Hypothesis A "multifaceted" intervention program aimed at increasing the responsiveness of care givers, the adequacy of care provided, and the efficacy of organisation of care, in presence of abnormal blood loss in the immediate post partum has more impact on the incidence of severe PPH and on the costs of care than the current methods of dissemination of clinical practice guidelines.

Intervention Intervention group. The intervention includes three components: (1) outreach visits with local presentation of evidence-based clinical practice guidelines for management of PPH, and discussion of their applicability in the context of local organisation; (2) during these educational visits, reminders - check list, "PPH emergency case" containing appropriate materials - to be used in case of PPH, will be proposed; (3) finally, cases of severe PPH will be reviewed during peer review sessions organized in each participating unit, to help identifying weaknesses in care provided, and needs for improvement.

Control group. The proposed guidelines for management of PPH will be disseminated through the participating perinatal networks; then each unit will be free to implement them at its own convenience.

Randomisation procedure The trial follows a cluster randomised trial design. Randomisation of maternity units will be stratified by region, status (public versus private) and size (annual number of deliveries). The stratified design will guarantee the comparability of the two arms of the trial at baseline.

Outcome measures The primary outcome is the incidence of severe PPH (number of severe PPH to number of deliveries). A severe PPH is defined as a PPH that was associated with one or more of the following: peripartum haemoglobin drop greater than 4g/dl, blood transfusion, arterial embolisation, surgical procedures such as hysterectomy or arterial ligation, transfer of the mother to intensive care unit, maternal death.

Secondary outcomes include the cost of care and the cost/efficacy ratio, and the incidence of adverse effects of uterotonic drugs.

Background Post partum haemorrhage (PPH) is estimated to occur in 5 to 10% of deliveries. In France this represents about 39 000 to 78 000 cases yearly, among which, 10 to 20% are life threatening events for the mothers. Moreover, PPH remains the leading cause of maternal mortality in France, contributing to a greater proportion of maternal deaths than in other comparable European countries.

According to the National Committee on Maternal Mortality, more than 80% of deaths due to PPH are " avoidable " deaths, involving unrecognised diagnosis, delay in providing care and underestimation of the severity of blood spoliation. These factors suggest weaknesses in the capacity for providing adequate care in presence of PPH, and they seem to be related to the organisation of care inside the maternity units.

On the other hand, although some specific clinical interventions have proved efficient in decreasing postpartum blood loss, thus resulting in recent national and international recommendations for management of PPH, integration of these guidelines into clinical practice still is to be achieved (preliminary unpublished results from a survey conducted in maternity units in France).

Available literature related to various health issues shows that the use of enhanced methods for dissemination of clinical practice guidelines can improve the translation of these guidelines into practice. Therefore, an intervention combining several enhanced dissemination methods may help improving the actual management of PPH cases.

The impact of the proposed intervention will be evaluated through a randomised trial. The target of the intervention is, in a given maternity unit, the entire group of health professionals involved in delivery care. Therefore, the unit of randomisation will be the maternity unit.

Hypothesis A "multifaceted" intervention program aimed at increasing the responsiveness of care givers, the adequacy of care provided, and the efficacy of organisation of care, in presence of abnormal blood loss in the immediate post partum has more impact on the incidence of severe PPH and on the costs of care than the current methods of dissemination of clinical practice guidelines.

Intervention Intervention group. The intervention includes three components: (1) outreach visits with local presentation of evidence-based clinical practice guidelines for management of PPH, and discussion of their applicability in the context of local organisation; (2) during these educational visits, reminders - check list, "PPH emergency case" containing appropriate materials - to be used in case of PPH, will be proposed; (3) finally, cases of severe PPH will be reviewed during peer review sessions organized in each participating unit, to help identifying weaknesses in care provided, and needs for improvement.

Control group. The proposed guidelines for management of PPH will be disseminated through the participating perinatal networks; then each unit will be free to implement them at its own convenience.

Randomisation procedure The trial follows a cluster randomised trial design. Randomisation of maternity units will be stratified by region, status (public versus private) and size (annual number of deliveries). The stratified design will guarantee the comparability of the two arms of the trial at baseline.

Outcome measures The primary outcome is the incidence of severe PPH (number of severe PPH to number of deliveries). A severe PPH is defined as a PPH that was associated with one or more of the following: peripartum haemoglobin drop greater than 4g/dl, blood transfusion, arterial embolisation, surgical procedures such as hysterectomy or arterial ligation, transfer of the mother to intensive care unit, maternal death.

Secondary outcomes include the cost of care and the cost/efficacy ratio, and the incidence of adverse effects of uterotonic drugs.

Number of patients required Estimation of the intracluster correlation coefficient: Since no estimation of this parameter was available in the literature, we based our estimation on results of an unpublished survey providing data on declared incidence of severe PPH in maternity units. Based on these results, the intracluster correlation coefficient is estimated to be 0.006.

Assuming an incidence of 1% for the primary outcome, and in order to detect a decrease to 0.6% (40% relative decrease), with 80% power, and 5% risk, and taking into account the average number of deliveries per unit and the estimated intracluster correlation, the required sample size is 58 945 deliveries by group and 117890 in total.

The 6 participating perinatal networks include 133 maternity units, performing 133 000 deliveries annually.

Eligibility criteria For maternity units All maternity units belonging to one of the six perinatal networks of the study are eligible.

For women All women delivering at a gestational age greater than 22 weeks, in one of the participating maternity unit during the study period, will be eligible.

Trial process Recruitment All maternity units pertaining to one of the 6 perinatal networks will be invited to participate. The trial protocol will be presented to obstetricians, anaesthetists and midwives.

Consent A written consent will be obtained from the head of the obstetrics department and from the administrative director of each unit.

Intervention In each participating perinatal network, a team pairing an obstetrician and a midwife will be in charge of the intervention program.

Start date: March 2005 Expected complete date: December 2007

Data collection Baseline data: The comparability of the two arms of the trial will be evaluated through data collected at the maternity unit level, including characteristics and resources of the unit, and characteristics of the women delivering.

Post-intervention data: Evaluation of the impact of the intervention on clinical practice related to PPH management will be based on individual data.

What ? Detailed information on components of care provided during labour, delivery and immediate post partum, related to prevention and management of PPH, collected in individual cases of PPH.

When collected ? Data collection will begin 3 months after the outreach visits, and will include all incident cases of PPH during one year.

How and by who ? Cases of PPH will be identified prospectively, according to the following definition:1) for a vaginal delivery, PPH is defined as an abnormal bleeding diagnosed within the 24 hours postpartum, which requires uterine revision or manual delivery of placenta, and /or resulting from vaginal or cervical tears, including haemorrhagic episiotomy; for a C section delivery, PPH is defined as an abnormal bleeding diagnosed by the team in charge; or 2) a peripartum haemoglobin drop greater than 2gr/dl.

PPH cases will be identified by the care providers and notified to the trial research assistant who will collect individual information from obstetrical records.

Outcome data Evaluation of the impact of the intervention on the incidence of severe PPH.

What ? Incident cases of severe PPH (see definition above) will be identified as a subgroup of incident cases of PPH.

When collected ? Data collection will begin 3 months after the outreach visits, and will include all incident cases of severe PPH during one year.

How and by who ? See above.

Inclusion Criteria:

For maternity units All maternity units belonging to one of the six perinatal networks of the study are eligible.

For women All women delivering at a gestational age greater than 22 weeks, in one of the participating maternity unit during the study period, will be eligible.

Exclusion Criteria:

-