Article Text
PDF
Abstract
Background Effective working in multidisciplinary teams (MDTs) is promoted as essential in ensuring good healthcare outcomes, suggesting that an understanding exists of the relationship between outcomes and the attributes needed by individuals to function effectively in the MDT. While the characteristics of effective teamwork have been described, the attributes needed by individual MDT members have not been investigated. To address this, the study focuses on newly qualified Foundation Year 1 (FY1) doctors, creating and testing a tool to evaluate their performance in the MDT.
Methods Repertory grid technique was used to elicit attributes needed by FY1 doctors to function effectively in the MDT. Study participants (all experienced MDT members) used these to evaluate MDT working by FY1 doctor colleagues. Data on 57 FY1 doctors were collected from 95 MDT members working in five hospitals. Participants also ranked the attributes in terms of importance for effective team functioning and rated an ‘Ideal’ FY1 doctor.
Results The repertory grid permitted differentiation between groups of FY1 doctors’ MDT performance. FY1 doctors who undertook interprofessional training were rated no differently than UK-trained graduates without such training. UK-trained graduates were rated significantly higher on all attributes than non-UK-trained graduates. Overall, FY1 doctors were rated lower than the ‘Ideal’. Factor analysis and rankings suggested tensions between clinical attributes needed for good team functioning and more ‘social’ attributes.
Conclusions This study demonstrates the potential of repertory grid methodology in eliciting attributes that are important for effective teamworking, and using these to evaluate MDT working by FY1 doctors.
- Attitudes
- Teamwork
- Health professions education
- Medical education
- Performance measures
Statistics from Altmetric.com
Introduction
In the complex world of healthcare, good teamwork among the members of the multidisciplinary team (MDT) that cares for patients is promoted as essential to ensure best outcomes, as well as best use of limited human and capital resources.1–3 This suggests that an understanding exists of the relationship between healthcare outcomes and particular MDT characteristics. It also implies that it is necessary to understand and develop the attributes that permit individuals to function effectively in the MDT.
In reality, ‘practical and well-evaluated plans for implementing teamwork are fairly rare’, and evidence of the effectiveness of teamworking is not measured and reported systematically by healthcare providers.4 The characteristics of effective MDTs have been described in overall terms,1–4 and several instruments have been described for assessing teamwork effectiveness5–8 though no single standardised instrument appears to be in widespread use. By contrast, the attributes needed by individual team members to work effectively in the MDT have not been widely investigated,2 and aside from one scale intended to assess nurses’ opinions toward other MDT members,9 studies have not sought to measure individual performance in the MDT, nor are instruments described for measuring such performance.
Annually in the UK, newly qualified doctors join MDTs in hospitals to undertake foundation training for their subsequent specialties, becoming responsible with their MDT colleagues for the delivery of safe and effective healthcare. Working in the MDT is generally a new experience because medical undergraduate training focuses dominantly on acquiring clinical competence. During the first 2 years of practice, the Foundation Programme ‘ensures that newly qualified doctors develop their clinical and professional skills in the workplace’.10 Though considered essential for good healthcare outcomes, MDT working skills of new doctors are, nevertheless, not formally assessed. In this study, we aimed to develop and test an instrument to assess the MDT performance of newly qualified doctors (Foundation Year 1 (FY1) doctors). We also wished to explore whether Hull York Medical School (HYMS) graduates, whose final year curriculum included 2 weeks of interprofessional training in a ward setting, were rated more highly in the MDT than graduates without such training.
Method
Repertory grid technique
Repertory grid technique (‘RGT’) is one of the methods of Personal Construct Psychology.11 We used this method because there seems to be no comprehensive listing of the attributes FY1 doctors require to work well in teams that has actually been drawn from those who would most appropriately be able to identify them, that is, other, more experienced members of MDTs who work with FY1 doctors. As ‘RGT’ is based on the precept that the researcher would conduct interviews with a sample taken from the target group to elicit such attributes,12 the method seemed ideal for this study.
RGT has been used in a very wide range of contexts, for example, organisational development, teacher development, psychotherapy, nursing, management development and so on,12, 13 to understand how people ‘construe’ (ie. see, understand, interpret) people, situations and things. Repertory grids are not standardised tests or questionnaires, so it is up to the researcher to design a grid to suit the particular research project.12 In this study, the purpose of the repertory grid was to see how members of MDTs construed FY1 doctors in the context of the attributes needed by such doctors to work well in MDTs.
