Introduction
Low birth weight (LBW) is a major problem in different societies. On average, 7.7% of infants weigh less than 2500 gr. Infants with birth weight less than 2500 gr are known as LBW infants, regardless of gestational age. According to a study, 10% of infants born in the United States are LBW infants (2). Several maternal factors such as age, anemia, etc. may affect LBW of the infants. Therefore, anemia during pregnancy can be considered as one of the major causes of infant LBW (1, 2). According to the data collected from the Center for Disease Control in 1989, if hemoglobin level in the first and the second trimester of pregnancy is less than 11 g/dL and is less than 10.5 g/dL in the third trimester, this condition is considered anemia (3). According to the recommendations of World Health Organization (WHO) in 1972, a pregnant woman is diagnosed with anemia when the hemoglobin level is below 110 g/l. Based on this report, more than 50% of women who do not take supplements suffer anemia (4). Statistics show that 14-62% of women in developing countries, and 16-29% in developed countries are suffering from anemia (5-7). LBW infants are more vulnerable and exposed to different health problems and complications compared with infants with normal weight. Based on other reports, there is a direct relationship between Low levels of hemoglobin during pregnancy and high birth weight (8, 9).
Considering the high prevalence of anemia among pregnant women aged 17-45, contradictory results regarding the effect of maternal anemia on birth weight and absence of a definite conclusion in obstetrics and genealogy books, we decided to investigate this topic using meta-analysis methods (10, 11). These methods are able to put together all the facts, thus presenting a more clear and accurate picture of the problem to provide a scientifically proved advice for pregnant women (13, 14).
Go to:
Materials and Methods
Data sources
This study was a systematic review of the literature and methods based on PRISMA guideline from the beginning of 1990 until April 2017 (15). Valid keywords were used for searching in national (Sid, Iran.doc, Iran medex and Magiran) and international (PubMed, Science Direct, Cochrane, Medline, Web of Science, Scopus, Springer, Embase, Google scholar) databases. In order to maximize the comprehensiveness of the study, general Persian keywords and all possible combinations (maternal anemia and LBW) were used in the search. We used English equivalents such as “anemia”, “LBW” and “pregnancy” in international databases and the “ either/or” structure was applied for further searching; references were also examined to find further necessary sources on the subject.
Study selection (Inclusion and exclusion criteria)
STROBE, which is a standard and famous international checklist, was used to determine the quality of the study (17). This checklist contains 22 sections, covering various aspects of the study including methodology, sampling, variables measurement, statistical analysis, modification of confounders and the reliability and validity of the tools used to evaluate the study objectives. 15.5 was considered the minimum attainable score. Finally, studies that attained the minimum score of the STROBE checklist entered the study and their data were analyzed using meta-analysis method.
Study selection
First, researchers prepared a list of topics and abstracts of each study at the beginning of study and studies that covered the relationship between maternal anemia and LBW based on pregnancy month entered the main study. Cohort, case-control and cross-sectional that evaluated the relationship between maternal anemia and LBW were analyzed in the present study. Relative risk with 95% confidence interval was used to assess the significance of the study.
Data extraction
Data extraction was performed by two researchers independently and data extraction form (containing name of the author, year of publication, country, continent, number of participants, trimester and LBW pregnancy outcomes) was used to minimize bias and error in collecting data. In cases of necessity, ambiguity of article information and some certain questions, we contacted the authors by E-mail. The data extracted by the two researchers was compared and discussed in case of possible discrepancies; the dispute was shared with a third person and eventual consensus was reached through re-examination of the issue.
Risk of bias assessment
One of the risks of bias assessment was the fact that data extraction was conducted by two researchers. There was also the risk of dispute.
Outcomes of interest
LBW defined as an infant born at less than 2,500 gr. And hemoglobin level in the first and the second trimester of pregnancy is less than 11 g/dL and is less than 10.5 g/dL in the third trimester, this condition is considered anemia (3).
Statistical analysis
In present study, the extracted data included study design, type of the study, name of the author, year of publication, continent, number of participants, trimester and the outcome of pregnancy (LBW). The level of hemoglobin was classified in three groups of <14, 11-13, 11> g/dL, while <11 was considered maternal anemia. LBW, as an outcome of pregnancy, was analyzed when the infant weighed less than 2500 grams. The effect of hemoglobin was evaluated using relative risk index (RRI) in three different phases (first, second, and third trimester) of pregnancy based on the weight of the infant. I2 index and Q test were used to assess the heterogeneity. Considering the heterogeneity of the studies, random effects model was used to combine different results. STATA Ver 3.2 was used for data analysis and p<0.05 was considered significant for all analyses.
