0
Articles |

Using a Drug Facts Box to Communicate Drug Benefits and Harms: Two Randomized Trials FREE

Lisa M. Schwartz, MD, MS; Steven Woloshin, MD, MS; and H. Gilbert Welch, MD, MPH
[+] Article and Author Information

ClinicalTrials.gov registration numbers: NCT00450931, NCT00753857.


From the Veterans Affairs Outcomes Group, White River Junction, Vermont, and the Center for Medicine in the Media at the Dartmouth Institute for Health Policy and Clinical Practice, Hanover, New Hampshire.


Note: Drs. Schwartz and Woloshin contributed equally to this article. The order of authorship is arbitrary.

Disclaimer: The views expressed herein do not necessarily represent the views of the Department of Veterans Affairs or the U.S. government.

Acknowledgment: The authors thank Andrew Oxman, MD, for advice on study design; Elliott Fisher, MD, MPH, and Robin Larson, MD, MPH, for helpful comments on an earlier draft; and the members of the Veterans Affairs Outcomes Group for their ongoing helpful feedback during all phases of the project.

Grant Support: Drs. Schwartz and Woloshin were supported by Robert Wood Johnson Generalist Faculty Scholars Awards. The study was supported by the National Cancer Institute (grant R01CA104721) and the Attorney General Consumer and Prescriber Education grant program.

Potential Financial Conflicts of Interest: Drs. Schwartz, Woloshin, and Welch are the authors of Know Your Chances: Understanding Health Statistics (Univ of California Pr, 2008). They have received no royalties or other payments for the book.

Reproducible Research Statement:Study protocols: Available at http://www.clinicaltrials.gov/ct2/showNCT00450931 and http://www.clinicaltrials.gov/ct2/showNCT00753857. Statistical code: Available from Dr. Woloshin (e-mail, steven.woloshin@dartmouth.edu). Data sets: Not available.

Requests for Single Reprints: Steven Woloshin, MD, MS, Veterans Affairs Outcomes Group (111B), Department of Veterans Affairs Medical Center, White River Junction, VT 05009; e-mail, steven.woloshin@dartmouth.edu.

Current Author Addresses: Dr. Schwartz, Woloshin, and Welch: Veterans Affairs Outcomes Group (111B), Department of Veterans Affairs Medical Center, White River Junction, VT 05009.

Author Contributions: Conception and design: L.M. Schwartz, S. Woloshin, H.G. Welch.

Analysis and interpretation of the data: L.M. Schwartz, S. Woloshin, H.G. Welch.

Drafting of the article: L.M. Schwartz, S. Woloshin.

Critical revision of the article for important intellectual content: L.M. Schwartz, S. Woloshin, H.G. Welch.

Final approval of the article: L.M. Schwartz, S. Woloshin, H.G. Welch.

Statistical expertise: L.M. Schwartz, S. Woloshin.

Obtaining of funding: S. Woloshin.

Administrative, technical, or logistic support: S. Woloshin.


Ann Intern Med. 2009;150(8):516-527. doi:10.7326/0003-4819-150-8-200904210-00106
Text Size: A A A

Background: Direct-to-consumer prescription drug ads typically fail to provide fundamental information that consumers need to make informed decisions: data on how well the drug works.

Objective: To see whether providing consumers with a drug facts box—a table quantifying outcomes with and without the drug—improves knowledge and affects judgments about prescription medications.

Design: Two randomized, controlled trials conducted between October 2006 and April 2007: a symptom drug box trial using direct-to-consumer ads for a histamine-2 blocker and a proton-pump inhibitor to treat heartburn, and a prevention drug box trial using direct-to-consumer ads for a statin and clopidogrel to prevent cardiovascular events.

Setting: National sample of U.S. adults identified by random-digit dialing.

Participants: Adults age 35 to 70 years who completed a mailed survey; the final samples comprised 231 participants with completed surveys in the symptom drug box trial (49% response rate) and 219 in the prevention drug box trial (46% response rate).

Intervention: In both trials, the control group received 2 actual drug ads (including both the front page and brief summary). The drug box group received the same ads, except that the brief summary was replaced by a drug facts box.

Measurements: Choice between drugs (primary outcome of the symptom drug box trial) and accurate perceptions of drug benefits and side effects (primary outcome of the prevention drug box trial).

Results: In the symptom drug box trial, 70% of the drug box group and 8% of the control group correctly identified the PPI as being “a lot more effective” than the histamine-2 blocker (P < 0.001), and 80% and 38% correctly recognized that the side effects of the 2 drugs were similar (P < 0.001). When asked what they would do if they had bothersome heartburn and could have either drug for free, 68% of the drug box group and 31% of the control group chose the proton-pump inhibitor, the superior drug (P < 0.001). In the prevention drug box trial, the drug box improved consumers' knowledge of the benefits and side effects of a statin and clopidogrel. For example, 72% of the drug box group and 9% of the control group correctly quantified the benefit (absolute risk reduction) of the statin (P < 0.001). Most of the control participants overestimated this benefit, and 65% did so by a factor of 10 or more.

Limitations: The trials tested drug boxes in only 4 direct-to-consumer ads. If other direct-to-consumer ads were to communicate outcome data better, the effect of the drug box would be reduced.

Conclusion: A drug facts box improved U.S. consumers' knowledge of prescription drug benefits and side effects. It resulted in better choices between drugs for current symptoms and corrected the overestimation of benefit in the setting of prevention.

Primary Funding Source: National Cancer Institute and Attorney General Consumer and Prescriber Education Program.

Editors' Notes
Context

  • Direct-to-consumer drug advertisements do not provide standardized information about the benefits and harms of drug therapies.

Contribution

  • These randomized trials tested whether adding a “drug facts box” to consumer prescription advertisements improved consumer knowledge and judgment. The facts box showed numbers of outcomes, including adverse events, which might occur with 2 alternative drug therapies. Consumers given advertisements that included the box had more accurate perceptions of drug benefits and side effects than consumers given advertisements without the box.

Implication

  • A standardized table quantifying outcomes that might occur with different therapies improved consumer knowledge of drug benefits and side effects.