A repertory grid consists of ‘elements’ (the column headers), ‘personal constructs’ (the labels that describe the rows of the grid) and a method for linking the two together—usually a rating scale.14 The components of a repertory grid are:
-
Elements. These are the people, situations or things construed. Elements must be carefully selected so that they are relevant for purpose of the grid.12 In this study, we were concerned with how participants (ie, members of MDTs) construed FY1 doctors, so the elements were FY1 doctors with whom the participants had worked, together with an ‘Ideal’ element.
-
Constructs. The means by which people construe are through their ‘personal constructs’. These are bipolar dimensions, such as kind-vs-cruel and lazy-vs-hardworking.12 We were interested in how participants would construe the elements on a range of constructs relating to the attributes required to work well in MDTs. An example of a construct used is the attribute poor team player-vs-good team player. The 15 constructs identified by the MDT members are listed in the grid page shown in online supplementary appendix 1. The words ‘construct’ and ‘attribute’ are used interchangeably in this article, and references to them are shown in italics for clarity.
-
Rating scale. The means by which a construct is ‘linked’ to an element is commonly a rating scale, for instance, of 1–7 as used in this study (see online supplementary appendix 1).12 A participant would rate the FY1 doctor (element) concerned on the construct does not reflect and learn from experience-vs-reflects and learns from experience on a scale of 1 (someone who does not reflect and learn from experience at all) to 7 (someone who very much does).
Developing the repertory grid
To elicit the constructs (attributes) needed by FY1 doctors to enable them to work well in MDTs, the authors conducted individual, confidential interviews with 10 health professionals representative of an MDT (3 staff nurses; 1 pharmacist; 2 physiotherapists; 2 occupational therapists; 2 senior doctors). In eliciting a representative sample of personal constructs from a homogeneous population in a particular setting (eg, hospital healthcare professionals), usually only small numbers need to be interviewed15 and, indeed, we found that the same constructs emerged repeatedly from different participants.
Interview participants wrote on index cards the initials of particular FY1 doctors with whom they worked (the ‘elements’). These were then presented to them in groups of three, and they were asked to think of ways in which, in the context of effective MDT working, two of the elements were alike, but different from the third. This gave rise to the constructs that formed the basis for the 15 attributes used in our study, for example, prioritises tasks efficiently-vs-not good at prioritising tasks efficiently. This ‘Triadic Method’ of construct elicitation is explained in detail in Fransella et al.12
To elaborate the meaning behind the construct labels, the techniques of ‘laddering’16 and ‘pyramiding’17 were also used in the interviews. A person's personal constructs do not exist in isolation; they are arranged in a hierarchical system with some being more important, value-laden constructs (‘superordinate’), and some being less important, more ‘concrete’ constructs (‘subordinate’). Laddering is used to elicit superordinate constructs, while pyramiding is used to elicit subordinate constructs. An example of a relatively superordinate construct is can be trusted to complete what they have agreed to do-vs-unreliable in delivering their undertakings. It is worth noting that participants nearly always said that the ‘top’ construct in their respective ‘ladders’ related to patient safety. The prioritising tasks construct mentioned above is an example of a more subordinate, concrete, construct.
The construct elicitation interviews created a ‘pool’ of approximately 200 cards with individual constructs written on them. We categorised these into ‘groups of like meaning or theme’,15 resulting in 15 bipolar constructs carefully selected to represent these groups. These were the constructs used in the repertory grid created to gather our study data. The final grid, a page of which is shown in online supplementary appendix 1, was presented to participants in a grid ‘booklet’.
The ‘elements’ in the grid booklet were the individual FY1 doctor/s known to each study participant with a separate page for each ‘element’.18 The participant rated the FY1 doctor concerned on each of the 15 constructs listed on the grid page on a scale of 1–7. To prevent participants developing a ‘mental set’, positive and negative poles were randomly allocated so that for certain constructs, 1 was the positive pole, and 7 the negative pole, and vice versa for others. An ‘Ideal’ FY1 doctor element was also included as one of the pages in the booklet. Participants were asked to rate the ‘Ideal’ FY1 doctor element on each construct. This was to provide an ‘anchor’ for the grid, making it clear what MDT members considered to be the ‘ideal’ level of FY1 performance in respect of each construct.
The final page of the booklet contained a table of the same 15 constructs used in the grid pages, and participants were asked to rank order these (1–15) to indicate their relative importance to them personally. Equal rankings were not permitted.
