The Florida Shock Anxiety Scale (FSAS) for Patientswith Implantable Cardioverter Defibrillators: TestingFactor Structure, Reliability, and Validity of a PreviouslyEstablished MeasureJESSICA FORD, M.A.,* JOHN F. FINCH, PH.D.,* LAWRENCE K. WOODROW, M.A.,*KATHERINE E. CUTITTA, B.A.,* JULIE SHEA, M.S.,† AVI FISCHER, M.D.,‡GARRETT HAZELTON, PH.D.,§ and SAMUEL F. SEARS, PH.D.¶,**From the *Department of Psychology, East Carolina University, Greenville, North Carolina; †Brigham and Women’sHospital, Boston, Massachusetts; ‡Mount Sinai Medical Center, New York, New York; §Duke University MedicalCenter, Durham, North Carolina; and ¶Department of Psychology, **Department of Cardiovascular Sciences, EastCarolina University, East Carolina Heart Institute, Greenville, North Carolina

Background: The implantable cardioverter defibrillator (ICD) reduces mortality in patients at riskfor life-threatening arrhythmias via high energy shock. The Florida Shock Anxiety Scale (FSAS) wasdeveloped to measure ICD patient shock-related anxiety. Initial psychometric evaluation revealed goodreliability and validity. The purpose of this study was to examine the psychometrics of the FSAS in a largeUS sample of ICD patients.

Methods: Participants were recruited via e-mail and the survey was completed online. Ultimately, 443ICD patients (359 male and 421 White) completed the 10-item FSAS.

Results: Means for FSAS were comparable to previously published data (M = 15.18, SD = 6.5). Interitemreliability was good (Cronbach’s α = 0.89). The FSAS was negatively correlated with single-item measuresof emotional well-being (r = –0.378, P < 0.01), sense of security (r = –0.365, P < 0.01), perceived generalhealth (r = –0.185, P < 0.01), and quality of life (r = –0.216, P < .01), demonstrating discriminant validity.Convergent validity was supported through significant correlations with number of shocks (r = 0.464, P< 0.01) and reported disruptiveness of shock (r = 0.484, P < 0.01). Confirmatory factor analysis revealedthat a single (second-order) factor model (χ2 [34] = 75.34, P < 0.05, comparative fit index = 0.98, rootmean-square error of approximation = 0.05) had the best fit.

Conclusions: Shock anxiety as a construct can be measured in a reliable and valid method by the FSAS.These nationally representative data suggest that a single score for shock anxiety is an easy to use andappropriate method of assessment. (PACE 2012;00:1–8)

implantable cardioverter defibrillator, anxiety, psychometrics, assessment, psychosocial

BackgroundThe implantable cardioverter defibrillator

(ICD) is implanted in patients at risk for life-threatening arrhythmias.1 It has been shown tosignificantly reduce mortality and even improvequality of life (QoL).2,3 The ICD’s primaryfunction is to monitor the cardiac rhythm and

Disclosures: Dr. Sears: Medtronic, St. Jude Medical, BostonScientific, and Biotronik. Dr. Fischer: Boston Scientific,Impulse Dynamics, Medtronic, St. Jude Medical, and Spectra-netics. Julie Shea: Medtronic, Janssen Pharmaceuticals, andSpectranetics. This manuscript and the research describedtherein were supported by funding from Medtronic. No otherauthors have disclosures to report.

Address for reprints: Samuel F. Sears, Ph.D., 104 RawlBuilding, East Carolina University, Greenville, NC 27858. Fax:252 328-6283; e-mail: searss@ecu.edu

Received January 17, 2012; revised March 14, 2012; acceptedApril 16, 2012.

doi: 10.1111/j.1540-8159.2012.03455.x

deliver a high energy shock to the myocardiumwhen a tachyarrhythmia is detected.1 Regrettably,the shock is often experienced as painful anddistressing. Patients typically rate the shock as 6 of10 on a pain scale (with 10 being the most pain everexperienced).4 Patients with ICDs already havea greater prevalence of anxiety than the generalpopulation5 and the experience of a shock orknowledge of this device function may result ina significantly increased and distressing concernabout future shocks. Ultimately, shock anxietymay generalize to a clinically significant anxietydisorder or avoidance of behavior associated withshock. The Florida Shock Anxiety Scale (FSAS)was developed to be a sensitive quantitativemetric of ICD shock-related anxiety for use inelectrophysiology clinics and research.6

Previous Measurement of Shock Anxiety withthe FSAS

The original FSAS validation demonstratedthat this unique measure of shock anxiety had

C©2012, The Authors. Journal compilation C©2012 Wiley Periodicals, Inc.

