Impact of a Sequential Intervention on Albumin

Utilization in Critical Care

Peter F. Lyu, MSPH1; Jason M. Hockenberry, PhD2; Laura M. Gaydos, PhD2; David H. Howard, PhD2;
Timothy G. Buchman, PhD, MD1,3; David J. Murphy, MD, PhD1,4

Objectives: Literature generally finds no advantages in mortality

risk for albumin over cheaper alternatives in many settings. Few
studies have combined financial and nonfinancial strategies to
reduce albumin overuse. We evaluated the effect of a sequential
multifaceted intervention on decreasing albumin use in ICU and
explore the effects of different strategies.
Design: Prospective prepost cohort study.
Setting: Eight ICUs at two hospitals in an academic healthcare
system.
Patients: Adult patients admitted to study ICUs from September
2011 to August 2014 (n = 22,004).
Interventions: Over 2 years, providers in study ICUs participated
in an intervention to reduce albumin use involving monthly feedback and explicit financial incentives in the first year and internal
guidelines and order process changes in the second year.
Measurements and Main Results: Outcomes measured were
albumin orders per ICU admission, direct albumin costs, and mortality. Mean (sd) utilization decreased 37% from 2.7 orders (6.8)
per admission during the baseline to 1.7 orders (4.6) during the
intervention (p < 0.001). Regression analysis revealed that the
intervention was independently associated with 0.9 fewer orders
per admission, a 42% relative decrease. This adjusted effect consisted of an 18% reduction in the probability of using any albumin

Emory Critical Care Center, Emory Healthcare, Emory University, Atlanta,

GA.
2
Department of Health Policy and Management, Emory University Rollins
School of Public Health, Atlanta, GA.
3
Department of Surgery, Emory University School of Medicine, Atlanta, GA.
4
Division of Pulmonary, Allergy, Critical Care and Sleep Medicine, Department of Medicine, Emory University School of Medicine, Atlanta, GA.
Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF
versions of this article on the journals website (http://journals.lww.com/
ccmjournal).
Dr. Buchmans institution received funding from the Society of Critical
Care Medicine for being Editor in Chief and from Philips Corp for attending a customer meeting. The remaining authors have disclosed that they
do not have any potential conflicts of interest.
For information regarding this article, E-mail: david.j.murphy@emory.edu
Copyright 2016 by the Society of Critical Care Medicine and Wolters
Kluwer Health, Inc. All Rights Reserved.
DOI: 10.1097/CCM.0000000000001638
1

Critical Care Medicine

(p < 0.001) and a 29% reduction in the number of orders per

admission among patients receiving any (p < 0.001). Secondary
analysis revealed that probability reductions were concurrent with
internal guidelines and order process modification while reductions in quantity occurred largely during the financial incentives
and feedback period. Estimated cost savings totaled $2.5M during the 2-year intervention. There was no significant difference in
ICU or hospital mortality between baseline and intervention.
Conclusions: A sequential intervention achieved significant reductions in ICU albumin use and cost savings without changes in
patient outcomes, supporting the combination of financial and
nonfinancial strategies to align providers with evidence-based
practices. (Crit Care Med 2016; XX:0000)
Key Words: evidence-based practice; fluid therapy; intensive
care unit; outcome and process assessment; physician incentive
plans; quality of health care

ritical care medicine accounts for a growing share of

hospital patient volume and cost. As much as 29% of
admitted patients receive critical care services, and the
costs of these services constitute 1738% of all hospital costs
(1, 2). Consequently, we expect critical care treatment patterns
to face increased scrutiny from payers and hospital administrators. Medical research continues to provide insight into better
quality and higher value treatment approaches (36). Despite
this research, evidence is not always translated into practice,
resulting in a high degree of practice variation (711). Novel
approaches to align provider behavior with evidence-based
standards are necessary.
Administration of IV fluids is one of the most common
interventions in ICU. Decisions related to IV fluid selection
(e.g., crystalloid vs albumin) represent an area of potential
improvement in evidence-based practice (1214). Clinical
research supports albumin use over crystalloids for specific
clinical indications (e.g., paracentesis) but not for most common critical care indications requiring IV fluid resuscitation
(4, 15, 16). For example, use in sepsis patients confers no mortality risk advantage over crystalloids (4). Substitution of crystalloids for albumin and other colloids is one of the American
www.ccmjournal.org

Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.

