Antiviral treatment for Bell s palsy(idiopathic facial paralysis)(review)

Cochrane Database of Systematic Reviews Antiviral treatment for Bell s palsy(idiopathic facial paralysis) (Review) GagyorI,MadhokVB,DalyF,SomasundaraD,SullivanM,GammieF,SullivanF GagyorI,MadhokVB,DalyF,SomasundaraD,SullivanM,GammieF,SullivanF. Antiviral treatment for Bell s palsy(idiopathic facial paralysis). Cochrane Database of Systematic Reviews 2015, Issue 11. Art. No.: CD001869. DOI: 10.1002/14651858.CD001869.pub8. www.cochranelibrary.com Antiviral treatment for Bell s palsy(idiopathic facial paralysis)(review) Copyright 2015 The Cochrane Collaboration. Published by John Wiley& Sons, Ltd.

T A B L E O F C O N T E N T S HEADER....................................... 1 ABSTRACT...................................... 1 PLAIN LANGUAGE SUMMARY.............................. 2 SUMMARY OF FINDINGS FOR THE MAIN COMPARISON................... 4 BACKGROUND.................................... 6 OBJECTIVES..................................... 6 METHODS...................................... 6 RESULTS....................................... 9 Figure 1...................................... 10 Figure 2...................................... 13 Figure 3...................................... 15 Figure 4...................................... 16 Figure 5...................................... 16 Figure 6...................................... 17 DISCUSSION..................................... 18 AUTHORS CONCLUSIONS............................... 19 ACKNOWLEDGEMENTS................................ 20 REFERENCES..................................... 20 CHARACTERISTICS OF STUDIES............................. 23 DATA AND ANALYSES.................................. 37 Analysis 1.1. Comparison 1 Antivirals plus corticosteroids versus corticosteroids plus placebo or no treatment, Outcome 1 Incomplete recovery at end of study........................... 38 Analysis 1.2. Comparison 1 Antivirals plus corticosteroids versus corticosteroids plus placebo or no treatment, Outcome 2 Motor synkinesis or crocodile tears........................... 39 Analysis 1.3. Comparison 1 Antivirals plus corticosteroids versus corticosteroids plus placebo or no treatment, Outcome 3 Adverse events.................................. 40 Analysis 2.1. Comparison 2 Antivirals versus corticosteroids, Outcome 1 Incomplete recovery at end of study.... 41 Analysis 2.2. Comparison 2 Antivirals versus corticosteroids, Outcome 2 Motor synkinesis and crocodile tears... 42 Analysis 2.3. Comparison 2 Antivirals versus corticosteroids, Outcome 3 Adverse events........... 42 Analysis 3.1. Comparison 3 Antivirals plus corticosteroids versus placebo, Outcome 1 Incomplete recovery at end of study..................................... 43 Analysis 3.2. Comparison 3 Antivirals plus corticosteroids versus placebo, Outcome 2 Adverse events...... 44 Analysis 4.1. Comparison 4 Antivirals versus placebo, Outcome 1 Incomplete recovery at end of study...... 44 Analysis 4.2. Comparison 4 Antivirals versus placebo, Outcome 2 Adverse events............. 45 Analysis 5.1. Comparison 5 Antivirals plus corticosteroids versus corticosteroids plus placebo or no treatment in severe cases, Outcome 1 Incomplete recovery at end of study..................... 45 ADDITIONAL TABLES.................................. 46 APPENDICES..................................... 48 WHAT S NEW..................................... 50 HISTORY....................................... 50 CONTRIBUTIONS OF AUTHORS............................. 52 DECLARATIONS OF INTEREST.............................. 52 SOURCES OF SUPPORT................................. 52 DIFFERENCES BETWEEN PROTOCOL AND REVIEW..................... 52 INDEX TERMS.................................... 53 i

[Intervention Review] Antiviral treatment for Bell s palsy (idiopathic facial paralysis) Ildiko Gagyor 1, Vishnu B Madhok 2, Fergus Daly 3, Dhruvashree Somasundara 2, Michael Sullivan 4, Fiona Gammie 2, Frank Sullivan 5 1 Department of General Practice/Family Medicine, University of Göttingen, Göttingen, Germany. 2 Centre for Primary Care and Population Research, Division of Clinical and Population Sciences and Education, University of Dundee, Dundee, UK. 3 Frontier Science (Scotland) Ltd, Kingussie, UK. 4 School of Clinical Sciences, University of Edinburgh, Edinburgh, UK. 5 Department of Family and Community Medicine, North York General Hospital, University of Toronto, Toronto, Canada Contact address: Ildiko Gagyor, Department of General Practice/Family Medicine, University of Göttingen, Humboldtalle 38, Göttingen, Lower Saxony, 37073, Germany. Ildiko.Gagyor@medizin.uni-goettingen.de. Editorial group: Cochrane Neuromuscular Group. Publication status and date: Edited (conclusions changed), published in Issue 11, 2015. Review content assessed as up-to-date: 7 October 2014. Citation: Gagyor I, Madhok VB, Daly F, Somasundara D, Sullivan M, Gammie F, Sullivan F. Antiviral treatment for Bell s palsy (idiopathic facial paralysis). Cochrane Database of Systematic Reviews 2015, Issue 11. Art. No.: CD001869. DOI: 10.1002/14651858.CD001869.pub8. Background A B S T R A C T Corticosteroids are widely used in the treatment of idiopathic facial paralysis (Bell s palsy), but the effectiveness of additional treatment with an antiviral agent is uncertain. Significant morbidity can be associated with severe cases of Bell s palsy. This review was first published in 2001 and revised several times, most recently in 2009. This version replaces an update of the review in Issue 7 of the Cochrane Library subsequently withdrawn because of an ongoing investigation into the reliability of data from an included study. Objectives To assess the effects of antiviral treatments alone or in combination with any other therapy for Bell s palsy. Search methods On 7 October 2014 we searched the Cochrane Neuromuscular Disease Group Specialized Register, CENTRAL, MEDLINE, EMBASE, LILACS, DARE, NHS EED, and HTA. We also reviewed the bibliographies of the identified trials and contacted trial authors and known experts in the field and relevant drug companies to identify additional published or unpublished data. We searched clinical trials registries for ongoing studies. Selection criteria We considered randomised controlled trials or quasi-randomised controlled trials of antivirals with and without corticosteroids versus control therapies for the treatment of Bell s palsy. We excluded trials that had a high risk of bias in several domains. Data collection and analysis Pairs of authors independently assessed trials for relevance, eligibility, and risk of bias, using standard Cochrane procedures. 1

