RESEARCH ARTICLE Open AccessA meta-analysis of stroke risk followingherpes zoster infectionFawziah Marra1* , Jeremy Ruckenstein1 and Kathryn Richardson2AbstractBackground: The incidence of herpes zoster (HZ) is increasing and poses a significant health concern to agingpopulations. Several studies suggest an increased risk of stroke following zoster infection, but the results are conflicting.We conducted a systematic review and meta-analysis to determine if stroke risk is increased following HZ infection.Methods: A search of MEDLINE, EMBASE, Google scholar, Web of Science, CAB Direct, Cumulative Index to Nursing andAllied Health Literature, and Evidence Based Medicine Reviews was conducted for observational studies of adults with HZinfection that examined stroke and TIA risk from January 1, 1966 to May 31, 2016. Adjusted relative risks reported forsimilar follow-up durations were pooled across studies separately using random-effects inverse variance models.Results: Data were pooled from nine studies. Relative risk for stroke after zoster was 1.78 (95% CI 1.70–1.88) for the firstmonth following herpes zoster, dropping progressively to 1.43 (95% CI 1.38–1.47) after 3 months, to 1.20 (95% CI 1.14–1.26) after 1 year. We found that stroke risk increases by a larger margin during the first month after a herpes zosterophthalmicus episode: relative risk 2.05 (95% CI 1.82–2.31). The risk remains elevated one year after the acute episode.Conclusions: Herpes zoster is an established risk factor for increasing the risk of stroke, especially shortly after infection.Vaccination should be encouraged in patients at high risk of cardiovascular disease.Keywords: Herpes zoster, Meta-analysis, Stroke, Transient ischemic attacks, TIA, AgingBackgroundVaricella zoster virus (VZV) is a member of the herpesvirus family with the characteristic capacity to persist inthe body after primary infection and cause latent infec-tion with the risk of reactivation [1]. Primary infectionwith varicella, also known as chickenpox, is a contagiousbut a relatively harmless disease that typically affectssusceptible children, although it can affect naïve adultsas a more severe condition [2, 3]. Viremia during pri-mary infection allows VZV to seed the cranial-nerves ordorsal-root ganglia and lie dormant until reactivation[1]. Reactivation and secondary VZV infection typicallyoccurs in older individuals, perhaps due to compromiseof the immune system through medical conditions ormedications, and results in herpes zoster (HZ) infection,commonly known as shingles [4]. In VZV reactivation,the virus multiplies and migrates along the nerve to thecorresponding dermatome. Postherpetic neuralgia, orpain persisting for 90 days or more after rash starts, isthe most common complication of HZ [5, 6]. Because ofits reactivation within the nervous system, HZ can causemany other neurological complications including ophthal-micus, a life threatening condition which frequentlyrequires hospitalization and antiviral treatment [7, 8].Ramsay Hunt syndrome, meningitis, encephalitis, andtransverse myelitis are also rare neurologic complicationsof zoster [7, 9, 10]. Recently, investigators have publishedseveral reports on patients with zoster being at higher riskfor ischemic or hemorrhagic strokes, however some stud-ies do not find long-term associations and disagree overthe time period at which zoster patients remain at risk ofstroke outcomes [11]. With over 90% of individuals in theUnited States showing serological evidence of exposure toVZV and because one in 3 individuals will experience zos-ter over their lifetime, [4] the potential for HZ reactivationto increase the risk of stroke is a relevant health concernin today’s society. The overall objective of our paper wasto conduct a systematic review and pool the results of the* Correspondence: fawziah@mail.ubc.ca1University of British Columbia, 2405 Wesbrook Mall, Vancouver, BC V6T 1Z3, CanadaFull list of author information is available at the end of the article© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe 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.Marra et al. BMC Infectious Diseases (2017) 17:198 DOI 10.1186/s12879-017-2278-zstudies evaluating stroke risk following an acute episodeof herpes zoster.MethodsThis systematic review and meta-analysis was reportedaccording to the MOOSE guidelines for the reporting ofobservational studies [12].Data sources and search strategyWe conducted a search of MEDLINE EMBASE, Googlescholar, Web of Science, CAB Direct, Cumulative Index toNursing and Allied Health Literature (CINAHL), and Evi-dence Based Medicine Reviews (EBM) for articles reportingon herpes zoster infection and stroke risk from January 1,1966 to May 31, 2016. EBM is a collection of seven differ-ent libraries including the American College of Physiciansjournal club, Cochrane Central, Cochrane Systematic Re-views, Cochrane Methodology, Database of Abstracts ofReviews of Effectiveness, Health Technology Assessments,and National Health Services Economic Evaluation Data-base. Search terms as keywords, Mesh terms and subjectheadings included: varicella OR zoster OR herpe* OR post-herpe* OR herpes ophthalmicus OR VZV OR shingle* ANDstroke OR TIA OR transient ischemic attack OR brainischemia OR cerebrovascular or cerebral ischemia. Afterpooling the articles and deleting duplicates, a manualreview of titles was conducted screening for relevant topicsand keywords. Another final manual review of article ab-stracts was conducted on shortlisted articles. If the articlewas a published abstract in a journal, it was captured in oursearch. The literature search was performed by two authors(JR and FM) and customized for each database with thehelp of a university