The impact of communicating information about air pollution events on public health

Science of the Total Environment 538 (2015) 478–491

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

The impact of communicating information about air pollution events on public health J. McLaren, I.D. Williams ⁎ Centre for Environment Sciences, Faculty of Engineering and the Environment, University of Southampton, Lanchester Building, University Rd., Highfield, Southampton, Hampshire SO17 1BJ, UK

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• Investigation of relationship between hospital admissions and poor air quality • Air quality forecasting service to reduce preventable hospital admissions evaluated • Impacts of air quality on emergency admissions quantified using relative risks • Pollution episodes cause admissions despite background concentrations in EU limits. • Air quality forecasting service proved ineffective at reducing hospital admissions.

a r t i c l e

i n f o

Article history: Received 12 May 2015 Received in revised form 22 July 2015 Accepted 30 July 2015 Available online xxxx Editor: D. Barcelo Keywords: Air pollution Pollution events Respiratory health Asthma Nitrogen dioxide Ozone Particulates Forecasting Coastal city

⁎ Corresponding author. E-mail address: [email protected] (I.D. Williams).

http://dx.doi.org/10.1016/j.scitotenv.2015.07.149 0048-9697/© 2015 Elsevier B.V. All rights reserved.

a b s t r a c t Short-term exposure to air pollution has been associated with exacerbation of asthma and chronic obstructive pulmonary disease (COPD). This study investigated the relationship between emergency hospital admissions for asthma, COPD and episodes of poor air quality in an English city (Southampton) from 2008–2013. The city's council provides a forecasting service for poor air quality to individuals with respiratory disease to reduce preventable admissions to hospital and this has been evaluated. Trends in nitrogen dioxide, ozone and particulate matter concentrations were related to hospital admissions data using regression analysis. The impacts of air quality on emergency admissions were quantified using the relative risks associated with each pollutant. Seasonal and weekly trends were apparent for both air pollution and hospital admissions, although there was a weak relationship between the two. The air quality forecasting service proved ineffective at reducing hospital admissions. Improvements to the health forecasting service are necessary to protect the health of susceptible individuals, as there is likely to be an increasing need for such services in the future. © 2015 Elsevier B.V. All rights reserved.

J. McLaren, I.D. Williams / Science of the Total Environment 538 (2015) 478–491

1. Introduction 1.1. Background Air pollution has recently been stated as ‘Britain's forgotten health crisis’ (CIWEM, 2013) and is becoming increasingly important for Local Authorities. Air pollution has serious short and long-term health and nuisance effects, particularly for susceptible individuals (Williams and McCrae, 1995; Brunekreef and Holgate, 2002; Bernstein et al., 2004; WHO, 2005) and causes 430,000 deaths per annum in UK urban areas (COMEAP, 1998). A significant AQ-related adverse health impact is asthma, which is a chronic inflammatory disorder of the airways that causes recurrent coughing, wheezing, chest tightness and dyspnea (Balmes et al., 2003). There are 5.4 million people receiving treatment for asthma in the UK, equating to 1 household in every 5 feeling the effects of asthma (Asthma UK, 2014). Similarly, chronic obstructive pulmonary disease

479

(COPD) is disease with significant extrapulmonary effects, characterised by progressive airflow limitation associated with an abnormal inflammatory response of the lung to noxious particles or gases (Rabe et al., 2007). Symptoms include increased breathlessness, persistent cough, frequent chest infections and wheezing. An estimated 3 million people suffer with the disease in the UK (NHS, 2012). Both conditions can result in impaired quality of life, morbidity, and mortality (Jacobs et al., 2001; Ringbaek et al., 2005; Chung and Marwick, 2010). Air pollution can cause the exacerbation of both diseases (Kelly and Fussell, 2011), as well as variables such as temperature and influenza (Johnston et al., 1996; McAllister et al., 2013). Nevertheless a wealth of evidence exists to support the relationship between poor AQ and its effects on asthma and COPD in both the long and short-term (Table 1). The prevalence of both conditions has increased in recent decades (Welte and Groneberg, 2006). This poses implications for an already financially stretched health service, as COPD is the second most common cause for emergency hospital admission and is costly in

Table 1 A summary of the long and short-term effects of ambient air pollutants on asthma and COPD. Author(s)

Andersen et al. (2011)

Atkinson (1999)

Beatty and Shimsack (2014) Delfino et al. (2002)

Halonen et al. (2008)

Hiltermann et al. (1999) Jacquemin et al. (2012). Li et al. (2011) Meng et al. (2013) Mortimer et al. (2002) Nadeau et al. (2010) Ostro et al. (1991)

Pope et al. (1991)

Qiu et al. (2012)

Samoli et al. (2011).

Schwartz et al. (1993) Silkoff et al. (2005) van der Zee et al. (2000) Wordley et al. (1997)

Main findings

Long-term exposure to traffic-related air pollution may contribute to the development of COPD with enhanced susceptibility in people with diabetes and asthma. Significant positive associations were found between all air pollution proxies and COPD incidence. There is a 3% increase in hospital admissions for a 31 μg m−3 increase in PM10. Significant associations were found between asthma visits and NO2 and PM10. No significant associations were found between respiratory complaints and O3. There is a statistically significant increase in children's respiratory treatments with marginal increases of O3. Pollutant associations with asthma symptoms were stronger in subjects not using anti-inflammatory medications. The strongest association between particulate concentrations and asthma symptoms was for a lag 0 1-hr max PM10 measurement. However more robust associations were shown for a 3-day moving average 8-hr max and 24-hr means for PM10. There are positive associations for pooled asthma-COPD hospital visits in the elderly for PM2.5, PM10 and NO2 at short lags. PM2.5 is associated with hospital admissions for asthma in children with a lag of up to 5 days. Ozone exposure produces a significant increase in the sputum levels of a number of inflammatory markers associated with the respiratory tract. There is a robust association between long-term exposure to O3 and PM10 and uncontrolled asthma. There is evidence of significant increases in daily hospital admissions for asthma associated with PM2.5. An increase of 10 μg m−3 of 2-day moving average concentrations of PM10 and NO2 corresponded to a 0.78% and 1.78% increase in COPD mortality, respectively. Each pollutant was associated with increased morning symptoms of asthma in children. Increasing exposure to ambient air pollution is associated with impairment in T-cell function and increasing asthma morbidity. PM2.5 is significantly associated with increasing coughing symptoms in asthmatics. The level of outdoor exposure and exercise intensity strengthened the association between PM2.5 and symptoms. A concentration of PM10 of 150 μg m−3 is associated with approximately 3 to 6% decline in lung function. This association was observed on the current day and lagged day of pollution event. Elevated levels of PM10 pollution are also associated with increases in symptoms of respiratory disease and use of asthma medication. A 10 μg m−3 increment of a 3 day exposure lag was associated with and increase in COPD admissions by 1.76% and 3.43% for NO2 and O3 respectively, all of which were statistically significant. No consistent modifications of weather factors were found for PM10. Generally it was found that gaseous pollutants increased COPD hospitalisations more in the cool season. An increase in concentration of 10 μg m−3 on the same day causes an increase in admissions by 2.54% and 1.10% for PM10 and NO2 respectively. There is a reduction in admissions by 3.07% for O3 over the year, however when this is reduced to the summer months, there is a 9.30% increase with 10 μg m−3 rise in O3. Daily emergency hospital admissions for people with asthma aged b65 were significantly associated with PM10 exposure on the previous day. There are significant detrimental effects on lung function PM10 and NO2 in the evening, which resulted in increasing medication use on the day of the pollution. Large decrements in morning peak expiratory flow in people aged between 50–70 are associated with PM10. 10 μg m−3 increase in PM10 associated with 2.4% increase in hospital admissions and 1.1% in mortality.

Disease studied

Pollutants studied O3

PM10

✓

✓

✓

✓

✓

✓

✓

Asthma

COPD

NO2

✓

✓

✓

✓

✓

✓

✓

✓

✓

✓

✓

✓

✓ ✓

✓

✓

✓

✓

✓

✓

✓ ✓

✓ ✓

✓

✓

✓ ✓

✓

✓ ✓ ✓

✓ ✓

✓

✓

✓

✓

✓

✓

✓

✓

✓

✓

✓

✓

✓

✓

✓ ✓

✓ ✓

PM2.5

✓

✓

✓

✓

✓ ✓

✓

480

J. McLaren, I.D. Williams / Science of the Total Environment 538 (2015) 478–491

terms of hospital care (DH, 2011). Asthma also poses resource pressures, costing the National Health Service (NHS) £1 billion per year (DH, 2012). Therefore it is paramount to reduce the number of hospital admissions, particularly for preventable causes such as air pollution to improve patient health and to relieve economic pressures. Negative impacts are also experienced by biodiversity and vegetation (Emberson et al., 2001; Lovett et al., 2009). This is recognised by the United Kingdom (UK) Government and forms the primary aim of the Air Quality Strategy, required by The Environment Act, 1995 (Defra, 2007), stating that: “…air quality objectives [are to provide] direct benefits to public health… and help protect our environment.” Air quality (AQ) monitoring systems are central to AQ management in the UK, which includes around 300 monitoring sites organised into networks. The largest is the Automatic Urban and Rural Network (AURN), which measures pollutants such as oxides of nitrogen (NOx) (the most important being nitric oxide (NO) and nitrogen dioxide (NO2)), sulphur dioxide (SO2), ozone (O3), and particulates (PM10 and PM2.5) (Defra, 2012). AQ is monitored for source identification (Viana et al., 2014), ensuring legislative compliance (Longhurst et al., 2009), monitoring policies e.g. the London Congestion Charge (Beevers and Carslaw, 2005), identifying trends (Ayres, 1997) and validating models (Oettl et al., 2001). Monitoring is an important requirement of The Environment Act, 1995 in the provision of Air Quality Management Areas (AQMAs) and the 2008 Ambient Air Quality and Clean Air for Europe Directive (2008/50/EC) which has been transposed into UK legislation by The Air Quality (Standards) Regulations, 2010. The regulations require Local Authorities to make AQ information publicly available and implement AQ action plans where AQ exceeds specific limit values. The current state of compliance in the UK varies with pollutants. As of 2012, annual limit values for SO2, PM10, PM2.5 and O3 were compliant. However, 79% of UK zones were non-compliant with the values for NO2 (Defra, 2013a), which highlights continuing difficulties for this pollutant (Williams and Carslaw, 2011). This has subsequently led to a time extension notification to extend the limit value of 40 μg m−3 annual mean deadline from 2010 to 2015. Yet, 17 areas will not achieve this until 2020, and 2025 for London (Barnes et al., 2013). Evidently AQ

