Parallel Program
A System Dynamics Approach to the
Hospital Nurse Turnover Problem in Taiwan
Showing H. Young, Associate Professor
Lih-Lian Hwang, Doctoral Student
Department of Business Management
National Sun Yat-Sen University, Kaohsiung, Taiwan, R.O.C.
ABSTRACT
The average nurse turnover rate in Taiwan’s hospitals was 31.56% in 1988. High turnover rate
has a detrimental effect on costs, staff morale, and patient care. Nursing staff turnover is a
critical personnel problem for nursing administrators and top hospital management. In order to
understand the underlying dynamic structure which causes such a high turnover rate of hospital
nurses in Taiwan, we developed a system dynamics model. The results of computer simulation
showed that the key solution to the hospital nurse turnover problem is not a quick-fix, but
rather a long-term committed and supportive hospital administration to make improving in the
nursing working conditions.
INTRODUCTION
In Taiwan, the average rates of nursing turnover in hospitals were 17.74% in 1983, 12.34-
17.74 in 1985, and 31.56% in 1988; 89% of 81 hospitals existed with nursing shortages in
1989. The supply of nurses were enough for hospitals to employ, but nurses did not like to join
hospitals. Nurses in hospitals complained of irregular working hours, unreasonable salaries,
inadequate benefits, heavy workload, and poor promotional opportunities. Since high turnover
rates and nursing vacancies had a detrimental effect on costs, revenues, staff morale, and
patient care, administrators in hospitals improved working conditions for nurses. Turnover
rates have declined in recent years, but newspapers reported the shortfall still were 14000 in
1993 (Shiau, Rong, Sheen, Yang, and Lan 1994; Hwang 1993; Chen, (Yu) Chao, Chiang,
Chen, and Chang 1992; Yang 1992; Chang, (Yu) Chao, Yang, and Chou 1987).
Hospital working conditions for nurses are frequently less attractive than those in other
areas of nursing practice and other professions. The shortage of registered nurses continues
nationwide. Periodically it is announced that this has become an acute crisis (Helmer and
McKnight 1988). Numerous studies have examined the reasons why staff leave their jobs, but
these have often produced conflicting findings and have frequently relied upon bivariate
996
System Dynamics '95 — Volume II
correlation or multiple regression techniques. The availability of analytical techniques such as
structural modelling provides an opportunity to examine several aspects of turnover behaviour
simultaneously, and overcome some of the methodological difficulties encountered by prior
research (Cavanagh and Coffin 1992).
Nursing staff turnover is a critical personnel problem for nursing administrators and top
hospital management (Mann 1989). Registered nurse turnover is a chronic problem in health
care (Wall 1988). So, the problem may be appropriate for a system dynamics approach. The
purpose of this research is to try to understand the underlying dynamic structure which causes
the problem and to propose suggestions.
MODEL DESCRIPTION
Sectors of the model
The model has three sectors: the patients sector, the nursing sector, and the management
sector. Patients are the demanders of nursing care, and nurses are the suppliers of nursing care.
Nurses leave hospitals due to irregular working hours, unreasonable salaries, inadequate
benefits, heavy workload, and poor promotional opportunities. Since high turnover rates and
nursing vacancies can have a detrimental effect on costs, revenues, staff morale, and patient
care, administrators in hospitals have to improve working conditions for nurses.
Causal Diagram
The causal diagram of this model is shown in Figure 1. In the causal diagram, there are five
feedback loops.
Loop 1 (turnover rate of hospital nurses---recruiting, selecting, and training of new nurses--
-workload of hospital nurses---turnover rate of hospital nurses) is a positive feedback loop. The
recruiting, selecting, and training of new nurses increases when the turnover rate of hospital
nurses grows, so that workload of hospital nurses rises. An increased workload for the nurses
who remain hospital, which in turn makes the turnover rate of hospital nurses to grow further.
Therefore, the turnover rate of hospital nurses becomes higher and higher in this positive
feedback loop.
