University of South CarolinaScholar Commons

Faculty Publications Management Department

Spring 1979

A Methodology for Nurse StaffingBruce M. MeglinoUniversity of South Carolina - Columbia, meglino@moore.sc.edu

Follow this and additional works at: https://scholarcommons.sc.edu/man_facpub

Part of the Management Sciences and Quantitative Methods Commons

This Article is brought to you by the Management Department at Scholar Commons. It has been accepted for inclusion in Faculty Publications by anauthorized administrator of Scholar Commons. For more information, please contact dillarda@mailbox.sc.edu.

Publication InfoCalifornia Management Review, Volume 21, Issue 3, Spring 1979, pages 82-93.http://cmr.berkeley.edu/© 1979 The Regents of the University of California

Bruce M. Meglino

A Methodology forNurse Staffing

It is obvious to even the casual observer thathospital costs have increased at an alarming ratein recent years. Most significant in this increasehas been the cost of hospital personnel. Person-nel payroll is responsible for over 60 percent ofthe total resources expended for health services.'Furthermore, it appears that these costs aregrowing at an increasing rate. Abernathy and co-authors state that "payroll expenses per em-ployee increased an average of 5.4 percent peryear during the period 1961 through 1968, and10.1 percent during 1968 alone in nonfederal,short-term hospitals." This increase in employeesalaries does not appear to be spurring a decreasein other expenses as "expenses per patient dayincreased on an average of 8.4 percent per yearduring 1961 through 1968. and were up 13.5percent in 1968."^ More recent figures indicatethat this trend is continuing, with a net increaseof 10.3 percent per year in earnings per em-ployee and a net increase of 16.0 percent peryear in total expenses for nongovernmental,nonprofit community hospitals for the years1969 to 1971.-'

ln addition to increases in employee salaries,there have also been significant increases in thenumber of personnel per 100 census. These in-creases have amounted to 2.9 percent per year

for nonprofit hospitals and 3.8 percent for pri-vate nongovernmental community hospitals forthe years 1969 through 1971.'*

One key to the reduction of personnel costs liesin the nursing area. Nursing staff salaries repre-sent the single most costly component of hos-pital operations, yet many hospitals employnurses inefficiently.^ In a study of nursing utili-zation a researcher discovered an average of 19percent "standby" time on two hospital units.^Another study examined forty-one nursing unitsduring a twenty-four hour period and found anoverall utilization rate of 78 percent.'' In addi-tion to obvious inefficiency, there is evidence tosuggest that additional problems arise as a resultof overstaffing. One nursing study discoveredthat overstaffing "increased boredom and rest-lessness, and resulted in a much more tiredfeeling."**

It seems clear that a systematic approach to theproblem of nurse staffing can result in signifi-cant decreases in hospital costs. This article de-scribes a protocol aimed at more efficient use ofnursing personnel and a method for a more ac-curate determination of patient load.

Present Staffing Practices

Much ol* the inefficiency involved in the alloca-

82 California Management Review

tion of personnel stems trom the fact that hos-pitals make a number of common errors withregard to staffing. The first error occurs at thetime forecasts are prepared. Forecasts are fre-quently prepared for the long term, and thus,"hospitals are unable to detect and respond toshort term fluctuations. As a result, actual occu-pancy may run well below forecast for severalmonths without adjustments being made."^ Insuch situations, individuals are added to the staffin anticipation of an increase in patient load.Patient load, however, does not automaticallyincrease at the time budgets are approved;rather, it increases gradually over the budgetperiod. This results in overstaffing. Anotherproblem occurs when, "in some instances thelevel of aggregation of forecast is too great for itto be particularly useful. A forecast of the over-all utilization rate is olten made for the entirehospital."'" In this case, while the overall fore-cast may indicate that patient load is increasing.some of this increase may occur within unitsthat can easily absorb additional loads withexisting personnel. Therefore, forecasting atsuch a level precludes the assignment of new per-sonnel to areas where they are truly needed.

Closely coupled with high aggregation is the factthat many hospitals forecast census rather thanpatient load. Abernathy and associates state,"We know of no attempt by hospitals to fore-cast patient load directly: there is almost single-minded attention to forecasts of census.""These census figures are usually multiplied bysome constant, such as 5 hours, 4 minutes arcused in one study,'^ or 3.2 hours, as used bymany hospitals.'^ The resulting number is thenused as a forecast of patient load. This estimatecan be very inaccurate, since investigators haveshown that load is not correlated with census.This appears to be a common problem not onlyin the construction of forecasts but also in theassignment of personnel.

