A Blood Ordering Strategy fo from a Computer Simulation. Hospital Blood Banks Derived

Transcription

1 A Bld rdering Strategy f frm a Cmputer Simulatin Hspital Bld Banks Derived BRUCE A. FRIEDMAN, M.D., RBERT D. ABBTT, PH.D., GERGE W. WILLIAMS, PH.D. The ultimate gal f a hspital bld bank inventry cntrl prgram is t reduce wastage f bld prducts and unnecessary use f labratry services withut jepardizing patient safety. The develpment f a practical bld rdering plicy at the hspital level is an integral part f any such prgram. In rder t explre varius bld rdering ptins in detail, a cmputer simulatin f a hspital bld inventry was used t assess the impact n bld bank perfrmance measures f reductins in grup and nn-grup- bld levels frm baseline levels, assuming bth a 21-day and 35-day shelf life. n the basis f data derived frm this study shwing that such inventry reductins accmpanied by partial prtectin f the grup inventry will nt result in significant shrtages, a practical strategy was develped fr establishing ptimal target inventry levels fr a hspital n an empirical basis. These target levels can serve as a guide fr subsequent bld rdering. A step-bystep apprach fr analyzing a hspital bld inventry cntrl prgram is then suggested, accmpanied by an actin plan fr implementing change which incrprates the experimentallyderived bld rdering strategy. Adherence t this plan shuld result in a lw utdate rate, a reductin in unnecessary crssmatching, and greater availability f bld fr thse patients with a legitimate need fr it. (Key wrds: Inventry cntrl; Bld rdering; Crssmatching; utdate rate; Cmputer simulatin; CPDA-1; Shelf life f bld; Bld shrtage; C/T rati) Am J Clin Pathl 1982; 78: ALTHUGH CMMNLY CNSIDERED a merely perfunctry exercise in hspital bld banks, the task f rdering bld frm a reginal bld supplier shuld be cnsidered as ne f the mst imprtant elements in an effective bld bank inventry cntrl prgram. Bld rders shuld be based n a peridic quantitative and qualitative assessment f the current bld inventry, adjusted t take int cnsideratin immediate and ptential demand fr bld amng hspital patients. Unfrtunately, bld rdering is ften perfrmed in many hspital bld banks in a mechanistic and inflexible manner, and based n target inventry levels which may have been develped many years previusly and may be far t high in view f current attitudes tward bld inventry. Excessive bld rdering ften results in an inflated inventry and bld wastage Received September 8, 1981; received revised manuscript and accepted fr publicatin December 21, Address reprint requests t Dr. Friedman: Department f Pathlgy, University f Michigan Medical Schl, 1335 E. Catherine St., Bx 45, Ann Arbr, Michigan /82/0800/0154 $01.15 Department f Pathlgy, The University f Michigan Medical Schl, Department f Epidemilgy, The University f Michigan Schl f Public Health, Ann Arbr, Michigan; Bimetrics Research Branch, Natinal Heart, Lung and Bld Institute, Bethesda, Maryland and Department f Bistatistics, The Cleveland Clinic Fundatin, Cleveland, hi thrugh utdating. Recent develpments in bld banking such as the extended shelf-life f bld, practical methds fr reducing excessive crssmatching, and the greater acceptance by clinicians f the transfusin f AB cmpatible bld in lieu f AB specific bld nw ffer bld bank wrkers the pprtunity t revamp their bld rdering prcedures and maintain fewer units in their bld inventries. Bld rdering, f curse, is nly ne facet f the larger issue f hspital bld bank inventry cntrl. It is impssible t cncern neself with bld rdering withut simultaneusly cnsidering parallel issues such as the number f units crssmatched r the time spent by such units in a reserved status. It is imprtant t state at the utset f this reprt that the ultimate gal f any bld bank inventry cntrl prgram, including the bld rdering functin, is t reduce wastage f bld prducts withut jepardizing patient welfare. It is becming increasingly apparent that the reductin f bld wastage and unnecessary bld transfusin is a mre cst-effective way t increase the bld supply than drawing mre bld dnrs. Tw majr issues relating t bld rdering and inventry cntrl were explred in this study using a cmputer simulatin mdeled n a hspital bld bank inventry. The first f these was the extent t which the amunt f grup bld in inventry can serve as a buffer against peridic shrtages f the ther bld types, thereby allwing the maintenance f a smaller running bld inventry. The secnd was the effect n bld bank perfrmance measures f the extensin f the shelf-life f bld frm 21 days t 35 days with CPD-adenine as the anticagulant-preservative slutin. Using data gleaned frm the cmputer mdel relating t bld rdering, a practical bld rdering strategy fr use by hspital bld bank persnnel was devised. :rican Sciety f Clinical Pathlgists 154

