Like other industries, healthcare is fascinated with the possibilities of using technology to help increase efficiencies and deliver solutions to populations. Replacing expensive physicians and other healthcare workers with apps and online tools is appealing. Unfortunately, our efforts to use health information technology (health IT) to reduce healthcare waste have not resulted in meaningful savings. Evidence-based medicine tools have not consistently improved diagnostic or treatment decisions. Congress, government agencies and others have tried to push health systems and payers to publish cost and quality information to help patients make more informed decisions with minimal impact. Highlighting data to promote shared decision-making have not reduced low-value care or increased value-concordant care plans. Prior authorization appears to reduce low-value care for some services at the cost of increasing the transaction costs for many other appropriate service requests.
Based on my own understanding of health IT and its applications, I would consider exploring opportunities:
- Data collection
- Evidence curation
- Individual preferences
- Real-world outcomes
- Co-morbidities and other contextual information
- Data presentation*
- Summarizing the medical literature
- Summarizing the real-world experience of peers
- Care coordination between healthcare interactions
- Suggesting high-value providers for discrete care episodes
- Remove the pressure of the decision to undergo a procedure led by a provider with an economic incentive to have the procedure done, and
- Consider care outside a medical center where usual care is typically received
- Following up on different elements of the care plan
- Sharing data across care settings
- Improving healthcare interactions themselves
- Telemedicine and audio-only visits
- Remote patient monitoring
- Reducing friction between payers and providers
This the evidence-based medicine approach of showing the effectiveness of different interventions for a particular condition. Historically, this information has been tailored for providers to view and interpret. Clinical decision support relies on providers entering information discretely into medication lists and problem lists to trigger algorithms to suggest new diagnostic or treatment options.
Health IT could collect patient preferences to help determine what diagnostic or treatment option might best match an individual’s wishes. Health IT could also track patient quality-of-life before and after a procedure or medical intervention.
Claims information and patient-reported outcomes could supplement clinical study findings for patient populations who might have been excluded from randomized, controlled trials. These data might include treatment effectiveness (e.g., quality-of-life), side-effects, and costs).
Some outcomes may vary by age, gender, socio-economic status and concurrent medical conditions. Collecting this information could allow providers and other patients to use these factors to make a healthcare decision.
Data presentations about a treatment’s effectiveness for providers include receiver operating characteristics (diagnostic test effectiveness), funnel plots (subgroup comparisons of two different treatments [usually one intervention against usual care]) and Kaplan-Meier survival curves. There are separate display formats for patients to consider a diagnostic or treatment decision based on health numeracy.
In the absence of peer-reviewed data about real-world outcomes, health IT specialists might consider using the same principles to summarize the medical literature for these data.
*The wellness industry relies on consumers interested in making some change in their behaviors to improve their health (e.g., tobacco, excessive alcohol use, diet, exercise). I think there is a need to provide evidence-based information to support individuals in making these types of changes, but the healthcare system has not meaningfully invested in this work to-date. Given the current payment criteria for healthcare services, I do not expect the healthcare industry to start delivering this type of information to patients anytime soon. We profit more from managing complications from poor healthcare choices than helping encouraging health choices initially.
Employers, payers and other purchasers have used healthcare entities who are paid to perform evaluations for back pain and other conditions independent of having a procedure performed. Dissociating the decision to have a procedure performed and actually performing the procedure allows patients and purchasers to:
Health IT could help identify patients who will be facing a decision to undergo a procedure in the near future, support patients in the decision-making process to undergo the procedure and then help select a high-value provider to perform the procedure.
Health IT could help patients track various items for follow-up from different healthcare encounters (e.g., abnormal test results, specialty recommendations).
New modalities for providers to interact with patients make economic sense if they replace higher-cost, lower-value interactions. Telemedicine and audio-only visits could reduce healthcare spending without a reduction in care delivery quality if those technologies replaced in-person office encounters. If providers managed these encounters within a payment framework that rewarded them for these interactions (e.g., value-based arrangement); patients, providers and payers would benefit.
The technology and reimbursement for remote patient monitoring (RPM) allow providers to bill payers and patients for these services currently. The literature supporting RPM to reduce healthcare expenditures is weaker (diabetes, heart failure, COPD). For many chronic diseases, less frequent check-ins with audio-only interactions might be more valuable than existing care delivery approaches.
Prior authorization and identifying in-network providers reflect interactions between payers and providers that could be considered “high friction.” Health IT could reduce that friction by presenting relevant information or updating queues to accelerate a coverage decision.
