Patients were found to understand medical risks and test results more accurately when the risks were presented with numbers instead of words, consistent denominators, absolute differences, and visuals showing part-to-whole relationships. Providing clinical context further improved interpretation, particularly for unfamiliar values.
Verbal probability words such as “rare” or “common” were applied inconsistently. A review of 24 studies showed wide variation, with the study participants assigning probabilities ranging from 0% to 80% for “rare” and from 10% to 100% for “common.” Pooled analyses demonstrated that verbal descriptions often led patients to overestimate likelihoods and influenced behavioral intentions. Most of the patients preferred numeric or combined numeric and verbal formats.
Risk estimates expressed as “1 in X” were commonly judged larger compared with equivalent per-1,000 formats. In an outpatient study of 633 female participants, 73% of them correctly identified that 8.9 per 1,000 was greater than 2.6 per 1,000, while only 56% did so when the same risks were presented as 1 in 112 vs 1 in 384.
Relative differences made treatments appear more effective than absolute differences. A review of 17 studies showed that relative formats increased behavioral intentions and perceived effectiveness. However, fewer than one-third of 55 cancer guideline recommendations presented absolute effects for both benefits and harms. More than 50% used asymmetric formats, reporting benefits as relative reductions while harms were expressed as absolute increases.
Further, visual formats influenced perception. Graphs that displayed both numerators and denominators such as icon arrays or stacked bar charts produced smaller perceived probabilities than numerator-only visuals. In a study of 1,931 US adults, numerator-only graphics led participants to rate disease likelihood and vaccine protection as higher compared with part-to-whole visuals.
Adding clinical context improved the participants' interpretation of laboratory results. In a study of 1,618 adults, including a threshold line on a creatinine chart reduced concern for near-normal values but didn't change responses to clearly abnormal ones. Clinical targets and thresholds for outcomes such as hemoglobin A1c and blood pressure also helped patients interpret the results more accurately.
“Poor number communication may lead to patient misunderstanding, confusion, and/or frustration,” said lead study author Brian J. Zikmund-Fisher, PhD, MA, of the University of Michigan, and colleagues.
In the analysis, the investigators drew from systematic reviews, randomized trials, guideline assessments, and national survey data. These included pooled findings incorporating thousands of patients across multiple settings.
The investigators acknowledged that the evidence base varied in quality and scope. Many studies focused on US patients, limiting generalizability. Several relied on hypothetical scenarios rather than real clinical decisions, which may have affected outcomes.
The findings supported using numbers with consistent denominators, presenting absolute rather than relative differences, and adding contextual anchors to improve patient comprehension and engagement in care.
Full disclosures can be found in the published study.
Source: JAMA Insights