Loop 2 (tumnover rate of hospital nurses---attractiveness of hospital to outer nurses---
nursing shortages---workload of hospital nurses---turnover rate of hospital nurses) is a positive
feedback loop. When turnover rate of hospital nurses increases, the attractiveness of hospital to
outer nurses decreases. Hospital faces the problems not only of retention but also of recruitment
of nurses, which makes the nursing shortages grows. Persistent nursing shortages of 20-25%
mean an increased workload for the nurses who remain hospital, which in turn affects morale.
997
Parallel Program
Eventually more nurses leave (Delamothe 1988). Therefore, the turnover rate of hospital nurses
becomes higher and higher in this positive feedback loop.
improve working
external working conditions for nurses “>
environment +
. aA tension of hospital
working conditions administrators
of hospital nurses
+
+f >)
costs +
turnover rate of
hospital nurses + close
wards
recruiting, selecting, and
training of new nurses
attractiveness _
of hospital to
outer nurses
patients
in hospital
nursing
shortages +
\ + people in need
workload of of treatment
hospital nurses
Figure 1. The causal diagram of the nursing turnover in hospital
Loop 3 (turnover rate of hospital nurses---recruiting, selecting, and training of new nur:
-costs---tension of hospital administrators---improve working conditions for nurses---working
conditions of hospital nurses---turnover rate of hospital nurses) is a negative feedback loop.
When the turnover rate of hospital nurses grows, the recruiting, selecting, and training of new
nurses increases. The costs of recruiting and orienting a professional nurse to an institution may
range from $2000 to $5000 (Helmer and McKnight 1988; Wall 1988; Hinshaw, Smeltzer and
Atwood 1987). When the recruiting, selecting, and training of new nurses rises, costs grow so
that the tension of hospital administrators increases. When the tension of hospital administrators
rises, the administrators then improve the working conditions for nurses, which makes the
998
System Dynamics '95 — Volume II
working conditions of hospital nurses a change for the better. When the working conditions of
hospital nurses are improved, the turnover rate of hospital nurses decreases.
Loop 4 (turnover rate of hospital nurses---attractiveness of hospital to outer nurses---
nursing shortages---tension of hospital administrators---improve working conditions for
nurses---working conditions of hospital nurses---turnover rate of hospital nurses) is a negative
feedback loop. The attractiveness of hospital to outer nurses decreases when turnover rate of
hospital nurses increases, so that the nursing shortages grows. Beds have to be shut down due
to nursing shortages or loss of experienced staff, resulting in a loss of bed revenues (Mann
1989; Delamothe 1988). When the nursing shortages rises, the tension of hospital
administrators increases, which makes the administrators improve working conditions for
nurses. When the working conditions of hospital nurses are improved, the turnover rate of
hospital nurses decreases.
Loop 5 (turnover rate of hospital nurses---attractiveness of hospital to outer nurses---
nursing shortages---tension of hospital administrators---close wards---patients in hospital---
workload of hospital nurses---turnover rate of hospital nurses) is a negative feedback loop. The
attractiveness of hospital to outer nurses decreases when turnover rate of hospital nurses
increases, so that the nursing shortages grows. When the nursing shortages rises, the tension
of hospital administrators increases, which makes the administrators close wards. When the
administrators close wards, patients in hospital decreases so that the workload of hospital
nurses declines. When the workload of hospital nurses reduces, the turnover rate of hospital
nurses decreases.
It is assumed that people in need of treatment and external working environment are
exogenous variables.
