You are here

Article Text

Article menu

Download PDFPDF

Original research
Validation of a teamwork perceptions measure to increase patient safety
  1. Joseph R Keebler1,
  2. Aaron S Dietz2,3,
  3. Elizabeth H Lazzara1,4,
  4. Lauren E Benishek2,3,
  5. Sandra A Almeida5,
  6. Phyllis A Toor6,
  7. Heidi B King6,
  8. Eduardo Salas2,3
  1. 1Wichita State University, Wichita, Kansas, USA
  2. 2Institute for Simulation and Training, University of Central Florida, Orlando, Florida, USA
  3. 3Department of Psychology, University of Central Florida, Orlando, Florida, USA
  4. 4University of Kansas School of Medicine Wichita, USA
  5. 5Army Patient Safety Program, US Army Medical Command, Fort Sam Houston, Texas, USA
  6. 6US Department of Defense Patient Safety Program, Defense Health Agency , Falls Church, Virginia, USA
  1. Correspondence to Dr Eduardo Salas, Institute for Simulation & Training, University of Central Florida, 3100 Technology Parkway, Orlando, FL 32826, USA; esalas{at}ist.ucf.edu

Abstract

Background TeamSTEPPS (Team Strategies and Tools to Enhance Performance and Patient Safety) is a team-training intervention which shows promise in aiding the mitigation of medical errors. This article examines the construct validity of the TeamSTEPPS Teamwork Perceptions Questionnaire (T-TPQ), a self-report survey that examines multiple dimensions of perceptions of teamwork within healthcare settings.

Method Using survey-based methods, 1700 multidisciplinary healthcare professionals and support staff were measured on their perceptions of teamwork. Confirmatory factor analysis was conducted to examine the relationship between the five TeamSTEPPS dimensions: Leadership, Mutual Support, Situation Monitoring, Communication, and Team Structure.

Results The analysis indicated that the T-TPQ measure is more reliable than previously thought (Cronbach's α=0.978). Further, our final tested model showed a good fit with the data (x2 (df) 3601.27 (546), p<0.0001, Tucker–Lewis Index (TLI)=0.942, Comparative fit index (CFI)=0.947, root mean square error of approximation (RMSEA)=0.057), indicating that the measure appears to have construct validity. Further, all dimensions correlated with one another, but were shown to be independent constructs.

Conclusions The T-TPQ is a construct-valid instrument for measuring perceptions of teamwork. This has beneficial implications for patient safety and future research that studies medical teamwork.

  • Teamwork
  • Surveys
  • Patient safety

https://doi.org/10.1136/bmjqs-2013-001942

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Background

    Despite the growing national recognition of the pivotal role of teamwork for safe, quality healthcare, the educational curricula and medical practices of few health professions integrate team training. In response to this critical gap, the Department of Defense Patient Safety Program (DoD PSP) in collaboration with the Agency for Healthcare Research and Quality (AHRQ) and a national team of subject matter experts launched a multi-year research and development effort to create TeamSTEPPS (Team Strategies and Tools to Enhance Performance and Patient Safety), releasing the programme in 2006 as a publicly available resource and the AHRQ national standard for medical team training.

    Evidence of TeamSTEPPS’ effectiveness across healthcare settings is beginning to accumulate, with research studies showing improvements in team skills such as leadership, situation monitoring, mutual support and communication,1 as well as reductions in medical errors related to communication, medication, needle-stick incidents2 and endotracheal intubation.1 Given that every state has implemented TeamSTEPPS and the Military Health System is one of the largest in the world, employing over 130 000 personnel and serving 9.6 million beneficiaries, TeamSTEPPS will probably sustain and continue to broaden.

