COMMENTS ON PCORI NATIONAL PRIORITIES
The PCORI team has surveyed a vast, critical and developing landscape with remarkable
concision. We present specific suggestions to complement or clarify the Priorities below,
but our major target is the larger framework: the research landscape the Program conveys
is broad, but does it adequately signal the need for “breakthrough” research that
challenges current paradigms? The breadth of the agenda indicates a willingness to
consider a range of proposals, but also suggests continuity, implicitly invoking past
research approaches. The Priorities and Research Agenda enumerate the many failings of
the current health care system but leaves implicit the overwhelming consensus that
immense changes are required to ensure effective, efficient patient-centered care. As this
critical new enterprise is launched, it is essential to distinguish between the less important
need, for more research, and the vastly more important one, for different research. In a
comprehensive survey such as this, space constrains opportunities to illustrate the failures
of conventional approaches and the urgent need for new ones, but omitting them also
sends a message. These comments highlight some potential opportunities to signal
dissatisfaction with current approaches and welcome alternatives.
Introduction
Comparative Assessments of Prevention, Diagnosis, and Treatment Options:
“distinct populations with specific health problems”: More accurately, patient populations
overlap and have multiple problems. This language invites specialized constituencies and
existing approaches.
However, mentioning some specific populations of interest could be helpful. Patients
who are very ill or who have multiple serious illnesses are understudied as a group, other
than from the perspective of advanced directives. Many solutions are proposed from the
perspective of primary care, which manages relatively few of these high-impact patients.
“emphasis on patient preferences and decision-making”: Are these concepts actually
meaningful to patients, or are they research constructs? Certainly, patients express
preferences, but they change over time. Is there a gold standard against which one can
compare observed patient decisions? Whether they are “informed” or not? Vast amounts
of information are potentially available to patients. Is there a correct amount or required
quality of information?
If we are concerned that patients are making decisions contrary to their interests, what
measurable outcomes might indicate poor decisions? Refusing recommended care?
Seeking alternate care? We require countable or measurable bad outcomes for such
research to develop. What is the relevant time frame in which to determine success? At
diagnosis, when the patient is “naïve” and thus an optimal source of community
preferences, or later, when he has encountered and adapted to altered expectations and
dysfunction? Some chronic diseases, such as cancer, result in highly stressful “crisis”
decisions, with anticipation of downstream decisions deferred not because they are
unlikely but because they are uncomfortable.
Finally, does the term “informed” imply only rational input? Seriously ill patients are
frightened, and their decisions are heavily influenced by emotion. These factors drive
overuse of medical care in seriously ill patients.
Improving Healthcare Systems:
“self-care”: This is the sole reference to the participation of patients in their care once
decisions have been made. (Other roles appear to be confined to the research agenda,
such as whether outcomes are meaningful to them and their contribution to research
design.) Why focus on so-called “physician extenders” rather than on patients’ roles in
their day-to-day care? Patients are certainly motivated and well placed to play a decisive
role in all medical activity that involves them. Their potential for preventing medical
errors is obvious, if underutilized, but larger roles are easy to identify. For example,
uncoordinated and discontinuous care is costly and dangerous. The patient may have the
greatest motivation of all stakeholders to avoid redundant diagnostic workups or
discontinuities in medication prescriptions that result from changes in treatment venue or
provider.
Communication and Dissemination:
Is the assumption that decisions are supported through providing “good quality”
information? Are there non-rational factors that distort outcomes? (See above)
Connotations of language are becoming more important (e.g., the proposal to remove
“cancer” from the terminology for low risk prostate neoplasms).
“ways to use electronic data (“e-health records”) to support decision-making”: There has
been little attention to how the EMR can disrupt decision-making, by producing
inaccurate information, useless warnings that produce “alarm fatigue” or create artificial
patient preferences at variance from decisions made in the absence of such exercises.
Virtually no clinicians rely on, for example, pain scores that are documented relentlessly,
and most progress notes contain inaccuracies produced by prepopulated fields or reliance
on editing copies of previous notes.
Addressing Disparities:
Ambiguity about the meaning of “disparities” confuses and undermines this objective.
The usual meaning is substandard care for patients who are “disadvantaged,” primarily
because of demographic differences from the culturally dominant group making medical
decisions. In this document, the meaning of “disparities” appears at times to mean “effect
modification,” which is almost certainly less important than bias in producing disparities
in patient outcomes.
