Socioecological risk factors associated with Malaria in Pregnancy in an endemic state of India: a retrospective study

Abstract

Introduction

Malaria in Pregnancy (MiP) has significant adverse effects on both mother and fetus. Pregnant women in regions with unstable malaria transmission are three times more vulnerable to infection. This study analysed malaria prevalence among pregnant women in Odisha, India from 2016 to 2020 and compared it with overall malaria rates. Socioecological factors potentially influencing MiP prevalence were also examined.

Methods

District-wise cases of malaria-positive pregnant women were analysed using a prevalence index named Malaria in Pregnancy rate (MiPr). The association of MiP with multidimensional poverty, forest cover and tribal population was studied. These three socioecological variables were compared with high and low MiPr (MiPr ≥1 or <1) respectively.

Results

A strong positive association was observed between the Annual Parasite Index (API) and the MiPr in 2016 (0.95), 2017 (0.97), 2018 (0.88), 2019 (0.97) and 2020 (0.97). The districts comprising a multidimensionally poor population of 45% or more accounted for 67% of the MiP cases in 2020. The odds of getting MiP (MiPr ≥1) were 82.5 times higher in the districts where the tribal population was ≥50% and 3.39 times higher in the districts where the forest cover was ≥40%. In 2020, two districts with high MiPr, Malkangiri (MiPr=5.61) and Rayagada (MiPr=3.24), were above the threshold for all three variables.

Conclusions

This work highlights an urgent need to increase awareness by the national control programme and the community in vulnerable regions through control and protection measures for pregnant women at higher risk of severe disease.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Pregnant women living in areas of unstable malaria transmission have a considerably higher susceptibility to malaria than non-pregnant women.

• Malaria in Pregnancy is likely dependent on varied sociocultural, economic and/or environmental factors where pregnant women reside, especially in malaria-endemic regions.

WHAT THIS STUDY ADDS

• The Annual Parasite Index (API) is commonly used to estimate malaria endemicity in the overall population of an area; similarly, we have used an index named Malaria in Pregnancy rate (MiPr) to analyse the temporal trend, spatial distribution and the intensity of malaria among the pregnant population subset in a highly malaria-endemic state of India.

• We analysed the district-wise association of MiPr with API and socioecological factors such as multidimensional poverty, forest cover and tribal population.

• This study shows that all three socioecological risk factors, that is, multidimensional poverty, tribal population and forest cover contribute to a higher number of malaria cases in pregnant women, highlighting their relevance in endemic settings.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Our work recommends enhancing research programmes in consonance with the social and ecological frameworks to achieve sustainable long-term outcomes for pregnant women.

• Since India is currently in the malaria elimination phase, early detection and prompt action of every case is essential along with appropriate vector control in endemic regions.

• Key interventions for poor and tribal population and those living in forested areas will be significant in achieving India’s malaria elimination target.

Introduction

Malaria remains a significant public health concern as the World Health Organization (WHO) estimated approximately 247 million malaria infections worldwide in 2021.¹ The South-East Asia region contributed around 2% of the overall burden, with India accounting for the majority of the cases (79%) in this region.¹ Pregnant women, children under the age of 5 years, and people with weak immunity are at a higher risk of contracting malaria and developing severe forms of this disease.² Sociodemographic factors such as the low age of the household head, occupation of the head and crowded households can contribute to an increased risk of malaria transmission.³ Additionally, cultural factors such as a belief in traditional medicines, fear of side effects of medicines, and lack of awareness about malaria and its vectors can influence malaria and outcome of malaria cases, particularly in women and children with poor access to education and healthcare resources.⁴ Women are socially and biologically more susceptible to malaria during their pregnancy, especially in developing countries, and the detrimental effects of malaria on the mother and fetus are well known. Malaria in Pregnancy (MiP) can lead to several adverse outcomes like fetal and maternal anaemia, preterm labour, pulmonary oedema, hypoglycaemia, low birth weight, stillbirth, and death.4,9 In areas of unstable malaria transmission, pregnant women are threefold more susceptible to malaria infection than non-pregnant women.⁷ MiP, a preventable disease, is dominated by Plasmodium falciparum (Pf) infections.^{5 6 10} Multiple studies in India have reported that pregnant women living in endemic regions are disproportionately affected by malaria.^{4 11 12} A study by Singh et al¹² reported that parasite densities were significantly higher in pregnant women as compared with non-pregnant women for both Pf and Plasmodium vivax (Pv) infections.¹² A significant difference was observed in the mortality rate among pregnant women infected with malaria in comparison with non-pregnant women and men.¹³ Pv is commonly reported in mothers with weakened acquired immunity.⁴ The general population exposed to malaria develops non-sterilising immunity by acquiring antibodies; however, a combination of factors in pregnant women leads to higher susceptibility to the infection. These factors include (i) Increased immune tolerance to protect the fetus, thereby compromising an effective immunological response against the malaria parasite, (ii) Maternal physiological changes, (ii) The existence of comorbidities, and (iv) The presence of placenta that poses a high risk of pregnancy-associated placental malaria leading to adverse outcomes for the mother and the fetus, especially during the first pregnancy. Pf-infected erythrocytes bind to chondroitin sulfate A (CSA) on the surface of the placenta via PfEMP1 VAR2CSA, (variant surface antigen 2 CSA).14,16 Multiple field have shown the presence of VAR2CSA antibodies during subsequent pregnancies in response to malarial infection in the placenta. Thus, enhanced acquired maternal immunity against pregnancy-associated placental malaria is observed during subsequent pregnancies since disease severity is lower when compared with the infection during the first pregnancy.¹⁶

