Full text of DOI:10.1001/jama.284.15.1926

LETTERS
Statin Drugs and the Risk of Fracture
Table. Frequency of Reported Fractures in the Scandinavian
Simvastatin Survival Study
To the Editor: Dr Meier and colleagues1 suggest that statins may
lower the risk of fracture. Their results are consistent with the
finding that statins increase bone morphogenetic protein-2
(BMP-2) in rodent and human cells and increase the rate of bone
formation in rodents by inhibiting 3-hydroxy-3-methylglutaryl
coenzyme A (HMG-CoA) reductase.2 However, pravastatin has
recently been shown to be ineffective at increasing BMP-2 or
stimulating bone formation.3 This may be due to the different
physicochemical properties and relative hepatoselectivity of
pravastatin.4 Indeed, pravastatin binds less strongly to plasma
proteins than do lovastatin, simvastatin, or fluvastatin.4 Exclud-
ing pravastatin from the analysis may increase the association
between statins and decreased risk of fracture. Although Meier
et al indicate that each of the statins was individually associated
with a decreased fracture risk it would be of interest to know
how the reduction in risk by pravastatin compares with the other
statins.
Meier et al examined the effects of fibrates as a separate class
of nonstatin lipid-lowering drugs, but examining the effects of
cholestyramine may be more informative. Unlike statins, which
inhibit HMG-CoA reductase, cholestyramine causes an in-
crease in the levels of this enzyme. This effect might be ex-
pected to increase, not decrease, the risk of fractures. Further-
more, unlike other lipid-lowering drugs, cholestyramine impairs
vitamin D absorption, leading to osteomalacia and increased
risk of fractures.5 Because Meier et al excluded subjects with
osteomalacia, this may have diminished the number of sub-
jects in the “other lipid-lowering drugs” group that used cho-
lestyramine. By both inhibiting vitamin D absorption and in-
creasing HMG-CoA reductase activity, cholestyramine may
particularly increase the risk of bone fractures.
Placebo Group
Simvastatin Group
Patients Patients
Patients Patients
Aged Յ60 y, Aged Ͼ60 y, Aged Յ60 y, Aged Ͼ60 y,
Type of
No. (%)
(n = 1223)
No. (%)
(n = 1000)
No (%)
(n = 1177)
No. (%)
(n = 1044)
Fracture
Hip or femur
Vertebral
Other
2 (0.16)
4 (0.33)
8 (0.80)
4 (0.40)
1 (0.08)
1 (0.08)
9 (0.86)
2 (0.19)
30 (2.45)
36 (2.94)
23 (2.30)
35 (3.50)
34 (2.89)
36 (3.06)
37 (3.54)
48 (4.60)
All
All, any age
71 (3.19)
84 (3.78)
We examined the frequency of reported fractures in the Scan-
dinavian Simvastatin Survival Study (4S).4 This was a random-
ized, double-blind, placebo-controlled multicenter trial of sim-
vastatin (20-40 mg/d) for patients aged 35 to 70 years with
coronary heart disease. During a median of 5.5 years of fol-
low-up the main end point, all-cause mortality, was reduced
by 30% relative to placebo. Patients were queried for adverse
experiences at each scheduled clinical visit, which occurred at
least every 6 months. All adverse experiences were recorded
on designated case report forms, regardless of whether they were
considered drug related.5 We identified 155 patients with re-
ports of fracture. The distribution is shown in the TABLE.
None of the differences between treatment groups was sta-
tistically significant. We also found no differences when the
data were analyzed by sex (19% of patients were women). We
could, therefore, not confirm any of the associations found in
the case-control studies.
The proposed risk reductions in the case-control studies were
substantial, with odds ratios (ORs) ranging from 0.48 to 0.55;
we therefore expected to find at least a trend in favor of sim-
vastatin in the 4S. Assuming a “true” fracture incidence of 3.5%
in the placebo group, the 4S sample size and number of all frac-
tures would have provided a statistical power of 94% to detect
a 50% relative risk (RR) reduction and a 78% power to detect
a 40% RR reduction.
Ahmed El-Sohemy, PhD
University of Toronto
Toronto, Ontario
1. Meier CR, Schlienger RG, Kraenzlin ME, Schlegel B, Jick H. HMG-CoA reduc-
tase inhibitors and the risk of fractures. JAMA. 2000;283:3205-3210.
2. Mundy G, Garrett R, Harris S, et al. Stimulation of bone formation in vitro and
in rodents by statins. Science. 1999;286:1946-1949.
3. Sugiyama M, Kodama T, Konishi K, Abe K, Asami S, Oikawa S. Compactin and
simvastatin, but not pravastatin, induce bone morphogenetic protein-2 in human
osteosarcoma cells. Biochem Biophys Res Commun. 2000;271:688-692.
4. Hamelin BA, Turgeon J. Hydrophilicity/lipophilicity: relevance for the pharma-
cology and clinical effects of HMG-CoA reductase inhibitors. Trends Pharmacol
Sci. 2000;19:26-37.
GUIDELINES FOR LETTERS. Letters discussing a recent JAMA article should
be received within 4 weeks of the article’s publication and should not exceed 400
words of text and 5 references. Letters reporting original research should not ex-
ceed 500 words and 6 references. All letters should include a word count. Letters
must not duplicate other material published or submitted for publication. Letters
will be published at the discretion of the editors as space permits and are subject
to editing and abridgment. A signed statement for authorship criteria and respon-
sibility, financial disclosure, copyright transfer, and acknowledgment is required
for publication. Letters not meeting these specifications are generally not con-
5. Knodel LC, Talbert RL. Adverse effects of hypolipidaemic drugs. Med Toxicol.
1987;2:10-32.
