Reading and skimming clinical information: insights from experiments examining medical students’ eye movement behaviour

Background: Reading times are shorter and comprehension is poorer during skim-reading compared to more careful reading for comprehension. Here we provide a novel examination of the effect of skimming on medical students’ reading of clinical texts. Three eye tracking experiments are reported. Each experiment manipulates the reading task (reading for comprehension, skimming for gist) and a key characteristic of the text (legibility, context, accuracy). Together the experiments provide key insights into how medical students read and skim clinical information.

Methods: Participants were fourth year medical students. In each experiment participants read for comprehension and skimmed for gist. Experiment 1 examined the effect of font legibility, comparing reading behaviour for vignettes presented in a legible vs. less legible font (n = 28). Experiment 2 examined the effect of contextual cues, comparing reading of clinical statements that were preceded by a neutral cue vs. a cue stating the diagnosis (n = 28). Experiment 3 examined the integration of the text with prior knowledge by comparing reading behaviour for statements that were accurate or inaccurate (n = 20). Eye movements were recorded to determine how reading processes differed according to reading strategy and the text manipulations.

Results: Across all three experiments skim-reading resulted in eye movement indices consistent with more superficial processing of text (shorter first-pass and re-reading times, ps < 0.001). There were fewer and shorter eye fixations during skimming compared to reading for comprehension (ps < 0.001) (Experiment 1). A less legible font was found to slow down reading (ps < 0.001), but did so similarly for skimming and reading for comprehension (Experiment 1). There were smaller effects of context (Experiment 2) and text accuracy (Experiment 3) for re-reading measures during skimming, indicating that skimming produces poorer integration of text with patient information or clinical knowledge.

Conclusions: The eye tracking results are consistent with previous work indicating that levels of comprehension can be reduced during skim-reading. The study also demonstrates that legibility and contextual cues (e.g., diagnosis sub-headings) are important for efficient reading. Especially when learning key concepts or making key decisions, medical students and healthcare practitioners should be aware that content may be missed or only superficially processed during skimming.