Economy & Finance

Computer-tailored health interventions delivered over the web: Review and analysis of key components

Computer-tailored health interventions delivered over the web: Review and analysis of key components
of 18
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
  Review Computer-tailored health interventions delivered over the web: Reviewand analysis of key components Mia Liza A. Lustria a, *, Juliann Cortese b , Seth M. Noar c , Robert L. Glueckauf  d a College of Information, Florida State University, 270 Louis Shores Building, Tallahassee, FL 32306-2100, United States b Department of Communication, Florida State University, United States c Department of Communication, University of Kentucky, United States d Department of Medical Humanities and Social Sciences, College of Medicine, Florida State University, United States 1. Introduction 1.1. What is tailoring and why tailor? In 2007, the United States spent close to $2.3 trillion onhealthcare or $7600 per person [1]. Within the next decade this isexpectedtoriseby6.7%yearly,reachinganestimated$4.3trillionin2017 [2]. The rising cost of healthcare presents a compellingargument for developing more effective health education andcommunication strategies aimed at improving health outcomesamong different segments of the population using a wide array of technologies and in different settings. A common misconception of publichealthprograms,however,istheassumptionthatpresentingfacts based on clinical and epidemiological research to individualsis enough to promote desired health behaviors or that one healthmessage can fit all types of audiences or set of circumstances [3].Over the years, we have come to learn that personalizinginformation or tailoring messages for each individual can be moreeffective than presenting generic information in terms of engagingindividuals, building their self-efficacy and improving healthbehaviors [4–9]. Rimer and Kreuter define tailoring as a processfor creating individualized communications by gathering andassessing personal data related to a given health outcome in ordertodeterminethemostappropriateinformationorstrategiestomeetthat person’s unique needs [10]. Compared to generic information,tailored information is more likely to be read, remembered andviewed as personally relevant [11,12]. Moreover, it enables indi-vidualizedfeedback,commandsgreaterattention,isprocessedmoreintensively, contains less redundant information, and is perceivedmore positively by health consumers [13]. Rimer and Kreuter [10] associate this process with Petty et al. [14] Elaboration LikelihoodModel(ELM).Accordingtothistheory,informationthatisperceived Patient Education and Counseling 74 (2009) 156–173 A R T I C L E I N F O  Article history: Received 5 February 2008 Received in revised form 10 August 2008 Accepted 30 August 2008 Keywords: Health educationComputer-tailoringComputer-assisted methodsAlgorithmsInternet A B S T R A C T Objective:  This systematic review explores how computer-tailored, behavioral interventions implemen-ted and delivered via the Web have been operationalized in a variety of settings. Methods:  Computer-tailored, online behavioral intervention studies published from 1996 to early 2007were selected and reviewed by two independent coders. Results:  Of 503 studies screened, 30 satisfied the selection criteria. The level of sophistication of theseinterventions varied from immediate risk/health assessment, tailored web content to full-blowncustomizedhealthprograms.Themostcommonvariablesfortailoringcontentwerehealthbehaviorsandstages of change. Message tailoring was achieved through a combination mechanisms including:feedback, personalization and adaptation. Conclusions:  Tailored, self-guided health interventions delivered via the Web to date have involved agreat diversity of features and formats. While some programs have been relatively brief and simple,othershaveinvolvedcomplex,theory-basedtailoringwithiterativeassessment,toolsfordevelopmentof self-regulatory skills, and various mechanisms for providing feedback. Practice implications:  Our ability to fully optimize the use of computer-assisted tailoring will depend onthe development of empirically based guidelines for tailoring across populations, health foci, healthbehaviors and situations. Further outcome research is needed to enhance our understanding of how andunder what conditions computer-tailoring leads to positive health outcomes in online behavioralinterventions.Published by Elsevier Ireland Ltd. * Corresponding author. Tel.: +1 850 644 6237; fax: +1 850 644 6253. E-mail address: (M.L.A. Lustria). Contents lists available at ScienceDirect Patient Education and Counseling journal homepage: 0738-3991/$ – see front matter. Published by Elsevier Ireland Ltd.doi:10.1016/j.pec.2008.08.023  to be personally relevant (as in the case of tailored information)enhances an individual’s motivation to elaborate on the message,and consequently, his/her receptivity to persuasion efforts.There is growing evidence for the use of tailoring in behavioralhealth interventions. First, tailored interventions evoke favorableperceptions from individuals [10,15–17]. Tailoring allows for verypersonal and direct content presentation based on elements suchas likes/dislikes, needs, and current health behaviors or behavioralintentions. Tailored interventions have led to positive outcomesacross a variety of interventions, such as binge drinking [18],nutrition and diet [19,20], and smoking [21]. Moreover, recent reviews [22,23] and meta-analyses [24,25] have reported incre- mental benefits for tailored versus non-tailored interventions inthe literature as a whole. 