Scope

The search of the literature for this review began with a broad reading around Narrative Medicine to achieve a good understanding of the theory and its suggested application in practice. The Cochrane and the evidence for policy and practice information and coordinating centre databases were searched, and it was found that no systematic reviews had previously been undertaken on the subject. Some literature reviews addressing the effectiveness of humanities in medical education were found but these did not explicitly explore Narrative Medicine. This is one of the first reviews to relate Narrative Medicine with compassionate care.

A systematic literature search was then performed using the databases of Sage publications, EBSCOhost and the Greenwich University library catalogue. Search terms were informed by the prior reading and included the following: Narrative Medicine and, in turn, creative writing or reflective writing or poetry or storytelling. The words ‘medical education’, ‘evaluation’ and ‘study’ were also used until saturation was reached and no new articles emerged. The search was limited to English language items published between 2000 and 2015 to ensure that up-to-date sources were obtained. Reference lists of identified material were also checked and a key author search was performed. This primary search identified 20 possible sources.

Inclusion and exclusion criteria

• Primary studies that demonstrated explicitly an attempt to evaluate an educational initiative related to Narrative Medicine. This could include a specific Narrative Medicine course or a creative and/or reflective writing initiative, including poetry.

• Studies targeted at clinicians and related to adult care only.

• Studies published after 2000.

• Studies that took the form of opinion or commentary pieces were excluded, as was grey literature or unpublished work.

• Other arts-based medical education such as film, photography, drama or theatre were excluded, as were any sources with no obvious educational element and any relating to children.

Nine studies were selected; one was quantitative, three combined quantitative and qualitative research and five were qualitative (box 1). The majority of the studies focused on medical students at different stages in their training as part of medical humanities education. There were three studies involving qualified physicians; however, there were none involving nurses or other members of the healthcare team as recommended in the literature. The majority of the research was from North America, but two studies were from the UK and one was from Taiwan, suggesting a wider appeal of Narrative Medicine than exclusively North America, although it remains limited at a global level.12 All the studies demonstrated positive outcomes associated with a Narrative Medicine intervention at a small scale and none included a patient-reported outcome measure. Each study was appraised using tools and guidance designed by the evidence for policy and practice information and coordinating centre14 and the critical appraisal skills programme.15

Quantitative study

In a quantitative study, Tsai and Ho16 used a quasi-experimental design to examine whether Narrative Medicine could enhance clinical performance as measured by objective, structured, clinical examination (OSCE). While this design lends itself to the comparison of a randomly selected intervention group with a case-matched control, this study did not employ a pre-intervention OSCE measurement in either group which might have strengthened the approach.17 A t-test of significance was conducted, which is suitable for a small sample with groups of unequal sizes.18 The results showed that the case group performed better on the two communication stations (p=0.03), but there was no difference across the 12 stations when taken as a whole (p=0.24). Tsai and Ho16 demonstrated that Narrative Medicine can improve performance in terms of communication and, critically, this is one of the few studies to have demonstrated an objective behavioural outcome. There are, however, significant limitations and questions regarding rigour, not least because the study is relatively brief. There is no explanation of volunteer recruitment, ethical approval and consent process or why a quasi-experimental design was chosen. Neither do the authors provide any details on how the training course was facilitated and by whom. There are no tables detailing the rating items or questions for the OSCE scores; hence, there is minimal ability to appraise the findings as reported. Adding a qualitative measure to the study might have increased the validity of the findings, incorporating students' experiences through an analysis of their narrative accounts. The authors recommend further studies to assess the impact of Narrative Medicine education on patient-reported outcomes and, on the basis of this study, plan to expand their training.

Combined quantitative and qualitative studies

In a combined study, Shapiro et al 19 mixed qualitative analysis of subjective, objective, assessment and plan notes (SOAP) with a quantitative analysis of OSCE outcomes. In contrast to Tsai and Ho,16 they measured learning rather than a behavioural outcome in the OSCEs. They used a five-point Likert scale response to one question concerning empathy and another addressing the impact on treatment plans. Likert scales are methods of ascertaining attitudes in research and they require 10–15 statements to be robust.17 Meanwhile, Shapiro et al 19 used only two statements, which is a significant limitation. Rather than t-test analysis, they use a non-parametric exact test (Wilcoxon signed rank test), which is also appropriate for a small sample but more appropriate where distribution is non-normal.17 This was also favoured by Misra-Hebert et al 20 in their analysis of empathy scores. The qualitative analysis appears robust with clear thematic analysis conducted and validated by both researchers. They were able to demonstrate statistical significance and suggested improved empathy from the SOAP notes analysis. However, the results must be interpreted with caution as their sample size was reduced by a change to the course set up during the study years. This prevented them from implementing the humanities education for the first half of the year in 2002–2003. The study was further limited by the absence of any control group and the lack of a long-term follow-up to measure any sustained impact. The results cannot be generalised beyond this particular medical school but are certainly transferrable. However, this research does seem to demonstrate an impact of a feasible amount of Narrative Medicine education on students' ability to empathise and plan care. Additionally, the study is well written with details and data presented in tables and appendices, thus allowing transparency in all aspects of the study. The authors sought appropriate ethics approval and acknowledged their funding source.