2 Vl. 78 N. 2 BLD RDERING STRATEGY 155 This strategy was then placed within the mre general cntext f an verall plan which can be implemented in rder t analyze a hspital bld bank inventry cntrl prgram and crrect any deficiencies nted. Materials and Methds The cmputer mdel used t simulate the cnsequences f the hspital bld bank inventry alteratins examined in this study has been previusly described in detail. 1 It was based n data cllected in University Hspital Bld Bank, University f Michigan, Ann Arbr, Michigan, during This bld bank serves a large tertiary-care teaching hspital in which 17,446 red units f bld (red bld cells plus whle bld) were transfused in the year that the mdel was develped. The stimulus fr the develpment f the cmputer mdel stemmed frm the perceived need t assess the relative benefits and risks f bld bank inventry cntrl plicy changes prir t their actual implementatin in the hspital setting. The simulatin was designed t take int accunt imprtant bld inventry characteristics cmmn t hspital bld banks nt previusly cnsidered in bld bank mdels such as the day-t-day influences within a week, the differing amunt f time that crssmatched units f bld are held in a crssmatched (reserved) status, the varying age f bld received by hspital bld banks frm their reginal bld suppliers, and the different utilizatin rates f red bld cells versus whle bld. The cmputer mdel which was develped incrprated the ptin t fill individual patients' bld rders with AB cmpatible bld when they culd nt be filled with AB specific bld due t bld shrtages, and t transfuse Rh-negative bld near utdate t Rhpsitive recipients. This inclusin f a bld switching strategy int the prcess f bld selectin required the develpment f hierarchical bld switching schemes which have been previusly detailed. 1 In the event that all pssible chices in a hierarchical bld switching list are exhausted during the search fr an apprpriate unit f bld fr a patient in the simulatin, the request is designated as a "unit refusal." Validatin f the cmputer mdel has been discussed previusly, 1 and was based n statistical agreement and crrespndence between the simulated data and the actual data which was cllected in the University Hspital Bld Bank. The cmputer simulatin was used t explre tw separate but related questins cncerning bld rdering and inventry cntrl. The first f these invlved an assessment f the impact n hspital bld bank perfrmance measures f varius reductins in the nngrup- bld inventry while hlding the grup inventry cnstant, varius reductins in the grup inventry while hlding the nn-grup- inventry cnstant, and, lastly, simultaneus reductins in the inventry f bth bld grups. The secnd questin dealt with was hw these same bld bank perfrmance measures wuld be affected by the extensin f the shelf-life f bld frm 21 days using CPD anticagulant-preservative slutin t CPD-adenine and a 35- day shelf-life. This latter questin was explred by cmparing the impact f the varius inventry reductins under bth a 21-day and 35-day utdate perid. The bld bank perfrmance measures examined in the study were the utdate rate, the percentage f unit rders that were refused due t nnavailability f grup specific r cmpatible bld, the mean age f transfused bld, and the percentage f crssmatched units which were grup. These perfrmance measures are analgus t the "measures f effectiveness" described previusly by Jennings. 10 The utdate rate was defined as 100 times the number f units f bld utdating divided by the sum f the number f units utdating plus the number f units transfused. The percentage f unit rders which were refused was defined as 100 times the number f units rdered and refused divided by the sum f the number f units rdered and refused plus the number f units crssmatched. The percentage f crssmatched units which were grup was defined as 100 times the number f units f grup bld which were crssmatched divided by the ttal number f units crssmatched. Grup- and nn-grup- inventries were assumed t encmpass bth Rh-psitive and Rh-negative units and bth red bld cells (packed cells) and whle bld. Fr the simulatins perfrmed using the 35-day shelflife, tw assumptins were made in rder t adapt the riginal mdel which had been develped assuming a 21-day shelf-life t the lnger shelf-life. First, it was assumed that the randm distributin f units f bld received in the hspital bld bank frm the reginal supplier did nt change with the lnger shelf-life. This was cnsidered t be a reasnable assumptin because less than tw per cent f the units f bld received by University Hspital Bld Bank during the perid mdeled were lder than 14 days. This assumptin wuld nt be valid if the bld bank had received large quantities f lder bld under a recycling prgram during the perid f data cllectin. The secnd assumptin made in rder t adapt the mdel t a 35-day shelf-life was that units f Rh-negative bld which were 33 days ld r lder culd be transfused t either Rh-negative r Rh-psitive recipients. This assumptin was necessary because a similar plicy was fllwed with regard t Rh-negative units reaching 19 days under the 21rday shelf-life mdel.

