7128-
A System Dynamics Model to Examine the Effectiveness of
Research and Education in Reducing Societal
Costs Due to Low Back Pain
William L. Cats-Baril, Ph.D.
University of Vermont
Wilfred Roth, Ph.D.
University of Vermont
Richard H. Zeliff
University of Vermont
ABSTRACT
Low Back Pain (LBP) is the most common cause of work loss after the
ordinary cold, and is the single.greatest source of compensation
payments. In the U.S., it is estimated that one million workers
sustain a low back injury every year, and that 217 million work days
are lost annually at a cost of 11 billion dollars for males aged 18-55
alone. In an effort to better understand how to control the economic
impact of this disorder, a System Dynamics model is being developed,
It is hoped that the model, by generating scenarios on the cost
effectiveness of different interventions, will provide useful insight
into specific policies to fund research addressing the causes of LBP
disability.
1, INTRODUCTION
It is not surprising that containment of health costs has been on top
of the political agenda for some time. Health costs, adjusted for
inflation, have increased continuously and dramatically over the last
25 years both on a per capita basis and as a percentage of the gross
national product (Economic Report of the President, 1983). Attempts to
curb costs have failed in the past and it is too early to tell whether
More recent strategies, like the Medicare new method of payment by
diagnosis-related groups, will be effective (Vladeck, 1984).
The basic question, of course, is whether we, as a society, are
healthier because of these expenditures. A no less relevant question
is whether the improvement in general health has been cost effective.
While statistics seem to answer the first question affirmatively, they
tend to deny the second. Indeed, as Luginbuhl, et al (1981) have
pointed out, the increase in health costs, as a percentage of the
gross national product, is disturbing since it suggests that
"investment" in health care may not return its costs in increased
productivity.
-129-
The notion of looking at health care under an investment criterion is
particularly relevant in an era of budgetary constraints. Still,
performing cost-benefit arialyses on specific illnesses and their
respective treatments is difficult. Assessing the impact of
intervention and rehabilitation efforts requires an analysis of not
only the medical costs of the intervention itself, but also of the
compensation and lost productivity costs that could be incurred.
The work presented here addresses these issues in the context of . low
back pain (LBP). The successful rehabilitation of LBP sufferers is a
matter of critical medical and socio-economic importance. This
disorder is mankind's major muscoloskeletal complaint, is the second
most common cause of work loss after the ordinary cold, and is the
single greatest source of compensation payments (Kelsey, et al, 1979).
Specifically, the questions we want to answer are whether we could
save society money in the long run by funding research into preventing
and treating LBP injuries and disabilities and, if so, how much? That
is, we want to determine the effect of several levels of funding,
coming from federal and private sources, on reducing the total costs
to society of LBP impairment and disability.
To explore these questions we developed a System Dynamic model that
simulates the migration of healthy individuals to various states of
disability. The model computes the net costs to society, assesses the
pressures that those costs may exert on research funding mechanisms,
assumes different efficiencies for medical and behavioral
interventions, and calculates the resulting number of disabled
individuals. Various scenarios can be developed by changing a variety
of societal factors.
The paper is organized in four sections. The first section examines
the importance of LBP in terms of its incidence and costs, and
discusses the difficulties involved in treating the disorder. The
second section decribes the System Dynamics model in detail. The third
section addresses the assumptions on which the model is built and the
range of possible scenarios that the model can generate. Finally, the
paper concludes with a note on the validity and reliability of the
model and some comments on future research,
2. BACKGROUND
The cumulative effects of LBP in America are staggering. Estimates of
the percentage of Americans who have had LBP at some point in their
lives range from 65% to 80% (Murphy and Cornish, 1984). According to
the National Center for Health Statistics (1981), impairments of the
back or spine (excluding spinal cord injury) are the major cause of
impairment in the U.S., after hearing and sight impairments, affecting
9,365,000 persons (4.41% of the population). Furthermore, disabling
impairments of the back or spine are the leading cause of disability
in this country, affecting 2,391,000 persons (1.13% of the population)
in 1977, Moreover, the rate of impairment is increasing.