Lyu et al

Society of Anesthesiologists Choosing Wisely recommendations for reducing low value care (17). Despite the evidence
that crystalloids are safe and effective for many critical care
patients, albumin has seen common usage in the ICU (18).
Albumin is also 20100 times more costly than crystalloids,
and differences in administered crystalloid volume do not offset higher per-unit albumin costs (15, 18).
Seeking to reduce avoidable albumin utilization, Emory
Healthcares Critical Care Center (ECCC) instituted a sequentially implemented multifaceted intervention to reduce albumin use in ICU patients. The program included unit-level
audit and feedback, provider financial incentives tied to unitlevel albumin use, institution-wide guideline development,
and systematic changes in the albumin ordering process. We
first report the results of a prospective prepost study evaluating the effects of the 2-year intervention on albumin use, albumin costs, and patient outcomes to offer a holistic perspective.
We then report the results of a secondary analysis examining
the relationship between different intervention strategies and
aspects of albumin utilization. Our study aims to evaluate the
effectiveness of these interventions on albumin use and provide
insight on translating evidence-based standards into practice.

METHODS
Study Setting and Population
The study was conducted in eight ICUs within the ECCC from
September 2011 to August 2014. The ICUs included 135 beds
across two hospitals in two medical, two cardiothoracic, two
neuroscience, one surgical, and one coronary unit. ECCC
units utilize a high-intensity staffing model with intensivists,
advanced practice providers (APPs) (i.e., nurse practitioners,
physician assistants), and physicians in training (19).
The studys main analysis followed a prepost study design,
examining albumin utilization and patient outcomes during a
baseline period (September 2011 to August 2012) and intervention period (September 2012 to August 2014). A secondary analysis also compared utilization from baseline to the
first intervention year (September 2012 to August 2013) and
from this first year to the second intervention year (September
2013 to August 2014) to assess how albumin use changed in
response to the implementation of specific components of
the intervention (Supplemental Fig. 1, Supplemental Digital
Content 1, http://links.lww.com/CCM/B733).
We included all patients admitted to study units during the
study period and restricted our observations to each patients
first ICU stay, as patients readmitted to the ICU are typically
sicker and potentially less responsive to standard processes
of care (20). We hypothesized that the strategies used in the
2-year intervention would result in reduced albumin use without increased mortality rates.
Intervention
In 2012, ECCC leaders developed a sequentially implemented
multifaceted initiative to reduce albumin use. The intervention
consisted of four components occurring at different points
2

www.ccmjournal.org

during the intervention period: monthly utilization reports, an

explicit financial incentive, internally developed albumin use
guidelines, and a modification to the ordering process.
During the first intervention year, ICUs received monthly
feedback reports including albumin utilization for their unit
and the other ICUs. Additionally, attending physicians and
APPs were offered a unit-based financial incentive to reduce
albumin utilization by 25% from baseline or maintain an average albumin cost of $15 per patient-day or less. These targets
were decided in collaborative discussions with each ICUs
medical director and lead APP. Attending physicians were
offered $250 per covered week. APPs were offered $2,500 for
the year. We sized these rewards to capture providers attention, but not so large that providers might avoid appropriate
albumin use. Both the financial incentive and feedback reports
ended after the first intervention year. Albumin orders for liver
transplant patients were excluded from albumin use and cost
measure calculations.
Early in the second intervention year, institutional albumin use guidelines were established by an interdisciplinary
taskforce representing groups with the most frequent albumin
use. Members evaluated research evidence and worked with
their respective groups and the taskforce to identify specific,
evidence-based indications for albumin use (Supplemental
Table 1, Supplemental Digital Content 2, http://links.lww.
com/CCM/B734). These indications were then incorporated
into a new computerized ordering process. Previously, providers were not required to specify patients indications for ordering albumin. Under the revised process, albumin orders were
grouped by indication based on Emory Healthcare guidelines,
and ordering providers were required to select an indication
from the list.
Dependent Variables
The primary dependent variable measured the number of
albumin orders (e.g., 5% and 25% albumin preparations) per
admission during the first ICU stay. Secondary dependent
variables were ICU and hospital mortality rates.
Independent Variables
In the main analysis, we aimed to observe a holistic, 2-year
program effect to account for the order and timing of different intervention strategies. Thus, the primary independent
variable captured whether the patient was admitted during
the baseline or 2-year intervention period. In the secondary
analysis, the primary independent variable identified whether
the patient was admitted during the baseline year, first intervention year, or second intervention year. All adjusted analyses
controlled for patient demographic characteristics (i.e., age,
gender, race, and type of insurance) and health status measures
(21). Health status variables included body mass index (BMI)
at ICU admission, Charlson Comorbidity Index (CCI) score
(2224), Sequential Organ Failure Assessment (SOFA) score at
ICU admission (Supplemental Table 2, Supplemental Digital
Content 3, http://links.lww.com/CCM/B735) (25), and ICU
length of stay.
XXX 2016 Volume XX Number XXX

Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.

Clinical Investigation

Statistical Methods
Data were extracted from electronic medical records for all
patients admitted to study ICUs during the 3-year study period.
Exploratory analysis of all variables and their distributions
guided our choice of statistical tests and regression models.
We tested unadjusted differences in patient characteristics and
albumin use using two-tailed t tests, proportions tests, and
chi-square tests.
We estimated the adjusted intervention effect on albumin
use in both the main and secondary analyses using two-part
regression models (26, 27). Adjusted estimates from the main
analysis compared albumin use between baseline and intervention periods, adjusting for age, gender, race, type of insurance,
BMI at admission, CCI score, SOFA score at admission, and
length of stay. In the secondary analysis, we evaluated albumin
use using the same models and covariates but compared baseline to the first intervention year and first intervention year to
the second intervention year.
The first part of the two-part models used a logistic model.
The second part used a generalized linear model with a negative
binomial distribution and a log link (28). A negative binomial
distribution was selected over Poisson because the distribution
of albumin use demonstrated significant overdispersion. The
two-part model captures two key dimensions of the albumin
decision-making process: first, whether to order any albumin
(i.e., probability), and second, how much albumin to order for
a patient who receives any albumin (i.e., quantity). The twopart model is appropriate for modeling irregular and skewed
distributions and recognizes the multidimensional nature of
resource use decisions (29).
We examined unadjusted changes in direct albumin costs
and ICU and hospital mortality from baseline to intervention periods using two-tailed t tests. Using Emory Healthcares
internal cost accounting system, we estimated that the direct

per-unit albumin cost was $170 (18). We multiplied this perunit cost by the number of albumin orders to estimate direct
albumin fluid costs in the baseline and intervention periods.
We estimated adjusted differences in ICU and in-hospital mortality rates using logit regression models, controlling for the
same covariates as in the utilization models.
All analyses were performed in Stata 12 SE (StataCorp,
College Station, TX). Effects with a p value of less than 0.05
were considered significant. This study was approved by the
Emory Institutional Review Board.

RESULTS
A total of 22,004 unique admissions were included, of which
7,303 were in the baseline year and 14,701 were in the 2-year
intervention period. Patient characteristics were similar between
the two periods (Table 1). Patients in the intervention period
were slightly more likely to be uninsured (p < 0.001). Although
intervention period patients also had a statistically significant
higher mean SOFA score (p < 0.001) at ICU admission, the magnitude of this difference is relatively small and is not necessarily
clinically meaningful. Patient characteristics had no significant
differences between the two intervention years.
Albumin Utilization
Unadjusted mean albumin orders per admission decreased
from 2.7 during the baseline to 1.7 during the intervention, a
36.2% relative reduction (p < 0.001) (Table 2). After adjusting
for differences in patient characteristics between the baseline
and intervention periods, the intervention was associated with
a 41.5% relative decrease in the number of albumin orders per
ICU admission (p < 0.001). This adjusted effect included a relative 18.2% reduction in the probability of the patient receiving
any albumin order (p < 0.001) and a relative 28.5% reduction

Table 1. Characteristics of Patients Admitted to Study ICUs at Baseline and During the
Intervention Period
Characteristic

Age (yr), mean (sd)

Baseline Period
(n = 7,303)

Intervention Period
(n = 14,701)

59.2 (16.5)

59.3 (16.6)

0.614

Male (vs female), n (%)

3,815 (52.2)

7,665 (52.1)

0.889

White (vs non-white), n (%)

3,630 (49.7)

7,231 (49.3)

0.529
< 0.001

Health insurance, n (%)

Private

2,028 (27.8)

