Table 1.

Articles on Verbal Probabilities and Characteristics Collected

Authors and year	Research question	Primary outcome measured	Sample size*	Sample description*	Health condition or situation	Severity of health event specified?	Study of EC† terms?	Risk of bias assessment
Lichtenstein and Newman 1967 27	To assess numerical estimates and symmetry of interpretation of “mirror image” pairs of terms (e.g., “quite likely”-“quite unlikely”)	Numerical estimates	188	Adult males	Not specified	No	No	A little concern
Budescu et al. 1985 28	To assess variability in the mapping of phrases to numbers	Numerical estimates	32	Faculty and graduate students of a university	Not specified	No	No	Moderate concern
Reagan et al. 198929	To map verbal probability words to numbers	Numerical estimates	100	Undergraduate students	Not specified	No	No	Moderate concern
Shaw and Dear 199030	To evaluate understanding of probability expressions and preference for receiving information	Numerical estimates, format preference	100	Adult female parents	Aspects of neonatal care	No	No	No concern
Weber and Hilton 199031	To examine the role of context in the interpretation of probability words	Numerical estimates	85	Undergraduate students	Varying disease types and side effects	Some specified as severe life-threatening events, others unspecified	No	Moderate concern
Freeman et al. 19926	To identify patients’ preferred risk language and physicians’ predictions about patient preferences	Format preference	208	Adult female patients with children from family practices	Vaccine risk	No	No	Moderate concern
Woloshin et al. 199432	To assess patients’ interpretation of probability terms	Numerical estimates; format preference	307	Adult patients from a family practice	Medication side effect or complication risk from procedure	Minor vs major complications	No	Moderate concern
Hallowell et al. 1997 33	To evaluate female patient preferences in formats used to present risk information during genetic counseling for breast and ovarian cancer	Format preference	43	Adult female patients presenting for genetic counseling in cancer clinic	Breast and ovarian cancer risks	No	No	A little concern
Franic et al. 200034	To evaluate format preference in patient medication package inserts	Format preference	74	Adult female patients from academic university	Adverse drug reactions	No	No	High concern
Biehl et al. 200135	To compare the interpretation of probability terms of adults with adolescents	Numerical estimates	34	Adults from a community center	Not specified	No	No	A little concern
Kaplowitz et al. 200236	To evaluate how patients want, request, and receive cancer prognosis information	Format preference	352	Patients from the American Cancer Society (ACS) mailing list in Michigan, US	Cancer prognosis information	No	No	A little concern
Berry et al. 200237	To assess the interpretation of verbal probability descriptors	Numerical estimates	268	Undergraduate and graduate students	Throat infection or ear infection; fictitious medication side effect	Mild vs severe side effects	Yes	Moderate concern
Berry et al. 200338	To compare the understanding of verbal and numerical descriptions of medication side effects	Numerical estimates	360	Adults from various public settings	Fictitious medication side effect	Mild vs severe side effects	Yes	A little concern
Budescu et al. 200339	To determine the directionality of probability phrases	Numerical estimates	27	Undergraduate students	Medical context; general medication administration	No	No	Moderate concern
Davey et al. 200340	To evaluate women’s understanding of diagnostic test results	Numerical estimates	37	Adult women who had previously participated in a population survey	Breast cancer risk	No	No	A little concern
Lobb et al. 200341	To evaluate how women wanted their risk of breast cancer to be described in consultation	Format preference	193	Adult women from cancer clinics	Breast cancer risk	No	No	Moderate concern
Berry et al. 200442	To evaluate people’s interpretation of EC verbal descriptors for medication side effect risks	Numerical estimates	188	Adults from various public places	Over-the-counter painkiller medication side effects	No	Yes	A little concern
Berry et al. 200443	To compare doctors’ and lay people’s interpretation of the EC verbal descriptors	Numerical estimates	134	Undergraduate and postgraduate students	Medication side effect	No	Yes	A little concern
