Devin M. Mann

Trends in Statin Use and Low-Density Lipoprotein Cholesterol Levels Among US Adults: Impact of the 2001 National Cholesterol Education Program Guidelines

Background Few data are available on the use of statins after publication of the National Cholesterol Education Program Third Adult Treatment Panel (ATP-III) guidelines in 2001. Objective To determine changes in statin use and its Impact on low-density lipoprotein cholesterol (LDL-C) control among US adults from 1999 to 2004. Methods High LDL-C levels and statin use among 1911 participants of the National Health and Nutrition Examination Survey (NHANES) 2003–2004 were determined and compared with 1770 and 2094 participants of NHANES 1999–2000 and NHANES 2001–2002, respectively. Statin use was obtained from review of participants’ drug containers. High LDL-C levels and LDL-C control were defined, using risk-specific cut-points from the ATP-III guidelines. Results: Statins were taken by 24 million Americans in 2003–2004, an increase from 12.5 million in 1999–2000. In 1999–2000, 2001–2002, and 2003–2004, statins were being used by 19.6%, 27.3%, and 35.9% of US adults with high LDL-C levels, respectively (p trend <0.001). Age-standardized mean LDL-C declined from 119.9 to 112.0 to 100.7 mg/dL among statin users between 1999–2000, 2001–2002, and 2003–2004. LDL-C control to ATP-III recommended targets was achieved by 49.7%, 67.4%, and 77.6% of statin users in 1999–2000, 2001–2002, and 2003–2004, respectively (p trend <0.001). Among US adults with high LDL-C, after multivariate adjustment, non-Hispanic blacks were 39% less likely (prevalence ratio = 0.61; 95 CI 0.39 to 0.97) than non-Hispanic whites to be taking statins. Conclusions: Statin use continues to increase among US adults and this has led to substantial improvements in LDL-C control. Nevertheless, suboptimal statin use, especially among racial/ethnic minorities, continues to prevent the maximal public health benefit from this effective drug class.

Predictors of Nonadherence to Statins: A Systematic Review and Meta-Analysis

Background: Nonadherence to statins limits the benefits of this common drug class. Individual studies assessing predictors of nonadherence haue produced inconsistent results. Objective: To identify reliable predictors of nonadherence to statins through systematic review and meta-analysis. Methods: Multiple databases, including MEDLINE, EMBASE, and PsycINFO, were searched (from inception through February 2009) to identify studies that evaluated predictors of nonadherence to statins. Studies were selected using a priori defined criteria, and each study was reviewed by 2 authors who abstracted data on study characteristics and outcomes. Relative risks were then pooled, using an inverse-variance weighted random-effects model. Results: Twenty-two cohort studies met inclusion criteria. Age had a U-shaped association with adherence; the oldest (≥70 years) and youngest (<50 years) subjects had lower adherence than the middle-aged (50-69 years) subjects. Women and patients with lower incomes were more likely to be nonadherent than were men (odds of nonadherence 1.07; 95% CI 1.04 to 1.11) and those with higher incomes (odds of nonadherence 1.18:95% CI 1.10 to 1.28), respectively. A history of cardiovascular disease predicted better adherence to statins (odds of nonadherence 0.68; 95% CI 0.66 to 0.78). Similarly, a diagnosis of hypertension or diabetes was associated with better adherence. Although there were too few studies for quantitative pooling, increased testing of lipid levels and lower out-of-pocket costs appeared to be associated with better adherence. There was substantial (l2 range 68.7-96.3%) heterogeneity between studies across factors. Conclusions: Several sociodemographic, medical, and health-care utilization characteristics are associated with statin nonadherence. These factors may be useful guides for targeting statin adherence interventions.

Evidence-based strategies in weight-loss mobile apps

BACKGROUND Physicians have limited time for weight-loss counseling, and there is a lack of resources to which they can refer patients for assistance with weight loss. Weight-loss mobile applications (apps) have the potential to be a helpful tool, but the extent to which they include the behavioral strategies included in evidence-based interventions is unknown. PURPOSE The primary aims of the study were to determine the degree to which commercial weight-loss mobile apps include the behavioral strategies included in evidence-based weight-loss interventions, and to identify features that enhance behavioral strategies via technology. METHODS Thirty weight-loss mobile apps, available on iPhone and/or Android platforms, were coded for whether they included any of 20 behavioral strategies derived from an evidence-based weight-loss program (i.e., Diabetes Prevention Program). Data on available apps were collected in January 2012; data were analyzed in June 2012. RESULTS The apps included on average 18.83% (SD=13.24; range=0%-65%) of the 20 strategies. Seven of the strategies were not found in any app. The most common technology-enhanced features were barcode scanners (56.7%) and a social network (46.7%). CONCLUSIONS Weight-loss mobile apps typically included only a minority of the behavioral strategies found in evidence-based weight-loss interventions. Behavioral strategies that help improve motivation, reduce stress, and assist with problem solving were missing across apps. Inclusion of additional strategies could make apps more helpful to users who have motivational challenges.

