BRITISH
334 7 August 1965
Raised Skin Temperature in the Early Diagnosis of Deep-vein
Thrombosis of the Legs
J. L. PROVAN,* M.B., B.SC., F.R.C.S.
Brit. med. J., 1965, 2, 334-337
The reported incidence of deep-vein thrombosis in hospital
patients varies from less than 1 % (Barker et al., 1940 ; Felder,
1949) to over 28% (Sevitt and Gallagher, 1959). This wide
variation reflects not only differences in the populations at risk
but also differences in the criteria of diagnosis and in the
vigilance with which they are sought. There is general
agreement that the clinical diagnosis of deep-vein thrombosis
may be difficult or impossible, and the occurrence of pulmonary
embolism without previous recognition of its cause is common
experience. The usually accepted signs of deep-vein thrombosis
are oedema, calf tenderness on pressure or on dorsiflex on of
the ankle, and tenderness over the femoral vein. Some authors
add cyanosis of the limb and dilated superficial veins, and a
few mention raised temperature of the affected limb. Delayed
cooling on exposure of the limb as an early sign of venous
thrombosis was first mentioned by Pilcher (1939), who stated,
"I do not remember any patient showing this sign who did
not develop others " (of deep-vein thrombosis).
This paper reports an attempt to assess the value of this
physical sign. The investigation was planned to answer three
questions: (1) What other factors than deep-vein thrombosis
can cause delayed cooling on exposure ? (2) In the absence of
such factors is delayed cooling a reliable early sign of deep-vein
thrombosis ? (3) Is delayed cooling always followed by other
physical signs of deep-vein thrombosis ?
Patients
During a period of six months 514 patients in an adult male
and an adult female ward were examined in the manner
described below, 3,950 observations being made. The patients
comprised a group of general surgical cases, including
emergencies and cold surgery, together with some thoracic
and vascular cases. Orthopaedic and genito-urinary cases were
not usually resident in these wards and only a few are included
in the series. Eight of the 28 beds in the male ward were
occupied by patients receiving radiotherapy for malignant
disease. Many of these were ill and confined to bed. Patients
with dressings or bandages on the legs were excluded.
Method
Each patient was examined at least every other day, th._
majority every day. The patient being supine, both legs
were extended with the feet wide apart and uncovered to above
the knee. Particular care was taken to ensure that the legs
were not crossed and that the pat-ent was not lying on his
side, and to see that there was no constriction of the thighs
by clothing. The legs were thus exposed for about 10 minutes,
after which the observer, using one hand, relt the skin around
both ankles and noted any difference of temperature. The
result was recorded simply as right warmer than left, or vice
versa, or no difference. The presence of associated local signs
of deep venous thrombosis was also recorded, together with the
8 Assistant, Surgical Unit, University College Hospital, London. Present
address: Department of Cardiovascular Research, Massachusetts
General Hospital, Boston 14, U.S.A.
MEDICAL JOURNAL
presence or absence of pedal pulses in those patients whose
legs showed a temperature difference. Finally, varicose ulcers,
veins, or eczema and any other leg lesion, particularly inflam-
matory ones, were looked for and noted. At the end of each
round the age, diagnosis, length of stay in hospital, and time
post-operatively for each patient was written down. The
observations were made by two observers only, and the results
of the previous day's observations were not known when the
patients were examined each morning at 8.30 a.m.
Results
Forty-s x patients were excluded from the survey because the
numbers of observations made were too few or too far
separated. Of the 468 patients remaining 266 were male and
202 female ; 317 had had operations and 151 had not.
For the purpose of this investigation it was decided to include
in the survey as positive results only those patients who showed
delayed cooling (hereafter called warm leg) on two or more
consecutive exposures within three weeks of admission to
hospital. It was felt that most patients who developed deep-
vein thrombosis would do so by this time. The exclusion of
single observations of a warm leg tended to eliminate any
observer error and also excluded those patients who might
have shown a false positive when insufficient time had elapsed
to allow the legs to equate their temperatures should they
have been crossed or should the patients have been lying on
their side prior to exposure. It was realized that this might
exclude some patients who had a transitory thrombosis if the
sign of a warm leg is reliable evidence of this. A patient was
assumed to have had a clinical deep-vein thrombosis if ho
developed one or more of the signs of local calf tenderness,
oedema, or femoral-vein tenderness. Dilated superficial veins
or cyanosis in the absence of one of the other local signs did
not place a patient in the clinical deep-vein-thrombosis
category.