3 156 FRIEDMAN ET AL. A.J.C.P. August 1982 Results The changes in the fur bld bank perfrmance measures during the 32 cmputer simulatins perfrmed in the curse f this study are displayed in Table 1. Each square in Table 1 crrespnds t a single simulatin, and is based n reductins f 20, 40, and 60 per cent in either r bth grup and nn-grup-0 bld inventry frm baseline levels, assuming a 21-day and 35-day shelf-life. Each simulatin was run fr the equivalent f a 27-week perid. The baseline levels fr the 21-day and 35-day simulatins are represented by the "index square" in the upper left-hand crner f each grid which crrespnds t a zer per cent reductin in bth grup and nn-grup- bld inventry. The mean daily bld inventry level in this baseline state was 256 units, f which 44 per cent was grup. The mean daily demand fr bld was 110 units. This demand figure was held cnstant during all simulatins, regardless f the simulated inventry manipulatins. f the mean number f units demanded daily, 43 per cent were fr patients whse actual bld grup was. The influence n perfrmance measures f reductins in the grup and nn-grup- bld inventries accmpanied by changing shelf-life f bld were evaluated statistically by weighted least squares. Specific statistical results are nt presented here because they add little additinal insight int the general impressins and recmmendatins presented in the remainder f this reprt. Cnsidering first the bld utdate rate with the varius inventry reductins under a 21-day shelf-life, it can be seen that islated reductins in the nn-grup- bld inventry reduce the utdate rate t a greater extent than reductins in the grup inventry. Therefre, marked reductins in the utdate rate can be achieved by preferentially reducing the number f units f A, B, and AB bld, as ppsed t grup units, because these frmer units are mre likely t utdate. 9 It might be argued that the greater reductin in utdating achieved by a reductin in the nn-grup- inventry is actually an artifact f the system using the fllwing line f reasning. Because 56 per cent f the ttal bld inventry n a daily basis were nn-grup- units, a percentage reductin f these bld grups wuld cause a greater decrease in the ttal number f units in inventry than wuld an identical percentage reductin in the grup inventry. This disprprtinate reductin f nn-grup- bld cmpared with that f grup bld wuld autmatically cause a greater reductin in the utdate rate which is quite sensitive t the size f the ttal bld inventry. Such a pssibility did nt prve t be valid because the difference in the decrease in the utdate rate due t identical percentage reductins in the grup versus nn-grup- units was t large t be ttally explained by this mechanism. The mst dramatic decrease in the utdate rate under a 21 -day shelf-life is achieved by a 60 per cent reductin f bth the grup and nn-grup- bld inventries, but the csts f implementing such reductins in terms f their impact n perfrmance measures ther than utdating is unacceptable. Nte in the instance just cited that the percentage f unit rders which are refused is per cent. A methd shuld be sught fr reducing the size f running bld inventries in hspitals while simultaneusly minimizing bld shrtages. Bld utdating becmes negligible (0.18 per cent) in the index square under a 35-day shelf-life with CPDadenine. Therefre, the varius simulated bld inventry reductins which serve t reduce utdating with a 21-day shelf-life frm the 5.15 per cent index value will nt prvide significant additinal reductins with the lnger shelf-life. Tw grups f investigatrs have als shwn that the intrductin f CPD-adenine will cause a dramatic effect n utdating, prducing apprximately a 50 t 79 per cent reductin in this perfrmance measure. 