Open Collections

UBC Faculty Research and Publications

Celebrities’ impact on health-related knowledge, attitudes, behaviors, and status outcomes: protocol… Hoffman, Steven J; Mansoor, Yasmeen; Natt, Navneet; Sritharan, Lathika; Belluz, Julia; Caulfield, Timothy; Freedhoff, Yoni; Lavis, John N; Sharma, Arya M Jan 21, 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-13643_2016_Article_395.pdf [ 967.53kB ]
JSON: 52383-1.0342299.json
JSON-LD: 52383-1.0342299-ld.json
RDF/XML (Pretty): 52383-1.0342299-rdf.xml
RDF/JSON: 52383-1.0342299-rdf.json
Turtle: 52383-1.0342299-turtle.txt
N-Triples: 52383-1.0342299-rdf-ntriples.txt
Original Record: 52383-1.0342299-source.json
Full Text

Full Text

PROTOCOL Open AccessCelebrities’ impact on health-relatedknowledge, attitudes, behaviors, and statusoutcomes: protocol for a systematic review,meta-analysis, and meta-regression analysisSteven J. Hoffman1,2,3*, Yasmeen Mansoor1,4†, Navneet Natt1,5†, Lathika Sritharan1, Julia Belluz6, Timothy Caulfield7,Yoni Freedhoff8,9, John N. Lavis2,3,10 and Arya M. Sharma11AbstractBackground: Celebrities are highly influential people whose actions and decisions are watched and often emulatedby wide audiences. Many celebrities have used their prominent social standing to offer medical advice or endorsehealth products, a trend that is expected to increase. However, the extent of the impact that celebrities have inshaping the public’s health-related knowledge, attitudes, behaviors, and status is unclear. This systematic review seeksto answer the following questions: (1) Which health-related outcomes are influenced by celebrities? (2) How large of animpact do celebrities actually have on these health-related outcomes? (3) Under what circumstances do celebritiesproduce either beneficial or harmful impacts?Methods: Ten databases were searched, including MEDLINE, EMBASE, PsycINFO, PubMed, CINAHL, CommunicationComplete, Sociological Abstracts, Social Sciences Citation Index, Journals @ Scholars Portal, and ProQuest Dissertations& Theses A&I. Two reviewers conducted title and abstract screening and full-text screening to identify primary studiesthat employed empirical methods (either quantitative or qualitative) to examine celebrities’ impact on health-relatedknowledge, attitudes, behaviors, or status outcomes.Discussion: The results of this review will contribute to our understanding of celebrity influences and how to designpositive evidence-based celebrity health promotion activities. In addition, these findings can help inform thedevelopment of media reporting guidelines pertaining to celebrity health news and provide guidance to public healthauthorities on whether and how to respond to or work with celebrities.Systematic review registration: PROSPERO CRD42015019268Keywords: Famous persons, Public health, Preventive medicine, Health policy, Attitude to health, Health behaviorBackgroundCelebrities can have a tremendous influence on the know-ledge we retain, the attitudes we adopt, and the decisionswe make, including those that affect our health [1–4]. Aprevious systematic meta-narrative analysis identified 14biological, psychological, and social mechanisms throughwhich celebrities influence our health-related behaviors, il-lustrating how the mechanics of celebrity and popular cul-ture is a serious public health issue (see Table 1 for the 14mechanisms) [5, 6]. This previous analysis identified those14 mechanisms by systematically reviewing and synthesiz-ing relevant research from economics, marketing, neuro-science, psychology, and sociology. The review ofeconomics literature showed that celebrities can catalyzeherd behavior, and help distinguish endorsed items fromcompetitors. Marketing studies explained that celebrities’characteristics are transferred to endorsed products whichlends credibility to these products. These findings were* Correspondence:†Equal contributors1Global Strategy Lab, Centre for Health Law, Policy and Ethics, Faculty ofLaw, University of Ottawa, Ottawa, Canada2Department of Health Evidence and Impact and McMaster Health Forum,McMaster University, Hamilton, 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 (, 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( applies to the data made available in this article, unless otherwise stated.Hoffman et al. Systematic Reviews  (2017) 6:13 DOI 10.1186/s13643-016-0395-1supported by emerging neuroscience research whichshowed that brain regions involved in making positiveassociations are activated by seeing or hearing celebrityendorsements. The review of psychology literatureshowed that people are conditioned to react positivelyto celebrity advice and that are subconsciously pushedto follow it to avoid cognitive dissonance and to be-come more like those celebrities they admire. Finally,the sociology literature explained how the spread of ce-lebrity advice through social networks increases its in-fluence and that people follow this advice to acquirecelebrities’ social capital [5, 6].Yet, despite this existing evidence about why peopletrust celebrities with their health, there is less eviden-ce—and, to the best of our knowledge, no synthesizedevidence—measuring the magnitude of this influenceand the conditions that mediate it across differentcontexts.Addressing this evidence gap is vitally important. Onthe one hand, celebrities may serve as an untapped re-source for public health promotion efforts, where their in-fluence could bring about positive changes in publicopinion and health-related behaviors. This positive celeb-rity health effect was witnessed as early as the 1990s, afterbasketball player Earvin “Magic” Johnson announced thathe was HIV-positive [7]. In the weeks following this dis-closure, the US Centers for Disease Control and Preven-tion’s National AIDS Hotline reported over 28,000 callsfrom people expressing an increased concern about HIV/AIDS and seeking AIDS-related information [8].Celebrity advocacy can also lead to the adoption ofcertain health prevention behaviors, as seen more re-cently with Angelina Jolie’s public announcement of herdouble mastectomy. Months later, studies recorded anincrease in the number of high-risk patient screeningsfor the impugned BRCA1 gene [9]. These studies suggestTable 1 Fourteen mechanisms explaining celebrity influenceDiscipline Mechanism DescriptionEconomics 1) Signals Celebrity endorsements act as markers that differentiate endorsed itemsfrom competitors.2) Herd behavior Celebrities activate people’s natural tendency to make decisions based onhow others have acted in similar situations.Marketing 3) Meaning transfer People consume items to acquire the endorsing celebrities’ traits, whichhave become associated with the product.4) Source credibility Celebrities share personal experiences and success stories associatedwith the endorsed item to be perceived as credible sources of healthinformation.5) Halo effect The specific success of celebrities is generalized to all their traits,biasing people to view them as credible medical advisors.Neuroscience 6) Neural mechanisms of meaning transfer Celebrity advertisements activate a brain region involved in formingpositive associations, indicating the transfer of positive memoriesassociated with the celebrity to the endorsed item.7) Neuropsychology of credibility Endorsements from celebrities activate brain regions associated withtrustful behavior and memory formation, thereby improving attitudestoward and recognition of the endorsed item.Psychology 8) Classical conditioning The positive responses people have toward celebrities come to beindependently generated by endorsed items.9) Self-conception People follow advice from celebrities who match how they perceive(or want to perceive) themselves.10) Cognitive dissonance People unconsciously rationalize following celebrity medical adviceto reduce the psychological discomfort that may otherwise resultfrom holding incompatible views.11) Attachment People, especially those with low self-esteem, form attachments tocelebrities who make them feel independent in their actions,supported by others, and competent in their activities.Sociology 12) Social networks Celebrity advice reaches large masses by spreading through systemsof people linked through personal connections.13) Commodification and social capital People follow celebrity medical advice to gain social status and shapetheir social identities.14) Social constructivism Celebrity medical advice may alter how people perceive healthinformation and how it is produced in the first place.Reproduced from Hoffman SJ, Tan C. Biological, psychological and social processes that explain celebrities’ influence on patients’ health-related behaviors. Archivesof Public Health. 