Search

Abstract Background Evaluating complex interventions in health services faces various difficulties, such as making practice changes and costs. Ways to increase research capacity and decrease costs include making research an integral part of health services and using routine data to judge outcomes. The purpose of this article is to report the feasibility of a pilot trial relying solely on routinely collected register data and being based on ordinary health services. Methods The example intervention was education to public health nurses (PHN) (childbirth classes) to reduce caesarean section rates via pre-delivery considerations of pregnant women. 20 maternity health centers (MHC) were paired and of each 10 pairs, one MHC was randomly allocated to an intervention group and the other to a control; 8 pairs with successful intervention were used in the analyses (1601 mothers). The women visiting to the study maternity centers were identified from the Customer Register of Helsinki City. A list of the study women was made using the mother's personal identification number, visit date, the maternity center code, birth date and gestation length. The mode of delivery and health outcomes were retrieved from the Finnish Medical Birth Register (MBR). Process data of the intervention are based on observations, written feedback and questionnaires from PHNs, and project correspondence. Results It took almost two years to establish how to obtain permissions and to actually obtain it for the trial. Obtaining permissions for the customer and outcome data and register linkages was unproblematic and the cluster randomization provided comparable groups. The intervention did not succeed well. Had the main aim of the trial been to cause a change in PHNs behavior, we would have very likely intensified the intervention during the trial. Conclusion Our experiences encourage the use of trials that obtain their outcomes from registers. Changing the behavior of ordinary health service providers is a challenging intervention. Trial registration number not registered (see Results)