Research highlights – 22 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.

Oral contraceptives and thrombosis

Two case-control studies in this week’s BMJ address the risk of non-fatal venous thromboembolism in women taking oral contraceptives that contain drospirenone versus those containing levonorgestrel. In the first, Susan Jick and colleagues looked at data from a US company and in the second they used the UK General Practice Research database. Both analyses found an increased risk associated with drospiredone compared with levonorgestrel—about double in the US study and triple in the UK study. Importantly, though, overall rates of events were still low.

In the online versions of the papers the authors explain how they add methodologically to two previous BMJ papers on oral contraceptives and thrombosis, published in 2009—both of which showed a small increase in risk for drospiredone compared with levonorgestrel. But we didn’t do a good enough job with one of those papers. In a recent editorial, Gerd Gigerenzer pointed out that while one made the absolute risks clear in the abstract the other reported that “oral contraceptives increased the risk of venous thrombosis fivefold” without giving the crude numbers.

Given alone, relative risks such as “fivefold” provide an incomplete and misleading message, because they don’t inform about the baseline risk. This point is well illustrated by an older example cited by Gigerenzer: in 1995, the UK Committee on Safety of Medicines issued a warning that third generation oral contraceptive pills increased the risk of thrombosis twofold, provoking great anxiety. Many women stopped taking the pill, leading to unwanted pregnancies and abortions, along with extra costs for the NHS. But the actual increase in risk was small: from one thrombosis per 7000 women taking second generation pills to two per 7000 for third generation pills.

Risk exaggeration can make for a more dramatic story, but journals have a responsibility to provide all the information and promote good risk reporting. Groups like Sense About Science are also backing a drive for the general public, as well as journalists, to become more stats-savvy. Worth bearing in mind when reading or submitting papers.

Cardiovascular risk from calcium supplements with or without vitamin D
Last year, Mark Bolland and colleagues reported on their meta-analysis of cardiovascular events in randomised controlled trials of calcium supplements (without co-administered vitamin D), which found that the supplements significantly increased the risk of myocardial infarction (BMJ 2010;341:c3691). Subsequently, the Women’s Health Initiative reported no adverse effect of co-administered calcium and vitamin D supplements on any cardiovascular end point in their large randomised controlled trial.  However, their results were complicated by 54% of their participants taking personal (non-protocol) calcium supplements at randomisation and 47% taking personal vitamin D supplements, effectively rendering the trial a comparison of higher dose and lower dose calcium and vitamin D for most of the participants.

Bolland and colleagues have now reanalysed the data from the Women’s Health Initiative study, isolating the data from women not using personal calcium supplements, and updated their meta-analysis. They conclude that calcium supplements, with or without vitamin D, modestly increased the risk of cardiovascular events, especially myocardial infarction, and they suggest that the use of calcium supplements in osteoporosis management in older people should be reassessed.

Relative risks from medical abortion in adolescents and adults
Medical termination of pregnancy has become widely used in the past decade and is particularly appropriate for the unplanned pregnancies that are all too common among teenage girls. However, few studies have specifically assessed the risks of medical abortion in adolescents. So Maarit Niinimäki and colleagues compared outcomes in 3024 adolescent women (<18 years old) and 24,006 adults (≥18 years) who underwent medical abortion between 2000 and 2006. The risk of adverse events (haemorrhage, incomplete abortion, infection) was similar or even lower in the adolescent cohort, indicating medical abortion to be at least as safe in adolescents as in adults.

The study is discussed further in the linked editorial by David Grimes and Elizabeth Raymond, including the “alarmingly high” rate of adverse events seen in both adolescent and adult groups, apparently some 20-100 times higher than in recent large studies with more specific outcome definitions. However, they say this should be interpreted with caution because some of the reported outcomes, particularly haemorrhage, were not strictly defined and could include many attendances by the worried well rather than validated complications.