p u b l i c h e a l t h 1 2 8 ( 2 0 1 4 ) 2 8 8 e2 9 6
Available online at www.sciencedirect.com
Public Health journal homepage: www.elsevier.com/puhe
Original Research
Prevalence of asthma and respiratory symptoms in 15e17 year-old Greek-Cypriots by proximity of their community of residence to power plants: Cyprus 2006e07 N. Middleton a,*, O. Kolokotroni a,b, D. Lamnisos a,b, P. Koutrakis c, P.K. Yiallouros b a
Department of Nursing, School of Health Sciences, Cyprus University of Technology, Cyprus Cyprus International Institute for Environmental & Public Health in Association with Harvard School of Public Health, School of Health Sciences, Cyprus University of Technology, Cyprus c Department of Environmental Health, Exposure, Epidemiology & Risk Program, Harvard School of Public Health, Boston, MA, USA b
article info
abstract
Article history:
Objectives: Numerous studies have reported adverse effects of traffic pollution on respira-
Received 10 June 2013
tory health. Exposure to power plants emissions has not been as comprehensively studied.
Received in revised form
The prevalence of asthma and respiratory symptoms was investigated among 15e17 year-
7 November 2013
olds in communities in the vicinity of power plants in Cyprus in relation to the rest of the
Accepted 14 November 2013
island.
Available online 3 March 2014
Study design: Cross-sectional study Methods: Based on responses of 5817 participants to the ISAAC questionnaire, study
Keywords:
outcomes were: active asthma (i.e. report of asthma and current symptoms), inactive
Asthma
asthma and respiratory symptoms without a diagnosis. Associations in terms of the dis-
Respiratory symptoms
tance of the participants’ community to any of the three power plants were investigated in
ISAAC
logistic models before and after adjusting for known confounders.
Air pollution
Results: At 7.4% (95% CI: 4.5, 11.3), the prevalence of active asthma in communities at 5 km
Power plants
of power plants appeared elevated but reduced to national levels of 5% at longer distances. Adjusted odds ratio for active asthma was 1.83 (95% CI: 1.04, 3.24) in the 5 km zone compared to 30 km away. No clear pattern was observed for inactive asthma while the odds ratio of respiratory symptoms in the absence of diagnosis was 0.76 (95% CI: 0.58, 1.01) in the affected communities. Conclusions: Higher prevalence of active asthma was observed in the vicinity of power plants, with no evidence of a distance-response relationship. With less than 5% of this agegroup residing in close proximity to power plants, this corresponds to a small fraction of active asthma attributable to power plant emissions. ª 2013 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved.
* Corresponding author. School of Health Sciences, Cyprus University of Technology, Vragadinou Str, Limassol, Cyprus. Tel.: þ357 25002796. E-mail address: [email protected] (N. Middleton). 0033-3506/$ e see front matter ª 2013 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.puhe.2013.11.004
p u b l i c h e a l t h 1 2 8 ( 2 0 1 4 ) 2 8 8 e2 9 6
Introduction Asthma is now the most common chronic disease in childhood1 with the highest prevalence recorded in affluent Western societies.2 In recent years, trends have levelled off in high prevalence countries, but low prevalence countries are still experiencing rises.3 In Cyprus, even though still relatively low compared to Northern and Western Europe,4 the prevalence of asthma is still on the rise.5 A range of factors related to urban environments and/or pertaining to the ‘Western lifestyle’ have been implicated, including: a more hygienic environment (e.g. fewer infections in early life),6 increasing obesity rates and reduced physical activity,7 caesarean sections,8 indoor environmental exposures (e.g. smoking and house allergens)9,10 and outdoor air pollution.11 With regards to the latter, numerous childhood asthma studies have linked various respiratory health outcomes to motor-vehicle pollution using different exposure measures, such as self-reported traffic density, proximity of residence to main or busy roads, pollutant concentrations or modelled estimates of emissions.12e14 There is also an increasing body of evidence from prospective studies that traffic-related pollution may not only contribute to the exacerbation of symptoms, but early-life exposures contribute to the development of asthma.15,16 While the European Environment Agency (EEA) in their publication ‘Air Pollution in Europe, 1990e2004’ estimated that the energy industry is the main contributor to outdoor air pollution in Europe, followed by the road transport, only a handful of studies have investigated the health risks of those exposed to emissions from power plant emissions. As part of a nationwide survey on the respiratory health of adolescents in Cyprus, the geographical patterning of asthma and related symptoms was examined among 15e17 year-olds and any evidence of excess morbidity in the vicinity of the three power plants on the island was investigated.
