Association of air pollution on birth outcomes in New Delhi – a pilot study on the potential of HMIS data for environmental public health tracking

Abstract

Objectives

The study objective was to assess the gaps in current hospital health management information systems (ie. paper based records of prenatal, delivery, neonatal, discharge data) for environmental studies. This study also considers the feasibility of linking patient-level hospital data with ambient air pollution data recorded in real time by air quality monitoring stations.

Methods

This retrospective hospital based cohort study used a semi-ecologic design to explore the association of air pollution with a neonate’s birth weight and gestational age. Maternal and neonatal data from 2007-2012 were encoded and linked with air pollution data based on distance to the nearest air quality monitoring station. Completeness and accuracy of neonatal anthropometric measures, maternal demographic information, nutritional status and maternal risk factors (gestational diabetes, anaemia, hypertension, etc.) were assessed.

Results

The records of 10,565 births in Sir Ganga Ram hospital in New Delhi were encoded and linked with real time air quality data. These were records of women who reported a New Delhi address during the time of delivery. The distance of each address to all the monitoring stations were recorded. Birth records were assigned pollution exposure levels averaged across records from monitoring stations within 10 kilometers of the address during the pregnancy period.

Conclusion

This pilot study will highlight the potential of hospital management information system in linking administrative hospital record data with information on environmental exposure. The linked health-exposure dataset can then be used for studying the impact of various environmental exposures on health outcomes. Mother’s educational attainment, occupation, residential history, nutritional status, tobacco and alcohol use during pregnancy need to be documented for better health risk assessments or case management. Health institutions can provide data for public health researchers and environmental scientists and can serve as the backbone of an environmental public health tracking system.

Background

With increasing economic development and industrial growth in countries like India, there is an emphasis on understanding the effects of environmental degradation on human health. There is a need for development, standardization, and application of instruments from the systematic evaluation and mitigation of possible adverse health effects due to changes in the ambient environment. According to the World Health Organization (1), 24 percent of India’s burden of disease can be avoided by modifying the environment. Environmental risk factors take a considerable toll on the Indian population with more than 2.6 million deaths per year attributed to them. The Global Burden of Disease project released in 2010 has documented the health losses that are caused by household (HAP) and ambient air pollution (AAP). It is estimated that in India, a million people die prematurely due to HAPs (from using solid cooking fuels) while more than half a million die early due to the air they breathe outside the home. Effective surveillance monitoring that is driven by reliable clinical data are needed to improve preventive measures and public health interventions. Environmental health researchers from the Public Health Foundation of India collaborated with the Neonatology and Obstetrics and Gynaecology Departments at Sir Ganga Ram Hospital to assess the feasibility of using delivery and neonatal outcomes data and air quality data collected through continuous ambient air quality monitoring stations (CAAQMS) operated by the Central Pollution Control Board. The investigators documented the types of data recorded, completeness, accuracy and utility for air pollution and health outcomes research.

Criteria Air Pollutants and Health

Air pollution and exposure to criteria air pollutants such as particulate matter (PM₁₀, PM_2·5), carbon monoxide (CO), sulfur dioxide (SO₂), and nitrogen dioxide (NO₂) have been linked to multiple adverse health outcomes, with cardiovascular mortality being one of the most comprehensively studied aspects. A harmful impact on fetal development in general and on birth outcomes in particular has been suggested to be a result of exposure to air pollutants. Different kinds of exposures and variables have been studied, although most focus on one or more component of ambient air environment.

The groups most vulnerable to air pollution are the elderly, children and pregnant women (2). In India, communities are exposed to high levels of pollution through outdoor pollution, but recent studies show that indoor environments coupled with women’s work may also lead to exposure to high levels of carbon monoxide (CO) and suspended particulate matter (3). “Developing individuals – including embryos, fetuses, newborns, infants, children, and adolescents – have unique and increased susceptibilities to adverse effects from environmental toxicants. This can result from greater exposure to environmental toxicants, increased exposure of individual organ systems, differences in distribution of toxicants, immaturity of metabolic pathways or excretory pathways, alterations in target organ susceptibility, critical periods of development, and a longer life span in which to express illness” (4). We are beginning to see how certain chronic childhood and adult diseases such as respiratory and cardiovascular diseases have their roots in-utero or early childhood (5,6).

