Is any one analgesic superior for episodic tension-type headache?

Methodological quality and data extraction

Two authors (LD with MB, BK, or AV) independently rated the methodological quality of the included trials using the Delphi list.¹⁰ The Delphi list is a generic criteria list developed by international consensus and consists of the following 9 items: 1) randomization; 2) adequate allocation concealment; 3) groups similar at baseline; 4) specification of eligibility criteria; 5) blinding of outcome assessor; 6) blinding of care provider; 7) blinding of patient; 8) presentation of point estimates and measures of variability; 9) intention-to-treat-analysis. One extra item was added: 10) withdrawal or dropout rate unlikely to cause bias. All selected methodological criteria were scored as yes (=1), no (=0) or don’t know (=0). A quality score of a trial was computed by counting the number of positive scores, with equal weights applied on all items. In case of a disagreement between the 2 authors, consensus was used to resolve disagreement. When consensus could not be reached, a third author made the final decision (MB or AV).

Extraction of data from the original reports was performed by 1 author (LD) and checked by a second (AV). Disagreements were resolved by consensus. Extracted information included (if available) demographic data, detailed description of the intervention and control (ie, dose given, study duration, rescue medication), data on pain relief or recovery, and information on adverse effects measured during a treatment period of 2 to 6 hours. When a trial protocol permitted the use of rescue medication prior to the outcome time (2 to 6 hours), then the latest outcome assessment not confounded by the use of rescue medication was extracted

Data analysis

A quantitative analysis was limited to clinically homogenous studies for which the study populations, interventions and outcomes were considered to be similar. For each study, the number of patients who were recovered (often defined as more than 50% pain relief) was used to calculate relative risk (RR) with 95% confidence interval (CI). RRs and 95% CI were presented using the random effects model. Data are presented as treatment success, indicating that an RR >1 represents a better outcome for the first mentioned medication group.

In parallel studies, when more than 1 comparison from the same study (ie, aspirin 650 mg vs placebo and ibuprofen 400 mg vs placebo) was used for the statistical pooling of NSAIDs vs placebo, the results from the placebo group were evenly spread out over the 2 comparisons and the number of patients in the placebo group was divided by 2 in order to prevent double counting (personal communication RJPM Scholten, Dutch Cochrane Centre).

Because only a subset of available trials provides sufficient data for inclusion in the quantitative analysis, also a qualitative analysis was performed. We summarized findings by strength of evidence, nature of intervention and control treatments. The evidence was judged to be strong when multiple high-quality trials produced generally consistent findings.¹¹ Results were considered consistent if over 75% of the studies reported similar results on the same outcome measure. It was judged to be moderate when multiple low-quality trials or one high-quality and 1 or more low-quality trials produced generally consistent findings. Evidence was considered to be limited when only 1 low-quality RCT existed and conflicting when the findings of existing trials were inconsistent. We arbitrarily regarded trials with methodological quality scores of 6 or more as of high quality.¹¹

Relation between funding source of the RCTs and conclusions

We extracted the sources of funding of the RCTs from the text, statements of sources of support, authors’ affiliations, and acknowledgements. Funding sources were classified as nonprofit organizations, not reported, both nonprofit and for-profit organizations, or for-profit organizations.¹² For-profit organizations were defined as companies that might acquire financial gain or loss depending on the outcome of the trial.¹² Funding included provision of grants, study material (drug, placebo), or manpower (authorship, statistical analysis, or other assistance).¹² We used the effect sizes between medication(s) and placebo to evaluate whether funding source affected outcome.

Results

Search results

A total of 1878 publications were identified by our search strategy. Finally, 41 RCTs met our inclusion criteria and 4 papers concerned double publications (FIGURE 1),^13-16 leaving a total of 41 trials which were included in this review. Thirteen of these RCTs used a crossover design.^15,7-27

FIGURE 1
How the 41 trials made our cut for the review

Description of studies

Full details of the included studies are presented in TABLE W1. The number of participants included in each trial ranged from 12 to 900 (mean=252.7 patients), with a total of 10,363 patients included. The mean percentage of participants who dropped out from the trials was 15.2% (range=0%–61.9%). Age of participants (for studies reporting this information) ranged from 18 to 87 years. Overall, the percentage of women was generally higher than men (mean=69.3%; range=36%–97%). Fifteen trials used the criteria of the International Headache Society to classify TTH,^{14,17,19-21,24,28-36} 12 trials used the Ad Hoc Committee’s criteria,^{13,23,26,37,45} while the remaining studies did not use a formal classification.