Bruce M. Meglino is Associate Professor of Managementat the University of South Carolina. His research interestsare in the areas of human behavior and organization de-sign. He has recently written articles on the impact oforganizational and psychological stress on employees'attitudes and performance.

The final error which many hospitals make isdemonstrated by the fact that personnel arecommonly budgeted at a level sufficient to sat-isfy peak demand. Demand, however, may Huc-tuate on both a daily and a seasonal basis. Inmany cases hospital units show predictablemonthly variations in patient load. Abernathyand associates point to the fact that budgetedpersonnel, "for a given month do not necessarilyrelate to that month's forecast, but to the fore-cast for the month in which peak demand isexpected to occur."''' With regard lo daily Huc-tuations Connor shows that "the nursing careload must be expected . . . to have a large varia-tion. The implication to the hospital administra-tors is that a large variation in the basic nursingtask must be expected, and some action shouldbe taken to meet this variation."'^ Many hos-pitals have no formal provision for accommo-dating Huctuations in patient load either on adaily or seasonal basis. This creates uncertaintyand, "faced with uncertainty as to what thespecific staff requirements of a nursing unit willbe on a given day, the staffing planner tends toanticipate the worst situation and staff for peaklevels."'^ In situations like this, "the conse-quences of failing to provide adequate patientcare during periods of peak demand loom farlarger in the supervisor's mind than the conse-quence of overstaffing.'"''

The protocol developed in this paper is aimed ateliminating the problems stated above. In thecase of forecasting, it will advocate forecastingon a hospital unit basis. In this way, forecastedload can be compared with present staffing onthe unit itself and an honest determination maderegarding increases in staff. In addition it willlook at weekly patient load figures for the pre-vious year. This will enable hospital managementto gain insight into the seasonal variation of loadon each unit. When seasonal variation is noted,methods can be devised to more equitably allo-cate personnel during such periods.

In the case of daily variation in patient load,management should think in terms of methodsthe hospital can use to adjust personnelcapacity. These methods include, but are notlimited to. the use of part-time employees andthe establishment of controlled variable staffingor tloat pools. With respect to part-time em-

SPRING / 1979 / VOL. XXI / NO. 3 83

ployees, a survey of nursing directors revealedthat "part-time nurses were reliable help anddedicated to nursing; that broader programs touse part-time nurses were needed; that manyhospitals could not be adequately staffed with-out part-time nurses, whose value outbalancedany extra administrative work they caused; andthat use of part-time nurses was not unfair tofull-time nurses."'^ While the use of part-timeemployees depends to a large extent upon theindividual characteristics of the hospital as wellas the community, the value of such a programis difficult to ignore.

The technique of controlled variable staffing in-volves a reduction in the number of full-timeemployees on a hospital unit to an amount re-quired for minimum loads. The day-to-day fluc-tuations above minimum load are then compen-sated using a pool of cross-trained float nurses.'^Economies in staffing are realized from such anarrangement because, as Connor observed, "it isindicated that the patient care loads are indepen-dent of one another, and therefore the averagedemand of four floors has a smaller variationthan it has for a single floor."^o Therefore, theuse of a float pool can substantially reduce theamount of overstaffing and thereby reducecosts. Although the use of such a system hasbeen shown to increase both morale and produc-tivity, several objections have been raised aboutits use.^' Abernathy and co-authors summarizesome of the problems:

Float nurses require more training and orientation. Theyare often uncomfortable in their jobs because they mayforego the opportunity to achieve continuity in provid-ing care to individuai patients or because they are un-familiar with the staff members with whom they work.Nursing supervisors sometimes feel that float and part-time help are inefficient. These objections, as well ascertain institutional and legal constraints, have tended tolimit the use of float, overtime, and part-time nurses inpractice.

However, they also go on to say:It is quite possible, however, that these problems occurbecause the variable staffing procedures are not imple-mented properly. A more careful selection of float orpart-time nurses, pay scales commensurate with the in-creased responsibilities of such nurses, and demonstra-tion of the benefits to all nurses in the hospital mayalleviate many of the objections. For float nurses them-selves, one might hypothesize that with careful attentionto the issues, they would be challenged by the greaterdiversification in their jobs. Also, the training of student

nurses in hospitals using controlled variable staffingwould occur under more stable conditions than in hos-pitals where the work load per nurse is more volatile.The important point, however, is that the benefits ofvarious short-term scheduling possibilities must be con-sidered separately from their collective disadvantages. Ifthe benefits are sufficiently attractive, then the adminis-trator can be assured that time consumed in search forcreative ways to overcome the difficulties will be wellspent."

