The imbalance of births in intervention and comparison communities in PRISM
was explained by fewer births in most rural LGAs and rapid population growth in a few metropolitan intervention areas (Victorian Perinatal Data Collection Unit, unpublished data.) The adjusted response fraction was slightly lower in PRISM
than in our earlier postnatal population surveys [1
], possibly because we could not afford to send a second copy of the questionnaire, but the differences in social characteristics between all eligible women and survey participants were very similar to prior surveys and the prevalence of probable depression was the same as in earlier surveys [1
]. As individual consent for participation had not been sought the adjusted response fraction does not demonstrate serious loss to follow-up but rather a relatively high response to receiving a mailed questionnaire 'out of the blue'.
Although the power calculation showed that a particular sample size would be required to identify a statistically and clinically important difference in the primary outcomes between the intervention and comparison groups the finding of no effect of the intervention is strongly based in the similarity of the proportions responding to the outcome questionnaire in the two arms of the trial, and the almost identical primary and secondary outcomes. Thus it is clear that the interventions in this trial did not have an impact on women's mental and physical health at six months after childbirth.
The other universal postnatal intervention trials, those recruiting women across the whole postnatal population, were all designed at the same time, with the exception of Gunn's trial which was a little earlier [42
]. All used the same mental health outcome measure (see Table ) and all but one also measured overall health status (mental and physical) with the Short Form 36. The interventions in the six trials were very diverse, although PRISM
and the trial of MacArthur and colleagues had some components in common. The similar timing of the six trials meant that they were not influenced by the others' findings. The lack of effectiveness of all the interventions implemented in these trials, except that of MacArthur and colleagues, is in contrast to the marked effectiveness of a wide range of postnatal counselling interventions, provided by a variety of practitioners, to women who had been diagnosed as being depressed or probably depressed. The pooled estimate of effect for those interventions is a large reduction in depression: with a relative risk of 0.52 (95% CI 0.40, 0.65) and no significant heterogeneity across the trials [50
Mental health outcomes in universal postnatal intervention trials
Our hypothesis from the beginning was that the inclusion of physical health as well as mental health and the community-based interventions would make a real contribution beyond the trials focused on individual women. That hypothesis was subsequently borne out by the outcomes of the trial of MacArthur and colleagues which was
effective in reducing depression [47
]. Distinguishing features of that trial were its use of existing staff and services to provide redesigned community postnatal care, the integration of their community midwives into primary services and their focus on women's individual physical and psychological health needs. Although there was substantial common ground between MacArthur's trial and PRISM
, including the finding of no effect on physical health in either trial, there were some differences which may have been important. The lack of integration of MCHNs with other primary care services (general practitioners) in Australia is one and the negative impact of a fee for service system on ready access to a GP in Australia is another. However, the success of MacArthur's trial raises the possibility that PRISM
could have been more effective, and we consider below a number of possible explanations for why it was not.
The impact of education and training on primary caregivers in PRISM
, assessed in terms of women's ratings of their care, was much less than we had hoped for. There was a real but small impact on GPs taking part in the education program [28
] but these were a small proportion of all GPs in participating communities, and academic detailing was limited. We saw the role of MCHNs in PRISM
as pivotal but recognised that the education and training in PRISM
involved a role shift from a focus on action around the health and well-being of babies, child health surveillance, immunisation and child protection, to a much more open-ended role involving 'active listening' to mothers, enhanced communication skills and much less certainty about what should be done [51
The CDOs had a five-day residential training program at the start of employment, eight all day meetings as a group with the research team, and three all day meetings with a range of participating community representatives in Melbourne, as well as having frequent email and telephone contact with each other and with the PRISM co-ordinators [22
]. However, their employment was for only two years which may not have been long enough, especially given the changes to local government (described below) which militated against community building.
The negligible impact of the whole intervention on women's partners was disappointing, and could have been a limiting factor in the effectiveness of the intervention in improving women's health outcomes.
One explanation for the lack of effect of the intervention might have been that the elements of PRISM
or other major alternative maternal health programs were implemented in comparison communities. Local government changes made that unlikely but we assessed the evidence in 2001–2 through 'unobtrusive monitoring' [52
] of, policies, programs and funding at local, state, and commonwealth government levels [53
], and an audit of GP Divisional projects, strategic plans and business plans. We also analysed systematic samples of local newspaper coverage of mothers and maternal health [54
] and surveyed the MCH team leaders in each comparison community to ask about specific local initiatives, finding almost none.
In the five years this trial was being planned (1993–1997) there were marked changes to local government implemented by the State government, including the dismissal of elected local councillors, appointment of commissioners, and the amalgamation of local councils from 210 to 78 [55
]. Service-contracting became a prominent feature of councils' operation for the first time, with a requirement that at least half of all municipal services be put out to compulsory competitive tendering – including, in most municipalities, the MCH Program [56
]. 'In-house' business units, comprising staff previously employed to provide the service directly, won some of the contracts for health and family support services. Some were won by community-based agencies, e.g. community health centres, some by hospitals, and some by private companies. Although the straitened funding co-incident with the reforms made the 50% chance of being provided with resource kits for mothers, professional development for MCHNs and a CDO for two years very attractive, the enforced competition was not the ideal context for a community intervention [57