High-Value, Cost-Conscious Care: Iterative Systems-Based … · 2017-11-01 · necessary testing, we aimed to decrease the amount of daily laboratory tests (for brevity, “labs”)

CLINICAL RESEARCH STUDY

High-Value, Cost-Conscious Care: IterativeSystems-Based Interventions to ReduceUnnecessary Laboratory Testing
Brett W. Sadowski, MD, Alison B. Lane, MD, MS, Shannon M. Wood, MD, MPH, Sara L. Robinson, MD, MS,Chin Hee Kim, MDDepartment of Internal Medicine, Walter Reed National Military Medical Center, Bethesda, Md.
Funding: NonConflict of InAuthorship:

manuscript in accReporting, Editingestablished by theviews expressed ireflect the officialfense, nor the US

Requests for rDepartment of MeBuilding 19 Room

E-mail address

0002-9343/$ -seehttp://dx.doi.org/1

Do

ABSTRACT

BACKGROUND: Inappropriate testing contributes to soaring healthcare costs within the United States, andteaching hospitals are vulnerable to providing care largely for academic development. Via its “ChoosingWisely” campaign, the American Board of Internal Medicine recommends avoiding repetitive testing forstable inpatients. We designed systems-based interventions to reduce laboratory orders for patients admittedto the wards at an academic facility.METHODS: We identified the computer-based order entry system as an appropriate target for sustainableintervention. The admission order set had allowed multiple routine tests to be ordered repetitively each day.Our iterative study included interventions on the automated order set and cost displays at order entry. Theprimary outcome was number of routine tests controlled for inpatient days compared with the precedingyear. Secondary outcomes included cost savings, delays in care, and adverse events.RESULTS: Data were collected over a 2-month period following interventions in sequential years andcompared with the year prior. The first intervention led to 0.97 fewer laboratory tests per inpatient day(19.4%). The second intervention led to sustained reduction, although by less of a margin than order setmodifications alone (15.3%). When extrapolating the results utilizing fees from the Centers for Medicareand Medicaid Services, there was a cost savings of $290,000 over 2 years. Qualitative survey data did notsuggest an increase in care delays or near-miss events.CONCLUSIONS: This series of interventions targeting unnecessary testing demonstrated a sustainedreduction in the number of routine tests ordered, without adverse effects on clinical care.Published by Elsevier Inc. � The American Journal of Medicine (2017) -, ---

KEYWORDS: Clinical decision making; Diagnostic tests; High-value care; Quality improvement

In an era of healthcare expenditures constituting more than17% of the US gross domestic product, determining inno-vative ways to lower costs is critical.1 Hospital care made up

e.terest: None.All listed authors have contributed to preparing theordance with the “Recommendations for the Conduct,and Publication of Scholarly Work in Medical Journals”International Committee of Medical Journal Editors. Then this manuscript are those of the authors and do notpolicy of the Departments of the Navy, Army, or De-Government.eprints should be addressed to Brett W. Sadowski, MD,dicine, Walter Reed National Military Medical Center,3547, 8901 Rockville Pike, Bethesda, MD 20889.: [email protected]

front matter Published by Elsevier Inc.0.1016/j.amjmed.2017.02.029

wnloaded for Anonymous User (n/a) at Walter Reed National Military MFor personal use only. No other uses without permission. C

nearly one-third of all healthcare spending in the UnitedStates in 2015, and hundreds of billions of dollars are wastedon care that fails to improve patient outcomes.2 This patternis particularly concerning within teaching hospitals preparingthe next generation of physicians to responsibly lead theprofession.2-4 With further restrictions on reimbursementfrom Medicare through revisions on the Clinical LaboratoryFee Schedule mandated by the Protecting Access to Medi-care Act of 2014, it will become even more important to helptrainees develop responsible ordering practices. By adoptingthese attitudes early, trainees may not only improve theirvalue-based practice but also enhance their ability to mentorthose who follow on cost-conscious care.5

The American Board of Internal Medicine has identifiedevidence-based recommendations from dozens of specialty

edical Center from ClinicalKey.com by Elsevier on July 13, 2017.opyright ©2017. Elsevier Inc. All rights reserved.

mailto:[email protected]

http://dx.doi.org/10.1016/j.amjmed.2017.02.029

2 The American Journal of Medicine, Vol -, No -, - 2017

societies to reduce waste in healthcare. Repetitive inpatientlaboratory testing in the face of clinical stability is one ofthe practices targeted by the American Board of InternalMedicine in its “Choosing Wisely” campaign.6 Althoughrecognized for decades,7 misuse of laboratory testing in theinpatient setting persists, resulting in increased costs, iat-

CLINICAL SIGNIFICANCE

� Cost-conscious decision making isbecoming more critical to include in thetraining of young physicians duringresidency training.

