schedule quality assessment metrics

Schedule QualityAssessment Metrics

Richard J. Tarpey, BBA; Millicent F. Nelson, PhD

Hospital unit staff scheduling continues to be a critical challenge directly impacting unit andfacility performance. The creation of a quality future work schedule that balances unit needs(providing adequate staff across multiple skills) versus employee needs (honoring as manyemployee requests as feasible) is the first step toward ensuring the most efficient and effectiveuse of staff resources. As an alternative to the use of expensive computer scheduling systems, thisarticle sets forth a set of easy-to-understand and easy-to-measure schedule quality metrics that canbe used via any basic spreadsheet application. The use of these metrics allows unit leaders toassess the ‘‘goodness’’ of a schedule and focus on identified opportunities toward improving theschedule resulting in a better match to the unit plan of expected volume/census. Higher qualityschedules lead to more efficient use of staff resources by ensuring that resources are fullyscheduled before reaching out to more expensive premium labor such as overtime or agency andto more effective use of resources by ensuring that schedules are as complete as possible to coverexpected volumes. Resulting decreases in last-minute staff adjustments via better planningcontribute to higher staff satisfaction and better patient care. Key words: efficiency, scheduleassessment, schedule quality, scheduling, staffing

ONE OF THE most significant challengesin managing hospital patient care units

is the pursuit of an effective balance betweenthe efficient utilization of resources and staffsatisfaction. Although there are several meth-ods toward approaching this workforce man-agement problem, the workforce schedule isone of the most critical tools available to uti-lize toward achieving this balance. The mainobjective of workforce scheduling attempts toprovide adequate staff coverage to accommo-date a predicted volume (patient load) coupledwith the attempt to provide a schedule withminimal negative impact on employees’ per-sonal lives. Poor workforce scheduling hasbeen contended to have a direct impact on unitand hospital performance through affectingquality of care, resource utilization, and staffretention.1,2 Given the impact of scheduling on

the quality of delivered care and the impor-tance of recruitment and retention of nursesthrough employee satisfaction,3,4,5 schedulequality should be a concern of every unit andfacility leader.

Although the workforce scheduling prob-lem exists in many industries, the resourceutilization and staff satisfaction balancing chal-lenge within a hospital setting is magnified,given the greater number of variables to befactored into the potential solution, such asthe need for 24-hour/7-day-a-week coverage,variability in patient volume patterns, vari-ability in shift lengths, variability in nursepreferences, and variability in degree of spe-cialization of tasks requiring differing skillsets and certifications. These variables repre-sent just a subset of items that need to beconsidered during the development of a fu-ture schedule. The additional complexity inthe health care setting has provided moti-vation for many authors to study the nursescheduling problem separate from the gen-eral workforce scheduling problem, leadingto many proposed solutions evolving into thedevelopment of several sophisticated com-puterized models to assist unit managers with

The Health Care ManagerVolume 28, Number 2, pp. 145–158Copyright # 2009 Wolters Kluwer Health |Lippincott Williams & Wilkins

Author Affiliations: Jennings A. Jones College of Business,

Middle Tennessee State University, Murfreesboro.

Corresponding author: Richard J. Tarpey, BBA, MBA

Candidate, Jennings A. Jones College of Bussiness,

Middle Tennessee State University, PO Box 450,

Murfreesboro, TN 37132 ([email protected]).

145

9Copyright @ 200 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

staff scheduling. Unfortunately, an issue withthese systems is that they are typically expen-sive to purchase and labor-intensive to setup/maintain to configure the automated sched-ule generation and optimization functionality,putting them out of reach of many resource-limited organizations in today’s health care en-vironment. In fact, many hospitals using thesesystems only use the basic functionality tocreate schedules and have not taken advantageof automated generation/optimization capa-bilities.6 This article recognizes these limita-tions and seeks to provide a methodology forevaluating the quality of a hospital staff work-force schedule across 6 basic dimensions thatcan be easily measured and evaluated for im-provement opportunities. Assuming the directimpact of schedule quality on unit and hospitalperformance, the provision of higher qualityschedules can facilitate improved performance.

