Optimal collaboration between professionals is important for offering continuous and coordinated care. A patient-care team that functions well is crucial for providing high-quality care for patients.1^–4 Team functioning is determined not only by structural determinants such as workload, team size or team composition,56 but also by team processes. There is a large body of research on the relationship between team processes and team effectiveness,7^–14 which shows that a team climate in which team members are encouraged to develop and implement new ideas can lead to better healthcare and healthcare outcomes.

West presents the most studied model of this “team climate aimed at innovation” and suggests that four climate factors, “vision,” “participative safety,” “task orientation” and “support for innovation,” are essential for developing and implementing innovations.12 This theoretical model led to the development of a questionnaire to measure team climate, the Team Climate Inventory (TCI). The TCI has been used as an improvement tool for assessing team function to identify areas that could be improved.1516 In addition, it has been used in research as an outcome measure of quality-improvement strategies or to predict the success or failure of such a strategy.51718

The TCI has been translated into several languages, including Norwegian,19 Swedish,20 Finnish,21 Italian22 and German,23 and it has been tested on many different teams (eg, management, social services, psychiatric, oil company, industrial and primary care teams; table 1). Given this utilisation of the TCI tool in multiple countries and industries, it may become a helpful vehicle in the translation of learning among countries and industries. However, none of these studies tested the questionnaire specifically for hospital teams.

Delivering integrated care is becoming increasingly important for hospital teams, and a good team climate is crucial for delivering high-quality care and quality improvement in healthcare. Therefore, it is important to determine the psychometric characteristics of the TCI in this specific setting.

The present study extends the research work on the TCI by investigating the psychometric properties of a Dutch version of the TCI on a population of hospital teams. In this paper, we describe how we systematically assessed the validity, reliability and discrimination capacity of the TCI in hospital teams.

METHODS

TCI questionnaire and Dutch translation

The original 116-item TCI of West underwent exploratory and confirmatory factor analyses, which resulted in a 44-item version.152425 The 44-item TCI includes four scales, “vision,” “participative safety,” “task orientation,” and “support for innovation” and 13 subscales (fig 1). The questionnaire includes a fifth scale that was designed to detect socially desirable answers. For the Dutch translation, four people used Beaton’s guidelines26 to systematically and independently translate and back-translate the TCI. One translator was a native-English speaker fluent in Dutch, and the other three were Dutch researchers fluent in English.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Structure of the 44-item Team Climate Inventory. © Anderson and West/ASE (1999). NFER-NELSON, Darvill House, 2 Oxford Road East, Windsor, UK.

Study population

On the basis of a 10:1 ratio of participants to items27 for each of the 38 items in the questionnaire (excluding the six items for social desirability), we aimed at including 380 healthcare professionals. Two types of hospital teams were invited to participate: monodisciplinary teams of nurses and multidisciplinary quality-improvement teams. We contacted all the head nurses of the nursing teams at one university hospital and invited their teams to participate in the study. We contacted the facilitators of quality-improvement teams participating in national breakthrough collaboratives. Participation was voluntary.

Procedure

The Dutch TCI was handed to the contact person of each participating team. All members completed the questionnaire individually and anonymously. The respondents were asked to indicate on a five-point scale to what extent they agreed with each item. Each respondent returned the survey in a sealed envelope, which was sent to the researchers.

Analyses

As in the preceding TCI psychometric studies, the six social desirability items were excluded from analyses.

Factor structure and intercorrelations

To test the validity of the TCI in healthcare teams, we performed exploratory (principal-component factor analysis, including a varimax rotation, SPSS version 12) and confirmatory factor analyses (using LISREL), to see which items cluster together. Confirmatory factor analyses tested the robustness of the factor solution. LISREL is a statistical program for structural equation modelling, a form of multivariate data analysis that tests the goodness of fit of the empirical data and a specified or hypothesised model. In order to evaluate the fit of the models, we examined the five most used criteria [the chi-square (χ², lower scores mean a better fit); the ratio of maximum-likelihood chi square (χ²/df, scores from 2 to 5), the Goodness-of-Fit Index (GFI>0.90), the Comparative Fit Index (CFI⩾0.95) and the root-mean-square error of approximation (RMSEA⩽0.10)]. The confirmatory factors for the total sample and for the nursing and quality-improvement teams were analysed separately. Once we had defined the structure, we computed the Pearson product moment correlations of the factors with the subscales that we identified.

Internal homogeneity

To assess the reliability of the TCI in healthcare teams, we calculated the internal homogeneity by calculating Cronbach alpha coefficients for the scales emerging from the factor analyses. A Cronbach alpha score of 0.7 or higher is usually regarded as indicative of acceptable reliability.28

Discriminating capacity

To see if the TCI discriminates between different types of healthcare teams, we tested its ability to discriminate between nursing teams and quality-improvement teams. We tested the difference in the TCI scores for nursing teams and quality-improvement teams with mixed model analysis in which we corrected for clustering in teams.

