Posts Tagged ‘ACA’
By: Fabrício Kury, M.D.
It is easy for a millennial – a person born between mid-1980’s to late-90’s – to be unaware of just how young the current methods used in health care research really are. Controlled randomized clinical trials (RCT), dating only from the late 40’s, are probably younger than most millennial’s grandparents. Case-control methodology and Kaplan-Meier curves only originated in the 1950’s, while meta-analyses were only accepted by medical researchers in the late 70’s. Step into the 80’s and early millennials are as old, if not older, than propensity scores and the concept that is today called cost-effectiveness research. The term “Evidence-Based Medicine” is as young as a millennial born in the early 90’s, while the late 90’s and 2000’s saw the explosion of genomics, proteomics, metabolomics, and other -omics research. Finally, the 2010s so far might be credited for when the term Data Science (“the fourth paradigm of science“) gained widespread notoriety, and established its modern meaning as – long story made short – the practice of producing knowledge out of data that had been created for other purposes.
While the second half of the 20th century transformed health care research into an ever more rigorous and technology-driven science, it also saw the cost of the health care sector of the U.S. unrelentingly grow from a comfortable 5% of the Gross Domestic Product in 1960 to a crushing 18% in 2015. Medical bills have become the leading cause of personal bankruptcies in the nation, while life expectancy, as well as other basic health indicators, depicted a country nowhere close to getting a similar bang for each buck as other developed nations. In 2009, the Obama administration prescribed to the health care sector a remedy that had previously brought efficiency and cost savings to every industry it had previously touched: information technology. The Health Information Technology for Economic and Clinical Health (HITECH) Act (part of the American Recovery and Reinvestment Act of 2009) literally gave away as much as $36.5 billion of taxpayers’ money to hospitals and physician practices for them to buy and “meaningfully use” electronic health records (EHRs). This outpouring of money was overseen by the Office of the National Coordinator of Health Information Technology (ONC), which had existed since 2004 as a presidential Executive Order, but became solidified as a legislative mandate via HITECH. This act fiercely transitioned the country from mostly paper-based health care in 2008 to near-universal EHRs adoption by 2015, giving electronic life, and potential reuse for research, to streams of health data previously dormant in paper troves.
Moreover, in March, 2010, the Patient Protection and Affordable Care Act (PPACA, a.k.a. “Obamacare”) was signed into law and, among so many other interventions, secured a few hundred million dollars for the creation of the Patient-Centered Outcomes Research Institute (PCORI). The mission of the PCORI is to do research that responds directly to real-life concerns of patients. For that purpose, among the first initiatives by the PCORI was the creation of PCORnet, a network of institutions capable of providing electronic health data for research. Most recently, in January 2015, President Obama announced the Precision Medicine Initiative (PMI). The PMI seeks to craft a nationwide and representative cohort of 1 million individuals, from whom a wealth of health data will be collected with no definitive goal besides to serve as a multi-purpose prime-quality dataset for observational electronic research. Meanwhile, private sector-led initiatives such as Informatics for Integrating Biology and the Bedside (i2b2) and Observational Health Data Sciences and Informatics (OHDSI) were also launched with the mission to access and do research on health care’s big data, and their publications can be easily found in PubMed.
These initiatives depict a political and societal hope – or hype? – that information technology, among its other roles in health care as whole, can make health care research faster, broader, more transparent, more reproducible, and perhaps also closer to the everyday lives of people. One premise is that by using existing EHRs for research, instead of data collected on-demand for a particular study, the researcher gets closer to the “real world” individuals that ultimately receive the treatments and conclusions produced by the study. In traditional clinical trials and other studies, the patients who participate are highly selected and oftentimes remarkably unrepresentative of the general population. Moreover, in EHR-based research there is also the potential to investigate more individuals than any previous method could possibly attempt. This broader reach makes rare conditions (or combinations of conditions) not so rare that they cannot be readily studied, and allows subtler variations in diseases to become detectable. On top of that, these studies can be done at the speed of thought. De facto, electronic health records-based clinical research has been recently published in the Proceedings of the National Academy of Sciences (PNAS) and evinced to be feasible at international, multi-hundred million patients scale at a breathtakingly swift time span. Altogether, one can sense in this picture that the millions of dollars spent on HITECH, PCORnet, PMI, and the NIH’s Data Science research grants might not have been just unfounded hype.
