O R I G I N A L A R T I C L E

Using lean techniques to define the platelet (PLT)transfusion process and cost-effectiveness to evaluate

PLT dose transfusion strategies_3539 1..11

William Riley, Benjamin Smalley, Shelley Pulkrabek, Mary E. Clay, and Jeffrey McCullough

BACKGROUND: Platelet (PLT) doses of 1.1 ¥ 1011,2.2 ¥ 1011, and 4.4 ¥ 1011/m2 body surface area areequally effective in preventing bleeding. These differentdose strategies involve different numbers of transfu-sions. We conducted a cost analysis of three separatePLT dose therapies.STUDY DESIGN AND METHODS: A process map ofpreparation and administration of a PLT transfusionidentified 46 steps (23 steps in the blood bank and 23steps on the care unit). Time studies were conductedfor these 46 steps. Supply costs and personnel costswere estimated based on time studies. We conducteda cost analysis of three separate treatment regimesinvolving 16 transfusions per patient for the low-dose,12 transfusions for medium-dose, and eight transfu-sions for high-dose regimes.RESULTS: The time and the cost of the transfusionprocess for the blood bank were 32.41 minutes and$21.93 per unit, and for the patient care unit, 58.36minutes and $57.71 per unit. The total cost for a courseof PLT therapy per patient ranged from $4503.77 to$7014.59 for three different PLT doses. For a simulatedbone marrow transplantation unit with 259 patientsannually, there would be approximately a $700,000difference among the clinically equivalent low-,medium-, and high-dose treatment options.CONCLUSIONS: The overall cost of transfusiontherapy is more influenced by the cost of the productthan the cost of providing the transfusion. Dependingon the cost adjustment by the supplier for differentdoses of PLTs, a low-dose transfusion strategy can beless costly.

The aim of comparative effectiveness research(CER) is to generate evidence that comparesthe benefits and harms of alternative clinicalmethods to improve the delivery of care.1-4 CER

consists of two separate and distinct components: clinicaleffectiveness and financial cost. While clinical effective-ness research methods are well established, the financialmodels to ascertain cost analysis for clinical processesneed further development. With the recent finding thatthe effectiveness of low-dose and high-dose platelet (PLT)transfusions is clinically equivalent,5 the cost of the indi-vidual transfusion becomes a factor in the overall cost ofPLT transfusion therapy.

Hersh and coworkers6 used a mathematical model topredict that giving smaller doses of PLTs more frequentlywould result in lower total PLT utilization. Since this wasa mathematical model, the effect on prevention or treat-ment of bleeding with smaller doses of PLTs was notknown. Klumpp and colleagues7 showed that using a largerthan usual dose resulted in fewer PLT transfusions butmore total PLTs used. In an uncontrolled observationalstudy, we8 also showed that giving large doses of PLTs sub-stantially extended the time to the next transfusion. None

ABBREVIATIONS: CER = comparative effectiveness research;

FTE = full-time equivalent; PEP = process efficiency percentage;

VSM = value-stream map.

From the Department of Health Policy and Management,

School of Public Health, and the Department of Laboratory

Medicine and Pathology, Medical School, University of Minne-

sota, Minneapolis, Minnesota.

Address reprint requests to: Jeffrey McCullough, MD, Uni-

versity of Minnesota, 420 Delaware Street SE, Minneapolis, MN

55455; e-mail: mccul001@umn.edu.

Supported by the University of Minnesota.

Received for publication February 21, 2011; revision

received November 30, 2011, and accepted December 4, 2011.

doi: 10.1111/j.1537-2995.2011.03539.x

TRANSFUSION **;**:**-**.

Volume **, ** ** TRANSFUSION 1

of these observations were very surprising in that if a largerdose of PLTs is transfused, the posttransfusion PLT countcan be expected to be much higher than from usual dosesand therefore extend the time to the next transfusion. Con-versely, smaller doses of PLTs should result in a lower incre-ment and a shorter time to the next transfusion. However,none of these studies evaluated the effect of different PLTdoses on prevention and control of bleeding. That has nowbeen done in two recent studies. In a small study, Heddleand coworkers9 found a 5% incidence of severe bleeding inpatients who receive PLT transfusions of 1.5 ¥ 1011 to3.0 ¥ 1011 compared to none in those who received thestandard dose of PLTs. Although there was no significantdifference between the two groups for Grade 2 or greaterbleeding, the trial was halted prematurely because of pre-determined stopping rules of excess Grade 4 bleeding inthe low-dose arm. In a larger study of 1251 patients, Slich-ter and colleagues5 found no difference in prevention orcontrol of bleeding in patients who received 1.1 ¥ 1011,2.2 ¥ 1011, or 4.4 ¥ 1011/m2 body surface area of PLTs. Thus,this definitive study seems to establish that doses ofPLTs ranging from half that currently used to twice thatcurrently used are equally effective in the managementof bleeding. However, patients receiving the half-dose ofPLTs receive transfusions more frequently (every day)compared with those who received twice the current usualdose who received transfusions only every 3 days. Sinceit appears that these different doses of PLTs are equallyeffective in the management of bleeding, this opens thepossibility that different dosing strategies could be used,but this in turn will affect the frequency and total numberof transfusions that a patient receives during a periodof thrombocytopenia. Presumably the cost of the PLTproduct will vary from the cost of the current standarddose, but also the overall transfusion costs will varydepending on the number of transfusions given. There-fore, we undertook an analysis to determine the cost ofgiving a PLT transfusion and used this to project the cost-effectiveness of different PLT dose strategies.