    In parallel, the National Strategy for Quality Improvement3 explicated measurement as one of the primary objectives for maximising results of patient safety advancements. The health professions education system has moved toward an outcomes/competency-based system, requiring that training demonstrates effectiveness by measuring performance, including patient safety competencies such as teamwork. Without adequate measurement, there will continue to be inaccurate estimates of medical errors and a lack of significant cultivations in patient safety initiatives.4 Despite the criticality of teamwork, there remains a paucity of validated metrics for team performance.5

    The TeamSTEPPS programme provides a survey-based metric of perceptions of teamwork—the TeamSTEPPS Teamwork Perceptions Questionnaire (T-TPQ). Unquestionably, surveys have inherent drawbacks, such as being susceptible to responder bias; however, evaluating perceptions of patient safety culture and teamwork does have merit. To illustrate, Mardon et al6 found that hospitals with higher patient safety culture survey scores had fewer complications and adverse events. In addition, Manser5 conducted a systematic literature review and found that perceptions of teamwork were related to the quality of patient care. Further, one study suggested that positive teamwork perceptions were related to lower mortality rates.7 When team members believe that the climate exists for open communication, valued opinions and trusted decisions, teams behave more effectively and efficiently.8 ,9 Others suggest that sources of errors include poor staff–staff and provider–patient communication, multiple handoffs, rapid decision-making, staff stress, fatigue and lack of appropriate education and training.10

    Unarguably, performance measurement is foundational for learning and behavioural change, but is often difficult to assess in the medical setting.11 Therefore, the T-TPQ measures an individual's perception of group-level teamwork skills within a medical unit or department.12 It was developed to align with the five core teamwork dimensions on which TeamSTEPPS is based: Team Structure, Leadership, Situation Monitoring, Mutual Support and Communication. These dimensions were synthesised from a comprehensive literature review of the evidence based on teamwork, patient safety and team training13 ,14 and represent the core teamwork competencies that most heavily affect team performance.15 These competencies are relevant to teams performing in dynamic, complex environments where the consequences of errors are high but the occurrence of errors are low.16

    The present paper contributes to the advancements of patient safety, and helps to address a current need by extending beyond previous work that described the initial creation and psychometric testing of the T-TPQ.12 In particular, it details the validation efforts of the T-TPQ.

    Rationale

    While previous efforts have provided initial evidence of the T-TPQ's reliability,12 there are several reasons underscoring the need for additional assessment of both its reliability and construct validity. First, there is insufficient evidence on its construct validity. Therefore, this paper will focus on an empirical approach to aid in establishing this. Throughout the remainder of this article, the term ‘validity’ will specifically refer to ‘construct validity’ unless otherwise noted. Although previous efforts have demonstrated that the T-TPQ is reliable (eg, Cronbach's α for each dimension ranged from 0.88 to 0.95), assessment of the measure's validity was limited in the initial validation effort.12 Thus, it remains unclear how well the T-TPQ measures its five teamwork dimensions. Second, previous validation efforts relied on a relatively small sample size of 169 participants.12 It has been suggested that, at a minimum, validation studies should include 200 participants.17 Third, the validation procedures adopted in previous investigations were limited in their approach to assessing validity. This study uses a more robust validation technique to better establish the reliability and validity of the T-TPQ through confirmatory factor analysis (CFA).18 CFA is a technique capable of describing exactly how well each of the T-TPQ items measures the dimension of teamwork that it is purported to measure. Lastly, the sample used in the previous effort was relatively homogeneous; 73.3% of participants were direct patient care providers.12 The relative absence of diversity from that sample limits the extent to which evidence on the reliability and validity of the T-TPQ will generalise to populations outside of direct patient care providers.

    Method

    The T-TPQ includes 35 items that target the five core teamwork dimensions discussed above (seven items per dimension). Responses are scored by asking participants to indicate their level of agreement with each T-TPQ item (ie, strongly agree, agree, neutral, disagree or strongly disagree). The US Army provided the data for this study and granted permission for analysis and publication of findings. The study sample excluded any identifying information that could be linked to participants. Staff members from the US Army medical facilities across the USA (N=1700) completed the T-TPQ (table 1). A CFA was conducted using Analysis of Moment Structure (AMOS) V.19. CFA is ideal for analyses involving large sample sizes and is uniquely suited to providing a more accurate account of the T-TPQ's reliability and validity than what was previously reported.12 For instance, each T-TPQ item can be assessed on how well it actually measures the dimension that it is purported to measure and whether items explain variance for multiple dimensions. This is deemed the model’s ‘fit’ and determines how well the measure assesses what it is intended to assess. Below we will discuss in more detail how the goodness-of-fit for a model represents its validity.