Terminology

This article includes discussion on severalscheduling and staffing concepts defined asfollows:

schedule period—short-term planning pe-riod over which staff must be scheduled(typically 4 weeks);skill mix—category of staff requiring differ-ing levels of certification, skill, or experi-ence (eg, registered nurse [RN], licensedpractical nurse [LPN], nurse technician,unit secretary, etc);shift—period of time with a defined startand end time in which employees work (eg,7 AM to 7 PM, 7 AM to 3 PM, etc);slot—position in schedule with a definedstart and end time to which an employeeis assigned to work;scheduling—process of future schedulecreation for a short-term planning periodresulting in the generation of a scheduleconsisting of designated slots with em-ployee names assigned to each slot;rostering—used interchangeably withscheduling;staff commitment—level of hours per weekagreed upon by staff to work (eg, 0.90employee has agreed to work 36 hours perweek based on 40 hours� 0.90 = 36 hours);

staff flexing—periodic evaluation and adjust-ment of number and skill mix of staff to covervolume fluctuations or staff call-ins; andnegative schedule occurrence—occurrencein a schedule potentially affecting qualitythat serves to move the quality metric awayfrom zero (eg, schedule slot not filled, staffnot scheduled to commitment level, an un-healthy shift pattern detected, etc).

WHY MEASURE WORKFORCESCHEDULE QUALITY

Workforce schedules are a critical compo-nent of a unit’s strategic plan for the provi-sion of the right number and skill mix ofstaff to provide quality care. In addition,schedules provide initial information to em-ployees concerning upcoming work commit-ments and timelines that will either contributeto or subtract from satisfaction. Therefore,work schedules can be direct contributorsto potential negative unit outcomes leadingto negative effects on unit performance, in turncontributing to negative hospital outcomesand performance as discussed by Silvestroand Silvestro.2 The measurement of the qualityof workforce schedules provides 2 importantcontributions toward positive unit perfor-mance. One is the provision of visibility andaccountability to the scheduling process, andthe other is the minimization of last-minutestaff adjustments or staff flexing.

The absence of scheduling evaluation andfeedback is one of the key components thatcontribute to the proliferation of poor unitoutcomes and dysfunctional behavior, such asinequality in shift allocation, preference ofstaff interests over unit interests, or poormonitoring of use of premium labor.2 Themeasurement and monitoring of schedulequality indicators provide visibility to thescheduling process and the opportunity foraccountability to focus on unit leaders pro-ducing poor schedules period after period.These leaders can then be given mentoringand assistance to obtain schedule quality im-provement, seeking to avoid the poor unitoutcomes listed above.

146 THE HEALTH CARE MANAGER/APRIL–JUNE 2009


Daily staff adjustment or ‘‘staff flexing’’ in-volves an evaluation of the number and skillset of staff (ie, nurse vs support staff) atregular intervals throughout the day to matchthe required workload of patient care accord-ing to the number of patients in the unit andthe severity of condition (acuity) per patient.Staff flexing results in calling/floating addi-tional staff into the unit or sending/floatingstaff out of the unit and can contribute to 3possible negative consequences. First, staffsatisfaction can be negatively impacted asstaff members are requested to work un-expectedly, floated to other units, or senthome when expecting to work. Requests towork unexpectedly or requests to go homeare obvious potential sources of dissatisfac-tion. The practice of floating or requiringstaff to work in a unit other than their normalwork unit has also been identified as a dissat-isfier.7 Second, unit costs can be negativelyimpacted by increased use of incentives inthe form of premium labor dollars to enticestaff to work a nonscheduled shift or bringin contract labor to cover open shifts. Third,management time spent on staff flexingpotentially takes time away from patient careor the supervision of staff providing patientcare.

Although a certain portion of staff flexingcannot be eliminated (ie, coverage of staffcalling in sick or unexpected increases/decreases in patient volume or acuity), thereis a portion that can be addressed proactivelyvia quality planning and scheduling servingto minimize the negative consequencesdiscussed above. Reasonable volume projec-tions, coupled with complete, quality sched-ules matched to volume expectations, serveto start a shift in a better state compared withschedules with holes in coverage or over-coverage, which essentially guarantee a cer-tain level of staff flexing. It is reasonable toassume that the closer a future schedulematches an accurate plan, the less scheduleadjustments will need to be made on a last-minute basis. The portion of staff flexingattributed to poor scheduling will be subse-quently eliminated, and the remainder willbe limited to the coverage of staff not re-

porting to work and unforeseen volume/acuity adjustments. One possible way toachieve a closer match between scheduleand plan is the measurement of the variousaspects of schedule quality in a manner thatcan be used by leaders during the schedulecreation process to identify improvementopportunities that can be exploited and ad-dressed before schedule posting.

