!
How$Could$We$Do$the$Most$Good$Here?$$$Using&the&ReThink&Health&Dynamics&Model&to&Evaluate&
&Combined&Intervention&Strategies&for&Local&Health&Systems&
Work&in&Progress!
Prepared by Jack Homer, PhD on behalf of the RTH Dynamics Team August 2013 $
The ReThink Health Dynamics Model is meant to help local leadership groups find a desirable
intervention strategy for improving the performance of their health systems. “Desirable” here
refers to an intervention strategy that meets the goals of the leadership group and is deemed
feasible in terms of cost and required implementation effort. The model simulates likely impacts
of proposed strategies along multiple outcome metrics, including health care costs, population
deaths, health inequity, and local economic productivity, out to the year 2040. A strategy
combining several different interventions may be necessary in order to achieve effective impact
across all of these outcome metrics simultaneously in a way that no single intervention can do.
The funding situation is an important consideration in identifying a desirable mix of
interventions. We have previously described how a community (“Anytown”) can be severely
constrained if all it has to work with is a one-time “innovation fund”; but how a mechanism of
“savings capture and reinvestment” can open up the possibility of action along multiple fronts
and substantially increase the combined impact (Homer, August 2013, “Introduction to the RTH
Dynamics Model”). With savings capture and reinvestment, the rapid cost savings from an
intervention like care coordination can become the engine that supports other interventions that
themselves are not rapidly cost saving but may offer other valuable benefits, like improving
health or reducing inequity. But that previous analysis also revealed that, even with savings
capture and reinvestment, funding can still become constrained when expensive interventions
Work&in&Progress! ! How!could!we!do!the!most!good?!&&!Page!2!!
that do not quickly pay for themselves are layered into the mix. Such overburdening of the
funding mechanism does not bring all progress to a halt, but it does have the effect of diverting
some funds away from other interventions, limiting their impact and also limiting the amount of
cost reduction that can generate savings for continued funding. In this way, the layering in of
any additional interventions that do not offer immediate cost savings of their own can be a
burden that, at least for some period of years, possibly causes overall health system performance
to become worse than it would have been without those additional interventions in place. Thus,
it is not in general the case that better outcomes can be achieved simply through the layering in
of more interventions; even with savings capture and reinvestment one must take care not to
spread available funds too thin.
We have also shown previously that the simulation-based ranking of individual
interventions along multiple outcome criteria depends somewhat on characteristics of the place
in question (Homer, August 2013, “Does Place Matter for Policy?”). These results helped us to
identify which interventions were likely most impactful, and in what sorts of places, but it did
not consider combined interventions, nor the role of funding limitations, nor how movement to a
new payment scheme (contingent global payments) or expansion of insurance eligibility under
ACA might affect the conclusions.
This bring us to the present analysis. Here we start by assuming the presence of savings
capture and reinvestment, so that several interventions may be combined in a powerful strategy
to address all outcome metrics of interest. The goal is to learn what the model can tell us about
which combined strategies are “best”, and how the selection of best might be affected by (a)
place (we again consider three: Anytown, Atlanta, and Morris County, New Jersey), and (b) four
possible scenarios of payment scheme and insurance eligibility.
Work&in&Progress! ! How!could!we!do!the!most!good?!&&!Page!3!!
Approach$to$Testing$
We developed a series of 24 simulations (base run + 23 intervention strategies) applied to each of
the three places, and for each of four scenarios for possible changes in payment scheme or
insurance eligibility. (Thus, a total of 24 x 3 x 4 = 288 simulations was performed.) This
breadth of testing was done to determine what combinations of interventions would be most
effective in terms of the model’s outcome metrics, and how the answer to that question might be
affected by choice of place or scenario.
As in previous analyses, all interventions are assumed to begin in 2012 and stay in place
through the end of the simulation in 2040. For all of these runs, it is assumed that 50% of any
cost savings against benchmarks for the insured population become available to the community
for reinvestment in the selected interventions.
We previously described, in “Does place matter for policy?”, the three places to which
the current version of the model (Version 2c; Reference Guide, Homer, May 2013) has so far
been calibrated. Anytown represents a national average, with relatively high prevalence fractions
of disadvantage, uninsurance, risky behavior, crime, and environmental hazards; perhaps
surprisingly high for a modern, developed nation. Atlanta has an identical fraction of
disadvantage but even more uninsurance, and a higher age-standardized death rate and
disadvantaged fraction of deaths. Morris County, in contrast, has low prevalence fractions of all
risk factors, and its age-standardized death rate and disadvantaged fraction of deaths are
accordingly low as well.
The four scenarios (S) defined by payment scheme and insurance eligibility are as
follows:
Work&in&Progress! ! How!could!we!do!the!most!good?!&&!Page!4!!
(S1) CGP0, ACA0: Continued fee-for-service payment; no expansion in insurance
eligibility;
(S2) CGP1, ACA0: Transition to Contingent Global Payment (CGP) starts in 2012 and
covers almost all of the insured population by 2016; no expansion in insurance eligibility;
(S3) CGP0, ACA1: Continued fee-for-service payment; expansion of insurance eligibility
causes two-thirds reduction of uninsurance starting in 2014 and is complete by 2016, as has been
projected to occur under full implementation of Affordable Care Act (ACA);
(S4) CGP1, ACA1: Transition to CGP; and expansion of insurance eligibility.
The$23$Intervention$Strategies$
The 23 intervention strategies are a sequence of model runs testing various combinations of
interventions for improving health care or for reducing health risks. In “Does Place Matter for
Policy?”, each of the 21 intervention types was tested individually. But, once one starts to
consider combinations, the potential number of multi-pronged strategies (two at a time, three at a
time, etc.) grows quickly into the thousands, and it is not practical to test them all. To limit the
task to something feasible and meaningful, we reviewed the results of our previous analyses and
decided to focus on the most promising interventions, eliminating some interventions from
consideration. This left us with ten initiatives (with one of them, called PrimaryCare, actually
comprising three separate interventions grouped as a single initiative) to be tested in various
combinations. These ten initiatives (I)—with descriptions, main effects, and relative intervention
cost—are as follows:
(I1) Coordination: includes technology assessment updating and also subsumes generic
drugs and shared decision making interventions—reduces health care costs; has moderate
intervention cost;
Work&in&Progress! ! How!could!we!do!the!most!good?!&&!Page!5!!