� Optimizing electronic order entry sys-tems can lead to re-evaluation of po-tential wasted care and instillresponsible ordering practices amongproviders.

� Sustainable impact on quality and per-formance is achievable once careful ex-amination of contributing factors tohealthcare waste are identified andaddressed at the institutional level.

rogenic anemia,8 and negativeimpacts on patient experience.Previous interventions targetingthis issue have included providereducation,9 price displays,10

alerts regarding redundanttesting,11-13 prompts on progressnotes,14 and unbundling of testpanels with multiple compo-nents.13 Unfortunately, educationof personnel alone often lackssustainability once trainingstops.15

We report a series of high-yield interventions implementedwithin an existing electronichealth record that were specif-ically designed to decreasewasted care, while also instillingcost-conscious attitudes amongphysicians at our training insti-
tution. With the overall goal of reducing the use of un-necessary testing, we aimed to decrease the amount ofdaily laboratory tests (for brevity, “labs”) by 10% overrepeated 2-month Plan-Do-Study-Act cycles withinteaching medicine services at a tertiary-care hospital. Byreporting on multiple cycles, we aimed to demonstratethat systems-based interventions are effective in imple-menting sustainable change and preparing house staff formore responsible practice when they move forward intheir medical careers.
Figure 1 Preintervention order set with availabilityon admission to the Internal Medicine Service, regar

Downloaded for Anonymous User (n/a) at Walter Reed National Military MFor personal use only. No other uses without permission. C

METHODS

Setting and DesignBoth iterations of the study took place at an academic,federally funded, tertiary-care facility in Bethesda, Mary-land. Patients were admitted to the teaching medicine ser-

of multiple laboratory testsdless of diagnosis.

edical Center from ClinicalKey.comopyright ©2017. Elsevier Inc. All r

vices on teams with 1 resident, 2interns, 2 to 3 students, and anattending physician. All orderswere placed via a computer-basedorder entry system, and all alter-ations required the approval of thechief of the affected services.

Intervention 1After identifying stakeholders andfactors contributing to high ratesof daily testing of the completeblood count, basic metabolicpanel, liver-associated enzymes,coagulation panel, and mineralpanels (magnesium and phos-phorus), we identified admissionorder sets as the highest-yieldtarget for potential intervention.As opposed to other contributingfactors, such as rotating trainees,

patients, laboratory technicians, and nurses, the order set is aconsistent factor on which an intervention could lead to asustainable difference. Before the first intervention, the or-der set used to admit patients listed the frequency of labo-ratory testing as “QAMLAB,” meaning that these studieswould be drawn each morning until discontinued(Figure 1). This allowed busy trainees to quickly submitadmission orders that mandate 6 studies to be drawn dailywithout requiring reconsideration, regardless of diagnosis.

to be ordered daily

by Elsevier on July 13, 2017.ights reserved.

Figure 2 Example of redundant laboratory test results after admission, resulting in limited relevant clinical data.

Sadowski et al Reducing Routine Inpatient Lab Testing 3

Additionally, with the automated comment of “and staton admission,” laboratory testing performed in theemergency room was often repeated once patientswere admitted, resulting in redundant blood work, typi-cally with negligible differences in clinical data(Figure 2).

We eliminated a step that led house staff to order labsindefinitely by altering order sets to allow labs to be drawnonly once at admission, if they had not already been drawnin the emergency department (Figure 3). House staffretained the ability to order labs daily as discrete ordersseparate from the admission order set, should the clinicalscenario require it. The number of labs ordered wastabulated over a 2-month period (January-February 2015)and controlled for inpatient days so that a comparison couldbe made with the same 2-month period the year prior,minimizing the likelihood that seasonal variation in admis-sion diagnoses would affect data and ensuring that traineeswould have a similar level of experience in both pre- andpostintervention time periods. Implementation occurred 1month before data collection to allow for house stafffamiliarization. To our knowledge no other simultaneouslarge-scale interventions on ordering practices or the orderset were made.