India targets to be malaria-free by 2027 and achieve malaria elimination by 2030. To provide a roadmap for progress towards malaria elimination, India’s National Center for Vector Borne Disease Control (NCVBDC) with support from WHO, developed a National Strategic Plan (NSP) for 2017–2022 by identifying the district in the country as operational unit. The NSP stratified districts into four categories based on their Annual Parasite Index (API) reported in 2015, that is, Category 0—Prevention of Re-establishment Phase (districts that reported no local transmission in the past 3 years), Category I—Elimination Phase (districts with API<1), Category II—Pre-elimination Phase (districts with API >1, <2 per 1000 population), and Category III (districts with API >2 per 1000 population).¹⁷ The API for a particular geographical area is calculated using the reported malaria-positive cases in a specific year, expressed per thousand population under surveillance.¹⁸ In 2016, 8 districts of Odisha state were in Category I, 1 in Category II and 21 in Category III. Whereas in 2020, 23 districts were in Category I, 2 in Category II and 5 in Category III. According to the NCVBDC, in 2020, the overall API of Odisha state was reported to be 0.9.

In this work, we intended to study the role of ecological (forest cover) and social (multidimensional poverty and tribal population) determinants in malaria transmission in pregnant women of Odisha state of India from 2016 to 2020. These three socioecological factors tend to determine the persistence of the disease among pregnant women. Further, we analysed the prevalence of malaria among pregnant women in relation to the overall population using 5-year retrospective data. The understanding and coexistence of the investigated socioecological aspects with MiP is crucial for developing and designing effective strategies to prevent and control malaria.

Methods

Study site

Out of 36 states and union territories of India, Odisha state is highly malaria endemic and approximately 46 million people are at risk of malaria infection.¹⁹ According to the NCVBDC, Odisha state, with its overall 4% share of the Indian population, was responsible for 16% of the country’s malaria burden and 14% of malaria-related fatalities in 2021.²⁰ Odisha is an east-situated state land-locked by five surrounding Indian states, namely, West Bengal, Jharkhand, Chhattisgarh, Telangana and Andhra Pradesh, and with the Bay of Bengal in its south-east direction (online supplemental figure S1). Odisha’s topography is broadly categorised into the northern upland, central river basin, hilly regions of the south-west and coastal plains.²¹ It is known for its tribal community, forested, hilly and inaccessible areas.²² The 30 districts of Odisha exhibit diverse characteristics such as tribal, forested or coastal. Additionally, it is primarily a state with a dominant rural population, with approximately 86% of its residents residing in rural areas.²³

The three socioecological variables—forest cover, multidimensional poverty and tribal population, were carefully selected, as these are independent but may be associated because (i) These cannot be easily manipulated by the residing population; (ii) These do not get affected by malaria control measures like the provision and usage of long-lasting insecticidal nets (LLINs) and other interventions, and (iii) These are broadly non-modifiable in the short-term duration and/or at an individual level. Also, the association between API and malaria in pregnancy rate (MiPr) was studied from 2016 to 2020 to emphasise the benefit of a declining trend in API percolating to pregnant women.