2015:73(3). doi:10.1186/2049-3258-73-3Hoffman et al. Systematic Reviews  (2017) 6:13 Page 2 of 13that celebrities can serve as agents of positive socialchange, erasing stigma associated with disease andprompting information-seeking and preventative behav-iors. However, at the same time, the power of celebritiesto sway public opinion can equally be a cause for con-cern [10–12]. Some authors criticize celebrities for pre-senting biased health information that evokes irrationalfear and persuades audiences to behave in a certain way,rather than educating patients [11]. For example, Sabeland Sin discovered that many of the women who con-sulted a surgeon for an elective bilateral mastectomy fol-lowing Jolie’s announcement were unsuitable candidatesfor the procedure following a thorough evaluation oftheir genome and family history [12]. Therefore, it maybe important to combine celebrity advocacy with expert-led patient education in order to provide complete andaccurate information about health issues and appropri-ately guide patient behavior [10–12].Celebrities may also negatively affect the public’shealth [13–15]. TV celebrity Jenny McCarthy’s anti-vaccine movement, for example, has captured signifi-cant public attention and roused concerns about vac-cine safety [16–18]. Such celebrity advocacy can becounterproductive to the efforts of public health orga-nizations that invest substantial resources in promot-ing the life-saving benefits of immunization [19].With social media, it is now easier than ever for ce-lebrities, journalists, and amateur bloggers to commu-nicate directly with the public to influence theirknowledge, attitudes, and/or behaviors; as technologyfurther develops, this ability of non-experts to reachthe masses will only increase and become more im-portant to understand [20–22]. For example, in 2012,pop and R&B sensation Beyoncé signed a $50 millionadvertising contract with Pepsi—news that made na-tional headlines [23]. This is not the first instancewhere famous musicians have been found to advertiseunhealthy food items to the public through mediaoutlets. A descriptive study by Bragg et al. found that81% of foods endorsed by musicians featured in theTeen Choice Awards were of poor nutritional quality[24]. Given that youth and adolescents are a primarytarget of these promotional advertisements and more-over a significant at-risk group for obesity, these find-ings suggest inherent value in establishing guidelinesthat limit child exposure to celebrity food advertise-ments [13, 24]. As such, there is a potential oppor-tunity for public health authorities to partner withcelebrities, provided we better understand the impactof celebrity involvement and the conditions that de-termine this impact.This planned review will compile empirical researchevidence that evaluates the impact of celebrity health ac-tivities and the conditions that influence the directionand magnitude of that impact. To the extent possible,we will use a meta-analytic approach to synthesize theresults from previous studies. Qualitative research willalso comprise a critical component of our analysis, as westrive to better understand the health-related attitudes,behaviors, and preferences that are pervasive in society,their sociocultural underpinnings, and how celebritiesmay influence these beliefs. Ultimately, this review willuse both quantitative and qualitative evidence to answerthe following questions:1) Which health-related outcomes are influenced bycelebrities? We will focus on identifying the impactof celebrities’ advice on various health-related know-ledge, attitudes, behaviors, and status outcomes.Health-related outcomes of interest will include aseries of short-term, intermediate, and long-termoutcomes, and are illustrated in the logic model (seeFig. 1).2) How large of an impact do celebrities actually haveon these health-related outcomes? We aim to un-cover both the directionality and the magnitude ofcelebrity impact on health by conducting a meta-analysis of quantitative studies.3) Under what circumstances do celebrities produceeither beneficial or harmful impacts? We willidentify underlying and/or contextual factors thatmay also influence the impact of celebrities byconducting subgroup analyses of factors (i.e.,demographical impact and location) as well as meta-regression analysis. Thematic synthesis of qualitativeresearch will also shed light on individual factorsthat may render certain groups of individuals moresusceptible to celebrity influence.Currently, there only exists a fragmented collectionof primary studies evaluating celebrities’ impact onhealth. Topics investigated by these primary studies in-clude body image, cancer screening, smoking, and sui-cide. In addition, outcomes are evaluated across variouspopulations and environments. Through the plannedsystematic review and meta-analysis, we will determinethe extent to which the health effects of celebrity activ-ities are consistent across the range of outcomes, popu-lations, environments, and interventions. In order toeffectively analyze such a heterogeneous pool of dataand offer meaningful conclusions, studies will be cate-gorized by common themes and outcomes. By encom-passing a wide range of quantitative and qualitativeevidence in our review, we aim to produce meaningfuleffect sizes to enhance our understanding of thepresent celebrity-health phenomenon and guide policy-makers in facilitating significant and positive publichealth initiatives.Hoffman et al. Systematic Reviews  (2017) 6:13 Page 3 of 13Overall, the results of this review should be helpful inunderstanding when to worry about negative celebrityinfluences and designing positive evidence-based celeb-rity health promotion activities. Findings could also helpinform the development of media reporting guidelinespertaining to celebrity health news and provide guidanceto public health authorities on whether and how to re-spond to or work with celebrities.MethodsStudy registrationWe will conduct this systematic review, meta-analysis,and meta-regression analysis adhering to the followingprotocol and will report any changes to the protocol thatarise as we proceed. The methods and design of this sys-tematic review protocol are in accordance with the Pre-ferred Reporting Items for Systematic Reviews andMeta-Analyses (PRISMA-P), available as an additionalfile to this protocol (see Additional file 1). This protocolis registered with PROSPERO (CRD42015019268).Types of study designsBoth quantitative and qualitative studies will be includedin this systematic review to identify all empirical re-search evidence pertaining to the impact of celebritieson health-related outcomes.All quantitative impact evaluations, including experi-ments (e.g., randomized controlled trials), quasi-experiments (e.g., interrupted time-series analyses), andobservational designs (e.g., pooled time-series and cross-sectional analyses) will be included. Based on