Table shows the four categories into which the 468 patients
were divided. These categories are further discussed below.
TABLE I.-Incidence of Warm Lefs (W.L.) and Deep-vein Thrombosis
(D.V.T.) in 468 Cases
W.L. with cause other than D.V.T. . 40
W.L. + signs of D.V.T... 36
No W.L. but signs of D.V.T. .. 12
No W.L. and no signs of D.V.T. .. 380
Table II shows those factors other than deep-vein thrombosis
which were presumed to be responsible for a difference in
leg temperature on exposure in 40 patients. No such factors
were present in the group of 36 patients in Table I who are
discussed below.
TABLE II. Cause for Differing Leg Temperature in 40 Patients
Without Deep-vein Thrombosis
Diagnosis No. of Patients
Ischaemic leg 17
Varicose veins, eczema, ulcer .. 11
Legs crossed 2
Bandage on leg ..1
Rheumatoid arthritis ..1
Inguinal or leg inflammation 4
Sympathetic block 1
Groin haematoma
Intravenous infusion in leg ..2
7 August 1965 Deep-vein Thirombosis-Provan BRITISH
MEDICAL JOURNAL 335
The varicose-vein group is probably abnormally small which males predominated because of the inclusion of the
because many patients were admitted in the evening for opera- radiotherapy cases, there were nine men and three women, and
tion the next morning. They were therefore often in the the right leg was found to be warmer seven times.
theatre or premedicated when the round was done. Of the
17 patients with ischaemic limbs it was not always the more TABLE IV.-Age Distribution of 36 Patients With a Warm Leg
affected leg which was the cooler. Gangrenous changes with
Decades 10- 20- 30- 40- 50- 60- 70- 80- ;90+ Total
mild infection often made this leg warmer. The varicose group
always showed the affected leg to be warmer ; with bilateral No. of patients 34 I61 159 62 83 184 130 14 1 428
No. with warm leg 1 2 1 3 9 8 6 6 00 36
varicose veins the worse-affected leg was usually warmer. The
other causes, with the exception of the two patients having an
intravenous infusion into one leg, also gave a warmer leg on
the affected side. This group of 40 patients may seem large Effect of Operation.-Table V shows the incidence of warm
in comparison with the other groups in Table I, but the wards legs in those patients undergoing operation. The group of
contained many patients with arterial insufficiency in the legs operated patients was divided into three subgroups by the
and varicose veins. severity of their operation. It can be seen that there were far
The following more detailed analysis concerns only the 36 hiore warm legs in the major group.
patients in Table I, in whom, in the absence of any other cause,
deep-vein thrombosis was presumed to account for a warm TABLE V.-Relation of a Warm Leg to Severity of Operation
leg. These patients had a warm leg on more than one con- No. Warm Leg Post-operatively
secutive occasion with or without the commonly recognized Type of Operation No. with
signs of deep-vein thrombosis. This group was analysed Waxrm Leg No. o/
further to decide if there was a statistically significant associa- Major
Intermediiate .. ..
, 130
157
18
5
15
4
11-5
2-5
tion between a warm leg and other signs of deep-vein Minor . .. .. 30 1
i2 0 0
thrombosis, and, if there was such an association, whether it Total .. .. 317 24 19 6-0
held good for other factors, such as operation or age.
Calculated by the exact single-tail test of probability in a 2 x 2
table, as shown in Table III, P = 0.00000000046 for the overall Time of Onset of Warm Leg and Signs of Deep-vein
results. Thus the probability of observing such extreme figures Thrombosis.-The Chart shows the time at which warm legs
by chance if a warm leg and deep-ve.n thrombosis were in fact and clinical signs of deep-vein thrombosis were first noted in
unrelated is very low indeed. the overall group and in those patients who had operations.
