Overcoming barriers to cancer clinical trial accrual

Impact of a mass media campaign

Authors


Abstract

BACKGROUND.

Annually, only 3% of adult patients participate in cancer clinical trials (CCT). Accrual barriers include lack of CCT awareness and uncertain third-party coverage. In 2002, a California law (SB37) required all insurers to reimburse costs related to CCT. The objective of the current study was to increase awareness of CCT and SB37 through a mass multimedia campaign (MMC) in the University of California (UC) Davis (UCD) Cancer Center catchment area. The authors assessed willingness to participate in and accrual to CCT.

METHODS.

Changes in CCT/SB37 awareness and willingness to participate were investigated before the MMC versus after the MMC and in UCD respondents versus UC San Diego (UCSD) catchment respondents—a control group that was not exposed to the MMC—by Pearson chi-square and logistic regression analyses.

RESULTS.

Of 1081 post-MMC respondents, 957 were from UCD, and 124 from UCSD. UCD respondents had a greater awareness of CCT (59% vs 65%; P < .01) and SB37 (17% vs 32%; P < .01) compared with UCSD respondents. Willingness to participate did not change in either cohort. Awareness level predicted willingness (odds ratio, 2.3; P < .01). Blacks, Asians, and lowest income (<$25 K per year) groups were the least willing to participate (P < .01, P < .04, and P < .02, respectively). The CCT accrual rate at UCD was unchanged.

CONCLUSIONS.

CCT and SB37 awareness increased significantly in the UCD cohort after the MMC. However, it was unclear whether this increase was attributable entirely to the MMC or to varying demographic variables. Enhancing patient willingness and accrual will require targeting other variables, such as physician or resource barriers, rather than just CCT and reimbursement awareness. Cancer 2008. © 2007 American Cancer Society.

In the United States, 1.2 million Americans are diagnosed with cancer annually, and approximately 550,000 die of cancer-related causes. Developing better therapeutic options for these patients requires well conducted cancer clinical trials (CCT). Unfortunately, only 2% to 4% of all adult cancer patients participate in CCT annually, and representation is low from elderly and minority populations.1–5 These numbers contrast with pediatric oncology, in which standard practice includes patient enrollment into peer-reviewed CCT resulting in an accrual rate exceeding 60%.6, 7 Low accrual rates often prolong the duration of trials, delay the analysis or publication of important results, and/or lead to early closure and failure of important studies.8, 9 Barriers to patient enrollment in adult CCT must be identified and overcome to increase trial participation.10–12

Previously, we reported that the type of patient insurance constituted a modifiable barrier to accrual.13 Since then, California has enacted legislation, Senate Bill 37 (SB37), that requires all third-party payers to reimburse the costs of CCT-related care. We hypothesized that a mass multimedia campaign (MMC) designed to increase awareness of CCT and SB37 would enhance willingness to participate and ultimately would improve accrual rates to CCT at the University of California (UC) Davis Cancer Center (UCDCC).

Before embarking on an MMC, we surveyed cancer patients and/or their families and friends regarding their baseline level awareness of CCT/SB37 and willingness to participate.14 We determined that CCT and SB37 awareness was associated significantly with willingness to participate in CCT. Reduced awareness and willingness to participate were especially apparent in respondents of minority and lower income/education groups. In this article, we report on the impact of an informational MMC on cancer patients and/or their families/friends, with special emphasis on CCT/SB37 awareness and accrual to CCT at UCDCC. A doubling in the accrual rate from 14% to 28% was considered a successful outcome.

MATERIALS AND METHODS

In the spring of 2004, an MMC was implemented to overcome the barrier of lack of information and fear of not obtaining insurance reimbursement through the dissemination of information about CCT and SB37 to cancer patients and their families/friends. The specific objective of the MMC was to disseminate information on the purpose of CCT, their value and necessity in the process of enhancing cancer care, and the role of SB37 in the coverage of the routine costs of these trials. Institutional Review Board approval was obtained for this project.