PACE, Vol. 00 2012 1

FORD, ET AL.

Figure 1. Consort flow chart demonstrating e-mails sent, e-mails opened, and surveys completed.

adequate reliability and validity.6 Two factorsaccounting for 66% of the variance were foundin factor analysis. The first factor was comprisedof items assessing fear or anxiety related to theconsequences of shock and the second factor wasassociated with fear or anxiety about triggering ashock.

Since its initial evaluation, the FSAS has beenused as a shock anxiety measure internationallyin Canada,7 Australia,8 and China.9 The measurehas also been evaluated in a primarily femaleparticipant sample,10 given the predominatelymale participant sample of the initial evaluationin the United States.6 The FSAS is also sensitiveto ICD experience such as recall and psychosocialintervention. A study of the psychological impactof ICD lead recall found that participants withfractured Medtronic Fidelis leads had significantlygreater FSAS scores than those without fracturedleads (P = 0.0016).7 In addition, several trialsof psychosocial intervention have shown theFSAS to be sensitive to reductions in shockanxiety.8,11 On the other hand, samples examiningthe psychometric properties of the FSAS havebeen small and geographically limited. This studyexamined the psychometric properties as well asthe factor structure, reliability, and validity ofthe FSAS in a large, national US sample of ICDpatients

MethodsProcedure

The survey was administered via e-mail toparticipants from an electronic database of singleand dual-chamber ICD recipients who indicatedinterest in participating in research maintained byMedtronic, a medical device company (MedtronicInc., Minneapolis, MN, USA). There were noexclusion criteria. None of the participants haddevices which provide cardiac resynchronizationtherapy. Informed consent was obtained beforethe survey was made electronically available to

the participants. As part of the consent process,potential participants were informed that thepurpose of the study was to examine patientattitudes, beliefs, and experiences with their ICD,including ICD shock. Participation in the studywas voluntary and no participant incentives wereprovided. The study was approved by the EastCarolina University Institutional Review Board.

Following informed consent, patients wereasked to complete the 42-question survey. Alldata were self-report responses to these questions.Using information from the ICD patient database,4,447 e-mail surveys were distributed and 563survey links were accessed, with 443 surveyscompleted, providing a completion rate of thesurvey at 78.69%, if opened and delivered. Nocompensation was provided for participation. Onehundred twenty surveys were incomplete. A flowchart which visually presents the number of e-mails sent in each wave, the number of e-mailsopened, and the number of surveys completed canbe seen in Figure 1.

Measures

The final sample of 443 patients completedthe e-mailed 42-question psychosocial ques-tionnaire “Shock 2010.” The battery includedempirically supported scales such as the FSAS,demographic questions, and single-item indicesassessing QoL, emotional well-being, subjectiveperception of general health, sense of security,and ICD-shock life disruption. Single-item indicesof these domains were used in order to reduceparticipant burden of time to complete the survey.

Single-Item Indices

Participants were asked to answer single-itemquestions regarding their QoL, emotional well-being, subjective perception of general health,sense of security, and ICD-shock life disruption.Participants rated items on scales that offered fromthree to seven response choices, depending on the

2 2012 PACE, Vol. 00

FLORIDA SHOCK ANXIETY SCALE (FSAS) PSYCHOMETRICS

Table I.

Means and Standard Deviations of the Total Score of the Florida Shock Anxiety Scale

Source Purpose Location n M SD

Kuhl et al., 20066 Initial development United States 72 15.4 7.6Vazquez et al., 200810 Female patients United States & Australia 88 16.4 6.7Kuhl et al., 200911 Intervention evaluation United States 30 15.5 6.9Vazquez et al., 20107 Female-specific intervention evaluation United States 29 Baseline: 21.2 9.7Chair et al., 20119 Validation China 85 19.9 7.1Keren et al., 20118 Device recall Canada 416 15.5 6.9

question, ranging from an extreme negative to ex-treme positive response. A higher score indicatesan extremely positive response and a lower scoreindicates an extremely negative response, exceptin the case of shock-life disruption where a higherscore is extremely negative and a lower scoreextremely positive. Questions were designed tocapture patient opinion on perceptions of currenthealth status, as well as a subjective opinion ofhow the ICD has affected them in specific areas,including anxiety levels.