Impact of A1C screening criterion on the diagnosis of pre-diabetes among U.S. adults.

OBJECTIVE New clinical practice recommendations include A1C as an alternative to fasting glucose as a diagnostic test for identifying pre-diabetes. The impact of these new recommendations on the diagnosis of pre-diabetes is unknown. RESEARCH DESIGN AND METHODS Data from the National Health and Nutrition Examination Survey 1999–2006 (n = 7,029) were analyzed to determine the percentage and number of U.S. adults without diabetes classified as having pre-diabetes by the elevated A1C (5.7–6.4%) and by the impaired fasting glucose (IFG) (fasting glucose 100–125 mg/dl) criterion separately. Test characteristics (sensitivity, specificity, and positive and negative predictive values) using IFG as the reference standard were calculated. RESULTS The prevalence of pre-diabetes among U.S. adults was 12.6% by the A1C criterion and 28.2% by the fasting glucose criterion. Only 7.7% of U.S. adults, reflecting 61 and 27% of those with pre-diabetes by A1C and fasting glucose, respectively, had pre-diabetes according to both definitions. A1C used alone would reclassify 37.6 million Americans with IFG to not having pre-diabetes and 8.9 million without IFG to having pre-diabetes (46.5 million reclassified). Using IFG as the reference standard, pre-diabetes by the A1C criterion has 27% sensitivity, 93% specificity, 61% positive predictive value, and 77% negative predictive value. CONCLUSIONS Using A1C as the pre-diabetes criterion would reclassify the pre-diabetes diagnosis of nearly 50 million Americans. It is imperative that clinicians and health systems understand the differences and similarities in using A1C or IFG in diagnosis of pre-diabetes.

Overweight, Obesity, and Elevated Serum Cystatin C Levels in Adults in the United States

BACKGROUND Although high body mass index (BMI) is a risk factor for hypertension, diabetes, and cardiovascular disease, limited data exist on the association of overweight and obesity with early stages of kidney disease. METHODS Cross-sectional data for 5083 participants of the nationally representative Third National Health and Nutrition Examination Survey with an estimated glomerular filtration rate > or = 60 mL/min/1.73 m(2) without micro- or macroalbuminuria were analyzed to determine the association between BMI and elevated serum cystatin C. Normal weight, overweight, class I obesity, and class II to III obesity were defined as a BMI of 18.5 to 24.9 kg/m(2), 25.0 to 29.9 kg/m(2), 30.0 to 34.9 kg/m(2), and > or = 35.0 kg/m(2), respectively. Elevated serum cystatin C was defined as > or = 1.09 mg/L (> or = 99th percentile for participants 20-39 years of age without diabetes, hypertension, micro- or macroalbuminuria, or stage 3-5 chronic kidney disease). RESULTS The age-standardized prevalence of elevated serum cystatin C was 9.6%, 12.9%, 17.4%, and 21.5% among adults of normal weight, overweight, class I obesity, and class II to III obesity, respectively (P trend < .001). After multivariate adjustment for demographics, behaviors, systolic blood pressure, and serum biomarkers, and compared with participants of normal weight, the odds ratio (95% confidence interval) of elevated serum cystatin C was 1.46 (1.02-2.10) for overweight, 2.36 (1.56-3.57) for class I obesity, and 2.82 (1.56-5.11) for class II to III obesity. CONCLUSION A graded association exists between higher BMI and elevated serum cystatin C. Further research is warranted to assess whether reducing BMI favorably affects elevated serum cystatin C and the development of chronic kidney disease.

