Table 1 Common problems in the reporting of acne trials
Problem | Description | References |
|---|---|---|
1. Insufficient power | Underpowered trials can produce false negative results in superiority studies or incorrect claims of equivalence | Pediatric Dermatology [4]; Indian Journal of Dermatology, Venereology and Leprology [5] |
2. Duplicate publication | Publication of the same trial more than once can artificially enhance its impact and distort subsequent meta-analyses | American Journal of Clinical Nutrition [8]; Journal of the American Academy of Dermatology [9]; Contraception [11]; Cutis [12] |
3. Incorrect statistical comparison | A "within-groups" comparison from baseline may give positive results when the correct "between-groups" comparison is negative | |
4. "Salami publication" | Splitting the results from a single trial to produce more than one publication can artificially increase its impact | |
5. Inferiority margin not pre-specified | In non-inferiority studies, lack of a pre-specified inferiority margin means that the margin might have been chosen in retrospect to fit the data | |
6. Two independent trials combined and reported as one | Independent trials should be analysed and reported separately before combination in any subsequent meta-analysis | Cutis [22]; Journal of the American Academy of Dermatology [23] |
7. Loss of masking due to trial therapies not considered in "double-blind" trials | Comparators with different physical characteristics or adverse effect profiles can cause loss of participant or investigator masking | European Journal of Dermatology[25]; International Journal of Cosmetic Science [26] |
8. Stating P values without publishing outcome data | P values can be misleading without confidence intervals and original outcome data | Indian Journal of Dermatology [27] |
9. Failure to account for all randomized participants | Absence of an intention-to-treat analysis raises the possibility of attrition bias due to loss of study participants before the primary endpoint | International Journal of Cosmetic Science [26] |
10. Selective outcome reporting | Multiple endpoints, rather than a single primary endpoint, allow "data fishing" in which only the positive outcomes are highlighted | Journal of Drugs in Dermatology [28] |
11. Treatment effects statistically significant but clinically insignificant | Highly significant P values may mask a small improvement in disease severity that is insufficient to be of clinical benefit to patients | Journal of Drugs in Dermatology [29] |
12. Odds ratios used to exaggerate treatment effect | Odds ratios can be misleadingly large when event rates are high - rate ratios give more understandable results | Contraception [11] |