Finally, in a recent survey of administrators ofvoluntary and municipal hospitals in New YorkCity, a researcher discovered that the administra-tors believed in fiexible assignment of personnel.They felt, however, that this was hindered byprofessionals confining their activities to profes-sional functions.^^

While the value of flexible assignment of person-nel (controlled variable staffing) has been estab-lished in many cases, it is recognized that allhospitals, for reasons mentioned above, may notwish to participate fully in such a program. Forthis reason a mechanism will be built into thefollowing protocol to allow for different degreesof participation.

General Protocol

The items listed below refer to steps in the pro-tocol shown in Figure 1. In hospitals where afloat pool of personnel exists, it will be assumedthat the present number of pool individuals isadequate for the daily functioning of the hos-pital; having been determined historically bymeans of trial and error. While references will bemade to patient load, a discussion of how thisload is determined will be postponed until theprotocol has been fully explained.

Step 1. As discussed previously, many hospitalsforecast personnel requirements at a high levelof aggregation and with little consideration ofseasonal variations. This results in a forecastwhich is not able to predict the number of per-sonnel required on each hospital unit. This stepas well as the next is specifically aimed at thisproblem. At this step, weekly load figures arecollected for the particular unit for the previousfifty-two weeks. These load figures are expressedin terms of the number of personnel hourswhich were required to meet the patient load onthat particular unit for the specific week in ques-tion. Again, a discussion of how this figure can

84 California Management Review

Figure 1. General Protocol

(1)Determineplanning periodfor each hospitalunit

(2)Hxamine repre-sentative periodwithin planningperiod

(3)Determine MPSLfor each shiftfor the planningperiod

(5)is

CSC > MPSLby amount ^ areadily assign-able person

,s X No

MPSL = CSC

Nochangesneeded Consider staff

reduction ortransfer forplanning period

(7)can

less than afull-time person

be assigned tounit

(6)is

MPSL>CSCby amount

hours

Add tull-timeperson to unitfor planningperiod

Add less than afull-time personto unit forplanning period

EitherReduce staff by onefull time person andadd to float pool forplanning period.*

OrReduce float pool andallow staff to remainat present level forplanning period.*

Add to floatpool forplanningperiod.*

*Note: Adjustments to float pool aremade by an amount equal toFPR(MPSL-CSC).

SPRING / 1979 / VOL. XXI / NO. 3 85

Table 1. Patient Load (Historical)

Su MDaily Hours(Patient Load)

1U5

Tu W Th

112 103 98

/ •

Days

Sii Su

90 85

A/

97

Tu

100w115

Th I-

122 115

Sa

M4

Table 2. Patient Load (Projected)

Days

Daily Hours(Patient Load)

Su

124M

!36

Tu

146 134

Th127

F

146

Sa

1 17

be arrived at must be postponed until later inthis article. For now, consider an array of fifty-two figures eacli ol which represents the amountof hours required to adequately care lor theactual census of patients for the previous fifty-two weeks on the specific hospital unit. Thesefigures are then examined in an attempt todetermine if significant seasonal shifts in patientload (amount of hours required) occurred duringthe previous year. If no significant shifts arenoted then the remaining steps in the protocolneed only be completed once for the entire year.In this case, the planning period is determined tobe one year. If significant shifts in patient loaddo. in fact, exist during the year then the re-maining steps in the protocol must be repeatedfor each period where a significant differencefrom another period is observed. For example,assume that a unit required approximately 700hours per week to care for patients for themontlis of January through June. During theperiod of July through September, however, thisload dropped to approximately 600 hours.Finally, the months of October through Decem-ber show a return of the required hours to 700per week. In such a case there are three planningperiods, the first consisting of six months dura-tion and the remaining two consisting of threemonths each. Tlie steps of the protocol, how-ever, need only be carried out twice since thereis no significant difference in patient load forthe months of January through June and themonths of October through December.

The determination of what constitutes a signifi-cant shift in load is somewhat difficult tospecify. Among other things, it will dependupon the actual amount o\ the load involved,the duration of that load, and the administrativepolicies of the particular hospital. A shift of 20

Su M Tu W Th F Sa

111 126 130 ISO 159 150 122

hours per week in the patient load for a periodof three weeks may be considered insignificantbecause it involves the relocation of one-half ofa luli-time equivalent employee (FTE) tor a per-iod of only three weeks. A redistribution of thissort may be more readily accomplished throughthe normal use of a float pool without having toresort to a permenent shift in personnel. On theother hand, a shift of 40 hours per week in thepatient load for a period of three monthsprobably constitutes a significant shift. Thiswould involve the relocation of an F.T.E. for aperiod of three months. Such a determination isbasically a policy decision to be made by hos-pital management. It should be kept in mind,however, that the degree to which such shifts areaccommodated will determine the degree towhich personnel are adequately utilized.