—The Editors

Direct-to-consumer advertising surpasses all other efforts to alert the U.S. public about prescription drugs. Industry spent more than $5 billion on these ads in 2007 (1)—more than twice the total U.S. Food and Drug Administration (FDA) budget (2). Although the ads are often justified as serving an educational purpose, they generally fail to provide the most fundamental information consumers need to make informed decisions: data on how well the drug works.

To address this problem, we developed a “drug facts box,” a 1-page summary of a drug's benefit and side effects. The central information is provided in a table that shows the chance of various outcomes for people who do and do not take the drug. In previous studies, we showed that consumers understand and value this information (34).

We performed 2 randomized, controlled trials in nationally representative samples of U.S. adults (the target audience for direct-to-consumer advertising) to test whether a drug facts box improves consumer knowledge and helps people make better choices.

Design

We conducted 2 randomized, controlled trials with the same basic design from October 2006 through April 2007: the symptom drug box trial and the prevention drug box trial. We conducted a pair of trials because consumers generally face decisions about 2 categories of drugs: those taken to treat current symptoms and those taken to reduce the risk for future events (prevention). In the symptom drug box trial, we used direct-to-consumer ads for drugs to treat heartburn, an outcome (heartburn relief) that is common and experienced directly. We chose to include a proton-pump inhibitor (PPI) and a histamine-2 (H2)–blocker because both drugs are used to treat current symptoms of the same disease and have similar side effect profiles, but PPIs clearly outperform H2-blockers; we could therefore assess whether the box helped consumers choose the superior drug. In the prevention drug box trial, we used advertisements for a statin and clopidogrel for secondary cardiovascular prevention. These drugs are used to reduce the risk for future events that are important but relatively rare (heart attack and death), so absolute effects can only be small. The goal of this trial was to see whether the box improved the accuracy of consumers' perceptions about drugs used to reduce cardiovascular risk. Figure 1 shows the ads used in each trial.

Grahic Jump Location
Figure 1.
Advertisements given to trial participants.

Appendix 1 and Appendix 2 provide full-size images of all 8 ads. PPI = proton-pump inhibitor.

Grahic Jump Location

Committees for the protection of human subjects at Dartmouth Medical School and the University of Massachusetts approved the trials. Trial protocols were registered with ClinicalTrials.gov before recruitment began.

Sample Selection

Figure 2 summarizes participant flow into the study.

Random-Digit Dialing Protocol

The Center for Survey Research, University of Massachusetts (who conducted both surveys), created 2 national, random-digit dial samples, each with about 3000 telephone numbers.

Trained interviewers used a standard protocol in which they called each telephone number at least 6 times (different days, weeks, and times). Whenever someone answered, the interviewer explained that “we are conducting an important study about how people feel about different ways to provide information about prescription drugs in advertising.” English-speaking adults age 35 to 70 years in the household were invited to participate (if >1 adult was eligible, the computer—which guided these screening interviews—randomly selected one).

Randomization and Mail Survey Protocol

Participants were randomly assigned to the drug box or control group by using a central computerized random-number generator (to ensure allocation concealment) and were mailed survey materials (2 drug ads, the survey, and a stamped return envelope) plus a prepaid $10 incentive with a promise of an additional $10 upon survey completion. Participants were not told about the alternate survey form. Nonresponders received reminder postcards after 2 weeks, a replacement questionnaire after 4 weeks, a telephone call after 6 weeks, and a questionnaire after 8 weeks.

Response Rates

We used standard Association for Public Opinion Research methods to calculate overall response rates (5). The numerator is completed surveys (231 symptom drug box trial participants [122 in the drug box group and 109 in the control group] and 219 prevention drug box trial participants [111 in the drug box group and 108 in the control group]). The denominator is individuals known to be eligible (289 in the symptom drug box trial and 290 in the prevention drug box trial) plus an estimate of eligible individuals in households that could not be screened. In the symptom drug box trial, this number—184—was calculated by applying the 44.4% eligibility fraction observed among screened households to the 414 unscreened households. The overall response rates were 49% in the symptom drug box trial (231/[289 + 184]) and 46% in the prevention drug box trial (219/[290 + 182]).

Completion rates were calculated by using completed surveys as the numerator and participants as the denominator. For the symptom drug box trial, the completion rate was 89% (231/258); the drug box group had a higher completion rate than the control group (95% [122/129] vs. 84% [109/129]; P = 0.010). In the prevention drug box trial, the completion rate was 84% (219/260) and did not differ between groups.

Study Materials

We modified actual drug ads by disguising the name of the drug and manufacturer to avoid preconceived notions that participants might hold. Appendix 1 and Appendix 2 include all materials.

Ad Image Page

All participants received the same “ad image page” (the colorful front page). Participants in the symptom drug box trial received ads for an H2-blocker (“Amcid”) and a PPI (“Maxtor”). In the prevention drug box trial, the ads were for a statin (“Concor”) and clopidogrel (“Pridclo”).

Second Page

In both trials, the second page differed according to randomization. For control participants, the second page was the actual ad's “brief summary”: the drug label excerpt that the FDA requires in drug ads. The drug box groups received drug facts boxes. These boxes (Figure 3 shows the Pridclo [clopidogrel] box) include a descriptive section with basic drug information (such as indications and other things to consider doing). The central element is a summary table of efficacy and side effects presented as the chance of various outcomes for people who do or do not take the drug. We completed the boxes by using data from individual trials because that is how the FDA approaches drug approval; they do not use meta-analysis to judge efficacy.

Grahic Jump Location
Figure 3.
Pridclo (clopidogrel) drug facts box.

FDA = U.S. Food and Drug Administration.

Grahic Jump Location

To find efficacy data for the boxes in the symptom drug box trial, we used the Cochrane Collaboration review (6) on short-term treatment of gastroesophageal reflux disease–like symptoms to identify all placebo-controlled trials of PPIs and H2-blockers for empirical therapy. For each drug, we chose the largest of the trials with same outcome: “sustained relief,” which is no heartburn episodes for 3 consecutive days.

