Prevalence and risk factors of retinopathy of prematurity in Iran: a systematic review and meta-analysis

Background Retinopathy of prematurity (ROP) refers to the developmental disorder of the retina in premature infants and is one of the most serious and most dangerous complications in premature infants. The prevalence of ROP in Iran is different in various parts of Iran and its prevalence is reported to be 1–70% in different regions. This study aims to determine the prevalence and risk factors of ROP in Iran. Methods This review article was conducted based on the preferred reporting items for systematic review and meta-analysis (PRISMA) protocols. To find literature about ROP in Iran, a comprehensive search was done using MeSH keywords in several online databases such as PubMed, Ovid, Science Direct, EMBASE, Web of Science, CINAHL, EBSCO, Magiran, Iranmedex, SID, Medlib, IranDoc, as well as the Google Scholar search engine until May 2017. Comprehensive Meta-analysis Software (CMA) Version 2 was used for data analysis. Results According to 42 studies including 18,000 premature infants, the prevalence of ROP was reported to be 23.5% (95% CI: 20.4–26.8) in Iran. The prevalence of ROP stages 1, 2, 3, 4 and 5 was 7.9% (95% CI: 5.3–11.5), 9.7% (95% CI: 6.1–15.3), 2.8% (95% CI: 1.6–4.9), 2.9% (95% CI: 1.9–4.5) and 3.6% (95% CI: 2.4–5.2), respectively. The prevalence of ROP in Iranian girls and boys premature infants was 18.3% (95% CI: 12.8–25.4) and 18.9% (95% CI: 11.9–28.5), respectively. The lowest prevalence of ROP was in the West of Iran (12.3% [95% CI: 7.6–19.1]), while the highest prevalence was associated with the Center of Iran (24.9% [95% CI: 21.8–28.4]). The prevalence of ROP is increasing according to the year of study, and this relationship is not significant (p = 0.181). The significant risk factors for ROP were small gestational age (p < 0.001), low birth weight (p < 0.001), septicemia (p = 0.021), respiratory distress syndrome (p = 0.036), intraventricular hemorrhage (p = 0.005), continuous positive pressure ventilation (p = 0.023), saturation above 50% (p = 0.023), apnea (p = 0.002), frequency and duration of blood transfusion, oxygen therapy and phototherapy (p < 0.05), whereas pre-eclampsia decreased the prevalence of ROP (p = 0.014). Conclusion Considering the high prevalence of ROP in Iran, screening and close supervision by experienced ophthalmologists to diagnose and treat the common complications of pre-maturity and prevent visual impairment or blindness is necessary.


Background
Retinopathy of prematurity (ROP) refers to the developmental disorder of the retina in premature infants and is one of the most serious and most dangerous complications in premature infants.
Embryonic retinal arteries start to grow in the third month of pregnancy and their development ends at birth. Therefore, the stages of evolution of the eye are defective in premature infants, and the growth of the vessels is either stopped or unusual, and ultimately, the vessels become very fragile, which can lead to visual impairment in severe cases [1].
Despite considerable progress made in the treatment of ROP, it is still a common cause of reduced vision in children in developed countries, and its prevalence is increasing [2][3][4]. This is a preventable disease and responds to treatments appropriately if diagnosed at early stages, but in case of delayed diagnosis and treatment, it may lead to blindness [5].
The first incidences of ROP were reported in the 1940s and 1950s, mainly as a result of the use of supplemental oxygen without supervision (first epidemic). Although the survival of premature infants improved in the following decades, and despite improved monitoring methods for oxygen supplements, ROP emerged with an increasing incidence (second epidemic) [6]. Over the past decade, the increasing incidence of ROP blindness has been recorded in low-income countries. Studies show that ROP is the leading cause of blindness in China, Southeast Asia, South America, Latin America, and Eastern Europe, especially in urban centers of newly industrialized countries, and this is referred to as the "third epidemic" [7].
ROP is a multifactorial disease and the most important risk factors are preterm delivery, especially before the 32nd week of gestation and birth weight less than 1500 g. Apnea, intraventricular hemorrhage, various maternal factors (diabetes, preeclampsia, mother's smoking), respiratory disorders, infection, vitamin E deficiency, heart disease, increased blood carbon dioxide, increased oxygen (O 2 ) consumption, decreased PH, decreased blood O2, bradycardia, transfusion, amount of received oxygen and duration of ventilation are other risk factors for ROP [8][9][10].
The prevalence of ROP in different regions of Iran is different and its prevalence is reported to be 1-70% in different regions [11][12][13][14]. Considering the abovementioned issues and the importance of the subject, as well as the diversity of reports in Iranian studies, it is necessary to carry out more extensive and precise studies. Meta-analysis is a method that collects and analyzes multiple research data with a common purpose to provide a reliable estimate of the impact of some interventions or observations in medicine [15,16]. Obviously, the sample size in meta-analysis becomes larger by collecting data from several studies and therefore the range of changes and probabilities will be reduced; therefore, the significance of statistical results increases [16,17]. This study aims to determine the incidence and risk factors for ROP in Iran.