Florida Shock Anxiety Scale

The FSAS is intended to measure ICD shockanxiety. The measure was designed by an in-terdisciplinary team including electrophysiologyand clinical psychology. Items were based onclinical experience with ICD patients. The scaleitself consists of 10 items which respondents rateon a five-point Likert scale ranging from 1 (notat all) to 5 (all of the time). See Appendix forthe FSAS items. Table I presents published FSASmeans and standard deviations (SD). These meansrepresent total scores for the FSAS, with highervalues representing greater shock anxiety andlower values representing lower shock anxiety.The score of the FSAS is determined by summingthe items.

Analyses

The present investigation examined thepsychometric properties of the FSAS includingreliability, validity, and factor structure.

Analysis of Reliability and Validity

Reliability was examined using Cronbach’salpha. Convergent validity of the measure wasevaluated by correlating the FSAS total score withnumber of shocks and disruption of shock to dailylife using Pearson product-moment correlationcoefficients. Correlation coefficients were alsoused to examine discriminant validity of the FSAS

with sense of security, QoL, emotional well-being,and general health.

Analysis of Factor Structure

Confirmatory factor analysis (CFA) was con-ducted using the software package Mplus 5.21(Mplus Software, Inc., Los Angeles, CA, USA).12

Three hypothesized model specifications werecompared: (1) a single latent exogenous factor, (2)two correlated exogenous factors, and (3) a second-order factor model. Briefly, unlike exploratoryfactor analysis, in which every observed variable isallowed to load on every latent factor, CFA allowsthe researcher to either constrain a given parame-ter (i.e. factor loading or factor covariance) to equalsome prespecified value (often zero), or to leavethat parameter free to be estimated. Because CFAallows for the a priori specification of zero andnon-zero factor loadings and factor covariances,the confirmatory model reflects a hypothesizedfactor structure that can be empirically tested forits adequacy. In exploratory factor analysis, sinceevery variable is free to load on every factor, theinvestigator is not afforded comparable theory-testing power.

Accordingly, the hypothesized measurementmodels estimated in this study specified a singlelatent exogenous factor model, Shock Anxiety,against a model specifying two competing models.In terms of the specification of the one-factormodel, (1) the matrix of factor loadings wasspecified such that each scale item had a nonzeroloading on the single factor, and (2) all off-diagonal elements of the matrix of errors ofmeasurement were fixed to zero, given theexpectation of uncorrelated indicator disturbanceterms. In addition, in order to identify the model,the variance of the single latent construct was fixedto a value of 1.0.13

There is some controversy regarding theadequacy of the numerous goodness-of-fitindices which have been proposed.13–15 Mostresearchers advocate using multiple fit indices

PACE, Vol. 00 2012 3

FORD, ET AL.

when evaluating model fit. In this study, wefollowed the recommendations of Hu andBentler16 and evaluated model fit using the rootmean-square error of approximation (RMSEA),17

and the comparative fit index (CFI).18 For theCFI, values of approximately 0.95 or higherare indicative of an acceptable fit. Browne andCudeck19 have suggested that values of the RMSEAabove 0.10 indicate room for improvement, valuesbetween 0.05 and 0.09 reflect an adequate fit,and values below 0.05 reflect close fit. Given thekurtosis values associated with the measuredvariables, estimates obtained using normal-theorymaximum likelihood (ML) were compared toestimates obtained using the robust ML estimatoroption in Mplus, and the Satorra-Bentler20 scaledχ2 (obtained using EQS version 6.1).21

ResultsA total of 443 ICD patients from geographi-

cally diverse locations across the United Statescompleted the survey. Eighty-one percent of thestudy sample (n = 359) was male. Ninety-fivepercent (n = 421) were White, 1.8% (n = 8)were Black/African American/Caribbean Ameri-can, 0.9% (n = 4) were Hispanic/Latino, 1.1%(n = 5) were Asian, 0.7% (n = 3) listed “other” fortheir race, and 0.5% (n = 2) were either missingthat data or selected “don’t know/refused.” Theage distribution of participants was 0.9% (n =4) between age 21–29, 1.1% (n = 5) 30–39 y/o,3.6% (n = 16) 40–49 y/o, 4.1% (n = 18) 50–54y/o, 22.1% (n = 98) 55–64 y/o, and 68.2%(n = 302) 65 years of age or older. Of thetotal population, 49.7% (n = 220) had receivedan ICD shock. Shocks were received by 50.1%(n = 180) of male participants, and 47.6% (n =40) of female participants. The overall mean andstandard deviation of the total score for the FSASwas comparable to other samples (M = 15.18, SD= 6.5).