Young and Rubicam (Y&R), a global advertising firm, was asked to develop a media plan for an advertising campaign in the Sacramento/Stockton/Modesto designated market area (DMA) to enhance awareness of CCT and SB37. The advertising agency was charged with creating and executing an MMC that was capable of reaching the entire DMA at exposure levels high enough to measurably elevate awareness. Y&R also was asked to design a campaign that could be adapted easily for use in other markets, would enhance CCT as a brand, and could be adopted by other cancer centers in support of their brands.

The UC Davis Public Affairs and Marketing office produced a 2-color patient booklet and developed a Web microsite with an online survey that asked visitors how they heard about the campaign. A user-friendly CCT search tool for the microsite and a mechanism for tracking hits to the site were developed. California State Senator Jackie Speier (author of SB37) provided a link from her Web site to the campaign microsite, as did the Association of Northern California Oncologists (ANCO) and California Department of Managed Healthcare. In addition, the public affairs office publicized the campaign through in-house media, including the Cancer Center's biannual magazine Synthesis, the health system's weekly e-mail newsletter and home page on the Web, and posters displayed throughout the Cancer Center. Contacts with local media were used to obtain extensive news coverage of the campaign in the Sacramento region.

To guide campaign development, Y&R developed a 1-sentence central message to be communicated in all aspects of the campaign: “If you have cancer, do everything you can do.” Individuals ages ≥35 years in the Sacramento/Stockton/Modesto DMA constituted the strategic target audience for the campaign. A smaller “media target audience” within the strategic target audience was identified for the selection of media outlets. Adults ages 35 to 64 years who were deemed most likely to have had a past cancer diagnosis were chosen as the primary media target audience. Use of any hospital within the past 3 years served as a surrogate for cancer diagnosis.

The communication goal set for the 12-week MMC was to achieve an average frequency of at least 4 “impressions” per individual within the target audience for each 4-week period of the campaign (each time an ad was seen or heard counted as 1 “impression”).

The CCT campaign revolved around a simple iconic concept arising from the phrase, “The Big C” (Fig. 1), a slang term for cancer. The Big C campaign in all its forms—television, radio, newspaper, online banners, booklet, posters, microsite—combined information that was designed to elevate awareness with a specific call to action. Viewers and listeners were urged to get more information by calling a toll free 800 number or by visiting the UCD website.

Figure 1.

The Big C campaign: This photograph depicts the print advertisement and poster for the Big C campaign. It features a large blue C against a black background. Copy inside the C communicates the key messages of the campaign: that cancer clinical trials offer patients a choice and a chance and that the trials nearly always are covered by insurance.

After the MMC, 1081 patients who were either newly diagnosed with cancer and were first-time visitors to the cancer treatment facilities or had a prior history of cancer but were presenting with a new stage of their disease and/or their family members/friends were surveyed. The same survey instrument that was used in the pre-MMC “awareness and willingness” surveys1, 14 was used to gauge the success of the informational effort. Surveys were conducted administratively through the UCD-Veterans Administration of Northern California (VA)-ANCO group (printed questionnaire) and the National Cancer Institute Cancer Information Service (CIS) (telephone interviews) between July 2004 and February 2005. Surveys were collected from the 11-county catchment area of the UCDCC. In addition, the CIS simultaneously accrued a cohort from the UCSD catchment area in the Southern California counties of San Diego, Imperial, and Riverside. These counties represented a geographic area remote from UCDCC and served as a control population. Survey data (completed questionnaires and/or telephone interviews) were collected either by electronic or postal means through designated site coordinators every 4 weeks and then were transmitted to a programmer. Analytic data sets were prepared by the programmer and transmitted to the Division of Biostatistics by secure file transfer protocol (FTP), with codebooks for each data set. Each respondent was assigned a study identifier; analytic data sets used only the study identifier and contained no personal identifiers.

Factors that affected patient accrual in CCT available at the UCDCC between October 2004 and December 2004 were tracked prospectively by performing an accrual survey similar to that used previously and validated by our group.11, 13 The available data were entered into a Microsoft Excel Database (Microsoft Corp., Redmond, Wash) and generated descriptive statistics to evaluate the associations of these characteristics with protocol accrual. Results from this accrual survey were compared with a prior pre-SB37 survey by using the chi-square test.13 Significance was set at P < .05.

