TH E AUSTRALIAN

JOURNAL OF PHYSIOTHERAPY
VOLUME 2 JULY, 1956 NUMBER I

RESEARCH IN PHYSIOTHERAPY.
By BRYAN GANDEVIA, M.D., M.R.A.C.P.

THE PRINCIPLES OF CLINICAL RESEARCH WORK.

Physiotherapy shares with medicine a Physiotherapy, as its name implies, is

common origIn, springing ultimately from entirely concerned with treatment; it may
man's need to do somethtng for himself or be of value to discuss the ways In which
fOl his fellows when inj ury or sickness different forms of treatment may be
leads to inability to fight or to work. Some assessed and compared. Basically, a satis-
idea of the relative value of various thera- factory therapeutic measure is one which
peutic measures was inevitably gained by produces improvement in the condition of
experience and the more lethal procedures the patient either more rapidly or more
were eliminated. Experience, however, can completely than would be expected if
be fallacious. Those of us who are charged the disease were allowed to follow its
with the care of the sick in any capacity natural course. That thought prompts us
whatever have a duty and responsibility to to pose the first problem: what is the
study and to investigate all possible methods natural history of the disease in question?
of treatment for the benefit of our patients. Usually we know whether death or
There are many facets to the study and recovery is likely; if we know that much
investigation of existent or proposed thera- definitely, a system of treatment which pro-
peutic prlJcedures. Many sciences may duces the reverse effect to that expected,
make contributions to our accumulated even in a single case, can readily be
knowledge of the normal structure and assessed. Thus the cure of one case of a
functions of an organ or a part of the body, disease which is inevitably fatal establishes
and to an understanding of the disorders the value of the treatment used; the first
consequent upon a certain disease. Some cure by penicillin of subacute bacterial
of our methods of treatment are soundly endocardItis established the value of peni-
based on this knowledge, but others remain cillin treatment in that disease. In the field
empirical at present; physiotherapists know of physiotherapy, however, the use of death
of procedures in each of these categories. as an index of therapeutic efficacy, for-
Once a method of treatment has been tunately, is rarely possible; usually the
devised, it remains to be shown that it is index will be the time taken for recovery,
safe, that it is effective, and that it is more or the degree of restoration of function
effective or less effective than are other which is obtained by the method of treat-
methods in use for the condition under ment adopted. In any given case these
consideration. Sometimes the study of the indices are but rarely predictable with any
way in which the treatment works throws accuracy when based on experience alone;
light on the mechanism which is producing if we are honest we cannot claim with cer-
the disorder. tainty that our treatment was helpful, or,
6 THE AUSTRALIAN JOURNAL OF PHYSIOTHERAPY