MATERIALS AND METHODS

This is a multimethod study using a lean process analysismethod to document key process steps, a financial assess-ment to determine standard costs, and a simulationmodel to assess cost-effectiveness. The study was con-ducted at the University of Minnesota Medical Center,Fairview, a tertiary teaching hospital associated with theUniversity of Minnesota.

LeanLean is a process engineering methodology to improveoperational effectiveness and efficiency that relies onan analysis of flow10 and value-stream mapping.11 Flow

consists of the methods and structure for how a processis designed. The ultimate goal of lean is to create flowthroughout the value stream by eliminating operationsthat do not add value. Lean methodologies have identifiedeight forms of waste (any factor that does not add value orcontribute to the efficient delivery of service) in healthcare.12 Used in this context, waste is not a pejorative termbut rather a technical term to identify specific inefficien-cies in a process.

The lean quality improvement methodology has beenused successfully in numerous laboratory managementapplications including redesigning laboratory workflow,13

creating a value-stream map (VSM) to reduce cycle timeand lead time,14 improving core laboratory operations bymodifying flow and reducing bottlenecks,15 developingpreanalytic processes,16 improving histopathology of ananatomical pathology laboratory,17 and reducing surgicalpathology in process-misidentification defects.18 Althoughsome blood banks have used lean techniques to improvetheir operation, there has not been a study applying value-stream mapping for CER in transfusion medicine.

Spaghetti chartA spaghetti chart is a diagram of patient flow and staff flow.Every process must operate within the space available andmany processes have numerous steps that involve traveland transport.19 A spaghetti chart visualizes the amount oftravel involved in providing patient care in a way that viewsspace against the goal of minimizing patient and staff time.We identified several steps in handling PLTs in the hospitalblood bank including receipt from the supplier and pre-paration of PLT product to release to the patient care unit(receive PLT order from nurse unit, prepare PLTs, andrelease PLTs to unit). Spaghetti charts were developed foractivities within the blood bank and on the patient careunit showing each of these steps (Table 1).

VSMA VSM is a diagrammatic depiction of all activities thatmust occur to provide a service to the patient. Valuestreams usually cross departmental boundaries and aretypically depicted two ways. The first depiction representsthe way a process operates currently (current state map)and the second portrays the desired process (future statemap). A VSM is a technique to identify efficiency improve-ments that will improve process flow in a way to enhanceperformance and lower cost. It is a waste visualizationmethod that focuses on one process that is mapped frombeginning to end to identify and decrease waste. Waste isany activity that does not add value19,20 to a patient service.A VSM visually shows the flow of material, information,and service from the time a product arrives at an organi-zation; goes through the processing steps; and is safely

RILEY ET AL.

2 TRANSFUSION Volume **, ** **

administered to a patient.10 Whenever a delay is identifiedbetween or within a process step, it is an indicator ofpotential inefficiency. In transfusion medicine, flow ofinformation is just as important as patient flow.

The VSM also uses several symbols to denote theflow of product and information. The lightning lines indi-cate the flow of information, while the arrows indicatethe flow of product from the supplier to the care unit. Theinformation flow is depicted by the lightning lines andcylinder symbolizing the order entry system. Informationflow for the PLT transfusion begins at the patient careunit when a decision is made to transfuse a patient basedon clinical factors or lab results. Orders for PLT trans-fusion are initiated and transmitted to the blood banklaboratory. A cylinder indicates a data base (which is arepository of information that transmits the PLT orderfrom a care unit to the blood bank) and rectangle boxesindicate the process step. At the bottom of the VSM, twotypes of periods are shown: the period to perform a nec-essary function and the period used for non–value-addedsteps, delays, and rework. A time line is drawn at thebottom of the map to indicate the beginning and end ofeach process step as well as the delay between processsteps. The actual time for each process step is written onthe time line below the corresponding process box. Wait

times are denoted on the time line between processboxes. The sum of all process times equals the timerequired to provide the service. Two of the steps dependon the patient care unit staff and are outside the controlof the blood bank staff.