Assessing the validity of self-reported medication adherence among inner-city asthmatic adults: the Medication Adherence Report Scale for Asthma

BACKGROUND A validated tool to assess adherence with inhaled corticosteroids (ICS) could help physicians and researchers determine whether poor asthma control is due to poor adherence or severe intrinsic asthma. OBJECTIVE To assess the performance of the Medication Adherence Report Scale for Asthma (MARS-A), a 10-item, self-reported measure of adherence with ICS. METHODS We interviewed 318 asthmatic adults receiving care at 2 inner-city clinics. Self-reported adherence with ICS was measured by MARS-A at baseline and 1 and 3 months. ICS adherence was measured electronically in 53 patients. Electronic adherence was the percentage of days patients used ICS. Patients with a mean MARS-A score of 4.5 or higher or with electronic adherence of more than 70% were defined as good adherers. We assessed internal validity (Cronbach alpha, test-retest correlations), criterion validity (associations between self-reported adherence and electronic adherence), and construct validity (correlating self-reported adherence with ICS beliefs). RESULTS The mean patient age was 47 years; 40% of patients were Hispanic, 40% were black, and 18% were white; 53% had prior asthma hospitalizations; and 70% had prior oral steroid use. Electronic substudy patients were similar to the rest of the cohort in age, sex, race, and asthma severity. MARS-A had good interitem correlation in English and Spanish (Cronbach alpha = 0.85 and 0.86, respectively) and good test-retest reliability (r = 0.65, P < .001). According to electronic measurements, patients used ICS 52% of days. Continuous MARS-A scores correlated with continuous electronic adherence (r = 0.42, P<.001), and dichotomized high self-reported adherence predicted high electronic adherence (odds ratio, 10.6; 95% confidence interval, 2.5-44.5; P < .001). Construct validity was good, with self-reported adherence higher in those saying daily ICS use was important and ICS were controller medications (P = .04). CONCLUSIONS MARS-A demonstrated good psychometric performance as a self-reported measure of adherence with ICS among English- and Spanish-speaking, low-income, minority patients with asthma.