at all events, we cannot precisely express for that dIsease, we would probably be
how helpful it was. Ideally, we need as prepared to take the slight risl( involved
a "base-line" an assessment of the dura- and would continue to prescribe radio-
tion and degree of recovery In a series therapy. It probably is of value, but there
of cases of a disease when no treat- is a difference of opinion on the point
ment was given. IncIdentally. we should amongst the authorities; unfortunately, no
remember to define the selected disease conclusive controlled investIgation to settle
carefully and we may have to divide it into the matter has ever been performed.
sub-groups; for example, sprained ankles Assuming for the moment that the risl{ is
should be divided into subgroups such as real and that the value of radiotherapy is
those involving the lateral ligaments only dubious, it is clear that there would be
and those associated with fracture of the justification for the treatment of some of
lateral malleolus. Nowadays this is usually the patients with a placebo to form a com-
unjustifiable on ethical grounds, as it would parative group; thus finality might be
mean the withholding of treatment which reached concernIng the advisability of con-
is firmly believed to be beneficial. Besides, tinuing to use radiotherapy for ankyloslng
an untreated group would not be ideal for spondylitis.
comparison with a series of treated patients; The history of medicine records the use
any mode of treatment, however irrational of thousands of futile remedies, including
and relatively useless, is usually better than such esoteric substances as unicorn horn
no treatment. Thus do the quacks prosper: and dragon's blood, as well as powdered
thus do we all find our symptoms a little Egyptian mummy and lTIOSS from a dead
better for some sympathy. The ideal group man's skull, each of which has been of
for the purposes of this comparison is a great repute for centuries before experi-
series of patients who have, in good faith, ence led to its dismissal at last. It would
undergone a form of treatment which is appear, then, that a new treatment should
unlikely to affect the course of the disease be subj ected to a thorough trial before
(placebo); but, again, unless the patient coming into general use and well before it
consents to take part in the experiment, this passes out of the spotlight of its newness
ideal is unattainable ethically. into the semi-obscurity of everyday use.
The decision to use a placebo is invari-
ably to be made by the doctor. On occasions Relative Merits of Two Methods of
this course is not only reasonable, but it Treatment.
may be the proper one to adopt. It is not Having considered in principle how the
proposed to discuss this question fully here, efficacy of a new therapy is gauged in com-
but to give an illustrative example. In a parison with ineffectual treatment which
chronic relapsing disease for which no cure does not disturb the natural course of the
or beneficial treatment is known, it may be disease, our next step is to plan the deter-
desired to find out whether a newly intro- mination of the relative merits of two
duced form of therapy is of value. It is methods of treatment. It can be shown that,
vital in a case such as that to obtain a in most circumstances, it is not reliable to
clear-cut answer if possible, otherwise new compare the actual results of one form of
hope and expensive treatment may be given therapy against the impression of results
to millions of patients who may ultimately obtained hitherto by some other method.
have to suffer disillusionment. Here Two series of patients should be treated
the finality of the results will justify a concurrently-one in the old way and the
properly controlled trial of the new treat- other in the new-and at the end of a cer-
ment. Evidence has recently been produced tain time a comparison of results should be
which suggests that patients with anky- made. No ethical problem arises if it is
losing spondylitis if treated by radiotherapy unlikely that the new therapy will be harm-
are a little more likely than are the members ful and also if it is reasonable to expect it
of the general population to develop to be beneficial. The great responsibility
leukremia. If ,ve were certain that this is to design the trial to ensure a definite and
form of treatment was of undoubted value final answer. It is preferable to replace the
 RESEARCH IN PHYSIOTHERAPY 7
fa1l1ble sense of impression which is called It may be desired that a number of thera-
experience by the less equivocal technique peutic systems should be compared at the
of a controlled clinical trial. Mere personal same time. In certain conditions, where
opinions do not and should not convince the purpose of treatment is to obtain tem-
others, but the results of a proper clinical porary relief of symptoms rather than to
trial are convincing. effect a complete cure with removal of the
It is more difficult to compare the effects underlying cause, as in asthma or arthritis,
Df two methods of treatment than it is to several forms of treatment may be tried in
compare the results of one method against succession on each patient. This method is
those in an untreated or placebo-treated ap~lic~bl~ wh~n treatment is effective only
grol1p. Since it is assumed that each of whl1e it IS beIng used, and when its with-
two effective methods will be beneficial drawal is followed by reversion to the
the differences between the results obtained previous state. Suppose four types of treat-
are likely to be smaller; it is therefore me~t are .to be t.ested: fOUf groups of
advisable to use larger groups to diminish patIents wIll be given the fOUf different
the risk a f chance occurrences and to ensure forms of treatment simultaneously; at the
t?at. the observed differences will carry end of a given time the groups WIll be
SIgnIficance. The number of cases required rotated and the rotation will continue until
depends on the nature of the problem which each group of patients has received each
is the subject of the triaL A large series type of treatment. This kind of trial has
is required when it is anticipated that the certain advantages: the total number of
differences \~l l1 be relatively small or when patients required for the test is smaller than
the possible methods of assessment lack for other kinds, and it is clearly reasonable
precision. Thus the cure of even one patient to ~raw comparisons, since the very same
with a disease which was previously invari- patIents have been subjected to each form
ably fatal is highly significant; but physio- of treatment.
therapists will seldom, if ever, be con- The Selection of a Research Project
fronted with such an easy problem. For
example, it would require a great many Having discussed the main principles
cases to assess significantly the relative underlying the clinical testing of methods
merits of prolonged postural drainage on of treatment and having emphasized the
the one hand or of brief periods of postural need for a controlled investigation, we will
coughing on the other in diminishing the pass to the consideration of the selection
number of attacks of pneumonia in bronchi- and definition of the problem to be '5tudied.
ectasis. To show which was the more effec- Although there are many projects for
tive in reducing sputum volume a smaller investigation in sciences such as anatomy
number of test cases would probably be and physiology, on which physiotherapy is
sufficient. Sputum will be present in some based, we shall continue to confine our-
degree in each case and it can be me::tsured selves to problems of treatment. These
easily and accurately; comparatively few of are the daily concern of all physio-
the patients would have attacks of pneu- therapists in the care of each patient. It
monia, and it might be difficult to decide is OU~ aim to show that this everyday
what constituted an attack of pneumonia experience could be translated into scientific
for the purpose of the test. When progress study at but little cost in time, money,
must be gauged by changes in subjective materials, and appliances.
sensations still greater difficulty is pre- It is easy to select a subj ect for investi-
sented. It is hard to assess quantitatively gation; a dozen will suggest themselves in
~uch things as pain or sense of well-being; a single day if one pauses to think and to
tn asthma, for example, comparisons of ask oneself the "how", "why", "when",
methods of treatment may depend on varia- and "where" types of questions and, for
tions in severity and in frequency or in our present particular purpose, "which?".
duration of the attacks. The problems One must be brutally frank about it. We
related to the assessment of progress will must refuse to accept as a satisfactory
be considered later. answer any statement of opinion or text-
8 THE AUSTRALIAN JOURNAL OF PHYSIOTHERAPY

book authority which is not supported by a series of questions framed in such a way
quoted factual evidence. Do not reject a that as little as possible is left to chance.
subject only because you feel that i~ is This facet of research work may best be
beyond your resources or your capacity; illustrated by consideration of some possible
some aspect of tIle problem can be found projects. We will take three and call them
to which even the humblest of us can con- respectively the varicose ulcer project, the
tribute something. Do not think that osteoarthritis project, and the poliomyelitis
because the subject is not a new one there project.
is nothing new to be found out about it.
It is given to very few to think of an The Varicose Ulcer Project.
entirely new idea or to devise a wholly Let us first imagine that it is desired to
original experiment. Most research, like know the relative merits of Method A and
most detective work, in real life, is based Method B in the treatment of varicose
on a routine series of inquiries, commencing ulcers. The question might be put: "Do
with the most obvious questions and build- varicose veins heal better on treatment by
ing the story up a step at a time. A or by B?" Unfortunately, simple as it
Other subj ects for investigation present sounds, it is the wrong question; it would
themselves as we read our journals. No probably get the answer, "Maybe". What
paper ever supplies the last word on its is meant by "better"? What is meant by
subject and, often indeed, the problems still "heal"? What is the answer if none of the
to be dealt with are indicated in the paper ulcers heal? Are we then to assume that
itself. It is quite justifiable to repeat the both methods are valueless? Or is it that
work done by someone else; all original treatment has been too brief? Let us avoid
work requires to be confirmed by indepen- these complications by posing a series of
dent observers before general acceptance. more specific questions at the outset. If
Profiting by published experience, one may two comparable groups of patients are
devise a simple experiment not only to treated, one by A and one by B, for one
confirm or deny the original conclusions, month, does one group show a significantly
but also, perhaps, to add some additional greater reduction in the total area of
information. It is worth while to remember ulceration than the other? Does one group
that a form of treatment which works in contain a significantly higher number of
England may not work in Australia; a patients in which the area of ulceration
therapeutic measure may have different diminishes-or increases? Does one group
values in different parts of the world for contain a significantly higher proportion of
a host of reasons, including differences in patients in whom no area of ulceration
climate, in the habits of the patients, or in remains?
the course of a particular disease. What we have termed the wrong question
it is best to regard as the general problem.
The Posing of Appropriate Questions. Our research proj ect is designed to provide
More important than the selection of a unequivocal answers to certain specific and
subject is the posing of the appropriate readily defined aspects of the problem: It
question to be answered by an experiment. will be seen that for "heal" and for "better"
Herein lies the secret of effective research we have substituted a measurable charac-
work, be it by a person working alone with teristic-the area of ulceration. We have
few facilities or by a research organization arbitrarily chosen a specific period of one
backed by all the resources of a modern month because we urgently require a guide
hospital. Precise questions upon the to the treatment to be adopted generally
selected subject must be formulated which pending further investigation, or for some
are capable of specific answers by delibe- other reason. The time chosen might
rately planned experimentation. A clear equally well have been three months or six
answer cannot be obtained to a question months; and we certainly would extend the
which is vague or lacking in definition. period of the trial if the results were incon-
Problems should therefore be analysed into clusive after one month.
 RESEARCH IN PHYSIOTHERAPY 9

