The Transparency Journey

In 2000, Beth Israel Deaconess Medical Center, where three years later I would start my medical training, launched one of the first patient portals. The portal provided patients with access to a copy of their record, hold one thing… notes. While my mentor Warner Slack, in 1972, predicted, “The medical record, traditionally closed to the patient, will be declassified at last and become a document developed jointly by patient and doctor,” notes were still considered in the realm of clinicians, written for one another rather than to be shared with patients. That would come eleven years later when the OpenNotes demonstration project studied what would happen when primary care notes were released to patients after signing. To give clinicians a chance to review and discuss results with patients, pathology and imaging were held for a few days. But even then, as the clinician with oversight over the portal, I began to field calls from patients about why their data was being withheld even for a few days, and not released immediately like lab values. The chorus was sounding, as e-patient Dave would say, to “give me my damn data.” Fast forward, Froedtert & the Medical College of Wisconsin went live with OpenNotes in 2017 broadly across our ambulatory clinics. And now, in 2020, shared clinical notes are the law.

At Froedtert & the Medical College of Wisconsin, as with most of our peer health systems nationally, we will be making changes to our portal and digital engagement platform to share data with patients and ensure that all information including medical notes are shared in real time.

What do we know about information transparency?

Patients who read notes report that they:

• have improved understanding of their health and medical conditions
• recall their care plan more accurately
• are better prepared for visits
• feel more in control of their care
• take better care of themselves
• take their medications as prescribed more frequently
• have more successful conversations and stronger relationships with their doctors
• Image for post

Data from the OpenNotes Demonstration Project

Beyond the demonstration study, additional research has identified that information transparency may mitigate disparities, improve safety, and also improve inpatient experience. Gerard M J Med Internet Res. 2018;20: e191.
Gerard M J Med Internet Res. 2018;20: e191.
Many initially thought that patients with higher levels of education would benefit disproportionately more than those with less education or other literacy barriers. In fact, residents who participated in our focus groups were quite concerned that this could worsen disparities. However, what we have learned is that all patients report benefits, and that patients with less education derive even more benefits, reporting access to notes as extremely important more than 3x that of the most educated. We have looked at the impact of safety and information transparency, noting in the Joint Commission’s journal: Open notes have the potential to connect patients and providers in new ways and to improve patient safety with a simple strategy that can be readily scaled up at relatively little cost. Although physicians express some concerns about sharing notes with patients, many agree that note transparency could improve patient safety and quality of care. Colleagues have looked more broadly that the impact of transparency on safety. Enabling patients to review notes helps them identify misunderstandings, clarity plans, and also the opportunity to correct anything that needs to be corrected in the chart for the next provider, such as medications. In the BMJ Quality and Safety, Bell and colleagues reported the data from a pilot program of requesting feedback from patients. 1 in 12 patients reported feedback, and 23% of feedback reported was perceived to be related to making a change that could impact the safety of patients. This means that involvement of patients into reading our notes could help with reducing patient harm and improving outcomes in the ambulatory setting. Laboratory results directly to patients also are a provision that can improve safety. Patients can provide a second set of eyes, and follow-up on abnormalities. Physicians have had concerns that information transparency could increase the risk of lawsuits. Seeing progress notes, and now and in the future other notes such as inpatient notes and operative notes, might alarm patients and identify errors to be the basis of a claim. When I have spoken with malpractice carrier leaders, usually the opposite is considered. When the patient has the ability to view alongside clinicians in the chart, the engagement is likely to reduce errors, and also reduce claims related to information not being shared with patients and families. For inpatients, research by Craig Weinert studied the impact of shared inpatient progress notes on hospitalized patients and families participating in a pilot study. He found: 80% of patients/families agreed that seeing doctors’ notes allowed better self- or family-provided care during the hospital stay 86% reported better understanding of health conditions 79% felt that notes helped understand plan of care better

Unintended Challenges

We have long known that information is asymmetric in healthcare. Clinicians go to medical school for four years, and complete at least an additional three years of post graduate residency to learn how to interpret and process facts, stories, images, and data values. Dave DeBronkart had summarized the patient need for information here: “Value in medicine depends on information — as I said in ‘Let Patients Help,’ ‘People perform better when they’re informed better.’ It follows that to make patients and families more effective in care, they need to know more.” What happens when patients see lab, imaging, or pathology results without a clinician present? Will patients be more anxious? Will patients worry unnecessarily if they see something that looks concerning but in reality isn’t? I love the image below, because it succinctly distinguishes the differences between data, information, knowledge, and wisdom Image for post Illustrated differences between the data to information to wisdom paradigm. Credit to Hugh MacLeod

We are not releasing knowledge per se, we are releasing data along with some information processing such as in radiology and pathology interpretations, or normal/abnormal markers for lab testing. Clinicians add knowledge to results, being able to contextualize and view patterns. Wisdom, coming from experience, highlights patterns that are more important or likely than others. By no means do I wish to discount any knowledge that patients gain through their experience. I know that many patients with chronic illnesses know more than me about their particular condition. Clinicians help patients to connect their experience and their results with broader knowledge and treatment networks. Clinicians add their experience through training and practice to help patients interpret their results and connect dots. What happens when patients see lab, imaging, or pathology results without a clinician present? Will patients be more anxious? Will patients worry unnecessarily if they see something that looks concerning but in reality isn’t? Two experiences are relevant here. The first is a study of laboratory releases done from University of Pittsburgh Medical Center.

