Research highlights – 8 March 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.

Malaria prophylaxis with co-trimoxazole in children exposed to HIV
More than two million infants are born to HIV infected African mothers each year, and the World Health Organization recommends giving these infants co-trimoxazole prophylaxis until breast feeding (and exposure to HIV) ends and HIV infection is excluded. The prophylaxis protects HIV infected children from common opportunistic infections, including malaria. More than 90% of these infants remain uninfected throughout breast feeding, but this cannot easily be confirmed because maternal HIV antibodies persist in infant blood until 18 months of age. Thus, many of these children continue taking co-trimoxazole prophylaxis until antibody testing for HIV infection becomes possible.

This led Taylor Sandison and colleagues to conduct a randomised clinical trial of co-trimoxazole prophylaxis against malaria in such HIV exposed, uninfected children in rural Uganda, where both malaria transmission intensity and resistance to antimalarial antifolate drugs are high. (They used the polymerase chain reaction to confirm absence of HIV DNA in the children, but that’s not readily available in rural Africa.) Continuing daily prophylaxis after end of breast feeding to 2 years of age decreased the incidence of malaria by 39% compared with stopping prophylaxis at the end of breast feeding. One big question, discussed in the full article, is why the protective efficacy of co-trimoxazole against malaria in this trial was so much lower than in other studies (with reports of 80-99% efficacy).

Steroids and bronchodilators for acute bronchiolitis in babies and toddlers
Doctors and patients often need to consider a wide range of treatments with different outcomes. Searching for robust evidence of which one is best will often yield only a single meta-analysis of treatment A versus treatment B or placebo, which may not do much to guide decision making in the real world. Increasingly, therefore, researchers are using complex methods to combine several systematic reviews on the same condition and to analyse outcome data for all relevant treatment comparisons. That’s what Lisa Hartling and colleagues have done in their systematic review and Bayesian network meta-analysis of bronchodilators and steroids, alone or combined, in the acute management of bronchiolitis in patients aged 24 months or less.  Previous studies have shown that this condition is treated in many different ways round the world, but here nebulised adrenaline given in the emergency department proved the best way to keep patients out of hospital on day 1, while combined nebulised adrenaline and oral dexamethasone most effectively reduced admission rates over the next week.

A couple of weeks ago we published a letter from Valeria Fadda and colleagues explaining how a single figure can show the comparative effectiveness of various treatments in a network meta-analysis (doi:10.1136/bmj.d1555). This paper on bronchiolitis comes with just such a figure.

Statin treatment after pneumonia
Two years ago a systematic review of observational studies suggested that statins might protect against sepsis and mortality. At around the same time Liam Smeeth and colleagues reported a cohort study using The Health Improvement Network (THIN) database of UK computerised primary care records, which suggested that statins might protect against pneumonia (doi:10.1111/j.1365-2125.2008.03308.x).  Building on that work, Smeeth and colleagues have now looked at mortality in the six months after diagnosis of pneumonia among patients aged over 40 who were either  users of statins (109/847; 13% died) or matched non-users (578/2927; 20% died).

The authors used modelling and a propensity score based method to control for differences between people prescribed and not prescribed statins. Given the inability of an observational study to completely rule out confounding and other biases, however, the authors say it is now time for a randomised controlled trial to test the association. They estimate, cautiously, that their result would equate to a trial derived number needed to treat (to prevent death within six months of an episode of pneumonia) of 15. In the meantime, this study—in which the main analysis covered nearly 5000 episodes of pneumonia with around 1000 in patients prescribed statins—provides much food for thought.

Effectiveness of antenatal corticosteroids in reducing respiratory disorders in late preterm infants
A randomised controlled trial by Ana Maria Feitosa Porto and colleagues showed that antenatal treatment with corticosteroids at 34-36 weeks of pregnancy did not reduce the incidence of respiratory disorders in newborn infants (doi:10.1136/bmj.d1696)

Reporting of eligibility criteria of randomised trials
Anette Blümle and colleagues found that the eligibility criteria reported in published articles differed from those given in the protocol for all 52 of the trials they looked at (doi:10.1136/bmj.d1828)

Understanding provision of chemotherapy to patients with end stage cancer
The findings of Hilde Buiting and colleagues’ qualitative study suggest that doctors may offer further chemotherapy to patients with end stage cancer out of the fear that “giving up hope” will harm their wellbeing (doi:10.1136/bmj.d1933).