Risk factors for the development of severe or very severe respiratory syncytial virus‐related lower respiratory tract infection in Indian infants: A cohort study in Melghat, India

Respiratory syncytial virus (RSV) is undoubtedly the single most important cause of severe lower respiratory tract infection (LRTI) globally. While new prevention measures in young infants have become available, their use in developing countries is likely many years away. While risk factors for severe or very severe RSV LRTI in impoverished rural areas likely differ to urban areas, there are very few studies, especially those conducted in India, the major country contributing to the global burden of disease.


INTRODUCTION
Respiratory syncytial virus (RSV) is the most common cause of acute lower respiratory tract infection (LRTI) in young children, with an estimate of 33 million cases globally in children younger than 5 years in 2019 [1].It is estimated that more than 95% of RSV-associated acute lower respiratory infection episodes and more than 97% of RSV-attributable deaths were in low-income and middle-income countries (LMICs) [1].Current estimations of the burden of RSV Sustainable Development Goals: Good Health and Well-being.DOI: 10.1111/tmi.14003morbidity and mortality in LMICs are limited by the lack of available surveillance data of RSV in the community and may underestimate it.Recent studies in Argentina, India, Pakistan and Zambia have shown a high and previously unmeasured burden of community RSV mortality among young infants in LMICs [2,3] highlighting the burden of disease in the community.
The anti RSV-specific humanised monoclonal antibody, palivizumab, is an effective prophylactic measure, but its use is restricted to a small number of infants who are at high risk for severe RSV disease.Newer effective RSV prevention treatments have become available.These include passive immunisation by maternal vaccination during pregnancy [4] and new long halflife prophylactic monoclonal antibodies that can be administered at birth [5].As these new prophylactic measures become widely available to all infants including healthy term infants in high income countries, affordability and access will likely limit their availability in LMICs, at least for the near future [6,7].It is important therefore to identify risk factors for RSV LRIs for young children in LMICs, to identify those with the higher risk of severe disease and to also identify risk factors that are amenable to environmental and behavioural interventions.
While hospital-based studies have identified risk factors on the severity of an RSV infection-including age, prematurity and underlying immune, pulmonary or cardiac disease [8,9]-these studies focused on the most severe infections while underestimating the substantial burden of disease within the community.There are few epidemiological and environmental community-based studies [4][5][6] and even fewer studies from LMIC countries studies identifying risk factors for an RSV infection [7,10].Previous studies have identified factors such as crowding, day-care, birth in the first half of the RSV season and tobacco smoke exposure that increase the severity of RSV infections [11].
The objective of this study was to identify demographic, socioeconomic, and environmental factors in India associated with developing RSV LRIs in children <2 years of age in a rural community with a high infant mortality rate and poor access to care.

Study design
This was a prospective active community and hospital surveillance study conducted in Melghat, India, between 1 September 2016 and 31 December 2020.The methods, described in detail elsewhere [2,12] are briefly summarised below.

Study population
Melghat is located in central Maharashtra, a forested hilly area with a predominantly rural community.Facility-based medical care for the 93 participating tribal villages is provided mainly by five primary health centres, a charitable trust Hospital (MAHAN) and two government district/ subdistrict hospitals.

Patient and public involvement
Previous survey-based analysis in Melghat performed in 2002-2003 indicated a very high mortality in children <5 years.It also revealed that the existing health care systems, and government interventions did not cater to the community's needs and socio-cultural practices, resulting in low health-care seeking behaviour.
Community participation and patient and public involvement (PPI) was critical to understand the causes for the high under five and infant mortality.Gramsabha/ community meetings were held in 2015-2016 with >60% of adults in the villages in attendance, where it was established that addressing the high infant and under 2 years mortality rate was a priority for the community.Hence, we focused on this as a primary outcome measure.The experience, preferences and health needs of community members were incorporated into the study design with study procedures respecting their culture, traditional practices, language and socioeconomic conditions and in close association with the traditional health system (i.e., birth attendants, community leaders, and traditional faith healers).A final study design and methodology was explained to the community by the principal investigator and study team and informed written consent was obtained from community members.