management in the UK is not yet fully effective in meeting annual targets, which will inevitably have penal and health implications. There are also short-term implications of AQ as urban areas can experience short-lived episodes of poor AQ particularly at certain wind speeds (Jones et al., 2010) and directions (Bigi and Harrison, 2010). In relation, limits are also on the number of times a threshold concentration is exceeded in a year, as short-term implications on health are experienced during pollution episodes (McCreanor et al., 2007). Health impacts can be reduced if AQ information is made publicly available. This is mandatory, yet historical presentation of information has often been complex, inconsistent and not easily understandable to meet public needs (Beaumont et al., 1999; Lindley and Crabbe, 2004). As a result, public awareness of AQ has been far from universal (Bickerstaff and Walker, 2001). Forecasting services now address these issues, predominantly in relation to susceptible individuals and their health. Defra's air pollution forecast on its website (Defra, 2013b) uses Air Quality Index (AQI) bands that synthesise AQ concentrations into 10 bands coloured in a traffic light system symbolising severity of health impacts (Table 2). Each band has accompanying health advice for the ‘at risk’ and general population. The information is provided to assist the user to avoid adverse health impacts. The service is based on regional forecasting, which may be unrepresentative of local AQ (Kelly et al., 2012). AQ information can be made more accessible and location specific by application software made available on an electronic device (“app”) such as the London Air App and airAlert. Both warn the user of poor episodes of AQ within the local environment. 1.2. airAlert airAlert is a free early warning service informing susceptible individuals of poor episodes of AQ to reduce emergency admissions to hospital. Subscribers receive information via text, phone call, email or internet regarding the severity of pollution (moderate, high or very high — in relation to Defra's AQI), the type of pollutant, location of concern and when the forecast is relevant (airAlert, 2013). Currently the service is only available in certain areas in England (Surrey, Sussex, Southampton and Sevenoaks). Southampton City Council (SCC) have shown evidence for and against increased asthma and COPD prevalence in relation to the proximity of AQMAs and have conducted a qualitative study to assess behaviour changes amongst subscribers during an alert (S. Hartill,

Table 2 Air Quality Index bands for each air pollutant and accompanying health advice (adapted from Defra, 2013b, 2013c). Pollutant and averaging time (μg m−3)

Index band

Low

Moderate

High

Very high

Health advice for

O3

NO2

SO2

PM10

PM2.5

8 hr running mean

Hourly mean

15 minute mean

Daily mean

Daily mean

1 2 3 4 5 6

0–33 34–66 67–100 101–120 121–140 141–160

0–67 68–134 135–200 201–267 268–334 335–400

0–88 89–177 178–266 267–354 355–443 444–532

0–16 17–33 34–50 51–58 59–66 67–75

0–11 12–23 24–35 36–41 42–47 48–53

7 8 9

161–187 188–213 314–240

401–467 468–534 535–600

533–710 711–887 888–1064

76–83 84–91 92–100

54–58 59–64 65–70

N240

N600

N1065

N100

N70

10

At-risk Individuals

The general population

Enjoy your usual outdoor activities.

Enjoy your usual outdoor activities.

Adults and children with lung problems, and adults with heart problems, who experience symptoms, should consider reducing strenuous physical activity, particularly outdoors. Adults and children with lung problems, and adults with heart problems, should reduce strenuous physical exertion, particularly outdoors, and particularly if they experience symptoms. People with asthma may find they need to use their reliever inhaler more often. Older people should also reduce physical exertion. Adults and children with lung problems, adults with heart problems, and older people, should avoid strenuous physical activity. People with asthma may find they need to use their reliever inhaler more often.

Enjoy your usual outdoor activities.

Anyone experiencing discomfort such as sore eyes, cough or sore throat should consider reducing activity, particularly outdoors.

Reduce physical exertion, particularly outdoors, especially if you experience symptoms such as cough or sore throat.

J. McLaren, I.D. Williams / Science of the Total Environment 538 (2015) 478–491

unpublished data). However little research has been done to evaluate the effectiveness of the service in preventing hospital admissions. It is advantageous to evaluate the service due to the potential health benefits and cost savings to NHS resources, which could be extended to other local authorities across the UK, should the service prove effective.

1.3. Aims and objectives Southampton is a coastal transport hub (Fig. 1), including an international airport, deep-water ports and good transport links to London via train and motorway, all of which contribute towards poor AQ in the city. In addition, there are two universities, two major shopping centres and a premier league football team that are of interest beyond the boundaries of Southampton. The city can become very busy, posing AQ problems due to congestion and use of Heavy Goods Vehicles (HGVs), which has caused 10 AQMA's for NO2 within the city (SCC, 2012). The city council is proactive in efforts to improve AQ and reduce the impacts on its citizens e.g., Southampton won Transport City of the Year in 2013 (SCC, 2013), SCC provides the airAlert service and encourages bus patronage and various traffic management schemes (SCC, 2010). This study investigated the relationship between episodes of poor AQ and daily hospital admissions in Southampton as an example of a coastal city. The study of coastal cities is important as over 50% of the world's population live near the coast (Sarkar, 1996). The effectiveness of the airAlert service in reducing daily hospital admissions of asthma and COPD will then be quantified to assess whether other local authorities can implement the service to provide benefits to all susceptible individuals.

481

2. Methodology 2.1. Site description Southampton has failed to meet annual NO2 limit values (Defra, 2013a) and will not do so until 2020 (Defra, 2011). This is due to congestion within the city. Generally, background AQ is within EU limit values, however, the city does experience poor episodes of AQ that are potentially detrimental to health, mainly localised to roadsides (SCC, 2012).

2.2. Data collection Hourly AQ data for 01/04/2008–31/04/2013 was acquired from the Defra website (Defra, 2013d) for NO2, O3, PM10 and PM2.5 for the Southampton Centre AURN site and where available, the Bitterne site (Hantsair, 2013). These monitoring stations are representative of urban background conditions. Meteorological data from Southampton Airport (in Eastleigh, just to the north-east of the city boundary) was obtained from Wunderground (2013). Data included mean daily wind speed and direction, temperature and relative humidity. This data is broadly representative of the conditions experienced at both monitoring stations because the site has no local influences and is within 40 km of the two sites (Met Office, 2010), although we acknowledge that: (a) there are many weather features that occur on smaller scales and (b) in general, the scale a site is representative of is dependent on, for example, the scale of local topography (valley, top of mountain, coast etc.), type or terrain (urban, forest, field etc.). In addition, hourly-modelled wind speed and

Fig. 1. a) The location of air quality monitoring stations within Southampton in relation to the major road network. b) A reference map to show the location of Southampton in the UK and c) the South coast of England.

482

J. McLaren, I.D. Williams / Science of the Total Environment 538 (2015) 478–491

direction for the Southampton AURN site were downloaded where available (Defra, 2013d). Daily hospital admissions data for asthma and COPD (only) was acquired from SCC, which represents the majority subscribers to airAlert. This was totalled to create a data set known as total daily admissions. SCC also provided airAlert dates, however it is not known which pollutant caused the alert to be issued. 2.3. Data analysis The openair package (Carslaw, 2013) for statistical software R was used to analyse trends in AQ. This application allows for detailed analysis tailored to the subject of AQ by the effective manipulation and visualisation of data (Carslaw and Ropkins, 2012). Correlations between AQ and hospital data were produced by a Spearman rank order correlation coefficient using statistical software SPSS. A Poisson regression model was constructed in R based on methodology in de Souza Tadano et al. (2012). It allows for the calculation of the relative risk of hospital admission whilst considering adjustments in confounding variables such as seasonality, meteorological variables and influenza epidemics (Sunyer et al., 1997; Atkinson et al., 1999; Halonen et al., 2008). Failure to account for confounding variables can lead to greater variability and larger standard errors of the relative risk (Schwartz et al., 1993). Confounding variables accounted for here are temporal trends, day of the week, national holidays, temperature, relative humidity and AQ. Age and smoking prevalence are not important as the distribution amongst the population do not vary day-to-day (Schwartz et al., 1996). However, the omission of influenza epidemics due to lack of data may lead to some uncertainties. Evaluation of airAlert was based on the methodology used in the AsPIRe project (Walton et al., 2012) and COMEAP (1998). Each day was classified as low, moderate, high or very high if either monitoring site met the criteria of the AQI for that pollutant. The average concentration of all low days was calculated and used as the baseline in which no negative impacts on health are experienced (Table 1). The average hospital admissions on these days were also calculated and used as the baseline admissions, which makes the results more specific to the study site as local hospital admissions will take into account local variables, which may be excluded by national estimates (used in the AsPIRe project). In addition, for each moderate, high and very high day, the mean concentration was calculated across the two sites. The difference between the baseline concentration on low days for that pollutant and the concentrations on pollution event days (i.e. moderate, high or very high) were then calculated. The difference is then multiplied by the relative risk of hospital admission (which has been scaled for an incremental increase in concentrations by 1 μg m−3). This equals the estimated percentage change of admissions on that pollution event day. This percentage change is then calculated for the baseline admissions on low days to give the change in admissions caused by AQ on pollution event days. The sum of the additional admissions gives the total additional admissions caused by AQ over the study period. The airAlert service was evaluated by calculating the likelihood of an additional admission caused by AQ being a subscriber to the service and comparing days where a pollution event occurred to when an airAlert was issued. 3. Results 3.1. Time variation plots Descriptive statistics of the main data set used in this study show little variation in the AQ and hospital data (Table 3). Generally, daily mean concentrations of pollutants did not deviate much over the study period, although there are clear peaks and troughs for all pollutants corresponding with seasons (Fig. 2). Background AQ concentrations are within the AQ objectives. Fig. 3 displays seasonal trends and provides a finer temporal resolution of pollutant trends, showing diurnal trends

