ORIGINAL ARTICLE
The Journal of Nursing Research h VOL. 23, NO. 1, MARCH 2015
Development and Validation of the Inventory of Perceptions of Medication Administration Errors for Nurses in Taiwan Chia-Chan Kao1 & Yu-Hua Lin2* & I Lee3 & Fan-Ko Sun2 Tzu-Chun Chang4 & Hsiu-Ping Li4 1
PhD, RN, Associate Professor, Department of Healthcare Administration, I-Shou University & 2PhD, RN, Professor, Department of Nursing, I-Shou University & 3PhD, RN, Assistant Professor, Department of Nursing, I-Shou University & 4 MSN, RN, Lecturer, Department of Nursing, I-Shou University.
ABSTRACT Background: Medication administration errors (MAEs) account for most medication errors, which not only threaten the safety of patients and increase hospital medical costs but also damage the personal and professional development of affected nurses. A feasible instrument measures the perceptions of committing an MAE that may provide support for nurses. Purpose: The purpose of this study was to conduct psychometric testing of the Inventory of Perceptions for Medication Administration Errors (IPMAE). Methods: Psychometric testing of the IPMAE used snowball sampling to collect data from nurse volunteers. Six hundred eighteen nurses completed the IPMAE and a personal profile. The construct validity and the Cronbach’s alpha of the inventory were tested. Results: The results indicated that the 12-item IPMAE consisted of four factors, including coping strategy, emotional reaction, fear of blame, and segregation behavior. These four factors accounted for 78.8% of the total variance. The IPMAE showed that both fit indices and Cronbach’s alpha coefficients (overall = .90 and subscales = .83Y.88) were acceptable. Conclusions/Implications for Practice: The IPMAE is a valid and reliable instrument for measuring perceptions of MAE occurring among nurses. Nursing directors may use the results generated by the IPMAE to help reduce the negative consequences of MAE events among nurses.
KEY WORDS: psychometric properties, medication administration errors, nurses.
Introduction Medication administration errors (MAEs) refer to ‘‘medication errors that occur during the process of administering a drug’’ (Lin & Ma, 2009) or ‘‘any deviation from administration
procedures, policies and/or practices for medication administration’’ (Drach-Zahavy & Pud, 2010). Errors in administration account for most medication errors (Haw, Dickens, & Stubbs, 2005), threaten patient safety, and increase hospital costs (Kohn, Corrigan, & Donaldson, 1999; Taiwan Joint Commission on Hospital Accreditation [TJCHA], 2013). Many studies have focused on preventing MAE and understanding the barriers to MAE reporting to create a safer environment for patients (Etchegaray & Throckmorton, 2009; Lin & Ma, 2009). Factors found to affect MAE prevention include the quality of medication administration procedures (O’Connell, Hawkins, & Ockerby, 2013), the willingness of hospital administration to identify risk factors (Lawton, Carruthers, Gardner, Wright, & McEachan, 2012; Teunissen, Bos, Pot, Piuim, & Kramers, 2013), and the organizational climate (Drach-Zahavy & Pud, 2010; Sheu, Wei, Chen, Yu, & Tang, 2009). However, few studies have examined the reaction of nurses who have committed an MAE. Schelbred and Nord (2007) used in-depth interviews to articulate three types of nurse reactions after committing an MAE. These included immediate reactions (e.g., shock or dread); emotional responses (e.g., shame or a posttraumatic stress episode); and changed relationships with patients, family, or coworkers. Adapting to an MAE incident may have a negative impact on the affected nurse’s personal and professional development (Crigger & Meek, 2007). Accepted for publication: February 27, 2014 *Address correspondence to: Yu-Hua Lin, No. 8, Yida Rd., Yanchao District, Kaohsiung City 82445, Taiwan, ROC. Tel: +886 (7) 615-1100 ext. 7719; Fax: +886 (7) 615-1100; E-mail: [email protected] Cite this article as: Kao, C. C., Lin, Y. H., Lee, I, Sun, F. K., Chang, T. C., & Li, H. P. (2015). Development and validation of the inventory of perceptions of medication administration errors for nurses in taiwan. The Journal of Nursing Research, 23(1), 41Y46. doi:10.1097/jnr.0000000000000048
41 Copyright © 2015 Taiwan Nurses Association. Unauthorized reproduction of this article is prohibited.
The Journal of Nursing Research
Crigger and Meek used a qualitative approach based on the theory of self-reconciliation to describe the psychological process to study individual MAE cases. This theory consists of four consecutive stages: realization (self-acknowledgement of the mistake), consideration (assessing the necessity of mistake disclosure), action (following the best response trajectory), and resolution (getting over the event and moving on). MAE incidence rates in Taiwan between 2005 and 2012 ranged between 2.9% and 16.6%, with an average of 10% (TJCHA, 2013). The Taiwan Patient Safety Reporting System, Taiwan’s clearinghouse for official MAE statistics, identified 1,760 MAE events at 66 hospitals in 2005 and 59,745 events at 698 hospitals in 2012 (TJCHA, 2013). However, because nurses are reluctant to report MAE because of administrative obstacles and fears of blame or lawsuits (Lin & Ma, 2009), these statistics may be significantly lower than actual MAE incident numbers. A survey instrument that measures nurse reactions after an MAE may allow hospitals to not only identify the key obstacles to MAE reporting but also develop support programs to facilitate the nurse’s adjustment and return to duty. To our knowledge, no instrument was available to measure nurse perceptions of MAE. Therefore, the purpose of this study was to describe the development and psychometric testing of the Inventory of Perceptions of Medication Administration Errors (IPMAE) for nurses in Taiwan.
Methods The study was conducted in two stages: (a) item generation and content validity and (b) testing of psychometric properties (exploratory factor analysis [EFA], confirmatory factor analysis [CFA], and internal consistency reliability).
