Association Between Adverse Pregnancy Outcomes and Preceding Risk Factors: an Analytical Cross-Sectional Study

Background: The health of women during the preconception phase though critical, is a comparatively ignored part in her life cycle. The presence of health risks is judged as hazardous to the wellbeing of women and their forthcoming progeny. The study aimed to estimate the prevalence of various pregnancy outcomes and assess the association between various risk factors and adverse outcomes. Methods: It was a population-based cross-sectional analytical study. The study was carried out in Nashik District, having a large mountainous area. It included two randomly selected blocks, one tribal and one nontribal, in which interventions were planned in the later stage. For comparison, two adjacent blocks, one tribal and one nontribal, were also included. All women who had a pregnancy outcome in the preceding 12 months (01 April 2017 to 31 March 2018) were interviewed. Trained Accredited Social Health Activists under the direct supervision of Auxiliary Nurse Midwives and Medical Ocers conducted the survey. Multivariate analysis was carried out to nd the adjusted risk ratio of having a particular adverse outcome because of the specied potential risk factors. Results: A total of 9,307 women participated in the study. The prevalence of adverse pregnancy outcomes was abortion 4.1%; stillbirth 1.6%; preterm birth 4.1%; LBW 11.2%; congenital physical defect 2.6%. Prevalence of consanguineous marriage, heavy work during the last six months of pregnancy, pre-existing illness, tobacco consumption, direct exposure to pesticides and domestic violence during pregnancy was 17.6%; 16.9%; 2.2%; 5.6%; 2.3%; and 0.8%, respectively. Risk factors that were signicantly associated with abortion include existing illness at the time of conception and performing heavy work in the last six months of pregnancy. Consanguinity, tobacco consumption during pregnancy and pre-existing illness were identied as risk factors for stillbirth. Signicant risk factors of LBW were heavy work in the last six months of pregnancy, pre-existing illness and residence in a tribal area. Conclusion: The survey showed that risk factors differentially affect outcomes of pregnancy. Preconception and antenatal care should include counselling about consanguineous marriages, identifying and managing a pre-existing illness, avoiding tobacco consumption in the prenatal second type included the following potential risk factors; consanguinity, heavy work in the last six months of pregnancy, tobacco consumption, alcohol consumption, exposure to a pesticide, domestic violence, existing illness. The outcome variables were abortion, stillbirth, preterm birth, low birth weight, congenital physical defect, and neonatal death. Standard denitions were used. Information about all these variables was based on responses given by the participants.


Study design
It was a population-based study conducted before the initiation of interventions. It was an analytical cross-sectional study.

Study setting
It was carried out in the Nashik district. It included two intervention blocks in which intervention was planned, one tribal (Peint) and one nontribal (Sinnar). It also included two adjacent comparison blocks, one tribal (Trimbakeshwar) and one nontribal (Niphad). Nashik district is located high on the Deccan Plateau; it is surrounded by the Sahyadri range of mountains. The selected tribal blocks are hilly, di cult to reach and are having high annual rainfall. The population of these four blocks as per the last census (2011) is 1,19,838 for Peint; 3,46,390 for Sinnar; 1,68,423 for Trimbakeshwar and 4,93,251 for Niphad [4]. The blockwise map of the district is given in Fig. 1.

Study period
The study was carried out in 2018-19. Actual data was collected from May to July 2018.

Tools and Data collection
The consultative process was adapted for designing the interview schedule. This predesigned tool was validated by experts, translated in the local language (Marathi) and pre-tested before it was used in the eld. The questionnaire included demographic information, details of pregnancy, its outcome and risk factors associated with adverse pregnancy outcome. Rigorous training was conducted on these tools for Accredited Social Health Activists (ASHA) who collected the data and Auxiliary Nurse Midwife (ANM) who supervised the activity. If a woman was not present during the home visit, she was visited later within the next month.

Participants
All women in the reproductive age group in these blocks were contacted. Women who had pregnancy outcome in the last 12 months, i.e., 01 April 2017 to 31 March 2018, were included in the study provided they met the inclusion criteria of being resident of the area (residing or intends to reside for more than six months). Those women who were unable to understand Marathi, Hindi or English or unable to respond due to psychotic illness were excluded.

