nursing manpower allocation in hospitals - home - הנדסת · pdf file ·...

Nursing Manpower Allocation - Assignments Vs. Quality 1

Nursing Manpower Allocation in HospitalsStaff Assignment Vs. Quality of Care

Issachar Gilad, Ohad KhabiaIndustrial Engineering and Management, Technion

Andris FreivaldsHal and Inge Marcus Department of Industrial & Manufacturing Engineering

Penn State University


Dilemmas

Can a hospital establishment be managed like an industrial or commercialservice organization?

In case we have a partial yes, than..

Can Industrial Engineering tools be used to enhance productivity measuresin daily management (like it does to industry)?

If possible, than..

Can the largest workforce in this operation - the nursing force, providebetter performance rates, according to economical criteria?

If yes, than..

Can an IE based methodology be used, to determine optimal nursingassignment in hospitals, to provide better quality care?


Dilemma – in practical terms

Can a pre-calculated performance time be a factor, to enhanceanonymous patient’s quality care?

orCan quality of nursing care be presented by calculated operationaltime - for nursing activities, at any given clinical ward in hospitals?

and

Is it practical to determined the best ‘mix of nurses’ for a random‘blend of patients’ in a ward?Can we then provide pre-determined values for quality of service?


There is an enormous demand for nurses in the immediate future, muchmore than can and will be available

Nurses FTE’s (Full Time Equivalent) doesn’t satisfy operational requirements

Nursing assignment are planed by the ‘head nurse’ according pastexperience – ‘guts feelings’

There is a non homogeneous distribution of work content among nurses

Quality of ‘nursing job’ does not play any factor in determination of nursingassignments

Forecasting the ward’s patients blend - doesn’t exist

Current state


Allocating a fixed number of nurses per shift while patient’s number and complexityvary from shift to shift

Nursing allocation pattern - Internal ward C, Sheba Med. Center

4

4.5

5

5

5

Morning

?23Friday

?23.5Monday

?24Sunday

?24Saturday

?23.5Tuesday - Thursday

Real needNightEveningDay / Shift

Current state - Example


� Develop a nursing manpower allocation model which dealswith workload, assignment and best fit between patient’sdemands and nurses availability - in clinical wards

� Establish a quantitative method that relates to productivityissues and to quality of care - for nursing activities

Objectives of study


� Redefine nursing activities to allocate productive ways to deal withdynamics of patients variability

� Define nursing skills suitable for proper handling at different nursinglevels for various patient’s needs

� Determine work load and nursing activities in a quantitative way - onbasis of performance measurement

� Exercise findings and study approach in 2 Internal Med. Wards and 1Surgery Ward in Sheba Med. Center

� Use simulation procedures to forecast best nurse and patient mixture– create a decision tool for management

Method


Method

Patientsclassification

Establishing time database for nursing activities

in target unit

Forecasting patientsblend

Nursing allocationpolicies definition

Converting the simulatorresults to a nursing FTE’s

Influence of the nursing FTE’son the nursing quality of care

Simulation


Design of study

Develop a patient’s classification routine

Establish time data base for nursing activities, to target departments

Develop a forecasting patients blend program

Introduce a nursing allocation policies

Simulate patients needs to nursing assignments along 4 steps

Integrate quality of treatment in a manpower allocation equation


Patients classification

Patient’s needs, in a time for care terms, varies as result of ‘nursingcomplexity’Nursing complexity, defines the scope of patient’s needs - equivalent to adiagnosis, related to the difficulty and measures of needed nursing careA five level scale to nursing complexity:

Level 1

Minimal nursing care – A patient that hisnursing needs are fulfilled by a routineactivity of the nursing staff

Partial nursing care – A patient with achronic disease that is not depended onthe nursing staff in implementation ofhis daily activities

Full nursing care – A patient that isconfined to his bed or depended on thenursing staff in implementation of hismajor daily activities

Complex nursing care – A patientthat needs an intensive nursing care

Special nursing care – A ventilatedpatient that his condition fits toemergency unit

Level 2Level 3

Level 4

Level 5


Nursing allocation policies, theoretical

ExpertNurseNurse

Levels4-5

Levels1-3

- PolicyNversion #1

ExpertNurse Nurse

Levels4-5

Levels1-3

N - PolicyVersion #2

Nurse

Levels1-5

- PolicyI


NurseExpertNurse

HeadNurse

Levels1-3

Levels4-5

M - Policy

ExpertNurseNurse

Level5

Levels1-3

Level4

W - Policy

Nursing allocation policies (cont.)