Stage 1: Item Generation and Content Validity The institutional review board of the authors’ school approved this study. After obtaining the written inform consent, the researchers asked four semistructured questions to 25 clinical nurses who had committed an MAE in hospitals. Their responses were used to derive 18 inventory items. All of the clinical nurses were women with a master’s degree. Ages ranged from 28 to 42 years (mean = 35 years), and length of work experience ranged from 8 to 16 years (mean = 14 years). The results of two previous qualitative studies (Crigger & Meek, 2007; Schelbred & Nord, 2007) provided the theoretical-based framework that guided the item generation process. The four questions were as follows: (a) How would you feel if or how do you feel when MAEs occur at work? (b) How do you handle an MAE event? (c) What is your primary concern after an MAE event? (d) What type of stress might or do you perceive after an MAE event? Nine experts were invited to examine the content validity of the resulting 18-item IPMAE. Two of the experts held a PhD degree in nursing and had 9 years of clinical practice
Chia-Chan Kao et al.
experience. The other seven experts were currently practicing clinical nurses with at least a university degree and 16 years of professional experience. All of the experts were asked to rate each item based on relevance, clarity, simplicity, and ambiguity on a 4-point scale (Lynn, 1986). The degree of agreement was defined as the sum score of content validity index for each item as divided by the number of items (Polit & Beck, 2006). These nine experts achieved 99.6% agreement with the 18-item scale, well above the .78 recommended as the minimum acceptable standard for the Content Validity Index (Lynn, 1986; Polit & Beck, 2006). Therefore, all 18 items were preserved and distributed to the participants. Each item was scored using an 8-point Likert scale (1 = never, 8 = always), with higher scores indicating higher level of negative perceptions of MAE.
Stage 2: Testing of Psychometric Properties Participants This study used a cross-sectional, snowball sampling method (Sheu et al., 2009), and 618 participants at several different hospitals completed the survey. Eligibility requirements were being a licensed nurse, a minimum of 3 months of hospital work experience, and willingness to participate. Study procedures Clinical practice teachers distributed the 18-item IPMAE and a demographic information to the participants. The participants, all volunteers, were then informed of the purpose, context, and procedures of the study. No identifiers were requested; all results were presented anonymously. The participants returned the questionnaires to the fourth and fifth authors directly in sealed envelopes. Data analysis Items were confirmed using content validity (LoBiondoWood & Haber, 2002). EFA, CFA, and internal consistency reliability were used to examine construct validity. To enhance measurement precision, all items that returned to either a ceiling or floor score or an itemYitem/itemYtotal correlation of greater than .70 or less than .40 were removed from the IPMAE (Stevens, 2009). EFA using principal axis factoring with an oblique promax rotation was used to assess the factor structure of the IPMAE. Initial eigenvalues were greater than 1, indicating an appropriate number of factors (Costello & Osborne, 2005). Items were retained if they loaded on any factor above .50, cross-loaded on different factors with greater than .10 differences, or indicated a meaningful interpretation (Stevens, 2009). A two-stage CFA, the first- and second-order confirmatory factor models, was performed to confirm the structure of the subscales produced by EFA. Items significantly loaded onto the latent variable; usually, the factor loading greater than .50 and acceptable goodness-of-fit index (GFI) indicated that the model fit the hypothesized model (Kline, 2005). A normed fit index, GFI, comparative fit index, and Tucker-Lewis index of
42 Copyright © 2015 Taiwan Nurses Association. Unauthorized reproduction of this article is prohibited.
Psychometric Testing of the IPMAE
greater than .90 indicate a good model fit (Kline, 2005). The sample data were randomly divided into two groups: one group for EFA and the other group for CFA (Costello & Osborne, 2005). Internal consistency reliability was confirmed by Cronbach’s alpha coefficients for the overall scale and each of the four subscales. An acceptable internal reliability was defined as an alpha value of greater than .70 (Nunnally & Bernstein, 1994). IBM SPSS Amos software version 18 (IBM, Armonk, NY, USA) was used to perform analyses.
Results Characteristics of Participants All participants were women (n = 618), with an average age of 29.88 years (SD = 7.49 years, range = 20Y57years), and had an average number of 7.91 years of clinical practice experience (SD = 6.94 years, range = 0Y35 years). Most had a university education or better (n = 460, 74.4%). Moreover, 293 of 593 (49%, missing responses = 25) reported having committed MAEs.
Exploratory Factor Analysis Because two items showed ceiling scores and three items had high itemYitem or itemYtotal relationships (r 9 .70), only 13 of the 18 items were retained for further EFA analysis. The KaiserYMeyerYOlkin was greater than .60 (.86), and Bartlett’s
VOL. 23, NO. 1, MARCH 2015
test of sphericity was significant (2 2 = 2123.246, df = 66, p G .001), indicating adequate sampling and an appropriate correlation matrix for the EFA (Stevens, 2009). One item was deleted from the EFA because of a factor loading of less than .50 (Stevens, 2009). The final solution yielded a construct with four factors: coping strategy, fear of blame, emotional reaction, and segregation behavior. Coping strategy (three items) represented psychological responses; fear of blame (three items) represented anxiety because of the rebuke of colleagues, supervisors, or patients/family members; emotional reaction (three items) represented psychological reactions such as anxiety, nonacceptance, and fear; and segregation behavior (three items) represented the personal response taken because of committing an MAE (i.e., self-isolation, more alert while administrating medication, or quit the professionals; Table 1). The four subscales accounted for 42.68%, 11.87%, 7.95%, and 6.30% of the variance, respectively.
Confirmatory Factor Analysis In the first-order confirmatory factor model, all 12 items significantly statistically loaded onto the subscales (all factor loadings 9 .50, 95% confidence interval [.59, .89], p G .001). The four subscales related to the latent variable in a statistically significant manner (p G .001). In addition, the acceptable fit index indicated that the first-order hypothesized model fit the data well. The correlations among the four latent constructs ranged from .39 to .51 (Table 2). In the secondorder confirmatory factor model (Figure 1), the acceptable
TABLE 1.
