Every year, winds from the Sahara-Sahel desert regions transport large amounts of dust around the world. Dust is transported across the Atlantic to the Americas, and across North Africa and the Mediterranean to Europe.1,2 Affected regions show increased ambient air dust concentrations that may last several days. In areas such as southern Europe, Saharan dust events are a recurrent air-quality problem.3
In urban areas, the fine fraction of airborne particulate matter (PM2.5, which is less than 2.5-μm aerodynamic diameter) is generated principally by combustion processes and traffic-related emissions. PM2.5 is associated with adverse health effects, including total mortality, cardiopulmonary mortality, and respiratory diseases.4 There is a growing body of evidence that coarse particles (PM10-2.5 with particulate matter size between 10 and 2.5 μm aerodynamic diameter) may also play a role in generating adverse health effects, independently of PM2.5, although these effects may depend on specific sources and composition.5,6 Saharan dust cover, which often includes a large proportion of PM10-2.5, increased pediatric asthma accident and emergency admissions on the Caribbean island of Trinidad. Saharan dust has also been shown to carry large amounts of biogenic factors, such as microbes and fungus, providing future biologic plausibility for triggering health effects.1,2,7,8
To our knowledge, no study on the health effects of Saharan dust has been conducted in Europe or the United States. The United States has not regulated PM10 since 2006, whereas proposed legislation in Europe would remove limits from so-called “natural” contributions to PM10, such as Saharan dust, under the assumptions that these contributions are harmless. Climate change is likely to increase the frequency, intensity, and distribution of Saharan and other desert dust outbreaks.9,10 Therefore, the health impact of dust outbreaks need to be further explored. This study explored the effects of Saharan dust, on mortality in Barcelona, Spain, especially in interaction with man-made pollution, specifically fine and coarse particulate matter (PM2.5 and PM10-2.5).
The city of Barcelona is located on the northeast coast of Spain. Saharan dust events generally occur 7 to 15 times per year, between spring and autumn, and last between 3 and 5 days. Outbreaks are caused in most cases by an atmospheric depression west or southwest of Portugal or by anticyclonic conditions over Algeria, which induce southern winds over North Africa, the Mediterranean, and the Iberian peninsula.3
We obtained daily mortality data on the Barcelona mortality registry for years 2003 and 2004. The outcome of interest was all-cause natural daily mortality (ICD code, 10th revision, ICD10: A00-R99). Deaths from external causes (including injury, poisoning, and accidents) were not included in the analysis.
Air Pollution and Chemical Composition
Daily aerosol particulate concentrations between March 2003 and December 2004 were obtained from measurements performed at a single urban monitor site in the city of Barcelona (69 m above mean sea level). We used a laser spectrometer to determine real time PM2.5 and PM10 measurements. Data were continuously validated and corrected by collecting multiple PM10 and PM2.5 samples (three 24 hours samples/wk for each size during the study period) using high-volume samplers (30 m3/h) and quartz fiber filters as substrates (Schleicher & Schuell QF20). Daily mass concentrations for coarse particles (PM10-2.5) were obtained by subtracting PM2.5 from PM10. All days for which both PM10 and PM2.5 were available were retained; days on which at least 1 fraction measurement was missing or presented improbable results (ie, PM10-2.5 greater than PM2.5), were discarded.
We obtained chemical composition of PM2.5 and PM10 particles from filters collected approximately once a week during the 2-year period. Samples were analyzed for nonmineral carbon (nmC), total carbon, crustal and marine aerosol elements (sodium and chloride), inorganic secondary components (sulfate, nitrate, and ammonium), and 46 additional metal and trace elements. Analytical techniques included a LECO analyzer for nmC, ion chromatography for anions, colorimetry-flow injection analysis for ammonium, and inductively coupled plasma atomic emission spectroscopy and inductively coupled plasma-mass spectrometry for major and minor elements, respectively. The chemical composition of PM10-2.5 was obtained by subtracting the chemical composition of PM2.5 from PM10.