Main results Ten trials, including 2280 participants, met the inclusion criteria and are included in the final analysis. Some of the trials were small, and a number were at high or unclear risk of bias. Other trials did not meet current best standards in allocation concealment and blinding. Incomplete recovery We found a significant benefit from adding antivirals to corticosteroids in comparison with corticosteroids alone for people with Bell s palsy (risk ratio (RR) 0.61, 95% confidence interval (CI) 0.39 to 0.97, n = 1315). For people with severe Bell s palsy (House-Brackmann scores of 5 and 6 or the equivalent in other scales), we found a reduction in the rate of incomplete recovery at month six when antivirals plus corticosteroids were used, compared to corticosteroids alone (RR 0.64, 95% CI 0.41 to 0.99, n = 478). The outcome for the participants receiving corticosteroids alone was significantly better than for those receiving antivirals alone (RR 2.82, 95% CI 1.09 to 7.32, n = 768). The treatment effect of placebo was significantly lower than that of antivirals plus corticosteroids (RR 0.56, 95% CI 0.41 to 0.76, n = 658). Antivirals alone produced no benefit compared with placebo (RR 1.10, 95% CI 0.87 to 1.40, n = 658). Motor synkinesis or crocodile tears In two trials comparing antivirals and corticosteroids with corticosteroids and placebo that assessed this outcome, we found a significant difference in long-term sequelae in favour of antivirals plus corticosteroids (RR 0.56, 95% CI 0.36 to 0.87, n = 469). Two trials comparing antivirals alone with corticosteroids alone investigating this outcome showed fewer sequelae with corticosteroids (RR 1.52, 95% CI 1.08 to 2.12, n = 472). We found no data on long-term sequelae for other comparisons. Adverse events Adverse event data were available in three studies giving comparison data on 1528 participants. None of the four comparisons (antivirals plus corticosteroids versus corticosteroids plus placebo or no treatment; antivirals versus corticosteroids; antivirals plus corticosteroids versus placebo; antivirals versus placebo) showed significant differences in adverse events between treatment and control arms. We could find no correlation with specific treatment within these results. Authors conclusions Low-quality evidence from randomised controlled trials showed a benefit from the combination of antivirals with corticosteroids compared to corticosteroids alone for the treatment of Bell s palsy of various degrees of severity. Low-quality evidence showed a benefit of combination therapy compared with corticosteroids alone in severe Bell s palsy. Corticosteroids alone were more effective than antivirals alone and antivirals plus corticosteroids were more effective than placebo or no treatment. There was no benefit from antivirals alone over placebo. Moderate-quality evidence indicated that the combination of antivirals and corticosteroids reduced sequelae of Bell s palsy compared with corticosteroids alone. We found no significant increase in adverse events from the use of antivirals compared with either placebo or corticosteroids, based on low-quality evidence. P L A I N L A N G U A G E S U M M A R Y Antiviral treatment for Bell s palsy Review question We reviewed the evidence about the effect of antiviral therapy alone or in combination with any other therapy, on Bell s palsy. Background Bell s palsy is a disease of the facial nerve that causes one side of the face to be paralysed. Some studies have suggested that it is caused by the same viral infections that cause cold sores (herpes simplex) or shingles (varicella zoster). If this is correct, antiviral drugs would be likely to help recovery. The paralysis is usually temporary even when left untreated, although without treatment about one person in five is left with permanent facial disfigurement or pain. A Cochrane review has already confirmed the effectiveness of corticosteroids in Bell s palsy. This review was first published in 2001 and has been revised several times, most recently in 2009. This version replaces 2

an update in Issue 7 of the Cochrane Library withdrawn because of an ongoing investigation into the reliability of data in one of the included studies. Study characteristics We identified 10 trials, which included 2280 participants with mild, moderate, or severe one-sided Bell s palsy of unknown cause. Participants were aged from 14 to 84 years. The trials compared antivirals in combination with corticosteroids to corticosteroids; antivirals alone to placebo or no treatment; antivirals alone to corticosteroid treatment alone; and antiviral treatment in combination with corticosteroids to placebo or no treatment. The duration of the included studies ranged from three months to 12 months. Key results and quality of the evidence Incomplete recovery According to low-quality evidence, combining antivirals with corticosteroids improved rates of incomplete recovery from Bell s palsy compared to corticosteroids alone. This finding was based on data from eight trials involving 1315 people with Bell s palsy of various degrees of severity. We were able to use data from four of the studies (478 participants) to examine treatment effects in severe Bell s palsy (that is people who have complete or almost-complete facial paralysis). The results showed that in this group, the addition of antiviral treatment to corticosteroids probably improved rates of incomplete recovery over corticosteroid treatment alone, although we assessed this evidence as low quality. In other analyses, we found the following: corticosteroids alone were more effective than antivirals alone in terms of incomplete recovery (768 participants); antivirals plus corticosteroids were more effective than placebo or no treatment (658 participants); and antivirals alone were less effective than placebo (although this difference was not significant) (658 participants). Long-term after-effects of Bell s palsy Three trials (941 participants) assessed the long-term after-effects of Bell s palsy, such as excessive tear production and synkinesis (involuntary movement of muscles occurring at the same time as deliberate movement). The combination of antivirals and corticosteroids reduced long-term after-effects of Bell s palsy compared with corticosteroids alone. The evidence was of moderate quality. There were fewer long-term after-effects of Bell s palsy after treatment with corticosteroids alone than with antivirals alone (two trials, 472 participants). Adverse events Based on the evidence from three trials (1528 participants), none of the treatments showed significant differences in adverse events, taking into account that the evidence for this outcome was low quality. The evidence in this review is current to October 2014. 3