Means, SDs, and Pattern Factor Loadings of the IPMAE Mean
SD
Factor Loading
Factor 1: Coping strategy 1 ‘‘Catharsis’’ would be a way to comfort myself if (when) medication errors occur at work. 2 ‘‘Someone taking care of me’’ would be a way to comfort myself if (when) medication errors occur at work. 3 ‘‘Monitoring clients’ conditions’’ would be a way to comfort myself if (when) medication errors occur at work.
5.58 5.99
1.93 1.97
.85
5.56
2.16
.77
6.83
1.59
.67
Factor 2: Fear of blame 4 Organizational disciplinary action 5 Clients’ and families’ blame 6 Managers’ reprimand
5.82 5.48 6.23 5.74
1.94 2.24 2.15 2.08
.88 .71 .66
Factor 3: Emotional reaction 7 I would feel ‘‘anxiety’’ if (when) medication errors occur at work. 8 I would feel ‘‘unacceptable’’ if (when) medication errors occur at work. 9 I would feel ‘‘frightened’’ if (when) medication errors occur at work.
5.84 6.24 5.17 6.11
1.82 1.89 2.34 1.99
.84 .82 .69
Factor 4: Segregation behavior 10 Self-isolation from colleagues 11 More alert about drugs 12 Quitting the profession
4.50 3.94 4.64 4.95
2.20 2.60 2.57 2.67
.94 .82 .52
Item
" .88
.86
.83
.83
Note. IPMAE = Inventory of Perceptions for Medication Administration Errors.
43 Copyright © 2015 Taiwan Nurses Association. Unauthorized reproduction of this article is prohibited.
The Journal of Nursing Research
Chia-Chan Kao et al.
TABLE 2.
TABLE 3.
Correlations Among the Four Subscales
Goodness-of-Fit Indices of CFA Models
Coping Strategy (1)
Fear of Blame (2)
Emotional Reaction (3)
Segregation Behavior (4)
1
.48*
.39*
.47*
.49*
.45*
(1) (2) (3)
1
1
(4)
.51* 1
*p G .001.
fit index indicated that the second-order hypothesized model fit the data well (Table 3). The four latent constructs contributed to the higher-order construct (all at a significant level, p G .001). The correlation among each of the four latent constructs and the higher-order construct ranged from .61 to .79.
Internal Consistency Reliability The Cronbach’s alpha coefficients were .90 for all items combined, .88 for the first factor, .86 for the second factor, and .83 for the third and fourth factors, indicating satisfactory internal consistency (Nunnally & Bernstein, 1994).
Discussion This study developed and evaluated the psychometric properties of the IPMAE to assess participants’ perceptions of an MAE occurrence. An EFA distributed the 12 items of this inventory into the four subscales of coping strategy, fear of blame, emotional reaction, and segregation behavior. CFA validated the four-factor structure of the IPMAE. Excellent
CFA Model
CFI
GFI
NFI
TLI
First order
.94
.91
.92
.92
Second order
.94
.92
.92
.92
Note. CFA = confirmatory factor analysis; CFI = comparative fit index; GFI = goodness-of-fit index; NFI = normed fit index; TLI = Tucker-Lewis index.
Cronbach’s alpha coefficients for the four subscales (9.80) indicated good internal consistency for the newly developed instrument (Nunnally & Bernstein, 1994). CFA and EFA results showed that the 12-item IPMAE attained a level of predictive power similar to the total score of the 18-item IPMAE (r = .94), supporting the concurrent validity of the instrument (LoBiondo-Wood & Haber, 2002). Items in the IPMAE may be calculated as either the total sum of all four subscales or as each subscale individually because of the similar GFI between the first- and second-order confirmatory factor models (Noar, 2003). Our results are congruent with those of Schelbred and Nord (2007) and Crigger and Meek (2007). However, whereas those two studies identified a distinct series of psychological responses, this study quantitatively expressed nurses’ perceptions after committing an MAE. The ‘‘coping strategy’’ subscale explained the manifest proportion of variance, indicating that ‘‘catharsis,’’ ‘‘others taking care of me,’’ and ‘‘monitoring clients’ conditions’’ may facilitate post-MAE emotional healing. Interactions among external stressors (e.g., the progress of the patient’s condition, whether to report the MAE event, and concerns about being reprimanded) and intrinsic pressures (e.g., self-condemnation and degraded professional confidence) define the immediate response to an MAE (Crigger & Meek, 2007;
Figure 1. A second-order confirmatory factor model of the Inventory of Perceptions for Medication Administration Errors. MAEs = medication administration errors.
44 Copyright © 2015 Taiwan Nurses Association. Unauthorized reproduction of this article is prohibited.
Psychometric Testing of the IPMAE
Schelbred & Nord, 2007). Therefore, the highest scores of participants were in the fear of blame and emotional reaction subscales. There is the risk that nurses who commit an MAE who do not receive appropriate support may face unresolved negative emotions that undermine work confidence over the long term, with the potential for resignation from the job (Crigger & Meek, 2007; Schelbred & Nord, 2007). This helps explain why segregation behavior (self-isolation, more alert about drugs, and leaving the profession) is the IPMAE subscale with the lowest variance and the lowest score.
Limitations This inventory was developed specifically for clinical nurses. Therefore, applicability to other health-related professionals is limited. The four subscales address a wide spectrum of assessment aspects that may help characterize MAE outcomes among nursing professionals. Continued work to explore the efficacy of using the four IPMAE subscales to assess health status, burnout, and professional commitment is recommended.
Implications for Practice Several factors were found to associate positively with relieving the effects of negative perceptions of MAE, including immediate involvement in patient treatment and communication with family members, receiving support rather than punishment, and receiving counseling. Continued periodic checks of these effects are recommended to prevent nurses leaving the profession because of MAE-related posttraumatic stress.
Conclusion Construct validity and internal consistency support the IPMAE as a valid and reliable instrument for measuring the perceptions of nurses toward committing MAEs. Nursing directors may apply the IPMAE to nurses involved in MAE events.