    View this table:
    Table 1

    Staff positions for study sample

    CFA allows hypothesis testing of the validity with which a test measures its underlying factors/constructs/dimensions. Specifically ‘in testing for the validity of factorial structure for an assessment measure, the researcher seeks to determine the extent to which items designed to measure a particular factor (ie, latent variable construct) actually do so’.19 The metric's subscales represent the underlying latent factors (in this case—the five dimensions of teamwork), and all items that are aimed at measuring that subscale should ‘load’ (ie, co-vary) with that factor and only with that factor. Therefore, the question ‘is this a good model?’, as tested through the methods of CFA, is answered by how well the CFA model fits the data. The better the fit, the more we can say that the model represents a metric that measures what it is intended to measure. A well-fitting model indicates that (a) the underlying factors account for the variance in the items within their subscale (ie, questions for a particular dimension only co-vary with that dimension) and (b) the underlying factors do not account for variance in other subscales. Further, it has been suggested that it's only appropriate to use CFA when the metric in question has been (a) theoretically derived and (b) previously tested empirically.19 Both of these requirements have been met for the T-TPQ, and, given that it has been implemented in situ in multiple operational environments (eg, hospitals), it appears to be a perfect candidate for validity testing using CFA.

    The analysis began by specifying a theoretical model in AMOS. In this model, the TeamSTEPPS’ dimensions are represented as ovals, and the indicators of each dimension (ie, each T-TPQ item) are represented as rectangles. Each item is directly connected to the dimension it is supposed to measure, and the relationships among the TeamSTEPPS’ dimensions are depicted as double-headed arrows. As illustrated in figure 1, for example, Communication is thought to be a unique and important TeamSTEPPS’ dimension, and indicators of Communication (Q29–Q35 of the T-TPQ) should only measure and represent Communication. Similarly, the correlations between Communication and the other TeamSTEPPS’ dimensions are thought to relate to one another in some way. In sum, the responses to the T-TPQ items (ie, the rectangles) are shown to be the product of the underlying TeamSTEPPS’ dimension (ovals).

    Figure 1
    Figure 1

    Final confirmatory factor analysis (CFA) model (model 5). Asterisks indicate significant relationships at p<0.001.

    Sample data were entered to estimate model fit.20 Essentially, a good fitting CFA model gives estimates that are very close to what could be expected from the actual study population (ie, extrapolating sample data to all individuals these data seek to account for). From this analysis, it can be concluded that: (1) the TeamSTEPPS’ dimensions are, in fact, unique and meaningful; (2) the T-TPQ items consistently measure what they are supposed to measure (ie, does an item seeking to measure Communication measure that dimension and only that dimension?).

    Ultimately, five models were assessed. At each stage, post hoc modifications were made to improve the fit of the model. Starting with the first theoretical model, parameter estimates and fit indices were evaluated to determine how well the data fit. Specifically, the analyses evaluated several fit indices commonly used to interpret CFA findings.19 The rationale for this approach was to demonstrate that our findings consistently demonstrate the same results across a variety of metrics. This process of model assessment answers the fundamental question: is it a good model?19 ,21

    Results

    The validity of the T-TPQ was found through modelling procedures of CFA. That is, the set of seven questions that accompany each dimension clearly represent that dimension. Also, the final analysed model demonstrates that each of the dimensions is a justifiable component of individual perceptions of teamwork. Many of the dimensions are strongly related to one another (eg, Situation Monitoring and Mutual Support), which is logical when one considers each dimension as an aspect of teamwork. As an example, it is impossible to monitor a situation as a team without mutual support from one's team members. Multiple individuals need to cooperate for effective situation awareness and monitoring, and this is evident in the data. To assess the strength of each model, three common fit indices were used: Tucker–Lewis Index (TLI), Comparative Fit Index (CFI) and root mean square error of approximation (RMSEA).19 Each of these fit indices has a set cut-off value, akin to the common p<0.05 of hypothesis testing. Specifically, when used in combinations (ie, more than one fit index for a model), the acceptable values for these indices are as follows: TLI ≥0.95, CFI ≥0.96 and RMSEA ≤0.06.22 Given this, the RMSEA is commonly accepted as one of the best fit indices available.23 Specifically, RMSEA provides a robust index of whether the tested model is of high quality, if the model is mis-specified or not, as well as providing a CI.23