To facilitate comparisons, we standardized the base rate of sustained relief by averaging the findings in the placebo groups of the 2 studies (23% in the H2-blocker study [7] vs. 15% in the PPI study [8], yielding an average base rate of 19%). We calculated the relative risk (drug vs. placebo) for “sustained relief” in the H2-blocker (relative risk, 1.87) and PPI (relative risk, 4.67) trials by using the observed findings. To get standardized rates, we multiplied the averaged placebo rate by each relative risk. To frame the outcome as the proportion of patients still having heartburn, we used the complement of these calculated rates.

Side effect data for the drug boxes were abstracted from the corresponding brief summaries (given to control participants) to ensure that both groups had access to similar information. Although many side effects are listed throughout the brief summary, the drug box includes only the most important and largest relative to placebo or other comparator (organized into life-threatening and symptom categories) to focus the reader's attention and minimize cognitive burden.

To find efficacy data for the boxes in the prevention drug box trial, we began with the currently approved label for clopidogrel (9). The label cites a single trial supporting the use of clopidogrel for heart attack prevention: CAPRIE (Clopidogrel vs. Aspirin in Patients at Risk of Ischemic Events) (10). To find a similar study for the statin (in terms of sample size, high-risk population, and outcome of heart attack), we reviewed a meta-analysis of cholesterol-lowering treatment (11). We chose the Medical Research Council Heart Protection Study of simvastatin (12).

To facilitate comparisons, we standardized the base rates of heart attack and all-cause mortality by taking the average risk in the comparator groups of the 2 studies. To get the standardized risk in both drug groups, we multiplied the standardized base rate by the respective study relative risks. The Medical Research Council study had 5 years of follow-up, and the CAPRIE study had 2 years of follow-up. To avoid extrapolating the effect of clopidogrel, we presented 2-year risks.

Side effect data were taken from the brief summaries given to the control group.

Measurement and Outcomes

Outcomes were assessed by using a mailed survey booklet. Surveys had been cognitively tested with 5 members of the target population to learn whether the survey questions meant the same things to respondents and investigators, and the survey was refined on the basis of a pilot study of the entire protocol with 18 participants. Surveys took about 30 to 45 minutes to complete.

Symptom Drug Box Trial

The primary outcome entailed directly comparing the 2 heartburn drugs; specifically, participants answered the question, “Imagine you had bothersome heartburn. If you could take [either of the 2 drugs] for free, which drug would you rather take? (Amcid, Maxtor, Neither drug).” Maxtor (the PPI) was considered the correct answer because it had substantially greater benefit than and similar side effects to Amcid (the H2-blocker). Secondary outcomes were comprehension of information contained in the drug box and usability ratings.

Prevention Drug Box Trial

The primary outcome was accurate perceptions of benefits and side effects. To quantify perceived benefit, participants were asked a multiple-choice question about baseline risk: “What percent of people who do not take a drug to lower their cholesterol have a heart attack over 2 years?” and 2 fill-in-the-blank questions for the percentage of people who would have a heart attack if given Concor (statin) or if given Pridclo (clopidogrel). To calculate perceived absolute risk reductions, we subtracted the fill-in-the-blank answers for each drug from the baseline risk.

Responses in the drug box group were considered correct only if participants gave exactly the right numbers (because these appear in the box); control participants' responses were considered correct if their answers were about the right order of magnitude (2% to 3% for the baseline risk and 0% to 5% for the absolute risk reductions). We used this expansive definition of “correct” to bias the results in favor of control participants.

In both trials, the drug box surveys were longer (50 vs. 40 questions in the symptom drug box trial and 55 vs. 52 questions in the prevention drug box trial) because we tested comprehension of information only contained in the drug box. The 2 surveys were otherwise identical.

Statistical Analysis
Symptom Drug Box Trial

We asserted that a 25% absolute difference in the proportion of persons who correctly chose the superior drug would be clinically important (50% of the control group vs. 75% of the drug box group). With power set at 0.90 and a 2-sided P value of 0.05, we calculated that 85 people would be needed in each group. To allow for nonresponse, we aimed to recruit 120 people in each group.

Prevention Drug Box Trial

Our sample size calculation was based on data from a previous drug box study (4) in which the average proportion of participants who rated a statin or clopidogrel as “very effective” in preventing heart attacks was 60% before they saw the box and about 20% afterward. To be conservative, we asserted that a 25% absolute difference mattered and calculated that, with a power of 0.90 and a 2-sided P value of 0.05, we needed 90 people in each group. To allow for nonresponse, we aimed for 120 participants in each group.

Both Trials

In both trials, blank answers were considered incorrect for the primary outcomes and all comprehension questions. For usability ratings, blank answers were considered missing (the largest item nonresponse was 2%). We used the chi-square test to compare differences in proportions and the t test to compare means. All comparisons were 2-sided, and P values less than 0.05 were considered statistically significant. All analyses were done by using Stata, version 10 (StataCorp, College Station, Texas).

Role of the Funding Source

This study was supported by the Robert Wood Johnson Foundation's Generalist Physician Faculty Scholars Program, National Cancer Institute, and the Attorney General Consumer and Prescriber Education grant program. The funding sources had no role in study design, data collection and analysis, manuscript preparation, or the decision to submit the manuscript for publication.

The Table shows that the demographic characteristics of participants were similar within and across the symptom and prevention drug box trials. Average participant age was 54 years in both trials; 64% of symptom drug box trial participants and 61% of prevention drug box trial participants were female, and 41% and 32%, respectively, had a college or higher education. Compared with the U.S. population age 35 to 70 years, trial participants were about 4 years older, were more likely to be female and white, and had somewhat more formal education (1415).

Table Jump PlaceholderTable.  Participant Characteristics
Usability and Comprehension of the Drug Box

In the symptom and prevention drug box trials, most participants in the drug box group said that the information in the box's data table was “very easy” or “somewhat easy” to find (68% and 71%, respectively) and that it was “very easy” or “easy” to understand (52% and 69%, respectively). Most participants (79% in each trial) also felt that it was “extremely” or “very important” to have a data table (such as the drug box) appear in ads.