References Costello, A. B., & Osborne, J. W. (2005). Best practice in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical Assessment, Research & Evaluation, 10(7), 1Y9. Retrieved from http://pareonline.net/getvn.asp?v=10&n=7
VOL. 23, NO. 1, MARCH 2015
Haw, C. M., Dickens, G. D., & Stubbs, J. (2005). A review of medication administration errors reported in a large psychiatric hospital in the United Kingdom. Psychiatric Services, 56(12), 1610Y1613. doi:10.1176/appi.ps.56.12.1610 Kline, R. X. (2005). Principles and practices of structural equation modeling. New York, NY: Guilford Press. doi:10.1080/10705511 .2012.687667 Kohn, L. T., Corrigan, J. M., & Donaldson, M. S. (1999). To err is human: Building a safer health system. Washington, DC: Institute of Medicine of the National Academy of Sciences. Lawton, R., Carruthers, S., Gardner, P., Wright, J., & McEachan, R. R. C. (2012). Identifying the latent failures underpinning medication administration errors: An exploratory study. Health Services Research, 47(4), 1437Y1459. doi:10.1111/j.1475-6773 .2012.01390.x Lin, Y. H., & Ma, S. M. (2009). Willingness to nurses to report medication administration errors in southern Taiwan: A crosssectional survey. Worldviews on Evidence-Based Nursing, 6(4), 237Y245. doi:10.1111/j.1741-6787.2009.00169.x LoBiondo-Wood, G., & Haber, J. (2002). Nursing research: Methods critical appraisal for evidence-based practice. St. Louis, MO: Mosby. Lynn, M. R. (1986). Determination and quantification of content validity. Nursing Research, 35(6), 382Y386. Noar, S. M. (2003). The role of structural equation modeling in scale development. Structural Equation Modeling, 10(4), 622Y647. doi:10.1207/S15328007SEM1004_8 Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.). New York, NY: McGraw-Hill. O’Connell, B., Hawkins, M., & Ockerby, C. (2013). Construct validity and reliability of the single checking administration of medication scale. International Journal of Nursing Practice, 19(3), 281Y287. doi:10.1111/ijn.12076 Polit, D. F., & Beck, C. T. (2006). The content validity index: Are you sure you know what’s being reported? Critique and recommendations. Research in Nursing and Health, 29(5), 489Y497. doi:10 .1002/nur.20147 Schelbred, A.-B., & Nord, R. (2007). Nurses’ experiences of drug administration errors. Journal of Advanced Nursing, 60(3), 317Y324. doi:10.1111/j.1365-2648.2007.04437.x Sheu, S. J., Wei, I. L., Chen, C. H., Yu, S., & Tang, F. I. (2009). Using snowball sampling method with nurses to understand medication administration errors. Journal of Clinical Nursing, 18(4), 559Y569. doi:10.1111/j.1365-2702.2007.02048.x
Crigger, N. J., & Meek, V. L. (2007). Toward a theory of self-reconciliation following mistakes in nursing practice. Journal of Nursing Scholarship, 39(2), 177Y183. doi:10.1111/j.1547-5069.2007.00164.x
Stevens, J. P. (2009). Applied multivariate statistics for social science (5th ed.). Mahwah, NJ: Lawrence Erlbaum Association.
Drach-Zahavy, A., & Pud, D. (2010). Learning mechanisms to limit medication administration errors. Journal of Advanced Nursing, 66(4), 794Y805. doi:10.1111/j.1365-2648.2010.05294.x
Taiwan Joint Commission on Hospital Accreditation. (2013). Patient safety in Taiwan. Retrieved from http://www.patientsafety.mohw.gov .tw/big5/Content/Content.asp?cid=4
Etchegaray, J. M., & Throckmorton, T. (2009). Barriers to reporting medication errors: A measurement equivalence perspective. Quality and Safety in Health Care, 19(6), 1Y4. doi: 10.1136/qshc.2008.031534
Teunissen, R., Bos, J., Pot, H., Piuim, M., & Kramers, C. (2013). Clinical relevance of and risk factors associated with medication administration time errors. American Journal of HealthSystem Pharmacy, 70(12), 1052Y1056. doi:10.2146/ajhp120247
45 Copyright © 2015 Taiwan Nurses Association. Unauthorized reproduction of this article is prohibited.
The Journal of Nursing Research
給藥錯誤感受量表心理計量
VOL. 23, NO. 1, MARCH 2015
護理人員給藥錯誤感受量表的發展與驗證 高家常1 林佑樺2* 李逸3 孫凡軻2 張智鈞4 李秀萍4 1
義守大學醫務管理學系副教授 2義守大學護理學系教授 3 義守大學護理學系助理教授 4義守大學護理學系講師
背 景
給藥錯誤除威脅病人安全與增加醫院成本外 ，也影響護理人員個人與專業的發展。經 由測量知覺給藥錯誤 ，有助於提供護理人員具體可行的協助。
目 的
發展護理人員給藥錯誤感受量表（Inventory of Perceptions for Medication Administration Errors, IPMAE）並檢測其信 、效度。
方 法
採用滾雪球抽樣方法 ，參與者為自願參加且工作達 3 個月以上的臨床護理人員 ，共有 618 位完成個人基本資料與 IPMAE 的填寫。採用探索性及驗證性因素分析評估 IPMAE 的建構效度 ，並以內在一致性信度檢測量表的信度。
結 果
IPMAE 包括 12 題與四個因素 ，四個因素為：因應策略 、情緒反應 、害怕受責備和隔離 行為 ，總解釋變異量為78.80%。總量表Cronbach’s α為.90及各次量表α介於.83至.88間 ， 顯示此量表具有良好的信度。
結 論 IPMAE 是一個有效和可靠的測量工具 ，用來了解護理人員給藥錯誤後的感受。依據 實務應用 IPMAE 結果的指引 ，可提供護理主管並減少護理人員因給藥錯誤所造成的負面結果。
關鍵詞：心理測量、給藥錯誤、護理人員。
接受刊載：103年2月27日 *通訊作者地址：林佑樺 82445高雄市燕巢區義大路8號 電話：（07）6151100－7719 E-mail: [email protected]
46 Copyright © 2015 Taiwan Nurses Association. Unauthorized reproduction of this article is prohibited.
The Journal of Nursing Research h VOL. 23, NO. 1, MARCH 2015
Development and Validation of the Inventory of Perceptions of Medication Administration Errors for Nurses in Taiwan Chia-Chan Kao1 & Yu-Hua Lin2* & I Lee3 & Fan-Ko Sun2 Tzu-Chun Chang4 & Hsiu-Ping Li4 1
PhD, RN, Associate Professor, Department of Healthcare Administration, I-Shou University & 2PhD, RN, Professor, Department of Nursing, I-Shou University & 3PhD, RN, Assistant Professor, Department of Nursing, I-Shou University & 4 MSN, RN, Lecturer, Department of Nursing, I-Shou University.
