OUP user menu

Incident reporting culture: scale development with validation and reliability and assessment of hospital nurses in Taiwan

Hui-Ying Chiang, Ya-Chu Hsiao, Shu-Yuan Lin, Huan-Fang Lee
DOI: http://dx.doi.org/10.1093/intqhc/mzr031 429-436 First published online: 13 June 2011

Abstract

Objective To examine the psychometric validity and reliability of the incident reporting culture questionnaire (IRCQ; in Chinese) following an exploration of the reporting culture perceived by hospital nurses in Taiwan.

Design Scale development with psychometric examination and a cross-sectional study.

Setting Ten teaching hospitals.

Participants A total of 1064 nurses participated with an average response rate of 83% between November 2008 and June 2009.

Main Outcome Measures The factorial construct, criterion-related validity, homogeneity and stability of the IRCQ were evaluated. The nurses’ perceptions of the IRCQ were also explored.

Results The four-factor structure of the 20-item IRCQ had satisfactory construct validity (explained variance: 49.37%), criterion-related validity (r = 0.42; P = 0.001), reliability (Cronbach's alpha: 0.83) and stability (3-week-interval correlation: r = 0.80; P = 0.001). These factors included ‘application of learning from errors’, ‘readiness to provide feedback on incident reports’, ‘collegial atmospheres of unpleasantness and punishment’ (CA) and ‘incident management: confidential and system driven’. The nurses perceived a moderate overall reporting culture (mean positive response = 49.25%; range: 67.2–24.94%). They weakly agreed on the CA factor of five items (mean positive response = 24.94%; range: 33.0–17.2%).

Conclusions This study provides empirical evidence for the psychometric properties of the IRCQ and the reporting culture which nurses perceive in Taiwan. To Taiwanese nurses, the reporting culture within their work environments especially as it relates to coworker relations, inter-professional collaboration and non-punitive atmosphere is their major concern. Healthcare administrators should consider nurses’ perceptions related to incident reporting when managing underreporting issues.

  • hospital incident reporting
  • safety management
  • organizational culture
  • nurses

Healthcare organizations have cultures of incident reporting, which enable frontline professionals to report incidents candidly, learn from errors openly, approach incidents effectively and commit to safety and quality wholeheartedly [1, 2]. As in the case with organizational culture, incident reporting culture represents the shared values, beliefs and principles among a group of professionals, which can influence their organizational communication, social relations and individual actions and motivations for reporting incidents such as errors, adverse events and near misses [3, 4]. Moreover, this culture encourages frontline practitioners to strengthen safety awareness and to acknowledge human vulnerability, while fostering a non-punitive learning culture and a willingness to report [1, 57]. Most incidents are recognized and reported by nurses who occupy 40–60% of the healthcare workforce and who play a major role in the surveillance of patient safety at work [3]. However, the underreporting of safety events is noted by nurses to be the result of unfavorable cultural and organizational factors in clinical environments [811]. In Taiwan, underreporting continues to present a problem for healthcare professionals [12]. Recently, a study showed that nearly 50% of Taiwanese nurses had failed to report medication errors in hospitals [9]. Nurses’ perceptions of organizational cultures and the way in which these perceptions affect incident reporting deserve deeper exploration. Such exploration can enhance the quality of voluntary incident reporting and inform changes in policy and practice.

In Taiwan, improving the quality of incident reporting has been recognized as crucial within the healthcare community since 2005 [13, 14]. Various initiatives have been adopted by healthcare organizations involving changes in leadership, government policy, research funding, quality management and hospital accreditation procedures to improve both patient safety and the incident reporting culture [14]. However, official reports have stated that medical errors and underreporting issues were underestimated among Taiwanese healthcare professionals [12]. Factors leading to nurses’ underreporting included burdensome processes of reporting, fear of reprisals, unsupportive attitudes from management, blame cultures and distrust of systematic correction generated from reported data [7, 9, 10]. These factors, influenced by traditional cultures of professionalism and organizational operations, are a manifestation of the reporting culture [11, 15, 16]. An applicable and valid measurement of the reporting culture can provide valuable insight into the organizational culture of incident reporting and can help to alleviate underreporting issues in the nursing workplace.