613 Cnsider next in Table 1 the percentage f unit rders which are refused. This is a measure f the extent t which the simulated inventry reductins serve t cause bld shrtages where they did nt exist previusly; that is, it depicts the extent t which a unit rder cannt be filled frm the bld inventry, even after switching a patient frm AB specific t AB cmpatible bld. A value f 0.10 per cent indicates that ne ut f every 1,000 units rdered cannt be prvided due t an insufficient bld supply. Nte first that the patterns fr this particular perfrmance measure are similar fr bth the 21-day and 35-day shelf-life perids. Such a finding is nt unexpected: bld shrtages are mst directly linked t the ttal number f units f bld in inventry and the distributin f the units by grup, bth f which remain the same under the 21-day and 35-day shelf-life simulatins. Nte als, and mre imprtantly, that the sharp reductins in the nn-grup- inventry fr either shelf-life perid have little adverse effect n the percentage f unit rder refusals. Fr example, a 60 per cent reductin in the nn-grup- inventry withut an alteratin f the grup inventry under bth a 21- day and 35-day shelf-life causes less than ne unit rder in every 100 t be refused. This lw shrtage rate can be explained by the fact that when nn-grup- bld is rdered and fund t be unavailable, the rder can be switched immediately t grup inventry, if available. A relatively high unit refusal rate is prduced when the grup inventry is reduced by 60 per cent

4 :ry Percentagi v * - Percentage Reductin in Nn-Grup- Inventry 05 W -i 3 gs A *. *». *. N N LU W ) y I p 4* L LA t b b N'-U b* ww -JN-- LA ^1 s l W s I ^- *J L N LA C09 ++ (- # 00 sr 3 00 t J W^jpw iu vciyi 4^ N t b\ s i w ^ l 4^ LA 00 N u w b ^j V b ^1 p k» ~- UJ M K) *- N> s S i- s b I*J yi ^~ S ^5 75 <* e 0 7 ^jyi t t t LA b b u> \ b LA LA t ** L J*. t N U) N ; _ w C\ i y> N _ N _ LA ^ A u r 00 -J ui *» % " 3 5 C a * 03 P 4* N) 00 W u> LA 4*. N t 0 ^ b b N t N LA L N N 4t t ^ w b b N N N N 4U. ^ N LA V N L 00 Percentage Reductin in Nn-Grup- Inventry N ^ ^ 3 s> S 3 i» = 2 8-S A * >< c is C. 3 3 a-q \ 00 p p b L* b b s 4^ (*- u> 4fc 0 N N N ' L - ** p p p 0i ^- b k> K> Wi K> -P. U ~ p p L U> U» 4*. 4^ ** p P '0 ' '< '<*> N> p "l N L L N 0 t W w b b U (7) ^ 0 4^ W N) W W t LA N N 0 S3 V 1 a 3 *< j gig. 3 3" N <0 4* U*N^-p ^ i b i t LA 4*. t ~ ^ I ^ 4^-J K) N t N t N LA t 4* N -J N LA LA K> "i c a c 3 a. 4* U NJ sjp u» t w 0-00 u> 00 4* i Itk. 4*. -J N p b i T- N U> N LA p ^J N U» N ^ U> -J N N LA v U b ' r N -J Lu LA N ^ Z.SI Anivyis Niii3a>i aia t N 8i PA

5 158 FRIEDMAN ET AL. A.J.C.P.. August 1982 and the nn-grup- inventry is held cnstant. f curse, the mst serius bld shrtages are prduced by 60 per cent reductins f bth grup and ringrup-0 units. Such a maneuver simultaneusly reduces the supply f bth grup specific and grup cmpatible units which are available fr a ptential bld recipient. Examining next the mean age f transfused bld in Table 1 fr the 21-day and 35-day shelf-life simulatins, it can be seen that increasing the shelf-life f bld by tw weeks generally causes a slight increase in the mean age f transfused bld fr mst f the inventry reductins examined. This change ccurs because many f thse units which wuld nrmally utdate under the 21-day shelf-life are retained in inventry and ultimately transfused under the 35-day shelf-life. This retentin and ultimate transfusin f units fr an additinal tw weeks serves t slight jy increase the mean age f transfused bld. There is thus a trade-ff assciated with the lnger shelf-life f bld a reductin f bld utdating is exchanged fr a slightly increased mean age f transfused units. Such an exchange is advantageus because there are n adverse clinical effects assciated with the transfusin f bld which is, n the average, abut ne day lder. After 60 per cent reductins f grup and nn-grbup-0 bld, the mean age f transfused bld is similar under bth the 21- day and 35-day simulatins. Bld shrtages cause units f bld t be transfused relatively quickly after they have been accepted int inventry, resulting in a lwered verall mean age f transfused units. The final perfrmance measure listed in Table 1 is the percentage f units crssmatched which are grup. This is an indicatr f the impact f inventry alteratins n bld switching activity. Reducing the nngrup- inventry while hlding the grup inventry steady causes a higher percentage f crssmatched units t be grup because patients are being switched t grup bld in the face f the increasing shrt supply f nn-grup- bld. As the grup bld inventry is reduced t 60 per cent f the baseline level with the nh-grup- inventry held cnstant, the percentage f crssmatched units which are grup is reduced because