Difficult news is information that is likely to substantially and negatively change how a person understands their situation or future.
In healthcare, this may include conversations about:
These conversations require more than simply communicating information accurately. Students must also judge how much information to provide, how to explain it clearly, how to respond to emotion and uncertainty, and how to adapt the conversation to the person in front of them.
Structured approaches such as SPIKES can help novice students organise a difficult conversation. However, a framework should act as a scaffold rather than a script. A student can follow every step of a protocol and still communicate poorly if they fail to listen, recognise distress or respond appropriately when the conversation develops in an unexpected direction.
The aim is therefore not simply to teach students how to deliver bad news. It is to prepare them to manage an emotionally difficult professional interaction.
Educational interventions have a large positive effect on medical students’ ability to communicate difficult news ➕➕➕➕➕ [1]. Across controlled studies, students who received targeted training performed substantially better than comparison groups, although effects varied considerably between studies.
Students benefit from opportunities to practise difficult conversations. Simulation and simulated-patient approaches consistently produce improvements in communication performance, although there is not enough comparative evidence to conclude that one form of simulation—or simulation itself—is consistently superior to other well-designed approaches [1,2].
Structured frameworks can help students organise these complex conversations, but there is no convincing evidence that one particular framework is best [1,2]. The most defensible approach is therefore to provide students with a clear structure, then use realistic rehearsal, reflection and feedback to help them apply it flexibly [1,4].
The strongest evidence comes from a 2026 systematic review and meta-analysis focused exclusively on medical students [1]. The review searched four major databases, used duplicate screening and extraction, assessed risk of bias using RoB 2, ROBINS-I and AHRQ tools, and included 27 studies, of which 17 contributed to meta-analysis. The studies included nine randomised controlled trials, two non-randomised controlled trials and a range of pre-post and observational designs.
Seven controlled studies with 1,164 unique participants contributed to the principal comparison. Educational interventions produced a very large improvement in difficult-news communication ability (SMD = 1.09, 95% CI 0.52–1.66), but heterogeneity was very high (I² = 89%). Six of the nine RCTs were rated at low risk of bias, although non-randomised and observational evidence was considerably weaker.
Although the review authors conclude that simulation-based training is particularly effective, the detailed subgroup results are not sufficiently consistent to establish simulation as superior. In the controlled studies, the estimated effect was actually larger for lecture/video interventions (SMD = 1.56) than for simulated-patient/simulation approaches (SMD = 0.67), but the lecture/video estimate was highly heterogeneous (I² = 94%) whereas the simulation estimate was much more consistent. We therefore interpret the evidence as supporting active rehearsal, without claiming that one teaching format is best.
An earlier systematic review examined 29 studies using simulated patients with undergraduate medical students [2]. It found consistently positive outcomes but insufficient high-quality comparative evidence to determine whether actors, peers, virtual patients or another simulation approach is most effective.
A broader 2018 meta-analysis, which combined medical students with physicians and trainees, similarly found a large improvement in observer-rated difficult-news communication (SMD = 0.74) but was not restricted to university students and therefore provides supporting rather than primary evidence for this summary [3].
Importantly, the 2026 evidence is almost entirely concerned with short-term performance in simulated or controlled settings. The authors found little evidence about retention or whether training ultimately improves patients’ psychological outcomes, satisfaction or experiences of care.
