Murtuza Bharmal

Recruitment of patients for participation in clinical trials is a major source of research expense. Estimates of recruitment costs vary by study and disease state; however, data suggests that patient recruitment costs range between less than $10 to more than $1300 per completed subject.

  Patient recruitment delays can compound costs by delaying the time to product approval, often decreasing the period of exclusivity post-launch. Studies, for example, have demonstrated that recruitment difficulties can account for up to 45% of study delays, and these often can last six months or more.

As sponsors search for more time-efficient, less-expensive approaches to conduct community outreach for clinical trials, Internet-based sources have gained more attention. Of particular interest has been the emergence of health-focused, online social networks. In theory, these online social groups, formed around specific medical conditions or health interests, are comprised of people that are highly engaged in their disease, and potentially, motivated to participate in clinical trials.  

While the number of marketing and research groups using online social networks has increased,

  few have published data that explores the potential effectiveness and costs of working with these groups, especially in comparison to outcomes associated with other techniques. In this article, we address this gap in the literature by presenting effectiveness and cost outcomes associated with three different patient outreach approaches. 

The objective of this study was to compare the effectiveness and costs of three direct-to-patient outreach strategies: direct mail outreach through use of a third-party mailing list; email outreach through use of a third-party email list; and email outreach through 

, a Web-based medication monitoring service that includes a social network of members interested in sharing feedback on their medications. 

This study assessed the ability of these three different outreach strategies to identify potential study participants for two clinical trials in rheumatoid arthritis and multiple sclerosis. These medical conditions were chosen because they comprise large numbers of patients suffering chronic illness with substantial potential for morbidity and mortality, and they are often the focus of many costly clinical trials. 

The outreach was conducted in the United States in June, July, and August 2009. Participants from any of the three outreach methods who completed an online survey received $10 in compensation for their time. The survey was blinded and did not contain any information on actual study medication, sponsor, or investigator site. Surveys conducted through 

 are covered under a blanket Institutional Review Board (IRB) assessment from Independent Review Consulting, Inc. IRB approval for the direct mail or email outreach was not seeked because the blinded nature of the questions (i.e., no clinical trial specifics included) met the criteria of 45 CFR 46.10 related to research on educational practices. 

The third-party direct mail and email outreach was conducted by Automated Information Systems, Inc. (Glen Ellyn, IL), a vendor chosen among other candidates for the size of their patient databases in these therapeutic areas; their ability to send both email and direct mail communications; and their ability to prevent duplication between the electronic and paper-based methods. Automated Information System's mailing list is derived from multiple sources, including managed care organizations and patient advocacy groups. Email communications were checked for accuracy using a matching system from Experian, Inc. (Costa Mesa, CA). 

The social network outreach was conducted among members of 

, a free, medication service that monitors the safety of prescription medicines, over-the-counter medicines, and health care supplements for over two million patients. 

 attracts members through online search engines as well as outreach efforts with physicians, pharmacies, and health-related Web sites. 

 members were randomly screened for participation in this study if their profile included an oral rheumatoid arthritis (azathioprine, hydroxychloroquine, leflunomide, methotrexate, or sulfasalazine) or multiple sclerosis (beta interferon or glatiramer acetate) medication. 

The effectiveness of each outreach strategy was measured by the proportion of those contacted who completed the online survey (i.e., the response rate) and the costs associated with generating one individual interested in the study (i.e., a function of response rate, unit costs, and clinical trial interest). The current study focused on outreach and did not follow patients through the investigator site screening process and enrollment. 

The unit cost for third-party mailing was $0.86 and $0.42  for third-party email. Actual marginal costs to contact the 

 members were negligible as the community and communication platform already existed. For the purposes of this analysis, however, the cost associated with 

 outreach was assumed to be equivalent to the third-party email cost of $0.42 per email contact. 

