Correlation of health morbidity to air pollutant levels in Bombay City: results of prospective 3 year survey at one year.
It is experimentally established that air pollutants like SO2, Ozone, oxides of nitrogen, benzopyrene and suspended particulate matter (SPM) lead to increased morbidity in the respiratory tract., , , , , , , ,  In health surveys, a slight relationship has been observed in the form of nonpersistent cough and sputum between urban-rural environments and air pollutants.,,  For SO2, high levels have been shown to cause greater chronic bronchitis, frequent colds, postnasal discharge, chest colds and a greater decline in lung function." But, in all these studies tobacco smoking was an important interacting factor., ,  However, in a long term study from U.S.A., Ferris et al have demonstrated that lung function is affected by long term SO2 levels beyond 35 µg/m3/day (but not at 10 µg). There was little difference in morbidity between the urban and rural areas except those attributable to smoking. Other workers have found lung function to be higher in rural group as compared to a matched urban group. There is some evidence that tropical weather with higher humidity allows effects of ozone and SO2 to concentrate on the upper airways.
With photochemical oxidants there is a correlation with daily symptoms and there was some evidence of later adaptation.
In Bombay City, a tropical island. showing high SPM and moderately raised SO2 levels, the complaints of the population were suspected to be related to these pollutants. The present survey was undertaken. to relate health morbidity to the prevailing air pollutant levels.
The Bombay city is a north-south island about 40 km long and 2-10 km wide [Fig. 1]. The central parts of the city have been shown to be highly polluted while pockets in the Eastern suburb are moderately polluted. Most areas of the Western suburb have comparatively lower pollution. From these areas, a population census of three communities (viz. -Lalbaug, Chembur and Khar [Fig 1] from reasonably stable enclosed colonies was undertaken [Table - 1].
While for socioeconomic purposes, the communities were broadly comparable, there were small differences. Thus, Lalbaug had older housing, greater congestion and multiple sources of pollution; Khar had more satisfactory housing, sanitation, more persons with higher incomes and hardly any industries; Chembur had multiple petrochemical industries, a thermal power station, but a reasonable standard of housing.
We also studied a rural community 40 km southeast of the city consisting of two agricultural village: and a commercial village [Fig. 1] [Table - 1].
The census was carried out in December 1976 and January 1977, when data regarding age, sex, income, smoking habits, occupation, housing and duration of residence was collected. We missed 41, 27, 50 and 4 families in the respective communities because of nonavailability or refusal. In all, 2514 families (11928 subjects) were enumerated.
The census population was divided age wise [Table 2] as well as income wise as per unit (adult male one unit; adult female 0.8 units; child between 7 and 12 years-0.6 units and child below 6 years-0.4 units) [Table 3]. The surveyed families were divided on the basis of their duration (in years) in the city [Table 3].
Groups were then selected from the census populations in the four communities by computer matching for age, sex, income and duration of residence. We chose family as a unit for matching. A study (index) population was derived from the computer selected families and from some voluntary subjects from the census populations though not selected by computer. After this, a study intake in each of 4 communities began, successively between February and July 1977. A full clinical proforma patterned after the MRC respiratory questionnaire (1971) was filled in for each subject, and details of housing, kitchen fuel, and occupation were collected. Haemoglobin, eosinophil count, urine for sugar, 70 mm radiograph of chest and lung function were done. The latter consisted of FVC, FEV1, MEFR and PEF by Wright's peak flow meter.
The lung function was not measured in many subjects below 10 years and radiographs were not taken in children below 5 years of age. In each area, about 400 subjects maintained daily health records for 2 years regarding frequency of colds, cough, dyspnoea and sickness rates. These subjects were followed clinically and with lung function tests done every 6 months.