TABLE III.-Time Relations Between Warm Leg and Deep-vein
There are two apparent waves in the warm-leg series-one
Thrombosis in Various Groups of Patients soon after admiss on or operation (these are not necessarily the
same), and the other, wider, wave occurring at 5 to 10 days,
D.V.T. Alone. when the incidence of deep-vein thrombosis might be expected
Group W.L. D.V.T. Before NoL
of Preceding W.L.
W.L. D.V.T. .. Total
P
to be at its highest. The patients with clinical deep-vein
Patients D.V.T. and W.L.
Together
No.T
D.T t thrombosis did not apparently show this same trend in the
All 14 114 20 380 428 0 00000000046 two groups, but this may be because the numbers in the deep-
Major ops.' 7 6
3
8
4
109
179
130
187
0-000073
0-1035
vein-thrombosis group are much smaller. It was noted in the
Other ops. 1
No ops. 6 5 8 92 111 0 00044 36 patients who had unexplained warm legs that 14 had a
warm leg only and 14 had a warm leg before developing the
* Calculated by the exact probability single-tail test in a 2 x 2 table. clinical signs of deep-vein thrombosis in the same leg, whereas
only three patients had signs of deep-vein thrombosis in one
The numbers in the groups of patients undergoing major and leg before showing an ipsilateral warm leg. In five patients
other operations are not widely different, whereas the figures are a warm leg and signs of deep-vein thrombosis were observed
more highly significant in the major-operation group. This together at the first examination. Table VI shows these rela-
implies that a warm leg is more closely related to deep-vein tions in the operated and non-operated groups. The period
thrombosis in the major-operation group. The figures indicate of time which elapsed between the onset of warm leg and the
that a warm leg is also more likely to occur in this group. appearance of the signs of deep-vein thrombosis varied. It
Age Distribution.-Table IV shows the age distribution of was one to six days in the non-operated group, the interval
the 36 patients. It can be seen that the incidence rises con- being one day in four instances, and two and seven days in
siderably after the fifth decade. Twenty-
nine of the 36 patients with warm legs ONSET Tr1ME OF WARM LEG AFTER APPEARANCE OF WARM LEG
were in the group of 212 aged 50 years ADMISSION RELATIVE TO TIME POST-OP
ALL PATIENTS (36) 19 WITH POST-OP WARM LEG
or more. Only 7 of the 216 patients 5
younger than 50 showed a warm leg. 4 _ A
Because the incidence of a warm leg and 3-t_ _
deep-vein thrombosis is so small in this
latter age group the test of probability was
applied to the age groups 10-59 years in-
clusive and 60-99 inclusive: P:=0.000023
21_
UI JJI4 litil n r{TIJr{Tt] m
.

APPEARAINCE OF SIGN OF D.V T. APPEARANCE OF SIGN OF D.V T.

and 0.000070 respectively for these groups, c RELATED TOILENGTH OF STAY IN HOSPITAIL RELATED TO TIME POST-OP
so that the same relation between a warm ,^ WARM LEG WITH D.V.T. EJ22 PATIENTS POST-OP. D.V.T. & WARM LEG = 19 PATIENTS
leg and deep-vein thrombosis mentioned 0 D.V. T. ONLY, NO WARM LEG_ 12 PATIENTS POST-OP D.V.T. NO WARM LEG _ 6 PATIENTS
previously still holds true. 0
5-
Sex Distribution and Side Affected.- 4-
There was no apparent preponderance of 3 *
sex or of one side in the warm-leg series.
The 24 patients who had had operations
consisted of 12 men and 12 women. The
right and left legs were also equally
2_
l
5
DAYS IN HO'SPITAL
10 IS
W UTMn 20 5
DAYS POST-OP.
10 IS 20
nFl-
affected. In the non-operated group, in Days on which patilents showed first evidence of either warm leg or clinical deep-vein thrombosis
after admission to hospital and after operation.
336 7 August 1965 Deep-vein Thrombosis-Provan
the operated group. In the latter the delay in appearance of
signs of deep-vein thrombosis was three days or less in five
of the eight patients.
TABLE VI.-Time Relations Between Warm Leg and Deep-vein
Thrombosis
No Operation Operation
W.L. only 5 9
|
W.L. preceding D.V.T.