289
level of parental education. BF% was measured with BioImpedance Analysers Tanita TBF-300 after breakfast using a standard protocol. Parental education was used as a proxy for socio-economic status since there is currently lack of generally accepted person- or area-based socio-economic indicators in Cyprus. Co-morbidity (i.e. eczema or hay fever) and family history of allergy i.e. report of asthma, eczema or hay fever amongst siblings or parents were also considered.
Geographical data and exposure assessment The main unit for distance calculations was the community of residence. Polygon maps are not available at a finer level of geographical aggregation i.e. postcodes. While in metropolitan areas, communities consist of several postcodes, there is generally a one-to-one match in rural areas. Due to lack of nationwide coverage of GIS data, automated geo-coding of street addresses was not possible.19 Hence Cyprus Postal Services link files were used to assign self-reported postcodes to a community. In case postcode was not reported, street address and community were considered in that order, as it is not uncommon for Cypriots to use unofficial names to identify their community (e.g. church catchment areas). Electrical power on the island is generated by three heavy fuel oil power plants, all situated close to small and otherwise rural communities on the south coast, namely Vasiliko (commissioned in 2000, production capacity 428 MW, generating 50% of total needs), Dekeleia (operating since the 1980s, capacity 360 MW, 40% of total needs) and Moni (in use since the 1960s, capacity 180 KW, 10% of total needs). Zones around the three plants were applied in increments of 5 km up to 15 km. Communities were ascribed to each zone based on the minimum distance to anywhere in that community. Hence, a community would be assigned to the 5-km zone if any of the plants was less than 5 km to anywhere in that community, irrespective of whether the community is not entirely contained within that radius. Two additional zones were considered e communities at 15e30 km and all the rest at distances of more than 30 km (reference group).
Methods Statistical analyses Study population, outcomes and risk factors Eligible participants in a 2006e2007 survey were all 15e17 yearold students attending state or private Lyceums in Cyprus.17 The Greek-version of the ISAAC questionnaire18 was distributed at the school setting and was self-completed by the participants after written parental consent. Using items of the ISAAC core questionnaire, study outcomes were defined as: (i) active asthma: positive report of ‘ever asthma’ and either wheeze and/or night time cough unrelated to colds in the past 12 months, (ii) inactive asthma: positive report of ‘ever asthma’ but negative report of symptoms and, (iii) respiratory symptoms without asthma: no report of asthma but positive report of either wheeze and/or night time cough unrelated to colds in the past 12 months. The following risk factors were considered a priori as potential confounders: age, gender, sibling order, active or passive exposure to tobacco smoke, ownership of pets or other animals in the household or yard, body fat percent and
Two types of analyses were performed: (a) an ecological analysis to depict the geographical variability in the study outcomes and (b) a person-based analysis to investigate evidence of increased morbidity by proximity to any of the plants. For mapping purposes, the prevalence ratio was calculated by indirect standardization as the ratio of observed cases in each community to expected cases, estimated by applying the gender-specific national prevalence to the gender-specific number of participants. Evidence of heterogeneity and/or clustering was investigated in Poisson models with (i) spatially unstructured (i.e. global between-area variability) and/or (ii) structured random effects (i.e. local variability to capture the tendency of neighbouring areas to exhibit similar levels in the study outcome). In the personbased analyses, odds ratios of the three study outcomes by increasing distance from power plants were estimated in logistic regression before and after adjusting for potential
290
p u b l i c h e a l t h 1 2 8 ( 2 0 1 4 ) 2 8 8 e2 9 6
confounders. Robust standard errors were calculated around these estimates in order to correct for intra-district (n ¼ 34) or intra-school (n ¼ 65) correlation. All models were repeated to exclude urban areas so that the communities in the vicinity of power plants are only contrasted to background rural communities in order to avoid any possible masking of the gradient produced by higher prevalence in urban areas. Distance calculations were performed in ArcView and statistical analyses were performed in STATA, with the exception of the Poisson random-effect models which were fitted in winBUGS using the built-in CAR models (conditional autoregressive models) for the spatially structured random effects with firstorder adjacency to define a community’s neighbours (i.e. those sharing a common border).