Basically, the reporting culture is dependent on patient safety culture and is commonly considered as a subculture or an index of patient safety performance [1517]. Therefore, reporting culture is seldom examined separately and thoroughly. In Taiwan, two translated instruments: the Safety Attitude Questionnaire [18] and the Hospital Survey on Patient Safety Culture [19] are generally accepted as valid measurements to investigate cultures of global patient safety containing frequency or willingness of reporting among professional and non-professional staff members. However, their dimensional constructs do not fully capture characteristics of the reporting culture such as shared beliefs, values and perceptions in a given working environment. They focus less on dynamic reporting processes such as barriers to making reports, quality improvements from reports, accessibility to the reporting system, administrator's attitudes toward error management and data confidentiality. Moreover, the reporting culture is interpreted and manifested differently depending on professional practices [2]. Studies have shown that healthcare practitioners’ attitudes and behaviors toward incident reporting vary by professional disciplines [57]. This indicates that the care practitioners’ perceptions of incidents and reporting are determined by their traditional professional cultures, work characteristics and practice environments. Yet, there is limited information on the extent to which frontline nurses perceive the incident reporting phenomena as part of their working environments. The aims of this study were to examine the psychometric properties of a newly developed incident reporting culture questionnaire (IRCQ; written in traditional Chinese characters) and to explore the nature of hospital nurses’ perceptions of the reporting culture in Taiwan. This exploration could be used as a baseline for understanding the reporting culture existing in Taiwanese hospital milieus.

Methods

This study was designed to develop an instrument following a cross-sectional survey to examine the scale's psychometric properties and its effectiveness in describing the reporting culture. Three groups of eight frontline nurses were invited to generate questionnaire items. Those nurses, senior nurses and shift leaders, not only provide direct nursing services but are also responsible for safety event surveillance in their wards. Thus, they were aware of incident reporting issues and had experience with reporting in their working hospitals. In the psychometric validation process, construct validity using exploratory and confirmatory factor analyses, criterion-related validity, stability and internal consistency were conducted.

Designed questionnaire

The IRCQ items were generated by group discussions with the frontline nurses. During the discussions, major issues discussed were: (i) experiences of making incident reports, (ii) attitudes and perceptions of incident reporting systems and (iii) feedback received from making incident reports in their working environments. For each group, the 2-h discussion content was taped and transcribed verbatim. The first author analyzed and identified repetitive keywords, phrases or concepts in the transcripts. From this, the initial items were determined. The second author used the same method to confirm the initial items separately. Any discrepancy within the confirmation process was discussed and resolved by the two authors. Initially, 30 items were generated and categorized under 5 subheadings (6 items in each) including: (i) encouragement to report, (ii) perceptions of organizational climate about incident management, (iii) quality improvement through reporting, (iv) feedback given following reports and (v) perceptions of the reporting process. Five nurses from the groups were randomly selected to review these items independently for word appropriateness and meaning clarity. After this, two items were deleted because of duplicate meanings and others underwent minor revision in wording. The final tool consisted of 28 items using a 5-point Likert-scale format (i.e. 1 = ‘strongly disagree’, 2 = ‘disagree’, 3 = ‘neutral’, 4 = ‘agree’ and 5 = ‘strongly agree’) which was applied to study construct validation. There were 19 positively worded items and nine negatively worded items, which were equally distributed on the subheadings.

The criterion-related validity was examined using correlation analyses between the reporting culture and working behaviors focusing on patient safety assurance. The organizational culture greatly influences healthcare practitioners’ commitment and behavior related to safety [4]. That is, the reporting culture is derived from the patient safety concept and corresponds to pervasive patient safety behaviors enacted by nurses. Therefore, the nine-item SOS with a 7-point Likert-type scale (from 1 = not at all to 7 = a very great extent) was administered to the nurses to measure their behaviors related to safety culture [20] and then treated as a criterion variable. The scale was confirmed by convergent, discriminant and criterion validities [20]. It had satisfactory reliability (Cronbach's alpha: 0.88) in the USA [20] and in Taiwan (Cronbach's alpha: 0.92) [21]. In addition, the nurses’ demographic characteristics were collected in this study.