there are simply fewer grup units available fr crssmatching. This latter inventry manipulatin als causes the unit refusal rate t g up dramatically. There is a parallel relatinship between the percentage f units crssmatched which are grup and the percentage f unit rders which are refused. Grup crssmatching activity is similar under bth shelf-life simulatins after cmparable reductins because neither the number nr distributin by grup f units changes in cnnectin with the lnger shelf-life. Discussin The first step in planning a hspital bld rdering strategy is t develp realistic target bld inventry levels cnsisting f the number f units f bld by grup and type which bld bank persnnel will attempt t maintain n a daily basis in the running bld inventry. The ptimal target inventry levels fr a given hspital are dependent n a number f variables such as the daily demand fr bld, cnsidered either in terms f units crssmatched r units transfused, the crssmatch t transfusin rati, and the time that units spend in a reserved status, and shuld theretically result in the lwest pssible shrtage and utdate rates fr that hspital. Because f the cmplexity and uncertainty f deriving target levels mathematically, mst hspital bld banks have arrived at individualized bld targets at levels n an empirical basis, ften erring n the side f excess bld inventry in rder t avid bld shrtages. In additin t the target inventry levels, bld bank wrkers ften keep tw ther levels firmly in mind when rdering and releasing bld fr transfusin. The first f these are the levels f the individual bld types at which additinal units are rdered frm the reginal bld supplier r drawn frm dnrs in rder t return t the ptimal pre-established target levels. Jennings 10 calls these levels the rerder pints. The secnd and mre critical levels are the emergency minimums fr the varius bld types, mst cmmnly established fr grup units, belw which elective surgery is ften cancelled and the remaining grup units reserved fr emergency use. Grups f investigatrs have published frmulas fr calculating ptimal target inventry levels fr hspital bld banks. Fr example, Brdheim, Hirsch, and Prastacs state that the rati f ttal units f bld by type in inventry t the mean number f units f that type which are crssmatched daily will be apprximately 3.1 and 2.3 in rder t yield shrtage rates between 5 and 10 per cent. 3 It was unfrtunate that these authrs required the selectin f an "acceptable" shrtage rate prir t the establishment f ptimal target inventry levels. Their reprt did little t assuage the anxiety f bld bank wrkers wh dread that any reductins in their hspital bld inventry levels will precipitate bld shrtages. Chen and clleagues in a subsequent publicatin als prvide a frmula fr calculating ptimal target inventry levels by type, but they emphasize that their system bviates the need t select acceptable shrtage rates. 5 Speaking in purely practical terms, many bld bank directrs and technical staff are reluctant t establish

6 Vl. 78 N. 2 BLD RDERING STRATEGY 159 ptimal target inventry levels fr their bld banks n the basis f a published mathematical frmula fr fear that sme significant variable which is particularly applicable t their wn hspital and bld supply may nt have been factred int the equatin. Fr example, their bld bank may be gegraphically islated frm the reginal supplier, r the reginal supplier may be undependable during certain perids f the year, r their hspital may be a referral center fr trauma cases, frcing the bld bank t maintain an apparently largerthan-necessary bld inventry as a buffer against shrtages. The target inventry level frmulas published t date, therefre, are used by bld bank wrkers mre cmmnly as a means t validate their pre-existing target levels established n an empirical basis than t establish reasnable inventry targets a priri within a hspital. This results in a tendency t justify discrepancies between calculated and bserved values n the basis f lcal circumstances rather than faulty inventry cntrl technics. The data presented in this study shwing the effects f varius inventry reductins n bld bank perfrmance measures underscre the extent t which the grup bld inventry serves as a buffer against shrtages, and