Patient recruitment
Before study start up, clearances were obtained from the Indian Council of Medical Research, Government of Maharashtra, National Institute of Virology Pune, COMIRB of University of Colorado, and MAHAN Institutional Review Board.An initial baseline population census of all households in the study population of 93 villages was conducted, collecting demographic and socioeconomic factors.This census identified all infants and children 0 to 24 months for recruitment into the initial cohort.
The study implemented trained, socially sensitive, female village health workers (VHWs) to recruit study participants from the community where they lived.Informed written consent was obtained from parents of all children <2 years old who participated in the trial, prior to any study related interventions.Initial recruitment occurred in September 2016 and was followed by the ongoing recruitment of all live-born neonates identified through community pregnancy monitoring [11] up until December 2020.
This enrolment strategy, with an initial baseline cohort and ongoing recruitment of all newborns throughout the study period ensured a consistent and representative distribution of enrolled children aged 0 to 24 months throughout the study period.Subjects were followed longitudinally until they reached 24 months of age, migrated permanently out of the study villages, the parents withdrew consent, the subject died or until the end of the study period.

Risk factor survey
At enrolment, a detailed parental questionnaire was administered capturing the child's birth and medical history, household demographics, socioeconomic data and household environmental factors.Data collection included housing characteristics (e.g., cooking fuel, type of walls, sanitation) and household ownership of durable assets (e.g., car, refrigerator, television).

Active surveillance for development of LRI
Trained VHWs were assigned to each village to monitor all enrolled subjects providing active surveillance of LRTIs in the community.The VHWs performed weekly home visits to assess for acute LRTI in subjects and if present, categorised the severity.Identified LRTI cases were assessed by a supervisor with another home visit [2,13].Trained counsellors assigned at the local hospitals and community health centres identified and assessed all enrolled subjects seeking medical care for an acute LRTI.
All VHW's, their supervisors and counsellors were trained, using a modified World Health Organisation (WHO) methodology, in the recognition of WHO defined Pneumonia and the WHO classifications of non-severe, severe and very severe categories of pneumonia [14].A nasopharyngeal swab was obtained in cases of severe LRTI in both the community and in all hospitalised children with an LRTI, since hospitalisation was considered to be a severe form of LRTI in the community.
VHWs were trained to record all respiratory symptoms, using the 2005 WHO classification to identify severe pneumonia.In our analysis, we defined severe pneumonia as the presence of cough or difficulty in breathing and tachypnoea (≥50 breaths per minute for children aged 2-11 months and ≥ 40 breaths per minute for children aged 12-59 months) plus lower chest indrawing, and very severe LRTI defined as the presence one or more of the observable danger signs: an inability to drink; convulsions; unconsciousness; cyanosis.
Since children may be seen successively by healthcare workers with data longitudinally collected, all assessments within 15 days of the initial assessment were grouped together as an episode of illness, using the worst categories of respiratory rate, chest drawing and danger signs to classify the illness.

Nasopharyngeal swab collection and sample processing
Flocked nasopharyngeal (NP) swabs were collected and transported in PrimStore MTM, (Longhorn Vaccines & Diagnostics, Bethesda, MD) to MAHAN hospital where they were stored at 4-8 C. Batches were transported to ICMR-National Institute of Virology Pune and tested for RSV and other respiratory viruses by real time PCR using standardised protocols [15,16].

Anthropometry
Anthropometry data was collected on study subjects on a regular basis during home visits.

Data entry
Data entry operators entered data into an access database with data quality checks performed by the data manager.Any incomplete data forms were sent back to the field for correction.The Indian PI (AS) sent data to UCD for analysis and monthly reviews.

