Nepal Journ

Copyright © 2021 CEA& INEA. Published online by NepJOL-INASP. www.nepjol.info/index.php/NJE Correspondence: Dr. Kapil Yadav, Room no. 37, Centre for Community Medicine, All India Institute of Medical Sciences, New Delhi-110029, India. Email: drkapilyadav@gmail.com Received 27 April 2021/Revised 07 June 2021/Accepted 14 June 2021 Citation: Sharma S, Khandelwal R, Yadav K, Ramaswami G, Vohra K. Effect of cooking food in ironcontaining cookware on increase in blood hemoglobin level and iron content of the food: A systematic review. Nepal J Epidemiol. 2021;11(2):994-1005. DOI: 10.3126/nje.v11i2.36682 This work is licensed under a Creative Commons Attribution 4.0 International License. Abstract:


Introduction
Polycystic ovary syndrome (PCOS) is a common endocrinological problem in the progenitive age group and the leading cause of infertility. PCOS expose women at high risk of developing complications, such as infertility, metabolic disorders such as type 2 diabetes mellitus, hypertension, other cardiovascular and cerebrovascular disorders, if it remains untreated [1]. Such complications impair social and mental well-being, and adversely impacted the quality of life of the patients. Long-term follow-up of 786 women with PCOS observed an elevated risk of endometrial cancer [2,3]. The prevalence of PCOS ranges from 2.2% to 26% worldwide [4]. Among Asian countries, the prevalence of PCOS ranges from as low as 2.4% in China [5] to 19.5% in Iran [6]. A recent metanalysis (2019) by Naz et al. concluded that there is a regional variation in the prevalence of PCOS among adolescents. Varying prevalence is due to different diagnostic criteria, i.e. National Health Institute (NIH) 1990 , Rotterdams (2003) and Androgen excess society (AES) 2006. The reported prevalence was 11.04% with Rotterdam criteria, 3.39% with NIH criteria, and 8.03% based on AES [7]. The World Health Organization (WHO) defines an adolescent as any person between 10 and 19 years. Due to the modern lifestyle, PCOS is commonly seen in the adolescent age group. The clinical manifestation of PCOS may be complete in adolescence, but the diagnosis is challenging as some features of PCOS overlap with the transitional characterstics from puberty to adulthood [8]. Hyperandrogenism and oligoanvolution criteria suggested by AES (2006) is not considered valid for early teenagers [9]. To date, available studies possessess limited sample size which failed to capture attention of policy makers. School health programme could have been the best platform to capture PCOS at an early age. Due to limited awareness of this issue among the school staff, health care workers and beneficiaries, prevailing cultural taboos, this issue remain unaddressed, neglecting its long term effect on women's physical and mental health. This systematic review and metaanalysis aimed to estimate the pooled prevalence of PCOS among adolescent girls (14-19 years) in India. Mean age of girls in India to start menstruration is 12 years and PCOS occurs after 2 years of menarche, therefore we chosen 14-19 years age group adoloscents.

Literature Searches
This review's conducting and reporting adhered to the PRISMA (Preferred reporting items for systematic reviews and meta-analyses) guidelines [10]. We extensively searched studies from 2011 onwards, three databases (PubMed, Embase, and Scopus) with Mesh terms related to polycystic ovarian syndrome and adolescence like subheadings B.M.I, Oligomenorrhea, Hirsutism, PCOS, Waist-Hip ratio. The search strategy for each database is in Annexure 1.

Inclusion and exclusion criteria
The observational studies with the following criteria: (1) Cross-sectional or cohort studies with participants as Indian Adolescent girls (14-19 years). (2) Outcome measured in terms of PCOS prevalence using NIH Criteria, AES criteria or Rotterdam's criteria. We extracted the data of age group 15-19 years from the study. However while searching databases we found studies related to age group 10-19 years. Onset of menarche more than 2 years was inclusion criteria for this studies and considering the mean age of menarche as 12, we decided to include this studies. There may be very negligible proportion below 14 years. We excluded the article with non-English language and non-availability of full text with our best efforts.