Response rates and demographic characteristics were compiled for each outreach strategy. Comparison between the strategies on gender was performed using Chi-square test, and on age using one-way analysis of variance (ANOVA), with Bonferroni correction to adjust for multiple comparisons. These analyses were conducted using SAS Version 9.1 for Windows from SAS Institute Inc. 

The third-party vendor sent 12,500 letters and 15,000 emails to patients with rheumatoid arthritis and multiple sclerosis (total 25,000 letters and 30,000 emails); these numbers reflected the vendor's minimum requirements. Overall costs associated with these contacts totaled $21,500 for the direct mail ($0.86 per letter) and $12,600 for the email contact ($0.42 per email). 

The response to direct mail outreach was 3.6% (905 of 25,000) as measured by participation in the online survey. Nearly all individuals responded within two weeks to the mail survey. Overall, 78% of individuals responding were interested in participating in the clinical trial, resulting in a cost per interested individual of $30.45. 

The third-party email outreach yielded a lower response rate than the direct mail, 0.5% (157 of 30,000). Most responses for email outreach occurred within 24 hours. The proportion interested in a trial, however, was 80%, very similar to the interest expressed by individuals contacted through direct mail. The resulting cost per interested individual for the direct email outreach was $100.32.  

During the June, July, and August 2009 time frame, 

 surveyed 3657 members treated with a medication for rheumatoid arthritis (n=3057) or multiple sclerosis (n=600). Of these 482 responded, yielding a response rate of 13.2%. Similar to email outreach, most responses to 

 outreach occurred within 24 hours. Of the individuals surveyed, 75% expressed interest in a clinical study, resulting in a cost per interested individual of $4.25 per response (assuming a cost per contact similar to charges for the third-party email outreach). 

The response rates and costs associated with each of the three outreach strategies are shown in Table 1, and demonstrate that the social network-based email strategy was substantially more effective and cheaper than either third-party direct mail or email. Contrary to industry assumptions, third-party email was substantially more expensive than direct, paper-based mailings, primarily driven by the lower response rate. Interestingly, clinical trial interest among patients motivated to respond to the survey was similar across all three outreach methods.  

Table 1. The social-network based recruitment strategy was substantially more effective and cheaper than third-party direct mail or email. 

Some differences were observed on gender and age between the three-outreach methods. 

 had a higher proportion of women respondents (84.9%) compared to other methods, however, women were generally more likely to respond in all groups (76% to 80%, see Table 2). Surprisingly, direct mail appeared to attract slightly younger respondents compared to both email approaches.  

Table 2. Women were more likely to respond to an invitation to participate than their male counterparts, regardless of outreach method. 

Patterns of response rates, proportion of individuals interested in a clinical trial, and costs per interested individual were similar for each of the three strategies when evaluated separately for rheumatoid arthritis and multiple sclerosis, with social network outreach through 

 yielding a substantially greater response rate and lower cost per response (see Table 3). Across all outreach strategies, patients with rheumatoid arthritis were less likely to respond than those with multiple sclerosis: third-party direct mail (1.4% vs. 5.8%), third-party email (0.3% vs. 0.7%), and social network outreach (11.2% vs. 19.5%).  

Table 3. iGuard.org returned the highest percentage of interested candidates among both Rheumatoid Arthritis and Multiple Sclerosis sufferers. 

Once an individual responded, however, interest in clinical trial participation was very similar across all outreach methods: 70% to 73% for rheumatoid arthritis, and 79% to 85% for multiple sclerosis. With respect to costs, the social network-based approach was substantially cheaper than both the third-party direct mail and third-party email outreach for rheumatoid arthritis ($4.82 vs. $86.28 and $195.65 per interested individual, respectively) and multiple sclerosis ($2.69 vs. $18.72 and $66.77 per interested individual, respectively). In these disease specific analyses, third-party email remained the most expensive option driven by the significantly lower response rate. 

