Four communities in Ontario were chosen as study sites, including urban, suburban and small-town practices, with patients from diverse socioeconomic and ethnic backgrounds. Family physicians, obstetricians and midwives were approached at rounds with information about the study and invited to participate. Interested health care providers were sent an introductory letter with a fax-back form, which was followed by a telephone call from 1 of the investigators to determine whether they would participate. Practitioners were eligible if they practised prenatal and intrapartum care or prenatal care with transfer of care for delivery after 28 weeks, provided care for 10 or more prenatal patients per year, and were not currently using any prenatal psychosocial screening tool other than the standard Ontario Antenatal Record.
To obtain a balanced sample, each participating provider was paired to the greatest extent possible with another provider by practice location, type of provider, sex and age. One member of each pair was randomly assigned to the ALPHA or control group by a biostatistician using computer-generated random numbers.
Providers were asked to enroll 5 consecutive pregnant women who were between 12 and 30 weeks' gestation, able to read and write English and give consent. Women were excluded if they were at high obstetric risk as defined by the Ontario Antenatal Record, such as those with pre-existing diabetes, renal disease, severe hypertension or heart disease.
Interested women received an explanatory brochure and consent form from their provider and a phone call from the study nurse to further explain the study and secure consent. Participants completed a questionnaire on demographic and obstetric details along with several psychosocial instruments (not reported in this paper).8,17,18,19,20,21,22
Intervention group providers attended a 1-hour workshop on the ALPHA form given by 1 or more of the investigators. This interactive session included a review of the evidence for the ALPHA form, specific interview questions, role play, management strategies for partner violence and a summary of community resources for psychosocial problems. Once trained, providers completed the ALPHA form with enrolled women at a prenatal visit of the provider's choice between 20 and 32 weeks' gestation. Risk factors were rated as being of concern if they raised concern in the woman, her family or the provider. Women whose providers were in the control group continued to receive usual care.
All of the providers completed a data collection sheet entitled “Psychosocial concerns” on each of the enrolled women within 1 month after the last woman delivered. They were asked whether they had any concerns about the women, with specific reference to the psychosocial risk factors on the ALPHA form. For each patient, providers identified whether each of the 15 risk factors raised concern and, using their clinical judgment, rated the level of concern as “low,” “some” or “high.” Providers were advised that concerns rated as “high” would be those they were more likely to act on. For the study, issues were considered to be of concern if the level of concern was rated as “some” or “high.” Both groups could refer to their antenatal records to fill out the form, and the intervention group could also refer to their antenatal ALPHA forms. Providers in the intervention group were also given a questionnaire about their experience using the ALPHA form.
At 4 months postpartum, the study nurse contacted all women in the trial to again complete a number of psychosocial instruments.17,18,19,20,21,22,23
Women with providers in the ALPHA group were asked to give feedback about the ALPHA form.
At completion of the study, ALPHA group providers received $50 per woman and control group providers $20 per woman for time spent completing the documentation. Women received $25 to help defray their expenses. The study took place between 1998 and 2002, with staggered participation by each site. Ethics approval was obtained from the University of Toronto Research Ethics Board and the McMaster University Research Ethics Board.
The intracluster correlation coefficient (ICC) in primary care settings for process variables is of the order of 0.05– 0.15,24
whereas the ICCs for outcome variables are generally lower than 0.05. We assumed an estimate of the ICC of 0.05 for the outcome variables. From previous work, it was estimated that a significant psychosocial concern would be detected in 5% of women. A 10% increase in detection (from baseline of 5%–15%) was considered clinically significant. A sample of 33 providers and 5 women per provider, which translates into 165 women in each group, was chosen to detect a 10% difference between the 2 groups (type I error = 0.05, power = 0.80) after adjustment for clustering of women by provider.25
We limited the analysis to patients who completed the study. However, we also did a sensitivity analysis by intention to treat to account for the 9 providers in the ALPHA group and the 3 providers in the control group who dropped out. Each missing provider was imputed with 5 patients, each with 0 psychosocial concerns.
All of the available data are reported for each question. χ2 tests were used to compare proportions, and t tests and nonparametric tests were used to compare means of continuous variables between the 2 groups. For the analysis of the primary study question and the response to ALPHA categories, a 2-sided p value less than 0.05 was taken to indicate a statistically significant finding, but for the analysis of the 15 risk factors, significance was set at 0.01. Hierarchical logistic regression was used to control for clustering of women per provider.