Effectiveness of Videos Improving Cancer PreventionKnowledge in People with Profound Hearing Loss

Philip Zazove & Helen E. Meador & Barbara D. Reed &

Ananda Sen & Daniel W. Gorenflo

Published online: 12 April 2012# Springer Science+Business Media, LLC 2012

Abstract Deaf persons have a poorer understanding of can-cer prevention, which is felt to be partly due to communicationbarriers. One hundred ninety-seven d/Deaf persons completeda survey and video on cancer prevention. Half viewed aspoken English program designed for hearing persons (controlgroup); the other half viewed an amended program that hadAmerican Sign Language, captions, and printed Englishoptions added (experimental group). Knowledge was mea-sured before and after the video, including 1 and 6 monthslater. Respondents were primarily Caucasian, had lowincomes, lost hearing at young ages, and had d/Deaf spouses.Although overall knowledge improved after viewing the vid-eo, the presence of culture-specific communications (Ameri-can Sign Language, captions) did not improve scorescompared to the control group, either immediately after theintervention or over time. Moreover, percentage correct on allpretest, and almost all post-test, questions was <50% for bothexperimental and control groups. For all subjects, regardlessof which group they were in, a hearing spouse (p<0.001) andmore healthcare information sources (p00.001) improvedknowledge, while African-Americans showed a trend tolesser improvement (p00.06). Using culture-specific lan-guage did not improve cancer prevention knowledge in thisd/Deaf population, and overall knowledge remained low.

More study is needed to determine the best way to increasecancer prevention knowledge in this population.

Introduction

Deaf and hard of hearing (D&HH) persons have a poorerunderstanding of health care, including preventive medicinerecommendations, compared to the general population [1–4]. Reasons for this are not well understood. Communica-tion barriers have been postulated as the major reason, butcultural and other factors may play a role [1, 5].

The D&HH population is a heterogeneous group; per-haps 5% have profound hearing losses and many are part ofthe Deaf community, a documented minority that identifythemselves as Deaf (with a capital D) versus deaf persons(lower case d) who also have profound hearing losses butidentify themselves as part of hearing society [1, 6]. Minor-ity populations have both lower knowledge about andunique attitudes toward health preventive recommendations[7–13]. This has also been shown in Deaf populations; [14]Deaf persons, in fact, are the non-English speaking minorityat greatest risk for physician–patient miscommunication[15].

Most healthcare information is transmitted by spoken orwritten language, and the written communication typicallyrequires a relatively high literacy level. Thus, writing isoften ineffective for minority populations; low literacy is aknown barrier for these groups [16–19], including both deafand Deaf persons, whose average reading level is at thefourth–sixth grade level [20, 21] (also personal communi-cation from Gallaudet Research Institute 2003; unpublishedinternal data). Specific programs for improving healthcareknowledge in low-literacy populations (including using lan-guage appropriate content) have been reported, with

P. Zazove (*) : B. D. Reed :A. Sen :D. W. GorenfloDepartment of Family Medicine,University of Michigan Health System,1018 Fuller Street,Ann Arbor, MI 48104-1213, USAe-mail: pzaz@med.umich.edu

H. E. MeadorPrivate Deaf Consultant,Ann Arbor, MI, USA

J Canc Educ (2012) 27:327–337DOI 10.1007/s13187-011-0292-1

variable effectiveness [22–26]. Language-concordance pa-tient–physician communication has been shown to be effec-tive for increasing appropriate use of preventive services forDeaf persons who use American Sign Language (ASL) [27,28].

Computer-based programs have been used to imparthealth information with some success [9, 29] includingminority populations [30]. Thus, adding a video of anASL interpreter, or closed captions, to a computer-basedprogram may be an effective means of transmitting infor-mation to both deaf and Deaf persons (often combined as d/Deaf). Preliminary work in this arena has been reportedelsewhere [29, 31] including using ASL videos to transmitknowledge about cancer [32]. The latter found significantimprovement in knowledge using videos; [32] however, thebaseline cancer knowledge scores were dramatically higherthan that found typical for d/Deaf persons elsewhere [2, 14,33, 34].

We decided to test such an approach in a communitybased group of d/Deaf persons by modifying an award-winning (multiple awards, including several at the Partner-ship for Networked Consumer Health Information Show-case and Games; http://odphp.osophs.dhhs.gov/confrnce/partnr98/Showcase/hovision.htm) cancer prevention videoproduced by the University of Michigan (Michigan Interac-tive Health Kiosk Demonstration Project). We took thevideo, put in a small window with a certified ASL interpret-er translating the information as well as adding appropriatereading level captions, then evaluated whether this modifiedvideo increased the cancer prevention knowledge of d/Deafpeople. This report summarizes our findings.