Study site and population

In Taiwan, three types of hospitals have been designated based on the number of hospital beds into three categories as described in the Taiwanese hospital accreditation program: medical center (large), regional hospital (medium) and district hospital (small). More specifically, the medical centers and the regional hospitals provide major healthcare and professional education services and thus are qualified as teaching hospitals. The participants were recruited from 10 teaching hospitals including 3 medical centers and 7 regional hospitals. Sample size was determined by factor analysis requirements and had a response rate of 80%. The rules of sample size estimation are at least 10–15 subjects per initial item for factory analysis and 500 subjects are considered to be very adequate [2224]. Nurses who directly provide nursing services were eligible to participate. For each study hospital, 20% of the nurse population in different nursing units was invited. The total survey sample of 1064 had an overall response rate of 83% (ranging from 66 to 90% in the study hospitals).

Data collection procedures

The study was initially approved by the ethical committee at the Chi Mei Medical Center in Taiwan. Prior to data collection, Human Subjects Research permission was obtained from study hospitals’ research committees. Survey packages were delivered to the nurses including a questionnaire, a cover letter explaining the research purposes and a return envelope. The time to complete the study questionnaire was 15–20 min. Once the participants completed the questionnaires, they were asked to seal them in the enclosed envelope and then return them. Consent to participate was assumed based on a returned questionnaire. Data collection took place from November 2008 to June 2009.

Data management and analysis

A total of 1064 surveys were randomly split into two subgroups using a computer program for cross-validation approach: exploratory factor analysis (EFA; n = 515) then confirmatory factor analysis (CFA; n = 489) (Table 1). Sixty surveys were eliminated from the factor analysis because of a substantial amount of missing data on the IRCQ items. Considering adequate sample sizes and random patterns of missing values, this elimination had the possibility to increase unbiased justification for factor analyses [22]. First, a hypothesized factor structure of the IRCQ was determined by using principal component analysis (PCA) extraction with varimax solution and item-to-factor loading of at least 0.40. The criteria for factor extraction were an Eigenvalue criterion of 1.0 or greater and scree plot. Second, CFA (using AMOS 6.0) was applied to examine whether the hypothesized factor structure provided a good fit for the research data. In order to evaluate the fit of the structure, the four most used criteria were examined: the chi-square test, the ration of maximum-likelihood chi square (χ2/df, score from 2 to 5), the adjusted goodness-of-fit index (AGFI ≥0.90), comparative fit index (CFI ≥0.90) and the root-mean-square error of approximation (RMSEA ≤0.08) [22, 25].

View this table:
Table 1

Characteristics of the participating nurses

VariableTotal group (n = 1064)EFA subgroupa (n= 515)CFA subgroupa (n= 489)
n%n%n%
Age group (year)
 21–2521119.8938.110321.1
 26–3044942.221241.221544.1
 31–3527826.214027.212225.0
 36–40736.9387.4336.7
 41–45373.5254.991.8
 46–5090.851.040.8
 Above 5050.520.420.4
Education
 Diploma39537.218837.518337.3
 Bachelor and above66762.832763.530662.7
Tenures of present work
 ≤1 year10910.3499.55310.8
 1–3 years24923.411522.312325.2
 3–5 years28626.913810.713327.2
 5–9 years19518.39618.78417.2
 9–11 years1049.85410.5469.4
 11 years above11711.06111.24910.0
 Not reported30.320.410.2
Specialty of nursing practice
 Medical-surgical42439.910938.630441.8
 Intensive care23121.710821.211323.1
 Pediatric524.9397.8377.6
 Operation room595.6336.5214.3
 Emergency757.1346.6357.2
 Psychiatric222.191.7112.2
 Outpatient323.0193.7102.0
 Others757.17114.05711.7
Failure to report at present job
 No58154.727052.427556.2
 Yes42640.121241.219439.7
 Not applicable504.7285.4204.1
 No response60.651.000
  • aRespondents without missing values of the IRCQ.

Third, the reliability of the IRCQ was assessed by test–retest reliability and internal consistency. A sample of 40 nurses who were recruited separately from the study hospitals completed the IRCQ again after a 3-week interval. Pearson's correlation analyses of these two surveys were performed. To determine internal consistency, the reliability of the total scale and subscales was evaluated using Cronbach's alpha coefficients. Fourth, Pearson's correlation analyses between the SOS and the IRCQ were performed to examine the criterion-related validity. In all correlation analyses, P-values <0.05 were considered statistically significant. Finally, descriptive statistics including means, ranges, standard deviations, frequency counts and percentages of all demographic variables and the IRCQ were performed according to data levels. The IRCQ items and subscales were analyzed by computing positive responses to explore the nurses’ agreements with incident reporting culture. The responses of ‘agree’ or ‘strongly agree’ were categorized as positive for the positively worded items and ‘disagree’ or ‘strongly disagree’ were categorized as positive for the negatively worded items. For each item, the number of positive responses was counted and then divided by the total number of the sample. The average positive response rates of each subscale and total scale were calculated. The statistical analysis was carried out using the SPSS version 14.0 and the AMOS version 6.0 statistical software.

Results

A total of 1064 female nurses participated with a mean age of 29.71 (SD = 3.84) and with 62.8% of them holding at least baccalaureate degrees. Six hundred and forty-four participants (60.6%) had tenures of <5 years and 426 (40.1%) of them had failed to report incidents at their present jobs (Table 1).

Construct validity

Prior to conducting EFA, five items with low internal consistency (item-to-total correlation coefficient <0.3) were deleted. The values of the Kaiser-Meyer-Olkin test of 0.87 (>0.80) and the Barlett's test of sphericity (χ2 = 2818.05, P = 0.001) were significant without violation in factor analysis and sampling requirements [22]. In Table 2, the four-factor solution accounted for 49.37% of the explained variance of the IRCQ after deleting three items. These three items with low item-to-factor loadings (<0.40) were excluded because they were not sufficient to be included in defining a factor [24]. Considering conceptual meaning and contribution to item-subscale correlation, Item A6 was related to Factor 1. These 20 items generated the four factors: ‘application of learning from errors’ (AL, five items), ‘readiness to provide feedback on incident reports’ (RF, six items), ‘collegial atmosphere of unpleasantness and punishment’ (CA, five items) and ‘incident management: confidential and system driven’ (IM, four items). These four factors had Eigenvalues >1.0 and explained variances ranging from 14.30 to 10.58% (Table 2). The items of the AL, RF and IM are positively worded and the items of the CA are negatively worded. This four-factor solution, a hypothesized structure, was analyzed using CFA to test the construct validity of the IRCQ. The CFA results showed that the four-factor structure had adequate fitness of the data based on the following fit indices: χ2 = 579.05 (df = 164, P = 0.000), χ2/df = 3.53, AGFI = 0.89, CFI = 0.82 and RMSEA = 0.07. Because the chi-square test is very sensitive to sample size (n > 250) to be significant, the other fit indices were considered to be acceptable for the fitness of the research data except the CFI value (<0.90).