To the Editor: Three recent case-control studies1-3 have sup- sidered. Letters will not be returned unless specifically requested. Also see Instruc-
ported an association between statin use and reduction in the
tions for Authors (July 5, 2000). Letters may be submitted by surface mail: Letters
Editor, JAMA, 515 N State St, Chicago, IL 60610; e-mail: JAMA-letters
risk of fracture. There appears to be an emerging consensus
that statins may increase bone mineral density and decrease
the risk of osteoporosis.
@ama-assn.org; or fax (please also send a hard copy via surface mail): (312) 464-5824.
Letters Section Editors: Stephen J. Lurie, MD, PhD, Senior Editor; Phil B.
Fontanarosa, MD, Executive Deputy Editor.
©2000 American Medical Association. All rights reserved.
(Reprinted) JAMA, October 18, 2000—Vol 284, No. 15 1921
Downloaded From: https://jamanetwork.com/ by a Carleton University User on 11/09/2020

LETTERS
There are some limitations to our analysis. Fractures result- low-up than the simvastatin group, because the placebo group
ing from severe traumas could not be distinguished from those was more likely to die (RR of death, 0.70; 95% CI, 0.58-0.85)
probably more directly related to osteoporosis. Fracture was and because the dropout rate was higher in the placebo group
not a predefined end point of the study and the study popula- (13% vs 10%). This may potentially decrease the number of
tion was not at particularly high risk of fractures, but neither fractures in subjects in the placebo group because they had less
were the populations reported in the case-control studies. We person-time of follow-up and therefore less chance to have a
strongly recommend that until dedicated prospective random- fracture.
ized controlled trials provide more reliable information on the
On the other hand, it is possible that the subanalysis of the
relationship between statin use and fracture risk, physicians 4S data is accurate. As we emphasized in our article, more con-
should not deviate from current treatment recommendations6 clusive data from controlled trials should be obtained before phy-
for people at risk of osteoporosis, regardless of whether those sicians change prescribing patterns for osteoporosis therapy. How-
patients may be taking statins.
ever, it is intriguing that 3 independent observational studies
found similar results2-4 and that statins have been shown to af-
fect bone density in both rodents5 and humans.6 Prospective con-
trolled trials with bone density measurements and fracture rates
as primary end points are needed to clarify whether statins are
indeed useful in preventing or treating osteoporosis.
Terje R. Pedersen, MD
Aker University Hospital
John Kjekshus, MD
The National Hospital
Oslo, Norway
for the 4S Study Group
Christoph R. Meier, PhD, MSc
Raymond G. Schlienger, PhD
Marius E. Kraenzlin, MD
Brigitta Schlegel, MD
University Hospital Basel
Basel, Switzerland
1. Meier CR, Schlienger RG, Kraenzlin ME, Schlegel B, Jick H. HMG-CoA reduc-
tase inhibitors and the risk of fractures. JAMA. 2000;283:3205-3210.
2. Wang PS, Solomon DH, Mogun H, Avorn J. HMG-CoA reductase inhibitors
and the risk of hip fractures in elderly patients. JAMA. 2000;283:3211-3216.
3. Chan KA, Andrade SE, Boles M, et al. Inhibitors of hydroxymethylglutaryl
coenzyme A reductase and risk of fracture among older women. Lancet. 2000;
355:2185-2188.
4. Scandinavian Simvastatin Survival Study Group. Randomised trial of choles-
terol lowering in 4444 patients with coronary heart disease: the Scandinavian Sim-
vastatin Survival Study (4S). Lancet. 1994;344:1383-1389.
Hershel Jick, MD
Boston University Medical Center
Lexington, Mass
5. Pedersen TR, Cook TJ, Færgeman O, et al. Safety and tolerability of choles-
terol lowering with simvastatin during 5 years in the Scandinavian Simvastatin Sur-
vival Study. Arch Intern Med. 1996;156:2085-2092.
6. National Osteoporosis Foundation. Physician’s Guide to Prevention and Treat-
ment of Osteoporosis. Belle Mead, NJ: Excerpta Medica Inc; 1998.
1. Scandinavian Simvastatin Survival Study Group. Randomised trial of choles-
terol lowering in 4444 patients with coronary heart disease: the Scandinavian Sim-
vastatin Survival Study (4S). Lancet. 1994;344:1383-1389.
2. Meier CR, Schlienger RG, Kraenzlin ME, Schlegel B, Jick H. HMG-CoA reduc-
tase inhibitors and the risk of fractures. JAMA. 2000;283:3205-3210.
3. Chan KA, Andrade SE, Boles M, et al. Inhibitors of hydroxymethylglutaryl-
coenzyme A reductase and risk of fracture among older women. Lancet. 2000;
355:2185-2188.
4. Wang PS, Solomon DH, Mogun H, Avorn J. HMG-CoA reductase inhibitors
and the risk of hip fractures in elderly patients. JAMA. 2000;283:3211-3216.
5. Mundy G, Garrett R, Harris S, et al. Stimulation of bone formation in vitro and
in rodents by statins. Science. 1999;286:1946-1949.