-130-
Indeed, while from 1974 to 1978 the general population increased 4.8%,
LBP related impairments went up by 21%. The Bureau of Labor Statistics
reported that one million workers sustained a low back injury in 1980
alone. Frymoyer, et al (1983) estimate that 217 million work days are
lost annually at a total cost of 11 billion dollars for males aged 18
to 55. The result in terms of economic impact is calculated to be in
excess of 20 billion dollars per year.
Despite these alarming statistics, the dynamics of LBP are still'not
well understood. Further research in at least two areas needs to be
done. First, the risk factors which can bring on, and/or increase the
severity and duration of, LBP have not been fully determined. Several
epidemiologic studies have identified specific factors, but do not
agree on their significance. This problem is due, in part, to the
large number of risk factors that have been proposed. Recently, for
example, a panel of experts identified 104 potential ones (Cats-Baril,
1984). These included the psychological profile of the individual, his
overall fitness, whether he had a previous back. injury, how
compensable his injury is, his age, his education, the amount of
lifting and vibration on the job, and the method of payment at work,
among others.
Second, the identification of the etiology of LBP complaints has also
proven to be difficult. Although certain congenital and acquired
lesions, acute trauma and other causes can be identified in some
cases, the majority of low back complaints have defied precise
structural diagnosis. Even with the use of the latest diagnostic
techniques, no more than 50% of LBP sufferers receive a definite
diagnosis.
Frymoyer (1984) has pointed out that because accurate diagnosis is so
elusive, it is not surprising that treatment programs and
rehabilitation efforts for LBP sufferers often fail. Indeed, when
treatment programs - e.g., physical therapy, pain medication, bracing,
electric stimulation, etc. - have been tested by prospective
randomized clinical trials, they have not shown a significant effect.
It is safe to assume that if treatments could be geared toward
specific and demonstrated factors causing LBP, their effectiveness
would be much higher.
The overview of LBP therefore portrays an extremely common, frequently
disabling clinical syndrome, often of undetermined etiology, for which
acute and chronic treatment programs may yield low symptomatic relief
and ineffective rehabilitation, particularly when a precise diagnosis
is not available. The high prevalence in the working population, the
resultant disability and its socio-economic impact suggest a critical
need for a better understanding of this problem.
In an effort to assess the potential impact of further research in LBP
- research that would develop more effective treatments and prevention
programs - and to determine the cost-effectiveness of funding such
research, a System Dynamics model was developed.
~ =131-
3. THE MODEL
The System Dynamics model consists of two major subsystems: the
problem subsystem which models the migration of individuals between
the various states of LBP disability, and the control subsystem which
comprises the costing and the funding and intervention policy sectors.
The causal diagram is shown in Figure l. :
+
Heaitry (> sp
Population 7) tojuries
+
Afflicted
Population
Recoveries/ e +
ty Rehabilitations io) 7 a
# ax Education/ Training
ES Disabled +, of Population on
=> Population How to Avoid Injuries
+
4 Povthanently
Not Costs Disabied
to Society Population (e)
NN, Pe he Need for
2 st Reduction in L8P
and LBP Coste
Reduction in
Treatment Coste
age is + Funding for
reinenta LBP Research/
gS Education
Advances in
Z
Medical Knowiedae/
Technology
FIGURE 1: Causal Loop
3.1. THE PROBLEM SUBSYSTEM
The problem subsystem consists of eight sectors. Each sector consists
of a level and all the inflowing and outflowing rates. The sectors are
the following: the Healthy Population, the Injured But Functionally
Recovered, the Short-Term Disabled, the Long-Term Disabled, the
Rehabilitated Population, the Short-Term Disabled * after
Rehabilitation, the Long-Term Disabled After Rehabilitation and the
Permanently Disabled. All population birth rates have been combined to
-132-
Hickman, 1972; White, 1969). So, while it is conceivable that some
individuals may decide to return to the work force, in one way or
another, after a disability of more than a year, the likelihood is
almost negligible. The model assumes that no individual who has been
classified as permanently disabled returns to work, i.e., the model
assumes that they will remain disabled until they die and computes
compensation and lost productivity costs accordingly.
3.2. THE CONTROL SUBSYSTEM
The Control subsystem consists of two sectors: the Cost Accumulation
sector, which comprises the equations for computing the annual
monetary costs incurred by society -due to LBP, and the Cost
Containment sector, which comprises the equations for determining the
treatment cost rates, the injury rates and the success rates of
rehabilitation and recovery. Both sectors are discussed below and the
Cost Containment sector is shown in Figure 3.
3.2.1. THE COST ACCUMULATION SECTOR
The Cost Accumulation sector is composed of four elements: the
aggregate treatment costs, the aggregate productivity costs, the
aggregate compensation costs, and the net costs to society which is
simply the sum of the three aggregate costs. In the model, all costs
are treated as auxiliaries
The treatment costs are derived from the levels in the problem
subsystem in which medical and rehabilitation costs are incurred.
Specifically, the treatment costs, ona per patient per year basis,
are accumulated from the LTDISR, LTDIS, STDISR and STDIS levels.
Productivity costs can be incurred for two reasons: either because of
time lost from work (e.g., individuals in the STDIS, LTDIS, STDISR,
LTDISR, and PERMD populations), and/or because of lower work
efficiency for a proportion of the IBFR and REHAB populations.
The compensation costs represent the total value of all transfer
payments made to people with compensable LBP injuries. Snook and
Jensen (1984) estimate that the mean compensation cost is $6,000. They
also estimate that disability payments constitute 67% of the total LBP
costs, with up to 90% being accounted for by the PERMD population.
3.2.2. THE COST CONTAINMENT SECTOR
One of the purposes of this model is to reflect the effects of
additional expenditures for research and education on LBP on the net
costs to society. A positive effect, i.e., lowering of costs, can be
achieved in at least four ways: by lowering the treatment costs, by
reducing the injury and reinjury rates, by increasing the
effectiveness of the rehabilitation and treatment efforts, and by
more strict and vigilant compensation laws.
-133-
form the Aggregate Birth Rate which flows only into the Healthy
Population level. Also, while each sector has a corresponding death
rate, the death rates, for simplicity sake, are not mentioned ia the
description of each of the sectors provided below. The flow diagram is
shown in Figure 2.
Rencteasaaes XC
Stam Oates
foe neha = Pesdatae (oto)
Popuistion (STDISR) ee
(F ‘om a = rine g
i t =
ii ote
: x 7
rag Tolored bat eg
ion (RPP : oe ene bess
Popul tx-AP Population USFAY re
i’ = we L a oe a
ca * te VE . fend &
eae ToneT 4
Tene Yor biseied | jae NI Pepwiston (ois)
‘Mier Renabitation
Popuiation (LTOISAD
T ‘Permanent Disabled | =
Loew oe (PERMO) ;
ee
FIGURE 2: Flow Diagram
3.1.1. THE HEALTHY POPULATION
Individuals in the Healthy Population (HP) are adults (defined as
older than 18 years old) who have never experienced LBP. They leave
the Healthy Population either by dying or by suffering their first
instance of LBP. Once an individual has suffered a LBP episode, he is
never again considered part of the Healthy Population.
This first LBP occurence can result in one of three specific
outcomes: a) no disability - the individual has some discomfort but
does not miss any work; b) short-term disability - the individual
misses up to three months of work as a result of the LBP episode; or
c) long-term disability - the individual is unable to show up for work
for more than three months. Each of these injury rates are established
in terms of annual percentages of the HP.
3.1.2. THE INJURED BUT FUNCTIONALLY RECOVERED POPULATION SECTOR
Individuals in the Injured But Functionally Recovered (IBFR)
population are those individuals who have suffered at least one
episode of LBP but have not had to change their employment situation
in any way. There are three paths leading to IBFR: a) from HP via a
first ever LBP episode with no disabling consequences; b) from the
Short-Term Disabled Population via a recovery from a LBP disabling
-134-
episode no longer than three months; or c) from the Long-Term Disabled
Population via a recovery from a LBP disabling episode longer than
three months.
Individuals in the IBFR population have a higher susceptibility to
LBP, thus, the reinjury rate of this population is higher than the
rate of first injury for the healthy individuals.