librarian. Uncertainty and revisionswere resolved by consensus.Inclusion and exclusion criteriaWe included all English studies which evaluated eitherstroke or transient ischemic attacks (TIA) in humans18 years of age or older following an acute episode ofherpes zoster or herpes zoster ophthalmicus (HZO). Weexcluded cases and case series reports, and literature re-views. For published abstracts found, we contacted theauthors to obtain the full manuscript, but if unavailable,we excluded the study.Data extraction study verification and quality assessmentData was extracted independently by two authors (FMand JR) using a standardized abstraction form. Discrep-ancies were resolved through discussion with anotherauthor (KR). Data extracted from the studies includedthe author, date of the study, type of study, inclusionand exclusion criteria, number of patients, HZ defin-ition, follow up period, confounders adjusted for, demo-graphics and study outcome data.Two authors (FM and JR) independently conducted thequality assessment of the studies in an unblinded fashionusing the Newcastle-Ottawa quality assessment scale(Additional file 1) [13]. Points were awarded to observa-tional studies for comparability if they controlled or ad-justed for age, sex, and cardiovascular disease, such ashypercholesterolemia, hypertension, diabetes, coronaryheart disease, atrial fibrillation, peripheral vascular disease,carotid stenosis, valvular heart disease, heart failure,chronic rheumatic heart disease, as these are consideredimportant risk factors for stroke. Self-controlled case-series studies were scored with the cohort studies, how-ever we interpreted ‘cohort representativeness’ as how rep-resentative the selected cases were of patients with strokein the community. Discrepancies were resolved throughdiscussion with another author (KR).Statistical analysisAs risk of stroke varied with time since herpes zoster in-fection we pooled adjusted relative risks (RR) reportedfor similar follow-up durations (e.g. first 1 month,3 months, 1 year and beyond 1 year since herpes zosterinfection, where possible) across studies separately usingrandom-effects inverse variance models. We also per-formed subgroup analyses by patients older and youngerthan approximately 40 years where possible. We chosethis cut-off as most studies reported stroke risk stratifiedby age less than and greater than 40 years. Unfortu-nately, there were not enough studies reporting resultsin the over 65 year age group to combine these findings.We measured heterogeneity across studies using the I2statistic, with higher values reflecting increasing hetero-geneity. We assessed sources of heterogeneity by sub-group analysis and publication bias by examining funnelplots. Stata version 12.1 (StataCorp, College Station, TX)was used for analysis. Statistical tests were two sidedwith p < 0.05 defining statistical significance.ResultsSearch results and study characteristicsFigure 1 identifies the search results and selection process.Of the 4478 articles identified through the database search1038 were removed due to duplicates and 46 articles werereviews of literature with topics such as pathophysiologyand clinical features as well as treatment of herpes zoster,and 3100 articles were considered as irrelevant leaving294 articles which were reviewed in detail by two individ-uals. Articles were deemed irrelevant if they didn’t meetthe inclusion criteria, i.e. examined varicella or chickenpoxinstead of shingles, examined encephalitis instead ofstroke with zoster, examined children primarily, examinedherpes simplex instead of varicella zoster, examined vac-cination, considered irrelevant outcomes (i.e. biomarkerstudies, cognitive impairment, stress). Of the 294 articlesMarra et al. BMC Infectious Diseases (2017) 17:198 Page 2 of 11considered for a detailed review, the majority consideredan irrelevant topic, while some were cases studies, non-English, or duplicates. This left 12 studies for inclusion inthe review, nine studies with a full published manuscript[14–22] and 3 abstracts. Since the necessary results fromthe abstracts were unavailable and investigators declinedto provide us with these when contacted, all 3 abstractswere excluded from analysis, leaving 9 studies for inclu-sion in our meta-analysis.The characteristics of the 9 studies included in themeta-analysis are displayed in Table 1. Two of the earli-est studies used the Taiwanese National Health Insur-ance Research Database (NHIRD) [15, 17]. Both Breur etal. [14] and Langan et al. [16] used the United King-dom’s Clinical Practice Research Datalink (CPRD) andThe Health Improvement Network (THIN) general prac-tice databases respectively. The European studies con-ducted by Sreenivasan et al. [21] and Sundstrom et al.[19] used the Danish and Swedish registries, respectively.The two USA-based studies used Medicare [18] andOlmsted County residents, [20] respectively and the lat-est study used the Korean Health Insurance database[22]. Two studies performed a self-controlled case seriesanalyses, [16, 18] while the other seven are retrospectivecohort in design [14, 15, 17, 19–22]. Follow-up time var-ied from 1 to 24 years, but most studies reported therisk over various time periods post zoster infection.All studies adjusted for age and sex. All studies exceptSundström et al. [19] adjusted for major cardiovascularrisk factors such as hypertension diabetes, congestiveheart failure, dyslipidemia, ischemic heart disease, atrialfibrillation, intermittent arterial claudication, carotidstenosis, and valvular heart disease. Only the study byBreuer et al. [14] was able to adjust for smoking andobesity (also considered as cardiovascular risk factors).However the two studies by Minassian [18] and