ABSTRACT Background: Medication administration errors (MAEs) account for most medication errors, which not only threaten the safety of patients and increase hospital medical costs but also damage the personal and professional development of affected nurses. A feasible instrument measures the perceptions of committing an MAE that may provide support for nurses. Purpose: The purpose of this study was to conduct psychometric testing of the Inventory of Perceptions for Medication Administration Errors (IPMAE). Methods: Psychometric testing of the IPMAE used snowball sampling to collect data from nurse volunteers. Six hundred eighteen nurses completed the IPMAE and a personal profile. The construct validity and the Cronbach’s alpha of the inventory were tested. Results: The results indicated that the 12-item IPMAE consisted of four factors, including coping strategy, emotional reaction, fear of blame, and segregation behavior. These four factors accounted for 78.8% of the total variance. The IPMAE showed that both fit indices and Cronbach’s alpha coefficients (overall = .90 and subscales = .83Y.88) were acceptable. Conclusions/Implications for Practice: The IPMAE is a valid and reliable instrument for measuring perceptions of MAE occurring among nurses. Nursing directors may use the results generated by the IPMAE to help reduce the negative consequences of MAE events among nurses.
KEY WORDS: psychometric properties, medication administration errors, nurses.
Introduction Medication administration errors (MAEs) refer to ‘‘medication errors that occur during the process of administering a drug’’ (Lin & Ma, 2009) or ‘‘any deviation from administration
procedures, policies and/or practices for medication administration’’ (Drach-Zahavy & Pud, 2010). Errors in administration account for most medication errors (Haw, Dickens, & Stubbs, 2005), threaten patient safety, and increase hospital costs (Kohn, Corrigan, & Donaldson, 1999; Taiwan Joint Commission on Hospital Accreditation [TJCHA], 2013). Many studies have focused on preventing MAE and understanding the barriers to MAE reporting to create a safer environment for patients (Etchegaray & Throckmorton, 2009; Lin & Ma, 2009). Factors found to affect MAE prevention include the quality of medication administration procedures (O’Connell, Hawkins, & Ockerby, 2013), the willingness of hospital administration to identify risk factors (Lawton, Carruthers, Gardner, Wright, & McEachan, 2012; Teunissen, Bos, Pot, Piuim, & Kramers, 2013), and the organizational climate (Drach-Zahavy & Pud, 2010; Sheu, Wei, Chen, Yu, & Tang, 2009). However, few studies have examined the reaction of nurses who have committed an MAE. Schelbred and Nord (2007) used in-depth interviews to articulate three types of nurse reactions after committing an MAE. These included immediate reactions (e.g., shock or dread); emotional responses (e.g., shame or a posttraumatic stress episode); and changed relationships with patients, family, or coworkers. Adapting to an MAE incident may have a negative impact on the affected nurse’s personal and professional development (Crigger & Meek, 2007). Accepted for publication: February 27, 2014 *Address correspondence to: Yu-Hua Lin, No. 8, Yida Rd., Yanchao District, Kaohsiung City 82445, Taiwan, ROC. Tel: +886 (7) 615-1100 ext. 7719; Fax: +886 (7) 615-1100; E-mail: [email protected] Cite this article as: Kao, C. C., Lin, Y. H., Lee, I, Sun, F. K., Chang, T. C., & Li, H. P. (2015). Development and validation of the inventory of perceptions of medication administration errors for nurses in taiwan. The Journal of Nursing Research, 23(1), 41Y46. doi:10.1097/jnr.0000000000000048
41 Copyright © 2015 Taiwan Nurses Association. Unauthorized reproduction of this article is prohibited.
The Journal of Nursing Research
Crigger and Meek used a qualitative approach based on the theory of self-reconciliation to describe the psychological process to study individual MAE cases. This theory consists of four consecutive stages: realization (self-acknowledgement of the mistake), consideration (assessing the necessity of mistake disclosure), action (following the best response trajectory), and resolution (getting over the event and moving on). MAE incidence rates in Taiwan between 2005 and 2012 ranged between 2.9% and 16.6%, with an average of 10% (TJCHA, 2013). The Taiwan Patient Safety Reporting System, Taiwan’s clearinghouse for official MAE statistics, identified 1,760 MAE events at 66 hospitals in 2005 and 59,745 events at 698 hospitals in 2012 (TJCHA, 2013). However, because nurses are reluctant to report MAE because of administrative obstacles and fears of blame or lawsuits (Lin & Ma, 2009), these statistics may be significantly lower than actual MAE incident numbers. A survey instrument that measures nurse reactions after an MAE may allow hospitals to not only identify the key obstacles to MAE reporting but also develop support programs to facilitate the nurse’s adjustment and return to duty. To our knowledge, no instrument was available to measure nurse perceptions of MAE. Therefore, the purpose of this study was to describe the development and psychometric testing of the Inventory of Perceptions of Medication Administration Errors (IPMAE) for nurses in Taiwan.
Methods The study was conducted in two stages: (a) item generation and content validity and (b) testing of psychometric properties (exploratory factor analysis [EFA], confirmatory factor analysis [CFA], and internal consistency reliability).