Statistical Considerations

The primary endpoint of this trial was to assess whether the intervention (ie, the MMC) had an impact on accrual rates into CCT. Statistical analyses addressed 5 key questions: 1) Did the intervention influence community awareness of CCT; specifically, had respondents heard of a clinical trial, and did they agree that it tested the safety and usefulness of a new drug against cancer and other diseases? 2) Did the intervention influence community awareness of California Law SB37? 3) Did the intervention influence community willingness to participate in CCT? 4) Are participants who are more aware of CCT and reimbursement issues also more willing to participate in them? 5) Did the intervention improve accrual to CCT at UCDCC?

Survey results initially were summarized by frequency tables, overall and cross-tabulated by site (Sacramento, San Diego) and by time period (pre-MMC vs post-MMC). Unadjusted comparisons by site and pre-MMC versus post-MMC time period demographics were analyzed with chi-square tests. The primary analysis for the first 3 questions was logistic regression; each model included a term for overall difference between Sacramento and San Diego (site effect), a term for post-MMC period (general time trend), and a term for post-MMC effects specific to Sacramento (thus, potentially attributable to the intervention). Regression models also were adjusted for demographic variables to address potential confounding. In secondary analyses, subgroup differences in intervention effect were examined. To address the fourth question, CCT awareness was included in the model as a predictor for willingness to participate. To address the fifth question, an accrual survey was conducted, and the results compared with pre-SB37 data using chi-square tests or the Fisher exact test for small cell sizes to compare proportions that participated in the 2 periods.13 All analyses were carried out using SAS/STAT software and R.15, 16

RESULTS

Pre-MMC Versus Post-MMC Awareness and Willingness Survey

A summary of the surveyed populations is presented in Figure 2. The post-MMC awareness and willingness survey was completed in February, 2005. In total, 1081 surveys were completed successfully: 957 from the UCDCC catchment area and 124 respondents from the UC San Diego (UCSD) catchment area.

Figure 2.

Summary of surveyed populations before (Pre) and after (Post) the mass multimedia campaign.

Respondent demographics are summarized in Table 1. Post-MMC respondents differed significantly from pre-MMC respondents with respect to type (the post-MMC group consisted of more patients than family members/friends; P < .01), age (the post-MMC group consisted of older respondents; P = .02), ethnicity (the post-MMC group included fewer blacks; P = .04), and income level (the post-MMC group consisted of more respondents with an income level ≥$75,000; P = .04). There were no significant differences between pre-MMC and post-MMC respondents with respect to sex, level of education, and type of insurance coverage.

Table 1. Respondent Demographics Before (Pre-) Versus After (Post-) the Mass Multimedia Campaign (MMC)
VariableNo. of respondents (%)P
Pre-MMCPost- MMC*
  • NS indicates not significant.

  • *

    Some percentages may not add to 100% because of rounding or when respondents were given the opportunity to select more than 1 choice (see ‡).

  • A printing error resulted in not collecting data on sex for all respondents.

  • Respondents could select more than 1 choice.

