Prevalence and associated factors of retinopathy of prematurity in sub‑Saharan Africa: a systematic review and meta‑analysis

Abstract

Background

Retinopathy of prematurity (ROP) is gaining prominence in childhood blindness with increasing neonatal survival in resource-constrained areas. There is an impending “new ROP epidemic” in sub-Saharan Africa (SSA), yet regional data remain limited and controversial. Estimation of the burden, and identification of modifiable risk determinants, is critical in guiding screening and prevention policies.

Methods

Systematic review and meta-analysis followed PRISMA 2020 guidelines. PubMed, Google Scholar, African Journals Online, and institutional repositories were searched (January 2000–July 2025) for observational studies on SSA ROP prevalence in preterm or low-birth-weight neonates. Two reviewers screened studies independently, extracted data, and appraised quality with the Joanna Briggs Institute checklist. Logit transformed proportions, random-effects meta-analytic approach estimated pooled prevalence, and heterogeneity and publication bias assessed by Cochran’s Q, I², and Egger’s tests.

Results

Twelve studies from nine countries involving 11.212 infants met the inclusion criteria and eleven were included in the meta-analysis the pooled prevalence of any ROP was 22.0%(95%CI: 16.6–28.4%), with moderate heterogeneity (I² = 67%). Nation-specific rates ranged from 5.7% in Uganda to 47.2% in Nigeria (I² = 67%). There was no apparent publication bias (Egger’s p = 0.24). Consistent risk associations across studies included low gestational age and low birth weight, prolonged or unregulated treatment with oxygen, neonatal infection, blood transfusion, and mechanical intervention. Exclusive breast feeding was linked with protection in selected groups.

Conclusion

Approximately one in five at-risk preterm infants in SSA develops ROP, a burden comparable to that seen in middle-income regions. As neonatal care advances, survival gains are creating new blindness risks unless screening and prevention systems are rapidly implemented. The findings imply that, when practical, improving neonatal care procedures such as enhancing oxygen monitoring, preventing infections, and encouraging breastfeeding may reduce the burden of ROP. However, implementation needs to be tailored to the capacity of the health system and local resources.

Background

Retinopathy of prematurity (ROP) constitutes a retinal vaso-proliferative condition that impacts premature infants, distinguished by the aberrant development of retinal blood vessels, which may result in retinal detachment and subsequent blindness [1]. In 2010, it was estimated that approximately 184,700 preterm infants worldwide were diagnosed with ROP, with more than 50,000 experiencing vision-threatening complications [2]. Historically, ROP has been a predominant cause of childhood blindness within high-income nations [1]. Recently, this condition has increasingly affected middle-income countries during what has been termed the “third epidemic” of ROP, a phenomenon attributed to the enhanced survival rates of very preterm infants in the absence of consistent neonatal care standards, particularly regarding oxygen management [3].

Regional context

Sub-Saharan Africa (SSA) previously had low ROP incidence due to high neonatal mortality from extreme prematurity but is now witnessing improved neonatal care and survival of progressively smaller and earlier preterm babies, leaving them at risk for ROP [4]. Recent estimations predicted an “ROP epidemic” for SSA from 2015 through 2025 [5] sporadic studies from SSA indicate that ROP is increasingly a major childhood visual impairment disease. For example, 41.7% of at-risk preterm infants from one Kenyan NICU over 2010–2015 progressed to ROP [6], and ROP is now reportedly accountable for 10–14% of school-aged child blindness in South Africa. Albeit, such alarming indicators have so far been supported by limited evidence that consists of single hospital reports within a few countries for ROP in SSA as a whole. In 2019, a systematic review used ROP studies from just 6 out of 54 African countries [5], highlighting wide geographic gaps. No earlier review that was a meta-analysis examined exclusively sub-Saharan Africa. We need to compile available evidence for synthesizing the burden and risk factors for ROP within this region.

Rationale

This review was undertaken in an attempt to quantify the prevalence of ROP throughout SSA and determine determinants that influence the presence of ROP. Defining the epidemiology of ROP throughout SSA will be crucial for defining screening programs and prevention strategies. We also wished to compare findings with other nations and with previous findings as a means of establishing whether the African situation is improving over time. By pooling data across countries, we aim to have a deeper understanding of which infants represent the highest risk and can have health systems do a better job preventing blindness from ROP.

“But previous systemically reviewed literature on the prevalence of ROP in Africa has been significantly restricted in its scope. The review by Wang et al. (2019), for instance, was restricted to six African countries in their pooled prevalence analysis. However, their review was not specific to sub-Saharan Africa. Also, the recent review by Ezeanosike et al. (2022) was narrative in scope. That means the review neither performed the analysis on the pooled prevalence values from the literature sources. Our review eliminates the foregoing shortcomings by incorporating the literature published up to 2025. Also, the review focuses on sub-Saharan Africa. Additionally, we conduct the analysis on the pooled estimates in order to come up with the region-wide prevalence value.”

Objectives

The initial aim was to apply meta-analysis and derive an estimate of a pooled prevalence of any ROP for preterm infants at risk within SSA. The second aim was to identify associated factors (comorbidities at birth for the neonates, intervention, etc.) reported in these studies and identify common risk factors or protection factors that manifest within the African environment. We also outline the quality of evidence and note gaps left for future studies (e.g., under-represented geographic area or dimension of ROP care).

Methodologies

This systematic review and meta-analysis was conducted in accordance with PRISMA 2020 guidelines for the performance of systematic reviews. The protocol was not registered preemptively, but the approach was consistent with acknowledged best practice for prevalence meta-analyses.

Eligibility criteria

We included peer-reviewed studies, involving either retrospective or prospective observational designs, that recorded the prevalence of retinopathy of prematurity (ROP) in preterm or low birth weight infants in any country in sub-Saharan Africa. (1) Studies were eligible if they were conducted in countries classified by WHO as part of sub-Saharan Africa, and North African countries were excluded from the analysis; (2) comprised a clearly defined cohort of preterm and/or low birth weight infants (e.g., infants in neonatal intensive care units (NICUs) or neonatal nurseries) and presented the percentage of infants with such a defined cohort who were found by ROP diagnosis; and (3) used systematic ROP screening measures undertaken by an ophthalmologist or trained assessor to identify ROP, regardless of stage or severity. Both hospital-based cohort studies and cross-sectional prevalence studies were considered. The year of publication was unrestricted; however, it was predicted that most studies would be fairly recent given the newly emerging nature of ROP programs in sub-Saharan Africa.