The study period 2016–2020 was purposively selected. India had pledged for malaria elimination in 2016 marked by the launch of the National Framework for Malaria Elimination (2016–2030). Technical guidelines in the form of NSP (2017–2022) were subsequently prepared in 2017. The national malaria control programme was entirely reoriented and realigned with a focus on malaria elimination, and several new strategies and activities, such as the provision of LLINs, were brought under the ambit of the national programme. Besides the national-level changes, Odisha state launched a programme called the Durgama Anchalare Malaria Nirakarana (DAMaN) in 2017 to focus on inaccessible high-burden districts. By 2020, up to 90% reduction in the malaria burden was observed in these districts, which was attributed to the DAMaN programme.

Data sources

The 5-year retrospective epidemiological data (2016–2020) on malaria-positive pregnant women was obtained from the state vector borne disease programme.²⁴ The data on the number of pregnant women were obtained from the state health authorities (Maternal and Child Health Division) on request. Malaria epidemiological data of the overall population were provided by the NCVBDC on request.²⁰ The epidemiological data included the API and the number of malaria cases reported.

Multidimensional poverty refers to the concept of poverty as a condition that is not only characterised by a lack of income or material resources but also by a lack of health facilities, education and standard of living. The Multidimensional Poverty Index (MPI) uses health, education and standard of living as three equally weighted dimensions to measure poverty in an area. These three dimensions are subdivided into the following indicators: (i) Health consisting of nutrition, child and adolescent mortality, and maternal health; (ii) Education consisting of school attendance and years of schooling; and (iii) Standard of living consisting of sanitation, drinking water, electricity, cooking fuel, housing, assets and bank account (online supplemental table S1). The percentage of multidimensionally poor people living in each district was calculated and represented as the MPI and published by the NITI Ayog (National Institution for Transforming India), a nodal agency of the Government of India, in association with the University of Oxford’s ‘Oxford Poverty and Human Development Initiative’, and released in 2021.²⁵

The district-wise percentage of the tribal population was extracted from district statistical handbooks published by the Directorate of Economics and Statistics, Odisha.²⁶ The estimated percentage of the tribal population of each district was based on surveys conducted by the government during 2016–2020.

The percentage of the geographical area under forest cover of each district of Odisha during 2019–2020 was obtained from the Forest Survey of India (FSI) report of 2021.²⁷ The FSI uses a hybrid approach to classify the satellite data, which includes several steps such as digitisation and visual examination of images, comparison after classification, validation, field evaluation by state forest departments, and corrections after field evaluation. The FSI publishes the results in the form of maps and area statistics using IRS LISS-III (Indian Remote Sensing Satellite -Linear Imaging and Self Scanning Sensor III) Resourcesat-2 satellite (spatial resolution of 23.5 m) data. In this study, we addressed the association between the percentage of forest cover and MiP.

Data analysis

Since the API is used for assessing malaria endemicity in an overall population, on a similar basis, the MiPr, was used as an indicator to determine the prevalence of malaria in pregnant women. MiPr was calculated as:

$${\displaystyle \mathrm{M}\mathrm{i}\mathrm{P}\mathrm{r}=\frac{\mathrm{N}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r}\mathrm{o}\mathrm{f}\mathrm{m}\mathrm{a}\mathrm{l}\mathrm{a}\mathrm{r}\mathrm{i}\mathrm{a}-\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{g}\mathrm{n}\mathrm{a}\mathrm{n}\mathrm{t}\mathrm{w}\mathrm{o}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{i}\mathrm{n}\mathrm{a}\mathrm{d}\mathrm{i}\mathrm{s}\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{c}\mathrm{t}}{\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{n}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r}\mathrm{o}\mathrm{f}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{g}\mathrm{n}\mathrm{a}\mathrm{n}\mathrm{t}\mathrm{w}\mathrm{o}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{i}\mathrm{n}\mathrm{a}\mathrm{d}\mathrm{i}\mathrm{s}\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{c}\mathrm{t}}\ast 1000}$$

# malaria-positive denotes both symptomatic and asymptomatic malaria cases.

In the present study we focused on the districts with MiPr ≥1 (one malaria-positive pregnant woman per thousand pregnant women), similar to the API-based categorisation of districts in the NSP.