Respondent type  <.01
 Cancer patient601 (51)601 (56)
 Family member or friend540 (45)453 (42)
 General public21 (2)3 (<1)
 Not reported26 (2)24 (2)
Sex  NS
 Men204 (34)321 (30)
 Women392 (66)561 (52)
Age, y  .02
 <188 (1)1 (<1)
 18–2443 (4)28 (3)
 25–3483 (7)58 (5)
 35–44160 (13)172 (16)
 45–54267 (22)253 (23)
 55–64263 (22)278 (26)
 65–74220 (19)183 (17)
 ≥75128 (11)103 (10)
 Not reported16 (1)5 (<1)
Race/ethnicity  .04
 White897 (76)838 (78)
 African American/black85 (7)55 (5)
 Hispanic84 (7)68 (6)
 Asian68 (6)67 (6)
 Other38 (3)40 (4)
 Not reported41 (3)38 (4)
Educational attainment  NS
 ≤Some high school91 (8)67 (6)
 High school graduate255 (21)211 (20)
 Technical school/some college426 (36)397 (37)
 College graduate228 (19)235 (22)
 Postgraduate166 (14)160 (15)
 Not reported22 (2)11 (1)
Annual income, US $  .04
 ≥100,000162 (14)160 (15)
 75–99,99999 (8)130 (12)
 50–74,999224 (19)189 (17)
 25–49,999286 (24)233 (22)
 <25,000245 (21)228 (21)
 Declined to state172 (14)141 (13)
Insurance coverage  NS
 None38 (3)32 (3)
 Employer pays382 (32)353 (33)
 Spouse's employer168 (14)179 (17)
 Self pay164 (14)142 (13)
 Medicare283 (24)266 (25)
 MediCal112 (9)113 (10)
 Military40 (3)49 (5)
 Do not know22 (2)11 (1)
 Not reported2 (<1)20 (2)

CCT and SB37 awareness survey results are summarized in Table 2. Respondents who had heard of the term clinical trial (Question 1 of the awareness survey) and agreed that a clinical trial tested how safe and useful a new drug was against cancer and other diseases (Question 3 of the awareness survey) were considered “aware of CCT.” In Sacramento, there was a significant increase in post-MMC respondents who had heard of a clinical trial (68% vs 76%; P < .001), which also translated into a significant increase in the proportion of those who were “aware of CCT” (59% vs 65%; P < .003). There were no significant differences observed in these categories for the control group in San Diego. However, this difference in respondents who were “aware of CCT” was attenuated and was no longer statistically significant after adjusting for demographic shifts between the pre-MMC and post-MMC cohorts (results not shown).

Table 2. Awareness Survey Results Before (Pre-) and After (Post-) the Mass Multimedia Campaign (MMC)
Query or statementPost/Pre-MMC, %
Sacramento-area respondentsSan Diego-area respondents
  • a

    Respondents who answered “yes” to the Question 1 and “agreed” to Question 3 were “aware of cancer clinical trials.”

  • b

    More than 1 response was allowed.

  • c

    Respondents who answered “agree” to question 5 were “aware of California Law SB37.”

  • §

    Respondents who answered “very likely” or “somewhat likely” to Question 7 were “willing to participate in cancer clinical trials.”

1. Have you heard of the term “clinical trial”?*
 Yes68/76 (P < .001)76/76
 No24/1723/23
 Not sure7/61/1
 Not reported1/10/0
2. What comes to mind when you think of a “clinical trial?”
 Test of a new drug or treatment70/77 (P = .001)38/40
 An experiment32/3424/25
 Test of a procedure in clinic29/326/3
 Cancer treatment18/203/4
 Legal or court case2/21/0
3. A clinical trial tests how safe and useful a new drug is against cancer and other diseases.*
 Agree72/7674/80
 Disagree3/33/1
 Not sure23/2023/18
 Not reported2/10/1
4. In a clinical trial, the sponsor pays for the new drug being tested. All other costs are billed to your insurance company.
 Agree33/3931/33
 Disagree19/1918/30
 Not sure46/4151/37
 Not reported2/10/0
5. There is a new California law that makes health insurers pay most of the costs for cancer patients involved in clinical trials.
 Agree17/32 (P<.001)28/31
 Disagree16/1520/27
 Not sure64/5250/42
 Not reported2/12/0
6. Have you, or anyone you know, ever been in a clinical trial?
 Yes18/1923/22
 No63/6673/76
 Do not know18/142/2
 Not reported1/12/0
7. If you had (or have) cancer and were asked to be in a clinical trial, how likely would you be to consider it?§
 Very likely36/3737/36
 Somewhat likely45/4547/48
 Not likely16/1410/10
 Not reported3/46/6

Respondents who agreed that there was a new California law that made health insurers pay most of the costs for cancer patients involved in clinical trials (Question 5 of the awareness survey) were considered “aware of SB37.” In Sacramento, there was a significant increase in post-MMC respondents who knew of California Law SB37 (17% vs 32%; P < .001). There was no significant difference observed in this category for the control group in San Diego (Table 2). Again, this difference in the proportion of respondents who were “aware of SB37” was attenuated and was no longer statistically significant after adjusting for demographic shifts between the pre-MMC and post-MMC cohorts (results not shown).

