Numbers game

February 26, 2006
The NHS's cancer failure

Just how good is the NHS? International comparisons exist in one area that really matters: survival rates from cancer. Our centrally planned system, unique in the developed world, does not come out of this well. Five-year survival rates show the percentage of those diagnosed who are still alive after five years. In England, for the most commonly diagnosed cancers, the rates are 36 per cent for men and 47 per cent for women (the disparity is mainly due to the different types of cancer men and women get). Across the EU as a whole the figures are 40 and 51 per cent respectively. And these include the new poor countries from the east. In France the averages are 45 and 59, in Germany 45 and 55, Italy 39 and 53, Spain 44 and 56 and the Netherlands 46 and 54.

American survival rates are often even better. For ovarian cancer, for example, the American Cancer Society in 2004 put the five-year survival rate at 52 per cent, the rate in England and Wales is just 31 per cent.

Modelling climate change

The world of climate change modelling has recently been thrown into confusion by two very important new discoveries. First, it turns out that aerosols, far from contributing to global warming, actually reduce it. Second, forests have been found to cause large amounts of methane emissions. And methane is believed to be a major factor in global warming.

Neither of these discoveries were factored into the complicated models of climate change that predict such dire consequences for the world over the next century. These omissions really ought to lead to a serious dose of humility among climate change modellers. But if the experience of economic models and forecasts is anything to go by, the scientists involved will claim even greater certainty.

The parallels with economic forecasting go even deeper. The basic data of these models, the global temperature, is based on uncertain estimates. These estimates for the past change from year to year, just like economic data. And whatever version of the data is used, it is prone to almost random fluctuation over time.

The most reliable data, published by the National Climatic Data Centre, goes back to 1880. This data suggests that the pace of global warming has slowed down. The average global temperature in 1880 is estimated to have been 13.67 degrees Celsius. By 1950 this had risen to 13.78, and even by 1975 only to 13.81. The 1975-90 period saw a jump to 14.24, around 0.8 degrees, but over the next 15 years from 1990 to 2005 the increase has only been 0.2 degrees.

Various datasets provide estimates that go back a thousand years. The recent climate has certainly been warm, matched only by the decades towards the end of the 11th century. But the data shows virtually no patterns, whether in terms of year on year changes or in terms of the regularities of fluctuations over longer periods of time.

The daddy of them all is the data based on drilling the Antarctic ice cap at Vostok, which gives temperature estimates for the past 420,000 years. There was a very marked warming as the last ice age ended. But since then there is no discernible trend up or down in the global temperature estimates. And, just as with the shorter data series, the changes over the past 12,000 years show very little evidence of the patterns that are necessary for accurate prediction of the future.