Study protocol
This review article was conducted based on the preferred reporting items for systematic review and meta-analysis (PRISMA) protocols [16]. The study was conducted in five stages: design and search strategy, a collection of articles and their systematic review, evaluation of inclusion and exclusion criteria, qualitative evaluation and statistical analysis of data. To avoid bias in the study, each of the above steps was carried out by two researchers independently. In case of differences in the results obtained by the two researchers, a third researcher intervened to reach an agreement.

Search strategy
To find literature about ROP in Iran, a comprehensive search was done using the terms (Retinopathy of Prematurity [MeSH]) AND ("Incidence" [MeSH] OR "Epidemiology" [MeSH]), OR ("Prevalence" [MeSH]) AND ("Iran" [MeSH]) in 7 international databases including PubMed, Ovid, Science Direct, EMBASE, Web of Science, CINAHL, EBSCO, and 5 national databases including Magiran, Iranmedex, SID, Medlib, IranDoc, as well as Google Scholar search engine until May 2017. References to all relevant articles were reviewed. Due to the inability of Iranian databases to search using Boolean operators (AND, OR and NOT), searches on these databases were only performed using the keywords.

Inclusion and exclusion criteria
Articles with the following characteristics were chosen for meta-analysis: 1. Original research papers published either in Persian or English; 2. Medical dissertations; 3. Review of the prevalence or risk factors for ROP. The exclusion criteria were: 1. Non-random sample for estimating the prevalence; 2. Being irrelevant to the topic; 3. Congress papers; 4. Sample size other than premature infants; 5. Non-Iranian studies; 6. Review articles, case reports, editorials; 7. Duplicate studies and 8. Lowquality studies.

ROP detection criteria
ROP was diagnosed by an expert through examination of retinas of infants using indirect ophthalmoscope.

Selection of studies
First, all related articles (articles with affiliations containing Iranian authors) were collected and a list of titles was prepared at the end of the search and removal of duplicates. After blinding the specifications of the articles by on researcher (Milad Azami), including the name of the journal and the name of the author, the full text of the articles was presented to the researchers. Each article was studied by two researchers independently (Gholamreza Badfar, Afsar Dastjani Farahani). If the article was rejected, the reason for this rejection was mentioned. In case of disagreement between the two authors, the article was judged by the team of researchers.

Quality of studies
Using the standard modified Newcastle Ottawa Scale (NOS) checklist [18], which included 8 sections. Thus, the minimum and maximum score available on this checklist were 0 and 8, respectively. Accordingly, the studies were divided into three categories: 1. low quality with a score less than 5; 2. moderate quality with a score of 5-6; and 3. high quality with a score of 7-8. Finally, the moderate to high quality studies were selected for the meta-analysis stage.