Stage 1: Item Generation and Content Validity The institutional review board of the authors’ school approved this study. After obtaining the written inform consent, the researchers asked four semistructured questions to 25 clinical nurses who had committed an MAE in hospitals. Their responses were used to derive 18 inventory items. All of the clinical nurses were women with a master’s degree. Ages ranged from 28 to 42 years (mean = 35 years), and length of work experience ranged from 8 to 16 years (mean = 14 years). The results of two previous qualitative studies (Crigger & Meek, 2007; Schelbred & Nord, 2007) provided the theoretical-based framework that guided the item generation process. The four questions were as follows: (a) How would you feel if or how do you feel when MAEs occur at work? (b) How do you handle an MAE event? (c) What is your primary concern after an MAE event? (d) What type of stress might or do you perceive after an MAE event? Nine experts were invited to examine the content validity of the resulting 18-item IPMAE. Two of the experts held a PhD degree in nursing and had 9 years of clinical practice
Chia-Chan Kao et al.
experience. The other seven experts were currently practicing clinical nurses with at least a university degree and 16 years of professional experience. All of the experts were asked to rate each item based on relevance, clarity, simplicity, and ambiguity on a 4-point scale (Lynn, 1986). The degree of agreement was defined as the sum score of content validity index for each item as divided by the number of items (Polit & Beck, 2006). These nine experts achieved 99.6% agreement with the 18-item scale, well above the .78 recommended as the minimum acceptable standard for the Content Validity Index (Lynn, 1986; Polit & Beck, 2006). Therefore, all 18 items were preserved and distributed to the participants. Each item was scored using an 8-point Likert scale (1 = never, 8 = always), with higher scores indicating higher level of negative perceptions of MAE.
Stage 2: Testing of Psychometric Properties Participants This study used a cross-sectional, snowball sampling method (Sheu et al., 2009), and 618 participants at several different hospitals completed the survey. Eligibility requirements were being a licensed nurse, a minimum of 3 months of hospital work experience, and willingness to participate. Study procedures Clinical practice teachers distributed the 18-item IPMAE and a demographic information to the participants. The participants, all volunteers, were then informed of the purpose, context, and procedures of the study. No identifiers were requested; all results were presented anonymously. The participants returned the questionnaires to the fourth and fifth authors directly in sealed envelopes. Data analysis Items were confirmed using content validity (LoBiondoWood & Haber, 2002). EFA, CFA, and internal consistency reliability were used to examine construct validity. To enhance measurement precision, all items that returned to either a ceiling or floor score or an itemYitem/itemYtotal correlation of greater than .70 or less than .40 were removed from the IPMAE (Stevens, 2009). EFA using principal axis factoring with an oblique promax rotation was used to assess the factor structure of the IPMAE. Initial eigenvalues were greater than 1, indicating an appropriate number of factors (Costello & Osborne, 2005). Items were retained if they loaded on any factor above .50, cross-loaded on different factors with greater than .10 differences, or indicated a meaningful interpretation (Stevens, 2009). A two-stage CFA, the first- and second-order confirmatory factor models, was performed to confirm the structure of the subscales produced by EFA. Items significantly loaded onto the latent variable; usually, the factor loading greater than .50 and acceptable goodness-of-fit index (GFI) indicated that the model fit the hypothesized model (Kline, 2005). A normed fit index, GFI, comparative fit index, and Tucker-Lewis index of
42 Copyright © 2015 Taiwan Nurses Association. Unauthorized reproduction of this article is prohibited.
Psychometric Testing of the IPMAE
greater than .90 indicate a good model fit (Kline, 2005). The sample data were randomly divided into two groups: one group for EFA and the other group for CFA (Costello & Osborne, 2005). Internal consistency reliability was confirmed by Cronbach’s alpha coefficients for the overall scale and each of the four subscales. An acceptable internal reliability was defined as an alpha value of greater than .70 (Nunnally & Bernstein, 1994). IBM SPSS Amos software version 18 (IBM, Armonk, NY, USA) was used to perform analyses.
Results Characteristics of Participants All participants were women (n = 618), with an average age of 29.88 years (SD = 7.49 years, range = 20Y57years), and had an average number of 7.91 years of clinical practice experience (SD = 6.94 years, range = 0Y35 years). Most had a university education or better (n = 460, 74.4%). Moreover, 293 of 593 (49%, missing responses = 25) reported having committed MAEs.
Exploratory Factor Analysis Because two items showed ceiling scores and three items had high itemYitem or itemYtotal relationships (r 9 .70), only 13 of the 18 items were retained for further EFA analysis. The KaiserYMeyerYOlkin was greater than .60 (.86), and Bartlett’s
VOL. 23, NO. 1, MARCH 2015
test of sphericity was significant (2 2 = 2123.246, df = 66, p G .001), indicating adequate sampling and an appropriate correlation matrix for the EFA (Stevens, 2009). One item was deleted from the EFA because of a factor loading of less than .50 (Stevens, 2009). The final solution yielded a construct with four factors: coping strategy, fear of blame, emotional reaction, and segregation behavior. Coping strategy (three items) represented psychological responses; fear of blame (three items) represented anxiety because of the rebuke of colleagues, supervisors, or patients/family members; emotional reaction (three items) represented psychological reactions such as anxiety, nonacceptance, and fear; and segregation behavior (three items) represented the personal response taken because of committing an MAE (i.e., self-isolation, more alert while administrating medication, or quit the professionals; Table 1). The four subscales accounted for 42.68%, 11.87%, 7.95%, and 6.30% of the variance, respectively.
Confirmatory Factor Analysis In the first-order confirmatory factor model, all 12 items significantly statistically loaded onto the subscales (all factor loadings 9 .50, 95% confidence interval [.59, .89], p G .001). The four subscales related to the latent variable in a statistically significant manner (p G .001). In addition, the acceptable fit index indicated that the first-order hypothesized model fit the data well. The correlations among the four latent constructs ranged from .39 to .51 (Table 2). In the secondorder confirmatory factor model (Figure 1), the acceptable
TABLE 1.