Our data suggest that patient outreach using the online social network, 

, is more effective and substantially cheaper than either third-party email or direct mail. This result is driven primarily by a substantially-higher response rate for the social network outreach: over four times higher than third-party mail and over 25 times higher than third-party email. Similar patterns were seen for responses in individual disease states, with the 

 email yielding over 35 times as many responses among rheumatoid arthritis patients as the third-party email, and more than 25 times the responses among multiple sclerosis patients as the third-party email. 

Although the proportion of individuals interested in the study was similar across methods, the outreach strategies differed substantially in their cost per interested individual. In this case, the social network outreach through 

 ($4.25 per respondent) was seven times cheaper than third-party direct mail ($30.45 per respondent) and nearly 25 times cheaper than third-party email ($100.32 per respondent). These results persisted when evaluated by disease state, where the 

 outreach was nearly seven (multiple sclerosis) to 18 times (rheumatoid arthritis) cheaper than the second best option, third-party direct mail, and faster as well (24 hours vs. two weeks to capture the majority of responses). 

These findings are consistent with the existing literature about variation in subject recruitment based on both approach and disease state. Data acquired in association with the Lung Health Study, for example, showed great variation in recruitment rates from various sources, ranging from 1.5% from direct mail to 46% from media sources.

  An unexpected finding of our study was that third-party email performed worse than third-party direct mail. Specifically, the third-party email response rate was 0.5% vs. 3.6% for direct mail, and costs per interested individual totaled $100 for email vs. $30 per respondent for direct mail. 

Importantly, our data extends existing knowledge about the use and relative cost-effectiveness of social networks for patient outreach. Although little has been published to date about the use of social networks in general and online social networks specifically for clinical trial outreach, our data appears to be consistent with the findings of one study conducted in nine community-based organizations. In this study, researchers successfully recruited participants for HIV testing through accessing peer (not online) networks at rates that exceeded other methods by five-fold.

 Furthermore, our findings about relative costs suggest that online social networks may not only increase response rate but may do so faster and at a lower cost than other approaches. 

Our data should be viewed in the context of some limitations. First, only two disease states were studied and it is possible that relative effectiveness and costs of these strategies could vary for different patient groups. The fact that our data showed similar relative response rates and costs when analyzed by specific disease states, however, suggests that the relative benefits of outreach through online social networks may hold in other disease states. 

Another limitation of our data is that we studied relative differences in outreach strategies based on responses to an online survey, not through actual clinical trial enrollment. While it is possible that the relative response and cost differences between strategies erode when actual participation in clinical trials is measured, the 

 experience with clinical trial recruitment in areas such as depression and gastroesophageal reflux suggests that the proportion of patients expressing interest that are ultimately enrolled in the study is somewhat consistent across outreach strategies. 

It is also possible that the quality of the third-party email list may have contributed to our findings. However, as this email list was checked against Experian, Inc., which most third-party vendors use for such validation, it is unclear whether a different result would have be obtained working with another third-party email vendor. There were also differences observed in gender and age among respondents to the three outreach methods. However, it is unclear whether these differences would be relevant in the context of recruitment for clinical trials. 

Finally, it is possible that unique characteristics of the 

 social network account for its relatively high response rates and low costs. As of August 2010, 

 had over 2.3 million members and is much larger than other online medically oriented social networks.

  While this may limit generalizations about the use of medically oriented social networks to recruit patients, the size of 

's membership allows its demographics to more closely approximate those of individuals seeking health information online when compared to smaller social network organizations. The outcomes associated with outreach through nonmedical social networks are unknown. 

The data suggests that patient outreach through an online social network has the potential to be more effective and less expensive than either third-party email or direct mail in the recruitment of subjects for clinical trials. Further studies are needed to quantify the benefit of collaboration with the growing number of social network sites for actual clinical trial enrollment when compared to traditional approaches to patient outreach. 

 66 Witherspoon Street, #262, Princeton, NJ 08542. 

, MD, is Senior Vice President, at Quintiles, Durham, NC.

*To whom all correspondence should be addressed.