Figure 3 Postintervention order set without a


The number of tests obtained in the study period wascollected through the electronic medical record. Patientswere admitted from the emergency room, clinic, or outsidehospitals and were more than 18 years of age. No admittingdiagnoses were excluded, to include patients who tradi-tionally have labs drawn serially to trend relevant markers(eg, gastrointestinal hemorrhage). Only data for patientsassigned to medicine, cardiology, and oncology serviceswere collected, because these were primarily affected by theorder set changes. No patients admitted to the intensive careunit were included, though patients subsequently transferredto the ward were.

Intervention 2After the initial modification to the order set substantiallyreduced the labs ordered as detailed in the Results section,we expanded our goal by displaying cost associated withtests. Residents at our institution reported limited knowl-edge regarding costs of labs, a finding that is not uniqueamong trainees.16 To counteract this, the second Plan-Do-Study-Act cycle included the addition of price displays forthe same labs as the first iteration. When a lab wasselected, the price was displayed on the same line as the

vailability to order daily laboratory tests.


Table 1 Summary of Electronic Ordering System Interventions Made in Both Iterations of Project

Intervention 1 Characteristic Intervention 2

January-February 2015 Intervention period January-February 2016January-February 2014 Comparison period January-February 2014Removal of option to order daily routine laboratory tests from automated admissionorder set

Intervention Price display of laboratory tests

Inpatient Wards for Internal Medicine, Cardiology, and Oncology Services affected All adult inpatient services


test name. With this intervention, the ordering providerwould visualize laboratory test prices while entering or-ders. The displayed price was chosen from publicallyavailable Centers for Medicare & Medicaid Servicesreimbursement data. Specific costs for components oflaboratory test panels (eg, platelets within a completeblood count) were not individually listed. Data werecollected over a 2-month period in the winter of 2016 andcompared with the 2 prior years. The number of labs wascontrolled by the number of inpatient days. A summary ofthe interventions made during this project is shown inTable 1.

Data AnalysisDaily counts of 6 tests were collected over a 2-monthperiod for 3 consecutive years, beginning in 2014. Thedata collected in the first year served as the control to thefirst intervention monitored in 2015, and then for bothinterventions in 2016. Incidence rates were created foreach laboratory test, whereby inpatient days was definedas the sum of daily ward census totals during the studyperiods within identified services. The sample was largeenough to compare control and intervention rates acrosslaboratory tests using the Normal-Theory test. Confidenceintervals were constructed for incidence rate ratios foreach test. All analyses were done in the statistical soft-ware R (R Foundation for Statistical Computing, Vienna,Austria), with figures constructed in Excel 2010 (Micro-soft, Redmond, WA). Alpha was set at .05 for allanalyses.

Table 2 Reduction of Total Number of Laboratory Tests per Inpatien

Parameter2014(Control)

2015(Intervention 1)

2016(Intervention 2)

PerDifAftInt

Inpatient bed days 2786 2905 3288 þTotal inpatientlaboratory tests

13,892 11,677 13,890 �1

Average laboratorytests per inpatient day

4.99 4.02 4.22 �1

In each of the years in which interventions were made, the total number ofthere was an increase in the total of inpatient days.

CI ¼ confidence interval.


RESULTS

Intervention 1The number of labs drawn in pre- and postintervention pe-riods was tabulated and controlled for with the total numberof inpatient days (2785 in 2014, 2850 in 2015). The totalnumber of preintervention labs drawn was 13,892 andsubsequently dropped to 11,677, indicating that despite in-creases in the census, the gross number of labs orderedduring the study period decreased. When correcting forinpatient days, the total number of labs per day droppedfrom 4.99 to 4.02, resulting in an incidence rate ratio (IRR)of 0.81 (95% confidence interval [CI], 0.79-0.83; P <.001)(Table 2). Significant reductions in IRR were observed forall studied tests, including coagulation panels (IRR 0.73;95% CI, 0.66-0.8; P <.001), phosphorus (IRR 0.76; 95%CI, 0.71-0.80; P <.001), magnesium (IRR 0.76; 95% CI,0.21-0.80; P <.001), complete blood counts (IRR 0.82;95% CI, 0.78-0.86; P <.001), liver-associated enzymes(IRR 0.83; 95% CI, 0.75-0.92; P <.001), and basic meta-bolic panels (IRR 0.89; 95% CI, 0.84-0.93; P <.001)(Table 3). The forest plot shown in Figure 4 demonstratesthe incidence rate ratios with confidence intervals.

