Predictors of Anemia Among Non Pregnant Married Women in a Peri-urban Coastal Slum of Karachi, Pakistan

Under-nutrition is a global public health threat, to which pregnant and lactating mothers are particularly susceptible. An estimated 468.4 million women (17%) of child bearing age have anemia, and this gure is as high as 50% of women of childbearing age in developing countries. Factors like substance abuse, nutritional habits, hygiene, reproductive health status and socio-cultural factors are associated with high prevalence of anemia among women. The aim of the study was to determine prevalence of anemia i.e. hemoglobin concentration of less than 12 gm/dl among non-pregnant married women of reproductive age (MWRA) and potential risk factors of anemia among these MWRA with a special focus on coastal slum. A analytical cross sectional study conducted at a coastal slum of Goth. We interviewed pregnant MWRA and their blood samples were assessed for anemia. Cox proportional hazard algorithm was used for determining the association of potential covariates with anemia for the analysis purpose.


Abstract Background
Under-nutrition is a global public health threat, to which pregnant and lactating mothers are particularly susceptible. An estimated 468.4 million women (17%) of child bearing age have anemia, and this gure is as high as 50% of women of childbearing age in developing countries. Factors like substance abuse, nutritional habits, hygiene, reproductive health status and socio-cultural factors are associated with high prevalence of anemia among women. The aim of the study was to determine prevalence of anemia i.e. hemoglobin concentration of less than 12 gm/dl among non-pregnant married women of reproductive age (MWRA) and potential risk factors of anemia among these MWRA with a special focus on coastal slum.

Methods
A analytical cross sectional study conducted at a coastal slum of Karachi, Rehri Goth. We interviewed N-554 non pregnant MWRA and their blood samples were assessed for anemia. Cox proportional hazard algorithm was used for determining the association of potential covariates with anemia for the analysis purpose.
Prevalence ratio of anemia among gutka consumers was 1.37 times (95%CI: 1.12 -1.68) higher than those who were not consuming it. Moreover, the prevalence ratio of anemia was 1.42 times higher in women who had not received formal education as compared to those who had received at least a primary education.(95% CI:1.13-1.81)

Conclusion
A high proportion of non pregnant MWRA of Rehri Goth were anemic. Women who were not literate were more likely to be anemic and were more likely to consume Gutka as a recreational substance.

Background
Anemia is a major public health hazard affecting millions in both the developed as well as the developing world. MWRA and children's are mostly affected in any stage of life, which mostly results in poor outcomes (1) The global estimates indicate that around 468.4 million (17%) non pregnant and 19% (16.2 million) of pregnant women between the ages of 18-49 years are anemic. (2) Sub Saharan African and South Asian regions are the most affected with high prevalence, of 47.5% and 35.7% respectively. (1,2) Similarly, the situation in the Eastern Mediterranean region is also alarming, which has the prevalence of around 32.4%. (1) The target population for this study were all MWRA (18-49 years). Around 12,000 married women of reproductive age are registered in the surveillance system. Our sampling frame was these 12,000 women. The basic line listing of registered women, maintained by local NGOs VPT in collaboration with Department of Pediatrics and Child Health, Aga Khan University (AKU), has been used to identify MWRA. This line listing is updated every two months and every married woman has been given a unique household identi cation number. The current system captures routine information on key vital events such as pregnancy, birth outcomes and deaths. For this study, women were randomly selected through computer generated simple random numbers.
Participant eligibility criteria MWRA were approached and screened regarding eligibility for participation, the inclusion and exclusion criteria provided in Table 1. For eligible participants, written informed consent was obtained in the local language.

Sample Size
OpenEpi software was used to calculate the sample size. Unpublished data from the antenatal services of VITAL Pakistan Trust suggests that around 50% of population consume gutka including about 30% of pregnant women. We assumed that MWRA in this community will have prevalence of anemia close to prevalence in Sindh province which is 62%. (12) Based on understanding through unpublished clinical data from VITAL Pakistan, we learnt that pregnant women who consumed gutka, has around 12% higher chances of hemoglobin level less than 12 gm/dl. Therefore, for sample size calculation, 12% prevalence difference of anemia between the population exposed to gutka compared to unexposed. The sample size of at least 510 women was estimated to nd the prevalence difference of 12% at p-value of signi cance < 0.05, power of 80%. Further, adjusting for expected overall refusal rate of 8.4%, which is based on our experience of 4% refusal with consent procedures and 4.4% refusals with blood specimens during antenatal clinic and missing data after consent procedure. Therefore, the in ated in sample size for this survey was '557 non pregnant MWRA'.