Selection of studies and Data Extraction
We exported eligible studies to Zotero version 5 and removed duplicates. Two researchers (MK and MS) trained in Systematic review independently screened each article using a three-step approach. First, we reviewed the titles and abstracts of studies and excluded studies that did not meet eligibility criteria. We further examined the retrieved full text of eligible studies and noted the reasons for exclusion. Discrepancy was resolved with discussion and a third researcher (VS). We Cross-referenced all included studies to find out missed out relevant studies, searched through Google Scholar. MK and MS independently extracted data with data extraction form. Data extraction form consisted of study design, sample size, place of study(urban/ rural), Setting (community/school/hospital), publication year, sampling technique, diagnostic techniques, and PCOS prevalence. We resolved the discrepancy in data extraction by discussion and built consensus.

Risk of Bias
We used the NIH quality assessment tool [11] to assess the quality and risk of bias of the included studies. The NIH tool consisted of 14 dichotomous (yes/no) questions, out of which Q.10 and Q.13 were not applicable for cross-sectional studies.
We rated the quality of each study based on our best judgment. If the answer to any question was "No", then examined the potential of bias in a study.  The random-effect model (Dersermonion and Laird method) with "dmetafor" function was used for analysis due to high heterogeneity. We assessed and quantified the risk of publication bias with the Funnel plot and Eggers regression test, respectively. We also found outlier using "dmetar" package in R. We performed a Sensitivity analysis to explore the influence of study setting and study design on PCOS prevalence. We have merged school/colleges as community-based studies for sensitivity analysis. This metaanalysis was registered at Prospero (Registration# CRD42020212443). Ethical approval was obtained from Institutional Ethical Committee prior to commencing the systematic review.

Results
The extensive search of potential studies in the three databases, yielded a total of 247 studies which were imported to Zotero. After removing the duplicates (n=64) and relevant 47 studies were subjected to full-text review, and finally included 12 articles in the present systematic review and metaanalysis. (Fig 1)

Characteristic of included studies
Characteristics of included studies is given in

Quality assessment
Cross-sectional studies scored between 6-10 out of 12, and cohort study scored 11 out of 14. Half of the studies (n=7) rated as fair quality and the later half as good (n=5). Except for three studies (14,15,23), all others have a low risk of bias. All studies reported a lower non-response rate. Studies done in a hospital setting had no recall bias as diagnosis mainly depends on the physician. Gupta [18] reporting of Hirsutism and androgen production was not based on the Ferryman-Gallewey model [24]. Participants selected their degree of hair growth across nine key anatomical sites based on pictographic representations, so we considered this study had a risk measurement bias. (Table 2) Outcome measures and Heterogeneity among included studies We used Rotterdam crtiteria to estimate the pooled prevalence as it was used in all the included studies. Pooled proportion of PCOS among adolescents girls (14-19 years) in Indian settings with random effect model was found to be 17.74 per 100 ( CI = 11.77-23.71) with I² statistics of 97% (Fig 2).

Sensitivity analysis
We conducted a sensitivity analysis to explore the influence of study setting and design on the proportion of PCOS. Figure 3 A shows pooled proportion according to different study setting. In a Community Setting with six studies, PCOS was 11% ( CI=5%-17%), I2 =96%, while in a hospital setting, the pooled prevalence was 25% ( CI= 12%-39%), I²=98% . Fig 3  B shows the pooled proportion of PCOS in cross-sectional studies was 16% (10%-22%), I²=97% and in one cohort study, the proportion was 36% (28%-44%). Fig 4 shows the funnel plot of 12 studies. Two studies Nair. M [13] and Balaji [16] are far from a funnel with large effects and small sample sizes. Gupta M. [18] had a low prevalence of 0.3 with small SE (large sample). Egger test intercept was 8.089 (3.71-12.47) and p-value < 0.05 depicting the presence of asymmetry in the funnel plot. Fig 5 shows five studies as outliers, and when removed, the heterogeneity was 0 % and pooled prevalence of 13.8% CI(12.37% to 15.22%).