View this table:
Table 2

Factorial structure of the IRCQ by using EFA (n= 515)

CodeItemsFactor loadinga
1234
Factor 1: Application of learning from errors
 A26Information about incidents is disclosed periodically that helps staff to deliver healthcare services mindfully.0.810.050.100.13
 A27Incidents that need to be reported are clearly defined and recognized in the hospital policy.0.800.160.090.12
 A21On this unit, the incident report is applied to identify problems for quality improvements.0.690.260.000.86
 A6Knowledge acquired through the incident report is used as in-service training.0.460.440.010.10
 A19To staff, their professional competences are advanced through the incident report mechanism.0.430.31−0.150.28
Factor 2: Readiness to provide feedback on incident reports
 A5Staff is willing to find out the cause of an incident.0.080.670.040.06
 A1The incident reporting system is easy to use.0.090.650.070.09
 A13The incident report is well documented for clinical practices.0.130.62−0.150.34
 A14This hospital facilitates system safety and care quality by incident report mechanisms.0.320.530.110.19
 A8Data reported are timely and accurate.0.380.460.030.23
 A10The unit that is involved in incidents will be notified soon after reporting.0.330.41−0.010.15
Factor 3: Collegial atmospheres of unpleasantness and punishmentb
 A18Reporting incidents could bring about distress in relationship among coworkers.0.000.010.78 0.12
 A28Reporting errors could cause tension and disharmony among departments.−0.190.150.720.12
 A15Staff who commit errors tend to be blamed.0.14−0.150.680.15
 A20The purpose and implementation of incident reporting systems are not addressed clearly.0.290.070.64−0.12
 A7Regardless of making reports, the improvement of the unit or department that incurs incidents is not effective.−0.010.370.50−0.16
Factor 4: Incident management: confidential and system driven
 A17The identity of people who made reports is confidential.0.070.15−0.010.78
 A9Reported data are anonymous prior to open discussion and disclosure.0.090.20−0.010.68
 A24Attitude toward error reporting is ‘system-driven’ not ‘individual-driven’.0.360.000.170.57
 A11Staff is comfortable to report a variety of safety events.0.090.390.170.47
Eigenvalue2.942.802.402.22
Percent of explained variancec14.3013.3311.4210.58
  • aPCA with varimax rotation and factor loading ≥0.40. bItems of Factor 3 with negatively worded items were scored inversely. cKaiser-Meyer-Olkin value: 0.87, Bartlett test χ2 = 2818.05, P < 0.01.

Additionally, for the total group, the inter-factor correlation coefficients ranged from 0.60 (P = 0.001) to 0.13 (P = 0.001) (Table 3). As to the criterion-related validity analysis, the correlations between the SOS and the IRCQ (r = 0.42) and the IRCQ factors (rs: 0.21–0.40) were significant (P = 0.001). The strength of correlation was located between low and medium. In summary, the IRCQ had satisfactory construct validity and criterion-related validity.

View this table:
Table 3

Reliabilities and correlation analysis of the IRCQ factors and the SOS (n = 1064)

Factorα1234SOSa
1. Application of learning from errors0.731.000.40**
2. Readiness to provide feedback on incident reports0.730.60**1.000.36**
3. Collegial atmospheres of unpleasantness and punishment0.700.13**0.20**1.000.24**
4. Incident management: confidential and system driven0.680.47**0.54**0.21**1.000.21**
IRCQb0.830.74**0.81**0.57**0.75**0.42**
  • aSOS with a 7-point Likert scale (1 = not at all to 7 = a very great extent; Cronbach's alpha: 0.91). bIRCQ with a 5-point Likert scale (1 = strongly disagree, 2 = disagree, 3 = neutral, 4 = agree and 5 = strongly agree).

  • *P < 0.05; **P< 0.01.

Reliability

Table 3 shows that the IRCQ had satisfactory reliabilities for the total scale (Cronbach's alpha: 0.83) and subscales (Cronbach's alphas: 0.68–0.73). The 3-week test–retest reliability revealed a correlation coefficient r = 0.80 (P = 0.001). The IRCQ had suitable internal consistency and stability.