3.1.3. THE SHORT-TERM DISABLED POPULATION SECTOR
The Short-Term Disabled (STDIS) population is made up of those
individuals who are unable to work for more than one day but less than
three months. While this definition is arbitrary, it has been
suggested as a good operational benchmark by physicians and physical
therapists (Cats-Baril, 1984).
The STDIS population is fed by two rates: the initial injury rate from
HP and the reinjury rate from IBFR. There are three ways of leaving
STDIS: a) through recovery from the LBP episode to the IBFR
population; b) through misdiagnosis - some patients remain disabled
longer than three months and should have been diagnosed as long-term
disabled in the first place - to the Long-Term Disabled population; or
c) through rehabilitation (assumed to last one year) to the
Rehabilitated population.
The distinction between recovery and rehabilitation is critical.
Recovery from LBP is spontaneous in at least 50% of the cases and
requires minimal treatment and medical intervention ( e.g., bed rest).
Recovery means that individuals can return to their former employment
status and to their former lifestyles with minimal, or no, changes.
Rehabilitation, on the other hand, usually occurs at a substantial
cost and has been defined to imply that the individuals must change
their employment status: either change jobs or drastically alter their
present workplace. Rehabilitation is much more costly than recovery.
It is important to note that once an individual has been rehabilitated
he 'forever' leaves the IBFR population sector to become part of the
Rehabilitated population.
3.1.4. THE LONG-TERM DISABLED POPULATION SECTOR
Individuals in the Long-Term Disabled (LTDIS) Population are those
indiviuals who have been unable to work for more than three months.
The three ways of reaching this population have been mentioned before:
a) through an initial serious injury from HP; b) through a reinjury
from IBFR; or c) through misdiagnosis of a short-term disabling injury
from STDIS.
Individuals can leave the LTDIS population by recovering from the
injury and going back to their former employment, and thus, back to
IBFR. They can also go through a rehabilitation effort (assumed to
-135-
last a year), change employment and join the Rehabilitated population.
Finally, they can deteriorate and become permanently disabled.
Typically, individuals who have been disabled for more than a year and
are getting some form of compensation seldom return to any form of
gainful employment.
3.1.5. THE REHABILITATED POPULATION SECTOR
The Rehabilitated (REHAB) population is made up of individuals who
have undergone some sort of rehabilitation because of a LBP
impairment. By definition, these individuals have changed jobs because
of their affliction, though they may be leading an otherwise normal
life.
REHAB can be reached not only from STDIS and LTDIS, as described
above, but also by recovering from short and long term reinjuries.
While people who have had a LBP injury severe enough to merit
rehabilitation are highly susceptible to recurrences, it is not clear
that the reinjury rates should be higher than those for the IBFR
population. On the one hand, individuals in the REHAB population have
fragile backs. On the other, they have been exposed to substantial
education efforts (e.g., back school) and they have curtailed
activities that would expose them to a reinjury.
3.1.6. SHORT-TERM DISABLED AFTER REHABILITATION POPULATION SECTOR
Individuals enter the Short-Term Disabled After Rehabilitation
(STDISR) Population by reinjuring themselves after having received
rehabilitation for a previous injury. As in the case of STDIS, most
individuals go back to work after three months, and thus, back to
REHAB, The small, misdiagnosed percentage that does not, is
transferred to the Long-Term Disabled After Rehabilitation population.
3.1.7. LONG-TERM DISABLED AFTER REHABILITATION POPULATION SECTOR
The Long-Term Disabled After Rehabilitation (LTDISR) Population
include those people who are disabled for more than three months and
who have been rehabilitated at least once before. Individuals can
either recover and go back to REHAB, or, if their disability lasts
more than one year, become permanently disabled. The proportion of
individuals from LTDISR who become permanently disabled is greater
than the proportion of individuals who become permanently disabled
from the LTDIS population due to a greater proportion of chronic LBP
sufferers in the REHAB population.