(I2) Behavior: healthier behavior intervention applied to all demographic groups—
reduces costs and deaths and inequity; has high intervention cost;
(I3) Environment: environmental hazards intervention applied to all demographic
groups—reduces costs and deaths and inequity; has high intervention cost;
(I4) PreScreen: pre-visit consultation intervention—reduces costs; has moderate
intervention cost;
(I5) Discharge: post-discharge care intervention—reduces costs; low intervention cost;
(I6) SelfCare: self-care intervention—reduces deaths and inequity, but may raise costs;
has high intervention cost;
(I7) PrimaryCare: combines preventive/chronic care intervention, PCP efficiency
intervention, and recruit PCP (FQHC) intervention—reduces deaths, but may raise costs and
inequity; has moderate intervention cost;
(I8) HospInfect: hospital-acquired infections intervention—reduces costs and deaths, but
may indirectly raise inequity; has low intervention cost;
(I9) FamilyPath: family pathways anti-poverty intervention—reduces inequity and raises
economic productivity; has very high intervention cost;
(I10) MentalDisadv: mental illness care intervention targeted to the disadvantaged—
reduces inequity and raises economic productivity; has high intervention cost.
The sequence of 23 strategies starts with Coordination (I1) alone, which is the single
most effective and rapid intervention for reducing health care costs, generating cost savings that
can be reinvested for funding other interventions. Behavior (I2) is then layered in: it also
reduces health care costs in both the short and longer term, and is one of the most important
interventions with regard to deaths, inequity, and economic productivity; Coordination plus
Behavior (“CB”) becomes the foundation for all strategies to follow. Environment (I3) is similar
Work&in&Progress! ! How!could!we!do!the!most!good?!&&!Page!6!!
in its effects to Behavior, though not quite as cost-effective; eleven of the 23 tested strategies
include Coordination plus Behavior plus Environment (“CBE”).1
Other initiatives that rapidly reduce health care costs are PreScreen (I4) and, somewhat
less powerfully, Discharge (I5). The fact that these two are of low or moderate cost to
implement means that they can be layered into a package of interventions without worrying that
they will constrain and dilute funds. Accordingly, PreScreen with or without Discharge appears
in eleven of the 23 tested strategies.
Combinations of the first five initiatives (I1 to I5) provide the cost savings that make it
possible to invest in other initiatives that are not cost saving (at least in the shorter term) but
provide other value. One of the most important of these is Self-Care (I6), which is relatively
expensive to implement and would by itself raise health care costs, but is (after Behavior) the
single most important intervention for reducing deaths, and is also important for reducing health
inequity and raising economic productivity. (In “Does Place Matter for Policy?”, we saw that
this importance of Self-Care held true in all three locations.). This initiative appears in sixteen of
the 23 tested strategies; in eight cases as adjunct to the CB combination (creating “CBS”), and in
eight cases as adjunct to CBE (creating “CBES”).
The remaining four initiatives may be useful when used as adjuncts to CBS or CBES.
PrimaryCare (I7) and HospInfect (I8) can help to further reduce the death rate, while
FamilyPath (I9) and MentalDisadv (I10) can help to further reduce inequity and increase
productive value.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!1 The importance of interventions to reduce behavioral and environmental risks, as part of a strategic package for
health system reform, has been been demonstrated previously through system dynamics modeling at the national level (Milstein et al. 2010; Milstein et al. 2011).
Work&in&Progress! ! How!could!we!do!the!most!good?!&&!Page!7!!
Each of the 23 intervention strategies tested includes no more than six of the ten
initiatives listed above. This limit was set in part for the sake of simpler analysis, but also on the
premise that it would be difficult for a local group of leaders to commit to a far-reaching strategy
that contains more than six components, from the standpoint of ongoing project management and
oversight, even if funding were adequate to do so. The management of even six concurrent
initiatives may be unrealistic for some locales; and indeed, when the RTH Dynamics model is
used in local workshops, an effective limit of fewer than six of the above initiatives is often set.2
Nonetheless, six initiatives may be what it takes to make a substantial impact across all outcome
metrics, and it is useful to ask how such a broadly effective approach can best be configured.
Numerical$Results$
Numerical results of testing are presented in Tables 1 to 4, one table for each of the four assumed
scenarios. For each of the three simulated places, four cumulative outcome metrics (2012
through 2040) are presented for the scenario’s base run: Average healthcare costs per capita;
Average death rate per thousand (age standardized); Disadvantaged fraction of deaths (a measure
of inequity); and Cumulative productive value of the employed population (in billions of
dollars). (These same metrics were reported in “Introduction to the RTH Dynamics Model” and
“Does Place Matter for Policy?”) The remainder of each column reports, for each of the 23
tested intervention strategies, the percentage change from the base run result, with one decimal
point of precision. A green shaded cell indicates the best (or tied for best) result among all
strategies along a particular metric. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!2 For the present analysis, “capture savings” is a given for all runs and does not count as an initiative. “CGP” also does not count as an initiative here, but is rather an exogenous scenario condition present in half of the simulations performed. In local workshops to date, these two conditions have usually been described as interventions rather than as givens. If six interventions are allowed in such a workshop, then a team selecting “capture savings” and “CGP” would have only four additional allowed interventions.
Work&in&Progress! ! How!could!we!do!the!most!good?!&&!Page!8!!