the resultsof our pilot search, we anticipate that most studies en-countered will likely be observational in nature (seeAdditional file 2 for pilot search strategy). Within obser-vational evaluations, both ecological studies (i.e., com-parisons of groups rather than individuals) andindividual-level studies (e.g., surveys of opinions) will beassessed. It is of value to include both ecological andindividual-level studies since ecological studies aloneface limitations with respect to identifying causal effects,whereas individual-level studies have an increased risk ofreporting bias [25].Fig. 1 Logic model of the systematic review. This logic model illustrates the rationale and the interaction among the health-related outcomemeasures that will be assessed by this reviewHoffman et al. Systematic Reviews  (2017) 6:13 Page 4 of 13Qualitative studies will complement the knowledgegained from quantitative analyses, especially in providingrich insights into the contexts in which celebrities affecthealth-related outcomes. Qualitative research evidencewill specifically allow us to better understand the mean-ing of the quantitative data, explore how individuals orgroups may perceive certain celebrity interventions, andbetter understand the biological, psychological, and so-cial mechanisms underlying the factors that mediate anyinfluence.Types of participantsWe will only include studies that evaluate the impact ofcelebrity health activities on individuals or groups of in-dividuals. Studies that evaluate the impact of celebritieson corporations, governments, organizations, or similarentities will not be considered.Types of settingsStudies from all settings and countries will be consid-ered, which is appropriate given that the influence of ce-lebrities manifests itself in different places and acrossgeography and culture.Types of interventionsInterventions include any health-related campaigns,news events, programs, or statements that primarily re-volve around a celebrity or several celebrities—whetheror not the celebrities intended to cause a health effect.Existing literature defines celebrity status in a number ofways, with a recent emphasis on celebrity as a type ofcapital that results from accumulated media visibilityand recognizability [26, 27]. We aim to be liberal in ourunderstanding of who are “celebrities,” with the recogni-tion that various cultures may endow celebrity status indifferent ways and this may or may not be linked tomedia presence. Therefore, we chose to define celebritiesas well-known and highly visible individuals in society,including but not limited to athletes, entertainmentstars, media personalities, politicians, religious leaders,and socialites. Interventions will be included regardlessof whether or not a celebrity played an active first-handrole in influencing a health-related outcome. In somecases, a celebrity may personally advocate for health-related behaviors; in other cases, a celebrity may be in-directly involved, such as a newspaper reporting on aparticular health condition with which a celebrity wasrecently diagnosed. The goal is to broadly evaluate theimpact of celebrities on health-related outcomes ratherthan only purposefully designed celebrity interventionsthat specifically aim to affect health-related outcomes.Randomized controlled interventions are not necessaryfor a study to be included in this review. Examples ofpossible control interventions include, but are notlimited to, comparator groups who were not exposed tothe celebrity health activity. Studies without a controlgroup will still be included in the review for the pur-poses of qualitative synthesis; however, only studies thatassess the intervention in comparison to a control groupwill be included in the meta-analysis.Type of outcomesThe outcomes of interest in this review can be dividedinto short-, medium-, and long-term effects of celebrityintervention on health-related outcomes. In the short-term, celebrities may alter the public’s health-relatedknowledge, such as by changing the public’s understand-ing of a certain disease’s etiology, risk, diagnosis, andtreatment. In the medium-term, celebrities may changethe health-related attitudes and/or behaviors of the pub-lic. The Health Belief Model proposed that individuals’perceptions of susceptibility, severity, benefits, and bar-riers related to health-related issues are precursors fortheir “readiness to act” and perform various health-related behaviors [28]. As such, we defined “health-re-lated attitudes” to encompass this range of perceptionsthat help to shape individuals’ intentions to performhealth-related behaviors. For instance, fear of needlesmay be viewed as a health-related attitude that will influ-ence whether an individual seeks vaccination. A healthbehavior was originally defined by Kebl and Cobb as anyactivity undertaken by a person who believes himself tobe healthy for the purpose of preventing disease or de-tecting disease in an asymptomatic stage [29]. For thepurposes of this review, we are interested in examiningboth positive and negative celebrity influences as it ap-plies to all individuals, both healthy and unhealthy, andthus defined a “health-related behavior” as any actionthat may promote or diminish one’s health status, bethat physical, mental, or emotional well-being. For ex-ample, the decision of a high-risk patient to be screenedfor cancer would be a positive health behavior that canpotentially benefit long-term health status, whereas de-clining treatment contrary to medical advice would be anegative health behavior that can lead to diminishedhealth status. Finally, in the long-term, health status out-comes such as the incidence and prognosis of a prevent-able disease will be considered. These long-term healthstatus outcomes are this review’s primary outcomes ofinterest; however, the secondary short- and medium-term outcomes serve as helpful secondary surrogatemeasures in the meta-analysis and potential mediatingfactors for study in the meta-regression analysis. Collect-ing evidence of celebrity influence on a wide range ofprimary and secondary outcomes from short- to long-term will allow for a comprehensive assessment of celeb-rity influence on health. The logic model illustrates thisHoffman et al. Systematic Reviews  (2017) 6:13 Page 5 of 13Table 2 Generic search strategy and adapted search strategy for MEDLINE yielding 5157 recordsGeneric search strategy (celebrit* or ((professional or elit* or famous or public or renown* or well-known or acclaim* or eminent orprominent or illustrious or recogniz* or reput* or influential or wealth* or power*) adj1 (person* or people orfigure* or leader or athlete* or player or bodybuilder or sport* or basketball or football or hockey or baseball orsoccer or Olympian or singer* or songwriter* or musician* or band or group or rapper* or artist* or actor* oractress or star or Hollywood or Bollywood or Nollywood or dancer or writer or author or comedian or performer ormodel* or supermodel* or chef or philanthropist or politic* or president or minister or king or queen or prince* ormonarch)))AND(health or wellness or wellbeing or aging or longevity or disorder* or disease* or cancer or epidemic orpandemic or disability or impair* or ill* or sick* or ailment or malady or syndrome* or infection* or mortality ormorbidity or death or dead or injur* or accident* or pain or incident*)AND(communicat* or promot* or endors* or advert* or convinc* or market* or persua* or dissua* or sell* or sale* orpublici* or awareness or campaign or ‘media coverage’ or announc* or message* or disclos* or advoca* oradvis* or advice or counsel* or educat* or instruct* or teach* or inform* or misinform* or prevent* or learn* orbehavior?r* or act* or practice* or habit* or lifestyle* or regime or choice* or decision* or prefer* or