Table 3 Descriptive statistics of the daily mean pollutant concentrations from Southampton AURN monitoring site and daily hospital admissions and meteorological data.

NO2 (μg m−3) O3 (μg m−3) PM10 (μg m−3) PM2.5 (μg m−3) Asthma admissions COPD admissions Total admissions Wind speed (m s−1) Wind direction (degrees) Temperature (°C) Humidity (%)

n

Min

Max

Mean

Standard deviation

1749 1766 1668 1413 1825 1825 1825 1820 1825 1820 1820

10 1 4 3 0 0 0 0 −1 −6 44

106 106 86 80 15 15 25 10.8 360 21 100

34.5 34.8 19.4 14.5 4.3 4.8 9.1 3.1 185.4 9.8 75.4

12.8 16.0 9.9 9.7 2.3 2.5 3.8 1.8 98.4 5.3 10.9

for each pollutant; NO2 and particulates show two peaks in the morning and evening, whereas O3 shows one peak in the evening. These peaks are less pronounced at weekends with the exception of O3. 3.2. Polar plots A negative relationship existed between pollutant concentrations and wind speed for NO2 and particulates i.e. the lower the wind speed, the higher the concentration (Fig. 4). PM10 showed some higher concentrations at high wind speeds from a south-westerly direction. Conversely, O3 concentrations were highest at high wind speeds. Each plot shows that higher pollutant concentrations occurred from differing wind directions with the exception of particulate concentrations, which occurred predominantly from easterly winds. The highest concentrations for NO2 occurred from all directions at low wind speeds and moderate concentrations occurred mainly from westerly and southwesterly direction. There is a multitude of wind directions that produce the highest O3 concentrations. 3.3. Polar annulus plot Seasonal and diurnal characteristics for each pollutant are displayed in Fig. 5. The highest concentrations for NO2 and particulates occurred from a south-easterly and easterly direction during the winter months, whereas O3 showed the highest concentrations from a number of directions during the spring/summer months. Each pollutant had a unique diurnal trend with the exception of particulate matter. Two peaks were associated with NO2 concentrations in the morning and evening, from a number of directions, but most strongly associated with northeasterly winds. One clear peak in O3 concentrations was evident in the afternoon and evening, which was strongest from a southerly wind direction. Particulates showed a trend similar to NO2 with two peaks in the morning and afternoon, but not as defined. The highest concentrations occurred from north-easterly and south-easterly directions in the evening. 3.4. Correlation and regression analyses There is no indication of strong relationships between the maximum AQ concentration with respect to individual averaging times from the AQI occurring in a day and daily hospital admission (Table 4). Both positive and negative relationships existed between the two variables, indicating that hospital admissions both increase and decrease with increasing concentrations. Three significant relationships occur in the low AQI and in general, the correlations at higher AQI bands are relatively stronger than those in the low AQI band. There are two significant Poisson regressions for total admissions with both NO2 and O3 (Table 5). The results are presented with pooled estimates used in the AsPIRe project derived from Anderson et al. (2007). Analysis in this study shows that NO2 had the highest relative

J. McLaren, I.D. Williams / Science of the Total Environment 538 (2015) 478–491

483

Fig. 2. Time series plot to show the daily mean concentrations of pollutants and the number of total respiratory daily admissions for asthma and COPD during 01/04/2008–30/03/2013. The pollutant concentrations are from Southampton AURN background site and show a 95% confidence interval smooth trend line.

risk of 0.9% increase in admissions for an increase of 10 μg m−3. Ozone and PM10 showed negative (i.e. when concentrations increase then admissions will reduce) and no risk relationships, respectively, when concentrations increased by 10 μg m−3. Results for PM2.5 showed a 0.1% increase in admissions when concentrations increased by 10 μg m−3, and are similar to PM10 results when considering the accompanying error values. Particulates displayed the highest error values with no statistical significance, whereas the relative risks for NO2 and O3 contain lower error values and are statistically significant. When comparing results for Southampton with those used in the AsPIRe project there is a small difference between the results for NO2, though the results vary for both O3 and PM10, however, there is some overlap within the error bars for PM10 indicating some similarities (Fig. 6). The relative risks used to calculate the total number of hospital admissions attributable to poor episodes of AQ were those used in the AsPIRe project due to marked differences with this study, which indicate a small, if not positive impact on health, which is evidently not the case. Additionally by using the relative risks of the AsPIRe project, the results are directly comparable.

3.5. Impact of AQ on admissions and effectiveness of airAlert Over the study period, 86 pollution events were recorded at the background sites in Southampton (44 excluding PM2.5) resulting in 40 hospital admissions for asthma and COPD sufferers (22 excluding PM2.5) (Fig. 7). The pollutant responsible for the most admissions was PM2.5; however, airAlerts are not currently issued for this pollutant, therefore by omitting PM2.5 the number of admissions attributable to PM10 increases from 10 to 18 as the analysis equates for double counting (see PM10 on Fig. 7). The most common pollution severity was moderate

with a total of 73 days (43 excluding PM2.5), followed by 12 high (1 excluding PM2.5) and 1 very high day (0 excluding PM2.5). airAlert began on 11/04/2010; therefore not all days in Fig. 7 were issued with an alert. Table 6 displays the data in Fig. 7 pre and postairAlert. The total number of subscribers as of 15/05/2013 was 197, which is assumed throughout the study period. The number of subscribers likely to be one of the 17 admissions post-airAlert, is calculated by first estimating the prevalence of subscribers within the respiratory disease population in Southampton. This was calculated by dividing the number of subscribers (197) by the number of people with asthma or COPD in Southampton (34,200 using national asthma Asthma UK, 2014 and COPD NHS, 2012 rates relative to the Southampton population SCC, 2014). This is a 0.6% prevalence, meaning that 0.1 people out of the 17 admissions post-airAlert were likely to be a subscriber, i.e. there is a 10% chance of an admission over the post-airAlert period being a subscriber (17 divided by 34,200, then multiplied by 197). Assuming that subscribers use airAlerts to 100% effectiveness, 1970 subscribers would be needed to prevent one hospital respiratory admission in the 3 years post-airAlert (one admission divided by 0.1 then multiplied by 197 subscribers). If airAlert provided a forecast for PM2.5, the chance of an admission being a subscriber increases from 10 to 17% and so reduces the number of subscribers needed to prevent one admission from 1970 to 1159; providing the airAlert service is 100% effective at forecasting poor episodes of AQ, which is not evident (Fig. 8). There are instances where the service over-predicts (see February), and under-predicts episodes of air pollution (see March). There are some correct predictions however, which proves the service is effective to some degree (see late May and July). The AQ values used in Fig. 8 are from background sites and do not include industrial or roadside monitoring stations, therefore the airAlert forecasting may be more accurate than exhibited. Fig. 8a

484

J. McLaren, I.D. Williams / Science of the Total Environment 538 (2015) 478–491

Fig. 3. Time variation plots to show the mean pollutant concentrations with respect to their averaging times for the AQI by hour of weekday, by hour, by month and by weekday during 01/ 04/2008–30/03/2013. The pollutant concentrations are from Southampton AURN background site and show a 95% confidence interval smooth line or box.

confirms that poor episodes of AQ occur at low wind speeds primarily from a south or easterly wind direction. 4. Discussion

that O3 enters a reaction with NOx, marked by the decline in O3 as NO2 concentrations increase, corresponding to the reaction with increased NO as a result from the daily rush hours. O3 concentrations are highest at weekends, in tandem with the lowest NO2 concentrations associated with reduced vehicular movements.