Means, SDs, and Pattern Factor Loadings of the IPMAE Mean
SD
Factor Loading
Factor 1: Coping strategy 1 ‘‘Catharsis’’ would be a way to comfort myself if (when) medication errors occur at work. 2 ‘‘Someone taking care of me’’ would be a way to comfort myself if (when) medication errors occur at work. 3 ‘‘Monitoring clients’ conditions’’ would be a way to comfort myself if (when) medication errors occur at work.
5.58 5.99
1.93 1.97
.85
5.56
2.16
.77
6.83
1.59
.67
Factor 2: Fear of blame 4 Organizational disciplinary action 5 Clients’ and families’ blame 6 Managers’ reprimand
5.82 5.48 6.23 5.74
1.94 2.24 2.15 2.08
.88 .71 .66
Factor 3: Emotional reaction 7 I would feel ‘‘anxiety’’ if (when) medication errors occur at work. 8 I would feel ‘‘unacceptable’’ if (when) medication errors occur at work. 9 I would feel ‘‘frightened’’ if (when) medication errors occur at work.
5.84 6.24 5.17 6.11
1.82 1.89 2.34 1.99
.84 .82 .69
Factor 4: Segregation behavior 10 Self-isolation from colleagues 11 More alert about drugs 12 Quitting the profession
4.50 3.94 4.64 4.95
2.20 2.60 2.57 2.67
.94 .82 .52
Item
" .88
.86
.83
.83
Note. IPMAE = Inventory of Perceptions for Medication Administration Errors.
43 Copyright © 2015 Taiwan Nurses Association. Unauthorized reproduction of this article is prohibited.
The Journal of Nursing Research
Chia-Chan Kao et al.
TABLE 2.
TABLE 3.
Correlations Among the Four Subscales
Goodness-of-Fit Indices of CFA Models
Coping Strategy (1)
Fear of Blame (2)
Emotional Reaction (3)
Segregation Behavior (4)
1
.48*
.39*
.47*
.49*
.45*
(1) (2) (3)
1
1
(4)
.51* 1
*p G .001.
fit index indicated that the second-order hypothesized model fit the data well (Table 3). The four latent constructs contributed to the higher-order construct (all at a significant level, p G .001). The correlation among each of the four latent constructs and the higher-order construct ranged from .61 to .79.
Internal Consistency Reliability The Cronbach’s alpha coefficients were .90 for all items combined, .88 for the first factor, .86 for the second factor, and .83 for the third and fourth factors, indicating satisfactory internal consistency (Nunnally & Bernstein, 1994).
Discussion This study developed and evaluated the psychometric properties of the IPMAE to assess participants’ perceptions of an MAE occurrence. An EFA distributed the 12 items of this inventory into the four subscales of coping strategy, fear of blame, emotional reaction, and segregation behavior. CFA validated the four-factor structure of the IPMAE. Excellent
CFA Model
CFI
GFI
NFI
TLI
First order
.94
.91
.92
.92
Second order
.94
.92
.92
.92
Note. CFA = confirmatory factor analysis; CFI = comparative fit index; GFI = goodness-of-fit index; NFI = normed fit index; TLI = Tucker-Lewis index.
Cronbach’s alpha coefficients for the four subscales (9.80) indicated good internal consistency for the newly developed instrument (Nunnally & Bernstein, 1994). CFA and EFA results showed that the 12-item IPMAE attained a level of predictive power similar to the total score of the 18-item IPMAE (r = .94), supporting the concurrent validity of the instrument (LoBiondo-Wood & Haber, 2002). Items in the IPMAE may be calculated as either the total sum of all four subscales or as each subscale individually because of the similar GFI between the first- and second-order confirmatory factor models (Noar, 2003). Our results are congruent with those of Schelbred and Nord (2007) and Crigger and Meek (2007). However, whereas those two studies identified a distinct series of psychological responses, this study quantitatively expressed nurses’ perceptions after committing an MAE. The ‘‘coping strategy’’ subscale explained the manifest proportion of variance, indicating that ‘‘catharsis,’’ ‘‘others taking care of me,’’ and ‘‘monitoring clients’ conditions’’ may facilitate post-MAE emotional healing. Interactions among external stressors (e.g., the progress of the patient’s condition, whether to report the MAE event, and concerns about being reprimanded) and intrinsic pressures (e.g., self-condemnation and degraded professional confidence) define the immediate response to an MAE (Crigger & Meek, 2007;
Figure 1. A second-order confirmatory factor model of the Inventory of Perceptions for Medication Administration Errors. MAEs = medication administration errors.
44 Copyright © 2015 Taiwan Nurses Association. Unauthorized reproduction of this article is prohibited.
Psychometric Testing of the IPMAE
Schelbred & Nord, 2007). Therefore, the highest scores of participants were in the fear of blame and emotional reaction subscales. There is the risk that nurses who commit an MAE who do not receive appropriate support may face unresolved negative emotions that undermine work confidence over the long term, with the potential for resignation from the job (Crigger & Meek, 2007; Schelbred & Nord, 2007). This helps explain why segregation behavior (self-isolation, more alert about drugs, and leaving the profession) is the IPMAE subscale with the lowest variance and the lowest score.
Limitations This inventory was developed specifically for clinical nurses. Therefore, applicability to other health-related professionals is limited. The four subscales address a wide spectrum of assessment aspects that may help characterize MAE outcomes among nursing professionals. Continued work to explore the efficacy of using the four IPMAE subscales to assess health status, burnout, and professional commitment is recommended.
Implications for Practice Several factors were found to associate positively with relieving the effects of negative perceptions of MAE, including immediate involvement in patient treatment and communication with family members, receiving support rather than punishment, and receiving counseling. Continued periodic checks of these effects are recommended to prevent nurses leaving the profession because of MAE-related posttraumatic stress.
Conclusion Construct validity and internal consistency support the IPMAE as a valid and reliable instrument for measuring the perceptions of nurses toward committing MAEs. Nursing directors may apply the IPMAE to nurses involved in MAE events.