Respondents who were very likely or somewhat likely to consider a clinical trial (Question 7 of the awareness survey) were considered “willing to participate in CCT.” In Sacramento, there was no difference in post-MMC respondents who were “willing to participate in CCT” (81% vs 82%; P = .16). This lack of difference persisted even after adjusting for demographic differences between the pre-MMC and post-MMC cohorts.

In general, respondents who were more “aware of CCT/SB37” were more “willing to participate in CCT,” even after adjustment for demographic differences (odds ratio, 2.34; 95% confidence interval, 1.76–3.09; P < .001). However, increased awareness of CCT/SB37 did not translate universally into increased willingness to participate in CCT in all subgroups, including those of lowest income and education. African Americans, Asians, and respondents from the lowest income group (<$25 K) were least willing to participate in CCT.

Accrual Survey

In total, 389 forms were completed and submitted by UCDCC physicians for evaluation. There was no cancer diagnosis for 91 patients. These patients were excluded; thus, 298 patients were identified as eligible for analysis. Patient characteristics are summarized in Table 3.

Table 3. Accrual Survey Patient Characteristics
CharacteristicNo. of patients (%)
  1. PMD indicates Professional Masters Degree.

Sex
 Men150 (50)
 Women148 (50)
Age range, y21–89
Race
 Caucasian188 (63)
 Hispanic16 (5)
 Asian17 (6)
 Black15 (5)
 Indian3 (1)
 Unspecified59 (20)
Insurance status
 Private160 (54)
 Medicare99 (33)
 MediCal24 (8)
 County8 (3)
 Self pay6 (2)
 Veterans Administration1 (<1)
Primary referral source
 Surgery79 (27)
 PMD55 (18)
 Medical oncology115 (39)
 Other49 (16)

Physicians did not consider clinical trials in 58% of patients (172 of 298). The primary reasons were no available protocol in 49 patients (28%), previous treatment in 30 patients (17%), patient not expected to return in 29 patients (17%), poor performance status in 23 patients (13%), and nonmeasurable disease in 21 patients (12%). Other less common reasons included no treatment indicated, unknown/simultaneous primary disease(s), and uncontrolled brain metastases.

Physicians considered 126 of 298 patients (42%) for clinical trials. This figure is significantly lower (P< .001) (Table 4) than the 62% reported in an accrual survey that was performed at our institution in the pre-SB37 era.13 Only 83 of 126 patients (66%) had protocols appropriate for site and stage available at the time of the survey, a higher figure than the 53% reported previously (P = .03). Seventy of 83 patients (84%) with available protocols met eligibility criteria for particular studies, a figure that was unchanged from the earlier report. Only 37 of 70 patients (53%) agreed to participate in a CCT, for an overall accrual rate of 12.4%. These overall results mirrored those from the earlier survey, because the increase in available trials compensated for a lower rate of physician referral. The remaining 33 of 70 patients (47%) who met eligibility criteria declined trial participation. The most common reasons were a desire for other treatment (9 of 33 patients; 27%), personal choice (5 of 33 patients; 15%), distance from the cancer center (4 of 33 patients; 12%), and no particular reason (6 of 33 patients; 18%).

Table 4. Accrual Survey Summary
VariableNo. of patients (%)P
Pre-SB37 surveyPost-MMC survey
  1. SB37 indicates California Law SB37; MMC, mass multimedia campaign.