The drug box groups demonstrated good understanding of information in the data tables. In the symptom drug box trial, the drug box group was asked 9 questions that required extracting data from the benefit or side effect sections of the 6-row–by–3-column data tables. The proportion of correct answers ranged from 65% to 89% (Appendix Table 1). In the prevention drug box trial, the drug box group was asked 12 comprehension questions requiring them to extract data from the 6-row–by–3-column and 9-row–by–3-column tables. The proportion of correct answers ranged from 55% to 93% (Appendix Table 2).

Table Jump PlaceholderAppendix Table 1.  Drug Box Comprehension in the Symptom Drug Box Trial*
Table Jump PlaceholderAppendix Table 2.  Drug Box Comprehension in the Prevention Drug Box Trial*
Participants' Understanding of Side Effects

The brief summaries and the drug box differed in how they presented side effect data. The drug boxes contain fewer elements and were presented in a highly structured format.

Symptom Drug Box Trial

The drug box decreased the perceived magnitude of side effects of both the H2-blocker and the PPI. For Amcid (H2-blocker), 92% of the drug box group and 42% of the control group said that its side effects were small, very small, or nonexistent (P < 0.001). For Maxtor (PPI), the corresponding numbers were 85% and 37% (P < 0.001). The drug box group's perceptions more accurately reflected the drugs' actual side effect profiles: The boxes reported no life-threatening side effects, and side effects occurred at the same rate as with placebo.

Prevention Drug Box Trial

The drug box group had more accurate knowledge about side effects of both drugs: 51% correctly answered questions about side effects (bruising, rash, and liver inflammation), compared with 16% of the control group. The drug box decreased the perceived magnitude of side effects for both drugs (Figure 4, left). For Concor (statin), 73% of the drug box group and 36% of the control group said that its side effects were small, very small, or nonexistent (P < 0.001). For Pridclo (clopidogrel), the corresponding numbers were 44% and 12% (P < 0.001).

Grahic Jump Location
Figure 4.
Effect of the drugs facts box on the perceived benefit and side effects in the prevention drug box trial.
Grahic Jump Location
Participants' Understanding of Benefit
Symptom Drug Box Trial

In contrast to the drug box, the brief summary pages contain no data on how well the drugs treat symptoms. The drug box decreased the perceived magnitude of the benefit of Amcid (H2-blocker): 16% of the drug box group and 46% of the control group said that the drug was “extremely” or “very” effective (P < 0.001). The drug box increased the perceived magnitude of the benefit of Maxtor (PPI): 73% and 53%, respectively, said that the drug was “extremely” or “very” effective (P < 0.001). The drug box group's perceptions more accurately reflected the actual efficacy of the drugs: Amcid (H2-blocker) has a moderate effect on heartburn, whereas the effect of Maxtor (PPI) is quite large.

Prevention Drug Box Trial

The drug box led to more accurate knowledge. For Concor (statin), 72% of the drug box group and 9% of the control group correctly quantified its absolute reduction (P < 0.001). For Pridclo (clopidogrel), the corresponding numbers were 74% and 7% (P < 0.001). Figure 5 shows that most of the control participants substantially overestimated the benefit: 65% (Concor [statin]) and 53% (Pridclo [clopidogrel]) overestimated it by a factor of 10 or more.

Grahic Jump Location
Figure 5.
Perceived absolute risk reduction in the prevention drug box trial.

The correct answers are a 0.8% absolute risk reduction with Concor (statin) and 0.6% with Pridclo (clopidogrel).

Grahic Jump Location

The right side of Figure 4 shows that the drug box decreased the perceived magnitude of the benefit of both drugs. For Concor (statin), 13% of the drug box group and 50% of the control group said that its benefits were “very large” or “large” (P < 0.001). For Pridclo (clopidogrel), the corresponding numbers were 14% and 27% (P < 0.001).

The reason for the more accurate knowledge seemed to be that the drug box helped people appreciate the magnitude of the baseline risk for disease. The drug box group was better able to quantify the baseline risk: 83% of participants, compared with 11% of the control group, correctly answered this multiple-choice question. Most control participants substantially overestimated this risk, and 59% overestimated it by a factor of 10 or more.

Participants' Overall Judgments
Symptom Drug Box Trial

Figure 6 shows that the drug box group was better able to choose the superior drug—the primary outcome of this trial. When asked which drug they would choose if they had bothersome heartburn and could have either drug for free, 68% of the drug box group and 31% of the control group chose the PPI (P < 0.001).

Grahic Jump Location
Figure 6.
Effect of the drug facts box on comparisons between drugs in the symptom drug box trial.

The bars show the proportion of participants who answered the question correctly. All comparisons were statistically significant (P < 0.001). PPI = proton-pump inhibitor.

* 14% of the drug box group and 23% of the control group said that they would not take either drug (typically because they wanted to try lifestyle treatments first or did not want to take a medicine).

Grahic Jump Location

Because completion rates differed between the drug box and control groups (95% vs. 84%; P = 0.008), we performed a sensitivity analysis to explore the extent to which this differential participation could introduce bias. Our main findings were robust even when we made worst-case assumptions: Assuming that every noncompleter in the drug box group answered the drug choice question incorrectly and every noncompleter in the control group answered it correctly, the drug box still results in a 22–percentage point absolute increase in the proportion of correct responses (64% of the drug box group vs. 42% of the control group; P < 0.001).

Prevention Drug Box Trial

The drug box group was slightly less likely to say that the benefits of either drug were worth the possible side effects for people with heart or vascular disease. For Concor (statin), 72% of the drug box group and 86% of the control group said the benefits were worth the side effects (P = 0.012). For Pridclo (clopidogrel), the corresponding numbers were 49% and 59% (P = 0.20). When asked what they would do if they had heart disease and could have either drug for free, more people in the drug box group than the control group chose “neither drug”: 32% and 19% (P = 0.028).