Air monitoring was done by locating one (in Lalbaugh and rural community) or two stations in each area in the residential communities at a height of 40-50 feet. In each urban area, SO2 (2 hourly), SPM (24 hourly) and N02 (4 hourly) were measured for full 24 hours every 5th day. Thus we covered every day of the week once every 4 weeks. In the rural area, one week-day was covered every season (4 months) in view of the low level of pollutants. For these measurements standard chemical methods, , ,  were used, taking adequate care for calibration, checking and standardisation. For correlation, monthly arithmetic means were derived as well as yearly averages, for each pollutant. These were then related to the various clinical parameters (standardised for age, sex, income and smoking) and lung function (after standardisation for age, sex and height).
The study (index) population was significantly different from the original census group and also to a degree from the computer selected ones due to the inclusion of 432 voluntary subjects (60, 101, 113 and 158 in respective areas). By matching the inter-community data, differences for age, income and residence were reduced but still remained significant. As rural incomes were not comparable for matching, incomes in Khar and rural groups were treated together. There was a greater proportion of younger subjects in Chembur and of older ones in Khar. The rural area showed a preponderance of young subjects too, as adults possibly migrated to the city. The families were smaller in urban areas especially at higher incomes. [Table - 1] shows the comparison between the census and index groups for the sex ratio.
[Table 2] shows the age distribution of the index population. [Table 3] shows the general characteristics of the index populations such as their duration of residence in the city and the monthly family incomes. [Table 4] shows housing and fuel patterns in the 4 communities. Major differences existed for housing at Lalbaug as compared to Chembur and Khar (p < 0.01). Basically, many (39%) rural houses were built kaccha. There were large differences in the use of wood and coal as kitchen fuel between the rural and 3 urban communities (p < 0.001) [Table 4].
There was preponderance of subjects with a Western Indian origin in all communities but in the rural communities there were hardly any southern Indians; in urban areas, the latter were 18 to 25 per cent. There was a slight excess of workers In dusty jobs in Lalbaug (12.9%) and Chembur (13.5%) as compared to other areas (p < 0.01) and the subjects in Khar had more sedentary habits (p < 0.01). In other parameters, the index populations in all communities were broadly similar.
[Table 3] shows the prevalence of tobacco habit in the males. There were significantly fewer bidee smokers in urban communities and fewer cigarette smokers in the rural community. Ex-smokers in Khar and rural communities were significantly fewer as compared to the other two communities. There were insignificant inter area differences for tobacco chewing and for current smokers. The rural smokers tended to smoke up to shorter butts (p < 0.01).
[Table 5] lists the prevalence of dyspnoea. It was similar in both sexes but in Lalbaug and Chembur females had a slightly higher prevalence. The prevalence was higher at older ages (p < 0.01) particularly in Lalbaug and rural communities. The frequency of paroxysmal dyspnoea was lower in Khar (p < 0.05) but similar in. other areas. The prevalence of exertional dyspnoea was higher in Lalbaug (p < 0.01) while Khar had the lowest prevalence (p < 0.05). Of these with exertional dyspnoea, 89% had gr II dyspnoea by British Medical Research Council criteria. The standardised prevalence-rates for 3 urban communities listed in [Table 5] show similar differences as the unstandardised rates.
[Table 5] also lists the details for chronic cough. The prevalence of dry cough in absence of colds was low. Khar had the lowest prevalence in comparison to other communities at significance levels shown in [Table 5]. The differences remain similar after standardisation. Thus the prevalence of chronic cough was 5.4, 3.0, 1.4 and 3.3 per cent in respective areas. When the prevalence at the age group 1-19 years was split into 0-4, 5-10 and 11-19 year groups, it was seen that it was highest in Lalbaug (6.8%) and lower for Chembur (3.5%) and rural (3.6%) communities up to the age 4 years (p < 0.05).
[Table 5] also shows the annual frequency of common colds and associated cough. The prevalence of frequent (>8) colds was high in Lalbaug (17.2%) and Chembur (21.8%) as compared to other two communities (11%) and the figures after standardisation showed the same trend. There were no differences for age or sex except at age below 5 years. Thus, of the latter group 27.3, 45.4, 13.7 and 22.9 percent in the respective communities had frequent (>8) colds per year. This prevalence was higher than at other ages (p < 0.01). In Chembur and rural areas, even children between 5 to 10 years showed a relatively higher prevalence. The total frequency for associated cough was highest in Lalbaug, followed by Chembur and was lowest In Khar.
