Socio-demographic factors associated with anaemia among non-pregnant and non-lactating women from low-income families in two selected districts of Madhya Pradesh state of India: a random forest analysis

INTRODUCTION

Anaemia is one of the most common and widespread disorders affecting around 800 million children and women in the world. Iron deficiency makes a large contribution to anaemia, global efforts to reduce the anaemia burden have mostly been directed towards increasing intake of iron through supplementation, food fortification, and diversification of the diet. It affects 1.62 billion people globally and South Asian countries, including India, have the highest prevalence of anaemia in the world. India contributes to this burden, with approximately 53% of women in reproductive age (WRA) who are anaemic.¹ According to the National family health survey-4, 52.5% of WRA in the state of Madhya Pradesh (MP) are anaemic; which is further observed to be 60% among rural women and 74% among tribal women with serious negative consequences and harmful effects on the physical and mental health of a person. Iron deficiency anaemia is the most common cause of anaemia globally, though other factors also contribute.² Anaemia can have adverse consequences and can be particularly detrimental to WRA who become pregnant, continuing the cycle of malnutrition to the next generation. There is also an increased perinatal risk of impaired foetal growth, risk of Low Birth Weight (LBW), and increased risk of infant mortality and maternal mortality among women with anaemia.³

Census 2011 UNFPA & NFHS-4, Government of India.^8,9

One of the significant steps to decrease the prevalence and incidence of anaemia initiated by the Government of India is the “National Nutritional Anaemia Control Program” and the more inclusive “National iron plus initiative program.” Under the former scheme, a life cycle approach was adopted to address anaemia.⁴ Other strategies are nutrition education and screening and treatment of severe anaemia.⁵ However, various studies have shown that compliance and program implementation, especially in remote rural areas, continue to remain a challenge.⁶ Since, inadequate dietary intake and lack of dietary diversification in food pattern are also a key contributor towards iron deficiency anaemia, especially among the lower socio-economic class. Thus, it is conceptualized that to mitigate the prevalence of iron deficiency anaemia (IDA), one of the most commonly accepted remedy is to improve the dietary intake of a population by ensuring that their daily food consumption consists of the adequate amount of iron. Fortification of the daily diet with iron and folic acid is one method which can address this public health issue with a broader reach, and provide a visible spatial impact on the prevalence of this medical condition. Fortification is the addition of micronutrients to a staple food to make it nutrient-dense. Some of the commonly fortified staples are rice, wheat, millet, and maize. These staples are consumed by the most vulnerable population groups who require these micronutrients the most.⁷

Nutrition International (formerly the micronutrient initiative) in collaboration with the Government of Madhya Pradesh is working towards decreasing the prevalence of anaemia. One of the approaches is to fortify staples such as wheat flour, but evidence is lacking on whether this is likely to have an impact on anaemia in the Indian context. Therefore, keeping this in mind, Nutrition International sought to assess haemoglobin levels, knowledge on anaemia, iron supplement consumption, general food habits, acceptability of fortified wheat flour and related environmental and socio-demographic factors among non-pregnant non lactating (NPNL) women of reproductive age from the lower socio-economic group living in the Vidisha and Raisen districts in Madhya Pradesh. According to the census of India, 2011, Vidisha and Raisen have a population density of 198 and 157 persons residing per sq. km respectively, with a population of 1.50 and 1.3 million respectively.⁸

According to the NFHS-4, 2015-16, in Vidisha, 44% NPNL women and 56% pregnant women were anaemic; while in Raisen, 51% of NPNL women and 54% of pregnant women were anaemic. Only 15% of mothers in Vidisha compared with 23% of mothers in Raisen reported to have consumed iron and folic acid tablets for 100 days or more. Between the districts, Raisen (61%) had a lower percentage of literate women compared with Vidisha (64%). In both the districts, 84-95% of households had electricity and an improved drinking water source. The household sanitation facilities were generally poor, in both districts, only 24% in Vidisha and 39% in Raisen had improved sanitation facility.