suggest a strategy fr reducing bld inventry levels, if necessary, while simultaneusly minimizing the risk f shrtages. Simply put, the strategy is t reduce the nn-grup- bld inventry initially by 10 per cent frm the preexisting baseline level, with the grup inventry maintained initially at the baseline level. After a trial f perhaps ne mnth at the new and lwer level fr nn-grup- bld and if n shrtages ensue, the nn-grup- inventry can be reduced by anther 10 per cent and the grup inventry can then be first reduced by 10 per cent. These 10 per cent decrements in inventry, with grup reductins lagging behind nn-grup- reductins, can be cntinued until the ptimal bld inventry target levels are achieved. This simple apprach t inventry reductin and the derivatin f ptimal target inventry levels has a distinct advantage ver the mathematical calculatin f target levels because it is based n gradualism; the prcess prceeds slwly and safely dwnward frm the bld bank's empirically derived, but excessive, baseline inventry levels. The grup inventry is partially prtected by the ne mnth lag perid shuld bld shrtages result frm the reductins and spradic bld switching becme necessary. Since the need fr reduced target inventry levels in a hspital bld bank is generally recgnized n the basis f excessive utdating f red bld cells and whle bld, the ptimal inventry target levels are reached when the utdate rate fr these prducts drps t an acceptable percentage. Althugh there is n cnsensus at the present time as t what cnstitutes an acceptable utdate rate fr "red units" (red bld cells plus whle bld), current practice dictates that it prbably shuld nt be greater than abut five per cent fr mst hspital bld banks, and prbably less. It must be kept in mind if a declining utdate rate is used as a measure f the success f an inventry reductin prgram that utdating can be mst efficiently reduced by inventry reductins nly when the crssmatch t transfusin rati (C/T rati) and the crssmatch reserve time are als at acceptable levels. Since it has been suggested abve that the prgress in revamping an inventry cntrl prgram shuld be measured in terms f a reductin in the red unit utdate rate, it must als be emphasized that there are parameters ther than the utdate rate which can be used t gauge the quality f such a prgram. In fact, it has been shwn that the utdate rate in the mdeled bld bank inventry presented in the current study drpped t belw ne per cent after the intrductin f bld with a 35-day shelf-life. Why, then, shuld standard inventry cntrl measures nt be abandned as the shelf-life f bld is extended? Why pursue an active curse if the inventry cntrl benefits can be derived passively? First f all, as demnstrated by the cmputer simulatin, bld bank perfrmance measures ther than the utdate rate such as the unit refusal rate will cntinue t be a prblem even after the intrductin f the units with a lnger shelf-life. Secndly, inventry cntrl technics such as the reductin f unnecessary crssmatching serve t reduce labratry fees billed t patients, a suitable gal even if a negligible utdate rate has already been achieved. Lastly, the baseline utdate rate in Table 1 under a 21-day shelf-life was relatively lw at the utset, s utdating became negligible after the intrductin f the lnger shelf-life. Hspital bld banks with relatively high utdate rates prir t the intrductin f CPD-adenine with a 35-day shelf-life may be unable t reduce their utdate rate t less than ne per cent n a passive basis simply by dint f intrducing a lnger shelf-life. Chen and clleagues have emphasized that inventry management cntrls must be maintained in rder t derive maximum benefit frm the 35-day shelf-life. 5 T cnclude and in summary, we can nw prceed t a cncise descriptin f a step-by-step apprach fr analyzing a hspital bld inventry cntrl prgram, accmpanied by an actin plan fr implementing change which incrprates the bld rdering strategy detailed in this reprt. Adherence t this plan shuld result in a lw utdate rate, a reductin in unnecessary

