Darm exercise 3 J-P Männikkö

From Testiwiki
Jump to: navigation, search


This analysis seeks to answer these questions:

How many times should action to reduce PM 2.5 by 2% be taken so that it would still be cost beneficial? Assumptions: cost of action 3 million euros, cost of death 1 million euros.

This analysis also seeks to address the following questions:

  • Mean PM2.5 concentration in 2015 in all three decisions (BAU, Option 1, Option 2)
  • Cost of health effects in all three scenarios
  • Comparison of health effect costs to different decision costs.
  • Sensitivity of the result to dose-response and to concentration uncertainties (thus, which uncertainty has more impact to the results).
  • Reviewers are lazy people so report should be short. Both report and R-code are submitted to individual Wiki page in HEANDE.


2015 PM2.5 levels 5.0 μg/m3 if BAU. Standard deviation of normal distribution 0.5.

Decision action would lower mean PM2.5 concentrations by 2 % and cost 3 000 000 euros, repeatable.

Dose-response data: Pope et al. 2002 epidemiological data. 10 ug/m3 change in PM 2.5 affects mortality by factor 1.06.

Background mortality is estimated to be 883 deaths per year in year 2015.

One death has economical value of 1 000 000 Euros.


If dose-response factor to 10ug/m3 change in PM2.5 is 1,06, 5 ug/m3 PM2.5 level would result in (5/10)*(1,06-1) = 0,03 (3%) increase in mortality. Based on this it is possible to calculate the effect of 2% decrease in PM 2.5 level with the cost of 3 million euros and compare it to cost of 1 million euros per death. Calculation is presented in the following table:

Scenario PM 2.5 ug / m3, 2015 Mortality Background 2015 Extra Deaths Cost of deaths, me Cost of reduction, me Total cost
BAU 5 0,03 883 26,5 26,5 0 26,5
1 4,9 0,0294 883 26 26,0 3 29,0
2 4,802 0,028812 883 25,4 25,4 6 31,4

Based on the calculations above, 3 million euros for a 2% reduction in PM 2.5 level would not be cost effective in any case compared to cost of 1 million euros per death.

1 : The calculations and conclusions are correct. The R code functionality is not a necessity, although we promote its use. --Jouni 10:19, 26 May 2011 (EEST)

2 : Sensitivity analysis is missing about which uncertainty (about dose-response or about exposure) is more important. --Jouni 10:19, 26 May 2011 (EEST)

Score: 4/5.

See also