Statistical analysis

The first association of MiPr with the percentage of population living under multidimensional poverty, the percentage of geographical area under forest cover and the percentage of tribal population was assessed using the Pearson coefficient of correlation. Further, each of these variables was compared in two categories of MiPr (<1, ≥1) in all districts of Odisha using the rank-sum test. Then, to identify a threshold value for each independent variable (multidimensional poverty, forest cover and the tribal population) for high malaria endemicity in pregnant women (MiPr ≥1), sensitivity and specificity tests using receiver operating characteristic (ROC) curves were done (online supplemental tables S2A-C). As per the ROC analysis, the point where we had the same (or close) sensitivity and specificity was considered a cut-off value (or point). Based on the ROC analysis, multidimensional poverty (<45% and ≥45%), tribal population (<50% and ≥50%) and forest cover (<40% and ≥40%) were split into two categories and compared between high (MiPr ≥1) and low (MiPr <1) using logistic regression. The magnitude of association was expressed in OR and CI. Statistical significance was defined as a value of p≤0.05 and a CI of 95%.

Data collected from different sources were compiled at the district level in Microsoft Excel. Further, data were coded and formatted to make them suitable for usage in the open-source geographic information system software i.e., QGIS²⁸ and statistical software (Stata V.15.0).²⁹ For a graphical impression, the relationship between MiPr and the three variables was shown using quadrant plots. The district-wise MiPr (2020) was plotted on the y-axis, and the percentage of the multidimensional poverty, tribal population and forest cover were plotted individually on the x-axis into separate graphs. The threshold values (40%, 45% and 50%) and MiPr≥1 were used for marking the four quadrants.

Results

Association between API and MiPr

Our study shows that malaria declined in the overall population and in pregnant women in Odisha state during 2016–2020. More than a 90% reduction in the state’s malaria burden was reported from 2016 (444 843 cases) to 2020 (41 739 cases). The state’s MiP burden declined by 88% from 4014 cases in 2016 to 483 in 2020. A strong positive correlation was observed between API and MiPr (0.95, 0.97, 0.88, 0.97 and 0.97 in 2016, 2017, 2018, 2019 and 2020, respectively) (figure 1).

Figure 1. Number of districts of Odisha state, India, in different categories of MiPr (Malaria in Pregnancy rate) from 2016 to 2020. Odisha demonstrated a remarkable 88% reduction in the incidence of malaria-positive cases among pregnant women from 2016 to 2020. In 2016, there were 17 districts with a MiPr≥1. The number of these districts decreased to 17, 11, 9 and 7 in 2017, 2018, 2019 and 2020 respectively. Over the same period, the number of districts with an MiPr=0 were 5, 3, 6, 4 and 11, indicating an increasing trend. The upward trend in districts with an MiPr=0 signifies an improvement in the situation of pregnant women in the state.

The number of districts of Odisha with API>1 decreased from 22 in 2016 to 7 in 2020. Seventeen out of 30 districts of the state had MiPr≥1 in 2016, which reduced to 7 districts by 2020. On the other hand, the number of districts with MiPr between 0 and 1 increased from zero in 2016 to 12 in 2020. The comparison between MiPr and API revealed that all districts had higher API than MiPr in 2016, except for Nuapada district. Notably, in Nuapada district, malaria prevalence was higher among pregnant women compared with the general population (figure 2). Though the MiPr and API reported a steep decline from 2016 to 2020, seven districts, Sundergarh, Keonjhar, Nawarangpur, Deogarh, Mayurbhanj, Bargarh and Angul, showed a higher prevalence of malaria among pregnant women compared with the overall population (figure 2).

Figure 2. Comparison of MiPr (Malaria in Pregnancy rate) of 2016 and 2020 with Annual Parasite Index (API) for 30 districts of Odisha state, India. Throughout the study period, a consistent positive correlation was observed between the API and MiPr. Strong positive correlations were found in each year: 0.95, 0.97, 0.88, 0.97 and 0.97 in 2016, 2017, 2018, 2019 and 2020 respectively. In most districts, the API was higher than the MiPr. However, in 2016 and 2020, Nuapada and Sundargarh districts, respectively, reported an MiPr value greater than the API. The pyramid graph illustrates a significant decrease in the number of cases in the general population as well as the pregnant population, indicating a substantial decline.