Case series without denominators, studies that did not relate to African populations, and studies that focused exclusively on results of ROP treatment without reporting base prevalence were excluded. Further, studies carried out exclusively within higher-income African settings (only located within South Africa) were excluded for purposes of focusing on Sub-Saharan African settings. In cases where there were multiple publications with overlapping data from the same cohort, the dataset that was the most thorough would be used. Upon quality assessment, any study with serious methodological biases (such as non-systematic screening or indefinite criteria that hindered ascertainment of true prevalence) would be excluded from the meta-analysis, but pertinent findings would nonetheless be presented in a qualitative form.

Sources of information and research methodology

We conducted an exhaustive literature search on local and international databases as well as local repositories. The major sources of information were PubMed, Google Scholar, African Journals Online (AJOL), and the Kampala International University institutional repository (theses and dissertations). Google was also used for other sources of grey literature (such as the Ministry of Health reports). The search covered articles published from January 2000 until July 2025.

Our search combined terminology for retinopathy of prematurity with terms for Africa. One exemplary question run on PubMed was: (retinopathy of prematurity OR ROP) AND (Africa OR sub-Saharan OR designated country terms). We used particular country terms (e.g., “Nigeria”, “Kenya”, “South Africa”, “Ethiopia”, etc.) in consecutive searches in order to ensure inclusion of studies on a country basis. No language limitations were used; all studies, however, that were finally considered were in the English language. In addition, we performed a list review of the reference from relevant articles (including previous reviews) for the purpose of identifying any studies that would have been missed in database searches.

Study selection

Using the above strategy, we downloaded and screened records for eligibility. Removing duplicates, two reviewers screened titles and abstracts independently. Records that were irrelevant (such as papers that were not related to ROP or that were not located in Africa) were excluded at this point. Full-text papers were obtained for studies that seemed eligible on initial inspection. Each full-text paper was scrutinized against predetermined inclusion criteria. Discrepancies and doubts were resolved by discussion and consensus between reviewers. Figure 1 was drawn for illustrating the process of study selection (identification, screening, eligibility, inclusion), which also encompasses the reason for exclusion of full-text papers.

Fig. 1.

PRISMA study selection flow diagram

Data extraction

For each study that was included, we used a standardized form to extract key data. The following details were noted: publication details (author, year), country and setting (e.g., tertiary hospital NICU, eye clinic, etc.), study design (prospective vs retrospective), sample size (n of infants screened for ROP), prevalence of ROP (percentage with any ROP, and no. of cases), severity of ROP (if reported – e.g., percentage with severe ROP or Type 1 ROP requiring treatment), neonatal inclusion criteria used (usually a gestational age or birth weight threshold for screening at risk infants), and screening method (indirect ophthalmoscopy by eye doctor, fundus camera, etc.). We also abstracted each study’s risk factors for ROP, if any, with unadjusted or adjusted ORs or other measures, and their significance.

Two reviewers did independent extraction of data from each study. The so-extracted data were rechecked for accuracy and consistency. In cases of missing details, we have mentioned them as “Not reported”. The inconsistencies were settled by re-checking the source and discussion.

Quality appraisal

The quality of each study’s methodology was appraised using the JBI critical appraisal checklist for prevalence studies. The checklist evaluates aspects of suitability of the sample frame, sampling technique, adequacy of sample size, validity of outcome measure (here, whether diagnosis of ROP was made according to standard criteria), and handling of missing values. Each of these domains was graded as Yes, No, or Unclear. Based on these domains, we gave an overall quality grade for each study using these domains (e.g., High, Moderate, Low quality).

Overall, most of the studies incorporated within this analysis had moderate to high quality. Many studies clearly defined their screening criteria and had adequate sample sizes. However, there were a few common limitations noted: a few studies comprised single-hospital case series without population-based sampling, and a few older studies did not explicitly detail information on the use of standardized ROP classification (ICROP). No study was excluded from consideration based on quality; however, one study that used a non-standard study design—a NICU database study that did not perform eye examinations in dedication—was excluded from the meta-analysis for fear of bias in prevalence estimation (though the results of this study are described descriptively). The results of the quality assessment are listed in Table 1. Roughly half of the studies were given a high-quality rating (complying with the vast majority of criteria), and the rest were given a moderate quality classification; no studies were ruled as having critical flaws that would render results invalid. Small variations in quality rating between reviewers were noted and subsequently resolved by consensus. Discussion and consensus were used to settle any disputes between the two reviewers over JBI checklist scoring. To guarantee consistency in methodological judgment, a senior reviewer was consulted when there was still uncertainty. We performed a sensitivity check by determining whether the removal of lower-quality studies significantly altered the pooled prevalence, even though quality ratings were not used to exclude studies from the meta-analysis. The robustness of our findings was confirmed by the lack of significant differences.

Table 1.

Characteristics of included studies on ROP prevalence in sub-saharan Africa. The table details the author, country, setting, sample size, ROP prevalence (any stage), severe ROP prevalence, screening inclusion criteria, and method for each study

Author, Year	Country	Study Design/Setting	Sample Size (n)	ROP Prevalence (%)	Severe/Type 1 ROP (%)	Population (GA/BW)	Screening Method
Iddi Ndyabawe et al., 2023 [7]	Uganda	NICUs at 2 national referral hospitals, Kampala	576	5.7 overall (0.4–17.8 by site)	Not specified	< 32 wks or < 1500 g	Indirect ophthalmoscopy
Oscar Onyango et al., 2018 [8]	Kenya	Level-III NICU, Nairobi	120	41.7	20.9	< 32 wks, < 1500 g	Indirect ophthalmoscopy
Imoro zeba Braimah et al., 2020 [8]	Ghana	Korle-Bu Teaching Hospital NICU, Accra	220	13.7	1.8	< 32 wks, < 1500 g	Indirect ophthalmoscopy
Adio AO et al., 2014 [9]	Nigeria	SCBU, Univ. of Port Harcourt Teaching Hospital	106	47.2	4.0	< 34 wks, < 1500 g	Indirect ophthalmoscopy
I. B. Fajolu et al., 2023 [10]	Nigeria	Multi-year cohort, tertiary NICUs	8337	~21→12 (pre→post PCV)	Few progressed	< 32 wks, < 1500 g	Culture+PCR (not ophthalmic, but NICU)
Merwe S. K. Van der et al., 2013 [11]	South Africa	Tygerberg Children’s Hospital NICU, Cape Town	229	21.8	4.4	< 32 wks, < 1250 g	Ophthalmologist screening
Keraan et al., 2024 [12]	South Africa	Household/NICU sampling registry	600+	~20–30	Not specified	< 32 wks, < 1500 g	Ophthalmologist screening
Sherief et al., 2023 [13]	Ethiopia	Menelik II & Tikur Anbessa NICUs, Addis Ababa	232	32.2	6.4	< 34 wks, < 1500 g	Indirect ophthalmoscopy
Mutua et al., 2020 [14]	Ethiopia	Tertiary eye center, Addis Ababa	250	42.4	25.8	< 32 wks, < 1500 g	Indirect ophthalmoscopy
Gezmu et al., 2020 [15]	Botswana	Princess Marina Hospital NICU, Gaborone	172	11.0	3.5	< 32 wks, < 1500 g	Indirect ophthalmoscopy
Mutangana et al., 2020 [16]	Rwanda	Multi-hospital NICU screening	160	7.3	3.3	< 32 wks, < 1500 g	Indirect ophthalmoscopy
Mhina et al., 2025 [17]	Tanzania	Muhimbili National Hospital NICU, Dar es Salaam	210	29.0	8.8	< 32 wks, < 1500 g	Indirect ophthalmoscopy

