On The Pulse - 24th May 2013

Prescribing antidepressants
General practitioners must feel that they can’t get it right where depression is concerned. Either they’re criticised for failing to recognise the diagnosis and leaving patients untreated, or they’re blamed for dishing out too many SSRIs. A pair of Head to Head articles in the BMJ ride familiar hobby-horses. One reckons that antidepressants are no more likely to be prescribed inappropriately than antibiotics or proton pump inhibitors, and that by demonising them we create a barrier to effective care. The other scorns the idea that unhappiness should be treated with pills, especially as antidepressants are often ineffective and have serious adverse effects.

Treating refractory asthma
A trial in the NEJM recruits 104 adults with elevated eosinophil levels and asthma that was poorly controlled by inhaled glucocorticoids and long-acting beta-agonists (LABAs), and randomises them to subcutaneous placebo or dupilumab, an antibody that inhibits interleukin-4 and interleukin-13 signalling. LABAs were then discontinued at 4 weeks, and steroids tapered over weeks 6–9. Over 12 weeks, asthma exacerbations occurred in 3 dupilumab patients (6%) vs 23 (44%) with placebo, and most measures of lung function and asthma control were also better with dupilumab. However, an accompanying Editorial warns that patients were highly selected, and the trial design hardly reflects clinical practice.

Trial size
Small trials tend to report greater treatment benefits than large trials, but how big does a trial have to be to give a reliable estimate? An analysis of 735 trials in the BMJ finds no dividing line: no matter how big a study is, a larger one would probably show a smaller treatment effect. On average, compared with trials of >1000 patients, treatment effects were 10% larger in trials with 500–999 patients, and nearly 50% larger in trials with <50 patients. Publication bias is only part of the explanation; larger trials are often better designed and have wider eligibility criteria.

Peripherally inserted central catheters
Peripherally inserted central catheters are easily introduced, even by less experienced staff, and are often perceived as less likely to induce complications, despite their longer track. However, in a meta-analysis in the Lancet of data from 11 comparison studies, they were associated with over twice the risk of deep-vein thrombosis compared to central venous catheters introduced through the internal jugular vein (OR 2.55, 95%CI 1·54–4·23). Data from non-comparison studies suggested greatest risk in critically ill patients and those with cancer, and an accompanying Comment also advises against them in patients with hypercoagulability or a personal or family history of venous thromboembolism.

Macular degeneration
A decade ago, a large trial in the US reported that a dietary supplement of zinc, beta-carotene and antioxidant vitamins slowed the progression of age-related macular degeneration. Multiple subgroup analyses and inadequate statistical power made the results less than compelling but people with the condition have generally been advised to eat a diet rich in these micronutrients anyway. A recent trial in JAMA finds no extra benefit from adding lutein, zeaxanthin and n-3 long-chain polyunsaturated fatty acids to the original cocktail, but the incidence of lung cancer was unexpectedly high in the subgroup of former cigarette smokers who were taking beta-carotene, suggesting that dietary advice isn’t always innocuous.

What sort of ICD?
The trials that established the value of implantable cardioverter defibrillators in the primary prevention of fatal arrhythmias almost all used single-chamber devices. In practice, however, many patients are fitted with dual-chamber devices, perhaps because they offer theoretical benefits. However, a large, retrospective, registry-based study in JAMA finds that dual-chamber devices carry a higher risk of complications without reducing rates of hospitalisation or death. Except for patients with indications for pacing, single chamber devices are probably superior.

Combining osteoporosis drugs
Anticatabolic osteoporosis drugs inhibit bone turnover while anabolic agents stimulate bone formation. Combining the two types seems sensible, but results to date have disappointed. However, in a trial reported in the Lancet randomising 100 postmenopausal women with osteoporosis to either teriparatide, denosumab or both, the combination substantially outperformed either monotherapy in increasing lumbar spine, femoral neck or total hip bone mineral density at 12 months. An accompanying Comment is impressed, but points out that teriparatide is currently only licensed for a maximum 24 months of treatment.

Evaluating healthcare processes
John Ioannidis is perhaps best known for using statistics to show that most published research findings are probably false. His thought-provoking Viewpoint article in JAMA Internal Medicine questions why we have been so slow to apply the standards of RCTs to healthcare system processes. He thinks we too easily believe the predictions of models, given the potential for unintended effects, overestimate the problems of designing suitable trials, and are insufficiently sceptical of before-and-after studies. As he says, we cannot afford merely to guess whether system interventions are effective. We need to design trials to find out for sure.