This classification of health IT highlights the differences in data collection and presentation from managing care between healthcare interactions and improving the actual healthcare interactions themselves. Berwick and Hackbarth‘s healthcare model of waste includes
- Administrative complexity
- Fraud and abuse
- Pricing failures
- Failures of care delivery
- Failures of coordinated care
This model does not integrate neatly into the health IT target categories I suggested above. I might reconcile the two paradigms as follows:
Largest opportunity: Care coordination between healthcare interactions
Managing information flows between providers is the least threatening way for a third-party to reduce healthcare waste. Dedicated teams could manage requests and responses between ambulatory specialists and primary care providers, could obtain discharge summaries and operative reports to facilitate follow-up of actions identified during a hospitalization or procedure, and could interact with various ancillary services to gauge a patient’s response to treatment. As the trust among the provider, patient and third-party increases, the third-party could provide information about where to receive the highest-value care, could offer suggestions for a second-opinion before considering a major procedure, and could help identify more optimal treatment options for the patient based on their specific preferences.
Shrank et al. suggested interventions to address failures of care coordination could result in $29.6-38.2 billion annually. If done right, health IT-based care coordination initiatives could address some portion of failures of care delivery ($44.4-97.3 billion), overtreatment ($12.8-28.6 billion) and pricing failure ($81.4-91.2 billion).
Next-largest opportunity: Improving healthcare interactions themselves
The COVID-19 pandemic has reinvigorated in interest in using technology to connect providers and patients without a face-to-face encounter. As several payers (including my employer, UnitedHealthcare) phase out parity payments for telemedicine encounters, we will have to see if telehealth utilization continues to rise. I am not aware of any group in academic or the payer community suggesting what encounters might be best served with telemedicine and what encounters might be better performed face-to-face. As a general internist, I think follow-up visits for chronic disease (e.g., hypertension, diabetes, heart failure) seem easily converted into telemedicine visits or virtual check-ins, but there may be other examples. As noted earlier, substituting in-office visits with health IT-enabled visits should be the approach that leads to the highest value for patients. Unlike urgent care conducted virtually, addressing longitudinal care with telemedicine and other innovations would be best conducted in conjunction with a provider office.
Health IT could also reduce the friction between providers and payers, but that work may not clearly lead to savings or higher value to the patient directly. It is possible providers and payers will need fewer staff to complete tasks, but I do not expect providers to be paid any less for the work they currently do or payers to reduce premiums because of a lower overhead required to manage provider interactions.
Small opportunity: Data presentation
Despite the tremendous amount of work done in data presentation to help shift provider and patient behavior, the research has not reliably led to providers or patients making value-based choices consistently when presented with information in a more meaningful way. For providers, seeing information about high-value specialists or procedures with a lower out-of-pocket cost to patients might nudge some decisions in a way to reduce healthcare waste.
For patients, data visualizations of other patients’ experiences, cost information and quality outcomes might make a difference in how patients choose where to receive their subsequent care. The evidence does not show a meaningful uptake in this information with a corresponding reduction in low-value care when presented to patients.
Is there an opportunity?: Data collection
Of all of the ways health IT can impact healthcare delivery, collecting data from patients wherever they are would seem to be the most accessible. Just open up a secure app and enter information about your [surgery, feelings, quality-of-life, (insert favorite metric here)]. One research group deployed an app to the third generation of the Framingham Heart Study to collect survey data along with a digital blood pressure cuff and Apple Watch. 675 of the 790 enrollees completed the baseline survey (85%). 241 (31%) and 306 (39%) completed the weekly digital BP and heart rate uploads over 12 weeks, respectively. 470 (59%) completed the three-month questionnaire.
Patients who view data collection as a way to avoid healthcare encounters may be more likely to enter data into a smartphone or web app, but that hypothesis has yet to be tested formally. Some patients may be specifically motivated to track health and cost data around a specific care episode (e.g., pregnancy, hip replacement, prostate cancer) in the hopes of helping others understand the nuances of treatment and surveillance for a given treatment choice. Unfortunately, in our current reimbursement system, these data points seem to be worth little to the current patient, the provider and the payer.
Most conversations about reducing waste in the American healthcare system appeal to clinicians or patients to “do the right thing” within an environment that makes it difficult to even know what the “right thing” might be. Should I challenge my provider about a prescription if a generic medication, a different class of medication or watchful waiting might be as effective? Does the health system where I receive most of my care really perform colonoscopies at a higher quality than competitors nearby? Will my specialist get their recommendations back to my primary care provider? I am not confident that any third-party (employer, payer, federal government) will have much success modulating the interaction between a provider and a patient. The best outcome any third-party might expect would be an opportunity to intercede at a decision point with long-term financial or quality-of-life implications (e.g., major surgery, decision to undergo dialysis, start an expensive medication that will be needed for life). Health IT could develop interventions to support care coordination, direct care delivery as well as data-based decision making within the context of that reality.