| Davis, 1988, UK | 100 patients >40 years with a 1st Colles’ fracture given flurbiprofen 50/100 mg twice daily or placebo for 2/52 | RCT | Functional recovery | Some early delay with active treatment (4–6/52) but no difference at 1 year | 1. High dropout rate (24%) 2. Possibility of self-administered confounding treatments not accounted for |
| Adolphson, 1993, Sweden | 42 post menopausal women with first Colles’ fracture given piroxicam 20 mg once daily or placebo for 8/52 | RCT | Radiological and functional recovery | No difference in recovery rate up to 12/52 | 1. Small study 2. Short follow up 3. Possibility of self-administered confounding treatments not accounted for 4. Randomisation process not specified and significant differences in demographics between groups |
| Giannoudis, 2000, UK | 99 patients who had undergone intramedullary nailing of femoral shaft fractures over a 6 year period. 32 patients had suffered from a non-union while 67 had successful bone healing. | Case control study | The patients were telephoned with a questionnaire about a number of factors which included NSAID use. Their notes were scrutinised for the type of operative procedure and device used (which was the primary variable under investigation). | A significant difference was found in NSAID use between the non-union (62.5%) and successful healing (13.4%) groups (p<0.0001). The non-union group tended to use NSAIDs for longer (21 v 1 week). | 1. Authors attributed a causal relation to the association found between non-union and NSAID use. Prolonged NSAID use may have been result, rather than cause, of non-union 2. No data given as to the degree of matching of the control cases. 3. Not a typical ED patient population |