Methods

Video Development

Making an educational video with ASL interpreters hasspecific potential issues. For example, in a signed multi-response question, each response option is signed but mustbe retained in memory while subsequent response optionsare signed [29]. It is also known that it can be difficult toconvey English into ASL since ASL has no written formand standardization of English questions in ASL is difficult[29, 31, 35]. Thus, it is important to pilot test any videoshowing ASL translation from English, as well as do back-translation [29, 31, 35], before one can test whether thisapproach actually improves Deaf persons’ preventive inter-vention knowledge. Considering this, the cancer preventionvideo was amended in several ways. The content was slightlyupdated to reflect the then current recommendations of the USPreventive Services Task Force and a questionnaire wasadded to gather comprehensive baseline information. In

addition, 12 survey questions based on the video contentwere added so that participants would complete it before(baseline knowledge), immediately after (looking forchanges in cancer prevention recommendations knowl-edge), and at one and 6 months (knowledge retention) afterthey viewed the video. A certified ASL interpreter signingthe entire video program, including the questionnaire andsurvey, was incorporated as a box covering 1/5 of thescreen; captions were placed below the interpreter. Priorto implementation, the entire program was verified by twoDeaf-person-focus groups, to ensure the intelligibility ofthe signs and captions. The second group indicated thesewere of optimal intelligibility.

Recruitment

The study was approved by the University of MichiganInstitutional Review Board and individuals with profoundhearing losses were recruited in Michigan’s lower peninsula.This is the part of Michigan where the majority of the state’spopulation lives, including most of the D&HH population;d/Deaf persons tend to live in the larger cities (Detroit, Flint,Lansing, Grand Rapids, etc.) which are mainly in the lowerpeninsula, though some do live in smaller towns. Potentialsubjects were identified at d/Deaf clubs and organizations inthe target area, existing contacts the researchers had withMichigan’s D&HH population, and contacts provided by theState of Michigan’s Division on Deafness.

All potential subjects, ages 18 and over, were offered theopportunity to participate. Approximately 20% agreed (lowparticipation is common in this population) [36]. The vastmajority of the subjects completed the initial screen, i.e.,the Gallaudet Hearing Loss Scale, on paper at the respectived/Deaf club/organization. If a profound hearing loss wasdocumented, they completed the consent form and wereenrolled. The program itself was almost always completedin the organization’s conference room or a similar largeroom; thus, there were multiple people in the room doingthe study at the same time. Each subject was given an iBookcomputer to use when participating in the program andcompleted it at their own pace. An estimated 5% of subjectscompleted the program on an individual basis, i.e., not in thelarger room. These individuals were either deaf, i.e., not partof the Deaf Community, or were interested in participatingbut unable to come to their respective d/Deaf organizationduring the scheduled times. Enrollment included both deaf(communicate via English and are not part of the Deafcommunity) and Deaf persons (communicate in ASL andare part of the Deaf community).

A small subset of individuals who declined to participatein the study agreed to complete demographic information.This was used to provide data whether nonparticipants dif-fered significantly from our participating subjects.

328 J Canc Educ (2012) 27:327–337

Data Collection

The questionnaire and videos were administered via aniBook computer. The questionnaire was completed firstand inquired about demographics, hearing loss variables(e.g., age of onset), family hearing loss history, languagehistory (e.g., current preference), healthcare utilization his-tory, presence of health problems, sources of healthcareinformation, and miscellaneous items related to health orhearing loss (e.g., smoking history). Then, subjects completedthe pretest survey (on the same iBook computer) assessingtheir cancer prevention recommendations knowledge for theirgender and age group. Skip logic was used in the computerprogram such that subjects only saw questions on the ques-tionnaire and survey that were appropriate for their age andgender. Eight of 12 survey questions were asked of all partic-ipants; these composed the “common cancer scale.” Of theremaining four questions, three were for females (cervical andbreast cancer) and one was for males (prostate cancer).Gender-specific questions were asked only of participants ofthe appropriate gender. Thus, females answered 11 surveyquestions total (the “female cancer scale”) and males an-swered nine survey questions total (the “male cancer scale”).

Subjects were randomly assigned to the experimental orcontrol group using a random numbers table. Controls viewedthe original video program (information transmitted by vary-ing combinations of voice, text and graphics) via an iBookcomputer, while cases (experimental group) viewed the sameprogram plus the simultaneous appearance of an ASL inter-preter and captions in addition to the voice throughout theentire video. Thus, each subject in the control group couldonly view the text and/or graphics (the assumption was thatnone could understand the voice based on the severity of thehearing loss level required to be in the study) to get whatinformation they could. Subjects in the experimental cohortcould, at their discretion, watch the ASL interpreter and/orcaptions that were translating the voice, in addition to viewingthe standard text and graphics. In both cases, the video wasplayed from beginning to end, though subjects had the oppor-tunity to repeat any areas they wanted to view again.

Immediately following the program, subjects completedthe same survey (post-test survey) on the iBook computerthey had completed before the program, to assess changes intheir understanding of age-specific cancer-preventionrecommendations.

The length of time to complete the consent process was5 min for most subjects, although a few (5–10%) took closerto 15 min due to their request to have the entire consent forminterpreted into sign language. The time to complete theprogram itself—including the questionnaire, video, and bothsurveys—ranged from 35 to 150 min; the mean was 50 min.During follow-up visits at 1 and 6 months, subjects againcompleted the survey. The mean time to complete those

surveys was 10 min. All participants received $20; $10 afterthe initial evaluation and $5 after each follow-up visit. Thisreport focuses on changes in participants’ knowledge ofcancer-screening recommendations as indicated by scoreson the pre-, post-, and 1 and 6 month follow-up surveys.

Analyses

Analyses were performed using SPSS 17.0 for the Mac (SPSSInc, Chicago, IL, USA). All continuous variable distributionswere tested for normality using the Shapiro–Wilk statistic. Chi-square tests for categorical data and t tests (or Wilcoxon–Mann–Whitney rank sum test if data were not normallydistributed) for continuous data were carried out to investi-gate any possible differences between the intervention andthe control group. Improvement in cancer knowledge dem-onstration was assessed for each cancer-related question bymeans of McNemar’s test applied to the proportion of sub-jects answering it correctly at pre- and post-test.

Correlational analyses were tested with Pearson’s coeffi-cient for normally distributed and Spearman’s rho for non-normally distributed data. Multivariate analyses were per-formed as appropriate to assess possible confounders, suchas age, education, age at loss of hearing, and use of ASL.Differences between the pretest and follow-up scores werecompared using paired-sample McNemar’s tests.

The percentage of correct responses at all four time points(pre, post, 1 and 6 months follow-up) was calculated based onthe number of correct answers divided by the total number ofquestions asked of each participant. Two separate analyses wereperformed using the percentage of correct responses. First apre–post difference in percent correct was computed and theresulting change score was compared across several demo-graphic characteristics separately either by means of t tests(for characteristics with two subgroups) or one-way analysisof variance (for characteristics with more than two sub-groups). Subsequently, a mixed model regression analysiswas performed with percent correct responses at all fourtime points as the outcome; age, gender, ethnicity, numberof information sources, having a hearing spouse, treatmentgroup assignment as between-subject predictors, time andtime–group interaction as within subject factors. The clus-tering within subjects was accounted for by a random sub-ject intercept. Model diagnostics was carried out by meansof residual plot.

Results

Sample Description

Two hundred and twenty-three participants entered thestudy (consented and completed the baseline survey); 195

J Canc Educ (2012) 27:327–337 329

completed both the initial pre- and post-test, 105 completedthe 1 month follow-up survey, and 39 completed the 6 monthfollow-up survey. There was no statistically significant de-mographic difference between those in the control and ex-perimental groups; the demographic information for the 195subjects who completed both initial pre- and post-tests issummarized in Table 1. Participants were primarily Cauca-sian, members of the Deaf community, lost their hearing at ayoung age, had a deaf spouse, and had low householdincomes. When communicating with d/Deaf persons,67.9% used ASL, but when communicating with hearingpeople, English was often used (41.3% spoken, 17.3%signed English—a word-for-word translation of English intosign, using English grammar which is very different fromASL—and 33.2% writing). When communicating with

physicians or nurses, 42.9% of subjects wrote in Englishand 30.3% used sign-language interpreters. Spoken Englishwas the most common language used at home as a child(58.3%) and with teachers at school (56.1%). No statisticallysignificant difference existed between nonparticipating andparticipating d/Deaf persons in age, gender, marital status,ethnicity, college degree, or use of ASL at home.