attitude* orknow* or belief* or perception* or view* or react* or response* or respond or stigma* or understand* oropinion* or litera* or illitera* or misunderstand* or misconcept* or misconstruct* or disbelief)AND(quantitative or qualitative or empirical or data or statistic* or evidence or stud* or “multi?method*” or survey*or interview or “self report*” or poll* or sampl* or experiment* or measure* or analyz* or analys* or ‘focusgroup*’ or question* or query or queri* or observation* or “field stud*” or phenomenolog* orphenomenograph* or ethnolog* or ethnograph* or “action research” or “grounded theory” or “case stud*”)AND(effect* or affect* or impact* or differ* or compliance or comply or adher* or implement* or influenc* orchang* or measure* or constrain* or screen* or control* or deter* or reduc* or increase* or decreas* or inflat*or vary or variation* or varie*)MEDLINE search strategy (yielding5157 records)1. celebrit*.ti,ab.2. ((professional or elit* or famous or public or renown* or well-known or acclaim* or eminent or prominent orillustrious or recogniz* or reput* or influential or wealth* or power*) adj1 (person* or people or figure* or leaderor athlete* or player or bodybuilder or sport* or basketball or football or hockey or baseball or soccer or Olympianor singer* or songwriter* or musician* or band or group or rapper* or artist* or actor* or actress or staror Hollywood or Bollywood or Nollywood or dancer or writer or author or comedian or performer ormodel* or supermodel* or chef or philanthropist or politic* or president or minister or king or queen orprince* or monarch)).ti,ab.3. exp Famous Persons/4. 1 or 2 or 35. exp Health Promotion/6. exp preventive health services/or exp health education/7. exp attitude to health/or exp health knowledge, attitudes, practice/8. exp Health Communication/9. exp health behavior/or exp information seeking behavior/10. exp information dissemination/or exp information literacy/or exp health literacy/11. exp Public Opinion/12. (communicat* or promot* or endors* or advert* or convinc* or market* or persua* or dissua* or sell* orsale* or publici* or awareness or campaign or “media coverage” or announc* or message* or disclos* oradvoca* or advis* or advice or counsel* or educat* or instruct* or teach* or inform* or misinform* or prevent*or learn* or behavior?r* or act* or practice* or habit* or lifestyle* or regime or choice* or decision* or prefer* orattitude* or know* or belief* or perception* or view* or react* or response or respond* or stigma* orunderstand* or opinion* or litera* or illitera* or misunderstand* or misconcept* or misconstruct* ordisbelief).tw.13. 5 or 6 or 7 or 8 or 9 or 10 or 11 or 1214. (health or wellness or wellbeing or aging or longevity or disorder* or disease* or cancer or epidemic orpandemic or disability or impair* or ill* or sick* or ailment or malady or syndrome* or infection* or mortality ormorbidity or death or dead or injur* or accident* or pain or incident*).tw.15. exp Public Health/16. 14 or 1517. (quantitative or qualitative or empirical or data or statistic* or evidence or survey or stud* or interview or‘self report*’ or poll* or experiment* or measure* or analyz* or analys* or ‘focus group*’ or question* or queryor queri* or observation* or ‘field stud*’ or phenomenolog* or phenomenograph* or ethnolog* or ethnograph*or ‘action research’ or ‘grounded theory’ or ‘case stud*’ or ‘multi?method*’).mp.18. exp empirical research/or exp qualitative research/19. exp health surveys/or exp interviews as topic/or exp focus groups/or exp questionnaires/or exp self report/or exp sampling studies/or exp sample size/or exp observation/20. 17 or 18 or 1921. (effect* or affect* or impact* or differ* or compliance or comply or adher* or implement* or influenc* orchang* or measure* or constrain* or screen* or control* or deter* or reduc* or increase* or decreas* or inflat*or vary or variation* or varie*).tw.22. 4 and 13 and 16 and 20 and 21Hoffman et al. Systematic Reviews  (2017) 6:13 Page 6 of 13rationale and the interaction between these outcomemeasures in determining short-, medium-, and long-term health-related outcomes (see Fig. 1).Search strategyWe constructed a generic search strategy and adapted itfor the MEDLINE database in consultation with healthsciences and social sciences information specialists atMcMaster University (see Table 2). With the goal ofconducting the most comprehensive review possible, weaimed to maximize the sensitivity of our search byencompassing a broad range of potential celebrity inter-ventions and health-related outcome measures in thesearch terms. We maintained specificity by taking a con-servative approach to defining the inclusion and exclu-sion criteria (see “Selection of studies” section below).We adapted the generic search strategy for the follow-ing ten electronic databases (see Additional file 3 for alladapted search strategies): CINAHL (1982 to July 2014);Communication Complete (1915 to July 2014); EMBASE(1947 to July 2014); Journals @ Scholars Portal (no daterestrictions); OVID MEDLINE In-Process & Other Non-Indexed Citations (1946 to July 2014); Proquest Disserta-tions & Theses A&I electronic database (all dates);PsycINFO (1806 to July 2014); PubMed (1966 to July2014); Social Sciences Citation Index (1976 to July2014); and Sociological Abstracts (1952 to July 2014).We searched a combination of health and social sciencesdatabases and additionally looked for gray literature. Allsearches were run on July 31, 2014. There were no lan-guage restrictions and the databases were searched fromtheir earliest date of inception.Selection of studiesTwo members of the research team (YM/NN) independentlyscreened the title and abstract of each study for inclusionbased on a pre-established eligibility screening form (Add-itional file 4). This eligibility criteria form was used to screenfor both quantitative and qualitative studies and, moreover,helped to categorize which studies belonged to which studydesign. Articles were selected during the title/abstract screen-ing if reviewers answered “yes” to all three questions: Independent variable: Does the study explore apotential association with celebrities as theintervention? Dependent variable: Does the study explore apotential association with outcomes associated withhealth-related knowledge, attitudes, behaviors, and/or status? (Note that health-related behaviors en-compass a wide range of actions that can either pro-mote or threaten one’s health) Method: Is the study an empirical evaluation?(Either quantitative or qualitative)Disagreements were resolved through discussion be-tween the two reviewers. If there had been a case inwhich a consensus could not be reached, the principalinvestigator (SJH) would have been consulted, althoughthat was never necessary. Studies that were eligible forinclusion based on title and abstract screening were thenreviewed in a full-text screening exercise. The same in-clusion criteria were used during full-text screening asthe title/abstract screening but with two adjustments:first, the full-text screening required a precise examin-ation of whether celebrity interventions were being eval-uated and if a health-related knowledge, attitude,behavior, or status outcome was being measured; andsecond, we narrowed our third eligibility criteria to onlyinclude empirical studies that had been peer-reviewed.Doing so enhanced the specificity of our search, allowingus to exclude studies that were likely to be less robustand credible. Again, the two independent reviewers usedan eligibility screening form to determine which studiesmet the full-text inclusion criteria (see Additional file 5for the full-text screening form). Disagreements were re-solved through discussion; if consensus could not bereached, the principal investigator would have been