Article type
Year
Abstract
Background:
Researchers have identified cases in which newspaper stories have put a spin on medical studies reported in original articles. Moreover, exaggeration sometimes starts with the journal articles themselves.
Objectives:
We examined how much spin there was in newspaper reports, and what proportion of the studies quoted in newspaper stories could be confirmed.
Methods:
We identified the newspaper stories that mentioned the effectiveness of certain treatments or preventions in the year 2000, and that were based on original studies from 40 main journals. For each original study, we searched subsequent studies with stronger research designs.
Results:
We found 164 original articles and randomly selected 100. The proportion of spin was 4.2% (95% confidence interval (CI) 1.2 to 7.3), and 78.0% (95% CI 67.8 to 85.7) of the original studies were confirmed. However, among the confirmed studies, the median standardized mean difference reported by the subsequent studies was 0.1 standard deviations (SD) smaller than that of the original studies. We considered only half to have been replicated both in the direction and magnitude of the treatment effect.
Conclusions:
Readers of newspapers should be aware that some of the claims made in highly circulated newspapers based on high-profile journal articles may still be overturned by subsequent studies.
Patient or healthcare consumer involvement:
None.
Researchers have identified cases in which newspaper stories have put a spin on medical studies reported in original articles. Moreover, exaggeration sometimes starts with the journal articles themselves.
Objectives:
We examined how much spin there was in newspaper reports, and what proportion of the studies quoted in newspaper stories could be confirmed.
Methods:
We identified the newspaper stories that mentioned the effectiveness of certain treatments or preventions in the year 2000, and that were based on original studies from 40 main journals. For each original study, we searched subsequent studies with stronger research designs.
Results:
We found 164 original articles and randomly selected 100. The proportion of spin was 4.2% (95% confidence interval (CI) 1.2 to 7.3), and 78.0% (95% CI 67.8 to 85.7) of the original studies were confirmed. However, among the confirmed studies, the median standardized mean difference reported by the subsequent studies was 0.1 standard deviations (SD) smaller than that of the original studies. We considered only half to have been replicated both in the direction and magnitude of the treatment effect.
Conclusions:
Readers of newspapers should be aware that some of the claims made in highly circulated newspapers based on high-profile journal articles may still be overturned by subsequent studies.
Patient or healthcare consumer involvement:
None.