The questions already set out are by no receive careful consideration in the initial
means the only ones that might be asked stage of planning.. The resources to be con-
in regard to the general problem, but they sidered should include manpower, time,
have been deliberately related to the finance, suitable patients, suitable diseases,
resources available in the conduct of the and appropriate technical methods. It may
experiment, say, by an individual physio- be difficult for a physiotherapist to organize
therapist working in a busy private practice.. and conduct clinical research work of this
Elaborate techniques and time-consuming nature without the collaboration of a doctor;
methods might be impossible in the circum- but it is not undesirable that she should
stances but, by basing the experiment on conduct it on her own account if she frames
comparatively simple, clear-cut questions, a the questions appropriately with due regard
useful return of information is assured for to the circumstances in which the work will
little, if any, extra outlay in time and be done..
money.. A more ambitious project would
probably have led to inconclusive results or The Osteoarthritis Project.
to a breakdown in administration, and thus Assuming for the moment that we can
to disappointment and a tendency to relapse define and measure the term "better" , a
into blind reliance on the opinions of other question such as "Is A better than B in the
people as our guide in practice.. treatment of osteoarthritis of the knees?"
is one which is likely to require the col-
. The random allocation of the patients
laboration of a physician with a physio-
lnto one group or the other is not trouble-
some-it can even be done by tossIng a COIn.
therapi~t if the conclusions are to carry
full weIght.. On the other hand, the solution
Having thus ensured that the two grottps
of the question "Does A produce greater
are comparable, it is no trouble to adhere
power in the quadriceps muscle than does
rigidly to one or other of the forms of
B in. a given p~riod of time ?" is purely a
~reatment which have been carefully planned
physIotherapeutIc problem. It is also an
In advance. As the two groups are treated easier question to solve in that the index
concurrently, the influence of such
of progress can be accurately measured in
extraneous and uncontrollable factors as
foot-pounds of work, whereas the standards
the state of the weather or the intervention
of "improvement" in osteoarthritis would
of a transport strike will affect each group be difficult to define.. Nevertheless the diffi-
equally. The element of personal bias in· culty alone should not deter a physio-
favour of one of the methods of treatment therapist from tackling the more general
is minimized by using something which can questIons under suitable conditions..
be measured accurately as the index of
progress. The Poliomyelitis Project.
When the opportunity arises to conduct ~inally, let us. discu~s a more complex
an e:x:periment of th~s kind in a teaching proJec! and o~thne brIefly. how a physio-
hospItal, some more dIfficult questions could therapIst workIng alone might approach it
be added, requiring more elaborate tech- and how it might be tackled when the
ni9ues. A pathological definition of healing resources of a department in a teaching
might be adopted, and a pathologist set the hospital are available. Let us suppose
task of deciding the degree of healing from th~t there are two schools of thought
a study of histological sections. The blood wIth. ~egard to .the management of polio-
flow to the limb before and after treatment myehtls : one whIch we may call the "active"
might be regarded as relevant in the assess- school, and one the "passive", dependent
ment of the efficacy of the respective on the physiotherapeutic regimen employed
methods of treatment. Finally, the inquiry by each. The general problem is to decide
might be extended to include the relapse which form of treatment produces the better
rate in the two groups over a lengthy period end results. Obviously that problem would
of not less than one year. require an elaborate research programme
The relation of the questions asked to the \vith a "follow-up" which would probably
resources available is one which must cover several years. As before, however"
10 THE AUSTRALIAN JOURNAL OF PHYSIOTHERAPY