Knapp et al. 200444	To explore whether the EC verbal descriptors effectively convey the risk of side effects	Numerical estimates	120	Adults from cardiac rehabilitation clinics following a recent admission	Medication side effects for cardiac medication	No	Yes	A little concern
Berry and Hochhauser 200645	To assess how verbal descriptors affect people’s perceptions of clinical trial participation risks	Numerical estimates	96	Adults from a train station	Fictional serious skin condition	No	No	A little concern
Hubal and Day 200646	To evaluate the understanding of verbal probability terms and effects of alternative formats	Numerical estimates	222	Undergraduate students	Medication side effect	No	No	Moderate concern
Young and Oppenheimer 200647	To assess how different formats of risk information influence medication compliance	Numerical estimates	120	Adult students from a university	Medication side effect	No	Yes	Moderate concern
France et al. 200848	To compare the understanding of frequency of side effects when expressed in percentages or descriptive language	Numerical estimates, % correctly identified	50	Patients in the chest pain unit of an urban emergency department who had one or more ischemic heart disease factors	Risks of treatment for acute myocardial infarction	Severe vs less severe side effects	No	No concern
Graham et al. 200949	To identify women’s preference and interpretation of language for description of the size of treatment complication risks	Format preference‡	262	Adult female patients undergoing routine follow-up visits for breast cancer	Breast cancer risk	No	No	A little concern
Knapp et al. 200950	To assess the effectiveness of presenting side effect risk information in different formats	Numerical estimates	148	Adult users of an online cancer information website	Medication side effect	No	No	No concern
Nagle et al. 200951	To evaluate female patients’ preference on risk of disease	Format preference	294	Adult female patients from a maternity unit	Down syndrome risk	No	No	No concern
Cheung et al. 201052	To compare patients’ preference for risk presentation in medications	Format preference	240	Adult patients from arthritis clinics in a hospital and outpatient practice	Pain relief medication	No	Yes	A little concern
Vahabi 201053	To evaluate whether format preference influences comprehension	Format preference	180	Adult female patients from various community settings	Breast cancer risk	No	No	High concern
Peters et al. 201454	To measure risk comprehension and willingness to use a medication when presented with different formats	Numerical estimates	905	Adult participants from a paid online questionnaire	Cholesterol medication	No	Yes	No concern
Knapp et al. 201455	To evaluate recommendations on communicating frequency information on side effect risk	Numerical estimates	339	Adult users of an online cancer information website	Medication side effects	No	No	A little concern
Webster et al. 201756	To assess how people interpret the EC verbal descriptors	Numerical estimates, % correctly identified	1003	Adult users of an online survey conducted by a market research company	Medication side effects	Mild vs severe side effects	Yes	A little concern
Carey et al. 2018 57	To assess patients’ interpretation of verbal descriptor chance of remission and preferences for format of risk communication	Numerical estimates, format preference	210	Adult medical oncology outpatients with a diagnosis of cancer	Cancer long-term side effects and chances of remission	No	No	A little concern
Wiles et al. 202058	To determine the perceived risk of surgical complication risk using verbal probability terms	Numerical estimates	290	Adult patients attending a pre-operative assessment in a clinic	Major adverse postoperative complication	No	No	No concern

*Several studies contained both subsamples that met our inclusion criteria (adult laypeople) and other subsamples that did not (physicians, adolescents). As described in the “METHODS” section, these studies were included if the results for the eligible subsample were reported separately. For these studies, we report the sample size and sample description of the subgroup that met our inclusion criteria

†EC = European Community

‡Graham et al. 2009 ⁴⁹ required respondents to choose from ordinal categories ranging from 1/100 to 1/10 000. The modal interpretation of “sometimes” was 1/100 (36% of women), “uncommon” 1/1000 (35%), “very uncommon” 1/10 000 (40%), “rare” 1/10 000 (58%) and “very rare” 1/10 000 (51%). Because of the categorical assessment and the fact that no larger numbers were provided to choose from, we did not average these results into the findings in Table 2.