con-sulted, although that was never necessary. Figure 2 visu-alizes the first part of a PRISMA flowchart showing thenumber of articles included and excluded at each stageof screening.Data extraction from quantitative studiesGiven the likely variability among studies and the chal-lenges we anticipate in developing a standardized formthat can be applied to all included studies, three stagesof data extraction will be conducted. In the first stage,general study information will be gathered including de-tails about participant demographics, interventions, andmethods employed (see Additional file 6 for a draft formfor stage one of data extraction). This will help makedata extraction more efficient by grouping together stud-ies with similar study designs. The next phase of data ex-traction will focus specifically on defining outcomevariables. This information will be used to further groupstudies by outcomes, thus facilitating the developmentof a third data extraction form specifically tailored toeach subgroup of studies. The third data extraction formwill focus on collecting precise quantitative data and keyfindings specific to the outcome measures reported inthe second data extraction form. As there will be a largenumber of studies in this review, data extraction will becompleted by two pairs of extractors (four reviewerstotal), with each pair randomly assigned half of thestudies.The three data extraction forms will be developedthrough a series of calibration exercises involving theinvestigators, data extractors, and a statistician.Hoffman et al. Systematic Reviews  (2017) 6:13 Page 7 of 13Specifically, each extractor will independently extractdata from three purposively selected studies with diversemethodologies and celebrity activities. After data extrac-tion is performed, discrepancies will be discussed untilfull agreement is reached among the extractors. We willalso ask each extractor for feedback to modify the dataextraction forms such that they are user-friendly and ap-plicable to the wide range of studies that are included.These calibration exercises will be repeated until all fourreviewers agree on a reliable set of data extraction formsand achieve near-perfect consistency in the extractionprocess. After finalizing the standardized data extractionforms, they will be implemented in Distiller SR onlinesoftware to maximize efficiency of the data extractionprocess. Similar to the calibration exercises, disagree-ments will be resolved by discussion between reviewersto achieve consensus and consultation with the principalinvestigator if necessary.Data extraction from qualitative studiesTwo extractors will independently describe the mainfindings from the included qualitative studies using astandardized form implemented in Distiller SR. Extrac-tors will record specific text fragments from the studieswherever possible and appropriate. In line with our thirdresearch question, we will also extract information onany factors identified as mediators of celebrities’ impacton health-related outcomes and make note of any themesor trends that are uncovered. Overall themes across all in-cluded qualitative studies will be allowed to emerge natur-ally from the data in keeping with thematic synthesismethodology; they will not be pre-identified by a priorihypotheses [30, 31]. Any discrepancies about the detailsand conclusions from qualitative studies will be addressedthrough discussion between reviewers, consulting theprincipal investigator if consensus is not possible.Assessment of risk of bias in included studiesStudies that met the inclusion criteria will be assessedfor bias using either the Cochrane Risk of Bias Tool forRandomized Trials or the Cochrane Risk of Bias in Non-Randomized Studies for Interventions (ROBINS-I) tool,depending on study design [32, 33]. Although the kindsof biases are broadly equivalent for randomized andFig. 2 PRISMA flowchart of the systematic review. This flowchart outlines the process of database searches, hand searches, title and abstractscreening, and full-text screening, and lists the number of studies included and excluded with reasonsHoffman et al. Systematic Reviews  (2017) 6:13 Page 8 of 13non-randomized studies—namely selection, confound-ing, group equivalence, spill-overs, and reporting bia-ses—there are important differences in theiroperationalization. Raters will resolve disagreements bydiscussion and through consultation with the principalinvestigator if necessary. Qualitative studies will beassessed for quality using the Critical Appraisal SkillsProgramme (CASP) approach, given its widely cited usein qualitative reviews and recommendation from theCochrane Qualitative Research Methods Group. We willexplore reporting biases such as publication bias byusing funnel plots and assessing the asymmetry bothvisually and using the Egger test [34]. In the presence ofvisual asymmetry, exploratory analyses will beperformed.Data analysisStudies will first be categorized by outcome domain(short-term, medium-term, and long-term health-relatedoutcomes). Within each outcome domain, three types ofanalysis will be conducted, resulting in at least three setsof results: 1) summaries of studies demonstrating thatcelebrities have positive, negative, mixed, or no impactson the health-related outcome; 2) synthesized quantita-tive measurements of the direction and magnitude ofany identified impact that celebrities have on the health-related outcome; and 3) evidence identifying the com-parative influence of different factors, circumstances, orconditions that mediate whether celebrities have positiveor negative effects on the health-related outcome. Ameta-analysis will facilitate all three lines of inquiry,while the subgroup analyses, meta-regression analysis,and the qualitative review will contribute to the thirdline of inquiry.Meta-analysisIn preliminary searches of health sciences and social sci-ences databases, it was found that studies evaluating ce-lebrities’ impact on health were very heterogeneous.Studies investigated a variety of outcomes, ranging fromHIV knowledge to opinions on different foods and tocancer screening behaviors. In addition, these outcomeswere evaluated across various populations, environ-ments, and interventions. Through the meta-analysis, wewill determine the extent to which the effects of celeb-rity influence on health are consistent across this rangeof outcomes, populations, environments, and interven-tions. For questions in which it is plausible, we will con-duct meta-analyses.Our aim is to be liberal in this judgment. We will poolresults from a relatively broad range of studies of differ-ent outcomes, populations, environments, and interven-tions. Having done that, we can examine the variabilityin results to determine the extent to which the datasupports the assumption regarding similar effects acrossoutcomes, populations, environments, and interventions.We anticipate substantial variability and will explore thisthrough pooling across studies that share similar out-comes and conducting subgroup analyses and/or meta-regression analysis (see below). This approach will allowus to estimate a broad summary effect of celebrities onhealth and also estimate more specific celebrity effects.For dichotomous outcomes, the risk ratio or odds ratioand 95% confidence intervals will be used for poolingthe data. For continuous outcomes, mean differencesand 95% confidence intervals will be used for poolingoutcomes reported on the same scales or measured inthe same units. Data will be transformed to allow ana-lyses with mean differences wherever possible. All for-mats of continuous outcome data will be extractedwhether reported as post-intervention or change frombaseline. We will consider using the R value, a correl-ation coefficient, for continuous variables when poolingsome studies that report