S U M M A R Y O F F I N D I N G S F O R T H E M A I N C O M P A R I S O N [Explanation] Antivirals plus corticosteroids compared to corticosteroids plus placebo or no treatment for Bell s palsy(idiopathic facial paralysis) Patient or population: People with one-sided mild, moderate, or severe Bell s palsy of unknown cause Settings: Primary and secondary care Intervention: Antivirals plus corticosteroids Comparison: Corticosteroids plus placebo or no treatment Outcomes Illustrative comparative risks*(95% CI) Relative effect (95% CI) Incomplete recovery at endofstudy House-Brackmann scale, Sunnybrook score, Yanagihara score Follow-up: 3 to 12 months Assumed risk Corticosteroids plus placebo or no treatment Corresponding risk Antivirals plus corticosteroids Study population RR 0.61 (0.39 to 0.97) 168per1000 102per1000 (66to163) Moderate 149per1000 91per1000 (58to145) No of participants (studies) 1315 (8 studies) Quality of the evidence (GRADE) low 1 Comments Motor synkinesis or crocodile tears Clinical assessment Follow-up:4to9months Study population RR 0.56 (0.36 to 0.87) 194per1000 109per1000 (70to169) 469 (3 studies) moderate 2 Moderate 228per1000 128per1000 (82to198) 4

Adverse events Follow-up: 6 to 12 months 108per1000 127per1000 (89to182) RR 1.18 (0.83 to 1.69) 877 (3 studies) *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention(and its 95% CI). CI: confidence interval; RR: risk ratio low 3,4 GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Lowquality:Furtherresearchisverylikelytohaveanimportantimpactonourconfidenceintheestimateofeffectandislikelytochangetheestimate. Very low quality: We are very uncertain about the estimate. 1 The benefit to people with Bell s palsy was greater from the combination of antivirals with corticosteroids than with corticosteroids alone or with placebo. We downgraded the quality of evidence for the outcome twice because the heterogeneity of the trials was high (inconsistency), the RR had wide CIs that included the possibility of very little effect(imprecision), and for risk of bias in the included studies. Heterogeneity may be due to clinical variation of participant characteristics, disease severity at baseline, delay in receiving treatment, different type of antiviral agent used, the choice of outcome assessment measures and recovery cut-off points, or the trial duration. 2 Weassessedthequalityofevidenceformotorsynkinesisorcrocodiletearsasmoderatebecauseonlyfourof10trialsprovideddata (selective reporting). 3 Oneofthetrialsdidnotreporthowadverseeventsweredocumented(Hato2007).Inthistrialthenumbersareverysmall. 4 Wedowngradedthequalityofevidenceforadverseeventstolow.Wedowngradedonceforpublicationbias,asonlythreeoftheeight studies in this comparison reported adverse events, and a second time for imprecision as the CIs were wide and encompassed the possibility of an effect in either direction. 5

