lack of scientific evidence for the treatment of lateral ... - Bone & Joint

1 downloads 0 Views 869KB Size Report
Diflunisal vs naproxen. No difference. No improvement. 11 solution as compared with. 13 treated by ionisation and a placebo, and these authors could not.
LACK

OF OF

SCIENTIFIC

EVIDENCE

LATERAL

EPICONDYLITIS AN

H. LABELLE,

We have

ATTEMPTED

R. GUIBERT,

From

the

Lateral and

for

lateral

epicondylitis results

185 articles

N. NEWMAN,

ofMontreal

epicondylitis

(LE)

published

TREATMENT

THE

ELBOW

META-ANALYSIS

J. JONCAS,

University

OF

THE

and

of the

or ‘tennis

elbow’

morbidity

and

in considerable

since

M. FALLAHA,

Sainte-Justine

elbow.

Of

cost.

In Sweden the prevalence is 1% to 3%, but increases 19% in men between 40 and 50 years of age (Allander 1 974).

The

involves

syndrome

is seen

repetitive

it is reported

mainly

movements

that

7.4%

in those

of the

forearm

of industrial

of tennis players affected by it (Dimberg

are

and at

to

work

and

workers

in the USA

50%

whose

185

RIVARD

Hospital

some

to

time

the scientific

evidence

78 discussed

treatment,

articles,

in Quebec

made

board

this

for

(Canadian) lost (1982 tion

Board Many

first

clinical

hand:

40%

In 198 1 , over 1 500 workers

1987).

1966

the

is common financial

to assess

C-H.

for methods of but since the natural history of the syndrome is uncertain we considered only those series with concurrent control groups. Only 18 of these were randomised and controlled studies. We then graded these papers for scientific validity, using the methods of Chalmers et al (1981). The mean score of the 18 articles was only 33%, with a range from 6% to 73%. A minimum of 70% is required for a valid clinical trial, and we therefore concluded that there was insufficient scientific evidence to support any of the current methods of treatment. There were too many methodological differences to allow a quantitative meta-analysis, but our qualitative review established the importance of the natural evolution of the syndrome and of the placebo effect of all treatments. Properly designed, controlled trials are needed. treatment

reviewed

FOR

claims

of Surgery,

University

Province

of Quebec,

statistics). papers have description

University

Service, Montreal,

Sainte-Justine Quebec, Canada

R. Guibert, MD, MSc, Medicine and Researcher Sante University of Montreal, Canada H3C 3J7.

Assistant Groupe

N. Newman, of Montreal

MD,

C, Assistant

H#{244}tel-Dieu

Hospital,

CanadaH2W

P0

FRCS 3840

physiotherapy

many

conventional of

MD,

Montreal,

Professor, de Recherche

Department of Interdisciplinaire

6128-Succ.

A,

Montreal,

ProfessorofSurgery, Street,

Sainte-Urbain

C, Assistant

Professor

Family en Quebec,

University Montreal,

British

$4442

Compensa-

on LE and

since

review

treatments

the

articles

used

either

anti-inflammatory techniques,

surgical

operations

such as radiotherapy,

cast and

less

acupuncture

vitamins.

The

choice

of

treatment

for

an

individual

case

remains controversial because it is empirical and based on the personal experience of the physician treating the patient. Our aim was to review the current literature to determine if there was any significant scientific evidence to favour a particular method of treatment for acute lateral

epicondylitis.

Quebec,

of Surgery,

University

We Hospital,

Quebec,

Correspondence

Canada

5415

L’Assomption

H1T 2M4.

should

be sent

Editorial

Society

030l-620X/92/5377 $2.00 J Bone Joint Surg [Br] 1992;

646

C#{244}teSainte-

of

MATERIAL FRCS

Maisonneuve-Rosemont

1992

3175

methods

lT8.

M. Fallaha, of Montreal

©

Box

Hospital, H3T 1C5.

1873),

These include

orthoses,

cost

of 62 days of work

published

drugs,

Montreal Orthopaedic Catherine,

been (Runge different

steroids,

compensation a

Workman’s

have reported over 40 alone or in combination.

and Surgery,

workers’

generating

per case and an average

immobilisation, H. Labelle, MD, FRCS C, Associate Professor of Montreal J. Joncas, RN, Research Assistant C-H. Rivard, MD, FRCS C, Professor of

to the

condition,

and

Joint

Surgery

1966 articles.

74-B

:646-51.

used

databases

surgery,

the

Medline

to make

French

to Dr H. Labelle. of Bone

Boulevard,

medical to

and

EMbase

a thorough

and paramedical

1990

(inclusive),

These

included

35 current

concepts

THE

JOURNAL

(Excerpta

review

literature

locating 78

of the

on

OF

and

BONE

studying treatment

articles,

AND

and

for the years

medical

or review

Medica) English

JOINT

27 on

SURGERY

185 or the

LACK

OF

SCIENTIFIC

Table I. The study on

the

EVIDENCE

FOR

design of 78 published of lateral epicondylitis

treatment

THE

TREATMENT

Case

series

without

Case

series

with

Controlled

historical

51

chronic

controls

2

trial

I

non-randomised trial

Controlled

randomised

single-blind

Controlled

randomised

double-blind

are two

important

history

the

in the II.

showing trial

Evaluation the I 5 items

form scored

Items I Selection

description

A, adapted to evaluate

from Chalmers the study design

Possible points

Items

0 to 3

9 Observer

blinded

et al (1981), of a clinical

0 to 3

log*

10

Prior

estimate

4 Control

regimen

appearance

0 to 3

1 1 Stopping

0 to 1 .5

12

to

of

rules

Testing

rate

ofrecovery

absence

tastet

6 Randomisation

blindability

blindability

*a full record

ofeligible

tconfirmation treatment

that does

the

not differ

compliance

0 to 3

0 to 8

15

Biological

0 to 3

Totalpossible

treated

that

the personal a few even

experience of earlier reports

Secondly, the results must remain doubtful

or improvement

III.

score

medication),

or the

and

groups

Possible points

Items

control

I classifies

design

We

the

78

reviewed

a minimum

sensation

of a

scoring therefore

intended

to

Possible points

sive

results. We used

a quantitative

2 Results

0 to 2

9 Statistical

0 to 4

and method photocopied,

0 to 4

published

analysis

4 Posttype2

Oto3

estimate 5 Confidence

limits

0 to 3

10 Withdrawals 11

Handling withdrawals

12 Side-effects

Oto4

were

conduct

and

analysis

VOL.