dichotomous outcomes andothers that report continuous outcomes. If so, resultsfrom individual studies will be converted to an R valuebefore the meta-analysis. The R value can be roughlyinterpreted as a small (R = 0.1), medium (R = 0.3), orlarge (R = 0.5) effect size. We plan to transform thepooled R to another statistic, such as an odds ratio, toaid in interpretation. For time-to-event data, the hazardratio, which is usually estimated from a Cox propor-tional hazards model, will be pooled using the genericinverse variance method.Since we know that the studies will measure outcomesof interest in different units, synthesized effect sizes ofcontinuous outcomes will additionally be measuredusing the standardized mean difference (SMD) (alsosometimes labeled the “effect size”). This will be calcu-lated by dividing the difference in mean values by thepooled standard deviation of the two groups (i.e., inter-vention versus non-intervention), as outlined in the fol-lowing equation:SMD ¼ DifferenceinmeanoutcomebetweengroupsStandarddeviationofoutcomeamongparticipantsThis measure will allow us to compare and pool acrossstudies where outcomes are reported in different units.The SMD will be expressed as a ratio of means to facili-tate meaningful interpretation of the effect size.In interpreting the SMDs, a rule of thumb suggeststhat 0.2 standard deviation units represents a small ef-fect, 0.5 a moderate effect, and 0.8 a large effect. Inaddition to this rule of thumb, to facilitate our interpret-ation, we will convert the SMDs to measures of effecttypically used for binary outcomes. In principle, this ap-proach assumes that data are normally distributed,Hoffman et al. Systematic Reviews  (2017) 6:13 Page 9 of 13allowing calculation of the probability that results aregreater than equal to a particular threshold. These prob-abilities allow calculation of odds ratios and risk differ-ences. There are a number of methods available to conductthe conversion from continuous to binary outcomes. Wewill use the approach described by Furukawa [35].Effect sizes for non-continuous outcomes will be mea-sured by combining risk ratios. Ninety-five percent con-fidence intervals will be calculated for all effectmeasures. We will be conservative in our methodologyby employing a random-effects meta-analysis as it con-siders both within- and between-study heterogeneity.We will statistically examine the extent of heterogeneityusing chi-square tests of heterogeneity and measure theextent of inconsistency with the I2 measure.Subgroup analysesSeveral subgroup analyses are planned to quantitativelymeasure the impact of different kinds of celebrity activ-ities depending on their source, message, and audience,assuming there are a sufficient number of studies withrelevant data to allow for it. These include the following:a) Source: celebrity type. At least one existing study hasshown that entertainment stars may be moreinfluential than politicians in terms of influencingcopycat suicide attempts [36]. This suggests thecelebrity’s occupation may be an importantmediating factor. Types of celebrities that will becompared include athletes, entertainment stars,media personalities, politicians, religious leaders, andsocialites.b) Message: communication channel. As forms of mediadiffer so greatly, so too may the impact of a celebrityhealth activity disseminated through differentchannels [37]. Media differ in their typical length ofexposure, sensory effects, and tone. Audiovisual(television/film), print, radio, and social media willbe compared.c) Message: time period. Are celebrities more influentialon our health in today’s age of social media thanever before? We will group and compare studiesthat evaluated celebrity health activities duringdifferent time periods. At the very least, we plan togroup studies conducted before and after 2004—theyear Facebook launched.d) Message: tone. The impact of celebrities may differbased on whether it is disseminated in a positive ofnegative tone. Positive dissemination involvesconveying a particular opinion or action in afavorable light, while negative dissemination warnsthe public against an opinion or action. For example,one study found that negative reporting of celebritysuicides was associated with 99% fewer copycatsuicides [37]. As possible, studies will bedichotomized according to whether they evaluatedpositively and negatively framed celebrity activitiesto measure the potential influence of tone.e) Message: condition/risk type. Given people oftenhave strong prior perceptions, the impact ofcelebrities may not only be due to their influencealone, but also the public’s existing views aboutparticular conditions and risks. As data allows,celebrities’ impact will be assessed across severalbroad classes of conditions and risk factors,hopefully including cancer, mental illness, physicaldisability, sexually transmitted disease, vaccination,and consumption of harmful products like alcohol,narcotics, and tobacco.f ) Audience: demographics. Certain populations may bemore susceptible to celebrity influence than others,on the basis of age, culture, education, gender, orsocioeconomic status [7]. Studies will be stratified bydemographic factors, depending on the availability ofdemographic data in the included studies. Studieswill also be stratified by World Bank regionalgroupings of countries to see whether the effect ofcelebrity influence differ across countries.We will additionally conduct a subgroup analysis ofthe highest-quality studies to calculate a more conserva-tive broad measure of celebrities’ impact on health-related outcomes.Meta-regressionMeta-regression will be used to explore the reasons fordifferent effect sizes across studies which will help us toquantify the specific influence of various factors thatmight determine celebrities’ impact. Depending on thenumber of studies available from which results can beanalyzed in this way, we will endeavor to consider asmany factors as possible from the list previously identi-fied for the planned subgroup analyses (see above) aswell as any others that might be relevant and feasiblesuch as the study’s design. This analysis will be especiallyhelpful for developing media reporting guidelines onhow to cover celebrity health news and providing guid-ance to public health authorities on whether and how torespond to or work with celebrities.Synthesis of qualitative resultsThematic synthesis will be used to synthesize the datafrom the qualitative studies. This approach combinesfree-coding, iterative categorization of text fragments,and reciprocal translational analysis from meta-ethnography with grounded theory’s inductive approachand constant comparison method [38, 39]. We will aimto synthesize the themes that emerge—particularly thoseHoffman et al. Systematic Reviews  (2017) 6:13 Page 10 of 13on the factors and mediators of celebrity influence onhealth-related outcomes—into a conceptual frameworkthat can inform both understanding and future inquiry.The conceptual framework will supplement the sub-group analyses and meta-regression analysis in order tomake recommendations on how to minimize the harmand maximize the benefits that celebrity health activitiesmay cause in a range of contexts.DiscussionGiven that this is the first review of its kind, we antici-pate several future challenges as we strive to evaluateboth the extent and magnitude of celebrities’ impact ondifferent health-related outcomes. Nevertheless, we haveproactively undertaken several steps to overcome theseanticipated challenges in order to conduct a rigoroussystematic review.Specifically, when defining our independent variable,we recognized that celebrities can include many differenttypes of public figures, such that we adopted a broaddefinition that can be operationalized across varied cul-tures and