Relationship Between Ambient Sulfur Dioxide Levels and Neonatal Mortality Near

We examined the association between neonatal mortality and ambient sulfur dioxide (SO2) levels in the neighborhood of Mt. Sakurajima, Yamashita public health district of Kagoshima City, during the period between 1978 and 1988. The analysis using Poisson regression models showed that the monthly average level of SO2 was positively associated with the neonatal mortality (P = 0.002). When the SO2 levels were categorized into four groups to estimate the relative risk (RR) of neonatal mortality using the lowest exposure category as a reference, the RR increased with elevated exposure levels (P for trend < 0.001) and was the highest in the group with the highest level of exposure (RR = 2.2, 95% confidence interval; 1.2-4.1). Other than SO2, we also examined the number of eruptions, the amount of ashfall, and the average level of suspended particulate matter. None of these factors was associated with neonatal mortality. Although the present study suggests that increase in SO2 levels has had an adverse effect on neonatal mortality in the neighborhood of Mt. Sakurajima, it is difficult to determine the source of the SO2. Further studies are necessary to elucidate the mechanisms of the excess neonatal mortality probably associated with the volcanic SO2 levels. J Epidemiol, 1999 ; 9 : 344-349

We examined the association between neonatal mortality and ambient sulfur dioxide (SO2) levels in the neighborhood of Mt.Sakurajima, Yamashita public health district of Kagoshima City, during the period between 1978 and 1988.The analysis using Poisson regression models showed that the monthly average level of SO2 was positively associated with the neonatal mortality (P = 0.002).When the SO2 levels were categorized into four groups to estimate the relative risk (RR) of neonatal mortality using the lowest exposure category as a reference, the RR increased with elevated exposure levels (P for trend < 0.001) and was the highest in the group with the highest level of exposure (RR = 2.2, 95% confidence interval; 1.2-4.1).Other than SO2, we also examined the number of eruptions, the amount of ashfall, and the average level of suspended particulate matter.None of these factors was associated with neonatal mortality.Although the present study suggests that increase in SO2 levels has had an adverse effect on neonatal mortality in the neighborhood of Mt.Sakurajima, it is difficult to determine the source of the SO2.Further studies are necessary to elucidate the mechanisms of the excess neonatal mortality probably associated with the volcanic SO2 levels.J Epidemiol, 1999 ; 9 : 344-349 neonatal mortality, sulfur dioxide, Mt.Sakurajima, volcano Mt.Sakurajima in Kagoshima, Japan, is one of the most active volcanoes in the world.In the 1970s and 1980s, the activity of the volcano was at its highest in the last 50 years.The downtown area of Kagoshima City, with a population of 550,000, is only 10 km away from Mt. Sakurajima, and the average monthly amount of ashfall in the area was 485 g/m2 in the 1980s.Although numerous studies have been conducted to examine the health effects of the volcanic activity in the neighborhood of Mt.Sakurajima 16), there is no clear evidence indicating that Mt.Sakurajima's ashfall causes serious health problems.Among the dust and gases disgorged by Mt.Sakurajima, ambient sulfur dioxide (SO2) is the substance that is most strongly suspected of having serious health effects.Wakisaka et al. examined the weekly numbers of all causes of death among residents of all ages in the neighborhood of Mt.Sakurajima from 1982 to 1984 and reported a sudden increase in the number of deaths occurring during the week immediately after several days when the average SO, levels per hour were 0.2 parts per million (ppm) or higher 6).Since it is possible that neonates are more susceptible to the detrimental health effects of volcanic activity than adults, we examined the relationship between neonatal mortality and SO2 levels in the neighborhood of Mt.Sakurajima.