Step 2. This step involves projecting the unitdaily patient load hours for each shift fora rep-resentative period during the planning period.While this step may appear to be quite involved,in practice the procedure is relatively simple.First, a representative period for each hospitalunit of approximately two weeks is chosen with-in the planning period. Since it must representthe entire period it should be chosen carefully sothat the peaks and valleys of the planning periodare adequately represented. Once this is chosenthe unit's patient load in hours is computedfrom historical figures for each day during thatperiod. Having determined these daily load fig-ures for the particular unit, they must then beadjusted by the projected change in census (ifany) for the next year's planning period. Forexample, assume that the figures shown in Table1 represent daily patient load figures for a hos-pital unit for a representative two weeks duringthe January through June 1979 planning period.

86 California Management Review

Table .1. Distribution of All Nursing Personnel

Day Shift

45%

Evening Shift

36.6%

Night Shift

18.47()

Let us further assume that the hospital antici-pates a 30 percent increase in census on thatunit for the same January through June planningperiod in 1980. In that case the projected figureswould be as shown in Table 2. These figures aresimply those in Table 1 increased by 30 percent.

The figures for each day shown in Table 2 repre-sent the patient load for a twenty-four-hour per-iod composed of three shifts. Since it is neces-sary to examine the staffing pattern on eachshift, these figures must be broken down intothe number of hours required on each shift.Determining the amount of daily hours whichare attributed to each shift may be done in twoways. The hospital itself may have data whichindicate the percentage of total hours expendedon each shift. If these percentages are availablethey may be used provided they were not ob-tained by examining previous staffing patterns.This would amount to a case of perpetuatingpercentages which may be inaccurate to beginwith. To be accurate the percentages should beobtained through the use of direct workmeasurement. If these data are not available afirst approximation may be made using the per-centages developed by the Commission for Ad-ministrative Services in Hospitals. They excludeNursery, Labor, and Delivery, and Intensive CareUnits. These percentages, shown in Table 3, rep-resent the recommended distribution of nursingpersonnel including charge nurses, nurses aides,orderlies and ward clerks.^'^ Using the previousadjusted daily figures in Table 2 it is now pos-sible to compute the number of hours requiredon each shift. In the case of the Nursery. Labor,and Delivery, and Intensive Care Units, an at-tempt should be made to obtain shift distribu-tion percentages from direct observation. Theseunits are unique in that there is an almost evendistribution o\' personnel across shifts. The pro-jected patient load for the day shift appears inTable 4.

Step 3- This point in the protocol calls for apolicy decision about the use of a float pool ofemployees to adjust for lluctuations in load. Inorder to illustrate this point it is best to graph

the data in Table 4, although this step need notbe taken in the actual protocol.

The extent to which a hospital wishes to utilizea tloat pool is directly related to the determina-tion of the minimum permanent staff level(MPSL) of a unit during a shift. If the number ofstaff assigned to the unit were ten, this wouldprovide an MPSL capacity of 80 hours duringthe day shift. In such a case, even if one em-ployee on the unit was absent, thus providing acapacity of 72 hours, the worst anticipated casewould have been met and a Hoat pool would beunnecessary, ln this case, however, assuming noabsence occurs, the amount of overstaffing (un-necessary personnel each day) would total 27man-days within the two-week period. If theMPSL was reduced to nine persons this woulddecrease the amount of overstaffing to 13 man-days. This case would either require the use of ailoat pool to take the place of individuals thatare absent or would leave the unit understaffedwhen a high toad and an absence occurred at thesame time. In effect, the administrator is tradingconvenience and continuity of care factors for asubstantial increase in staffing efficiency. Fin-ally, if the MPSL level is set at seven persons theamount of overstaffing would decrease to a max-imum of 3 man-days. This would have to resultin an increase in the float pool capacity over thelevel needed for an MPSL level of nine persons;however the tloat pool increase would be muchless than the difference in personnel (twopeople) required at each level. This occurs be-cause, as described earlier, the average patientload across many units has a smaller variationthan it has for a single unit.