4.1. AQ trends and relationships 4.2. AQ relationship with wind speed and direction Evidently AQ demonstrates seasonal trends in Southampton (Figs. 2 and 3). The highest concentrations for NO2 and particulates in winter are likely due to the impact of colder temperatures, as faster conversion of NO to NO2 occurs in winter smog episodes (Bower et al., 1994; Harrison and Shi, 1996). Particulate concentrations are high during winter and spring potentially due to the re-suspension of road surface particles and long-range transport, which are a main source of particulates in the UK at this time (Latha and Highwood, 2006; Sanchez-Reyna et al., 2006; Harrison et al., 2012). Diurnal patterns show similar results to Bigi and Harrison (2010), which links NO2 and particulate concentrations to vehicular rush hours. This may further explain higher winter concentrations as increased emissions are observed when engines are cold (Gao et al., 2012; Dardiotis et al., 2013). However, particulate concentrations peak after NO2 in the evening (Fig. 3), indicating a shift from vehicular to domestic sources (e.g. heating) and increased condensation of volatile particulates during the night (Harrison et al., 2012). Weekly profiles of both pollutants show a reduction in concentrations at the weekend. In contrast, O3 concentrations are highest during the spring/summer (Figs. 2 and 3) as background O3 concentrations in the Northern Hemisphere are highest in spring (Munir et al., 2013). Diurnal trends in O3 show no signs of rush-hour peaks, but peaks in the evening due to increased solar activity during the middle of the day and the nature of the 8-hr mean. It is evident from morning and evening concentrations

Wind speed showed an inverse relationship with NO2 and particulates, and a positive relationship with O3 (Fig. 4). This is primarily due to the lack of dispersion of pollutants at lower wind speeds, allowing concentrations to accumulate (Jones et al., 2010). High concentrations at wind speeds b6 ms−1 are indicative of low-level local sources such as road traffic, whereas speeds N6 ms−1 indicate elevated sources such as stacks or long-range transport (Harrison et al., 2012; Malby et al., 2013). The direction of the wind is also indicative of pollutant source. Therefore the highest concentrations of NO2 and particulates occur from low-level sources in close proximity to the monitoring site (Figs. 4 and 5), verifying trends in Fig. 3. However NO2 shows moderate concentrations occurring from a westerly direction at speeds of up to 10 ms−1, which could be indicative of movement of NO2 from the local port and heavy goods vehicles (HGV) activity on a nearby major road (the A33), which are recognised for the production of NO2 prerequisites (Beevers et al., 2012). In addition, an area of moderate concentration occurs from a northerly direction at 6 ms−1, possibly a result of heavily congested areas (Beevers and Carslaw, 2005) close to the airport and around the gateway to the urban area of Eastleigh. Both areas identified as the cause of these concentrations are designated as AQMAs (SCC, 2012). Similarly, PM10 concentrations were moderate when wind speeds were high from a south-westerly direction. It is unlikely

J. McLaren, I.D. Williams / Science of the Total Environment 538 (2015) 478–491

485

Fig. 4. Polar plot to show the daily mean concentrations of each pollutant by wind speed (m s−1) and direction at Southampton AURN background site during 01/04/2008–30/03/2013.

due to the same sources as NO2 because Figs. 4 and 5 show differing profiles from this direction. Fig. 5e and g specifically show the lack of the rush-hour trend for particulates. The Marchwood Energy Recovery facility is located outside the boundary of Southampton to the southwest and emits a high amount of particulates which may contribute to the observations in Fig. 4. However, the plant emits more NOx than it does particulates (VES, 2014), which should produce similar profiles in Fig. 4a and c, which it does not. Therefore, the cause of this trend is likely due to maritime wind-blown particulates from the Atlantic (Abdalmogith and Harrison, 2005). Highest PM10 concentrations occur from an easterly direction at low to moderate wind speeds. This is a trend shared by PM2.5 to a lesser extent and is exhibited around the UK (strongest in the south-east), linked to emissions and precursors of secondary particulate matter from Europe (Witham and Manning, 2007; Harrison et al., 2012) and natural contributions such as Saharan dust (Viana et al., 2014). Although this is a significant source to urban particulate concentrations, higher concentrations are generally a result of local anthropogenic sources (Turnbull and Harrison, 2000). It is therefore reasonable to assume the highest concentrations for NO2

and particulates are from similar combustion sources and deviate above 2 ms−1. Conversely, O3 shares a positive relationship with wind speed from all directions (Fig. 4). Rural sites show short-term elevations in O3 concentrations during spring and summer as a consequence of photochemical processing and transport from Europe (Jenkin et al., 2002). It should be noted that Fig. 4b is almost the exact inverse of Fig. 4a, as urban areas are characterised by higher concentrations of NOx relative to rural areas. Therefore there is an increased loss of O3 to scavenging by the NO/O3 reaction (Jenkin, 2008). Consequently, at higher wind speeds increased concentrations of O3 are blown into Southampton from surrounding rural areas. The highest concentrations occur from an easterly wind direction, indicating long-range transport from Europe (Figs. 4b and 5b). 4.3. Relationship between AQ and health in Southampton Hospital admissions of both asthma and COPD in Southampton increased over the study period, although AQ concentrations have remained stable (Fig. 2). Well-defined seasonal and weekly trends

486

J. McLaren, I.D. Williams / Science of the Total Environment 538 (2015) 478–491

Fig. 5. Seasonal (mean concentrations displayed by time of year and wind direction) and hourly (mean concentrations displayed by time of day and wind direction) Polar Annulus plots for each pollutant over time periods 01/04/2008–30/03/2013 and 01/08/2010–30/03/2013 respectively from Southampton AURN background site.

exist for admissions, highest in the winter and early weekdays. The seasonality trend is linked to colder temperatures that can bring on exacerbations of respiratory disease (McAllister et al., 2013). Colder temperatures increase the likelihood of influenza epidemics (Hussain et al., 2005), which can further exacerbate underlying respiratory conditions; in Southampton asthma admissions are strongly correlated with influenza epidemics (Johnston et al., 1996). The lower weekend admissions correspond with lower air pollutant concentrations (Fig. 3), however the trend is unlikely to be caused by AQ, as weak relationships are shown in Table 4 and contractual arrangements have reduced the amount of out-of-hours-care general practitioners offer (Saxena et al., 2009). This along with further changes to the NHS, such as the enforcement of the 4 hr wait target, changes to care pathways and NHS direct telephone service which encourages the caller to seek medical advice, are believed to be increasing emergency hospital admissions (Gill

Table 4 Spearman Correlation for maximum daily pollutant concentration vs. admissions for respiratory diseases per air quality index band. Pollutant

AQI

n

Asthma

COPD

Total admissions

NO2

Low Moderate High Low Moderate High Low Moderate High Low Moderate High Very High

1710 – – 1708 12 – 1639 28 1 1343 57 12 1

0.04 – – 0.003 0.277 – 0.041 0.302 – 0.031 0.117 −0.068 –

0.075b – – −0.044 −0.05 – 0.034 0.065 – 0.027 −0.162 0.218 –

0.074b – – −0.026 0.03 – 0.052a 0.267 – 0.042 −0.045 0.226 –

O3

PM10

PM2.5

a b

Signifies a significant correlation (p b 0.05). Signifies a highly significant correlation (p b 0.001).

et al., 2013). These changes occurred before the study period, so it is hard to prove the causation of the increased hospital admissions, however it is reasonable to assume a combination of the factors have contributed to trends observed in Fig. 2. Although, AQ has little effect on long-term trends in hospital admissions, poor episodes of short-term AQ may have negative health effects (Table 2). The correlations shown in Table 4 indicate a weak positive relationship for the majority of cases. The strongest correlation is found between asthma and PM10 at moderate concentrations, which corresponds with relative risk analysis in the UK (Sunyer et al., 1997). As the AQI band increases from low to high the relationship between the two variables becomes stronger, but less significant due to the low number of correlating points. Negative correlations occur for some pollutants such as O3, potentially due to lack of correlating points, seasonal trends or time lags. This is common when correlating yearly data for O3 (Anderson et al., 2007), but obviously not accurate, as lab tests prove that exposure to high O3 concentrations can negatively affect respiratory health (Hiltermann et al., 1999; Nadeau et al., 2010). Correlating admissions and O3 during the spring/summer months achieves stronger positive relationships, which are more representative of true conditions (Samoli et al., 2011). Table 5 Relative risks of hospital admissions for respiratory disease when maximum pollutant concentrations increase by 10 μg m−3 for Southampton and the ASPIRE project. Pollutant

NO2 O3 PM10 PM2.5

Total admissions

ASPIRE total admissions

Relative risk

95% CI lower

95% CI upper

Relative risk

95% CI lower

95% CI upper

1.0086a 0.9901a 1.0002 1.0011

1.0002 0.9808 0.9834 0.9823

1.0171 0.9994 1.0174 1.0202

1.0015b 1.0063b 1.0171b –

0.9992 1.0009 1.0119 –

1.0038 1.0118 1.0223 –

CI signifies 95% confidence interval. a Signifies a significant regression (p b 0.05). b Signifies an unknown significance level.

J. McLaren, I.D. Williams / Science of the Total Environment 538 (2015) 478–491

487

Table 6 The number of hospital admissions attributable to poor episodes of air quality (the number of pollution event days) pre and post-airAlert service for the current airAlert pollutants and hypothetical scenario where airAlert forecasted for PM2.5.

Current airAlert pollutants

airAlert pollutants including PM2.5

Fig. 6. A comparison of the relative risk of hospital admissions for respiratory disease changing with an increase in max pollutant concentrations by 10 μg m−3 for Southampton and the ASPIRE project.