Data extraction
The raw data of the prepared articles were extracted using a premade checklist. The checklist includes the name of the authors, published year the year of study, the location of the study, the study design, quality score, sample size, the prevalence of ROP, the ROP detection criteria, the prevalence of ROP based on gender (ROP) and ROP risk factors.

Statistical analysis
In each study, the prevalence of ROP was considered as the probability of binomial distribution. To evaluate the heterogeneity of the studies, Cochran's Q test and I 2 index were used [19]. There are three categories for the I 2 index: heterogeneity lower than 25%, heterogeneity between 25% and 75% and heterogeneity more than 75%. Considering the heterogeneity of the studies, a random effects model was used to combine ROP prevalence. For ROP risk factors, the fixed effects model and the random effects model were used, respectively in the case of low heterogeneity and high heterogeneity in the metaanalysis [20,21]. Sensitivity analysis was performed to identify the influence of a single study on the combined result incidence or any risk factors (with ≥ 7 studies). In order to identify the cause of heterogeneity of ROP prevalence, sub-groups analysis of ROP were carried out based on geographical region, province and quality of studies, while the metaregression model (method of moments) was carried out based on the year of studies [22]. Egger and Begg's tests were used to identify publications bias. Data analysis was performed using Comprehensive Meta-Analysis Software Version 2 and the significance level in the tests was considered to be lower than 0.05.

Search results and characteristics
In the initial search, 452 studies were found to be related to the topic. Two independent researchers reviewed the title and the abstract. If the title or abstract was likely to be related to the topic, the full text was reviewed. After reviewing the full text of 74 relevant articles, 30 articles were omitted due to lacking the necessary criteria and finally 44 qualified studies entered the qualitative assessment stage (Fig. 1).  (84), respectively (Fig. 2).

Sensitivity analysis and cumulative analysis for ROP
The sensitivity analysis of the prevalence or risk factors of ROP and its 95% confidence interval (CI) was estimated simultaneously regardless of one study and the results showed that the incidence or risk factors of ROP were not significantly changed before and after the deletion of each study. (Fig. 3a). Cumulative analysis for incidence of ROP based on the year of publication is shown in Fig. 3b.  Fig. 2 The prevalence of retinopathy of prematurity in Iran. Random effects model  Subgroup analysis of ROP prevalence based on province Table 2 and Fig. 4 show the prevalence of ROP based on Iran's provinces. The highest prevalence was in provinces of Mazandaran (34.8%) and Khuzestan (32%), and the lowest prevalence was in the provinces of Semnan (1.1%) and Zanjan (1.2%).

Subgroup analysis of ROP prevalence based on the quality of studies
The prevalence of ROP in moderate and high-quality studies was 23.5% (95% CI: 16.6-28.0) and 23.5% (95% CI: 19.1-28.7), respectively, and the difference was not statistically significant (p = 0.995) ( Table 2).

The prevalence of ROP based on stage
The prevalence of stages 1, 2, 3, 4 and 5 were reported in 10, eight, nine, five, and five studies, respectively. Fig. 5 shows the prevalence of ROP at different stages. The

Meta-regression
Meta-regression model in Fig. 6 shows that the incidence of ROP is increasing according to the year of study, and this relationship is not statistically significant (meta-regression coefficient: 0.034, 95% CI -0.016 to 0.085, P = 0.181).

Publication bias
The significance level of publication bias in the reviewed studies was 0.003 and 0.002 according to Egger and Begg's tests, respectively, which is shown in Fig. 7.