6. Edwards CJ, Hart DJ, Spector TD. Oral statins and increased bone-mineral den-
sity in postmenopausal women. Lancet. 2000;355:2218-2219.
In Reply: Dr El-Sohemy comments on recent findings that
pravastatin does not stimulate bone formulation in vitro by in-
creasing BMP-2. This mechanism is a current focus of interest
in the search of a possible pharmacologic explanation for the
observed effect of statins on bone, but it is not yet proven.
We stratified statin users and assessed the risk of having a
fracture in association with current exposure to simvastatin or
pravastatin, the 2 statins most often used in our study sample.
Compared with nonusers of any lipid-lowering drugs, ORs for
current use of simvastatin (90 patients and 671 controls) and
Problems in Pharmacoeconomic Analyses
pravastatin (10 patients and 131 controls) were 0.61 (95% con- To the Editor: The pharmaceutical industry in Australia is
fidence interval [CI], 0.47-0.79) and 0.34 (95% CI, 0.17- certainly aware that pharmacoeconomic analyses are com-
0.65), respectively. There was insufficient cholestyramine use plex and difficult to evaluate, as stated in the article by Dr
in our sample to perform a separate analysis of its effect on frac- Hill and colleagues,1 but we disagree with its conclusions. It
tures.
is also not obvious from the article that at least 2 of the
Drs Pedersen and Kjekshus report the findings of a sub- authors were involved previously in more than 1 stage of the
analysis of the 4S trial1 suggesting that simvastatin had no ef- analysis of each submission. This brings their perspective
fect on the risk of developing fractures. Reanalyses of “old” data into question.
sets that have been assessed with different goals in mind may
As pointed out in the Editorial by Drs Rennie and Luft,2 gov-
be misleading, unless it is certain that all outcomes were com- ernments have biases, too. Kassirer and Angell3 have pointed
prehensively recorded at the time of the study, even if they were out that cost-effectiveness analyses share many of the features
not the primary end point and not considered to be of interest of a review article. Both allow numerous judgment calls by the
in the context of the study hypothesis at that time. Another analyst, all of which may be biased.
possible explanation for their negative results may be the pos-
In determining the relevance of the highlighted problems,
sibility that the placebo group had less person-time of fol- it would have been helpful for Hill et al to describe their rela-
1922 JAMA, October 18, 2000—Vol 284, No. 15 (Reprinted)
©2000 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ by a Carleton University User on 11/09/2020

LETTERS
tionship to the recommendation of the Pharmaceutical Ben- nificant, its associated incremental CE [cost-effectiveness] would
efits Advisory Commitee (PBAC). Although they list this in- be comparable with many interventions.”
formation in their Table 1, the authors supply no data that would
have put in perspective the “problems with the interpretation
of pharmacoeconomic analyses” within the unique Australian
system and their relevance to the reimbursement decision.
The PBAC has a strong preference for economic evalua-
tions based on head-to-head randomized clinical trials,4 so it
is not surprising that the authors found that 62% of submis-
sions had “uncertainty” in the estimates of comparative clini-
cal efficacy, and 28.5% had “modelling issues.” The Austra-
lian pharmaceutical industry suggests that many of these
uncertainties and issues are matters of interpretation or a re-
flection of the uniqueness of the Australian system in contrast
to the global drug development environment.
Second, in handling quality-adjusted survival, there is gen-
eral agreement with Weinstein et al4 that utility values should
reflect those of society and not of individual patients. In prac-
tice, however, this recommendation may result in endorsing
poor-quality analyses. One such example is a study by Leung
et al5 in which the societal utility for 3 different treatments for
breast cancer was constructed on the basis of the assessments
made by a small group of patients (n=25) and health care pro-
viders (n=25), who examined slides describing the adverse ef-
fects of the treatments. Surprisingly, these values were a cru-
cial component of the cost-utility analysis.
These studies illustrate the kinds of problems that we encoun-
tered in interpreting pharmacoeconomic analyses. While we are
confident that analyses supporting statistically ineffective treat-
ments will become less frequent in the future, we fear that the
assessment of utility will remain a source of poor-quality stud-
ies. New guidelines need to be issued with regard to the collec-
tion of utility data and the application of standard clinical re-
search statistical testing to the measurements of utility and quality-
adjusted survival. The Italian Society of Pharmacoeconomics6 has
recently issued recommendations along these lines.
The role of economic analysis is to assist the decision-
making process, not to supply perfect data unique to the Aus-
tralian system.
Michael Stafford, MBA
Health Economics Committee
Australian Pharmaceutical Manufacturers Association
North Sydney, Australia
Disclaimer: Mr Stafford is a nonvoting member of the ESC and secretary of the
Health Economics Committee (HEC). The HEC of the Australian Pharmaceutical
Manufacturers Association (APMA) consists of 17 representatives from member
companies of the APMA.
Andrea Messori, PhD
Sabrina Trippoli, PhD
Azienda Ospedaliera Careggi
Firenze, Italy
Paola Dæalessandro, MA
Domenico Di Giorgio, PhD
Francesca Tosolini, PhD
Dipartimento per la Valutazione dei Medicinali
e la Farmacovigilanza
1. Hill SR, Mitchell AS, Henry DA. Problems with the interpretation of pharma-
coeconomic analyses: a review of submissions to the Australian Pharmaceuticals
Benefits Scheme. JAMA. 2000;283:2116-2121.
2. Rennie R, Luft H. Pharmacoeconomic analyses: making them transparent, mak-
ing them credible. JAMA. 2000;283:2156-2160.