Performance time

The daily nursing work is a very complex work which characterized witha great random activitiesMore than 65% of the work performed by a nurse influenced by others,including physicians, administrators or professionals from otherdisciplinesOn average 74 different activities are executed by a nurse in oneshift (Murphy et al. ,1997)

Establishing a nursing time values (for core activities) can bemade by the following techniques:

Time studyWork samplingPMTSEvaluation


Collecting core time values


Procedures for collecting time data

Operativevalue

Simulator

currentstate

Evaluation

classicapproach

Worksampling

classicapproach

Time studyIncl. PMTS

Managerstool ?

Implementation

Level ofaccuracy

Appliedor used

Characterization

Technique

was notdone high

based ondata

very slow

varies

fairly fast

fast

justifiesitself

bad

problematic

good

non

quantitative

promising

problematic

Yes

(for pros.)

proposedstate

Classical approaches vs. current and proposed state

done inpast

usedoften

was notdone

low

intuitive


The core activities approachUnlike traditional industries, measuring outputs in the health care industryis very difficultHigh variability, random and frequency of calls for nursing care, contributeto complex definitions for ‘quality of care’ - such will not be adoptedA different approach is proposed, based on “time for core activities” whichwill spread along time from selected activities to all activitiesThe term for ‘quality of nursing care’ will gradually expand by self learning

Nursing coreactivities

Nursing activities


Time study technique was chosen as the best technique for establishingnursing time foundation

A time study analyst observes Nurse / Expert Nurse / Head Nurse / Aid Nurse andmeasures the actual time taken to perform an activityThe description for each activity and its duration are recorded in a time study form (Excel)

The nursing activities were divided into 2 groups:

Direct nursing activities (A): tasks that can be connected to the assigned patientcaring handled by his nurse such as: bandaging, medication, family interaction, patienthygiene etc.Indirect nursing activities (B): tasks that can not be connected to the assigned patientcaring handled by his nurse: unit related tasks (administration, clerical, team meetings etc.)co-ordination of care, breaks etc.

Direct time distributions were derived for each level of nursing complexity(for 3 shifts)

Establishing nursing time database


Simulation

A simulation model is developed using the Matlab software

Patients classification, nursing time database, patient’s blend forecastingand 5 nursing allocation policies are the basic “ingredients” in this Matlabsimulation program

GUI (Graphic User Interface) to provides an easy way for inserting datawhich determine both the planning period and the nursing policy

The simulator provides the following results:1. FTE’s for the nursing mix2. Total forecasted direct time3. Total number of forecasted patients


ThursdaySunMonTueWedThuFriSatPlanning day

Method implementation

Phase 1 => Backward assignment

Manpower planning using current method, by Head NurseAt end of week re-planning, by simulatorEvaluate the differences

Simulation


Phase 2 => Parallel assignment

Manpower planning, by Head Nurse and by Simulator - in parallelEvaluate the differences by the two planeres

Simulation




Phase 3 => Assignment by simulator

Manpower planning, by SimulatorEvaluate outcomesDebugging

Simulation




Introducing Quality

Phase 4 => Plan according quality

Plan according quality – use of quality measures while planingEvaluate results

Simulation+ Quality



Patient’s blend forecastingNursing staff mix will depend on the blend and volume of patientsadmitted and applicable nurses - demanding complexitymanpower allocation procedure occurs in a 2 weeks planning horizonA model has been developed in order to forecast the patient mixFor patients blend we define 3 patients ‘populations’:

Released – patients that were released during the shift from the ward to the community(home or institute), to other ward or to vacationAdmissions - patients that were accepted during the shift to the ward from another ward orfrom ERVeterans – patients that are not related to the 2 above populations

In order to forecast the patients blend, in a given shift, a 9 stagesstatistical analysis procedure was implemented


Patient’s blend forecasting1. Define the parameters that need to be forecasted(1) Released patients(2) Admissions(3) Veteran patients

2. Mine historical data for the parametersData from 1996 till 2006 was gathered

3. Divide the data to 2 groups: (1) Learning, and (2) Testing

(1) Learning group – data from 1996 till 2005 ; (2) Testing group – data from 2006

4. Search for any patterns in the learning data

Season patterns for the weeks days and exceptional patterns for some of the holidaysperiods were revealed