Objectives

The objectives of the study were to examine the association between different socio-demographic characteristics and the prevalence of anaemia among non-pregnant non-lactating women of reproductive age from low-income families in two selected districts of Madhya Pradesh state in India.

METHODS

We conducted a cross-sectional survey in two districts of Madhya Pradesh in India between January and March 2017. Data were collected from 4,170 non-pregnant and non-lactating women 15-45 years of age (NPNLW) from low-income homes (BPL) (2,082 in the Vidisha district and 2,088 in Raisen district); and from 217 decentralized millers, fair price shop owners and warehouse managers.

Sampling

NPNLW were selected using a two-stage sampling design. Villages were selected from the two districts using probability proportionate to size (PPS) method. In each selected village, segments of approximately 200 households were identified based on natural boundaries. One segment was selected at random using a computer-generated random number table. In the selected segment, all households were listed to prepare the sampling frame for eligible low-income households having at least one NPNLW. The required number of NPNLW were selected from these lists by systematic random sampling.In order to understand, the distribution mechanism of wheat and micronutrient pre-blend, mixing of pre-blend in wheat flour and knowledge and awareness on anaemia and benefits of providing Iron Folic Acid-fortified food, chakki operators, fair price shop owners, warehouse in-charges and district-level officials were interviewed. One each chakki operator/owner and the fair price shop owner was interviewed in every second village. At the district level, one staff each from the food and supply department was interviewed.

Data analysis

Individual-level datasets were analyzed using PASW Statistics 18, Release 18.0 software to analyze NPNLW characteristics and status indicators; and to conduct multiple binary logistic regressions to explore factors associated with adherence to iron folic acid supplements or presence of anaemia. The dependent variable was considered as the prevalence of any anaemia, which was coded as 0 if the women had any anaemia and 1 if she did not. The independent covariates included in the multiple binary logistic regression analyses were; socio-demographic characteristics like; place of residence, age, type of family, social status, literacy, type of house, sanitation facilities, source of drinking water, electricity and knowledge about anaemia. Apart from canvassing background information and knowledge levels on anaemia, haemoglobin concentration was estimated using capillary Hemocue. In addition, random forests (RF) analysis was conducted, using data from NPNLW, to identify independent variables importance and model accuracy using Python 3.6 software. Random forest is a recently developed machine learning technique that deals with classification and clustering of data non-parametrically. It is an ensemble method that combines several trees by taking the same number of bootstrap samples from the original data and growing a tree on each bootstrap sample. Tree implementations are very simple and user-friendly and require fewer techniques from the investigator. The individual trees in a random forest are not pruned and used for decision in classification or clustering. Random forest uses a randomly selected subset of predictors for splitting the root nodes into new daughter nodes for each split. From all trees grown in this process based on the bootstrap samples, we generate a forest. From the complete forest, the response variable for instance, is predicted as an average or majority vote of the predictions of all trees. RF can highly increase the prediction accuracy compared to an individual tree, as the ensemble reduces the variance. RF is one of the most effective machine learning models for predictive analytics. The model feature importance from sklearn random forest was used to calculate feature importance. Random forest uses gini importance or mean decrease in impurity (MDI) to calculate the importance of each feature. Gini importance is also known as the total decrease in node impurity. This is how much the model fit or accuracy decreases when you drop a variable. The larger the decrease, the more significant the variable is. Here, the mean decrease is a significant parameter for variable selection.

RESULTS

Haemoglobin concentration and knowledge of anaemia

The results revealed that both the districts have a high prevalence of anaemia ranging from 60.8% (95% CI: 58.6%, 62.9%) in Vidisha to 63.7% (95% CI: 61.6%, 65.7%) in Raisen district amongst NPNLW. The mean±standard deviation haemoglobin concentrations were 11.28±1.93 g/dl in Vidisha and 11.24±1.71 g/dl in Raisen (Figure 3).