D.V.T. preceding W.L. .I
6
1
8
2
W.L. and D.V.T. simultaneous .. 0 5
Pulmonary Embolism.-Two patients of the 514 observed
had pulmonary emboli. Both occurred in the operated group,
and one had already had signs of deep-vein thrombosis in the
legs for six days. The second patient developed a warm leg
with no other signs of deep-vein thrombosis the day after the
pulmonary embolism.
Discussion
This investigation was prompted by the generally accepted
estimate that 5000 of pulmonary emboli occur in patients with-
out signs of venous thrombosis in the legs (Cosgriff, 1947).
Sevitt and Gallagher (1961) state that deep-vein thrombosis
is symptomless in two-thirds of cases. In 1959 they demon-
strated that where thrombosis has been diagnosed in one leg
necropsy may show that it is bilateral. Whatever policy is
adopted to deal with the hazard of pulmonary embolism,
earlier and more frequent recognition of deep-vein thrombosis
is a worth-while aim.
The common site for the beginning of deep-vein thrombosis
is the calf muscles (Gibbs, 1957), whence it may or may not
spread to the main deep veins of the limb. It is only when
such spread occurs that the full clinical picture with oedema
and thigh tenderness will develop. The structure of pulmonary
emboli suggests that the propagation and separation of the
clot is often rapid, and many believe that when the signs of
thrombosis are gross embolization is less likely. The stage
at which diagnosis is needed is while thrombosis is confined
to the calf and before occlusion of main deep veins has
occurred.
As an explanation of delayed cooling on exposure it is
postulated that thrombosis in deep veins results in shunting
of the venous return into superficial veins and that increased
flow in these raises skin temperature. The presence of dilated
veins below the knee has been noted by other observers
(Homans, 1934 ; Felder, 1949) in association with deep-vein
thrombosis. Such dilated veins are most easily recognized in
thin subjects, but many patients with deep-vein thrombosis are
fat. A rise of skin temperature has also been noted by Homans
(1934, 1947), Felder (1949), and Short (1952). Felder found
a raised leg temperature in 54 out of 105 legs, accompanied by
tenderness and swelling in 42 of them. He also states, however,
that swelling often occurs without heat. This does not
invalidate the explanation of the sign, since with the progress
of deep-vein thrombosis sufficient to cause oedema the super-
ficial venous return will also be slowed and the warming effect
of increased flow will be lost. It is suggested, therefore, that
the sign of delayed cooling on exposure is most likely to be
found early in the process and may disappear as it either
extends or resolves. This would explain the lack of a constant
association between a warm leg and other signs of deep-vein
thrombosis in Tables I and VI.
The method of carrying out the test as described in the
present paper is an attempt to exclude increase in leg tempera-
ture due to crossing of the legs or other postural factors while
the patient is in bed. The period of 10 minutes was adequate
for cooling to an equal temperature to occur in normal legs,
but not in those where a pathological condition of the leg was
BRITISH
MEDICAL JOURNAL
present. The simplicity of the test is its great advantage. It
can be carried out quickly and easily, and a "leg round " on
48 patients takes about half an hour.
It is difficult to decide unequivocally that the physical sign
of a warm leg is indicative of venous thrombosis in that leg
in the absence of the recognized physical signs. Phlebography
Total is notoriously unreliable as well as difficult and time-consuming
14 to perform. Post-mortem examinations have obvious limita-
14
3
tions in a vital study such as this, and it was often found that
5 the sign of a warm leg, as well as the usual signs of thrombosis,
disappeared after a few days. This may be due to thrombolysis
or to the development of deep venous collaterals, but if the
former is true necropsy evidence may have little value.
The statistical evidence given does not prove that the rela-
tion between a warm leg and deep venous thrombosis is causal,
but in the absence of any other suggested factor it seems very
likely to be so. Additional support for the theory is given by
the relation of the sign of a warm leg to age, time of onset, and
type of operation. This conforms closely to the findings of
Barker et al. (1940, 1941) at the Mayo Clinic for deep venous
thrombosis.
