Direct-to-Patient Enrollment and Data Collection

For many conditions, patients can self-identify easily and volunteer to participate in health research studies that are seeking patients with their characteristics, with the caveat that the patient has self-identified as having the condition being studied and this may or may not have been validated. For some conditions, self-identification can be very important because patients may know about their condition and be at high risk for serious consequences, but not have presented for medical attention. For example, it is important to study fetal exposures in early pregnancy when the risk of teratogenesis is high, yet many women do not present for medical attention until later in pregnancy and are unaware of which exposures may be risky.⁴ Not all conditions are as straight forward and easy for patients to self-identify as pregnancy. A limitation of direct-to-patient enrollment is that it assumes that the patient is an accurate, reliable and precise reporter of his or her condition. Depending on the condition and research question, it may be desirable to obtain clinical validation of the enrollment criteria, particularly the diagnosis or condition of interest.

In a recent internet-based pilot study to test new methods of conducting pharmacovigilance from the Pharmacoepidemiological Research on Outcomes of Therapeutics by a European Consortium (PROTECT), pregnant women were invited to participate in a study of medication use and pregnancy outcomes in four European countries (Denmark, the Netherlands, the United Kingdom and Poland).⁵

Figure 4-1 shows the process for study recruitment and followup.

Figure 4-1

Pharmacoepidemiological Research on Outcomes of Therapeutics by a European Consortium (PROTECT) enrollment and followup. EOD = expected date of delivery; IVRS = Interactive Voice Response System ©Nancy A Dreyer, Stella CF Blackburn, Shahrul Mt-Isa, (more...)

Pregnancy was chosen as the test condition for this project since, if successful, direct-to-patient research could offer an effective way of learning about potentially teratogenic exposures that occur early in pregnancy.

This study showed that women could indeed correctly report most of the prescription medications that they took. In addition, roughly 25 percent of them reported taking non-prescription medications, some of which were not noted in their electronic health records. Also women reported rich information about personal habits including smoking, alcohol, herbal medications and use of illicit and recreational drugs. They also reported on vaccinations and experience with anesthesia as well as other details not available or easily extractable or linkable from existing databases.⁶

Together Rheumatoid Arthritis (RA) is an example of a direct-to-patient registry that collected both patient-reported information as well as biological specimens. The objective of this pilot program was to study whether a direct-to-patient research approach could be used to complement conventional clinical research methods. A variety of digital approaches were used to promote recruitment including patient communities and social media outreach programs conducted by a large clinical research organization and funded by a pharmaceutical company. Potentially eligible patients with rheumatoid arthritis were invited to access study details, consent to participate, and to be screened for eligibility. The first 1,000 eligible, consenting patients were enrolled in a study that included two Web-based surveys. Participants also were required to submit an authorization for medical record release and a saliva sample using a kit that was supplied, including return packaging with paid postage. After receipt of the signed authorization for medical record release, a copy of the patients’ medical record was obtained from his or her physician and chart data abstraction was performed.

Over the 18 week enrollment period, 22,855 patients visited the study Web site and 8,142 (36%) attempted to screen for the study. Nineteen percent (n=4,289) completed the screener with a self-reported RA diagnosis. Only 1,421 (6%) met the study enrollment criteria based on self-reported RA diagnosis; previous exposure to an anti-tumor necrosis factor (TNF) α; age 21–75; Caucasian; and located in any U.S. State except New York or Maryland. One thousand patients proceeded to complete enrollment by consenting to provide medical record release and a DNA saliva sample for genetic analysis. Eighty-two percent (n=818) of enrollees provided lab data and genotyping was completed for 80 percent (n=798). In addition, 59 percent (n=591) of patients’ medical records were retrieved.⁷ Overall, data for all three aspects of this pilot study (PROs, lab data for genetic analysis and medical record review) were completed for 48 percent of enrolled patients, with collection of medical record data being the most difficult to complete.

Direct-to-Patient Enrollment and Data Collection Supplemented With Existing Data

In some cases, even when direct-to-patient enrollment and data collection is preferable overall, there is still certain clinical information that is best provided by a medical care provider.⁸ For example, in a study of birth outcomes, a trained clinician would be better able to provide a clinical description of most birth anomalies than would a patient. When medical care provider input is critical to meeting the study objectives, direct-to-patient registry recruitment can include patient authorization for targeted data collection from the patient’s medical record and/or design the study such that medical provider sites may be integrated into the recruitment strategy.