1.2. Rationale and research questions Despite these promising results, we still have limited knowl-edge about how tailoring works and what moderates the effects of tailoring on health outcomes. In a recent meta-analysis of tailoredprint-based behavioral interventions, Noar et al. found that whiletailored print interventions outperformed generic or ‘‘targeted’’messages in effecting behavior change, the overall effect size wasslightlylessthansmallmagnitudewithan r   = .074(Cohen’s d  = .15[24]). Noar et al. also found that tailoring is a product of a complexprocess and that its effects may be influenced by a number of moderating variables, including: (a) type of comparison condition,(b) health behavior, (c) type of participant population, (d) type of print material, (e) number of intervention contacts, (f) length of follow-up, and (g) demographic factors, behaviors, and theoreticalconcepts used for tailoring.While the literature on computer-tailored interventions hasgrowngreatly,withapplicationstomorethan20healthbehaviors,research on computer-tailored interventions delivered over theInternet is in the early stage of development [26]. These new Internet-based applications hold promise for a number of reasons.First,computingtechnologieshavecontributedsignificantlytothelevel of sophistication of tailoring. Powerful expert systems nowmakeitpossibletoautomatethecollectionofpersonalinformationwhich then can be used to provide individualized feedback and todesign personalized health programs. Tailored interventionssupported by expert systems enable tailoring on a variety of factors that influence behaviors, approaching a level of persona-lization that, in the past, could only be achieved through face-to-face interactions with experts or trained health care providers.Second, the Internet is potentially a more powerful channel fordelivering tailored messages and improving access to expert careand feedback. Interactive web technology not only provides anopportunity to present tailored information in a number of for-mats, but it also provides the greatest ability to toggle betweenmodalities,furtherenhancinguser’sexperienceandunderstandingof the material. Moreover, synchronous and asynchronousmodalities facilitate scheduling of interactions and the timelydelivery of reminders and messages to help motivate patients andto ensure their steady progress through their health programs.Combining computer-based tailoring with web delivery,therefore, holds promise for boosting the efficacy of tailoredbehavioral interventions. At the same time, this introduces aheightened level of complexity that makes it more difficult todetermine how tailoring works, and to tease out the differentialeffects of multiple modes of delivery. Before we speculateabout the efficacy of computer-tailored online behavioral inter-ventions, it is important to understand the nature of these typesof interventions and what makes them different from the firstgeneration of tailored print interventions.This study examines how computer-tailored, online beha-vioral interventions have been operationalized and implementedin a variety of settings through a systematic review of efficacystudies. Specifically, this study seeks to answer the followingresearch questions: (1) How are computer-tailored behavioralinterventions implemented and delivered via the Web? and (2)What mechanisms and criteria are used to individualize healthmessages within computer-tailored behavioral interventionsdelivered via the Web?Implementation and delivery of computer-tailored behavioralinterventionsonlinecanvaryconsiderablycomparedtouni-modalprint-based behavioral interventions. The term, implementationstrategies, refers to the level of sophistication of the behavioralintervention, scheduling of individual assessment and delivery of tailored components, level of user control and the type and natureof contact with experts or therapists. Online behavioral interven-tions also include a varietyof componentsin different formats andinteractive activities supporting the achievement of behavioralgoals. While the basic process of developing tailored messages issimilar to that used in print-based interventions, we expect thatmore sophisticated and automated assessment methods, and theweb environment will provide additional options for presentingand delivering tailored messages and health programs.The primary aim of this article is to determine the variety of ways computers are being used to create and deliver tailoredhealth messages in online environments. Additionally, this studyseeks to describe and assess this growing body of research and toconsider future directions and development of this area. Only bygainingbetterinsightintohowcomputersandwebdeliverycanbeused together for tailored interventions will we be able to fullyappreciate the tailoring process and in turn, to maximize thepotential for improving the efficacy of behavioral health inter-ventions. 