it can be resolved into a series of more servative regimen or to practise some

specific questIons providing a variety of reasonable compromise.. The history of
aspects each of which might fall into the medicine is full of opposing theories which
province of a different specialist. have resulted in the adoption of some
middle course which turns out to be nearer
The lines along which this proj ect mIght
the truth than either theory.
be developed are worth indicating here, as
this is the type of project upon which every However, by posing an entirely different
major department should be engaged, set of questions, the lone physiotherapist
especially when the work in the department may make a significant contribution to the
deals predominantly with one disease or a general problem as well as to the solution
group of similar diseases. A physician or of her personal dilemma" A difference in
team of physicians may be asl<:ed to grade the efficacy of the two methods of treat-
the improvement of the patients, taking ment may become apparent within a few
into account factors such as general health, months, as manifested by a difference of
activities, and residual signs and symptoms.. improvement in muscle power" On these
The grading would be done periodically, assumptions a relatively simple question
say at the end of three, six, and twelve may be based: "Is improvement in the
months, two years, and five years. The power of paresed or paralysed muscles at
speci~list in physical medicine may assess the end of three months greater in a group
physIcal status, posture, and deformity.. of patients treated by the 'active' method
The orthopredic surgeon may be called upon than it is in a comparable group treated by
to analyse the corrective surgical procedures the 'passive' method?"
r~quired i? each group of patients.. Ques- Let us examine this question more closely..
tIons bearlng on the progress of the indivi- "Improvement" implies assessment before
duals could be devised to be answered by a and after treatrnent~ and indicates that we
radiologist, a psychologist, and a school- are concerned not with the absolute values
Inaster. Many difficulties would arise in in each instance but with the differences
deciding upon the indices of progress which between them. Thus It may be assumed
should be used, but it is probable that those tentatively that any discrepancy in the
difficulties would be surmounted.. Clearly, degree of paralysis existing in each group
change of muscle power, one of the chief of before treatment will not influence the
these, would be within the province of the ·results of the test. It is possible that this
physiotherapist.. The latter might also conclusion may prove to be unjustified; we
supply information based on electromyo- may have to compare only muscles which
graphic and other recordings. are paralysed to a similar extent in each
grottp at the time of commencement of the
The weakness of an elaborate investiga- comparison of the two methods. "Power"
tion of this nature lies in its complexity and is capable of accurate measurement in
its diffuseness.. Its success would depend terms of the work each muscle can do..
chiefly upon the care with which the sub- Measurement in "foot-pounds" or similar
sidiary questions were framed, upon absolute units is to be preferred to the
detailed planning in advance, and upon system of "grading" muscular power; the
efficient administration. In other words, the latter system brings in the subjective
rules are exactly like those we have dis- element of personal bias, though in some
cussed for the simpler studies. circumstances "grading" may be more prac-
But what form of treatment is the ticable. It may be easier to measure the
i~di~idual therapist to adopt before the pub- change in the power of a movement rather
l~catlon of the result of the big investiga- than of a single muscle; for example, the
tIon by the team of experts? Having made power of flexion at the elbow joint may be
a critical appreciation of the two systems used as an index of the functional power
under consideration from knowledge of the of the biceps muscle, though, of course,
natural history of poliomyelitis and of other muscles assist the biceps in this
previously published work, the physio- movement. The term "paresed or paralysed
therapist is likely to follow the more con- muscles" may well be qualified by speci-
 RESEARCH IN PHYSIOTHERAPY II

fying certain muscles or muscle groups to influence the recovery of paralysed muscles.
which the investigation is limited because Even when we have fairly large numbers
they are muscles whose power and function of patients in the groups, comparability in
can be readily assessed. Alternatively, or terms of extent and severity of paralysis
as a complementary approach, the investiga- will not necessarily be achieved; to some
tion could be based solely upon muscles extent tl1is is obviated as a disability by the
that are completely paralysed at the outset. fact that we are comparing differences in
This has the advantage of the "yes, no" improvement and not differences in the
type of answer-"some recovery" or "no final strength of the muscles concerned.
recovery"-but the disadvantage that most Further, our question is related to the
paralysed muscles are likely to recover some change in power of muscles and not to the
function. In these circumstances the general improvement of the patients; the
expected differences between the two groups experiment is not designed or intended to
become smaller, and a correspondingly deal directly with the latter aspect of the
larger number of patients will be required progress of the cases.. It is limited to
to get the clear-cut conclusion of a signifi- lTIuscles and not extended to patients on
cant difference. Nevertheless it is worth purpose, as we recognize that one physio-
while to remember to analyse the final therapist cannot treat more than a given
figures in this way speculatively. The number of patients in the time available
"active" and the "passive" methods of for the experiment. By dealing with
treatment must be defined carefully and we muscles only it is likely that the number of
must adhere to the planned regimens observations made and available for COlTI-
strictly. The period "three months" is an parison will be substantially increased. It
arbitrary one; assessment could well be may be reasonable for us to hope that our
made at one, three, and six months, to cover results for individual muscles or muscle
the possibility that one method might pro- groups will be applicable to all muscles
duce improvement faster than the other. and thus to patients as a whole.
The precise definition of the time at which By reduction of the question to its com-
treatment begins is perhaps more important ponent parts in this manner it can be
than the duration of the observations; it demonstrated that the project is not beyond
might be defined as Hwhen referred by the the scope of an experienced physiotherapist
physician" for the purposes of the physio- ~orking alone. The fireside task of plan-
therapist, but it would be preferable to give nIng and organizing is practically the only
~h~ . medical status of the patient at the addition to the work of treating the patients.
InItIal stage, for example, in terms of the The planning of the experiment would be
temperature and pulse charts. Whatever sounder and the results would be more con-
criteria are adopted, we must make sure vincing if a colleague, unaware of the
that they are applied similarly to each method of treatment used for individual
group. The term "comparable groups" has patients, performed the muscle testing
already been mentioned and will be specifi- independently, and thus eliminated the
cally discussed later on. Comparability of element of bias which is always liable to
the groups to a reasonable degree will prob- infl uence our findings. ..
ably be. achieved in regard to age, sex, The project which has just been described
occupatIon, and severity of initial illness, would not supplant the lTIOre elaborate
by the random allocation of the patients to investigation previously outlined. Inevit-
one or other of the groups. If that should ably it would contribute something to the
prove not to be the case because the number solution of the major problem, it would act
of patients is insufficient, the experiment is as a guide to immediate action, and it would
not automatically invalidated; it means prevent the wide adoption of a form of
merely that, when we draw conclusions treatment, which might be useless or
from our results, the possible influence of positively harmful, through misguided
the factor of unequal distribution must be enthusiasm"
taken into account. It is conceivable for The complex problem of poliomyelitis
instance, that age of the patient doe; not was selected deliberately because it bristles
12 THE AUSTRALIAN JOURNAL OF PHYSIOTHEF..APY