To realize these 2 contributions to unitperformance, the criteria of schedule qualityand a methodology for measuring the criteriamust exist. The objective of this article is toprovide criteria and a methodology that isadaptable to the various processes of sched-uling whether from computer systems ormanually generated but remain basic enoughto be used and understood by unit andfacility leaders for easy operationalization.Simplicity has been favored over complexgranularity to provide tools that are mean-ingful to the shift-to-shift time-constrainedleader without the use of expensive andproprietary software applications. The crite-ria are defined in basic terms with measure-ment formulas that can be used via anyspreadsheet application to measure the qual-ity of schedules against a ‘‘perfect state’’where projected needs of the unit are bal-anced with the needs of the employeesresulting in the provision of adequate re-quired staff coverage and skill mix for anaccurately predicted volume, as well as theaccommodation of all employee needs andpreferences in a fair and equitable schedule.Schedules meeting these criteria will providea better starting point for unit leaders tobegin each shift facilitating the minimizationof last-minute staff adjustments to accommo-date unbalanced schedules and provide amechanism to add accountability and feed-back to the scheduling process.

CRITERIA FOR MEASURINGSCHEDULE QUALITY

The concept of schedule quality includes3 measurable components of criteria tobe used toward defining the quality of aschedule (see Table 1). The first component

Schedule Quality Assessment 147


involves the prospective comparison of thefuture schedule to projected unit needs (esti-mated unit staffing needs based on a projectedvolume and ideally estimated accounting forshift, day-of-week, and seasonality variability).6,8

The comparison determines if the scheduleeffectively and efficiently meets the projectedneeds. The objective of the effectiveness mea-surements (total completeness, licensed com-pleteness, and nonlicensed completeness) is toassess the completeness of the schedule towardfulfilling the number of slots required to ac-commodate the projected volume of patientsfor a given shift on a given day. The objective ofthe efficiency measurement (commitments) isto assess the efficient scheduling of existing staffresources to full commitment before reachingout to more expense overtime or contract la-bor. The second component of criteria con-cerning the quality of a schedule involves theprospective comparison of the future scheduleto employee needs through measurement ofcriteria such as unhealthy shift patterns, week-end shift scheduling, and preferences honored(healthiness, weekend commitments, and pref-erences). A schedule that measures well acrossthe metrics within the first 2 components ex-hibits a level of balance between unit demandand employee needs. The third component ofcriteria involves the retrospective comparison,after the schedule period has closed, of sched-uled hours to actual hours used in the unit. Thiscomponent is important in determining if theunit plan is effectively projecting unit demandand therefore providing for the scheduling of

adequate staff levels. The creation of a needs-balanced schedule resulting in a close match toan inaccurate plan will serve no better purposethan an unbalanced schedule that does notmatch unit needs, employee needs, or both.

Several of the following proposed metricshave been developed and adapted from theprecedent work of Oldenkamp5; however, themethodology and mathematical calculationshave been adapted and simplified to providemeasurements more easily applied within aday-to-day operational setting. There are sev-eral data points required to be able tomeasure the quality of a schedule via theproposed metrics in this article as given inTable 2. Each of these data points is typicallyreadily available information for any hospitalunit and therefore allows for the application ofthe proposed metrics to any unit future sched-ule regardless of the schedule creation processin use.

Individual metric acceptance levels

Each of the following 8 quality metrics pro-vides a numerical representation of schedulequality for the respective indicator. Formulasfor each metric were derived to provide aconsistent acceptance level based on a ‘‘good,fair, and poor’’ rating system corresponding tothe score ranges given in Table 3.

Research has not revealed any standard orconsistent scoring methodology concerninghospital staff scheduling quality. Our objectiveis to provide consistent scoring that indicates a

Table 1. Components of schedule quality

Component Definition Metric

Schedule to unit needs Measures the schedule prospectiveeffectiveness and efficiencytoward meeting unit demand(provision of staff for aprojected volume)

Total completeness EffectivenessLicensed completeness EffectivenessNonlicensed completeness EffectivenessCommitments Efficiency

Schedule to employee needs Measures the schedule againstemployee needs and preferences