Evidence-Based Risk Communication: A Systematic Review

Shared decision making is a collaborative process that allows patients and medical professionals to consider the best scientific evidence available, along with patients values and preferences, to make health care decisions (1). A recent Institute of Medicine report concluded that although people desire a patient experience that includes deep engagement in shared decision making, there are gaps between what patients want and what they get (2). For patients to get the experience they want, providers must effectively communicate evidence about benefits and harms. To improve the decision-making process, the Institute of Medicine recommended development and dissemination of high-quality communication tools (2). New tools, however, must match patients numerical abilities, which are often limited. For example, in one study, as many as 40% of high school graduates could not perform basic numerical operations, such as converting 1% of 1000 to 10 of 1000. This collective statistical illiteracy is a major barrier to the interpretation of health statistics (3). Physicians may also find statistical information difficult to interpret and explain (4). Existing literature about methods of communicating benefits and harms is broad. One review, based on 19 studies, concluded that the choice of a specific graphic is not as important as whether the graphic frames the frequency of an event with a visual representation of the total population in which it occurs (5). Another review, involving a limited literature search, found that comprehension improved when using frequencies (such as 1 in 5) instead of event rates (such as 20%) and using absolute risk reductions (ARRs) instead of relative risk reductions (RRRs) (6). The review did not assess affective outcomes, such as patient satisfaction, and behavioral outcomes, such as changes in decision making. Yet another review identified strong evidence that patients misinterpret RRRs and supported the effectiveness of graphs in communicating harms (7). However, they did not examine the comparative effectiveness of such approaches. More narrowly focused Cochrane reviews examined the communication of risk specific to screening tests (8, 9); numerical presentations, such as ARRs, RRRs, and numbers needed to treat (NNTs) (10); and effects of decision aids (11). An expert commentary about effective risk communication recommended using plain language, icon arrays, and absolute risks and providing time intervals with risk information (12). A group of experts identified 11 key components of risk communication, including presenting numerical estimates in context with evaluative labels, conveying uncertainty, and tailoring estimates (13). The aim of this systematic review is to comprehensively examine the comparative effectiveness of all methods of communicating probabilistic information about benefits and harms to patients to maximize their understanding, satisfaction, and decision-making ability. Methods We developed and followed a plan for the review that included several searches and dual abstraction of study data using standardized abstraction forms. Data Sources and Study Selection We searched PubMed (1966 to March 2014), CINAHL, EMBASE, and the Cochrane Central Register of Controlled Trials (1966 to December 2011) using keywords and structured terms related to the concepts of patients; communication; riskbenefit; and outcomes, such as understanding or comprehension, preferences or satisfaction, and decision making. Supplement 1 shows the detailed search strategy. Supplement 1. Search Strategies We included cross-sectional or prospective, longitudinal trials that were published in English and had an active control group that recruited patients or healthy volunteers and compared any method of communicating probabilistic information with another method. We focused on different methods of communicating the same specific probabilities to eliminate any independent effects that could result from different probabilities being studied (for example, different magnitudes or directions of effect). Studies of personalized risks, which may vary from person to person, were included when participants were randomly assigned. When studies of personalized risks were not randomized, the risks were considered to differ between the groups and were excluded. No limits were placed on study size, location, or duration or on the nature of the communication method. When needed, we reviewed sources specified in the articles, such as Web sites, to directly review the interventions and determine whether probabilistic information was addressed. Studies of medical students, health professionals, and public health or mass media campaigns were excluded. One independent reviewer screened each title and abstract and excluded citations that were not original studies or were unrelated to probabilistic information. Two independent reviewers screened the full text of the remaining citations to identify eligible articles. Disagreements between the 2 reviewers were resolved by consensus, with a third reviewer arbitrating any unresolved disagreements. Data Extraction and Quality Assessment Two reviewers independently abstracted detailed information about the study population, interventions, primary outcomes, and risk of bias from each included study using a standardized abstraction form, which was developed a priori (Supplement 2). A third reviewer resolved any disagreements. We categorized outcomes in 1 of 3 domains: cognitive (or understanding, such as accuracy in answering questions related to probabilistic information, or general comprehension of the probabilistic information), affective (such as preferences for or satisfaction with the method of communicating probabilistic information), and behavioral (such as real or theoretical decision making). Supplement 2. Abstraction Form Risk of bias in randomized, controlled trials was assessed on the basis of adequacy of randomization, allocation concealment, similarity of study groups at baseline, blinding, equal treatment of groups throughout the study, completeness of follow-up, and intention to treat (participants analyzed in the groups to which they were randomly assigned) (14). Risk of bias in observational studies was assessed with a modified set of criteria adapted from the NewcastleOttawa Scale (15). Data Synthesis and Analysis Data were tabulated, and the frequency of all head-to-head comparisons in studies was assessed to identify clusters of comparisons. In many instances, several interventions were bundled in a single study group (such as event rate plus icon array, or event rate plus natural frequencies plus ARRs). Bundles were not separated or combined with similar interventions because it could not be determined which component of the bundle drove the intervention. Descriptive statistics were used. We decided a priori not to do meta-analysis because of study heterogeneity. We emphasized findings from randomized studies as well as nonrandomized studies when findings were supported by more than 1 study. Role of the Funding Source No funding supported this study. The authors participated within their role on the Evidence-Based Medicine Task Force of the Society of General Internal Medicine. Results The initial search through December 2011 retrieved 22103 citations (16661 from PubMed, 1194 from CINAHL, 2861 from the Cochrane Central Register of Controlled Trials, and 1387 from EMBASE), and 20076 remained after removing duplicates. We updated the PubMed search through 30 March 2014, yielding 6529 additional citations; 5970 remained after removing duplicates, for a total of 26046 citations for review. A total of 630 articles were selected for full-text review and 84 were included, representing 91 unique studies (1699). Reasons for exclusion are noted in Figure 1, and study details are provided in Supplement 3. Figure 1. Summary of evidence search and selection. Supplement 3. Details of All Included Studies Seventy-four (81.3%) of the 91 included studies were randomized trials, most with cross-sectional designs. The median number of participants in randomized trials was 268 (range, 31 to 4685), and the median in all studies was 268 (range, 24 to 16133). Thirty-three studies (36.3%) included patients at specific risk for the target condition of interest. Forty-eight studies (52.7%) presented probabilistic data about benefits of a therapy or intervention (with 7 [14.6%] also presenting harms), 21 (23.1%) presented data only on harms, and 9 (10%) involved screening tests. Forty-nine studies (54.4%) delivered interventions on paper and 39 (42.9%) on a computer, typically over the Internet. The characteristics of study participants are presented in Tables 1 and 2. Table 1. Characteristics of Study Participants Table 2. Proportion of Studies Including Participants at Risk Versus Not at Risk for Target Condition Risk of bias for the included randomized trials was moderate (Figure 2). Randomization was adequate in 32 trials (42.7%), inadequate in 3 (4.0%), and unclear in 40 (53.3%). Allocation concealment was not stated in 55 trials (73.3%). Similarity of groups at baseline was adequate in 37 trials (49.3%) and unclear in 32 (42.7%). Blinding, equal treatment, and intention-to-treat items were similarly difficult to assess from reported information. Figure 2. Risk of bias for randomized, controlled trials (n = 74). Adapted from reference 100. Study Interventions and Comparators A frequency table (heat map) of all study intervention comparisons was created to identify clusters of comparisons (Supplement 4). The heat map represents study group comparisons, so one study may contribute several comparisons. The most commonly studied numerical presentations of data were natural frequencies, defined as the numbers of persons with events juxtaposed with a baseline denominator of persons (for example, 4 out of 100 persons had the outcome); event rates, defined as the proportions of persons wi