    Model 1: Our first model contained the five dimensions and their representative questions and error terms. Each dimension had seven associated questions. This model showed a reasonable fit: x2 (df) 4829.299 (550), p<0.0001, TLI=0.919, CFI=0.925, RMSEA= 0.068. Owing to high modification indices between some of the error terms resulting from this procedure, it was decided that a second model be analysed in which the error terms for questions 29 and 31were correlated, both of which were under the dimension of Communication, and which contained the highest modification index (410.369).

    Model 2: After correlation of the error terms between items 29 and 31, the resultant model had the following fit indices: x2 (df) 4375.639 (549), p<0.0001, TLI=0.928, CFI=0.933, RMSEA=0.064. Again, the modification indices were examined, and it was observed that there was a high modification index (305.513) between the error terms for questions 12 and 13, both under the dimension of Leadership. Therefore, a third model was run which included a correlated error term for these items.

    Model 3: After correlation of the error term between items 12 and 13, the resulting model had the following fit indices: x2 (df) 4043.456 (548), p<0.0001, TLI=0.934, CFI=0.939, RMSEA=0.061. Modification indices were again examined, and it was found that the error terms between questions 26 and 27, which fall under the dimension of Mutual Support, had a high modification index (297.472). A fourth analysis was conducted with correlated error terms for items 26 and 27.

    Model 4: After correlation of the error term between items 26 and 27, the resulting model had the following fit indices: x2 (df) 3724.847 (547), p<0.0001, TLI=0.940, CFI=0.945, RMSEA=0.058. This model shows a good fit, yet modification indices demonstrated that the error terms for items 22 and 23, which were also both under Mutual Support, had a high modification index (118.023). A fifth and final analysis was conducted using a model including all three previous error terms and the error term between 22 and 23.

    Model 5: This included four sets of correlated error terms. All of these error terms were between items within dimensions, and included: questions 12 and 13 under Leadership; questions 22 and 23 under Mutual Support; questions 26 and 27 under Mutual Support; and questions 29 and 31 under Communication. Upon examination of the correlated items, it was apparent that these questions contained highly similar content (eg, staff ‘speak up’ when they have a concern about the patient), which would therefore lead to their correlated errors. The fit indices for this final model were good (x2 (df) 3601.27 (546), p<0.0001, TLI=0.942, CFI=0.947, RMSEA=0.057). The final model is shown in figure 2, and the correlations between latent variables are shown in table 2.

    View this table:
    Table 2

    Estimated correlations between dimensions (N=1700)

    Figure 2
    Figure 2

    The five dimensions of teamwork measured by the TeamSTEPPS Teamwork Perceptions Questionnaire (T-TPQ). The construct of team structure is represented by the Patient Care Team encircling the other four dimensions.

    The hypothesised model containing five dimensions (Team Structure, Leadership, Situation Monitoring, Mutual Support and Communication) fits the data very well. In other words, the theoretical model we set out to validate was predicted by the 1700 data points we collected from healthcare workers. The final model clearly demonstrated that all five dimensions of the T-TPQ are important components for measuring individual perceptions of teamwork. Specifically, all fit indices within our final model (ie, Model 5) demonstrated borderline to good fit given index standards in CFA: TLI=0.942 (≥0.95), CFI=0.947 (≥0.96) and RMSEA=0.057 (≤0.06; CI 90=0.056–0.059)22 (see table 3 for a summary of all fit indices and acceptable value ranges). Also, and just as important, measures of one dimension did not correlate with or predict other dimensions. Therefore, each of the dimensions is an independent, yet integral, component of the T-TPQ. Given this outcome, we can posit with certainty that the T-TPQ can be used as a valid measure in healthcare settings. Below, the reliability (the consistency with which the measured variables are assessed by the T-TPQ) and validity (whether the measure is representing the dimensions or not) of the T-TPQ will be reviewed.