The National Amyotrophic Lateral Sclerosis (ALS) Registry is an example of a registry that collects data in two ways, from existing data and from direct enrollment. The first approach used four existing national administrative databases (maintained by Medicare, Medicaid, the Veterans Health Administration, and the Veterans Benefits Administration) to identify cases of ALS. The second approach used a secure Web portal (www.cdc.gov/als) where patients self-enroll after answering six validation questions. These questions were developed by the Veterans Administration and were found to be 93 percent accurate when reviewed by a neurologist for an ALS diagnosis. The decision to use the national databases was prompted by pilot projects to evaluate the feasibility of identifying ALS cases. An algorithm was developed using International Classification of Diseases (ICD)-9 codes, frequency of visits to neurologists, prescriptions for Riluzole, and death certificates listing ALS as cause of death to identify cases as either “definite ALS,” “possible ALS,” or “not ALS.” The algorithm has a sensitivity of 87 percent and specificity of 85 percent.⁹

When patients enroll directly their self-reported data are linked to existing data. Direct-to-patient enrollment is accomplished via an online portal that also uses data acquired from national administrative sources including the Centers for Medicare and Medicaid Services (CMS) and the Veterans Health Administration. This federally mandated program is used to describe the incidence and prevalence of ALS in the United States, characterize the demographics of those with ALS, and examine potential risk factors that may lead to disease development. Patients are identified using a two-pronged approach: (1) through existing national administrative databases on the basis of services received and (2) using a secure Web portal where patients self-enroll after answering six validation questions. A total of 12,187 persons were identified as “definite ALS” across the four national databases and through Web portal registration from October 19, 2010–December 31, 2011.¹⁰ The majority of patients (70%) were identified via national databases. This direct-to-patient approach has allowed for the administration and completion of more than 50,000 followup surveys to date among all enrollees to better understand potential risk factors and disease etiology.

These surveys represent the largest collection of ALS risk factor data ever assembled and analyses are ongoing. By leveraging existing national administrative databases and using an online secure Web portal, the Registry has been able to estimate the first-ever national prevalence of ALS in the United States and plans to continue to release future prevalence, incidence, and mortality estimates as subsequent calendar years are analyzed. This ability to follow the patient across changes in medical provider allows for the long-term followup needed to provide valuable insights into disease etiology.

The Registry has also developed an email research notification mechanism which alerts eligible patients to potential research studies after the research plans have been reviewed and approved by an Institutional Review Board.¹¹ Over 96 percent of enrollees have elected to be notified about research opportunities. To date, 21 studies have used the Registry for recruitment purposes, and researchers report that the ability to connect directly with patients was critical to successful recruitment. Another interesting aspect of this registry is its national biorepository, which is currently being tested through a pilot study. The goal is to have a nationally representative registry that will contain biologic specimens (e.g., blood, tissue) from patients enrolled in the National ALS Registry. As of the date of this writing, the biospecimen study is attempting to collect blood, urine, hair and fingernail clipping samples from 300 people with ALS in their homes. Additional specimen collection includes postmortem specimens of brain, spinal cord, cerebral spinal fluid, and pieces of muscle, skin, and bone from ALS. Data from the biorepository will be paired with the completed surveys in the Registry and hopefully will help researchers learn more about disease pathology and pathways.¹²

Site-Based Patient Enrollment and Data Collection Supplemented With Patient-Reported Data

In some cases it may also be desirable to identify and enroll patients from medical provider sites but contact them directly for followup data. This patient-centric study design is particularly attractive when patients may not return to the original site where they were recruited, thereby impeding the collection of followup data required to achieve the registry objectives. This may occur for example in long-term followup after bariatric surgery to evaluate sustained weight loss, evaluation of long-term success after implantation of a medical device, or treatment satisfaction and safety of cosmetic treatments.