1. L. Gren, K. Broski, J. Childs et al., "Recruitment Methods Employed in the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial," 

2. S. Roy, S. Patel, K.H. Sheehan, J. Janavs, D. Sheehan, "Efficacy of Print Advertising for a Bipolar Disorder Study," 

4. D. Borfitz, "Social Networking Sites Have Myriad Trial-Related Uses," Cambridge Healthtech Institute, March 2009, 

http://www.ecliniqua.com/2009/03/02/social-networking-recruitment-sites.html~

6. W.M. Bjornson-Benson, T.B. Stibolt, B.A. Manske et al., Monitoring Recruitment Effectiveness and Cost in a Clinical Trial,

7. L.W. Kimbrough, H.E. Fisher, K.T. Jones et al., "Accessing Social Networks with High Rates of Undiagnosed HIV Infection: The Social Networks Demonstration Project," 

8. M. Allison, "Can Web 2.0 Reboot Clinical Trials?" 

Articles

A comparison of the yields and costs of online outreach methods to other recruitment techniques.

Direct-to-Patient Enrollment Strategies

Top Related Articles: Two Decades of Change for PROs

Putting the "e" in FDA's Draft PRO Guidance 

 The growing need to establish a drug's safety and long-term efficacy in a given population has increased the number of late-phase studies performed by the pharmaceutical industry over the past decade. The competitive drug market has driven companies toward product differentiation and positioning, both of which may be determined through postmarketing studies. Strategic guidance in areas such as reimbursement, franchise development, and marketing is provided through late-phase studies.

PHOTOGRAPHY: WESTEND61, PHOTOALTO/ERIC AUDRAS, GETTY IMAGES    ILLUSTRATION: PAUL A. BELCI 

Late-phase trials are postmarketing studies and classified as Phase IIIb or IV. The term Phase IIIb is used by the industry to describe additional therapeutic exploratory and confirmatory studies conducted after the primary regulatory dossier has been submitted by the manufacturer. They are typically conducted on marketed drugs to look for a new indication. Ziprasidone hydrochloride (Geodon), a commonly used antipsychotic introduced to the market for the treatment of schizophrenia in February 2001, received approval for treatment of bipolar mania in August 2004 as a result of Phase IIIb research.

Phase IV studies are typically conducted to capture physician prescription patterns, evaluate adherence to treatment guidelines, and conduct safety surveillance after the approval of a drug. Phase IV postmarketing research has evolved into patient registries, risk management programs, postmarketing surveillance programs, observational studies, disease management programs, and care quality improvement initiatives.

  The Phase IV research environment is largely driven by the changing regulatory environment, increasing safety concerns of new medicines, and the need for real-world data on the marketed drugs' safety and efficacy.

  The association between long-term combination Hormone Replacement Therapy use and a heightened risk of breast cancer was discovered during a Phase IV trial partially funded by Wyeth and conducted by the National Institutes of Health.      

Figure 1. Some of the main factors that lead to treatment ineffectiveness. 

Decision-making by health care professionals in the United States is primarily based on patient visits, is highly decentralized, and driven by professional autonomy to a large extent.

  The prevailing model of health care delivery is complicated, comprised of layers of processes and handoffs that result in slow and expensive health care. The system often fails to consider the importance of tailoring care based on patients' needs. While health care delivery needs to be timely and safe, the overall effectiveness of treatment can be optimized through patient-level feedback. 

Some of the main causes for treatment ineffectiveness have been described in Figure 1. Patient noncompliance, driven in part by lowered treatment satisfaction, has emerged as one of the leading causes for treatment inefficacy. High patient noncompliance leads to annual economic losses of up to $100 billion in the United States. Studies have shown that noncompliance causes 125,000 U.S. deaths annually, and hospital costs due to noncompliance have been estimated to be $8.5 billion annually.

Since the 1990s, studies have shown that systematic attention to patient feedback on health care outcomes and behavioral change lead to improved treatment outcomes.

  According to an IBM publication on health care in the 21st century,

  a patient-centric approach would ensure an improvement in health; a reduction in costs and premiums by payers; improvements in the efficiency, quality, and safety of care; and a reduction of the financial burden in the industry.  