Data collection
The study participants were MDT members working on wards at five hospitals in the northeast of England. Patients were not involved. Approval for the study was granted by Hull and East Yorkshire NHS Hospitals Clinical Governance Unit, North Lincolnshire and Goole Hospitals NHS Foundation Trust Clinical Audit Unit, North and East Yorkshire Alliance Research and Development Unit, and Hull York Medical School Ethics Committee. All participants provided informed consent. FY1 doctors’ anonymity was preserved by assigning to each a numeric code that was stored securely on a password-protected database. Their medical school of graduation was obtained from their hospitals.
Nurse managers were asked for permission to approach ward MDT staff to explain the study. Staff expressing interest were consented, given a grid booklet to complete, and an envelope for return. Some MDT members completed the grids at once. Others returned them by an agreed day. All data were collected between November 2010 and April 2011.
Results
A total of 95 participants rated between 1 and 5 FY1 doctors (elements) with whom they worked. In all, 205 ratings were completed on 57 different FY1 doctors. The participants’ professional groups comprised 58 nurses, 16 senior doctors (registrar level and above), 8 physiotherapists, 3 pharmacists, 3 care assistants, 2 dieticians, 2 ward clerks, and 3 who did not specify profession. These numbers are representative of hospital staffing proportions, and of the typical makeup of the multidisciplinary teams, and samples came from different hospitals (tertiary and district), and across different teams including acute admitting units, surgical teams, anaesthetic, general and specialty medical teams.
Most participants (n=85; 89%) rated the ‘Ideal’ FY1 element. Eighty participants (84%) correctly completed rankings of the 15 constructs. They comprised 45 nurses, 13 senior doctors, 8 physiotherapists, 3 occupational therapists, 2 care assistants, 2 pharmacists, 2 dieticians, 2 ward clerks and 3 profession-unspecified participants.
The FY1 doctors were grouped by medical school of graduation of which 21 were from HYMS and 25 were graduates of 8 other UK medical schools, 9 were graduates of non-UK schools and 2 were unknown.
Analysis of the repertory grid ratings of FY1 doctors
We examined the ratings to determine whether participants from different professional groups construed FY1 doctors differently in terms of the team work attributes in the grids. Owing to small numbers in some of the professional groups of participants, we limited this analysis to nurses (n=58) and senior doctors (n=16), and results are shown in table 1. Because individual FY1 doctors had different numbers of ratings, the statistical analyses were made using both the full dataset and the average of all ratings given for a particular FY1, so that s/he had a single mean rating for each construct. The patterns of results were the same when using this average, so all analyses in this paper are presented using the full dataset. All analyses were undertaken using SPSS V.18.
Average rating of FY1 doctors on each construct by staff nurses and senior doctors*
In table 1, positive numbers (closest to +3) represent best levels of performance, and negative numbers (closest to −3) represent poorer performance. Staff nurses and doctors did not rate the FY1 doctors significantly differently on any construct (p values range from 0.09 to 0.95), suggesting agreement on the overall preparedness, or lack of it, of the FY1 doctors in terms of attributes (constructs) relating to good multidisciplinary team-working. The other MDT groups were very similar in their average ratings, but rated too few FY1 doctors to support statistical comparisons between them.
Ratings for different FY1 graduate groups were examined. Graduates were identified as being from one of nine UK medical schools, including HYMS, or as non-UK graduates. Each UK school had from one to 89 rating grids completed for their graduates with an average of nine grids. There were 89 grids for HYMS and 85 for other UK schools (total=174 grids). The non-UK graduate group had 27 grids in total. For four grids, school of graduation could not be determined.
An overall analysis of variance (ANOVA) comparing the average ratings of different graduate groups (HYMS, other UK medical schools, non-UK graduates) was highly significant (F=37.61, p<.0001). A posthoc analysis using Bonferroni adjustments for multiple comparisons indicated that there were no statistical differences between HYMS graduates and those from other UK medical schools. Thus, for subsequent analysis, the UK graduates were amalgamated in a single ‘UK graduate’ group. The UK and non-UK graduate groups were compared, and the mean ratings on individual constructs in the grid are shown in figure 1. They indicate that non-UK graduates had lower ratings than UK graduates, particularly for attributes relating to decisiveness, evaluating patient risk and prioritising tasks. Both UK and non-UK FY1 groups were rated lower than the ‘Ideal FY1'on all the attributes.