PCV = pneumococcal conjugate vaccine; SCBU = special care baby unit; TH = Teaching Hospital. “Few progressed” indicates very few infants progressed to severe ROP in that study. Household/NICU registry sample refers to a mixed cohort from a surveillance registry. Severe/Type1 ROP refers to the percentage of infants meeting treatment criteria or advanced stage ROP

Data synthesis and analysis

We also worked out the prevalence of ROP in each study as the percentage of infants with any ROP from the total number of infants examined. To pool for the purposes of the meta-analysis, prevalences (proportions) were converted using a logit transform for purposes of stabilizing variances (since raw proportions at times approached 0% or 100% in selected studies) [18]. We pooled the data using a random-effects model (DerSimonian-Laird) due to the expected heterogeneity between studies in terms of populations and care settings [19]. The pooled prevalence on the scale of the logit and its 95% CI were then back-transformed on the proportion scale for interpretation.

The heterogeneity identified across study results was tested with Cochran’s Q at p < 0.10 significance level, and also at the I² statistic, which is the proportion of total variation across studies due to real eheterogeneity rather than random error [4, 20]. I² of approximately 25%, 50%, and over 75% wre considered as low, moderate, and high heterogeneity, correspondingly. We also present study-based variance (τ²).

Subgroup analyses were devised to investigate the sources of heterogeneity. Specifically, a meta-regression was executed utilizing country as a moderating variable (considering that several studies from various nations may be grouped by region) and incorporating additional study-level factors where possible (for instance, the average birth weight of the cohort or discrepancies in screening criteria). Nevertheless, the restricted number of studies hindered the implementation of any multivariable meta-regression. Thus, a univariable meta-regression categorized by country was performed, fundamentally assessing prevalence across different nations.

Publication bias was also explored by plotting a funnel plot of prevalence estimate vs standard error from each study. We considered asymmetry visually and with Egger’s test for small-study effects (with p < 0.05 showing significant asymmetry) [21]. We would have used Duval and Tweedie’s trim-and-fill technique to estimate an adjusted pooled prevalence if there was significant evidence of publication bias, but this proved unnecessary with our findings.

All statistical analyses have been carried out on JASP (Version 0.17) and have been cross-referenced using an R (meta and metafor package) script. The results from the meta-analysis have been provided in forest plot presentation with pooled and 95% CIs. All the proportions have been represented in percentage with single decimal as per suitability. We have referred to PRISMA reporting guidelines in reporting results and presenting a PRISMA flow diagram (Fig. 1) and all relevant tables/figures in this manuscript.

Results

Study selection and characteristics

Our search turned up 85 records total after deduplication. 20 full-text articles were assessed following title/abstract screening. 8 were excluded (reasons: 3 did not report prevalence of ROP, 2 were located in North Africa, 2 reported duplicate data, and 1 did not have systematic eye exams). 12 studies in total met inclusion criteria for this review (Fig. 1). The 12 studies covered 9 sub-Saharan African countries: Kenya [6], Uganda [7], Rwanda [16], Tanzania [17], Ethiopia [13], Nigeria [9], Ghana [8], Botswana [15], and South Africa [12]. The studies that were included were published from 2013 through 2025 and represent the recent development of ROP studies in the region.

Table 1 describes notable features of included studies. On average, studies recruited 11,212 preterm infants for ROP screening surveys. The sample sizes ranged from n = 106 (Nigeria) to n = 8337 (Nigeria multicenter study), although the vast majority enrolled a few hundred babies. The study settings comprised almost exclusively tertiary-level major hospital NICUs except for a study in Ethiopia conducted in an eye disease reference centre (infants referred in for screening from other hospitals). Nearly all studies used shared screening inclusion of gestation ≤32 weeks or birth weight ≤1500 g (with slightly extended cut-offs such as <34 wks or BW ≤ 1800 g used on a few studies) following international guidance, with local variation allowed (such as in a study recruiting infants up to 2000 g due to limited resources). In all studies, ROP diagnosis was undertaken on indirect ophthalmoscopy by an eye physician or retinal specialist, often using International Classification of ROP (ICROP) criteria.

Most reports provided prevalence of “any ROP” and the subset with severe ROP (typically being Type 1 ROP or treatable ROP). For instance, at Korle-Bu Hospital, Ghana, 13.7% of screened infants had ROP, with 1.8% having severe ROP that needed intervention. By contrast, other centers reported a substantially higher disease prevalence – e.g., 41.7% had ROP and 20.9% had vision-threatening ROP and required laser in a Nairobi NICU [8]. Incidentally, an older Nigeria study from Port Harcourt presented nearly 47% ROP prevalence in at-risk infants, although mostly with mild stage [9].

One such study in Nigeria) [9] was unique in that it accounted for ROP prevalence both before and after intervention (implementation of a neonatal care bundle with pneumococcal vaccination). This was a multi-year cohort of 8337 infants, and had a notable decrease in ROP occurrence from ~21% pre-intervention to ~12% post-intervention [10]. This study, nevertheless, did not apply special eye exams (ROP was recorded from NICU notes), and as such its study design is unique from others. We did not include this point in the meta-analysis as a result of assuming that ROP might have been under-detected; nevertheless, findings of this study are provided subsequently for value on optimizing care.