The relationship of IT and health care must, however, recognize its rather long history of frustrated expectations. In 1968, for example, Dr. Laurence Weed – the father of today’s prevailing paradigm of patient notes – predicted that in the future all text narratives present in electronic health records would be entered in structured form that enables scientific analysis. Today, to say the minimum, we have become less confident about whether such change is feasible or even desirable to begin with. In 1987, Barnett and colleagues believed that “relatively simple computational models” could be used to construct “an effective [diagnostic] assistant to the physician in daily practice” and distributed nationwide, but such assistant is yet to arrive at your physician’s office downtown (although, truth be recognized, it might be around the corner). While presently teaming with excitement and blessed with incentives, the journey of IT into health care and health care research is invariably one of uncertainties and risks. Health information technology has been accused of provoking life-threatening medical errors, as well as – like previous technological breakthroughs along the history of Medicine, including the stethoscope – harming the patient-physician relationship and the quality of care. The editors of the New England Journal of Medicine early this year went as far as to state that data scientists are regarded by some clinical researchers as “research parasites.”
Moreover, the Federal Bureau of Intelligence has investigated that medical information can be sold on the black market for 10 times more than a credit card number, while at the same time cybersecurity experts are stunned by the extreme vulnerability of current U.S. health care facilities. This provides sensible ground for concern about patient privacy violation and identity theft once the health records have moved from papers into computers. Unlike a credit card, your medical and identity information cannot be cancelled over the phone and replaced by a new one. Patient matching, i.e. techniques for recognizing that data produced at separate sites refer to the same person, oftentimes confronts blunt opposition by civil opinion, while the ultimate ideal of a National Patient Identifier in the U.S. is explicitly prohibited by present legislation (HIPAA). Such seamless flow of interoperable health data between providers, however, is the very first recommendation expressed in 2012 by the Institute of Medicine for realizing the Learning Health Care System – one that revolves around the patient and where scientific discovery is a natural outgrowth of patient care.
With or without attaining the ideal of a Learning Health Care System, the U.S. health care system will undergo transformation sooner or later, by intervention or by itself, because the percentage of the GDP that is spent on health care can only continuously increase for so long. Information technology is at minimum a sensible “bet” for improving efficiency – however, the power of IT for improving efficiency lies not in greasing the wheels of existing paradigms, but in outclassing them with novel ones. This might be part of the explanation for the resistance against IT, although there does exist some evidence showing that IT can sometimes do more harm than good in health care, and here the word “harm” sometimes can mean patient harm. The cold truth is that, in spite of decades of scientific interest in using computers for health care, only very recently the health care industry became computerized, so we remain not far from the infancy of health care informatics. Nevertheless, Clinical Informatics has been unanimously approved in 2011 as a board-certified physician subspecialty by the American Board of Medical Specialties, signaling that the medical community sees in IT a permanent and complex duty for health care. Similarly, the NIH has in late 2013 appointed its first Associate Director for Data Science, also signaling that this novel field holds importance for health care research. Finally, there might be little that can be done with the entire -omics enterprise, with its thousands over thousands of measurements multiplied by millions of patients, that does not require data-scientific techniques.
The first cars were slower than horses, and today’s high-speed, road-only automobiles only became feasible after the country was dependably covered with a network of roads and freeways. Such a network was built not by the automobile producers, but by the government upon recognition that it would constitute a public good. The same principle could very well be the case of health care IT’s important issues with privacy, security and interoperability, with the added complication that it is easy for an EHR producer to design a solution but then block its users from having their system interact with software from competing companies. Now that health care records are electronic, we need the government to step in once again and build or coordinate the dependable freeways of health care data and IT standards, which will also constitute a public good and unlock fundamental potentials of the technology. Health care, on top of its humanitarian dimension, is fundamentally intensive in data and information, so it is reasonable to conjecture that information technology can be important, even revolutionizing, for health care. It took one hundred years for Einstein’s gravitational waves to evolve from a conjecture based on theoretical fundaments to a fact demonstrated by experiments. Perhaps in the future – let us hope not a century from today! – some of the data-scientific methods such as Artificial Neural Networks, Support Vector Machines, Naïve Bayes classifiers, Decision Trees, among others, in the hands of the millennials will withstand the trial of time and earn an entry at the standard jargon of medical research. Just like how, in their generations, meta-analyses, case-control studies, Kaplan-Meier curves, propensity scores, and the big grandpa of controlled randomized trial were similarly accepted.
By: Felisa Gonzales, Ph.D.
The appropriate role of science in policy making has been debated for centuries. Most theories of decision-making posit that decisions, including policy decisions, are based on beliefs and values. How best to incorporate scientific knowledge into policymakers’ beliefs and values is unclear, and doing so is particularly difficult when the science is not definitively conclusive. The challenges inherent to the use of science to inform policy were clearly demonstrated when the Patient Protection and Affordable Care Act (ACA) mandated free health insurance coverage for certain preventive services based on the science-based recommendations of the United States Preventive Services Task Force (USPSTF). The USPSTF is an independent, volunteer panel of experts and clinicians from the fields of preventive medicine and primary care charged with evaluating the scientific evidence regarding the benefits and harms of clinical preventive services. The target audience for the USPSTF recommendations is primary care clinicians, not policymakers. Because the use of their recommendations has been codified into law, the scientifically and clinically oriented USPSTF is now occupying a policy role for which it was neither designed nor intended. The scientific conclusions of the USPSTF may not match the beliefs and values of democratically elected policymakers, raising the question: Is USPSTF the appropriate body to be put in the position of determining coverage policy?