In 2 nationally representative, randomized trials, we found that replacing the brief summary of direct-to-consumer ads with a drug facts box improved consumer knowledge and judgments of drug benefit and side effects.

In the symptom drug box trial, the drug box reduced perceptions of benefit for the moderately effective drug (H2-blocker) but increased them for the highly effective drug (PPI). The study also showed that the box helped patients make better decisions: The drug box group was more than twice as likely as the control group to choose the superior drug.

The prevention drug box trial confirmed the effects of the drug box on consumer judgments. Consumers who viewed standard ads overestimated disease risk, treatment benefit, and side effects; the drug box corrected these overestimations. We were also reassured about the concern that motivated this trial: that consumers might dismiss the value of small absolute effects on important health outcomes. Nearly three quarters of the drug box group still felt that the benefits of the statin—an absolute risk reduction in death of less than 1%—were worth the side effects for people with heart or vascular disease.

At the same time, although the drug box group was better informed about the data, participants were somewhat less enthusiastic about both drugs. Fewer felt that the drug's benefit outweighed its side effects, and more said they would not take either drug. This reduced enthusiasm can be interpreted in 2 ways. First, informed consumers may reasonably decide that the drugs are not worth taking. Alternatively, consumers may be insufficiently calibrated in judging what constitutes an important effect size. For example, statins result in a small yet important mortality reduction in high-risk patients, but people may dismiss this reduction as trivial because it is numerically small. They may not appreciate that the absolute magnitude of the effect is necessarily small because the risk being reduced—death—is itself small, and few risk-reducing drugs have been shown to reduce death. This finding underscores the need for physicians to provide patients with context, such as the magnitude of the benefit of other preventive interventions, to help them decide whether a drug is worth taking.

Our studies have 3 main strengths. First, we know of no other published trials evaluating formats to improve the presentation of outcomes data to consumers in drug ads. An English-language MEDLINE search through January 2009 shows that studies of direct-to-consumer advertising have largely focused on documenting problems with their educational content. The ads typically do not quantify drug benefit or do so using exaggerated formats (for example, relative risk reductions without reporting the base rate) (1619). Other researchers have criticized the typical presentation of side effects: long, hard-to-read lists that do not distinguish between clinically important and unimportant problems (16, 1821). Second, our studies went beyond simple comprehension of the material being presented. The symptom drug box trial showed that the box helped people perform a difficult task: choosing the superior drug. Finally, we used a national sample, which should make our results more generalizable than those of trials in which participants are drawn from a convenience sample.

Our studies also have limitations. First, as is always the case with random-digit dialing, it was impossible to verify the eligibility of all households, and some eligible individuals declined to participate. Consequently, participants were slightly older, slightly more likely to be non-Hispanic white, and slightly more educated than the general population. It is unlikely that these small differences introduced important bias; in other surveys, the amount of bias introduced by nonresponse has been shown to be minimal (2224). Second, not all participants found the information in the drug box to be accessible. Because finding the relevant information does entail some effort, it is not surprising that comprehension was not perfect (responses ranged from 55% to 93% correct). This may reflect the public's limited experience working with medical data. We believe that understanding will improve with increased exposure to standardized data presentations, such as the drug box, and better education; we have shown elsewhere that basic educational materials (25) can improve people's understanding of risk (26). It is reassuring that the box was effective even for those with the lowest level of education. For example, among people with a high school education or less, 50% of the drug box group versus 8% of the control group chose the superior drug in the symptom drug box trial and 79% versus 6%, respectively, accurately estimated the baseline heart attack risk in the prevention box trial. Finally, we tested the drug box in only 4 direct-to-consumer ads. If other ads communicate data better, the effect of the drug box would be reduced. This is unlikely, however, because very few direct-to-consumer drug ads quantify benefit (16, 19). Moreover, using 3 other ads (for rofecoxib, pravastatin, and clopidogrel), we demonstrated that a simpler version of the drug box had very similar effects on perceived benefit (4).

Some may wonder whether we simply proved the obvious: One group was provided with the “right answers,” whereas the other was not. However, that is precisely the goal of the drug box—it provides the data needed to make informed decisions. Without these data, people can only guess, and their guesses are most likely based on the information that appears in the ads. The H2-blocker ad is definitive (“Just one and heartburn's done”). The PPI ad is uninformative (“It's time for a Maxtor summer”). Neither ad made it possible to see that the choice was straightforward, because one drug really outperforms the other. Without the data, most participants in the control group made the wrong choice, but when given the data, most made the right choice. Similarly, the statin and clopidogrel ads heighten vulnerability (for example, the clopidogrel ad says “At 6'4”, 220 pounds, Bob is a formidable man. But he's no match for something one millionth his size. A clot.”) without educating consumers about the actual size of the risks they face. Exposed only to the ads, most control participants dramatically overestimated risk. But when given data, most assessed risk accurately.

Our findings have implications for policy. The FDA has recognized the need to improve direct-to-consumer ads. In 2004, the FDA issued voluntary guidance to the pharmaceutical industry about the readability of the brief summary (27). In 2007, it issued legislation requiring pharmaceutical companies to submit television ads before airing and giving the FDA the authority to levy fines for misleading ads (28). But neither the guidance nor the legislation addresses the routine provision of efficacy data or standards for presenting side effect data in print ads. The drug facts box is a viable way to disseminate these data.

Although drug boxes could be produced by the proposed Center for Comparative Effectiveness Research (29) or existing, independent organizations (such as the Drug Effectiveness Review Project [30]), we believe that the FDA should produce and routinely update them.

Given its central role in summarizing drug information, the FDA is the most important leverage point in getting balanced drug information to physicians and consumers. Moreover, the FDA drug reviewers are uniquely suited to creating boxes: They are independent experts trained to assess drug performance, and no one knows new drugs better or has better access to the totality of published and unpublished drug data. An Institute of Medicine workshop (31) called on the FDA to pilot test a drug facts box. We are now conducting such a pilot and found that FDA medical reviewers can produce drug facts boxes and have been helpful in developing a handbook to standardize the process. New FDA policy, a potentially large hurdle, will be required for the FDA to take on production of drug boxes. Such policy would directly respond to the recent Institute of Medicine report calling for better communication of drug benefit and side effects (32).