In both the districts, majority of the respondents were moderately anaemic (31% in Vidisha and 34% in Raisen) while 24% and 26% were detected with ‘mild’ anaemia in Vidisha and Raisen respectively. Severe anaemia was observed only in <5% respondents. Less than 5% of respondents with anaemia reported consuming iron supplements, as a majority (95.7%) of the respondents tested to be anaemic in the study area were not taking any iron supplements. Majority of the population lived in rural areas (81% in Vidisha and 77% in Raisen), with the mean age being similar in both districts (31.14±9.88 in Vidisha and 30.82±9.97 in Raisen).

Raisen had a higher percentage of literacy (65%) as compared to Vidisha (54%). Majority of the respondents were observed to be living in a nuclear family and belonged to the other castes (non-ST/SC) category in both districts. We also observed, the 81% and 90% respondents in Vidisha and Raisen respectively had heard about anaemia. Other factors potentially affecting anaemia, like sanitation and drinking water, fared differently. Sanitation was better in Raisen (61%) than in Vidisha (49%), though in both the districts <5% of the respondents had “piped water into dwelling.” Majority of the respondents had “other sources of drinking water” (96% in Vidisha and 97% in Raisen).

Association between socio-demographic factors and anaemia

The findings of the multiple binary logistic regression indicated that those who were from the scheduled caste (SC) [AOR 2.162 (95% CI: 1.535-3.045, p=0.000) in Raisen] and other (non SC/ST) castes [AOR 1.951 (95% CI: 1.429-2.664, p=0.000) in Raisen] were less likely to be anaemic compared to those who were from scheduled tribe (ST), which was observed in Raisen district. In Vidisha district, those respondents consuming iron supplements were less likely to be anaemic [AOR 1.502 (95% CI: 2.007-2.239, p=0.046)]. The random forest analysis (Figure 2) revealed that the age and caste of the woman were important variables in the model.

Knowledge and awareness among fair price shop (FPS) owners

Out of the FPS owners interviewed in Vidisha and Raisen, almost all of them were aware of anaemia. The knowledge about the fortification of foods was poor (about 45% among decentralized millers and 58.5% among FPS owners in Vidisha), while in Raisen it was even lower at 20.4% among FPS owners and 17.2% among decentralized millers. More than 90% of the FPS owners and 80% of the decentralized millers in both districts stated their willingness to participate in a wheat flour fortification intervention.

DISCUSSION

We conducted a multiple binary logistic regression adjusting for specific variables to understand factors affecting anaemia. The results of our study should be interpreted in the context of socio-demographic factors. Our analysis showed that education and consumption of iron supplements influence predisposition towards anaemia. In addition, we also observed that age, religion, and type of household structure also affect an individuals’ likelihood of being anaemic. The prevalence of anaemia in Vidisha and Raisen districts is higher than the prevalence of anaemia in Madhya Pradesh and India.² Majority of the respondents were moderately anaemic as per the WHO categorization, similar to other studies.^10-21 The target group for NFHS-4 was both rural and urban population, while our study sampled only the rural population. This observation is in line with other studies, which also report rural population to have a higher chance of being anaemic compared to the urban population. A multitude of factors such as poverty, access to food and health services, lower literacy levels, and poor hygiene and sanitation could be reasons for higher anaemia rates.^6,10,11,13 First, illiteracy as a predictor of anaemia, corroborates with other studies.^11-15 Results from the referenced studies align with our study, confirming that education is an influential factor in the prevention of anaemia. The studies recommend improving maternal knowledge to support sustained positive changes in maternal health and nutrition. Empowering women with knowledge on nutrition will guide them to make healthier choices, which in turn has a bearing on the health and nutrition of the entire family.¹⁵ Second, consumption of iron supplements is inversely related to anemia. Study results revealed that those not consuming iron supplements had a higher likelihood of being anaemic, again, this is in line with other studies.^11,14 Current study results show a high level of non- consumption of iron supplements among the anaemic population, which probably reflects low demand among other factors as this could be either because the respondents were not aware of being anemic or it is not considered as a serious life threatening medical condition. Despite the GoI providing IFA supplementation and it being one of the longest-running programs of the government, India has still not been able to sufficiently tackle the anaemia burden.