In their study, Misra-Hebert et al 20 applied the Jefferson scale of empathy (JSE) to 40 physicians pre-Narrative Medicine, intra-Narrative Medicine and post-Narrative Medicine education. This is a validated 20-item scale designed to assess empathy in medical students.21 It has been extensively tested and is therefore considered an appropriate tool for this study. The non-parametric method of statistical analysis suits the small sample and the findings showed significantly increased scores on the JSE in the intervention group but not in either of the control groups (p=0.02). However, this must be interpreted with caution as it is plausible that the JSE scores may have improved purely as a result of peer support from the Narrative Medicine group sessions. Credible qualitative analysis was made using a grounded theory approach involving a comprehensive iterative analysis and coding system of the reflective writings by all three researchers. The participants received Continuing Professional Development (CPD) accreditation. It is a small sample (n=40) within one institution and cannot therefore be generalised but instead potentially transferrable to other contexts as is the case with qualitative research. More significant results might be generated by a larger sample, making differences between the intervention group and the controls more apparent. The study is transparent with tables and appendices showing the data, coding form and details of reflective sessions offered. However, the authors do not express their personal interest in Narrative Medicine nor do they state in what way they are qualified to facilitate reflective writing sessions. Although ethics approval was granted, the authors do not specifically indicate that consent was sought from participants to publish extracts of their writing within the paper.

In another study, Lancaster et al 22 used a nominal group technique to evaluate a special study module (SSM) for medical students in literature and medicine. This is a robust method of evaluation involving group decision-making in social research.23 Although labour-intensive, this method increases the validity, as it has both quantitative and qualitative elements. However, the study is small with a self-selected sample of five cases. The use of ranking in the evaluation process results in an average view as opposed to a consensus, which limits the findings. In a similar way to the other studies involving modules for medical students, this study relates to a short intervention with no long-term follow-up or ability to demonstrate a behavioural outcome in students. The university clearly benefited from having a visiting lecturer with comprehensive experience of teaching Narrative Medicine. This raises feasibility issues for anyone seeking to replicate such an initiative and is likely to have favoured a more positive evaluation by the students. This study is unique in making specific mention of social theoretical perspectives that relate to Narrative Medicine and the potential to include this in future module design. The article includes comprehensive tabular information about the evaluation process and themes, thus affording transparency, although the authors do not make specific reference to their personal interest in Narrative Medicine.

Qualitative studies

In a qualitative study, Clandinin and Cave24 used parallel reflective charts in their study with four medical students. This approach is advocated by Charon9 as a powerful reflective tool for promoting Narrative Medicine. They describe and analyse one parallel chart using narrative methodology to demonstrate the outcomes of the study. Narrative research is rooted in contemporary humanism and suits the analysis of individual case studies such as this.25 Throughout the analysis, reference is made to the parallel chart and the dialogue from the discussion as evidence of the findings—many examples are quoted in an attempt to validate themes. The authors suggest that Narrative Medicine education for doctors facilitates reflection in practice and helps them develop their own professional identity, thus contributing to their personal growth. The study contains a small sample size and the researchers interpreted their data by ‘close reading’. The researchers are keen to promote narrative inquiry and are themselves skilled in narrative reflective practice, thus demonstrating the traditions in qualitative research of acknowledging and accepting that bias can exist. They make specific reference to supporting doctors with regard to the ethical concerns raised by the Narrative Medicine process. This is important and reflects the findings of other studies reporting the need for small groups well facilitated by skilled educators.26 ,27

In their study, Lazarus and Rosslyn28 evaluated a SSM where a significant component was devoted to poetry and literature and students were required to keep reflective journals. This is the study in which there was an unusual emphasis on strengthening students' ‘knowledge and understanding of people’s experiences and emotions’ when they are ill.28 In the first year, evaluation was conducted by group discussion which was used to develop a five-point Likert scale questionnaire for the second year (but with only five statements). However, the small number of evaluations in the second year (n=10) meant that no statistical analysis was possible, suggesting that the development of a Likert scale during the first year was an oversight as the small numbers would have been known in advance. Instead, the authors interpreted their data qualitatively by using a thematic analysis. The authors clearly state their interest and previous experience in relevant Narrative Medicine education and acknowledge a grant from the university with which to fund the project. No ethical approval is mentioned nor is there evidence of explicit consent from the students to include excerpts from their contributions in the paper. There is a clear recommendation, however, for long-term follow-up of educational initiatives like these and the measurement of actual rather than potential impact in clinical practice.