Weekend commitmentsSchedule healthiness

PreferencesSchedule to actual hours Measures planned hours to

actual hours to assess theaccuracy of the plan

Daily flexing



reasonable level of acceptance for negativeschedule occurrences contributing to lowerquality schedules based on the understandingof the complex criteria and contributing fac-tors affecting the ability to provide a qualityschedule. Each negative schedule occurrencecontributes to moving the individual metricscore away from zero, representing an oppor-tunity for improvement in a schedule. Becauseof the lack of precedent information, a startingpoint was determined for the acceptance levelswith further scrutiny and analysis expected. Toestablish this starting point, a level of 10% orless opportunity was determined to be good, alevel greater than 10% but less than 25% wasdetermined to be fair, and opportunity exceed-ing 25% was determined to be poor. Ourexpectation is that any opportunities represent-ing more than 10% of a schedule should bereviewed and analyzed for improvement possi-bilities before a schedule is posted.

Schedule to unit needs

There are 4 metrics measured to comparefuture schedules to unit needs (total complete-ness, licensed completeness, nonlicensed com-pleteness, and commitments). Each metric is

prospective in nature and assesses the schedulebased on its ability to effectively or efficientlymeet a unit’s projected volume or demand. Thecompleteness metrics assess effectiveness inthe provision of adequate staff to cover pro-jected unit needs. The commitments metricassesses efficiency in the full scheduling of ex-isting staff resources before scheduling over-time or contract labor.

Completeness metrics (total completeness,licensed completeness, and nonlicensed com-pleteness) all measure the degree to whichrequired unit needs are met by representinga measure of how many schedule slots are‘‘unfilled’’ or ‘‘overfilled’’ in relation to thetotal number of projected required slots in theschedule based on projected volume. Ideally,every forecasted slot in the schedule is as-signed to a staff member, indicating the unit’sexpected demand is met with the appropriateamount of staff. In this state, the unit’s com-pleteness metric equals 0.0000. Each of the 3completeness views—total completeness (allstaff regardless of skill mix), licensed com-pleteness (licensed patient care staff), andnonlicensed completeness (support staff)—iscalculated the same way but considers a dif-ferent set or subset of staff to provide unitleaders direction toward identifying existingopportunities. It is important to note that thesemetrics only measure the completeness of thenumber of slots in the schedule to the plan (ie,number of staff needed for a projected patientvolume) and not the accuracy of the plan (ie,accuracy of projection of patient volume).

Measuring total staff versus licensed staffversus nonlicensed staff allows a unit leaderto quickly identify where opportunities exist

Table 3. Individual metric score rangesand ratings

Score Rangea Rating

0.0000 � Si < 0.1000 Good0.1000 � Si < 0.2500 Fair0.2500 � Si Poor

aSi = metric score.

Table 2. Data elements required for schedule quality assessment

Data Element Description

Future schedules Copy of schedule before start of schedule period (ie, 4- or 6-wk schedule)Staffing grids Tables indicating recommendations of the number of each skilled staff required

to take care of a given number of patientsEmployee commitment Number of hours per week agreed upon by each scheduled staff member

to workPatient volume projections Assumption or projection of patient volumes per unit per shiftActual hours worked Actual hours worked in each unit after the schedule period closes



in either overscheduling or underscheduling.Licensed staff includes those individuals withlicensures requiring certifications for provid-ing care, such as a RN, LPN, licensed vocationalnurse, and so forth. Nonlicensed staff mem-bers work in support positions that do notrequire specific patient care licenses, such asnurse assistants, unit clerks, technicians, andso forth. It is expected that there will be somedebate in grouping multiple skill sets togetherfor a metric score centering on the aspectthat many of the skill sets cannot necessarilybe used interchangeably. It is important to re-member, however, that the intent of thesemetrics is to identify areas of opportunity topoint leaders toward further investigationrather than detail all issues within a schedule.A system to detail and identify opportunity ateach level of skill set by department is left tothe more complex automated applications.

All 3 completeness metrics are calculatedfor each shift in the same manner using thesame formulas, which compare the number ofschedule slots to be filled to the number ofschedule slots actually filled on the schedule.The difference for each metric is the subset ofstaff slots being considered (ie, all, licensed, ornonlicensed). For each day i in the givenschedule period of N days, the number ofslots required for an expected volume (Ri) iscompared to the number of slots filled (Si).Every required slot that is not filled and everyslot filled but not required are a schedulevariance. The number of schedule variancesfor each day in the schedule period (Vi) issummed for all of the days in the scheduleperiod and divided by the sum of the number ofrequired slots in all days of the schedule period(see equations [1] to [3]). Therefore, the metricvalue represents the total number of schedulevariances as a percentage of total requiredscheduled slots for a given volume projection.