7 1 60 FRIEDMAN ET AL. A.J.C.P. August 1982 crssmatching, and greater availability f bld fr thse patients with a legitimate need fr it. (1) Calculate the C/T rati fr all patients discharged frm the hspital and fr perated versus nnperated patients cnsidered separately in rder t determine whether excessive crssmatching is present in the hspital. If excessive crssmatching is present, it may be useful and necessary t pinpint the exact lcus' f excessive bld rdering by calculating the C/T rati fr individual peratins, diagnsis grups, r physicians admitting patients t the hspital. Excessive crssmatching can be reduced r eliminated rather quickly and efficiently with a maximum surgical bld rder schedule (MSBS) r a type and screen prgram. 2 ' 7,8,11 (2) Next examine the crssmatch reserve time. The majrity f crssmatched untransfused units remaining after cmpletin f a surgical prcedure shuld be rapidly cnverted t unassigned inventry. If this is nt ccurring, hspital plicies shuld be develped which ensure the rapid return f crssmatched untransfused units frm the perating rm r prvide the authrity t dereserve crssmatched units held in the bld bank fr surgical patients. With regard t bedside transfusin, a maximum f tw units shuld generally be held in reserve fr patients wh are hemrrhaging acutely. This maximum can be increased if the patient is bleeding s rapidly that a reserve cannt be maintained. Fr nnsurgical patients wh are nt actively bleeding, crssmatch nly the number f units that the physician intends t transfuse immediately. Bld can be supplied rather quickly using the standard crssmatch r emergency technics such as the immediate spin crssmatch, 12 shuld bld transfusin be required n an emergency basis in the perating rm r at the bedside. (3) After reducing excessive crssmatching and the time that units remain in a reserved status, the utdate rate fr the hspital shuld fall t an acceptable level. If the utdate rate des nt fall t such a level, the bld bank is prbably maintaining an excessive number f units f bld in the running bld inventry and there is a need fr lwer target inventry levels. Inventry levels shuld be reduced by 10 per cent decrements in nn-grup- and grup units with a lag in the grup reductins t frestall serius shrtages during the re-equilibratin perid. The red unit utdate rate will eventually fall t an acceptable level. (4) Having cmpleted these three steps, the bld bank inventry cntrl prgram shuld be functining at an ptimal level and thse hspital plicies established t achieve this end shuld be maintained, including the newly revised target inventry levels. Peridic changes intrduced int the system such as a new patient mix r the additin f new physicians t the medical staff may necessitate ccasinal fine-tuning f the system. References 1. Abbtt RD, Friedman BA, Williams GW: Recycling lder bld by integratin int the inventry f a single large hspital bld bank. Transfusin 1978; 18: Bral L, Henry JB: The type and screen: a safe alternative and supplement in selected surgical prcedures. Transfusin 1977; 17: Brdheim E, Hirsch R, Prastacs G: Setting inventry levels fr hspital bld banks. Transfusin 1976; 16: Chen MA, Pierskalla WP: Target inventry levels fr a hspital bld bank r a decentralized reginal bld banking system. Transfusin 1979; 19: Chen MA, Pierskalla WP, Sasetti RJ, Cnsl J: An verview f a hierarchy f planning mdels fr reginal bld bank management. Transfusin 1979; 19: Chen MA, Pierskalla WP, Sasetti RJ: The impact n adenine n bld inventry management (abstract). Transfusin 1980; 20: Friedman BA, berman HA, Chadwick AR, Kingdn K.I: The maximum surgical bld rder schedule and surgical bld use in the United States. Transfusin 1976; 16: Friedman BA: An analysis f surgical bld use in the United States hspitals with applicatin t the maximum surgical bld rder schedule. Transfusin 1979; 19: Graf Z, Katz A, Mrse E: Bld inventry distributin by type in a ttal bld supply. Transfusin 1972; 12: Jennings JB: An analysis f hspital bld bank whle bld inventry cntrl plicies. Transfusin 1968; 8: Mintz PD, Nrdine RB, Henry JB, Webb WR: Expected hemtherapy in elective surgery. NY State J Med 1976; 76: berman HA, Barnes BA, Friedman BA: The risk f abbreviating the majr crssmatch in urgent r massive transfusin. Transfusin 1978; 18: Taswell HF, Smith JR, Fss ML: The impact f CPDA-1 n the lss f bld due t utdating (abstract). Transfusin 1980; 20:616