Reliability and Validity with OtherPsychosocial Variables

Alpha coefficients suggested good reliabilityof the total FSAS (Cronbach’s α = 0.89). Table IIpresents correlations used to evaluate convergentand discriminant validity of the FSAS. There was amedium negative correlation between FSAS scoreand emotional well-being (r = –0.378, n = 443, P< 0.01), with higher levels of anxiety associatedwith lower levels of emotional well-being. Therewas also a medium negative correlation betweenFSAS score and sense of security (P = –0.365, n= 443, P < 0.01), with higher levels of anxietyassociated with lower sense of security. There wassmall negative correlation between FSAS scoreand QoL (r = –0.216, n = 443, P < 0.01), with

Table II.

Pearson Product-Moment Correlations between FloridaShock Anxiety Scale Scores and Quality of Life Measures

Florida ShockAnxiety Scale

Self-report of emotional well-being −0.378*Self-report sense of security −0.365*Self-report of quality of life −0.216*Self-report of general health −0.185*Number of shocks received 0.464*ICD-shock life disruption 0.482*

*P < 0.01 (two-tailed).

higher levels of anxiety associated with lowerQoL. There was also a small negative correlationbetween FSAS score and general health (r =–0.185, n = 443, P < 0.01), with higher levelsof anxiety associated with lower levels of generalhealth. There was a medium positive correlationbetween FSAS score and the number of shocksreceived (r = 0.464, n = 443, P < 0.01), with higherlevels of anxiety associated with a greater numberof shocks received. For those participants whohad received a shock, the relationship betweenshock anxiety and the disruptiveness of the shockexperience was also investigated. There was amedium positive correlation between FSAS scoreand disruptiveness of shock (r = 0.482, n = 220,P < 0.01), with higher levels of anxiety associatedwith higher reported disruptiveness of the shockexperience.

Factors

The suitability of data was assessed priorto conducting factor analysis. The Kaiser-Meyer-Olkin value was 0.925 and Bartlett’s test ofsphericity reached statistical significance, thussupporting the factorability of the correlationmatrix. In order to determine if the two-factorexploratory factor analysis model found by Kuhland associates during the initial development ofthe FSAS6 could be replicated, an exploratorymodel specifying the two factors was initially es-timated. Estimation of the two-factor exploratorymodel with factors obliquely rotated using promaxrevealed a different pattern of loadings. Specif-ically, FSAS items 3, 6, and 10 loaded on onefactor whereas the remaining items loaded on thesecond factor. The fit of this model was quiteadequate as judged by the Bentler and Bonnett14

non-normed fit index (NNFI = 0.975). Next, weused CFA to fit the two-factor model we hadinitially estimated, since this provides a more

4 2012 PACE, Vol. 00

FLORIDA SHOCK ANXIETY SCALE (FSAS) PSYCHOMETRICS

Figure 2. Comparison of the second-order and single-factor models of the Florida Shock Anxiety Scale. The one-factor confirmatory model is on the left. The second-order confirmatory factor analysis of the factor structure is onthe right. All coefficients are standardized (*P > 0.05; **P > 0.01; ***P < 0.001).

stringent test of the two-factor hypothesis. The fitof the two-factor CFA model was quite good (χ2

(34) = 82.1, P < 0.05, CFI = 0.97, RMSEA = 0.06).The standardized factor loadings ranged from 0.45to 0.80 with the majority above 0.6, and all theloadings were highly significant (all P’s <0.05).The very substantial correlation between the twofactors (r = 0.83, P < 0.05), however, suggestedthat a second-order factor might underlie these twohighly correlated primary dimensions.

Accordingly, we respecified the model as asecond-order factor model to determine how thefit of a second-order model compared to a CFAmodel just specified by two correlated primarydimensions. In this second-order model, the twoprimary dimensions of Shock Anxiety, namelyConsequences of Shock and Triggering Devicesof Shock, were both specified to load on a singlesecond-order factor (see Fig. 2). When a second-

order factor model was fit to the data, the fit ofthis model was very acceptable (χ2 (34) = 75.34,P < 0.05, CFI = 0.98, RMSEA = 0.05), and yieldedthe lowest χ2 and the best fit of any of the modelsestimated. As seen in Figure 2, the standardizedloadings on the two primary dimensions ofConsequences of Device Shock and TriggeringDevice Shock on the underlying second-orderfactor were substantial and uniformly statisticallysignificant (0.99 and 0.96, respectively, both Pvalues <0.05).