A final point which should be made is that forillustration purposes the MPSLs in the previousexample were set at increments of 8 hours. Thisneed not be the case. It is possible for the MPSLto be set at any level. This could be arrangedthrough the use of part-time employees, overlap-ping shifts, permanent assignment of an indivi-dual to more than one unit, or permanent assign-ment of a IToat pool person for a portion of ashift. For this reason, a policy could be estab-lished on how the MPSL level is to be set. Forexample, one might decide to set the MPSL levelat 90 pffcent of the mean of projected patientload hours per shift. In the case of Table 4 the

SPRING / 1979 / VOL. XXI / NO. 3 87

Table 4. Patient Load (Projected), Day Shift

Su MDaily Hours 5, ^,(Patient Load)

Tu W Th

51

/ • •

66

Days

Sa

53

Su

50

M

57

Tu

59 68

Th

72

/ • •

68 55

mean is equal to 61; therefore the MPSL levelwould be set at (61 x 0.90) or 55. It is alsopossible to set the MPSL level in relation to themean and variance of the projected patient loadhours per shift. Regardless of how the level isestablished, the remaining steps of the protocolare then followed for each shift.

Step 4. At this point the already establishedMPSL for each shift is compared to the currentstaff capacity (CSC) or the present capacity ofthe personnel assigned to that particular shift.

This comparison attempts to determine if thecurrent staff capacity is above or below thestaffing policy for the shift as expressed in theMPSL. If CSC exceeds MPSL then the protocolproceeds to Step 5. If MPSL is greater than CSCthen it proceeds to Step 6. If both are equalthen no further steps need be taken since it wasassumed at the beginning of the protocol thatthe float pool was adequate for the daily func-tioning of the hospital.

Step 5. Here the degree to which CSC exceedsMPSL is examined. If CSC exceeds MPSL by anamount approximately equal to or greater than areadily assignable person, then strong considera-tion should be given to a reduction or transfer ofpersonnel for the planning period. A readilyassignable person normally means a full timeemployee assigned to the particular shift on aregular basis. It can also mean any individualwho is normally assigned to the particular shiftfor less than 8 hours. Such situations wouldoccur with part-time employees, those withmultiple shift or unit assignments or individualswho are permanently assigned for only a portionof a shift. In effect, this means that cases mayarise which warrant a reduction or transfer ofpersonnel where CSC does not exceed MPSL byan amount equal to at least 8 hours.

Where CSC exceeds MPSL by less than 8 hoursand there are no readily assignable persons withwork hours approximately equal to the amountof the reduction desired then a decision must bemade. Either the staff must be reduced by one

full-time person with a corresponding increase inthe float pool or the float pool must be appro-priately reduced and the staff allowed to remainat its present level. As with previous decisions,this relates to the staffing philosophy of the par-ticular hospital.

Step 6. This step determines the extent to whichMPSL exceeds CSC. If the excess is greater thanor equal to 8 hours, a full-time person should beadded to the unit for the planning period. If theexcess is less than 8 hours, however, the proto-col proceeds to Step 7.

Step 7. This last step in the protocol considersthe possible permanent assignment of less thanfull time personnel for the planning period. Asmentioned earlier, this is possible in the case ofpart-time employees, overlapping shift assign-ments, or the permanent assignment of an indivi-dual to a unit for only a portion of the day. Ifthis cannot be readily done, an addition shouldbe made to the float pool.

Float Pool

Implementation of the protocol described abovewill yield the number of permanent employeesto be assigned to each shift of a particular hospi-tal unit as well as information on the adjust-ments to be made to the float pool. However, asnoted earlier, additions or decreases to the floatpool should not be made at an amount equal tothe difference between MPSL and CSC. Thefloat pool, because of variation in unit load fluc-tuations, makes more efficient use of personnel.Therefore, these adjustments should be made ata rate far less than the difference between MPSLand CSC. This rate can be readily estimated withavailable data. It is assumed that the hospital hassome historical data on the appropriate size ofthe float pool. The hospital also has historicaldata on patient load lor the representative peri-ods as calculated in Step 2 of the protocol.Using both of these data it is possible to calcu-late the float pool ratio (FPR). This figure canthen be used as a guide for determining both the

California Management Review

size and the rate at which adjustments are madeto the float pool.

Since historical load figures for a representativeperiod were calculated in Step 2 of the protocoland since the current staff capacity (CSC) is alsoknown, it is possible to determine, for all units,the number of hours during the representativeperiod that load exceeded CSC. Historically, thisis a measure of the work load which was ab-sorbed by the float pool. This need only be donefor one particular shift (such as day shift) forthe entire period. From these data it is possibleto calculate the FPR using the following form-ula:

Number of hours in float poolduring the representative period

Float pool =ratio (FPR) Number of hours load exceeded

CSC for all units during therepresentative period

The number of hours in the float pool for therepresentative period is determined by obtaininga cumulative total of the number of people inthe noat pool for each day of the representativeperiod and multiplying that number by 8 hours.Again, this number need only be calculated forthe same daily shift during the entire period;however, it is essential that the shift used be thesame as the shift used to calculate the denomi-nator of the FPR equation.