Accelerating Patient-Centered and Methodological Research:
What is the conceptual link between patient-centered and methodological research? What
makes research “patient-centered”?
External CER Priorities appear to avoid direct inclusion of quality improvement, but
rather error prevention (safety) and avoiding overuse. At a minimum it ignores 1/3 of
Wennberg’s taxonomy: effective care, patient-sensitive care and demand-sensitive care.
One could argue that PCORI’s overarching goal is determining which care is effective,
but without development that is just a slogan, not a program with specific content.
Proposed Research Agenda
The comments above are relevant to the more detailed presentations and language of this
section.
General Comments:
Quality improvement per se is not addressed, although medical errors and practice
variations, which characterize poor care, are mentioned. Practice variation is introduced
in a constricted sense: “Does this area currently show wide variations in clinical practice?”
This leads to a narrow focused on 2 potential causes: lack of clear evidence for a
preferred choice or lack of awareness of evidence. Wennberg’s distinctions between
effective care, preference-drive care and demand driven care, which inform the concept
of practice variation, do not address additional procedures for identifying inferior
techniques or incorporating new information and technology (the latter was an IOM
Priority). The conventional pivotal trials framework addresses research “strategies” but
not the “tactics” of carrying them out; Wennberg’s analytic approach produced the
empirical argument that 1/3 of health care expenditures do not contribute to health, and
the Dartmouth Atlas documents variations in medical resource use, but the approach does
not identify useless care or suboptimal approaches.
Provider morale and medical education are both highly relevant to transformational
change and are not mentioned.
The choice to avoid of a “restricted set of conditions or treatments” misses an opportunity
to critique the unjustified dominance of the primary care paradigm. There is little
evidence that primary care provides an optimal fulcrum for coordinating complicated,
specialized care, and the experience with primary care “gatekeeping” in the 80’s and 90’s
is a strong counter-example. Therefore, it is unclear why primary care and the diseases it
primarily manages should be considered privileged perspectives. There are other
paradigms than that of diabetes/hypertension chronic care. Cancer features a crisis
management paradigm, with decisions made with a near-exclusive focus on the shortterm, a very different decision-making process highly relevant to the patient-cantered,
patient-focused emphasis of PCORI.
Comparative Assessments of Prevention, Diagnosis, and Treatment Options:
Patient-centered/decision-making: research should address patient adaptation and
learning in decision-analytic modeling. The theoretical basis of decision analysis requires
using the naive patient as the decision-maker and source of preferences and, to calculate
long-term outcomes, assuming no adaptation (at least in terms of preference
ascertainment). Yet patients do clearly adapt, becoming less naive and more “in control”
as they learn about their disease and adopt an updated perspective over time, just as aging
requires ongoing “new normal” perspectives. Failure to include the information that
patients adapt must result in less well-informed decisions. The longitudinal perspective
may be the most useful paradigm for studying patient decision-making in novel, useful
ways, rather than continuing to refine existing methodology focused on short-term patient
satisfaction. This static concept of the patients limits research.
Improving Healthcare Systems:
If we require “breakthrough” changes, this language may invoke too incremental changes.
This should be an area where the challenge to conventional wisdom and attachments to
current paradigms and techniques is the strongest. 1) Proposals to improve the use of IT,
not just expand it. “Meaningful use” criteria have no proven relationship with improved
patient outcomes; 2) There is an unacknowledged conflict between “overuse” of medical
technology and the global assumption that “more” IT is beneficial to medicine; 3) We
should exploration of “bad” uses of technology (e.g., potential for creating and
perpetuating erroneous information, too much information – complicating finding useful
information – and distraction).
Communication and Dissemination Research:
This should be explicitly linked to QA/QI programs? How does CME fit into this
objective? My guess is that it has little role in PCORI thinking, but it should be clear;
huge resources are invested in this model. There is a conflict between modeled decisionmaking, which usually assumes a parametric, multiple regression model and the way
patients think, which is largely binary. Research should emphasize decision thresholds
for using a variety of data, especially patient-reported, which is being incorporated into
the Labs section of patient medical records but has modest value as currently used.
Standards for public presentation of outcomes data would be useful?