B A C K G R O U N D Description of the condition Bell s palsy is an acute, generally unilateral paralysis or weakness of facial musculature consistent with peripheral facial nerve dysfunction, of no detectable cause (Niparko 1993). Additional symptoms frequently include pain around or behind the ear on the affected side, sometimes extending into the occipital or cervical regions. Impaired tolerance to ordinary levels of noise and disturbed sense of taste on the affected side may also be present (Burgess 1984). Epidemiological studies have reported an annual incidence of 23 to 25 per 100,000 per year, but a recent study using a general-practice database suggests it may be even higher, at 37 per 100,000 per year (Victor 1994; Martyn 1997; Morales 2013). The condition affects men and women more or less equally, and was once thought to be most common in the 30- to 45-year age group (Peitersen 1982; Katusic 1986; Yanagihara 1988; Bateman 1992; Brandenberg 1993; Peitersen 2002). However, a recent primary-care database study suggests a second peak in the over-70s (Morales 2013). Bell s palsy presents disproportionately amongst pregnant women and people who have diabetes, influenza, a cold, or some other upper respiratory ailment. On average, a British general practitioner will see one person who has developed the condition every 12 to 18 months. Both sides of the face are affected equally often (Prescott 1988). The aetiology of Bell s palsy is still debated. A viral infection, vascular ischaemia, autoimmune inflammatory disorders, and heredity have been proposed as underlying causes (Adour 1982; Burgess 1984; Lorber 1996; Lackner 2010). A viral aetiology has gained popularity since the isolation of herpes simplex virus type 1 genome from the saliva and facial nerve endoneurial fluid of people with this condition (Murakami 1996; Lackner 2010). On the whole, the prognosis is favourable, though a significant proportion of people who are left untreated have residual symptoms. One of the largest series of people with Bell s palsy, which included people receiving no treatment, showed that 85% of people began to recover within three weeks after onset (Peitersen 1982). Partial recovery usually occurred within three to six months for the remaining 15%. The same series showed that normal facial expression reappeared in 71% of cases, 13% had insignificant sequelae (longterm after-effects), and the final 16% had permanently diminished function with aberrant innervation (expressed as motor synkinesis or autonomic dysfunction), and postparalytic spasms. Description of the intervention The treatment of Bell s palsy was highly controversial until 2008, at which time corticosteroids alone were shown to be effective in treating the condition (Salinas 2010). Previous Cochrane reviews on the treatment of Bell s palsy examined the effectiveness of oral prednisolone and aciclovir or valaciclovir (Allen 2004; Lockhart 2009). Several studies excluded from the analysis in these reviews either failed to randomise participants or, when correctly randomised, results were erroneously interpreted in a favourable light (May 1976; Wolf 1978). High-dose corticosteroid therapy has several potential side effects, including peptic ulceration, hypertension, and confusional states. Antiviral therapy was considered expensive, and treatment was reserved for circumstances in which a clear benefit appeared likely.previous recommendations suggested that antivirals needed to be started within 48 hours, although a study of viral replication in participants with Bell s palsy suggests that the window might be extended (Abiko 2002). Why it is important to do this review The 2009 update of this Cochrane review included large-scale randomised controlled trials (RCTs) of antivirals (A) and corticosteroids (S) that had been published in the years before the review. The authors stated that, compared to placebo, A did not contribute to a significant improvement in the rate or extent of recovery of trial participants (Lockhart 2009), but that when added to S, the rate of recovery was slightly better than S alone. The current review updates these findings based on three additional studies. An update of the review published earlier this year (most recently as Gagyor 2015b) also included Abdelghany 2013, a trial currently undergoing investigation for potential data unreliability. We requested withdrawal of that version of the review and here republish our analyses having removed the data from that trial pending the outcome of the investigation. O B J E C T I V E S To assess the effects of antiviral treatments alone or in combination with any other therapy for Bell s palsy. M E T H O D S Criteria for considering studies for this review Types of studies We searched for all randomised controlled trials (RCTs) or quasi- RCTs (which are trials that employ alternate or other systematic allocation) involving aciclovir, valaciclovir, or famciclovir alone or in combination with any other therapy in the treatment of Bell s palsy, published in any language. The duration of studies included in this review ranges from 3 to 12 months; the minimum study 6

duration was 3 months. As in previous versions of the review, we used study quality as an exclusion criterion, excluding trials with a high risk of bias in several domains. Types of participants We considered all trials where participants had a diagnosis of unilateral facial paralysis of unknown cause and satisfied the trial authors requirements for eligibility and inclusion. Types of interventions We considered all trials of treatment with any oral antiviral licensed for the treatment of herpes simplex infections in immunocompetent participants. The list comprised aciclovir; valaciclovir, which is a prodrug of aciclovir; and famciclovir, which is a prodrug of penciclovir. We considered trials where participants received antiviral therapy versus placebo or any other treatment. Types of outcome measures We have described changes to outcome measures in this and previous versions of the review in Differences between protocol and review. Primary outcomes Incomplete recovery of facial function at the end of study measured using a validated rating scale. Secondary outcomes 1. Motor synkinesis or crocodile tears at the end of the study. 2. Adverse events. 3. Incomplete recovery at month six in severe cases. Some trials have reported other symptoms (pain, discomfort, and embarrassment) as outcomes, but we did not consider them in this review. Search methods for identification of studies Electronic searches On 7 October 2014, we searched the Cochrane Neuromuscular Disease Group Specialized Register, the Cochrane Central Register of Controlled Trials (CENTRAL; 2014, Issue 9), MED- LINE (January 1966 to September 2014), EMBASE (January 1980 to September 2014), and LILACS (January 1982 to September 2014). On the same date we also searched the National Institute for Health Research Database of Abstracts and Reviews of Effects (DARE), Health Technology Assessment (HTA) database, and the National Health Service Economic Evaluation Database (NHS EED). We searched for registered trials in ClinicalTrials.gov (www.clinicaltrials.gov) and in the World Health Organization International Clinical Trials Registry Platform (www.who.int/ictrp/ en/). We have provided detailed search strategies in the appendices: Cochrane Neuromuscular Disease Group Specialized Register ( Appendix 1), CENTRAL (Appendix 2), MEDLINE (Appendix 3), EMBASE (Appendix 4), and LILACS (Appendix 5). Searching other resources We also reviewed the bibliographies of the identified trials and contacted trial authors and known experts in the field and relevant drug companies to identify additional published or unpublished data. Data collection and analysis Since a significant benefit of S for the early management of Bell s palsy has been previously demonstrated (Lockhart 2009; Salinas 2010), the main focus of this review was to determine the effect of A in combination with S. Throughout we utilised the following notation: AO: antiviral treatment alone or in combination with placebo AS: antiviral treatment in combination with corticosteroids OO: placebo or no treatment only OS: corticosteroid treatment alone or in combination with placebo Using these notations we conducted four comparisons altogether: AS versus OS AO versus OO AO versus OS AS versus OO For an overall comparison of treatment with and without A we used the following notations: A: antiviral treatment in any combination (i.e. AO or AS); not-a: treatment not including antiviral treatment (i.e. OS or OO). Selection of studies Six review authors working in pairs (FG and VM, DS and MS, IG and FS) scrutinised titles and abstracts to determine eligibility for review of the full-text article. At least two review authors independently assessed each full-text paper for relevance, eligibility, and quality. We had no disagreements about inclusion. Two review authors independently analysed each full-text report and selected studies for inclusion. 7