countries. Additionally, we acknowledged thatcelebrity interventions could assume many differentforms. Given these many considerations, we opted for abroad definition of celebrity interventions in order to en-sure a more comprehensive review of the literature andretrieve as many relevant studies as possible. We tookthe same approach when assigning definitions for thedependent variables of interest, including what com-prises health-related attitudes and behaviors. The use ofbroad definitions allowed us to conduct a highly sensi-tive literature search that yielded 19,365 records thatwere each reviewed.One future obstacle that we foresee involves the ana-lysis of both quantitative and qualitative studies in ouranalysis. In order to maximize the relevance of this studyfor policymakers in their decision-making processes, wefeel it necessary to incorporate diverse forms of evi-dence. While quantitative studies provide a highly pre-cise means to assess celebrity impact, we cannotdiscount the importance of ethnographic research andother qualitative findings since celebrity influence is aculturally based and sociologically rooted phenomenon.A significant challenge therefore lies in finding the mosteffective way to synthesize our qualitative and quantita-tive findings to answer the three research questions. Wewill address this issue by developing two independentstrategies for data analysis—one for quantitative studiesand the other for qualitative studies. As possible, quanti-tative evaluations will be subject to meta-analysis so asto pool different effect sizes and determine the magni-tude of celebrity influence for different health-relatedoutcomes. Meta-regression and subgroup analyses willoffer additional insights as to which contextual factorseither enhance or diminish the aforementioned celebrityeffect. Qualitative studies will complement the know-ledge gained from quantitative analyses and provide add-itional understanding of the factors affecting thecelebrity-health phenomenon. For example, by analyzingthe outcomes of focus groups and surveys, we can drawupon individual perspectives and community experi-ences to better understand the types of attitudes thatexist toward various health topics, the factors that mayshape these attitudes, and the exact motives driving cer-tain health-related behaviors. Conducting two distinctmethods of analysis will preserve the unique benefits of-fered by quantitative and qualitative methodologies, andwill allow us to draw from different pools of findings tofacilitate a more critical analysis of our researchquestions.Finally, there are many considerations to take into ac-count when ensuring our review has sufficient power.From our initial pilot search (Additional file 2), we un-covered a vast amount of literature examining celebrityimpacts on different health topics such as smoking,body image, suicide, and cancer screening. In order toeffectively analyze such a heterogeneous pool of dataand provide meaningful conclusions, we must carefullycategorize all included studies by themes and outcomesbefore performing any statistical tests. This has beenaddressed by conducting data extraction in three dis-tinct stages, with the first two stages analyzing thestudy designs and outcome measures in order to groupthem before proceeding with the final phase of data ex-traction. While organizing the studies in this way iscrucial to achieving a meaningful analysis, it may de-crease the number of studies available for statisticaltesting with respect to each outcome. However, byencompassing a wide range of quantitative and qualita-tive evidence in our study, we aim to produce meaning-ful findings to enhance our understanding of thepresent celebrity-health phenomenon and guide policy-makers in facilitating significant and positive publichealth initiatives.In order for this review to be leveraged to have ameaningful impact on public health, our researchteam is committed to engaging in a series of know-ledge translation activities. By disseminating our find-ings to a wide range of stakeholders in the publichealth arena, health systems can involve celebrities intheir efforts to promote positive health messagesthrough opportunistic moments while working inpartnership to mitigate the potentially negative impactthat celebrities might have. These knowledge transla-tion activities will be specifically targeted toward thedevelopment of updated media reporting guidelines,partnerships with celebrities, and future public healthinterventions.Hoffman et al. Systematic Reviews  (2017) 6:13 Page 11 of 13Additional filesAdditional file 1: Populated PRISMA-P (Preferred Reporting Items forSystematic Review and Meta-Analysis Protocols) checklist. Recommendeditems to include in a systematic review protocol. (DOCX 102 kb)Additional file 2: Pilot search strategy. Reports the search phrases andthe databases on which the pilot search was conducted. (DOCX 117 kb)Additional file 3: Adapted search strategies for each database.(DOCX 147 kb)Additional file 4: Title and abstract screening form. (DOCX 80 kb)Additional file 5: Full-text screening form. (DOCX 81 kb)Additional file 6: Data abstraction form —phase 1. The first dataabstraction form that will be utilized with all studies in order to collectpopulation, intervention, methodology, and preliminary outcomeinformation. (DOCX 111 kb)AbbreviationsCINAHL: Cumulative Index to Nursing and Allied Health Literature; HIV/AIDS: Human immunodeficiency virus infection and acquired immunedeficiency syndrome; SMD: Standardized mean differenceAcknowledgementsWe thank Laura Banfield and Olga Perkovic for providing feedback on oursearch strategy, and Lucy Turner for advising on our data analysis plan.FundingThis study was funded by the Canadian Institutes of Health Research as apart of its Knowledge Synthesis Grant: Fall 2014 competition (#KRS-339658).SJH is additionally financially supported by the Research Council of Norwayand the Trudeau Foundation.Availability of data and materialsNot applicable to this protocol.Authors’ contributionsSJH conceived the review, led its design as principal investigator, andsupervised the implementation. NN led the development of the searchstrategy and conducted all the searches. YM led both stages of articlescreening and led the logistical planning for data extraction, with guidancefrom SJH and LS and assistance from NN. LS and LT advised on theapproach to data analysis. SJH, YM, NN, and LS led the preparation of thismanuscript. LT, JB, TC, YF, JNL, and AMS helped design the review protocoland critically revised this manuscript for important intellectual content. Allauthors will play important roles in the analysis, interpretation, manuscriptpreparation, and knowledge translation of this review in the future. Allauthors read and approved the final manuscript.Authors’ informationSJH is an Associate Professor of Law, Medicine and Public and InternationalAffairs and Director of the Global Strategy Lab at the University of Ottawawith courtesy appointments as an Associate Professor of ClinicalEpidemiology and Biostatistics (part-time) at McMaster University, AdjunctFaculty with the McMaster Health Forum, and Adjunct Associate Professor ofGlobal Health and Population at Harvard University.YM is a medical student at the University of British Columbia.NN is a medical student at the University of Toronto.LS is a Research Coordinator with the Global Strategy Lab at the Universityof Ottawa.JB is a National Magazine Award-winning journalist covering medicine,health policy, and public health for Vox Media Inc.TC is a Professor in the Faculty of Law and the School of Public Health at theUniversity of Alberta and holds a Canada Research Chair in Health Law andPolicy.YF is the Medical Director of the Bariatric Medical Institute, an AssistantProfessor of Family Medicine at the University of Ottawa, and a boardmember of the Canadian Obesity Network.JNL is a Professor in the departments of Health Evidence and Impact andPolitical Science, Director of the McMaster Health Forum, Associate Directorof the Centre for Health Economics and Policy Analysis, and CanadaResearch Chair in Evidence-Informed Health Systems at McMaster University,and Adjunct Professor of Global Health at the Harvard School of PublicHealth.AMS is a Professor of Medicine and Chair in Obesity Research andManagement at the University of Alberta and Scientific Director of theCanadian Obesity Network.Competing interestsThe authors declare that they have no competing interests.Consent for publicationNot applicable to this protocol.Ethics approval and consent to participateNot applicable to this protocol.Author details1Global Strategy Lab, Centre for Health Law, Policy and Ethics, Faculty ofLaw, University of Ottawa, Ottawa, Canada. 