The FPR calculated above should be a fractionand will represent that portion of understaffedhours (total amount of hours that load fallsabove permanent staff) which must be containedin the float pool. For example, assume thatFPR = 1/3. If the hospital anticipates that allunits on a particular shift (such as day shift) willbe understaffed by an average of 3000 hoursduring any given week then it must provide1000 hours in its day shift float pool each week.In addition, when adjustments to the float poolare indicated in the protocol, the amount of theadjustment should be as follows:

Adjustment = FPR (MPSL - CSC)The methods described above for establishingthe size of the float pool are intended only asgeneral guides. It is anticipated that as the hospi-tal gains experience with making adjustments tothe tloat pool it will arrive at a more accurateFPR.

Determination of Patient Load

As described earlier, many hospitals attempt to

determine patient load directly from censusthrough the use of a single constant. The impre-cision of such a system was recognized by Con-nor who noted that such methods of staffingresulted in wide variations in the degree ofpatient care. For this reason, another system wasdeveloped which attempted to classify patientsinto categories so that a smaller variation in eachgroup would result. The product was a classifica-tion system which divided patients into threegroups: Class I. Class II, and Class HI. Theseclasses have come to be known as minimal care,intermediate care, and complete care, respec-tively. The first step in obtaining an accurateestimate of patient load is to divide the censuson each hospital unit into these three categories.This can be accomplished using the followingrules.

Class I. Any of the following combinations:a. Ambulatory, or up in chair—self; Feeding self, or re-quires food cut; Bathing in bathroom, or at bedside-partial self.b. Ambulatory- with assistance; Up in chair-self; Bath-ing in bathroom, or at bedside—partial self.c. As in a and b, with Vision inadequate; oxygen ther-apy; Intravenous feeding; but no two of these factorssimultaneously.

Class II. Any of the following combinations:

a. Ambulatory-with assistance; Bathing in bathroom,or at bedside-partial self feeding-complete assistance(except I.V. feeding); Vision inadequate (optional); Oxy-gen therapy (optional).b. Up in chair—self; Bathing at bedside—complete assis-tance; Feeding self, or requires food cut or I.V. feeding;Oxygen therapy (oiJional); Vision inadequate(optional).c. As in b, with the following changes: Up in chair-withassistance; Bath at bedside.d. Up in chair-with assistance; Bath at bedside—partialself; Feeding-complete assistance; Vision inadequate(optional); Oxygen therapy (optional).e. Getting special care of necessity, 'Note: Any patient who otherwise falls into categories Ior 11, but who is in isolation or is incontinent or mark-edly emotionally disturbed will be dropped to the nextcategory.

Class III. All combinations not previously men-tioned.^^

When personnel are familiar with this scheme,the classification of patients can be accomp-lished relatively quickly. Connor's experienceindicates that patients on a twenty-nine-bed unit

SPRING / 1979 / VOL. XXI / NO. 3 89

Table S. Minutes of Direct Patient Care Per Shift

Class I Class II Class HI

Mean (Connor) 28.3 58.3 139-6

Mean (Moon) 38 53 96

Table 6. Extension Factors for Specialty Units

Figure 2. Average Nursing Hours/Patient/Patient Day byType High. Low, and Mean Values

Hospital Unit

PediatricsICU/CCLIPostpartumPsychiatricMain NurseryIntensive Care NurseryDelivery Room

Extension Faclor1.333.00

1.50.88

1,502.00

can be categorized in about two minutes each.As individuals gain an intuitive feel for eachcategory it is likely that this can be accomp-lished even faster without reference to the itemsin the previous list.

It is clear that Class I, II. and III patients shouldeach require different amounts of care. This factlias been statistically verified by many re-searchers, among them Connor and Moon. Bothresearchers measured the amount of direct pa-tient care in minutes for each category of pa-tient. Tlieir conclusions appear in Table 5. ^̂ '̂ ^While there is some variation between the sets offigures, it appears that each category does dis-criminate well along the dimension of direct pa-tient care.

Althougli the data in Table 5 illustrate the dif-ferences in direct care for each class of patient,in actual practice hospital unit personnel haveother responsibilities. To obtain an accuratemeasure of patient load, standards must bedetermined whicii reflect these additional activ-ities. Obviously, the ideal situation would be foreach hospital to determine its own load stan-dards for each class of patient by direct observa-tion. If tiiis is not feasible, it is possible to esti-mate patient load using the figures presentedbelow. This estimate can then be revised throughactual experience. Successive refinements ofthese standards should yield accurate values forpatient load.