Data extraction and management Two review authors extracted data onto a data extraction form and double-checked data extraction in pairs. Two review authors (IG and VM) agreed data input into the Cochrane authoring and statistical software, Review Manager 5 (RevMan 2014). Any disagreements were discussed with FD to reach a resolution. We arranged translation of any papers where necessary. Assessment of risk of bias in included studies Two review authors (IG and VM) independently assessed the risk of bias in included studies using the Risk of bias methodology described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). FD reviewed the Risk of bias assessments. The review authors considered methods of randomisation and allocation concealment, blinding (of treatment administrator, participants, and outcome assessors), selective outcome reporting (for example failure to report adverse events), and incomplete outcome data (that is dropouts). We assessed each trial as at high, low, or unclear risk of bias for each of these criteria. Measures of treatment effect All our outcomes were dichotomous. We analysed the data as risk ratios (RRs) with corresponding 95% confidence interval (CI). When comparing studies that used differing symptom scores to assess outcome, we used the House-Brackmann scale when available, as this was used most widely and could be compared with other scales. When assessing adverse events, we used the number of participants affected as opposed to the number of events, to facilitate data comparison. Summary of findings table We created a Summary of findings table for the comparison Antivirals plus corticosteroids versus corticosteroids plus placebo or no treatment using the following outcomes. 1. Incomplete recovery at the end of the study (House- Brackmann scale) 2. Motor synkinesis or crocodile tears at the end of the study (House-Brackmann scale) 3. Adverse events We used the five GRADE considerations (risk of bias, inconsistency of effect, imprecision, indirectness, and publication bias) to assess the quality of a body of evidence (studies that contribute data for the prespecified outcomes). We employed methods and recommendations described in Section 8.5 and Chapter 12 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011) using GRADEpro software (GRADEpro 2008). We justified all decisions to down- or upgrade the quality of studies using footnotes. Unit of analysis issues Each of the included studies carried out randomisation at the participant level. Nine trials used a simple parallel-group design (Adour 1996; Li 1997; De Diego 1998; Hato 2007; Kawaguchi 2007; Vázquez 2008; Yeo 2008; Lee 2013) and two trials used a factorial design (Sullivan 2007; Engström 2008). For studies with a factorial design, we aggregated groups according to whether or not antivirals were administered. We described details in the Results. Dealing with missing data We contacted authors of two studies (Kawaguchi and Engström) for additional data that were required for analysis but which were not provided in their papers, and one (Engström) responded with data. We contacted a previous review author (Lockhart) for information on other studies and we received a response. We used the Kawaguchi 2007 data provided in previous versions of this review. Assessment of heterogeneity We used the I 2 statistic to assess heterogeneity among the included studies in each analysis. If we found substantial unexplained heterogeneity, we reported it and explored possible causes. Assessment of reporting biases As a result of the small number of included trials, we were not able to produce meaningful funnel plots to assess the likelihood of publication bias (Egger 1997). Data synthesis We calculated a treatment effect using the Mantel-Haenszel method (Egger 2007). We used the random-effects model where we found marked heterogeneity (I 2 greater than 40%) between studies. If we had found little or no heterogeneity, we would have used a fixed-effect analysis. The main outcome in all trials was complete recovery. For this review, the review authors calculated the number of participants with incomplete recovery by subtracting the number of participants with complete recovery from the number of participants in the reference group. Subgroup analysis and investigation of heterogeneity We calculated a subgroup analysis for incomplete recovery of people with severe Bell s palsy at baseline using data from four trials (Hato 2007; Sullivan 2007; Engström 2008; Lee 2013). Severe Bell s palsy was defined as severe to complete facial paralysis, graded as equal to or greater than V in the House-Brackmann grading system (Table 1), equal to or less than 20 in the Sunnybrook score (Table 2), and equal to or less than 20 in the Yanagihara score (Table 3). 8

Sensitivity analysis We used sensitivity analyses to assess the effects of combining trials with and without additional treatments in an analysis of A versus S and to assess the impact of length of follow-up on the results of the meta-analysis. We performed sensitivity analyses to assess the potential difference in participant response to aciclovir versus valaciclovir and further sensitivity analyses to investigate whether the exclusion of trials not meeting current best standards (that is those that had a high or unclear risk of bias in more than five categories or trials with less than 200 participants) would influence the results. An analysis investigated whether our conclusions were altered when we excluded the study with a follow-up of less than six months (Adour 1996). Furthermore, in the analysis of AS versus OS in severe cases, we performed a sensitivity analysis to assess differences between trials in which only a subgroup of participants had severe Bell s palsy versus the trials where almost all participants had severe Bell s palsy. R E S U L T S Description of studies Results of the search For this review, we found 375 references in MEDLINE, 102 in EMBASE, 4 in LILACS, 28 in the Cochrane Neuromuscular Disease Group Specialized Register, and 39 in CENTRAL. We also found 12 additional trials in the Database of Abstracts of Reviews of Effects (DARE), one in the Health Technology Assessments (HTA), one in the Economic Evaluation Database (EED) and two by hand searching. We found two ongoing studies in the World Health Organization International Clinical Trials Registry Platform (IRCT201109187575N1; IRCT2012062210087N1). See Figure 1 for a flow chart of the study selection process. 9

