B) “likely” PH, grouped based on WHO groups. Groups are not based on hemodynamic criteria but solely based on the underlying risk population. Sickle cell disease (dark purple); HIV, human immunodeficiency virus (light green); rheumatic heart disease (light red); heart failure (dark red); COPD, chronic obstructive pulmonary disease (light blue); tuberculosis (dark blue); schistosomiasis (dark green); chronic kidney disease (light purple); PH, pulmonary hypertension. *Selection bias and other limitations associated with extrapolating prevalence estimates from targeted cohorts to the broader population may substantially increase the uncertainty surrounding these estimates.

4. Discussion

We provide a contemporary epidemiologic report of PH from a very large global region that is underrepresented in the published literature. Based on the totality of our calculations that included extrapolation, our results suggest that PH might likely affect ~1.2% of the SSA population, and although PH prevalence varied substantially across different at‐risk subgroups and was dependent on diagnostic criteria, our overall observation places newfound emphasis on the public health importance of PH in SSA. We also present contemporary PH prevalence estimates for widely understudied and underpublished clinical populations endemic to LMICs, including HIV, HHF, and RHD. Using a two‐tier PH diagnostic system that focuses on echocardiography likely provided a reasonable and accurate approach to prevalence reporting in LMICs, which categorically lacked RHC data. Taken together, this study identifies PH as a highly prevalent disease and major public health problem in SSA.

Elevated PAP is independently associated with increased morbidity and early mortality across the medical spectrum, including chronic lung [40] and kidney disease [41], but particularly evident in diseases endemic to SSA like HIV [7] and TB [8], as well as in cardiovascular diseases [42]. The later was, in fact, the greatest contributor to NCD deaths in this region in 2021 [2]. Importantly, treating diseases associated with the development of PH often fails to mitigate PH and subsequent risk for adverse clinical outcomes [13, 43]. Alternatively, accumulating data identify PH as a clinically actionable finding to guide patient care. For example, elevated PAP is a surgical indication in patients with rheumatic mitral stenosis [44], and a longitudinal decrease in PAP from pre‐ to post‐procedure for RHD was independently associated with improved survival [45]. For these collective reasons, PH is regarded as a specific condition of global relevance and, thus, it became important to understand the prevalence of PH in SSA per se.

The majority of published reports on PH originate from high‐income countries, using RHC to diagnose and risk stratify PH. More than 75% of the world's population resides, however, in LMICs, with widely divergent geopolitical, economic, healthcare, and demographic profiles. One major aspect of PH care in LMICs is the virtual absence of RHC, which was evident in our data, in which 98% of reports used echocardiography to assess PAP. In addition to lack of availability, RHC is an invasive test and thereby subject to strong selection bias. Population studies that focus on PH epidemiology in LMICs should therefore prefer echocardiography, which is the noninvasive modality of choice to establish high clinical suspicion of PH as well as allow initial risk stratification [46]. To overcome limitations in the specificity for echocardiography in diagnosing PH [47], we implemented a methodology that leverages two PAP thresholds, based on the current international consensus statement [15], to categorize PH likelihood. This approach allowed us to report two evidence‐based PH estimates. The inclusive criteria suggest that PH may affect up to ~22.7 million people in SSA and that > 50% of patients with certain high‐risk subtypes such as heart failure and RHD could in fact harbor the diagnosis of PH. Even after choosing stricter criteria, likely PH remained a frequent finding, potentially affecting ~14.3 million people. These results reinforce the need to consider PH as a unique clinical problem in SSA based on data using echocardiography.

A previous review highlighted the unique epidemiological profile of PH in SSA [48]; and prior meta‐analytic data on PH prevalence in SSA are limited mainly to two published studies. The first included HIV patients only [49] and the second included a smaller number of patients but represented the first systematic report across all PH groups [50]. To overcome the challenge of limited available studies, we searched for all echocardiographic and RHC studies published in the SSA region, thus not limiting the search to PH‐centric reports only. Using this comprehensive approach, we identified a total of 105 reports. To further increase the accuracy and generalizability of our reports, we stratified PH prevalence based on referral patterns for PAP assessment. This is important, considering the observation that PH prevalence was 8‐fold higher in HIV patients presenting to a cardiology clinic due to cardiopulmonary symptoms as compared to screening of asymptomatic volunteers in HIV outpatient clinics (17% vs. 2%, respectively). Based on this result, population screening of PH in HIV may be more feasible in symptomatic patients only, as compared to the totality of people living with HIV. By contrast, we did not observe a PH prevalence difference based on referral in other subgroups. For example, the PH rate in SCD was similar for referral and nonreferral populations (13% vs. 14%, respectively), suggesting that indications for echocardiography in this diseases used commonly to evaluate alternative processes (e.g., chest pain crises) may have also led a diagnosis of PH. Alternatively, the lack in statistical difference may be due to high within‐group heterogeneity or relatively small study sample size for referred patients (n = 3) as compared to nonreferred studies (n = 19). Taken together, these observations might allow for clinical care patterns in SSA for PH detection that vary by predisposing disease to optimize efficiency and resource utilization.