Standard costUsing information obtained from the spaghetti chart todocument movement and VSM to document time, wecalculated the standard PLT cost (all costs associatedfrom the blood bank) and transfusion costs (all costsassociated with patient care unit). To conduct the analy-sis, we used several managerial accounting reports fromthe patient care unit and blood bank to arrive at a stan-dard cost for all patient care unit expenses and all bloodbank expenses for each transfusion administered. Thestandard cost calculation included the direct labor costfor the blood bank and the patient care unit. We did notincorporate the institutional indirect costs because theyare not related to a specific PLT transfusion regime anddid not vary with respect to treatment arms in this study.The objective was to calculate the direct cost associatedwith each PLT transfusion in the study to accurately

TABLE 1. Steps in the blood bank for receipt and storage, preparation, and release of PLTs to the patientcare unit

Process step Station Activity Staff person Time (min) Total elapsed time (min)

Receive product activity1 A Open boxes Lab Tech 2.45 2.452 A Throw away box inserts Lab Tech 0.38 2.833 B Prepare for check-in Lab Tech 0.50 3.334 B Organize products into categories Lab Tech 1.83 5.175 B Check into computer Lab Tech 0.35 5.526 C and D Irradiate bags Lab Tech 3.10 8.627 E Give to components to inventory Lab Tech 0.78 9.408 B Titer “O” PLTs (5 bags of 18 bags) Lab Tech 4.50 13.909 B Document that titer was done Lab Tech 13.52 27.42

10 E Take PLTs to components for inventory Lab Tech 3.42 30.8311 E Inventory product Lab Tech 3.58 34.41

Total time 34.41Time per unit 3.44

Prepare order12 F Order entry received Lab Tech 2 2.90 2.9013 F Place in basket Lab Tech 2 2.93 5.8314 G Pick up order from basket Lab Tech 3 0.17 6.0015 H Identify appropriate PLT product to give Lab Tech 3 2.32 8.3216 I Find PLT product from storage Lab Tech 3 0.13 8.4517 H Reserve PLT product for patient in computer Lab Tech 3 0.55 9.0018 I Return PLT product to storage Lab Tech 3 1.10 10.10

Total time 10.10Release products to patient care unit

19 J Review order Lab Tech 3 0.78 0.7820 I Pick up PLTs Lab Tech 3 0.18 0.9721 J Issue order for patient Lab Tech 3 4.65 5.6222 K Prepare for pneumatic tube transport Lab Tech 3 0.57 6.1823 L Send product Lab Tech 3 0.53 6.72

Total time 6.72

Total time blood bank 20.26

PLT TRANSFUSION COSTS

Volume **, ** ** TRANSFUSION 3

simulate the cost of each dose strategy. We accounted for“nonproductive” time such as staff meetings, in-servicetraining, and work breaks, by adding 60% to the directlymeasured activity time based on a recent nursing timeand motion study of 36 hospitals.20

We used three steps to calculate a standard cost forPLT transfusion. In the first step, we analyzed the totaldirect cost per PLT transfusion (nonproduct) by separatelycalculating total cost in the blood bank and the patientcare unit. The second step involved estimating the cost ofthe PLT products and the third step involved adjustmentsbased on PLT doses.

RESULTS

Within the blood bank laboratory, two stages wereidentified for spaghetti charting—product processing andpatient-specific processing (Fig. 1). The product stage

involved all the activities to receive, inventory, and storethe PLT product (Table 2). The patient stage involved allthe activities to process a specific patient PLT orderreceived from the patient care unit and send the productto the patient care unit. The blood bank workflow for PLTtransfusion processing through these two stages was con-structed by a nonparticipant observer following the flowof PLTs in each stage. This process identified 12 stationsand 23 total steps for both stages within the blood bank(Table 1). For the product stage, there were five stations,11 steps that consumed 34.41 minutes of time (not includ-ing waiting or “down” time between steps; Table 1).However, this time covered the product time for a batch of14 PLT units. Therefore, we arbitrarily divided the totaltime (34.41) by 10 and used a time of 3.44 minutes perunit. For the prepare order and product release stages,seven stations and 12 steps were involved, which con-sumed 16.82 minutes (10.10 + 6.72 min). Thus, there was a

PT

S

172 Floor refrig

Irradiator Plt rotator

167

Freezer

166

H. freeze

205

19

1

18

5

256

175

257

177

286

262

28

7

Com

pon

en

ts

312

Phone (277)

A

B

D

C

I

E

F

G

J

KL

H

Lab Tech 1

Lab Tech 2

Lab Tech 3

Fig. 1. Spaghetti diagram of blood bank workflow for PLT transfusions: from product receipt to release to patient care unit.