Multidimensional poverty

As per a report published in 2021, 11%–59% of the population in Odisha was multidimensionally poor. Around 45% of the population of the six districts located in the south-western region of Odisha were multidimensionally poor and accounted for 67% of MiP cases in 2020 (table 1).

Table 1. Districts of Odisha state, India, falling above the calculated threshold and MiPr 2020.

District	Malaria in pregnancy rate (MiPr) (2020)	Multidimensional poverty (≥ 45%)	Tribal population (≥ 50%)	Forest cover (≥ 40%)
Malkangiri	5.6	●	●	●
Rayagada	3.2	●	●	●
Kalahandi	2.4	●
Kandhamal	2.3		●	●
Koraput	1.4	●	●
Nawarangpur	1.4	●	●
Sundargarh	1.3		●	●
Boudh	0.9			●
Keonjhar	0.5
Bargarh	0.5
Mayurbhanj	0.4	●	●
Deogarh	0.4			●
Gajapati	0.3		●	●
Nuapada	0.2
Sambalpur	0.2			●
Ganjam	0.2			●
Jharsuguda	0.1
Angul	0.1			●
Balangir	0.1
Nayagarh	0.0			●
Bhadrak	0.0
Jagatsinghpur	0.0
Puri	0.0
Balasore	0.0
Jajpur	0.0
Kendrapada	0.0
Subarnapur	0.0
Khurda	0.0
Cuttack	0.0
Dhenkanal	0.0

Among the thirty 30 districts in the eastern state of Odisha, India, only two districts—Malkangiri and Rayagada—displayed all three variables (multidimensional poverty, tribal population, and forest cover) surpassing the predetermined threshold. These districts, represented by red dots in the table, reported the highest MiPr values among all districts. Six districts, indicated by black dots, had two variables exceeding the threshold, resulting in MiPr values ranging from 0.3 to 2.3. On the other hand, seven districts, depicted by blue dots, had only one variable surpassing the threshold, with MiPr values ranging from 0 to 2.5. The remaining sixteen 16 districts, falling below the threshold values for each variable, reported significantly low or zero MiPr.

Significantly higher MiPr (p<0.001) was observed in districts where the percentage of multidimensional poverty was more than 45% (MiPr, p50 (p25 to p75): 1.86 (0.95 to 2.82)) as compared with the districts where the multidimensionally poor population was lower than 45% (MiPr, p50 (p25 to p75): 0.03 (0 to 0.22)) (table 2). Five districts with multidimensional poverty ≥45% had high MiPr (≥1), whereas only 2 districts among the remaining 25 districts with multidimensional poverty below 45% had MiPr≥1. Since no district showed MiPr<1 and multidimensional poverty ≥45%, OR could not be calculated. Thus, we reported the value of p from the χ² test and marked it as ‘NA’ in the OR column (table 2).

Table 2. Association between Malaria in Pregnancy rate for the year 2020 (MiPr 2020) and the selected variables in districts of Odisha state using logistic regression.

Variables	N	MiPr 2020		OR (95% CI)	P value
		<1	≥1
		N (%)	N (%)
Multidimensional poverty (%)					0.002
<45%	24	23 (95.8)	1 (4.17)	NA
≥45%	6	0 (0.0)	6 (100.0)
Tribal population (%)
<50%	22	21 (95.4)	1 (4.5)	Ref	0.002
≥50%	8	2 (25.0)	6 (75.0)	82.5 (4.8 to 820)
Forest cover (%)					0.139
<40%	20	17 (85)	3 (15)	Ref
≥40%	10	6 (60.0)	4 (40.0)	3.39 (0.6 to 22.0)

Multidimensional poverty: All six districts with multidimensional poverty ≥45% (ie, 100%) had MiPr≥1. Out of the 24 districts, 23 of them (ie, 95.8%) with multidimensional poverty <45% reported MiPr<1. The Odds ratioOR could not be calculated, as the P value of p was non-significant. Tribal population: Out of the eight districts, six of them (ie, 75%) having a tribal population ≥50% reported MiPr≥1. 2Twenty-three3 out of 24 districts (ie, 95.4%) with a tribal population less than <50% reported MiPr lower than <1. The pregnant women living in the districts with a tribal population ≥50% were 82.5% more susceptible to malaria infection, as compared tocompared with its the reference group. Forest cover: Out of ten 10 districts, four 4 districts districts (ie, 40%) having a forest cover ≥40% reported MiPr≥1. Seventeen17 out of 30 districts (ie, 56.5%) with a tribal population less than <40% reported MiPr lower than< 1. The pregnant women living in the districts with a tribal population ≥50% were 3.39 times more susceptible to malaria infection, as compared tocompared with its the reference group.