3. Kassirer JP, Angell M. The Journal’s policy on cost-effectiveness analysis. N Engl
J Med. 1994;331:669-670.
4. Commonwealth Department of Human Services and Health. Guidelines for the
Pharmaceutical Industry on Preparation of Submissions to the Pharmaceutical Ben-
efits Advisory Committee: Including Major Submissions Involving Economic Analy-
ses. Canberra: Australian Government Publishing Service; 1995.
Ministero della Sanita´
Rome, Italy
1. Hill SR, Mitchell AS, Henry DA. Problems with the interpretation of pharma-
coeconomic analyses: a review of submissions to the Australian Pharmaceuticals
Benefits Scheme. JAMA. 2000;283:2116-2121.
2. Trippoli S, Messori A. Cost-effectiveness analyses of statistically ineffective treat-
ments. JAMA. 1998;280:1992-1993.
3. Hillner BE, Agarwala S, Middleton MR. Post hoc economic analysis of temo-
zolomide versus dacarbazine in the treatment of advanced metastatic melanoma.
J Clin Oncol. 2000;18:1474-1480.
4. Weinstein MC, Siegel JE, Gold MR, Kamlet MS, Russell LB. Recommendations
of the Panel on Cost-effectiveness in Health and Medicine. JAMA. 1996;276:
1253-1258.
5. Leung PP, Tannock IF, Oza AM, Puodziunas A, Dranitsaris G. Cost-utility analy-
sis of chemotherapy using paclitaxel, docetaxel, or vinorelbine for patients with
anthracycline-resistant breast cancer. J Clin Oncol. 1999;17:3082-3090.
6. Societa´ Italiana di Farmacoeconomia. Linee guida per la conduzione di studi
sull’efficacia e sul costo dei trattamenti farmacologici (Versione 1.1, Novembre 1999).
Giornale Farmacoeconomia. 1999;3:147-153.
To the Editor: The experience of the Comitato Inter-
Ministeriale Programmazione Economica (CIPE) of the Ital-
ian Ministry of Health (the body responsible for national drug
reimbursement negotiations) is similar to that of Dr Hill and
colleagues.1 From January 1999 to April 2000, we reviewed 212
applications by the pharmaceutical industry, of which fewer
than 10% included a pharmacoeconomic analysis. We identi-
fied 2 consistent aspects of inadequate pharmacoeconomic meth-
odology.
First, when a comparison between a new treatment and a
reference treatment does not find a significant clinical superi-
ority of the new treatment, pharmacoeconomic studies can con-
struct an analysis based on an unproven difference in effec- In Reply: Mr Stafford’s main criticism appears to be that the
tiveness. Even though this strategy has been discouraged,2-4 it conclusions of our study have been biased by a government or
nonetheless still occurs. Although we cannot disclose the de- health system perspective and the authors’ involvement in vari-
tails of such studies submitted to the CIPE, it is not difficult to ous aspects of the decision-making process (listing drugs on
find examples in the published literature. For example, in com- the Australian Pharmaceutical Benefits Schedule).
paring 2 treatments for metastatic melanoma, Hillner et al3 state
During the study, recommendations to the Australian fed-
that “although the base-case efficacy of TEM [temozolomide] eral health minister regarding which drugs in the schedule
compared with DTIC [dacarbazine] was not statistically sig- should be subsidized were made by 2 advisory committees com-
©2000 American Medical Association. All rights reserved.
(Reprinted) JAMA, October 18, 2000—Vol 284, No. 15 1923
Downloaded From: https://jamanetwork.com/ by a Carleton University User on 11/09/2020

LETTERS
prising more than 20 members. Assessments of companies’ sub- note on the first page of their article reveals that 1 of the au-
missions were made by a group of 16 evaluators. We analyzed thors was from the Pharmaceutical Evaluation Section, Phar-
maceutical Benefits Branch, Department of Health and Aged
Care. Furthermore, it is clearly stated that “The database of sub-
missions and the committee reports were scrutinized by 2 of
[the] authors (S.R.H. and D.A.H.).” As was made clear in an-
other footnote, the study received financial support from the
Commonwealth Department of Health and Aged Care. We be-
lieve that this was full disclosure. We point out that only those
involved in the analyses could be in a position to conduct such
an overview because the individual analyses, under the se-
crecy provisions of the Australian National Health Act, are kept
confidential.
It is ironic that a representative of the pharmaceutical in-
dustry is asking for more public disclosure. The pharmaceu-
tical industry may wish to further its interests by keeping ev-
erything secret, but, as we said in our Editorial,2 especially where
the public purse is involved, the need for secrecy is counter to
the interests of the public. The drugs in question had already
received official approval for clinical use. We believe that the
public’s health and money would have been far better served
had the entire submissions to the PBAC and the detailed analy-
ses by the PBAC been freely available to the public by putting
them on the Web. Mr Stafford’s complaint that facts were hid-
den from the reader will be more compelling when the indus-
try Mr Stafford represents agrees to adopt that open and public-
spirited approach.
the evaluations and the committees’ reviews systematically, us-
ing objective criteria. The results and conclusions reflect the
views of a wide group of professionals involved in the evalu-
ation process and not just those of the authors. While our ar-
ticle undoubtedly reflects the experiences of those who pur-
chase or subsidize the use of pharmaceuticals (rather than those
who manufacture them), we do not think this has a bearing
on the main conclusions of the article, which are that phar-
macoeconomic analyses are valuable, but complex and error
prone, that each analysis requires detailed evaluation to avoid
critical errors, and that traditional peer-review standards are
unlikely to lead to detection of many of these problems.