Methods

The researchers encoded paper based delivery records stored by the medical records office of Sir Ganga Ram Hospital in New Delhi. The delivery records from 2007-2012 included maternal demographic information, obstetric history, estimated date of delivery, date of delivery, maternal morbidity markers, neonatal information such as birthweight, gestational age, length and head circumference. Continuous ambient air quality data from ten monitoring stations were collected through the CPCBs online data repository [http://www.cpcb.gov.in/CAAQM/frmUserCompReportCriteria.aspx].

The distance of each maternal residential address to the air monitoring stations was determined through Batchgeo, an online mapping software. Addresses within 10 kilometers (km) of monitoring stations were assigned average pollution exposure data from those stations. The monitoring stations recorded varied types of pollutants of interest (PM, SO₂, CO, Ozone, etc.) (Table 1).

Table 1. Distance of Residential Addresses/data points to CAAQMS.

Monitoring Station	Location			Patient addresses within 10 kilometers (km) of monitoring station
	Zone	Latitude	Longitude	2km	5km	10km
Shadipur	Karol Bagh	N28° 39.551′	E077° 09.429′	1201	4511	6186
Dwarka	Nazafgarh	N28° 36.635′	E077° 02.117′	146	802	1730
Punjabi Bagh	West	N28° 39.514′	E077° 09.435′	385	2787	5771
DCE	Rohini	N28° 40.218′	E077° 07.890′	6	167	1103
Mandir Marg	Central Zone	N28° 38.220′	E077° 12.058′	75	1966	5123
Pragati Maidan (ITO)	Central Zone	N28° 37.444′	E077° 14.463′	73	965	4727
IHBAS	Shahadra North	N28° 41.160′	E077° 18.258′	82	492	1067
Anand Vihar	Shahadra South	N28° 38.856′	E077° 18.950′	70	669	1158
Civil Lines	Civil Line	N28° 40.959′	E077° 13.466′	155	1818	4995
RK Puram	South	N28° 36.600′	E077° 02.066′	138	554	2946

Not all criteria pollutants which may impact the health outcomes of the neonates were recorded during the specific time period by all the monitoring stations (during 2007-2012). Only the Pragati Maidan station had the most complete and comprehensive data set among all the stations. Except for a few monitoring stations installed in primary residential communities, most of the stations are located in mixed residential-commercial areas (Table 2). There is a need to install additional stations to increase coverage and produce robust and sufficiently representative data.

Table 2. Pollutant Data Reported by Central Pollution Control Board CAAQM Stations in New Delhi.

Monitoring Station	Pollutants
	CO	O3	SO2	NH3	NO2	NOx	PM10	PM2.5
Shadipur	2008-2012	2008-2012	2008-2012	-	2008-2012	2008-2012	2008 - 2012	-
Dwarka	-	2009-2012	2009-2012	-	2009-2012	2009-2012	2009 - 2012	-
Punjabi Bagh	2011-2012	2011-2012	2011-2012	2011-2012	2011-2012	2011-2012	-	2011-2012
DCE	2006-2012	2006-2012	2006-2012		2006-2012	2006-2012	2006 - 2012
Mandir Marg	2011-2012	2011-2012	2011-2012	2011-2012	2011-2012	2011-2012	2011 - 2012	2011-2012
Pragati Maidan (ITO)	2006-2012	2006-2012	2006-2012	-	2006-2012	2006-2012	-	2006-2012
IHBAS	2009-2012	-	2009-2012	2009-2012	2009-2012	2009-2012	2010 - 2012	-
Anand Vihar	-	2012	2012	2012	2012	2012	-	2012
Civil Lines	2010-2012	2010-2012	2010-2012	-	2010-2012	2010-2012	-	-
RK Puram		2011-2012	2011-2012	2011-2012	2011-2012	2011-2012	2011-2012	2011-2012

Linkage of individual level medical records to air quality data

For this semi-ecological design, we used patient-level hospital records and linked it with area-level air pollution data. Recorded patient addresses were considered as the residence of the mother during the pregnancy period. Using the Batchgeo software, the research team geocoded addresses which helped visualizing these locations on a map and identifying distances in relation to the ten monitoring stations. Records on pollutant concentrations from montoring stations that are within a prespecified distance, e.g. 10 kms, of a particular address will be combined using approprate metrices to represent maternal exposure during each trimester and throughout the pregnancy. Most of the patients came from Karol Bagh, Central and West zones of New Delhi. For this data linkage process, gestational age and date of birth of the neonate and proper residential address data are important variables.