    View this table:
    Table 3

    CFA fit indices for each model.

    The overall reliability of the T-TPQ was excellent (Cronbach's α=0.978). Compared with previous reliability analyses, this large sample analysis demonstrates a higher reliability than previously concluded (see figure 2 and table 4). Also, the reliability of the individual dimensions was exceptional, and each exceeded an acceptable level of 0.9 or more. The reliabilities for each dimension are listed in table 4. Given these reliability estimates, it seems that the T-TPQ has a very high internal consistency (ie, the items are related in what they measure) and is therefore a reliable measure of individual perceptions of teamwork. When the outcome of this CFA is examined in relation to the previous T-TPQ analysis, it is clear that the reliability of the measure is better than previously suspected (see figure 2). Specifically, the seven items used to measure each of the five dimensions are, in fact, measuring their targeted teamwork dimension. Also, none of the items cross-correlate, which indicates that none of the T-TPQ items measure more than one dimension. This point is important and noteworthy because, if items measure multiple dimensions, it can be argued that the number of dimensions is incorrect or needs to be reduced. There is no basis for this phenomenon concerning the metric being discussed here. Therefore, according to our analysis, it can be concluded from these data that all five dimensions are important facets of teamwork, and that, across 1700 healthcare professionals and support staff, all five dimensions are consistent components of individuals’ perceptions of teamwork.

    View this table:
    Table 4

    Summary of reliability for T-TPQ items and dimensions

    Discussion

    This effort was a large sample follow-up to previous reliability analyses that used a much smaller sample size12 and adds to the previous work by showing that the T-TPQ is more reliable and that it demonstrates validity, compared with previous analyses. Our CFA model demonstrates a strong case that the T-TPQ is actually measuring its intended dimensions. Further, the CFA demonstrated remarkable fit across multiple standardised indices. These fit indices provide clear support that the T-TPQ can be used as a valid tool to measure individual staff members’ perceptions of group-level teamwork within their unit or department across multiple healthcare settings, professions and work positions. Also, reliability analyses demonstrated high reliability for both the entire measure and each of its sub-dimensions.

    Although the analyses presented here tested the structure of the T-TPQ (ie, construct validity) as a measurement tool of perceptions of teamwork, they in fact did not test ‘criterion validity’ or ‘predictive validity’. In other words, this analysis only shows to what extent the survey measured perceptions of teamwork and does not directly link the perceptions of teamwork to external performance criteria (eg, patient safety outcomes). Other research has shown that good teamwork is associated with increased patient safety, specifically lower mortality rates,24 but this paper does not provide empirical data demonstrating that relationship. Given these results, this CFA demonstrated that the T-TPQ is: (a) a valid measure, at least construct-wise, of its individual dimensions; (b) reliable, both as an overall survey and at the level of each dimension.

    Assessing teamwork is an integral component of patient safety. Without proper teamwork, it is difficult, if not impossible, to increase patient safety outcomes.25 The T-TPQ is a valid and reliable measure of individual staff members’ perceptions of the quality of teamwork within their work units. Although it does not measure actual teamwork behaviours, it is a relatively time- and cost-effective survey that can evaluate medical team members’ perceptions: Team Structure, Leadership, Situation Monitoring, Mutual Support and Communication. In addition, the T-TPQ is a viable alternative for gaining insights into teamwork when assessment of teamwork behaviours (eg, attitudes, behaviours and cognitions) is impractical or impossible. When used in addition to objective measures of teamwork behaviours, such as direct observations, the T-TPQ can be a valuable component of a comprehensive teamwork assessment and performance-improvement strategy. For those healthcare organisations that have implemented TeamSTEPPS, the T-TPQ is a construct-valid tool for assessing and continually improving TeamSTEPPS’ training, implementation and sustainment. Further, the T-TPQ would probably also be a useful tool for evaluating other performance-improvement interventions that focus on enhancing these specific teamwork dimensions. Conversely, because it is designed to assess just this set of five teamwork dimensions, the T-TPQ may not be appropriate for assessment of interventions that are aimed at improving other facets of teamwork (eg, collective efficacy and team psychological safety). In these instances, the T-TPQ may underestimate the intervention impact because of incongruences between the teamwork dimensions targeted by the intervention and those measured by the T-TPQ. In short, users should consider these to be potential limitations for assessing whether the T-TPQ is an appropriate measure for evaluating their interventions; otherwise they may make erroneous judgements about the interventions’ effectiveness.