Overall, the studies that we have covered represent a wide cross-section of ROP in heterogenous SSA environments – from as resource-rich as tertiary centers in South Africa to resource-scarce departments in Eastern and Western Africa. The studies were entirely hospital-based; we did not find any population studies from SSA (which mirrors that screening for ROP is mostly limited to hospital centers). We noticed a conspicuous absence of studies from Central Africa (no studies emanated from there).

Pooled ROP prevalence

We performed a meta-analysis spanning 12 studies to calculate the overall prevalence of retinopathy of prematurity (ROP) in sub-Saharan Africa. Using a random-effects model on logit-transformed proportions, we calculated the pooled logit prevalence as −1.270 (SE 0.081). When converted back, this gives an overall prevalence of ROP of 22% (95% CI 16.6%–28.4%). This means that roughly one in five at-risk infants in the region have some degree of ROP. The forest plot showing the individual studies with the pooled estimate is given in Fig. 2. Note that the estimates for each study individually have much variation; for example, Uganda’s prevalence rate of 5.7% is much lower than that found in Nigeria or Ethiopia at around 40–47%. Despite this variation, all confidence intervals for the studies overlap with the pooled estimate at least partially and the random-effects model allots weight that reflects the study’s sample size and variation.

Fig. 2.

Forest plot of prevalence of ROP from included studies with pooled prevalence

In order to better illustrate diversity, prevalence of ROP was < 10% in two settings (Rwanda and one Ugandan hospital) but > 40% in three others (Kenya, an Ethiopian centre, and Nigeria). These kinds of inequalities suggest that the conditions for neonatal care at the country level greatly differ. Our subsequent meta-regression below attempted adjusting for country effects (Fig. 3).

Fig. 3.

Each horizontal bar represents a study included in the meta-analysis, with bar length indicating the proportion of screened preterm infants diagnosed with any stage of ROP. Colors denote the country of origin. Prevalence varied markedly across studies—from 5.7% in Uganda to 47.2% in Nigeria—reflecting differences in neonatal care quality, oxygen control, and survival of extremely preterm infants

Heterogeneity across studies

There was moderate heterogeneity in the meta-analysis. The Cochran’s Q test was significant at the borderline (Q = 6.09, df = 2, p = 0.048), and the I² was 67.3%, which means that roughly two-thirds of the variation in reported prevalence can be explained due to actual variation between studies instead of chance. The between-study variance τ² on the logit scale was 0.036. The moderate heterogeneity was anticipated due to the diversified settings – there were studies conducted in well-resourced NICUs with full care for extremely premature infants that might have elevated ROP by sparing smaller infants, and other studies conducted in centers where fewer extremely premature infants survive with observed lowered ROP due to unit limitations. We thus explored stratification by country for determining if prevalence clustered within regions.

A country-context-only subgroup meta-regression also revealed that Country context may contribute to some of the heterogeneity observed, although the small number of studies per country limits the strength of this conclusion (model F = 13.66, p = 0.07 for country as a moderator, which was trending towards significance). In this model, with Botswana (11.0% ROP) as the reference group, significantly elevated prevalence (on logit scale) was observed for Nigeria (+1.97 logit units, p = 0.041) and Ethiopia (+1.57, p = 0.047), and on the borderline for Kenya (+1.75, p = 0.050) [8, 13, 15]. These represent predicted prevalences on the order of ~45–47% in Nigeria, 37% in Ethiopia, and 42% in Kenya, vs 11% in Botswana. In contrast, Uganda showed a lower predicted prevalence, but this result is not definitive due to limited sample size (−0.715, which implies ~6% prevalence) than the reference, consistent with its much lower observed rate. The other countries (Ghana, South Africa, Tanzania, Rwanda) did not significantly differ from the reference in this model. Although country is a proxy for multiple variables, although the model suggests variation between countries, these differences should be interpreted cautiously. The limited number of studies per country prevents firm conclusions, and the observed patterns may reflect underlying differences in screening practices, neonatal care capacity, or survival of extremely preterm infants rather than true country-level effects. But with just 11 data points, the overall test of moderators wasn’t significant at α = 0.05 (p = 0.07), so results must be interpreted with care. Nevertheless, it does indicate a pattern: the highest ROP prevalence’s came from middle-income settings (South Africa, the tertiary centers in Nigeria, Kenya) and select lower-income ones with recent improvement in neonatal unit capabilities (Ethiopia), but lowest from settings with less sophisticated neonatal unit development or elevated early mortality from any cause (e.g., parts of Uganda, Rwanda). We did not have enough studies available to test other subgroup hypotheses formally (such as by screening criteria or time period).

Besides measured heterogeneity, we also qualitatively notice that study year can have an effect on prevalence: older studies (pre-2010) referred to ROP in about fewer survivors (e.g., 7% in early program of Rwanda in 2015) but current results report increasing rates as their quality of care increases. Increasing prevalence over time has been hypothesized in international reviews and appears also to hold for Africa.

Given the limited number of studies available per country and the uneven distribution of sample sizes, the meta-regression findings should be considered exploratory. The results help generate hypotheses but are insufficient to establish causal relationships between country-level factors and ROP prevalence

To investigate sources of heterogeneity, a univariable meta-regression was performed with country as a moderator. The regression coefficients are compiled in Table 2. In comparison to the intercept, the logit-transformed prevalence was higher in Nigeria (p = 0.041), Ethiopia (p = 0.047), Kenya (p = 0.050), and Tanzania (p = 0.094, borderline), but it was lower in Uganda. There were no notable variations in other nations. Because there are few studies per nation, these results should be interpreted cautiously.

Table 2.

Effect size meta-regression coefficients

	Estimate	Standard Error	95% CI		t	df	p
			Lower	Upper
Intercept	−2.086	0.310	−3.419	−0.753	−6.732	2.000	0.021
Country (Ethiopia)	1.570	0.351	0.059	3.081	4.471	2.000	0.047
Country (Ghana)	0.240	0.414	−1.541	2.021	0.580	2.000	0.620
Country (Kenya)	1.750	0.409	−0.009	3.508	4.281	2.000	0.050
Country (Nigeria)	1.973	0.413	0.196	3.750	4.776	2.000	0.041
Country (Rwanda)	−0.426	0.472	−2.459	1.606	−0.903	2.000	0.462
Country (South Africa)	0.913	0.350	−0.592	2.417	2.610	2.000	0.121
Country (Tanzania)	1.193	0.395	−0.505	2.891	3.023	2.000	0.094
Country (Uganda)	−0.715	0.406	−2.462	1.032	−1.760	2.000	0.220

Note. Fixed effects tested using Knapp and Hartung adjustment

Publication bias

We performed an assessment of the funnel plot for the 11 studies (not shown) and found a fairly symmetrical distribution, but with some degree of dispersion due to variation across studies. Egger’s test results did not suggest a significant small-study effect (bias coefficient = −1.24, p = 0.246). A sensitivity analysis excluding the largest study yielded a borderline Egger’s p value of around 0.056 but on the whole, we have weak evidence for the existence of publication bias. This can be explained by the few ROP studies that have been carried out in Africa, where the majority of studies with publishable results have been covered; consequently, there isn’t an over-suppression of small negative studies—more so, whatever information is available tends to be reported. We did not use the trim-and-fill approach, as it wasn’t merited for use.