3.1.8. THE PERMANENTLY DISABLED POPULATION SECTOR
The Permanently Disabled (PERMD) population consists of those
individuals who have not worked for at least one year. PERMD can be
entered only from LTDIS and LTIDISR. Evidence shows that the
probability of symptomatic relief and return to work at one year is
20%, and for more than a year the probability is nil (Beals and
-136-
The Cost Containment sector is "driven" by a combination of the total
cost to society and the number of permanently disabled individuals
(i.e., the size of the PERMD population). This combination affects the
level of funding made available for research and education. Different
levels of funding achieve different increases in the effectiveness of
all treatments and rehabilitation programs, including better
prevention. The table functions linking total costs and the number
permanently disabled to funding levels, and funding levels to increase
effectiveness of treatments have been assumed, initially, to be
linear.
Qa
= =) 4 a=
Ee roe fi
+ FIGURE & Cost Containment Sector (Original Assumptions)
4. ASSUMPTIONS
The purpose of the model is to explore the economic impact of
different policies to fund research and education to prevent and
lessen the incidence of LBP. In order to achieve this purpose, the
model permits changing several types of assumptions. Different
scenarios can be developed by changing the assumptions of: a) the
demographic, occupational and lifestyle profile of the population, b)
the effect of differently designed workplaces, c) the driving forces
of federal and private funding for research and education in LBP, and
d) the relation between funding and the increased effectiveness of
various interventions.
~137-
The assumptions behind changes in the demographics of the population
and the advent of the factory of the future affect directly the rates
at which individuals are hypothesized to get injured, reinjured and
disabled. For example, since the general population is getting older
as a whole, and since the likelihood of having LBP increases with
age, it is fair to assume that the incidence of LBP will tend to
increase. On the other hand, the population is much more aware of the
importance of fitness, and since level of fitness is negatively
correlated to the likelihood of LBP disability, the trend toward
lighter foods, less smoking and drinking, and more exercise could
reduce the incidence of LBP.
Changes in the occupational profile of the population and the general
working conditions may also have an effect on LBP. For example, asa
greater proportion of the population is employed in delivering
services, and as the industrial workplaces continue to be changed by
automatization, robotics, and better ergonomic design of workstations
and machinery, it can be expected that LBP disability from injuries on
the job will tend to go down. Also, the likelihood of LBP disability
has been negatively correlated with years of education. Accordingly,
as the educational level of the general population increases, a
decrease in disability due to LBP can be expected.
An important factor influencing the rate of rehabilitation, and thus
disability, is the compensation law. The more generous’ the
compensation and the more lax the criteria to qualify for it, the less
likely it is that people will go back to work. Given the nature of
LBP, malingerancy is fairly common. Consequently, it can be expected
that other economic factors may affect the "willingness" of an
individual to become "disabled". For example, the rate of unemployment
(the more difficult it is to find a job, the less likely that someone
would let it go) and the federal and state deficits (the larger the
deficits, the more stringent the compensation laws) can be expected to
influence the disability rates.
Another relevant set of assumptions is the one dealing with the forces
that drive the funding of research in this area. At present, the model
assumes that funding is a function of the total costs incurred by
society and the total number of permanently disabled individuals. The
initial form of this function consists of an "urgency" index
determined through a weighted average which assigns, quite
arbitrarily, total costs four times as much weight as it does to the
number of permanently disabled individuals. A delay of five years has
been introduced to reflect the lag between the time the "urgency"
index is known and the time when the funds are made available (see
Figure 3). A factor that needs to be incorporated in the model is the
impact of government deficits, unemployment rate, and other economic
activity indicators on the political will to allocate funds for LBP
research,
~138-
Funding of LBP research is assumed to increase the effectiveness of
not only medical treatments but also rehabilitation programs,
educational efforts, and preventive interventions. As _ the
effectiveness of these programs increases, the injury, reinjury, and
permanent disability rates decrease, and the time needed for
rehabilitation is shortened. A delay of five years between the time
when the funds are made available and the time when the population
starts feeling the effects of improved treatments, has been
incorporated.