RILEY ET AL.

4 TRANSFUSION Volume **, ** **

total blood bank time of 20.26 minutes per unit (Table 2).The blood bank activity used for cost calculations was32.41 minutes (20.26 min + 60%) or 32.32 minutes or0.54 hours.

The spaghetti chart in Fig. 2 depicts the process flowon the patient care unit beginning at the initial review ofthe patient’s PLT count through the ordering and trans-fusing of PLTs until the point of flushing the patient’sintravenous line after transfusion. The spaghetti chartcaptures the activity of three individuals: the health unitcoordinator, the primary nurse, and the second nursewho performs a double check on the patient and PLTunit identity. The chart excludes all the movementbetween steps that are not related to the specific trans-fusion. The chart shows that there are 13 stations with 23total steps and a time analysis indicating 36.47 minutesneeded (Table 2). The patient care unit activity used forcost calculations was 58.36 minutes (36.47 min + 60%) or0.97 hours.

Figure 3 shows a VSM for the blood bank activitiesdepicted in the spaghetti chart (shown in Fig. 1). Value-added time is a term used to indicate time needed forprocess steps that ensure quality and safety for thepatient. Non–value-added time consists of unnecessarysteps and delays that do not make incremental improve-ment to quality or safety of PLT transfusion. Lead time isthe time to process one PLT order through the blood bank.Lead time and value-added time are used to calculatethe process efficiency percentage (PEP), defined as theamount of the process that is value added. The PEP is

calculated by dividing the value-added time by the leadtime and multiplying by 100 as follows:

PEP = Value-added time/Lead time

minutes minutes= ×51 23 145 1. / 000 35 3= . %.

In Fig. 3, the value-added time and lead time are cal-culated to be 51.23 minutes. However, as stated earlier, thiscan vary greatly depending on the actions of the patientcare staff. In our study, the wait times were estimated.

Nonproduct cost per transfusion

Patient care unit cost per transfusionSince a large portion of PLT transfusions in our medicalcenter are given to patients undergoing hematopoieticstem cell transplantation, we used the staffing and costs ofthat patient care unit for these calculations. This includesthe total monthly direct costs (labor, supplies, and otherdirect operating costs) to operate the patient care unit andis divided by the total number of full-time equivalent(FTE) employees multiplied by the monthly hours eachFTE works. This yields an accurate hourly cost of one FTE,which is then multiplied by the total FTE time required totransfuse PLTs.

Total Monthly Labor Expenses

Total FTEs Total MonthlyHours

×pper FTE

Total Transfusion Time

Patient Unit Cost per Tran

×

= ssfusion

TABLE 2. Steps on the patient care unit for transfusion of 1 unit of PLTsStep Station Activity Staff person Time (min) Total elapsed time (min)

1 A Read patient labs Nurse 1 3.00 3.002 B Order PLTs HUC 3.00 6.003 B Look for confirmation by Blood Bank HUC 2.00 8.004 B Send release to Blood Bank HUC 3.00 11.005 C Inform Nurse 1 of PLT arrival HUC 2.00 13.006 D Remove PLTs and check information Nurse 1 3.50 16.507 E Retrieve supplies Nurse 1 0.28 16.788 F Travel Nurse 1 0.23 17.029 F Wash hands Nurse 1 0.32 17.33

10 F Distraction Nurse 1 0.50 17.8311 G Put on sterilization clothes Nurse 1 0.52 18.3512 H Enter patient room Nurse 1 0.50 18.8513 I Take patient vitals Nurse 1 1.18 20.0314 J Confirm correct patient and PLT Nurses 1 and 2 1.37 21.4015 K Wait for other caregiver Nurse 1 0.78 22.1816 I Take patient vitals Nurse 1 2.28 24.4717 L Talk with patient Nurse 1 1.63 26.1018 L Prepare tubing Nurse 1 0.85 26.9519 L Hang PLTs Nurse 1 4.47 31.4220 L Talk with patient Nurse 1 0.58 32.0021 K Document vitals for second time Nurse 1 0.47 32.4722 I Check vitals after 10 min Nurse 1 2.00 34.4723 M Takedown of product and dispose Nurse 1 2.00 36.47

Total Patient care unit 36.47

PLT TRANSFUSION COSTS

Volume **, ** ** TRANSFUSION 5

Example:

$ ,

.. $ . .

475 997

50 14 1600 97 57 71

×× =hours

Blood bank cost per transfusionThe total monthly labor expense incurred in the bloodbank was calculated by summing all PLT products pro-cessed annually and divided by all products processed bythe blood bank. The cost for the blood bank was deter-mined similar to the patient care unit by the formula

Total Monthly Labor Expenses

Total FTEs Total MonthlyHours

×pper FTE

Total Process andRelease Time

Blood Bank Cost per

×

= Unit

Example:

$ ,. $ . .