The patient-centered Function and Outcomes Research for Comparative Effectiveness in Total Joint Replacement (FORCE-TJR) Registry, funded by the Agency for Healthcare Quality and Research, FORCE-TJR uses direct-to-patient data collection supplemented with data obtained from medical chart review. FORCE-TJR is a national registry established from a comparative effectiveness research network of community-based orthopedic offices that are representative of contemporary TJR surgeons and their patients (e.g., 75% are community practices).¹³ The goal of this program, like all quality improvement programs, is to reduce medical errors and adverse events and improve patient outcomes. In addition, FORCE-TJR a research program to develop new knowledge about best TJR surgical practices.¹⁴ A unique feature of FORCE-TJR is that is collects patient-reported outcomes (PRO) before and after joint replacement with a focus on general health, the frequency and severity of joint pain, and the ability to walk or climb stairs and walk distances, before and after the surgery.¹⁵ At the same time that patients complete the PRO, they are also asked to report any visits to an emergency room, hospital admission, or inpatient or ambulatory surgical procedure related to their knee or hip implant during the first six months following TJR surgery. Annual PRO surveys inquire about revision surgeries or any operative procedures related to the implant. The Clinical Data Team investigates all patient-reported events by reviewing the medical records from the facility listed on the report. FORCE-TJR’s timeline for data capture is illustrated in Figure 4-2. This systematic approach to data collection permits estimation of readmission rates within 30 days of discharge from TJR surgery and 90-day complication rates.

Figure 4-2

Function and Outcomes Research for Comparative Effectiveness in Total Joint Replacement (FORCE-TJR) data capture and timeline.

The majority of the FORCE-TJR data are obtained from the patient, particularly through questionnaires summarizing pain and function (Knee Injury and Osteoarthritis Outcome Score [KOOS], Hip Disability and Osteoarthritis Outcome Score [HOOS]). In addition, patients report key data including medical and musculoskeletal co-morbidities. The ability to capture patient-centric outcomes over long-term followup is valuable for surgical procedures and implants since these are often intended to be long-lasting treatments. To obtain these PROs, FORCE-TJR built a centralized information technology system to automate timely distribution of surveys via secure email with an individualized Web link or mailed scannable paper, automated reminders, tracking for completion, and personal reminders as needed to assure complete followup. This flexibility in method of survey administration based on patient preference is practical with the direct-to-patient design. In fact, thus far FORCE-TJR has collected PROs from more than 85 percent of the registry patients.

These patient-reported data are then supplemented by the electronic medical record (EMR) information for patients who report post-operative adverse events and surgical data for all patients. Medical chart reviews are conducted through either electronic data capture or manual chart review depending on site capability. In both scenarios a standardized clinical review is performed to apply pre-defined standardized definitions and algorithms for each diagnoses. For example, the diagnosis of infection is not a code from the EMR. Clinic notes, labs, and treatment records are reviewed to assure standard definitions independent of the data were sent.

Further, CMS administrative data are collected annually and used to validate post-operative events which are reported for patients who are over 65 years of age at the time of surgery which represent 50 percent of the registry patients.¹⁵ FORCE-TJR submits a “finder file” and CMS returns matched data, recognizing CMS will not be able to provide any data for patients under age 65, since they are not eligible for medical coverage by CMS. Those CMS data are then linked with the PRO data collected annually by FORCE-TJR.¹⁶ Over time, FORCE-TJR patients age and become CMS beneficiaries, so the administrative data will allow for validation of long-term revisions and complications to supplement patient-reported events.

One of the strengths of using patient-reported complications and outcomes data is that it improves complete capture of post-operative events and assures consistent clinical definitions. On average 25 percent of all readmissions and complications in the 90-day post-TJR period occur at hospitals or emergency rooms other than where the TJR was originally performed. While payer data such as CMS would capture these events, the surgical hospital does not have this information.¹⁶ In addition, this approach avoids extensive chart reviews on patients without reported post-operative events and focuses staff effort on those patients who report a suspected complication. Finally, this approach allows FORCE-TJR to use consistent clinical definitions for adverse events which minimizes the impact of varied definitions that result from differing administrative coding practices across hospitals. For example, the definition of “deep vein thrombosis” will vary across hospitals, but centralized registry staff can impose consistent clinical criteria.

FORCE-TJR plans to track patients for decades, even if patients no longer have a relationship with the original surgeon, have moved, or have different insurers/health care system. The direct-to-patient approach not only helps FORCE-TJR maintain a relationship over time, but also allows the collection of PROs, revisions and adverse events data much longer than 90 days after surgery, information which is of great interest to patients, surgeons and medical insurers.