Graph of average ratings of individual constructs by place of graduation. * All constructs were significantly different between all UK graduates (n=46 FY1 doctors with 178 grids) and non-UK graduates (n=9 FY1 doctors with 27 grids) at p<.0001; all constructs were significantly different between all FY1 doctors and the ideal FY1 doctor (n=85 grids) at p<.0001. Note on table layout: For presentational purposes only, the graphical layout has been adapted to illustrate the bipolarity of the constructs. Accordingly, the 1–7-point rating scale was converted into a −3 to +3 seven-point rating scale, with zero representing the ‘4’ on the original 1–7 scale. As space does not permit both labels to be shown on the tables and figures, only the labels for the positive poles of the constructs are shown, and positive ratings are reflected in the (+) side of the rating scale. This transformation of the data permits use of horizontal bar charts shown in figure 1 which optimally illustrate whether a mean rating falls on the positive or negative pole of a given construct.
Analysis of the ranking of constructs
Eighty MDT members ranked the constructs in order of importance to them personally (most important=1 to least important=15, with no equal rankings allowed).
Figure 2 shows the mean ranking of each of the 15 constructs, ordered from most important to least important.
Graph of mean ranking of constructs.
An overall ANOVA for each construct did not indicate that MDT professional groups ranked the constructs significantly differently. Differences in the mean rankings cannot be considered robust for the smaller groups of MDT professionals. t-tests comparing the two largest groups, nurses (n=45) and doctors (n=13), did not show any constructs with mean rankings that were statistically different (p values from 0.20 to 0.87).
A factor analysis was undertaken to examine the inter-relationships among construct rankings. A principal component analysis was run with a varimax rotation which resulted in 6 factors with eigenvalues greater than 1 accounting for 64.7% of the variance. The factor loadings are shown in table 2.
Factor analysis of construct rankings
Examining items that load at least 0.5 on the first four factors, marked by an asterisk in the table, some consistencies emerge. The first factor, accounting for 14.5% of the variance, contains five items: deals with events in a rational/decisive manner, anticipates risks/safety issues and clinically capable which are negatively correlated with makes effort to be sociable and enthusiastic about work.
The second factor, accounting for 11.5% of the variance, again trades-off the ranking of prioritises tasks efficiently and acknowledges importance of all opinions, while the third factor similarly trades-off can be trusted to complete undertakings and takes into account all aspects of care. The fourth factor positively relates values expertise of others and understands expertise of team.
Considering the factors extracted, two constructs making up factors 5 and 6 consist of one item each which contribute to explaining the variance, but do not seem to relate consistently to the ranking of other items. One item, good team player, had moderate loadings on four of the six factors, indicating that it was positively related to the more ‘social’ constructs, but was not as closely linked to particular items as others were.
Discussion
This study demonstrates that RGT can be used successfully both to elicit the attributes (constructs) required by FY1 doctors for good MDT performance, and to evaluate FY1 doctors on those constructs to indicate how they are performing, as judged by their MDT colleagues. Rank ordering the constructs permitted their relative importance to MDT members to be quantified.
Professional group comparisons
There were no major differences between senior nurses’ and senior doctors’ ratings for the FY1 doctors with whom they worked suggesting that these professionals have similar judgements and expectations of FY1 doctors’ MDT performance. Using the ‘Ideal FY1’ to anchor ratings, participants rated FY1 doctors with whom they worked less highly than the ‘Ideal FY1’ suggesting that they used the grid mindfully, and could judge where an individual FY1 doctor fell short of the expected ideal and, indeed, where the absolute ideal, in terms of the highest possible rating, might not actually be necessary or expected. This suggests a realistic view that FY1 doctors are commencing training and, therefore, are still learning these skills.
FY1 between groups comparisons
There were no statistical differences between FY1 HYMS graduates and those from other UK medical schools suggesting that interprofessional training had not affected performance based on the perceptions of the MDT team members for the attributes measured. Graduates from different UK medical schools were all rated similarly. However, there were significant differences between UK and non-UK graduates’ ratings for multiple constructs. The number of non-UK graduates was very small, so this finding would need further research in different contexts, and using much larger samples to support it. Other studies have also reported differences in performance between locally trained graduates and those who trained in other countries.19, 20 In terms of testing the grid, a primary aim of this study, a demonstration of its capacity to distinguish between performances is informative, suggesting that it is potentially of value in distinguishing areas of strength in FY1 doctors’ MDT performance from those needing remediation.