[Table 6] shows the prevalences of various minor abnormalities. The complaint of frequent study, chocked nose was commoner in Lalbaug (19%) and Chembur (24%: p < 0.05). A history of cutaneous allergy in the form of eczema, urticaria or rashes was also commoner in these areas (p < 0.05) and lowest in the rural area (p < 0.01). The history of allergy in family showed similar differences. On auscultation, abnormal lung findings like prolonged expiration and foreign sounds were lower in Khar (p < 0.05) only. In all areas, the frequency in females was insignificantly lower.
The clinical history of tuberculosis was more frequent in Lalbaug (2.7%) but for cardiac complaints, Khar showed the highest (3.9%) and rural area the lower (0.5%) frequency. In subjects in whom radiographs were done in respective areas, 1.0, 0.8, 0.2 and 1.0 per cent showed evidence of old or recent tuberculous shadows.
The frequency of blood eosinophilia (> 2000 cells per cmm) was high in rural community (9%) as compared to all urban areas (p < 0.01).
Headache was complained oftener by females of all 4 areas but the prevalence was highest in Lalbaug and lowest in Khar for both sexes. The frequency of eye irritation was lowest in the rural and highest in Lalbaug community (p < 0.05). For chest pain, the prevalence was greater in Lalbaug and Chembur as compared to the other two communities (p < 0.05).
The history of work absentee rates in males for previous one year showed a frequency of 19.2 per cent in Lalbaug, 7.4% in Chembur and 3.1% in Khar (significantly lower at < 0.01). Roughly for two-thirds of these, the duration was about a week.
[Table 7] gives details of clinical diagnoses. The prevalence of chronic bronchitis/asthma was 4.5% in Lalbaug and Chembur, 5% in rural community and 2.3% in Khar. The frequency of tuberculous shadows (possibly active) was 0.8, 0.2, 0.5 and 0.1 per cent in respective communities. The prevalence of eosinophilia was significantly higher (2.5 Jo) in rural community. A previous history of cardiac disease was obtained in 3.5°/o Khar, 2 to 2.6% Lalbaug and Chembur and 0.5% rural subjects. The prevalence of cardiac abnormality on the interrogated evidence was 8.2% in Khar and 2.7% in rural area. General abnormalities were higher (9.1%) in Chembur. The difference from other areas was significant (p < 0.05).
The levels of casual blood pressures in urban males were between 115 and 121 mm.Hg. systolic and between 74 and 77 mm.Hg. diastolic; the same for females were 111-118 mm.Hg. systolic and 72 to 75 mm.Hg. diastolic. The values in rural communities were 107 to 109 mm.Hg. systolic and 72 mm. Hg diastolic for both sexes. The systolic values were significantly lower (p < 0.01). The prevalence of levels beyond 140/90 mm was 4.4, 1.8, 5.9 and 1.6 per cent in respective communities (Khar/others: p < 0.05).
General basic factors and lung function
There were small differences regarding height and weight below the age of 5 years between Chembur and other urban subjects but at later ages all urban groups were similar. The rural subjects had significantly lower height and weight (usually 2-6 cm and 2-10 kg difference respectively).
The total of 4129 study subjects comprised of 3697 randomly selected and 432 voluntary subjects. These two groups were analysed and compared separately for clinical morbidity and lung function. While the voluntary subjects were normal oftener, the differences were insignificant. For lung function after standardisation for age, sex and height, the differences were not clearcut. So random and voluntary groups were analysed together as stated earlier.