The standards shown in Figure 2 represent anextensive study of a variety of hospitals.^^ Thefigures were determined using engineering stan-dards for direct care items while making allow-ances for indirect care based upon actual time

Day

r Pa

tien

in

Hou

8.0 •

7.0 •

6.0 •

5,0 -

4.0 -

3.0

2.0 -

1.0 -

3.51

2.66

1 1.75

3

~ —

.00

47

41

4

3

001

97

06

4r~

3

-I

98\

62

54

MinimalCare

Inter-mediate

CarePatient

CompleteCare

Type

AveragePatient

taken. Since the standards shown represent arange of values, it is necessary for hospital per-sonnel to estimate the level of patient care of-fered by their particular institution. If, forexample, the level of care is determined to beaverage, the formula for calculating patient loadwould be:Load = (# of Class I patients) {2.66 hours) + t # o f Class

II patients) (3.47 hours) + ( # of Class III pa-tients) (4.97 hours).

In the case of liistorical data, it may be impos-sible or difficult to discover the census of ClassI, II, and III patients on each unit at a specifictime. In this situation it is possible to liave nurs-ing personnel estimate the percent of Class I, II,and III patients that are nounally present on theunit. In this situation the fonnula for load be-comes:Load = l(% of Class I patients) (2.66 hours) + (% of

Class II patients) (3.47 hours) + (% of Class 111patients) (4.97 hours)] X Census.

This second equation is less accurate than thefirst but represents a considerable improvementover previous techniques.

The measurement technique for load shownabove is suitable for both Medical and Surgicalunits. These units experience patients of differ-ent classifications. Other hospital units (sucli asIntensive Care) tend to be composed of a morehomogeneous group of patients tliat require

90 California Management Review

Table 7. Daily Staffing Personnel Guide

Medical/Surgical, tiyn. Pediatric, Post Partum, Orthopedic (fcixcludes Nursery, Labor and Delivery, and intensive Care)

Total DailyStaffing

5

6

7

RN

1

I

1

1

DaysLVN

1

I

1

Au\

I

1

11

I

RN

1

1

I

1

Lvenin^sLVN

1

1

Au\

t1

1

1

RN

1

I

I

I

NightsLVN Aid

1

1

9

Ul

II

12

1 1

I I

1 1

2 1

I 1

1 1

n14

15

16

17

18

19

20

31

22

23

24

25

26

27

28

2 1

2 1

2 1

2 1

3 2

3 2

3 2

3 2

4 2

4 2

4 2

22

2

2

2

2

2

2

2

2

3

3

3

3

3

3

11

1

1

1

1

1

1

2

2

2

2

2

2

2

2

22

2

3

3

4

4

4

4

4

5

4

4

4

5

5

1 2

2 I

2 1

2 1

Distribution of total is made by logic for low census levels and on the basis of a recommended mix of approximately 30% RNs and 20%LVNs wliert' census level allows.Note: KN requirements are inclusive of Head or Charge Nurse positions. Auxiliary requiiements are inclusive of Nurses Aides, Orderlies,and Ward Clerks.

similar levels of care. Determining the load onthese units requires a different set of standards.These standards, shown in Table 6. representrelative care loads lor specialized hospitalunits. ̂ ^

Since the types of patients on these units arevery similar, the standard for the average pa-tient, shown in Figure 2, is used. In the case ofan average level of care, the formula for patientload on these units becomes:

Load = (3.62 hours) (Extension Factor) (Census).

Mix of Personnel

Use of the protocol will indicate the number of

personnel to be assigned to each hospital unitfor each shift. The makeup of personnel on eachshitt (number of R.N.s, L.P.N.s, and auxiliarypersonnel) is another question which should beconsidered. This is an important considerationbecause an imbalance in the mix of personnelwill cause an inequitable distribution of work aswell as affect the quality of patient care,-^" Ifthere is a shortage of R.N.s. patient care willsuffer. If there is an excess of R.N.s, there willbe a reluctance to complete tasks which do notrelate to either supervision or the importantaspects of patient care. One researcher discov-ered that "when the floors were staffed with a

SPRING / 1979 / VOL. XXI / NO. 3 91

large number of persons, three-fourths of whomwere staff nurses, the *aide-type' of work stillremained pretty much with the aides."^* Thisfinding is supported by another study whichdiscovered that a staff composed of two R.N.sand one L.P.N. accomplished more than a staffcomposed of three R.N.s.^^

A general guide for the distribution or mix ofpersonnel on each shift is provided by the Com-mission for Administrative Services in Hospitals.This guide is shown in Table 7.^^ According tothe guide if the protocol indicates that thenumber of permanent staff assigned to the dayshift should be nine, then the distributionshould be three R.N.s, two L.P.N.s and fourauxiliary personnel. While the policy of the par-ticular hospital will have an important impact onthe distribution of personnel on each shift, thefigures in Table 7 can provide a valuable guidefor the hospital administrator.