116 91718 160

0 54 21 93×

× =hours

Thus, the total nonproduct cost per unit was deter-mined as follows:

Total cost per unit transfused Blood bank cost unit

Total

+=

/

nnonproduct cost unit/ .

For example,

$ . $ . $ . .57 71 21 93 79 64+ =

Fig. 2. Spaghetti diagram of activity on patient care unit from receipt of PLT unit to completion of transfusion.

RILEY ET AL.

6 TRANSFUSION Volume **, ** **

Total product cost per transfusionThe product cost was calculated based on the cost of 1apheresis unit and multiplied by the amount of productrequired. Volume discounts or different pricing by thesupplier were not taken into consideration. An arbitraryprice of $390 was used for one standard apheresis unitbecause this is the current cost to our hospital. This wasdoubled for high-dose transfusion and halved for a low-dose transfusion.

Adjustments to the transfusion product cost andnonproduct cost depending on doseFor small-dose apheresis units, an adjustment was madeto account for the additional time required to split theunit in the hospital blood bank. Because at the time ofobservation, bag splitting was not being performed, anestimate of 9 minutes was provided by employees as tothe time required to split a bag of PLTs. For a large dose,an adjustment was made in the patient care unit causedby additional nursing time to administer a second bag.A third additional cost was when an adverse reactionoccurs, thus involving physician time, nurse time, andtechnologist time. This is based on the assumption thata larger number of transfusions will lead to more trans-fusion reactions.

Small-dose adjustment

Total time to split bag

Blood bank technologist cost per h

×oour

Split results in two bags

Additional cost for smaller = uunits.

0 15 40 602

3 04. $ .

$ . .hours per hour× =

Large-dose adjustmentFor the high-dose transfusions, we presumed that twoseparate bags of PLTs would be administered. We attrib-uted the time required for a second bag only for high-dosetransfusions. If the supplier provided high-dose products,in one bag, this extra cost would not be incurred.

Additional time for second bag Cost per hour× $

0 23 59 33 13 38. $ . $ . .hours × =

Cost of reaction per transfusionThe probability of an adverse reaction is approximately1.7%.21 To be conservative, we estimated the adverse reac-tion rate at 1%. Patient care and transfusion physicians,nurses, and blood bank laboratory staff are involved in atransfusion reaction.

Fig. 3. VSM for blood bank.

PLT TRANSFUSION COSTS

Volume **, ** ** TRANSFUSION 7

Total Physician Time Physician Cost per Hour

Total Tech Tim

×ee Technologist Cost per Hour

Total Nurse Time Nurse Cost p

×× eer Hour

Total Cost per Reaction

⎫⎬⎪

⎭⎪

=

Likelihood of reaction Total cost per reaction

Cost per un

×= iit

We estimated that each reaction consumed a total of1 hour of physician time (patient care physician and initialtransfusion medicine physician interpretation and finalinterpretation by transfusion medicine physician) and1 hour of blood bank technologist time and two-thirds ofan hour of the patient care nurse’s time.

1 111 31 111 31hour physician× =$ . $ . ( )

1 39 00 39 00hour technologists× =$ . $ . ( )

0 67 59 33 39 75. $ . $ . ( )hours nurse× =

Total cost per reaction = $ .190 06

1 190 06 1 90% $ . $ . .likelihood of reaction per transfusion× =

We analyzed three separate dose scenarios using twosimulation trials to assess the financial impact of each dosealternative. Although other kinds of patients receive PLTtransfusions, each analysis included 259 patients that rep-resent the appropriate number of annual bone marrowtransplantation (BMT) patients at the medical center.Based on our specific experience in the PLADO study,5 wepresumed that patients in the low-, medium-, and high-dose strategies would receive 16, 12, and 8 transfusions,respectively, during their course of therapy. The low-,medium-, and high-dose transfusions would equal 0.5, 1,and 2 standard apheresis units. Table 3 simulates thetotal PLT transfusion cost based on these assumptions.Summing the product and nonproduct costs per patient

shows a cost range of $4503.77 to $7014.59, indicating thatthe high-dose scenario costs $2510.82 more per patient or$650,000 more than the low-dose scenario ($1,812,000compared with $1,164,000) for the overall patient popula-tion of the simulated BMT program.

Next we reversed the analysis to estimate the numberof transfusions to equalize the cost for each of the three PLTdoses (Table 4).The purpose of this simulation was to iden-tify the number of transfusions per patient that could beadministered to equalize the total cost across treatmentarms. The analysis shows that, if total annual cost isnearly equivalent for the three arms, the low-treatmentarm would necessitate 19 transfusions per patient, themedium-treatment arm 12 transfusions, and the high-treatment arm seven transfusions.

