Page 116 - Human Lyme Neuroborreliosis

100 David S. Younger

Practice Parameters

The first of two evidenced-based Practice Parameters published by the
AAN in 1996 [2] was based upon a literature search that yielded fifty-nine
relevant articles, only a few of which compared different treatment regimens
for LNB [3-5]. The authors concluded that CNS involvement probably
required parenteral antimicrobial therapy with a third generation
cephalosporin, while limited European data [6] suggested that oral regimens
might be equally efficacious in acute meningitis.

The literature search forming the second Practice Parameter [7] published
in 2007 disclosed thirty-seven articles with assessable data, eight of which [6,
8-14] provided data on patients with definite LNB, and all were European-
based. An aggregated analysis failed to demonstrate any difference in outcome
whether patients received oral doxycycline or parenteral beta-lactam therapy.
With an overall response rate of doxycycline to parenteral penicillin or
ceftriaxone of 98.6% and associated narrow confidence intervals, there were
no apparent clinically or statistically significant differences between the oral
and parenteral regimens.

However, there were obvious problems in this analysis that pointed to a
lack of epidemiological rigor and generalizability of their findings from the
European to the U.S. patients. First, all of the eight studies [6, 8-14] were
Class III or Class IV with regard to clinical outcome according to a four-tiered
classification-of-evidence scheme [15]. In that classification [15], Class I
studies were judged to have a high quality and low risk of bias; Class II were
of moderate quality and risk of bias. Class III studies had a high risk of bias.
Class IV studies were those with very highest risk of bias. French and
Gronseth [15] and the Editors of the journal Neurology [16] recognized the
importance of this evidence-based classification noting that the evidence
classification did not relate to the studies per se, but rather to the questions
addressed by them. Thus, as the same study could contain a high level of
evidence (Class I) for one question but a low level (Class III or IV) for
another, two of the eight European studies [6, 12] neither of which were Class
I or II with regard to clinical outcomes, were both Class II in at least one of
their predetermined objective measures of disease activity. With low levels of
evidence for the European studies suggesting a high risk of intrinsic bias, it
would be appropriate to question the validity of a meta-analysis claiming
comparable response rates for doxycycline to parenteral antibiotic therapy for
the outcome of neurological disease. While U.S. patients with the same
manifestations of neuroborreliosis might be similarly responsive to

Complimentary Contributor Copy