In identifying areas where FY1 doctors’ MDT performance deviated from that anticipated of the ‘Ideal FY1’ and in highlighting differences in performance between FY1 graduate groups in interactions with the MDT, the grid is potentially of assistance in prioritising FY1 training needs, and in focusing educational and social support where it is most required. This information could contribute to induction training programmes, such as that currently under development by the UK General Medical Council.21
Relative importance of constructs—the ranking exercise
While participants expressed great difficulty in ranking constructs because they felt they were all important, the mean rankings indicate that being clinically capable in terms of knowledge, skills and other ‘clinical’ constructs was considered more important than ‘social’ constructs, such as makes an effort to be sociable and enthusiastic about work. The factor analysis revealed a negative correlation between clinical and social constructs of performance, suggesting that when forced to decide, participants prioritised clinical competence over ‘softer’ social skills. Further research is required to reveal whether the two categories of team-working constructs (clinical and social) are truly trade-offs that are in tension with each other and/or whether their relative importance changes in different situations.
Limitations of the study
The primary aim of this study was to create an instrument to gather preliminary validation data for the use of RGT in evaluating MDT working. Further studies could identify whether different constructs (attributes) are needed by members of MDTs who are not FY1 doctors, to enable them to work well in MDTs, and to see whether there are ‘team’ constructs that all MDT members need for good team working, as well as to test the current grid in other settings. Work is also required to explore the rationale underlying the intriguing finding that the team considered all constructs to be important, but when forced into a ranking, chose the clinical competencies as more important. We believe that it is important to explore further what would be lost if these constructs were missing from a team (eg. because clinical competence is seen as the only important attribute in practice). For instance, in one interview, a senior nurse stated that she strongly believed that a construct such as understanding the expertise in the team is critical if team members are to support each other and utilise the strengths of the members effectively. Another important issue is what constructs (attributes) might be missing from the list identified in this research. For instance, there is nothing in the list of constructs we identified about ‘having confidence in reporting mistakes-vs-being afraid to own up to errors I have made’.
Conclusions
RGT has for long been used in connection with the education of professionals22 and team development23 but not, as far as we can determine, for MDT development, though Mazhindu and Pope24 used RGT to explore the case for interprofessional education for nurses. Its flexibility is such that it could be used to create a team development tool for health professionals working in any setting. As we have illustrated, RGT may be used to create a standardised competence-assessment instrument that could guide the development of MDT skills for a particular professional group, FY1 doctors. Alternatively, because the composition of multidisciplinary teams and the cultures of hospitals in which they work may vary, another application of our research is in using RGT to create customised instruments that individual teams then use themselves to identify, evaluate and develop the specific attributes (constructs) needed for good team working in their particular multidisciplinary team.
As stated in a policy synthesis review in Canada, ‘Teams function differently depending on where they operate ... This wide variety of settings and tasks means that transferability of processes is not always straightforward.’25 Similarly, attributes needed by new doctors may vary somewhat, or be expressed differently by teams. By using RGT, or by adapting the grid in this study to local requirements or language, the information thereby collected could be used as a sophisticated method to help build an optimally functioning team, with feedback and training adapted to their own needs. The instrument would permit individual multidisciplinary teams to explore (1) what the attributes so identified mean to them, (2) what ‘level’ of each attribute each team member possesses and (3) how the team might cooperatively develop an educational plan to improve the working of the team and to identify support and training needs for individual MDT members.
Acknowledgments
Our thanks are due to the staff, past and present, at Ward 2 in Goole and District Hospital, where HYMS students undertook interprofessional training, and where our consideration of assessment of FY1 doctors’ MDT performance began; to the MDT members who assisted in construct elicitation; to the clinical audit teams and MDT participants in the study hospitals; and to HYMS for partial financial support.
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online appendix
Footnotes
-
Contributors PM and JM initiated the study. NR provided expertise throughout on repertory grid methods and analysis. PM, JM, NR, SH and CD-W conducted interviews and categorised into final constructs. SH, CD-W and TM collected the majority of grid data under the supervision of PM. JM conducted the statistical analysis. PM, JM and NR wrote the initial draft of the paper and all authors contributed comments to final version.
-
Funding A small amount of development funding for evaluation and research was given by Hull York Medical School.
-
Competing interests None.
-
Ethics approval North and East Yorkshire Alliance Research and Development Unit, Hull and East Yorkshire NHS Hospitals Clinical Governance Unit and North Lincolnshire and Goole Hospitals NHS Foundation Trust Clinical Audit Departments.
-
Provenance and peer review Not commissioned; externally peer reviewed.