2. Methods  2.1. Identification of studies Weconductedanextensivesearchforcomputer-tailoredonlinebehavioral health intervention studies published in English-language peer-reviewed journals from 1996 to February 2007.Becauseoftheinterdisciplinarynatureofthisresearcharea,atotalof 23 scholarly databases were searched (see Table 1) using thefollowing main terms in various combinations: ‘Internet’, ‘Web’,‘health intervention’, and ‘tailoring’. The main search terms wereselected based on their high recall ratio (or ability to retrieverelevant articles). Secondary search terms that also yieldedrelevant citations were: ‘health education’, ‘behavior changeprograms’, and ‘computer-tailoring’. The initial search yieldedthousands of citations—503 of these were potentially relevantstudies and were archived in an Endnote (ver. 10) database.Studies were screened in several stages using explicit inclusionandexclusioncriteria(seeFig.1).First,thecitationswerereviewedto judge their general eligibility for the systematic review. Thecitations were evaluated and generally ineligible studies (i.e.,studiesthat didnotincludeabehaviorchange component;studiesabout online health information seeking, telemedicine, and web-based continuing medical education) were excluded ( n  = 257).Abstractsoftheremainingstudies( n  = 246)werefurtherevaluatedand studies that were not randomized controlled trials (RCTs) orquasi-experimentalstudies,didnothaveawebcomponent,and/ordid not incorporate any tailoring also were excluded ( n  = 169).Since we were interested in coding and comparing studies on awiderange of characteristics,comprehensivenessof reporting wasalso a key consideration in selection of articles. M.L.A. Lustria et al./Patient Education and Counseling 74 (2009) 156–173  157  Full-textcopiesoftheremaining77potentiallyrelevantstudieswere examined more closely by two independent coders toconfirm whether computer-tailoring was used in the intervention.‘Computer-tailoring’ was operationally defined as the use of automated methods for collecting, assessing and providing indi-vidualized feedback to patients. Of the 77 online behavior changeinterventions that reported some evidence of tailoring, only 30clearly reported using computer-assisted methods for assessmentand message tailoring. The 47 excluded studies typically achieved‘‘tailoring’’ through manual assessment by experts. Individualizedfeedback may have then been provided synchronously (e.g., viachat, telephone or face-to-face) or asynchronously (e.g., via e-mailor discussion board, or mail). Studies that only focused on processoutcomes (e.g., attitude change and knowledge) and did not targetbehavior change were likewise excluded.  2.2. Coding of computer-tailored intervention characteristics Tworesearcherscoded the eligible studiesindependentlyon anumber of variables including: health focus, length of interven-tion, study design, implementation strategies (sophistication of tailoring method, timing, level of user control, type and nature of expert contact, modalities, tools for building self-regulatoryskills), and message tailoring strategies (i.e., tailoring criteria,tailoringmechanisms).Operationaldefinitionsweresummarizedin a codebook to ensure that categorization was consistently andaccurately applied throughout the coding process. Satisfactoryinter-coder reliability was established with an average percentagreement across all categories of 85% [27]. Any disparities in  judgment that emerged during the coding process were resolvedthrough discussion. 3. Results  3.1. General characteristics of studies A total of 30 online behavioral interventions that clearlyreportedusingcomputer-tailoringmethodswereselectedforthisreview. Table 2 summarizes general characteristics of thesestudies.Theyspannedfourbroadhealthareasincluding:nutritionand diet ( n  = 10), physical activity ( n  = 7), alcoholism ( n  = 3),smoking cessation ( n  = 7) and one study each for encopresis,eating disorders, and general risk behaviors. The majority of thestudies focused on risk prevention (e.g., obesity, cardiovasculardisease), cessation (e.g., alcohol abuse and smoking) andhealth maintenance (e.g., nutrition, diet and exercise), while onlyone dealt with disease management and treatment (e.g., encopr-esis).We began this review with a general idea of the features of computer-tailored behavioral interventions. After immersingourselves in the literature, however, we discovered how morecomplex tailoring is when combined with web delivery methods.Tailoring is a multi-step and multi-dimensional process thatinvolves assessing individual characteristics, creating individua-lized messages, and then, delivering these messages using avariety of appropriate strategies and channels. Due to the varietyofapproachesforcomputer-tailoringtestedwithinvarioushealthareas, we started with two broad categories to help organize thisstudy:(1)generalimplementationanddeliverystrategies,and(2)message tailoring strategies. Fig. 2 presents an organizingheuristic that summarizes the general implementation and keymessage tailoring strategies that may be used in computer-tailored behavioral interventions.The following section discusses how computer-tailoring iscurrently being achieved for a variety of health conditions. Westart with an overview of how tailored interventions are typicallyimplemented online and will present key examples to illustratethese (see Table 3 for a summary). We will then drill down to thetypical strategies for generating tailored messages using compu-terized expert systems and subsequently, present specific exam-ples (see Table 4 for a summary).  