Research highlights – 1 April 2011

Research questions“Research highlights” is a weekly round-up of research papers appearing in the print BMJ. We start off with this week’s research questions, before providing more detail on some individual research papers and accompanying articles.

Cancer recurrence after surgery with epidural block
Apart from the risk of releasing cancer cells into the circulation, surgical removal of cancer has several immunosuppressive effects that could increase the risk of recurrence. Surgery itself depresses cell mediated immunity; general anaesthesia impairs many immune functions; and opioids used for postoperative pain relief inhibit both cellular and humoral immune function. Regional anaesthesia might avoid such problems by reducing the need for opioids and by suppressing the response to surgical stress. Animal studies support the theory, but the few, small, observational studies in humans have conflicting results.

Paul Myles and colleagues now report on their long term follow up (up to 15 years) of 503 patients who underwent potentially curative abdominal surgery for cancer and were randomised to receiving general anaesthesia with or without epidural block for at least three postoperative days. They found no effect on cancer-free survival. The authors, and the linked editorial by Tsui and Green, acknowledge the study limitations that could have hidden a real effect—both groups of patients received immunosuppressive general anaesthesia and some opiates, and the heterogeneous nature of the cancers included in the study would have disguised any cancer specific effect. However, the fact that research on this subject reaches back to the 1970s with no conclusive result is not encouraging.

Impact of doctors’ shorter working hours on patients’ outcomes and postgraduate training
Junior doctors’ working hours have been getting shorter over the past 20 years in the United States and Europe. Has this adversely affected patients’ outcomes or objectively measured outcomes of doctors’ training? Not according to the published evidence, as systematically reviewed by S R Moonesinghe and colleagues. They analysed the results of 72 studies from the US and UK and found no adverse effects associated with reducing working hours to 80 or fewer a week. The evidence, particularly from the UK, was of relatively low quality, however. So the authors call for bigger and better studies, particularly on the impact of European legislation limiting working hours to 56 or 48 a week. And, given that one of the fundamental principles behind these reforms was to improve patient safety, editorialist Leora I Horwitz from Yale University Medical School asks why they have not benefited patients.

Industry funding and reporting of subgroup analyses in trials
The CONSORT 2010 statement on reporting parallel group randomised controlled trials asks authors to report “Results of any other analyses performed, including subgroup analyses and adjusted analyses, distinguishing prespecified from exploratory” and warns that “Multiple analyses of the same data create a risk for false positive findings. Authors should resist the temptation to perform many subgroup analyses. Analyses that were prespecified in the trial protocol are much more reliable than those suggested by the data, and therefore authors should report which analyses were prespecified … Selective reporting of subgroup analyses could lead to bias” (

Several studies have shown associations between industry funding or sponsorship and biases in trial design, conduct, and reporting, but does the same apply to the use of subgroup analysis? Indeed it does, say Xin Sun, from the Chinese Evidence-Based Medicine Center in Chengdu, and colleagues from North America and Europe. They reviewed 1140 papers reporting randomised controlled trials in 118 core clinical journals in 2007; half from high impact journals and half from lower impact journals. Randomised controlled trials published in high impact journals, with larger sample size, studying non-surgical topics, and with industry funding were associated with more frequent reporting of subgroup analyses if the primary outcome was not statistically significant. It’s well worth reading the authors’ discussion of their findings in the full paper on, where they also note that “Industry funded trials, regardless of the statistical significance of primary outcomes, less often prespecify subgroup hypotheses and less often use the interaction test for analyses of subgroup effects compared with trials that are not funded by industry.” This study was funded by the National Natural Science Foundation of China: let’s hope it reaches researchers in China, given that so many industry trials are now conducted there.

Group therapy for self harming adolescents
J M Green and colleagues found that a targeted group therapy programme was not effective for young people who repeatedly self harmed, although outcomes for the cohort as a whole were better than expected (doi:10.1136/bmj.d682).

Statins for primary prevention of vascular disease
J P Greving and colleagues say that in daily practice, statin treatment may not be cost effective for primary prevention of vascular disease in populations at low risk, despite low costs of generic pills. They suggest that improved adherence to statins would enhance cost effectiveness (doi:10.1136/bmj.d682).