Results synthesis
The authors first extracted the relevant data from the selected studies, entered the data into Microsoft Excel 2016, and used R statistical software to assess the outcomes. A table and a forest plot were used to represent and report the results. By applying the binomial distribution formula, the standard error of the incidence of anemia for each original study was calculated. Using the I2 test, any potential heterogeneity was investigated in the stated prevalence of the studies. A random effects model was employed to estimate the Der Simonian and Laird’s pooled effect.
Study selection
A literature search revealed a total of 102 studies on various kinds of anemia risk factor prevention, etc. Due to duplication, 10 of them were eliminated. Another 26 studies were eliminated after the titles and abstracts were screened because the content was irrelevant to the goal of our review, as we were bound to include studies on IDA prevalence in females. We eliminated 37 of the 66 preselected papers because they did not match the selection criteria.
Study characteristics
Two National Health Surveys were considered: NNS 2011 and NNS 2018. Eleven studies on nonpregnant women/adolescent girls with 2,294 participants were included, and sixteen studies on IDA incidence among pregnant women with 9,263 participants were included in this systematic review. The data were obtained from seven studies conducted in Sindh, ten studies from Punjab, five studies from KPK, two studies from Baluchistan, two studies from AJK, and two national nutritional health surveys that show prevalence throughout the whole country. According to the WHO criteria, 12 mg/dl is the cutoff for pregnant/adolescent females, and 11 mg/dl is the cutoff for pregnant females [9].
Quality assessment of studies
The studies included in the current review were cross-sectional. The New Castle Ottawa tool was used to assess overall quality. Among the group of adolescent girls, 11 studies were included. Of these 11 studies, five were identified as high-quality, with scores of 09 and 07, four studies were classified as moderate with a score of 5, and two studies were classified as low-quality, with scores of 04. Among the second group of pregnant females, 16 studies were included, among which six studies had high-quality scores of 7 and 8, seven had moderate quality scores of 05 and 06, and three had low-quality scores of 04 and less than 04.
Figure 1 shows the selection process for the studies involved. The following databases were used for the search: Google Scholar, PubMed, Scopus, and ScienceDirect.
Flowchart describing the details of the papers included in the review
Data were extracted from studies conducted in different geographical regions of Pakistan by using various key terms from the PICOS. The frequency, rates, prevalence, proportion, and magnitude of iron deficiency in young females are shown in Table 1.
Table 2 shows the results of two national nutrition surveys, which depict anemia incidence from various geographical locations in Pakistan. In both surveys, the population residing in rural settings had a greater prevalence than the females residing in urban areas, and a similar pattern was observed in pregnant and nonpregnant individuals in both groups. A survey in the 2011 Sindh Province revealed a greater prevalence of anemia in both groups, with 59.7% in pregnant females and 62.0% in the adolescent/WRA group; however, in the survey in the 2018 Baluchistan Province, the highest prevalence was 53.9% and 61.8%, respectively.
Table 3 presents the characteristics of a total of 11 studies on adolescents/nonpregnant individuals, with 2,294 sample size studies representing mostly large cities and their peripheries; no single study from Gilgit Baltistan has been conducted on IDA prevalence to date to provide figures.
In the northern areas of Pakistan, a study conducted in Peshawar had a sample size of 259 reported 23.9% and the study conducted in Mardan had a sample size of 50 reported 10% showing less prevalence; however, studies performed in Lahore and Rawalpindi have shown a higher prevalence of anemia in the non-pregnant/adolescent group (84% and 71.2%, respectively).
Meta-analysis
To provide a visual summary of the data, a forest plot (Fig. 2) was generated to evaluate the pooled effect size of each study with their respective confidence intervals (CIs). The forest plot of the 11 studies included in this analysis revealed that the pooled prevalence of anemia among nonpregnant adult/adolescent females was 54.6% (95% CI: 0.422, 0.669). We detected significant heterogeneity among the included studies, as indicated by the I2 statistic (I2 = 97.8%, p value < 0.001). Because the DerSimonian and Laird random effects model provides more conservative effect sizes, we used it to estimate the total pooled prevalence of anemia among nonpregnant adult/adolescent females.
Forest plot showing the pooled prevalence among nonpregnant females
Subgroup analysis
The pooled prevalence of anemia across studies with various sample sizes was revealed by subgroup analysis by sample size (Table 4). With a broad confidence interval indicating considerable uncertainty, the estimated pooled prevalence for studies with sample sizes under 200 was greater at 60.5%. The pooled prevalence is also lower—44.4%—for studies with sample sizes over 200—again, with a confidence interval that considers this heterogeneity.
Significant variability among studies within each category is indicated by the subgroups’ notably high I2 values (97.23% and 97.26%) and extremely low p-values (0.001). The socioeconomic status subgroup analysis sheds light on how anemia prevalence changes between various socioeconomic categories. In this analysis, studies with participants from mixed backgrounds showed a higher prevalence. There was considerable heterogeneity among the studies within each of the three categories, as indicated by the large I2 values and extremely low p-values (0.001) for all three subgroups.
Table 5 represents a total of sixteen studies with 9,263 participants on the incidence of anemia in pregnant females in Pakistan by year, population, SES, and technique used for Hb estimation. The same pattern is observed here; almost all studies represent large cities or their peripheries. Most of these studies do not use Hb estimation techniques. A study from Rawalpindi showed the lowest prevalence rate of anemia which was 46.7%, with a sample size of 860, and the highest prevalence was reported by a study conducted in AJK, which was 91% in pregnant females with a sample size of 1000. The prevalence of anemia by city of residence showed significant differences when compared with studies conducted in comparatively small cities, as studies conducted in Karachi, Lahore, Islamabad, and Quetta showed a lower frequency when compared with small cities and peripheries.
Meta-analysis of studies on pregnant women
To offer a visual summary of the data in this meta-analysis, a forest plot (Fig. 3) was generated to evaluate the pooled effect size of each study with their respective confidence intervals (CIs). The pooled prevalence of anemia among pregnant women by using the Der Simonian and Laird random-effects model, was 70.4% (95% CI: 0.619, 0.789) as shown by the forest plot of the 39 studies included in this analysis. We discovered significant heterogeneity among the included studies, (I2 = 99.0, p-value = 0.001).
Forest plot showing the pooled prevalence among pregnant females
Subgroup analysis
The subgroup analysis by region reported in Table 6 highlights the prevalence of anemia differs across the different regions. The pooled prevalence estimate, confidence interval, and measures of heterogeneity differ for each region. A higher prevalence was reported in Punjab Province, among other regions77.4 (0.630,0.918). Strong I2 values and extremely low p-values (0.001) illustrate the significant heterogeneity among the studies within each region. The Baluchistan region is the exception: no observed heterogeneity exists (I2 = 0.00%). Trimester-wise subgroup analysis of the second trimester reported a greater pooled prevalence.
Add Comment