Discussion
The present study is the first systematic and metaanalytic review on the prevalence and risk factors of ROP in Iran. The results of this meta-analysis showed that the prevalence of ROP in 18,000 Iranian premature infants was 23.5%, and the prevalence for stages 1, 2, 3, 4 and 5 was 7.9%, 9.7%, 2.8%, 2.9% and 3.6%, respectively. In this study, the level of heterogeneity was high for ROP studies (95.6%). The results of the subgroup analysis showed that geographic regions and the provinces could be a cause of high heterogeneity. However, this difference can be a reflection of studies conducted on different samples based on the GA or neonatal weight.
ROP is still a major cause of potentially preventable blindness around the world [23]. According to guidelines published by the American Academy of Ophthalmology, the American Academy of Children, and the American Association for Ophthalmology for Children and Strabismus for ROP screening, infants weighing less than 1500 g or GA ≤ 30 weeks, and infants weighing between 1500 and 2000 g or GA > 30 weeks with an unstable clinical course should receive dilated ophthalmoscopy examinations for ROP [24].
The prevalence of ROP in various studies is mainly due to differences in mean GA and birth weight of infants in each study. Based on GA, the prevalence of ROP significantly decreases from 77.9% in GA 24-25 to 1.1% in GA 30-31, which indicates the direct role of GA in ROP incidence. These results are completely consistent with the data published in other literature [25][26][27][28][29][30][31]. Moreover, in a meta-analysis study in Iran, the  prevalence of prematurity was reported to be 9.2% (95% CI: 7.6-10.7) [32]. Therefore, the high prevalence of ROP in Iran (23.5%) can be explained by the high prevalence of prematurity.
In a study by Tabarez-Carvajal et al. among 3018 premature infants, the incidence of stages 1, 2, 3, 4, and 5 was reported to be 8.34%, 8.78%, 1.9%, 0.03%, and 0.30%, respectively [33]. In another study by Abdel HA et al.,  [34]. But in the present study, the prevalence of ROP stages 4 and 5 was higher. ROP is a multi-factorial disease, and in the present study, the strongest risk factor for ROP was prematurity and low birth weight. Most studies have demonstrated that prematurity and low birth weight are the strongest predictive factors of ROP, which indicates the crucial role of factors associated with the progression of the ROP disease [35][36][37][38][39][40][41][42][43][44][45].
After low birth weight and prematurity, exposure to oxygen for a long period and saturation over 50% were the most important risk factors for ROP in this study, which was consistent with the results of many other studies [42][43][44][45][46][47]. Due to inadequate antioxidant defense system, premature infants are not evolved to live in an oxygen-rich ectopic environment [48,49]. Oxidative stress is the result of various organs' exposure to free radicals of oxygen after being exposed to high concentrations of oxygen, which can lead to the progression of many pathogens such as ROP, necrotizing enterocolitis, IVH, bronchopulmonary dysplasia, and periventricular leukomalacia [50,51].
In this study, other significant relationships with ROP were also found, including frequency and duration of blood transfusion, phototherapy, septicemia, apnea, IVH, and RDS. The comparison between the risk factors in our study and other reports is shown in Table 4.

Conclusion
Finally, it can be concluded that the present systematic review and meta-analysis summarizes the results of previous studies and provides a comprehensive view of ROP in Iran. Although the prevalence of ROP in Iran is similar to some developing countries, it is much higher than some other countries. Therefore, this fact highlights the importance of preventing and treating ROP and its following complications. To achieve a more favorable level and reduce the prevalence in the coming years, screening and close monitoring by experienced ophthalmologists are essential to diagnose and treat the common complications of prematurity and prevent visual impairment or blindness. Authors' contributions MA was involved in study concept and design, acquisition of data, search, quality evaluation of studies, drafting of the manuscript, analysis and interpretation of data, critical revision of the manuscript for important intellectual content, approval of final version, and accountable for accuracy and integrity of the work. ZJ was involved in search, interpretation of data, acquisition of data, quality evaluation of studies, drafting of the manuscript, and approval of final version. ShR was involved in search, analysis and interpretation of data, quality evaluation of studies, drafting of the manuscript, and approval of final version. GhB was involved in study concept and design, acquisition of data, quality evaluation of studies, drafting of the manuscript, critical revision of the manuscript for important intellectual content, approval of final version, administrative, technical or material support and accountable for accuracy and integrity of the work. ADF was involved in search critical revision of the manuscript for important intellectual content, and approval of final version.