Table 1 Academic databases searched.Databases1. AIDS & Cancer Research2. ArticleFirst3. CINAHL and Pre-CINAHL 4. Communication Abstracts5. ContentsFirst6. Educational Research Abstracts Online (ERA)7. Emerald Library E-Journals (MCB Publications)8. ERIC (FirstSearch)9. Expanded Academic ASAP10. IngentaSelect11. ISI Web of Knowledge12. JSTOR 13. Kluwer (E-Journals)14. Medline (via PubMed)15. PsycARTICLES16. PsycINFO17. ScienceDirect18. Social Sciences Citation Index19. Sociological Abstracts20. Springer LINK21. Synergy22. Wiley Interscience Journals23. Wilson Science Complete Fig. 1.  Summary of selection process. RCT = Randomized Controlled Trial. M.L.A. Lustria et al./Patient Education and Counseling 74 (2009) 156–173 158   Table 2 Summaries of the efficacy studies of computer-tailored online behavioral interventions selected for this review.Randomizedcontrolled trialHealth focus & targetpopulationLength TheoreticalframeworkDescription and effects1. Block et al. [75] Nutrition/Diet(employees ata corporateworksite,21–63-year-old)12 weeks Stages of change This study was a nonrandomized pilot study of the WorksiteInternet Nutrition (WIN) program, a fully automated assessmentand tailoring program delivered weekly via email. The programassessed the individual’s stage within the Stages of Change model atbaseline and asked that participants choose a dietary emphasis fortheir 12-week program (either reducing fat intake or increasing fruitand vegetable intake), then delivered weekly emails based on eachindividual’s stage and specified emphasis. Each email containednutrition information, individually tailored dietary tips, and goals forthe following week. No comparison group was used. Results indicatedincreased fruit and vegetable consumption, decrease in fat intake,and positive movement through the stages of change.2. BruningBrown et al. [76]Eating disorders(10th gradefemales &69 parents)8 weeks (girls),4 weeks(parents)Not indicated Participants were assigned to one of two groups for this pre-test/post-test study: the intervention (Student Bodies internet deliveredprogram;  n  = 102) and a comparison group ( n  = 51; the details of the comparison group are not clear, but may have received basichealth information in class). Tailoring was achieved by assessingattitudes and behaviors regarding weight and shape followed byimmediate feedback indicating whether or not participants may beat risk of an eating disorder. Results indicated that students usingthe intervention reduced eating restraint and had greaterincreases in knowledge regarding eating disorders. However, nosignificant differences were found at follow-up. (It appears as thoughonly the student intervention portion of this study was tailored andnot the parent portion. Parents also decreased their critical attitudestoward weight and shape.)3. Buller et al. [45] Smoking(6th–9th graders)6 class meetings(45–60 minsessions)Social cognitivetheoryConsider this, a 6-module tailored interactive program on smokingprevention was presented to 6th–9th graders in the US (1234)and Australia (2077) to help reduce expectations of smokingand 30-day smoking prevalence. The study used pair-matched,group-randomized, pretest-posttest controlled design, with schoolas the unit of randomization comparing intervention (AU InterventionGroup = 754; US Intervention Group = 640) and a standard healtheducation group) AU Control Group = 756; US Control Group = 364).Tailoring was assessed using motivational interviewing technique a andwas based on levels of experience with smoking. Interactive contentavailable in different media formats were provided in 6 modules(delivered in six 45–60 min weekly sessions) as students progressedthrough the program. The program reported lower expectations of smoking in the future only among American children and sufferedproblems of low exposure among classes.4. Chiauzzi et al. [18] Alcoholism(college students)12 weeks (four20-minsessions/week)Stages of change ortrans-theoreticalmodelThis 12-week study evaluated the efficacy of,a brief tailored web-intervention designed to provide motivationalfeedback to college students with a high risk for binge drinking.The study used a single-blind randomized, controlled design tocompare the intervention ( n  = 131) with a text-based education-onlyWeb site ( n  = 134) on the alcohol use, alcohol-related problems, andreadiness to change of students who binge drink. Immediate tailoredfeedback was based on students’ behaviors and risk perceptions andmade available for print out. Students then had access to a healthwebsite with tailored content representing personal risk factors andvarious skills building tools. The intervention involved minimal expertcontact with trained clinical evaluators who administered the DailyDrinking Questionnaire to the students at each collection point. Thetailored web intervention was found to significantly reduce alcoholconsumption especially among female students and persistent bingedrinkers.5. Cobb et al. [77] Smoking (websiteregistrants, 37.3years mean age)3 months Stages of change Survey study evaluating the free version of QuitNet, a smokingcessation website. No comparison group was used. Participants indicatedrisk behaviors and stage of change, which was used to provided tailoredcontent. Personalized program content was provided to the participantsby providing a quit calendar (with coping strategies) based on theindividual’s quit