Key variables and equations
Turnover rate determination function
Naturally, the extent of functional turnover (turnover that actually is beneficial to the
organization) is unknown. Presumably, it would vary depending on the nature of the
organization (Dalton and Krackhardt 1982). It is economically dysfunctional when it reduces
the capacity of hospital units (Mann 1989). When workload and environment gap increases, the
turnover rate rises. This model deduces the determination of turnover rate as follows:
TR = separation/NCN eq.1
NCN = EN+NN eq.2
separation = INT(EN*(MIN(1,(FENTR+DENTR)))) eq.3
EN(t) = EN(t - dt) + ( US- separation) * dt eq.4
NN(t) = NN(t - dt) + (selection - US) * dt eq.5
999
Parallel Program
DENTR = DELAY(EEGTR+EWTR,0.5) eq.6
EEGTR = GRAPH(EG)
(-0.05, 0.00), (-1.39e-18, 0.00), (0.05, 0.05), (0.1, 0.25), (0.15, 0.375), (0.2, 0.45),
(0.25, 0.48), (0.3, 0.505), (0.35, 0.52), (0.4, 0.525), (0.45, 0.525) eq.7
EG = EWE-WC eq.8
EWE = 1*(1+GREWE)\(TIME-1981) eq.9
where
TR: turnover__rate(dimensionless)
NCN: number_of_current_nurses(persons)
EN: experienced_nurses(persons)
NN: new__nurses(persons)
separation(persons/year)
FENTR: functional_experienced_nursing_turnover_rate(dimensionless)
DENTR: dysfunctional_experienced_nursing_turnover_rate(dimensionless)
US: up_to_speed(persons/year)
dt: delta time, simulation solution interval(years)
EEGTR: effect_of_environment_gap_on_turnover rate(dimensionless)
EWTR: effect_of_workload_on_turnover rate(dimensionless)
selection(persons/year)
EG: environment_gap(dimensionless)
EWE: external_working_environment(dimensionless)
WC: working_conditions(dimensionless)
GREWE: growth_rate_of_external_working_environment(dimensionless)
Nursing shortage determination function
When turmover rate of hospital nurses increases, the attractiveness of hospital to outer
nurses decreases. Hospital faces the problems not only of retention but also of recruitment of
nurses, which makes the nursing shortages grows. This model deduces the determination of
nursing shortage as follows:
NS = recruitment-selection eq.10
recruitment = NAN-NCN + separation eq.11
selection = MIN(recruitment,respondents) eq.12
respondents = recruitment*DELA Y(AHON,0.5) eq.13
AHON = GRAPH(TR)
(0.2, 1.00), (0.23, 1.00), (0.26, 0.98), (0.29, 0.925), (0.32, 0.825), (0.35, 0.7),
(0.38, 0.6), (0.41, 0.535), (0.44, 0.505), (0.47, 0.485), (0.5, 0.485) eq.14
where
NS: nursing_shortage(persons)
recruitment(persons)
NAN: number_of_authorized_Nurses(persons)
respondents(persons)
AHON: attractiveness_of_hospital__to_outer_nurses(dimensionless)
Tension of improving working conditions determination function
High turnover rate has a detrimental effect on costs, staff morale, and patient care. The
turnover rate rises, then the tension of hospital administrators increases, which makes the
administrators improve working conditions for nurses. When the working conditions of
1000
System Dynamics '95 — Volume II
hospital nurses are improved, the turnover rate of hospital nurses decreases, which in turn
reduces the tension of improving working conditions. This model deduces the determination of
tension of improving working conditions as follows:
TIWC(t) = TIWC(t- dt) + (IT - RT) * dt eq.15
IT = GRAPH(TR)
(0.2, 0.00), (0.22, 0.03), (0.24, 0.115), (0.26, 0.45), (0.28, 0.8), (0.3, 0.925),
(0.32, 0.985), (0.34, 1.00), (0.36, 1.00), (0.38, 1.00), (0.4, 1.00) eq.16
RT = IF(RP=1)AND(TR<0.28))THEN(TIWC)ELSE(0.2*TIWC) eq.17
RP = DELAY (IF(TIWC>1)THEN(1)ELSE(0),1) eq. 18
where
TIWC: Tension_of_improving_working_conditions(dimensionless)
TT: increasing_tension(dimensionless)
RT: tension_reduced(dimensionless)
RP: retention_policy(units)
Working conditions determination function
Administrators in many hospitals have not always been responsive to suggestions from
nurses about ways to improve working conditions (Helmer and McKnight 1988). This model
deduces the determination of working conditions as follows:
WC(t) = WC(t - dt) + (improving) * dt eq.19
improving = WC*IRWC eq.20
IRWC = IF(RP=1)THEN(GREWE+0.03)ELSE(GREWE-0.01) eq.21
where
improving(dimensionless)
IRWC: improving_rate_of_working_conditions(dimensionless)
SIMULATION
As shown in the causal diagram of Figure 1, if people in need of treatment and external
working environment do not change, the feedback loops will remain steady and no fluctuation
will occur. We will change people in need of treatment (PNT) in Test 1 and growth rate of
external working environment (GREWE) in Test 2 to some extent and see the patterns of
simulated outcomes. Time for simulation is 40 years.