Informed consent procedure
For the eligible participants, written informed consent was obtained by the research team in a local language (mostly Urdu and where required, in the languages Sindhi and Pashto). The team members explained the details of the survey, including the purpose, specimen collection, and other related processes. If a participant was eligible and agreed to undergo the procedure as explained, the research team gave the consent form to the participant or decision maker, of if they were unable to read it, a team member read it word-by-word for them in Urdu or the local language. The participants were allowed to ask any questions related to the consent form and trial procedure. If the participant/decision maker required additional time to make more informed decisions, the team also allowed this opportunity and waited until the nal voluntary decision was made. If the participant voluntarily agreed to participate, the participant signed the consent form in the presence of a witness; either the form was duly signed, or a thumb impression was provided by participant and the witness. The ethics committee approved the use of a thumb impression by the participant and witness if they were unable to read or write. A copy of informed consent was provided to all participants and attached in the le with the study ID.
Data collection procedure Every household where any MWRA resides in the surveillance area of Rehri Goth was numbered with a unique ID. This line listing formed the sampling frame for the selection of the study sample i.e. sample of 554 households were randomly selected from this list. A community health worker with one senior research assistant reached the selected women for introduction of the study, eligibility, and consent. As the initial step, eligibility assessment was performed, and if found eligible, the women were explained about the research and written informed consent was taken. The interviews were conducted in local language followed by a collection of blood sample for testing for hemoglobin levels. Each interview lasted for approximately 30-40 minutes. All the Interviews and blood sample collection were carried out at the household with privacy.
The pilot study was conducted using the around 10% of the actual sample size (n=50), at the adjacent community of Ibrahim Haidri, a neighboring coastal village with identical socioeconomic status and cultural characteristics. The purpose of the pilot was to assess the feasibility and robustness of consent, tools, and study procedures. With the outcomes of the pilot phase, we re ned our study tool. The actual data collection was conducted between April 10 to May 10, 2018.

Data collection tool
A screening questionnaire was used to identify the eligibility criteria of the selected woman from the random list, i.e. MWRA 18-49 group with no exclusion criteria. If the selected woman did not meet any of inclusion criteria and/or had any of the exclusion criteria, she was considered ineligible. A close-ended questionnaire, especially designed to capture the key information of interest, was used in the interview to collect data. Questionnaire was comprised of important demographic information related to characteristics of the women, screening for eligibility, household related information, reproductive history, socio-economic status, personal history such as Gutka consumption, food consumption and frequency using last four week recall, and standard variable used for food insecurity. We have also used standard questions of food insecurity adopted from the Household Food Insecurity Access Scale for Measurement Food Access: Indicator Guide Version 3. (31)

Specimen collection
All specimens which were collected on daily basis, were transported to Koohi Goth Hospital Laboratory, Research and Training Center, following the proper procedure of transportation using carrier. The blood test for hemoglobin was run on the same day of sample collection and reports were generated and delivered to the research team on the next day. The standard operating procedure for blood specimen collection were adopted from 'WHO Guidelines on Drawing Blood: Best Practices in Phlebotomy'. (32) The report contained complete blood count indicators and a copy was provided to the family by the team.