Before considering what these results suggest about possible best strategies across
multiple metrics simultaneously, it is useful to look at the results one metric at a time. Here are
some observations on the varied impacts in Tables 1 to 4:
1. For all places and scenarios, the most important initiatives for healthcare cost reduction
are Coordination and Behavior, after which Environment, PreScreen, Discharge, and HospInfect
also contribute but by smaller amounts.
2. CGP reduces healthcare costs in the base run and also magnifies the impact of
Coordination on cost reduction. Because CGP is implemented starting in 2012, and healthcare
costs are assumed to be benchmarked to 2010, the additional healthcare cost reduction from CGP
translates into more cost savings available to the community for reinvestment. The additional
funds generated by CGP can make it easier to implement certain costly initiatives, such as
FamilyPath, MentalDisadv, and PrimaryCare, without diverting funds from other initiatives.
Thus, CGP makes it possible to include more initiatives in a combined strategy without causing
adverse effects due to spreading funds too thin.
3. Absent CGP, insurance expansion slightly magnifies the impact of cost reduction
initiatives, thereby enabling more cost saving and more funds available for reinvestment. With
CGP, this impact of insurance expansion disappears.
4. For all places and scenarios, the most important initiative for death reduction is
Behavior, followed by Environment, SelfCare, and HospInfect. The PrimaryCare initiative can
also reduce deaths significantly, but only when the payment scheme is CGP. CGP generates
more reinvestment funds to help with the implementation of PrimaryCare, as noted above, and it
also provides greater incentive to physicians to comply with preventive and chronic care
Work&in&Progress! ! How!could!we!do!the!most!good?!&&!Page!9!!
guidelines and to improve the efficiency of their practices. Absent CGP, the addition of
PrimaryCare to the CBS strategy can have neutral or even adverse marginal impact on deaths,
possibly doing more harm (spreading limited funds too thin) than good.
5. The initiative with the greatest potential for inequity reduction and productivity
improvement is FamilyPath, a potential that is, however, only fully realized in the presence of
CGP. Absent the additional cost savings generated by CGP, the high implementation cost of
FamilyPath leads to a shortage of funds, thereby blunting its impact on inequity and productivity
and doing more harm than good with regard to healthcare costs and death rate.
6. Aside from FamilyPath under CGP, the most important initiative for inequity
reduction is Behavior, followed by MentalDisadv, SelfCare, and Environment. The
MentalDisadv initiative, like FamilyPath, is specifically targeted to the disadvantaged, fully
aimed at reducing inequity. The Behavior, SelfCare, and Environment initiatives address all
people who can benefit from them, but because disadvantaged people tend to need more such
help than advantaged people do, these initiatives also have the effect of reducing inequity.
7. Aside from FamilyPath under CGP, the most important initiative for productivity
improvement is Behavior, followed by SelfCare, MentalDisadv, HospInfect, and Environment.
The MentalDisadv initiative increases productivity primarily by reducing absenteeism and
presenteeism due to mental illness; the SelfCare and Environment initiatives do so by preventing
severe chronic illness and death; the HospInfect initiative does so by preventing death.
$ $
Work&in&Progress! ! How!could!we!do!the!most!good?!&&!Page!10!!
Best$Strategy$Candidates$Based$on$Pareto$Efficiency$
If the results presented in Tables 1 to 4 pointed to one strategy that had top results across all four
outcome metrics, then the choice of best strategy would be clear. But such a fully dominant
strategy does not exist among the 23 strategies tested under any of the scenarios. One strategy is
consistently best for reducing healthcare costs (CBE + PreScreen + Discharge), but this strategy
is somewhat inferior along the other three metrics because it lacks key components, such as
SelfCare or MentalDisadv, that improve those metrics. But adding such components would
detract from the strategy’s cost-cutting ability, either by raising cost directly (as SelfCare does)
or by spreading funds too thin and forcing diversion of effort (as MentalDisadv does).
The field of multiple-criteria decision making (MCDM) addresses itself to such problems
in which there is no single strategy or option that is dominant across all outcome criteria of
interest. Qualitative and quantitative MCDM techniques are well developed and frequently used
for real-life decision problems. Well-known examples of such techniques include the Analytic
Hierarchy Process (Saaty 1980; Forman and Gass 2001; Bhushan and Rai 2004), Multiattribute
Utility Analysis (Keeney and Raiffa 1980), and Interactive Multiobjective Optimization
(Miettinen et al. 2008).
These MCDM techniques all start with the identification of a subset of strategies known
as Pareto efficient or non-dominated strategies. A Pareto efficient strategy is one for which no
other strategy under consideration dominates it—i.e., that is superior (or in a tie) with it across
all criteria of interest. The next step, after identifying the Pareto efficient strategies, is to elicit
preferences from the real-world decision makers that allow a single strategy to be selected as
“best”. This may be done most simply by presenting the strategies directly to the decision
makers and hoping they can reach a conclusion through discussion and voting; or, rather (as all
Work&in&Progress! ! How!could!we!do!the!most!good?!&&!Page!11!!
the formal techniques do) by asking various hypothetical choice questions that allow a
computable inference to be drawn about which of the Pareto efficient strategies should, in theory,
be most preferred by the decision makers.
Table 5 presents the subset of Pareto efficient (non-dominated) strategies, among the 23
strategies tested, for each combination of place and scenario. These were identified by pairwise
comparison of strategies with respect to the simulated outcomes (percentage change relative to
the base run, with single decimal point precision) reported in Tables 1 to 4. Pareto efficient
strategies are marked in Table 5 with a plus (+) sign. Green shading indicates a Pareto efficient
strategy that additionally is best (or tied for best) among all Pareto efficient strategies along at
least one outcome metric. Here are some observations from Table 5 (informed also by the
details in Tables 1 to 4):
1. Of the 23 strategies, 14 are Pareto efficient for at least one combination of place and
scenario, and two (CBE + PreScreen + Discharge, and CBES + Prescreen + Discharge) are
Pareto efficient for all 12 combinations of place and scenario. Four other strategies are Pareto
efficient for most combinations of place and scenario.