Stafford suggests that the Australian government has unique
requirements for the evaluation of cost-effectiveness. A num-
ber of national and regional governments and agencies have
introduced similar requirements.1,2 It is inevitable that there
will be some variation as a result of differences in health care
systems and economies, but the basic approach to the ap-
praisal of evidence appears quite common. For instance, the
evidentiary requirements of the Australian Pharmaceutical Ben-
efits Scheme3 and the United Kingdom National Institute for
Clinical Excellence (NICE)4 are similiar.
We are grateful for the comments of Dr Messori and col-
leagues, who present a similar government perspective from
Italy on the evaluation of cost-effectiveness data. We think it
likely that the use of clinical trial data in pharmacoeconomic
analysis will improve as standards for the conduct of such analy-
ses are more widely promulgated. We share their concerns about
the derivation of utility data from small and unrepresentative
groups of individuals, and we are working with the pharma-
ceutical industry in Australia to provide better guidelines for
this important aspect of pharmacoeconomic analysis.
Mr Stafford further states that the authors found so many of
the drug companies’ submissions to be inadequate because of
the PBAC’s preference for randomized controlled trials using
“head-to-head”comparisons. In essence, he is saying, to quote
Voltaire, that “the best is the enemy of the good.” In this case,
we support the PBAC, which produces plenty of evidence that
the best is even more the enemy of the inadequate, the unsub-
stantiated, and the unbelievable. We would think that Austra-
lian taxpayers would be delighted to see the rigor with which
their interests are being protected.
Suzanne Hill, MBBS, PhD
David Henry, FRCP, MBChB
School of Population Health Sciences
University of Newcastle
Drummond Rennie, MD
Harold S. Luft, PhD
Newcastle, New South Wales, Australia
Institutes for Health Policy Studies
University of California, San Francisco
Andrew Mitchell, BPharm, MMedSci
Commonwealth Department of Health and Aged Care
Woden, Australian Capital Territory
1. Hill SR, Michell AS, Henry DA. Problems with the interpretation of pharma-
coeconomic analyses: a review of submissions to the Australian Pharmaceutical
Benefits Scheme. JAMA. 2000;283:2116-2121.
2. Rennie D, Luft HS. Pharmacoeconomic analyses: making them transparent, mak-
ing them credible. JAMA. 2000;283:2158-2160.
1. Glennie JL, Torrance GW, Baladi JF, et al. The revised Canadian Guidelines for
the Economic Evaluation of Pharmaceuticals. Pharmacoeconomics. 1999;15:459-
468.
2. Preparatory Committee on Guideline Development. Dutch Guidelines for Phar-
macoeconomic Research. Amstelveen, the Netherlands: Ziekenfondsraad; 1999.
3. Australian Department of Health and Aged Care. Available at: http://
www.health.gov.au. Accessibility verified September 19, 2000.
4. National Institute for Clinical Excellence. Available at: http://www.nice.org.uk/
appraisals/appraisals.htm. Accessibility verified September 19, 2000.
Clinical Diagnosis of Carpal Tunnel Syndrome
To the Editor: To establish sensitivity, specificity, and likeli-
In Reply: Mr Stafford believes that Dr Hill and colleagues’1 con- hood ratios for a diagnostic test, the investigation must in-
nections to, and relationships with, the Australian Pharmaceu- clude both patients with and without the disease. However, in
tical Benefits Scheme brings their subsequent analysis into ques- their review of carpal tunnel syndrome (CTS), Drs D’Arcy and
tion. He states that the authors failed to reveal their involvement McGee1 define likelihood ratios as the “odds that a given find-
in these official government-sponsored analyses. In fact, a foot- ing would occur in a patient with CTS as opposed to one with-
1924 JAMA, October 18, 2000—Vol 284, No. 15 (Reprinted)
©2000 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ by a Carleton University User on 11/09/2020

LETTERS
out CTS,” but they excluded articles with asymptomatic con-
Carpal tunnel syndrome is experienced by patients who have
trol groups. The use of asymptomatic control subjects ensures sensory changes in a median nerve distribution reproducible
that patients are included who do not have any evidence of nerve by the Tinel and Phalen signs. This is the result of mechanical
compression. Thus, the 14 studies that D’Arcy and McGee ex- deformation of the median nerve in the carpal tunnel. With pro-
cluded on this basis should have been included in the assess-
ment of the CTS diagnostic criteria.
longed and persistent compression, nerve conduction is slowed
and, ultimately, irreversible injury occurs.
We also question their anatomic criteria. The hand diagram
for classic CTS as shown in Figure 3 in their article describes
symptoms involving the fourth and fifth digits. By contrast, clas-
sic anatomists describe the median and ulnar nerve sensory ter-
ritories as splitting the fourth digit.2 D’Arcy and McGee ex-
clude symptoms in the palm because this area is innervated from
a branch of the median nerve proximal to the carpal tunnel.
Yet, they appear to endorse a hand diagram that would imply
the presence of CTS if symptoms involve the little finger (ul-
nar nerve sensory distribution) and the upper extremity above
the wrist.
