`DOI 10.1186/s12917-016-0703-y
`
`BMC Veterinary Research
`
`Antiepileptic drugs' tolerability and safety —
`a systematic review and meta-analysis of
`adverse effects in dogs
`
`CrossMark
`
`
`
`Marios Charalambous', Sara K Shivapour2, David C. Brodbelt3 and Holger A. Volk4
`
`Abstract
`
`Background: The safety profile of anti-epileptic drugs (AEDs) is an important consideration for the regulatory
`bodies, owners and prescribing clinicians. Information on their adverse effects still remains limited. A systematic
`review including a meta-analytic approach was designed to evaluate existing evidence for the safety profile of AEDs
`in canine patients. Electronic searches of PubMed, CAB Direct and Google scholar were carried out without date or
`language restrictions. Conference proceedings were also searched. Peer-reviewed full-length studies reporting
`adverse effects of AEDs in epileptic and healthy non-epileptic dogs were included. Studies were allocated to three
`groups based on their design. Individual studies were evaluated based on the quality of evidence (study design,
`study group sizes, subject enrolment quality and overall risk of bias) and the outcome measures reported
`(proportion of specific adverse effects for each AED, prevalence and 95 % confidence interval of the affected
`population in each study and comparative odds ratio of adverse effects for AEDs).
`Results: Ninety studies, including six conference proceedings, reporting clinical outcomes of AEDs' adverse effects
`were identified. Few studies were designed as blinded randomised controlled clinical trials. Many studies included
`low canine populations with unclear criteria of subject enrolment and short treatment periods. Direct comparisons
`suggested that imepitoin and levetiracetam might have a better safety profile than phenobarbital, whilst the latter
`might have a better safety profile than potassium bromide. However, none of these comparisons showed a
`statistically significant difference. Comparisons between other AEDs were not possible as a considerable amount of
`studies lacked power calculations or adequate data to allow further statistical analysis. Individual AED assessments
`indicated that levetiracetam might be one of the safest AEDs, followed by imepitoin and then phenobarbital and
`potassium bromide; these findings were all supported by a strong level of evidence. The safety profile in other
`AEDs was variable, but weak evidence was found to permit firm conclusions or to compare their safety to other
`AEDs.
`Conclusions: This systematic review provides objective evaluation of the most commonly used AEDs' adverse
`effects. Adverse effects usually appeared mild in all AEDs and subsided once doses and/or serum levels were
`monitored or after the AED was withdrawn. Although phenobarbital might be less safe than imepitoin and
`levetiracetam, there was insufficient evidence to classify it as an AED with a high risk of major adverse effects. It is
`important for clinicians to evaluate both AEDs' effectiveness and safety on an individual basis before the selection
`of the appropriate monotherapy or adjunctive AED therapy.
`
`Keywords: Systematic review, Meta-analysis, Epilepsy, Canine, Antiepileptic drugs, Safety, Side effects
`
`Correspondence: marios.charalambous.15@ucl.ac.uk
`'Faculty of Brain Sciences, UCL Institute of Neurology, University College
`London, London WC1E 6BT, UK
`Full list of author information is available at the end of the article
`
`$ (cid:9)
`
`EXHIBIT
`
`efiZ/1— 1/43
`-/z) -/6
`
`1
`Its.
`1
`
`BioMed Central
`
`0 2016 Charalambous et al. Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
`International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
`reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
`the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
`(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
`ARGENTUM Exhibit 1053
` Argentum Pharmaceuticals LLC v. Research Corporation Technologies, Inc.
`IPR2016-00204
`
`Page 00001
`
`(cid:9)
`(cid:9)
`
`
`Charalambous et al. BMC Veterinary Research (2016) 12:79 (cid:9)
`
`Page 2 of 44
`
`Background
`In human medicine, a plethora of new antiepileptic
`drugs (AEDs) have been developed over the years for
`use either as monotherapy or adjunctive therapy [1].