    Practically speaking, the T-TPQ is a short, construct-valid survey that can aid frontline clinicians, patient safety professionals, researchers, medical educators and healthcare administrators in gathering insightful information to evaluate teamwork and the associated safety climate within their medical units and departments. This survey provides a reliable and (construct) valid instrument to assess teamwork perceptions based on a prevalent training, which is a worthwhile step in improving team-training initiatives, teamwork and ultimately patient safety. Owing to the difficulty of collecting medical professional behaviours in situ, a valid survey that can assess perceptions in proxy of actual behaviours allows researchers to understand major dimensions of teamwork in an easily accessible, non-intrusive way. Future research is needed to further understand the T-TPQ in relation to relevant patient safety outcomes such as safety culture, patient harm and medical errors.

    References

    1. ↵
      1. Capella J,
      2. Smith S,
      3. Philip A,
      4. et al
      . Teamwork training improves the clinical care of trauma patients. J Surg Educ 2010;67:439–43.
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Deering S,
      2. Rosen M,
      3. Ludi V,
      4. et al
      . On the front lines of patient safety: implementation and evaluation of team training in Iraq. J Qual Patient Saf 2011;37:250–7.
      OpenUrl
    3. ↵
      U.S. Department of Health and Human Services. National Strategy Quality Improvement in Health Care 2012 Annual Progress Report to Congress. http://www.ahrq.gov/workingforquality/nqs/nqs2012annlrpt.pdf Published 2012. (accessed 10 Jun 2012).
    4. ↵
      1. Rosen AK
      . Are we getting better at measuring patient safety? Agency for Healthcare Research and Quality website. http://webmm.ahrq.gov/perspective.aspx?perspectiveID=94 Published November 2010. (accessed 9 May 2012).
    5. ↵
      1. Manser T
      . Teamwork and patient safety in dynamic domains of healthcare: a review of the literature. Acta Anaesthesiol Scand 2009;53:143–51.
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Mardon R,
      2. Khanna K,
      3. Sorra J,
      4. et al
      . Exploring relationships between hospital patient safety culture and adverse events. J Patient Saf 2010;6:226–32.
      OpenUrlCrossRefPubMed
    7. ↵
      1. Wheelan SA,
      2. Burchill CN,
      3. Tilin F
      . The link between teamwork and patients’ outcomes in intensive care units. Am J Crit Care 2003;12:527–34.
      OpenUrlAbstract/FREE Full Text
    8. ↵
      1. Edmondson AC
      . Psychological safety and learning behavior in work teams. Adm Sci Q 1999;44:350–83.
      OpenUrlCrossRefWeb of Science
    9. ↵
      1. Hackman JR,
      2. Oldham GR
      . Motivation through the design of work: test of a theory. Organ Behav Hum Perform 1976;16:250–79.
      OpenUrlCrossRefWeb of Science
    10. ↵
      1. Leape LL
      . Error in medicine. JAMA 1994;272:1851–7.
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      1. Galvin RS,
      2. McGlynn E
      . Using performance measurement to drive improvement: a road map for change. Med Care 2003;41:48–60.
      OpenUrl
    12. ↵
      American Institutes for Research. TeamSTEPPS teamwork perceptions Questionnaire (T-TPQ) manual. Washington, DC: American Institutes for Research, 2010.
    13. ↵
      1. Baker DP,
      2. Gustafson S,
      3. Beaubien J,
      4. et al
      . Medical teamwork and patient safety: the evidence-based relation. Washington, DC: American Institutes for Research, 2003.
    14. ↵
      1. Baker SP,
      2. Salas E,
      3. Battles JB,
      4. et al
      . The relation between teamwork and patient safety. In: Carayon P. ed. Handbook of human factors and ergonomics in health care and patient safety. 2nd edn. Boca Raton, FL: CRC Press, 2012:185–98.
    15. ↵
      1. Salas E,
      2. Sims DE,
      3. Burke CS
      . Is there a “Big Five” in teamwork? Small Gr Res 2005;36:555–99.
      OpenUrlAbstract
    16. ↵
      1. Baker DP,
      2. Day R,
      3. Salas E
      . Teamwork as an essential component of high-reliability organizations. Health Serv Res 2006;41:1576–98.
      OpenUrlCrossRefPubMedWeb of Science
    17. ↵
      1. Gorsuch RL
      . Factor analysis. 2nd edn. Hillsdale, NJ: Erlbaum, 2003.
    18. ↵
      1. Blunch NJ
      . Introduction to structural equation modeling using SPSS and AMOS. Thousand Oaks, CA: Sage Publications LTD, 2008.
    19. ↵
      1. Byrne BM
      . Structural equation modeling with AMOS. 2nd edn. New York, NY: Routledge, 2010.
    20. ↵
      1. Kohn LT,
      2. Corrigan JM,
      3. Donaldson MS
      . eds. To err is human: building a safer health system. Committee on Quality of Health Care in America, Institute of Medicine. Washington, DC: National Academy Press, 2000.
    21. ↵
      1. Tabachnick BG,
      2. Fidell LS
      . Using multivariate statistics. Boston, MA: Allyn & Bacon, 2001.
    22. ↵
      1. Hu LT,
      2. Bentler PM
      . Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives. Struct Equ Modeling 1999;6:1–55.
      OpenUrlCrossRef
    23. ↵
      1. MacCallum RC,
      2. Austin JT
      . Applications of structural equation modeling in psychological research. Annu Rev of Psychol 2000;51:201–26.
      OpenUrlCrossRef
    24. ↵
      1. Neily J,
      2. Mills PD,
      3. Young-Xu Y,
      4. et al
      . Association between implementation of a medical team training program and surgical mortality. JAMA 2010;304:1693–700.
      OpenUrlCrossRefPubMedWeb of Science
    25. ↵
      1. Firth-Cozens J
      . Cultures for improving patient safety through learning: the role of teamwork. Qual Health Care 2001;10: 26–31.
      OpenUrlWeb of Science
    View Abstract