Saharan Dust Days
We identified Saharan dust outbreaks using a 2-step process. First, back-trajectory analysis (Hysplit model) was performed using information obtained from NRL,11 SKIRON,12 and BSC-DREAM dust maps,13 and satellite images provided by the NASA SeaWiFS project.14 These tools made it possible to identify days on which air masses from the Sahara-Sahel region were transported to Northeastern Spain. Second, days on which air mass transport occurred were classified as Saharan dust days in Barcelona, if levels of PM10 concentrations measured at a reference remote rural monitoring site reached at least 50% of the PM10 levels measured at the urban sampling site in Barcelona. The rural monitoring site (Montseny) is located approximately 60 km north of the city of Barcelona, (700 m above sea level). As for the urban site in Barcelona, we used laser spectrometers to determine real time PM2.5 and PM10 measurements.
Design and Statistical Analysis
The association of daily concentrations of PM2.5 and PM10-2.5 with daily mortality was investigated using a case-crossover design. This design uses the day on which the outcome of interest (mortality) occurs as a case day. Exposure on case days is compared with exposure on days on which the outcome of interest does not occur (control days).15 We selected a time-stratified approach to represent exposure on control days. Control days were selected from the same day of the week, month, and year as case days. With this approach, bias from time trends in the exposure series and from other short-term time-varying confounders was minimized.16
The association between PM and mortality was estimated by odds ratios (ORs) with 95% confidence intervals (CIs) using conditional logistic regression, adjusted for 4-day average temperature and humidity (day of exposure and 3 days before exposure), and for case days that occurred during a bank holiday, flu epidemic week, or heat wave day. Adjustment was accomplished by creating dummy indicators for these variables. Flu epidemic weeks were selected as weeks with incidence rates above basal levels based on local information.17,18 Heat wave days were defined as days between the periods of 10 June to 1 July and 8 July to 30 August with daily average temperatures above 30°C. Between June and August 2003, record-breaking high temperatures were reported across Europe including Spain.19 We performed sensitivity analysis to assess the impact of changing the definition of heat wave days, and of adjusting models for exposure to daily 24-hour maximum and 8-hour mean ozone levels measured at 2 monitoring stations in Barcelona (Eixample and Gracia Sant Gervasi).
The effects of exposure to PM2.5 and PM10-2.5 were examined for the same day (lag 0) to 4 days after exposure (lag 4). Effect modification by Saharan dust outbreaks was examined by creating a dummy variable for the presence or absence of Saharan dust at exposure days. The potential for effect modification by age was examined in 2 ways: by including an interaction term in the multivariate models for Saharan dust days and by stratification for age. (age <75 years and >75 years). Models that tested the interaction effects of Saharan dust were further adjusted by using both PM2.5 and PM10-2.5 concentrations (2-pollutant models as opposed to single-pollutant models). Comparison of changes in adjusted mass and chemical concentration during Saharan and non-Saharan dust days was carried out by multivariate linear regression.
Air Pollution and Mortality
Air pollution data were retained for 602 days (93% of all days with data). Considering all days, PM2.5 average concentrations were 24.9 μg/m3, and PM10-2.5 average concentrations were 15.1 μg/m3 (Table 1). In comparison, average levels for the same period measured at the rural sampling location (Montseny), not directly influenced by traffic emissions, were 12.7 μg/m3 and 5.6 μg/m3 for PM2.5 and PM10-2.5, respectively. In Barcelona, the correlation coefficient between PM2.5 and PM10-2.5 was 0.34. Our findings are very similar to PM concentrations recorded at urban sites in Barcelona since 1999.20
During the study period, there were 24,850 natural deaths. On average, there were 38.5 deaths per day, with 70% of all deaths occurring among persons age 75 years or older. The average number of deaths per day was similar during Saharan dust days (38.5) and non-Saharan dust days (38.4). Mortality rate increased to 45.6 deaths per day during heat wave days.