The initial lung function was not recorded in 447 subjects due to young age. In 138 (3.3%) subjects, there was refusal to undergo the test [significantly higher proportions (6.3%) in rural area than in urban areas (1.6 to 2.8%) p < 0.05]. Such refusals were less frequent in the later follow-ups. In 141 subjects (3.4%) the test results were unsatisfactory. This proportion of unsatisfactory test results was higher (5.5%) in rural area. At one year, proportions for unsatisfactory tests dropped to 2.8%.
Tables 8 and 9 give details of the initial lung function for all age groups. The values were standardised for age, sex and height for adults above age 19 years, only, as only in adults this can be reliably done. It can be seen from [Table 8] that the values for Khar are higher than for other areas for PEF and MEFR (significant at 5 and 1 per cent levels). In adults too, similar differences exist for both these tests (p < 0.01: [Table 9]). Thus in urban areas lower air pollution levels may be associated with higher lung function at all ages. The values for rural area are lower despite standardisation. It is possible that general nutrition, food and water contamination and housing may have contributed to this difference.
While it is too early to look at sequential changes at one year, it was seen that decline in FVC was smaller for adults of both sexes in Khar as compared to Chembur or Lalbaug (p < 0.05) ; for FEV, the decline was lower for Lalbaug and Khar as compared to Chembur (p < 0.05).
Early trends in clinical morbidity over first year
We have followed 2819 (68%) subjects at 6 months and 2645 (65%) at one year. The major causes of drop outs, except in Lalbaug (where shifting outside the city was important) were unwillingness and long absence from the area. The success in the follow up was 75% in Lalbaug, 69% in Chembur, 71% in Khar and 46% in the rural area.
[Table 10] lists the prevalences in those followed at 6 months (winter) and at one year (summer). It is seen that the prevalences remained broadly similar to those seen initially. But Khar was lower for cough and colds (more so in summer). In all urban areas, the frequency in summer is insignificantly lower. But in rural area, the prevalence for all complaints were similar in both seasons. For frequent (8+) common colds in winter for urban areas. Chembur had the highest prevalence and Khar, Lalbaug were intermediate. In summer, Chembur and Lalbaug showed a higher prevalence as compared to other two areas (p < 0.05). The trends in other abnormalities (not tabulated) also showed Khar to be the best and Lalbaug the worst in summer; but in winter Chembur was the worst.
The prevalences of clinical diagnoses during one year indicated the same trends. The frequency of chronic bronchitis was 4.4-4.6% in Lalbaug and Chembur; in Khar it was 2.3 to 2.8% and for rural area the figure was 6-6.2% Thus Khar had the lowest frequency for bronchitis but for cardiac causes it had the highest (7.8%) prevalence.
When standardised, prevalences in 3 urban areas for these abnormalities were compared for 6 month and one year stage, these inferences and their significances remained unchanged. Thus Khar was better for frequent colds (p < 0.05), chronic cough and dyspnoea (p < 0.01 for both) than Chembur and Lalbaug.
While 300-600 subjects in each areas have maintained a health diary for 1 year, it is too early to correlate the symptom frequency (cough, colds, dyspnoea) with air pollution levels. But trends in morbidity indicate higher levels in Chembur, Lalbaug and the lowest levels in rural area.
Air Pollutant monitoring
From the samples collected for 6-8 days each month, a monthly mean was derived for each station. From this an yearly mean was also obtained. The mean values based on once a month readings for the period 1970-73 are shown in [Table 11]. The levels for 1978 are generally lower.
For SO2 it is seen that the highest levels were in January, February and July (141-164 µg/M3) at Lalbaug and the lowest levels in November (54 µg). At Chembur, the highest levels were seen in June-July (81-105 µg) and lowest in November (25 µg) and March (27 µg). At Khar the lowest level of (8-10 µg) was seen in August while the highest levels were in January-February (57-61 µg). In rural area, mean levels varied from 6-10 µg. In 3 urban areas, 73%(Khar), 54% (Chembur) and 16% (Lalbaug) of the total SO2 values measured were below 35 ug. Above 75 µg level, the respective percentages were 10%, 17% and 58% of samples. The levels above 150 ug were seen in 2, 6 and 20 per cent of samples respectively. These differences among the 4 areas are highly significant.
For suspended particulate matter (SPM) [Table 11] shows the levels to be 241, 221, 189 and 228 µg/day in the respective 4 areas. These differences are insignificant, but Khar tends to have lower levels. The levels were lower at all stations in monsoon. In Lalbaug the levels were higher in January, March (458-518 µg). In Chembur, higher levels were seen in May, June and July (450 513 µg). In Khar higher levels were seen in February-March (361-388 µg). A quantitative estimation of SPM showed low lead or cadmium levels but nitrate or sulphate levels were high, especially in Chembur.
For NO2 all urban areas showed higher levels than in the rural area and in winter these levels were the highest (32-42 µg) for Lalbaug. In Chembur a high level of 31 ug was seen in June while in Khar, levels of 31-36 ug were seen in January or February. Though the pollutant levels were generally lower in 1978 than in 1970-73 as the means derived form the earlier study were based on once a month (too few) readings, this comparison may not reflect a real change.
Air pollutant levels and clinical morbidity Taking standardised clinical prevalences and pollutant levels we attempted to obtain regression coefficients [Table 12]. It is seen that chronic cough with sputum had a significant correlation with all 3 pollutants, particularly with SO2. For paroxysmal dyspnoea, there was a relation only with SPM, but for exertional dyspnoea, there was a significant correlation with all 3 pollutants, which was particularly high for SO2. For frequent colds with cough, the correlation was just significant for NO2 and SPM (r = 0.31 and 0.36 respectively). In comparison for frequent colds alone only NO2 showed some correlation (r = 0.40).
Thus there seems to be a distinct correlation between the increasing pollutant levels (especially S02) and cough or exertional dyspnoea.
When standardised lung functions in urban citizens were compared with those with other Indian norms, from less polluted city, (Madras) or U.S. norms (Ferris et al8), we found that for PEF, Bombay population had lower values by about 65 litres, at all ages, (U.S. and Madras norms were similar) in both sexes. For FEV1 while U.S. norms were higher, values in Madras and Bombay were similar for both sexes. For FVC, for both sexes, Indian norms were lower than the U.S. norms and our study population had lower values particularly in older subjects. Thus, there was a suggestive evidence that the lower values of FVC and PEF in our city's population could be due to increased air pollution.
This is the first comprehensive survey to assess the effects of air pollutants in an urban-rural population in the developing world. This paper lists some of the findings at the initial and one year stages. A significant correlation has been obtained between the respiratory morbidity, lung function and urban air pollutant levels. However, the rural population shows a morbidity similar to a moderately polluted area in several respects. This situation may be as a result of poorer housing, non-existent sanitation, unprotected water supply, poor medical care, use of wood as kitchen fuel and contamination of food and soil with parasites. The rural subjects were smaller in size, had lower lung function for their body stature, had a higher prevalence of cough, dysponea, eosinophilia, parasitism but a lower incidence of common colds, skin rashes, eye irritation, chest pain and cardiac diseases.
Our efforts at matching or selection of the enclosed communities can be criticised but we did reduce the existing differences. An inclusion of 10% voluntary subjects in the study did not seem to affect the conclusions to a significant extent. The study population had 48-58% non-smokers in adult males of age below 45 years in various areas while in older subjects (above 45 years) the proportions were 26-41 per cent. When the lung function and clinical morbidity were compared in urban areas after standardising for height, weight, age and smoking, the differences in the 3 areas persisted. This may indicate a major role of the prevailing differences in the air pollution levels. Slightly lower morbidity during one year of follow up may be due to a greater loss of abnormal subjects.
With significantly lower pollution levels in Chembur, the clinical morbidity is similar in many respects to that seen in Central Bombay. This may be due to the fact that Chembur is a newer suburb and there is known lack of adaptation for SO2 and O3 for clinical symptoms, in newer residents.