Langan[16] also provided strong control for confounding by usinga self-controlled case-series design which only makeswithin-person comparisons. Antiviral therapy varied be-tween studies with two studies including cases that wereall taking antivirals, [18, 21] while Langan [16] and Lin[17] reported that 55% and 24% of their patients receivingantiviral therapy respectively. The mean age of the partici-pants in each study ranged from 47 to 80 years. Langan etal. [16], Minassian et al. [18], and Yawn et al. [20] includedthe oldest participants with mean ages of 77 (median), 69,and 80 years, respectively. The proportion of women washigher than men in most of the studies.Four studies examined the combined outcome of strokeor TIA, [15, 17, 21, 22] whereas the other five studies ex-amined the risk of stroke only [14, 16, 18–20]. All studiesexamined stroke risk following zoster except Lin et al. [17]who examined stroke risk after herpes zoster ophthalmi-cus only. An additional four studies also included an ana-lysis of stroke risk following ophthalmicus [14–17, 22].Sreenivasan et al. [21] used acyclovir prescriptions filled ina certain strength and quantity to identify cases of HZ butused ICD10 codes to identify incident stroke and TIA.The rest of the studies used diagnosis codes to identifytheir HZ cases and cerebrovascular events.Risk of biasAssessment of study validity using the Newcastle-OttawaScale revealed a low risk of bias amongst studies (Additionalfile 1: Table S1). Among the nine studies three scored a full9/9, five studies scored 8/9, and one study scored 6/9.Herpes zoster infection and risk of strokeOur meta-analysis indicates an elevated risk of stroke afterzoster (Fig. 2). The risk of stroke was greatest during theFig. 1 Study selection for inclusion into the meta-analysisMarra et al. BMC Infectious Diseases (2017) 17:198 Page 3 of 11Table1CharacteristicsofincludedstudiesBreuerKangKwonLanganLinMinassianSreenivasanSundströmYawnYear201420092016201420102015201320152016DesignRetrospectiveMatchedCohortRetrospectiveMatchedCohortRetrospectiveMatchedCohortSelf-ControlledCaseSeriesRetrospectiveMatchedCohortSelf-ControlledCaseSeriesRetrospectiveCohortRetrospectiveCohortRetrospectiveMatchedCohortDataSource,CountryTheHealthImprovementNetwork(THIN)generalpracticedatabase,UnitedKingdomTaiwanNationalHealthInsuranceResearchDatabase(NHIRD),TaiwanKoreaHealthInsuranceDatabase,KoreaUKClinicalPracticeResearchDatalink(CPRD)generalpracticedatabase,UnitedKingdomTaiwanNationalHealthInsuranceResearchDatabase(NHIRD),TaiwanMedicareadministrativedataclaims,USADanishCivilRegistrationSystem(CRS),DenmarkVästraGötalandCountyPrimaryHealthCareRegisterandtheSwedishPatientRegister,SwedenRochesterEpidemiologyProject,USAHZdiagnosisperiod2002–20101997–20012003–20131987–20122003–20042006–20111995–20082008–20101986–2011NumberofCases(HZandHZO)106,6017,76070,4246,58465842,954117,92613,2964,478NumberofControls213,20223,280695,75501,97404,503,054~1,500,00016,800Follow-UpPeriod24years1year11years1year1year1year14years1year3yearsGender,%female595249575171516062Meanageatzosterdiagnosis,years594741Medianageatstrokeonset77years5780Notreported5968%ReceivingAVTherapyNotreportedNotreportedNotreported55%24%ofcases100%50%NotreportedNotreportedCaseInclusionCriteriaAdults18yearsofageorolderdiagnosedwithincidentHZorHZO(indexdateasrecordedindatabase)Adults18yearsofageorolderpresentingtoambulatoryclinicwithincidentHZorHZO(indexdateasrecordedindatabase)Adults18yearsofageorolderwithincidentHZAdults18yearsofageorolderwithincidentHZorHZOandincidentstrokeAdults18yearsofageorolderpresentingtoambulatoryclinicwithincidentHZO(indexdateasrecordedindatabase)Adults65yearsofageorolderwithevidenceofincidentHZorHZOandanincidentischemic/nonspecificstrokeAdults18yearsofageorolderwhoreceivedacyclovir800mginpacksof35AllindividualsrecordedineitherofthetwodatabaseswithincidentHZAdults50yearsandolderwithincidentHZCaseExclusionCriteriaPatientswhoexperiencedcardiovascular/strokeevent(MI,TIA,stroke)beforeindexPatientswhohadbeendiagnosedwithstrokebeforetheindexdatePatientswhohadbeendiagnosedwithstrokebeforeHZPatientswithevidenceofHZ,postherpeticneuralgia,orstrokebeforethestudyperiod.PatientswithPatientsdiagnosedwithHZOduringtheprevious1-yearperiod.PatientsdiagnosedwithanytypeofstrokepriortoIndividualswithevidenceofvasculareventsorHZbeforeobservationperiod.SecondaryPatientswhoreceivedasecondacyclovirprescriptionofsamestrengthandpacksize.PatientswithastrokeorTIAPatientsdiagnosedwithHZduringtheprevious1-yearRecurrentherpessimplexinfection,historyofstrokegreatthan1monthMarra et al. BMC Infectious Diseases (2017) 17:198 Page 4 of 11Table1Characteristicsofincludedstudies(Continued)dateandthosewithrecurrentHZincidentepisodesofTIAandsubarachnoidhemorrhageorriskfactorsforsubarachnoidhemorrhage.Patientswithencephalitis12monthsafterstroke.Patientswithnonspecificcerebralaneurysms.indexambulatorycarevisit.Patientswithsystemiclupuserythematosus,rheumatoidarthritis,multiplesclerosis,HIV,malignancy,anduseofsteroidsorimmunosuppressantsformorethan1monthwithin1yearpriortoindexdate.inpatientdiagnosisofzosterorcardiovascularevents.Individualswithsubarachnoidhemorrhage(orestablishedriskfactorsof)orencephalitisdiagnosedupto12monthsafterstrokediagnosesbeforethestartofthefollowupperiodaspertheICD10codesbelowaswellasICD8codes430-438beforeindexdate,MIbeforeindexdateHZDefinitionReadcodescorrespondingtodiagnosesofincidentHZ(indexdateasrecordedindatabase)IncidentHZdiagnosisusingICD9codes053x(indexdateasrecordedindatabase)ICD10codescorrespondingtodiagnosesofincidentHZ(indexdateasrecordedindatabase)ReadorICD10codescorrespondingtodiagnosesofincidentHZN/AHZdiagnosisusingICD-9codes053xANDantiviraltherapy7daysbeforeorafterdiagnosisPatientswhofilledasingleprescriptionofacyclovir800mgfor35tabletsIncidentHZdiagnosisusingICD-10codeB02.XIncidentHZdiagnosisusingICD-9codes,andconfirmedbymedicalrecordreviewHZODefinitionReadcodescorrespondingtodiagnosesofincidentHZOIncidentHZOdiagnosisusingICD9code053N/AReadcodescorrespondingtodiagnosesofincidentHZOIncidentHZOdiagnosisbasedontheICD9code053.2HZOdiagnosisusingICD-9codesANDantiviraltherapy7daysbeforeorafterdiagnosisN/AN/AN/AStrokeDefinitionReadcodescorrespondingtostroke,diagnosisStroke/TIAdiagnosisasperICD9codes430-438Stroke/TIAdiagnosisasperICD10codesReadorICD10codescorrespondingtostrokediagnosisStroke/TIAdiagnosisasperICD9codes430-438Ischemic/nonspecificstrokediagnosisasperICD9codes436,433x1,or434x1Diagnosisinhospitalofstroke/TIAasperICD10I60-64andG45StrokediagnosisasperICD-10codesI61-I64(excludingI62)StrokediagnosisusingICD-9codesControlselectionPatientswhohadnorecordofHZ,matched(2:1)byage(+/−2years),sex,andgeneralpracticePatientswithnoHZorstrokebefore2001,matched(3:1)onageandsex,anddefinedtheirindexdateastheirfirstambulatorycarevisitin2001PatientswithoutHZ,matchedonagegroupSelfcontrolledSelectedfromremainingpatients,matched(3:1)onagegroupandgender,anddefinedtheirindexdateastheirfirstambulatorycarevisitin2004SelfcontrolledPatientswhohadnopriorhistoryofacyclovir,valacyclovir,orfamcicloviruse(asaproxyforunexposed)TotalremainingpopulationwithoutHZMatchingeachpatientwithHZwithupto4patientswhosebirthdaywas+/−1year,whowerethesamesex,andnoHZinthepast5yearsConfounders(Adjustedfor)Age,sex,obesity,smoking,highcholesterolAge,sex,hypertension,diabetes,coronaryAge,malegender,hypertension,ConfoundersareimplicitlycontrolledforAge,sex,hypertension,diabetes,ConfoundersareimplicitlyAge,sex,calendarperiod,acuteMI,atrialfibrillation,AgeandsexAge,sex,hypertension,dyslipidemia,Marra et al. BMC Infectious Diseases (2017) 17:198 Page 5 of 11Table1Characteristicsofincludedstudies(Continued)recording,hypertension,diabetes,ischemicheartdisease,atrialfibrillation,intermittentarterialclaudication,carotidstenosis,andvalvularheartdiseaseheartdisease,hyperlipidemia,renaldisease,atrialfibrillation,heartfailure,heartvalve/myocardiumdisease,carotid/peripheralvasculardisease,monthlyincome,urbanizationlevel,andgeographicalregionhyperlipidaemia,ischaemicheartdisease,diabetes,heartfailure,peripheralvasculardisease,arterialfibrillationoratrialflutter,renaldiseaseandvalvularheartdiseaseduetostudydesignhyperlipidemia,coronaryheartdisease,chronicrheumaticheartdisease,otherformsofheartdisease,andmedicationhabitscontrolledforduetostudydesigneducation,cancer,medications(antihypertensives,drugsusedtotreatdyslipidemiaandatrialfibrillation,immunosuppressivedrugs)coronaryarterydisease(includingMI),arrhythmias,congestiveheartfailure,diabetes,depression,chronicobstructivepulmonarydisorder,vasculopathies,stroke,andanxietyRelevantstudyoutcomesStrokeHRafter>1yearfollowup,withanalysesstratifiedbyage<40and≥40years,andbyHZO.StrokeHRafter1yearfollowupsinceHZ,withanalysesstratifiedbyage(<45and≥45years),genderandbyHZOStrokeHRafter11yearsoffollowup,stratifiedbyage18–30,30–40,40–50,50–60,60–70,and>70yearsStrokeIR1–4,5–12,13–26,and27–52weeksafterHZorHZO+/−headandneckinvolvement,withanalysesstratifiedbythosewhoreceivedantiviraltherapyandbyHZOStrokeHRafter1yearfollowupsinceHZO,withanalysesstratifiedbythosewhoreceivedantiviraltherapyStrokeIRat1,2–4,5–12,13–26,and27–52weekssinceHZdiagnosis,withanalysesstratifiedbygender,HZOStrokeIRRafter<2weeks,2–52weeks,and>1yearfollowupsinceHZ,withanalysesstratifiedbyage<40,40–59,and≥60yearsandbygenderStrokeIRRafter1yearfollowup,withanalysesstratifiedbyage<40,40–49,50–59,60–69,70–79,and≥80yearsandgenderStrokeOR3and6monthsand1and3yearsafterzoster.HZherpeszoster,HZOherpeszosterophthalmicus,MImyocardialinfarction,TIAtransientischemicattack,HIVhumanimmunodeficiencyvirus,BMIbodymassindex,IQRinterquartilerange,ICDInternationalClassificationofDiseases,HRhazardratio,IRincidenceratio,ORoddsratioMarra et al. BMC Infectious Diseases (2017) 17:198 Page 6 of 11first month following the HZ episode RR 1.78 (95% CI:1.70-1.88 I2 = 0%), dropping to 1.43 (95% CI: 1.38-1.47, I2= 0%) over the first 3 months following HZ, to 1.20 (95%CI: 1.14–1.26, I2 = 55%) in the first year after the HZ epi-sode, and finally to 1.07 (95% CI: 0.99–1.15, I2 = 91%) over3 or more years. For the studies with one year follow-up,the considerable heterogeneity was due to the smaller ef-fect size reported by Minassian et al. [18]. This could bedue to the self-controlled case-series design, the relativelyolder age of the patients and the specific outcome of is-chemic stroke. For the studies with three or more years offollow-up, heterogeneity was introduced by the larger ef-fect size reported by Kwon et al. [22]. This may reflect dif-ferences in the study population in the South Koreanstudy or study design, such as misclassification of the TIAdiagnosis. We did not find evidence of publication bias,however there were few studies.The risk of stroke following herpes zoster ophthalmi-cus appeared stronger compared to an episode of HZonly (1 month: RR 2.05 95% CI: 1.82-2.31, I2 = 0%) asseen in Fig. 3. The risk also remains high in the first yearafter the HZ episode (RR 2.26, 95% CI: 1.35–3.78), butthere was considerable heterogeneity (I2 = 91%). Thisheterogeneity may be explained by the Langan et al. [16]and Minnassian et al. [18] studies using a self-controlledcase series design, thus potentially having better con-founding control than the other two cohort studies [15,17]. Kang et al. [15] and Lin et al. [17] likely producedsimilar effect estimates as their data was ascertainedfrom the same Taiwanese Health Insurance database, al-beit in different patients over different time periods. Fi-nally, the Breuer et al. [14] study found no excess riskover 