Table 5 indicates the major cost components of thetwo simulations to illustrate the cost structure. The coststructure of the corresponding treatment arms remainsthe same for both analyses. In the low-dose scenario,69% of the costs are from the PLT product with 31% ofthe cost coming from transfusions. The medium-doseshows an 83% cost for PLTs and a 17% cost for transfu-sions. For the high dose, 89% of the cost can be attrib-uted to the cost of the PLTs with the remaining 11%attributed to transfusion cost. Results are similar for bothdose strategy simulations.

The results are especially sensitive to the productcosts. Depending on what kind of pricing is used by PLTsupplier, the results could dramatically differ and thelowest cost scenario could shift. For example, if we makethe product cost for the low-dose scenario to be 70% ofthe cost of a medium dose (as opposed to the 50% weassumed in the previous tables), the medium dosebecomes less expensive (Table 6).

Another alternative to be considered is the potentialto receive a discount from the PLT supplier if the high-dose scenario became the preferred treatment method.In other words, rather than assume the high-doseproduct would cost 100% more than the medium dose,what if the high dose was only 60% more cost than the

medium? The results of this scenariomakes the medium cost more expen-sive than the high dose and the lowdose remains as the preferred methodfrom a purely cost standpoint (Table 7).

DISCUSSION

The aim of cost-effectiveness researchis to evaluate the comparative impactof alternative clinical measures.1-4 Thisstudy compares the total cost of threeclinically equivalent PLT dose strategiesusing four cost simulations for theannual patient base of a major BMT

TABLE 3. PLT therapy of different dose regimens using number oftransfusions per dose

Category

Simulation 1

Low Medium High

Number of transfusions/patient 16 12 8Number of units/transfusion

(1 unit = one apheresis procedure)0.5 1 2

Total PLT cost/transfusion $195.00 $390.00 $780.00Total transfusion cost/transfusion $86.49 $81.54 $96.82Total PLT cost/patient $3,120.00 $4,680.00 $6,240.00Total transfusion cost/patient $1,383.77 $978.48 $774.59Total cost/patient $4,503.77 $5,658.48 $7,014.59

Total annual cost* $1,163,504 $1,461,812 $1,812,149

* Based on 259 patients in simulated BMT program.

RILEY ET AL.

8 TRANSFUSION Volume **, ** **

center. The findings show that the low-dose PLT therapy is also the lowest costtherapy, approximately $2500 less perpatient or $650,000 less than the highdose per course of therapy for the simu-lated transplant program. This findingis surprising considering the low-dosetherapy involves more transfusions perpatient than the high-dose, medicallyequivalent therapy. This is explained bythe cost structure of the product and thenonproduct transfusion costs. The find-ings indicate that the nonproduct costsfor the high-dose regimen have econo-mies of scale (specifically, the averagecost decreases as volume increases)while the product costs have a disec-onomy of scale (the average costincreases as volume increases). In otherwords, the product is the cost driverdespite the efficiencies gained in thenonproduct costs of the medium-doseand high-dose therapies, in comparisonto the low-dose therapy. While the priceof the PLT product used in this studyis lower than other published prices,use of a higher price would only furtheraccentuate the results of this study,which are that the product price ratherthan the cost of giving the transfusion isthe major influence on the overall costof PLT therapy. Annually for a simulatedprogram transplanting 259 patients peryear, there would be a 56% cost differ-ence ($1.164 million vs. $1.812 million)between two therapies that providesimilar control of bleeding.5 In thisstudy, we focused on personnel costs ofadministering the PLT transfusions anddid not include supplies, overhead forthe facilities, or indirect costs sincethese latter two are not likely to be influ-enced by the number of PLT transfu-sions given. The cost of supplies for thetransfusion would be quite small andnot influence the conclusions of thestudy. We did presume that more trans-fusions might lead to more transfusionreactions and estimated a cost for this.However, that cost was small and didnot meaningfully affect our results.Some high-dose transfusions might beone high-dose product from one donorwhile others might involve two productsfrom different donors. Larger plasma in

TABLE 4. PLT transfusion cost based on cost-neutral simulation

Category

Simulation 2

Low Medium High

Number of transfusions/patient 19 12 7Number of units/transfusion

(1 unit = one apheresis procedure)0.5 1 2

Total PLT cost/transfusion $195.00 $390.00 $780.00Total transfusion cost/transfusion $86.49 $81.54 $96.82Total PLT cost/patient $3,705.00 $4,680.00 $5,460.00Total transfusion cost/patient $1,643.22 $978.48 $677.77Total cost/patient $5,348.22 $5,658.48 $6,137.77

Total annual cost* $1,381,660 $1,461,812 $1,585,631

* This exercise determines the number of transfusions for each dose if the objectiveis to provide a course of therapy with similar costs for 259 patient simulated BMTprogram.