with difficulties, and thus is as complIcated we have to assume that osteoarthritis gives
a problem as we are likely to encounter. rise to fewer symptoms when the quadriceps
We have tried to show that even those muscle is strengthened, or some such
difficulties can be overcome by reduction assumption must be made. This is in fact
of the question posed to a level at which the hypothesis upon which this form of
the project could be tackled confidently by treatment is based; it seems to be a reason-
any of us. able theory, but it would require quite a
different experiment from the one described
The Limitations of Experiments. above to prove its truth. A similar assump-
Reference has been made earlier to some tion is required to enable us to apply the
of the ethical and practical limitations to results of the experiment of the lone-handed
experiments involving human beings. It physiotherapist to the general management
is necessary here to emphasize that research of muscles damaged by poliomyelitis.
work does not modify in any way our These limitations are perhaps obvious,
approach to an individual patient, though it but they are emphasized because it is sur-
affects our approach to the management of prising how frequently enthusiastic inve"ti-
a group of them. Some limitations have gators translate their experimental results
already been indicated in discussing diffi- into sweeping conclusions. These con-
culties that have to be faced. One further clusions are usually printed in a summary
point requires emphasis: the answers to the at the end of a paper; as the summary may
questions we pose are merely answers to be the only section of the paper which is
those questions; they do not provide auto- read by uncritical persons, the stated con-
matic and direct answers to what we have clusions may be given more weight than
termed the major problem. To derive the they deserve.
answer to the major problem from our There is an objection which the writer
findings it is usually necessary to make a senses is already rising in the reader's
further assumption with the risk that it mind : "You would need a lot of cases."
Inay not prove to be justifiable. Suppose However, we have already pointed out that
that we have found that varicose ulcers the number of cases depends on the size
show a greater diminution in area of ulcera- of the difference between the two groups..
tion in one month when treated by Method A large difference will be demonstrated in a
A than they do when treated by Method B smaller series, but a small difference will
under comparable conditions. If it is need a larger number of cases before it
assumed that decrease in size of an ulcer tecomes demonstrable. In the words of
means better healing eventually, we can John Hunter, "Don't think; try the experi-
answer the "major problem" with the state- ment". If at the end of your planned work
ment: "Better healing of varicose ulcers is there is a difference which is not quite
achieved by Method A than by Method B". great enough to be significant, continue the
It will be recalled that we avoided the words experiment. And if the experiment gives
"better" and "healing" in planning the the disappointing or the unexpected result,
experiment chiefly by making this very so much the better, for the next experi-
assumption. The possible invalidity of the ment - to find out why - is immediately
assumption was implied in the suggestion before you.
that a more elaborate investigation should
include an assessment of the relapse rate The Planning and Administration of
at the end of a year. Clinical Trials
Again, suppose that it is easy to prove Continued emphasis has been laid on the
that Exercises A produce greater improve- importance of the planning and administra-
ment in the power of the quadriceps muscle tion of these clinical trials because little
than was obtained by means of Exercises B additional work beyond the routine is
over a given period of time. To make the required for the actual experiment. Treat-
further claim that the findings show that ment is carried out as usual, although two
Exercises A are to be preferred to Exer- or more methods are being used concur-
cises B in the treatment of osteoarthritis rently. Perhaps progress is assessed more
 RESEARCH IN PHYSIOTHERAPY 13

meticulously than would otherwise be affected in a certain area in a given tilne.

necessary or customary, though in other Even then our results would only be valid,
respects no special techniques may be in theory, for that area at that time; and
employed.. The value of the results obtained the number of cases might easily be
and the validity of the conclusions depend influenced by the time limit or the season
almost entirely on the care with which the selected. In our planning, however, we
trial is planned. must be practical; since we cannot study all
Though the writer disclaims any detailed of the cases we want to find a representa-
knowledge of statistics he would strongly tive sample after the style of "cross sec-
recommend that the advice of a statistician tion" used for the Gallup polls. This again
or of an experienced biological research may not be entirely practicable, as some
worker be obtained to assist in the selection has already been effected in the
planning of any controlled clinical triaL process by which the patients are referred
We have already indicated why this form for physiotherapy. Inevitably, therefore,
of trial is to be preferred, andall that need we have available only a portion of the
be said here is that its use leads us to a {(population" with varicose ulcers or osteo-
definite rather than to a vague answer to a arthritic knees; and that portion may not
given question. Statistical methods are a be a true usample" of the "population".
means to an end, not an end in themselves; The significance of our results will be
they are only as valid as the data to clear provided that we recognize these
which they are applied, so we must make selective factors and define the way in which
certain that the data used are accurate selection is operating. FaIlure to allow for
representations of the true situation. this pre-selection may lead to error when
We repeat that the first steps of planning we apply our findings; we would be liable
are the selection of a problem and of the to claim them as applicable to all ulcers
specific questions to which we require instead of restricting the conclusions we
answers. The questions are related to the have made to the comparatively severe
resources and limitations of the individual examples of varicose ulceration which are
or of the staff of a department who will directed to the special department of a hos-
be conducting the experiment. The essen- pital. I f we find in our series of ulrers that
tial feature of a controlled trial is com- the average healing time is three lTIonths,
parison. The requisites for comparison it does not mean that the average healing
are: (I) that the two or more groups to be time of all ulcers is three months, even if
tested are suitably homogeneous in all treated in the same way by the same people
respects other than the one variable which in the same place. Having shown that
is to be compared; and (2) accurate assess- Method A is superior to Method B, it may
ment of the influence of the variable on be reasonable to assume that Method A is
the progress of the case.. We will now better than Method B for the treatment of
proceed to discuss in more detail the com- all ulcers under the conditions of the
parability of the groups and the assessment experiment. If those conditions are altered
of progress; these two factors are to be the assumption may be invalidated; thus a
considered as vital to the success of the system of treatment effective in a hospital
trials. may be relatively ineffective in the con-
ditions of general practice.
The Comparability of the Groupso It must also be appreciated that failure
The two chief factors in ensuring com- to study this process of selection may result
parability of the groups are the selection of in a failure to recognize that the composi-
cases and the mode of conduction of the tion of one of the groups of cases is being
experiment itself.. influenced more than that of the other. If
the patients attending on Monday are all
The Selection of Cases. treated by Method A and those attending
Ideally, in our experiment on varicose on Tuesday are all treated by Method B, a
ulcers or on osteoarthritic knees, we would pair of groups may not be comparable; for
like to be able to treat all the patients example, the surgeon attending on Monday
14 THE AUSTRALTAN JOURNAL OF PHYSIOTHERAPY

