“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.

• Do variants of the cytochrome P450 genotype affect the clinical efficacy of clopidogrel in patients with coronary artery disease?
• Is an additional oxytocin infusion beneficial in elective caesarean section?
• Is antidepressant treatment associated with adverse outcomes in older people?
• What are the health risks and benefits of travel using a bicycle sharing scheme?
• Are public health programmes that prevent cardiovascular disease in entire populations cost effective?

Genetic sensitivity to clopidogrel
CYP2C19 might read like a car number plate, but in fact it denotes a gene related to the cytochrome P450 gene, and both of them code for enzymes that metabolise drugs. Clopidogrel is a good drug, but it doesn’t work in 20% of patients. Previous studies have suggested that CYP2C19 might be implicated, leading to intense debate and a warning from the US Food and Drug Adminstration.

Tim Bauer and colleagues have now scrutinised the association between variants of the CYP2C19 gene and the clinical efficacy of clopidogrel.  Their systematic review and meta-analysis  found, however, no robust evidence to recommend individualised clopidogrel treatment driven by CYP2C19 genotype.

For those who want to consider the bigger picture, Michael V Holmes and colleagues’ linked editorial discusses the wider challenges of pharmacogenetic research.

Bicycle sharing schemes
Last year, the BMJ published a systematic review that looked at whether community interventions to promote cycling were successful (BMJ 2010;341:c5293). The authors concluded that such schemes had the potential to get people on their bikes—but what are the actual benefits to health?

This week, a modelling study by David Rojas-Rueda and colleagues attempts to answer this question in the context of a bike sharing scheme—“Bicing”—in Barcelona, Spain. They estimated that as a result of the physical activity involved in using these bikes, 12.46 deaths would be avoided per year—a large benefit, compared with the risks from inhalation of air pollutants and road traffic incidents—and that the scheme also reduced carbon dioxide emissions. The reliability of the conclusions is limited by the assumptions made in the model, but in sensitivity analyses the authors found a net benefit for Bicing users in all the scenarios they tested.

In a rapid response to the article online, Naveen Dutt of the Government Medical College and Hospital, Chandigarh, India, suggests that non-fatal cycle accidents and other ailments related to bicycle use are important to consider, in addition to the three primary outcome measures used in the study (all cause mortality owing to physical activity, air pollution, and road traffic accident mortality), for an unbiased assessment of the benefits of cycling.

Use of antidepressants in older people
How do you weigh up the risk of antidepressants in older people? Trials are often not designed to address the harms of treatment in older patients with complex conditions. Yet clinicians need guidance to deliver quality care to such patients. So other avenues of evidence must be explored.
Carol Coupland and colleagues say that prescriptions for antidepressants have increased by 35% over five years. A systematic review showed that SSRIs and tricyclic antidepressants have similar benefits in older patients, but more patients discontinue tricyclics. These authors asked: in the over 65s, what is the association between class of antidepressant prescription for a new episode of depression and a variety of expected adverse events? They conducted a large primary care database study drawing on WHO International Classification of Disease codes and information from death certificates to investigate outcomes.

We hope that their results are useful to working doctors. However, as the authors caution, interpreting their findings is complicated. Observational studies are prone to confounding. Choice of antidepressant is not random. Patient preference, comorbidities, and doctors’ judgment mean that there are likely to be systematic differences between patients who are, and are not, prescribed a particular class of drug.

So what should clinicians do? In his commentary, Professor Ian Hickie gives a few observations and pointers. Are we prescribing too much medication? Ninety per cent of patients received a prescription. Are non-pharmacological approaches being used appropriately? He was struck by the observation that most adverse events occurred in the first month after starting or stopping an antidepressant, and he calls for weekly monitoring of patients during these periods.

Self correction of refractive error among young people in rural China
Mingzhi Zhang and colleagues found that over 96% of young people in rural China with poor vision in at least one eye could improve their vision to ≥6/7.5 in the better seeing eye by self refraction with cheap adjustable spectacles.

How to obtain the confidence interval from a P value
Some published articles report P values, but do not give corresponding confidence intervals—or vice versa. A pair of Statistics Notes articles by Doug Altman and Martin Bland explain how to work out either of these missing statistics based on the one available (doi:10.1136/bmj.d2090, doi:10.1136/bmj.d2304)