0 to 2

13 Retrospective evaluation

7 Timing

0 to 4

T otalpossible

74-B,

No.

ofevents

5, SEPTEMBER

1992

0 to 3

sco re 40

using which

of papers

by

research carried out. It the results appropriately,

technique

and results so that the

the paper, to the

the evaluators The

in Tables 6 Time series analysis

Chalmers’

unknown

among 0 to 3

meta-

is

results between two sizes and inconcluwith

great

care.

Two

evaluation forms, with 28 individually scored items, allowed for a maximum score of 100. These forms were completed independently by four trained health professionals (HL, JJ, NN, MF), with adequate experience in clinical trials. To reduce observer bias, only the material

0 to 2

0 to 4

validity.

study of the 18 selected papers by Chalmers et al (1981)

the design,

8 Regression! correlation

end-points

the

We made a trials in which

of internal

make

the quality of the possible to interpret

0 to 2

3 Major

on

but found it impossible to group the studies of the considerable variation in treatment, criteria and measures of efficacy. We therefore

1 Dates of study description of

differing

articles

from further analysis. of the 18 controlled

had

initially

analysis because selection

evaluates

from Chalmers to evaluate the

Items

randomisation

syndrome

The

selection of cases from one study to another must lead to potentially important differences in the stage of evolution of the disease, and make such uncontrolled studies

made a qualitative the system proposed

60

for the trial

Evaluation form B adapted showing the I 3 items scored of a clinical trial

et al (1981) data analysis

ofthe

is unknown.

especially when there are conflicting studies, or studies with small sample Table

First, ignortreatment

METHOD

14 Testing

between

consequences. means

of LE

0 to 3

0 to 10

(of oral

enor-

the study

0 to 3

taste

varied

0 to 3

blinding

not accepted

and

the and

therefore excluded more detailed study

1 3 Testing

population

the

0 to 3

0 to I .5

0 to 8

on

treatment of LE according to the design of the study. Most of them did not include control groups, and were

equivalence 8 Observer

articles

0 to 4

blinding 7 Patient

unclear

of treatment

randomisation S Control

to

six on

inconclusive. Table

numbers 3 Therapeutic definition

letters and

Possible points

results 2 Reject

were

of the natural

since

Table

24

which could explain definitions of acute

rather than on scientific evidence. of uncontrolled therapeutic trials

13

trial

disease, instruments

seems to be based on either the different authors, or on

1

trial

the

647

ELBOW

aspects of the syndrome. information in these

no

epicondylitis There

3

randomised

THE

mously.

ance Controlled

of

on measuring

natural course of the disease variations in treatment. The

7 controls

OF

pathology

15 articles

the biomechanical We found

Number reports

EPICONDYLITIS

and

editors,

elbow

Case

LATERAL

diagnosis

articles of the

Study

OF

contents

III.

of each paper names, the journal

the discussion researcher.

were ofthe

II and

sections authors’

resolved first

the conclusions

Differences

evaluation The

and

at group forms

form

were that

in scores

discussions.

are summarised

examined

the

study

design by giving particular importance to blinding patients, physicians and measuring observers, and presence

and

method

of randomisation

of the

patients

of the

648

H. LABELLE,

and of the physicians evaluated the quality

when applicable. The second form of the data analysis, the statistical

analysis

and

the

presentation

revision

of Chalmers’

in 1984

(personal

this

is the

first

orthopaedic

R. GUIBERT,

of results.

original

methods

communication), use

of this

J. JONCAS,

We

used

of 1981 and

we

evaluation

N. NEWMAN,

ultrasound a

in the

literature.

over

rest

that

there

possible

score,

applicable excluded The minimum and

give our evaluations results as percentages and

allowing

to every

for

study

and

from the calculations. average score for the of6%

of 73%

for the of the

items

which

which

were

18 trials

for the weakest

a maximum

for the

exceeding

for

our

1 8 studies

was

33%,

very

score

not

with

(Kivi

a

results, perspective. conflicting We

discussion This

a quality

and conclusion is

index

sections

particularly

results reported further classified

in different the papers

useful

et al 1985

Lundeberg

Stratford

Halle

et al 1988

et al 1989

et al 1986

studies

low,

only

; it

the

of a study

into

in

not

of 27%, accord-

tion

76

Ultrasound

48

Ultrasound ultrasound

with

with

to have

of 30%,

no

in

with

confirmed

the

difference placebo ultrastimulation

nerve

with hydrocortisone in samples were too small

have a therapeutic effect: from baseline values was but whether this effect is

significant

unclear. ointment

advantages.

The did More

(Table

V).

Three

reported

a significant

a diclofenac

examined

0.01). studies.

This was Vecchini

showed

only

treated

only and

with

partially Grossi

a score

ionisaa placebo

confirmed by (1984), with a

therapeutic

ionisation

and

effect

diclofenac

vs placebo

vs placebo vs rest

Quality score(%)

Conclusions Ultrasound (p