Finally, to exclude the possibility that a one-factor model fits significantly better than either(1) the two-factor CFA model or (2) the second-order model, we estimated a one-factor modelin which all the variables loaded on one factor.When the hypothesized one-factor, 10-indicatormeasurement model was fit to the data, theresulting fit was acceptable (χ2 [35] = 109.2, P <

PACE, Vol. 00 2012 5

FORD, ET AL.

0.05, CFI = 0.94, RMSEA = 0.97). This estimatedmodel is presented in Figure 2 for comparison,although the fit of this model was significantlyworse than that of the second-order factor model.

The results of the Mplus CFA using MLanalysis suggested that the measurement structureof FSAS can be well-represented by a second-orderfactor structure. It should be noted that inspectionof the indicators of Shock Anxiety used in thisstudy revealed that the distributions of some of themeasured variables were skewed and leptokurtic.Although ML estimation is relatively robust toviolations of the assumption of multivariatenormality, a comparison of estimates obtainedusing both (1) normal-theory ML and (2) MLwith robust standard errors were compared. ML-robust does not require multivariate normalityof the measured indicator variables. Estimationwith ML-robust yielded estimates that werequite comparable in magnitude to those obtainedunder normal-theory ML. All factor loadings inthe hypothesized second-order model remainedhighly significant. The model estimated usingrobust ML yielded a very acceptable fit to the data(χ2 [34] = 72.01, P < 0.05, Normed Fit Index [NFI]= 0.95, NNFI = 0.95).

DiscussionThis study confirms that the FSAS is a reliable

and valid measurement of the construct of shockanxiety. The sample included in this study waslarger and more geographically representative ofICD patients living within the United States. Thevalidity of the FSAS was evaluated by examiningits correlation with several hypothetically similarand opposing variables. The FSAS was found tobe sensitive to the number of shocks experienced,with those experiencing the greatest number ofshocks having the highest FSAS shock anxietyscores. The more shock anxiety participantsreported on the FSAS, the greater the rated dis-ruptiveness of shocks to everyday life. However,FSAS scores were negatively associated withreported emotional well-being, general health,sense of security, and QoL.

The initial factor structure of the FSASwas also evaluated. Although the CFA employedidentified two subscales, these were heavily inter-correlated, and a second-order factor model bestfit the data. This study suggests that the first-orderfactors, shock trigger and shock consequences, canbe explained by their common association with ageneral second-order factor, called shock anxiety.Each of the initial factors retained some uniquevariance, but the magnitude of this variance wassmall and the substantial loadings on the generalsecond-order construct suggest that the initialfactors are indicators of a more abstract theme

of shock anxiety. Altogether, this indicates thatlittle information is gained from scoring the FSASsubscales and that the total score can be used inisolation for a measure that is easy to use andinterpret.

Limitations

Although the sample studied in the currentanalysis is much larger than any previous FSASsample, there are several limitations that shouldbe noted. Specifically, as approximately 21.31%of surveyed patients did not respond to thesurvey e-mail, we cannot exclude the possibilityof response bias. In addition, the survey requiredresponse via the Internet and as such excludedpatients without Internet access or a valid e-mail address. Although the number of patientsstudied was large, the sample was ethnicallyhomogenous and as such may not generalize toother ethnic groups. Finally, single-item measures(i.e. emotional well-being, sense of security,general health, and QoL) were used to reducethe patient burden associated with completingthe survey. Although these were helpful inreducing participant dropout, the psychometriccharacteristics of single-item indices may not beas robust as multi-item tools.

Future Directions

Further examination of the FSAS in moreethnically diverse samples will be of value asmore specific norms tables can be developedfor comparison of ICD patients’ shock anxietyto that of their peers. Additionally, the FSASshould be validated with standardized measures,especially well-established measures of anxiety todemonstrate that shock-specific anxiety is related,but distinct from general anxiety. Finally, itwould be interesting to ascertain whether theFSAS is associated with biological endpoints,such as cardiac morbidity and mortality. Futurestudies examining the FSAS in ethnically diversepopulations, with generalized anxiety measuresand with biological endpoints, will further definethe construct of shock anxiety and can be used toimprove the quality of care ICD patients receive.