Next steps

After the perinatal and the air quality data are linked, the team will conduct regression analysis to examine associations between levels of pollutant exposures and adverse pregnancy outcomes, such as low birth weight, preterm delivery and small for gestational age. Available data on maternal risk factors (ex. maternal morbidity, cardiac and renal problems, socio-economic status, parity) will be included in the model to control for confounding. Meterological variables, such as temperature and relative humidity, will be accounted for in the modeling exercises.

Results

Maternal nutritional status such as anemia or vitamin deficiencies were recorded, in addition to delivery procedures, maternal morbidity such as infections, peripartum and hypertensive conditions. The study team noted the lack of socio-economic information, maternal anthropometric information such as height and pre-pregnancy weight which exert confounding effect on a neonate’s birthweight. Substance use such as alcohol consumption, smoking behavior or exposure to second hand tobacco smoke during pregnancy were not recorded as well. Based on the encoded delivery data, there were very few missing birthweight and gestational age data. On the other hand, about 20% of the data on neonatal head circumference and length were missing. The documentation of obstetric notes was not standardized which made assessment of its comprehensiveness challenging.

Discussion

Documentation of health determinants such as a mother’s socio-economic position, education level, substance use, exposure to second hand tobacco smoke, and nutritional status are crucial to identifying clinical and environmental covariates that cause negative maternal and neonatal outcomes. The shift from paper based records to electronic health records will lead to better data quality and accuracy once a standardized structure and format is followed. But to undergo this shift without substantial technical support and funding will prove daunting for many health institutions (7). Behavioural changes on the side of hospital staff will also occur given enough time and space. Convergence of information within the hospital through coordinated data management and reporting procedures will result in meaningful use of data.

Conclusion

Linking clinical data with air quality data can become the basis for surveillance systems to track occurrence of environmental exposures to health risks. Potential health outcomes that can be monitored at the community level include hospital visits due to asthma, lead poisoning of children through inhalation or ingestion, or emergency room visits due to heat stress, etc. Such linkages of information including birth outcomes data, identification of underserved areas, or monitoring of patient outcomes and and quality of care are just a few examples of information that can guide and formulate policy and action. Systematic reporting of birth records at the population level will yield important insights into specific risks and outcomes faced by Indian women and their infants. The use of electronic health records offers great potential in linking health care providers and institutions to environmental and public health researchers for understanding disease trends, or documenting health intervention impacts, while maintaining the confidentiality of patients.

Table 3. Mean (Standard Deviation) of Birth Weight and Gestational age by Social and Demographic Characteristics for births in New Delhi, 2007-2012.

Variables	Number of records with birth weight	Number of records with gestational age	Mean birth weight (SD) [grams]	Mean gestational age (SD) [weeks]
Maternal age (years)
20 and below	210	206	2657.4 (522.0)	37.7 (2.0)
21-25	2437	2413	2752.6 (523.2)	37.6 (2.1)
26-30	4786	4742	2837.1 (537.4)	37.5 (2.0)
31-35	2462	2436	2827.9 (559.8)	37.2 (2.1)
36-40	509	507	2764.0 (573.6)	36.9 (2.2)
41-45	40	39	2705.2 (816.3)	36.1 (3.6)
46 and above	79	76	2772.7 (541.7)	37.1 (2.2)
Sex of neonate
Male	5629	5568	2841.5 (559.0)	37.3 (2.1)
Female	4890	4841	2768.1 (523.0)	37.5 (2.0)

List of abbreviations used

PM10, PM2.5	Ultrafine particulate matter with diameter ≤ 10 μm and diameter ≤ 2.5 μm)
CO	Carbon monoxide
SO2	Sulfur dioxide
CPCB-CAAQMS	Central Pollution Control Board - Continous ambient air quality monitoring system
O3	Ozone
NO, NO2, Nox	Nitrogen oxide, nitrogen dioxide, mono-nitrogen oxides