R ESPONSE R ATE

are explored, and the strong and weak collaboration ties within and between health professionals in the OR are identified. These results are presented in section 8.3 Assessment of Relational Coordination. Third, analyses of the health professionals’

attitudes toward safety culture are conducted, in which any changes over time are explored, and the strong and weak dimensions of safety culture are identified. These results are presented in section 8.4 Assessment of Safety Culture. Fourth, a comparative analysis between the construct of RC and the construct of safety culture is conducted, in which the correlation between the concepts is explored. These comparative analyses generated knowledge about the connection between health professionals’ attitudes toward the interdisciplinary collaboration they are part of and their attitudes toward the safety culture, which the collaboration is embedding within.

These results are presented in section 8.5 Comparing Relational Coordination and Safety Culture. Finally, the analyses and results are discussed, and the third research question is addressed, in section 8.5 Interpretation and Discussion.

8.1 Response Rate

The data derived from all the analyses conducted during the assessments included the completed responses. There were 150 invited respondents at both Time 1 and Time 2, and 137 invited respondents at Time 3, giving a total at 437 invited respondents to the RC Surveys, and a total of 287 invited respondents in the SAQ-DK surveys. The response rates varied over time. At Time 1, 92 respondents completed the survey, at Time 2, 74 respondents completed the survey, while at Time 3, 48 completed the survey. In total, 214 responses were completed, resulting in a response rate¹⁸ of 49%.

A total of 222 responses were deleted due to being missing (51%). The percentage of partial responses was 3.2%, as shown in Table 21.

Responses to the Surveys Over Time

Time 1 Time 2 Time 3 Total

n % n % n % n %

Invited respondents 150 100 150 100 137 100 437 100

Partial responses 4 2.6 2 1.3 8 5.8 14 3.2

No responses 54 36.0 74 49.3 81 59.1 209 47.8

Complete responses 92 61.3 74 49.3 48 35.0 214 49.0

Table 21 Distribution of surveys and responses for all the workgroups over time. The RC Survey was distributed at Time 1, Time 2, and Time 3, while the SAQ-DK was distributed at Time 2 and Time 3.

The number of completed responses decreased from 61.3% (Time 1) to 35% (Time 3), which is considered to be a huge change over time in terms of the response rate. The response rates decreased over time for all the workgroups, except for the coordinating nurses. A table showing the response rates distributed by the workgroups over time is presented in Appendix 8 (Table A). Thus, the distributions of the invited respondents to the workgroups varied over time, as did the response rates. When comparing over time and across workgroups, the differences in the sample distributions and response rates may be considered. The respondents invited to the surveys were participants in one of three large workgroups (OR nurses, surgeons, AN nurses), or in one of three small workgroups (anesthesiologists, nurse assistants, coordinating nurses). To obtain a more detailed view of the variations, the percentage distribution of the invited respondents is compared to the percentage of respondents who completed the survey in each of the workgroups included in the surveys. Comparing the distribution of

“respondents invited” with the distribution of “respondents responding” by workgroups at Time 1, only a small variation was visible, as illustrated in Figure 28. This supports the representativeness of the sample distribution when compared to the population of interest (health professionals employed in the surgical unit who were invited to

18The response rate is defined as the number of respondents who completed the survey divided by the number of respondents invited, expressed as a percentage.

participate in the survey). At Time 2 and Time 3, larger variations were visible when comparing these distributions. Resulting in a possible response bias that must be considered (Draugalis et al., 2008).

Figure 28 Distribution of respondents invited and distribution of respondents responding at Time 1, Time 2, and Time 3 by workgroups.

The distribution of responses by workgroups at Time 2 differed in that proportionately more OR nurses were responded to the survey than surgeons and AN nurses, increasing from OR nurses delivering 23% of the completed responses at Time 1 to them delivering 34% of the completed responses at Time 2 (Appendix 8, Table A). The distribution of responses by workgroups at Time 3 also differed in that proportionately less surgeons responded to the survey than OR nurses and AN nurses, increasing from surgeons delivering 38% of the completed responses at Time 1 to them delivering 29% of the completed responses at Time 3. Looking at the large workgroups included in the survey, the surgeons also exhibited the largest change in response rate over time, decreasing from 56% at Time 1 to 26% at Time 3. The change in the distribution of respondents by workgroups over time resulted in proportionately fewer responses from surgeons being included in the measurements at Time 3, which could influence any changes in RC over time, depending on whether or not there were differences in the measures of RC across the workgroups. In order to correct for unequal and changing response rate a weighting of RC scores for site-level aggregation are calculated, in which individual responses are weighted according to the size of their function in the unit, described in the Guidelines for Theory, Measurement and Analysis (Gittell, 2012a). Only marginal differences were found between weighted and unweighted site-level RC scores at all times (Appendix 8, Table P).

To explore for potential response bias comparisons between respondents answering the survey only once and respondents answering the survey more than once were

Respondents Invited & Distribution of Responses by Workgroups over Time

OR nurse AN nurse SUR ANE NUASS COORNU

respondents who answered the survey once and the respondents who answered more than once, as shown in Appendix 8 (Table B). Thus, the respondents who responded more than once to the survey did not answer the questions more positively or more negatively than the respondents who only responded to the survey once. Therefore, the risk of response bias attributed to the decreasing response rate was assumed to be minimized.

The study sample of health professionals who completed the RC Survey more than once was also used to examine any statistical differences in the RC index and in the RC dimensions over time. The number of completed responses for this study sample varied slightly: 47 respondents completed the survey both Time 1 and Time 2, 33 respondents completed it at both Time 2 and Time 3, and 36 respondents completed it at both Time 1 and Time 3. A total of 26 respondents completed the survey all three times. A table showing the response rate for the respondents who responded more than once distributed by the workgroups over time is presented in Appendix 8 (Table C).

Further considerations about the sample distribution and changes in the response rate over time are discussed later in section 8.7 Strengths and Limitations.

8.2 Test of Reliability and Validity