Study Group
The study was carried out in 2008 in all the public hospitals--1 general hospital, 1 teaching hospital, and 1 university hospital--located in the metropolitan center of Konya Province, Turkey. Konya is one of the most important cities in central Anatolia. It is world renowned for its history and culture and is famous as an international tourist city.
More than 90% of healthcare services in Turkey are provided by public enterprises. Out of all public hospitals, 775 are general, 59 are teaching, and 57 are university hospitals. Certain general, teaching, and university hospitals have the greatest bed capacity; therefore, this study was carried out in the hospitals that represent these three types of large-capacity hospitals. About 96% of the people in Turkey have social security insurance that provides free health care, and patients are free to choose their hospitals.
Efforts were made to include physicians and nurses--both permanent and temporary--who have direct interaction with patients. The temporary group includes nurses, emergency technicians, and midwives whom they work on nursing services. These workers operate under the direction of supervisors and are under contract for provisional time periods. The size of the study group was determined by a formula that enabled a comparison of the predicted mean score of 63 ± 10 [18] for positive perception of the overall patient safety culture in the hospital within an SD of 5 points, at a 95% confidence level and with a 0.80 power. Thus, the goal was to reach at least 63 people in each group. However, considering the careful attention paid to representing as many units as possible and the likelihood of a low rate of response, the surveys were delivered individually many of the available staff. The response rate varied from 51% to 91%, depending on the profession. The response rate was 51% for doctors, 64% for temporary nurses, and 91% for registered nurses. By hospital type, response rate was 56% for university hospitals, 72% for general public hospitals, and 86% for teaching hospitals. A total of 309 questionnaires were completed and returned as follows: 102 physicians, 135 nurses, and 72 temporary nurses from three public hospitals. The demographics and professional characteristics of the participants are shown in Table 1.
Table 1 The demographic and professional characteristics of the participants
Measurement
Hospital Survey on Patient Safety Culture (AHRQ)
A Turkish version of by AHRQ's "Hospital Survey on Patient Safety Culture" was developed [16]. The survey was composed of items measuring the dimensions of patient safety culture at work area/unit and hospital levels and included questions measuring outcome variables. Moreover, there were 7 items that asked for personal information. Eighteen of the 42 items belonging to 12 dimensions were worded negatively. The five-level Likert scale was employed for the responses. The scales used for the questions were (1) strongly disagree, disagree, neither, agree, or strongly agree and (2) never, rarely, sometimes, most of the time, or always. Structural validity of each safety culture dimension was partially associated with all others and was reflected by composite scores. This relationship was shown with a correlation between 0.23 and 0.60. Highest correlations were calculated for result variables of "overall perceptions of patient safety" and "patient safety grade" (r = 0.66, p < 0.001). The lowest relationship was revealed by "frequency of events reported" and "non-punitive response to error" (r = 0.23, p < 0,001). The differing nature of safety culture composite scores between hospitals was shown with ANOVA [16].
Preparing the Turkish version of the survey
The investigators performed initial translation of the survey into Turkish. An expert in the English language whose native language is Turkish reviewed this translation. Then an independent translator who had not seen the original questionnaire [19] translated it back into English. Sorra, who is one of the authors of the original survey, checked the last translated questionnaire. The Turkish questionnaire was revised according to suggestions on 3 of the 42 items that might have led to differing interpretations. The items of the translated survey were categorized under 10 factors. This might be caused by ambiguity of the translation or cultural differences. Three items were relocated between the dimensions. It was assumed that this state would not affect the general score of the culture of patient safety.
Pilot study
The comprehensibility of the survey was tested on 5 physicians and 8 nurses who had not been included in the study group and their opinions were used to prepare the final version of the survey.
Data collection
An assistant researcher made an in-person presentation about the surveys to the study group. After participants gave verbal consent, each participant was asked to fill out the survey without discussion.
Ethical considerations
The researchers obtained the approval of the ethics committee and obtained institutional permits and consent of the participants.
Statistical Analyses
The researchers checked to make sure the data set was complete and to determine if the inter-item correlations were sufficient (Bartlett's test χ2 = 4238; df = 861; p < 0.001). There was no high correlation between items. In addition, the Kaiser-Meyer-Olkin measure of sampling adequacy was determined. The KMO-score was 0.9, far above Kaiser's criterion of 0.5. The pre-analyses demonstrated that the data could be used for factor analysis. A confirmative factor analysis was performed (with Varimax rotation) to investigate whether the factor structure of the AHRQ survey could be used with Turkish data. In addition, it was studied using explorative factor analysis to check whether the items formed different factors in the Turkish situation. The construct validity was studied by calculating scale scores for every factor (after any necessary reverse coding) and the correlation coefficients between the scores. The internal consistency of the factors was calculated with Cronbach's alpha.
For each positively worded item, the percentage of positive responses was calculated--that is, the percentage of respondents answering the question by checking "strongly agree" and "agree" or "always" and "most of the time" [16]. The data obtained was summarized as percentage, mean, and SD values. Categorical values were examined using the chi-square test. ANOVA and t tests were applied to compare the mean values, and P < 0.05 was recognized as statistically significant.