SUBJECTS AND METHODS
We studied neonatal mortality in the 1970s and 1980s, when Mt. Sakurajima was at its most active during the last four decades (Figure).We obtained monthly data on live births and neonatal deaths occurring within the first 28 days of life from the annual statistics report published by the Yamashita health center in Kagoshima City 7).We calculated the neonatal mortality per month as follows: neonatal mortality in the X,6 month = neonatal deaths in X,1, --{[sum of live births in (X-1)th and X,h months] /2).
The number of Mt.Sakurajima eruptions per month was obtained from annual reports published by the Kagoshima Prefecture municipal government.The amount of ashfall, measured at a monitoring station, were also obtained from the Kagoshima Prefecture municipal government.Data for SO2 and suspended particulate matter (SPM) levels measured at automatic monitoring stations were obtained from the General Air Pollution Monitoring Station, Report of Monitoring ').SO2 was measured using the conductimetric method.SPM was measured using the light-scattering method from 1979 to 1984, and the *-ray absorption method was used after 1984.The interchangeability of data obtained from these two methods was confirmed.We used the monthly average and maximum hourly average levels of SO,_ and SPM in our study.
The distributions of the amount of ashfall and the number of eruptions per month were positively skewed with a long upper tail.In the correlation analysis, the amount of ashfall, the num-ber of eruptions, and the monthly levels of SO2 and SPM were categorized into quintiles, on the basis of their frequency distributions, and Kendall's rank correlation coefficients were calculated 9).We calculated maximum likelihood parameter estimates of relative risk (RR) in Poisson regression models 10) and Waldtype confidence intervals (CIs).All P values were two-sided.

RESULTS
Table 1 shows yearly and seasonal data for the mean monthly ashfall and monthly average levels of SO, and SPM measured in the Yamashita district, the downtown area of Kagoshima City, during the study period.Kendall's rank correlation coefficients among these parameters are shown in Table 2.The level of SO2 was not related to the number of Winter is December , January, and February.eruptions or the amount of ashfall.However, the SO2 level was significantly related to the SPM level.
The neonatal mortality in Yamashita district, which was 3.3 per 1,000 births during the study period, was not associated with gender, calendar year, or season (Table 3).As shown in Table 4, neonatal mortality was associated with the monthly average level of SO2 (P = 0.002), but not with its maximum hourly average levels.The amount of ashfall, the number of eruptions, and the SPM levels did not affect neonatal mortality.We categorized the monthly average levels of SO2 into four groups, and we estimated the RR of neonatal mortality by using the lowest exposure category (Group 1) as a reference  (Table 5).The RR of neonatal mortality increased with the monthly average level of SO2(P for trend < 0.001).The months of highest exposure (Group 4) had the highest RR of neonatal mortality among the four groups (RR = 2.2; 95%CI, 1.2-4.1).These results did not change significantly after adjusting for either gender, calendar year, or season.We could not examine the causes of neonatal death because the published data lacked this information.