In the meantime, Arntfield et al 26 thoroughly evaluated a Narrative Medicine module for fourth-year medical students. The module was facilitated by lecturers experienced in Narrative Medicine and 12 students participated. The lectures included close reading of literature, reflective writing exercises and discussion within small groups. The authors evaluated the intervention using grounded theory involving an initial anonymous survey followed by a focus group. They used the emergent themes to develop a ‘concept map’, which is in keeping with a grounded theory approach in generating new concepts and theories from the data.17 They uniquely attempted some long-term follow-up using an email survey, although the response rate was low at 25%. They used an iterative thematic analysis of all the data together, coding independently and then collectively returning to the raw data to resolve any discrepancies. They increased the validity by using triangulation, returning to transcripts and notes taken during the groups. Uniquely, they explicitly identified the need to overcome a perceived counterculture. For example, Narrative Medicine was considered ‘fluffy’ and non-essential by peers, yet those who undertook the course unanimously believed that Narrative Medicine training would make them better doctors. The findings are consistent with the evidence of the benefits of Narrative Medicine education, corroborated in this literature review. The study accomplishes sound ethical considerations mentioned throughout the paper, including participants' consent, ethics approval and confidentiality within the group discussions. Funding is also acknowledged. The authors highlight the need to measure actual rather than perceived changes in attitude and behaviour but suggest that the appetite for this will depend on the extent to which Narrative Medicine education is incorporated into curricula.

In another study, Gull et al 29 conducted a pilot study evaluating creative writing workshops with hospital staff in the UK. They freely admit that they had no experience of delivering such an initiative but were seeking a more creative approach to teach medical students in future. Thus, a participant observation methodology is suitable for gaining insights into the first-hand experience.18 Uniquely, the participants were interdisciplinary, which the authors reported as a strength in their findings, echoing Charon's9 claim that Narrative Medicine fosters community and understanding of others. Interestingly, the most favoured workshop concerned expressions of illness and death in which the authors noted much self-reflection. Rather than seeing this as a strength, they report that it is something to be cautious of in future initiatives, citing the difficulties of managing emotions raised by Narrative Medicine methods. They warn against too much reflection within the process, even though this was highly valued in their findings.27 They conceded some difficulties in managing a relatively large mixed group, stating the need to revisit initial ground rules to help manage this. The researchers achieved some soundness in their results by having participated in, recorded, discussed and coded their themes; furthermore, the participants viewed their paper and were able to make comments and modifications prior to finalisation. There is no clear mention of whether ethics approval was sought or any funding allocated. As a result, the main lesson learnt here is that Narrative Medicine initiatives are enjoyable but must be well planned and carefully implemented.

Finally, DasGupta and Charon27 evaluated six reflective writing seminars undertaken with second-year medical students. The seminars were facilitated by DasGupta and evaluation was by questionnaire after the last seminar. No specific methodology is stated but a robust iterative thematic analysis was undertaken by the authors. Once again, the seminars were well received, a now common thread across most studies. Both positive and negative themes emerged in the analysis, for example, words such as ‘enlightened’, ‘relief’ and ‘healing’ and also ‘embarrassing’, ‘confusion’ and ‘vulnerability’. Most reported a perceived enhanced empathy, having gained insights from both patients and their own experience shared within the group. The sessions were recommended as part of the curriculum but within small safe groups, which is another recurrent finding in the literature. The study's size comprised n=11 questionnaires. Additionally, there was an increased response rate in the second year. Of note, all the participants were women although, in mitigation, the authors report that findings from other comparable seminars involving both genders have reported similar outcomes. Ethical considerations are emphasised in respect of ground rules and confidentiality afforded to participants in the groups, but there is no mention of consent to include quotes from their evaluations.

Thematic analysis

Four key themes emerged from the literature reviewed here and were consistently reported among the findings and discussion of the studies. They also reflect the broader literature on Narrative Medicine and its potential for developing a more person-centred approach to healthcare.