The direct linkage of the USPSTF’s science-based recommendations to health insurance coverage came to public attention in 2009 when the panel changed its breast cancer screening recommendation. Based on data from randomized trials of mammography screening, the USPSTF made age-specific recommendations that advised against routine breast cancer screening for women between the ages of 40-49 and called for women to weigh “the potential benefit against the potential harms” before deciding to initiate mammography before the age of 50. Although women older than 40 have long heard messages such as “screening saves lives” and “take the test, not the chance”, researchers had been expressing their uncertainty about the benefits of mammography for women in their 40s since at least 1993. Nevertheless, the 2009 recommendation was criticized as “gender genocide”, “incredibly flawed”, “disastrous for women’s health”, and “callous and poorly conceived”. Despite the backlash, the USPSTF reiterated the same recommendation in January of 2016.
A review of the available evidence in 2016 indicated that for every 10,000 women ages 40-49 screened for breast cancer, approximately 1,212 false-positives will result, 164 biopsies will be conducted, 10 cancers will be missed. With repeat screening over 10 years, only 4 breast cancer deaths among women ages 40-49 will be avoided. Based on this information, the USPSTF gave mammography for women ages 40-49 a grade of “C”, which indicates that “there is at least moderate certainty that the net benefit [i.e., the degree to which the benefits outweigh the harms] is small”. Only clinical preventive services with “A” or “B” grades, which indicate a moderate to high degree of certainty that the benefits outweigh the harms of a procedure by a moderate to substantial margin, are required to be completely covered by health insurers under ACA. The USPSTF grade definitions include assessments of certainty because science is not often absolutely conclusive. Commenting on the role and responsibility of expert bodies, the Organisation for Economic Co-Operation and Development Committee for Scientific and Technological Policy notes, “the policy and societal context for scientific advice is challenging, not only because the stakes are high, but also because the general expectation is that science can provide clear and unambiguous answers. The reality is that the results of scientific research are often provisional and sometimes heavily contested…” The fact that the USPSTF recommendations are based on the best available science and are of superior quality was not enough to convince policymakers that they were sufficient to be the sole determinant of coverage for mammography.
The “C” rating for mammography among women ages 40-49 was not a recommendation against screening or against coverage, but because the ACA linked the USPSTF recommendations to coverage decisions, some incorrectly interpreted it this way. As a “C” rating would result in mammography not being covered as a preventive service under ACA, Senators Barbara Mikulski (D-MD) and David Vitter (R-LA) drafted amendments requiring insurance plans to pay for annual mammograms for women ages 40 and older and not restrict mammography based on USPSTF recommendations. Mikulski’s amendment also included screenings for ovarian and lung cancer screening despite a lack of evidence of any benefit, and concerns about substantial harms, associated with these procedures. Members of the House and Senate have proposed additional actions including eliminating funding for future USPSTF recommendations and requiring people who are not experts in prevention or evidence-based medicine to serve on the panel (for example representatives from patient groups, specialty physicians, and relevant stakeholders from the medical products manufacturing community). Experts are wary of these efforts, noting that “political interference with science can discourage shared decision-making, increase harms from screening, and foster public doubt about the value and integrity of science.” These tensions highlight differences in scientists’ and policymakers’ beliefs and values despite a shared commitment to improved public health.
The USPSTF “is committed to using the best science to identify the most effective preventive services to improve the health of the public,” but carrying out this mission is much more complicated now that their recommendations are used to dictate health insurance coverage. One proposed solution, favored by the USPSTF and its critics, is the creation of a separate independent panel to be charged with reviewing the USPSTF screening recommendations as well as other considerations important for public policy, such as cost, context, and feasibility. As the current chair and members of the USPSTF have noted, “the science on effectiveness – although foundational – is only one factor that needs to be considered in developing policy coverage.” Others closely associated with the USPSTF have warned that “limiting first-dollar coverage to services supported by strong evidence of effectiveness, as determined by one panel, is potentially harmful for public policy and threatens the USPSTF and other independent panels like it.” The linkage of the USPSTF recommendations to health insurance coverage policy reminds us that scientists are not policymakers, and policymakers are not scientists. The development of evidence-based policy requires scientific advice that is “scientifically sound and politically suitable and legitimate at the same time.” In the absence of an independent, intermediate body that can consider both scientific beliefs and prevailing societal values in health insurance coverage decisions, we risk building walls rather than bridges between science and policy.