Intervention 2The number of labs drawn in the postintervention periodwas tabulated and controlled for the total number of inpa-tient days as compared with 2 years prior (2785 in 2014,3288 in 2016). During this time, intervention 1 remained inplace. The total number of labs drawn in the study period

t Day

centageferenceerervention 1

Incidence RateRatio AfterIntervention1 (95% CI, P Value)

PercentageDecreaseAfterIntervention 2

Incidence RateRatio AfterIntervention 2(95% CI, P Value)

4.3 þ18.06.0 �0.01

9.4 0.81 (0.79-0.83,<.001)

�15.3 0.85 (0.83-0.87,<.001)

laboratory tests was less than the preintervention time period, although


Table 3 Change in the Number of Individual Laboratory Tests per Inpatient Day

Laboratory Test2014 (Control;2786 Bed Days)

2015 (Intervention 1;2905 Bed Days)

2016 (Intervention 2;3288 Bed Days)

Incidence Rate RatioAfter Intervention 1(95% CI, P Value)

Incidence Rate RatioAfter Intervention 2(95% CI, P Value)

BMP 3375 3120 3626 0.89 (0.84-0.93, <.001) 0.91 (0.87-0.95, <.001)LAE 842 729 805 0.83 (0.75-0.92, <.001) 0.81 (0.74-0.89, <.001)CBC 3286 2816 4051 0.82 (0.78-0.86, <.001) 1.04 (1.00-1.09, .067)Coagulation panel 1009 765 756 0.73 (0.66-0.80, <.001) 0.63 (0.58-0.70, <.001)Magnesium 2722 2152 2367 0.76 (0.72-0.80, <.001) 0.74 (0.70-0.78, <.001)Phosphorus 2685 2095 2285 0.76 (0.71-0.80, <.001) 0.73 (0.69-0.77, <.001)

BMP ¼ basic metabolic panel; CBC ¼ complete blood count with differential; CI ¼ confidence interval; LAE ¼ liver-associated enzymes.


was 13,890, indicating that although inpatient daysincreased by more than 500, the gross number of laboratorytest orders remained constant. The number of labs perinpatient day was 4.22 during this period, a decrease of15.3% when compared with the preintervention data,translating to an IRR of 0.85 (95% CI, 0.83-0.87, P <.001).There were further decreases in IRR that exceeded theimpact of intervention 1 alone for liver-associated enzymes(0.81; 95% CI, 0.74-0.87, P <.001), coagulation panels(0.63; 95% CI, 0.58-0.70, P <.001), magnesium (0.74; 95%CI, 0.70-0.78; P <.001), and phosphorus (0.73; 95% CI,0.69-0.77; P <.001). The impact on basic metabolic panels,although still reduced from the preintervention levels, wasless than it was after the first intervention (IRR 0.91; 95%CI, 0.87-0.95; P <.001). There was an increase in the IRRfor complete blood counts after intervention 2, but this wasnot statistically significant (IRR 1.04; 95% CI, 1.00-1.09;P ¼ .067). The forest plot shown in Figure 5 demonstratesthe IRRs with CIs. Table 4 displays estimations of costsavings both in total and individual laboratory tests.

Balancing MeasuresAt the conclusion of intervention 1, we surveyed the housestaff to identify whether there were any balancing measures

Figure 4 Incidence rate ratios (IRR) of laboratory orderingby test studied when comparing intervention 1 with the controlperiod from 1 year prior. All IRRs demonstrated statisticallysignificant declines in the intervention period. BMP ¼ basicmetabolic panel; CBC ¼ complete blood count with differen-tial; CI ¼ confidence interval; Coag ¼ coagulation panel;LAE ¼ liver-associated enzymes/hepatic function panel.


in terms of impact on patient care, via SurveyMonkey.Residents and interns rotating on affected wards during thestudy period responded at a rate of 72% (28 of 39), with adistribution between residency classes of 10 interns, 12 juniorresidents, and 6 senior residents. In all, 82.1% of respondentsindicated that they experienced no delays in patient care. Themost commonly cited adverse event was repeated phlebot-omy due to a missed laboratory test order (5). No near-missor sentinel events were recorded via this survey.