4,034 (27.4)

Public

4,800 (65.7)

9,562 (65.0)

390 (5.3)

981 (6.7)

85 (1.2)

124 (0.8)

28.9 (8.8)

28.7 (8.3)

0.205

Charlson Comorbidity Index score, mean (sd)

3.1 (2.5)

3.2 (2.5)

0.653

Sequential Organ Failure Assessment score, mean (sd)

5.2 (3.5)

5.5 (3.6)

< 0.001

First ICU length of stay (d), mean (sd)

4.1 (5.9)

4.2 (6.0)

0.083

Uninsured
Other
Body mass index at admission, mean (sd)

Critical Care Medicine

www.ccmjournal.org

Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.

Lyu et al

Unadjusted and Model-Adjusteda Baseline Period Means, Intervention Period

Means, and Marginal Change Between Periods in Albumin Utilization
Table 2.

Albumin Utilization Outcomes

a

Baseline
Period
Mean

Intervention
Period
Mean

Unadjusted no. of orders per admission

2.7

1.7

Proportion with any order

0.33

No. of orders among patients with 1 order

Marginal
Change (Relative
Change, %)

1.0 (36.2)

< 0.001

0.29

0.04 (11.8)

< 0.001

8.3

6.0

2.3 (27.9)

< 0.001

Model-adjusted no. of orders per admission (among all patients)

2.1

1.2

0.9 (41.5)

< 0.001

Part 1: probability of any order

0.31

0.26

0.06 (18.2)

< 0.001

Part 2: no. of orders among patients with 1 order

7.8

5.5

2.2 (28.5)

< 0.001

Model estimates adjust for age, sex, race, health insurance status and source, body mass index, Charlson Comorbidity Index scores, Sequential Organ Failure
Assessment scores, and first ICU length of stay.

in the number of orders per admission among patients receiving at least one order (p < 0.001).
Figure 1A depicts a steady decline in the overall utilization
of albumin orders per ICU admission. Figure 1B shows that the
reduction in the likelihood of any albumin orders per patient
decreased in intervention year 2. In contrast, Figure 1C demonstrates that the number of albumin orders among patients
receiving at least one order decreased in intervention year 1.
When analyzed by intervention year (Table 3), the mean
number of albumin orders per admission in intervention
year 1 decreased by 23.9% compared with the baseline period
(p < 0.001). Although the probability of receiving any albumin
did not decrease in intervention year 1 (p = 0.318), the number of
orders among patients who received at least one order decreased
by 25.9% relative to baseline (p < 0.001). Reductions in overall
utilization continued into intervention year 2, with 44.7% fewer
albumin orders per admission compared with intervention year 1
(p < 0.001). Although utilization among patients with any orders
continued to decrease (relative, 7.6%; p = 0.015), most of the
reduction in albumin use in the second intervention year was due
to a 40.1% decrease in the probability of receiving any albumin
relative to baseline and intervention year 1 (p < 0.001).
Cost and Mortality
We estimate that reductions in albumin orders translated
into mean savings of $171 per admission among all patients
(95% CI, 197 to 144; p < 0.001) (Table 4). Aggregate direct
cost savings were $0.8M in the first year and $1.7M in the second year, totaling nearly $2.5M over the 2-year intervention
period. Neither unadjusted ICU mortality rate (p = 0.135) nor
in-hospital mortality rate (p = 0.187) demonstrated statistically significant differences between baseline and intervention
periods. Similarly, adjusted analysis found no difference in
ICU mortality (p = 0.959) or in-hospital mortality (p = 0.958).

DISCUSSION
We report the results of a sequential intervention that decreased
overall albumin utilization by 36% and lowered direct albumin costs by $2.5M in a system of eight ICUs over 2 years. The
4

