A qualitative, descriptive, multiple-case research design was used [21 ]. The cases comprised the health facilities providing perinatal services in four health regions in the province of Quebec: two urban regions, one mixed urban-rural region and one rural region [18 ]. The regions were selected because they present different models of continuity of care, depending on whether responsibility for postnatal follow-up had or had not been transferred to community partners. Each region has one or two hospitals offering perinatal care, an average of six community health centres, several private medical clinics, and voluntary organizations providing postnatal support. In addition, each region has a regional agency, whose mission includes ensuring the integration and consistency of services between the various healthcare facilities on its territory. These facilities were expected to implement collaboration following introduction of the early discharge of newborns, defined as a postpartum stay of approximately 48 hours (2 days) for an uncomplicated vaginal delivery and approximately 96 hours (4 days) for an uncomplicated caesarean [3 (link)-5 (link)]. This practice requires linking hospital care with primary-care services.
The data were collected through 33 semi-structured interviews with healthcare managers and professionals working in the four regions (Table1 ). An analysis of written material (coordination agreements, protocols, etc.) was also conducted. The sampling was purposeful; respondents were selected to represent management and the different types of professionals. An interview plan based on D'Amour's model of collaboration was developed to guide the interviews. All interviews were recorded on audio tape and transcribed in full.
The data analysis combined two complementary strategies [22 ,23 ], deduction and induction. We thus based our analysis on the theoretical proposals of the analytical model of interprofessional collaboration but left room for new elements to emerge. Each stage in the analysis was conducted independently by three investigators who then compared their findings. The analysis comprised two main stages: an internal analysis of each case followed by a cross-sectional analysis of all the cases [24 ]. The first step involved three levels of analysis: condensation, organization and interpretation [23 ]. The units of meaning were thus coded in terms of the indicators of the model of collaboration or of the emerging themes. A history of collaboration and the factors influencing it was then written for each case and the cross-sectional analysis was then undertaken. The study was submitted for approval to the ethics committees of the University of Montreal and of the participating facilities, and all participants signed a consent form.
The data were collected through 33 semi-structured interviews with healthcare managers and professionals working in the four regions (Table
The data analysis combined two complementary strategies [22 ,23 ], deduction and induction. We thus based our analysis on the theoretical proposals of the analytical model of interprofessional collaboration but left room for new elements to emerge. Each stage in the analysis was conducted independently by three investigators who then compared their findings. The analysis comprised two main stages: an internal analysis of each case followed by a cross-sectional analysis of all the cases [24 ]. The first step involved three levels of analysis: condensation, organization and interpretation [23 ]. The units of meaning were thus coded in terms of the indicators of the model of collaboration or of the emerging themes. A history of collaboration and the factors influencing it was then written for each case and the cross-sectional analysis was then undertaken. The study was submitted for approval to the ethics committees of the University of Montreal and of the participating facilities, and all participants signed a consent form.