Saharan Dust Days
A total of 90 days were classified as Saharan dust days for the study period (Table 1). On Saharan dust days, mean concentrations were 1.2 times higher for PM2.5 and 1.1 times higher for PM10-2.5 than on non-Saharan dust days, with mean concentrations reaching 29.9 μg/m3 for PM2.5 and 16.4 μg/m3 for PM10-2.5 on Saharan dust days. Correlation coefficients were the same for Saharan dust days as those for all days (R = 0.33). The ratios of the mean PM concentration between Saharan dust days and non-Saharan dust days for PM2.5 were 1.2 (case days) and 1.3 (control days). For PM10-2.5, the ratio was 1.1 for both case days and control days. These data suggest that changes in mean PM concentration between Saharan and non-Saharan dust days are similar for fine and coarse fractions.
Total Mortality Risks
In single-pollutant models, exposure to PM2.5 was associated with an increase in mortality for lag 0, lag 1, and lag 2, indicating that effects were observed for the same day to 2 days after exposure (Fig. 1). Increased risks were similar across lags, with the strongest effects observed at lag 1 (OR =1.04 [95% CI 1.023–1.058] per 10 μg/m3). Exposure to PM10-2.5 was associated with a similar increase in mortality, although effects were not quite as large or for PM2.5 and the maximum risk at lag 1 (1.027 [1.008–1.046] per 10 μg/m3) was smaller than that for PM2.5. In 2-pollutant models, exposure to PM2.5 was associated with an increase in mortality at lag 1 (1.032 [1.015–1.05] per 10 μg/m3), as was exposure to PM10-2.5 (1.016 [0.996–1.036] per 10 μg/m3) (Table 2).
Results from the 2-pollutant models (Table 2) show that during Saharan dust days, a daily increase of 10 μg/m3 in PM10-2.5 increased daily mortality by 8.4% (95% CI 1.5%–15.8%) compared with 1.3% (−0.8% to 3.4%) during non-Saharan dust days (P for interaction = 0.05). The increase was smaller for exposure to PM2.5. A daily increase of 10 μg/m3 in PM2.5 increased daily mortality by 5.0% (95% CI 0.5%–9.7%) during Saharan dust days compared with 3.5% (1.6%–5.5%) during non-Saharan dust days(P for interaction=0.56). Results show some evidence of effect modification by Saharan dust days for exposure to PM10–2 5 in all age groups (P for interaction for age group <75 = 0.19 and for age group >75 = 0.14). There was no evidence of effect modification for exposure to PM2.5 (P value for interaction for age group <75 = 0.44 and for age group >75 = 0.86). Figure 2 summarizes the results of the 2-pollutant models for PM10-2.5 and PM2.5 during Saharan and non-Saharan dust days by age group at lag 1.
A total of 89 samples were analyzed for chemical composition of both size fractions. During non-Saharan dust days, 80 samples were collected, and 9 samples were collected during Saharan dust days.
Figure 3 presents the average concentrations of the 4 groups of major elements found in the particulate matter. During Saharan dust days, PM2.5 was dominated by nonmineral carbon (40% of the particulate matter; 11.3 μg/m3) and secondary inorganic aerosols (34%; 9.8 μg/m3), with lesser amounts of crustal elements (23%; 6.5 μg/m3). In contrast, PM10-2.5 was dominated by crustal elements (65%; 13.1 μg/m3), with lesser amounts of secondary aerosols (18%; 3.7 μg/m3), marine aerosols (8%; 1.5 μg/m3), and nonmineral carbon (8%; 1.6 μg/m3). Undetermined mass (mostly attributed to water) accounted for 19% of PM2.5 and 17% of PM10. Comparison of the chemical composition of both size fractions during Saharan and non-Saharan dust days showed that ratios of mass concentrations were near unity for carbon, secondary aerosols, and marine elements, but increased for crustal elements to 1.6 in PM2.5 and 1.5 in PM10-2.5.