TABLE 5. Cost structures of PLT transfusion cost simulations

Dose category

Simulation 1 Simulation 2

Low Medium High Low Medium High

Number of transfusions/patient 16 12 8 19 12 7Number of units/transfusion 0.5 1 2 0.5 1 2Total PLT cost/patient 69% 83% 89% 69% 83% 89%Total transfusion cost/patient 31% 17% 11% 31% 17% 11%Total cost/patient 100% 100% 100% 100% 100% 100%

TABLE 6. PLT cost scenarios assuming cost of low-dose units of 70%standard-dose product

Category Low Medium High

Number of transfusions/patient 16 12 8Number of units/transfusion 0.5 1 2Total PLT cost/transfusion $273.00 $390.00 $780.00Total transfusion cost/patient $86.49 $81.54 $96.82Total PLT cost/patient $4,368.00 $4,680.00 $6,240.00Total transfusion cost/patient $1,383.77 $978.48 $774.59Total cost/patient $5,751.77 $5,658.48 $7,014.59

Total annual cost $1,485,912 $1,461,812 $1,812,149

TABLE 7. PLT cost scenario assuming cost of high-dose units are60% more than the standard-dose product

Category Low Medium High

Number of transfusions/patient 16 12 8Number of units/transfusion 0.5 1 2Total PLT cost/transfusion $195.00 $390.00 $624.00Total transfusion cost/patient $86.49 $81.54 $96.82Total PLT cost/patient $3,120.00 $4,680.00 $4,992.00Total transfusion cost/patient $1,383.77 $978.48 $774.59Total cost/patient $4,503.77 $5,658.48 $5,766.59

Total annual cost $1,163,504 $1,461,812 $1,489,741

PLT TRANSFUSION COSTS

Volume **, ** ** TRANSFUSION 9

high-dose transfusions might also lead to an increase inreactions. We did not account for that possible increase intransfusion reactions.

The costs of giving the transfusion were smaller thanwe expected and, as a result, the product costs are thedriving factor in the overall cost of PLT therapy. The totalnonproduct costs for the low-dose strategy was only $609more than the nonproduct cost of the high-dose strategyand this was more than offset by the lower cost of the PLTproduct. Because these simulations are very sensitive tothe product cost, less of a reduction in price of the low-dose product has a huge impact on these scenarios. Weprojected a product cost of one-half or double the stan-dard product cost for the low- and high-dose products,respectively. It seems likely that suppliers would not splitstandard-dose units and this would be done at the hospi-tal. If so, the assumption seems reasonable that the cost ofa low-dose product would be half the cost of a standardproduct plus the costs of splitting the unit, which we haveincluded. It is possible that blood suppliers would use adifferent price adjustment and this would influence thestudy conclusions.

Although the specific steps in handling units of PLTswithin the hospital blood bank and on the patient careunit may differ in other institutions, we believe that theconclusions of this study should generalize to other insti-tutions. There were a large number of steps within theblood bank and on the patient care unit suggesting thatthis operation was not highly efficient. This was alsosupported by the VSM that showed a considerable amountof non–value-added time. Thus, if other institutions aremore efficient, the costs of giving the transfusion might beeven lower than suggested here further accentuating theimpact of the product costs rather than costs of giving thetransfusion. If the costs of the product are higher thanused in this study, this also increases the impact of theproduct cost instead of the transfusion process.

The absolute costs of providing the PLT transfusionwill vary depending on local factors and we do not suggestthat the costs reported here will be applicable elsewhere.However, the activities in other institutions could be com-pared with the specific steps listed here to provide ageneral idea of whether the process in those institutions ismore or less efficient than that described here. From that arough comparison could be made with the direct times of20.26 minutes in the blood bank and 36.47 minutes on thepatient care unit observed in this study. We applied a 60%factor for nonproductive time and did not includesupplies, overhead, or indirect costs; those assumptionsmight be different in other institutions. However, supplycosts are not likely to vary much in different institutionsand indirect or overhead costs are not likely to changedepending on the number of transfusions given, so itseems unlikely that these factors will have a major influ-ence on the general results of this study—that the product

costs rather than the transfusion costs are the drivingfactor in PLT therapy.

We did not consider giving transfusions in the outpa-tient setting. More patients are being managed in thatsetting and more transfusions given there. Thus, if thehigh-dose strategy was used there and this reduced thenumber of clinic visits, this could affect our analysis andthe conclusions of this study. In conclusion, the low-dosetransfusion strategy is substantially less costly thancurrent or high-dose strategies but only if the supplierprices the low-dose product substantially less than thecurrent standard-dose product or that a standard productis split at the hospital blood bank.

CONFLICT OF INTEREST

None.

