Achieving a Climate for Patient Safety
The parent study used a quasi-experimental pre–post survey study design. Envelopes containing invitation letters and questionnaires were placed in employee mailboxes. Subsequent reminders were sent out ~2–3 weeks later. The implementation of the IPMPC was staggered to healthcare teams in four of the five hospital campuses (one of the campuses did not participate), and evaluation data were collected at three time points. Time Zero (T0) data were collected before the inter-professional practice model was implemented, Time One (T1) data were collected 6 months after the implementation of the model by the last team (12–18 months after the model was introduced to the first teams) and Time Two (T2) data were collected 6 months after T1.
All subjects from the population of health professionals were invited to participate in the parent study, except for nurses. Due to their much larger numbers, only a random sample of nurses was recruited. Table 1 displays the response rates by profession for all three data collection periods.
Inter-professional collaboration was measured with a 17-item, 5-point Likert-type scale with higher numbers indicating higher perceptions of collaboration. This instrument was created specifically for this project and was based on a conceptualization of professional collaboration by D'Amour. Cronbach's alpha (α) was 0.94 for this study.
Inter-professional conflict was measured with a 7-item, 5-point Likert-type scale with higher numbers indicating higher perceptions of inter-professional conflict. This instrument was also developed specifically for this project (α = 0.84).
Respect was measured with three items on a 5-point Likert-type scale taken from the effort–reward imbalance questionnaire. The three items measure respect, which is conceptualized in the instrument as a component of esteem, and thus a type of reward. These three items have been used to measure nurses' perceptions of respect in previous work. There were five options for each item, with the first being 'Agree', but the other four options ranged from 'Disagree, but I am not at all distressed' to 'Disagree, and I am very distressed'. Given that the instrument did not measure respect on the same type of ordinal scale as the other instruments, the scale was dichotomized prior to analysis so that option one was left as 'agree', but options two to five were collapsed and recoded as 'disagree' (α = 0.83).
Patient safety climate was measured with an 8-item, 5-point Likert-type subscale taken from the Safety Attitudes Questionnaire. The safety climate subscale assesses the degree to which individuals perceive that their organization has made a commitment to safety; higher numbers reflect higher perceptions of patient safety climate. Seven items came from version one of the safety climate subscale whereas item three came from version two of the same subscale, for a total of eight items (α = 0.79).
For our secondary analysis, health professionals were grouped into three categories: physicians, nurses, and other health professionals. Given that individuals worked on multiple units (e.g. physicians and other health professionals) or worked on one unit but sometimes transferred to other units (e.g. nurses), aggregation to the unit level was not appropriate. Campus (i.e. Civic, General, Riverside, Rehabilitation Centre) was used as an aggregating variable in analysis, because the commitment that is needed to establish a patient safety climate is a site-level (i.e. campus) mandate. A series of generalized estimating equation (GEE) models were generated to account for the clustering of responses by campus. Where there was empirical justification, we added control variables that proposed relationships to any concepts in the model. Thus, age was included because of research demonstrating a positive relationship between age and work (including respect). We included gender because of a recent meta-analysis, which showed gender differences in conflict resolution skills. Marital status and education were included because of research suggesting a relationship between these variables and work values. Interaction terms (collaboration × conflict; respect × conflict) were entered into the models to test for moderation. Data analysis was conducted using the SAS software, Version 9.2, by SAS Institute, Inc. (Cary, NC, USA). All cases with missing values were deleted prior to analysis, and a P-value of ≤0.05 was considered significant.
Methods
Procedures
The parent study used a quasi-experimental pre–post survey study design. Envelopes containing invitation letters and questionnaires were placed in employee mailboxes. Subsequent reminders were sent out ~2–3 weeks later. The implementation of the IPMPC was staggered to healthcare teams in four of the five hospital campuses (one of the campuses did not participate), and evaluation data were collected at three time points. Time Zero (T0) data were collected before the inter-professional practice model was implemented, Time One (T1) data were collected 6 months after the implementation of the model by the last team (12–18 months after the model was introduced to the first teams) and Time Two (T2) data were collected 6 months after T1.
Sample
All subjects from the population of health professionals were invited to participate in the parent study, except for nurses. Due to their much larger numbers, only a random sample of nurses was recruited. Table 1 displays the response rates by profession for all three data collection periods.
Instruments
Inter-professional collaboration was measured with a 17-item, 5-point Likert-type scale with higher numbers indicating higher perceptions of collaboration. This instrument was created specifically for this project and was based on a conceptualization of professional collaboration by D'Amour. Cronbach's alpha (α) was 0.94 for this study.
Inter-professional conflict was measured with a 7-item, 5-point Likert-type scale with higher numbers indicating higher perceptions of inter-professional conflict. This instrument was also developed specifically for this project (α = 0.84).
Respect was measured with three items on a 5-point Likert-type scale taken from the effort–reward imbalance questionnaire. The three items measure respect, which is conceptualized in the instrument as a component of esteem, and thus a type of reward. These three items have been used to measure nurses' perceptions of respect in previous work. There were five options for each item, with the first being 'Agree', but the other four options ranged from 'Disagree, but I am not at all distressed' to 'Disagree, and I am very distressed'. Given that the instrument did not measure respect on the same type of ordinal scale as the other instruments, the scale was dichotomized prior to analysis so that option one was left as 'agree', but options two to five were collapsed and recoded as 'disagree' (α = 0.83).
Patient safety climate was measured with an 8-item, 5-point Likert-type subscale taken from the Safety Attitudes Questionnaire. The safety climate subscale assesses the degree to which individuals perceive that their organization has made a commitment to safety; higher numbers reflect higher perceptions of patient safety climate. Seven items came from version one of the safety climate subscale whereas item three came from version two of the same subscale, for a total of eight items (α = 0.79).
Secondary Data Analysis
For our secondary analysis, health professionals were grouped into three categories: physicians, nurses, and other health professionals. Given that individuals worked on multiple units (e.g. physicians and other health professionals) or worked on one unit but sometimes transferred to other units (e.g. nurses), aggregation to the unit level was not appropriate. Campus (i.e. Civic, General, Riverside, Rehabilitation Centre) was used as an aggregating variable in analysis, because the commitment that is needed to establish a patient safety climate is a site-level (i.e. campus) mandate. A series of generalized estimating equation (GEE) models were generated to account for the clustering of responses by campus. Where there was empirical justification, we added control variables that proposed relationships to any concepts in the model. Thus, age was included because of research demonstrating a positive relationship between age and work (including respect). We included gender because of a recent meta-analysis, which showed gender differences in conflict resolution skills. Marital status and education were included because of research suggesting a relationship between these variables and work values. Interaction terms (collaboration × conflict; respect × conflict) were entered into the models to test for moderation. Data analysis was conducted using the SAS software, Version 9.2, by SAS Institute, Inc. (Cary, NC, USA). All cases with missing values were deleted prior to analysis, and a P-value of ≤0.05 was considered significant.
SHARE