Statistical methods
Data were double-entered into a Microsoft Access database.Statistical analysis was performed using Stata statistical software, version 14.2.Differences in demographics were tested for significance using chi square and t-tests.Univariate logistic regression analyses were performed to find predictors for the first documented RSV LRI.Only univariate predictors with p < 0.20 significance level were considered for inclusion in the logistic multivariate models.For groups of colinear variables, only those with the strongest univariate association were included.A stepwise method was used, starting with the variable with the highest test statistic.The results are reported as adjusted odds ratios (aOR) with 95% confidence intervals (95% CI).
A wealth index score was derived from the household's durable assets and housing characteristics, as described elsewhere [12].It is a composite measure of a household's socioeconomic status, giving a wealth index score which can be used to determine the subject's wealth quintile (poorest/ poor/medium/wealthy/wealthiest) compared to the general rural Indian population.
A mean weight-for age z-score was calculated for each child to provide an overall measure of nutritional status from 0 to 24 months of age that is adjusted for age and sex.Anthropometry data collected during a respiratory illness was disregarded, to ignore acute changes in a child's weight during an illness.

RESULTS
The initial recruitment enrolled 3611 children under two of years age, and a further 9707 newborns were enrolled over the next 4 years.Among these 13,318 subjects, there were 3290 severe or very severe index LRTI cases of which 483 (14.6%) were RSV-associated, with 198/483 (41%) hospitalised.The first index episode of an RSV-associated LRI occurred at a mean age of 9.4 months (range: 0-24), with an overall incidence of 3.6% and 16 RSV+ deaths.
Table 1 shows the demographics of the total cohort and three comparison groups: the 483 RSV-positive severe or very severe LRTI subjects, the 2807 subjects with RSV-negative severe or very severe LRTI and 10,028 control subjects.
Control subjects were defined as children who did not have a severe or very severe LRTI.Overall, 6783 (50.9%) boys were enrolled in the study.Parental characteristics, such as age, education and employment were similar for the three groups.The socioeconomic status of the study population compared to the rural population in India is represented as a wealth quintile distribution, shown in Figure 1.The study population showed more subjects in the lower wealth quintiles with 35.6% in the 'poorest' category and 32.8% in the 'poor' category compared to the 25% expected for each wealth quintile, if the population was representative of the general rural population.
The total cohort had a median weight-for-age z-score of À1.9 (SD = 1.2) with 3680 (27.6%) subjects having a z-core ≤À2 (2 SD below the reference median).The detailed household socioeconomic and environmental characteristics for each group is shown Table S1.For the total cohort, housing consisted of buildings constructed with mud/cow dung/ stone (35.9%) or thatch (33.5%) rather than cement/bricks (28.4%).The housing had flooring that is predominantly mud/cow dung (91.6%) with tin roofing (65.4%).Only 24.8% had their own household water tap and few dwellings had a home toilet (14.2%).The most common source of drinking water was from a public tap (39.9%).The majority of households (81.1%) used a water purification method for drinking water, with chlorination the most common method (59.3%).For cooking, 92.3% of households used firewood.Other fuel sources were gas (4.7%), cow dung (0.6%) and kerosene (0.1%).
Table 2 shows the univariate risk factor analysis for RSV-associated severe or very severe LRTI with the associated odds ratios (ORs).
Significant risk factors include low gestational age at birth, having a weight-for-age z score <À2, the mother or father with primary education or less and the father employed in unskilled labour.Home environment risk factors include the use of wood or kerosene as a cooking fuel, obtaining drinking water from a public tap and having no household toilet facility.The use of a purification method for drinking water was protective.
A multivariate analysis gave adjusted odds ratios (aOR) showing weight-for-age z-score ≤À2, the use of kerosene or wood for cooking, obtaining drinking water from a public tap and low gestational age at birth significantly increased the risk of an RSV LRTI (Table 3).Comparison with non-RSV LRTI showed male sex as an additional risk factor and did not find higher wealth index score to be significant.The analysis highlighted, the risk of kerosene use [aOR = 17.8 (3.0-104.4),( p ≤ 0.001)] for RSV LRTIs.