Test 1
Suppose test 1 input conditions are as follows:
PNT = 480+PULSE(120,1987,100)
GREWE =0
The PULSE function generates a pulse input. Figure 2 shows the system behavior under
test 1 input conditions. In Figure 2, workload rises and declines sharply because people in need
of treatment pulses. The rapid growth workload in turn increases quickly the turnover rate of
Parallel Program
hospital nurses. Tension_of_improving_working_conditions increases sharply because
turnover rate grows rapidly. When turnover rate rises, the attractiveness of hospital to outer
nurses decreases (not shown). Hospital faces the problems not only of retention but also of
recruitment of nurses, which makes the nursing shortages ascends. Workload increases again
due to high turnover rate and nursing shortages, which in tum grows turnover rate again.
Tension_of_improving_working_conditions decreases due to the rate of tension reduced greater
than the rate of increasing tension (not shown). Turnover rate rises again results in the rate of
increasing tension greater than the rate of tension reduced (not shown), which makes
tension_of_improving_working_conditions ascends again.
1: workload 2: environment gap 3: turnover rate 4: nursing shortage
5: Tension of improving working conditions
=
1: 136
i oe
5: 0.40
O00 2 2
A . 2
3: E
1
5: 0.20
dy 1.21 ~
a 18 >
4 0.00 po ——— 1345
> 0.001981.00 1990.75 4 2000.50 3 4 2010.25 2020.00
Years
Figure 2. System response under the test 1 input conditions
Test 2
Suppose test 2 input conditions are as follows:
PNT = 480
GREWE = 0.05
Figure 3 shows the model response to test 2 input conditions. In Figure 3, environment gap
rises because external working environment grows by a constant 5 percent per year. The
increased environment gap in turn ascends the turnover rate of hospital nurses. The recruitment
and selection of new nurses rises (not shown) when the turnover rate of hospital nurses grows,
so that workload of hospital nurses increases. When the turnover rate rises, the tension of
hospital administrators increases, which makes the administrators improve working conditions
for nurses. When turnover rate rises, the attractiveness of hospital to outer nurses decreases
(not shown). Hospital faces the problems not only of retention but also of recruitment of
1002
System Dynamics '95 — Volume II
by 90 from 1995.
PNT = 480+STEP(90,1995)
Following the STEP change, the results indicated in Figure 5 still show fewer fluctuations
than the behavior resulting from the original system (as in Figure 3).
031 1: turnover rate 2: nursing shortage
2: 15.07
: 0.24
st
I O17 F 2 12 L2
2: 0.00}981.00 1990.75 2000.50 2010.25 2020.00
Years
Figure 4. System response under the conditions of PNT = 480, GREWE = 0.05, and new
policy from 1992
ai 1: turnover rate 2: nursing shortage
2 15.07 a
1: 0.24,
2 7:54 van an an Laman am an
1: 0.17]
2: 0.001981.00
1990.75 2000.50
Years
Figure 5. System response under the conditions of PNT = 480+STEP(90,1995), GREWE =
0.05, and new policy from 1992
2010.25 2020.00
SUMMARY
1003
Parallel Program
The problem of turnover rate fluctuations of hospital nurses has been examined and the
dynamic structure of information and decision flows causing that problem has been described.
In the model, the structure has been changed and tested. Hospital nurse turnover problem
appears to have been eliminated in the new structure.
One general conclusion that may be drawn from this study, however, is that fluctuations are
often not externally caused but are internally generated. Hospital administrators and directors of
nursing used to blame outside forces for the nursing turnover problem in Taiwan. The results
of computer simulation based on a system dynamics model show the key solution to the
hospital nurse turnover problem is not a quick-fix, but rather a long-term committed and
supportive hospital administration to make improving in the nursing working conditions. .
Our proposed suggestion to the complex problem of turnover seem very straightforward.
Perhaps the greatest obstacle to implementing this suggestion is that hospital administrators,
just like many other major institutions in our society, and many other societies in the world as
well, have to undergo a paradigm shift in management attitudes and practices.