DISCUSSIONThis study demonstrates that high-yield interventions aimedat optimizing electronic order entry settings led to reductionsin daily laboratory tests ordered per inpatient day. Changingthe way physicians practice within their personal routines is achallenge often faced during quality improvement initiatives,including with regard to laboratory testing.17 By first care-fully examining the reasons that tests were ordered with suchfrequency at our institution, we identified ways we could

Figure 5 Incidence rate ratios (IRRs) of laboratory orderingby test studied when comparing serial implementation of bothinterventions with the control period. The IRRs for coagulationpanels, magnesium, phosphorus, and liver-associated enzymesall continued to decline. Basic metabolic panel orderingincreased between year 1 and year 2 but still had sustainedstatistically significant reduction as compared with the control.The CBC ordering demonstrated a statistically insignificantincrease as compared with the control period. BMP ¼ basicmetabolic panel; CBC ¼ complete blood count with differen-tial; CI ¼ confidence interval; Coag ¼ coagulation panel;LAE ¼ liver-associated enzymes/hepatic function panel.


Table4

Estimated

Cost

Saving

sover

Test

Perio

dandYear

AfterEach

Intervention

Was

Implem

ented,

asEstimated

byCM

SClinical

Diagno

stic

Labo

ratory

FeeSchedu

leListing

Labo

ratory

Test

2017

CMS

Clinical

Diagno

stic

Labo

ratory

FeeSchedu

leListing($)

Redu

ctionin

Labs

Controlled

forInpatientDa

ys(Intervention1)

Estimated

Redu

ced

Cost

perInpatient

DayAfter

Intervention

1($)

Estimated

Redu

ced

Cost

forTest

Perio

d(Intervention1)

($)

Estimated

Yearly

Cost

Saving

sAfter

Intervention

1($)

Redu

ctionin

Labs

Controlledfor

InpatientDa

ys(Intervention2)

Estimated

Redu

ced

Cost

perInpatient

DayAfter

Intervention

2($)

Estimated

Redu

ced

Cost

forTest

Perio

d(Intervention2)

($)

Estimated

Yearly

Cost

Saving

sAfter

Intervention

2($)

BMP

11.60

�0.14

�1.60

�464

4.89

�27,86

9.36

�0.11

�1.26

�415

9.30

�24,95

5.79

LAE

11.49

�0.05

�0.74

�216

3.07

�12,97

8.41

�0.06

�0.83

�273

9.68

�16,43

8.09

CBC

10.66

�0.21

�2.24

�651

9.52

�39,11

7.11

0.05

0.56

1828

.34

10,970

.05

Coagulation

panel

8.24

�0.10

�0.82

�236

8.80

�14,21

2.80

�0.13

�1.09

�358

6.34

�21,51

8.06

Magnesium

9.19

�0.24

�2.17

�631

6.15

�37,89

6.92

�0.26

�2.37

�778

0.45

�46,68

2.71

Phosph

orus

6.50

�0.23

�1.52

�440

3.93

�26,42

3.59

�0.26

�1.69

�554

4.91

�33,26

9.44

Total

�9.09

�26,41

6.37

�158

,498

.19

�6.69

�21,98

2.34

�131

,894

.04

BMP¼

basicmetabolic

panel;CB

C¼

completebloodcoun

twithdifferential;CI

¼confi

denceinterval;CM

S¼

CentersforMedicare&MedicaidServices;Labs

¼labo

ratory

tests;

LAE¼

liver-associated

enzymes.



improve our order entry platform to influence orderingbehavior. Sedrak et al16 recently reported that the vast ma-jority of residents recognize that they order unnecessary labs,partly attributing this to practice habits, lack of cost trans-parency, and ease of ordering repeating labs in the electronichealth record, among others. Our findings support these as-sertions and demonstrate that by addressing each factor inturn, sustained change is achievable.

By first reducing the ease of ordering automated daily labsthrough the admission order sets, we encouraged and enabledproviders to make clinical decisions regarding each patient’scase and not rely on automated orders for routine tests. Nextwe made prices of routine tests more transparent, therebyallowing easily accessible, previously under-publicized data tobe utilized by providers when making decisions aboutnecessary testing. The first intervention yielded a reduction inlaboratory testing of more than 19%, which was sustainedover the course of the subsequent data collection period, albeitwith more variation among the individual tests. The lack ofimpact of the second intervention breaks from a priorcontrolled study that had shown substantial reduction in theordering rates with price displays alone.10 The second iterationitself had modest yield, but the impact of this intervention waslikely blunted by the effect magnitude of the preceding cycleand a lack of formal mandated acknowledgment of the priceby the ordering provider. Obtaining data regarding the mostaccurate price of a laboratory test is difficult because pub-lished values vary widely.18 Additionally, because trainees atour institution know that beneficiaries are not billed directly,this knowledge may be more easily ignored. Although wechose to display the Centers for Medicare & Medicaid Ser-vices rate, if pricing available from other sources such as themuch higher “Fair Price” in the Healthcare Blue Book wasshown, perhaps house staff would have been more judiciousin ordering, even with the knowledge that their patient wouldnot bear the full financial burden.19