Inadequate access to information, lack of counseling on benefits of supplementation and absence of programs to reach NPNLW, especially those who do not fall within the purview of 6 month post-partum, could be reasons for non-consumption of IFA tablets.^10,14

However, the Government of Madhya Pradesh is intensifying its efforts to address anaemia and conducts bi-annual health camps where the haemoglobin levels of NPNLW are assessed and those who are anaemic are provided with iron supplements. In addition, the importance of consuming IFA tablets needs to be reiterated while educating women on health and nutrition in order to achieve sustained progress in tackling anaemia. Third, engaging to improve the sanitation facilities is essential. Poor hygiene and sanitary conditions significantly contribute to anaemia incidence, which could have adverse consequences, especially, on pregnant women and young children.¹⁶ The Government of India has undertaken a ‘total sanitation campaign’ to end the practice of open defecation and in turn, improve hygiene and sanitation, which will eventually lead to improved health in the population.¹⁷ However, even though there is sufficient evidence to link sanitation with anaemia, in reality, improving sanitation still remains a major challenge. Sensitizing and educating people and ensuring provision for appropriate sanitation facilities is important while addressing the challenge of anaemia.

Dependent variable: prevalence of any anaemia: coded as 0 if the women had any anaemia and 1 if she did not.

Apart from the above mentioned three factors, which can be addressed through public health interventions, age, religion, and type of household also emerged as predictors of anaemia. Anaemia is observed from a young age in most females, and studies confirm that younger women have a higher chance of being anaemic.^10,18 Our results showed that those in the 20 to 29 years and 40 to 49 years age bracket had the highest risk of being anaemic. As observed in other studies, anaemia from a young age could also be related to gender bias, poverty and lack of nutrition education, thus making it imperative that public health interventions start right from young age.^14,18 Along with the above observations, our study also reported low levels of knowledge on fortification among FPS owners and decentralized millers. The low level of knowledge can be corroborated with studies conducted in India on the knowledge of fortification of foods.^19,20 Poor knowledge could partially be due to fortification in India not being mandatory.

Since it still is voluntary, there has been very little to almost no communication on the benefits of fortification.¹⁹As part of our study, we also observed that almost everyone has access to mobile phones and televisions. It is vital that while planning any behavior change communication campaigns, to try and include as much messaging as possible through these two media sources, as these are most accessible to everyone. Hence, communicating the benefits of any public health intervention, while initiating as well as during the intervention is essential to its success.

CONCLUSION

The results of our study revealed that education and consumption of iron supplements are important determinants of anaemia. Other determinants which emerged were age and type of household. Of all the determinants, two of them (education and iron supplementation) can be addressed through public health interventions. Along with the National Iron Plus Initiative Program, Poshan Abhiyan (Campaign) and the Total Sanitation Campaign, it is imperative that women and the population, in general, be sensitized and educated on anaemia. Explaining to them how all factors work together to affect their health should be the strategy for these education sessions. However, along with this, ensuring proper access to health facilities and supporting programs which provide nutrition education, livelihood skills, and livelihood training are also essential in order to achieve and maintain progress. The World Health Organization (WHO) recommends a comprehensive approach of a mix of nutrition-specific and nutrition-sensitive interventions to address the burden of anaemia, two of which are (i) dietary diversification and food fortification and (ii) food supplementation. Food fortification is stated by The Copenhagen Consensus (2008) as a cost-effective intervention to address malnutrition/iron deficiency anaemia. Along with its cost-effectiveness, fortifying a staple food is also a sustainable way of countering the malnutrition challenges faced by a population. The next steps could be towards creating awareness on the benefits of fortification and ways in which wheat flour can be fortified, followed by generating demand for fortified foods

ACKNOWLEDGEMENTS

Authors are thankful and acknowledge the support received from the officials of the Government of Madhya Pradesh, survey respondents, service providers, and other key stakeholders.