Results Of the 6393 students who participated in the original health survey, 5817 (91%) were successfully ascribed to one of 248 communities (62% of all communities). As this survey
involved a very narrow age range, several rural communities were not represented while the median number of participants across communities was 5 (min 1, IQR 2e14, max 823). Table 1 presents the characteristics of the participants with regards to geo-coding success. While no statistical difference was observed in terms of level of parental education, the 9% of the participants whose address information was insufficient or incorrect to be geo-coded accurately were more likely to be male and less likely to report co-morbid conditions or family history of allergy. Table 1 also presents the characteristics of those residing in immediate vicinity of the power plants compared to the rest. They were more likely to come from families with lower educational background, to have older siblings suggestive of bigger families and to have higher body fat % measurements but they were as likely to report eczema, hay fever or family history. Fig. 1 shows raw and locally-smoothed maps of prevalence ratios, depicting areas with lower and higher prevalence than the national average in varying lightness of blue and orange respectively. Areas with no participants or those not
Table 1 e Participant characteristics (A) in the overall sample, (B) in terms of geocoding success and (C) in terms of the proximity to any of the power plants. A. Overall
Males Age 16 years >16 years Not specified Older siblings No Yes Not specified Parental education Both primary At least one secondary At least one tertiary Not specified Active smoker Yes Not specified Passive smoking Yes Not specified Animal at home/yard Yes Not specified Body fat %b 10 (M), 16 (F) 10e20 (M), 16e30 (F) 20e24 (M), 30e34 (F) >24 (M), >34 (F) Not measured Co-morbidityc Family historyd of allergy a b c d
B. By geocoding success
C. By proximity to power plants Pa
5 km n ¼ 5561
Available online at www.sciencedirect.com
Public Health journal homepage: www.elsevier.com/puhe
Original Research
Prevalence of asthma and respiratory symptoms in 15e17 year-old Greek-Cypriots by proximity of their community of residence to power plants: Cyprus 2006e07 N. Middleton a,*, O. Kolokotroni a,b, D. Lamnisos a,b, P. Koutrakis c, P.K. Yiallouros b a
Department of Nursing, School of Health Sciences, Cyprus University of Technology, Cyprus Cyprus International Institute for Environmental & Public Health in Association with Harvard School of Public Health, School of Health Sciences, Cyprus University of Technology, Cyprus c Department of Environmental Health, Exposure, Epidemiology & Risk Program, Harvard School of Public Health, Boston, MA, USA b
article info
abstract
Article history:
Objectives: Numerous studies have reported adverse effects of traffic pollution on respira-
Received 10 June 2013
tory health. Exposure to power plants emissions has not been as comprehensively studied.
Received in revised form
The prevalence of asthma and respiratory symptoms was investigated among 15e17 year-
7 November 2013
olds in communities in the vicinity of power plants in Cyprus in relation to the rest of the
Accepted 14 November 2013
island.
Available online 3 March 2014
Study design: Cross-sectional study Methods: Based on responses of 5817 participants to the ISAAC questionnaire, study
Keywords:
outcomes were: active asthma (i.e. report of asthma and current symptoms), inactive
Asthma
asthma and respiratory symptoms without a diagnosis. Associations in terms of the dis-
Respiratory symptoms
tance of the participants’ community to any of the three power plants were investigated in
ISAAC
logistic models before and after adjusting for known confounders.