2Department of Health Evidenceand Impact and McMaster Health Forum, McMaster University, Hamilton,Canada. 3Department of Global Health and Population, Harvard T.H. ChanSchool of Public Health, Boston, MA, USA. 4Faculty of Medicine, University ofBritish Columbia, Vancouver, Canada. 5Faculty of Medicine, University ofToronto, Toronto, Canada. 6Vox Media, Washington, DC, USA. 7Health LawInstitute, University of Alberta, Edmonton, Canada. 8Bariatric Medical Institute,Ottawa, Canada. 9Department of Family Medicine, Faculty of Medicine,University of Ottawa, Ottawa, Canada. 10Centre for Health Economics andPolicy Analysis and Department of Political Science, McMaster University,Hamilton, Canada. 11Canadian Obesity Network and Faculty of Medicine,University of Alberta, Edmonton, Canada.Received: 21 July 2016 Accepted: 7 December 2016References1. Viale PH. Celebrities and medicine: a potent combination. J Adv PractOncol. 2014;5(2):82–4.2. Tanne JH. Celebrity illnesses raise awareness but can give wrong message.BMJ. 2000;321(7269):1099.3. Ransohoff DF, Ransohoff RM. Sensationalism in the media: whenscientists and journalists may be complicit collaborators. Eff Clin PractECP. 2001;4(4):185–8.4. Caulfield T. Biotechnology and the popular press: hype and the selling ofscience. Trends Biotechnol. 2004;22(7):337–9.5. Hoffman SJ, Tan C. Biological, psychological and social processes thatexplain celebrities’ influence on patients’ health-related behaviors. Arch PubHealth. 2015;73:3.6. Hoffman SJ, Tan C. Following celebrities’ medical advice: meta-narrativeanalysis. BMJ. 2013;347(f7151):1–6.7. Brown WJ, Basil MD. Media celebrities and public health: responses to“Magic” Johnson’s HIV disclosure and its impact on AIDS risk and high-riskbehaviours. Health Commun. 1995;7(4):345–70.8. Kalichman SC, Hunter TL. The disclosure of celebrity HIV infection: its effectson public attitudes. Am J Public Health. 1992;82(10):1374–6.9. Lee J. Celebrity impact. Benefits, risks seen in hype over Jolie’s disclosure.Mod Healthc. 2013;43(20):10–1.10. Caulfied T. Jolie can start a health conversation, but she mustn’t be the lastword. 2015. 28 June 2016.11. Larson RJ, Woloshin S, Schwartz LM, Welch HG. Celebrity endorsements ofcancer screening. J Natl Cancer Inst. 2005;97(9):693–5.12. Sabel MS, Cin SD. Trends in media reports of celebrities’ breast cancertreatment decisions. Ann Surg Oncol. 2016. doi:10.1245/s10434-016-5202-7.13. Dixon H, Scully M, Wakefield M, Kelly B, Chapman K, Donovan R. Parent’sresponses to nutrient claims and sports celebrity endorsements on energy-dense and nutrient-poor foods: an experimental study. Public Health Nutr.2011;14(6):1071–9.14. Martin G, Koo L. Celebrity suicide: did the death of Kurt Cobain influenceyoung suicides in Australia? Arch Suicide Res. 1997;3(3):187–98.Hoffman et al. Systematic Reviews  (2017) 6:13 Page 12 of 1315. Maltby J, Giles DC, Barber L, McCutcheon LE. Intense-personal celebrityworship and body image: evidence of a link among female adolescents.Br J Health Psychol. 2005;10(1):17–32.16. The Economist. Of vaccines and vacuous starlets: a rise in anti-vaccinesentiment has put everyone at risk. 2015. Accessed 6 July 2015.17. Rochman B. Jenny McCarthy, Vaccine Expert? A quarter of parents trustcelebrities. In: Time Magazine. 2011. 6 July 2015.18. Kata A. Anti-vaccine activists, Web 2.0, and the postmodern paradigm—anoverview of tactics and tropes used online by the anti-vaccinationmovement. Vaccine. 2012;30(25):3778–89.19. Public Health Agency of Canada. Canadian immunization guide. 2013. Accessed on 6 July 2015.20. Friedman SM, Dunwoody S, Rogers CL, editors. Communicating uncertainty:media coverage of new and controversial science. Mahwah, N.J: Routledge;1999.21. Grilli R, Ramsay C, Minozzi S. Mass media interventions: effects on healthservices utilisation. Cochrane Database Syst Rev. 2002. doi:10.1002/14651858.CD000389.22. Wakefield MA, Loken B, Hornik RC. Use of mass media campaigns tochange health behaviour. Lancet. 2010;376(9748):1261–71.23. Casserly M. Beyoncé’s $50 Million Pepsi deal takes creative cues from Jay Z.2012. 28 June 2016.24. Bragg MA, Miller AN, Elizee J, Dighe S, Elbel BD. Popular music celebrityendorsements in food and nonalcoholic beverage marketing. Pediatrics.2016;13b(1). doi:10.1542/peds.2015-3977.25. MacArthur GJ, Wright M, Beer H, Paranjothy S. Impact of media reporting ofcervical cancer in a UK celebrity on a population-based cervical screeningprogramme. J Med Screen. 2011;18(4):204–9.26. Celebrity RC. The Wiley Blackwell encyclopedia of consumption andconsumer studies. 2015. p. 1–3.27. Driessens O. Celebrity capital: redefining celebrity using field theory. TheorySoc. 2013;42(5):543–60.28. Janz NK, Becker MH. The health belief model: a decade later. Health EducBehav. 1984;11(1):1–47.29. Kasl SV, Cobb S. Health behavior, illness behavior, and sick-role behavior.Archives Environ Health. 1965;12(4):531–41.30. Bauer MW, Gaskell G. Qualitative researching with text, image and sound: apractical handbook. London: Sage Publications; 2000.31. Strauss A, Corbin J. Basics of qualitative research: techniques and proceduresfor developing grounded theory. California: Sage Publications; 1998.32. Higgins JP, Altman DG, Gøtsche PC, Jüni P, Moher D, Oxman AD, Savovic J,Schulz KF, Weeks LL, Sterne JAC. The Cochrane Collaboration’s tool forassessing risk of bias in randomized trials. BMJ. 2011;343:d5928.33. Sterne et al. ROBINS-I: a tool for assessing risk of bias in non-randomisedstudies of interventions. BMJ. 2016;355:i4919.34. Egger M, Smith GD, Schneider M, Minder C. Bias in meta-analysis detectedby a simple, graphical test. BMJ. 1997;315:629–34.35. Furukawa TA. From effect size into number needed to treat. Lancet. 1999;353:1680.36. Niederkrotenthaler T, Fu K, Yip PSF, Fong DYT, Stack S, Cheng Q, et al.Changes in suicide rates following media reports on celebrity suicide: ameta-analysis. J Epidemiol Community Health. 2012. doi:10.1136/jech-2011-200707.37. Stack S. Suicide in the media: a quantitative review of studies based onnonfictional stories. Suicide Life Threat Behav. 2005;35(2):121–33.38. Thomas J, Harden A. Methods for the thematic synthesis of qualitativeresearch in systematic reviews. BMC Med Res Methodol. 2008;8:45.39. Barnett-Page E, Thomas J. Methods for the synthesis of qualitative research:a critical review. BMC Med Res Methodol. 2009;9:59.•  We accept pre-submission inquiries •  Our selector tool helps you to find the most relevant journal•  We provide round the clock customer support •  Convenient online submission•  Thorough peer review•  Inclusion in PubMed and all major indexing services •  Maximum visibility for your researchSubmit your manuscript your next manuscript to BioMed Central and we will help you at every step:Hoffman et al. Systematic Reviews  (2017) 6:13 Page 13 of 13


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items