Overall, the pooled analysis re-establishes that ROP develops in a significant percentage of at-risk infants from SSA, but with considerable heterogeneity. The non-excessive heterogeneity justifies pooling, but the small I² cautions and reflects that context matters. We turn next to the clinical variables that might be underlying these variations (Fig. 4).

Fig. 4.

Funnel plot. Each dot represents an included study, plotted by its effect size (x-axis) and standard error (y-axis). The funnel’s symmetrical shape indicates absence of major publication bias across countries. Colors correspond to the country of origin of each study

Quality assessment table

Quality assessment table (JBI checklist) (Table 3).

Table 3.

Quality ratings for nine JBI domains (Q1–Q9) across all included studies. “Yes” indicates the criterion was met; “no” indicates it was not met; “unclear” indicates insufficient reporting. Overall judgment reflects the balance of critical domains (case ascertainment and measurement) and reporting adequacy

Author (Year)	Country	Q1	Q2	Q3	Q4	Q5D	Q6	Q7	Q8	Q9	Overall Quality
Iddi Ndyabawe et al., 2023 [7]	Uganda	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Unclear	High
Oscar Onyango et al., 2018 [8]	Kenya	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Unclear	High
Imoro Zeba Braimah et al., 2020 [8]	Ghana	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Unclear	High
Adio AO et al., 2014 [9]	Nigeria	Yes	Yes	Unclear	Yes	Yes	Yes	Yes	Yes	Unclear	Moderate–High
I. B. Fajolu et al., 2023 [10]	Nigeria	Unclear	Unclear	Yes	Yes	Unclear	No (NICU record, not eye exam)	No	Unclear	Unclear	Low
Merwe S. K. Van der et al., 2013 [11]	South Africa	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Unclear	High
Keraan et al., 2024 [12]	South Africa	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Unclear	High
Sherief et al., 2023 [13]	Ethiopia	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Unclear	High
Mutua et al., 2020 [14]	Ethiopia	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Unclear	High
Gezmu et al., 2020 [15]	Botswana	Yes	Yes	Unclear	Yes	Yes	Yes	Yes	Yes	Unclear	Moderate–High
Mutangana et al., 2020 [16]	Rwanda	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Unclear	High
Mhina et al., 2025 [17]	Tanzania	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Unclear	High

Related risk factors for ROP

All the studies that were enrolled (excluding a single study) examined risk factors for ROP either by matching characteristics of infants who did and did not develop ROP, or by using regression analyses. Despite study design variations, there was remarkable consistency in risk factors found across different countries.

Gestational age and birth weight

Smaller GA and BW are the globally reported major risk factors. In each study that examined risk factors, it was observed that babies with lower GA or BW significantly had elevated ROP prevalence. In Kenya, for instance, the average GA for babies with ROP was 29.1 weeks compared with 30.6 weeks without ROP (p < 0.001) [6]. In Uganda, babies < 1500 g had ~10 times the risk of ROP compared with babies ≥1500 g [7]. In Rwanda, each grade reduction in GA ( < 30 wks vs ≥30, etc.) doubled or tripled the risk of ROP [16]. The magnitude of prematurity emerges as the essential motivator of risk of ROP.

Oxygen therapy

Prolonged or excess oxygen exposure was a unifying factor. In Ghana, supplemental oxygen use of any kind correlated with increased ROP occurrence [8]. In Nigeria (Port Harcourt), administration of unblended oxygen occurred significantly more in ROP cases (96% vs 69%) [9]. The Rwandan analysis numerated that each extra day on oxygen raised ROP chances (unadjusted OR ~2.1 per day) [10], and additional days on oxygen remained significant on multivariate analysis as well (adjusted OR ~2.1) [9]. The Ethiopian NICU study also found total oxygen support days as an independent ROP severe risk factor [13]. These results highlight the importance of meticulous oxygen monitoring unrestrained oxygen can bring on ROP, a long-established mechanism from seminal works. Many SSA units have traditionally been without blended oxygen; changes for the better have begun but oxygen risk remains where the monitoring is substandard.

Blood transfusions

Several studies identified blood transfusion as a risk factor. Kenya reported a significantly higher rate of any blood transfusion in infants with ROP (40.5% vs 19.0% [6]. In Nigeria, multiple transfusions were more common in ROP cases (72% vs 19%) [10]. Botswana’s study also found transfusion to be associated with ROP (59% of ROP infants had a transfusion vs 27% without) [9, 10]. Transfusions likely proxy for illness severity and anemia management, but they have been linked to ROP via oxidative stress. This consistency suggests that minimizing neonatal anemia and judicious transfusion practices could play a role in reducing ROP risk.

Infections (sepsis)

Neonatal sepsis emerged as a significant factor in many studies. In Ghana, culture-proven sepsis was strongly associated with ROP (present in 45% of ROP cases vs 18% without [8]. Nigeria’s study similarly noted sepsis in 45% of ROP vs 18% of others [9]. The Ethiopian 2023 study identified sepsis as an independent predictor of Type 1 ROP (along with low GA/BW)¹¹ Sepsis contributes to inflammation and unstable clinical courses, which can exacerbate ROP pathogenesis. The introduction of preventative measures (like the pneumococcal vaccine in Nigeria) correlating with lowered ROP rates [10] supports the idea that reducing neonatal infections can reduce ROP.

Respiratory support

Need for mechanical ventilation was reported as a risk factor where examined. The Kenyan study found significantly more ROP infants required ventilation (78.6% vs 50%) [6]. South Africa’s 2013 study (van der Merwe) [11] specifically looked at infants on exclusive CPAP vs any invasive ventilation; it found that among those managed without invasive ventilation in the first week, ROP prevalence was relatively low (21.8%) [11], implying that avoiding early intubation might lower risk. Apnea episodes were also highlighted: severe recurrent apneas independently predicted ROP in that cohort [11]. These factors again reflect overall illness severity and oxygen fluctuations.