24 years of follow-up.Figure 4 details stroke risk after HZ in different agegroups. In younger adults (less than 40 years) there is a sig-nificant increased risk of stroke within a year after zosterepisode (RR 2.96, 95% CI: 1.05-8.41, I2= 86%), but signifi-cant heterogeneity in these results. The larger effect size bySundstrom et al. [19] could be due to not adjusting for anyconfounders beyond age and sex. The study cohort by Sree-nivasan and colleagues [21] differed by being based on atreated population as they used use of antivirals as a proxyfor identifying their HZ cases. For the younger aged cohort,the risk remains significant in the long-term, that is beyond11 years after the acute HZ episode (RR 1.39, 95% CI: 1.27–1.52, I2= 0%). More studies evaluated stroke risk in older co-horts, and like the younger cohorts, we found that olderadults also demonstrate a significant increase in stroke riskone year after zoster (RR 1.19, 95% CI: 1.13–1.24, I2= 44%),with this risk being greatest shortly after the acute zoster in-fection. The risk remains elevated beyond 3 years of followup (RR 1.09, 95% CI: 1.01–1.17, I2= 86%), but with consider-able heterogeneity in the results in line with the overall re-sults where Kwon et al. [22] reported a greater effect size.DiscussionThis is the first study to systematically review and per-form a meta-analysis of the risk of stroke following aNOTE: Weights are from random effects analysis....1 month follow-upLangan et al 2014Minassian et al 2015Subtotal (I-squared = 0.0%, p = 0.390)3 month follow-upLangan et al 2014Minassian et al 2015Yawn et al 2016Subtotal (I-squared = 0.0%, p = 0.714)1 year follow-upKang et al 2009Langan et al 2014Minassian et al 2015Sreenivasan et al 2013Sundstrom et al 2015Yawn et al 2016Subtotal (I-squared = 55.2%, p = 0.048)3+ years follow-upBreuer et al 2014Kwon et al 2016Sreenivasan et al 2013Yawn et al 2016Subtotal (I-squared = 90.8%, p = 0.000)Study1.63 (1.32, 2.02)1.79 (1.70, 1.89)1.78 (1.70, 1.88)1.49 (1.31, 1.70)1.42 (1.37, 1.47)1.53 (1.01, 2.32)1.43 (1.38, 1.47)1.31 (1.07, 1.61)1.21 (1.12, 1.30)1.14 (1.12, 1.17)1.24 (1.16, 1.31)1.34 (1.11, 1.61)1.04 (0.79, 1.36)1.20 (1.14, 1.26)1.02 (0.98, 1.07)1.16 (1.13, 1.20)1.05 (1.02, 1.09)1.02 (0.86, 1.21)1.07 (0.99, 1.15)ES (95% CI)5.6394.37100.006.7192.650.65100.005.5621.8137.0925.466.693.40100.0028.5130.1329.7711.59100.00Weight%1.43 2.32Fig. 2 Effect of herpes zoster on stroke risk by length of study follow-upMarra et al. BMC Infectious Diseases (2017) 17:198 Page 7 of 11zoster infection. We found an elevated risk of strokeafter pooling the results of nine observational studies.The risk was greatest shortly after the acute zoster epi-sode but diminished slowly over time, although the riskwas still significant after the first year. The risk of strokewas more pronounced in patients with herpes zosterophthalmicus infection.Varicella zoster virus is a highly neurotropic DNAvirus that infects more than 95% of the world population[1]. Varicella infection or chickenpox most commonlyoccurs in children after which viral latency is establishedwhereby the virus resides, but remains dormant, in thecranial nerve, dorsal root, and autonomic ganglionicneurons. During latency, varicella zoster virus transcrip-tion is limited and without production of virions. Theincreased stroke risk after reactivation of the varicellazoster virus may be due to the vasculopathy characteris-tically caused by this pathogen [23, 24]. By looking atbrain tissue of individuals who have died from varicellainfection, investigators believe the virus migrates trans-axonally from the trigeminal nerves to cranial vascula-ture and spreads transmurally through the tunicaadventitia, media, and intima, causing inflammation andthickening of the intima, reducing the media, and dam-aging the inner elastic layer of the vessels [24]. The pres-ence of varicella zoster virus in intracerebral arteries isseen shortly after the acute infection and as late as10 months after, allowing for the possibility that the riskof stroke could be present for up to a year after the ini-tial HZ infection [25, 26].The second possible mechanism is related to inflamma-tion associated with systemic infection that can create astate in which the blood is more prone to clotting. Releaseof cytokines such as TNF-alpha and interleukin-2 duringinflammation or stress leads to endothelial dysfunctiondisruption of atheromatous plaques and hypercoagulabil-ity, all leading to acceleration of atherosclerosis [27, 28]. Anumber of studies have suggested microbes play some rolein inflammation, thereby accelerating atherosclerosis [29].For example, Wang et al. conducted a meta-analysis thatshowed Helicobacter pylori infection contributes to risk ofischemic stroke,[30] however a more recent analysis re-futed this [31]. In their meta-analysis, Chen et al. found anincreased risk of cerebrovascular disease with the presenceof IgG for Chlamydia pneumonia [32]. In a self-controlledcase-series of over 50,000 patients, Smeeth et al., demon-strated a relationship between recent respiratory tract in-fection and myocardial infection (IR 4.95, 95%CI: 4.43–5.53) as well as stroke (IR 3.19, 95%CI: 2.81–3.62) [33].The risks for both events were highest in the first 3 days,then gradually decreased to baseline the weeks followingthe acute infection.We found evidence that the risk of stroke was presentin both younger and older individuals but was more pro-nounced in the younger patients. However, this analysiswas limited by only four studies that had one year follow-up data in patients less than 40 years of age, [15, 19, 21]and the results were considerably heterogeneous. One rea-son for this discrepancy could be that Sundstrom et al.