In the original article by Katz and Stirrat3 in which the hand
diagram was first described to assist in the diagnosis of CTS,
the authors define “classic” symptoms as “tingling, numbness
or decreased sensation with or without pain in at least 2 of dig-
its 1, 2, or 3. Palm and dorsum of the hand excluded; wrist pain
or radiation proximal to the wrist allowed.”3 In the original de-
scription of the hand diagram, these authors did not include
the fifth digit in the “classic” pattern, as is recommended in
other consensus criteria.4 In the article cited by D’Arcy and Mc-
Gee, Katz et al5 included the fifth digit because they had noted
that many patients with CTS complain of fifth digit symp-
toms. We suggest that these symptoms involving the fifth digit
are not the result of CTS (median nerve compression) but rather
concomitant ulnar nerve compression.
The clinician, as opposed to the physiologist, is interested in
2 outcomes: relief of the symptoms and prevention of irrevers-
ible nerve damage. Symptomatic patients with negative nerve con-
duction studies still may have CTS as determined by the bed-
side diagnostic criteria of Phalen.2 They do not need surgery for
prevention of irreversible injury. Conversely, a comatose pa-
tient cannot have CTS (an experience of conscious human be-
ings) but may have demonstrable median nerve compression in
the carpal tunnel as shown by electrodiagnosis.
D’Arcy and McGee would have clinicians reserve the term
“carpal tunnel syndrome” for patients with abnormal electro-
diagnostic test results. In their clinical example, however, they
diagnosed and treated what they might call “the condition for-
merly known as carpal tunnel syndrome” without nerve con-
duction tests. The physical findings of motor weakness and loss
of pain sensation are useful because they predict an abnormal
electrodiagnostic test result and identify patients who need sur-
gical intervention. To propose the electrodiagnostic test as the
gold standard for the diagnosis of the clinical syndrome is, how-
ever, a significant conceptual error.
Physicians should identify patients with symptomatic CTS and
treat them with conservative measures such as splints and ste-
roid injection to prevent median nerve injury. The clinical defi-
nition of CTS should be preserved. There are already adequate
descriptors of the more specific condition of median nerve com-
pression neuropathy in the carpal tunnel. Patients will be best
served if we maintain this distinction.
The appropriate management of patients with CTS requires
an accurate diagnosis and should be based on patient symp-
toms involving the median nerve distribution, physical find-
ings, and confirmation with nerve conduction studies.
Richard F. LeBlond, MD
Department of Internal Medicine
University of Iowa Health Care
Iowa City
Susan E. Mackinnon, MD
Christine B. Novak, PT, MS
Division of Plastic and Reconstructive Surgery
1. D’Arcy CA, McGee S. Does this patient have carpal tunnel syndrome? JAMA.
2000;283:3110-3117.
2. Phalen GS. The birth of a syndrome, or carpal tunnel revisited. J Hand Surg
Am. 1981;6:109-110.
William M. Landau, MD
Department of Neurology
Washington University School of Medicine
St Louis, Mo
In Reply: In answer to Dr Mackinnon and colleagues, we chose
to exclude studies using asymptomatic control subjects be-
cause such studies tend to inflate the specificity of a diagnos-
tic test and overestimate its accuracy.
1. D’Arcy CA, McGee S. Does this patient have carpal tunnel syndrome? JAMA.
2000;283:3110-3117.
2. Anderson JE. Grant’s Atlas of Anatomy. 7th ed. Baltimore, Md: Williams & Wilkins;
1978.
3. Katz JN, Stirrat CR. A self-administered hand diagram for the diagnosis of car-
pal tunnel syndrome. J Hand Surg. 1990;15A:360-363.
4. Rempel D, Evanoff B, Amadio PC, et al. Consensus criteria for the classification
of carpal tunnel syndrome in epidemiologic studies. Am J Public Health. 1998;
88:1447-1451.
5. Katz JN, Stirrat CR, Larson MG, et al. A self-administered hand symptom dia-
gram for the diagnosis and epidemiologic study of carpal tunnel syndrome. J Rheu-
matol. 1990;17:1495-1498.
Clinicians have no difficulty distinguishing a normal, asymp-
tomatic hand from one with classic CTS. Instead, they are in-
terested in how well bedside findings distinguish CTS from other
diagnoses that cause hand dysesthesias. If the patient has no
symptoms, physical examination for CTS is unnecessary. In ad-
dition, we believe it is important to distinguish patients’ sub-
To the Editor: I suggest that Drs D’Arcy and McGee1 should jective symptoms (eg, Katz hand diagram) from the objective
have entitled their article, “Does This Patient Have Abnormal sensory findings (eg, hypalgesia). Katz et al included the fifth
Median Nerve Conduction?”
finger in their “classic” pattern because they, like others, noted
©2000 American Medical Association. All rights reserved.
(Reprinted) JAMA, October 18, 2000—Vol 284, No. 15 1925
Downloaded From: https://jamanetwork.com/ by a Carleton University User on 11/09/2020

LETTERS
that patients with CTS frequently complain of symptoms in the terpretation of studies of acute stroke treatment. All recent stud-
fifth digit.1 We included their hand diagram in our analysis be- ies use time to presentation as a crucial variable, with the un-
cause it is a bedside finding that was clearly defined, and our
analysis shows it has modest diagnostic accuracy. Whether other
rating schemes of hand diagrams have better accuracy is un-
known. Nonetheless, in contrast to the subjective phenom-
enon of symptoms in the fifth finger, sensory loss is rarely dem-
onstrated in the fifth finger in CTS,1 consistent with the classic
anatomic teaching mentioned by Mackinnon et al.