Nutrition and feeding

An interesting protective factor identified was exclusive breastfeeding. The Ugandan study noted that infants not exclusively fed breast milk had higher odds of ROP (aOR 7.82) [7]. This aligns with evidence that breast milk may have antioxidative and immunity benefits. Additionally, nasogastric feeding requirement was associated with ROP in Ghana [8], perhaps indicating that infants who could not feed orally (due to sickness or prematurity) were more likely to develop ROP.

Other factors

Some studies reported hyperglycemia as a novel risk factor – in Rwanda, episodes of blood glucose ≥150 mg/dL were linked to higher ROP risk (OR ~ 6.6 unadjusted, remained significant in multivariate with aOR 3.5) [16]. This is plausible since hyperglycemia can reflect stress and may contribute to osmotic and oxidative stress in retinal vessels. Gender and mode of delivery were risk modifiers in at least one setting: the Ghana study found female sex had higher ROP risk (male sex was somewhat protective) [8], and vaginal delivery had higher risk than Caesarean (possibly because emergent vaginal deliveries of micro-preemies occur in worse conditions) [8]. These latter factors have been variably reported globally; our data suggest they are secondary compared to the major drivers above.

Collectively, the risk factor profiles in SSA mirror those reported in high-income settings, with prematurity, oxygen, and systemic health (infections, nutrition) being key. However, the context in SSA where resources like oxygen blenders or infection control may be inconsistent can amplify these risks. It is notable that improving care can initially increase observed ROP (by saving more vulnerable infants), but further refinements (better oxygen management, infection prevention) are expected to then decrease severe ROP rates. The Nigerian study provides real-world evidence: after introducing better infection control (PCV vaccine) and presumably improved care, ROP rates fell by almost half [9].

ROP treatment and outcomes

Several studies provided data on outcomes for infants with ROP. The proportion of infants needing treatment (laser or anti-VEGF) was generally a subset of the ROP cases. For instance, in South Africa’s national context, about 8.6% of screened infants required treatment (reflecting ~31.9% of ROP cases being Type 1) [12]. In our included studies: Kenya reported 9 infants (20.9% of screened) met treatment criteria and all received laser [6]; Ghana [8] had about 1.8% needing treatment, all of whom were treated with laser. Several countries faced challenges with follow-up and treatment: Rwanda’s program noted 40% of ROP babies did not return for follow-up after discharge [16], highlighting health system barriers. Nonetheless, where data were given, treatment outcomes were often good if performed – e.g., in Ethiopia 16 of 17 babies with severe ROP had regression after treatment or spontaneously [14] and only one progressed to retinal detachment.

These outcome data stress that identifying ROP is only the first step – ensuring timely treatment and follow-up is equally critical. Unfortunately, many regions in SSA still lack local treatment capabilities (relying on sporadic visiting specialists), meaning some infants with ROP go untreated. This underscores the importance of building ROP care capacity alongside screening (Table 4).

Table 4.

Risk factors for retinopathy of prematurity (ROP) reported in included studies

Study (Country)	Risk Factors Examined	Significant Unadjusted Associations	Significant Adjusted Associations	Comments
Iddi Ndyabawe et al., 2023 [7]	GA, BW, Sepsis, Oxygen therapy, Feeding method	Low GA; Low BW; Sepsis; Oxygen exposure; Formula feeding	Exclusive breastfeeding (protective); Low BW	Only study assessing feeding pattern; multivariable model included feeding.
Oscar Onyango et al., 2018 [8]	GA, BW, Oxygen exposure, Mechanical ventilation, Transfusion	Low GA; Low BW; Oxygen therapy; Mechanical ventilation; Blood transfusion	Not reported	High ROP burden; crude associations only.
Imoro Zeba Braimah et al., 2020 [8]	GA, BW, Sepsis, Oxygen therapy, Sex, Delivery mode	Low GA; Low BW; Sepsis; Oxygen therapy; Female sex	Low GA; Low BW	Included demographic risk modifiers.
Adio AO et al., 2014 [9]	Oxygen exposure, Sepsis, Blood transfusion	Unregulated oxygen; Sepsis; Transfusion	Not reported	Early work; no multivariate model.
I. B. Fajolu et al., 2023 [10]	GA, BW, Sepsis, Quality-of-care bundle	Lower GA; Lower BW; Sepsis	Sepsis remained significant	Study based on NICU registry diagnosis, not systematic eye exams.
Merwe S. K. Van der et al., 2013 [11]	Ventilation mode, Apnea, GA, BW	Invasive ventilation; Recurrent apnea	Not reported	Focused on respiratory instability rather than classical neonatal risk factors.
Keraan et al., 2024 [12]	GA, BW, NICU factors	Low GA; Low BW	Not reported	Population/registry-based sampling.
Sherief et al., 2023 [13]	GA, BW, Sepsis, Oxygen therapy	Low GA; Low BW; Sepsis; Oxygen exposure	GA; BW; Sepsis	Strongest multivariable model among included studies.
Mutua et al., 2020 [14]	GA, BW, Oxygen duration, Sex	Low GA; Low BW; Prolonged oxygen	GA; BW	Centre with highest severe ROP rate.
Gezmu et al., 2020 [15]	GA, BW, Transfusion, Oxygen	Low BW; Blood transfusion	Not reported	Prospective NICU cohort.
Mutangana et al., 2020 [16]	GA, BW, Oxygen duration, Sepsis, Hyperglycemia	Low GA; Low BW; Duration of oxygen; Hyperglycemia	Hyperglycemia (aOR 3.5); Oxygen days	Only African study identifying hyperglycemia as an independent predictor.
Mhina et al., 2025 [17]	GA, BW, Oxygen therapy, Sex	Low GA; Low BW; Oxygen exposure	Not reported	Findings consistent with regional patterns.

