Background and Prior Research Rationale

Illustrative examples - Perinatal care trial

‘The evaluation of the medical care in general and of the area of perinatology in particular, shows alarming results. It has been seen that only 15% of the health related practices are based on valid scientific evidence. This situation is even more serious in Latin American countries, where a significant proportion of the care administered to women during the prenatal period, delivery, and the postpartum are scientifically verified as ineffective or harmful.

Several trials have been performed in North America and Europe to evaluate strategies to change behaviors of birth attendants. Lomas et al. (1991) conducted a trial in Canada with 76 physicians to test three interventions to increase the number of vaginal births after C-section. The interventions compared were (1) distribution of educational materials, (2) local opinion leaders and distribution of educational materials, and (3) audit and feedback and distribution of educational materials. The use of opinion leaders was significantly more effective than audit and feedback and than educational materials alone in increasing the number of women offered a trial of labor and increasing the number of vaginal births.

Another trial performed in Canada (Hodnett et al., 1996) compared a strategy based on the influence of nurse opinion leaders to no intervention. It was hypothesized that the strategy would result in lower rates of epidural analgesia through increasing the amount of support nurses provided to their patients. Other outcomes included rates of narcotic analgesia, episiotomy, and operative delivery. The results showed that the strategy was unsuccessful in improving intrapartum nursing care. The individual marketing approach used in the intervention de-emphasized information about research and did not include institutional changes. The authors suggested that this might partly explain the lack of success of the intervention.

One trial comparing the introduction of a database of systematic reviews, the Cochrane Pregnancy & Childbirth Database, with a control group has been conducted recently (Wyatt et al., 1998). This trial has compared the introduction of the Cochrane database through a single educational visit (outreach visit) conducted by an expert obstetrician with a control group that did not receive any intervention. The educational visit involved meeting with the lead obstetrician and nurse of the labor ward and demonstrating how to select and interpret Cochrane reviews. Twenty-five obstetric units within a small geographic area in England were included in this trial. There was a significant increase in the use of only one of the four recommended practices (vacuum extraction). However, this trial has been conducted in a region where awareness of evidence-based medicine and Cochrane reviews increased significantly during the trial period in both the study and control hospitals, reducing the magnitude of any possible impact of the intervention.

Another cluster randomized trial performed in the United States evaluated the effect of an “active dissemination” strategy compared to a “usual dissemination” strategy, on the use of corticosteroids for fetal lung maturation prior to a preterm birth in 27 hospitals (Leviton et al., 1999). The active strategy consisted of the use of a local opinion leader, grand round lectures given by respected national experts, chart reminders, group interactive discussions, and audit and feed back. In usual dissemination hospitals, the use of corticosteroids increased from 33% to 58% or by 75% over baseline. The active dissemination hospitals increased from 33% to 68% or by 108% over baseline.

Very few trials of a similar nature have been performed in developing countries, and the results of the few that have been performed are inconclusive. Two trials utilizing educational outreach visits were conducted in Indonesia. Santoso et al. (1996) compared two interventions (outreach visits or a formal seminar) to a nonintervention control to improve drug use in the management of acute diarrhea in children. They reported that outreach visits resulted in a significant 24% relative reduction in antimicrobial use compared with the control group. There was a significant 40% relative reduction in the use of antidiarrheals compared to the control group. The authors also reported that the seminar resulted in significantly greater changes from the baseline period than did the outreach group. The use of oral rehydration agents was not significantly improved after either intervention (9% reduction). Ross-Degnan and colleagues (Ross-Degnan et al., 1996) reported that outreach visits to pharmacists/owners, coupled with a small group session with counter attendants, significantly increased sales of oral hydration salts by 40% and reduced antidiarrheal sales by 35% for the treatment of diarrhea.

In sum, multiple strategies to change medical behaviors have been used, with various degrees of success. Few trials have been done in obstetrical settings or in developing countries.

