How to make sense of the effects of policy on HIT
Here’s a checklist for digital health policy, derived from lessons from three decades of policy engagement.
I’ve spent my career tackling complex health policy initiatives to change provider or industry behavior. Reflecting on this experience, on what works and doesn’t, I’ve developed a checklist of points for policymakers to consider as they develop or revise policies.
Such questions are especially timely as the federal government is considering policies for its health IT incentive programs; health information exchange and interoperability, “information blocking,” and health IT certification; digital health product safety; and evaluation and management codes.
These are key points to track:
Policy views are theories subject to verification
Too often, by the time policies are proposed, assertions that began in the literature as theories tested against data no longer reflect underlying uncertainties. By their nature, laws and regulations assume unwarranted confidence about cause and effect and what people and organizations want and will do. In developing interoperability policies, for example, it is important to know: do doctors want all the information in the Consolidated CDA document; are providers more focused on cross-organization rather than internal interoperability as suggested by health IT incentive programs, and do patients really want electronic access to all of their electronic health data? Maybe yes, maybe no, maybe sometimes.
This shift from theory to certainty is inevitable in advocacy and policymaking; nonetheless, policy development should acknowledge and reflect uncertainty. It can set more modest aims or build-in opportunities to assess, validate, and adjust. Policymakers can use agile policy and implementation models that allow for stakeholder feedback (and adjustment), to continually test policies against data and experience, to use a scientific model.
Evidence of a problem does not equal evidence that policy intervention is justified, will work or is worth the cost
The currency of health policy is technical analysis, with many footnotes, that documents a problem and ends with policy recommendations that usually call for new programs and requirements that people or organizations do certain things. Unfortunately, documentation of a problem is usually more robust than evidence for the feasibility or net benefits of interventions.
Unintended consequences are inevitable
It is a truism that unintended consequences are inevitable. Agile and iterative approaches to policymaking and implementation can help reduce such risks. The most critical way to mitigate this risk is to anticipate, to think analytically about what could occur, to listen to affected parties, to listen some more, to imagine what might happen, and to have the courage to act on these insights. More generally, it is not a bad idea to be modest and prioritized in policy aims. I have witnessed several unintended consequences:
“Market failure” is in eye of the beholder
Market failure is a classical rationale for healthcare regulation. The argument is often made that “we will only try to regulate where the market won’t produce the right outcome or action”. The reality is usually less rigorous, “they won’t do what I think they should, so the market has failed”. Policymakers should resist the temptation to use regulation to make people do what you think they should do. Markets, however imperfect, can allow for incremental adjustment to individual preferences that regulation does not. Always ask: why aren’t they doing what I think they should do and what are the costs and prospects for potential corrective policies?
Logrolling leads to policy bloat
Logrolling is the political concept of “you add my policy, I’ll add yours and we’ll both be happy.” Intentions are usually good and policymakers typically well-meaning but, see “unintended consequences.” It is often used in legislative bodies and committees (e.g., federal advisory committees) where support of others is needed to move one’s own agenda. Examples of negative consequences are the breadth and depth of Meaningful Use Stages 1 and 2 and almost any major health policy bill. Logrolling is inevitable and necessary to legislating and policymaking but it is also critical to recognize its risks and try to limit its impact, for example as the Administration evaluates advisory committee recommendations.
Compliance eats policy intentions for lunch
A familiar pattern in public policy is that initiatives are distorted when compliance enters, often by organizations and individuals who do not “own” the initial policy (e.g., auditors, inspectors general, etc.). E&M documentation guidelines became rules that became the basis for audits and technology (e.g., design of EHR documentation) and process initiatives (e.g., healthcare organization compliance programs). Similarly, Meaningful Use incentives to adopt technology became a complex compliance risk for providers via Medicare and state Medicaid audits. Investments in compliance generate costs and waste and discourage participants. Policymakers should anticipate compliance implications and design to minimize these costs, for example by ensuring that provider guidance can reduce compliance risks while enhancing compliance program efficiency.
We are not all alike: segmentation works in marketing and public policy
Laws, regulations and interest groups seek uniform policy implementation, with variations limited to size (e.g., small practices) or location (e.g., rural practices). But clinicians and health care organizations vary along many dimensions, not just economic and demographic. Drawing on a marketing concept, segmentation by shared interests and behaviors is more robust than simple geographic or demographics classification. Policymakers should seek solutions that allow for and can reflect varying (and often unknown) needs, interests, and priorities. The CMS Center for Medicare and Medicaid Innovation (CMMI) seems to be following this approach with its multiple, mostly voluntary initiatives. CMMI’s early models demonstrated the importance of testing a diverse set of approaches “. . ., [b]ecause every provider, practice, and system has a unique set of concerns, priorities, and resources, the diversity of CMMI’s models enabled significant learning and innovation,” according to this article.
