Patient Reported Outcomes For Registries: Should The Treating Physician See The Data, And If So, When?

Eric Gemmen, MA, Senior Practice Leader, Epidemiology & Outcomes Research, Real-World & Late Phase Research, Quintiles; Michelle Leavy, MPH, Manager, Health Policy, Real-World & Late Phase Research, Quintiles; Erin Tomaszewski, MPH, Epidemiologist, Real-World & Late Phase Research, Quintiles

The application of electronic patient reported outcome (ePRO) data collection and direct-to-patient reporting has enabled patients participating in registries to report their symptom or treatment experiences between physician visits, with minimal patient burden. These independent reporting approaches – described in Chapter 5 of the Agency for Healthcare Research and Quality’s ‘Registries for Evaluating Patient Outcomes: A User’s Guide’[1] – are often applied in real world research not only to collect data directly from the patients outside of an office visit, but also to support patient retention over long follow-up periods.

A hallmark of patient registries is their naturalistic follow-up, where the intent is that the existence and design of the registry in no way interferes with the treatment the patient receives over the course of the study, i.e., ‘real-world.’ For some institutional review boards and ethics committees, the mere collection of patient reported data via patient reported outcome instruments (PROs) in a registry is enough ‘interference’ with the natural follow-up to tip the scales towards classifying the registry as an interventional study.

This article discusses three key areas to be considered when determining whether or not to collect PRO data in an observational study: 1) the lack of guidelines regarding PRO data collection in observational studies; 2) types of PRO data that are considered warning flags for patient’s health status and are applicable to report to the treating physician; and 3) whether or not treating physicians should have access to the PRO data collected in an observational study.  At the end of this article, a checklist of considerations is included regarding when to provide PRO data to treating physicians.

Need for guidelines on availability of PRO data to treating physicians

Evidence-based guidelines for collection of PRO data in registries do not exist, and there is substantial variation in when and how patient registries incorporate PROs.[2] In particular, there is a lack of guidance on whether PROs collected as part of a registry but completed outside of a physician’s presence (i.e., not completed at the physician’s office and not entered by the registry site) should be shared with the physician, and, if so, when that data should be shared.

There is a need for guidelines to articulate ethical, effective, and transparent distribution of PRO data collected in patient registries. In particular, guidelines should address when PRO data must be shared promptly with the treating physician, when the PRO responses could be provided only upon physician request, and when the data may not need to be shared, depending on the study objectives, physician preferences or patient preferences regarding confidentiality.

Guidelines are also needed to help set appropriate expectations with physicians regarding access to and interpretation of PRO data during an observational study.

And finally, the implementation of guidelines that address how to include PRO data in patient charts or electronic medical records (EMRs) could facilitate and enhance PRO data analysis in retrospective chart review and observational study execution. This does not remove the potential for the treating physician to alter patient care based on PRO responses, but if certain PRO instruments become standard of care, they will then be accessible to the physician whether or not a registry is being conducted. Moreover, the potential benefits of the PRO data for improving patient care and patient satisfaction with care could exceed any potential cost of sacrificing the naturalistic purity of the data.

PROs as warning flags

In some cases, PRO instruments include items that could be answered in a way that indicate an immediate patient safety issue. The application of these specific instruments should be carefully considered in relationship to the study objectives and appropriate alerts to medical professionals should a safety signal be detected.

Examples of PRO instruments that potentially include a safety signal include:

• Suicide Ideation: Any PRO response that is suggestive of suicidal thoughts or intentions must be reported immediately to the treating physician. An example would be the Beck Depression Inventory-II item on “suicidal thoughts or wishes”, with an alerting response option of “I would kill myself if I had the chance.”[3]
• Adverse Events: The Patient-Reported Outcomes Safety Event Reporting (PROSPER) guidelines on reporting of PRO adverse events (AEs) seek to support the wider use of PRO-AEs and ensure that the patient 'voice' and perspective feed appropriately into collection of safety data.[4] One of the most important aspects of patients reporting on their own experience is the removal of the clinician detection of the AE, which sometimes leads to underestimation. As an example, the Blood Pressure Success Zone Longitudinal Study of Success (BPSZ-BLISS) registry aimed to collect patient satisfaction with treatment for hypertension using the Treatment Satisfaction Question for Medication (TSQM), but there were concerns that patient responses to the side effects domain would interfere with the desired naturalistic follow-up of the registry.[5] The registry proceeded to collect the TSQM after dropping the side effects domain, resulting in development of the abbreviated, nine-item TSQM-9.[6]
• Adherence: Medication Adherence Report Scale 5 (MARS-5) and other PRO measures of adherence are additional examples where the PRO should be sent to the physician when the patient indicates that they have stopped taking a critical or life-sustaining medication.

Physician Access to PRO data

Physician knowledge of PRO data could potentially alter the naturalistic follow-up in an observational study, especially if the data prompts a change in the patient’s treatment. For example, for studies of fixed duration, such as those that are examining treatment effectiveness, failure to communicate PRO data that may be a safety signal may carry ethical implications. On the other hand, advocates of centering care around the patient’s needs may argue that the treating physician should have access to all available patient information, in as close to real time as possible. In some cases, PROs collected as part of the registry are already routinely recorded as part of the standard of care, and physicians are accustomed to reviewing these data. However, in most cases, PROs are not part of the standard of care and may only be completed to comply with registry procedures.

Checklist: When to provide PRO data to physicians?

In conclusion, when considering providing PRO data collected in a registry to physicians, researchers should:[7]

• Determine what PRO data will be collected that could be used to inform clinicians, bearing in mind that there may be several different objectives for the data
• Set expectations on how the PRO data will be collected and which elements will be reported to the clinician
• Set expectations on transparency and timeliness of PRO data collection during planning of the registry protocol and project and data management approaches
• Involve clinicians in the registry’s planning and design
• Plan for training of clinicians on interpretation of PRO data
• Develop a plan for how clinicians will use PRO data in managing care
• Be aware and plan for the possibility that use of PROs may increase patient awareness of their disease during the study and potentially foster response bias.

Conclusion

Clear guidelines are needed to support ethical, effective, and transparent use of PRO data collected in patient registries. In particular, guidelines should address when PRO data must be shared promptly with the treating physician, as in the case of a safety signal, and adequate training of the physicians who are reviewing the PRO data. In addition, guidelines should address when PRO data should not be shared with the treating physician.

References