`Many of these drugs are now also used in veterinary
`medicine. This has led to an increase in the arsenal of
`AEDs used to treat canine epilepsy. As a rule, AEDs are
`evaluated on the grounds of their effectiveness and
`safety through clinical trials and experimental laboratory
`studies before they are approved for use in patients by
`the regulatory authorities, e.g. the European Medicines
`Agency (EMA) or the US Food and Drug Administration
`(FDA) [2]. The safety profile of drugs is an important
`consideration for their approval by the authorities and
`use by prescribing clinicians on their clients' animals [2, 3].
`It affects clinicians' decisions to prescribe specific
`AED(s), as serious adverse effects can lead to chronic
`complications or even death. Less serious, but none-
`theless important, adverse effects can significantly im-
`pact quality of life, leading to systematic illness which
`may increase the overall cost of treatment [3, 4].
`Ultimately, the benefits of an effective AED may be
`outweighed by its adverse effects, and the latter should
`be always taken into consideration.
`Many potential adverse effects for AEDs have been re-
`ported, but the evidence behind the severity of these ef-
`fects or the likelihood of their occurrence has not been
`systematically compiled [5, 6]. Randomised clinical trials
`(RCTs) are a considerable source of evidence for some
`common or expected adverse effects [4]. However, infor-
`mation about serious, rare, and/or long-term adverse
`effects can typically be found in studies such as case
`reports, case series and observational studies [7, 8].
`Consequently, the clinician will need to search for infor-
`mation from sources other than RCTs [7, 8]. Identifica-
`tion of all relevant studies can be time-consuming and
`for a busy practitioner it may be more effective to review
`this information via a systematic review. Systematic re-
`views are one of the most powerful and reliable tools to
`assess the severity and the probability of occurrence of
`AEDs' adverse effects across the spectrum of primary
`literature [9-12].
`Although evidence for AEDs' efficacy has been re-
`cently reported and evaluated in a systematic review
`[13], it has been suggested that, apart from the efficacy,
`the selection of the appropriate AED should be also
`largely influenced by its safety profile [14]. To our know-
`ledge there is only one systematic review in the field of
`canine epilepsy which evaluated the safety profile of a
`single AED, potassium bromide, across species and aeti-
`ology of seizures [15]. However, a systematic review of
`the adverse effects observed during treatment with any
`AED(s) in dogs, as well as AEDs' safety profile compari-
`sons, has not been reported. The aim of this systematic
`
`review was to perform an objective analysis of AEDs' ad-
`verse effects in dogs, in order to provide evidence-based
`information on AEDs' safety profiles.
`
`Methods
`Search strategy
`The literature search aimed to identify all studies asses-
`sing or reporting the adverse effects of an AED in dogs.
`Specifically, studies were evaluated based on the inclu-
`sion criteria below:
`
`• Criterion 1-Type of study: Peer-reviewed studies in
`English (or translated). Experimental laboratory ani-
`mal studies, clinical trials, observational and descrip-
`tive studies were included.
`• Criterion 2-Case definition: For the clinical studies,
`dogs with IE were included as previously defined
`[13]. Briefly this required dogs within a certain age
`range, unremarkable interictal neurological status
`and diagnostic investigation for seizures. For the ex-
`perimental laboratory animal studies (ELAS), healthy
`non-epileptic dogs were also included; for the latter
`a clear diagnostic investigation or health statement
`should have been reported in the study to exclude
`the possibility of underlying diseases.
`• Criterion 3-Treatment: Dogs treated with any AED
`available used in canine IE were included. Doses and
`serum concentrations of AEDs, frequency of drug
`administration and treatment period were consid-
`ered important information to record. Dogs treated
`with methods other than pharmacological interven-
`tion, e.g. homoeopathy methods, surgery, food trials,
`nerve stimulation, were excluded.
`• Criterion 4-Outcome: Studies had to assess or report
`adverse effects following administration of AED(s) in
`canine subjects. Studies were conducted either to
`specifically assess or report AED(s)' safety (primary
`evidence studies) or to assess an outcome other than
`AED(s)' safety (i.e. efficacy), while also reporting ad-
`verse effects (supportive evidence studies). Assess-
`ment of the adverse effects should have been
`performed by the investigators or owner.
`
`Search strategies included use of electronic search en-
`gines for publication databases, searching of reference
`lists of published papers and proceedings of relevant sci-
`entific conferences. Electronic databases used were Pub
`Med (www.ncbi.nlm.nih.gov/PubMed), CAB Abstracts
`(www.cabdirect.org) and Google Scholar (www.scholar.-
`google.com). Final electronic searches were carried out
`on 30 February 2015 by the primary and the second
`author independently, with no date or language res-
`trictions. The search terms used in both search engines
`were as follows: (dog OR dogs OR canine) AND
`
`Page 00002
`
`
`
`Charalambous et al. BMC Veterinary Research (2016) 12:79 (cid:9)
`
`Page 3 of 44
`
`[(phenobarbital OR phenobarbitone OR primidone OR
`PBr OR KBr OR potassium bromide OR bromide OR
`nimodipine OR zonisamide OR ELB138 OR imepitoin
`OR levetiracetam OR verapamil OR gabapentin OR gaba
`OR topiramate OR felbamate OR pregabalin) OR [(treat-
`ment OR management) AND (epilepsy OR seizures)]
`OR (anti-convulsant OR anti-seizuring OR anti-epileptic
`OR AED) AND (safety OR safe OR adverse-effect OR
`adverse-effect OR effect OR undesirable effect OR toler-
`ability OR toxicity OR drug toxicity OR reactions OR
`disease). Hand searching for articles from the reference
`lists of publications and searching major veterinary
`neurology conference meeting proceedings from 1970 to
`2015 and relative textbook chapters was carried out by
`the primary and second authors independently. Confer-
`ence proceedings were searched for the Annual Con-
`gresses of the European Society and College of
`Veterinary Neurology (ESVN/ECVN) and the American
`College of Veterinary Internal Medicine (ACVIM).
`Other conference proceedings were searched only if the
`reference list of identified publications indicated this. All
`items returned by the search engines, hand searches and
`correspondence were recorded and entered into the
`screening process.
`
`non-randomised ELAS (NRELAS), uncontrolled clinical
`trials (UCTs) and uncontrolled ELAS (UELAS), cohort,
`case—control and cross sectional studies and lastly case
`series and reports [16-18]. Accordingly, the studies were
`allocated based on their design to one of three groups,
`i.e. bRCTs, bRELAS, nbRCTs and nbRELAS (first group),
`NRCTs, NRELAS, UCTs, UELAS, cohort, case—control
`and cross-sectional studies (second group) and case series
`and reports (third group).
`As a general rule, the studies in the first group (bRCTs
`and bELAS in particular) were considered to provide
`higher quality evidence, followed by the studies in the
`second and third group. In addition, a three-part system
`of evidence quality assessment to indicate the strengths
`and weaknesses of each study within each group was
`used [13, 19]: (a) study group sizes, (b) subject enrol-
`ment quality and (c) overall risk of bias based on
`Cochrane [20] and Syrcle's [21] 'risk of bias' assessment
`tool in order to provide an indicator of confidence asso-
`ciated with the findings of each study. For instance,
`bRCTs or bRELAS with large group sizes, clear inclusion
`criteria, thorough diagnostic investigations and low over-
`all risk of bias were considered to provide the highest
`available quality of evidence.
`
`Study selection
`Restrictions based on publication date or language were
`not imposed. Studies written in non-English language
`were assessed initially based on an English translation
`(Google Translate software) and then verified by a veter-
`inarian fluent in the language of publication.
