Table 1

Example of an evidence profile quality assessment given to the evaluators for them to grade in the pilot study. Example question: Should depressed patients in primary care be treated with SSRIs rather than tricyclics?

Outcome: Depression severity (measured with Hamilton Depression Rating Scale)

Studies
Design
Quality
Consistency
Directness

8 trials Citalopram
38 trials Fluoxetine
25 trials Fluvoxamine
2 trials Nefazodone
18 trials Paroxetine
4 trials Sertaline
4 trials Velafaxine
RCTs
No serious flaws
No important inconsistency
Some uncertainty about relevance (outcome measure)

Outcome: Transient side effects (drop-out from 6 week treatment)

8 trials Citalopram
50 trials Fluoxetine
27 trials Fluvoxamine
4 trials Nefazodone
23 trials Paroxetine
6 trials Sertaline
5 trials Velafaxine
RCTs
No serious flaws
No important inconsistency
Some uncertainty about relevance (outcome measure)

Outcome: Poisoning fatalities

Office for National Statistics (British)
Observational data (national statistics)
Serious flaw, population based
Reporting bias
Only one study
Direct

Atkins et al. BMC Health Services Research 2005 5:25   doi:10.1186/1472-6963-5-25