Air pollution
Results: At 7.4% (95% CI: 4.5, 11.3), the prevalence of active asthma in communities at 5 km
Power plants
of power plants appeared elevated but reduced to national levels of 5% at longer distances. Adjusted odds ratio for active asthma was 1.83 (95% CI: 1.04, 3.24) in the 5 km zone compared to 30 km away. No clear pattern was observed for inactive asthma while the odds ratio of respiratory symptoms in the absence of diagnosis was 0.76 (95% CI: 0.58, 1.01) in the affected communities. Conclusions: Higher prevalence of active asthma was observed in the vicinity of power plants, with no evidence of a distance-response relationship. With less than 5% of this agegroup residing in close proximity to power plants, this corresponds to a small fraction of active asthma attributable to power plant emissions. ª 2013 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved.
* Corresponding author. School of Health Sciences, Cyprus University of Technology, Vragadinou Str, Limassol, Cyprus. Tel.: þ357 25002796. E-mail address: [email protected] (N. Middleton). 0033-3506/$ e see front matter ª 2013 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.puhe.2013.11.004
p u b l i c h e a l t h 1 2 8 ( 2 0 1 4 ) 2 8 8 e2 9 6
Introduction Asthma is now the most common chronic disease in childhood1 with the highest prevalence recorded in affluent Western societies.2 In recent years, trends have levelled off in high prevalence countries, but low prevalence countries are still experiencing rises.3 In Cyprus, even though still relatively low compared to Northern and Western Europe,4 the prevalence of asthma is still on the rise.5 A range of factors related to urban environments and/or pertaining to the ‘Western lifestyle’ have been implicated, including: a more hygienic environment (e.g. fewer infections in early life),6 increasing obesity rates and reduced physical activity,7 caesarean sections,8 indoor environmental exposures (e.g. smoking and house allergens)9,10 and outdoor air pollution.11 With regards to the latter, numerous childhood asthma studies have linked various respiratory health outcomes to motor-vehicle pollution using different exposure measures, such as self-reported traffic density, proximity of residence to main or busy roads, pollutant concentrations or modelled estimates of emissions.12e14 There is also an increasing body of evidence from prospective studies that traffic-related pollution may not only contribute to the exacerbation of symptoms, but early-life exposures contribute to the development of asthma.15,16 While the European Environment Agency (EEA) in their publication ‘Air Pollution in Europe, 1990e2004’ estimated that the energy industry is the main contributor to outdoor air pollution in Europe, followed by the road transport, only a handful of studies have investigated the health risks of those exposed to emissions from power plant emissions. As part of a nationwide survey on the respiratory health of adolescents in Cyprus, the geographical patterning of asthma and related symptoms was examined among 15e17 year-olds and any evidence of excess morbidity in the vicinity of the three power plants on the island was investigated.
289
level of parental education. BF% was measured with BioImpedance Analysers Tanita TBF-300 after breakfast using a standard protocol. Parental education was used as a proxy for socio-economic status since there is currently lack of generally accepted person- or area-based socio-economic indicators in Cyprus. Co-morbidity (i.e. eczema or hay fever) and family history of allergy i.e. report of asthma, eczema or hay fever amongst siblings or parents were also considered.
Geographical data and exposure assessment The main unit for distance calculations was the community of residence. Polygon maps are not available at a finer level of geographical aggregation i.e. postcodes. While in metropolitan areas, communities consist of several postcodes, there is generally a one-to-one match in rural areas. Due to lack of nationwide coverage of GIS data, automated geo-coding of street addresses was not possible.19 Hence Cyprus Postal Services link files were used to assign self-reported postcodes to a community. In case postcode was not reported, street address and community were considered in that order, as it is not uncommon for Cypriots to use unofficial names to identify their community (e.g. church catchment areas). Electrical power on the island is generated by three heavy fuel oil power plants, all situated close to small and otherwise rural communities on the south coast, namely Vasiliko (commissioned in 2000, production capacity 428 MW, generating 50% of total needs), Dekeleia (operating since the 1980s, capacity 360 MW, 40% of total needs) and Moni (in use since the 1960s, capacity 180 KW, 10% of total needs). Zones around the three plants were applied in increments of 5 km up to 15 km. Communities were ascribed to each zone based on the minimum distance to anywhere in that community. Hence, a community would be assigned to the 5-km zone if any of the plants was less than 5 km to anywhere in that community, irrespective of whether the community is not entirely contained within that radius. Two additional zones were considered e communities at 15e30 km and all the rest at distances of more than 30 km (reference group).