Cancer Screening Survey Results

Overall Performance

When analyzed separately by treatment and control group,the presence of the ASL interpreter and captions in theintervention group did not improve long-term knowledge

Table 1 Demographic charac-teristics of participants (p>0.05unless otherwise indicated)

*p00.08aNineteen participants did notgive household income databInterquartile range

Total (N0195) Intervention (N097) Controls (N098)

Age in years (mean±SD) 55.1±16.6 54.5±15.7 55.8±17.6

Income (mean $±SD)a 20,834±21,230 19,761±19,116 21,920±23,429

Income (median $)b 15,000 (28,825) 16,000 (28,300) 14,500 (39,488)

% (N) % (N) % (N)

Female gender 117 (60%) 59.8 (58) 60.2 (59)

Ethnic background

Caucasian 76.5 (150) 76.3 (74) 76.5 (75)

African-American 17.3 (34) 16.5 (16) 18.4 (18)

Hispanic 0.5 (1) 1.0 (1) 0 (0)

Native American 3.1 (6) 4.1 (4) 2.0 (2)

Asian American 0.5 (1) 1.0 (1) 0 (0)

Other 2.0 (4) 1.0 (1) 3.0 (3)

Employed 44.4 (87) 48.5 (47) 39.8 (39)

High school degree or greater 81.0 (158) 82.5 (80) 79.6 (78)

College degree 21.4 (42) 23.7 (23) 19.4 (19)

Married 58.2 (114) 60.8 (59) 55.1 (54)

Spouse

Deaf 44.4 (87) 48.5 (47) 39.8 (39)

Hard of hearing 5.1 (10) 5.2 (5) 5.1 (5)

Hearing 8.7 (17) 7.2 (7) 10.2 (10)

No spouse 41.8 (82) 39.2 (38) 44.9 (44)

Member Deaf community* 79.6 (156) 74.2 (72) 84.7 (83)

Age lost hearing

<1 Year 45.4 (89) 47.4 (46) 43.9 (43)

1–4 36.7 (72) 36.1 (35) 36.7 (36)

5–19 12.2 (24) 13.4 (13) 11.2 (11)

≥20 5.6 (11) 3.1 (3) 8.2 (8)

Reason lost hearing

Meningitis 11.2 (22) 9.3 (9) 13.3 (13)

Rubella 12.8 (25) 13.4 (13) 12.2 (12)

Birth complications 12.2 (24) 12.4 (12) 12.2 (12)

Heredity 11.2 (22) 12.4 (12) 10.2 (10)

Other 52.6 (103) 52.6 (51) 52.01 (51)

330 J Canc Educ (2012) 27:327–337

outcomes. Of interest, each group separately had significantimprovement at the initial post-test (p≤0.001) from thebaseline pretest answers, but there was no difference be-tween them. Moreover, although scores trended higher at the1 and 6-month follow-ups compared to the baseline, inneither group was this statistically significant. The percent-age correct over time—i.e., at the pretest, the post-test, andat the 1- and 6-month follow up evaluations—are shown inTable 2. There were no demographic differences betweenthe experimental and control groups at this follow-up surveytime periods. Of interest, although neither the experimentalnor the control group individually showed a sustained im-proved knowledge over time, as a group (i.e., experimentaland control subjects combined), there was improvement notonly on the post-test (p≤0.001), but also at 1 month (p≤0.01) and 6 months (p≤0.05).

Table 3 shows the extent of improvement on the initialpost-test for each cancer-screening question separately.Responses to the age-directed questions were evaluated byage subgroups. The percentage correct for each cancerscreening pretest question was less than 50% and on thepost-test, only two questions were answered correctly over50% of the time There is a statistically significant (p<0.05)improvement at the initial post-test for eight of the 12questions. However, the absolute size of the increase incorrect responses was small, and ranged from 8% to 18%across these eight questions.

Demographic Variables

The degree of improvement among various demographiccategories was assessed (Table 4). Regardless of whetherthe subject was in the experimental or control group, subjectimprovement in scores was greater among those who wereCaucasian versus other ethnic groups (p00.012), were mar-ried (p00.01), and used computers (p00.01). There was nodifference in levels of improvement related to gender, in-come, and level of education.