Funding: Global Affairs of Canada

Conflict of interest: None declared

Ethical approval: The study was approved by the Institutional Ethics Committee

REFERENCES

1. 1. Benoist B, McLean E, Egli I, Cogswell M. Worldwide prevalence of anaemia 1993-2005: WHO global database on anaemia. Available at: https://www.who.int/publications/i/item/9789241596657. Accessed on 20 November 2021.
2. 2. National Family Health Survey 2017 (NFHS-4). Available at: https://www.rchiips.org/nfhs/nfhs4.shtml. Accessed on 20 November 2021.
3. 3. Kalaivani K. Prevalence and consequences of anaemia in pregnancy. Indian J Med Res. 2009;130(5):627-33.
4. 4. Special Bulletin on Maternal Mortality in India 2014-16. Available at: http://www.censusindia.gov.in/vital_statistics/SRS_Bulletins/MMR%20Bulletin-2014-16.pdf. Accessed on 20 November 2021.
5. 5. Kumar A. National nutritional anaemia control programme in India. Indian J of Public Health. 1999;43(1):3-5.
6. 6. Rai RK, Fawzi WW, Barik A, Chowdhury A. The burden of iron-deficiency anaemia among women in India: how have iron and folic acid interventions fared?. WHO South-East Asia J Public Health. 2018;7(1):18-23.
7. 7. Horton S, Mannar V, Wesley A. Second copenhagen consensus: micronutrient fortification best practice. Available at: https://www.copenhagenconsensus.com/. Accessed on 20 November 2021.
8. 8. India: Census 2011 office of the registrar general & census commissioner, India. Available at: http://censusindia.gov.in/. Accessed on 20 November 2021.
9. 9. District level population projections in eight selected states of India 2006-2016. Available at: https://District_level_population_projections_in_selected_staties_of_India_-_2006_to_2016. Accessed on 20 November 2021.
10. 10. Little M, Zivot C, Humphries S, Dodd W, Patel K, Dewey C. Burden and determinants of anemia in a rural population in south india: a cross-sectional study. Anemia. 2018;2:34-8.
11. 11. Lokare PO, Karanjekar VD, Gattani PL, Kulkarni AP. A study of prevalence of anemia and sociodemographic factors associated with anemia among pregnant women in Aurangabad city, India. Ann Nigerian Med. 2012;6:30-4.
12. 12. Titaley CR, Rahayu E, Damayanti R, Dachlia D, Sartika RAD, Ismail A, et al. Association between knowledge and compliance of taking iron/folic acid supplements during pregnancy. Asian J Pharm Clin Res. 2017;22:159-65.
13. 13. Bentley ME and Griffiths PL. The burden of anemia among women in India. EJCN. 2003;57:52-60.
14. 14. Gautam VP, Bansal Y, Taneja DK, Saha R. Prevalence of anemia amongst pregnant women and its socio-demographic associates in a rural area of Delhi. IJCM. 2002;27(4):157-60.
15. 15. Balarajan Y, Fawzi W, Subramanian SV. Changing patterns of social inequalities in anaemia among women in India: cross-sectional study using nationally representative data. CMJ. 2013;3:23-9.
16. 16. Coffey D, Geruso M, Spears D. Sanitation, disease externalities and anaemia: evidence from Nepal. EJ. 2017;12:1395-432.
17. 17. Patil SR, Arnold BF, Salvatore AL, Criceno B, Ganguly S, Colford JM, et al. The effect of India’s total sanitation campaign on defecation behaviors and child health in rural Madhya Pradesh: A cluster randomized controlled trial. PLOS. 2017;11(8):e1001709.
18. 18. Lone FA, Qureshi RN, Emanuel F. Maternal anemia and its impact on perinatal outcome. Trop Med Int Health. 2004;9(4):12-8.
19. 19. Arora H, Dixit V, Srivastava N. Evaluation of knowledge, practices of vitamin D and attitude toward sunlight among Indian students. Asian J Pharm Clin Res. 2016;9(1):23-9.
20. 20. Ahluwalia N. Intervention strategies for improving iron status of young children and adolescents in India. Nutr Res. 2002;60(5):23-9.
21. 21. Haemoglobin concentrations for the diagnosis of anaemia and assessment of severity. Available at: https://www.who.int/vmnis/indicators/haemoglobin.pdf. Accessed on 20 November 2021.