INSIGHTS | PERSPECTIVES
720 13 FEBRUARY 2015 • VOL 347 ISSUE 6223 sciencemag.org SCIENCE
The medical profession has long recog- nized the importance of randomized evaluations; such designs are com- monly used to evaluate the safety and efficacy of medical innovations such as drugs and devices. Unfortunately,
innovations in how health care is delivered
(e.g., health insurance structures, interventions to encourage the use of appropriate
care, and care coordination approaches) are
rarely evaluated using randomization. We
consider barriers to conducting
randomized trials in this setting
and suggest ways for overcoming
them. Randomized evaluations of fundamental issues in health care policy and delivery
should be—and can be—closer to the norm
than the exception.
There is particular interest in improving
delivery of health care in the United States,
where the health care sector accounts for
almost one-fifth of the economy. The newly
created Patient-Centered Outcomes Research Institute is providing an estimated
$3.5 billion in research grants, and the latest round of Center for Medicare and Medicaid Innovation Health Care Innovation
Awards provides about $1 billion in research
grants—much of it aimed at improving the
delivery of U.S. health care.
Studies of U.S. health care delivery typically rely on a range of observational and
quasi-experimental methods. These can be
extremely valuable for learning as much as
possible from existing historical data and for
studying questions that are not amenable to
randomized designs. For prospective evaluation of new interventions, however, it is often
possible to use a randomized design without
adding substantially to the cost or difficulty
Randomize
evaluations
to improve
health care
deliver y
HEALTH CARE POLICY
By Amy Finkelstein1,2,3*
and Sarah Taubman2
POLICY
Administrative data and
experimental designs lead
the way
dicting enantioselectivities based on simple
molecular descriptors (such as vibrational
frequencies and dipole moments) that characterize the reactants (7). These data are
easily obtained, obviating the need to compute all possible transition states for each
catalyst-substrate combination. Moreover,
they showed that classical physical organic
techniques can be effectively combined
with modern data analysis tools to yield
insights into the mechanisms of catalyzed
reactions and the role of noncovalent interactions in enantioselectivity.
To demonstrate the power of their approach, Milo et al. tackled a particular example of chiral anion catalysis, in which
enantioselectivity is induced by
the noncovalent association of
a cationic intermediate with a
chiral, anionic catalyst (see the
figure) (8, 9). To understand
these reactions, they synthesized and tested a library of catalysts exhibiting a broad range
of enantioselectivities. These
experiments provided a wealth
of data regarding the impact of
steric and electronic factors on
enantioselectivity, which was
then distilled into predictive
mathematical models through multivariate
regressions. These models unveiled subtle
factors that control the enantioselectivity
of these reactions and, ultimately, lead to
the design of better catalysts.
By embracing modern data analysis techniques to enhance the more traditional
tools of physical organic chemistry, Milo
et al. have provided a way to harness the
power of noncovalent interactions for the
design of enantioselective catalysts. Importantly, their approach is general and should
be applicable to a wide range of catalytic
reactions. This expands the power of the
simple linear free-energy relations that
have long been the workhorse of physical
organic chemistry, and provides a key step
toward a future in which big data can be
used to design small catalysts. ■
REFERENCES
1. A.Milo, A.J.Neel,F.D. Toste, M.S.Sigman, Science 347,
737 (2015).
2. E. H. Krenske, K. N. Houk, Acc. Chem. Res. 46, 979 (2013).
3. R. R. Kno wles, E. N. Jacobsen, Proc. Natl. Acad. Sci. U.S.A.
107, 20678 (2010).
4. S. E. Wheeler, J. W. G. Bloom, J. Phys. Chem. A 118, 6133
(2014).
5. K. N. Houk, P. H.-Y. Cheong, Nature 455, 309 (2008).
6. K. N. Houk, P. Liu,Daedalus 143, 49 (2014).
7. A. Milo, E. N. Bess, M. S. Sigman, Nature 507, 210 (2014).
8. M. Mahlau, B. List, Angew. Chem. Int. Ed. 52, 518 (2013).
9. A. J. Neel, J. P. Hehn, P. F. Tripet, F. D. Toste, J. Am. Chem.
Soc. 135, 14044 (2013).
10.1126/science.aaa5624
desired pathways (3). The benefit of such
an approach, which is typical of enzyme
catalysts, is that it should lead to greater
overall catalytic activity while retaining selectivity. Moreover, recent advances in our
understanding of noncovalent interactions,
including π-stacking and cation-π interactions, appear to have laid the groundwork
for the exploitation of these interactions in
rational catalyst design (4).
However, harnessing the power of noncovalent interactions for enantioselective
catalysis has proved difficult. Chief among
the reasons is the relatively weak, nondirectional nature of these interactions, necessitating the introduction of numerous
interactions that must operate in concert
to effectively stabilize the desired reaction
pathway (3). Rationally designing catalysts that achieve such coordinated effects
is fraught with difficulties. Indeed, even
identifying the noncovalent interactions responsible for selectivity in existing catalytic
reactions, which is a prerequisite for rational catalyst design, is often not straightforward based only on experimental data.
Computational quantum chemistry, in
which quantum mechanics is used to predict molecular properties by describing the
electronic motion, has proved invaluable for
understanding chemical reactions and even
designing catalysts (5, 6). It is routinely used
to understand enantioselectivities by predicting the structures and energies of the
operative transition states, while also quantifying the impact of noncovalent interactions on these structures. Unfortunately, for
many catalytic reactions, there are simply
too many potential transition-state structures (possibly hundreds) for such analyses
to be practical. For example, in noncovalent
catalysis, the catalyst and substrate can interact in a myriad of ways, and many such
transformations are not amenable to computational study with this direct approach.
Milo et al. have effectively circumvented
this problem by providing a means of pre-
“To demonstrate the power of their
approach, Milo et al. tackled a
particular example of chiral anion
catalysis, in which enantioselectivity
is induced by the noncovalent
association of a cationic intermediate
with a chiral, anionic catalyst.”
Department of Chemistry, Texas A&M University, College
Station, TX 77842, USA. E-mail: wheeler@chem.tamu.edu