[19] did not adjust for any confounders beyond age andNOTE: Weights are from random effects analysis....1 month follow-upLangan et al 2014Minassian et al 2015Subtotal (I-squared = 0.0%, p = 0.771)3 month follow-upLangan et al 2014Minassian et al 2015Subtotal (I-squared = 89.0%, p = 0.003)1 year follow-upKang et al 2009Langan et al 2014Lin at al 2010Minassian et al 2015Subtotal (I-squared = 91.1%, p = 0.000)24 years follow-upBreuer et al 2014Subtotal (I-squared = .%, p = .)Study1.82 (0.81, 4.09)2.06 (1.82, 2.32)2.05 (1.82, 2.31)2.94 (2.00, 4.32)1.60 (1.47, 1.74)2.10 (1.16, 3.80)4.28 (2.02, 9.07)1.72 (1.32, 2.24)4.52 (2.45, 8.33)1.22 (1.16, 1.28)2.26 (1.35, 3.78)1.03 (0.77, 1.38)1.03 (0.77, 1.38)ES (95% CI)2.1897.82100.0044.9955.01100.0018.7028.7221.5930.99100.00100.00100.00Weight%1.11 9.07Fig. 3 Effect of herpes zoster ophthalmicus on stroke risk by length of study follow-upMarra et al. BMC Infectious Diseases (2017) 17:198 Page 8 of 11sex and given that comorbidies would play a role in zosterincidence in the younger population, their results shouldbe interpreted with caution. Further, in the younger popu-lation many risk factors may not be recorded or reportedby patients and therefore despite adjustment, residual con-founding may still be present. We were able to pool sixstudies to evaluate stroke risk in older patients one yearafter the acute HZ episode, giving less heterogeneousresults. Despite differences in study populations, the stud-ies were consistent in the magnitude of the risk beingaround 20% higher within the first year post HZ.We observed a greater than 2-fold increase in strokerisk in the first year following herpes zoster ophthalmicusor zoster with trigeminal nerve involvement which as withgeneral HZ decreased over time. The increased risk wasnot surprising given that HZO arises when latent VZVNOTE: Weights are from random effects analysis..1 year follow-upKang et al 2009Sreenivasan et al 2013Sundstrom et al 2015Subtotal (I-squared = 85.7%, p = 0.001)11+ years follow-upBreuer et al 2014Kwon et al 2016Sreenivasan et al 2013Subtotal (I-squared = 0.0%, p = 0.448)Study1.02 (0.49, 2.12)2.87 (1.93, 4.28)10.30 (3.86, 27.51)2.96 (1.05, 8.41)1.74 (1.13, 2.67)1.39 (1.26, 1.54)1.27 (1.00, 1.61)1.39 (1.27, 1.52)ES (95% CI)33.1737.4929.34100.004.4681.1214.42100.00Weight%1.0364 27.5NOTE: Weights are from random effects analysis....1 month follow-upLangan et al 2014Minassian et al 2015Subtotal (I-squared = 0.0%, p = 0.390)3 month follow-upLangan et al 2014Minassian et al 2015Yawn et al 2016Subtotal (I-squared = 0.0%, p = 0.714)1 year follow-upKang et al 2009Langan et al 2014Minassian et al 2015Sreenivasan et al 2013Sundstrom et al 2015Yawn et al 2016Subtotal (I-squared = 44.3%, p = 0.110)3+ years follow-upKwon et al 2016Sreenivasan et al 2013Yawn et al 2016Subtotal (I-squared = 86.2%, p = 0.001)Study1.63 (1.32, 2.02)1.79 (1.70, 1.89)1.78 (1.70, 1.88)1.49 (1.31, 1.70)1.42 (1.37, 1.47)1.53 (1.01, 2.32)1.43 (1.38, 1.47)1.31 (1.06, 1.62)1.21 (1.12, 1.30)1.14 (1.12, 1.17)1.21 (1.14, 1.29)1.34 (1.11, 1.62)1.04 (0.79, 1.36)1.19 (1.13, 1.24)1.14 (1.11, 1.18)1.05 (1.02, 1.09)1.02 (0.86, 1.21)1.09 (1.01, 1.17)ES (95% CI)5.6394.37100.006.7192.650.65100.004.0620.4742.2525.515.092.63100.0043.5443.4513.00100.00Weight%1.43 2.32abFig. 4 a Effect of herpes zoster on stroke risk by length of study follow-up and age (younger adults, aged 40 years or less). b Effect of herpes zos-ter on stroke risk by length of study follow-up and age (older adults, aged >40 years)Marra et al. BMC Infectious Diseases (2017) 17:198 Page 9 of 11infection of the trigeminal ganglion becomes reactivatedand involves the ophthalmic division of the trigeminalnerve, giving rise to a number of head and neck-relatedcomplications, including stroke or TIA [9, 26].The pathogenesis behind increased stroke risk couldalso lead to increased risk for a myocardial infarction(MI). In our meta-analysis we did not address this riskand only two studies examined this outcome. Using aretrospective cohort of 106,601 HZ cases and 213,202controls matched on age, sex, and general practice from aUK general practice database, Breuer et al. examined therisks of MI [14]. They found an increased risk overall (HR1.10), but more pronounced in those less than 40 years ofage (HR 1.49). Yawn et al. evaluated MI risk in 4,454 indi-viduals with herpes zoster and 16,740 individuals withoutherpes zoster, all with no history of prior MI before theirindex date [20]. They found a greater risk of MI at3 months post HZ (HR 1.68), but this reduced over timeto a non-significant findings at 3 years (HR = 1.17). Weagree with the authors of the two studies, that there isneed for more studies to be conducted using large datasetsto further examine the causal association.There were too few studies for us to be able to exam-ine stroke risk according to gender however the threestudies that provided their results stratified by sex, foundno differences in risk for men or women [15, 19, 20].Clinical trials show that antivirals hasten the resolutionof lesions, reduce the formation of new lesions and viralshedding as long as they are taken within 72 h after theonset of the rash [1]. The greatest benefit for theseagents are in those at greatest risk for complicationssuch those aged 50 years or older, have moderate to se-vere rash and/or pain, involvement of the face or eye, orare immunocompromised due to medications or disease.Unfortunately, there were too few studies for us to beable to examine stroke risk according to receipt of anti-viral therapy, however, this would be an important pieceof information as provision of antiviral therapy by clini-cians may be of use for those adults with a high cardio-vascular risk after zoster infection; further research iswarranted on