Analysis of Newspaper Research Report Results

This paper is going to present discuss from a statistic point a view a health related newspaper article in which a research study is mentioned and summarized. I will also look into how the conclusions are presented, and if the general approach was correct and suitable as far as statistics is involved.

The article is named: “Pills protect against ovarian cancer: Study”, it was written by Helen Branswell, and was published by Toronto Star on January 24th, 2008 in its Living/Health section. (See references for the link)

The original study was published in a medical journal, The Lancet, by Dr. Valerie Beral of Cancer Research UK epidemiology unit at Oxford University. The study is in reality a meta-study as “is actually a pooling of data from 45 previous studies”. The total number of the women followed is very large: over 110,000, with about 40,000 of them had used oral contraceptives. The large size of the sample can make the results of the meta-study very precise, as long as the selection of the participants followed the same statistical rigorous procedure.

The article tries to give a background of this medical problem, and cites The Canadian Cancer Society when saying that it is estimated 2,400 women were diagnosed with ovarian cancer in Canada in 2007, and 1,700 women died from it. These numbers were probably determined using inferential statistical method (recording the occurrence and fatality in a small sample, generalizing for the whole country after that).

Most probably the same statistical procedure was used when the study estimated the worldwide numbers: 200,000 cases of ovarian cancer and 100,000 deaths from it.

The article states that the protective effect has been recorded regardless of ethnicity, education level, age, medical history, weight, height, smoking or drinking history. This statement makes me think that there were studies where the subjects were all the same ethnicity, education, age, and so on.

On the other hand, when The Canadian Cancer Society says that ovarian cancer “ranks fifth, after lung, breast, colorectal and pancreatic cancer” a cause of death, this is calculated after recording the death causes for the whole population, and checking the frequency for each cause.

Building on the results found until then, the article mentions that considering that the numbers of women using oral contraceptives have raised in the past decades, an estimated number of 30,000 deaths due to ovarian cancer has been prevented. This number was probably determined as follows:

1. Determine the fatality rate without any pills being taken

2. Estimate the reduced fatality rate for women taking the pills

3. Estimate (extrapolate current trends) the percentage/number of women that