In addition, impacts of AQ on the respiratory system may be lagged (Pope et al., 1991; Schwartz et al., 1993; Halonen et al., 2008), which is not accounted for in this study, therefore stronger associations may be found on lagged days of AQ. Similar results are shown in Table 5, indicating that the likelihood of admission declines as O3 concentrations increase. This is an evident limitation, as influenza data could not be obtained. Conversely, NO2 and particulates show that as concentrations increase, so does the relative risk of a respiratory hospital admission. The confidence intervals for NO2 and O3 are similar, whereas those for particulates are double the latter, due to the absence of statistical significance. Sunyer et al. (1997) also found NO2 and O3 to be significant

Pollutant

Number of respiratory hospital admissions (number of pollution days) Pre-airAlert

Post-airAlert

O3 PM10 Total O3 PM10 PM2.5 Total

0.2 (2) 3.5 (7) 3.7 (9) 0.2 (2) 1.4 (3) 7.8 (16) 9.4 (21)

2.4 (10) 14.8 (25) 17.2 (35) 2.4 (10) 8.6 (16) 19.3 (39) 30.3 (65)

whereas PM10 was not. In comparison with the relative risks used in Walton et al. (2012) derived from Anderson et al. (2007), there is considerable variation (particularly for O3). Although confidence levels overlap, the high uncertainty associated with the regression in this study is large. Due to the extensive literature review conducted in Anderson et al. (2007), it is likely the figures used in the AsPIRe project are more accurate due to multi-city studies, larger sample sizes, inclusion of additional confounding variables and statistical significance and therefore used in the calculation of the impact of AQ on respiratory admissions in Southampton. 4.4. Quantification of the impact AQ has on respiratory admissions in Southampton The AsPIRe project does not quantify the impact of PM2.5, although it is calculated here due to its importance for respiratory outcomes (Atkinson et al., 2010). Fine and course particulates must be studied independently in research and epidemiology as they are classified in separate pollutant classes (Wilson and Su, 1997). The relative risks for PM10 were used for PM2.5; this is a reasonable assumption as PM10 has a similar, if not stronger short-term effect on respiratory admissions than PM2.5 (Brunekreef and Forsberg, 2005; Ferguson et al., 2013). This is because PM2.5 causes a reduced inflammatory response in the lung due to its higher solubility and different surface properties compared to PM10 (Happo et al., 2007). The impact of PM2.5 is greater than

Fig. 7. The number of hospital admissions attributable to poor episodes of AQ [the number of pollution event days] for each pollutant and AQI band. Pollutant days were established by an aggregate dataset of Southampton AURN and Southampton Bitterne background sites. PM10* symbolises days where there were no pollution event days for PM2.5 but one for PM10.

488

J. McLaren, I.D. Williams / Science of the Total Environment 538 (2015) 478–491

Fig. 8. Calendar plot to show a) the days in which air quality for both O3 and PM10 at both Southampton AURN and Bitterne monitoring sites was low, moderate or high, annotated by wind angle scaled to wind speed i.e. the longer the arrow, the higher the wind speed and b) the days in which an airAlert was issued in 2012.

PM10 when studying coronary events (Cesaroni et al., 2014) because increased blood coagulation occurs in the alveoli from finer particles (de Hartog et al., 2003). The majority of respiratory hospital admissions are attributable to particulates (Fig. 7). Comparatively, O3 caused three admissions; however this pollutant may become an increasing issue in the future. It is predicted that emissions of its precursors will increase over the next 20–50 years, leading to significant increases in concentrations (Cape, 2008), likely to be intensified by the reduction of NOx concentrations associated with meeting EU targets (Defra, 2011). In London it is predicted that annual NO2 concentrations will reduce by 5.3%, whereas O3 and PM10 concentrations will increase by 20% and 2.6% respectively by the end of the century compared to a late 1900s baseline (Athanassiadou et al., 2010). London is representative of other UK cities in terms of pollutant mix and concentrations (Bigi and Harrison, 2010); it is therefore reasonable to assume similar, if not worse increases in Southampton, because O3 (Jenkin, 2008) and particulate (Harrison et al., 2012) concentrations are worse in the south-east. If the AsPIRe methodology to calculate baseline admissions had been applied here, total admissions would reduce by 5, which may indicate an underestimation of admissions by not using admissions data relevant to the study site. AQ cased 40 emergency admissions over the period, which is negligible considering a total of 16,529 admissions were recorded; however, they were likely preventable and had financial implications. 4.5. Evaluation of the airAlert service The airAlert service started on 11/04/2010, so not all admissions in Fig. 7 would be affected. A total of 17 admissions post-airAlert are attributable to air pollution events (Table 6). Of these admissions 0.1 were likely to be prevented, provided they followed airAlert advice to 100%

effectiveness and the forecast was 100% accurate. This indicates airAlert had a negligible effect on admissions caused by AQ in Southampton. The service costs £10,000 per annum which can be recouped by preventing one COPD admission which costs the NHS between £3000 a day over a course of 3–5 days; admissions of asthma cost between £1500 and £2000 for admission and one day in hospital (Jenkins, 2013). As airAlert is statistically unlikely to prevent one admission, the service is not cost efficient as it stands. To improve cost effectiveness the number of subscribers needs to be increased to 5910 of the respiratory disease population to statistically prevent one admission per year. Offering the service to individuals with cardiovascular problems can help this, as airAlert is currently limited to respiratory diseases. Alternatively, broadening the forecasting capabilities to include PM2.5 can improve cost effectiveness and reduce the number of subscribers needed to prevent one admission. By forecasting PM10 some of the admissions attributable to PM2.5 are accounted for (Fig. 7). However, by excluding PM2.5 from analysis 42 moderate days are not alerted to the public. If forecasts had been provided for PM2.5 the number of admissions likely to have been a subscriber increases to 0.17, which improves cost efficiency, but is still statistically unlikely to be cost beneficial. However, it does reduce the number of subscribers needed to prevent one admission per year to 3282 of the respiratory disease population. In a best-case scenario where all of the respiratory population subscribed to airAlert, the maximum saving to the NHS would be £255,000 if all admissions were COPD patients staying for a maximum period. This is an overestimate, as AQ brings forward admissions that would have occurred in the near future (COMEAP, 1998). Though in this case airAlert is still beneficial, as it may prevent a bulk of admissions from occurring at once which inundates NHS resources, as predicted in April 2014 due to very high UK particulate concentrations (Perry and Edgar, 2014).

J. McLaren, I.D. Williams / Science of the Total Environment 538 (2015) 478–491

All of these calculations assume 100% prediction accuracy and effective use of the information by subscribers. Evidently this is not the case (Fig. 8). Subscribers have noticed this trend and have suffered negative health effects as a result (S. Hartill, unpublished data). This is common with AQ forecasts, particularly for O3 concentrations (Eder et al., 2006, 2009). Over-predictions will not have negative implications on health compared with under-predicting, however if too many ‘false alarms’ occur, the public may not respond to future alerts (Teisberg and Weiher, 2009). Currently 16% of subscribers do not modify their behaviour during an alert (S. Hartill, unpublished data). These inaccuracies and behavioural issues reduce the effectiveness of the service for preventing hospital admissions. Nevertheless, the service still has potential. Whilst the number of subscribers remains small, the forecasting technology will improve in time. airAlert will be needed more in the future as AQ is predicted to worsen and stems from sources not under governmental control. There is also potential to adapt airAlert to warn subscribers of heat waves (Åström et al., 2012) and is currently being used for cold temperatures as coldAlert (Jenkins, 2013). In addition, alerts should be issued for a number of days to ensure any corresponding days of poor AQ are covered, which is not always the case (Fig. 8). 5. Conclusions Background AQ in Southampton is within EU limits although there are certain conditions in which poor episodes of AQ occur. Whilst there are current concerns over annual NO2 concentrations, there are no pollution events at background sites. The pollutants that have caused emergency respiratory admissions in Southampton are O3 and particulates, even though background concentrations do not exceed annual limits. The majority of emission sources causing the highest concentrations of certain pollutants are local, although it is believed that emission sources from outside national boarders have an effect. This is concerning as current poor episodes of AQ are associated with negative health effects in Southampton, placing greater emphasis on the need for AQ forecasts and the provision of public information, particularly to those with underlying respiratory diseases. However, under current conditions the airAlert service provided by Southampton City Council is not statistically likely to have an impact on the number of emergency admissions of asthma and COPD. In addition, it is not cost beneficial, which is important when considering strained local authority resources. Nevertheless, the effectiveness of the airAlert service in reducing hospital admissions can be improved provided a number of changes are made, including broadening the forecasting to include PM2.5; increasing the number of days alerts are active for; and increasing the number of subscribers. Offering the service to individuals with other diseases exacerbated by air pollution (mainly cardiovascular disease) will be beneficial to both health and cost effectiveness of airAlert. Out of the pollutants studied, particulates have the greatest impact on respiratory admissions in Southampton; therefore the impact is potentially much greater for cardiovascular disease, which can be improved by the airAlert service. Short-term pollution episodes cause respiratory admissions of asthma and COPD, despite background AQ concentrations within EU limit values. To reduce these admissions, a number of modifications to the airAlert service are required. Acknowledgements Thanks to S. Hartill at Southampton City Council for providing the airAlert and admissions data. References Abdalmogith, S.S., Harrison, R.M., 2005. The use of trajectory cluster analysis to examine the long-range transport of secondary inorganic aerosol in the UK. Atmos. Environ. 39, 6686–6695.