The analytical depth of risk factors reported in the included studies varied. While some studies used multivariable logistic regression, others only reported crude (unadjusted) associations. Because adjusted analyses take neonatal variable confounding into account, they were interpreted with more weight. All risk factors analyzed across studies are compiled in Table 4, which makes a clear distinction between unadjusted and adjusted associations

Discussion

This systematic review and meta-analysis provides the first comprehensive synthesis (to our knowledge) of ROP prevalence in sub-Saharan Africa. Our findings confirm that ROP is no longer rare in African neonatal units – about 22% of high-risk preterm infants develop ROP, on average, in the studies analyzed. This prevalence is substantial, albeit somewhat lower than pooled estimates reported in Asia or Latin America in recent decades (often 30–35%) [22], and lower than an earlier Africa-wide estimate of ~30% which included North Africa [22]. The slightly lower pooled rate in SSA might reflect that many extremely premature infants still do not survive in the most resource-limited settings, keeping observed ROP rates moderate. However, the wide range we found (5% to 47%) indicates two different realities: in some African hospitals, ROP is currently an infrequent issue (likely due to high early mortality or limited neonatal intensive care), whereas in others with more advanced care, ROP incidence approaches that seen in middle-income countries.

Comparison with previous reviews

A 2019 systematic review by Wang et al. [5] noted that published ROP data were available from only 6 African countries and highlighted ROP as an emerging problem with improved neonatal survival [5]. Our review adds newer studies from additional countries (e.g., Rwanda, Botswana, Uganda, Tanzania) and uses meta-analytic techniques to derive an overall burden estimate. Another recent review focusing on incidence and screening in Africa reported that the incidence of ROP was rising and called for expanded screening programs [14]. Our results align with these qualitative conclusions and put numbers to the trend. We observed that studies published after ~2015 generally show higher ROP rates than those before 2010, supporting an upward trend as predicted by the concept of a third epidemic in Africa [18]. For example, South African data from early 2000s (Kalafong Hospital, 2002) showed ~24% ROP [19], whereas more recent South African audits report ~27–33% in some units [19], and we found ~42% in a 2018 Kenyan cohort – all suggesting increasing prevalence. Our meta-regression hinted that country-level differences in ROP prevalence exist within SSA: East African and West African NICUs with improved care (Kenya, Nigeria, Ethiopia) have caught up to (or surpassed) the ROP rates seen in South Africa, whereas places with nascent NICUs (Uganda’s general hospital, Rwanda early in program) still report lower rates (since the most fragile infants likely do not survive long enough to manifest ROP).

The low prevalence reported in Uganda may reflect differences in NICU survival of extremely preterm infants, screening timing, oxygen delivery practices, and potential under-detection in one of the participating units. It is possible that sites with more advanced neonatal care report higher ROP prevalence because more extremely preterm infants survive long enough to be screened, but this relationship cannot be confirmed with the available data.

Clinical and policy implications

The confirmation that roughly one-fifth of preterm infants in SSA develop ROP, with up to one-quarter of those cases being severe, has important implications. It means that in many African NICUs, for every 100 infants meeting screening criteria, about 20 will get ROP and perhaps 3–5 will need treatment to prevent blindness. Extrapolated to the wider region, where an estimated 16,000 infants annually are at risk in South Africa alone [12], the scale of potential ROP-related blindness is considerable. Historically, childhood blindness programs in Africa focused on infectious and congenital causes; ROP now demands attention as a preventable cause of blindness that is directly linked to healthcare improvements – a bitter irony that saving a premature infant’s life can endanger their sight if proper measures aren’t in place.

Findings suggest that strengthening neonatal care practices, including oxygen monitoring, infection prevention, and timely ophthalmic screening, may help reduce severe ROP. However, the feasibility of these recommendations varies considerably across health systems, and implementation should be context-specific [19]. The World Health Organization recommends that all “at-risk” babies (typically < 32 wks or < 1500 g) undergo their first retinal examination by 4–6 weeks of age. Implementing this requires training ophthalmologists in ROP screening (and treatment) and integrating screening into neonatal care pathways. Our review shows that where screening has been introduced (e.g., tertiary centers in our included studies), a significant burden of ROP requiring intervention is uncovered, and treatment (laser/anti-VEGF) can be successfully delivered in many cases preventing blindness. However, gaps like poor follow-up (as seen in Rwanda’s 40% loss to follow-up) and lack of local treatment expertise remain challenges. Health ministries should prioritize building capacity: this might include training pediatric ophthalmologists in ROP, investing in equipment (indirect ophthalmoscopes, laser machines, oxygen blenders), and developing retinopathy surveillance registries (such as the ROPSA register in South Africa) [19] to monitor outcomes.

Risk factor insights

The consistency of risk factors identified (low GA/BW, oxygen, sepsis, etc.) indicates that known preventive strategies are applicable in Africa. Emphasis on strict oxygen titration (avoiding hyperoxia) in NICUs is paramount – many African units have now acquired oxygen blenders and pulse oximeters, but training and adherence need strengthening. Infection prevention (hand hygiene, antibiotic stewardship, early immunizations like PCV) could yield dual benefits of reducing mortality and severe ROP [4]. Encouraging practices like kangaroo mother care and feeding with breast milk may also confer protection against ROP as suggested by the Uganda study (breast milk’s protective effect) [7]. Notably, our results highlight that improving overall neonatal care can reduce ROP in the long term – the Nigerian multi-center study showed a drop in ROP after quality improvements, echoing experiences in middle-income countries where “ROP epidemics” were curbed by better neonatal practices.

Why the variability?

We should discuss why Uganda’s ROP prevalence was so low (5.7%). That study spanned two large hospitals; interestingly, one hospital (Mulago Women’s) had 17.8% ROP while the other (Kawempe) had only 0.4% [7]. The authors attributed this to differences in care level: Mulago is a referral NICU with more small babies surviving (hence more ROP), whereas Kawempe had limited respiratory support leading to higher early mortality and almost no ROP among survivors [7]. This stark within-country contrast exemplifies how ROP rates initially rise when a NICU improves survival of tiny infants, and may remain low where such infants simply do not survive long. Similarly, Rwanda’s 7.3% in 2020 likely reflects that ROP screening had just begun after many improvements, and earlier many preemies might not have survived or been screened. As NICU capacity expands in those settings, ROP rates could climb if proactive measures aren’t taken.

Although our review identifies a number of modifiable neonatal care practices linked to ROP, the viability of putting these strategies into practice varies significantly throughout sub-Saharan Africa. Staffing, equipment, oxygen-blending capacity, and follow-up systems are all issues that many neonatal units must deal with. Therefore, rather than being applied consistently, policy and clinical recommendations should be tailored to the local context. Instead of being prescriptive guidelines, the evidence from this review should be interpreted as supporting potential strategies.