Policy deadlines are made to be broken, and we all know it
Those who make and are affected by health policy have become accustomed to delayed regulations and missed deadlines. Repeated delays waste effort and investment, penalize and discourage early adopters, and breed cynicism and complacency. Examples include the false starts for implementation of ICD-10 codes, postponed implementation of the ONC 2015 EHR certification, and delays in Stage 3 of Meaningful Use and related programs. Given the likelihood of such delays, policymakers should anticipate objections and roadblocks to their proposals and avoid those likely to require significant and costly timetable adjustment. Above all, they should listen, with an open mind and open heart, to those likely to be affected by a policy and those able to predict its course. One way to avoid the need for delays and to minimize their costs is a sequenced and iterative approach to policy development and implementation of “crawl, walk, and run”.
Policy innovations run their course, as do the organizations charged with their execution
Part of the challenge faced by policymakers is when to realize that an initiative is nearing end of life, and what to do about that. Policies, like organizations, have lifecycles felt both by the organization and those it serves. Congress and CMS leaders have recognized this dynamic for the EHR incentive programs, framed as “meaningful use fatigue,” reinventing “Meaningful Use” as Advancing Care Information and then Promoting Interoperability, narrowing focus on interoperability and patient data access rather than a broad set of health IT-related activities, and adjusting organizational focus.
Emphasize “diffusion of innovations” in lieu of or along with detailed regulation
Health policies aim to change behaviors of organizations, clinicians, and patients. In doing so, they face daunting challenges, including the headwinds reflected in this checklist. In essence, they are pursuing diffusion of innovations, such as enhanced documentation, use of health IT, greater health data exchange, increased quality, etc. Innovations diffuse over a long time, and unevenly within populations (see segmentation). There is extensive literature on how innovations (including healthcare) spread. Policymakers should incorporate these perspectives into policy design and harness what we know about diffusion of innovations to achieve lasting, organic, and durable changes instead of “check the box” responses.
This perspective was reflected in a December 2017 report by the JASON federal advisory committee, “Artificial Intelligence for Health and Health Care.“ JASON noted that “[t]he process of developing a new technique as an established standard of care uses the robust practice of peer-reviewed R&D, and can provide safeguards against the deceptive or poorly-validated use of AI algorithms (p 2). Dearing and Cox, in an excellent review of diffusion of healthcare innovations state that “[p]urposive dissemination, or designing for diffusion, means taking additional steps early in the process of creating an innovation to increase its chances of being noticed, positively perceived, adopted, adapted, and implemented—and, thus, successfully crossing the research-to-practice chasm.”
Conclusion
Regulations work, but at a cost, and often with fragile results. For example, Meaningful Use and related initiatives spurred a digital healthcare environment, opening opportunities for new solutions and applications and to data use, but at the cost of significant concerns about usability and clinician burden, with data quality still not where we would like it to be. In the framework of diffusion of innovations, this program shifted out the cost and effectiveness curve for the economic buyers of EHRs but not for many end users. A more modest and focused approach could also have achieved high levels of digitization, but with more organic clinician uptake.
More recently, we have seen how a targeted and flexible approach to health IT policymaking, well aligned with this checklist, has led to rapid roll-out of API-based interoperability using the HL7 FHIR® standard, driven by modest investments by the Office of the National Coordinator for Health IT, functional requirements on what APIs must do and how they should be deployed rather than prescriptive technical standards, funded and focused industry engagement, high-level policy and regulatory support, and resulting commercial responses. In essence, this more recent approach to policymaking has emphasized “what” rather than “how,” with an agile technical and policy mindset. Such a model is likely to more durable and less dependent on federal rules, funding, or timetables than prior attempts to change health IT directions.
Such questions are especially timely as the federal government is considering policies for its health IT incentive programs; health information exchange and interoperability, “information blocking,” and health IT certification; digital health product safety; and evaluation and management codes.