Vi ¼ jSi � Rij ðfor each day iÞ ð1Þ

PNi¼1

Vi

PNi¼1

Ri

i ¼ 1; :::;N ð2Þ

orPNi¼1

jSi � Rij

PNi¼1

Ri

i ¼ 1; :::;N ð3Þ

In equation (3) above, for the total complete-ness metric, the numerator represents theabsolute value of the sum of the differencesbetween the number of all staff slots filled/scheduled (Si) and the number of all scheduleslots required (Ri) for each day or, in simplerterms, the total number of schedule variancesacross the schedule period. Similarly, for thelicensed completeness metric, the numeratorrepresents the total number of schedule vari-ances across the schedule period consideringlicensed staff only. Finally, for the nonlicensedcompleteness metric, the numerator representsthe total number of schedule variances acrossthe schedule period considering nonlicensedstaff only. In all 3 metric scores, the denomina-tor represents the number of required slots tobe scheduled for the respective set of skills.

An example for the licensed completenessmetric is illustrated in Figure 1.

Assume:� 14-day schedule period for 5 employees

in pattern below.� D indicates day work shift; O, scheduled off.� Tech indicates technician (nonlicensed).� NA indicates nurse assistant (nonlicensed).Let N be 14 days in the schedule period

and Vi be the daily difference between filledlicensed schedule slots and required licenseslots. Then using equation (2):

(a) Sum of the absolute value of the variances

Vi orPNi¼1

Vi is 7.

(b) Sum of the required non-licensed slots to

be filled Ri orPNi¼1

Ri is 28.

(c) Therefore, licensed completeness met-ric = 7/28 = 0.2500.

An example for the nonlicensed complete-ness metric is illustrated in Figure 2.


in pattern below.� D indicates day work shift; O, scheduled off.



Let N be 14 days in the schedule period andVi be the daily difference between filled non-licensed schedule slots and required nonli-censed schedule slots. Then using equation (3):

(a) Sum of the absolute value of the

variances Vi orPNi¼1

Vi is 3.

(b) Sum of the required nonlicensed slots

to be filled Ri orPNi¼1

Ri is 14.

(c) Therefore, nonlicensed completenessmetric = 3/14 = 0.2143.

Metric values approaching 0.0000 indi-cate higher quality with required slots filledand nonrequired slots not used appropri-ately. Unassigned slots (underscheduling),nonrequired slots (overscheduling), licensedslots that are filled with nonlicensed staff(for the licensed completeness metric), andnonlicensed slots filled with licensed staff

Figure 2. Nonlicensed completeness metric. LPN indicates licensed practicing nurse; NA, nurse assistant;RN, registered nurse; Tech, technician.

Figure 1. Licensed completeness metric. LPN indicates licensed practicing nurse; NA, nurse assistant; RN,registered nurse; Tech, technician.



(for the nonlicensed completeness metric)all contribute toward moving the metricscores away from 0.0000 indicating lowerquality.

Commitment metric measures the degreeto which staff is scheduled to meet hourcommitments as agreed upon by the staff andthe unit or facility. Staff members are hiredinto commitment or complement levels de-fining the expectation of the number ofhours per week to be worked based on apercentage of a standard 40-hour work week(ie, a 0.90 staff member is expected to work36 hours per week) defined by:

ð4Þ40 hours � 0:90 ¼ 36 hours perweek

This metric measures how many occur-rences exist where employees are not sched-uled enough hours to meet commitmentlevels within the schedule period.

For each employee (x) in a given unit of(E) employees and for each week (y) in aschedule period of (P) segments, let (Mxy) bethe total number of occurrences where thenumber of hours scheduled (H) is less than the

staff member’s commitment (C). The metricscore is:

PPy¼1

PEx¼1

f ðMxyÞ ¼1; Hxy < Cx

0; Hxy < Cx

�

E � Pð5Þx ¼ 1; . . . ;E and y ¼ 1; . . . ;P

A metric value of 0.0000 results from aschedule where every staff member is sched-uled for hours that equal, at a minimum, the staffmember’s respective hour commitment duringeach defined segment of the schedule period.Nonzero values indicate lost opportunities inscheduling hours that can lead to staff dissatis-faction, as well as potentially be replaced withmore expensive overtime or contract labor.Nonproductive scheduled time such as vaca-tion, jury duty, and so forth is used to contributetoward the fulfillment of scheduling the com-mitment illustrated in Figure 3.


in pattern below (two 1-week periods).� D indicates day work shift of 12 hours;

V, vacation; O, scheduled off.

