On the Pulse - 19th June 2009

Metoclopramide is safe to use in pregnancy
Morning sickness in the first trimester of pregnancy is common and often severe enough to make women miserable. Doctors, however, are reluctant to offer drug treatment for fear of teratogenesis. Israel seems to be an exception and there metoclopramide is frequently prescribed for nausea and vomiting in pregnancy, despite being an off-label indication. The NEJM has a retrospective study of over 80,000 infants, which shows that this practice is safe. Rates of congenital malformation, low birth weight, premature delivery and perinatal death were no higher in those who had been exposed to metoclopramide in the first trimester.

Self-testing for Alzheimer’s disease
A paper in the BMJ has attracted plenty of media attention, offering a self-administered test for early Alzheimer’s disease. The TYM test is simple and quick to complete, its sensitivity (93%) and specificity (86%) are a welcome improvement over the mini-mental state examination, and an accompanying Editorial is broadly supportive. The problem, as an analysis by natural numbers shows, is that, on average, out of every 100 people aged 65-69 taking the test, one will receive an early warning, and 11 will incorrectly conclude they have Alzheimer’s disease. Perhaps not wholly suitable for self-administration, after all?

Optimum treatment for people with diabetes and coronary heart disease
Everyone knows that having diabetes increases risk of cardiovascular events but no one knows the best way to mitigate it. Is it better to intervene early with revascularisation, or stick to medical treatment? And, since higher insulin levels are associated with adverse outcomes, might it be better to achieve glycaemic control with insulin sensitising drugs rather than by insulin provision? A large RCT in the NEJM uses a factorial design to test all combinations of these options and finds almost no difference between them for the primary outcomes of death or major cardiovascular events. Adverse events were similar in the different groups too, with the exception of hypoglycaemia, which was commoner in people treated with insulin.

Lipoprotein(a)
Lipoprotein(a) consists of an LDL particle bound to the plasminogen-like glycoprotein, apolipoprotein(a). Several studies have shown that high levels are associated with a modestly increased risk of myocardial infarction but whether this association was causal remained in doubt. A Danish study in JAMA, using the technique of Mendelian randomisation, pretty well settles the matter. It did 3 things: first, it confirmed that lipoprotein(a) levels were associated with risk of myocardial infarction; second, it identified polymorphisms of the KIV-2 gene that influenced lipoprotein(a) levels ; and third, it showed that these polymorphisms were themselves associated with risk of myocardial infarction. The accompanying Editorial explains how Mendelian randomisation can be thought of as a randomised controlled trial of genetic variants.

Corticosteroids in septic shock
Experiments in animal models of sepsis show that corticosteroids prolong survival. But trial evidence in humans is inconsistent, and current international guidelines recommend steroids only in patients who fail to respond to fluid replacement and vasopressors. A systematic review in JAMA puts the results of 20 randomised or quasi-randomised trials together and finds no overall benefit of steroids on mortality. However, there was considerable heterogeneity among the trials, and when the analysis was restricted to those that investigated a prolonged course ( 5 days) of low-dose corticosteroid treatment ( 300 mg of hydrocortisone or equivalent), there was a significant reduction in both 28-day all-cause mortality and in-hospital mortality.

Independent midwives
A study in the BMJ compares outcomes from births assisted by independent midwives vs NHS services, using anonymized record matching to account for socioeconomic and other differences. Many of the outcomes favour the independent midwives: lower rates of premature birth and admission to neonatal intensive care, less perineal trauma, more spontaneous labour and lower doses of analgesia. However (and it’s a big however), perinatal mortality was lower in the NHS group. The authors think that mothers who seek out independent care may be less likely to agree to aggressive treatments or interventions when things go awry. An accompanying Editorial weighs the findings.

Progesterone for twins
Rates of neonatal death remain much higher for multiple births than for singleton pregnancies, and prematurity is the main cause. A trial in the Lancet tests the hypothesis that such deaths can be lowered by treating the mother with antenatal progesterone, which is known to be of benefit in high-risk singleton pregnancies. Unfortunately, the composite outcome of risk of delivery or intrauterine death before 34 weeks of pregnancy remained undiminished. An accompanying Editorial suggests that the fetus may benefit nonetheless, and that longer-term outcomes should be monitored in survivors.

Epidemics of type 1 diabetes?
The inadvisability of extrapolating logarithmic curves to predict impending doom should be well known. But a paper in the Lancet does exactly that: plots data on the incidence of type l diabetes in Europe from 1989 to 2003, fits them to an exponential function, derives an annual rate of increase of 3.9%, and predicts that we’ll be overwhelmed by an epidemic of type I diabetes by 2020. Absurdly, confidence intervals are given for the observed rates, but not for the extrapolations, where they would surely be enormous. An accompanying Editorial warns of a “devastating effect on health-care costs”. Yes, and Stephen Fry will take over the planet in March 2010.