Although our study found ARS to be the second most common cause of medical visits (after malaria) in areas around the volcanoes such as the city of Goma, as well as in the highlands between 1500 to 2500 m (asl), we found little if any convincing associations between the incidence of ARS (at least as assessed in our study) and the occurrence or intensity of the Nyamulagira and/or Nyiragongo volcanic eruptions (as assessed by SO2 emissions) during the observation period (2000–2010). However, the absence of a clear associations does not allow us to conclude that the volcanic eruptions did not affect respiratory health in the area.
Our study is the first to have investigated the possible health effects of volcanic eruptions with pronounced degassing of SO2 in the region. A major strength of our study is the use of morbidity data collected over a whole decade among several hundred thousand inhabitants of a large area. Nevertheless, our findings suggest that the impact of volcanic degassing of the Virunga volcanoes on human respiratory health is not straightforward. Similarly to our findings, Tam et al. failed to demonstrate (in a small community of 1957 people) an association between respiratory health problems, such as asthma, chronic persistent wheeze or bronchitis, and chronic exposure to acid fog from a high degassing volcano, the Kilauea on the Island of Hawaii . In a study conducted in the Azores, the authors highlighted that the volcanogenic air pollutants (i.e. non-eruptive active volcanism) have a high potential to cause lung injury in the long term , but detailed studies of chronic respiratory morbidity are difficult to perform in an under-resourced region such as eastern DRC. In Japan, Iwasawa  found that short-term acute exposure concentrations contribute more to irritation symptoms than average exposure concentrations. Indeed, the exposure duration might not be a determining factor because responses occur very rapidly, within the first minutes following the beginning of inhalation; apparently, further exposure does not increase effects . However, a review by Gudmundsson suggested that acute and chronic respiratory effects varied with the ash composition, which differed from volcano to volcano, as well as from eruption to eruption . Thus, Ishigami et al. reported a significant exposure-response relationship between volcanic SO2 and ARS over time among 611 health volunteers newly arriving after 2 years of the Miyakejima volcano eruption in Japan . Wakisaka et al. found an association between monthly reported clinical respiratory cases and levels of SO2 rather than total suspended particles, after adjusting for seasonality . This suggests not only a possible synergistic effect between SO2 and suspended particles , but also the fact that rainy season could potentiate SO2 toxicity even though no adjustment for meteorological factors was performed.
The Nyiragongo eruption of 2002, its worst eruption in living memory, and the associated high peak in the incidence of ARS (Figs. 4 and 5a) deserves a detailed discussion. This catastrophic event, with lava reaching the city of Goma, had serious short-term and long-term impacts on the city’s economy and its environment (vegetation and water) . Its public health consequences have been addressed in a report to the WHO, which described a high number of consultations for eye and respiratory problems in the days following the eruption. The peak of attendances could hardly be linked to the volcanic degassing emitted during this eruption (17th January) because the estimated total SO2 emission (15–48 kT; Table 1)  was a magnitude lower than the typical daily discharge observed during an eruption of the neighbouring Nyamulagira. The 23rd to 26th January morbidity peak was more likely linked to the collapse of the crater floor which occurred during the 22nd January’s night. Then, hot ash fell over the village of Rusayo (8 km SW of the summit) and accumulated to a 10 cm-thick ash layer, whereas lighter ash fell over Goma and Gisenyi . However, as Horwell underlines, the incidence of acute respiratory symptoms (e.g. asthma, bronchitis) varies greatly after ashfalls, from very few, if any, reported cases to population outbreaks of asthma . It is, therefore, difficult to conclude to any relationship. Other gases may have been partly responsible for this peak: abnormal smells of hydrocarbon gas and numerous gas bursts were reported to have occurred in Goma after the eruption and the lava flow invasion, principally on January 20th to 22nd [29, 31]. However, although reports indicate that air quality was affected by smoke and particulate matter in the first days after the eruption, it soon returned to normal .