Generalizability

A strength of patient registries is that they are more generalizable than randomized controlled trials, since registries reflect real-world behavior and practices and have fewer inclusion and exclusion criteria. However, as with any study design, thoughtful selection of the patient population is critical to enhance generalizability and reduce the potential for systematic error (bias). It can be particularly challenging to defend generalizability for direct-to-patient registries, since patients are recruited directly in situations where the underlying sampling frame is unknown. The concern to evaluate the representativeness of any cohort study conducted without benefit of an underlying sample frame is indeed challenging. In these situations, generalizability is addressed in terms of characterizing the population that has been recruited (accessible population) and comparing their demographic and other characteristics with what is known about the target population from other sources, e.g., case series and/or national data. For example, those who participate in Internet studies are often more educated. Also it is not uncommon to find racial and ethnic differences. At the very least, a direct-to-patient registry should report its findings by major subgroups and characteristics of interest, and leave it to others to determine whether the patterns observed in these cohorts are similar to other patient populations of interest. Thus the basic scientific process of replication and confirmation holds true for registries as with other forms of scientific inquiry.

In the PROTECT pregnancy study, there was concern that requiring study participation via the internet would discourage women with low income (and no internet access) from participating, which might bias the results. In an effort to ensure accessibility to women of various social classes and income levels, the study was designed to collect data in two modalities: by Internet or by interactive voice. However, Internet data collection was overwhelmingly the most popular choice, with only one woman completing the baseline pregnancy questionnaire by interactive voice; she subsequently dropped out of the study (Figure 4-1). PROTECT study participants were described by age, ethnicity, parity and residence within country, and those characteristics were then compared with known information about each country’s population. Although this descriptive approach cannot guarantee that all eligible subjects have been recruited or that the same is truly representative of the target population, it is often sufficient to characterize a population and provide a rigorous study of those participants, even if not 100 percent complete or representative.¹⁷

In contrast, consider the FORCE-TJR example where the registry staff monitor surgery logs each week to assure that all TJR patients are referred for invitation to the registry. The operating room schedule was used to identify eligible patients at sites with both phone-based and on-site recruitment. Case identification through operating room schedules avoided the potential bias of surgeons not enrolling more complex cases. In this example, the characteristics of participating patients can be compared to those who do not participate, using the surgery logs to enumerate the entire target population. Further, since FORCE-TJR has a followup rate of more than 85 percent among enrolled patients, reportedly the highest among all joint replacements registries in the United States, their data are likely to be less biased than other datasets (http://www.force-tjr.org/overview.html).

Validity

Data collection from patients should be approached differently from data collected from medically trained personnel. First, patients working independently may not have as much patience or persistence in completing a long, detailed questionnaire as they might if someone were present and coaching them. Second, a patient cannot be expected to report exposures and medical events using the same terminology and response choices as a trained medical professional.

For example, in PROTECT, patients were asked if they had various medical conditions (e.g., respiratory conditions) and if they used any medications to treat those conditions. If they answered affirmatively, they were offered a choice of the most common prescription medications for that condition, along with an opportunity to use text entry if they could not find their medication already listed. In many cases, women provided information on their medication use through free text fields even when the medication was available in the prepopulated list, suggesting either that patients did not recall the indication for which a particular medication was prescribed or that they had forgotten to respond to the condition-specific question when it was first mentioned and did not want to look back through all the indications listed to find where their drug belonged. This finding may, of course, have been related to the technology used for this study, and technology advances in data collection may simplify this type of data collection.

There have been technological advances in how information on adverse drug effects are collected from patients. MyMeds and Me is a good example of a patient-friendly approach that starts with a picture of the full body and asks patients to place the mouse cursor on “where it hurts.”¹⁸ From there, it takes them through guided, well-illustrated pictures with the end result being medically coded adverse event data that can be used for regulatory submission to satisfy the pharmacovigilance obligations of pharmaceutical companies who have marketed products.

While this is an outstanding example of how patient-friendly information can be translated into medically useful data, it was developed for a large commercial market (for use in call centers that are maintained by all pharmaceutical companies with marketed products), a situation that is far different from an individual set of researchers tackling various problems of public health interest. In the absence of such data collection and coding tools, it is important to ask patients to report information in terms that can be used to screen important medical events of interest, such as hospitalizations or illness requiring expensive medical care or support, and then flag those events for followup with medical trained personnel. That type of targeted medical followup can be used as an efficient approach to obtain clinical information to support drug benefit and risk assessments.