Figure 1. Study selection flow diagram. 10

Included studies We added three RCTs with 293 participants in total to the previous version of this review, which had seven trials and 1987 participants. The newly added trials were Lee 2013, Li 1997, and Vázquez 2008. Two of these were published after the search in the 2009 review (Lee 2013; Vázquez 2008), and we identified Li 1997 via a Google search that was additional to our prespecified searches. The 10 studies in the current review included 2280 participants (see Characteristics of included studies). Engström 2008 recruited 829 participants who were treated within 72 hours of onset and randomised by a computerised mechanism in a two-stage process into four treatment groups: valaciclovir with prednisolone or valaciclovir with placebo or placebo with prednisolone or double placebo in a factorial design. The trial was double blind (administrator and participant) for assessment of recovery status until the end of follow-up. Participants were assessed at onset, after 2 weeks (11 to 17 days), and after 1, 2, 3, 6, and 12 months. Disease status was measured using the House-Brackmann grading system and the Sunnybrook scale. Complete recovery status was defined by a Sunnybrook score of 100 and a House-Brackmann grade of 1. Time to recovery was estimated. Data analysis included an assessment of treatment interaction. For this review, we aggregated groups and analysed AS plus AO versus OS plus OO as A versus not-a in order to achieve the most powerful comparison for the effect of treatment with valaciclovir on recovery rates at 12 months. We analysed the recovery rates 12 months after the onset of facial palsy using the Sunnybrook definition and defined complete recovery as a House-Brackmann grade of I. Hato 2007 randomised 296 participants within seven days of onset using sealed envelopes into two treatment groups: valaciclovir with prednisolone or placebo with prednisolone, to provide a comparison AS versus OS. The final analysis included 221 participants. Administrators were not blinded to the treatment allocation, but participants were blinded to treatment received. Those assessing recovery status were not blinded to treatment. Participants disease severity was assessed using the Yanagihara 40-point scoring system; participants were considered completely recovered if attaining a score greater than 36. Participants were assessed at onset and monthly thereafter for six months or until completely recovered if recovery occurred before six months. Sullivan 2007 recruited 551 participants who were treated within 72 hours of onset and randomised by a dedicated remote telephone-computerised mechanism in a two-stage process into four treatment groups: aciclovir with prednisolone or aciclovir with placebo or placebo with prednisolone or double placebo in a factorial design. The trial was blinded for administrator, participant, and assessment of recovery status until the end of follow-up. Participants were assessed at onset, after three months, and if still unwell at three months, after nine months. Recovery status was measured using the House-Brackmann scale, with complete recovery defined by House-Brackmann grade I. Data analysis included an assessment of treatment interaction. Sullivan 2007 reported final outcomes on 496 completed participants at three months and nine months in treatment groups that were aggregated as for Engstrom. Yeo 2008 recruited 91 participants who were randomised to receive either aciclovir and prednisolone or prednisolone alone, to provide a comparison AS versus OS. Participants also received physical therapy and plasma volume expanders as adjuncts. The trial was double blind and investigators followed up participants for six months or until complete recovery. Recovery was assessed using the House-Brackmann scale and defined as a House-Brackmann score of II or less. Li 1997 recruited 51 participants within four days of onset of Bell s palsy. Participants were randomly assigned into two groups, receiving either aciclovir plus prednisolone or prednisolone. Good recovery was defined as a House-Brackmann score of II or I at month six. Li 1997 reported outcomes for 46 participants; five were lost to follow-up. Lee 2013 was a RCT in participants with severe to complete Bell s palsy that used the House-Brackmann score for assessment. The trial included 201 participants with a score of V or more. After randomisation into two groups, participants received either famciclovir plus prednisolone simultaneously or prednisolone alone. Recovery rate was designated as scores of I or II on the House- Brackmann scale at month six. Vázquez 2008 included 42 participants and reported outcomes at 6 and 12 months using the Sunnybrook Facial Grading System. Scores of greater than 90 were defined as a satisfactory recovery. Participants in the intervention group were treated with prednisone and valaciclovir and in the control group with prednisone and placebo. The main study outcome was rate of participants with total recovery at six months follow-up in each group and average time to recovery in each group. Adour 1996 recruited 119 participants, of whom 99 were included in the published analysis. The study was double blind and placebo controlled. Participants were recruited within three or days or less of onset of paralysis and received either aciclovir and prednisolone or placebo and prednisolone. The study duration was four months; participants were reviewed at two weeks, two months, and four months. This was a single-centre study. The Facial Paralysis Recovery Index (FPRI) was used to measure facial function; the primary trial outcome was incomplete recovery defined by a FPRI of 7 or less. De Diego 1998 recruited 113 participants, including 101 in the final analysis. Participants were randomly assigned to treatment. Baseline assessment was carried out within 48 hours of onset of 11