`A two-stage screening process was used [13] and the
`process was performed by the primary author. Firstly,
`studies of relevance to the systematic review objectives
`were identified (stage 1) and, secondly, studies likely to
`provide evidence of the highest available quality and suf-
`ficient detail for assessing the outcome measures and
`methodology were selected (stage 2). Stage 1 of the
`screening process identified from the total search results
`any studies that: (a) fulfilled inclusion criterion 1 and (b)
`reported findings related to the adverse effects and safety
`of AEDs administered in dogs. Stage 1 assessment evalu-
`ated the retrieved papers' titles and abstracts only. At
`stage 2, papers were selected for full data extraction ac-
`cording to the inclusion criteria 2, 3 and 4 and were
`evaluated in detail on the grounds of the quality of evi-
`dence and outcomes by MC.
`
`Assessment of quality of evidence
`Blinded RCTs (bRCTs) and blinded randomised ELAS
`(bRELAS) were considered most likely to produce higher
`quality evidence, followed by non-blinded RCTs
`(nbRCTs) and non-blinded randomised ELAS (nbRE-
`LAS), then non-randomised clinical trials (NRCTs) and
`
`Study group sizes
`This characteristic was categorized for each study using
`the following system [13, 19]: (a) >50 subjects per group
`('good' number of subjects), (b) 20-50 subjects per
`group ('moderate' number), (c) 10-19 subjects per group
`('small' number) and (d) <10 subjects per group ('very
`small' number).
`
`Assessment of subject enrolment quality
`Data on investigations to reach the diagnosis of IE were
`retrieved to evaluate the quality of subject enrolment in
`each study as 'well characterized', 'fairly characterized',
`'poorly characterized' or 'unclear.' Well characterized
`diagnoses were defined as diagnostic investigations that
`included clinical signs and thorough test results consist-
`ent with the diagnosis of IE; specifically, the signalment,
`the absence of neurological deficits between the ictal
`phases, unremarkable routine biochemical and haemato-
`logical blood tests and imaging results (including brain
`MRI and/or CT) and/or normal cerebrospinal fluid
`(CSF) analysis for all cases of the study. Fairly character-
`ized, used for intermediate situations, were defined as
`diagnostic investigations that were based on signalment,
`clinical examination and basic diagnostic investigation
`(i.e. blood tests only), with only some study cases having
`had advanced brain imaging and/or CSF analysis. Poorly
`characterized were defined as diagnostic investigations
`that were based on signalment, clinical examination
`and/or basic diagnostic investigation (i.e. blood tests)
`
`Page 00003
`
`
`
`Charalambous et al. BMC Veterinary Research (2016) 12:79 (cid:9)
`
`Page 4 of 44
`
`only. Unclear related to reports where the approach to
`diagnosis of IE was not clearly stated (e.g. when clinical
`signs were not stated and insufficient or no details of
`diagnostic tests were provided or when dogs with IE
`were included without reporting details on diagnostic
`investigation).
`For the ELAS, which included non-epileptic healthy
`animals, 'clearly characterized' were the studies that de-
`fined diagnostic investigations and thorough test results
`to exclude any systemic illness; 'unclear' were character-
`ized when diagnostic investigations to rule out diseases
`were not clearly stated or when dogs were included and
`considered healthy without reporting details on diagnos-
`tic procedures.
`
`Assessment of overall risk of bias
`The overall risk of bias in the clinical trials was assessed
`based on the criteria of the Cochrane 'risk of bias' as-
`sessment tool [20]. Syrcle's 'risk of bias' assessment tool
`[21] was used to assess the overall risk of bias in ELAS.
`The latter tool is an adapted version of the Cochrane
`one and was designed to facilitate critical appraisal of
`evidence from ELAS.
`Each of the following study components was catego-
`rized as presenting a 'high', 'low' or 'unclear' risk of
`introducing bias to the study findings: random sequence
`generation, allocation concealment, blinding of partici-
`pants and personnel, blinding of outcome assessment,
`completeness of outcome data, selective reporting of
`outcomes and other sources of bias. For ELAS, two fur-
`ther components-random housing and baseline charac-
`teristics of dogs - were also assessed and mentioned as
`part of the "other sources of bias" section. Case series
`and reports as well as observational studies were consid-
`ered to be of high overall risk of bias.