www.ccmjournal.org

proportion of patients given any albumin decreased by 18%,

and the quantity of orders among patients receiving any albumin decreased by 29%. These reductions were achieved without any significant adverse effect on ICU or hospital mortality
rates. The findings are consistent with existing literature and
further support the effectiveness of multifaceted interventions
for driving evidence-based and higher value care (3034).
By observing different dimensions of utilization in our
study, we found that the feedback and financial incentive intervention was associated with quantity reductions but did not
seem to affect decisions on which patients should receive albumin. There was no difference in the probability of prescribing
albumin to a given patient during this first intervention year,
while number of orders among patients with at least one order
significantly decreased. These pattern differences are observed
in (Fig.1, B and C). Neither the incentives nor feedback
reports signaled what types of patients should receive albumin.
Therefore, these interventions may have simply led providers to
pay greater attention to general albumin overutilization without changing how they identified which patients to administer
albumin. Furthermore, changes in the trajectory of orders at the
end of the baseline period indicate a possible Hawthorne effect
as providers became broadly aware of the upcoming incentives
(35). Still, further studies are needed to explore this behavior
change and better understand the type of impact that incentivebased strategies have on utilization decisions.
In intervention year 2, institutional guidelines and an order
process change focused on standardizing albumin use for a list
of specific indications. These strategies were designed to encourage providers to evaluate the appropriateness of albumin for each
patient. As expected, the proportion of patients prescribed any
albumin dropped significantly and to a much larger degree than
relative decreases in quantity ordered per admission. This finding is consistent with past literature that has studied computerized
decision support systems for promoting evidence-based practices
or guidelines (3639). For example, one study of RBC transfusion
guidelines found a significant drop in the proportion of admissions with inappropriate transfusions, but a smaller reduction in
mean number of RBC units transfused per admission (37).
XXX 2016 Volume XX Number XXX

Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.

Clinical Investigation

Figure 1. Albumin ordering patterns throughout study period along three

dimensions of utilization.

Our study contributes to the broader literature on financial

incentives in healthcare delivery. The larger pay-for-performance debate has mostly focused on the role of incentives in
promoting underused services in outpatient settings (4045).
Few studies have evaluated the impact of financial incentives
on reducing low value or overused services in the inpatient setting (46), or on the use of discrete medical resources such as
albumin along two separate dimensions (47, 48).
Our study also demonstrates that targeting low value treatments that are frequently prescribed can reap substantial
Critical Care Medicine

cost savings without adversely affecting mortality outcomes.

Although the typical unit cost of albumin is significantly less
than that of many diagnostic tests, medical devices, and surgeries, aggregate cost savings can be large given the high frequency
of orders and the relatively large cost difference between albumin and crystalloids (18). Given our findings and mounting
evidence that support more conservative use, albumin and
fluid therapy represent an area with promising opportunities
to improve value in care delivery (4, 15, 16).
This study has a number of potential limitations. First,
we did not have a concurrent control group, so our estimate
of the intervention effect may be biased by other factors that
affected trends in the use of albumin. For example, large, complex health systems are constantly evolving and implementing
quality improvement programs, and it is difficult to identify
appropriate comparison groups to account for external factors
that may impact study outcomes. Additionally, our available
data did not include the indication for albumin administration or need for fluid resuscitation, and thus our study is limited in its ability to examine the appropriateness of individual
albumin orders. Still, we observed few significant differences in
patient case mix between the baseline and intervention periods
and controlled for patient characteristics in adjusted analysis.
Although cluster randomized trials and other study designs
may further decrease risks of bias, such studies on practice
change interventions in ICUs are often challenging to conduct
given the severity of patients illnesses.
Second, it is impossible to determine whether changes in
albumin use in the second intervention year were due to components that began in the second year or delayed effects of
components implemented in the first year. It is also possible
that the impact of the initial intervention component faded in
the second year, in which case reductions in albumin use in
the second year were mainly due to the components implemented at that time. Furthermore, we cannot determine which
intervention strategy, if any, had the more dominant effect on
ordering behavior. Still, sequential implementation offers a
valuable perspective into differential intervention effects less
available with simultaneous implementation and raises important questions that must be explored in future studies.
Third, we did not observe substitute fluids, and data on
related medications were not readily available. Including these
costs could allow more comprehensive cost-effectiveness analysis. Still, our study provides meaningful estimates on direct
savings from reduced albumin use. Furthermore, the large unit
cost differences between albumin and crystalloids suggest that
an extraordinary increase in crystalloid use would be required
to offset cost savings from reduced albumin use (18).
Fourth, due to limited data availability, we were not able
to capture non-mortality patient outcomes relevant to fluid
administration. Although unobserved patient morbidity may
have changed over the study period, the lack of changes in
mortality rates is consistent with existing evidence supporting
more conservative albumin administration (4, 15, 16).
Fifth, the intervention was implemented in the critical care
center of an academic medical center, potentially limiting
www.ccmjournal.org

Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.