These are key points to track:- Policy views are theories subject to verification
- Evidence of a problem does not equal evidence that a policy intervention is warranted, will work, or is worth the cost
- Unintended consequences are inevitable
- “Market failure” is in eye of the beholder
- Logrolling leads to policy bloat
- Compliance eats policy intentions for lunch (apologies to Peter Drucker)
- We are not all alike: segmentation works in marketing and public policy
- Policy deadlines are made to be broken, and we all know it
- Policy innovations run their course, as do the organizations charged with their execution
- Emphasize “diffusion of innovations” in lieu of or along with detailed regulation
Policy views are theories subject to verification
Too often, by the time policies are proposed, assertions that began in the literature as theories tested against data no longer reflect underlying uncertainties. By their nature, laws and regulations assume unwarranted confidence about cause and effect and what people and organizations want and will do. In developing interoperability policies, for example, it is important to know: do doctors want all the information in the Consolidated CDA document; are providers more focused on cross-organization rather than internal interoperability as suggested by health IT incentive programs, and do patients really want electronic access to all of their electronic health data? Maybe yes, maybe no, maybe sometimes.
This shift from theory to certainty is inevitable in advocacy and policymaking; nonetheless, policy development should acknowledge and reflect uncertainty. It can set more modest aims or build-in opportunities to assess, validate, and adjust. Policymakers can use agile policy and implementation models that allow for stakeholder feedback (and adjustment), to continually test policies against data and experience, to use a scientific model.
Evidence of a problem does not equal evidence that policy intervention is justified, will work or is worth the cost
The currency of health policy is technical analysis, with many footnotes, that documents a problem and ends with policy recommendations that usually call for new programs and requirements that people or organizations do certain things. Unfortunately, documentation of a problem is usually more robust than evidence for the feasibility or net benefits of interventions.
Unintended consequences are inevitable
It is a truism that unintended consequences are inevitable. Agile and iterative approaches to policymaking and implementation can help reduce such risks. The most critical way to mitigate this risk is to anticipate, to think analytically about what could occur, to listen to affected parties, to listen some more, to imagine what might happen, and to have the courage to act on these insights. More generally, it is not a bad idea to be modest and prioritized in policy aims. I have witnessed several unintended consequences:
- CMS/AMA Documentation Guidelines for evaluation and management codes (E&M) moved from guidelines, their initial purpose, to rules for audits to drivers of EMR design (“too many bullets”) to physician unhappiness. Addressing these issues has been a top CMS and ONC initiative in their broader project to reduce physician burden. Indeed, CMS recently proposed a major overhaul in E/M documentation and payment policies. Notably this proposal, which reduces the importance of E/M billing documentation by collapsing payments for all but the lowest level of code to a single payment level is ripe with the potential for unintended consequences and has sparked significant criticism.
- Electronic Clinical Quality Measures (eCQMs): We have seen a shift from “use data already in the EMR” to “add structured data elements to the EMR to support eCQMs” to physician burnout and hospital dissatisfaction. CMS has responded with its “meaningful measures” initiative.
- CMS/ONC Meaningful Use and certification programs: These have moved from “incentivize use of fully functioning health IT” to “require vendors and physicians to do what we think they should do” to EHR design by regulation and committee to reduced usability.
“Market failure” is in eye of the beholder
Market failure is a classical rationale for healthcare regulation. The argument is often made that “we will only try to regulate where the market won’t produce the right outcome or action”. The reality is usually less rigorous, “they won’t do what I think they should, so the market has failed”. Policymakers should resist the temptation to use regulation to make people do what you think they should do. Markets, however imperfect, can allow for incremental adjustment to individual preferences that regulation does not. Always ask: why aren’t they doing what I think they should do and what are the costs and prospects for potential corrective policies?
Logrolling leads to policy bloat
Logrolling is the political concept of “you add my policy, I’ll add yours and we’ll both be happy.” Intentions are usually good and policymakers typically well-meaning but, see “unintended consequences.” It is often used in legislative bodies and committees (e.g., federal advisory committees) where support of others is needed to move one’s own agenda. Examples of negative consequences are the breadth and depth of Meaningful Use Stages 1 and 2 and almost any major health policy bill. Logrolling is inevitable and necessary to legislating and policymaking but it is also critical to recognize its risks and try to limit its impact, for example as the Administration evaluates advisory committee recommendations.
Compliance eats policy intentions for lunch
A familiar pattern in public policy is that initiatives are distorted when compliance enters, often by organizations and individuals who do not “own” the initial policy (e.g., auditors, inspectors general, etc.). E&M documentation guidelines became rules that became the basis for audits and technology (e.g., design of EHR documentation) and process initiatives (e.g., healthcare organization compliance programs). Similarly, Meaningful Use incentives to adopt technology became a complex compliance risk for providers via Medicare and state Medicaid audits. Investments in compliance generate costs and waste and discourage participants. Policymakers should anticipate compliance implications and design to minimize these costs, for example by ensuring that provider guidance can reduce compliance risks while enhancing compliance program efficiency.