489

airAlert, 2013. What is airAlert?. [Online]. Available at:, http://www.airalert.info/Sussex/ AboutairAlert.aspx (Date accessed 03 February 2014) Andersen, Z.J., Hvidberg, M., Jensen, S.S., Ketzel, M., Loft, S., Sørensen, M., Tjønneland, A., Overvad, K., Raaschou-Nielsen, O., 2011. Chronic obstructive pulmonary disease and long-term exposure to traffic-related air pollution: a cohort study. Am. J. Respir. Crit. Care Med. 183 (4), 455–461. Anderson, H.R., Atkinson, R.W., Bremner, S.A., Carrington, J., Peacock, J., 2007. Quantitative systematic review of short term associations between ambient air pollution (particulate matter, ozone, nitrogen dioxide, sulphur dioxide and carbon monoxide), and mortality and morbidity. [Online]. Available at:, https://www.gov.uk/government/uploads/ system/uploads/attachment_data/file/215975/dh_121202.pdf (Date accessed 26 March 2014). Asthma UK, 2014. Asthma facts and FAQs. [Online]. Available at:, http://www.asthma.org. uk/asthma-facts-and-statistics (Date Accessed: 01 February 2014). Åström, C., Orru, H., Rocklöv, Strandberg, G., Ebi, K.L., Forsberg, B., 2012. Heat-related respiratory hospital admissions in Europe in a changing climate: a health impact assessment. BMJ 3 (1), 1–7. Athanassiadou, M., Baker, J., Carruthers, D., Collins, W., Girnary, S., Hassell, D., Hort, M., Johnson, C., Johnson, K., Jones, R., Thomson, D., Trought, N., Witham, C., 2010. An assessment of the impact climate change on air quality at two UK sites. Atmos. Environ. 44 (15), 1877–1886. Atkinson, R.W., 1999. Short-term associations between outdoor air pollution and visits to accident and emergency departments in London for respiratory complaints. Eur. Respir. J. 13 (2), 257–265. Atkinson, R.W., Strachan, D.P., Bland, J.M., Bremmer, S.A., Ponce de Leon, A., 1999. Shortterm associations between outdoor air pollution and visits to accident and emergency departments in London for respiratory complaints. Eur. Respir. J. 13 (2), 257–265. Atkinson, R.W., Fuller, G.W., Anderson, H.R., Harrison, R.M., Armstrong, B., 2010. Urban ambient particle metrics and health: a time-series analysis. Epidemiology 21 (4), 501–511. Ayres, J.G., 1997. Trends in air quality in the UK. Allergy 52 (38), 7–13. Balmes, J., Becklake, M., Blanc, P., Henneberger, P., Kreiss, K., Mapp, C., Milton, D., Schwartz, D., Toren, K., Viegi, G., 2003. American thoracic society statement: occupational contribution to the burden of airway disease. Am. J. Respir. Crit. Care Med. 167 (5), 787–797. Barnes, J., Hayes, E.T., Longhurst, J., 2013. Is local air quality management a successful strategy in achieving selected EU limit values? FET Doctoral Exchange 2013 (July 2013) Beatty, T.K.M., Shimsack, J.P., 2014. Air pollution and children's respiratory health: a cohort analysis. J. Environ. Econ. Manag. 67 (1), 39–57. Beaumont, R., Hamilton, R.S., Machin, N., Perks, J., Williams, I.D., 1999. Social awareness of air quality information. Sci. Total Environ. 235 (1–3), 319–329. Beevers, S.D., Carslaw, D.C., 2005. The impact of congestion charging on vehicle emissions in London. Atmos. Environ. 39 (1), 1–5. Beevers, S.D., Westmoreland, E., de Jong, M.C., Williams, M.L., Carslaw, D.C., 2012. Trends in NOx and NO2 emissions from road traffic in Great Britain. Atmos. Environ. 54 (July 2012), 107–116. Bernstein, J.A., Alexis, N., Barnes, C., Bernstein, I,.L., Nel, A., Peden, D., Diaz-Sanchez, D., Tarlo, S.M., Williams, P.B., 2004. Health effects of air pollution. J. Allergy Clin. Immunol. 114 (5), 1116–1123. Bickerstaff, K., Walker, G., 2001. Public understandings of air pollution: the ‘localisation’ of environmental risk. Glob. Environ. Chang. 11 (2), 133–145. Bigi, A., Harrison, R.M., 2010. Analysis of the air pollution climate at a central urban background site. Atmos. Environ. 44 (16), 2004–2012. Bower, J.S., Broughton, G.F.J., Stedman, J.R., 1994. A winter NO2 smog episode in the U.K. Atmos. Environ. 28 (3), 461–475. Brunekreef, B., Forsberg, B., 2005. Epidemiological evidence of effects of coarse airborne particles on health. Eur. Respir. J. 26 (2), 309–318. Brunekreef, B., Holgate, S.T., 2002. Air pollution and health. Lancet 360 (9341), 1233–1242. Cape, J.N., 2008. Surface ozone concentrations and ecosystem health: past trends and a guide to future projections. Sci. Total Environ. 400 (1–3), 257–269. Carslaw, D.C., 2013. The openair manual — open-source tools for analysing air pollution data. Manual for version 0.8–0. King's College London. Carslaw, D.C., Ropkins, K., 2012. Openair- and R package for air quality data analysis. Environ. Model. Softw. 27–28, 52–61. Cesaroni, G., Forastiere, F., Stafoggia, M., Andersen, Z.J., Badaloni, C., Beelen, R., Caracciolo, B., de Faire, U., Erbel, R., Eriksen, K.T., Fratiglioni, L., Galassi, C., Hampel, R., Heier, M., Hennig, F., Hilding, A., Hoffmann, B., Houthuijs, D., Jöckel, K.H., Korek, M., Lanki, T., Leander, K., Magnusson, P.K.E., Migliore, E., Ostenson, C.G., Overvad, K., Pedersen, N.L., Pekkanen, J.J., Penell, J., Pershagen, G., Pyko, A., Raaschou-Nielsen, O., Ranzi, A., Ricceri, F., Sacerdote, C., Salomaa, V., Swart, W., Turunen, A.W., Vineis, P., Weinmayr, G., Wolf, K., de Hoogh, K., Hoek, G., Brunekreef, B., Peters, A., 2014. Long term exposure to ambient air pollution and incidence of acute coronary events: prospective cohort study and meta-analysis in 11 European cohorts from the ESCAPE Project. BMJ 348 (f7412), 1–16. Chung, K.F., Marwick, J.A., 2010. Molecular mechanisms of oxidative stress in airways and lungs with reference to asthma and chronic obstructive pulmonary disease. Ann. N. Y. Acad. Sci. 1203 (1), 85–91. Committee on the Medical Effects of Air Pollutants (COMEAP), 1998. Quantification of the effects of air pollution on health in the United Kingdom. [Online]. Available at:, http:// www.comeap.org.uk/images/stories/Documents/Reports/quantification%20report% 201998.pdf (Date Accessed 27 March 2014). Dardiotis, C., Martini, G., Marotta, A., Manfredi, U., 2013. Low-temperature cold-start gaseous emissions of late technology passenger cars. Appl. Energy 111, 468–478. de Hartog, J.J., Hoek, G., Peters, A., Timonen, K.L., Ibald-Mulli, A., Brunekreef, B., Heinrich, J., Tiittanen, P., van Wijnen, J.H., Kreyling, W., Kulmala, M., Pekkanen, J., 2003. Effects of