Total no. of patients276298 
Overall accrual39 (14)37 (12).62
Patients considered for study171 (62)126 (42)<.001
Appropriate study available91 (53)83 (66).03
Patients eligible for study76 (84)70 (84)>.99
Patients agreed to participate39 (51)37 (53).87

DISCUSSION

In a prospective study of patient accrual patterns in CCT performed at our institution between 1997 and 2000,13 we prospectively tracked factors that potentially affected patient accrual into CCT. In that trial, the overall accrual rate into CCT was 14%. Patients with private insurance were significantly less likely to enroll in CCT compared with those who had government insurance.

In 2002, SB37 took effect and formed the basis of another UCD accrual study in 2004,11 in which we assessed the impact of SB37 on accrual. We observed that although the accrual rate of 16% was not significantly different from 14% in the pre-SB37 era, a larger proportion of patients who were offered a clinical trial agreed to participate. No patient declined participation because of insurance limitations. Furthermore, insurance type no longer represented a significant factor in whether patients agreed to enroll in a trial. It was unclear whether increasing SB37 awareness through an MMC subsequently would increase accrual.

We observed that CCT and SB37 awareness increased in the target population after the MMC. However, it was unclear whether this was attributable wholly to the MMC or to varying demographic variables between the pre-MMC and post-MMC groups. Furthermore, awareness of CCT and SB37 did not translate universally into willingness to participate in all subgroups. Finally, the informational MMC had little impact on actual accrual into cancer studies at UCDCC. It is important to note that, although there was an increase in the number of available trials, a lower rate of physician referral resulted in similar accrual rates (Table 4). Hence, if physicians had referred at the same level pre-MMC, then an accrual increase may have occurred.

We also performed focus group interviews as part of this project with the objective of understanding factors that could improve communication between patients, family, healthcare providers, and the clinical trial research team to ultimately enhance trial accrual. The results of that study have been published in abstract form,17 and details will be presented in a future report. Preliminary analysis reveals that discussion with an oncologist was the most important factor in understanding of and accrual to trials. In contrast, access to the Internet, mass media, and general information from cancer organizations were ranked among the least important or influential elements. These results also partly explain the inability of an MMC to increase willingness and accrual substantially despite enhanced awareness.

Our results are supported by other recently reported studies highlighting the importance of the physician's recommendation as the critical determinant in CCT participation.18–20 Willingness to participate in CCT was influenced strongly by patients' trust in their physician and by the quality of the information presented to them. Several other studies looked at physicians' proposals of CCT to eligible patients in which only 30% of patients were considered for CCT.21, 22 Patient characteristics that were predictive of a trial being offered, irrespective of eligibility criteria, included age (older patients were less likely to be offered a trial),21–23 ethnic background (minority patients were less likely to be offered a trial),21 and disease stage/prognosis (the higher the stage and the worse the prognosis, the less likely patients would be offered a trial).24

Some studies have raised the issue that physicians are not necessarily prone to participating in CCT. This is related in part to the finding that CCT have the potential to affect physicians' autonomy and, ultimately, their relationships with patients.25, 26 Investigators demonstrated that physicians' decisions regarding their CCT involvement depended more on their evaluation of specific protocols than on their general attitude toward clinical research.13, 20, 27, 28 Physicians who doubt the benefits of CCT and believe that the risks overweigh the benefits may not end up enrolling patients.

It also has been demonstrated that the impact of the relation between physicians, investigators, and members of the CCT support team affects patient accrual. In 1 study, physicians who had formal support from a cooperative group were more likely to refer patients to trials.22 Similarly, another study identified the presence of a multidisciplinary forum in the hospital as a significant determinant of increased accrual into CCT.2

In light of the data reported here and in previous reports, it appears that enhancing accrual to cancer trials will require more than just information dissemination to patients and/or their family and friends. Instead, a culture change within cancer physicians, along with enhancements in the supporting clinical trials infrastructure, will be required to increase trial participation meaningfully. Future accrual enhancement strategies should focus on physician and infrastructure barriers rather than solely on patient barriers.

Acknowledgements

We acknowledge the contributions of Jose Gonzalez of the Association of Northern California Oncologists; Lisa Montell, Manager of University of California Davis Market Research; and Claudia Morain of the University of California Davis Cancer Center.

Ancillary