Statistics is basically an exercise in probability. It is something like the arguments about correlation versus causation. There's a stronger statistical correlation when P equals or is less than 0.05 versus P equals or is less than 0.10 by a factor of at least 2. The usual scientific convention is to use P equals or is less than 0.05, i.e. picking random samples you would expect an erroneous result less than one out 20 samples. Any corrections will be appreciated.
Obesity, Smoking Speed Aging Process
FReepmail me if you want on or off my health and science ping list. Every puff and added pound makes cells old before their time, study finds
MONDAY, June 13 (HealthDay News) -- Researchers are providing one more reason to drop excess weight and quit smoking: a new study finds that both accelerate human aging.
The new study included more than 1,100 British women between 18 and 76 years of age. The women filled out a questionnaire on their smoking history and provided blood samples, which were also tested for concentrations of a body fat regulator called leptin and for telomere length.
Telomeres cap the ends of chromosomes in cells and protect them from damage. However, each time a cell divides -- and as people age -- these caps get shorter, so decreases in telomere length have long been associated with the aging process.
Reporting June 14 in the early online edition of The Lancet, the British researchers found that telomeres of obese women and smokers were much shorter than those of lean women and those who'd never smoked. In contrast, lean women had much longer telomeres than moderately overweight women who, in turn, had longer telomeres than obese women.
Each pack-year (the number of cigarettes smoked per day times the number of years of smoking) smoked was equivalent to an 18 percent telomere shortening, in addition to normal telomere shrinkage, the study found.
Overall, obese women aged an additional 8.8 years -- based on telomere length -- compared to lean women, the researchers reported. A current or previous history of smoking entailed an average 4.6 year increase in aging compared to never-smokers, while those with long-term smoking habits -- a pack-a-day for 40 years -- added an additional 7.4 years of aging to their life compared to those who stayed away from cigarettes completely, the study found.
"Our results emphasize the potential wide-ranging effects of the two most important preventable exposures in developed countries -- cigarettes and obesity," researcher Tim Spector of St. Thomas' Hospital, U.K., said in a prepared statement.
More information
The Alliance for Aging Research has more about aging.
SOURCE: The Lancet, news release, June 14, 2005
FReepmail me if you want on or off my health and science ping list.
When a (classical) statistical judgment is made (for example, A > B), a confidence interval is supposed to be given also. It is a measure of "how confident" one is in the result. (I won't go into the philosophy.) The primary factor in determining confidence intervals is the number of samples; more samples generally mean more confidence.
A confidence interval of 5% basically means that, "Were many randomly chosen samples measured, 5% would have worse results than those of the actual sample." The 5% is arbitrary; it stems from the early 1900s when computation of confidence intervals was done with pencil and paper (or chalk and slate). Tables were published at the 5% level and this became the norm.
The criticism of the published result is that it used a 10% confidence level. This means that the results are more "uncertain" or that the conclusion is "weaker" than normal.
There is some criticism that the observations have systematic (rather than sampling) errors. I haven't read the original work, so I don't know about the results. I just like to criticize methodology.