The figures in Table 7 are not intended for usein Nursery, Labor, and Delivery and IntensiveCare Units. As mentioned earlier, these areunique units and the distribution of personnelshould be based upon the historical informationof the particular hospital. Also, as the table indi-cates, the figures for auxiliary personnel includenurses aides, orderlies and ward clerks. In thecase of these personnel the hospital will, again,have to decide their distribution based upon thehistorical or desired mix.

The protocol and procedures described aboveare intended as guides to hospital planners andadministrators. It is felt that a systematic ap-proach to the problem of nurse staffing will leadto increased efficiency and better patient care.While the standards presented call for successiverefinement through actual experience, they,along with the procedures outhned, represent asubstantial improvement over methods in cur-rent use by many hospitals.

Acknowledgment: This article was prepared under agrant from Prospective Payment Pilot Project, SocialSecurity Administration, SSA-PMB-74-1 75.

REFERENCES

1. K. G. Bartscht, "Hospital Staffing Methodologies-Concepts, Development, and Mszs" Journal of IndustrialEngineering (December 1967), pp. 708-717.

2. W. J. Abernathy, N. Baloff, and J. C. Hershey, "The

Nurse Staffing Problem: issues and Prospects," SloanManagement Review (Fall 1971), pp. 87-99.

3. J. Pattengill, "The Financial Position of PrivateCommunity Hospitals, \96\-l\," Social Security Bulle-tin (November 1973), pp. 3-19.

4. Ibid.

5. R. C. jelinek, T. K. Zinn and J. K. Brya, "Tell theComputer How Sick the Patients Are and lt Will TellYou How Many Nurses They Need," Modern Hospital(December 1973), pp. 81-85.

6. B. Steiner and N. Lindquist, "Surprise Finding ofStaffing Study: Nurses Have Too Much Spare Time,"Modern Hospital (February 1970), pp. 108-1 10.

7. "Nurse Staffing and Utilization Program," (Charl-otte, N.C: Carolinas Hospital Improvement Program,August 1972).

8. P. Kong-ming, G. Nite, and J. Callahan, "Too ManyNurses May Be Worse Than Too Few," Modern Hospital(October 1959), pp. 104-108.

9. Abernathy et al., op. cit.

10. Ibid.

11. Ibid.

12. M. K. Aydelotte, Nurse Staffing Methodology(Washington: U.S. Department of Health, Education andWelfare), Publication No. (NIH) 73-433.

13. M. D. Fottler, "Administrative View of ManpowerUtilization in the Hospital Industry," Personnel Journal(July 1972), pp. 505-510.

14. Abernathy et al., op. cit.

15. R. J. Connor, A Hospital Inpatient ClassificationSystem, Unpublished Doctoral Dissertation, IndustrialEngineering Department, Johns Hopkins University,Baltimore, 1960.

16. D. H. Harris, "Staffing Requirements," i/ojp//a/.i,J.A.H.A, (April 1970), pp. 64-70.

17. Abernathy et al., op. cit.

18. E.G. Stewart, "Part-Time Nurses: EmploymentConditions," American Journal of Nursing (October1971), p. 1957.

19. J. C. Hershey, W. J. Abernathy, and N. Baloff,"Comparison of Nurse Allocation Policies-A MonteCarlo Model," Decision Sciences (1974),.pp.58-72.

20. Connor, op. cit.

21. Harris, op. cit.

22. Abernathy, op. cit.

23. G. Nadler and V. Sahney, "A Descriptive Model ofNursing Care," American Journal of Nursing (February1969), pp. 336-341,

24. "Staff Utilization and Control Program Orienta-tion Report-1967," (Los Angeles: Commission forAdministrative Services in Hospitals).

25. Connor, op. cit.

92 California Management Review

26- Ibid.

27. V^.R. Moon, A Study of a Patient ClassificationSystem, Unpublished Master's Thesis, University ofMichigan, 1964.

28. "Nurse Staffing and Utilization Program," op. cit.

29. "Productivity Audit Review of the Carolinas,"(Charlotte, N.C.: Carolinas Hospital ImprovementProgram).

30. S. Miller and W. D. Bryant, "How Minimal Can"̂""̂ ^ Staffing Be?," Modern Hospital (September

1964), pp. 111-113.

31. Kong-ming, op. cit.

32. Miller, op. cit.

33. "Staff Utilization and Control Program Orienta-tion Report-1967." op.,cit.

SPRING / 1979 / VOL. XXI / NO. 3 93