Table 3 presents the average mass concentrations for selected metals during Saharan and non-Saharan dust days. During Saharan dust days, average mass concentrations of iron, copper, titanium, phosphorus, manganese, barium, antimony, zircon, and chromium were higher or slightly higher in PM10-2.5 than PM2.5. These metals are typically associated with crustal elements. Data showed that in PM2.5, the ratios of mass concentrations of all metals between Saharan dust days and non-Saharan were above 1, except barium (0.6) and zinc, lead, and arsenic (near 1). In PM10-2.5, ratios of all metals were above 1, except zinc and antimony (near 1). Zinc and antimony are typical indicators of road dust. Data also showed that in PM2.5 crustal elements, iron, titanium, and vanadium concentrations increased substantially during Saharan dust days. In PM10-2.5, major increases were detected only for titanium and manganese. Crustal elements, iron, titanium, and vanadium are typical constituents of Saharan dust.
This study demonstrates that Sahara dust outbreaks increase daily mortality. In particular, results show that the effects of short-term exposure to PM10-2.5 on daily mortality are stronger during Saharan dust days than non-Saharan dust days. The effects of short-term exposure to PM2.5 did not change during Saharan dust days. Lack of statistical power may explain why we found no increased mortality from PM10-2.5 short-term exposure during non-Saharan dust days.
The major strengths of this study are the quantification of independent effects of different PM fractions, the large distribution of the absolute difference between PM concentrations on case days and control days, which allowed for sufficient statistical power to detect marginally significant effects,21 and the restrictive definition used to retain Saharan dust days, which minimized misclassification. Limitations include the low number of Saharan dust days during the study period, which prevented analysis by cause-specific mortality, and the potential confounding effect of the summer 2003 heat wave. Sensitivity analysis showed that changing the definition of the heat wave did not change results. Similar results were obtained when the heat wave was defined as all days from 1 June to 30 August on which the temperatures were above 30°C, or by varying the potential lag of this variable. Sensitivity analysis also showed that incorporating daily 24-hour maximum and 8-hour mean concentrations of ozone into the models did not confound results, with correlation coefficients between the daily mean concentrations of the 2 size fractions and the different metric of ozone levels ranging between −0.24 and 0.15.
Past studies of the health effects of dust in arid regions have produced varied results. A study in Spokane, Washington that examined windblown dust found no association between days with high PM10 and mortality.22 A study in British Columbia that examined the impact of the 1998 Gobi dust event on hospital admissions also found no evidence of an association between PM10 and mortality.23 Two studies in Taiwan and Korea investigated the health effects of Asian dust storms and found an increase of approximately 1% in total mortality per each 10 μg/m3 increase of PM10,24,25 although those studies lacked power.
Additional studies have examined the health effects of various size fractions. An initial study in the Coachella Valley in California found an increase of approximately 1% in total mortality per each increase of 10 μg/m3 PM10.26 Follow-up studies showed an association of PM10 and PM10-2.5 with cardiovascular mortality, and also between PM2.5 and total mortality.27 Another study in the city of Phoenix in Arizona, showed a strong association of PM10 and PM10-2.5 with cardiovascular mortality, and a weak association of PM10 and PM10-2.5 with total mortality.28 A strong association was also found between PM2.5 and cardiovascular mortality. Factor analysis of PM2.5 showed that elements with terrestrial origin (aluminum, silicium, potassium, calcium, manganese, iron, strontium, and rubidium) had no association with cardiovascular mortality, whereas combustion-related pollutants and secondary aerosols did. No factor analysis was performed for constituents of PM10-2.5. These results are intriguing because the studies were conducted in urban cities located in arid regions, where there are both natural and man-made dust, suggesting that sources of dust play a role in influencing health effects, which may be especially relevant in arid regions.
To our knowledge, only 1 study has investigated the health effects of Saharan dust.29 This study found that Saharan dust cover increased pediatric asthma accident and emergency admissions on the island of Trinidad. The measure of exposure was based on dust cover and not PM level. The authors concluded that irritants or allergens in Saharan dust may have caused the increased asthma.