Many registries use selective validation to assure completeness and accuracy of reporting. For example, in the ALS registry, three state and eight metropolitan-area surveillance projects were funded for this purpose and several reports have been issued, describing results of local projects which can then be compared with national registry data.^19–21 In the case of FORCE-TJR, which supplemented patient-reported data with existing data, various methods were used to validate patient-reported data. For example, operative data about the implant were used to verify that the patient had a primary (or revision), unilateral (or bilateral) total knee or hip replacement procedure. The implant component type and volume support the coded procedure. FORCE-TJR also verified 30-day readmissions and 90-day complications with review of medical records and CMS administrative data. Medical record review for all patients (CMS and under 65 years) allowed FORCE-TJR to assure consistent event definitions were applied and consistent with national professional practice standards. A yearly review of CMS claims for patients over 65 years of age allows verification of revision surgery and timing, if present.

Nonetheless, it is important to keep in mind that there is no incentive for patients to tell the truth (or the complete truth) beyond altruism. Since patients are not, as a group, trained in the art of observation and reporting, there may be some suspicion that patients will report what they expect the study team would want to hear.

Patient Reimbursement, Recruitment, and Retention

Methods for patient recruitment including reimbursement vary based on the target patient population, study design, budget and timeline considerations. Methods should be carefully selected to ensure timely recruitment of the desired sample size in a manner that supports generalizability and minimizes the potential for selection bias. Ethical considerations require that any reimbursement be appropriate for time spent and not be considered any inducement to use a particular product not already prescribed by his or her medical care provider (e.g., by providing free drug). Loss to followup is another challenge faced by many long-term registries and can lead to bias by selective reporting from patients with favorable (or unfavorable) outcomes, or who are concerned about the health effects of selective exposures, which could increase reporting experience from people who had those exposures.

Reimbursement for Participation

Patients appreciate some consideration for the time and effort required for them to participate in research,²² particularly noninterventional research like patient registries where they are not receiving benefits as would be available in a clinical trial (e.g., more comprehensive testing, special treatments and/or other procedures that are not part of standard clinical care). Even in non-interventional studies, patients may receive reimbursement for parking or transportation for participation in traditional office-based registries. However, patients are rarely compensated for time spent completing their forms.

In the PROTECT pregnancy study, there was no reimbursement for patient surveys. A small focus group was convened to understand the high loss to followup. Pregnant women reported that while they would consider participating in a study like this purely for altruistic reasons, one of the most frequent comments was that some form of modest compensation, preferably a cash payment, would enhance the appeal of participation.

When cash payment s are used, these reimbursements should be nominal, i.e. in the range of $5–$20 U.S. or country equivalent per assessment, depending on length, user interface and estimated time for completion. The amounts should be clearly noted in the informed consent form, confirm how the reimbursement will be made, and reimburse directly in an electronic formats in order to best maintain patient confidentiality.

Recognition for registry participation can take many forms besides cash payments. FORCE-TJR did not compensate patients, but instead offered small thank you gifts such as sticky notes and pens with the FORCE-TJR logo at the time of enrollment. In addition, each year the participants receive a newsletter of “lessons learned” to reinforce the value of their participation.

Recruitment and Retention

Done well, there is tremendous value in being able to recruit patients via the internet and using other strategies that support broad recruitment and minimize the need for in-person contact. For example, a Danish study of characteristics that influence fertility and fecundity used an internet-based study approach to recruit women to a study about pregnancy and time to conceive. Internet-based data collection was chosen as the means for data collection because it afforded women privacy to disclose their intent and behavior and did not require them to share this information in any face-to-face contact with an interviewer or health care provider. The “Snart Gravid” [“Soon Pregnant”] pregnancy planning study started recruitment in 2007. These researchers used a popular Web site in Denmark (same as was used later in the PROTECT study) and 2,368 participants were enrolled in six months. After 54 months of recruitment, 5,920 women were enrolled in the cohort.²³ The TogetherRA example described earlier showed that a large number of RA patients were enrolled in only four months, with the requirement that they not only provide answers to questionnaires but that they also provide a signed release to allow medical record review and provide a saliva sample for genetic analysis. Overall, use of social media and support groups, with or without the use of paid advertisements, can be a useful tool for patient recruitment.²⁴