^*Percentages are given row-wise.

NA, not applicable; Ref, reference group

The quadrant analysis of MiPr and multidimensional poverty in 2020 revealed quadrant I containing districts of concern, where five districts have high MiPr (≥1) and ≥45% multidimensional poverty. Quadrant II indicates low poverty and high MiPr and contains two districts, while the remaining 23 districts lie in quadrant III, characterised by low MiPr (<1) and low multidimensional poverty (<45%) (figure 3A).

Figure 3. Quadrant distribution of MiPr (Malaria in Pregnancy rate) in 2020 with the percentage share of the three selected socioecological variables in each district of Odisha state, India. The districts analysed in this study were categorised into different quadrants based on their levels of MiPr and the percentage share of the variables. The following descriptions outline the distribution of districts across the quadrants for each variable: (A) MiPr with multidimensional poverty: In quadrant I, six districts had high MiPr and a high variable percentage. Quadrant II had one district with high MiPr but a low variable percentage. Quadrant III consisted of 23 districts with low MiPr and a low variable percentage. No districts were into quadrant IV, indicative of low MiPr and a low variable percentage. (B) MiPr with tribal population: Quadrant I had six districts with high MiPr and a high variable percentage. One district was in Quadrant II, representing high MiPr but a low variable percentage. Quadrant III included 20 districts with low MiPr and a low variable percentage. Three districts were situated in quadrant IV, indicating low MiPr and a low variable percentage. (C) MiPr with forest cover: Four districts were classified in Quadrant I, representing high MiPr and a high variable percentage. Quadrant II contained three districts with high MiPr but a low variable percentage. Quadrant III had 17 districts with low MiPr and a low variable percentage. Six districts were placed in Quadrant IV, suggesting low MiPr and a low variable percentage.

Tribal population

Approximately 50% or more of the total population in eight districts of Odisha state is tribal-dominated (table 1). MiPr was observed to be significantly higher (p<0.001) in these eight districts (MiPr, p50 (p25 to p75): 1.38 (0.85 to 2.78)) in comparison with the remaining districts (MiPr, p50 (p25 to p75): 0.03 (0 to 0.22)). The odds of malaria cases in pregnant women (MiPr≥1) were 82.5 times higher (95% CI 4.8 to 820, p=0.002) in districts comprising a tribal-dominated population (table 2). The 95% CI has a wide range since there were only 30 observations, and the proportion of the tribal population (≥50%) varied greatly between the two groups of MiPr. As a result, there was a higher standard error (SE) and, therefore wider CI.

Additionally, quadrant analysis for the tribal population ≥50% and MiPr≥1 revealed six districts in quadrant I that show high MiPr (≥1) and high tribal population (≥50%). Quadrant II contains only one district, with MiPr≥1 and a tribal population <50%. Twenty-one districts in quadrant III had low MiPr (<1), and the percentage of the tribal population was lower than the calculated threshold (<50%). The remaining two districts are in quadrant IV and have a low number of malaria cases but a high percentage of the tribal population (figure 3B).

Forest cover

A comparison between districts with forest cover <40% (MiPr, p50 (p25 to p75): 0 (0 to 0.15)) and ≥40% (MiPr, p50 (p25 to p75): 0.36 (0.17 to 2.33)) revealed that the districts with forest cover >40% had higher malaria cases (p=0.005). Also, the odds of getting higher malaria cases in pregnant women (MiPr≥1) were almost 3.39 (95% CI 0.6 to 22.0, p=0.139) times higher in these districts (table 2).

In quadrant I, three districts were identified, as having high MiPr (MiPr≥1) and ≥40% area under forest cover (figure 3C). Quadrant II also comprised of three districts that represent high MiPr (≥1) and low forest cover (<40%). Six districts were in quadrant IV, depicting a high percentage of forest cover (≥40%) but low MiPr. The remaining 18 districts in quadrant III represent low MiPr (<1) and low forest cover (<40%).