Variables
The variables were of three types. The rst type included the following demographic variables; age, family type, education, occupation, and place of usual residence. The second type included the following potential risk factors; consanguinity, heavy work in the last six months of pregnancy, tobacco consumption, alcohol consumption, exposure to a pesticide, domestic violence, existing illness. The outcome variables were abortion, stillbirth, preterm birth, low birth weight, congenital physical defect, and neonatal death. Standard de nitions were used. Information about all these variables was based on responses given by the participants.

Sample size estimation
As per District Level, Household Survey 4, stillbirth and abortion rate for the rural area in Nashik District was 1.3% and 6.1%, respectively [ 13 ]. For estimation of the stillbirth rate with 95% con dence and an accepted difference of 0.26%, 7,600 pregnant women were required to be surveyed. In the year 2016-17 and 2017-18, each year, more than 22,000 pregnancies were registered in these four blocks [ 14 ]. Hence authors conducted a survey covering entire rural areas of these four blocks.

Data Analysis
Statistical analysis was conducted using a Statistical Package for Social Sciences version 25.0. Chi-square test was applied wherever applicable. We rst carried out the univariate then multivariate analysis. The authors preferred calculation of the adjusted risk ratio (ARR) to adjusted odds ratio because of relating a particular adverse outcome to the presence of the known potential risk factors prior to the outcome. The level of signi cance was decided at P<0.01.

Results
In the study area, 9,307 (tribal=3298, non-tribal 6009; study=4766, control=4541) women reported pregnancy outcome in the aforesaid period. The nonresponse rate was less than 10% except for birth weight and gestation at the time of delivery; it was about 15% and 37.26%, respectively.
Their socio-demographic details are shown in Table 1. The overall mean age was 23.91+3.23 years. The age group distribution among tribal and nontribal women was different. The age of tribal women ranged from 17-45 years with a mean of 23.72 (SD ± 3.21), and for the nontribal area, age ranged from 17-43 years with a mean of 24.02 (SD ± 3.23). Many women were married before the legal age. Teenage pregnancy was more in the tribal block. Most of the women (70.03%) from both areas were living in a joint family. Overall, 26.02% of women had completed their Secondary School Certi cate (SSC); 13.32% were illiterate, and only 2.18% were postgraduate or having a professional degree. Educational status was signi cantly different in both study and control group as well tribal and nontribal blocks; however, the educational status of women from tribal blocks was less compared to nontribal, and the difference was highly signi cant (p<0.001). Almost 80.0% of tribal women were working, whereas 50% of nontribal women were homemakers (p<0.001). The majority of tribal women were involved in farming. Very poor educational status in Peint block and better in Niphad block and similarly majority women working from Peint block and not working in Niphad block, in uenced the block performance.
Details of adverse outcomes are given in Table 2. Overall, abortion was reported by 4.28% of women. We did not differentiate whether abortion was spontaneous or induced. The difference was observed across all the blocks. It was least in Peint Block and maximum in Niphad Block. Stillbirth was reported by 1.70% of women; preterm birth by 4.32%; low birth weight by 15.16%; congenital physical defect by 2.84%, and neonatal death by only 0.59%. In tribal blocks, three maternal deaths were reported, and none by nontribal block. The exact duration of gestation in weeks could not be told by 37.30% of women.
The association of the potential risk factors during pregnancy and with adverse pregnancy outcomes are given in Table 3 and Table 4, respectively. The distribution of risk factors among women in study and control areas as well as among women from tribal and nontribal areas is given in Table 3. There were differences between the study and control area to some extent, but the distributions of all risk factors between tribal and nontribal areas were profoundly different. Excepting existing illness, all risk factors were more among women from tribal areas.
We have calculated the adjusted risk ratio. Abortions were reported by 5.83% of women performing heavy work in the last six months of pregnancy as against 3.74% who were not performing it. Only 3.82% of women who did not have any pre-existing illness experienced abortion against 15.20% who had an illness.
Both heavy work in the last six months and pre-existing illness were identi ed as signi cant risk factors for abortion. Consanguineous marriage and consumption of tobacco or alcohol were not associated with abortion. More abortions were reported from the nontribal area as compared to tribal (5.08% vs 2.33%).
Among 3.55% of tobacco consumers, stillbirth was reported compared to 1.56% among non-consumers. In women having a pre-existing illness, 6.36% had stillbirth compared to 1.59% among those who did not have any such illness. Other factors were not associated with stillbirth.
The proportion of preterm birth was 2.73% among tribal and 4.76% among nontribal women. It was 4.94% among women performing heavy work in the last six months of pregnancy, compared to 4.03% among those who did not perform heavy work.
Low birth weight (LBW) was reported by 17.03% of mothers performing heavy work in the last six months of pregnancy, compared to 12.09% who did not perform heavy work. It was also reported by 23.33% of women having pre-existing illness as against 13.03% who did not have any such illness. LBW was reported by 17.57% tribal and 11.00% nontribal women.
The congenital physical defect was observed among 4.86% of off-springs from consanguineous marriages and 2.15% of non-consanguineous marriages. The proportion of congenital physical defect was higher (4.43% vs 2.28%) among mothers having a history of performing heavy work in the last six months of pregnancy. It was observed that among women consuming tobacco, 5.21% of babies were having congenital physical defect, whereas the proportion was 2.62% among non-consumers.
Early neonatal death was reported by 0.8% tribal and 0.5% nontribal women. None of the potential risk factors mentioned above was found to be associated with early neonatal death.

