Influence of Weather and Atmospheric Pollution on Physical Activity in Patients With COPD

Ayedh D. Alahmari; Alex J. Mackay; Anant R.C. Patel; Beverly S. Kowlessar; Richa Singh; Simon E. Brill; James P. Allinson; Jadwiga A. Wedzicha; Gavin C. Donaldson


Respiratory Research. 2015;16(71) 

In This Article


Patient Recruitment

The London COPD cohort is a group of approximately 200 COPD patients under longitudinal observation at the Centre for Respiratory Medicine, University College London. This cohort was started in 1995 for the prospective investigation of COPD exacerbations. Patients who withdraw or die are replaced on a rolling basis. COPD is defined as a Forced Expiratory Volume in 1 s (FEV1) ≤ 80 % of a normal value predicted from age, height, and sex and a FEV1/Forced Vital Capacity (FVC) ratio < 0.7. Patients enrolled in the cohort complete daily diary cards and are seen in clinic every 3 months if stable and annually undergo a comprehensive medical review. Patients were also seen at exacerbation and most were prescribed oral corticosteroids and/or antibiotics. Patients with any other primary respiratory diseases or who are unable or unwilling to complete daily diary cards were excluded.

In April 2011, there were 199 patients enrolled in the London COPD cohort. 24 patients were ineligible as they used a walking support (cane or frame) or were confined to a wheel chair or used ambulatory oxygen cylinders and 30 refused. We eventually provided pedometers to 145 patients. Data was successfully acquired from only 73 patients for the following reasons a) 21 patients once issued refused to use the pedometer, b) 19 patients lost their pedometers, c) 23 patients recorded less than 35 days of data whilst stable due to repeated exacerbation and, d) 9 pedometers malfunctioned. The study ended in March 2013.

A full medical and smoking history was taken and measurements of FEV1 and FVC made with a Vitalograph Gold Standard spirometer (Vitalograph Ltd, Maids Moreton, UK). Body mass index (BMI) was calculated from height and weight.

Monitoring. Patients were educated to use diary cards at the recruitment visit and re-educated as needed when visiting the clinic. The diary cards also have instructions (how to fill the card and how use the pedometer) and contact numbers on the back of every card. All patients kept a daily diary card on which they recorded any worsening in their respiratory symptoms, the number of hours spent outside their home and their daily peak expiratory flow (PEF) measured with a mini-Wright meter (Clement-Clark International, Harlow, UK) once a day at morning. Patients were instructed to wear a pedometer (Yamax Digi-Walker SW-200) on left side of body all the time, except when sleeping or showering. Patients recorded daily step counts on written daily diary cards. This pedometer has been shown to accurately measure steps in free-living individuals[18,19] and in normal and moderately obese patients and[20] detected differences in physical activity of COPD patients.[21] Patient also completed a daily COPD Assessment Test (CAT) questionnaire after first being trained in clinic. Pedometry data collected over the initial 7 days were discarded to avoid any learning effects and only patients who had recorded more than 35 days of data were included in this analysis.

Exacerbations. Exacerbations were identified according to our usual criteria of increases in any two major symptoms (dyspnoea, sputum volume or sputum purulence) or one major and one minor symptom (nasal congestion, wheeze, cough, sore throat) over two consecutive days.[22] Data recorded two weeks either side of the onset of an exacerbation were excluded from the analysis.

Ethics. The study was approved by the London-Hampstead research ethics committee and all patients gave written informed consent (REC 09/H0720/8).

Temperature and Pollution Data. Daily data for atmospheric PM10 and ozone (O3) were obtained for Bloomsbury Square, Central London from the Air Quality Information Archive databases ( Data from the archive is reported as μg/m3 . The conversion factor of ozone is 1 ppb = 1.9957 μg/m3 at 20 °C and 1013 millibar atmospheric pressure. We did not use data from the monitoring site in Hackney that would be closer to our patients because it did not record data on PM10 which we have previously shown to increase dyspnoea.[14]

Weather data was the average of hourly readings over 24 h at Heathrow Airport and obtained from the British Atmospheric Data Centre ( A dry day was defined as zero precipitation[23] and a sunny day arbitrarily defined as a day when the sun shone for a minimum 0.1 h or more.

Statistical Methods. Patient characteristics are summarised as appropriate by a mean and standard deviations or standard errors, or a median and inter-quartile ranges, or as a percentage.

Unadjusted Analysis. Generalised estimating equations (GEE) were used to model the effects of weather and pollution on daily step count, PEF, CAT scores (assuming their Gaussian distribution), time (hours) outdoors (as Poisson distributed) or worsened dyspnoea (with a Bernoulli distribution) on days with temperatures ≤22.5 °C. It was an a priori decision that a cut-off would be necessary as relatively hot weather can reduce time spent outdoors.[24] To identify the inflexion in the relationship between activity and temperature we plotted mean daily step count against temperature in 0.25 °C intervals. After inspection, a cut-off of 22.5 °C was chosen as daily step count was highest at this temperature and decreased with temperatures below or above 22.5 °C. GEE models were used to examine our panel data as they correct the standard errors and p-values for the various regression coefficients for the correlation structure between the repeated measurements on the same patient. We used the xtgee command in Stata with the robust option as this would produce valid standard errors even if our assumption of an independent correlation structure was incorrect.

Comparisons between daily step count of a sunny compared to a dull day, or a dry versus rainy day were made by paired t-test, after first obtaining the average for each patient under the various conditions.

Analysis of variance (ANOVA) was used to determine the effect of day of the week on daily step count, hours outdoors, O3 and PM10 . Post-hoc comparisons were made between Sunday and Saturday, and between Saturday and Friday.

Adjusted Analysis. GEE regression models were used to assess the independent effects of climate and pollution on daily step count and the other outcome measures. These models included a linear term to adjust for age related decline, sine and cosine terms with periods of 12, 6 and 4 months to allow for seasonal changes, and a variable for day of week with Monday as the first day of the week. The covariates also included daily temperature, wind speed, rainfall, hours of sunshine and day-length, PM10 and O3 as independent variables. We did not examine any lagged effects of climate or pollution. The analysis was repeated using data collected during week-days only (Monday-Friday) and during week-ends (Saturday and Sunday) since activity was markedly dissimilar in these periods. The analysis was also repeated with an auto-regressive term (the previous day value of the dependent variable) in the model to adjust for autocorrelation in the dependent variable. We also repeated the analysis of daily step count and the pollutants with time outdoors included as an independent variable.

Distance to Pollution Monitoring Site. The coordinates of Bloomsbury Squares and the patient's home (defined as the centre of their post-code) were obtained from the National Statistics postcode directory database. The straight-line distance between the two sets of coordinates was calculated by Pythogras' theorem.