490

J. McLaren, I.D. Williams / Science of the Total Environment 538 (2015) 478–491

fine and ultrafine particles on cardiorespiratory symptoms in elderly subjects with coronary heart disease: the ULTRA study. Am. J. Epidemiol. 157 (7), 613–623. de Souza Tadano, Y., Ugaya, C.M.L., Franco, A.T., 2012. Methodology to assess air pollution impact on human health using the generalized linear model with Poisson regression. In: Khare, Mukesh (Ed.), Air Pollution — Monitoring, Modelling and Health. InTech. ISBN: 978-953-51-0424-7 http://dx.doi.org/10.5772/33385. Delfino, R.T., Zeiger, R.S., Seltzer, J.M., Street, D.H., McLaren, C.E., 2002. Association of asthma symptoms with peak particulate air pollution and effect modification by antiinflammatory medication use. Environ. Health Perspect. 110 (10), 607–617. Department for Environment Food and Rural Affairs (Defra), 2007. The Air Quality Strategy for England, Scotland, Wales and Northern Ireland: Volume 1. [Online]. Available at:, http://archive.defra.gov.uk/environment/quality/air/airquality/strategy/documents/ air-qualitystrategy-vol1.pdf (Date Accessed 11 January 2014). Department for Environment Food and Rural Affairs (Defra), 2011. Air Quality Plan for the achievement of EU air quality limit values for nitrogen dioxide (NO2) in Southampton Urban Area (UK0019): September 2011. [Online]. Available at:, http://uk-air.defra. gov.uk/library/no2ten/ (Date Accessed 12 February 2014). Department for Environment Food and Rural Affairs (Defra), 2012. Monitoring networks: automatic urban and rural network. [Online]. Available at:, http://uk-air.defra.gov.uk/ networks/network-info?view=aurn (Date Accessed 11 January 2014). Department for Environment Food and Rural Affairs (Defra), 2013a. Air pollution in the UK 2012. [Online]. Available at:, http://uk-air.defra.gov.uk/library/annualreport/air_ pollution_uk_2012_issue_1.pdf (Date Accessed 13 January 2014). Department for Environment Food and Rural Affairs (Defra), 2013b. About air pollution: what is the daily air quality index?. [Online]. Available at:, http://uk-air.defra.gov.uk/ air-pollution/daqi?view=more-info&pollutant=pm10#pollutant (Date Accessed 18 January 2014) Department for Environment Food and Rural Affairs (Defra), 2013c. About air pollution: daily air quality index. [Online]. Available at:, http://uk-air.defra.gov.uk/airpollution/daqi (Date Accessed 18 January 2014). Department for Environment Food and Rural Affairs (Defra), 2013d. Data archive: data selector. [Online]. Available at:, http://uk-air.defra.gov.uk/data/data_selector (Date Accessed 04 September 2013). Department of Health (DH), 2011. An outcomes strategy for Chronic Obstructive Pulmonary Disease (COPD) and asthma. [Online]. Available at:, https://www.gov.uk/ government/uploads/system/uploads/attachment_data/file/216139/dh_128428.pdf (Date Accessed 03 February 2014). Department of Health (DH), 2012. An outcomes strategy for COPD and asthma: NHS companion document. [Online]. Available at:, https://www.gov.uk/government/uploads/ system/uploads/attachment_data/file/216531/dh_134001.pdf (Date Accessed 03 February 2014). Directive 2008/50/EC of the European Parliament and of the council of 21 May 2008 on ambient air quality and cleaner air for Europe. Eder, B., Kang, D., Mathur, R., Yu, S., Schere, K., 2006. An operational evaluation of the EtaCMAQ air quality forecast model. Atmos. Environ. 40 (26), 4894–4905. Eder, B., Kang, D., Mathur, R., Pleim, J., Yu, S., Otte, T., Pouliot, G., 2009. A performance evaluation of the National Air Quality Forecast Capability for the summer of 2007. Atmos. Environ. 43 (14), 2312–2320. Emberson, L.D., Ashmore, M.R., Murray, F., Kuylenstierna, J.C.I., Percy, K.E., Izuta, T., Zheng, Y., Shimizu, H., Sheu, B.H., Liu, C.P., Agrawal, M., Wahid, A., Abdel-Latif, N.M., van Tienhoven, M., de Bauer, L.I., Domingos, M., 2001. Impacts of air pollutants on vegetation in developing countries. Water Air Soil Pollut. 130 (1–4), 107–118. Ferguson, M.D., Migliaccio, C., Ward, T., 2013. Comparison of how ambient PMc and PM2.5 influence the inflammatory potential. Inhal. Toxicol. 25 (14), 766–773. Gao, Z., Kim, M.Y., Choi, J.S., Daw, C.S., Parks, J.E., Smith, D.E., 2012. Cold-start emissions control in hybrid vehicles equipped with a passive adsorber for hydrocarbons and nitrogen oxides. Proc. Inst. Mech. Eng. D J. Automob. Eng. 226 (10), 1396–1407. Gill, P.J., Goldacre, M.J., Mant, D., Heneghan, C., Thompson, A., Seagroatt, V., Harnden, A., 2013. Increase in emergency admissions to hospital for children aged under 15 in England, 1999–2010: national database analysis. Arch. Dis. Child. 98 (5), 328–334. Halonen, J.I., Lanki, T., Yli-Tuomi, T., Kulmala, M., Tiittanen, P., Pekkanen, J., 2008. Urban air pollution, and asthma and COPD hospital emergency room visits. Thorax 63 (7), 635–641. Hantsair, 2013. Air quality in Southampton. [Online]. Available at:, http://www.hantsair.org. uk/hampshire/asp/home.asp?la=Southampton (Date Accessed 02 September 2013). Happo, M.S., Salonen, O.R., Hälinen, A.I., Jalava, P.I., Pennanen, A.S., Kosma, V.M., Sillanpää, M., Hillamo, R., Brunekreef, B., Katsouyanni, K., Sunyer, J., Hirvonen, M.R., 2007. Dose and time dependency of inflammatory responses in the mouse lung to urban air course, fine, and ultrafine particles from six European Cities. Inhal. Toxicol. 19 (3), 227–246. Harrison, R., Shi, J., 1996. Sources of nitrogen dioxide in winter smog episodes. Sci. Total Environ. 189–90 (1), 391–399. Harrison, R.M., Laxen, D., Moorcroft, S., Laxen, K., 2012. Processes affecting concentrations of fine particulate matter (PM2.5) in the UK atmosphere. Atmos. Environ. 46 (1), 115–124. Hiltermann, J.T.N., Lapperre, T.S., van Bree, L., Steerenberg, P.A., Brahim, J.J., Sont, J.K., Sterk, P.J., Hiemstra, P.S., Stolk, J., 1999. Ozone-induced inflammation assessed in sputum and bronchial, lavage fluid from asthmatics: a new noninvasive tool in epidemiologic studies on air pollution and asthma. Free Radic. Biol. Med. 27 (11–12), 1448–1454. Hussain, S., Harrison, R.M., Ayres, J., Walter, S., Hawker, Wilson, R., Shukur, G., 2005. Estimation and forecasting hospital admissions due to influenza: planning for winter pressure. The case of the West Midlands, UK. J. Appl. Stat. 32 (3), 191–205. Jacobs, J.E., van de Lisdonk, E.H., Smeele, I., van Weel, C., Grol, R.P.T.M., 2001. Management of patients with asthma and COPD: monitoring quality of life and the relationship to subsequent GP interventions. Fam. Pract. 18 (6), 574–580.

Jacquemin, B., Kauffmann, F., Pin, I., Le Moual, N., Bousquet, J., Gormand, F., Just, J., Nadif, R., Pison, C., Vervloet, D., Künzli, N., Siroux, V., 2012. Air pollution and asthma control in the Epidemiological study on the Genetics and Environment of Asthma. J. Epidemiol. Community Health 66 (9), 796–802. Jenkin, M.E., 2008. Trends in ozone concentration distributions in the UK since 1990: local, regional and global influences. Atmos. Environ. 42 (1), 5434–5445. Jenkin, M.E., Davies, T.J., Stedman, J.R., 2002. The origin and day-of-week dependence of photochemical ozone episodes in the UK. Atmos. Environ. 36 (6), 999–1012. Jenkins, N., 2013. Protecting people's health and improving air quality through initiatives and policy. [Online]. Available at:, http://www.london.gov.uk/sites/default/files/ archives/cleaner_air_conference_sussex_220113.pdf (Date Accessed: 27 March 2014). Johnston, S.L., Pattemore, P.K., Sanderson, G., Smith, S., Campbell, M.J., Josephs, L.K., Cunningham, A., Robinson, B.S., Myint, S.H., Ward, M.E., Tyrrell, D.A., Holgate, S.T., 1996. The relationship between upper respiratory infections and hospital admissions for asthma: a time-trend analysis. Am. J. Respir. Crit. Care Med. 154 (3), 654–660. Jones, A.M., Harrison, R.M., Baker, J., 2010. The wind speed dependence of the concentrations of airborne particulate matter and NOx. Atmos. Environ. 44 (13), 1682–1690. Kelly, F.J., Fussell, J.C., 2011. Air pollution and airway disease. Clin. Exp. Allergy 41 (8), 1059–1071. Kelly, F.J., Fuller, G.W., Walton, H.A., Fussell, J.C., 2012. Monitoring air pollution: use of early warning systems for public health. Respirology 17 (1), 7–19. Latha, K.M., Highwood, E.J., 2006. Studies on particulate matter (PM10) and its precursors over urban environment of Reading, UK. J. Quant. Spectrosc. Radiative Transf. 101 (2), 367–379. Li, S., Batterman, S., Wasilevich, E., Wahl, R., Wirth, J., Su, F.C., Mukherjee, B., 2011. Association of daily asthma emergency department visits and hospital admissions with ambient air pollutants among the paediatric Medicaid population in Detroit: timeseries and time-stratified case-crossover analyses with threshold effects. Environ. Res. 111 (8), 1137–1147. Lindley, S.J., Crabbe, H., 2004. What lies beneath? — issues in the representation of air quality management data for public consumption. Sci. Total Environ. 334–335, 307–325. Longhurst, J., Irwin, J., Chatterton, T., Hayes, E., Leksmono, N., Symons, J., 2009. The development of effects-based air quality management regimes. Atmos. Environ. 43 (1), 64–78. Lovett, G.M., Tear, T.H., Evers, D.C., Findlay, S.E., Cosby, B.J., Dunscomb, J.K., Driscoll, C.T., Weathers, K.C., 2009. Effects of air pollution on ecosystems and biological diversity in the Eastern United States. Ann. N. Y. Acad. Sci. 1162, 99–135 (April 2009). Malby, A.R., Whyatt, J,.D., Timmis, R.J., 2013. Conditional extraction of air-pollutant source signals from air-quality monitoring. Atmos. Environ. 74 (1), 112–122. McAllister, D.A., Morling, J.R., Fischbacher, C.M., MacNee, W., Wild, S.H., 2013. Socioeconomic deprivation increases the effect of winter on admissions to hospital with COPD: retrospective analysis of 10 years of national hospitalization data. Prim. Care Respir. J. 22 (3), 296–299. McCreanor, J., Cullinan, P., Nieuwenhuijsen, M.J., Stewart-Evans, J., Malliarou, E., Jarup, L., Harrington, R., Svartengren, M., Han, I.K., Ohman-Strickland, P., Chung, K.F., Zhang, J., 2007. Respiratory effects of exposure to diesel traffic in persons with athma. N. Engl. J. Med. 357, 2348–2358. Meng, X., Wang, C., Cao, D., Wong, C.M., Kan, H., 2013. Short-term effect of ambient air pollution on COPD mortality in four Chinese cities. Atmos. Environ. 77 (1), 149–154. Met Office, 2010. Observations: National Meteorological Library and Archive Fact sheet 17 — weather observations over land. [Online]. Available at:, http://www. metoffice.gov.uk/media/pdf/k/5/Fact_sheet_No._17.pdf (Date Accessed: 15 February 2014). Mortimer, K.M., Neas, L.M., Dockery, D.W., Redline, S., Tager, I.B., 2002. The effect of air pollution on inner-city children with asthma. Eur. Respir. J. 19 (4), 699–705. Munir, S., Chen, H., Ropkins, K., 2013. Quantifying temporal trends in ground level ozone concentration in the UK. Sci. Total Environ. 458–460 (1), 217–227. Nadeau, K., McDonald-Hyman, C., Noth, E.M., Pratt, B., Hammond, S.K., Balmes, J., Tager, I., 2010. Ambient air pollution impairs regulatory T-cell function in asthma. J. Allergy Clin. Immunol. 126 (4), 845–856. National Health Service (NHS), 2012. Chronic obstructive pulmonary disease — symptoms. [Online]. Available at:, http://www.nhs.uk/Conditions/Chronic-obstructivepulmonary-disease/Pages/Symptoms.aspx (Date Accessed: 01 February 2014). Oettl, D., Kukkonen, J., Almbaier, R.A., Sturm, P.J., Pohjola, M., Härkönen, J., 2001. Evaluation of a Gaussian and Lagrangian model against a roadside data set, with emphasis on low wind speed conditions. Atmos. Environ. 35 (12), 2123–2132. Ostro, B.D., Lipsett, M.J., Wiener, M.B., Selner, J.C., 1991. Asthmatic responses to airborne acid aerosols. Am. J. Public Health 81 (6), 694–702. Perry, K., Edgar, J., 2014. Sahara dust storm prompts ‘serious’ health warning for asthmatics. [Online]. Available at:, http://www.telegraph.co.uk/earth/environment/ 10739019/Sahara-dust-storm-prompts-serious-health-warning-for-asthmatics.html (Date Accessed: 06 April 2014). Pope, C.A., Dockery, D.W., Spengler, J.D., Raizenne, M.E., 1991. Respiratory health and PM10 pollution: a daily time series analysis. Am. Rev. Respir. Dis. 144 (3), 668–674. Qiu, H., Yu, I.T.S., Wang, X., Tian, L., Tse, L.A., Wong, T.W., 2012. Season and humidity dependence of the effects of air pollution on COPD hospitalisations in Hong Kong. Atmos. Environ. 76 (1), 74–80. Rabe, K.F., Hurd, S., Anzueto, A., Barnes, P.J., Buist, S.A., Calverley, P., Fukuchi, Y., Jenkins, C., Rodriguez-Roisin, R., van Weel, C., Zielinski, J., 2007. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease: GOLD executive summary. Am. J. Respir. Crit. Care Med. 176 (6), 532–555. Ringbaek, T., Seersholm, N., Viskum, K., 2005. Standardiese mortality rates in females and males with COPD and asthma. Eur. Respir. J. 25 (5), 891–895.