We agree with Dr LeBlond that using electrodiagnosis as the
diagnostic standard has shortcomings and that there is still a
tension between clinical diagnosis and electrodiagnosis of CTS.
We discussed this problem fully in “The Diagnostic Standard
for CTS” section of our article.
Nonetheless, we are unaware of any other diagnostic standard
that is superior to the combination of compatible symptoms and
abnormal median nerve conduction. Perhaps our results could
best be regarded as identifying those bedside findings that dis-
tinguish electrodiagnostically confirmed CTS from other com-
mon electrodiagnoses that produce hand dysesthesias, such as
polyneuropathy, ulnar neuropathy, and cervical radiculopathy.
derlying assumption that early intervention with a vaso-
occlusive clot may restore circulation before irreparable tissue
destruction occurs. Time lag from symptom onset is believed to
be a marker for progression of the infarction cascade; thus, on
average, thrombolysis administered late causes more hemor-
rhage. But this assumption may be false. It may be that early pre-
sentation really signifies that, on average, a smaller volume of
cortex has been and will be infarcted, regardless of any inter-
vention and that it is the volume of infarcted cortex that deter-
mines outcome. The use of diffusion and perfusion magnetic reso-
nance imaging scans to show infarcted and at-risk cortex might
therefore be much more prognostic than time to presentation.
While no study has explicitly addressed this issue, the data
are already extant in the placebo arms of the large stroke stud-
ies; one needs merely to calculate mean National Institutes of
Health Stroke Scale score as a function of time to presenta-
tion. We hope that this important work will be undertaken and
published in the near future.
Darren Orbach, MD, PhD
David Levine, MD
Department of Neurology
New York University Medical Center
New York, NY
Christopher A. D’Arcy, MD
Division of Rheumatology
Steven McGee, MD
Department of Medicine
University of Washington
Seattle
1. Sherman DG, Atkinson RP, Cippendale T, et al. Intravenous ancrod for treat-
ment of acute ischemic stroke: the STAT Study: a randomized controlled trial. JAMA.
2000;283:2395-2403.
2. Jorgensen HS, Nakayama H, Raaschou HO, Olsen TS. Acute stroke: prognosis
and a prediction of the effect of medical treatment on outcome and health care
utilization. Neurology. 1997;49:1335-1342.
1. Katz JN, Stirrat CR, Larson MG, et al. A self-administered hand symptom dia-
gram for the diagnosis and epidemiologic study of carpal tunnel syndrome. J Rheu-
matol. 1990;17:1495-1498.
3. The National Institute of Neurological Disorders and Stroke rt-PA Stroke Study
Group. Tissue plasminogen activator for acute ischemic stroke. N Engl J Med. 1995;
333:1581-1587.
Outcomes of Ancrod in Acute Ischemic Stroke
4. International Stroke Trial Collaborative Group. The International Stroke Trial
(IST): a randomised trial of aspirin, subcutaneous heparin, both, or neither among
19435 patients with acute ischaemic stroke. Lancet. 1997;349:1569-1581.
5. CAST (Chinese Acute Stroke Trial) Collaborative Group. CAST: randomised pla-
cebo-controlled trial of early aspirin use in 20,000 patients with acute ischaemic
stroke. Lancet. 1997;349:1641-1649.
6. Clark WM, Wissman S, Albers GW, Jhamandas JH, Madden KP, Hamilton S,
for the Alteplase Thrombolysis for Acute Noninterventional Therapy in Ischemic
Stroke Group. Recombinant tissue-type plasminogen activator for ischemic stroke
3 to 5 hours after symptom onset: the ATLANTIS Study: a randomized controlled
trial. JAMA. 1999;282:2019-2026.
To the Editor: We noted a surprising result in Table 3 in the
article by Dr Sherman and colleagues1 discussing ancrod for
management of acute ischemic stroke. Patients in the placebo
group who presented before 3 hours had elapsed from the on-
set of stroke symptoms fared better than patients presenting
later, whether the latter were given placebo or ancrod.
One possible interpretation is that the time lag between symp-
tom onset and presentation to the hospital is actually an indirect
marker for stroke severity. There is a known correlation be- To the Editor: The article by Dr Sherman and colleagues1 on
tween initial stroke severity and prognosis.2 One might imagine the results of the STAT trial showed an advantage of ancrod
that patients with severe impairment from a large stroke would given within 3 hours from onset of ischemic stroke. However,
present later, on average, either because they are more likely to the comments made by these authors about the results of Eu-
be anosognosic or because they find it more difficult to summon ropean Stroke Treatment with Ancrod Trial (ESTAT), an on-
help. However, there are no empirical data addressing this ques- going European trial with ancrod administered 6 hours after
tion. The National Institute of Neurological Disorders and Stroke stroke onset, require clarification.