The U.S. public currently lacks accessible and accurate information about prescription drug efficacy and side effects (33). Instead, people are exposed to billions of dollars in marketing designed to generate enthusiasm for new products (4, 1617, 19, 34), leaving them vulnerable to persuasive marketing techniques and selective presentations of information. It is not surprising that the control groups dramatically overestimated drug benefit. Our studies show that the drug facts box helped consumers appreciate how well drugs work, resulted in better choices between drugs for current symptoms, and corrected overestimation of benefit in the setting of drugs for prevention.

Spending review. DTC Perspectives. 2008;7:10. Accessed athttp://www.dtcperspectives.com/images/email/DTCPJUNE08.pdfon 3 July 2008.
 
U.S. Food and Drug Administration.  Office of management budget formulation and presentation. Summary of full costs. Accessed athttp://www.fda.gov/oc/oms/ofm/budget/2007/HTML/Tables/SummaryofFullCosts.htmon 3 July 2008.
 
Schwartz LM, Woloshin S, Welch HG.  The drug facts box: providing consumers with simple tabular data on drug benefit and harm. Med Decis Making. 2007; 27:655-62. PubMed
CrossRef
 
Woloshin S, Schwartz L, Welch H.  The value of benefit data in direct-to-consumer drug ads. Health Affairs. 2004;Suppl Web Exclusives:W4-234-45, 2004 Jan-Jun.
 
The American Association for Public Opinion Research.  Standard Definitions: Final Dispositions of Case Codes and Outcome Rates for Surveys. 5th ed. Lenexa, KS: AAPOR; 2008. Accessed athttp://www.aapor.org/uploads/Standard_Definitions_04_08_Final.pdfon 3 July 2008.
 
van Pinxteren B, Numans ME, Bonis PA, Lau J.  Short-term treatment with proton pump inhibitors, H2-receptor antagonists and prokinetics for gastro-oesophageal reflux disease-like symptoms and endoscopy negative reflux disease. Cochrane Database Syst Rev. 2001; CD002095. PubMed
 
Rush DR, Stelmach WJ, Young TL, Kirchdoerfer LJ, Scott-Lennox J, Holverson HE. et al.  Clinical effectiveness and quality of life with ranitidine vs placebo in gastroesophageal reflux disease patients: a clinical experience network (CEN) study. J Fam Pract. 1995; 41:126-36. PubMed
 
Johnsson F, Hatlebakk JG, Klintenberg AC, Román J.  Symptom-relieving effect of esomeprazole 40 mg daily in patients with heartburn. Scand J Gastroenterol. 2003; 38:347-53.
 
U.S. Food and Drug Administration.  Plavix (clopidogrel bisulfate). NDA no 020839. Approved 9/20/2007. Accessed athttp://www.fda.gov/cder/foi/label/2007/020839s038lbl.pdfon 9 July 2008.
 
.  A randomised, blinded, trial of clopidogrel versus aspirin in patients at risk of ischaemic events (CAPRIE). CAPRIE Steering Committee. Lancet. 1996; 348:1329-39. PubMed
 
Cholesterol Treatment Trialists' (CTT) Collaborators.  Efficacy and safety of cholesterol-lowering treatment: prospective meta-analysis of data from 90,056 participants in 14 randomised trials of statins. Lancet. 2005; 366:1267-78. PubMed
 
Heart Protection Study Collaborative Group.  MRC/BHF Heart Protection Study of cholesterol lowering with simvastatin in 20,536 high-risk individuals: a randomised placebo-controlled trial. Lancet. 2002; 360:7-22. PubMed
 
Schwartz LM, Woloshin S, Black WC, Welch HG.  The role of numeracy in understanding the benefit of screening mammography. Ann Intern Med. 1997; 127:966-72. PubMed
 
National Center for Health Statistics.  National Health and Nutrition Examination Survey 2003-4. Accessed athttp://www.cdc.gov/nchs/about/major/nhanes/nhanes2003-2004/nhanes03_04.htmon 9 July 2008.
 
The dataferrett for browsing the dataweb. Accessed athttp://dataferrett.census.gov/on 3 February 2009.
 
Bell RA, Wilkes MS, Kravitz RL.  The educational value of consumer-targeted prescription drug print advertising. J Fam Pract. 2000; 49:1092-8. PubMed
 
Frosch DL, Krueger PM, Hornik RC, Cronholm PF, Barg FK.  Creating demand for prescription drugs: a content analysis of television direct-to-consumer advertising. Ann Fam Med. 2007; 5:6-13. PubMed
 
Kallen A, Woloshin S, Shu J, Juhl E, Schwartz L.  Direct-to-consumer advertisements for HIV antiretroviral medications: a progress report. Health Aff (Millwood). 2007; 26:1392-8. PubMed
 
Woloshin S, Schwartz LM, Tremmel J, Welch HG.  Direct-to-consumer advertisements for prescription drugs: what are Americans being sold? Lancet. 2001; 358:1141-6. PubMed
 
Davis J.  The effect of qualifying language on perceptions of drug appeal, drug experience, and estimates of side-effect incidence in DTC advertising. J Health Commun. 2007; 12:607-22. PubMed
 
Davis JJ.  Consumers' preferences for the communication of risk information in drug advertising. Health Aff (Millwood). 2007; 26:863-70. PubMed
 
Biener L, Garrett CA, Gilpin EA, Roman AM, Currivan DB.  Consequences of declining survey response rates for smoking prevalence estimates. Am J Prev Med. 2004; 27:254-7. PubMed
 
Curtin R, Presser S, Singer E.  Changes in telephone survey nonresponse over the past quarter century. Public Opin Q. 2005; 69:87-98.
 
Keeter S, Kennedy C, Dimock M, Best J, Craighill P.  Gauging the impact of growing nonresponse on estimates from a national rdd telephone survey. Public Opin Q. 2006;70(Special Issue):759-79.
 