Discussion
Adverse pregnancy outcomes can be in uenced by various factors that are present prior to or emerged during pregnancy. The present study was a survey carried out for measuring the prevalence of pregnancy outcomes and its association with potential risk factors.
The socio-demographic pro le of women is certainly different for women residing in the tribal and nontribal area. This is important to note because the outcomes are also in uenced by these factors. Although rarely practised in western countries, consanguineous marriages are seen throughout India. As shown by the India Human Development Survey (IHDS, 2004(IHDS, -2005 data, they were predominant in the Southern States ranging from 28% to 38% [ 15 ]. Prevalence in Maharashtra was also high (28%). In the present study, the prevalence of consanguineous marriage was 17.6% which is lower than the reported studies from rural South India, ranging from 20.3 to 36% [ 16 , 17 , 18 ]. This lower prevalence in the present study may be due to a declining trend consequent to education and modernisation.
In the present study, only 5.6% of women reported tobacco consumption during pregnancy which is less than other studies [ 19 , 20 , 21 ]. Amongst all forms of tobacco, women from India indulge more in applying Mishri (roasted tobacco) on gums and teeth, which was also observed in this study. Alcohol consumption in Indian women is less compared to women from western countries, which decreases further in pregnancy [ 22 , 23 ]. Alcohol consumption was very less in the study, which was lower than the consumption reported by Bellad  Farming is a major occupation in rural and tribal parts of India, and various pesticides are commonly used to protect the crops. Pesticide spraying on grapes is a highly prevalent seasonal work in the area. Exposure to various pesticides during pregnancy has been identi ed as a risk factor for adverse pregnancy outcomes; however, results are not consistent [ 24 , 25 , 26 , 27 ]. Association was dependent on various factors like the type of pesticide, the period of gestation, duration and amount of exposure. In this study, only 2.3% of women reported direct exposure to pesticides during pregnancy. The reason may be the noninvolvement of women in the actual spraying of pesticides. This exposure to pesticides was not found associated with any of the adverse pregnancy outcomes.
Domestic violence during pregnancy was reported by only 0.8% of women, which was lower than previous reports from different states of India that ranged from 7.1% -18% [ 28 , 29 , 30 , 31 ]. Considering Indian culture, women do not divulge such information, so that these gures may be just the tip of the iceberg. The pre-existing illness like endocrine abnormalities, heart disease, liver disease, etc., in the pregnant woman, can have adverse pregnancy outcomes over and above risk to maternal health [ 32 , 33 , 34 ]. In the present study, 2.2% of women had reported pre-existing illness; however, there is a dearth of information about the prevalence of pre-existing illness in pregnant women in the community.
When the distribution of these potential risk factors was compared in tribal versus nontribal area, consanguineous marriage, heavy work in the last six months of pregnancy, tobacco and alcohol consumption, exposure to pesticides, and domestic violence were signi cantly more in the tribal area, which maybe because of the socio-cultural differences in these areas. Some studies have compared the socio-demographic pro le of women from the tribal and nontribal area; however, a comparison of the distribution of such potential risk factors in these two groups of women are extremely rare. Alcohol consumption was higher in tribal women, which is similar to the observation of Mohan D et al. [ 35 ] who categorised tribal women in the high-risk group for alcohol consumption.
Only pre-existing illness at the time of conception was more in nontribal area than tribal which may be because of better availability and accessibility of health care services in nontribal area than tribal which has led to more detection of existing illnesses.

