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 Characteristics

The 73 COPD patients (51 Male, 22 Female) studied had moderate to very severe COPD (Table 1). There were no significant differences in the patient characteristics between the 73 patients involved in this study and 126 patients excluded for reasons described in the methods. The patients recorded daily step count on 16,478 days with an average per patient of 267 days (range 29–658). Of these, 3020 days were excluded, as exacerbations commenced either two weeks before or after.

The patients lived on average 7.39 km (SD 4.70) from the Bloomsbury Square site. Of the 73 patients, 59 lived north-east, 7 north-west, 2 south-east and 5 south-west of Bloomsbury.

The average of 225 days of pedometry readings per patient (SD 139; range 29–578); 459 days of PEFR readings per patient (SD 139; range 124–768); 463 days per patient of whether or not dyspnoea was worse than usual (SD 138; range 124–680) and 70 days with a CAT score per patient (SD 87; range 0–356). During the week, when patients might be at work, the mean of each patient's average time outside the home per day was 3.05 h (SD 1.51; range 0.52 to 7.3 h). Over the whole week, there was a strong relationship between the average number of steps per day and the average time spent outdoors (regression coefficient =671 steps per day per hour outdoors; intercept = 1804 steps per day; p = 0.001; see Fig. 1).

Figure 1.

Relationship between the average steps per day for each patient and the average hours spent outside the home during the whole week

Unadjusted Analysis

Warmer weather was associated with increased daily step count (Fig. 2). A 1 °C rise in temperature increased the count by 43 steps per day per °C (95 % CI 2.14 to 84.4; p = 0.039). However, when the temperatures exceeded 22.5 °C, patient activity appeared to decrease and steps per day fell by -891 per 1 °C rise (95 % CI -1735 to -47; p = 0.038).

Figure 2.

Relationship between daily step count and daily temperature; data is averaged in 1°C intervals. Bars are standard errors

Physical activity was higher on days with sunshine or without rain (Fig. 3). The mean of patient's average step count on sunny days was 3938 per day (SD 2447) compared to 3596 per day (SD 2260) on overcast days (paired t-test; p < 0.0010). Similarly, on dry days the mean of each patient's average step count was 3999 per day (SD 2507) compared to 3771 per day (SD 2349) on days with rain (p < 0.0001).

Figure 3.

a Daily step count on Overcast versus Sunny days. b Daily step count on Dry versus Wet days. Data are means ± standard errors of the average for each patient; p-values by paired t-test

The day of week effected both daily step count and hours outside. A post-hoc analysis of variance showed that daily step count was 434 steps per day lower on Sunday than Saturday (p < 0.001) and 353 steps per day lower on Saturday than Friday (p < 0.001). Similarly, time outdoors was 0.55 h lower on a Sunday compared to Saturday (p < 0.001) and by 0.09 h lower on Saturday compared to Friday (p < 0.001) (see Additional file 1: Figure S1

Adjusted Analysis

Table 2 shows results from the GEE models with daily step count data recorded during either (a) the whole week and (b) over Monday to Friday (weekdays). Daily step count increased significantly with warmer, sunny weather and fell with wet weather. Over the whole week, higher O3 levels were associated with decreased activity (p = 0.005) but not with PM10 (p = 0.112). Conversely, over just weekdays, PM10 was associated with reduced activity (p = 0.018) but not O3 (p = 0.239). There were no significant seasonal effects (sine and cosine terms) with temperature included in the model. With inclusion of an autoregressive term, over the whole week, rise in O3 was still associated with reduced daily step count (p = 0.008) and rise in PM10 also significantly and independently associated with reduced daily step count (p = 0.047). Inclusion of time outdoors as an independent variable in the regression model, eliminated the effect of O3 on daily step count over the whole week (regression coefficient = -3.9; 95 % CI -8.8 to 0.9; p = 0.113) and similarly between step count and PM10 over weekdays only (regression coefficient = -4.4; 95 % CI -10.4 to 1.5; p = 0.147).

Table 3 shows only the effects of the two pollutants (PM10 and O3) on the various outcome measures over the whole week (Table 3); over week-days (see Additional file 1: Table S1 and over the week-end (see Additional file 1: Table S2

Time spent outdoors fell with higher O3 levels (p < 0.001) for data collected over the whole week, just weekdays (p = 0.001) and at weekends (p < 0.001). PM10 show no effects on time spent outdoors on either whole week (p = 0.275) or weekdays (p = 0.217) or weekends (p = 0.502). Dyspnoea increased and PEF fell with higher levels of O3 over the whole week (p = 0.015 and p = 0.054 respectively) and for weekdays only (p= 0.017 and p = 0.040 respectively) but not at weekends. No effects of PM10 were observed on daily dyspnoea or PEF. No effects of either pollution were seen on daily CAT score.

Figure 4 shows the residuals after fitting the climatic and other variables plotted against PM10 and O3 . The plots show little effect of the pollutants on daily step count, time outdoors, PEFR and dyspnoea until they exceed around 60–70 μg/m3 . CAT score appears unrelated throughout the range of pollutants.

Figure 4.

Residuals from a GEE model that included temperature, wind speed, rainfall, hours of sunshine, day length, season and linear trend, plotted against daily PM10 and Ozone (O3) levels; data are averaged over 10 μg/m3 intervals; bars as ± standard error

Figure 5 shows that O3 concentration was significantly higher by 4.6 μg/m3 and PM10 levels 1.73 μg/m3 lower during the weekend (p = <0.001 and p = 0.057 respectively).

Figure 5.

PM10 and O3 concentrations during the week between 7th April 2011 and 31 March 2013