The possibility that Saharan dust may contain irritants or allergens is supported by several studies. A study in the Virgin Islands showed that samples collected during Saharan dust outbreaks carried 3 times more microbes and fungi than normal samples.1 The authors suggested that active biologic agents may be transported in Saharan dust, shielded from inactivation by ultra violet light by attaching to crevasses within coarse particles. Numerous other species of fungi, bacteria, and viruses have also been found in other desert dust samples.2 Toxicology studies have shown that endotoxin and other biologic compounds found in PM10-2.5 activates inflammatory responses.30,31 A study in North Carolina found that ambient PM10-2.5 exacerbated the response of allergic individuals to airborne bacteria; 13% of their mass was composed of pollens, spores, and bacteria.31 In study of 6 European cities, the highest inflammatory effect in Barcelona was for the PM10-2.5 fraction.30 Samples were collected during 2 episodes of Saharan dust events, although biologic composition of the samples was not available.30 Taken together, these studies suggest that increased mortality detected during Saharan dust outbreaks may be related to biogenic factors associated with coarse particles.
Chemical analysis provides further indirect evidence about the health effects of Saharan dust. Analysis showed that chemical composition varied equally in PM2.5 and PM10-2.5 during Saharan dust days. In addition, metals involved in oxidative stress pathways, such as iron, copper, lead, and zinc32 were similarly abundant during Saharan dust days and non-Saharan dust days in PM10-2.5. This comparison indicates that some factor associated with PM10-2.5, though not detected by chemical analysis, may be responsible for increased mortality. Biogenic factors carried by coarse particles in Saharan dust are a possible explanation of observed effects. In addition, there may be other chemicals not measured in this study, such as pesticides or industrial byproducts, transported within Saharan dust. Past studies have shown that soluble and insoluble chemical constituents may affect health in different ways,33,34 and these constituents may vary with different size fractions.35 In our study, only total metal mass amounts were available for analysis. Further research is needed to investigate the biologic and chemical composition and associated allergenic and inflammatory properties of Saharan dust, and the effects of Saharan dust on cause-specific morbidity and mortality. Furthermore, to comply with international public health guidelines, man-made sources of particulate pollution may have to be rigorously controlled in arid regions and areas affected by Saharan dust, particularly if dust outbreaks increase with climate change.
We thank Bruno Schull for help with editing the manuscript.
1. Griffin D, Kellogg C, Shinn E. Dust in the wind: long range transport of dust in the atmosphere and its implications for public and ecosystem health. Global Change Human Health
2. Griffin DW. Atmospheric movement of microorganisms in clouds of desert dust and implications for human health. Clin Microbiol Rev
. 2007;20:459 –77; table of contents.
3. Rodriguez S, Querol X, Alastuey A, et al. Events affecting levels and seasonal evolution of airborne particulate matter concentrations in the Western Mediterranean. Environ Sci Technol
. 2003;37:216 –222.
4. Pope CA III, Dockery DW. Health effects of fine particulate air pollution: lines that connect. J Air Waste Manag Assoc
. 2006;56:709 –742.
5. Brunekreef B, Forsberg B. Epidemiological evidence of effects of coarse airborne particles on health. Eur Respir J
. 2005;26:309 –318.
6. Sandstrom T, Nowak D, van Bree L. Health effects of coarse particles in ambient air: messages for research and decision-making. Eur Respir J.
7. Kellogg CA, Griffin DW. Aerobiology and the global transport of desert dust. Trends Ecol Evol
8. McCarthy M. Dust clouds implicated in spread of infection. Lancet
9. IPCC. Intergovernmental Panel on Climate Change. Climate Change 2007: Impacts, Adaptation and Vulnerability. Summary for Policymakers; 2007.
10. Prospero JM, Lamb PJ. African droughts and dust transport to the Caribbean: climate change implications. Science
. 2003;302:1024 –1027.
11. NRL. Naval Research Laboratory. Available at: http://www.nrlmry.navy.mil/aerosol
. Accessed February 15, 2007.
12. SKIRON. Available at: http://www.forecast.uoa.gr
. Accessed February 15, 2007.
13. DREAM-BSC. Dust Regional atmospheric Model-Barcelona Supercomputing Center. Available at: http://www.bsc.es/projects/earthscience/DREAM
. Accessed February 15, 2007.
