Setting
The context of this study was the implementation of IPS in Bodø city in Nordland county in Norway. Nordland County is geographically placed in the northern part of Norway, and Bodo is the capital and largest city in the region, with 50 000 inhabitants. Other municipalities in Nordland county, where IPS was not yet implemented, were included as reference group. Nordland County has 44 municipalities, with 43 of these having their own statutory Labour and Welfare Administration offices. These offices represent Norway´s Public Employment Services, functions defined at EU level: (https://ec.europa.eu/social/main.jsp?catId=105&langId=en)
In 2013, there were 450 employees in these offices, which had increased to 480 employees in 2017. Traditionally there has been little coordination between these labour and welfare offices and mental health services to support individuals with severe and enduring mental health conditions into work. The traditional way of working has been a step-wise train and place approach [26], similar to the control arm in IPS efficacy trials. Prior to 2013, there was no IPS activity in Nordland County. In 2013, the Labour and Welfare Administration office, along with the mental health service in the municipality of Bodø, began implementing IPS and employed three full-time IPS employment specialists. Between 2012 and 2017, all the employees both in health and NAV in Bodø were trained in the principles of IPS, and several local meetings and seminars about work and mental health were organized. The IPS program had three independent IPS fidelity reviews during this period: first in September 2013 scoring fair fidelity, and the second and third scored good fidelity. The rest of the county did not start implementing IPS before 2017.
Questionnaire
A questionnaire was developed based on the eight key principles of IPS. The questionnaire began with a case vignette about a hypothetical patient “Line” and continued with statements with reference to this vignette (text box 1). Eight paired statements were developed according to the eight key principles for IPS (Figure 1). For each pair, one statement was in accordance to an IPS principle whilst the other statement opposed it. The statements opposing IPS were generally in line with current practice in the Labour and Welfare Administration, and in accordance with ‘train-and-place’ approaches to vocational rehabilitation.
Insert Text box
For example, for the IPS principle “Rapid job search” we developed the statement: “Line wants to work now. This means that the job search should start as soon as possible” paired with the divergent statement: “We need adequate time for work preparation and treatment before looking for competitive employment”. Responses were collected on a six-point Likert scale defined by each pair of statements. The respondents were asked to read the case vignette and respond to the statements marking on the Likert scale which best indicated their view. On two of the questions, 2 and 7, a score of six on the scale indicated attitudes in accordance with IPS, whereas a score of six on questions 1,3,4,5,6 and 8, indicated attitudes more in accordance with usual practice. To check conceptual validity, as part of questionnaire development a focus group was conducted with representatives from the local labour and welfare offices which further refined the questionnaire.
Insert Figure 1
Design and participants
The survey was conducted first between September and October 2013, and then again in May 2017. Survey data were collected in the context of meetings and seminars, or distributed by the office managers at each of the 43 offices. At the seminars, one of the authors collected the responses with help from personnel from NAV. When the survey was completed in the local offices, the local office managers sent the responses to the researchers by post. Completing the survey was voluntary and anonymous.
Statistics
Data were analysed by Stata version 15 [27]. Scores for each of the statements, as well as a sum score for overall attitude towards the IPS principles, were computed. The internal consistency of this scale was analysed by Cronbach’s coefficient alpha. The independent samples t-test was used to analyse differences between responses from the geographical area where IPS was implemented and areas without IPS experience. This was executed for the total sum-score and for each of the individual items at two timepoints. Multiple linear regressions with attitudes as the dependent variable were also used to test the hypothesis that attitudes changed over time, dependent on exposure to IPS. The hypothesis for interaction was tested by an interaction term (time by site) where both were coded as dichotomous indicating 2013 versus 2017 and IPS versus no IPS implementation area.