Figure 3. Commitments metric. LPN indicates licensed practicing nurse; NA, nurse assistant; RN, registered nurse;Sched, schedule; Tech, technician.



� Commitment hours for each employee is36 hours per schedule week.

� Defined segment period is 1 week.Let P be 2 (1-week) segment periods in

the schedule period, E be 5 employees, Hxy

be the number of hours scheduled in each1-week period for each employee, and Cxy

be 36 hours (commitment level) for eachemployee. Then using equation (5):

(a) Sum of the number of commitments not

met Mxy orPPy¼1

PEx¼1

f ðMxyÞ1;Hxy < Cx

0;Hxy � Cx

�

is 1.(b) Number of employees E is 5 and the

number of 1-week periods in theschedule is 2.

(c) Therefore, commitments metric = 1/(5 �2) = 0.1000.

The commitment metric calculation canbe adapted for facilities that measure commit-ments over multiple weeks (eg, where a payperiod consists of 2 weeks). In these circum-stances, multiple schedule weeks are aggre-gated to compare against the totalcommitments over the defined segment of thesame number of weeks. Assuming a 2-week payperiod, 2 weeks of the schedule is used tocompare against each employee’s commitmentlevel for 2 weeks instead of one. Equation (4)above becomes:

80 hours � 0:90 ¼ 72 hours perweek

The number of weeks in the definedsegment can be configured at the unit orfacility level to customize this metric appro-priately.

Schedule to employee needs

There are 3 metrics used to compare fu-ture schedules to employee needs (weekendcommitments, schedule healthiness, and pre-ferences). Each metric is prospective innature and assesses the schedule’s qualityon the basis of ability to meet a componentof employee needs. The major areas of as-sessment include scheduling of weekendassignments against weekend policies, sched-uling of unhealthy shift patterns, and sched-uling of employee preferences, honored ornot honored.

Weekend commitment metric measures thedegree to which weekend days are equitablyassigned to staff in accordance with facility orunit weekend assignment policy or individualweekend commitment. The metric measuresthe number of occurrences in the scheduleperiod where a staff member is scheduled forless weekend shifts compared with policy orcommitment as a percentage of the number ofemployees in the unit.

For each employee (x) in a given unit of(E) employees, let (Wx) be the total num-ber of occurrences where the number ofweekend shifts scheduled (Bx) is less thanthe staff member’s required number of week-end shifts to be worked (Dx). The metricscore is:

PEx¼1

f ðWxÞ ¼1;Bx � Dx > 1

0;Bx � Dx � 1

�

E

x ¼ 1; . . .E ð6Þ

Weekend shifts that are assigned according topolicy/commitment to each staff member overthe entire schedule period result in a perfectstate and metric score of 0.0000. Occurrencesof employees that are assigned to less week-end shifts than required move the metric scoreaway from 0.0000 as illustrated in Figure 4.


in pattern below.� D indicates day work shift of 12 hours;

V, vacation; O, scheduled off.� Unit policy is for each employee to work

2 weekend shifts per schedule period.Let E be 5 employees, Bx be the number of

weekend shifts scheduled in the scheduleperiod, and Dx be 2 weekend shifts requiredfor each employee. Then using equation (6):

(a) Sum of the number of employees whereweekend requirement is not scheduledPEx¼1

f ðWxÞ¼1;Bx � Dx > 1

0;Bx � Dx � 1

�

Ex ¼ 1; . . .E is 3.

(b) Number of employees E is 5.(c) Therefore, weekend commitment met-

ric = 3/5 = 0.6000.Schedule healthiness metric measures the

degree to which staff members are assigned



unhealthy shift patterns. These patterns aredefined as follows (but can be adjusted perfacility):

� assuming 12-hour shifts—working morethan 3 consecutive days,

� assuming 10-hours shifts—working morethan 4 consecutive days,

� assuming 8-hour shifts—working morethan 5 consecutive days,

� rotating shift within same 24-hour pe-riod (multiple shifts in the same day),

� rotating shift without adequate numberof shifts/days rest in between (minimumof 2 times the number of hours in theemployee’s standard shift with exceptionof 12-hour shifts).