may have some odd notion such as refer- some of the reference books. Occasionally,
ence for physiotherapy for varicose ulcera- ClrCtlmstances may require the use of a
tiOll only of those patients with ulcers on method of pairing whereby the inclusion
the medial side of the leg; and perhaps of one case in one group of the trial calls
Tuesday's surgeon refers only those who for the inclusion of a case which is similar
have failed to derive benefit from other in respect of some defined characteristics in
measures. The possibilities are countless the other group. This method of pairing
and all that need be said is that the problem might conceivably be employed in the com-
of choice of patients for the groups to be plex poliomyelitis trial we have discussed
compared must be studied carefully for already, but its use is uncommon and it
each experiment. Even the use of volun- should not be undertaken except on the
teers instead of conscripts introduces diffi- advice of a statistician.
culties, as volunteers may be expected to While random allocation to groups within
have a different outlook from conscript'3 the series to be studied is essential, it is
on life in general and treatment in par- of course reasonable to select on specified
ticular. If one wished to select a repre- grounds exactly what type of patient and
sentative group from a class of students it what sort of disease will be included.
would be most unwise to pick the front two Indeed, what is to be included must be
rows; it has been said that, relative to the precisely defined to let others understand
remainder of the class, they may be
clearly the exact problem which is being
expected to be half-deaf, half-blind, or tackled. If we need a quick answer to the
over-anxious to please.
problem of the treatment of varicose ulcera-
In practice it is usually adequate to tion, it might be decided beforehand, for
recognize the existence of selection in the example, to exclude ulcers complicated by
cases presenting for study, to define this peripheral arterial disease, or ulcers of more
selection, and to take it into account in than five years' duration. We have already
drawing conclusions from the data col- indicated that a physiotherapist working
lected. Within the series of patients to be alone on the problem of poliomyelitis would
studied, however, it is vital that the design be wise to select for study only those
of the experiment provides that selection muscles whose performances can be readily
of patients for one or other treatment group assessed. This aspect of selection is merely
is entirely a random selection. Thus chance a matter of deciding with exactitude what
will ultimately ensure that the groups are question we propose to attempt to answer.
comparable not only in respect of size of But it is necessary to remember to recog-
ulcer but also with regard to age, sex, nize that our results apply to that question
habits, regularity of attendances for treat- only; we may perhaps assume that the
ment, care with which they carry out
results are applicable to all ulcers or to all
instructions, and a hundred and one other
muscles, but the qualification of the assump-
characteristics. More important still, par- tion must be clearly defined.
ticularly when the groups are small, it
ensures that no bias is introduced into the It must be pointed out that it is not
groups even unintentionally, as would occur, essential in every investigation of the kind
for instance, if Tuesday's rather special under consideration that the groups should
ulcers were to be compared with Monday's be comparable in every conceivable charac-
ordinary ones. teristic, though random selection would
Random selection means, then, that no ensure this if the groups were sufficiently
such selective factor can be operating. large. It may be safely assumed that cer-
Where only two groups are required the tain characteristics would not have any
selection can be made readily by tossing a influence on the healing of ulcers or on the
coin. Allocation of patients to one out of recovery of muscles, and a chance unequal
two or more groups by rotation is likely distribution of those attributes may be
to be fallacious in certain circumstances; ignored. Alternatively, the data for the
allocation is best done by referring to tables two groups may be analysed statistically
of random numbers which are published in to see whether this or that feature influences
 RESEARCH IN PHYSIOTHERAPY IS
the results significantly. If a difference is active therapeutic principles are deleted. If
thus revealed unexpectedly, a careful search the efficacy of a new method is to be tested
must be undertaken to ascertain whether it is essential theoretically to use a control
there is an uneven distribution of any group; any favourable significant difference
characteristics which were not expected to may thus be credited to the new method
influence the results when the" experiment Of, alternatively, its potential harmfulness
was planned" Whenever there are differ- may be exposed. If the new form of treat-
ences in the results, and especially when ment compares favourably wIth a placebo
those differences are unexpected, the con- system, an experiment should be arranged
stitution of the groups under comparison to compare it with a standard method.
should be scrutinized for the presence of As we have pointed out before, this means
SOlne unevenness of distribution, including that the difference between the two sets of
in the search features which it was not findings may be small and thus hard to
anticipated would influence the results. determine accurately unless large numbers
of patients are treated.
The Mode of Conduction of the Experi- When two groups of patients are sub...
ment. j ected to two entirely different forms of
Once the groups themselves are con- treatment, both of which are genuinely
sidered to be comparable, it is necessary to regarded as efficacious, it is not necessary
make certain that their management in the to impose on each patient an imitation of
experiment is also comparable. One group the opposite regimen in addition to the
must not receive more careful or thorough effective therapy. If that were done, how-
treatment than the other one. If the work ever, psychological influences inherent in
the regimen might be eliminated; in prac-
is being done by more than one therapist,
personal bias in favour of a method of tice this facet is usually only of academic
treatment may be avoided by arranging interest, which is fortunate, for in the field
of physiotherapy it would frequently be
that each of the physiotherapists treats
found to be impl'acticable to include it.
patients in each group or, alternatively,
It is patently impossible to make sure that
that the supporters of one regimen should
a group of persons with sprained ankles
practise it alone on a group and that equally
treated in plaster casts is comparable with
enthusiastic exponents of the rival method
another group treated with ultra-violet
should treat the patients in the other group.
irradiation in any sense other than that the
When we described earlier the usefulness methods of treatment are applied in appro-
of untreated controls, emphasis was placed priate amounts to patients selected at
on the value of the administration of a random under otherwise comparable con-
plac~bo which outwardly closely resembled ditions. In such an experiment we must
actu?,l treatment. Subsequently, the dis- beware of the possibility that any observed
CUSSIon has centred around the comparison difference may be due not to the respective
of two methods of treatment, but it is forln of treatment per se) but to the ten-
apparent that one of the groups could well dency for patients in plaster casts to be
be a placebo-treated one if ethical con- given seats on trams.
siderations allow. This principle can be In designing the experiment, a fair trial
extended to certain procedures in physio- of the systems of treatment in terms of
therapy. Simple methods of massage, or dosage and duration n1ust be assured.
pills and potions, are likely to be at a Otherwise there may justly be criticism on
distinct disadvantage when undergoing the grounds that one group had too little
comparison with systems of treatment time under the lamp or the other had too
invoking the use of flashing lights, whirring few visits to have the desired result. In
noises, pleasurable warmth, or Heath- most therapeutic trials it is essential to
Robinsonian appliances. Some experiments include one standard method of treatment
will be made sounder scientifically by means of wide acceptance; the other may be as
of the inclusion of a control group in which experimental as one likes within ethical
the stage effects are reproduced but the limits. Thus we have not only a useful
16 THE AUSTRALIAN JOURNAL OF PHYSIOTHERAPY

