LBW is considered as a problem great enough to trigger public health action when its incidence exceeds 15% 
. Consequently, with the incidence of 16.5%, LBW requires focused intervention in the locality.
Due to the community-based nature of the study it was only possible to weigh newborns within 72 hours of birth. As birthweight is known to decline by 5–7% in the first three days of life 
, the study might have underestimated the birthweight with equivalent fraction.
In Ethiopia very few studies attempted to determine the prevalence of LBW. Most were conducted in major referral hospitals. The reported prevalence figures ranged from 8.6 to 15.4% 
. The current study witnessed relatively higher prevalence probably due to the reason that it was conducted in rural area where the burden of the problem is expected to be high.
The study did not witness significant association between prenatal zinc status and infants' birthweight. The finding is in confirmation of the conclusion a meta-analysis that maternal zinc supplementation does not enhance birthweight 
. However several observational studies conducted in the developed and developing world reported positive association between maternal zinc status and birthweight 
. Nevertheless, nearly all of the studies reported the positive association merely based on correlation or t-test analyses. Among the studies, only few 
applied multivariate analyses to control potential confounders. Accordingly the overwhelmingly positive association reported in the literatures can be due to the effect of confounding factors as most of the studies did not adequately control potential extraneous variables.
Observational studies on the relationship between maternal vitamin A and birthweight in apparently health subjects concluded divergently. A study in Israel reported that cord retinol along with gestational age explained more than a quarter of the variability of birthweight 
. Studies in UK 
and Guatemala 
found a significant positive correlation and a study in India linked maternal night-blindness to LBW. However in the current study, parallel to studies conducted in Bangladesh 
and USA 
, no significant association was witnessed. Systematic review of the available observational studies may help to resolve the existing conflicting findings.
In the current study the concurrent presence of VAD and ZD during pregnancy showed marginally insignificant effect on LBW (95% CI of 1.00–3.11). Nevertheless, it is important to interpret the finding in consideration of the fact that the sample size of the study was only calculated for the evaluation of the independent effect of the deficiencies on birthweight. Hence, in the measurement of the joint effects of the deficiencies on birthweight, the power of the study might have been compromised. Accordingly, in this area, further studies with optimal sample size are evidently required.
The fact that maternal zinc and retinol concentrations were measured only once during the entire pregnancy can potentially over or under estimate their association with birthweight as the exposure status to the deficiencies might not be fixed. Nevertheless, among women who were at their second gestational trimester during the baseline survey, follow-up survey conducted during their third trimester showed no significant change in their dietary diversity, food frequency and consumption of animal source foods. This might be taken as supporting evidence that their exposure status had not been assorted remarkably.
The linear regression model explained merely 20% of the variability in birthweight. This might have happened as some of the key predictors of low birthweight like weight gain during pregnancy, pre-pregnancy weight and malaria during pregnancy were missed from the model. A study showed that in the developing world the aforementioned factors can roughly explain 40% of the variability of birthweight 
Further, both in the linear and log binomial models, gestational duration has not been included as a covariate despite the fact that it is an important predictor of birthweight. This was due to the obvious difficulty of measuring gestational age in rural areas of developing countries where access to ultrasound is limited and estimation based the date of last menstrual period lacks accuracy. The exclusion of the variable from the regression analyses might have compromised the comprehensiveness of the models.
Several studies witnessed the positive contribution of superior household wealth status in reducing the burden of LBW 
. However, it was not the case in the current study. The unanticipated finding might have to do with the fact that wealth was quantified using a relative scale as measurement based on actual household income was not feasible. The relative scale might not be powerful when the wealth status of the population is reasonably homogenous.
In the current study male babies weigh more than females by about 100 gm and the risk of LBW was 1.8 times raised in females. Previous studies in Ethiopia and abroad concluded likewise 
. A systematic review concluded that on average, males weigh 150 g heavier than females 
Both in the log-binomial and linear regression models maternal MUAC was a strong correlate of birthweight. Especially in the linear model, removing the variable from the equation reduces the r-squared value from 21% to 12%. Even the study might have underestimated the association as pregnant women are commonly enrolled into supplementary and therapeutic feeding programs based on their MUAC. Studies in India 
, Brazil 
and Bangladesh 
also reported parallel findings.
Maternal height was also a strong predictor of birthweight. This is consistent to the understanding that in the developing world approximately 12% of low birthweight can be explained by maternal short stature 
. Studies in India 
and UK 
concluded likewise. Though the exact mechanism how maternal height affects birthweight is not clear, it might be due to the mixed manifestation of genetic and intergenerational effect of malnutrition 
Unlike several studies that witnessed the beneficial effect of ANC in reducing the burden of LBW 
, the current study did not observe parallel findings. This might have to do with the inferior quality of ANC in the locality where only 42% and 76% of the ANC attendants received iron supplementation and nutrition education, respectively.
In the linear regression model elevated CRP during pregnancy was an important predictor of birthweight. The finding is in confirmation of the result of a systematic review that in the developing world approximately 4% of the burden of LBW is attributed to general illnesses during pregnancy and an extra 10% can be linked to malaria