Lyu et al

Relative Change in Model-Adjusteda Albumin Utilization Outcomes From the

Baseline Period to the First Intervention Year and the First Intervention Year to the
Second Intervention Year
Table 3.

Intervention Year 1 vs
Baseline
Marginal Change
(Relative Change, %)

Overall: no. of orders among all patients

0.5 (23.9)

< 0.001

0.7 (44.7)

< 0.001

Part 1: probability of any orders

0.01 (2.7)

0.318

0.13 (40.1)

< 0.001

Part 2: no. of orders among patients with 1 order

2.0 (25.9)

Albumin Utilization Outcomes

a

Intervention Year 2 vs
Intervention Year 1
Marginal Change
(Relative Change, %)

< 0.001

0.4 (7.6)

0.015

Adjusted effect estimates adjusted for age, sex, race, health insurance status and source, body mass index, Charlson Comorbidity Index scores, Sequential
Organ Failure Assessment scores, and first ICU length of stay.

Unadjusted Differences in Direct Albumin Costs and Unadjusted and Adjusteda

Mortality Rates Before and During the Intervention Period

Table 4.

Cost and Mortality Outcomes

Baseline Period

Intervention Period

Marginal Change (95% CI)

297 (785)

171 (197 to 144)

< 0.001

396 (1,056 to 985)

< 0.001

Direct albumin fluid costs, mean $ per admission (sd)

Among all patients

467 (1,160)
1,416 (1,652)

1,021 (1,175)

ICU

4.7

5.1

0.5 (0.1 to 1.1)

0.135

Hospital

5.2

5.6

0.4 (0.2 to 1.1)

0.187

ICU

3.0

3.0

0.01 (0.39 to 0.41)

0.959

Hospital

3.4

3.4

0.01 (0.45 to 0.42)

0.958

Among patients with 1 order

Mortality rate, %
Unadjusted

Adjusted

Adjusted effect estimates adjusted for age, sex, race, health insurance status and source, body mass index, Charlson Comorbidity Index scores, Sequential
Organ Failure Assessment scores, and first ICU length of stay.

a

generalizability. The center organization and attributes of the

medical center may amplify the impact of individual practice
change interventions (49, 50).
Despite these limitations, our study was able to adjust for
observed patient characteristics by using a baseline comparison group and multivariate regression models. Furthermore,
we included five different types of ICUs, offering greater generalizability. As current research on overused medical services is limited (46), findings from quasi-experimental study
designs can guide future randomized controlled design studies. The savings observed in our study suggest that albumin is
a meaningful target for future research on higher value care.

CONCLUSIONS
A sequential intervention involving feedback reports, financial incentives, internal guidelines, and an order process
modification significantly decreased albumin utilization in
the ICU, saving $2.5M in direct costs over 2 years. Different
intervention components also impacted different dimensions
of provider behavior change, but future research is needed to
6

www.ccmjournal.org

confirm these patterns of impact. As the focus on high value

care grows, research evidence on best practices will continue
to evolve. Translating this evidence into practice requires that
we better understand how interventions impact provider
behaviors.

ACKNOWLEDGMENTS
We thank Albumin Utilization Task Force for developing albumin guidelines and participating ICUs for their dedication to
continually improving quality of care for patients.

REFERENCES

1. Halpern NA, Pastores SM: Critical care medicine in the United States
2000-2005: An analysis of bed numbers, occupancy rates, payer mix,
and costs. Crit Care Med 2010; 38:6571
2. Coopersmith CM, Wunsch H, Fink MP, et al: A comparison of critical
care research funding and the financial burden of critical illness in the
United States. Crit Care Med 2012; 40:10721079
3. Pearse R, Dawson D, Fawcett J, et al: Early goal-directed therapy
after major surgery reduces complications and duration of hospital
stay. A randomised, controlled trial [ISRCTN38797445]. Crit Care
2005; 9:R687R693
XXX 2016 Volume XX Number XXX

Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.