control, but a familiar yardstick is also stages after treatment commences the
provided for those people who will read the grading process may be repeated. This type
report when it is published. of assessment is relatively inaccurate, but
often it must be used, though its limitations
The Assessment of Progress. must never be overlooked. If it is used,
The problem of assessment of progress particular care must be taken when the
has already been mentioned several times, experiment is designed to make sure that
but it is of sufficient importance to merit the groups of patients are genuinely com-
separate discussion. parable. It is also in this special type of
The simplest methods of assessing pro- test that assessment before and after the
experiment should be made by independent
g:ess ale those based on the patient's
h.istory and on the physiotherapist's impres- observers who are quite unaware of the
group to which the patient being assessed
s.lons. Unfortunately these methods are
belongs.
lIkely to be fallacious. The patient is by no
means. reliable. as a ~it~ess. The physio- When the patients have some chronic dis-
therapIst who is the originator of an experi- order susceptible to treatment but liable to
ment may readily be biased in the role of relapses, it is sometimes possible to assess
assessor but is still likely to gain useful progress by another method. The patients
objective impressions. Impressional assess- are divided into groups for each method of
ment by patient or therapist rudY expose the treatment to be compared and into an
results of the experiment to adverse additional placebo-treated group. Each
c!iticism. .Assessments based on impres- group then receives in rotation each system
SIons are dlfficult to treat statistically and of therapy, thus acting in rotation also as
cannot be compared accurately. They need the control group. Assessments may be
not be avoided entirely, but wherever pos- made by the impressional methods we have
sible they should be supplemented or just outlined or, alternatively, each patient
replaced by objective methods which are may be requested to rank the systems of
capable of mathematical expression. When treatment strictly in order of effectiveness
the object of treatment is the relief of a based on his own experIence of them.
subjective symptom such as pain a mathe- In most experiments, however, it is pos-
matical assessment cannot be made and we sible to obtain some more precise measure-
must make the best use open to us of ment of progress. Sometimes, as with
assessment by jmpressions. Before the ulcers, the diseased area may be directly
expe"riment con;mences, for example, we measured. Sometimes, as in the muscular
requlre the patlents to state, as definitely weakness of poliomyelitis, the deficiency
as they can, the number of attacks of pain p:oduced by the disease may be measured
per day or per week, and the duration dlrectly.. In other conditions the number
radiation, and severity.. They tnay be asked of degrees of flexion at a joint, or the cir-
to. obser.ve th~se points specifically over a cumference of a joint, may be measured.
trial perIod .prlor to the testing experiment, When direct measurement is not possible,
though ethIcally pain beyond trivial pain some allied or related characteristic may be
must be relieved without delay. The dosage measured; as an example, we may measure
of analgesics to keep the pain under control the power of the quadriceps muscle in
may become valuable in assessment of pro- osteoarthritis. Assessment in osteoarthritis
gre~~. The physiotherapist may be in a would otherwise fall into the category in
posItion to record observations of the which considerable reliance must be placed
sy.mptoms in the preliminary period along on the opinions of the patients, though, of
wlth the reports of the patients on their course, we might check these opinions by
symptoms. It is helpful if some mathe- asking each patient to register each day how
matical expression can be given to these often the joint "gave on him" or "locked"
~taternents, such as gradation of severity vlhen he tried to stand up.
into extrelne, moderate, mild, or absent; Considerable ingenuity may be given to
these ~rades may be noted numerically, the devising of what we may loosely refer
respectIvely, as 3, 2, I, and o. At fixed to as an index of progress. An example
 RESEARCH IN PHYSIOTHERAPY 17

which comes to mind is the counting of recovery, or cure can be clearly defined..
handkerchiefs used daily by patients with To show that the number of ayes and noes
hay fever. The assumption there is that in one group have statistical significance
the number of handkerchiefs used would be when compared with those in another is a
constantly related to the amount of nasal simple calculation. But, unfortunately,
discharge, although that relationship would satisfactory criteria for this purpose are
vary from patient to patient; however, it not always available. In most of the illus-
may further be assumed that, in two groups trations we have used the comparisons
selected at random, the variation would be would be in terms of means, or averages,
similar. As the variation is similar, the but the statistical techniques for establishing
two groups are comparable before and after significance are simple to apply.
treatment by means of this admittedly
rough but indubitably ready index.. Another Departmental Research Plans.
ingenious index was devised to assess the In addItion to indicating the principles of
degree of weakness of the muscles of the a controlled investigation of a therapeutic
eyelids as they became too weak to hold regimen we have attempted to show that
the eyes open in myasthenia gravis. A ver- this type of research work is within the
tical scale of letters was placed in front of capacity of an individual physiotherapist
the patient, seated in a fixed position, with or of the staff of a large department. All
the chin in a rest. Before treatment he that is required in either instance is careful
could not raise his eyelids sufficiently to planning in advance based on questions
see any other than the lowest letters.. appropriate to the problem and to the
Within a few minutes of the injection of a available resources. We believe that work
physostigmine analogue he could read letters of this nature should be in progress in every
at progressively higher levels. Thus power department of physiotherapy. Special
in the muscles of the eyelids could be departments and teaching units which have
expressed in terms of an angle of elevation. particular opportunities denied to others
Numerical methods of assessment possess have a special responsibility in this respect,
the advantages that they are relatively for it is to these places that the members
objective and that they can be subjected to of the profession of physiotherapy look for
statistical analysis. The units or methods advancement of their subject and for guid-
used do not matter much, provided that the ance in their work.
same units and methods are employed Because of the heavy load of routine
before, during, and after treatment for each work in a department, research projects
of the groups undergoing the trial. But cannot be undertaken as the main objective.
precision of definition of the units and Thus opportunities for purely academic
methods is essential. Thus it is useless to research work are limited, but that does not
discuss changes in chest circumference mean that the scientific study of treatment,
without reference to the level at which the with which we have been deliberately
measurement is taken. The more accurate dealing almost exclusively, should not or
the method is, the better it is, but an cannot be undertaken. New methods of
Inaccurate method is not necessarily useless, treatment must surely be tried in each
as any errors occurring in the measure- department if it is not to become stagnant.
ments may be assumed to operate equally It has been our aim to show how this
in all the groups.. Differences between trial may be done scientifically to provide
groups may still be of significance in these reasonably dogmatic conclusions. In fact
circumstances. it is becoming a duty to tacl{le problems in
The method of assessment which we have this manner, for it is indisputable that
called the "yes, no" type is easy and rapid- conclusions n1ay be reached which are quite
"healing, no healing", "recovery of muscle erroneous if control groups are not used
power, no recovery", "cure, no cure"" This and assessments of progress are merely
affords an unequivocal nlcans of estimating based on impressions..
the relative merits of two or more thera- In physiotherapy very few of the methods
peutic systems when the criteria of healing, adopted have been tested adequately in this.
18 THE AUSTRALIAN JOURNAL OF PHYSIOTHERAPY