� excessive consecutive hours includingboth scheduled work and call hours, and

� scheduling of overtime.The metric measures the number of occur-

rences of unhealthy shift patterns scheduled asa percentage of the number of employeesmultiplied by the number of days within theschedule period.

For each employee (x) in a given unit of(E) employees and for each day (i) in the

schedule period of (N) days, let (Uyn) be thetotal number of occurrences of unhealthyshift patterns. The metric score is:

PEx¼1

PNi¼1

Uxi

E � N x ¼ 1; . . .E and i ¼ 1; . . .N ð7Þ

Each occurrence of one of these unhealthy

shift patterns moves the metric value away

from 0.0000 indicating ‘‘unhealthiness’’ or

lower quality as illustrated in Figure 5. A metric

value of 0.0000 results from a schedule that

does not contain any unhealthy shift patterns.Assume:� 14-day schedule period for 5 employees

in pattern below.� D indicates day work shift of 12 hours;

O, scheduled off.

Let N be 14 days in the schedule period and

Uxi be the number of occurrences of unhealthy

shift pattern each day. Then using equation (7):(a) Sum of the number of occurrences

of unhealthy shift patterns Uxi or

PEx¼1

PNi¼1

Uxi is 6.

Figure 4. Weekend commitments metric. LPN indicates licensed practicing nurse; NA, nurse assistant; RN, registerednurse; Tech, technician.



(b) Number of employees E is 5 and thenumber of days in the schedule periodis 14.

(c) Therefore, healthiness metric = 6/(14 �5) = 0.0857.

Preferences metric measures the degree towhich employee requests and preferencesare honored in the schedule. This metric isassumed to be one of the key componentstoward employee satisfaction with schedul-ing. The metric measures the number ofemployee schedule requests that are nothonored within the schedule as a percentageof the number of employees in the unitmultiplied by the number of days in theschedule period.

For each employee (x) in a given unit of

(E) employees and for each day (i) in the

schedule period of (N) days, let (Rqxi) be the

total number of occurrences of employee

day-off requests not honored. The metric

score is:

PEx¼1

PNi¼1

Rqx;i

E � N x ¼ 1; . . . ;E and I ¼ 1; . . . ;N ð8Þ

A metric value of 0.000 results when every

staff member’s schedule request has been

honored within the schedule. Schedule re-

quests not honored move the metric value

away from 0.0000 as illustrated in Figure 6.

Figure 6. Preferences metric. LVN indicates licensed vocational nurse; RN, registered nurse.

Figure 5. Shift pattern healthiness metric. LPN indicates licensed practicing nurse; NA, nurse assistant;RN, registered nurse; Tech, technician.




in pattern below.� D indicates day work shift of 8 hours.Let N be 14 days in the schedule period

and Rqxi be the number of occurrences ofemployee day-off requests not honored.Then using equation (8):

(a) Sum of the number of occurrences ofnot honored day off requests Rqxi orPEx¼1

PNi¼1

Rqxi is 2.

(b) Number of employees E is 5 and thenumber of days in the schedule periodis 14.

(c) Therefore, preferences metric = 2/(5 �14) = 0.0286.

Schedule to actual hours worked

There is one metric measured to retrospec-tively compare future schedules to actualschedules (daily flexing). This metric is retro-spective in nature and measures the sum of thedifferences between the number of hoursscheduled and the number of actual hoursused for each day in the schedule period, in-dicating the schedule hours overage or short-age as a percentage of the number of hoursscheduled. The measure represents the num-ber of hours that a manager needed to adjust

based on actual events and therefore providesan indication of how closely the plannedschedule accommodated the actual needs ofthe unit.

Daily flexing metric measures the degreeto which the unit either ‘‘flexed’’ hours up or‘‘flexed’’ hours down. Units call in additionalstaff hours to cover for individuals that didnot show up for work (call in sick) or tohandle an unexpected increase in patientvolume or severity (acuity). Conversely, unitssend staff hours out of the unit or send staffhome when patient volume is too low tojustify having the current number of staff onthe unit. The metric measures the differencebetween scheduled hours and actual hoursas a percentage of total scheduled hours usedover the schedule period.