Meta-Analysis Impact of Drug Class on Adherence to Antihypertensives

Background— Observational studies suggest that there are differences in adherence to antihypertensive medications in different classes. Our objective was to quantify the association between antihypertensive drug class and adherence in clinical settings. Methods and Results— Studies were identified through a systematic search of English-language articles published from the inception of computerized databases until February 1, 2009. Studies were included if they measured adherence to antihypertensives using medication refill data and contained sufficient data to calculate a measure of relative risk of adherence and its variance. An inverse-variance–weighted random-effects model was used to pool results. Hazard ratios (HRs) and odds ratios were pooled separately, and HRs were selected as the primary outcome. Seventeen studies met inclusion criteria. The pooled mean adherence by drug class ranged from 28% for &bgr;-blockers to 65% for angiotensin II receptor blockers. There was better adherence to angiotensin II receptor blockers compared with angiotensin-converting enzyme inhibitors (HR, 1.33; 95% confidence interval, 1.13 to 1.57), calcium channel blockers (HR, 1.57; 95% confidence interval, 1.38 to 1.79), diuretics (HR, 1.95; 95% confidence interval, 1.73 to 2.20), and &bgr;-blockers (HR, 2.09; 95% confidence interval, 1.14 to 3.85). Conversely, there was lower adherence to diuretics compared with the other drug classes. The same pattern was present when studies that used odds ratios were pooled. After publication bias was accounted for, there were no longer significant differences in adherence between angiotensin II receptor blockers and angiotensin-converting enzyme inhibitors or between diuretics and &bgr;-blockers. Conclusion— In clinical settings, there are important differences in adherence to antihypertensives in separate classes, with lowest adherence to diuretics and &bgr;-blockers and highest adherence to angiotensin II receptor blockers and angiotensin-converting enzyme inhibitors. However, adherence was suboptimal regardless of drug class.Background— Observational studies suggest that there are differences in adherence to antihypertensive medications in different classes. Our objective was to quantify the association between antihypertensive drug class and adherence in clinical settings. Methods and Results— Studies were identified through a systematic search of English-language articles published from the inception of computerized databases until February 1, 2009. Studies were included if they measured adherence to antihypertensives using medication refill data and contained sufficient data to calculate a measure of relative risk of adherence and its variance. An inverse-variance–weighted random-effects model was used to pool results. Hazard ratios (HRs) and odds ratios were pooled separately, and HRs were selected as the primary outcome. Seventeen studies met inclusion criteria. The pooled mean adherence by drug class ranged from 28% for β-blockers to 65% for angiotensin II receptor blockers. There was better adherence to angiotensin II receptor blockers compared with angiotensin-converting enzyme inhibitors (HR, 1.33; 95% confidence interval, 1.13 to 1.57), calcium channel blockers (HR, 1.57; 95% confidence interval, 1.38 to 1.79), diuretics (HR, 1.95; 95% confidence interval, 1.73 to 2.20), and β-blockers (HR, 2.09; 95% confidence interval, 1.14 to 3.85). Conversely, there was lower adherence to diuretics compared with the other drug classes. The same pattern was present when studies that used odds ratios were pooled. After publication bias was accounted for, there were no longer significant differences in adherence between angiotensin II receptor blockers and angiotensin-converting enzyme inhibitors or between diuretics and β-blockers. Conclusion— In clinical settings, there are important differences in adherence to antihypertensives in separate classes, with lowest adherence to diuretics and β-blockers and highest adherence to angiotensin II receptor blockers and angiotensin-converting enzyme inhibitors. However, adherence was suboptimal regardless of drug class. # Clinical Perspective {#article-title-43}

The Statin Choice decision aid in primary care: a randomized trial.