fashion. The omission of scientific trials Summary..

has led recently to criticism of some of the I. A subject for a research project sug-
recognized procedures in the correspond- gests itself with disconcerting frequency.
ence columns of two British medical 2. Attention is given to the selection of
journals. This challenge to physiotherapists specific questions relating to the major
must not be allowed to pass unheeded if problem.. These questions should be
full professional status is to be retained. designed to obtain a definitive answer
Is it possible that research projects will within the limitations imposed by the avail-
immediately be developed in each depart- ability of time, resources, and personneL
ment of physiotherapy after careful review
3.. Three proj ects are examined in det.ail
-of its interests and its resources? Ideally,
to illustrate the posing of appropriate
there should be some rna j or topic upon
questions in a variety of circumstances.
which the members of the staff of each
department will work perhaps for years 4.. The modifications dependent on
.and perhaps in collaboration with other whether the proj ect is to be undertaken by
departments in the hospitaL In addition, an individual or a departmental staff are
each physiotherapist should be encouraged discussed.
to investigate for herself certain aspects of 5- The limitations of these experiments
some subject of interest. Students can be are briefly reviewed"
readtly included in the plans, as they can be 6. The operation of selection in cases
expected to perform some of the work, presenting for study is described, and the
such as rOl.ltine assessments, with the neces- importance of random selection to obtain
sary care and ability. Evaluation of the comparable groups is stressed.
power of the quadriceps muscle, to which 7- Factors which may influence the com-
reference has been made, is well suited to parability of the groups are outlined.
the capacity of students with initial instruc- 8. Effective comparison is achieved in
tion and some supervision.. Every student therapeutic trials by the inclusion of a group
asks questions; often the question asked for placebo-treatment or treatment by a
could be suitably re-stated in a form to method which is recognized to be a standard
which the answer could be provided by a one..
simple experiment. The spirit of inquiry 9. Different ways of assessing progress
gives life to a department immersed in are examined, and particular reference is
routine.. Its influence on students is greatly made to the use of numerical methods.
to be preferred to the stilted atmosphere of 10. Stress is laid on the responsibility of
didactic instruction which often has to be the staff of a physiotherapy department to
endured in the class-room9 promote research work.
After the experiment is done, the results
analysed, their significance evaluated, and Annotated References..
the conclusions drawn, one important matter CURRENT COMMENT (1952), "The ClInIcIan and
must receive attention.. The material must Chnical Research", M J Australia, 2" 863
be prepared for publication.. This final step ThIs brIef reVIew 0 £ several papers contaIns
is for many the hardest task of all, but its much that IS relevant to the present dIscussion
If for "chulclan" we read "physlotheraplsf'8
completion is an experience which is its Parttcular attentlon IS paId to a paper by Pro-
.own reward. In our experience physio- fessor Bradfol d Hill and the article IS sum-
therapists in Australia or abroad have not marized It provIdes a simple exposItIon of
yet seized the magnificent opportunities for all aspects of the conduct of a controlled
InvestIgatIon
research work which lie open to them. A
series of articles published in THE Aus- HILL, A" BRADFORD (1952), "The Chnical TrIal",
New England J Med 2 113 j
TRALIAN JOURNAL OF PHYSIOTHERAPY
This IS the artlcle summarized In "Current
based on controlled and properly conducted Comment" supra; It IS the text of an address at
investigations would place our journal in the Harvard MedIcal SchooL
the forefront of the world's journals of HILL, A BRADFORD (1955), "PrincIples of Medical
physiotherapy. StatistIcs", SIxth Edition, "Lancet", London
 RESEARCH IN PHYSIOTHERAPY

MORONEY, M. ]. (1953), "Facts from Figures", This ::,hort paper, whIch IS easy to understand,
Second EdIt1on, Pengtnn Books, Middlesex, ampltfies some of the pOInts made in regard to
England. the collection of data and Interpretation thereof,
ThIs httle book and that of Professor Bradford and It includes some elementary tests of statistical
HtlI, supra, provIde for readers a more detailed significance
Introduction to experimental desIgn and the use WHITE, C (1953), "Samphng In MedIcal
of statistics than 15 obtainable In the shorter Research", Brtt M J, 2 1284
articles mentioned herein This article deals In some detaIl wIth problems
RUSSELL, C. SCOTT (1955), "Proof in Chnical of bIas and selectIon, and some useful references
Research", Lancet, 2 335 are supphed