Moreover, in the HIS database, a peak of attendances was not only observed for respiratory diseases. In fact, fearing a cholera epidemic, a programme of free health care and drug supply was implemented to support the primary health centres in Goma for a six-week period following the 17th January eruption, followed by another six-week period with reduced prices (0.2 US$ instead of 1 US$ for ambulatory care services, including drugs) . As a consequence, the epidemiological surveillance programme showed a large increase in total attendances at the two hospitals and 18 functioning primary health care centres after the eruption. This apparent increase in overall morbidity must be attributed to the free treatment offered in the wake of the 2002 Nyiragongo eruption. Attendances dramatically decreased, when the free health care programme stopped (1st of March) , notably because not only people living within the Goma health district had come to be freely treated, but also those living in bordering health districts . Thus, a higher proportion of patients came from neighbouring districts in January (14.6%) and February (18.6%) than in March (13.6%).
Looking at the Nyamulagira eruptions, the risk of developing ARS during the period of the 2006 event appeared to be high, and increasing with the distance to the volcanoes. Tremendous amounts of volcanic discharges during this short eruption might have quickly travelled westwards [33,34,35,36]. In addition, this area is at a high altitude and this factor could also contribute to the higher IRR registered following the November 2006 eruption. As Delmelle explains , the local topography exerts a strong influence on plume dispersal, and hilltops are particularly prone to fumigation and thus, to high ambient SO2 levels. On the contrary, it could also explain the low IRR obtained for the population living at lower altitudes less than 25 km from the volcanoes. However, more detailed analyses would be required. Our findings could strengthen the role played by anthropogenic factors, as well as by climate, in the amplification of exposure due to volcanic degassing. Indeed, outside of the city of Goma, the degree of poverty is high, which means that poverty and its covariates (such as malnutrition, use of biomass fuel etc.) could be predictors of respiratory diseases and could constitute potential confounding. This has been partially controlled by comparison with the reference year (1999) but could be investigated in more details.
In contrast, the high IRRs registered between 2008 and 2009 are unlikely to have resulted from volcanic emissions, as these two years were free of eruptions. These high IRRs may be attributed to the insecurity that prevailed from mid-2008 to end-2009, leading groups of internally displaced persons (IDPs) to settle around Goma. As observed through the reported ARS cases, during this period, an increased number of people attended health centres outside their area of origin. In addition, during these periods of insecurity, international NGOs started to provide free medical care; anecdotal reports mention an extra work load because many people used health care services, even though their physical or mental conditions did not always require any treatment . The high IRRs reported in 2008 and 2009 could, therefore, be due to the massive presence of IDPs around Goma fleeing from conflict situations. As this security crisis phase was used as the reference period (November 2008 to February 2009), it could in turn explain the “low” ARS IRR of January 2010. This illustrates the complexity of investigating the health impact of eruptive events since our IRRs were sometimes affected by inappropriate reference periods.
In our research, because of lack of data, the possible effects of particulates were not considered, although several studies  identify particulate matter as a major cause of ARS, especially in developing countries, where the main cooking fuel remains wood and charcoal. In addition, our study did not take into consideration the weather variations throughout the year, nor the day and night temperatures fluctuations, which can be strong compared to the relatively stable annual temperatures. Whilst temperatures do not have a direct impact on ARS, people do change their behaviour according to temperature and weather.
We recognize that uncertain quality and completeness of the health data – from collection to encoding – might represent a significant limitation of our study. Although the increased number of registered ARS cases possibly reflects an improvement of the HIS and data collection over time, routine registration was not accurate in every health centre. Moreover, the denominator (population covered by the health centre) was imprecise due to difficult local conditions, such as internal migration because of war and violence . Consequently, we strongly recommend improving the data collection system and monitoring its quality to allow research based on reliable information. Besides, not all people have the capacity to go to health centres when health problems occur, mainly because they lack resources or because of long distances. This leads to incomplete morbidity data in the HIS used in this study. A household-based approach would provide complementary data. This approach should also be associated with actual measurements of the various components of the ground level volcanic degassing in the field. The long-term perspective would involve the combination of SO2 and PM measurements with regularly collected meteorological data, not available today. Lastly, local institutions assessing risks associated with the volcanoes and providing precautionary options remain weak. The study of Cuoco et al.  following the 2010 Nyamulagira eruption suggested that the impact on the environment – through contaminated drinking water (exposure to toxic metals and chemical species) – was substantial and could have consequences in terms of public health, such as renal, gastrointestinal, cardiovascular and neurological diseases, as well as birth defects, thyroid disease, cancers, and osteoporosis. However, as exposure to pollutants is largely beyond the control of individuals, it requires action from the public authorities .