`
`Level of the studies' evidence
`The level of evidence provided for the safety profile of
`each AED was based on the overall quality of evidence
`of the studies. The level of evidence was allocated ac-
`cording to a previous similar system [13, 19] which was
`extensively modified for the needs of the current study:
`'strong' evidence was provided for the safety profile
`when at least one bRCT and/or bRELAS reported or
`assessed the adverse effects of an AED; 'weak' evidence
`was provided for the safety profile when bRCTs and/or
`bRELAS were not available.
`
`Assessment of outcome measures
`The outcome measure of this study was the evaluation
`of the safety profile of AED(s) administered in dogs. The
`adverse effects were organized by body system (e.g.
`neurological, gastro-intestinal, dermatological, etc.) and
`types, including type I (dose dependent and predictable)
`
`and type II (idiosyncratic-dose independent and unpre-
`dictable). Different terms used by the studies but de-
`scribing the same adverse effects (e.g. drowsiness and
`somnolence, wobbly gait and ataxia, lethargy and sed-
`ation, etc.) were considered synonymous and only one
`term was selected for use in the analysis. The outcome
`measure was assessed according to the methods below:
`
`Proportion of specific adverse effects for each AED
`This was expressed as a percentage and calculated for
`each AED by dividing the number of studies that re-
`ported a specific adverse effect by the total number of
`the studies for this AED. If an AED was used as a mono-
`therapy and adjunctive therapy, further calculations were
`also performed for each sub-category.
`
`Prevalence and 95 % confidence interval of the affected
`population in each study
`Prevalence was expressed as a percentage and calculated
`for each study by dividing the number of subjects that
`developed adverse effects during the specified study
`period by the total size of the study population. The
`95 % confidence interval (95 % CI) of the proportion of
`study animals that developed adverse effects related to
`the AED(s) was calculated by standard methods [22].
`This was used as a further indicator of an AED's safety
`profile. If the 95 % CI of affected dogs (based on 95 %
`CI calculations) were 50 %, then it was considered that
`the majority of the study population experienced adverse
`effects.
`For each study, the period of treatment, AED's doses
`and serum levels were reported with the aim to evaluate
`the association of these values with the prevalence of
`each AED's adverse effects.
`
`Statistical analysis
`For the comparison groups' studies, a further approach
`was conducted to identify statistical differences between
`studies with regards to reported adverse effects. For each
`AED study, the total number of patients experiencing
`adverse effects and/or the number of patients experien-
`cing specific adverse effects (e.g. sedation, ataxia, poly-
`uria, etc.) in all therapeutic groups were retrieved. The
`odds ratio (OR) was then estimated in order to indicate
`the increased or decreased odds of observing a specific
`adverse effect(s) in total for an AED compared to its
`control group (comparison AED or placebo or untreated
`animals). Statistical analysis was undertaken following
`the guidelines of the Handbook of the Cochrane Collab-
`oration 5.0. The OR for dichotomous data was calcu-
`lated using the random-effects model in Review
`Manager 5.3. Heterogeneity between studies was calcu-
`lated using the Chi square test and was considered to be
`heterogeneous when P 0.1. 12 values of no more than
`
`Page 00004
`
`
`
`Charalambous et al. BMC Veterinary Research (2016) 12:79 (cid:9)
`
`Page 5 of 44
`
`25, 26 to 74 % and no less than 75 % were considered as
`"low", "moderate" and "high" heterogeneity, respectively.
`Associations were considered to be statistically signifi-
`cant at P < 0.05. P values between 0.05 and 0.1 were con-
`sidered as statistical trends of potential interest.
`
`Results
`Description of studies
`By 29 December 2015, the search strategy had identified
`a total of 368 unique citations; 347 from the electronic
`searches of PubMed, CAB Abstracts, Google Scholar
`and manual searches from the publications' reference
`lists, 16 from manual searching of major conference pro-
`ceedings and 6 unpublished studies included as part of
`published data. Two hundred ninety two items fulfilled
`stage 1 screening criteria. Of these, 90 final studies (pub-
`lished between 1981 and 2015) also fulfilled stage 2
`selection criteria and were thus selected for review.