To our knowledge, no randomized controlled trial has been performed to evaluate the effectiveness of an intervention to implement evidence-based birth practices in Latin American countries.’ (CLAP Trial - go to protocol)

Illustrative examples - Magpie trial

Pre-eclampsia, a multisystem disorder of pregnancy usually associated with raised blood pressure and proteinuria, complicates 2-8% of pregnancies.  Although outcome is often good pre-eclampsia is a major cause of morbidity and mortality for the woman and her child.  It accounts for an estimated one fifth of antenatal admissions, two thirds of referrals to day care assessment units, and a quarter of obstetric admissions to intensive care units.  Eclampsia is rare, affecting around 1 in 2000 deliveries in the UK for example, but is associated with a considerably higher morbidity and mortality. 

Anticonvulsants may be administered to women with pre-eclampsia in the belief that these reduce the risk of a seizure, and so improve outcome.  There is considerable variation in clinical practice but if an anticonvulsant is used, magnesium sulphate has recently emerged as the most rational choice.  As for all prophylaxis, however, there is a particular responsibility to ensure that it does more good than harm.  If magnesium sulphate does reduce the risk of eclampsia, the potential for benefit is relatively low (unless there are other beneficial effects), as most women would not have convulsed anyway.  This is important as very large numbers of women and babies could be exposed to prophylactic anticonvulsants.  In the US, for example, 5% of pregnant women receive magnesium sulphate before delivery.  If the same usage was applied in the UK, 35,000 women would be treated each year.  However, the incidence of eclampsia is only about 400 women/year without the use of magnesium sulphate. So, even if magnesium sulphate halves this risk, 200 women might be protected from fitting but 34,800 would have been exposed to the possible adverse effects.

The first question to be answered is whether magnesium sulphate reduces the risk of eclampsia.  Even if it does do so, before it can safely be introduced into clinical practice more information is required about: (a) the size of the risk reduction; (b) effects on other important outcomes for the woman and child; (c) disease severity at which benefits outweigh the risks and (d) the cost consequences of different forms of care.  Magnesium sulphate is cheap and relatively, easy to administer, but there could be a considerable cost for women, children and the health services if it is used routinely for pre-eclampsia without proper evaluation.

Systematic review of anticonvulsants for women with pre-eclampsia  A recent systematic review of randomised trials demonstrates that, currently, there is insufficient evidence to either support or refute the use of prophylactic anticonvulsants.  Twelve trials have evaluated anticonvulsants for women with pre-eclampsia.  Two small studies were excluded from the review because either no clinical outcomes were reported, or they were not reported separately for pre-eclampsia and eclampsia.  Two trials have compared magnesium sulphate with no anticonvulsant (228 women, South Africa; 64 women, Taiwan) and two have compared magnesium sulphate with placebo (822 women, South Africa; 135 women, USA).  One quasi-randomised study (59 women, Tanzania compared oral diazepam with no anticonvulsant.  The remaining trials compared magnesium sulphate with diazepam (38 women, Mexico; 28 women, Malaysia) or with phenytoin (2138 women, USA; 115 women, USA; 54 women, USA).

Results  Overall, the methodological quality of these trials was average to poor.  In most, concealment of the allocation at trial entry was inadequate, and two studies excluded from their analyses >10% of women randomised.

In the comparison of magnesium sulphate with no anticonvulsant/placebo, three women allocated magnesium sulphate had a fit, compared to 13 amongst those allocated no anticonvulsant/placebo (Relative Risk 0.33; 95% CI 0.11-1.02).  There was also a non-significant trend towards a small increase in the risk of Caesarean section for women allocated magnesium sulphate (RR 1.04; CI 0.92-1.17).  There is little information about possible effects on other important outcomes.  In the comparison of magnesium sulphate with phenytoin, women allocated magnesium were less likely to develop eclampsia (RR 0.09; CI 0.01-0.72), but more likely to have a Caesarean section (RR 1.21; CI 1.05-1.41).  There were no statistically significant differences in stillbirths (RR 0.62; CI 0.27-1.41) or neonatal deaths (RR 0.84; CI 0.41-1.74) amongst the babies allocated magnesium sulphate rather than phenytoin in utero.  There is little information about other relevant outcomes.  No woman in the comparison of magnesium sulphate with diazepam developed eclampsia.  No trials have reported follow-up of the children beyond the perinatal period, an economic evaluation, or an assessment of the costs to the health services.