We are not all alike: segmentation works in marketing and public policy
Laws, regulations and interest groups seek uniform policy implementation, with variations limited to size (e.g., small practices) or location (e.g., rural practices). But clinicians and health care organizations vary along many dimensions, not just economic and demographic. Drawing on a marketing concept, segmentation by shared interests and behaviors is more robust than simple geographic or demographics classification. Policymakers should seek solutions that allow for and can reflect varying (and often unknown) needs, interests, and priorities. The CMS Center for Medicare and Medicaid Innovation (CMMI) seems to be following this approach with its multiple, mostly voluntary initiatives. CMMI’s early models demonstrated the importance of testing a diverse set of approaches “. . ., [b]ecause every provider, practice, and system has a unique set of concerns, priorities, and resources, the diversity of CMMI’s models enabled significant learning and innovation,” according to this article.
Policy deadlines are made to be broken, and we all know it
Those who make and are affected by health policy have become accustomed to delayed regulations and missed deadlines. Repeated delays waste effort and investment, penalize and discourage early adopters, and breed cynicism and complacency. Examples include the false starts for implementation of ICD-10 codes, postponed implementation of the ONC 2015 EHR certification, and delays in Stage 3 of Meaningful Use and related programs. Given the likelihood of such delays, policymakers should anticipate objections and roadblocks to their proposals and avoid those likely to require significant and costly timetable adjustment. Above all, they should listen, with an open mind and open heart, to those likely to be affected by a policy and those able to predict its course. One way to avoid the need for delays and to minimize their costs is a sequenced and iterative approach to policy development and implementation of “crawl, walk, and run”.
Policy innovations run their course, as do the organizations charged with their execution
Part of the challenge faced by policymakers is when to realize that an initiative is nearing end of life, and what to do about that. Policies, like organizations, have lifecycles felt both by the organization and those it serves. Congress and CMS leaders have recognized this dynamic for the EHR incentive programs, framed as “meaningful use fatigue,” reinventing “Meaningful Use” as Advancing Care Information and then Promoting Interoperability, narrowing focus on interoperability and patient data access rather than a broad set of health IT-related activities, and adjusting organizational focus.
Emphasize “diffusion of innovations” in lieu of or along with detailed regulation
Health policies aim to change behaviors of organizations, clinicians, and patients. In doing so, they face daunting challenges, including the headwinds reflected in this checklist. In essence, they are pursuing diffusion of innovations, such as enhanced documentation, use of health IT, greater health data exchange, increased quality, etc. Innovations diffuse over a long time, and unevenly within populations (see segmentation). There is extensive literature on how innovations (including healthcare) spread. Policymakers should incorporate these perspectives into policy design and harness what we know about diffusion of innovations to achieve lasting, organic, and durable changes instead of “check the box” responses.
This perspective was reflected in a December 2017 report by the JASON federal advisory committee, “Artificial Intelligence for Health and Health Care.“ JASON noted that “[t]he process of developing a new technique as an established standard of care uses the robust practice of peer-reviewed R&D, and can provide safeguards against the deceptive or poorly-validated use of AI algorithms (p 2). Dearing and Cox, in an excellent review of diffusion of healthcare innovations state that “[p]urposive dissemination, or designing for diffusion, means taking additional steps early in the process of creating an innovation to increase its chances of being noticed, positively perceived, adopted, adapted, and implemented—and, thus, successfully crossing the research-to-practice chasm.”
Conclusion
Regulations work, but at a cost, and often with fragile results. For example, Meaningful Use and related initiatives spurred a digital healthcare environment, opening opportunities for new solutions and applications and to data use, but at the cost of significant concerns about usability and clinician burden, with data quality still not where we would like it to be. In the framework of diffusion of innovations, this program shifted out the cost and effectiveness curve for the economic buyers of EHRs but not for many end users. A more modest and focused approach could also have achieved high levels of digitization, but with more organic clinician uptake.
More recently, we have seen how a targeted and flexible approach to health IT policymaking, well aligned with this checklist, has led to rapid roll-out of API-based interoperability using the HL7 FHIR® standard, driven by modest investments by the Office of the National Coordinator for Health IT, functional requirements on what APIs must do and how they should be deployed rather than prescriptive technical standards, funded and focused industry engagement, high-level policy and regulatory support, and resulting commercial responses. In essence, this more recent approach to policymaking has emphasized “what” rather than “how,” with an agile technical and policy mindset. Such a model is likely to more durable and less dependent on federal rules, funding, or timetables than prior attempts to change health IT directions.
More for you
Loading data for hdm_tax_topic #better-outcomes...