`The vast majority of studies were allocated in the sec-
`ond (i.e. non-blinded, non-randomised and uncontrolled
`studies) and third (i.e. retrospective case series and
`reports) group. A few studies included more than one
`sub-study (i.e. a clinical trial and/or ELAS and/or retro-
`spective case series part); accordingly, such studies were
`included in more than one group. Therefore, study
`designs represented were five bRCT [23-27], two
`nbRCT [28, 29] and seven nbRELAS [25, 30-35] in the
`first group, six NRCTs [36-43 11 NRELAS [42-52], 22
`UCTs [44, 48, 53-71], six UELAS [34, 72-76] and one
`cross sectional study [3] in second group, and 19
`retrospective case series [77-95] and 16 case reports
`[96-111] in the third group. In addition, five unpub-
`lished studies described adverse effects and were re-
`ported briefly in EMA report; thus all these were
`considered as one study [112] and were not included in
`any category as there was insufficient information as far
`as their design was concerned.
`Overall, the 90 selected studies reported 12 AEDs. In
`all studies but one [43], the AEDs were orally adminis-
`tered. Within each study, one or more AEDs were evalu-
`ated as a monotherapy and/or adjunct to other AEDs.
`
`Disease characterisation
`In the majority of the studies, the inclusion criteria for
`diagnosing IE were not well characterized. According to
`the described grading system for subject enrolment qual-
`ity, 16 studies [27, 33, 53, 61, 63, 64, 68, 80, 81, 90, 96,
`97, 100, 101, 110, 111] enroled treatment groups of well
`characterized IE, 13 studies [3, 44, 48, 54-57, 62, 66, 67,
`77-79] enroled treatment groups of fairly characterized
`IE, and 14 studies [23-26, 39, 58, 59, 65, 74, 82, 84, 88,
`108, 109] enroled treatment groups of poorly character-
`ized IE. In 26 studies [28, 29, 36-38, 50, 60, 69, 70, 75,
`
`83, 85-87, 89, 91-94, 98, 99, 102-106], the diagnostic
`procedures for enrolment of cases with IE were unclear.
`As far as the ELAS including healthy animals were
`concerned, eight [31, 36, 45, 46, 50-52, 73] enroled
`treatment groups of clear and 14 [25, 30, 32, 34, 35, 37,
`38, 42, 43, 47, 49, 72, 76] enroled treatment groups of
`unclear or unknown diagnostic investigation for ruling
`out other diseases. In one report, a dog was non-
`epileptic and was treated with phenobarbital and chlo-
`mipramine due to anxiety and aggression, but the diag-
`nostic investigation for this was unclear [107].
`
`Study group sizes
`The vast majority of studies reported the total number
`of dogs evaluated. The majority of studies evaluated
`small or very small study size groups. Thirteen studies
`[25, 26, 40, 50, 62, 69-71, 75, 82, 88, 90, 113] evaluated
`groups with a good number of dogs, 13 studies [23, 24,
`32, 37, 39, 65, 74, 77, 79, 80, 91, 94, 95] evaluated groups
`with a moderate number of dogs, 26 studies [3, 28, 34,
`36, 38, 44-46, 48, 53, 54, 56-61, 63, 64, 66, 70, 75, 81,
`83, 84, 114] evaluated groups with a small number of
`dogs and 38 studies [33-35, 39, 42, 43, 51-53, 55, 67,
`68, 72, 73, 76, 78, 85, 87, 89, 92, 93, 96-100, 102-111,
`115, 116] evaluated groups with a very small number of
`dogs. In two studies, the study group size was unclear
`[47, 49].
`
`Signalment and baseline characteristics of study subjects
`Baseline characteristics (such as breed, age and sex) of
`total enroled dogs were reported to some extent for all
`90 studies. Clear presentation of statistical comparison
`of intervention groups with respect to signalment and
`baseline disease characteristics was not commonly
`encountered.