Of all investigated disease cohorts, PH prevalence estimates were elevated particularly across HF subpopulations, ranging from 50% to 68% for TRV > 2.8 m/s and 26% to 66% for TRV > 3.4 m/s. Specifically in RHD, PH prevalence was 3‐fold greater (78% vs. 26%) in patients with a surgical indication for RHD as compared to individuals in whom RHD was considered to be mild. This observation is clinically relevant, since PH is independently associated with worse outcome in RHD [10, 11], but surgical correction and consequent decrease in PAP have shown to improve outcomes [45], and referral for correction prior to the development of severe PH may improve outcomes [11]. These data emphasize that severity of the underlying cardiac disease influences PH prevalence, which is supported by our finding that PH prevalence correlated inversely with LVEF in heart failure subpopulations, although the observed effect sizes were low, as well as previous findings [51]. Our report further advances the PH field by providing the first ever prevalence estimates for PH in HHF, which is particularly important considering the fact that ~1 in 10 people suffer from hypertension in SSA [52] and HHF is a common disease in SSA [53].

PH is a heterogenous disease, marked by substantial pathophenotypic overlap, which makes any estimation of the total number of PH cases in SSA difficult. However, based on our calculations, we found that ~1.2% of the SSA population may have ‘likely’ PH. This aligns with previous assumptions for the global prevalence of PH of ~1% [6, 54]. We found that an even bigger universe of ~1.9% may have “at risk” PH. Although this TRV threshold is often considered only moderately elevated, significant changes of right heart morphology and function as well as independent associations with increased mortality risk have been observed in a large US‐based cohort [55], emphasizing the potential clinical relevance of this epidemiological finding.

5. Limitations

Not all countries were represented equally in the literature search, which is mainly due to a paucity of published data, and it is unclear whether our assumptions apply to all SSA countries equally. In addition, there were insufficient data on the PH prevalence in schistosomiasis, TB, CKD, and COPD that originated from SSA, which necessitated using these data from other global areas. In this approach, we made an assumption that the prevalence of PH in patients with specific PH‐predisposing diseases is similar across geographic regions, as has been reported previously. For example, a similar PH prevalence (assess by echocardiography) is observed in asymptomatic individuals living with HIV in this study (2.3%), Spain (2.5%) [56], and Brazil (2.1%) [57]. Importantly, our approach was not used to determine the prevalence of PH in SSA per se; those data were calculated using estimates of population density and rates of PH‐predisposing diseases in SSA. Here, we aimed to reduce bias by using either comprehensive meta‐analytic reports (CKD and COPD) or data from other LMIC regions (TB and schistosomiasis). Our extrapolation data relies on reports from the WHO and Global Burden of Disease Study, although these are considered reliable sources by the medical community. In addition, we cannot exclude that there may have been overlap of individuals between studies over time (e.g., HIV and concomitant heart failure) or across centers. The share of PH attributed to TB may be underestimated, which is derived by active TB cases reported by the WHO although some patients post‐TB may have PH, especially if they developed significant TB‐related lung fibrosis. Furthermore, the population of SSA is growing rapidly, so prevalence estimates will change (most likely increase) over time. We accounted for frequent diseases that are associated with PH and therefore rare cases including idiopathic PAH and chronic thromboembolic PH (accounting for WHO Group 4) were not included. Due to the selection bias of the published literature which do not necessarily represent the true population of SSA, our results may be subject to overestimation of PH prevalence, which may be amplified by the extrapolation methods of this study. Nevertheless, we believe the novelty of these results, their potential implications for clinical research and health initiatives as well as the large sample size, and novel, more detailed method to report prevalence estimates in PH outweigh these limitations.

6. Conclusion

Our data suggest that PH may be present in up to ~1.2% of the SSA population, and that up to ~1.9% of the SSA population may be at risk for PH, after accounting for referral bias and disease severity. These data establish an important epidemiological framework profiling PH in a large LMIC region for which prevalences of NCDs are generally under‐reported. In doing so, it adds an important step toward unraveling the global epidemiology of PH. Importantly, our report also highlights the great gap of invasive and noninvasive data available in this global region and the critical need for primary epidemiology reports in SSA, particularly for under‐reported diseases including COPD, TB, and schistosomiasis. Our findings demonstrate the urgent need to build population health initiatives in SSA focusing on PH prevention, detection, and management that can mitigate the increased clinical risks associated with PH in affected patients.

Author Contributions

Formulating research question: Katarina Zeder and Bradley A. Maron. Development of design: Katarina Zeder and Bradley A. Maron. Database search, study screening, and data extraction: Katarina Zeder, Veranyuy Ngah, Armella Santi, and Eric W. Robbins. Data analysis: Lanjing Wang and Guoqing Diao. Manuscript writing: Katarina Zeder and Bradley A. Maron. Editing of the manuscript: Lanjing Wang, Veranyuy Ngah, Armella Santi, Eric W. Robbins, Ana O. Mocumbi, Judith Namuyonga, Peter Nyasulu, Simon Stewart, Geoff Strange, Gabor Kovacs, Marc Humbert and Guoqing Diao. All authors approved the final version of the manuscript.

Disclosure

Katarina Zeder and Bradley A. Maron are guarantors.

Ethics Statement

No ethical approval was required as all data used were published before.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File

Data Availability Statement

Data sharing is not applicable to this article as no data sets were generated or analyzed during the current study.