In 2020, 7 out of 30 districts of Odisha state reported MiPr>1. Among these seven districts, only two districts, Malkangiri and Rayagada, surpassed the threshold for all three socioecological factors, that is, percentage of multidimensional poverty (≥45%), percentage of forest cover (≥40%) and tribal population (≥50%) (figure 4). It is noteworthy that these two districts reported the highest MiPr in 2020. Also, four districts, Kandhamal, Koraput, Nawarangpur and Sundargarh, surpassed the threshold for two out of three socioecological factors. Finally, Kalahandi district lied above the threshold in only multidimensional poverty.

Figure 4. Maps representing the districts of Odisha state that lie above the calculated threshold values for three socioecological variables (A) The south-eastern region of the state encompasses six districts characterised by multidimensional poverty levels of ≥45%. (B) Within Odisha, eight districts have a tribal population of ≥50% of the total. Remarkably, six of these districts are situated in the south-east region of the state. (C) The map illustrates nine centrally located districts where forest cover spans ≥40% of the geographical area. These districts stretch from the north to the south of the state. (D) The map highlights two districts, namely Malkangiri and Rayagada, which surpass the threshold for all three variables studied. These districts have a history of significant malaria endemicity, as indicated by the numerical labels representing Malaria in Pregnancy rate (MiPr) in 2020.

Discussion

This study investigated the district-wise association of MiPr with API in Odisha state from 2016 to 2020. We also evaluated the impact of three socioecological variables, namely, multidimensional poverty, tribal population and forest cover on the prevalence of MiP. Our findings revealed a strong positive correlation between API and MiPr, indicating that the intensity of malaria was equally severe among the pregnant women section and the overall population. The districts with higher percentage of the three investigated socioecological factors had significantly higher MiPr. Thus, pregnant women living in those districts were more susceptible to malaria infection. Based on these findings, we have identified the districts of concern and here provid insights into areas that require further interventions to control malaria incidence in pregnant women.

Similar to the findings of our study, Garg et al³⁰ observed a positive correlation between MiP and malaria cases in the overall population in Chhattisgarh, another malaria-endemic state in the country. The study suggested that the pregnant population can be treated as a sentinel population to effectively predict the malaria situation in the overall population.³⁰

Our study indicates that pregnant women living in forested and tribal-dominated districts of the state were significantly more susceptible to malaria infection. Ranjha and Sharma³¹ suggested that forested districts contribute to approximately 32% of malaria cases in India, whereas only around 6% of the country’s total population lives in these districts. Inhabitants of forested districts may either be tribal or migrant communities and are often missed by the surveillance network.³¹

Limited access to healthcare in marginalised or impoverished regions of a country poses a significant risk for pregnant women.⁴ Multidimensional poverty is indicative of a region’s socioeconomic status since it considers various aspects like health, education and the standard of living. Access to healthcare is essential for ensuring positive maternal and child health outcomes, especially during pregnancy. However, pregnant women living in marginalised or impoverished areas often face numerous barriers to accessing healthcare services. Also, undernutrition in pregnant women that live in poverty can worsen their susceptibility to malaria infection, leading to fetal malnutrition and adverse maternal and perinatal outcomes.³² The nutrition status of the mother determines the impact of malaria on a newborn’s birth weight and growth. Lack of access to antenatal care, literacy skills, clean water, hygiene and sanitation further affect maternal health.^{33 34} Since half the population of the districts with high MiP incidence was also multidimensionally poor, the health and status of pregnant women gets impacted further. The tribes living in remote villages of some districts of Odisha prefer their traditional healers for malaria and other ailments due to the lack of transport facilities to access modern healthcare, among other factors.³⁵ Healthcare providers must actively promote modern healthcare access and educate pregnant women on malaria prevention for their safety.