`In all studies reporting baseline data, the recruited
`dogs represented multiple breeds, both sexes and a wide
`range of ages at study entry (median 5, mean 4, range
`0.5-7 years). Major affected breeds were crossed-breeds
`and pure breeds such as Labrador and Golden Retrievers
`followed by German Shepherd dogs, Beagles, Boxers and
`Poodles. In the majority of the studies more males were
`affected compared to females, though these differences
`were not evaluated statistically.
`
`Methodological quality of included studies
`The vast majority of studies revealed high and/or un-
`clear risk of bias for all the components (Fig. 1). As
`stated in the methods, retrospective case series and re-
`ports were not included in the methodological quality
`assessment as these were considered to be at an overall
`high risk of bias.
`
`Page 00005
`
`
`
`Charalambous et al. BMC Veterinary Research (2016) 12:79
`
`Page 6 of 44
`
`Random sequence generation (selection bias(
`
`Allocation concealment (selection bias)
`
`Blinding of participants and personnel (performance bias)
`
`Blinding of outcome assessment (detection bias)
`
`Incomplete outcome data , attinion bias)
`
`Selective repotting r e porting bias) III (cid:9)
`
`Other bias
`
`10.% (cid:9)
`
`25%
`
`MI=
`
`
`
`n LOw riSI, of bias
`
`[[Unclear risk of bias (cid:9)
`
`•High nsk of bias
`
`Fig. 1 Risk of bias. Risk of bias assessment presented as percentages across all included studies
`
`Randomization and allocation concealment
`Studies in group A used randomization to allocate the
`dogs and were considered to provide a low risk of bias.
`Eight studies [25, 28, 29, 32-35, 68] did not offer enough
`detail to confirm that allocation concealment was per-
`formed. Five studies stated that randomization was
`concealed without further details. Two studies [23, 27]
`assigned by random blocking (random allocation to
`blocks of 10 and 6 respectively). One study [27] used
`pre-defined randomization lists under double-blinded
`conditions. One study [31] used drawing lots and two
`studies [24, 30] used a computer-generated list of
`random numbers.
`
`Blinding of outcome assessment
`Only in five studies [23-27], in group A, blinding was
`clearly described; these were also considered to be at
`low risk. In these five studies, blinding was applied to
`all participants, personnel and outcome assessment.
`In one of them [23] all but the primary investigator
`were blinded.
`
`Incomplete outcome data
`Ten studies presented outcome data from all enroled
`dogs in the treatment group to which they were origin-
`ally allocated and there were no losses between enrol-
`ment and evaluation [30, 33, 42, 51, 53, 59, 61, 67, 73,
`78]. The same studies were considered to be at low risk
`of bias. In two studies, it was unclear whether all dogs
`completed the study, as inadequate information was pro-
`vided [60, 64]. Across the remaining studies, there were
`dogs that were euthanized or excluded due to poor seiz-
`ure control, owner request or for unidentified reasons;
`thus there were losses between the initial study popula-
`tions and the final number of the dogs.
`
`Selective reporting
`It was difficult to assess selective reporting as study pro-
`tocols were not sought beyond the information pub-
`lished. In two reports [29, 112] further information was
`
`attempted to be retrieved but complete protocols were
`never obtained.
`
`Acknowledgment of other sources of bias
`Twelve studies reported financial support [24, 26-28, 41,
`44, 53, 58, 61, 68, 73, 79] but there was not adequate
`evidence to support whether this biased the results. One
`study [54] clearly mentioned that there was no financial
`support, while the remaining studies failed to report fi-
`nancial support.
`In two studies [28, 78] the statistical analysis was not
`clarified. In one study [25], many dogs were excluded
`from both groups mainly for treatment-related reasons
`(post-randomisation bias). Six studies [29, 49, 60, 64, 70,
`86] were conference abstracts, thus no further informa-
`tion could be retrieved. One dog in one study [66] and
`two dogs in two studies [28, 63] were diagnosed with
`symptomatic epilepsy (i.e. a cause was identified); this
`could potentially affect the final results on AED safety
`profile. Conflict of interest was clearly stated in one
`study [25].