5. Select appropriate forecasting methods3 forecasting methods were selected : (1) Simple moving average ; (2) Weightedmoving average ; and (3) Exponnential smoothing

6. Select a measure for evaluation the forecasting goodness

MAPE - Mean AbsolutePercentage Error

7. Choose the best forecasting methodThis was done according to the value of the MAPE

8. Forecast to the test periodA forecast was made to 2006

9. Evaluate the forecasts goodnessThis was done by the following measures : (1)Forecast average errors ; (2) Forecast standarddeviation error ; and (3) Forecast error distribution


Quality of nursing care

109876

#

Measuring pulse,temperature and bloodpressure

Falling risk estimationLansky estimationPain estimationNorton estimation

Core activity

Admissions ward explanation5Patients classification4Measuring pain in a daily basis3Filling nursing report2Wearing identifying brace1

Core activity#

Quality of nursing care’s grade is a weighted averageThis will be calculated by the relative importance of each nursing coreactivity’s grade (AHP by Satty)10 nursing core activities were selected:


The Quality equation

The result of a nursing core activity is the percentage of patients for whom themeasured activity was fulfilled

Step 1 – converting the result of a nursing core activity to a nursing quality grade

The nursing quality grade for a nursing core activity result is computed as follows:

Step 2 – computing the weighted average grade for the Quality of nursing care (Satty, 1994)

The average grade for the quality of nursing care is computed as the sum of the products of eachnursing quality grade and its relative weight

The relative weight for each nursing quality grade determined by the PairedComparisons technique

)( iQ )( iR

∑=

⋅=10

1iii WQQ

( )iQ ( )iW

=∀≥

=Otherwise

iRRQ iii 0

10,...,1%80


Findings

2 Internal Wards (C and D) and 1 Surgery ward in the Sheba Med. Centerwere in the scope of a pilot studyAssignments, as determined for internal ward C and for internal ward D, are:

333434Night

454465Evening

56-58-Morning

Aid nurseNurseHead nurseAid nurseNurseHead nurseInternal ward DInternal ward C

Time study in the Sheba medical center


Patients blend

Blend of patients in C & D wards

18%

43%

27%

7% 5%

32%23%

38%

7%0%

0%

10%

20%

30%

40%

50%

1 2 3 4 5nursing level

Internal ward C Internal ward D

Based on the historical data an average percentages of each nursing level ofcare for both internal wards were calculated


Nursing operation time – pilot studyMaximal valueMininmal valueStandard div.Average (minutes)nDistributionNursing levelShift

29.312.66.619.51012 + 18 * Beta (0.644, 0.791)1Morning

41.712.39.424.92012 + 30 * Beta (0.654, 0.862)265.613.616.033.91813 + Weibull (21.9, 1.18)397.924.327.252.1724 + Gamma (39.6, 0.709)4

138.084.124.9101.0484 + 54 * Beta (0.255, 0.348)536.37.37.919.719Triangular (7, 15, 37)1Evening

39.97.47.622.437Normal (22.4, 7.54)238.116.26.726.12116 + 23 * Beta (0.844, 1.11)381.013.517.838.311Normal (38.3, 17)492.923.128.556.15Uniform (23, 93)511.20.82.75.124Normal (5.06, 2.64)1Night

20.81.94.79.3401 + Weibull (9.36, 1.88)220.75.04.612.0335 + Weibull (7.28, 1.21)378.812.124.941.36Uniform (12, 79)493.89.624.338.4119 + Weibull (30.8, 1.16)5

Maximal valueMininmal valueStandard div.Average (minutes)nDistributionAdmissions/ReleaseShift

59.715.213.430.81015 + Exponential (15.8)(*) AdmissionsMorning23.613.34.218.3713 + 11 * Beta (0.72, 0.779)Release59.715.213.430.81015 +Exponential (15.8)AdmissionsEvening

21.82.45.39.715Normal (9.73, 5.13)Release73.524.116.648.810Uniform (24, 74)AdmissionsNight


Selecting the forecasting days


Selecting a month


Selecting a predicted day


Selecting nursing policy


Findings of simulation - pilot study


-' פנימית ג

-' פנימית ד

Simulating – Wards C & D - morning shift


-' פנימית ג

-' פנימית ד

Simulating – Wards C & D - Eveninging shift


-' פנימית ג

-' פנימית ד

Simulating – Wards C & D - Night shift


So far…

nursing manpower allocation in hospitals - home - הנדסת · pdf file ·...

Documents