For each day (i) in the given schedule periodof (N) days, let (V ) be the absolute value of thevariance between the number of hours sched-uled for each day in the schedule period (Si)and the number of actual hours used for eachday in the schedule period (Hi). The metricscore is:

PNi¼1

Vi

PNi¼1

Hi

1 ¼ 1; . . . ; N ð9Þ

Figure 7. Daily flexing metric. LPN indicates licensed practicing nurse; NA, nurse assistant;RN, registered nurse; Tech, technician.



or

PNi¼1

jSi � Aij

PNi¼1

Hi

i ¼ 1; . . . ; N ð10Þ

Metric values approaching 0.0000 indicateless ‘‘flexing’’ of hours as illustrated in Figure 7and therefore less management time spent ondaily staffing functions. High values of thismetric can indicate a poor schedule, a poorplan, or both. If at the same time, the scheduleto plan metric performance is good, the unitmanager should review the projected volumeand staffing needs to reevaluate the unit plan.


in pattern below.� D indicates day work shift of 12 hours.Let N be 14 days from the schedule period

and Vi be the daily variance between schedulehours and actual hours used. Then usingequation (10):

(a) Sum of the absolute value of thevariances Vi or PN

i¼1

Viis 108.

(b) Sum of the actual hours used in theschedule period Hi or PN

i¼1

Hiis 456.

(c) Therefore, daily flexing metric = 108/456 = 0.2368.

CONCLUSION

The Scheduling Quality Assessment Metricsproposed in this document are intended tobe a starting point for the measurement ofschedule quality. Within a health care setting,many facilities lack a system of monitoringschedule quality.1 Although a large volume ofresearch and proposals concerning the im-provement of schedule quality via automated,

systematic forms of linear programming orheuristics has been published and severalcommercial systems exist, a gap remains inmany facilities. We believe that the concept ofschedule quality evaluation and improvementneeds to exist to contribute to the improve-ment of unit and facility performance. There-fore, this article proposes a measurementmethodology to provide a means for quantify-ing quality and identifying improvement op-portunities without the requirement of anexpensive commercial scheduling system.

The proposed metrics in this article pro-vide a simplified approach to quantifying staffschedule quality to provide an avenue forimproving schedules toward the perfect stateof balance between unit and employee needs.The metrics are intended to provide a mech-anism for measuring and improving schedulequality with the goal of the day-to-day oper-ationalization of the concept of schedulequality. It is envisioned that unit managerswho use these metrics during the schedulecreation process in an effort to improve qualityscores will produce higher quality schedules,which in turn will lead to higher staff satis-faction via more schedule stability and loweruse of premium labor dollars in the form ofincentives, overtime, and contract labor and freeup more manager time to focus on patient careand the supervision of staff providing patientcare. The metrics provide a unit-level view ofquality indicators intended to uncover trendsand potential issues affecting schedule quality.They should be used as guiding indicators tolead toward areas of attention and focus.Although a more detailed level of assessmentcould be achieved by accounting for morevaried and complex scenarios, the ability tooperationalize the metrics would be negativelyimpacted by the higher level of complexityrequiring a correspondingly higher level ofeffort for understanding and utilization.

REFERENCES

1. Silvestro R, Silvestro C. An evaluation of nurse ros-

tering practices in the National Health Service. J Adv

Nurs. 2000;32:525-535.

2. Silvestro R, Silvestro C. Towards a model of Strate-

gic Roster Planning and Control: an empirical study

of nurse rostering practices in the UK National

Health Service. Health Serv Manage Res. 2008;21:

93-105.



3. Cooperrider F. Staff input in scheduling boosts

morale. Hospitals. 1980;54:58-61.

4. Miller ML. Implementing self-scheduling. J Nurs Adm.

1984;14:29-48.

5. Oldenkamp JH. Quality in Fives: On the Analysis,

Operationalization and Application of Nursing

Schedule Quality [dissertation]. The Netherlands:

University of Groningen; 1996.

6. Burke EK, De Causmaecker P, Berghe GV, Van

Landeghem H. The state of the art nurse rostering.

J Sched. 2004;7:441-499.

7. Clark PF, Clark DA. Challenges facing nurses’ asso-

ciations and unions: a global perspective. Int Labour

Rev. 2003;142:29-48.

8. Bradley D, Martin J. Continuous personnel scheduling al-

gorithms: a literature review. J Soc Health Syst. 1990;2:8-23.



schedule quality assessment metrics

Documents