    Footnotes

    • Correction notice This article has been corrected since it was published Online First. Figure legends 1 and 2 have been transposed.

    • Contributors Each author provided a unique and needed insight and effort into the completion of this manuscript, from original conception through iterative editing and final proofing. JRK provided a substantial contribution through analysis and interpretation of the data, outlining and drafting the article, and providing tasking and substantial writing of portions of the manuscript. ASD contributed through aiding in all aspects of the analysis, as well as writing portions of the method and conducting portions of the literature review. EHL contributed through writing major portions of the introduction, conducting portions of the literature review, and providing expertise in AMA formatting, and creation of many of the tables and figures. LEB contributed through providing major portions of writing of the rationale section, and through writing portions of the method. PAT, SAA and HBK were responsible for conception, design and acquisition of the datasets. They all contributed substantially to both initial writing efforts, many of the following edits, and handling the manuscript as it passed through the DoD approval process, which was a necessary and vital step to be able to publish these data. They were also responsible for giving access to the data and ensuring that the manuscript was a significant contribution to the TeamSTEPPS programme and patient safety in general. ES provided expert management and guidance of the writing team through his expertise in teamwork. ES reviewed, contributed to and edited many iterations of the manuscript. Further, he provided invaluable insights on measurements of teams, and aided as a liaison between the Army Patient Safety programme and the University of Central Florida throughout the entire process of writing the manuscript.

    • Funding This publication was prepared by Booz Allen Hamilton under contract to TRICARE Management Activity, Department of Defense (DoD) Contract No W81XWH-08-D-0025, Task Order No 0015. The views herein are those of the authors and are not to be construed as official or as reflecting the views of TRICARE Management Activity or the Department of Defense.

    • Competing interests None.

    • Ethics approval University of Central Florida, US Army, US Department of Defense.

    • Provenance and peer review Not commissioned; externally peer reviewed.