J. McLaren, I.D. Williams / Science of the Total Environment 538 (2015) 478–491 Samoli, E., Nastos, P.T., Paliastos, A.G., Katsouyanni, K., Priftis, K.N., 2011. Acute effects of air pollution on pediatric asthma exacerbation: evidence of association and effect modification. Environ. Res. 111 (1), 418–424. Sanchez-Reyna, G., Wang, K.Y., Gallardo, J.C., Shallcross, D.E., 2006. Association between PM10 mass concentration and wind direction in London. Atmos. Sci. Lett. 6 (4), 204–210. Sarkar, D., 1996. The ocean blues. Navigating the course of population growth. ZPG Report. 28 (1), 1–4. Saxena, S., Bottle, A., Gilbert, R., Sharland, M., 2009. Increasing short-stay unplanned hospital admissions among children in England; Time trends analysis '97–'06. PLoS ONE 4 (10), e7484. Schwartz, J., Slater, D., Larson, T.V., Pierson, W.E., Koenig, J.Q., 1993. Particulate airpollution and hospital emergency room visits for asthma in Seattle. Am. Rev. Respir. Dis. 147 (4), 826–831. Schwartz, J., Spix, C., Touloumi, G., Bachárová, Barumamdzadeh, T., le Tertre, A., Piekarksi, T., Ponce de Leon, A., Pönkä, A., Rossi, G., Saez, M., Schouten, J.P., 1996. Methodological issues in studies of air pollution and daily counts of deaths or hospital admissions. J. Epidemiol. Community Health 50 (1), 3–11. Silkoff, P.E., Zhang, L., Dutton, S., Langmack, E.L., Vedal, S., Murphy, J., Make, B., 2005. Winter air pollution and disease parameters in advanced chronic obstructive pulmonary disease panels residing in Denver, Colorado. J. Allergy Clin. Immunol. 115 (2), 337–344. Southampton City Council (SCC), 2010. Air quality action plan: progress report. [Online]. Available at:, http://www.southampton.gov.uk/Images/Air%20Quality%20Action% 20Plan%20Progress%20Report%202010_tcm46-273646.pdf (Date Accessed 10 February 2014). Southampton City Council (SCC), 2012. 2012 Air Quality Detailed Assessment Report Southampton City Council: in fulfilment of Part IV of the Environment Act 1995: Local Air Quality Management April 2012. [Online] Available at:, http://www. southampton.gov.uk/Images/SCC%20Detailed%20Assessment%20%20Report%2021% 205%2012_tcm46-336008.pdf (Date Accessed 11 February 2014). Southampton City Council (SCC), 2013. Southampton wins Transport City of the Year award. [Online]. Available at:, https://www.southampton.gov.uk/news-events/ latest-news/transportaward.aspx (Date Accessed 10 February 2014). Southampton City Council (SCC), 2014. City statistics and research. [Online]. Available at:, http://www.southampton.gov.uk/living/statsresearch/ (Date Accessed 01 February 2014). Sunyer, J., Spix, C., Quénel, P., Ponce de Leon, A., Pönka, A., Barumandzadeh, T., Touloumi, G., Bacharova, L., Wojtyniak, B., Vonk, J., Bisanti, L., Schwartz, J., Katsouyanni, K., 1997. Urban air pollution and emergency admissions for asthma in four European cities: the APHEA Project. Thorax 52 (9), 760–765. Teisberg, T.J., Weiher, R.F., 2009. Background paper on the benefits and costs of early warning systems for major natural hazards. [Online]. Available at:, https://gfdrr.org/ sites/gfdrr.org/files/New%20Folder/Teisberg_EWS.pdf (Date Accessed 31 March 2014).

491

The Air Quality Standards Regulations 2010. HMSO, London, UK, (2010). The Chartered Institution of Water and Environmental Management (CIWEM), 2013. Clearing the air: priorities for reducing air pollution in the UK. CIWEM, London. The Environment Act 1995. HMSO, London, UK, (1995). Turnbull, A.B., Harrison, R.M., 2000. Major component contributions to PM10 composition in the UK atmosphere. Atmos. Environ. 34 (19), 3129–3137. van der Zee, S.C., Hoek, G., Doezen, M.H., Schouten, J.P., van Wijnen, J.H., Brunekreef, B., 2000. Acute effects of air pollution on respiratory health of 50–70 year old adults. Eur. Respir. J. 15 (4), 700–709. Veolia Environmental Services (VES), 2014. Energy recovery: emissions to air data. [Online]. Available at:, http://www.veoliaenvironmentalservices.co.uk/Hampshire/ Energy-recovery/Marchwood/Emissions/March-20111112/ (Date Accessed 25 March 2014). Viana, M., Pey, J., Querol, X., Alastuey, A., de Leeuw, F., Lükewille, A., 2014. Natural sources of atmospheric aerosols influencing air quality across Europe. Sci. Total Environ. 472, 825–833 (15 February 2014). Walton, H., Fuller, G., Baker, T., Science Policy Group, Environmental Research Group, 2012. Air Pollution Intervention Research (AsPIRe) — prediction of possible effectiveness and assessment of intervention study feasibility. [Online]. Available at:, http:// www.londonair.org.uk/london/asp/LAQNSeminar/pdf/June2012/Heather_Walton_ ASPIRE_the_health_impact_of_air_pollution_episodes.pdf (Date Accessed December 06 2013). Welte, T., Groneberg, D.A., 2006. Asthma and COPD. Exp. Toxicol. Pathol. 57 (2), 35–40. Williams, M.L., Carslaw, D.C., 2011. New directions: science and policy — out of step on NOx and NO2? Atmos. Environ. 45 (23), 3911–3912. Williams, I.D., McCrae, I.S., 1995. Road traffic nuisance in residential and commercial areas. Sci. Total Environ. 169, 75–82. Wilson, W.E., Su, H.H., 1997. Fine particles and coarse particles: concentration relationships relevant to epidemiologic studies. J. Air Waste Manag. Assoc. 47 (12), 1238–1249. Witham, C., Manning, A., 2007. Impacts of Russian biomass burning on UK air quality. Atmos. Environ. 41 (37), 8075–8090. Wordley, J., Walters, S., Ayres, J.G., 1997. Short term variations in hospital admissions and mortality and particulate air pollution. Occup. Environ. Med. 54 (2), 108–116. World Health Organisation (WHO), 2005. WHO Air quality guidelines for particulate matter, ozone, nitrogen dioxide and sulfur dioxide:Global update 2005. [Online]. Available at:, http://whqlibdoc.who.int/hq/2006/WHO_SDE_PHE_OEH_06.02_eng.pdf (Date Accessed: 01 August 2013). Wunderground, 2013. Welcome to weather underground. [Online]. Available at:, http:// www.wunderground.com (Date Accessed 02 September 2013).