Description of the incident reporting culture

In Table 4, the mean scores of the subscales were located near medium. Taking into consideration that the scale format of the IRCQ is a 5-point Likert type, an appropriate interpretation of the scale is to compute positive responses by item. As mentioned previously in the analyses of the IRCQ (see Methods section), the AL, RF and IM with positively worded items were recoded directly and the CA with negatively worded items was recoded inversely. This could provide insights into the nurse's agreement and disagreement with a particular statement or factor in the reporting culture. Moreover, we proposed to make a positive response rate of 50% as acceptable. According to recent surveys in Taiwan, healthcare professionals reported around 50–60% of satisfaction with overall safety culture [18, 19] but nurses had substantial reporting barriers [26, 27] and estimated that 60% of medication errors had been reported [28]. Thus, 50% of the nurses holding positive attitudes toward the IRCQ factors might be considered reasonable. The average positive response rate of the total scale was 49.25% (item range: 67.6–17.2%) with the AL factor (64.84%) and the RF factor (54.60%) being the top two. That is, using information related to incident reports to improve care quality, in-service education and professional competence was positive. Probably, under the reporting mechanism, the nurses were more favorable to hospital-dominated facilitations of safety and quality, timely and accurate reporting and receiving prompt notification of incident occurrences except for reporting documentation.

View this table:
Table 4

Descriptive statistics of the IRCQ (n = 1064)

FactorMeanSDScore rangePositive response rate (%)Mean of positive response rate (%)
IRCQ68.757.4839.00–97.0049.25
Factor 1: ALa18.372.2711.00–25.0064.84
 A263.720.6467.6
 A273.710.6367.0
 A213.700.6167.6
 A63.670.6663.7
 A193.570.6958.2
Factor 2: RFa21.092.8211.00–29.0054.60
 A53.390.7546.7
 A13.480.7654.2
 A133.250.7136.1
 A143.670.6864.0
 A83.630.6763.1
 A103.660.7063.5
Factor 3: CAb15.582.926.00–25.0024.94
 A183.050.8927.7
 A282.910.8733.0
 A153.270.9220.9
 A203.290.8317.2
 A73.060.8225.9
Factor 4: IMa13.702.474.00–20.0052.67
 A173.400.9251.0
 A93.410.8952.1
 A243.640.7463.8
 A113.260.9143.6
  • aPositively worded items with ‘agree’ or ‘strongly agree’ responses were categorized as positive responses; bNegatively worded items scored inversely with ‘disagree’ or ‘strongly disagree’ responses were categorized as negative responses.

The CA factor (24.94%) and its items (five items <50%) had the lowest positive response rates. Obviously, the nurses expressed disagreement with overall incident reporting culture especially the dissatisfaction about working atmospheres surrounding colleagues and departments after reporting and being blamed for errors. These unpleasant situations included having distressful peer relationships (A18), causing inter-departmental conflicts (A28), being blamed (A15), not knowing about the purpose and implementation of incident reporting (A20) and ineffective improvements (A7). Moreover, there were three other items with positive response rates <50% including willingness of conducting error analyses (A5), documentation in incident reports (A13) and being comfortable in reporting all incidents (A11). In summary, the Taiwanese nurses evaluated the overall incident reporting culture as modest. Noticeably, they perceived punishment for errors and a collegial atmosphere of unpleasantness after reporting.

Discussions

Overall, the IRCQ had acceptable reliability and stability despite the IM subscale having a slightly lower coefficient alpha (0.68). Considering newly developed scales, using large sample sizes and a smaller number of items in a subscale, a coefficient alpha >0.65 is allowable [23]. Additionally, items of the IM subscale had median to high factor loadings (0.47–0.78), revealing that the items were highly correlated with this subscale. Since coefficient alphas are determined by the number of items and inter-item correlations, adding high congruent items could improve subscale homogeneity [29]. Thus, the dimension of incident management in a way of being confidential and system-driven deserves further explorations among nurses in Taiwan.