To tackle the malaria situation in Odisha, ‘Durgama Anchalare Malaria Nirakarana (DAMaN)’ was initiated by the state NCVBDC in mid-2017 in 17 high-endemic districts of the state. The aim was to eliminate malaria in tribal-dominated areas using efficient malaria control measures by conducting three surveillance rounds per annum.³⁶ The human-mosquito-human cycle was broken to control the disease by distributing and promoting LLINs.³⁷ The results of this disease intervention programme indicated a positive outcome as it reduced the disease burden among pregnant women. The state’s malaria burden reduced from 37% in 2015 to only 12% in 2019, and was credited to the DAMaN initiative.³⁸ The overall 90% reduction in the malaria case incidence among pregnant women of Odisha during the study period can be attributed to the intervention by the state and national programmes.²² Malaria reduced significantly (89%) after the mass distribution of LLINs and the introduction of DAMaN in Koraput district of Odisha as reported by Sahu et al. ³⁷ Preventive measures and control of malaria incidence can be successfully managed with the help of ASHA (Accredited Social Health Activists) workers.³⁹ To support the public health delivery system, India’s National Rural Health Mission introduced the ASHA workers in 2005, covering about 1000 persons for each worker. Starting in 2010, these ASHAs were supplied with rapid diagnostic test (RDT) kits and antimalarial drugs and trained to diagnose, treat and report malaria using NCVBDC standards.⁴⁰ These ASHA workers are female volunteers working as health activists at the grassroots level and are accepted by the community to bridge the gap between the community and the health system.⁴¹ ASHAs participate in active surveillance to identify people with fever, and test them using RDT kits and suitably treat them. ASHAs also encourage the communities to support and use appropriate interventions.

India’s target of eliminating malaria by 2030 may prove to be challenging in tribal communities which live in predominantly inaccessible forested areas in plain, hilly and mountainous terrains. The NSP has identified states like Odisha with hilly, tribal areas and large forests as major contributors of malaria in their report ‘National Strategic Plan Malaria Elimination in India 2017–2022’.¹⁷ According to this report, LLINs are distributed to pregnant women on a priority basis in high-risk areas. The NSP targeted the tribal population with a forest-based economy, outdoor sleeping habits and poor health-seeking behaviours.

To address the significant burden of MiP in Odisha, especially among the poor and tribal population, it is essential to research and introduce effective interventions to reduce the prevalence of malaria. The urgency of this issue cannot be overemphasised, and with the looming threat of population growth and climate change, it is even more critical to identify appropriate strategies for the prevention and treatment of an infectious disease like malaria.⁴² Therefore, further research is needed to determine the at-risk population, identify the most vulnerable groups, and develop targeted approaches for protecting vulnerable sections of the population.

Limitations of the study

The data available do not provide the type of infection, that is, Pv or Pf, or the age of malaria-infected pregnant women. These details may have helped to further assess the distribution patterns of MiP in the districts of Odisha state. Moreover, the data are available at the district level, limiting the scope and outcome of the study. Also, we lack an understanding of the patterns and determinants at the micro level for MiP. Micro-level data are crucial for identifying health disparities and understanding the critical contributing factors. Thus, it may be difficult to identify groups exposed to a higher risk of malaria and develop targeted interventions to address disparities at the macro level. In addition, we have not considered the interventions deployed by the state government for communities (general and specifically for pregnant women) in these districts.

Conclusion

Malaria declined in the overall population and among pregnant women in the Odisha state of India during 2016–2020. Our study here shows a strong positive correlation between API and MiPr. Districts showed high MiP where the investigated socioecological factors such as tribal population, forest cover and/or multidimensional poverty were above the threshold value. A multidisciplinary approach that combines the expertises of epidemiology, ecology and sociology is required to tackle MiP.

For pregnant women and infants living in poverty, in forested areas or those belonging to the tribal community, malaria infection is a serious health problem and a continuous risk. This work recommends a diversified elimination programme in consonance with the social and ecological framework of pregnant women for sustainable results on a long-term basis. There is an urgent need to improvise awareness among the community towards personal protection measures, especially among pregnant women. Use of insecticide-treated nets/LLINs among pregnant women has significantly reduced the susceptibility to malaria infection. The tribal community primarily resides in forests in inaccessible areas with poor communication, transport and health facilities. Improved public healthcare facilities are required in tribal areas to prevent the population from visiting untrained and unlicensed practitioners. In India’s elimination phase of malaria, early detection and prompt action to every case is essential, along with appropriate vector control in the region. Key disease control interventions in the low-income and impoverished tribal regions of India will be important in achieving the country’s malaria elimination targets.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.