Data Analysis
Data analysis was conducted using Stata version 12. Descriptive analysis was carried out for basic demographic covariates by generating frequencies and percentage. For inferential analysis, a single binary variable 'hemoglobin level 12 gm/dl or greater' was considered as normal and any value less than 12 gm/dl was de ned as anemia. Unadjusted prevalence ratio was calculated using univariate cox proportional hazard analysis with 95% CI. All variables which were found to be signi cant at univariate analysis at p-value less than 0.25, were selected to be analyzed in the multivariate model by stepwise regression. Variables with p-value less than 0.05 were found to be statistically signi cant; prevalence ratios were recorded at each model building and reported as adjusted prevalence ratio.
The main exposure variables in the analysis were Gutka consumption, women's education, seafood consumption, consumption of iron rich diet, type of toilet which the woman uses, regular fecal disposal at household level, parity, gravidity, total number of children under ve years of age, and other socioeconomic indicators. Further, using international analysis guideline of Household Food Insecurity Access Scale (HFIAS) for Measurement Food Access: indicator Guide Version 3, the food security condition of each household was calculated. (31) The HFIAS module yields information on food insecurity (access) at the household level. Four types of indicators were calculated to understand characteristics of and changes in household food insecurity (access) among the participants. The responses from the household food insecurity (access) were entered into an excel sheet and tabulation was done for these indicators and status of household food insecurity was nally measured i.e. whether the woman is residing in food secure or insecure household. The data was then incorporated with main dataset for analysis.

Quality assurance and data management
All lled questionnaires were passed through an astringent quality control mechanism, such as the questionnaire was lled again through a senior research staff to see difference between two of them and 100% quality checking was performed by senior research staff. Discrepancies were solved by revisiting the household, as and when required. Further, 20% of the data collected was rechecked at the eld by PI by household visit on the next day to check the data validity.

Results
Total 613 non pregnant MWRA (18-49 years) were approached from the random list to achieve the sample size of 557. Out of 613, 13 (2.1%) refused for eligibility assessment and 600 MWRAs were screened for eligibility. Non-pregnant MWRA with inclusion and no exclusion were 589, out of which 32 (5.4%) refused to give consent. Total MWRA consented and enrolled in the survey were 557. Figure 1 is showing the study ow. The mean age of the women was 31.3 years (SD 7.9). Women with age equal and greater than 36 years, had higher prevalence of anemia of 73.15% than others. The other age categories i.e. younger than 25 years and 25-35 had an anemic prevalence of 67.47% and 64.15% respectively. The mean hemoglobin level was 11.2 gm/dl (SD 1.6) and overall prevalence of the anemia in the sample was calculated to be 68.04% (95 % CI: 64.15 -71.92). Table 2 is representing anemia prevalence among the sample population.
Gutka consumption was common; 38.24% of the women used it regularly at an average intake of 2.8 times per day. Among these women, the prevalence of anemia was 83.57%. In the 61.75% of the sample who did not consume Gutka, the prevalence of anemia was lower at 58.43%. Additional analysis of the data revealed that only 32% of the participants mentioned that they have at least primary education or more. Among these participants, the prevalence of anemia was 53.07%. However, women with no formal education suffer a much higher prevalence i.e. 75.13%. Only 7% of of participants were found to be engaged in some kind of employment, and among them 62.50% were anemic. Prevalence of anemia also had some correlation with the wealth quantiles of the participants. 20.10% of these MWRA belonging to the poorest wealth quantile had an anemia prevalence of 71.43%. The prevalence of anemia among the rest of the wealth quantiles which were poor (19.92%), middle (20.10%), rich (19.92%) and richest (19.92%) households was 70.27%, 63.39%, 68.47%, and 66.67% respectively.
Approximately 59.42% of the households where these MWRA reside were food insecure (mild to severe) and prevalence of anemia among these food insecure households was 68.28%. Seafood intake was high among the participants; 77.01% of the participants consumed seafood ( sh and/or prawns) at least once a month. The prevalence of anemia among participants was 68.99%. Those who did not eat seafood at all had a slightly lower prevalence of anemia at 64.84%. Furthermore, about 50.80% of participants only consumed seafood. With no other source of meat (red meat or poultry), 73.85% of them were anemic, compared to 62.04% prevalence in participants who consume all types of meat in a month. A very large number of women i.e. 65.70% mentioned that they did not consume red meat at all and the prevalence of anemia among them was found to be 72.95%. In comparison, 34.29% of participants who consumed meat at least once in a month had a 58.63% prevalence of anemia. Table 3 presents the descriptive and prevalence.
Further analysis using cox proportion revealed Gutka and education status of non-pregnant MWRA as signi cant predictors of anemia among the study participants. The adjusted prevalence ratio of anemia among those women who consume Gutka was 1.37 times more, as compared to the women who did not consume Gutka at all (95 % CI: 1.12 -1.68). Additionally, the adjusted prevalence of anemia among women who had no formal education was 1.42 times higher in comparison to those who had at least primary education (95 % CI: 1.13 -1.81). The rest of the variables which were found signi cant at univariate analysis became insigni cant at multivariate analysis in the model.