Outcomes that matter to patients following treatment can be gauged most accurately using patient-reported outcomes (PRO) measures. The term PROs has evolved to include any endpoint derived from patient reports, whether collected in the clinic, using a diary or by other means, including single-item outcome measures, event logs, symptom reports or longer questionnaires. PROs are typically used to measure health-related quality of life (HRQoL), health status, symptoms, adherence, and treatment satisfaction.

 The selection of appropriate PRO instruments depends on the study design, nature of the intervention, and the target population. 

Results from psychometrically validated PRO measures can help us learn about overall patient treatment satisfaction, treatment effectiveness, and the impact of the drug on patients' day-to-day lives (see Figure 2). Physicians can gain useful feedback on prescribed treatments and could alter their course of treatment to improve the care delivered. 

The recent FDA and EMEA guidelines providing clarifications on the use of PRO endpoints from clinical trials to support label claims have resulted in a significant increase in the use of PRO measures in clinical studies.

  While PRO measures used in early-phase studies evaluate a drug's safety and efficacy in a small group of patients, PRO measures in late-phase studies are used on a larger patient population and in real-world settings. Compared to the controlled settings used in early-phase studies, the data provided in late-phase research, particularly related to PROs, has better external validity. The observational nature of most late-phase studies requires following patients through their normal course of treatment, which enables better determination of the factors leading to treatment ineffectiveness.  

Figure 2. Mediators contributing to improving treatment effectiveness and safety. 

Since early-phase studies are designed to specifically address drug safety, efficacy, and dose adjustment issues, they do not allow for head-to-head comparisons between drugs. Late-phase studies provide more comprehensive data about overall treatment satisfaction and patient HRQoL in real-world settings (see Table 1). 

A study designed to investigate the scope of HRQoL data from oncology studies showed that more than 80% of oncologists believed that HRQoL information collected prior to the commencement of treatment would improve overall treatment outcomes.

  Health-related quality of life data from late-phase studies can set the stage for physicians to tailor treatment regimens based on patient needs and ensure the provision of timely safe and high-quality health care. 

However, care must be taken to make study design adjustments in the context of observational studies or registries, where patient assessment using a PRO measure has the potential to alter routine clinical care.  

Table 1. Examples of why to use PROs and the different ways the data can be captured. 

Provider attitudes, decision-making, and health care delivery styles can influence treatment outcomes to a great extent. Physician characteristics such as lower practice volume, previous primary care or interviewing training, satisfaction with personal autonomy, and physician specialty are associated with higher doctor/patient participatory decision-making style ratings.

PRO measures from observational studies can provide useful information on the effect of physician attitudes on patient outcomes. Researchers agree that low levels of treatment satisfaction among gastroesophageal reflux disease patients are often a result of inadequate communication between physicians and patients, and believe that an understanding of a patient's HRQoL during the course of the disease would help address this problem.

PRO measures have been used to establish long-term treatment effectiveness and the negative impact of treatment on patients' HRQoL. The Prostate Cancer Outcomes Study (PCOS) was initiated in 1994 in collaboration with six cancer registries that were part of the National Cancer Institute's Surveillance, Epidemiology, and End Results program. Their aim was to obtain information on specific diagnostic procedures, clinical status, and details of treatments using patient surveys, medical records, and registry data.

HRQoL measures such as the SF-36 were used to obtain information on patient treatment satisfaction and disease symptoms. The PCOS generated over 25 publications, and one of the study findings was that with aggressive treatment there was a reduction in absolute risk of dying from prostate cancer but at the cost of a significant decrease in disease-specific HRQoL.

 PRO endpoints in late-phase trials support treatment benefit claims that describe a patient's symptoms or ability to function.

  They can help establish secondary indications for an approved product label. 