2. All 14 of the strategies that are ever Pareto efficient include both the Coordination and
Behavior (CB) initiatives, 12 include SelfCare (as part of CBS), 9 include PreScreen, 8 include
Environment (as part of CBE or CBES), 5 include Discharge, 4 include HospInfect, and 3
include MentalDisadv.
3. Most of the strategies that are ever Pareto efficient are well balanced with respect to
the four outcome metrics, but there is one notable exception. In the absence of CGP (Scenarios 1
and 3), the CBS + FamilyPath strategy overburdens the funds available and for many simulated
Work&in&Progress! ! How!could!we!do!the!most!good?!&&!Page!12!!
years constrains investments that can reduce healthcare costs and deaths. The reason that this
strategy is nonetheless Pareto efficient in 5 of the 6 columns for Scenarios 1 and 3 (all but
Atlanta-Scenario 1) is that its contribution to productive value (by gradually moving people from
disadvantage to advantage and more gainful employment) is still superior to that of any other
strategy, despite the budget-constrained path in getting there.
4. There are from 6 to 9 Pareto efficient strategies for any given combination of place
and scenario. Absent CGP (under Scenarios 1 and 3), the PrimaryCare initiative does not appear
in any of the Pareto efficient strategies, and the MentalDisadv initiative also sometimes falls out
of contention. With CGP (under Scenarios 2 and 4), sufficient funds are generated so that the
inherent value tradeoffs among the initiatives (cost vs. death vs. inequity and productivity) are
allowed to emerge without being masked by problems of insufficient funds. In Anytown and
Atlanta, where there are significant environmental hazards, this means that all four of the CBES
+ PreScreen–based initiatives are Pareto efficient. In Morris, where environmental hazards are a
lesser concern, this means that all seven of the CBS + PreScreen-based initiatives (that is, all of
the combined approaches using six interventions) are Pareto efficient.
Possible$Extensions$
The analysis here contributes to our understanding of intervention strategy for local health
systems, and puts the ReThink Health modeling team in a position to provide further-reaching
and more definitive answers than we have given previously. Nonetheless, there is still more
work that could be done along a few basic fronts:
Work&in&Progress! ! How!could!we!do!the!most!good?!&&!Page!13!!
1. The analysis here culminated with the identification of several possible candidates for
best strategy for any given combination of place and scenario, rather than identifying a single
winner. That last step would require the elicitation of preferences either by vote of decision
makers or through a formal quantitative technique of multicriteria optimization.
2. Only 23 interventions strategies were considered, a subset of the many strategies
possible. Not included at all in these strategies were the interventions of Crime Reduction,
Hospice, Malpractice Reform, Medical Home, Recruit PCP (General), and Student Pathways.
Further testing might include some of these interventions selectively to see whether they might,
in some instances, lead to additional Pareto efficient strategies. For example, perhaps Crime
Reduction could be part of a Pareto efficient strategy in Anytown, where crime is relatively high.
Or, perhaps Malpractice Reform could be part of a Pareto efficient strategy in Morris, where it
was one of the top interventions for reducing healthcare costs in the “Does Place Matter for
Policy?” analysis.
3. The analysis here considered only the three places to which the current version of the
model has so far been calibrated. When the model is calibrated to additional places, this sort of
strategic analysis should be extended to those places as well.
4. All analyses done to date have been deterministic, assuming baseline values for all
intervention-related constants: effect sizes, time constants, and implementation costs. A further
step could be to subject all identified Pareto efficient strategies to probabilistic Monte Carlo
testing to establish confidence intervals around each outcome metric. This would involve
sampling from probability distributions using the “min” and “max” values for the intervention-
related constants specified in the model’s Reference Guide (Table 3).
Work&in&Progress! ! How!could!we!do!the!most!good?!&&!Page!14!!
References$
Bhushan N, Rai K. Strategic Decision Making: Applying the Analytic Hierarchy Process. Springer-Verlag: London, 2004.
Forman EH, Gass SI. The analytical hierarchy process—an exposition. Operations Research 2001; 49(4): 469–486.
Homer J. ReThink Health: a simulation model of health system transformation. Reference guide for model version 2c, “Anytown, USA” calibration. For Fannie E. Rippel Foundation, Morristown, NJ; May 2013.
Homer J. Introduction to the ReThink Health Dynamics model: simulating local health reform in “Anytown USA”. For Fannie E. Rippel Foundation, Morristown, NJ; August 2013.
Homer J. Does Place Matter for Policy? The effect of local characteristics on intervention priorities in the ReThink Health Dynamics model. For Fannie E. Rippel Foundation, Morristown, NJ; August 2013.
Keeney R, Raiffa H. Decisions with Multiple Objectives: Preferences and Value Tradeoffs. New York: Wiley: New York, 1980.
Miettinen K, Ruiz F, Wierzbicki AP. Introduction to multiobjective optimization: interactive approaches. Lecture Notes in Computer Science 2008; 5252: 27-57. Chapter 2 in Multiobjective Optimization: Interactive and Evolutionary Approaches, eds. Branke J, Deb K, Miettinen K, Slowinski R; Springer, 2008.
Milstein B, Homer J, Briss P, Burton D, Pechacek T. Why behavioral and environmental interventions are needed to improve health at lower cost. Health Affairs 2011; 30(5). DOI: 10.1377/hlthaff.2010.1116.
Milstein B, Homer J, Hirsch G. Analyzing national health reform strategies with a dynamic simulation model. Am J Public Health 2010; 100(5):811-819.
Saaty TL. The Analytic Hierarchy Process: Planning, Priority Setting, Resource Allocation. McGraw-Hill: New York, 1980.
!Work&in&Progress! ! Title!&&!Page!15!!!