REFERENCES

1. Tunis SR, Benner J, McClellan M. Comparative effective-

ness research: policy context, methods development and

research infrastructure. Stat Med 2010;29:1963-76.

2. Lauer MS. The historical and moral imperatives of

comparative effectiveness research. Stat Med 2010;29:

1982-4.

3. Federal Coordinating Council. Report to the President and

the Congress on comparative effectiveness research. Feb-

ruary 17, 2010. [cited 2010 Feb 17]. Available from: URL:

http://www.hhs.gov: http://www.hhs.gov/recovery/

programs/cer/execsummary.html

4. Committee on Comparative Effectiveness Research

Prioritization. Initial national priorities for comparative

effectiveness research. Washington (DC): The National

Academies Press; 2009.

5. Slichter J, Kaufman R, Assmann SF, McCullough J, Triulzi

DJ, Strauss RG, Gernsheimer TB, Ness PM, Brecher ME,

Josephson CD, Konkle BA, Woodson RD, Ortel TL, Hillyer

CD, Skerrett DL, McCrae KR, Sloan SR, Uhl L, George JN,

Aquino VM, Manno CS, McFarland JG, Hess JR, Leissinger

C, Granger S. Dose of prophylactic platelet transfusions

and prevention of hemorrhage (the PLADO trial). NEJM

2010;362:600-13.

6. Hersh J, Hom E, Brecher M. Mathematical modeling of

platelet survival with implications for optimal transfusion

practice in the chronoically platelet transfusion-dependent

patient. Transfusion 1998;38:637-744.

7. Klumpp TR, Herman JH, Gaughan JP, Russo RR, Christman

RA, Goldberg SL, Ackerman SJ, Bleecker GC, Mangan KF.

Clinical consequences of alterations in platelet transfusion

dose: a prospective, randomized, double-blind trial.

Transfusion 1999;39:674-81.

8. Goodnough L, Kuter D, McCullough J. Prophylactic platelet

transfusions from healthy normal apheresis platelet donors

RILEY ET AL.

10 TRANSFUSION Volume **, ** **

undergoing treatment with thrombopoietin. Blood 2001;97:

1346-51.

9. Heddle NM, Cook RJ, Tinmouth A, Kouroukis CT, Hervig T,

Klapper E, Brandwein JM, Szczepiorkowski ZM, AuBuchon

JP, Barty RL, Lee KA; SToP Study Investigators of the BEST

Collaborative. A randomized controlled trial comparing

standard- and low-dose strategies for transfusion of

platelets (StoP) to patients with thrombocytopenia. Blood

2009;113:1564-73.

10. Smith A, Barry R, Brubaker C. Going lean: busting barriers

to patient flow. Chicago (IL): Health Administration Press;

2008.

11. Zidel T. A lean guide to transforming healthcare: how to

implement lean principles in hospitals, medical offices,

clinics and other healthcare organizations. Milwaukee

(WI): American Society for Quality Press; 2006.

12. Black J, Miller D. The Toyota way to healthcare excellence:

increase efficiency and improve quality with lean. Chicago

(IL): Health Administration Press; 2008.

13. Garikes R. Lean lab design. Med Lab Obs 2004;36:30-4.

14. Joseph T. Design a lean laboratory layout. Med Lab Obs

2006;38:24-31.

15. Rutledge J, Xu M, Simpson J. Application of the Toyota pro-

duction system improves core laboratory operations. Am J

Clin Pathol 2010;133:24-31.

16. Persoon TJ, Zaleski S, Frerichs J. Improving preanalytic

processes using the principles of lean production

(Toyota production system). Am J Clin Pathol 2006;

125:16-25.

17. Raab S, Grzybicki D, Condel J. Effect of lean method imple-

mentation in the histopathology section of an anatomical

pathology laboratory. J Clin Pathol 2008;61:1193-9.

18. Zarbo R, Tuthill M, D’Angelo R, Varney R, Mahar B,

Neuman C, Ormsby A. The Henry Ford production system

reduction of surgical pathology in-process misidentifica-

tion defects by bar code-sepcified work process standard-

ization. Am J Clin Pathol 2009;131:468-77.

19. Ohno T. Toyota production system: beyond large-scale

production. Portland (OR): Productivity, Inc.; 1988.

20. Hundrich A, Chow M, Skierczynski BA, Lu Z. A 36-hospital

time and motion study: how do medical-surgical nurses

spend their time? Perm J 2008;12:25-34.

21. Tormey C, Sweeney J, Champion MP, Snyder E, Wu Y.

Analysis of transfusion reactions associated with

prestorage-pooled platelet components. Transfusion

2009;49:1242-7.

PLT TRANSFUSION COSTS

Volume **, ** ** TRANSFUSION 11