In the PROTECT study,²² the internet face of the study was a multi-lingual Web site that explained the study and showed friendly pictures of well-respected country-lead researchers for each country/language combination. Frequently asked questions were addressed and those who were interested were invited to provide informed consent and enroll in the study. Multiple patient outreach methods were utilized, including both low cost (posts on pregnancy e-forums, hyperlinks on pregnancy-related Web sites, leaflets and posters at community pharmacies and/or obstetric/midwifery units, social media profile on Facebook) and higher cost methods (large digital banners or hyperlinks on pregnancy specific Web sites, emails to registered users of popular pregnancy-related Web sites, and paid advertising on a social media site). After initially focusing on low cost methods, higher cost advertising and paid recruitment activities were implemented and determined to be essential to achieving a reasonable study size.²⁵

FORCE-TJR used two methods of patient recruitment, and noted that in-person enrollment by office staff during the pre-TJR visit resulted in somewhat higher recruitment levels compared to phone recruitment in the low volume sites, but both were highly successful. On average, in-person enrollment exceeded 90 percent of all English and Spanish speaking patients (other languages were not available), while phone recruitment to the distributed network averaged 70–75 percent enrollment. Of note, when patients were invited by the surgeon at the community sites, and accepted the recruiter’s phone call, enrollment exceeded 90 percent. However if the surgeon did not mention the registry, patients were less inclined to answer the call from the central enrollment staff. When patients answered the recruitment call, enrollment exceeded 80 percent. In summary, at either in-person or phone recruitment has been highly effective for enrolling patients in a registry. Key office procedures associated with high recruitment rates are (1) an invitation from the treating physician and (2) systematic identification of all eligible patients.

Patient motivation is a key factor influencing patient recruitment and retention. In the ALS Registry, the seriousness of the disease and the relatively short expectancy after diagnosis may be sufficient motivation for enrollment. In FORCE-TJR, patients reported that they want to know how they are doing and where they stand regarding pain relief and functional gain after surgery compared to other patients. They may be a highly motivated population since a successful surgical experience may return them to greater mobility and function than before surgery. In addition, the Registry is used to provide feedback to surgeons about the outcomes of specific devices and surgical approaches. The Registry consent/enrollment process emphasizes that their data are shared with the surgeon and are an important part of the Registry program. A newsletter is distributed to patients annually, and it demonstrates how patient data are used to inform new strategies to improving care.

It is important to keep in mind that no matter what reimbursement is used to enhance recruitment and retention, patients are unlikely to complete questionnaires which they find confusing or lengthy. In the PROTECT study, women who completed informed consent but who did not complete the baseline questionnaire were asked for their reasons for not continuing in the study. Seventeen percent (6/34) of patients indicated that the baseline questionnaire was too long. When assessing patient burden, it is important to consider factors that may affect time required for survey completion, including the educational level of the target population, number of comorbidities and medications, and level of detail requested.

Creating Standing Cohorts

The concept of a standing cohort is that a group of patients with a characteristic in common, e.g., a particular disease or condition, are enrolled in a registry with ongoing basic data collection that can be utilized on its own or in combination with linked data or supplementary data collection to address a specific research question. Recruitment from an established patient network or standing cohort, which may contain patients with a particular condition or exposure that are eager to participate in research, can also increase the efficiency of patient recruitment and enrollment. For example, a standing cohort of pregnant women could be created, with rolling admission and study completion after pregnancy has ended. Such a cohort would provide a rich source of information about changes in medication use during pregnancy and pregnancy outcome.

Standing cohorts can also be created from existing networks which have not been created for a specific study purpose. For example, PCORI has launched the National Patient-Centered Clinical Research Network (PCORnet) to increase speed, efficiency, and relevance of clinical research by funding both Clinical Data Research Networks (CDRNs) and PPRNs. Currently they have funded 23 PPRNs that are effectively standing cohorts, operated and governed by patients, advocacy organizations, and their clinical research partners. These networks are tasked with enrolling >0.5 percent of the U.S. population with the specified condition with a minimum of 50,000 patients for most common conditions. The PPRNs are developing a governance structure and operating policies that engage patient participants to generate and prioritize research questions. These networks are collecting patient-generated health information suitable for research from >80 percent of their membership. These research networks are exploring mechanism for patients to obtain electronic health data directly from administrative claims or electronic health records in addition to standardizing the collection of PROs. PPRNs are in the process of becoming a sustainable national resource.