References Costello, A. B., & Osborne, J. W. (2005). Best practice in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical Assessment, Research & Evaluation, 10(7), 1Y9. Retrieved from http://pareonline.net/getvn.asp?v=10&n=7
VOL. 23, NO. 1, MARCH 2015
Haw, C. M., Dickens, G. D., & Stubbs, J. (2005). A review of medication administration errors reported in a large psychiatric hospital in the United Kingdom. Psychiatric Services, 56(12), 1610Y1613. doi:10.1176/appi.ps.56.12.1610 Kline, R. X. (2005). Principles and practices of structural equation modeling. New York, NY: Guilford Press. doi:10.1080/10705511 .2012.687667 Kohn, L. T., Corrigan, J. M., & Donaldson, M. S. (1999). To err is human: Building a safer health system. Washington, DC: Institute of Medicine of the National Academy of Sciences. Lawton, R., Carruthers, S., Gardner, P., Wright, J., & McEachan, R. R. C. (2012). Identifying the latent failures underpinning medication administration errors: An exploratory study. Health Services Research, 47(4), 1437Y1459. doi:10.1111/j.1475-6773 .2012.01390.x Lin, Y. H., & Ma, S. M. (2009). Willingness to nurses to report medication administration errors in southern Taiwan: A crosssectional survey. Worldviews on Evidence-Based Nursing, 6(4), 237Y245. doi:10.1111/j.1741-6787.2009.00169.x LoBiondo-Wood, G., & Haber, J. (2002). Nursing research: Methods critical appraisal for evidence-based practice. St. Louis, MO: Mosby. Lynn, M. R. (1986). Determination and quantification of content validity. Nursing Research, 35(6), 382Y386. Noar, S. M. (2003). The role of structural equation modeling in scale development. Structural Equation Modeling, 10(4), 622Y647. doi:10.1207/S15328007SEM1004_8 Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.). New York, NY: McGraw-Hill. O’Connell, B., Hawkins, M., & Ockerby, C. (2013). Construct validity and reliability of the single checking administration of medication scale. International Journal of Nursing Practice, 19(3), 281Y287. doi:10.1111/ijn.12076 Polit, D. F., & Beck, C. T. (2006). The content validity index: Are you sure you know what’s being reported? Critique and recommendations. Research in Nursing and Health, 29(5), 489Y497. doi:10 .1002/nur.20147 Schelbred, A.-B., & Nord, R. (2007). Nurses’ experiences of drug administration errors. Journal of Advanced Nursing, 60(3), 317Y324. doi:10.1111/j.1365-2648.2007.04437.x Sheu, S. J., Wei, I. L., Chen, C. H., Yu, S., & Tang, F. I. (2009). Using snowball sampling method with nurses to understand medication administration errors. Journal of Clinical Nursing, 18(4), 559Y569. doi:10.1111/j.1365-2702.2007.02048.x
Crigger, N. J., & Meek, V. L. (2007). Toward a theory of self-reconciliation following mistakes in nursing practice. Journal of Nursing Scholarship, 39(2), 177Y183. doi:10.1111/j.1547-5069.2007.00164.x
Stevens, J. P. (2009). Applied multivariate statistics for social science (5th ed.). Mahwah, NJ: Lawrence Erlbaum Association.
Drach-Zahavy, A., & Pud, D. (2010). Learning mechanisms to limit medication administration errors. Journal of Advanced Nursing, 66(4), 794Y805. doi:10.1111/j.1365-2648.2010.05294.x
Taiwan Joint Commission on Hospital Accreditation. (2013). Patient safety in Taiwan. Retrieved from http://www.patientsafety.mohw.gov .tw/big5/Content/Content.asp?cid=4
Etchegaray, J. M., & Throckmorton, T. (2009). Barriers to reporting medication errors: A measurement equivalence perspective. Quality and Safety in Health Care, 19(6), 1Y4. doi: 10.1136/qshc.2008.031534
Teunissen, R., Bos, J., Pot, H., Piuim, M., & Kramers, C. (2013). Clinical relevance of and risk factors associated with medication administration time errors. American Journal of HealthSystem Pharmacy, 70(12), 1052Y1056. doi:10.2146/ajhp120247
45 Copyright © 2015 Taiwan Nurses Association. Unauthorized reproduction of this article is prohibited.
The Journal of Nursing Research
給藥錯誤感受量表心理計量
VOL. 23, NO. 1, MARCH 2015
護理人員給藥錯誤感受量表的發展與驗證 高家常1 林佑樺2* 李逸3 孫凡軻2 張智鈞4 李秀萍4 1
義守大學醫務管理學系副教授 2義守大學護理學系教授 3 義守大學護理學系助理教授 4義守大學護理學系講師
背 景
給藥錯誤除威脅病人安全與增加醫院成本外 ，也影響護理人員個人與專業的發展。經 由測量知覺給藥錯誤 ，有助於提供護理人員具體可行的協助。
目 的
發展護理人員給藥錯誤感受量表（Inventory of Perceptions for Medication Administration Errors, IPMAE）並檢測其信 、效度。
方 法
採用滾雪球抽樣方法 ，參與者為自願參加且工作達 3 個月以上的臨床護理人員 ，共有 618 位完成個人基本資料與 IPMAE 的填寫。採用探索性及驗證性因素分析評估 IPMAE 的建構效度 ，並以內在一致性信度檢測量表的信度。
結 果
IPMAE 包括 12 題與四個因素 ，四個因素為：因應策略 、情緒反應 、害怕受責備和隔離 行為 ，總解釋變異量為78.80%。總量表Cronbach’s α為.90及各次量表α介於.83至.88間 ， 顯示此量表具有良好的信度。
結 論 IPMAE 是一個有效和可靠的測量工具 ，用來了解護理人員給藥錯誤後的感受。依據 實務應用 IPMAE 結果的指引 ，可提供護理主管並減少護理人員因給藥錯誤所造成的負面結果。
關鍵詞：心理測量、給藥錯誤、護理人員。
接受刊載：103年2月27日 *通訊作者地址：林佑樺 82445高雄市燕巢區義大路8號 電話：（07）6151100－7719 E-mail: [email protected]
46 Copyright © 2015 Taiwan Nurses Association. Unauthorized reproduction of this article is prohibited.
