Local Anesthetics for the Treatment of Neuropathic Pain:
On the Limits of Meta-Analysis
James P. Rathmell, MD*, and Jane C. Ballantyne, MD, FRCA†
*University of Vermont College of Medicine, Fletcher Allen Health Care Center for Pain Medicine, Burlington; and
†Harvard Medical School, Massachusetts General Hospital Pain Center, Boston
studies, the methodology, and the interpretation of
both contributing data and analytic results. When high
quality, homogeneous trials are combined using
sound methodology, meta-analysis can be a powerful
predictive tool (1). When high-quality homogeneous
trials are not available, as is often the case in the pain
literature, the validity of a meta-analysis may depend
more on the analysts’ interpretation of available data
than on the strength of the methodology (2–5). In this
issue of Anesthesia & Analgesia, Tremont-Lukats et al.
(6) present a systematic review and meta-analysis ex-
amining the effectiveness of systemic administration
of local anesthetics for relief of neuropathic pain. The
methodology used by these authors is flawless, but
have the analyses contributed new and clinically use-
Research begins with an observation about possible
cause and effect; the observation leads to formulation
of a hypothesis that then becomes the basis of the
research question. Meta-analysis should be no differ-
ent from other forms of research, i.e., a question
should be posed, and then the data should be sought
to fit the question; the question should not be manip-
ulated to fit the available data. Yet, when the available
data are scanty, the temptation to manipulate the
question is great. One way to ensure a result (and a
publication) is to combine studies with unreasonable
heterogeneity. Random effects statistical models used
in meta-analysis are designed to account for heteroge-
neity, but they cannot account for excessive heteroge-
neity (7–9). When excessively heterogeneous trials are
combined in a meta-analysis, complex issues tend to
s meta-analysis a powerful predictive tool or a de-
ceptive device that leads the average practitioner
astray? The answer depends on the contributing
be oversimplified, and conclusions are misleading.
The classic example in the anesthesia literature is the
meta-analysis by Rodgers et al. (10) that finds a 30%
reduction in mortality overall when neuraxial anesthe-
sia is used during surgery. The conclusion of these
authors is that the findings “support more widespread
use of neuraxial blockade anesthesia.” Yet reduced
mortality is found in only two surgical subsets, and
the authors’ conclusion is grossly misleading (9). In
this case, the question “Is neuraxial anesthesia always
better than no neuraxial anesthesia?” was tailored to
fit the vast amount of available data. Thus, it was not
a reasonable question.
In the present meta-analysis, the question “Is sys-
temic administration of local anesthetic effective for
the relief of neuropathic pain?” seems reasonable, but
what does the question really entail? The effectiveness
of an analgesic is not simply a matter of whether the
analgesic can alter a pain score to a statistically signif-
icant extent. More important is whether the analgesia
is clinically meaningful and whether the side effects
are tolerable. The analyses of pain scores presented by
Tremont-Lukats et al. (6) convincingly demonstrate
that both short-term infusions of IV lidocaine and
longer-term oral mexiletine administration produce a
small improvement in pain (11 mm on a 0–100 mm
scale) compared with placebo. Lidocaine, tocainide,
and mexiletine were no better than other analgesics in
206 patients. The authors recognize that there is sig-
nificant heterogeneity; in particular, differences in
pain causes are an important source of clinical and
statistical heterogeneity. They describe several subset
analyses that highlight sources of heterogeneity. They
also discuss the more difficult question of whether a
mean difference of 11 mm represents a true clinical
difference for patients. Although the comparisons be-
tween systemic local anesthetics and other analgesics
may not have been worth combining in a meta-
analysis, one could not wish for a better culling of the
literature, better analyses, or a better interpretation of
the data and analyses with regard to the analgesic
Accepted for publication June 15, 2005.
Address correspondence and reprint requests to James P. Rath-
mell, MD, Center for Pain Medicine, 62 Tilley Dr., South Burlington,
Vermont 05403. Address e-mail to firstname.lastname@example.org.
©2005 by the International Anesthesia Research Society
Anesth Analg 2005;101:1736–70003-2999/05
efficacy of systemic local anesthetics versus placebo.
But what about side effects? Here, the present analyses
fall down, and the quality of the contributing studies
lets down the meta-analysts. All they can do is count
the number of patients reporting side effects because
this is the only information available in most studies.
More patients receiving systemic local anesthetics re-
port side effects than patients receiving placebo but
not compared with patients receiving alternative an-
algesics. Yet, without knowing the nature, severity,
and interference with treatment of these side effects,
these results are essentially meaningless. For example,
nausea is often less well tolerated than pain, so this
side effect may severely limit the clinical utility of
systemic local anesthetics. And the small improve-
ment in pain may not be worth having for patients
experiencing troublesome side effects.
How much do we gain from this type of analysis?
The average practitioner wants to know how and
when to use these drugs in the care of their patients.
There are few good quality or reasonably sized ran-
domized trials to guide their choices. The present
study demonstrates modest pain reduction using ei-
ther lidocaine or mexiletine for a wide range of neu-
ropathic pain syndromes. But the limitations of the
contributing studies preclude drawing useful conclu-
sions about the adverse effect profiles of these drugs.
Has it been worth it, then, to invest enormous effort
into a meta-analysis that cannot answer the research
question “Is systemic administration of local anesthet-
ics effective for the relief of neuropathic pain?”
1. Lau J, Antman EM, Jimenez-Silva J, et al. Cumulative meta-
analysis of therapeutic trials for myocardial infarction. N Engl
J Med 1992;327:248–54.
2. Sacks HS, Berrier J, Reitman D, et al. Meta-analyses of random-
ized controlled trials. N Engl J Med 1987;316:450–5.
3. Jadad AR. Meta-analysis in pain relief: a valuable but easily
misused tool. Curr Opin Anaesthesiol 1996;9:426.
4. Naylor CD. Meta-analysis and the meta-epidemiology of clini-
cal research: meta-analysis is an important contribution to re-
search and practice but it’s not a panacea (Editorial). BMJ 1997;
5. Souter MJ, Signorini DF. Meta-analysis: greater than the sum of
its parts (Editorial)? Br J Anaesth 1997;79:420–1.
6. Tremont-Lukats IW, Challapalli V, McNicol ED, et al. System-
atic administration of local anesthetic agents to relieve neuro-
pathic pain: systematic review and meta-analysis. Anesth
7. Mantel N, Haenszel W. Statistical aspects of the analysis of data
from retrospective studies of disease. J Natl Cancer Inst 1959;
8. DerSimonian R, Laird N. Meta-analysis in clinical trials. Contr
Clin Trials 1986;7:177–88.
9. Ballantyne JC. Does epidural analgesia improve surgical out-
come (Editorial)? Br J Anaesth 2004;92:1–3. Abstract.
10. Rodgers A, Walker N, Schug S, et al. Reduction of postoperative
mortality and morbidity with epidural or spinal anaesthesia:
results from overview of randomised trials. BMJ 2000;321: