THIS ANALYSIS IS TENTATIVE AND BOTH THE ANALYSIS AND ITS CONCLUSIONS MAY CHANGE

In particular, we are trying to understand the possibly conflicting information in ...

Effects of Air Pollution

In order to guide us in understanding the significance (or otherwise) of the readings from the St. Marys Lane sensor we looked for the single figure that best represented the average, overall effect of the sort of pm2.5 and pm10 concentrations we saw (i.e. pm2.5 values around 5 - 15 μg/m3).

  1. We found estimates of the effects of air pollution on several individual medical conditions, but fastened on long-term, all-age, all-cause, death rate, as being more generally applicable. We hoped this could be used to understand and check...
    1. community action or policy; and
    2. individual decisions, such as whether to stay indoors on polluted days.
  2. Since the smaller PM2.5 particle concentration was described as having the stronger correlation with all-cause, death rate, and since we could find no mention of advantages of combining this with PM10 values, we decided to base our figure on the PM2.5 values alone.
  3. Since the St. Mary Lane sensor does not have NOx, O3 or other pollution sensors, we thought it better to use the PM2.5 value to represent overall air pollution. To do this, we concentrated on research which did not control for other air pollution variables, only for social, economic, geographical, and other non-air pollution variables.
  4. Even then, we found that different studies implied widely different effect sizes (see table 1.1). We eventually selected the 0.6% change in death rate for each 1μg/m3 of pm2.5 implied by the UK Committee On the Medical Effects of Air Pollutants as being both UK specific and authoritative.
In comparing the above figure with that in the published studies, note that ...
  1. Some studies quote the effect of a larger, 10ug/m^3 change in pm2.5, which (since we assume linearity), should be divided by 10.
  2. Other studies quote the differences in death rates between co-cohorts having even larger differences in air pollution. Confusingly, these are sometimes reported without specifying the difference in pm2.5 which is producing the effect.
  3. Still other studies report the effect of pm2.5 on specific diseases. Apart from the very common cardiovascular diseases, these can be lower, or very much higher without affecting the all-cause death rate given here.
  4. The long-term effect is also very different from the short-term effect of changes in pm2.5, which are better estimated by "time-series studies" of, for instance, pollution peaks.
> In support of this effect size, we quote ...

enable pictures to see the all cause death hazard ratio against PM2.5

This curve shows a (normalised) death rate varying with the average PM2.5 level in μg/m3 shown along the bottom. Our straight line shows our simplifying assumption. It could be argued that our assumed linear relation overestimates the effect of changes in PM2.5 in low pollution areas. However, this may not be wise. Firstly, some studies have failed to find a safe threshold for PM2.5, and actually show more sensitivity at low pollution levels, with the effect of PM2.5 leveling off at higher levels. Secondly, because particulate levels which are not great enough to kill could nevertheless have extremely deleterious effects on quality of life.

Examples of Pollution Reduction

Enable pictures to see a map of average UK pm2.5
Enable pictures to see a map of average UK pm2.5
Enable pictures to see a map of average UK pm2.5
  1. Over an average year life span, the 0.6% reduction in death rate associated with a 1μg/m3 reduction in pm2.5 would result in an increase in life span of 81.1 * 0.6% = 0.49 years or about 6 months. If, as seems likely, improvements in air pollution improve health as well as death, this could be an extra half-year of good health.
  2. To see the implications of this for the appropriately 7,000 residents of Ecclesfield Village, we considered the effect of a village-wide 1μg/m3 reduction in PM2.5. Assuming that the UK average death rate of around 1% per year applies to Ecclesfield, this would reduce the expected 7000 * 1% = 70 by 70 * 0.6% = 0.42 deaths/year. If this could be maintained over an 80 year lifetime, then it could give the average resident in the village half a years extra life. However, the feasibility of any such reduction must be limited by the fact that only an estimated 55% of UK PM2.5 originates in the UK, the rest originating from Continental Europe and elsewhere. Something of this can be seen from the average PM2.5 map on the right.
  3. To see what progress has already been made, we looked at the UK between 1970 and 2010, and found that regulation and technology improvements reduced population average PM2.5 concentration from about 16μg/m3 to about 7.5μg/m3. This was said to have eliminated the cause of 6.62% of all deaths. If this is right, then it seems to explain most of the 80.40 - 71.97 = 8.43 years of extra life expectancy during this period.

Comparison with other harms

Some Sources of PM2.5

Note that once generated, any pollution may become trapped by temperature inversions - e.g. when warmer air lies over colder valleys, hidden from the sun. Alternatively, air pollution may be rapidly diluted by high wind speeds.

Some Problems and Qualifications

In spite of these reservations, we still suggest that the pollution figures given on the pm10 measurement page together with an assumed effect size of 1μg/m3 to 0.6% death rate is a useful measure of the importance of reducing PM2.5 in the general case where there is no more specific information.

Pollution Events

Possible Objections

Although we lack training in epidemiology or medicine, in writing the above, we tried to take account of arguments to the effect that there is no significant causal link between PM2.5 and death rate. However, the objections we found did not seem strong enough to override what appeared to the the consensus. Some of the objections seemed to centre on the actions which California has already taken to reduce air pollution. Sadly, arguments seem inevitable in the softer sciences, and become prominent objections whenever large commercial interests are threatened. The mere fact of these objections does not therefore add information or change our conclusion above. We found the web site of one skeptic however, which although it contained video and other emotional material which might be capable of doing more damage to one's peace of mind than the PM2.5s themselves, also contained material we could not find the full-text of elsewhere. Having provided this health-warning we therefore link to http://www.scientificintegrityinstitute.org.

Historical Data

Luftdaten is an "Open Data" project. Their historical data is therefore available for download, which can be done from archive.luftdaten.info. Click on the folder with required date to see the list of files for that day. The St. Marys Lane data, seems to be stored in file *sds011_sensor_31702.csv.