Discussion  To date, 1200 women with pre-eclampsia have been entered into trials comparing an anticonvulsant with none.  The results of these trials, when taken together, are promising in terms of a reduction in the risk of eclampsia associated with the use of anticonvulsants.  These data should be interpreted with caution, however, as the number of events was small and the largest study had a large proportion of post randomisation exclusions.  Also, apart from the suggestion of a small increase in the risk of Caesarean section associated with the use of magnesium sulphate, there is little information about possible effects on other important outcomes, including toxicity and side effects.  Nearly 2400 women have been randomised into trials comparing different anticonvulsants, most of whom were in one study comparing magnesium sulphate with phenytoin.  Women allocated magnesium sulphate were less likely to fit than those allocated phenytoin but this study provides no insight into whether giving magnesium sulphate is preferable to withholding it.  Also, the number of events was small (0 versus 10) and, apart from an increase in the risk of Caesarean section, there are few data on other measures of maternal morbidity.  Finally, 1% of the women allocated phenytoin developed eclampsia.  This is an unusually high incidence for the reported severity of disease (only 18% had ≥2+ proteinuria and 4% had been given an antihypertensive) and may, at least in part, be due to chance.

In conclusion this review, combined with evidence that magnesium sulphate is the drug of choice for women with eclampsia, supports magnesium sulphate as the best choice of anticonvulsant to evaluate for women with pre-eclampsia.  There is a suggestion that magnesium sulphate may be associated with an increase in the risk of Caesarean section, which may be a tocolytic effect.  If true, this tocolytic action might have other consequences such as an increase in length of labour, postpartum haemorrhage and retained placenta. ( Magpie trial - go to protocol).

Illustrative examples - ISAT trial

Aneurysmal subarachnoid haemorrhage (SAH) is a significant cause of death and continuing disability in relatively young patients with an annual incidence of between 6 and 12 per 100,000 population in most western countries.  The natural history of the disease is such that over 30% of patients will die within 24 hours of the bleed and a further 25- 30% will succumb in the next four weeks without some form of surgical intervention.

The publication of the International Co-operative Study on the Timing of Aneurysm Surgery provided the most extensive data to date on the results of modern management in a large number of patients treated in experienced neurosurgical centres.  Over 3,000 patients were entered in a prospective observational study from 1980 to 1983.  The results showed that the overall mortality rates in the surgical patients were between 20% and 28% depending on the timing of surgery and the grade of the patient.  Patients in grade 3/4 (stuporose or comatose) had mortality rates between 39% and 79%.  Overall good outcomes occurred in approximately 60% of patients, leaving approximately 20% of patients with residual morbidity.  Of the patients planned for surgery between 11 and 14 days, 14% rebled pre-operatively.  Many of these rebleeds were either fatal or resulted in a significantly poorer outcome.  More recent surgical series of selected patients from experienced centres have suggested lower rates of serious morbidity and mortality following surgery. e.g. 15% for grades 1 and 2 in posterior circulation aneurysms operated acutely, but 19% overall for all grades.

A recent multicentre prospective randomised study examining the prevention of vasospasm with cisternal rTPA, showed a 3 month mortality of 19% and a good outcome in only 50% of patients with severe SAH operated within 48 hours of the haemorrhage (some of these patients were Grade 5).  Three patients rebled between 14 days and 3 months after surgery. This emphasises the still serious outcome in the condition even with modern surgical management techniques in large centers’ (ISAT Trial - go to protocol)