DISCUSSION
This study, conducted in a poor rural community, enrolled over 13,000 newborns, infants, and children, and followed them with active surveillance for the first 2 years of life.We Birth weight (g) 1.0 (1.0-1.0)NS 1.0 (1.0-1.0)NS Gestational age at birth (weeks) 0.9 (0.9-1.0) ≤0.001 0.9 (0.9-1.0) ≤0.001 Wealth index score-rural 0.9 (0.8-0.9) ≤0.001 0.9 (0.9- Gestational age at birth (weeks) 0.9 (0.9-1.0) ≤0.05 1.0 (0.9-1.0) ≤0.001 Mother-primary education or less 1.1 (0.9-1.4) NS 1.1 (1.0-1.2) ≤0.05 Father-unskilled labour 1.1 (0.9-1.4) NS 1.0 (0.9-1.1) NS identified the use of either kerosene or wood for cooking in the child's household, as a risk factor for RSV LRTI.Infant and child acute respiratory illness has been associated with biomass burning for cooking, in developing countries [17,18], but the association with RSV LRTI has not been well studied.The effect of kerosene fuel and its association with RSV LRTI is also not well defined, but kerosene is often assumed to be "cleaner burning" than biomass fuels, such as wood and dung, and regarded as a "step up the energy ladder" and provide less indoor air pollution in comparison to wood burning.However, the magnitude of emissions and level of indoor air pollution exposure from kerosene combustion can vary greatly, depending on the stove design, operation, and fuel grade [19].Unfortunately, it was beyond the scope of this study to collect this level of detail and obtain levels of indoor air pollution.
Generally, household air pollution has been previously recognised as a risk factor for LRTIs and associated with the substantial burden of disease in LMICs [17].But there is sparse data on the effect of household air pollution on RSV LRTIs and in particular in LMICs, where the factors affecting indoor air quality will differ substantially from an industrialised country.Studies that can quantify the exposure to emissions from both wood and kerosene fuel, providing a dose-based effect on LRTIs in young children are important to guide interventions that reduce risk.A community-based study in Guatemala [20], performed active surveillance for pneumonia in children <18 months while collecting regular blood carbon monoxide levels as a measure of exposure to emissions from wood fuel.They found that providing a chimney stove reduced carbon monoxide exposure by 50% leading to a significant reduction in severe pneumonia, but not for RSV LRTI.
Further support on the connection between air quality and RSV LRTIs comes from studies on outdoor air pollution.A study based in the USA found some evidence of increased risk of bronchiolitis attributable to chronic traffic-derived particulate matter exposure particularly for infants born just before or during peak RSV season [21].Wrotek et al. [22] concluded that air pollutants play a significant role in RSV hospitalizations in Polish children in their multivariate analysis.
It is theorised that ambient air pollution, induces host susceptibility to a respiratory viral infection [21,23].Air pollution may induce proinflammatory mediators and altered immune function that, increase lung susceptibility to virally induced bronchiolitis [24,25].The combustion of both wood and kerosene for cooking has been shown to increase levels of carbon monoxide and fine particulate matter in the air [26].The combustion of kerosene indoors can also expose occupants to air contaminants, including particulate matter (PM), carbon monoxide, polycyclic aromatic hydrocarbons and volatile organic compounds, even during normal operation [23].The fine PM2.5 (particulate matter with an aerodynamic diameter ≤10 and 2.5 μm) emitted from kerosene combustion enables inhaled particles to be deposited in the deep lung [22].As infants and young children breathe faster than adults, they can inhale more pollutants in proportion to their body weight than adults at a time when their immune systems and organs are still developing.
Kerosene is used minimally in this study population (0.1%) but is an important finding as it still has widespread use in many developing countries, replacing the use of biomass fuels, especially in urban populations where electricity and liquefied pressured gas (LPG) can be expensive or unavailable [27].A similar surveillance study in Indonesia, where kerosene is the predominant fuel used for cooking, also found kerosene a significant risk factor for RSV LRTIs (aOR = 2.16) along with other factors that affected indoor air quality, such window ventilation [13].
