OBJECTIVE

To determine the rate of early discontinuation and non-publication of randomised controlled trials involving patients undergoing surgery.

DESIGN

Cross sectional observational study of registered and published trials.

SETTING

Randomised controlled trials of interventions in patients undergoing a surgical procedure.

DATA SOURCES

The ClinicalTrials.gov database was searched for interventional trials registered between January 2008 and December 2009 using the keyword "surgery". Recruitment status was extracted from the ClinicalTrials.gov database. A systematic search for studies published in peer reviewed journals was performed; if they were not found, results posted on the ClinicalTrials.gov results database were sought. Email queries were sent to trial investigators of discontinued and unpublished completed trials if no reason for the respective status was disclosed.

MAIN OUTCOME MEASURES

Trial discontinuation before completion and non-publication after completion. Logistic regression was used to determine the effect of funding source on publication status, with adjustment for intervention type and trial size.

RESULTS

Of 818 registered trials found using the keyword "surgery", 395 met the inclusion criteria. Of these, 21% (81/395) were discontinued early, most commonly owing to poor recruitment (44%, 36/81). The remaining 314 (79%) trials proceeded to completion, with a publication rate of 66% (208/314) at a median time of 4.9 (interquartile range 4.0-6.0) years from study completion to publication search. A further 6% (20/314) of studies presented results on ClinicalTrials.gov without a corresponding peer reviewed publication. Industry funding did not affect the rate of discontinuation (adjusted odds ratio 0.91, 95% confidence interval 0.54 to 1.55) but was associated with a lower odds of publication for completed trials (0.43, 0.26 to 0.72). Investigators' email addresses for trials with an uncertain fate were identified for 71.4% (10/14) of discontinued trials and 83% (101/122) of unpublished studies. Only 43% (6/14) and 20% (25/122) replies were received. Email responses for completed trials indicated 11 trials in press, five published studies (four in non-indexed peer reviewed journals), and nine trials remaining unpublished.

CONCLUSIONS

One in five surgical randomised controlled trials are discontinued early, one in three completed trials remain unpublished, and investigators of unpublished studies are frequently not contactable. This represents a waste of research resources and raises ethical concerns regarding hidden clinical data and futile participation by patients with its attendant risks. To promote future efficiency and transparency, changes are proposed to research governance frameworks to overcome these concerns.