Implications for Clinical Practice

Not every patient with an ICD will havesignificant shock anxiety. Shock anxiety maybe influenced by a variety of psychological,personality, and biomedical factors. The FSASrepresents a tool which can be used to identifythose patients with the greatest shock-specificanxiety and thus the greatest need for intervention.

The average rating for each question inthe initial validation6 and our sample was 1.5(between “one = not at all” and “two = rarely”)

6 2012 PACE, Vol. 00

FLORIDA SHOCK ANXIETY SCALE (FSAS) PSYCHOMETRICS

on the scale of one to five for each item. Thisindicates that on average, patients with ICDsdo not have significant shock anxiety. However,as shock anxiety total scale score increases, sodoes fear of decrement to emotional well-being,perception of general health, sense of security, andQoL. Therefore, there are patients with significantanxiety related to ICD shocks, indicating that theFSAS is a valuable screening tool. The total scoreaverage was 15.4 in the initial sample and 15.18in the present sample. Based on the results ofthis further psychometric evaluation of the FSAS,it is recommended that the total score of theFSAS be used to compare ICD patients with theirpeers on level of shock anxiety. It is important tounderstand that shock anxiety (while significantand distressing) is not equivalent to an anxietydisorder. The FSAS is device-specific and onlytargets anxiety related to ICD shocks. Althoughshock anxiety may predispose those patients likelyto develop an anxiety disorder or be a specificsymptom of a broader mental health problem,patients with significant shock anxiety do notnecessarily meet diagnostic criteria for a mentalhealth diagnosis. Thus, screening with the FSAScan be used in a fashion whereby patients withexcessive shock anxiety can be targeted for morein-depth education about the low probability forfuture shocks and for the development of a “shockplan.” If shock anxiety persists, more intensive

clinical intervention and consideration for referralto a mental health professional may be warranted.

ConclusionsPrevious evaluations of the psychometric

properties of the FSAS found it to be a valid andreliable measure of shock anxiety with two factors,described as fear of triggering shock and fear of theconsequences of shock. The current evaluation,using a larger and more geographically representa-tive group of ICD patients living within the UnitedStates, corroborated findings that the measure isreliable and valid. While factor structure wasreplicated, a second-order factor indicated thatthe total scale score may be more clinicallyuseful. Based on our findings, we recommendthat the FSAS be used to evaluate ICD patients’level of shock anxiety as this device-specificanxiety is associated with worse emotional well-being, perceived general health, QoL, and sense ofsecurity. Patients with particularly elevated FSASscores should be provided additional educationand support related to their device. Those whodo not respond to this enhanced education and/orthose with elevated scores on the FSAS andgeneralized anxiety measures may warrant referralto a mental health professional. Overall, the FSASrepresents a valid and reliable clinical tool whichcan be used to improve care of patients with ICDs.

References1. Moses HW, Mullins JC (eds.): A Practical Guide to Cardiac Pacing.

Philidelphia, PA, Lippincott Williams & Wilkins, 2007.2. Bardy GH, Lee KL, Mark DB, Poole JE, Packer DL, Boineau R,

Troutman C, et al. Amiodarone or an implanted cardioverterdefibrillator for congestive heart failure. N Engl J Med 2005;352:225–237.

3. Connolly SJ, Dorian P, Roberts RS, Gent M, Bailin S, Fain E, ThorpeK, et al. Comparison of β-blockers, amiodarone plus β-blockers,or sotalol for prevention of shocks from implantable cardioverterdefibrillators: The OPTIC study: A randomized trial. JAMA 2006;295:165–171.

4. Ahmad M, Bloomstein L, Roelke M, Bernstein AD, Parsonnet V.Patients’ attitudes towards implanted defibrillator shocks. PacingClin Electrophysiol 2000; 23:934–938.

5. Magyar-Russell G, Thombs BD, Cai JX, Baveja T, Kuhl EA, Singh PP,Barroso MMB. The prevalence of anxiety and depression in adultswith implantable cardioverter defibrillators: A systematic review. JPsychosom Res 2011; 71:223–231.

6. Kuhl EA, Dixit NK, Walker RL, Conti JB, Sears SF. Measurementof patient fears of implantable cardioverter defibrillator shock: Aninitial evaluation of the Florida Shock Anxiety Scale. Pacing ClinElectrophysiol 2006; 29:614–618.