(NINDS) trial3 did not stratify patients who presented within 6
It is correct that an interim futility analysis based on 673 pa-
hours of symptom onset, nor did the International Stroke Trial4 tients followed up for 3 months showed that the adjusted prob-
or the Chinese Acute Stroke Trial,5 and too few patients were ability of success with the full projected sample of 1680 pa-
treated within the first 3 hours for meaningful subgroup analy- tients was too low to justify continuation of the trial and also
sis in the Alteplase Thrombolysis for Acute Noninterventional that the 90-day mortality was higher in the ancrod group. How-
Therapy in Ischemic Stroke study.6
ever, this second issue requires some clarification, especially since
The possibility that time to presentation is a proxy for stroke this information is not presented on the BASF Web site (Sher-
severity is important because, if true, it would confound the in- man et al,1 reference 31). The ESTAT Independent Data and Safety
1926 JAMA, October 18, 2000—Vol 284, No. 15 (Reprinted)
©2000 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ by a Carleton University User on 11/09/2020

LETTERS
Monitoring Board used a group-sequential design2 to compare
overall mortality (not only 90-day mortality) between the an-
crod and placebo treatment groups and at no stage was the bound-
ary of safety crossed. In other words, while there was a differ-
ence in the 90-day mortality in the patients included in the interim
analysis, this is not true of total mortality in the sample. The se-
quential monitoring included 1221 patients at the time of study
discontinuation, and there was never a statistically significant
excess mortality rate in the ancrod group.
1. Sherman DG, Atkinson RP, Cippendale T, et al. Intravenous ancrod for treat-
ment of acute ischemic stroke: the STAT Study: a randomized controlled trial. JAMA.
2000;283:2395-2403.
2. Whitehead J. Application of sequential methods to a phase III clinical trial in
stroke. Drug Inf J. 1993;27:733-740.
3. The National Institute of Neurological Disorders and Stroke rt-PA Stroke Study
Group. Tissue plasminogen activator for acute ischemic stroke. N Engl J Med. 1995;
333:1581-1587.
4. Hacke W, Kaste M, Fleschl C, et al. Intravenous thrombolysis with recombi-
nant tissue plasminogen activator for acute hemispheric stroke. JAMA. 1995;274:
1017-1025.
5. Hacke W, Kaste M, Fleschl C, et al. Randomised double-blind placebo-
controlled trial of thrombolytic therapy with intravenous alteplase in acute ische-
mic stroke (ECASS II). Lancet. 1998;352:1245-1251.
We await the analysis of the full data set from ESTAT, with
the follow-up data from all patients enrolled, to gain a clear
understanding of the possible benefits and hazards of ancrod
in acute ischemic stroke. Any conclusions in advance of such
an analysis would be inappropriate. Premature speculation on
an increased death rate in ESTAT will only cast doubts on the
validity of the positive STAT results and prejudice the poten-
tial therapeutic usefulness of ancrod in this setting. A possible
explanation for the failure of ESTAT to show a benefit of an-
crod as opposed to STAT is the longer time window. Indeed,
the only successful trial in acute ischemic stroke with another
agent, tissue plasminogen activator, was also obtained with a
time window of 3 hours,3 while 2 subsequent studies with a
time window of 6 hours failed to demonstrate a benefit.4,5
We await with great interest the analysis of the complete
ESTAT data and their comparison and incorporation with the
data from STAT to allow a more complete understanding of
the effects of ancrod in the treatment of acute ischemic stroke.
In Reply: We appreciate Drs Orbach and Levine highlighting
the worse outcomes of patients who received placebo with in-
creasing time to treatment, as this is an observation that also
struck us. We can verify their explanation that those who re-
ceived placebo and entered our study later after stroke onset
had worse stroke severity than those entering earlier. Specifi-
cally, the proportion of the placebo group with Scandinavian
Stroke Scale scores of 30 to 39 (the group with the mildest defi-
cits permitted in the study) decreased from 46.2% among those
enrolled within 2 hours of stroke onset to 39.2% among those
enrolled at 2 to 3 hours of onset, and further to 35.0% among
those enrolled more than 3 hours after stroke onset. While the
number of patients enrolled at less than 2 hours (n=13) and
more than 3 hours (n=40) is relatively small, the impact of stroke
severity on outcome is profound and, as suggested by Orbach
and Levine, this probably explains the observed worsening
outcome.
Two additional points deserve brief attention: first, as the
NINDS trial1 strictly adhered to a 3-hour time limit, the im-
plicit criticism that the NINDS team failed to stratify patients
within 6 hours of stroke onset seems unwarranted. More im-
portantly, our study has no data that will permit us to explore
in more detail reasons for the change in stroke severity with
increasing time to presentation.
Jean-Marc Orgogozo, MD
Service de Neurologie A
Hôpital Pellegrin
Bordeaux, France
Marc Verstraete, MD
Centre de Biologie Vasculaire et Mole´culaire
Universite´ de Louvain
Leuven, Belgium
Richard Kay, PhD
Parexel International
Sheffield, England
Independent Data and Safety Monitoring Board for ESTAT
David G. Sherman, MD
Division of Neurology
University of Texas Health Science Center
San Antonio
Michael Hennerici, MD
Department of Neurology
University of Heidelberg
Mannheim, Germany
David E. Levy, MD
Knoll Pharmaceutical Co
Mount Olive, NJ
1. The National Institute of Neurological Disorders and Stroke rt-PA Stroke Study
Group. Tissue plasminogen activator for acute ischemic stroke. N Engl J Med. 1995;
333:1581-1587.
Gian-Luigi Lenzi
Dept Sienze Neurologische
Universita di Roma “la Sapienza”
Rome, Italy
for the Steering Committee for ESTAT
The letter from Dr Orgogozo et al also was shown to Drs Sherman and Levy.—ED.
©2000 American Medical Association. All rights reserved.
(Reprinted) JAMA, October 18, 2000—Vol 284, No. 15 1927
Downloaded From: https://jamanetwork.com/ by a Carleton University User on 11/09/2020