Abortion
In the present study, 4.1% of women reported abortion which is consistent with estimated 47·0/1000 women abortions in India in 2015 [ 36 ]. Performing hard manual work, including lifting heavy weights during early pregnancy, is associated with abortion which is once again con rmed in this study. In the present study, existing maternal illness had three times the risk of abortion, which was also identi ed as a risk factor for abortion in other studies [ 37 , 38 , 39 ]. More abortions occurred in the nontribal area compared with tribal (5.1% vs 2.3%), and the difference was statistically signi cant. Residence in the tribal area has a protective factor. This may be due to more availability of various means of abortion like access to abortifacient medicines, health facilities, and affordability to avail these facilities. However, this could not be con rmed, as we did not differentiate between spontaneous and induced abortion.  49 , 50 ]. The meta-analysis by Marufu TC et al. 2015 reported maternal smoking during pregnancy increased the risk of stillbirth by 47% and con rmed a strong dose-response effect [ 51 ]. Many studies have identi ed alcohol consumption during pregnancy as a risk factor for stillbirth, which was not seen in this study [ 52 , 53 ]. It may be because of a small number of women consuming alcohol in the present study. Existing maternal illnesses like thyroid dysfunction or diabetes mellitus are associated with stillbirth, which is similar to the present study [37,42,54 ]. Domestic violence was not associated with stillbirth, which is not consistent with previous studies [28,55].

Preterm
In the present study, reported preterm birth was 4.1% which was lower than the range of 9% to 18% reported by various Indian studies [ 55 , 56 , 57 ], and the global estimate of 10.6% [ 58 ]. One of the reasons may be that these studies are either past studies or not from a progressive state. The reporting of the preterm birth rate was less than expected, which may be because of the recall bias and inability of mothers in reporting the exact pregnancy duration. An overall high proportion of women (37.26%) were unable to quantify the period of gestation in weeks, and particularly among women from tribal areas, the proportion was 46.63%. Less preterm births were reported in tribal women compared to nontribal; the difference was statistically signi cant. Although the non-response rate was more among tribal women, the explicit reasons for the lower prevalence of preterm babies could not be elicited.
Consanguineous marriage has been found associated with preterm births in previous studies; however, it was not observed in the present study [ 59 , 60 ]. In the present study, heavy work during the last six months of pregnancy was identi ed as a risk factor for preterm birth. Knudsen  the present study, tobacco consumption was not associated with preterm birth. The role of alcohol consumption in preterm birth is controversial, and no association was found in this study [43,65 ]. Few studies have identi ed exposure to pesticides as a risk factor for preterm birth; however, ndings are not consistent [24, 66 , 67 ]. The present study did not nd any association of exposure to pesticides with preterm birth. A meta-analysis by Shah PS et al. 2010 reported a 46% risk of preterm birth in women exposed to domestic violence during pregnancy; however, it was not observed in the present study [ 68 ]. Existing maternal illnesses like diabetes mellitus or liver disorder are found to be associated with preterm birth, but no such association was found in the present study [32,39,69 ].

Low Birth Weight
The study reported 11.2% LBW, which was lower than the global estimate of 14.6% [ 70 ], and the range of 17% to 36.8% reported from rural or tribal settings of India [ 71 , 72 , 73 ]. Performing heavy work during the last six months of pregnancy was associated with LBW, which was similar to the observation made by Kumar M et al. [71]. Use of any form of tobacco, including smokeless tobacco, was associated with harmful effects in the form of LBW; however, this study did not nd such association [63,74 ]. Similarly, in the present study, no association of maternal alcohol consumption during pregnancy was found with LBW. Like premature birth, exposure to various pesticides during pregnancy has been identi ed as a risk factor for LBW; however, results are not consistent [24,25]. This study also did not nd such an association. The association between pesticides and LBW depends on various factors like the type of pesticide, the period of gestation, duration and amount of exposure.
Gebremedhin M et al., in their study, had observed ve times the risk of LBW in women having chronic medical illness during pregnancy [ 75 ]. In the present study, such a risk was 1.8 times high. Domestic violence was found to be associated with LBW in some studies; however, it is not observed in this study [70,76]. Residence in the tribal area was identi ed as a risk factor for LBW, which was similar to the nding of Niswade A et al. [40]. This may be because of various socio-demographic and environmental factors like maternal age, education, nutrition, Ante Natal Care (ANC) visits, availability and accessibility of health care facility etc.