7. Vazquez LD, Conti JB, Sears SF. Female-specific education,management, and lifestyle enhancement for implantable car-dioverter defibrillator patients: The FEMALE-ICD Study. PacingClin Electrophysiol 2010; 33:1131–1140.

8. Keren A, Sears SF, Nery P, Shaw J, Green MS, Lemery R, Gollob MH,et al. Psychological adjustment in ICD patients living with advisorfidelis leads. J Cardiovasc Electrophysiol 2011; 22:57–63.

9. Chair SY, Lee CK, Choi KC, Sears SF. Quality of life outcomesin Chinese patients with implantable cardioverter defibrillators.Pacing Clin Electrophysiol 2011; 34:858–867.

10. Vazquez LD, Kuhl EA, Shea JB, Kirkness A, Lemon J, Whalley D,Conti JB, et al. Age-specific differences in women with implantablecardioverter defibrillators: An International Multi-Center Study.Pacing Clin Electrophysiol 2008; 31:1528–1534.

11. Kuhl EA, Sears SF, Vazquez LD, Conti JB. Patient-assistedcomputerized education for recipients of implantable cardioverterdefibrillators: A randomized control trial of the PACER program. JCardiovasc Nurs 2009; 24:225–231.

12. Muthen LK, Muthen BO. Mplus 5.21 User’s Guide: The Comprehen-sive Modeling Program for Applied Resarchers. Los Angeles, CA,Mplus Software, Inc., 2007.

13. Hoelter JW. The analysis of covariance structures: Goodness-of-fitindexes. Sociol Method Res 1983; 11:325–344.

14. Bentler PM, Bonnett DG. Significance tests and goodness of fit in theanalysis of covariance structures. Psychol Bull 1980; 88:588–606.

15. Tucker LR, Lewis PM. A reliability coefficient for maximumlikelihood factor analysis. Psychometrika 1999; 38:1–10.

16. Hu L, Bentler PM. Cutoff criteria for fit indexes in covariancestructure analysis: Conventional criteria versus new alternatives.Struct Equ Modeling 1999; 6:1–55.

17. Steiger JH. Structural model evaluation and modification: An inter-val estimation approach. Multivar Behav Res 1990; 25:173–180.

18. Bentler PM. Comparative fit indexes in structural models. PsycholBull 1990; 107:238–246.

19. Browne MW, Cudeck R. Alternative ways of assessing model fit. In:Bollen KA & Long JS (eds.), Testing Structural Equation Models,Newbury Park, CA, Sage, 1993, 136–162.

20. Satorra A, Bentler PM. Scaling corrections for chi-square statisticsin covariance structure analysis. In: von Eye A, Clogg CC (eds.).Proceedings of the American Statistical Association. Alexandria,VA, American Statistical Association, 1988 pp. 308–313.

21. Bentler PM. EQS 6 Structural Equations Program Manual. Encino,CA, Multivatiate Software, 2006.

PACE, Vol. 00 2012 7

FORD, ET AL.

Appendix: Florida Shock Anxiety Scale (FSAS)

I am scared to exercise because it may increase my heart rate and cause my device to fire.1 2 3 4 5

Not at all Rarely Some of the time Most of the time All the timeI am afraid of being alone when the ICD fires and I need help.

1 2 3 4 5Not at all Rarely Some of the time Most of the time All the time

I do not get angry or upset because it may cause my ICD to fire.1 2 3 4 5

Not at all Rarely Some of the time Most of the time All the timeIt bothers me that I do not know when the ICD will fire.

1 2 3 4 5Not at all Rarely Some of the time Most of the time All the time

I worry about the ICD not firing sometime when it should.1 2 3 4 5

Not at all Rarely Some of the time Most of the time All the timeI am afraid to touch others for fear I’ll shock them if the ICD fires.

1 2 3 4 5Not at all Rarely Some of the time Most of the time All the time

I worry about the ICD firing and creating a scene.1 2 3 4 5

Not at all Rarely Some of the time Most of the time All the timeWhen I notice my heart beating rapidly, I worry that the ICD will fire.

1 2 3 4 5Not at all Rarely Some of the time Most of the time All the time

I have unwanted thoughts of my ICD firing.1 2 3 4 5

Not at all Rarely Some of the time Most of the time All the timeI do not engage in sexual activities because it may cause my ICD to fire.

1 2 3 4 5Not at all Rarely Some of the time Most of the time All the time

8 2012 PACE, Vol. 00