symptoms. Participants received either aciclovir for 10 days or prednisolone for 16 days (reducing dose). Reviews were scheduled for 1, 3, 6, and 12 weeks after initial contact, with further contact if persistent incomplete recovery was noted. The primary study outcome was recovery as defined by a House-Brackmann score of II or less or a Facial Paralysis Recovery Profile of 8 or more. The report did not give the final length of follow-up but stated that it continued until complete recovery or stabilization of the paralysis. Kawaguchi 2007 recruited 150 participants who were treated within seven days of onset and randomised using sealed envelopes into two treatment groups: valaciclovir plus prednisolone or prednisolone alone. There was inadequate blinding of the clinician. Kawaguchi et al. provided unpublished data on incomplete recovery for the previous update. Participants were assessed at 1, 2, 3, 4, 5, and 6 months after inclusion using the Yanagihara scale. We could not contact the authors to obtain data for this review, but we have included the data previously published in this review. Excluded studies Authors of the previous version of this review excluded Antunes 2000 because of incomplete data in 44 participants. Despite our attempts to contact the authors, there was still insufficient information for the data to be usefully included in the analyses. The authors of a previous version of this review reassessed the inclusion of the two studies that were awaiting assessment (de Aquino 2001; Roy 2005). Dr. D. Allen, the author of a previous version of this review, attempted to contact the lead author of the former paper for clarification of the data, but this was not forthcoming and so we have excluded this trial because of a lack of adequate information. The latter study recruited 82 participants, of whom 18 dropped out, and compared aciclovir plus methylprednisolone to methylprednisolone alone, reporting no benefit from the addition of aciclovir. The authors did not provide outcome data in the abstract, which appeared in a journal supplement, and according to the search strategies employed, the trial has not been published as a full paper to date. We excluded this trial due to a lack of adequate information. A further study was classified as awaiting assessment in the previous version of this review (Inanli 2001). This paper was included in a systematic review and a meta-analysis (de Almeida 2009; Goudakos 2009). We excluded the paper from the current review because we could not find it in print or electronic form. See Characteristics of excluded studies. Overall, we excluded five studies for not being RCTs (Ibarrondo 1999; Axelsson 2003; Hato 2003; Hultcrantz 2005; Ahangar 2006); two for having a very short follow-up (Zhou 1999; Chen 2005); three because reports provided insufficient information to assess methods or outcomes (Antunes 2000; de Aquino 2001; Roy 2005); four because of high risk of bias in several domains: Minnerop 2008 (inadequate randomisation, unblinded, 50 of 167 participants lost to follow-up), Ramos Macias 1992 (only abstract available, 15 of 45 participants had Ramsay Hunt syndrome and were given intravenous therapy), Roy 2005 (lack of outcome data, as described above), Shahidullah 2011 (inadequate randomisation, unblinded, 39 of 107 participants lost to follow-up); and one paper was not available (Inanli 2001). Ongoing studies We identified two trial reports in a search of ClinicalTrials.gov just prior to completion of the review. The current status of these trials is unknown; we will assess them fully in a future update of the review. See Ongoing studies. Studies awaiting classification Abdelghany 2013 is awaiting classification pending the results of an investigation into the reliability of the trial data. Risk of bias in included studies We have summarised our Risk of bias assessments in Figure 2. 12

Figure 2. A summary of review authors Risk of bias assessments for included studies. Red = high risk of bias; yellow = unclear risk of bias; green = low risk of bias. 13

Allocation Three studies were at low risk of selection bias, being adequately randomised with allocation concealment (Adour 1996; Sullivan 2007; Engström 2008). Two other studies were at low risk of bias from randomisation but at high risk of bias from inadequate allocation concealment (Kawaguchi 2007; Lee 2013). In the remaining studies, the risk of bias from the method of randomisation was unclear, and the risk of bias for allocation concealment was either unclear, in Li 1997, Hato 2007, Vázquez 2008, and Yeo 2008, or high, in De Diego 1998. Blinding Adour 1996, Li 1997, Sullivan 2007, Engström 2008, and Vázquez 2008 were described as double blind and placebo controlled. Lee 2013 did not use a placebo to ensure blinding. De Diego 1998, Hato 2007, and Kawaguchi 2007 did not describe blinding or placebo use. The Yeo 2008 report stated that the study was double blind, however, the text does not describe this, so we assessed the risk of bias for blinding as unclear. Incomplete outcome data All studies except Yeo 2008 reported frequencies, and often reasons, for failure to complete follow-up. Most trials reported a dropout rate of 10% or less (Li 1997; De Diego 1998; Kawaguchi 2007; Sullivan 2007; Engström 2008; Vázquez 2008; Yeo 2008). Exceptions were Lee 2013 with a dropout rate of 13.1%, Adour 1996 with a dropout rate of 17%, and Hato 2007 with a dropout rate of 25%. Selective reporting All studies except Adour 1996 and Kawaguchi 2007 reported all their intended primary outcomes. Adour 1996 failed to report on audiometry and stapedial reflex testing. Engström 2008 reported all primary outcomes; secondary outcomes were reported in later published papers (Axelsson 2012; Berg 2012). Five studies did not report adverse events (Li 1997; De Diego 1998; Kawaguchi 2007; Yeo 2008; Lee 2013). Other potential sources of bias Diagnostic criteria Nine studies gave adequate information (Li 1997; De Diego 1998; Hato 2007; Kawaguchi 2007; Sullivan 2007; Engström 2008; Vázquez 2008; Yeo 2008; Lee 2013). All studies explicitly mentioned a diagnosis of Bell s palsy and stated that they had considered and excluded other causes of facial palsy. Two trials, Hato 2007 and Kawaguchi 2007, retrospectively excluded participants on the basis of positive serology for herpes simplex or varicella zoster viruses. Lee 2013 excluded participants who did not fulfil inclusion criteria, without specifying the reasons. Two studies mentioned referral to specialists for diagnostic confirmation (Sullivan 2007; Engström 2008). The remaining study, Adour 1996, stated participants were diagnosed with Bell s palsy but did not give any further information. Outcome criteria All studies used referenced facial function scoring systems to grade recovery from facial paralysis. Adour 1996 and De Diego 1998 used the Facial Paralysis Recovery Profile and Adour 1996 also used the Facial Paralysis Recovery Index. Hato 2007 and Kawaguchi 2007 used the Yanagihara scoring system (Yanagihara 2003), which has a validated system for conversion to the House- Brackmann scale (House 1985). Li 1997, Sullivan 2007, Engström 2008, Yeo 2008, and Lee 2013 presented results using the House- Brackmann scale (House 1985). Engström 2008 supplemented this with use of the Sunnybrook scale to minimise the effects of interrater variability (Ross 1996). Vázquez 2008 used a facial grading scale related to Sunnybrook (Ross 1996). Statistical analysis Nine of the 10 studies analysed gave adequate detail; they clearly stated and then used appropriate statistical tests. Only Hato 2007 scored unclear in this category, as it did not state the tests used. Baseline differences between groups Eight of the 10 trials were adequate in this category. De Diego 1998 found a significant difference in rates of hypertension between the two groups, but further analysis revealed that there was no significant difference in trial outcomes as a result. Kawaguchi 2007 reported a significant difference between mean ages of the treatment groups, but further analysis of the age distribution using Chi 2 test revealed no significant difference. Lee 2013 reported a later onset of the treatment in the combination treatment without significance. Effects of interventions See: Summary of findings for the main comparison Antivirals plus corticosteroids compared to corticosteroids plus placebo or no treatment for Bell s palsy (idiopathic facial paralysis) 14