Porfimer was approved for a second indication of endobronchial cancer based on two randomized Phase IIIb studies comparing the safety and efficacy of porfimer versus laser therapy alone for the reduction and palliation of symptoms in the patients. Patient symptoms were evaluated using a PRO, the pulmonary symptom severity scale. 

 The most efficacious combination of drugs can often vary among patient populations, and it becomes necessary to conduct the assessment from the patient's perspective, using PRO measures. 

The postmarketing ADORE trial studied the relative efficacy of etanercept alone versus combination therapy of etanercept and methotrexate for patients with rheumatoid arthritis. It established that both treatment options were effective in reducing symptoms associated with rheumatoid arthritis and improving general patient health, by using endpoints from PRO measures like the EQ-5D, the VAS, and the Stanford Health Assessment Questionnaire.

 Treatment modifications, symptom control, and side-effect prevention techniques can be evaluated in late-phase clinical trials using PRO data. 

For example, an open-label observational study used the Short Form-12 (SF-12) to evaluate changes in symptomatology for patients whose antihypertensive drug therapy was modified due to tolerability issues.

  Instruments used in osteoporosis studies, such as the Osteoporosis Functional Disability questionnaire, were developed to assess disability in patients with osteoporosis and back pain due to vertebral fractures. This has led to interventions using fracture prevention techniques in clinical practice.

 PRO-reported side effects can present a comprehensive picture of drug-related safety. Health care researchers use PRO instruments to study side effects in late-phase studies, and experts believe the combined use of side-effect questionnaires with health status instruments could yield a broader understanding of treatment outcomes. 

The National Cancer Institute's Common Terminology Criteria for Adverse Events (CTCAE) is comprised of over 1000 individual items related to specific adverse events based on their degree of severity, and has been mandated for identifying toxicity-related symptom burden from chemotherapy patients.

  While the CTCAE was designed for clinician reporting, the rating scale was adapted by researchers for patient self-reporting and to evaluate the efficiency of the electronic collection of PROs.

Online patient reporting was demonstrated to be a feasible strategy for chemotherapy toxicity symptom monitoring, and could improve overall treatment safety and satisfaction with care. Another questionnaire, the Side Effect and Life Satisfaction inventory was useful in the evaluation of side effects among epilepsy patients.

  PRO data in real-world settings assessing side-effects often form the basis for dose modifications or the introduction of supportive medications. 

 Health care researchers and policymakers can use PRO data from late-phase trials to study the burden of disease on the targeted patient population. 

Resource allocation, drug costs, and premium reductions can be made through the extraction of information from postmarketing surveillance data. Using short-term data from the late-phase PROactive trial, a health outcome model was updated to project long-term clinical and cost outcomes associated with pioglitazone versus placebo in patients with type 2 diabetes.

  However, for modeling HRQoL based on Quality-Adjusted Life Years, health state utilities derived from another source was modeled into the event data from the PROactive trial. 

In the future, late-phase studies could incorporate simple preference-based PRO measures such as the EQ-5D to enable researchers to directly obtain information on HRQoL without the need to use multiple data sources. 

While manufacturers and researchers use late-phase clinical trials as instruments to corroborate the safety and long-term effectiveness of their drug, incorporating PRO endpoints could ensure focus on patient-centered health care delivery. PRO data from late-phase trials should percolate through the layers of the health care system to support the industry's growing commitment to patient-centric health care and reinforce the principles of safety, effectiveness, timeliness, efficiency, and equity in health care delivery. 

1. I.F. Moodie, "Emerging Role of Postmarketing Clinical Research," 

Strategic Research: A Practical Handbook for Phase IIIB and Phase IV Clinical Studies

3. D. Polygenis and G. Radulescu, "The Changing Face of Post-Marketing Research," Bioscienceworld, 

http://www.bioscienceworld.ca/TheChangingFaceOfPostMarketingResearch

Patient-level feedback from later studies provides a multitude of valuable answers to all stakeholders.

Murtuza Bharmal

Articles

Direct-to-Patient Enrollment Strategies

Late Phase Patient Reported Outcomes