RESULTS

Study population

The study participants consisted of 424 healthcare workers: 355 were part of one of the 22 monodisciplinary hospital teams of nurses, and 69 were part of one of the 14 multidisciplinary quality-improvement teams. The nursing teams came from 22 specialities in surgery, internal medicine and childcare, and each team consisted of 10 or more people (mean 16; range 6–36). The improvement teams were breakthrough collaborative teams aiming at improving care for patients with diabetes, perioperative care and care for patients with head and neck cancer. The multidisciplinary teams consisted of a minimum of three people (mean 5; range 3–13), and each included at least one nurse and one physician.

Factor structure and correlations

Exploratory factor analysis

The principal-component factor analysis indicated that there should be five factors instead of the original four factors. These five factors together accounted for 60% of the total variance. Table 2 presents the items of the scales and their factor loadings on the five-factor solution, after varimax rotation. Overall, all items from the scales “vision,” “support for innovation” and “task orientation” load on their theoretical scale except for item 38 (“provide each other useful ideas”) which loaded higher on the scale “participation safety.” The original scale “participation safety” fell apart into two scales: the first part contained all items of the subscales “interaction frequency,” “information sharing” and item 3 (“we all influence each other”); the second part consisted of all items of the subscales “safety” and “influence” except for item 3. Two items besides item 38 had high factor loadings (r>0.4) on one other factor than their theoretical scale, namely, item 1 (“share information rather than keeping it to oneself”) and item 7 (“people feel understood and accepted by each other”).

Confirmatory factor analysis

The results of the confirmatory factor analysis with LISREL indicated that the five-factor model has the best fit (table 3). The fit indices show that the five factors fit the model moderately. Further LISREL analyses of the models for nursing teams and improvement teams also suggested that a five-factor solution provides the best fit to the data for both groups. However, all fit indices of the improvement teams were lower than those for the nursing teams.

Intercorrelations

All factors or scales are significantly and positively correlated (table 4). Scale correlations range from 0.47 (“vision” and “interaction and information sharing”) to 0.72 (“support for innovation” and “participation safety”). The interitem correlations show adequate levels of interscale correlation for the factors “vision,” “support for innovation” and “participation safety” (table 4). The items of the factor “interaction and information sharing” all show high interitem correlations (⩾0.37) except for item 3 (“we all influence each other’; score 0.18). The lowest interitem correlation within the factor “task orientation” was caused by item 38: “providing useful ideas and practical help.”

Internal homogeneity

The Cronbach alpha analysis of the five scales revealed alphas between 0.83 and 0.93 (table 4), which indicates good reliability for all five factors of the instrument.

Discriminating capacity

The analysis of the difference between nursing and quality-improvement teams revealed significant TCI-score differences (table 5). The mean scores of the improvement teams were significantly higher on all scales than those of the nursing teams. The greatest difference was on the scale “vision,” and the least on “participation safety.”

DISCUSSION

This is the first study in which the Team Climate Inventory (TCI), an instrument for measuring team climate, has been studied exclusively in a population of hospital teams. We found five factors in exploratory and confirmatory factor analyses, as did Anderson and West.24 However, their fifth scale consists of all items from the subscale “interaction frequency,” while our scale “participative safety” fell into two subscales, namely, “information sharing and interaction frequency” and “safety and influence.” The Cronbach alpha scores in our study of 0.83 or more indicated good reliability, and the significant differences found between nursing teams and quality-improvement teams show that all the TCI scales have a discriminating capacity.

A five-factor solution was found in all the validation studies, including Anderson and West’s, except from the study in Sweden.19^–23 However, all the studies that found a fifth factor identified it as the subscale “interaction frequency.” Our study of healthcare teams is the first study to found another partitioning of the items. Further research should be done to see if this is specific for hospital teams. High scores on internal homogeneity that are as good or better than those of other studies highlight the reliability of our factor solution.19^–23 The discriminating capacity of the TCI has not been tested before.

A limitation of this study is that we studied only two types of hospital teams, namely, nursing teams and quality-improvement teams. It is unclear whether the differences in scores of these kinds of teams are influenced by their size or composition. All nursing teams were large monodisciplinary teams (consisting of more than 10 people each), and all improvement teams were small, multidisciplinary teams. Further research should be done to test the TCI on more hospital teams with various sizes and compositions.

We also recommend research into shortening the TCI. The high interitem correlations may mean that different items actually measure the same phenomenon to a significant degree. For example, “How far are you in agreement with the objectives of the team?” and “To what extent do you think these objectives are useful and appropriate?” have an intercorrelation of 0.80. A short version of the Finnish TCI has already been developed29 but needs further testing in healthcare teams.

Patient care teams are key components in the delivery of high-quality patient care.430 Studies show that team climate is an important characteristic of successful teams, either working in a microsystem of healthcare, as teams of nurses do, or coming together as a quality-improvement team.3132 Our study shows that the TCI is a valid, reliable and discriminating measure of team climate among hospital teams. Further research within the healthcare setting is needed to estimate its usefulness as a quality-improvement tool or as a predictor of quality-improvement outcome.