Table 1. Comparison of 23 intervention strategies along four cumulative outcome metrics through 2040, assuming continued fee-for-service payment scheme with no expansion of insurance eligibility (Scenario 1) Green shaded cell indicates best (or tied for best) result among all strategies along a particular metric.
Anytown Atlanta Morris Anytown Atlanta Morris Anytown Atlanta Morris Anytown Atlanta MorrisBase Run (Scenario 1) $9,535 $9,173 $9,215 7.543 7.946 6.185 46.6% 58.2% 21.8% $192.5 $1,110.2 $681.8
CARE COORDINATION, HEALTHIER BEHAVIORS, AND ENVIRONMENTAL SAFETY
Coord (with assessment updating) -4.0% -4.1% -5.3% -0.1% -0.2% -0.1% -0.5% -0.5% -0.7% 0.3% 0.4% 0.1%
Coord + Behavior (CB) -7.6% -7.7% -8.1% -6.4% -6.5% -5.6% -2.0% -2.1% -2.1% 1.4% 1.5% 0.6%
Coord + Behavior + Environment (CBE) -8.4% -8.2% -8.3% -8.7% -8.1% -5.9% -2.8% -2.6% -2.3% 1.6% 1.6% 0.6%
FURTHER FOCUS ON COST
CB + PreScreen -8.4% -8.7% -8.7% -6.4% -6.5% -5.6% -2.0% -2.2% -2.1% 1.4% 1.5% 0.6%
CB + PreScreen + Discharge -8.9% -9.2% -9.0% -6.4% -6.5% -5.5% -2.1% -2.2% -2.2% 1.5% 1.6% 0.6%
CBE + PreScreen -9.2% -9.2% -8.8% -8.7% -8.1% -5.8% -2.9% -2.7% -2.3% 1.6% 1.6% 0.7%
CBE + PreScreen + Discharge -9.7% -9.7% -9.1% -8.7% -8.0% -5.8% -2.9% -2.8% -2.4% 1.7% 1.7% 0.7%
FURTHER FOCUS ON HEALTH
CB + SelfCare (CBS) -5.8% -5.6% -6.9% -9.8% -9.7% -8.5% -3.2% -2.9% -4.0% 1.6% 1.6% 0.8%
CBS + PrimaryCare -5.4% -5.3% -5.6% -10.2% -9.6% -7.5% -2.5% -2.2% -2.4% 1.6% 1.4% 0.6%
CBS + HospInfect -6.1% -5.8% -7.1% -11.3% -11.0% -10.0% -3.1% -2.5% -3.8% 1.8% 1.7% 0.9%
CBE + SelfCare (CBES) -6.4% -5.9% -6.9% -11.0% -10.4% -8.2% -3.7% -3.2% -4.0% 1.7% 1.6% 0.8%
CBES + PrimaryCare -5.8% -5.5% -5.4% -11.0% -9.9% -7.1% -2.9% -2.4% -2.3% 1.5% 1.4% 0.6%
CBES + HospInfect -6.7% -6.1% -7.1% -12.5% -11.7% -9.8% -3.6% -2.8% -3.8% 1.8% 1.7% 0.9%
FURTHER FOCUS ON EQUITY & PRODUCTIVITY
CBS + FamilyPath -4.0% -3.4% -5.4% -3.3% -2.4% -3.7% -3.5% -2.3% -5.2% 2.2% 1.6% 1.0%
CBS + MentalDisadv -5.3% -4.8% -6.5% -9.3% -8.6% -7.9% -3.9% -3.4% -4.8% 1.9% 1.8% 0.8%
CBES + MentalDisadv -5.8% -5.1% -6.5% -10.3% -9.1% -7.6% -4.3% -3.6% -4.7% 1.9% 1.7% 0.8%
COMBINED APPROACHES USING SIX INTERVENTIONS
CBS + PrimaryCare + PreScreen + Discharge -6.7% -6.7% -6.3% -10.5% -9.8% -7.7% -2.8% -2.4% -2.6% 1.7% 1.6% 0.7%
CBS + HospInfect + PreScreen + Discharge -7.4% -7.1% -8.0% -11.3% -9.9% -9.9% -3.3% -3.1% -3.9% 1.9% 1.7% 0.9%
CBS + MentalDisadv + PreScreen + Discharge -6.6% -6.3% -7.3% -9.6% -9.0% -8.0% -4.2% -3.7% -4.9% 2.0% 1.9% 0.8%
CBES + PrimaryCare + PreScreen -6.4% -6.1% -5.7% -10.9% -9.8% -6.8% -2.9% -2.4% -2.2% 1.6% 1.4% 0.6%
CBES + HospInfect + PreScreen -7.4% -7.0% -7.6% -12.5% -11.6% -9.6% -3.7% -2.9% -3.8% 1.9% 1.8% 0.9%
CBES + MentalDisadv + PreScreen -6.5% -5.8% -6.9% -10.3% -9.0% -7.4% -4.3% -3.6% -4.6% 1.9% 1.8% 0.7%
CBES + PreScreen + Discharge -7.7% -7.4% -7.7% -11.3% -10.6% -8.2% -4.0% -3.5% -4.1% 1.8% 1.7% 0.8%
Percent change from Base Run
Death Rate per 1,000 stdized. Inequity (disadv. death %) Productive Value ($ billion)Healthcare costs per capita
Work&in&Progress! ! How!could!we!do!the!most!good?!&&!Page!16!!
Table 2. Comparison of 23 intervention strategies along four cumulative outcome metrics through 2040, assuming transition to Contingent Global Payment scheme with no expansion of insurance eligibility (Scenario 2)
Green shaded cell indicates best (or tied for best) result among all strategies along a particular metric.