Limitations

First, this review was not registered in PROSPERO because it was initiated retrospectively after data extraction had commenced. Although this may introduce a risk of reporting bias, we followed PRISMA 2020 guidance to ensure transparency in methods and reporting our analysis has limitations. Second, the studies (and countries) remain few − 12 studies from 9 countries can hardly represent all of sub-Saharan Africa. Much of Central and West Africa had no data; as such, our pooled estimate must be generalized with extreme caution. Second, heterogeneity was moderate and other sources of variation (other than country) may have been due to variations in screening criteria (some studies screened slightly larger babies), exam time, and study year (earlier vs later in an NICUs evolution). Our meta-regression could only study country variations due to the few-sample of studies. Third, publication bias is hard with few studies, but it’s possible that smaller clinics with no ROP never published data (though we suspect this is negligible, as any ROP results tend to have been reported due to interest in Africa). Fourth, study quality was variable; some were retrospective chart reviews, which would probably underestimate cases of ROP if follow-up was incomplete. However, we endeavored to include only studies where screening was at least somewhat systematic. Excluding the Nigeria data from pooling might be seen as a limitation, but we felt that its study methodology would underestimate actual prevalence (since it used as reference the discharge diagnosis from the NICU, which would miss acute ROP cases that didn’t have an eye exam) – therefore excluding it probably improved the accuracy of our pooled estimate for actual ROP prevalence.

Furthermore, we did not measure and combine risk factor information for pooling due to reporting heterogeneity, which involved definition and measure variation; but the qualitative consensus supports the accuracy of such findings.

Additionally, differences in examination schedules, screening thresholds, and versions of the International Classification of Retinopathy of Prematurity (ICROP) may have contributed to variability across studies. Few studies used prospective multicentre designs, and there were no data from Central or Francophone Africa, limiting regional representativeness.

Future directions

We need more studies in less-explored geographic regions – and in particular, Central Africa (like DR Congo, Cameroon) and more studies from African countries that have not been represented yet like Nigeria (more than one center) or Francophone countries. Multi-center prospective studies would also be helpful for establishing incidence and outcomes of ROP on a broader spectrum of African NICUs. Cost-effectiveness analyses of ROP screening across Africa would also be helpful for policymakers; preliminary indications are that investment in ROP programmers is cost-effective with the high lifelong cost of blindness prevented.

Clinically [4], the results demand co-operation between neonatologists and eye specialists throughout Africa. National screening recommendations for ROP have been published in South Africa and Kenya [6] other nations must follow with a description of which infants to screen, where and when treatment is required. Training extra personnel is required: task-shifting towards training pediatric specialists or eye clinical officers in first-stage ROP screening could extend coverage where there is a scarcity of eye specialists as a temporizing measure. Telemedicine with retina imaging (e.g., RetCam) would also extend coverage from a distance but at a cost.

Lastly, there is a need for advocacy: African health worker and public awareness of ROP is low. Both parents and health workers do not recognize that oxygen can blind preemies in excess [14]. Our review can aid advocacy by placing a quantity on the issue – demonstrating that a high percentage of babies are at risk and that effective measures can lower risk. Incorporation of ROP training in neonatal care training programs and inclusion of ROP in neonatal ICU protocols can be taken as short-term measures.[20]

In addition, there were significant differences in ROP screening procedures between sites. While some facilities relied on NICU chart documentation, others employed standardized indirect ophthalmoscopy performed by qualified ophthalmologists. Heterogeneity in disease staging and case ascertainment may have been brought about by variations in examination schedules, screening standards, and classification schemes (ICROP versions). Lastly, the lack of prospective multicenter studies in a larger geographic area, such as Francophone and Central Africa, emphasizes the critical need for more representative and methodologically sound data.

Conclusion

Retinopathy of prematurity (ROP) is increasingly acknowledged as a critical public health concern in sub-Saharan Africa. Our systematic review discloses that around twenty percent of high-risk premature infants in this area develop ROP, with a fraction of such cases progressing to severe types that endanger eyesight. While overall prevalence (around 22%) is slightly lower than international norms, it hides wide geographic variations dependent on neonatal care quality provided. The disease burden remains highest in places where neonatal intensive care advances have allowed for the survival of a larger cohort of preterm infants a positive trend that incidentally exposes a correspondingly larger cohort of infants to adverse risk from ROP. Importantly, the major risk factors for ROP that have been identified in Africa notably prematurity, oxygen therapy, sepsis, and anemia are receptive to intervention through proven measures.

To prevent an increase in ROP-related childhood blindness within Africa, These results demonstrate the potential benefits of taking ROP into account when planning neonatal care.

This involves the introduction of ROP screening programs within all tertiary neonatal centers, training health workers on effective administration of oxygen therapy and detection of ROP, and ensuring the availability of treatments (such as laser or anti-VEGF therapy) for infants with advanced ROP. Developing context-appropriate ROP screening strategies may be beneficial, although feasibility varies widely across sub-Saharan health systems [18] and incorporate these within routine neonatal care procedures. Simple improvements, such as the use of blended oxygen, the introduction of infection control procedures, and promotion of exclusive breastfeeding, can have the ability to significantly reduce the occurrence of severe ROP [21, 23]. The example of a Nigerian unit that managed to cut the occurrence of ROP by half after improving care (including infection prevention and control demonstrates that improvement is within reach.

In short, ROP in sub-Saharan Africa can no longer be described as exotic and confined to academic discussion it’s a presence within neonatal services and on the rise as survival of early-birth children increases. As neonatal survival improves across parts of the region, consideration of ROP within routine neonatal care may become increasingly relevant through the use of ROP screening and preventive care within health systems for neonates, improved detection and timely management could potentially prevent some cases of ROP-related vision loss, but the magnitude of achievable impact depends on local infrastructure and follow-up systems. What initially was a disease of the industrialized world has struck firm roots on African soil; with collaborative effort, this region can avoid the full force of an ROP blindness pandemic. Our review provides evidence towards policy intervention and resource distribution and with special emphasis on the point that the survival and eyesight of each and every child born preterm matter.

Author contributions

Dr.Mohamed Farah Ismail conceived and planned the study, conducted the literature search, executed data extraction, conducted statistical analysis, and prepared the manuscript. Prof. Intisar Khalafalla critically reviewed the methodology, validated the data, and revised manuscripts. All authors read and approved the final manuscript. Acknowledgments The authors appreciate the support from the Department of Ophthalmology, Kampala International University Teaching Hospital.

Funding

This study did not get any particular financial support from public, commercial, or non-profit funding agencies.

Data availability

All data generated or analyzed throughout this study are contained in this published article.

Declarations

Ethics approval and consent to participate

Not applicable, as this study is based solely on analysis of previously published data and does not involve any human participants or identifiable individual data.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.