Clinical Investigation
4. Caironi P, Tognoni G, Gattinoni L: Albumin replacement in severe sepsis or septic shock. N Engl J Med 2014; 371:84
5. Olfson M, Marcus SC: Decline in placebo-controlled trial results suggests new directions for comparative effectiveness research. Health
Aff (Millwood) 2013; 32:11161125
6. Rouette J, Trottier H, Ducruet T, et al; Canadian Critical Care Trials
Group; PALISI Network: Red blood cell transfusion threshold in postsurgical pediatric intensive care patients: A randomized clinical trial.
Ann Surg 2010; 251:421427
7. Balas E, Boren, SA (Eds): Managing Clinical Knowledge for Health
Care Improvement. Yearbook of Medical Informatics 2000: Patientcentered Systems. Stuttgart, Germany, Schattauer, 2000, pp 6570
8. Berenholtz S, Pronovost PJ: Barriers to translating evidence into practice. Curr Opin Crit Care 2003; 9:321325
9. Hewson-Conroy KM, Burrell AR, Elliott D, et al: Compliance with processes of care in intensive care units in Australia and New ZealandA
point prevalence study. Anaesth Intensive Care 2011; 39:926935
10. Cook DJ, Meade MO, Hand LE, et al: Toward understanding evidence
uptake: Semirecumbency for pneumonia prevention. Crit Care Med
2002; 30:14721477
11. Grol R, Grimshaw J: From best evidence to best practice: Effective implementation of change in patients care. Lancet 2003; 362:12251230
12. Tanzi M, Gardner M, Megellas M, et al: Evaluation of the appropriate use of albumin in adult and pediatric patients. Am J Health Syst
Pharm 2003; 60:13301335
13. Yim JM, Vermeulen LC, Erstad BL, et al: Albumin and nonprotein colloid solution use in US academic health centers. Arch Intern Med
1995; 155:24502455
14. Seymour CW, Angus DC: Making a pragmatic choice for fluid resuscitation in critically ill patients. JAMA 2013; 310:18031804
15. Annane D, Siami S, Jaber S, et al; CRISTAL Investigators: Effects of
fluid resuscitation with colloids vs crystalloids on mortality in critically
ill patients presenting with hypovolemic shock: The CRISTAL randomized trial. JAMA 2013; 310:18091817
16. Finfer S, Bellomo R, Boyce N, et al; SAFE Study Investigators: A comparison of albumin and saline for fluid resuscitation in the intensive
care unit. N Engl J Med 2004; 350:22472256
17. ABIM Foundation and American Society of Anesthesiologists: Five
Things Physicians and Patients Should Question. 2013. Available at:
http://www.choosingwisely.org/wp-content/uploads/2015/02/ASAChoosing-Wisely-List.pdf. Accessed February 29, 2016
18. Lyu PF, Murphy DJ: Economics of fluid therapy in critically ill patients.
Curr Opin Crit Care 2014; 20:402407
19. Pronovost PJ, Angus DC, Dorman T, et al: Physician staffing patterns
and clinical outcomes in critically ill patients: A systematic review.
JAMA 2002; 288:21512162
20. Rosenberg AL, Watts C: Patients readmitted to ICUs: A systematic
review of risk factors and outcomes. Chest 2000; 118:492502
21. Spencer CS, Gaskin DJ, Roberts ET: The quality of care delivered to
patients within the same hospital varies by insurance type. Health Aff
(Millwood) 2013; 32:17311739
22. Charlson ME, Pompei P, Ales KL, et al: A new method of classifying
prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis 1987; 40:373383
23. Deyo RA, Cherkin DC, Ciol MA: Adapting a clinical comorbidity index
for use with ICD-9-CM administrative databases. J Clin Epidemiol
1992; 45:613619
24. Romano PS, Roos LL, Jollis JG: Adapting a clinical comorbidity index
for use with ICD-9-CM administrative data: Differing perspectives.
JClin Epidemiol 1993; 46:10751079; discussion 1081
25. Vincent JL, Moreno R, Takala J, et al: The SOFA (Sepsis-Related Organ
Failure Assessment) score to describe organ dysfunction/failure. On behalf
of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine. Intensive Care Med 1996; 22:707710
26. Friedberg MW, Schneider EC, Rosenthal MB, et al: Association between
participation in a multipayer medical home intervention and changes in
quality, utilization, and costs of care. JAMA 2014; 311:815825
27. Trasande L, Chatterjee S: The impact of obesity on health service
utilization and costs in childhood. Obesity (Silver Spring) 2009;
17:17491754