Anytown Atlanta Morris Anytown Atlanta Morris Anytown Atlanta Morris Anytown Atlanta MorrisBase Run (Scenario 2) $9,217 $8,846 $8,862 7.360 7.742 6.035 46.3% 57.9% 21.6% $193.6 $1,117.6 $683.7
CARE COORDINATION, HEALTHIER BEHAVIORS, AND ENVIRONMENTAL SAFETY
Coord (with assessment updating) -7.1% -7.1% -8.2% -0.2% -0.3% -0.1% -0.9% -1.0% -1.1% 0.6% 0.7% 0.2%
Coord + Behavior (CB) -11.0% -11.0% -11.4% -6.5% -6.6% -5.6% -2.5% -2.6% -2.5% 1.7% 1.8% 0.7%
Coord + Behavior + Environment (CBE) -12.1% -11.6% -11.6% -8.9% -8.3% -5.9% -3.4% -3.2% -2.7% 1.9% 1.9% 0.7%
FURTHER FOCUS ON COST
CB + PreScreen -12.3% -12.3% -12.2% -6.5% -6.6% -5.6% -2.6% -2.8% -2.6% 1.8% 1.9% 0.7%
CB + PreScreen + Discharge -12.9% -13.0% -12.7% -6.5% -6.6% -5.6% -2.7% -2.9% -2.7% 1.8% 1.9% 0.7%
CBE + PreScreen -13.3% -13.0% -12.4% -9.0% -8.3% -5.9% -3.5% -3.4% -2.9% 2.0% 2.0% 0.7%
CBE + PreScreen + Discharge -13.9% -13.6% -12.9% -9.0% -8.3% -5.9% -3.6% -3.5% -2.9% 2.1% 2.1% 0.7%
FURTHER FOCUS ON HEALTH
CB + SelfCare (CBS) -9.5% -9.3% -10.5% -11.0% -11.6% -9.6% -4.3% -4.1% -5.2% 2.1% 2.1% 0.9%
CBS + PrimaryCare -9.4% -9.5% -9.9% -13.0% -13.6% -11.5% -4.1% -4.0% -4.8% 2.3% 2.3% 1.0%
CBS + HospInfect -9.8% -9.4% -10.7% -12.2% -12.6% -10.8% -4.2% -3.8% -5.0% 2.2% 2.2% 1.0%
CBE + SelfCare (CBES) -10.5% -9.9% -10.6% -13.2% -13.1% -9.8% -5.2% -4.7% -5.5% 2.3% 2.2% 1.0%
CBES + PrimaryCare -10.3% -10.0% -10.0% -15.0% -15.0% -11.7% -4.9% -4.5% -5.0% 2.4% 2.4% 1.1%
CBES + HospInfect -10.7% -10.1% -10.8% -14.3% -14.1% -11.0% -5.0% -4.3% -5.2% 2.4% 2.4% 1.1%
FURTHER FOCUS ON EQUITY & PRODUCTIVITY
CBS + FamilyPath -9.3% -9.0% -10.4% -9.5% -9.4% -9.1% -10.0% -9.1% -11.8% 5.7% 5.6% 2.2%
CBS + MentalDisadv -9.3% -8.9% -10.3% -11.6% -12.3% -9.8% -5.5% -5.4% -6.7% 2.5% 2.6% 1.0%
CBES + MentalDisadv -10.2% -9.5% -10.5% -13.6% -13.7% -10.1% -6.3% -5.9% -6.9% 2.7% 2.7% 1.0%
COMBINED APPROACHES USING SIX INTERVENTIONS
CBS + PrimaryCare + PreScreen + Discharge -11.1% -11.2% -11.1% -13.1% -13.6% -11.4% -4.4% -4.2% -5.0% 2.4% 2.4% 1.1%
CBS + HospInfect + PreScreen + Discharge -11.6% -11.5% -11.9% -12.2% -12.6% -10.7% -4.5% -4.1% -5.2% 2.4% 2.4% 1.1%
CBS + MentalDisadv + PreScreen + Discharge -11.1% -10.9% -11.6% -11.6% -12.3% -9.8% -5.8% -5.6% -6.9% 2.6% 2.8% 1.1%
CBES + PrimaryCare + PreScreen -11.4% -11.1% -10.7% -15.1% -15.0% -11.7% -5.1% -4.7% -5.1% 2.5% 2.5% 1.1%
CBES + HospInfect + PreScreen -11.9% -11.4% -11.6% -14.4% -14.2% -11.0% -5.2% -4.5% -5.4% 2.5% 2.5% 1.1%
CBES + MentalDisadv + PreScreen -11.5% -10.9% -11.3% -13.7% -13.8% -10.1% -6.5% -6.1% -7.0% 2.8% 2.8% 1.1%
CBES + PreScreen + Discharge -12.3% -11.9% -11.9% -13.3% -13.1% -9.8% -5.4% -5.0% -5.6% 2.4% 2.4% 1.0%
Percent change from Base Run
Death Rate per 1,000 stdized. Inequity (disadv. death %) Productive Value ($ billion)Healthcare costs per capita
Work&in&Progress! ! How!could!we!do!the!most!good?!&&!Page!17!!
Table 3. Comparison of 23 intervention strategies along four cumulative outcome metrics through 2040, assuming continued fee-for-service payment scheme with expansion of insurance eligibility (Scenario 3) Green shaded cell indicates best (or tied for best) result among all strategies along a particular metric.
