RSS Feeds
Locating and Appraising Systematic Reviews
The Editors welcome submissions for possible publication in the Letters section. Authors of letters should:
•Include no more than 300 words of text, three authors, and five references
•Type with double-spacing
•Send three copies of the letter, an authors' form signed by all authors, and a cover letter describing any conflicts of interest related to the contents of the letter.
Letters commenting on an Annals article will be considered if they are received within 6 weeks of the time the article was published. Only some of the letters received can be published. Published letters are edited and may be shortened; tables and figures are included only selectively. Authors will be notified that the letter has been received. If the letter is selected for publication, the author will be notified about 3 weeks before the publication date. Unpublished letters cannot be returned.
Annals welcomes electronically submitted letters.
TO THE EDITOR:
Hunt and McKibbon make a serious mistake in their discussion of what they call the test for homogeneity [1]. The authors' decision to use this term rather than the more appropriate term, the test for heterogeneity, underscores their failure to account for the type II error inherent in tests of significance based on P values [2]. The null hypothesis in the test for heterogeneity is that the different studies have underlying populations with the same mean and variance for the outcome under study. Although a P value less than 0.05 may be cause to reject this null hypothesis, a P value of 0.05 or greater is not sufficient evidence to accept the null hypothesis. A nonsignificant P value tells us only that we cannot be more than 95% sure that the underlying populations are different. Under this stringent criterion, we can say only that the differences between groups may be due to chance, not that they are due to chance. This lack of symmetry of statistical testing with P values must always be kept in mind.
More important, the problem of deciding whether it is appropriate to combine different studies in a systematic review is similar to the assessment of confounding due to bias. In both cases, determining whether differences are important does not depend on whether they are due to chance; it depends on how much they affect the results [3]. There is no rule for how much difference is important. The authors' responsibility is to convince the readers that any differences are not likely to be material. One effective method may be to note any studies that seem discrepant and analyze the results with or without them. Readers can then judge for themselves whether the authors are justified in their use of multiple studies in a meta-analysis.
Carl D. Atkins, MD
South Shore Hematology-Oncology Associates; Rockville Centre, NY 11570
The Editors welcome submissions for possible publication in the Letters section. Authors of letters should:
•Include no more than 300 words of text, three authors, and five references
•Type with double-spacing
•Send three copies of the letter, an authors' form signed by all authors, and a cover letter describing any conflicts of interest related to the contents of the letter.
Letters commenting on an Annals article will be considered if they are received within 6 weeks of the time the article was published. Only some of the letters received can be published. Published letters are edited and may be shortened; tables and figures are included only selectively. Authors will be notified that the letter has been received. If the letter is selected for publication, the author will be notified about 3 weeks before the publication date. Unpublished letters cannot be returned.
Annals welcomes electronically submitted letters.
- Copyright ©2004 by the American College of Physicians