date and a list of steps for smoking cessation. Tailoredcontent also focused, gender, age, quit history, and medication status.The site also includes social support and mediated contact with experts.Results indicated that sustained use of QuitNet, especially use of support tools, had positive effects on abstinence rates. M.L.A. Lustria et al./Patient Education and Counseling 74 (2009) 156–173  159   Table 2 ( Continued ) Randomizedcontrolled trialHealth focus & targetpopulationLength TheoreticalframeworkDescription and effects6. Etter [30] Smoking (smokersand ex-smokers;33.8 years mean age)10 weeks Stages of change ortrans-theoreticalmodelVisitors of, a French-language anti-smoking websitethat provides information, advice, and support to smokers andex-smokers were randomized either to an srcinal tailored healthintervention site based on psychological and addiction theory( n  = 5966) or to an intervention site focusing on nicotine replacementtherapy ( n  = 6003). Tailoring was based on demographiccharacteristics, smoking status, stage of change, level of tobaccodependence, attitudes towards smoking, self-efficacy, use of self-change strategies and coping methods, and intention to usenicotine replacement therapy (NRT). Results showed that smokersin the contemplation stage and former smokers showed higherabstinence rates using the srcinal program than when using themodified program (differences in the modified program: shorterassessment questionnaire, more information on nicotine replacementand dependence and less on health risks/coping strategies).7. Frenn et al. [61] Nutrition/diet/exercise(7th grade students)8 sessions Stages of change ortrans-theoreticalmodelThis study evaluated an Internet/video-delivered intervention designedto increase physical activity and reduce dietary fat amonglow-income, culturally diverse, seventh-grade students. The study useda pair-matched, group-randomized, pretest-posttest controlled design,with class section as the unit of randomization comparing effectsof a computer-tailored educational site ( n  = 43) versus a standardscience education site ( n  = 60). Students in the intervention groupwere exposed to eight 40-min sessions with individualized feedback.Computer-generation of tailored feedback was based on Stages of Change. The tailored intervention was found to be effective inimproving diet and physical activity among the students.8. Gold et al. [35] Nutrition/diet(overweight andobese adults; 46.5years mean age)48 weeks Not indicated This randomized controlled study compared effects of a therapist-led,structured behavioral weight loss online health intervention(VTrim;  n  = 62) to a self-help commercial weight loss website(;  n  = 62) over a 4-week intervention period. Tailoring forthe VTrim intervention was essentially expert-led through provisionof e-mailed feedback for weekly e-journal entries. Tailoring for theeDiets website was computer-generated based on weight lossprogression and food preferences. The therapist-led online healthprogram was more successful at helping patients lose weightcompared to the commercial weight loss website.9. Hagemanet al. [32]Physical activity(50–69-year-oldhealthy women)12 weeks Social cognitivetheoryThis 12-week randomized controlled study compared the effectsof a computer-tailored behavioral intervention using tailored onlinenewsletters ( n  = 15) versus standard online newsletters ( n  = 16) onthe physical activity of women ages 50–69 years. Tailoring was basedon level of self-reported physical activity, benefits and barriers toactivity, and self-efficacy and initial goals for activity. Both groupsimproved in flexibility but neither group increased in self-reportedtime in physical activity. Women in the tailored intervention seemedto improve more on perceived behavioral influences (increasedperceived benefits of exercise and self-efficacy to exercise, anddecreased perceived barriers to exercise) but there were no over-allsignificant differences between women in the tailored versusstandard interventions on main outcomes.10. Hageret al. [67]Physical activity(universityemployees;42 years mean age)10 weeks Stages of change ortrans-theoreticalmodelThe study evaluated the efficacy of two computer-tailored onlineinterventions designed to increase physical activity among employeesof a large university. Participants were randomly assigned to one of three groups: a stage-based tailored intervention ( n  = 175); anaction-message tailored intervention ( n  = 175); or a control group( n  = 175). The two intervention groups received either tailoredstage-based or tailored action messages at baseline followed by 5weekly tailored motivational e-mail messages while the controlgroup received 5 weekly general e-mails encouraging proper nutritionand no baseline message. The online assessment included thefollowing items: demographic variables, staging questionnaire andalgorithm, assessment of physical activity, leisure time activity,occupational activity, and self-efficacy assessment. The stage-basedtailored information group showed only increases in leisure timephysical activity while the action-message group showed increasesin leisure time activity, occupational activity, and daily energyexpenditure. Readiness to change increased in all conditions. M.L.A. Lustria et al./Patient Education and Counseling 74 (2009) 156–173 160
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!