OBJECTIVE To assess the impact of a decision aid on perceived risk of heart attacks and medication adherence among urban primary care patients with diabetes. METHODS We randomly allocated 150 patients with diabetes to participate in a usual primary care visit either with or without the Statin Choice tool. Participants completed a questionnaire at baseline and telephone follow-up at 3 and 6 months. RESULTS Intervention patients were more likely to accurately perceive their underlying risk for a heart attack without taking a statin (OR: 1.9, CI: 1.0-3.8) and with taking a statin (OR: 1.4, CI: 0.7-2.8); a decline in risk overestimation among patients receiving the decision aid accounts for this finding. There was no difference in statin adherence at 3 or 6 months. CONCLUSION A decision aid about using statins to reduce coronary risk among patients with diabetes improved risk communication, beliefs, and decisional conflict, but did not improve adherence to statins. PRACTICE IMPLICATIONS Decision aid enhanced communication about the risks and benefits of statins improved patient risk perceptions but did not alter adherence among patients with diabetes.

Evidence-based risk communication

Shared decision making is a collaborative process that allows patients and medical professionals to consider the best scientific evidence available, along with patients values and preferences, to make health care decisions (1). A recent Institute of Medicine report concluded that although people desire a patient experience that includes deep engagement in shared decision making, there are gaps between what patients want and what they get (2). For patients to get the experience they want, providers must effectively communicate evidence about benefits and harms. To improve the decision-making process, the Institute of Medicine recommended development and dissemination of high-quality communication tools (2). New tools, however, must match patients numerical abilities, which are often limited. For example, in one study, as many as 40% of high school graduates could not perform basic numerical operations, such as converting 1% of 1000 to 10 of 1000. This collective statistical illiteracy is a major barrier to the interpretation of health statistics (3). Physicians may also find statistical information difficult to interpret and explain (4). Existing literature about methods of communicating benefits and harms is broad. One review, based on 19 studies, concluded that the choice of a specific graphic is not as important as whether the graphic frames the frequency of an event with a visual representation of the total population in which it occurs (5). Another review, involving a limited literature search, found that comprehension improved when using frequencies (such as 1 in 5) instead of event rates (such as 20%) and using absolute risk reductions (ARRs) instead of relative risk reductions (RRRs) (6). The review did not assess affective outcomes, such as patient satisfaction, and behavioral outcomes, such as changes in decision making. Yet another review identified strong evidence that patients misinterpret RRRs and supported the effectiveness of graphs in communicating harms (7). However, they did not examine the comparative effectiveness of such approaches. More narrowly focused Cochrane reviews examined the communication of risk specific to screening tests (8, 9); numerical presentations, such as ARRs, RRRs, and numbers needed to treat (NNTs) (10); and effects of decision aids (11). An expert commentary about effective risk communication recommended using plain language, icon arrays, and absolute risks and providing time intervals with risk information (12). A group of experts identified 11 key components of risk communication, including presenting numerical estimates in context with evaluative labels, conveying uncertainty, and tailoring estimates (13). The aim of this systematic review is to comprehensively examine the comparative effectiveness of all methods of communicating probabilistic information about benefits and harms to patients to maximize their understanding, satisfaction, and decision-making ability. Methods We developed and followed a plan for the review that included several searches and dual abstraction of study data using standardized abstraction forms. Data Sources and Study Selection We searched PubMed (1966 to March 2014), CINAHL, EMBASE, and the Cochrane Central Register of Controlled Trials (1966 to December 2011) using keywords and structured terms related to the concepts of patients; communication; riskbenefit; and outcomes, such as understanding or comprehension, preferences or satisfaction, and decision making. Supplement 1 shows the detailed search strategy. Supplement 1. Search Strategies We included cross-sectional or prospective, longitudinal trials that were published in English and had an active control group that recruited patients or healthy volunteers and compared any method of communicating probabilistic information with another method. We focused on different methods of communicating the same specific probabilities to eliminate any independent effects that could result from different probabilities being studied (for example, different magnitudes or directions of effect). Studies of personalized risks, which may vary from person to person, were included when participants were randomly assigned. When studies of personalized risks were not randomized, the risks were considered to differ between the groups and were excluded. No limits were placed on study size, location, or duration or