Validation of the Safety Attitudes Questionnaire for Assessing Patient Safety Culture in Critical Care Settings of Three Selected Ugandan Hospitals

Figures & data

Table 1 Verbatim Comments and Suggestions by the Experts in Face Validation of the Tool and Actions Taken to Address the Comments

Panel Member	Comments and Suggestions Verbatim
FPM 3	Fellow physician and resident physician not clear among “position” options. Underlines “peds” under “primarily” options
FPM 4	Consider separating hospital and unit management items
FPM 6	How many levels are there? (item 29). Not clear - if I am between 6 to 11 months, how many? (years in specialty). Underlines “speak up” in item 2, “errors” in items 11 & 13, “info” in item 28, “discipline” in item 32, and “breakdowns” in item 36. Why make me to read all this when I am going to pick one? (under “position” options)? Font size too small, difficult to read. Shading (with pencil) takes a lot of time, consider ticking instead. Instructions on how to fill the form are not clear, consider having a Likert scale for each item.
FPM 7	I think it would be more clear if this instruction started with the part of “please complete this survey with respect to ….” then continue with “I work in ….” (opening instruction). Item 24–28 seem to need their separate instructions; it has appeared to me somehow incomplete. I suggest that background information is moved to the section above the tool items (ie just before item 1). Rated attending/staff physician and physician assistant/nurse practitioner as not clear under position options.
FPM 8	The words are too small, please increase font size and space. You may need additional instructions for items 24–28. For item 25, put the word “knowingly” at the end; items 27 and 28 are not complete colloquially. In item 34, the word “staff” may be confusing, it was not used in items 33 and 35 which read more clearly. The section for years in specialty, some of the answer options have units in months and not years. The section on “primarily” is not clear.

Abbreviation: FPM, Face validity panel member.

Figure 1 Graphic display of the mean scores by tool item.

Figure 2 Comparison of frequency of missing data between unit-level and hospital-level management-related items.

Table 2 The Construct Validity Model from Confirmatory Factor Analysis: Specifications, Goodness of Fit Indicators, and Internal Reliability Coefficients

Properties/Features	First Model (Based on a Priori Model)	Final Model (After Construct Modification)
Tool items
Overall	Items: 1–13, 15–32	Items: 1, 3–14, 16–24, 26–32
Teamwork climate	Items: 1–6	Items: 1, 3–6
Safety climate	Items: 7–13	Items: 7–13, 3
Job satisfaction	Items: 15–19	Items: 16–19
Stress recognition	Items: 20–23	Items: 20–23
Unit management	Items: 24a–28a	Items: 24a, 26a–28a, 14, 32
Working conditions	Items: 29–32	Items: 29–32, 4, 19
Goodness of fit parameters
Chi-square: value (p-value)	694.28 (p <0.001)	424.98 (p=0.002)
Root mean squared error of approximation (RMSEA)
Value (pclose)	0.051 (pclose=0.427)	0.030 (pclose=1.000)
90% CI	0.044–0.057	0.019–0.039
Comparative factor index (CFI)	0.884	0.966
Tucker–Lewis index (TLI)	0.871	0.960
Coefficient of determination (CD)	0.999	1.000
Internal reliability: Cronbach’s alpha (no. of items)
Overall scale	0.8881(31)	0.8967(29)
Teamwork climate	0.6187(6)	0.6397(5)
Safety climate	0.7874(7)	0.7951(8)
Job satisfaction	0.7374(5)	0.7641(4)
Stress recognition	0.8106(4)	0.8106(4)
Unit management	0.7532(5)	0.8151(6)
Working conditions	0.7244(4)	0.7689(6)