Critical Care Medicine

28. Wooldridge JM. Introductory Econometrics: A Modern Approach.

Fifth Edition. Mason, OH, South-Western, 2013
29. Bahn GD, Massenburg R: Deal with Excess Zeros in the Discrete
Dependent Variable, the Number of Homicide in Chicago Census
Tract. Joint Statistical Meeting, Denver, CO, August 37, 2008
30. Curtis JR, Treece PD, Nielsen EL, et al: Integrating palliative and
critical care: Evaluation of a quality-improvement intervention. Am J
Respir Crit Care Med 2008; 178:269275
31. Garrouste-Orgeas M, Soufir L, Tabah A, et al; Outcomerea Study
Group: A multifaceted program for improving quality of care in
intensive care units: IATROREF study. Crit Care Med 2012;
40:468476
32. Ilan R, Doan J, Cload B, et al: Removing nonessential central venous
catheters: Evaluation of a quality improvement intervention. Can J
Anaesth 2012; 59:11021110
33. Jain M, Miller L, Belt D, et al: Decline in ICU adverse events, nosocomial
infections and cost through a quality improvement initiative focusing on
teamwork and culture change. Qual Saf Health Care 2006; 15:235239
34. Scales DC, Dainty K, Hales B, et al: A multifaceted intervention for
quality improvement in a network of intensive care units: A cluster
randomized trial. JAMA 2011; 305:363372
35. De Amici D, Klersy C, Ramajoli F, et al: Impact of the Hawthorne effect
in a longitudinal clinical study: The case of anesthesia. Control Clin
Trials 2000; 21:103114
36. Hibbs SP, Noel S, Miles D, et al: The impact of electronic decision
support and electronic remote blood issue on transfusion practice.
Transfus Med 2014; 24:274279
37. Rana R, Afessa B, Keegan MT, et al; Transfusion in the ICU Interest
Group: Evidence-based red cell transfusion in the critically ill: Quality improvement using computerized physician order entry. Crit Care
Med 2006; 34:18921897
38. Rothschild JM, McGurk S, Honour M, et al: Assessment of education
and computerized decision support interventions for improving transfusion practice. Transfusion 2007; 47:228239
39. Sucher JF, Moore FA, Todd SR, et al: Computerized clinical decision
support: A technology to implement and validate evidence based
guidelines. J Trauma 2008; 64:520537
40. Eijkenaar F, Emmert M, Scheppach M, et al: Effects of pay for performance in health care: A systematic review of systematic reviews.
Health Policy 2013; 110:115130
41. Van Herck P, De Smedt D, Annemans L, et al: Systematic review:
Effects, design choices, and context of pay-for-performance in health
care. BMC Health Serv Res 2010; 10:247
42. Bardach NS, Wang JJ, De Leon SF, et al: Effect of pay-for-performance incentives on quality of care in small practices with electronic
health records: A randomized trial. JAMA 2013; 310:10511059
43. Chen JY, Tian H, Taira Juarez D, et al: The effect of a PPO pay-forperformance program on patients with diabetes. Am J Manag Care
2010; 16:e11e19
44. Chung S, Palaniappan LP, Trujillo LM, et al: Effect of physician-specific pay-for-performance incentives in a large group practice. Am J
Manag Care 2010; 16:e35e42
45. Petersen LA, Simpson K, Pietz K, et al: Effects of individual physicianlevel and practice-level financial incentives on hypertension care:
Arandomized trial. JAMA 2013; 310:10421050
46. Korenstein D, Falk R, Howell EA, et al: Overuse of health care services in the United States: An understudied problem. Arch Intern Med
2012; 172:171178
47. Armour BS, Pitts MM, Maclean R, et al: The effect of explicit financial incentives on physician behavior. Arch Intern Med 2001; 161:
12611266
48. Clarke PM, Glasziou P, Patel A, et al; ADVANCE Collaborative Group:
Event rates, hospital utilization, and costs associated with major complications of diabetes: A multicountry comparative analysis. PLoS
Med 2010; 7:e1000236
49. Ayanian JZ, Weissman JS: Teaching hospitals and quality of care:
Areview of the literature. Milbank Q 2002; 80:56993, v
50. Allison JJ, Kiefe CI, Weissman NW, et al: Relationship of hospital
teaching status with quality of care and mortality for Medicare patients
with acute MI. JAMA 2000; 284:12561262
www.ccmjournal.org

Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.