The IRCQ factorial structure was a reflection of the feature of incident reporting culture and the approach to human error. The reporting culture is multi-dimensional, having four factors: application of learning from errors, readiness to provide feedback on incident reports, collegial atmospheres of unpleasantness and punishment and incident management: confidentiality and system driven. This four-factor structure clearly indicates the positive elements within incident reporting systems such as using information about incidents for safety improvement, learning from errors by system-based approaches, knowledge of human error management, no-blame culture, positive and timely feedback and confidentiality [57, 30]. As noted, error reporting not only provides safety defenses but also yields information to correct system impairment through in-service education, quality improvement and professional competence enhancement [1, 10, 31]. Moreover, an ideal climate for reporting can be characterized as fair treatment after reporting, confidentiality and ease of use and timely feedback [4]. Timely and valid feedback is necessary to encourage individual and departmental involvement in the reporting, identifying causative factors of error and seeking solutions to prevent incidents [30]. Alternatively, work environments featuring culture of blame, negative responses from coworkers and departments and lack of confidentiality could inhibit organizations from creating an agreeable climate for reporting to take place [7, 10, 32]. Nursing requires strong collaboration and interaction with colleagues, peer relationships that directly or indirectly influence nurses’ attitudes, behaviors and working climates in clinical settings. Besides, maintaining harmony and favorable interpersonal relationships is one important trait of an organizational culture of collectivism such as exists in Taiwan [33]. Employees in collectivist societies are more concerned about harmonious interpersonal relations because of the high value on ingroup solidarity and intragroup harmony in workplaces [33, 34]. The myth of zero-error performance is rooted in traditional cultures of nursing professionalism. Thus, making mistakes, a symbol of incompetence, is apt to cause blame and inculcation [6]. This is a plausible explanation for why Taiwanese nurses recognize the collegial atmospheres of unpleasantness and punishment as one factor of the reporting culture. It is clear that these unfavorable and blame-like attributes within the CA factor, measured by negatively worded items, need to be further validated with regard to methodological concerns such as social desirability and acquiescence bias and theoretical concerns such as cognitional and conceptual equivalences between positively and negatively phrased items [23, 29, 35, 36].

The IRCQ has an acceptable criterion-related validity using the SOS as a criterion variable. As indicated in the literature, the safety culture could foster the staff's compliance with safety-related practices [4]. Previous studies showed that nurses’ perceptions of safety cultures influenced their safety-related behaviors such as awareness of safety risks, involvement in quality improvement and open discussion of errors [18, 31, 37]. As expected, the relationship between the IRCQ and the SOS was significant with modest correlation (r = 0.42). Since employees’ behavioral performances related to safety are the result of organizational culture, management policy, professional competence and individual cognition simultaneously, using more criterion variables for empirical validation is recommended.

The nurses’ assessment of the reporting culture was unsatisfactory because the average positive response in the total scale was <50% especially in the CA subscale. Other items with low-positive response rates (<50%) were A13, A5 and A11. These results reflect recent studies of reporting barriers targeting frontline nurses in Taiwan [9, 26, 27]. After reporting, nurses struggled with distress concerning colleague relationships, fear of punitive action, negative feedback from the administration and disappointment at quality improvement in work settings. Plausibly, an unfavorable reporting climate could result in underreporting by nurses. These reporting obstacles likely originate from the organization itself and then affect practitioners’ perceptions, beliefs and attitudes toward incident reporting practices. The above explanation requires further examination of the practical significance existing in the incident reporting rate, nurse reporting behavior and cutoff of the IRCQ. In brief, Taiwanese nurses’ disagreements on incident reporting culture are substantial, particularly collegial atmospheres of unpleasantness and punishment.

Conclusions

The IRCQ is a useful instrument with satisfactory reliability and validity. Further field testing on different healthcare professionals following item modification is needed. Moreover, this study provides empirical information about the incident reporting culture based on nurses’ perspectives in Taiwan. Nowadays, the global patient safety culture is maturing gradually through technological and engineering improvements to ensure safety in the healthcare practices and systems in Taiwan [18]. Overtly, there is room to improve the reporting culture. Hospital administrators should invest more efforts in this culture both through rebuilding and internalization. It is also necessary to be concerned with coworker relations, inter-professional collaboration and punitive atmosphere in situations in which nurses have underreporting issues.

Funding

This study was, in part, supported by the Chi Mei Medical Center, Tainan, Taiwan (grant no: CMFHR9659).

References

View Abstract