As the included trials reported different intervals and lengths of follow-up, we performed the analyses on data reported at three months (De Diego 1998), six months (Li 1997; Hato 2007; Kawaguchi 2007; Vázquez 2008; Yeo 2008; Lee 2013), nine months (Sullivan 2007), or 12 months (Engström 2008) after the start of treatment. For the subgroup analysis of incomplete recovery in participants with severe Bell s palsy at onset, we either extracted data at month six, in Hato 2007 and Lee 2013, or imputed data to month six, in Sullivan 2007 and Engström 2008. Antivirals plus corticosteroids versus corticosteroids plus placebo or no treatment (AS versus OS) Incomplete recovery This comparison contained eight studies with 1315 participants in total (Adour 1996; Li 1997; Hato 2007; Kawaguchi 2007; Sullivan 2007; Engström 2008; Vázquez 2008; Yeo 2008). We excluded Lee 2013 from this analysis due to its selected participants (severe cases only). The risk ratio (RR) of incomplete recovery at the end of the study showed a statistically significant difference between AS and OS treatments, favouring AS over OS. The RR of incomplete recovery was 0.68 (95% confidence interval (CI) 0.52 to 0.89). Heterogeneity was substantial when we used the fixed-effect model (Chi 2 = 13.19, df = 7 (P value = 0.007), I 2 = 47%); we used the randomeffects model to partially correct for this (RR 0.61, 95% CI 0.39 to 0.97) (Analysis 1.1, Figure 3). Figure 3. Forest plot of comparison: 1 Antivirals plus corticosteroids versus corticosteroids plus placebo or no treatment, outcome: 1.1 Incomplete recovery at end of study. Motor synkinesis or crocodile tears Adour 1996 and Engström 2008 provided data for the outcome motor synkinesis or crocodile tears at the end of the study (Analysis 1.2). The analysis included 469 participants and showed a significant difference between AS and OS (RR 0.56, 95% CI 0.36 to 0.87). Adverse events Adverse events were not significantly less likely in AS versus OS (RR 1.18, 95% CI 0.83 to 1.69). This analysis included data from 877 participants in three studies (Hato 2007; Sullivan 2007; Engström 2008) (Analysis 1.3). Antivirals versus corticosteroids (A versus S) Incomplete recovery This comparison contained three studies (De Diego 1998; Sullivan 2007; Engström 2008). All studies (768 participants) provided data for our primary outcome of incomplete recovery at the end of the study. Incomplete recovery was significantly less common in the participants treated with A versus those treated with S. Initial calculations using the fixed-effect model showed a RR of 1.96 (95% CI 1.48 to 2.59), but with a high degree of heterogeneity (Chi 2 = 8.78, df = 2, P value = 0.01, I 2 = 77%). We repeated the analysis using the random-effects model to partially correct for this, and the RR was 2.82 (95% CI 1.09 to 7.32) (Analysis 2.1, Figure 4). 15

Figure 4. Forest plot of comparison: 2 Antivirals versus corticosteroids, outcome: 2.1 Incomplete recovery at end of study. Motor synkinesis or crocodile tears De Diego 1998 and Engström 2008 provided data for the outcome motor synkinesis or crocodile tears at the end of the study (Analysis 2.2). The analysis included 472 participants and showed significantly fewer sequelae in S than in A (RR 1.52, 95% CI 1.08 to 2.12). Adverse events Fewer participants experienced adverse events in the antiviral group than the corticosteroids group (RR 0.85, 95% CI 0.57 to 1.28, Analysis 2.3), but the CIs included the possibility of the opposite effect. Antivirals plus corticosteroids versus placebo (AS versus OO) Incomplete recovery This comparison contained two studies and outcome data on 658 participants (Sullivan 2007; Engström 2008). Incomplete recovery at the end of the study was significantly much less common with AS versus OO (RR 0.56, 95% CI 0.41 to 0.76) (Analysis 3.1, Figure 5). Figure 5. Forest plot of comparison: 3 Antivirals plus corticosteroids versus placebo, outcome: 3.1 Incomplete recovery at end of study. Motor synkinesis or crocodile tears Not reported Adverse events Adverse events (among 649 participants) were slightly but not significantly more common with AS than with OO (RR 1.14, 95% CI 0.79 to 1.65) (Analysis 3.2). Antivirals versus placebo (AO versus OO) Incomplete recovery For two trials (658 participants) that compared antivirals versus placebo without any complicating additional treatment, the RR of incomplete recovery was non-significant (RR 1.10, 95% CI 0.87 to 1.40) (Sullivan 2007; Engström 2008). The heterogeneity was 16