this topic. Similarly only one study exam-ined the effect of vaccination on stroke risk after zoster,but had insufficient numbers of patients vaccinated inorder to demonstrate a significant difference in strokerisk [18]. However, other studies have demonstrated thatvaccination does reduce HZ incidence, hence theyshould be recommended for individuals at high vascularrisk or with significant comorbidities [34]. Our findingsfocus solely on results from observational studies, al-though we acknowledge that randomized controlled tri-als would not be ethically possible in this setting.Observational studies are limited by their lack ofrandomization and inherent risk of bias. However, allnine studies were rated as having a low risk of bias andadjusted for important confounders such as demograph-ics and a range of cardiovascular risk factors. Two stud-ies implemented a self-controlled case series design,which potentially gives even stronger confounding con-trol [16, 18]. However, we cannot rule out the possibilityof some residual confounding due to stress, mentalhealth or life events. The study by Sundström [19] had agreater risk of bias due to its primary aim being to assessHZ incidence, hence the recording of methods used toassess stroke risk were limited.Our study is not without limitations. Occasionally we ob-served heterogeneity amongst studies results potentiallydue to different study designs, as previously mentioned. Allof the studies except Yawn et al. [20], relied on electronicmedical records to ascertain both the HZ and the strokeoutcome. Administrative data has been reported to over-estimate herpes zoster by 10–15%, [35] but other studiesdid make attempts to reduce HZ miscoding, by for ex-ample, excluding recurrent HZ, which may be confusedwith herpes simplex [14] or requiring concurrent antiviraltherapy [18]. Regarding the accuracy of stroke recording, ina study of the UK databases nearly 90% of patients withstroke had diagnoses confirmed in their written medicalrecord [36]. Studies varied as to whether their stroke defin-ition included TIA or not, and those including TIA wouldhave potentially overestimated risk as TIA is more commonthan stroke [14]. 10 or READ/10 or READ codes to identifyzoster cases, study outcomes and confounders, except Yawnet al., [20] who used medical records, and Sreenivasan etal., [21] who used antivirals as a proxy of zoster infection.Given that the use of antivirals as surrogate for zoster infec-tion has not been validated, we performed a sensitivity ana-lysis and excluded this study from our analyses; however itdid not change our results.ConclusionsStroke risk is significantly increased shortly after acutezoster infection and remains elevated for up to one year.Herpes zoster ophthalmicus increases stroke risk by alarger magnitude. Increased efforts should be made toprovide vaccinations to at-risk individuals, especiallythose with a high cardiovascular risk.Additional fileAdditional file 1: Assessment for Risk of Bias. Risk of Bias Assessment forCohort Studies Included in the Meta-Analyses Using the Newcastle-Ottawa Quality Assessment Scale. (DOCX 21 kb)AbbreviationsHZ: Herpes zoster; HZO: Herpes zoster ophthalmicus; PHN: Post-herpeticneuralgia; TIA: Transient ischemic attack; VZV: Varicella zoster virusFundingThis study is unfunded.Marra et al. BMC Infectious Diseases (2017) 17:198 Page 10 of 11Availability of data and materialsThe datasets supporting the conclusions of this article are available in theindividual studies which are already published and in the public domain.Authors’ contributionsFM was responsible for the design, implementation, data extraction andinterpretation and drafted the manuscript. KR was responsible for the studydesign, carried out the statistical analysis and wrote the results portion of themanuscript. JR helped with the search strategy, data extraction,interpretation and helped to draft the manuscript. All authors read andapproved the final manuscript.Competing interestsNone of the authors have received reimbursements, fees, or salary from anorganization that may in any way gain or lose financially from thepublication of the manuscript, either now or in the future. None of theauthors hold stocks, shares or patents in an organization that may in anyway gain or lose financially from the publication of the manuscript.Consent for publicationOur manuscript does not contain data from any individual person, andtherefore this section is not applicable.Ethics approval and consent to participateEthics approval was obtained for this study from the University of BritishColumbia’s Behavioural Ethics Committee.DisclosureDr. Marra has received grants from Merck Canada Inc and Pfizer Canada forresearch studies related to epidemiology of herpes zoster and invasivepneumococcal disease, respectively. She has received honoraria from servingon the scientific advisory board of Sanofi Pasteur.Dr. Richardson reports no disclosures. Mr. Ruckenstein reports no disclosures.Author details1University of British Columbia, 2405 Wesbrook Mall, Vancouver, BC V6T 1Z3, Canada.2University of East Anglia, Norwich, United Kingdom.Received: 24 September 2016 Accepted: 21 February 2017References1. Cohen JI. Clinical practice: Herpes zoster. N Engl J Med. 2013;369(3):255–63.2. Boëlle PY, Hanslik T. Varicella in non-immune persons: incidence,hospitalization and mortality rates. Epidemiol Infect. 