Woloshin S, Schwartz LM, Welch HG.  Know Your Chances: Understanding Health Statistics. Berkeley: Univ of California Pr; 2008.
 
Woloshin S, Schwartz LM, Welch HG.  The effectiveness of a primer to help people understand risk: two randomized trials in distinct populations. Ann Intern Med. 2007; 146:256-65. PubMed
 
U.S. Food and Drug Administration.  Guidance for Industry: brief summary: disclosing risk information in consumer-directed print advertisements. Rockville, MD: Center for Drug Evaluation, U.S. Food and Drug Administration; 2004.
 
Public Law 110-85. The Food and Drug Administration Amendments Act of 2007. 121 STAT. 823. Accessed athttp://frwebgate.access.gpo.gov/cgi-bin/getdoc.cgi?dbname=110_cong_public_laws&docid=f:publ085.110.pdfon 9 July 2008.
 
Comparative Effectiveness Research Act of 2008. Vol. S.3408. Accessed athttp://thomas.loc.gov/cgi-bin/query/z?c110:S.3408: on 3 February 2009
 
Drug effectiveness review project (DERP). Oregon Health & Science University. Accessed athttp://www.ohsu.edu/ohsuedu/research/policycenter/DERP/index.cfmon 8 July 2008.
 
Institute of Medicine.  Understanding the Benefits and Risks of Pharmaceuticals: Workshop Summary. Washington, DC: National Academies Pr; 2007.
 
Committee on the Assessment of the US Drug Safety System.  The Future of Drug Safety: Promoting and protecting the health of the public. Washington, DC: National Academies Pr; 2006.
 
Avorn J.  Drug warnings that can cause fits—communicating risks in a data-poor environment. N Engl J Med. 2008; 359:991-4. PubMed
 
Kravitz RL, Epstein RM, Feldman MD, Franz CE, Azari R, Wilkes MS. et al.  Influence of patients' requests for direct-to-consumer advertised antidepressants: a randomized controlled trial. JAMA. 2005; 293:1995-2002. PubMed
 

Figures

Grahic Jump Location
Figure 1.
Advertisements given to trial participants.

Appendix 1 and Appendix 2 provide full-size images of all 8 ads. PPI = proton-pump inhibitor.

Grahic Jump Location
Grahic Jump Location
Figure 3.
Pridclo (clopidogrel) drug facts box.

FDA = U.S. Food and Drug Administration.

Grahic Jump Location
Grahic Jump Location
Figure 4.
Effect of the drugs facts box on the perceived benefit and side effects in the prevention drug box trial.
Grahic Jump Location
Grahic Jump Location
Figure 5.
Perceived absolute risk reduction in the prevention drug box trial.

The correct answers are a 0.8% absolute risk reduction with Concor (statin) and 0.6% with Pridclo (clopidogrel).

Grahic Jump Location
Grahic Jump Location
Figure 6.
Effect of the drug facts box on comparisons between drugs in the symptom drug box trial.

The bars show the proportion of participants who answered the question correctly. All comparisons were statistically significant (P < 0.001). PPI = proton-pump inhibitor.

* 14% of the drug box group and 23% of the control group said that they would not take either drug (typically because they wanted to try lifestyle treatments first or did not want to take a medicine).

Grahic Jump Location

Tables

Table Jump PlaceholderTable.  Participant Characteristics
Table Jump PlaceholderAppendix Table 1.  Drug Box Comprehension in the Symptom Drug Box Trial*
Table Jump PlaceholderAppendix Table 2.  Drug Box Comprehension in the Prevention Drug Box Trial*

References

Spending review. DTC Perspectives. 2008;7:10. Accessed athttp://www.dtcperspectives.com/images/email/DTCPJUNE08.pdfon 3 July 2008.
 
U.S. Food and Drug Administration.  Office of management budget formulation and presentation. Summary of full costs. Accessed athttp://www.fda.gov/oc/oms/ofm/budget/2007/HTML/Tables/SummaryofFullCosts.htmon 3 July 2008.
 
Schwartz LM, Woloshin S, Welch HG.  The drug facts box: providing consumers with simple tabular data on drug benefit and harm. Med Decis Making. 2007; 27:655-62. PubMed
CrossRef
 
Woloshin S, Schwartz L, Welch H.  The value of benefit data in direct-to-consumer drug ads. Health Affairs. 2004;Suppl Web Exclusives:W4-234-45, 2004 Jan-Jun.
 
The American Association for Public Opinion Research.  Standard Definitions: Final Dispositions of Case Codes and Outcome Rates for Surveys. 5th ed. Lenexa, KS: AAPOR; 2008. Accessed athttp://www.aapor.org/uploads/Standard_Definitions_04_08_Final.pdfon 3 July 2008.
 
van Pinxteren B, Numans ME, Bonis PA, Lau J.  Short-term treatment with proton pump inhibitors, H2-receptor antagonists and prokinetics for gastro-oesophageal reflux disease-like symptoms and endoscopy negative reflux disease. Cochrane Database Syst Rev. 2001; CD002095. PubMed
 
Rush DR, Stelmach WJ, Young TL, Kirchdoerfer LJ, Scott-Lennox J, Holverson HE. et al.  Clinical effectiveness and quality of life with ranitidine vs placebo in gastroesophageal reflux disease patients: a clinical experience network (CEN) study. J Fam Pract. 1995; 41:126-36. PubMed
 
Johnsson F, Hatlebakk JG, Klintenberg AC, Román J.  Symptom-relieving effect of esomeprazole 40 mg daily in patients with heartburn. Scand J Gastroenterol. 2003; 38:347-53.
 
U.S. Food and Drug Administration.  Plavix (clopidogrel bisulfate). NDA no 020839. Approved 9/20/2007. Accessed athttp://www.fda.gov/cder/foi/label/2007/020839s038lbl.pdfon 9 July 2008.
 