DISCUSSION
Among the substances emitted by Mt.Sakurajima, SO2 is most strongly suspected of having serious health effects.Wakisaka et al. reported a sudden increase in deaths during the week immediately after several days when the average SO2 level per hour was 0.2 ppm or higher 6), suggesting that the health effects of SO2 occurred immediately after the exposure to SO2.Since it is possible that neonates are more susceptible to the detrimental health effects of volcanic activity than adults, we examined the relationship between the monthly neonatal mortality and SO2 levels in the corresponding month in the vicinity of Mt.Sakurajima.Although it would have been desirable to use week-specific mortality data in order to confirm the results reported by Wakisaka et al., those data were not available to us.There was a statistically significant relationship between neonatal mortality and the monthly average SO2 levels in Yamashita district, where residents experienced unpleasant effects of Mt.Sakurajima's volcanic activity.On the other hand, there was only a weak and statistically insignificant association between monthly infant mortality and SO, levels (data not shown).Hence, we conclude that exposure to SO2 during the early stages of life causes some acute effects on neonatal health, and results in elevated neonatal mortality.
Wakisaka et al. reported the effect of SO2 on the mortality of adults in Kagoshima Prefecture 2).Although they investigated mortality for all causes, asthma, bronchitis, emphysema, pneumonia, and acute bronchitis, only bronchitis showed a positive association with SO2 levels.Lacking published cause-specific mortality data for neonates, we could not study this finding in greater detail.Further work related to cause-specific neonatal mortality is needed.According to published data on infant deaths, the proportions due to pneumonia or bronchitis and the deaths due to all causes from 1978 to 1988 in Yamashita district and Japan were similar, 4.6% and 4.4%'• 11), respectively.
Although the number of live births is usually used as the denominator when calculating neonatal mortality, we used the mean of live births in (X-1)th and Xth, months.When defining mortality, we assumed that the denominator is the number of people at risk during the period, this is usually the population at the midpoint of the study period 12).Therefore, we used the mean of live births in the (X-1),h and X,h months to approximate the number of live neonates at the midpoint of the X,h month.Conventional calculation gave us results similar to those obtained from our calculation.We examined only the period from May 1978 to December 1988, because official records of ashfall began in May 1978, and sex-specific neonatal mortality data were not available after January 1989 .High levels of ambient SO2 cause various health problems in children 13.14).However, only a few studies have reported an association between SO2 levels and the mortality of children and infants.One such study by Bobak et al .found a weak positive association between neonatal mortality and SO2 levels 15) .They have reported that the crude RR of neonatal mortality in the highest exposure group (SO2 > 0 .02ppm) is 1.20 (95% CI: 1.03-1.39),using the lowest exposure category as a reference .
Our study is the first to suggest that SO, levels are related to neonatal mortality in the vicinity of a volcano .Neonatal mortality in the Yamashita district during the study period was 3 .3per 1,000 births, lower than the 4.0 per 1,000 for Japan as a whole7, 16) during the same period .Perhaps the number of risk factors affecting neonatal mortality in the study district were relatively few when compared to other areas , and, therefore, the effects of SO, on neonatal mortality became apparent in our study.
Although numerous studies have reported the health effects of Mt.Sakurajima's volcanic activity 1-6) , data on childhood health are limited.Wakisaka et al. reported an increased prevalence of asthma cases at primary schools in the region near Mt.Sakurajima, as ascertained by physical examina-tions17).Recently, Uda et al. reported an increase in allergic rhinitis in children living in the neighborhood of Mt.Sakurajima a.However, the proportion of children with asthma-like diseases did not appear to differ in the ash-exposed and control areas of their study.
Some researchers have reported a positive correlation between SPM levels and infant mortality due to respiratory disease or pneumonia ").We did not observe any significant associations of neonatal mortality with SPM levels or the amount of ashfall.Our observations are consistent with results reported by Yano et al., who have concluded that most ash particulate is too large to be respired and is, relatively speaking, biologically inert).
Although we expected that the levels of SO2 would be associated with the amount of ashfall and the number of eruptions, SO2 levels were associated with neither index.Possible explanations for the above results include: i) volcanic gases are frequently dispersed by the wind and diluted in the atmosphere d ownwind 21-24), since volcanic SO, gases are released from volcanic vents even without eruptions and ashfall; and ii) grains of ash, that are larger and heavier than SPM and volcanic gases fall relatively close to Mt. Sakurajima's crater, whereas the smaller and lighter content from the volcanic vents is carried to places far from the crater by the force of the ejection and by the wind.
An interesting finding in our study was the positive relationship between SO2 and SPM levels.Kinoshita et al. have reported that most of the high SO2 concentration events ( > 0.1 ppm) in Kagoshima City are accompanied by events of high SPM concentration, which they attribute to the volcanic activity of Mt.Sakurajima .Therefore, our observation of a correlation between SO2 and SPM levels indicates that the SO2 levels in this district are affected by Mt.Sakurajima s volcanic activity.However, it is difficult to estimate the magnitude of the contribution of Mt.Sakurajima s volcanic activity to ambient SO2 levels in this area.The monthly average levels of SO2 in Yamashita district are not markedly higher than those in other urban areas in Japan M. Perhaps high SO2 concentration events in Kagoshima City affect the health of the residents in the neighborhood of Mt.Sakurajima, not the monthly average levels.To clarify the health outcomes of SO2 exposure, further studies are needed to evaluate the extent of the SO2 derived from Mt. Sakurajima's volcanic activity.
In conclusion, the results of our study suggest that increases in ambient SO2 levels adversely affect neonatal mortality in the neighborhood of Mt.Sakurajima.However, we plan to conduct further studies to elucidate the mechanisms of excess neonatal mortality that may be associated with volcanic SO2 levels.
*Spring is March , April, and May.Summer is June, July, and August .Fall is September, October, and November.

Table 2 .
The matrix correlation coefficients of volcanic activity *The number of eruptions per month , amount of asgfall per month, and monthly levels of SO2 and SPM were categorized into quintiles on the basis of their frequency distributions.

Table 3 .
The means of neonatal mortality by sex, calendar year, and season.
*1 Since neonatal mortality in the Xth month = neonatal deaths in Xth* ([sum of live births in (X-1)th and Xth months]/2), the number of births is not an integer.*2 May-December 1978

Table 4 .
The relationship between neonatal mortality and indices.

Table 5 .
Estimated relative risk of neonatal mortality by SO2 level.