Methods Statistical analyses Study population, outcomes and risk factors Eligible participants in a 2006e2007 survey were all 15e17 yearold students attending state or private Lyceums in Cyprus.17 The Greek-version of the ISAAC questionnaire18 was distributed at the school setting and was self-completed by the participants after written parental consent. Using items of the ISAAC core questionnaire, study outcomes were defined as: (i) active asthma: positive report of ‘ever asthma’ and either wheeze and/or night time cough unrelated to colds in the past 12 months, (ii) inactive asthma: positive report of ‘ever asthma’ but negative report of symptoms and, (iii) respiratory symptoms without asthma: no report of asthma but positive report of either wheeze and/or night time cough unrelated to colds in the past 12 months. The following risk factors were considered a priori as potential confounders: age, gender, sibling order, active or passive exposure to tobacco smoke, ownership of pets or other animals in the household or yard, body fat percent and
Two types of analyses were performed: (a) an ecological analysis to depict the geographical variability in the study outcomes and (b) a person-based analysis to investigate evidence of increased morbidity by proximity to any of the plants. For mapping purposes, the prevalence ratio was calculated by indirect standardization as the ratio of observed cases in each community to expected cases, estimated by applying the gender-specific national prevalence to the gender-specific number of participants. Evidence of heterogeneity and/or clustering was investigated in Poisson models with (i) spatially unstructured (i.e. global between-area variability) and/or (ii) structured random effects (i.e. local variability to capture the tendency of neighbouring areas to exhibit similar levels in the study outcome). In the personbased analyses, odds ratios of the three study outcomes by increasing distance from power plants were estimated in logistic regression before and after adjusting for potential
290
p u b l i c h e a l t h 1 2 8 ( 2 0 1 4 ) 2 8 8 e2 9 6
confounders. Robust standard errors were calculated around these estimates in order to correct for intra-district (n ¼ 34) or intra-school (n ¼ 65) correlation. All models were repeated to exclude urban areas so that the communities in the vicinity of power plants are only contrasted to background rural communities in order to avoid any possible masking of the gradient produced by higher prevalence in urban areas. Distance calculations were performed in ArcView and statistical analyses were performed in STATA, with the exception of the Poisson random-effect models which were fitted in winBUGS using the built-in CAR models (conditional autoregressive models) for the spatially structured random effects with firstorder adjacency to define a community’s neighbours (i.e. those sharing a common border).
Results Of the 6393 students who participated in the original health survey, 5817 (91%) were successfully ascribed to one of 248 communities (62% of all communities). As this survey
involved a very narrow age range, several rural communities were not represented while the median number of participants across communities was 5 (min 1, IQR 2e14, max 823). Table 1 presents the characteristics of the participants with regards to geo-coding success. While no statistical difference was observed in terms of level of parental education, the 9% of the participants whose address information was insufficient or incorrect to be geo-coded accurately were more likely to be male and less likely to report co-morbid conditions or family history of allergy. Table 1 also presents the characteristics of those residing in immediate vicinity of the power plants compared to the rest. They were more likely to come from families with lower educational background, to have older siblings suggestive of bigger families and to have higher body fat % measurements but they were as likely to report eczema, hay fever or family history. Fig. 1 shows raw and locally-smoothed maps of prevalence ratios, depicting areas with lower and higher prevalence than the national average in varying lightness of blue and orange respectively. Areas with no participants or those not
Table 1 e Participant characteristics (A) in the overall sample, (B) in terms of geocoding success and (C) in terms of the proximity to any of the power plants. A. Overall
Males Age 16 years >16 years Not specified Older siblings No Yes Not specified Parental education Both primary At least one secondary At least one tertiary Not specified Active smoker Yes Not specified Passive smoking Yes Not specified Animal at home/yard Yes Not specified Body fat %b 10 (M), 16 (F) 10e20 (M), 16e30 (F) 20e24 (M), 30e34 (F) >24 (M), >34 (F) Not measured Co-morbidityc Family historyd of allergy a b c d
B. By geocoding success
C. By proximity to power plants Pa
5 km n ¼ 5561