REFERENCES
Cavanagh, S. J. and D. A. Coffin. 1992. Staff Turnover Among Hospital Nurses. Journal of
Advanced Nursing. 17(11):1369-1376.
Chang, M., Y. M. (Yu) Chao, C. L. Yang, and C. H. Chou. 1987. Survey on Nurses
Licensed in Recent 10 Years-1974-1983. The Journal of Nursing. 34(1): 53-61.
Chen, Y. C., Y. M. (Yu) Chao, T. L. Chiang, S. G. Chen, and D. J. Chang. 1992. A Study
of the Supply of Nursing Manpower. The Journal of Nursing. 39(1): 35-45.
Dalton, D. R., W. D. Todor and D. M. Krackhardt. 1982. Turnover Overstated: The
Functional Taxonomy. Academy of Management Review. 7(1): 117-123.
Delamothe, T. 1988. Nursing Grievances, I: Voting with their feet. British Medical Journal.
296(6614): 25-28.
Helmer, F. T. and P. McKnight. 1988. One More Time-Solutions to the Nursing Shortage.
Journal of Nursing Administration. 18(11):7-15.
Hinshaw, A. S., C. H. Smeltzer and J. R. Atwood. 1987. Innovative Retention Strategies for
Nursing Staff. Journal of Nursing Administration. 17(6): 8-16.
Hwang, Lih Lian. 1993. A System Dynamics Approach to the Hospital Nurse Turnover
Problems. Unpublished Master’s Thesis. Institute of Business Management, National
Sun Yat-Sen University. Kaohsiung, Taiwan, R.O.C.
Mann, E. E. 1989. A Human Capital Approach to ICU Nurse Retention. Journal of Nursing
Administration. 19(10): 8-16.
Shiau, S. J., J. R. Rong, Y. T. Sheen, L. L. Yang, and C. F. Lan. 1994. Issues and
Prospects for Nursing Education System in Taiwan. The Journal of Nursing. 41(2): 52-
61.
Wall, L. L. 1988. Plan Development for a Nurse Recruitment-Retention Program. Journal of
Nursing Administration. 18(2): 20-26.
Yang, K. P. 1992. The Current Nursing Shortage at Hospitals in Taiwan--- A Study on
Retention of Nurses in Taiwan: A Positive Approach to the Nursing Shortage(Part I). The
Journal of Nursing. 39(1): 47-53.
1004
System Dynamics '95 — Volume II
nurses, which makes the nursing shortages grow. The environment gap declines when the
working conditions of hospital nurses are improved, so that turnover rate of hospital nurses
decreases.
1: workload 2: environment gap 3: turnover rate 4: nursing shortage
1.62 + 5: Tension of improving working conditions
4 04 r
3: 35. is
4 4165 F
5: 2.78
1: «1.42
z 09
2 0688
5: 1.39
1: (1.217
2 8 rf
4 0.00 :
5: 0.00 981.00 1990.75 2000.50 2010.25 2020.00
Years
Figure 3. System response under the test 2 input conditions
Policy design for improving behavior
The problems are manifested as undesirable oscillations in the behavior of the system. This
situation may be improved by changing policies used by management. The new policy results
described below are derived through analysis and trial of various policies. Although only the
final results are presented here, the process to obtain these is long and far from simple.
Instead of responding to turnover rate in making decisions about improving working
conditions for nurses, management should improve working conditions for nurses actively.
The former eq.21 is modified:
IRWC = IF(RP=1)THEN(GREWE+0.03)ELSE(IF(TIME<1992)THEN(GREWE-
0.01)ELSE(GREWE+0.001))
The results indicated in Figure 4 show fewer fluctuations than the behavior resulting from
the original system (as in Figure 3). Turnover rate has a peak value in Figure 4 of 0.31 in 1988
and gradually returns to its normal value of 0.17 without much oscillation.
Testing scenarios in policy design
Hospitals experience many other types of external environment changes. A good policy
cannot be chosen until policy performance is evaluated under the conditions the hospital is
likely to face. If the people in need of treatment remains steady in 1981-1994, it then increases
1005