Standing cohorts can also be created from existing registries. For example, the ALS Registry’s outreach program, which gives participants the opportunity to designate that they are interested in hearing about potential research projects, allows the Registry to then create standing cohorts of ALS patients with various characteristics who can then be questioned and followed over time to learn more about treatment and prognosis.

Direct-to-Patient Recruitment Material

Ethics review is required of all materials used in patient registries, including any posters, brochures, educational newsletters, and thank-you gifts. Recruiting and informational materials must be carefully crafted, taking into consideration variation in the education and reading level of the target population, presentation of information in an assessable way for non-medical personnel, and ensuring that information is presented clearly and concisely. Patient incentives for study participation should be critically examined and thought through as they are often a concern of ethics committees.

Informed Consent for Participation

Informed consent requirements may vary depending on study design and country. A flexible approach to informed consent may be needed, especially when conducting multi-country and multi-stakeholder studies.

For example, even though PROTECT was a fairly benign study in terms of its questions, there was great variability in the response of four Ethical Review Committees (the European equivalent of Institutional Review Boards). The most frequently cited issues of concern to these Ethics Committees related to data protection, which may be different from U.S.-based studies since the European Union has very strict restrictions on “cross-border” sharing of patient data. There was also variability in how Ethics Committees wanted to document informed consent.⁵ One Ethics Committee asked to have the informed consent form printed and mailed; others accepted informed consent by voice and/or by e-consent. Moreover, PROTECT was conducted through a public-private partnership, which complicated all aspects of data collection including safeguards for data protection including data transfer, storage, access and overall liability.⁵

In contrast, consider the Informed Consent used by the ALS Registry.²⁶ This single-page document uses plain language to explain the purpose and importance of this registry, showing that the informed consent document does not have to be a long, detailed document.

The process for managing informed consent is also important. Enrollment in FORCE-TJR requires office staff to provide an information packet to all patients who schedule TJR and to ask patients to provide their preferred contact telephone number. FORCE-TJR staff then contacts the patient by phone within two days of having mailed the information to review the material and invite participation. This process assures that consistent information is provided prior to obtaining consent to participate and meant that sites do not need to train research staff on site and do not need to worry about a back-up of eligible patients that need to be consented and enrolled. FORCE-TJR also uses the consent process as an opportunity to help patients understand the importance of the program, including the value of long-term followup. Patients understand they will be asked about their experience of TJR surgery over time and that the data have the potential to benefit others.

Patient Consent for Access To Medical Records

Confirmation or supplementation of patient-reported information with data from medical records can be critically important to direct-to-patient research. For example, in the PROTECT pregnancy study, it was difficult to interpret patient-reported information about birth defects. In the FORCE-TJR example, a great deal of information was collected about the device and the surgical procedure for the surgeon’s records. In the ALS registry, valuable medical information was obtained from health care providers that substantially enriched the value of the patient-reported data.

In studies where patient-reported information need to be supplemented with medical records, patient consent to access medical records is required. Depending on study design, obtaining consent to access medical records at the time of informed consent for the study is the preferred approach, where feasible, due to its efficiency.

In crafting a medical release form, it is important to consider where the data are being requested from (hospital or specialist vs. primary care) and length of access needed (one-time vs. long-term). FORCE-TJR obtains patient consents and medical record releases at the time of enrollment to specifically allow FORCE-TJR to review patients’ medical records and obtain CMS utilization data over time. However, hospitals are often slow to respond to approved requests for record access. Medical records office staff may insist that the medical record release is valid for one year only. In this situation (and whenever data are requested for new sites, such as where a patient may go for a revision or for treatment of a complication), FORCE-TJR is required to obtain a new release from the patient, which involves another step and is cumbersome. Fortunately, less than 8–10 percent of patients report events at each survey so efforts to obtain medical records are focused on these patients.

In some studies, a patient’s primary care physician may be asked to obtain other medical data as needed, such as the discharge diagnosis for a hospitalization, or a specialist’s report providing additional diagnostic information. Although it may be tempting to simply ask the primary medical care contact for the registry to seek these additional data without any compensation, those type of requests are rarely given priority and this approach is likely to diminish the amount of supplementary data that are made available to for study purposes. It is preferable to provide compensation for this extra collection and reporting using Fair Market Value for time spent.