`In the ELAS, specifically, nine studies [25, 30, 33, 34,
`42, 44, 51, 52, 68] reported details for the experimental
`dogs' housing. Random housing of the dogs occurred in
`all but nine studies [34, 35, 37, 42-44, 49, 72, 73]. The
`baseline characteristics of the dogs were reported in nine
`studies [25, 30, 33, 42, 44, 51, 52, 68, 73] and were
`similar for all the experimental groups in seven studies
`[25, 30, 33, 42, 44, 51, 52].
`
`AEDs safety profile
`A) Safety profile for each AED individually
`Proportions of adverse effects for each AED are summa-
`rized in the text and presented in Figs. 2, 3, 4, 5, 6, 7, 8,
`9, 10, 11. Details of doses and serum concentrations of
`AED(s), period of treatment as well as prevalence of ad-
`verse effects and 95 % CI of the proportion of affected
`cases (included type and most frequently occurred) for
`each study are summarized in the text and provided in
`detail in Tables 1, 2, 3, 4, 5, 6, 7.
`
`Page 00006
`
`(cid:9)
`(cid:9)
`
`
`Charalambous et al. BMC Veterinary Research (2016) 12:79
`
`Page 7 of 44
`
`Phenobarbital - type 1
`
`111111111
`
`High ALT
`Sedation Niior (cid:9)
`70174004/6 MIME
`Polyorta NINE.
`Ataxia NNW
`
`Polyphag.
`/1,41wroot sole syndrome 11111111111.111111111
`
`livocoacton4 111MINEM
`
`1600 7-GT
`
`LCw albunon
`
`holosterol
`
`DWI-hoed
`
`Von0Ong
`
`pruritus
`
`Milllt GLI/11
`
`Hien AST Min
`
`Chronic Alhicai 044/44‘7030100/0,7tY 1111111
`
`High We xtda
`
`Ogiorsuon
`Dizobedtence MI
`
`AlkeredMA
`
`tllah biloubin
`
`Low 0041 puotem
`
`Fig. 2 Proportion of specific type I adverse effects for phenobarbital. Each adverse effect represents the percentage of studies that reported this
`specific adverse effect for phenobarbital monotherapy
`
`0 (cid:9)
`
`10 (cid:9)
`
`40 (cid:9)
`
`50 (cid:9)
`%stdies
`
`60 (cid:9)
`
`70 (cid:9)
`
`On (cid:9)
`
`v0 (cid:9)
`
`100
`
`Phenobarbital
`There was an overall strong level of evidence provided
`for the phenobarbital safety profile. Forty-three studies
`[3, 23, 25, 27-29, 31, 33, 36-40, 42, 43, 45, 46, 49, 50,
`52, 53, 60, 69, 71, 74-76, 79-81, 83-85, 87, 93-95, 98,
`100, 103, 107, 109, 111] presented data about the safety
`
`profile of phenobarbital as a monotherapy agent, giving
`a combined sample size of 1003 dogs.
`Twenty-seven studies reported type I adverse effects
`(dose dependent/predictable), including neurological
`signs and clinical pathological findings as the most com-
`mon (Table 1). Specifically, these adverse effects most
`
`Phenobarbital - type II
`
`Blood dyscrasias
`
`Pancreatitis
`
`Necrolytic dermatitis
`
`Dyskinesia
`
`Fig. 3 Proportion of specific type II adverse effects for phenobarbital. Each adverse effect represents the percentage of studies that reported this
`specific adverse effect for phenobarbital monotherapy
`
`0 (cid:9)
`
`10 (cid:9)
`
`20 (cid:9)
`
`30 (cid:9)
`
`40 (cid:9)
`
`50 (cid:9)
`
`60
`
`% studies
`
`Page 00007
`
`(cid:9)
`
`
`Charalambous et al. BMC Veterinary Research (2016)