2002;129(3):599–606.3. Choo PW, Donahue JG, Manson JE, Platt R. The epidemiology of varicellaand its complications. J Infect Dis. 1995;172(3):706–12.4. Gnann JW, Whitley RJ. Clinical practice. Herpes zoster N Engl J Med. 2002;347(5):340–6.5. Kost RG, Straus SE. Postherpetic neuralgia–pathogenesis, treatment, andprevention. N Engl J Med. 1996;335(1):32–42.6. Kawai K, Gebremeskel BG, Acosta CJ. Systematic review of incidence andcomplications of herpes zoster: towards a global perspective. BMJ Open.2014;4(6):e004833.7. Lam FC, Law A, Wykes W. Herpes zoster ophthalmicus. BMJ. 2009;339:b2624.8. Yawn BP, Wollan PC, St Sauver JL, Butterfield LC. Herpes zoster eyecomplications: rates and trends. Mayo Clin Proc. 2013;88(6):562–70.9. Gilden DH, Kleinschmidt-DeMasters BK, LaGuardia JJ, Mahalingam R, CohrsRJ. Neurologic complications of the reactivation of varicella-zoster virus. NEngl J Med. 2000;342(9):635–45.10. Angles EM, Nelson SW, Higgins GL. A woman with facial weakness: a classiccase of Ramsay Hunt syndrome. J Emerg Med. 2013;44(1):e137–8.11. Wise J. Shingles is linked to increased risk of cardiovascular events. BMJ.2015;351:h6757.12. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, et al.Meta-analysis of observational studies in epidemiology: a proposal forreporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE)group. JAMA. 2000;283(15):2008–12.13. Wells GA, Shea B, O’Connell D, Peterson J, Welch V, Losos M, et al. TheNewcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomisedstudies in meta-analyses. http://www.ohri.ca/programs/clinical_epidemiology/oxford.htm14. Breuer J, Pacou M, Gauthier A, Brown MM. Herpes zoster as a risk factor for strokeand TIA: a retrospective cohort study in the UK. Neurology. 2014;82(3):206–12.15. Kang J-H, Ho J-D, Chen Y-H, Lin H-C. Increased risk of stroke after a herpeszoster attack: a population-based follow-up study. Stroke J Cereb Circ. 2009;40(11):3443–8.16. Langan SM, Minassian C, Smeeth L, Thomas SL. Risk of stroke followingherpes zoster: a self-controlled case-series study. Clin Infect Dis Off PublInfect Dis Soc Am. 2014;58(11):1497–503.17. Lin H-C, Chien C-W, Ho J-D. Herpes zoster ophthalmicus and the risk ofstroke: a population-based follow-up study. Neurology. 2010;74(10):792–7.18. Minassian C, Thomas SL, Smeeth L, Douglas I, Brauer R, Langan SM. AcuteCardiovascular Events after Herpes Zoster: A Self-Controlled Case SeriesAnalysis in Vaccinated and Unvaccinated Older Residents of the UnitedStates. PLoS Med. 2015;12(12):e1001919.19. Sundström K, Weibull CE, Söderberg-Löfdal K, Bergström T, Sparén P, Arnheim-Dahlström L. Incidence of herpes zoster and associated events includingstroke–a population-based cohort study. BMC Infect Dis. 2015;15:488.20. Yawn BP, Wollan PC, Nagel MA, Gilden D. Risk of Stroke and MyocardialInfarction After Herpes Zoster in Older Adults in a US CommunityPopulation. Mayo Clin Proc. 2016;91(1):33–44.21. Sreenivasan N, Basit S, Wohlfahrt J, Pasternak B, Munch TN, Nielsen LP, et al.The short- and long-term risk of stroke after herpes zoster - a nationwidepopulation-based cohort study. PLoS One. 2013;8(7):e69156.22. Kwon SU, Yun S-C, Kim M-C, Kim BJ, Lee SH, Lee S-O, et al. Risk of strokeand transient ischaemic attack after herpes zoster. Clin Microbiol Infect OffPubl Eur Soc Clin Microbiol Infect Dis. 2016.23. Nagel MA, Traktinskiy I, Azarkh Y, Kleinschmidt-DeMasters B, Hedley-WhyteT, Russman A, et al. Varicella zoster virus vasculopathy: analysis of virus-infected arteries. Neurology. 2011;77(4):364–70.24. Nagel MA, Gilden D. The relationship between herpes zoster and stroke.Curr Neurol Neurosci Rep. 2015;15(4):16.25. Case Records of the Massachusetts General Hospital articles: The NewEngland Journal of Medicine [Internet]. New England Journal of Medicine.[cited 2016 May 4]. Available from: http://www.nejm.org/medical-articles/case-records-of-the-massachusetts-general-hospital?topic=1&articletype=case-records-of-the-massachusetts-general-hospital&page=826. Nagel MA, Traktinskiy I, Stenmark KR, Frid MG, Choe A, Gilden D. Varicella-zoster virus vasculopathy: immune characteristics of virus-infected arteries.Neurology. 2013;80(1):62–8.27. Elkind MSV, Ramakrishnan P, Moon YP, Boden-Albala B, Liu KM, Spitalnik SL,et al. Infectious burden and risk of stroke: the northern Manhattan study.Arch Neurol. 2010;67(1):33–8.28. Xu J, Lupu F, Esmon CT. Inflammation, innate immunity and bloodcoagulation. Hamostaseologie. 2010;30(1):5–6. 8–9.29. Guiraud V, Amor MB, Mas J-L, Touzé E. Triggers of ischemic stroke: asystematic review. Stroke J Cereb Circ. 2010;41(11):2669–77.30. Wang ZW, Li Y, Huang LY, Guan QK, Xu DW, Zhou WK, et al. Helicobacterpylori infection contributes to high risk of ischemic stroke: evidence from ameta-analysis. J Neurol. 2012;259(12):2527–37.31. Yu M, Zhang Y, Yang Z, Ding J, Xie C, Lu N. Association betweenHelicobacter pylori infection and stroke: a meta-analysis of prospectiveobservational studies. J Stroke Cerebrovasc Dis Off J Natl Stroke Assoc. 2014;23(9):2233–9.32. Chen J, Zhu M, Ma G, Zhao Z, Sun Z. Chlamydia pneumoniae infection andcerebrovascular disease: a systematic review and meta-analysis. BMC Neurol.2013;13:183.33. Smeeth L, Thomas SL, Hall AJ, Hubbard R, Farrington P, Vallance P. Risk ofMyocardial Infarction and Stroke after Acute Infection or Vaccination. N EnglJ Med. 2004;351(25):2611–8.34. Oxman MN, Levin MJ, Johnson GR, Schmader KE, Straus SE, Gelb LD, et al. Avaccine to prevent herpes zoster and postherpetic neuralgia in older adults.N Engl J Med. 2005;352(22):2271–84.35. Yawn BP, Wollan P, St. Sauver J. Comparing Shingles Incidence andComplication Rates From Medical Record Review and Administrative DatabaseEstimates: How Close Are They? Am J Epidemiol. 2011;174(9):1054–61.36. Gibbs RG, Newson R, Lawrenson R, Greenhalgh RM, Davies AH. Diagnosisand initial management of stroke and transient ischemic attack across UKhealth regions from 1992 to 1996: experience of a national primary caredatabase. Stroke J Cereb Circ. 2001;32(5):1085–90.Marra et al. BMC Infectious Diseases (2017) 17:198 Page 11 of 11