The minimum and maximum relative risks (RR) or odds ratio (OR) at upper and lower limits of 95% confidence intervals were extracted for each study. In studies, for which this index was not extracted (articles without the OR to use the formula), the index was calculated using OR=ad/bc formula and OR or RR logarithm were used to symmetrize the effect size. Error standard was calculated using the following formula: SE= 1n (upper 95% CI /lower 95% CI) /(2×1.96). If only upper and lower limit had been reported in studies, OR= ad/bc was used. Meta-regression was used to investigate the relationship between the year of study, effect size and causes of heterogeneity. Beggs Funnel Plot was used to examine the publication bias. Sensitivity analysis was used to assess the effect of each study on the overall result.
To analyze sensitivity, the effect of each article was excluded and the result was combined with other studies. This shows whether exclusion of this study has significant effect on the overall result or not. In the case of our study, the effect was not statistically significant.
Go to:
Results
Overall, 2454 relevant articles were identified by searching databases. Studies that did not yield any helpful and relevant information were excluded; 76% of them were cohort studies with an age range of less than 30 years old. 70% of articles were conducted in Asia. (Diagram 1, Table I)
Table I
Characteristics of studies that entered meta-analysis
Diagram 1
Entry procedures of studies into meta-analysis study
The relationship between maternal anemia in the first trimester of pregnancy and LBW
By type of study
As illustrated in Figure 1 and table II, there were 9 cohort studies in which the relationship between maternal anemia in the first trimester and LBW was investigated. After combining cohort studies using random effects model, a no significant relationship was observed between maternal anemia in the first trimester of pregnancy and LBW (relative risk, 1.20 [95% CI: 0.98 to 1.46]) (8, 18, 19, 22, 23, 24, 28,29,33). It seems that Levy’s study has estimated this relationship with a relatively high RR (8). There were 2 case-control studies about the relationship between maternal anemia and LBW indicating a significant relationship with a RR of 2.40 (95% Cl: 1.62-3.55) (11, 12).
Table II
The relationship between maternal anemia and LBW in the first trimester based on the continent and study type
Figure. 1
The relationship between maternal anemia and LBW in the first trimester based on the type of study
It seems that Hamalainen’s study has estimated this relationship with relatively high risk (12). And There were 1 cross-sectional studies about the relationship between maternal anemia and LBW indicating a no significant relationship with a RR of 0.72 (95% Cl: 0.52-0.99) (30). However, the total combination of cohort, case-control and cross-sectional studies using random effects model showed a significant relationship between maternal anemia and LBW in the first trimester with a RR of 1.26 (95% Cl: 1.03-1.55) (8, 10, 11, 18, 19, 22-24, 28-30, 33).
As illustrated in Figure 1, I2 index showed that the results of 9 cohort studies were heterogeneous and this heterogeneity was significant (I2 Index=69.8%, p=0.001). The heterogeneity of 2 case-control studies was 0.0% and this heterogeneity was not significant (p=0.481). Ov each article was measured individually and it was found that their weight is calculated when the data is combined. Among cohort studies, Levy’s study had the highest weight because of large sample size (8). Kidnato’s study had the highest weight among case-control studies (11).
By type of continent
Table II and Figure 2 shows the data analysis based on the continent; about 10 studies were conducted in Asia, 1 in Europe and 1 in Africa. RR, confidence interval, and P value were measured through random effects model, that combining Asian studies using a random effects model and this relationship was no significant (relative risk, 1.13 [95% CI: 0.93 to 1.37]). The data related to European studies (relative risk, 3.14[95% CI: 1.35 to 7.29]) and African studies 2.23 [95% CI: 1.44 to 3.46]) are shown in Table II and Figure 2 separately. Levy’s study in Asia had the highest weight (8).
Figure 2
The relationship between maternal anemia and LBW in the first trimester based on continent
By year of publication
According to Figure 3 and Table III, there is not significant relationship between publication year and logarithm size and the the regression equation was estimated to be y=41.52-0.02x. Since p-value was found to be 0.43 for the slope of the line.
Table III
The relationship between maternal anemia and LBW in the first trimester by continent based on year of publication and sample size
Figure 3
The relationship between maternal anemia and LBW in the first trimester based on the year of publication
By Begg’s funnel plot
Figure 4 was used to find the direction of publication bias, according to which the publication direction effect was significant (p=0.02).
Figure 4
Begg’s funnel plot for relationship of maternal anemia with low birth weight in the first trimester of pregnancy
By Sensitivity analysis
Sensitivity analysis showed that the elimination of a study can either change the results or does not affect the final direction at all and it shows the influence of studies in meta-analysis research. Sometimes, excluding a study can change the overall effect of meta-analysis significantly. In this study, as shown in the figure 5, excluding a study had no effect on the overall results.
Figure 5
Sensitivity analysis for relationship between maternal anemia with low birth weight in the first trimester of pregnancy