Educational initiatives to promote responsible orderinghave been used within our institution as well as in otherlarge medical centers and have met variable success.9,13,14

The most powerful component of our sequential systems-based interventions is the sustainability that has beendemonstrated despite the turnover of hundreds of housestaff. By making interventions on a system level, lessvariation due to personnel shifts is experienced.

The reduction in orders resulting from intervention 1 wasnot uniform across tests, and there was a more robust impacton coagulation and mineral panel rates as compared withbasic metabolic panels, liver-associated enzymes, and com-plete blood counts. Although all of these tests had identicalorder set frequency options in the pre- and postinterventionperiods, we hypothesize that clinicians still made decisionsto order some routine labs more often than others, dependingon the patient-centered scenario. Because discrete order entrycontinued to allow for some degree of automation, clinicianswere permitted to select the tests they wanted to be seriallyrepeated (eg, basic metabolic panel during active diuresis).Additionally, some tests may have been ordered at higher



unnecessary rates during the control period, leading to anaugmentation of the effect size. One limitation is that no datawere collected on admitting diagnoses during each of thestudy periods, which could confound some of the results,although to likely a limited extent.

Two laboratory tests (basic metabolic panel and completeblood count) were ordered at higher rates per inpatient dayafter the second intervention (cost display) than after the first(removal of QAMLAB from admission order set), and thelatter was ordered more in the second year than before anyintervention. There was a robust response to the initialintervention, which caused drops in all labs by between7.6% and 25.1%, despite an increase in patient census (ie,before even controlling for inpatient days). Once the patientcensus increased by a substantial margin in the seconditeration (þ18.1% from preintervention census), these 2tests were ordered at higher rates, which could be explainedby either overcorrection of unnecessary labs during the firstintervention, or if specific patient factors during the morethan 500 additional bed days necessitated clinically indi-cated increases in these ordering rates. One of the strengthsof this study is that we did not exclude any diagnosis whenexamining the data, even though some diagnoses requiringserial laboratory monitoring would lead to increases in thefrequency of laboratory evaluation.

The implications of these data support that when devisingstrategies to change practice patterns within an institution,careful examination of the root causes of wasted care isrequired to make high-yield interventions. In our study weidentified the automation of order sets as a driving factorbehind unnecessary testing, and addressing this directlycreated a substantial improvement in laboratory utilization.Although the addition of price displays during the seconditeration helped consolidate efficiency gains in some tests,the result was not as prominent, at least in part because ofthe particular situation and setting in which it was imple-mented (ie, federally funded institution) and potentiallyowing to the magnitude of decrease resulting from the initialintervention. With this knowledge, we are moving forwardwith plans to adjust order sets throughout the hospital forother services outside of our department, as well as limitingthe ability for discrete orders to be repeated indefinitely.Further analysis in regard to iatrogenic anemia rates andtransfusion requirements can also be assessed moving for-ward. Additionally, a larger-scale effort to examine the ef-fects of encouraging responsible resource utilization in thetrainees of today on their practice patterns later in theircareer would guide efforts to instill cost-conscientiousnessin the undergraduate and graduate medical education arenas.

CONCLUSIONThrough a series of iterative interventions on a pre-existingorder entry platform, we demonstrated sustainable re-ductions in the ordering of routine labs within the medicinewards at an academic center while not adversely affectingpatient care outcomes.


ACKNOWLEDGMENTSWe thank Sorana Raiciulescu, MSc, and William Shimeall,MD, MPH, FACP, for contributions made in the biostatis-tical analysis and preparation of the manuscript.

References1. The World Bank. Health expenditure, total (% of GDP). Available at:

http://data.worldbank.org/indicator/SH.XPD.TOTL.ZS?locations¼US.Accessed December 26, 2016.

2. Weinberger SE. Providing high-value, cost-conscious care: a criticalseventh general competency for physicians. Ann Intern Med.2011;155(6):386-388.