14. NASA. National Aeronautics and Space Administration. SeaWIFS project. Available at: http://seawifs.gsfc.nasa.gov/SEAWIFS
. Accessed February 15, 2007.
15. Sunyer J, Basagana X. Particles, and not gases, are associated with the risk of death in patients with chronic obstructive pulmonary disease. Int J Epidemiol
. 2001;30:1138 –1140.
16. Janes H, Sheppard L, Lumley T. Case-crossover analyses of air pollution exposure data: referent selection strategies and their implications for bias. Epidemiology
17. Actividad de la gripe en la temporada 2003–2004 en España y en Europa. Actualización. Boletín Epidemiológico Semanal. España, Centro Nacional de Epidemiología. 2003;11:325–336.
18. Larrauri A, Oliva J, deMate S. Vigilancia de la gripe en España con sistemas centinela. Actividad gripal en la temporada 2004–2005. Boletín Epidemiológico Semanal. España, Centro Nacional de Epidemiología. 2005;13:205–216.
19. Simón F, Lopez-Abente G, Ballester E, et al. Mortality in Spain during the heat waves of summer 2003. Euro Surveill
. 2005;10:156 –161.
20. Querol X, Alastuey A, Viana MM, et al. Speciation and origin of PM10 and PM2. 5 in Spain. J Aerosol Sci
21. Kunzli N, Schindler C. A call for reporting the relevant exposure term in air pollution case-crossover studies. J Epidemiol Community Health
22. Schwartz J, Norris G, Larson T, et al. Episodes of high coarse particle concentrations are not associated with increased mortality. Environ Health Perspect
23. Bennett CM, McKendry IG, Kelly S, et al. Impact of the 1998 Gobi dust event on hospital admissions in the Lower Fraser Valley, British Columbia. Sci Total Environ
. 2006;366:918 –925.
24. Chen YS, Sheen PC, Chen ER, et al. Effects of Asian dust storm events on daily mortality in Taipei, Taiwan. Environ Res
25. Kwon HJ, Cho SH, Chun Y, et al. Effects of the Asian dust events on daily mortality in Seoul, Korea. Environ Res
26. Ostro BD, Hurley S, Lipsett MJ. Air pollution and daily mortality in the Coachella Valley, California: a study of PM10 dominated by coarse particles. Environ Res
27. Ostro BD, Broadwin R, Lipsett MJ. Coarse and fine particles and daily mortality in the Coachella Valley, California: a follow-up study. J Expo Anal Environ Epidemiol
28. Mar TF, Norris GA, Koenig JQ, et al. Associations between air pollution and mortality in Phoenix, 1995–1997. Environ Health Perspect
. 2000; 108:347–353.
29. Gyan K, Henry W, Lacaille S, et al. African dust clouds are associated with increased paediatric asthma accident and emergency admissions on the Caribbean island of Trinidad. Int J Biometeorol
30. Happo MS, Salonen RO, Halinen AI, et al. Dose and time dependency of inflammatory responses in the mouse lung to urban air coarse, fine, and ultrafine particles from six European cities. Inhal Toxicol
. 2007;19: 227–246.
31. Alexis NE, Lay JC, Zeman K, et al. Biological material on inhaled coarse fraction particulate matter activates airway phagocytes in vivo in healthy volunteers. J Allergy Clin Immunol
. 2006;117:1396 –1403.
32. Schlesinger RB, Kunzli N, Hidy GM, et al. The health relevance of ambient particulate matter characteristics: coherence of toxicological and epidemiological inferences. Inhal Toxicol
33. Imrich A, Ning Y, Kobzik L. Insoluble components of concentrated air particles mediate alveolar macrophage responses in vitro. Toxicol Appl Pharmacol
. 2000;167:140 –150.
34. Roberts JR, Taylor MD, Castranova V, et al. Soluble metals associated with residual oil fly ash increase morbidity and lung injury after bacterial infection in rats. J Toxicol Environ Health A
35. Heala MR, Hibbs LR, Agiusb RM, et al. Total and water-soluble trace metal content of urban background PM10, PM2. 5 and black smoke in Edinburgh, UK. Atmos Environ