.  A randomised, blinded, trial of clopidogrel versus aspirin in patients at risk of ischaemic events (CAPRIE). CAPRIE Steering Committee. Lancet. 1996; 348:1329-39. PubMed
 
Cholesterol Treatment Trialists' (CTT) Collaborators.  Efficacy and safety of cholesterol-lowering treatment: prospective meta-analysis of data from 90,056 participants in 14 randomised trials of statins. Lancet. 2005; 366:1267-78. PubMed
 
Heart Protection Study Collaborative Group.  MRC/BHF Heart Protection Study of cholesterol lowering with simvastatin in 20,536 high-risk individuals: a randomised placebo-controlled trial. Lancet. 2002; 360:7-22. PubMed
 
Schwartz LM, Woloshin S, Black WC, Welch HG.  The role of numeracy in understanding the benefit of screening mammography. Ann Intern Med. 1997; 127:966-72. PubMed
 
National Center for Health Statistics.  National Health and Nutrition Examination Survey 2003-4. Accessed athttp://www.cdc.gov/nchs/about/major/nhanes/nhanes2003-2004/nhanes03_04.htmon 9 July 2008.
 
The dataferrett for browsing the dataweb. Accessed athttp://dataferrett.census.gov/on 3 February 2009.
 
Bell RA, Wilkes MS, Kravitz RL.  The educational value of consumer-targeted prescription drug print advertising. J Fam Pract. 2000; 49:1092-8. PubMed
 
Frosch DL, Krueger PM, Hornik RC, Cronholm PF, Barg FK.  Creating demand for prescription drugs: a content analysis of television direct-to-consumer advertising. Ann Fam Med. 2007; 5:6-13. PubMed
 
Kallen A, Woloshin S, Shu J, Juhl E, Schwartz L.  Direct-to-consumer advertisements for HIV antiretroviral medications: a progress report. Health Aff (Millwood). 2007; 26:1392-8. PubMed
 
Woloshin S, Schwartz LM, Tremmel J, Welch HG.  Direct-to-consumer advertisements for prescription drugs: what are Americans being sold? Lancet. 2001; 358:1141-6. PubMed
 
Davis J.  The effect of qualifying language on perceptions of drug appeal, drug experience, and estimates of side-effect incidence in DTC advertising. J Health Commun. 2007; 12:607-22. PubMed
 
Davis JJ.  Consumers' preferences for the communication of risk information in drug advertising. Health Aff (Millwood). 2007; 26:863-70. PubMed
 
Biener L, Garrett CA, Gilpin EA, Roman AM, Currivan DB.  Consequences of declining survey response rates for smoking prevalence estimates. Am J Prev Med. 2004; 27:254-7. PubMed
 
Curtin R, Presser S, Singer E.  Changes in telephone survey nonresponse over the past quarter century. Public Opin Q. 2005; 69:87-98.
 
Keeter S, Kennedy C, Dimock M, Best J, Craighill P.  Gauging the impact of growing nonresponse on estimates from a national rdd telephone survey. Public Opin Q. 2006;70(Special Issue):759-79.
 
Woloshin S, Schwartz LM, Welch HG.  Know Your Chances: Understanding Health Statistics. Berkeley: Univ of California Pr; 2008.
 
Woloshin S, Schwartz LM, Welch HG.  The effectiveness of a primer to help people understand risk: two randomized trials in distinct populations. Ann Intern Med. 2007; 146:256-65. PubMed
 
U.S. Food and Drug Administration.  Guidance for Industry: brief summary: disclosing risk information in consumer-directed print advertisements. Rockville, MD: Center for Drug Evaluation, U.S. Food and Drug Administration; 2004.
 
Public Law 110-85. The Food and Drug Administration Amendments Act of 2007. 121 STAT. 823. Accessed athttp://frwebgate.access.gpo.gov/cgi-bin/getdoc.cgi?dbname=110_cong_public_laws&docid=f:publ085.110.pdfon 9 July 2008.
 
Comparative Effectiveness Research Act of 2008. Vol. S.3408. Accessed athttp://thomas.loc.gov/cgi-bin/query/z?c110:S.3408: on 3 February 2009
 
Drug effectiveness review project (DERP). Oregon Health & Science University. Accessed athttp://www.ohsu.edu/ohsuedu/research/policycenter/DERP/index.cfmon 8 July 2008.
 
Institute of Medicine.  Understanding the Benefits and Risks of Pharmaceuticals: Workshop Summary. Washington, DC: National Academies Pr; 2007.
 
Committee on the Assessment of the US Drug Safety System.  The Future of Drug Safety: Promoting and protecting the health of the public. Washington, DC: National Academies Pr; 2006.
 
Avorn J.  Drug warnings that can cause fits—communicating risks in a data-poor environment. N Engl J Med. 2008; 359:991-4. PubMed
 
Kravitz RL, Epstein RM, Feldman MD, Franz CE, Azari R, Wilkes MS. et al.  Influence of patients' requests for direct-to-consumer advertised antidepressants: a randomized controlled trial. JAMA. 2005; 293:1995-2002. PubMed
 

Letters

NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Comments

Submit a Comment
Submit a Comment

Summary for Patients

Clinical Slide Sets

Terms of Use

The In the Clinic® slide sets are owned and copyrighted by the American College of Physicians (ACP). All text, graphics, trademarks, and other intellectual property incorporated into the slide sets remain the sole and exclusive property of the ACP. The slide sets may be used only by the person who downloads or purchases them and only for the purpose of presenting them during not-for-profit educational activities. Users may incorporate the entire slide set or selected individual slides into their own teaching presentations but may not alter the content of the slides in any way or remove the ACP copyright notice. Users may make print copies for use as hand-outs for the audience the user is personally addressing but may not otherwise reproduce or distribute the slides by any means or media, including but not limited to sending them as e-mail attachments, posting them on Internet or Intranet sites, publishing them in meeting proceedings, or making them available for sale or distribution in any unauthorized form, without the express written permission of the ACP. Unauthorized use of the In the Clinic slide sets will constitute copyright infringement.

Toolkit

Want to Subscribe?

Learn more about subscription options

Advertisement
Related Articles
Related Point of Care
Topic Collections
PubMed Articles
Forgot your password?
Enter your username and email address. We'll send you a reminder to the email address on record.
(Required)
(Required)