3. Centers for Medicare & Medicaid Services. National health expendi-ture data: historical. Available at: www.cms.gov/research-statistics-data-and-systems/statistics-trends-and-reports/nationalhealthexpenddata/nationalhealthaccountshistorical.html. Accessed December 26, 2016.

4. Institute of Medicine. The Healthcare Imperative: Lowering Costs andImproving Outcomes. Washington, DC: National Academies; 2010.

5. Centers for Medicare & Medicaid Services. Medicare clinical diag-nostic laboratory tests payment system final rule. Available at: www.cms.gov/newsroom/mediareleasedatabase/fact-sheets/2016-fact-sheets-items/2016-06-17.html. Accessed December 26, 2016.

6. Bulger J, Nickel W, Messler J, et al. Choosing wisely in adult hospitalmedicine: five opportunities for improved healthcare value. J HospMed. 2013;8(9):486-492.

7. Griner P, Litzin B. Use of the laboratory in a teaching hospital: im-plications for patient care, education and hospital costs. Ann InternMed. 1971;75(2):157-163.

8. Thavendiranathan P, Bagai A, Ebidia A, Detsky A, Choudhry N. Doblood tests cause anemia in hospitalized patients? The effect of diag-nostic phlebotomy on hemoglobin and hematocrit levels. J Gen InternMed. 2005;20:520-524.

9. Thakkar RN, Kim D, Knight AM, Riedel S, Vaidya D, Wright SM.Impact of an educational intervention on the frequency of daily bloodtest orders for hospitalized patients. Am J Clin Pathol. 2015;143(3):393-397.

10. Feldman LS, Shihab HM, Thiemann D, et al. Impact of providing feedata on laboratory test ordering: a controlled clinical trial. JAMA InternMed. 2013;173(10):903-908.

11. Bates DW, Kuperman GJ, Rittenberg E, et al. A randomized trial of acomputer-based intervention to reduce utilization of redundant labo-ratory tests. Am J Med. 1999;106(2):144-150.

12. Procop GW, Yerian LM, Wyllie R, Harrison AM, Kottke-Marchant K.Duplicate laboratory test reduction using a clinical decision supporttool. Am J Clin Pathol. 2014;141(5):718-723.

13. Neilson EG, Johnson KB, Rosenbloom ST, et al. The impact of peermanagement on test-ordering behavior. Ann Intern Med. 2004;141(3):196-204.

14. Wertheim BM, Aguirre AJ, Bhattacharyya RP, et al. An educationaland administrative intervention to promote rational laboratory testordering on an academic general medicine service. Am J Med.2017;130(1):47-53.

15. May T, Clancy M, Critchfield J, et al. Reducing unnecessary inpatientlaboratory testing in a teaching hospital. Am J Clin Pathol. 2006;126:200-206.

16. Sedrak M, Pateil M, Ziemba J, et al. Residents’ self-report on why theyorder perceived unnecessary inpatient laboratory tests. J Hosp Med.2016;11(12):869-872.

17. Solomon D, Hashimoto H, Daltroy L, Liang M. Techniques to improvephysicians’ use of diagnostic tests: a new conceptual framework.JAMA. 1998;280:2020-2027.

18. Riggs K, DeCamp M. Providing price displays for physicians: whichprice is right? JAMA. 2014;312(16):1631-1632.

19. Healthcare Bluebook. Find your fair price. Available at: https://www.healthcarebluebook.com/page_ConsumerFront.aspx. AccessedJanuary 4, 2017.


http://data.worldbank.org/indicator/SH.XPD.TOTL.ZS?locations=US

http://data.worldbank.org/indicator/SH.XPD.TOTL.ZS?locations=US

http://refhub.elsevier.com/S0002-9343(17)30254-1/sref2



http://www.cms.gov/research-statistics-data-and-systems/statistics-trends-and-reports/nationalhealthexpenddata/nationalhealthaccountshistorical.html





http://www.cms.gov/newsroom/mediareleasedatabase/fact-sheets/2016-fact-sheets-items/2016-06-17.html












































https://www.healthcarebluebook.com/page_ConsumerFront.aspx

https://www.healthcarebluebook.com/page_ConsumerFront.aspx

Documents

High-Value, Cost-Conscious Care: Iterative Systems-Based … · 2017-11-01 · necessary testing, we aimed to decrease the amount of daily laboratory tests (for brevity, “labs”)