Discussion
The prevalence of anemia in non-pregnant MWRAs in Rehri Goth was found to be 68.04%. The estimated prevalence is greater than the overall prevalence of Sindh province in Pakistan. (12) A study from Indian coastal belt identi ed that around 89% of the MWRA were anemic. (33) This indicates that health status of MWRA in the costal slums is at dismal. (14) The high prevalence in married women may be associated with many factors, which may be associated with household characteristics. (34) Maternal education also plays a pivotal role in improving the health status of women. (35,36) An educated woman from low income settings can take better and informed decision for her health related issues, (35,36) which is something that is lacking in Rehri Goth. Awareness about good nutrition and knowledge about health risks of other poor behaviors are improved if women are educated. (37) This may reduce the chances of anemia among these women. (37) In this study, it is found that anemia prevalence is high among MWRAs with no formal education and most of them did not have formal education.
Linking low education status to poor behaviors which are prevalent in these coastal slums is crucial; the use of gutka being one of the emerging health risks. (38) South Asians have the highest prevalence of gutka consumption with increasing trends among women, (39) ignoring the health risk associated with the use of it. (40) The use of gutka among MWRA in slums is strongly associated with wealth, knowledge, and education status. (41) This study suggests that use of gutka in Rehri goth is strong predictor of anemia.
This study did not nd any signi cant association of anemia with other factors, but other studies have shown signi cant correlation of anemia with food insecurity, gravidity, birth interval, use of contraception and wealth quantiles. (26,42) However, this study successfully highlighted the magnitude of maternal anemia, which is also very high in this coastal slum. Further, we used pre-validated variables in the questionnaire to maintain rigor of the study, including food frequency questionnaire, variables for measuring household food insecurity, wealth quauntiles, and other potential associated factors. This assured the validity and generalizability of the study.
One of the study limitation was that a cross-sectional study was used to investigate the prevalence and risk factors of anemia among non pregnant women in urban slum; such a study design does not give clear indication of the sequence of events hence, we cannot draw causal inferences from the ndings. As most of the responses were based on maternal recall, there may be chance of recall bias in the study. To counter these problems, research team was speci cally trained in probing against each variable.

Conclusion And Recommendations
Anemia is one of the major public health concerns for health care providers and also for policy makers in low and middle-income countries. This study has determined the association between Gutka consumption and anemia. To gain a better understanding and insight into important determinants of anemia, the community and individual level factors must be studied. Hence, awareness and proper family planning guidance programs must be proposed regarding the health bene ts of small family size in such urban slum areas of Pakistan. Additionally, education of women is necessary to take care of their health and its important in nurturing of the family. If a woman is educated, she is less likely to follow the poor lifestyle and negative health behaviors like Gutka consumption. We strongly recommend raising awareness about a balanced diet and ill-health effects of Gutka in these communities. At the policy level, such poor communities should be provided with safety nets for better nutrition to prevent malnutrition, not only in the current generation but also in future generations.

HFIAS
Household Food Insecurity Access Scale MWRA Married women of reproduction age

Declarations Competing interests
The authors declare that they have no competing interests.

Funding
This study was not funded by any external sponser. However, author received support by VITAL Pakistan Trust through existing maternal, neonatal and child health program to conduct laboratory assessment.
Ethics and consent to participate Ethical approval has been taken from 'Ethical Review Committee' at Aga Khan University (Refrence number: 5090-CHS-ERC-17). Written informed consent was administered in a local language to eligible participants and explained about the research. In cases where the MWRA was not educated, consent was documented by a thumbprint in the presence of literate witness.

Availability of the data and material
Deidenti ed dataset are available from the corresponding author on reasonable request.

Consent publication
Not applicable Inclusion criteria Exclusion criteria · Married women of eligible age criteria who were not pregnant or had delivered at least 6 months back from the day of data collection · Women who were permanent resident of Rehri Goth since last one year at least.

·
Women who gave a written consent to participate in the study and for specimen collection.

·
Any known diagnosed case of blood disorder