on the nature of the communication method. When needed, we reviewed sources specified in the articles, such as Web sites, to directly review the interventions and determine whether probabilistic information was addressed. Studies of medical students, health professionals, and public health or mass media campaigns were excluded. One independent reviewer screened each title and abstract and excluded citations that were not original studies or were unrelated to probabilistic information. Two independent reviewers screened the full text of the remaining citations to identify eligible articles. Disagreements between the 2 reviewers were resolved by consensus, with a third reviewer arbitrating any unresolved disagreements. Data Extraction and Quality Assessment Two reviewers independently abstracted detailed information about the study population, interventions, primary outcomes, and risk of bias from each included study using a standardized abstraction form, which was developed a priori (Supplement 2). A third reviewer resolved any disagreements. We categorized outcomes in 1 of 3 domains: cognitive (or understanding, such as accuracy in answering questions related to probabilistic information, or general comprehension of the probabilistic information), affective (such as preferences for or satisfaction with the method of communicating probabilistic information), and behavioral (such as real or theoretical decision making). Supplement 2. Abstraction Form Risk of bias in randomized, controlled trials was assessed on the basis of adequacy of randomization, allocation concealment, similarity of study groups at baseline, blinding, equal treatment of groups throughout the study, completeness of follow-up, and intention to treat (participants analyzed in the groups to which they were randomly assigned) (14). Risk of bias in observational studies was assessed with a modified set of criteria adapted from the NewcastleOttawa Scale (15). Data Synthesis and Analysis Data were tabulated, and the frequency of all head-to-head comparisons in studies was assessed to identify clusters of comparisons. In many instances, several interventions were bundled in a single study group (such as event rate plus icon array, or event rate plus natural frequencies plus ARRs). Bundles were not separated or combined with similar interventions because it could not be determined which component of the bundle drove the intervention. Descriptive statistics were used. We decided a priori not to do meta-analysis because of study heterogeneity. We emphasized findings from randomized studies as well as nonrandomized studies when findings were supported by more than 1 study. Role of the Funding Source No funding supported this study. The authors participated within their role on the Evidence-Based Medicine Task Force of the Society of General Internal Medicine. Results The initial search through December 2011 retrieved 22103 citations (16661 from PubMed, 1194 from CINAHL, 2861 from the Cochrane Central Register of Controlled Trials, and 1387 from EMBASE), and 20076 remained after removing duplicates. We updated the PubMed search through 30 March 2014, yielding 6529 additional citations; 5970 remained after removing duplicates, for a total of 26046 citations for review. A total of 630 articles were selected for full-text review and 84 were included, representing 91 unique studies (1699). Reasons for exclusion are noted in Figure 1, and study details are provided in Supplement 3. Figure 1. Summary of evidence search and selection. Supplement 3. Details of All Included Studies Seventy-four (81.3%) of the 91 included studies were randomized trials, most with cross-sectional designs. The median number of participants in randomized trials was 268 (range, 31 to 4685), and the median in all studies was 268 (range, 24 to 16133). Thirty-three studies (36.3%) included patients at specific risk for the target condition of interest. Forty-eight studies (52.7%) presented probabilistic data about benefits of a therapy or intervention (with 7 [14.6%] also presenting harms), 21 (23.1%) presented data only on harms, and 9 (10%) involved screening tests. Forty-nine studies (54.4%) delivered interventions on paper and 39 (42.9%) on a computer, typically over the Internet. The characteristics of study participants are presented in Tables 1 and 2. Table 1. Characteristics of Study Participants Table 2. Proportion of Studies Including Participants at Risk Versus Not at Risk for Target Condition Risk of bias for the included randomized trials was moderate (Figure 2). Randomization was adequate in 32 trials (42.7%), inadequate in 3 (4.0%), and unclear in 40 (53.3%). Allocation concealment was not stated in 55 trials (73.3%). Similarity of groups at baseline was adequate in 37 trials (49.3%) and unclear in 32 (42.7%). Blinding, equal treatment, and intention-to-treat items were similarly difficult to assess from reported information. Figure 2. Risk of bias for randomized, controlled trials (n = 74). Adapted from reference 100. Study Interventions and Comparators A frequency table (heat map) of all study intervention comparisons was created to identify clusters of comparisons (Supplement 4). The heat map represents study group comparisons, so one study may contribute several comparisons. The most commonly studied numerical presentations of data were natural frequencies, defined as the numbers of persons with events juxtaposed with a baseline denominator of persons (for example, 4 out of 100 persons had the outcome); event rates, defined as the proportions of persons wi