Anytown Atlanta Morris Anytown Atlanta Morris Anytown Atlanta Morris Anytown Atlanta MorrisBase Run (Scenario 3) $9,499 $9,209 $9,193 7.449 7.890 6.116 46.5% 57.9% 21.7% $193.6 $1,118.4 $682.7
CARE COORDINATION, HEALTHIER BEHAVIORS, AND ENVIRONMENTAL SAFETY
Coord (with assessment updating) -4.9% -5.4% -5.6% -0.1% -0.2% 0.0% -0.6% -0.6% -0.7% 0.4% 0.5% 0.1%
Coord + Behavior (CB) -8.4% -8.8% -8.3% -6.3% -6.5% -5.4% -2.1% -2.2% -2.0% 1.3% 1.4% 0.6%
Coord + Behavior + Environment (CBE) -9.1% -9.3% -8.4% -8.5% -8.1% -5.6% -2.9% -2.8% -2.3% 1.5% 1.5% 0.6%
FURTHER FOCUS ON COST
CB + PreScreen -8.9% -9.7% -8.7% -6.3% -6.6% -5.4% -2.1% -2.2% -2.1% 1.3% 1.5% 0.6%
CB + PreScreen + Discharge -9.4% -10.2% -9.0% -6.2% -6.5% -5.3% -2.2% -2.3% -2.1% 1.4% 1.5% 0.6%
CBE + PreScreen -9.6% -10.2% -8.8% -8.5% -8.1% -5.6% -2.9% -2.8% -2.3% 1.5% 1.5% 0.6%
CBE + PreScreen + Discharge -10.1% -10.7% -9.1% -8.5% -8.0% -5.6% -3.0% -2.8% -2.4% 1.6% 1.6% 0.6%
FURTHER FOCUS ON HEALTH
CB + SelfCare (CBS) -6.5% -6.8% -7.0% -9.8% -9.9% -8.0% -3.5% -3.1% -3.9% 1.6% 1.6% 0.7%
CBS + PrimaryCare -6.1% -6.6% -5.3% -10.6% -10.3% -6.4% -2.9% -2.6% -2.0% 1.6% 1.5% 0.6%
CBS + HospInfect -6.8% -7.0% -7.3% -11.1% -11.1% -9.6% -3.3% -2.7% -3.7% 1.7% 1.7% 0.9%
CBE + SelfCare (CBES) -7.0% -7.1% -7.0% -11.0% -10.6% -7.6% -4.0% -3.5% -3.9% 1.6% 1.6% 0.7%
CBES + PrimaryCare -6.6% -6.9% -5.2% -11.7% -10.9% -5.9% -3.4% -2.8% -1.9% 1.6% 1.5% 0.5%
CBES + HospInfect -7.3% -7.3% -7.3% -12.5% -11.8% -9.2% -3.8% -3.1% -3.7% 1.8% 1.7% 0.8%
FURTHER FOCUS ON EQUITY & PRODUCTIVITY
CBS + FamilyPath -5.6% -5.0% -5.3% -5.0% -3.1% -3.2% -5.4% -3.1% -4.7% 3.3% 2.1% 0.9%
CBS + MentalDisadv -6.1% -6.2% -6.6% -9.6% -9.2% -7.3% -4.3% -3.8% -4.6% 1.9% 1.8% 0.7%
CBES + MentalDisadv -6.6% -6.5% -6.6% -10.7% -9.8% -7.0% -4.6% -4.0% -4.5% 1.9% 1.8% 0.7%
COMBINED APPROACHES USING SIX INTERVENTIONS
CBS + PrimaryCare + PreScreen + Discharge -7.0% -7.7% -6.0% -10.8% -10.5% -6.6% -3.0% -2.7% -2.2% 1.6% 1.6% 0.6%
CBS + HospInfect + PreScreen + Discharge -8.0% -8.3% -8.0% -11.2% -11.0% -9.5% -3.4% -2.8% -3.9% 1.8% 1.8% 0.9%
CBS + MentalDisadv + PreScreen + Discharge -7.3% -7.5% -7.4% -9.7% -9.3% -7.4% -4.4% -3.9% -4.8% 1.9% 1.9% 0.7%
CBES + PrimaryCare + PreScreen -7.0% -7.3% -5.3% -11.7% -11.0% -5.6% -3.4% -2.9% -1.8% 1.6% 1.5% 0.5%
CBES + HospInfect + PreScreen -8.1% -8.1% -7.6% -12.5% -11.7% -9.1% -3.9% -3.1% -3.7% 1.8% 1.7% 0.8%
CBES + MentalDisadv + PreScreen -7.2% -7.1% -6.9% -10.7% -9.7% -6.8% -4.7% -4.0% -4.4% 1.9% 1.8% 0.7%
CBES + PreScreen + Discharge -8.3% -8.4% -7.7% -11.3% -10.6% -7.7% -4.2% -3.6% -4.0% 1.7% 1.6% 0.7%
Percent change from Base Run
Death Rate per 1,000 stdized. Inequity (disadv. death %) Productive Value ($ billion)Healthcare costs per capita
Work&in&Progress! ! How!could!we!do!the!most!good?!&&!Page!18!!
Table 4. Comparison of 23 intervention strategies along four cumulative outcome metrics through 2040, assuming transition to Contingent Global Payment scheme with expansion of insurance eligibility (Scenario 4)
Green shaded cell indicates best (or tied for best) result among all strategies along a particular metric.