Congenital Physical Defect
Congenital physical defects were present in 2.6% of babies, which is slightly more than the estimated national pooled prevalence of 184.48 per 10,000 births reported by Bhide P et al. 2018 in his meta-analysis [ 76 ]. In the present study, the perception of women about congenital physical defects pertained mostly to physical defects. Actual assessment of genetic or metabolic physical defects is di cult to assess in the eld. Similar to previous studies, the present study found consanguineous marriage as a risk factor for a congenital physical defect in the baby [47, 77 , 78 ]. The proportion of congenital physical birth defects may be reduced by creating awareness about the effects of consanguineous marriages. This intervention is not attempted and probably may require minimal resources. Moreover, it will have collateral bene ts on biochemical and functional congenital anomalies.
It was also found to be associated with heavy work during the last six months of pregnancy. Tobacco consumption was identi ed as a risk factor for a congenital physical defect. A meta-analysis by Little J et al. 2004 has reported an association of maternal smoking with congenital oro-facial clefts [ 79 ], whereas Hackshaw A et al., in his systematic review, has found a signi cant association with many other birth defects like a heart defect, musculoskeletal defect, oro-facial clefts etc.
[ 80 ]. Alcohol consumption during pregnancy was identi ed as a risk factor for a congenital physical defect in off-springs [ 81 , 82 ]; however, it was not observed in this study. Existing illness in pregnant women has been identi ed as a risk factor for congenital physical defects in a few studies but was not seen in this study [77,83 ].

Early Neonatal Death
Early neonatal death was 0.59% which was lower than the estimated early neonatal death rate of 20 per 1000 live births in India for the year 2017 [ 84 ]. This may be due to lower rates of preterm deliveries and low birth weight babies and a high rate of institutional deliveries in the study area. Ahmed et al. reported a 2.3 times high risk of neonatal mortality among mothers who experienced violence during pregnancy [28]. No such association was found in the present study.

Strengths And Limitations
It was a large study involving more than nine thousand women from four blocks. It included various adverse pregnancy outcomes as well as various risk factors. It was conducted with the help of existing health care workers. However, the overall prevalence of socio-demographic characteristics and risk factors considered in the present study may not be a true re ection of the community in general because 35% of the women in the study were from tribal area. It also implies that it is pertinent for states having a high tribal population. Although the non-response rate was more among tribal women, the explicit reasons for the lower prevalence of preterm babies could not be elicited. Details regarding pre-existing illness were not studied. The details of abortion were not collected.

Conclusion
The present study con rms a higher rate of consanguineous marriage, heavy work during the last six months of pregnancy, tobacco and alcohol consumption among tribal women. The existing illnesses were more common in the nontribal area. Of the studied potential risk factors, performing heavy work during the last six months of pregnancy was associated with abortion, preterm birth, LBW and congenital physical defect. The pre-existing maternal illness was associated with abortion, stillbirth and LBW. Thus, there is clear evidence that pre-existing illness is associated with adverse pregnancy outcomes, but unfortunately, there is a dearth of information about the prevalence of existing illness in the preconception period in the community. Consanguineous marriage and tobacco consumption during pregnancy were associated with stillbirth and congenital physical defect.
Mass awareness regarding the adverse effects of consanguineous marriage should be done to avoid such marriages in future. This study high lights the need for physical examination and necessary investigations of women in the preconception period. Advocacy of prevention of tobacco consumption in prenatal and natal period is very important and should be included in preconception care. Advice regarding adequate rest and not performing heavy work during pregnancy should be given to the pregnant women by health care provider during the ANC visits. Data availability data and material Data used in the analysis are available from the corresponding author on reasonable request.
Competing interests Authors declared that there were no competing interests.
Funding The study was funded by United Nations Children Funds (UNICEF) through the Government of Maharashtra.
Authors' contributions PPD, SHP, JSG, AVP, PDP, AVD, KKB, MVK and ANS contributed to the overall design and methodology of assessing the PCC programme in the Nashik District. PPD, SHP, APC, MVK, and PDP trained the ASHAs for data collection and monitored data collection. SHP and PPD performed data analysis; SHP wrote the manuscript's initial draft and was nalized by PPD. KKB revised the manuscript extensively. All authors approved the nal version for publication.