The finding of warm legs from days 1 to 3 after admission
is in keeping with the evidence of Browse (1964) that hos-
pitalization alone increases the period of a patient's inactivity
from 33 % to 63 % per 24 hours. It may also increase his
tendency to venous thrombosis at that time.
The number of pulmonary emboli in the series is too small
for any conclusions to be drawn, but, using a warm leg with
or without the generally accepted signs as evidence of throm-
bosis, the diagnosed incidence of deep venous thrombosis in
this series (including those patients without a warm leg) is
10.500. If patients with a warm leg only are excluded the
incidence of deep venous thrombosis falls to 7.3 It remains
to be seen how often pulmonary embolism follows the finding
of a warm leg alone.
Summary
A clinical investigation is described of delayed cooling (warm
leg) as an early physical sign of unilateral deep venous throm-
bosis of the lower extremities in 468 patients. Thirty-six of
them had an otherwise inexplicable unilateral warm leg, and
statistical analysis of these patients shows that there is a highly
significant correlation between the presence of a warm leg and
the signs of venous thrombosis in that leg. The same
statistical correlation is present when these patients are sub-
divided for age and type of operation, and the results are more
significant in patients undergoing major operations and in
patients over the age of 59 years.
Fourteen of the 36 patients showing a warm leg subsequently
developed other clinical signs of deep venous thrombosis in
that leg. The interval between the appearance of a warm leg
and other signs of thrombosis varied from one to seven days,
but was usually three days or less. Fourteen patients who
developed a temporary warm leg did not produce other signs.
Only two patients in the series had pulmonary emboli; only
one patient had leg signs first, but the numbers are too small
to assess the relation between a warm leg and pulmonary
embolism. The use of this sign may draw early attention to
venous thrombosis, and, in this series, raises the diagnosis rate
from 7.3 % to 10.5 %.
I wish to thank Professor R. S. Pilcher, who originally suggested
this investigation, for his help in the preparation of the paper.
Sister Mauger, Sister Robb, and the nursing staff of wards 24 and
21 showed unfailing patience and assistance during the survey. Mr.
L. E. Edwards, a senior medical student, assisted with the observa-
tions during his elective period. Mr. I. D. Hill, of the Medical
Research Council Statistical Research Unit, has been responsible
for the statistical work. To all these I express my gratitude.
 7 August 1965 Deep-vein Thrombosis-Provan
REFERENCES
Barker, N. W., Nygaard, K. K., Walters, W., and Priestley, J. T. (1940).
Proc. Mayo Clin., 15, 769.
~~ (1941). Ibid., 16, 1.
-~
Browse, N. (1964). Brit. med. 7., 1, 669.
Cosgriff, S. W. (1947). Amer. 7. Med., 3, 740.
Felder, D. A. (1949). Surg. Gynec. Obstet., 88, 337.
Use of Propranolol in Angina Pectoris
P. M. S. GILLAM,* M.B., M.R.C.P.; B. N. C. PRICHARD,t M.Sc., M.B., B.S.
Brit. med. J., 1965, 2, 337-339
Pronethalol, an adrenergic beta-receptor-blocking drug (Black
and Stephenson, 1962), has been shown to relieve the pain
of angina pectoris in a double-blind trial (Prichard, Dickinson,
Alleyne, Hurst, Hill, Rosenheim, and Laurence, 1963). Its use
in angina was, however, discontinued following a report of its
tumour-producing activity in mice (Paget, 1963). A further
disadvantage of pronethalol was that the therapeutic dose was
close to that which produced side-effects.
Black, Crowther, Shanks, Smith, and Dornhorst (1964) have
described the pharmacology of propranolol (Inderal), a beta-
receptor-blocking agent in animals and man with about ten
times the potency of pronethalol. It is non-carcinogenic in mice
and other experimental animals. In view of the effectiveness of
pronethalol it was thought that the closely related drug
propranolol might be of value in angina. The present paper
reports a double-blind trial in out-patients with angina pectoris.