Analyses of anthropometry data collected throughout the study showed the population to be significantly underweight with a median weight-for-age z-score (WFA) of À1.9 and with 28% scoring ≤À2 z-scores; an indicator of being undernourished.Low body weight showed to be a significant factor in both univariate and multivariate analyses for both RSV associated LRTI and non-RSV LRTI (aOR = 2.2 [1.8-2.6],p ≤ 0.001) and (aOR = 1.7 [1.5-1.8],p ≤ 0.001), respectively, suggesting low body weight has more impact on the risk of RSV LRTI than LRTI due to other causes.This finding is supported by a study in the Philippines [28] but contrasts with studies finding RSV children are relatively better nourished than other pneumonias [29][30][31].Our finding may reflect the overall poor nutritional status of this study population and the lack of large-scale communitybased studies in similar populations.
A somewhat unexpected finding was the association between drinking water source and the decreased risk when a water purification method was used for drinking water.As the RSV pathogen and most other respiratory infections are not contracted through water, these measures are likely proxies for the sanitation level in the household.Obtaining drinking water from a public tap likely correlates with a lack of an at-home water supply and leading to increased difficulty in maintaining basic sanitation at the home.Lack of home water service has been previously associated with higher RSV LRTI hospitalisation rates in children <5 years, in a study in rural Alaska [32], when comparing hospitalisation rates in regions with low and higher proportion of home water service to regions.
Other measures of sanitation; handwashing after using the toilet and having no at-home toilet facility reached statistical significance in the univariate analyses for RSV LRTI but not in the multivariate analyses, adding to our hypothesis that sanitation at home is an important factor.This is supported by a systematic review and meta-analysis, concluding that interventions promoting handwashing with soap can reduce acute respiratory illness in LMICs, and could help to prevent the large burden of respiratory disease [33].Our finding that collecting water from a public tap is a risk factor may also be due to social behaviour, as in this community women will often congregate at the public tap, bringing their young children which would increase their exposure to respiratory pathogens.
The relatively high mean age of the first episode of severe RSV LRTI has been has been reported [12].Most hospitalbased studies show a younger age for severe disease, but active surveillance in the community, captures the entire burden of disease.A community-based study from Indonesia [13] and older community-based studies [34][35][36] showed a similar distribution of disease.Interestingly, we did not find low birth weight to be correlated with increased risk of RSV LRTI, but rather poor infant growth.A large cohort study in Philippines found that poor infant growth increases the risk for severe RSV infection hospitalisation [27].
Risk factors for RSV LRTI identified in other studies [37,38] did not reach statistical significance in our analyses.This includes male sex and other young children in the household.It appears that in this population the identified environmental risk factors play a larger role in increasing the risk of developing an RSV LRTI.
The comparison of RSV LTRI cases with non-RSV LRTI shows which risk factors appear to be specific to RSV.However, non-RSV LRTI cases will, by definition, be due to multiple pathogens, both viral and bacterial and therefore be a less homogeneous group as this should be considered in interpretation of results.
Limitations to our study include the possibility of multicollinearity due to the type of cooking fuel used in households being related to their poverty index score, as kerosene users are likely to reside in lower socioeconomic neighbourhoods than LPG/electricity users [27].Both measures can be seen as proxies for socioeconomic status.Also, while the study collected simple measures of indoor exposure to air pollution and found wood and kerosene fuel to be contributing factors to the risk of RSV LRTI, it is not able to directly quantify the level of emissions contributing to indoor air pollution or the level of outdoor pollution.The nutritional status of children in the different cohorts were compared as mean WFA z-scores representing nutritional status from 0 to 24 months which may fail to reflect the nutritional status of the RSV LRTI children just before infection.

CONCLUSIONS
This study shows that nutritional status and environmental air quality are predisposing factors for developing an RSV LRI, factors that are amenable to environmental and behavioural interventions.
Demographics of study population.
T A B L E 1 T A B L E 2 Univariate risk factors for severe/very severe RSV LRTI and non RSV LRTI.
Multivariate risk factors for RSV LRTI and non RSV LRTI.