Anytown Atlanta Morris Anytown Atlanta Morris Anytown Atlanta Morris Anytown Atlanta MorrisBase Run (Scenario 4) $9,202 $8,919 $8,840 7.272 7.691 5.970 46.3% 57.6% 21.6% $194.7 $1,124.8 $684.6
CARE COORDINATION, HEALTHIER BEHAVIORS, AND ENVIRONMENTAL SAFETY
Coord (with assessment updating) -7.1% -7.1% -8.1% -0.2% -0.2% -0.1% -0.9% -0.9% -1.0% 0.6% 0.6% 0.2%
Coord + Behavior (CB) -11.0% -11.0% -11.3% -6.3% -6.5% -5.4% -2.4% -2.5% -2.4% 1.5% 1.6% 0.7%
Coord + Behavior + Environment (CBE) -12.0% -11.6% -11.4% -8.7% -8.2% -5.6% -3.3% -3.1% -2.7% 1.7% 1.7% 0.7%
FURTHER FOCUS ON COST
CB + PreScreen -12.2% -12.4% -12.0% -6.4% -6.6% -5.4% -2.6% -2.7% -2.5% 1.6% 1.7% 0.7%
CB + PreScreen + Discharge -12.9% -13.1% -12.5% -6.3% -6.6% -5.3% -2.7% -2.8% -2.6% 1.6% 1.7% 0.7%
CBE + PreScreen -13.2% -13.1% -12.2% -8.8% -8.3% -5.6% -3.5% -3.3% -2.8% 1.8% 1.8% 0.7%
CBE + PreScreen + Discharge -13.8% -13.7% -12.7% -8.7% -8.3% -5.6% -3.6% -3.4% -2.8% 1.9% 1.8% 0.7%
FURTHER FOCUS ON HEALTH
CB + SelfCare (CBS) -9.4% -9.4% -10.4% -10.8% -11.5% -9.4% -4.3% -4.0% -5.3% 1.9% 1.9% 0.9%
CBS + PrimaryCare -9.3% -9.7% -9.6% -12.9% -13.6% -11.2% -4.2% -3.9% -4.7% 2.1% 2.1% 1.0%
CBS + HospInfect -9.6% -9.5% -10.6% -12.0% -12.6% -10.6% -4.2% -3.7% -5.0% 2.0% 2.1% 1.0%
CBE + SelfCare (CBES) -10.3% -10.0% -10.6% -13.0% -13.0% -9.6% -5.2% -4.6% -5.5% 2.1% 2.0% 0.9%
CBES + PrimaryCare -10.1% -10.2% -9.7% -14.8% -15.0% -11.4% -5.0% -4.5% -4.9% 2.2% 2.2% 1.0%
CBES + HospInfect -10.5% -10.2% -10.8% -14.1% -14.1% -10.8% -5.0% -4.3% -5.2% 2.2% 2.2% 1.0%
FURTHER FOCUS ON EQUITY & PRODUCTIVITY
CBS + FamilyPath -9.4% -9.1% -10.2% -9.3% -8.8% -8.8% -9.8% -8.4% -11.6% 5.4% 5.1% 2.2%
CBS + MentalDisadv -9.2% -9.1% -10.3% -11.4% -12.2% -9.6% -5.5% -5.3% -6.7% 2.3% 2.4% 1.0%
CBES + MentalDisadv -10.1% -9.7% -10.5% -13.4% -13.6% -9.8% -6.3% -5.8% -6.9% 2.5% 2.5% 1.0%
COMBINED APPROACHES USING SIX INTERVENTIONS
CBS + PrimaryCare + PreScreen + Discharge -10.8% -11.3% -10.8% -12.9% -13.6% -11.2% -4.4% -4.1% -4.9% 2.2% 2.2% 1.0%
CBS + HospInfect + PreScreen + Discharge -11.5% -11.6% -11.8% -12.0% -12.6% -10.5% -4.5% -3.9% -5.2% 2.2% 2.2% 1.0%
CBS + MentalDisadv + PreScreen + Discharge -11.1% -11.1% -11.5% -11.5% -12.2% -9.6% -5.8% -5.5% -6.8% 2.4% 2.5% 1.0%
CBES + PrimaryCare + PreScreen -11.1% -11.2% -10.5% -14.9% -15.0% -11.4% -5.1% -4.6% -5.0% 2.3% 2.2% 1.0%
CBES + HospInfect + PreScreen -11.8% -11.5% -11.5% -14.2% -14.2% -10.8% -5.2% -4.4% -5.4% 2.3% 2.3% 1.0%
CBES + MentalDisadv + PreScreen -11.4% -11.0% -11.2% -13.5% -13.7% -9.8% -6.5% -6.0% -7.0% 2.6% 2.6% 1.0%
CBES + PreScreen + Discharge -12.2% -12.0% -11.8% -13.1% -13.1% -9.6% -5.5% -4.9% -5.6% 2.2% 2.2% 1.0%
Percent change from Base Run
Death Rate per 1,000 stdized. Inequity (disadv. death %) Productive Value ($ billion)Healthcare costs per capita
Work&in&Progress! ! How!could!we!do!the!most!good?!&&!Page!19!!
Table 5. Pareto efficient (non-dominated) strategies for the four scenarios Plus (+) sign indicates a Pareto efficient strategy among the 23 considered. Green shading indicates a Pareto efficient
strategy with a best (or tied for best) result among all Pareto efficient strategies along at least one outcome metric.
Anytown Atlanta Morris Anytown Atlanta Morris Anytown Atlanta Morris Anytown Atlanta MorrisCARE COORDINATION, HEALTHIER BEHAVIORS, AND ENVIRONMENTAL SAFETY
Coord (with assessment updating)
Coord + Behavior (CB)
Coord + Behavior + Environment (CBE)
FURTHER FOCUS ON COST
CB + PreScreen
CB + PreScreen + Discharge
CBE + PreScreen + + +CBE + PreScreen + Discharge + + + + + + + + + + + +FURTHER FOCUS ON HEALTH
CB + SelfCare (CBS) +CBS + PrimaryCare
CBS + HospInfect + +CBE + SelfCare (CBES)
CBES + PrimaryCare
CBES + HospInfect + +FURTHER FOCUS ON EQUITY & PRODUCTIVITY
CBS + FamilyPath + + + + + + + + + + +CBS + MentalDisadv
CBES + MentalDisadv + +COMBINED APPROACHES USING SIX INTERVENTIONS
CBS + PrimaryCare + PreScreen + Discharge + +CBS + HospInfect + PreScreen + Discharge + + + + +CBS + MentalDisadv + PreScreen + Discharge + + + + + + + + +CBES + PrimaryCare + PreScreen + + + + + +CBES + HospInfect + PreScreen + + + + + + + + + +CBES + MentalDisadv + PreScreen + + + + + + + + +CBES + PreScreen + Discharge + + + + + + + + + + + +
Scenario 2 (CGP1, ACA0) Scenario 3 (CGP0, ACA1) Scenario 4 (CGP1, ACA1)Scenario 1 (CGP0, ACA0)