Design of Trial
The design of the trial closely followed that described by
Prichard et ad. (1963). The original aim was to compare
maximum tolerated doses of propranolol with placebo, but this
was modified in that an upper dose limit of 100 mg. q.d.s. was
later fixed ; this dose represented 40 tablets a day. During the
" run-in period" the initial dosage was 10 mg. q.d.s. Incre-
ments of 10 mg. per dose per week were made up either to 100
mg. q.d.s. or to the maximum tolerated dose, when this was less.
This variable dosage was selected because of the need to take
into account individual variability, and also because benefit at a
lower dosage might be largely a placebo effect, which may also
result from increased frequency of out-patient visits. This
latter factor was mitigated by having a long run-in period
before the trial proper, three months or more, a period further
extended owing to delay in obtaining identically tasting placebo
tablets. During this time patients were seen every two weeks
and the dosage was adjusted.
The period of the actual trial was eight weeks. Patients were
given placebo and propranolol each for two periods of two
weeks' duration; they were identical in taste and appearance.
Randomization of these four periods was made so that the six
possible orders were achieved for every six patients admitted to
the trial.
Patients were considered suitable for admission to the trial if
they were having at least two typical attacks of angina pectoris
each week. The pain had to be of characteristic nature, site,
and radiation, brought on by exertion, relieved by rest and
* Formerly Senior Registrar, Medical Unit, University College Hospital,
London; now Senior Registrar, Whittington Hospital, London.
t Lecturer in Applied Pharmacology and Therapeutics, Medical Unit,
University College Hospital Medical School, London.
MEDIBALJURNAL 337
Gibbs, N. M. (1957). Brit. 7. Surg., 45, 209.
Homans, J. (1934). New Engl. 7. Med., 211, 993.
(1947). Ibid., 236, 196.
Pilcher, R. (1939). Lancet, 2, 629.
Sevitt, S., and Gallagher, N. G. (1959). Ibid., 2, 981.
(1961). Brit. 7. Surg., 48, 475.
Short, D. S. (1952). Brit. med. Y., 1, 790.
glyceryl trinitrate, and usually lasting for one to three minutes.
The purpose of the trial was explained to the patients, who were
told that the new drug would be compared with identical
dummy tablets in order to try to eliminate bias. Sixteen patients
were started on the run-in period. One 71-year-old woman
developed signs and symptoms of heart failure two days after
starting 10 mg. q.d.s. The drug was stopped; the symptoms
cleared without further treatment in three days, as previously
reported by Prichard and Gillam (1964). She subsequently
experienced an exacerbation of her angina. She has been put on
chlorothiazide with potassium supplement and digitalized.
Starting as an in-patient, she has been given propranolol again,
when she tolerated 50 mg. q.d.s. without signs or symptoms of
heart failure, with an improvement of her angina. A 63-year-
old man developed visual hallucinations after five weeks on
100 mg. q.d.s.; these subsided at a reduced dosage, but his
dosage was not readjusted in time for inclusion in the double-
blind trial. The remaining 14 patients entered the trial proper
and completed it.
Assessment
Patients were asked to take glyceryl trinitrate for pain, but
not prophylactically, and to record their attacks of angina and
glyceryl trinitrate consumption on record sheets that were
provided. The sheets were divided into four periods for each
day, and patients were asked to fill in their sheet at the end of
each period, to record the time of onset of angina, its duration,
severity, and the number of glyceryl trinitrate tablets consumed,
and to note any comments they desired. Patients were supplied
with a known number of tablets, and those taken from the
bottle were checked with the number recorded as consumed.
Patients were seen fortnightly by the same clinicians at each
visit. The physician who assessed each patient did not know
whether propranolol or placebo was being taken and did not
refer to previous assessment sheets. Patients were asked standard
questions and their subjective impressions recorded.
Results
The results are summarized in Table I. All 14 patients con-
sumed fewer glyceryl trinitrate tablets and had fewer attacks of
pain on propranolol than on placebo. The probability that all
14 patients should produce the same result due to chance is
P = 0.00012, which is highly significant. The subjective
reports of the patients were assessed as +2 (much better), + 1
(better), 0 (same), - 1 (worse), -2 (much worse). The score on
placebo was subtracted from that on the drug so that a positive
score indicated subjective benefit from propranolol. The maxi-
mum score in any patient depended on the number of changes
between drug and placebo, which was determined by the