An earlier study for 3 years to correlate air pollutants and clinical symptoms by Cassell et al revealed a complex relationship with weather also. Their findings indicated an increased prevalence of common colds with the pollutant profile. Cohen et al could not correlate chronic oxidant levels with clinical or functional status. However, acute or subacute exposure causes these changes in normals , and more so in asthmatic subjects.
A study of daily symptoms in students showed an increase of 16.7% in upper respiratory symptoms due to multiple air pollutants. A similar record kept by nurses showed a relationship to oxidant levels and an increase in headache, eve irritation, chest pain, cough and breathlessness. Such increase has been noted in Bombay suggesting a need for oxidant :monitoring.
In an urban-rural comparison, it has been shown that urban subjects had a lower lung function and higher school absentism. In a study on school children, MEFR at 50% was seen to be correlated to air pollutants; temperature and humidity raised the effects of pollutants on the upper airway. There is evidence that some subjects (especially abnormals) react more sensitively and there may be synergism among various pollutants. Prevailing high humidity, SPM levels and tropical temperatures in Bombay may aid conversion of SO2 to SO3 and show a greater morbidity due to pollutants at relatively lower levels.
While only 1% of dry SO2. may reach beyond oropharynx, a reflex bronchoconstriction caused by it may reduce the midexpiratory flow rates. This is confirmed in our initial values of expiratory flow rates of urban areas, which were related to the levels of air pollution. In relation to SO2, extensive spot studies in U.S.A. have shown that the levels of pollution in Salt lake city were similar to those in Chembur. The prevalence of cough and dysponea in non-smokers were slightly lower in Salt Lake City than in Chembur. There was 40-50% excess in lower respiratory infections and chronic bronchitis in relation to SO2, and SPM levels. In New-York, with SO2 of 51 µg the prevalence of bronchitis was 9-11% in polluted and 2.6% in cleaner areas. The effect of tobacco smoking was twice as that of air pollution. They found that SO2 had a threshold level of 95 µg and SPM of 100 µg. In Bombay, while we have yet to finally work out such levels, our above data suggests that these are likely to be significantly lower.
Levy et al found in Canada that occupation was not as important as age; SO2. and SPM may act synergistically with humidity. This may be the reason for seeing a higher morbidity in Bombay. The use of kitchen fuel may be an important factor in this situation as shown by Melia et al. They found an increase in bronchitis of 2.6%, cough (4.7%), chest colds (3.8%) and dyspnoea (0.9%) with use of butane instead of electricity. This may be a major factor in causing morbidity in our rural area and central Bombay where wood is largely used.
In a study comparable to ours, Ferris et al,  correlated the urban rural morbidity and lung function. In their results, there was no correlation between urban SO2 levels and respiratory morbidity. The rural areas with no pollution had lower (by 1/3) morbidity. In a follow up they found that lung function declined less when SO2 levels decreased from 35 µg to 10 µg. Thus, they concluded that at the latter level, there was no demonstrable effect on lung function but it was evident at the former level. Thus there is suggestive evidence of relationship of health morbidity in urban residents to air pollutant levels, and our study adds to this data. But there is still no convincing evidence in the world literature of a good correlation after accounting for socioeconomic factors and smoking. However, there is sufficient evidence of acute increase in morbidity and deaths with acute rise in SO2 or S.P.M. levels .
By comparing 2 urban areas with raised pollution levels and two areas (one rural and one urban) with a lower pollution level we have obviated many interacting factors. Our population also had a large proportion of non-smokers. Our results do indicate that a big chunk of upper and lower respiratory morbidity may be related to air pollution. The relative importance of other factors like nutrition etc. will be known after a further follow up.
We thank Dr. C. K. Deshpande, Dean, Seth G.S. Medical College and K.E.M. Hospital, Mr. V. D. Desai, Mr. J. R. Patwardhan and Mr. V. B. Shirodkar for the help and guidance. The help of municipal air monitoring cell and NEERI's zonal laboratory in doing the air monitoring is appreciated. Several personnel in field work have been very useful in ensuring good co-operation of the community.