| Ref ID | 493 |
| First Author | J. P. Ioannidis |
| Journal | BMJ |
| Year Of Publishing | 2010 |
| URL | https://pubmed.ncbi.nlm.nih.gov/20837576/ |
| Keywords |
• Pharmacological • Open data • Multiplicity • Publication bias • Pre-specification • External validity |
| Problem(s) |
• Ignores setting or context of included studies which limits review applicability • Overly stringent inclusion criteria affecting external validity • Perpetuates citation of poor quality primary study data |
| Article Type | Editorial |
| Article Subtype | Discussion piece |
| First Author Country | United States |
| Aim | An editorial analysis illustrating how systematic reviews comprised of trials with early stopping rules can lead to overestimated treatment effects, particularly when the same drug is being examined in multiple indications, and particularly when interventions are only compared to placebo. Examples, including one of anti-TNF agents, are discussed. The authors searched MEDLINE and ClinicalTrials.gov (April 2010) for trials and for published systematic reviews to illustrate how reliance on trials which are stopped early can lead to regulatory approval of expensive drugs with low survival benefits. |
| Level of Investigation | Descriptive |
| Summary of Findings | All systematic reviews dealt with a single indication (for either one agent or multiple agents) but none considered several indications. 188 relevant trials were identified for 42 indications-pairs. Only 5 compared head to head different anti-TNF agents, and another 32 trials compared an anti-TNF agent with one or more active comparators (only 4 published at time of writing). All other trials were comparisons against placebo or no treatment. Overall, the published trials represented only 34% of the total evidence (46% if ongoing trials were excluded). |
| Number of systematic reviews included | 34 |
| Number of eligible systematic reviews assessed | 34 |
| Treatment impacted | No |
| Treatment impacted description | |
| Interpretation impacted | Not Applicable |
| Interpretation impacted description |