In 2016 in Pittsburgh, Francesca Pillemer, Ateeve Mehrotra, and team looked the impact of lab release to patients through interviews of patients and clinicians as well as by following the data to see what happened. They looked specifically at HbA1c for diabetes control, where interpretation is based upon a person’s goal and whether, or not, they have diabetes; they also looked at pap smear results as an example of a pathology/cytology report. What they found is interesting, and likely bears learning from as we release more broadly and immediately. 80% of portal users viewed test results 83% of portal users who responded to a survey reported that test results were a very useful feature Of all patients with a test result, 43% had at least one abnormal value In interviews, patients reported “a strong sense of ownership of their results and nearly all thought they should have direct access.” They also noted that “Getting results first… gave [one respondent] time to do research and come to [the] appointment prepared. Using a difference-in-differences before and after automated test release went into effect, data showed that viewing test results “was associated with an increase in the number of patients with an office or telephone visit (3.7% vs 4.6% p<0.001). With manually released results, these increases were not seen. For the second case, we can look at pathology reports. Pathology reports can be nuanced, and clinicians must often confer with reading pathologists to either more fully understand results, discuss the need for additional tests or stains. As one example, a report for dermatopathology may include terms like “r/o melanoma” which means that the clinician is looking to rule out melanoma, but melanoma is not necessarily the diagnosis. Another version could be “melanoma in situ” which would be contained and treated with the biopsy. Both of these could cause alarm to readers, without context of how to interpret these studies. But there’s also another side of this. The patients that are relieved by a CT scan that shows that a tumor is shrinking, or a tumor marker that is falling. Or a diagnostic mammogram that is normal.

Future Practice

We must recognize that these new changes may further re-shape the relationship between patients and physicians. More importantly, the new regulations and changes across the country can be considered a catalyst for new approaches and improved experience for both patients and physicians. In other words, the new regulation can produce side-effects and nudge new behaviors. Let me consider two below. As I started, I see that we will come to a new equilibrium, one that takes advantage of the engagement and ownership that patients are bringing into the equation and nudges newer ways of partnering around lab release to work on mitigating the challenges above. I think there’s a human change and a technological change that might help us all make the most of these changes. Human Changes We, like many healthcare organizations, are looking at our processes and see how we may provide better services to our patients. In some cases, that may mean reading results and acting on abnormal result, such as a patient with an abnormal ultrasound and having the radiologist review and share with the patient in real-time. We may need to have better counseling and discussion of lab results with patients before testing. For example, if we order a biopsy, what might we see, and what will be do if the test is abnormal? It’s really best practice for lung cancer screening through computed tomography (CT) to have a pre-scan shared decision-making session to review potential incidental findings and discuss. Similarly, primary care physicians like myself routinely discuss the implications of prostate cancer screening with prostate specific antigen (PSA) testing before sending the result. This upstream counseling can be helpful for providing that context to patients before they see results. It also makes the conversations smoother on the other end. Image for post Good Practices for sharing notes. Adapted from Jared Klein et al For clinical progress notes, we continue to encourage our clinicians to be factual and descriptive rather than judgemental, be clear and succinct, and to, where possible, encourage patients in the note-writing process. (We will be enabling patients to share their history and contribute to their notes in the near future.) Encouraging all patients to read notes as standard practice may also provide important safety benefits while ensuring that implicit bias does not mediate access to notes. We have previously found that clinician encouragement is one of the most important factors in accessing medical records. We are also looking at providing patients handouts or other information about common tests. What exactly is an MCHC anyway, and what if mine is out of range? But I think the real improvement in this area will come with some of the technological changes. Technology Changes To be clear, there is no technological solution that replaces having a conversation between a patient and a clinician, either before or after a test has resulted — particularly if it reflects a new diagnosis. Technology, in the form of better human-computer interactions, visualization, and information presentation, though, may aid patients in interpreting their information, understanding where their results fall in terms of other patients, (remember, this is usually a Gaussian distribution, rather than a binary normal/abnormal), and to assist with terminology. Image for post Follow work by B Zikmund-Fisher (like this one) to look through examples of where visualizations can help with report interpretation A few studies by Brian Zikmund-Fisher and colleagues illustrated the point that information display can really determine how a patient perceives their result. For example, a hemoglobin A1c level of 6.2% would be flagged as “elevated.” Image for post Copyright authors/ Creative Commons But, for a patient with diabetes on insulin therapy, it may represent that they are too low, and at risk of hypoglycemia that could be addressed by lowering insulin dosing. The interpretation, thus, is in need of patient specificity. Setting that context, such as defining goals of therapy, can then result in more clear result visualizations. Overall, these approaches are helpful for patients across a wide range of numerical literacy. Image for post Prototype of a laboratory test. Open source code at: https://github.com/smart-on-fhir/cardiac-risk-app Interactive reports that provide richer information would also be helpful. For example, plugging in the Smart on FHIR app that shares information about cholesterol and lipid interpretation could provide more useful information about cardiovascular risk than high/low flags on cholesterol subtypes. AI and Machine Learning Our Research Lab is also engaged in using natural language processing to be able to parse reports and ‘translate’ them for patients. Appended below is an early concept that we worked on a few years ago. I’d also encourage you to read Nathan Chapman’s entry on our blog about his internship experience with our lab on conversational AI. In summary, I see transparency as being good overall for our patients and our field. It serves the Platonic goal of healing and informing those we serve. It may have some bumps, and will require some changes to how we practice. In the OpenNotes group, we often discussed this as a forcing function, something that will nudge us to do what is right. It may come with some more work, and some unintended challenges. But I predict we will rise over these concerns, change practices where needed, and overall provide better clinical care.