Discussion
The pooled proportion of PCOS among adolescents girls age group (14-19 years) according to Rotterdam [18]. A meta-analysis done by Naz MS 2019 [7] included studies worldwide, including four from India, five from Iran and one each from Thailand and Southern California respectively. All these studies were communitybased, and the finding was comparable to our pooled community-based studies i.e 11 per 100 (0.05-0.17). The participants' mean age in the current review was 16.8 years (SD=2.059), and in Naz was 16.99 years (16.46-17.52). The pooled prevalence as per NIH criteria was 3.39% (0.28-9.45%), and AES was 8.03% ( 6.24-10.01%). Publication bias was not seen in Naz study. We have not conducted a pooled analysis of prevalence as per NIH and AES criteria as it was estimated respectively in two studies and one study only [12,14,16]. Further in our meta-analysis, publication bias was seen. Two hospital-based studies [13,16] (6) The prevalence among women is higher than the adolescent age group because, first, PCOS is often overlooked in adolescent age. Secondly, it is challenging to diagnose as a menstrual pattern is difficult to distinguish from anovulation associated with puberty due to maturating hypothalamic pituitary ovarian axis [27]. We found a wide prevalence of PCOS within India. Metaanalysis by Ding and colleagues, reported prevalence rate in Chinese women was 5.6% (based on the Rotterdam criteria) and in Middle Eastern countries was 16% [28]. These differences in the prevalence of PCOS across geographical location, racial or ethnic groups may be attributed to environmental or genetic influence, study design, study setting, different diagnostic criteria and pattern of utilization of health care services. However, we do not attribute variation in our study to geographical location, as confirmed by Wolf,et al. [29]. Hospital based studies [12,13,16,19,21,22] [23], questionnaire was self administer resulting in low prevalence.

Expert opinion
Diagnosis of PCOS in adolescence is a challenge because of overlapping symptoms of PCOS with normal pubertal changes in adolescents. Incidence of PCOS among adolescence has been increasing due to modern lifestyle. Early diagnosis is important to instill early lifestyle modifications which will prevent metabolic and reproductive complications. Health education and screening for PCOS need to be incorporated in School or community based adolescent health programs. An assessment of target groups by simple menstrual history could detect possible PCOS during early adolescence to facilitate appropriate early intervention.
We also recommend open discussion on menstruation problems in adolescent groups should be encouraged in schools and aaganwadi centres. Health talk should be established between mother and daughter and break the hesitance to seek timely medical advice.

Limitation of the study
The main limitation of our review is searching only three databases for screening and identifying studies. However, we conducted, a thorough search of the cross-references to include any eligible studies via google scholar. Epidemiology study design , varied study population and methodology added to heterogeneity. We could find only six community-based studies with a limited sample size may not be adequate to comment on PCOS prevalence. Publication bias was noted with five outliers. Hospital-based study may have distorted the prevalence to the higher side, which was later rectified by sensitivity analysis showing comparatively lower prevalence.

Conclusion
It can be inferred that cooking food in iron pot escalates the levels of blood hemoglobin and iron content of the food, and thus reduces the incidences of iron deficiency anemia. The bioavailability of food containing heme iron increases more when cooked in iron pot than food having non-heme iron form. Also, the content of iron in the food was found to be increased by cooking acidic food with iron ingot. Very limited research trials are available on this topic that warrants a careful interpretation of results inferred and a considerable need of larger population-based studies and randomized controlled trials for better outcomes. Acknowledgement None Authors' contribution All authors have made substantial contributions to all of the following: (1) the conception and design of the study, or acquisition of data, or analysis and interpretation of data, (2) drafting the article or revising it critically for important intellectual content, (3) final approval of the version to be submitted.

Competing interests
There is no conflict of interest between any of the authors.

Source of Support:
No financial support was received for the present study from any of the public, commercial, or not-for-profit sector agencies. Authors' affiliation: 1,4