Finally, the model assumes that the birth and death rates are equal
for all the populations with the exception of the death rate for the
Permanent Disabled (PERMD) population. While this population is less
exposed to life threatening situations (e.g.,driving), they tend to
exhibit higher rates of depression, alcohol and drug abuse, and low
cardiovascular fitness leading to ahigher death rate for this
population.
5. CAVEATS
We are in the process of refining the model and are starting to
compare different policies. The model has proven to be well behaved
and stable. It has been tested under extreme conditions and has shown
remarkable robustness.
However, trying to understand the economic effect of increased funding
for research and education in LBP is fraught with inherent
difficulties. To a large extent, these difficulties are due to the
lack of accurate data relating to the various injury rates, to the
costs attributable to the different categories of disability, to the
effect of better prevention, treatment and rehabilitation on the
specific populations, etc.. While establishing base-rate data is
essential in this type of modeling, very little information is
available, and what is, is often in formats that do not match and is
often contradictory.
Another peculiarity that makes modeling LBP difficult, is the
smallness of a number of critical coefficients leading to specific
rates. For example, the rate of injury from the Healthy Population
(HP) to Long-Term Disabled (LTDIS) is only .0008. A minor change in
this constant will appreciably alter the LTDIS population and thus
the costs associated with that population. In turn, this change will
produce a large change in the funding available for research and
education and this will affect other variables as the various
feedback loops adapt to the altered injury rate.
Until accurate data are available for all the sensitive constants in
the model, the simulation should be studied with regard to general
system behavior rather than be used to forecast specific numerical
results,
-139-
REFERENCES
Beals, R. K. and N. W. Hickman, “Industrial Injuries of Back and
Extremeties: Comprehensive Evaluation - Aid in Prognosis and
Management: Study of One Hundred and Eighty Patients," Journal of Bone
Joint Surgery, Volume 54A, Number 10, 1972, pp. 1593-1611.
Cats-Baril, W. L., "Developing a Predictive Model of Low Back Pain: A
Deductive Methodology," Vermont Rehabilitation Engineering Center,
Working Paper, November 1984,
Economic Report of the President, U.S. Print Office, Washington, D.C.,
1983, pp. 172-179.
Frymoyer, J. W., State-of-the-Science: Low Back Pain, Manuscript,
Department of Orthopaedic Medicine, University of Vermont College of
Medicine, 1984. (C.V. Mosby Co., in press).
Frymoyer, J. W., Pope, M. H., Clements, J. H., Wilder, D. G.,
MacPherson B., and T. Askikaga, "Risks Factors in Low-Back Pain. An
Epidemiological Survey," Journal of Bone Joint Surgery, Volume 65A,
Number 2, 1983, pp. 213-218.
Kelsey, J. L., White, A. A. III, Pastido, H. and G.E. Bisbee, "The
Impact of Musculoskeletal Disorders on the Population of the United
States," Journal of Bone Joint Surgery, Volume 61A, Number 7, 1979,
pp. 959-964.
Luginbuhl, W. H., B. R. Forsyth, G. B. Hirsch, and M. R. Goodman,
"Prevention and Rehabilitation as a Means of Cost Containment: The
Example of Myocardial Infarction," Journal of Public Health Policy,
Volume 2, Number 2, 1981, pp. 103-115.
Murphy, K. A. and R. D. Cornish, "Prediction of Chronicity in Acute
Low Back Pain," Arch Phys Med Rehabil, Volume 65, Number 2, 1984, pp.
334-337.
National Center for Health Statistics, Series 10, No.134, Prevalence
of Selected Impairments, United States 1977, DHSS publication (PHS)
81-1562, Hyattsville, MD, 1981.
Snook, S. H., and R. C. Jensen, "Cost," In Pope, Frymoyer and
Andersson (eds.), Occupational Low Back Pain, Praeger Press, New York,
N.Y., 1984,
Vladeck, B. C., "Medicare Hospital Payment By Diagnosis-Related
Groups," Annals of Internal Medicine, Volume 100, Number 4, 1984, pp.
576-591.
White, A. A., “Low Back Pain in Men Receiving Workmen's compensation,"
Canadian Medical Association Journal, Volume 101, Number 1, 1969, pp.
61-67.
CC BY-NC-SA 4.0