Hearing Loss Variables

Regardless of whether the subject was in the experimental orcontrol group, participants with a hard of hearing or ahearing spouse were more likely to improve compared tothose with a d/Deaf spouse or no spouse (p00.008), whereasthose whose communication with their physician was ratedas poor tended to be more likely to improve (p00.06). Noassociation was found between improvement on the post-test and participants’ age of onset of hearing loss, being partof the Deaf community, hearing status of the father ormother, or etiology of hearing loss.

Language Use

The type of language used in communication with physi-cians (Spoken English, ASL, Signed English, or SignedContact Language [SCL]) was not associated with improve-ment between the pre- and post-test. Those who used SCL athome were more likely to improve than if they did not useSCL (p00.03); there was no association of greater improve-ment with all other languages used at home (ASL, SpokenEnglish, Signed English, or writing). There was no associ-ation of language used with hearing persons with improve-ment in test scores. When communicating with other d/Deafpeople, however, subjects who used SCL were more likelyto improve (13.2 vs. 6.8, p00.03).

Healthcare Status, Knowledge, and Utilization

Eighty-nine percent of women had had a Pap smear in thepast and 55% of men had had a prostate examination; thesesubjects had no differences in improvement of scores com-pared to those who had not had the preventive interventions.Only 20.4% had smoked more than 100 cigarettes ever andonly 5.6% were current smokers (compared to the average∼25% smoking rate for the state of Michigan at the time[37]). Moreover, 86% of respondents reported that smoking

Table 2 Longitudinal change in the pattern of response (percent correct)

Group Pretest (n0195) Post-test (N0195) 1 Month follow-up (N0105) 6 Month follow-up (N039)% (N) % (N) % (N) % (N)

Overall 22.8 (17.8) 31.5 (21.9)* 30.9 (19.9)** 29.8 (21.5)***

Treatment group 23.4 (18.9) 33.4 (21.4)* 30.3 (20.5) 30.5 (18.6)

Control group 22.1 (16.7) 29.5 (22.4)* 31.8 (19.3) 29.1 (24.8)

Table entries are mean (SD) of percentage of correct responses. P values are based on a mixed model analysis of percentage of correct responses onthe pretest, post-test, 1- and 6-month follow up, with age, gender, ethnicity, number of information sources, having a hearing spouse, and treatmentgroup as between-subject predictors; time, and time×group interaction as within-subject predictors* p≤0.001 compared to pretest scores, ** p≤0.01 compared to pretest scores, *** p<0.05 compared to pretest scores

J Canc Educ (2012) 27:327–337 331

is bad for one’s health; smaller percentages indicated itcaused lung disease (68.3%), cancer in general (65.3%),premature death (37.7%), dirty teeth (35.3%), or heartattacks (31.7%). There was no association between im-provement in post-test scores and whether the participanthad ever smoked at least 100 cigarettes, was a currentsmoker, had agreed with the statement that smoking is badfor one’s health, or his/her level of comfort with his/herphysician during the last visit.

Healthcare Information Source

Participants reported getting their healthcare informationfrom physicians (62.0%), family (38.5%), books (36.2%),friends (29.9%), the Internet (27.1%), TV (21.7%), news-papers (20.4%), and nurses (19.0%). The average number ofsources of information selected was 2.7±1.9 (median 2,mode 1), with 42% of participants naming only one sourcecategory. As the number of information sources increased,

post-test scores increased compared to the pre-test scores(p00.001). Respondents showed more improvement on thepost-test when they had obtained their healthcare informa-tion from nurses (15.2% improvement for those whosehealthcare information came from nurses, compared to7.2% improvement for those not getting healthcare informa-tion from nurses, p<0.01), friends (12.5% versus 7.0, p00.03), or the Internet (13.2% versus 7.0, p00.02). Therewere no differences in improvement on the test for thosewho did or did not obtain their healthcare information frombooks, doctors, newspapers, Deaf clubs, TV, family mem-bers, or other sources of information.

Multivariate Analysis

A longitudinal analysis of pattern of change in the overallpercent of correct responses was performed using a linearmixed regression model. The percentage correct at the pre-test, the post-test, and at the 1- and 6-month follow up

Table 3 Cancer knowledge demonstrated at pre- and post- test (proportion of subjects answered correctly)

N Pretest Post-test P valuea

% %Correct Correct

What should be the Pap smear frequency at their age?b 118 24.6 33.9 0.052

Age >50 years and older 70 32.9 44.3 0.096

Age 18–49 49 12.2 18.4 0.508

What should be the mammogram frequency at their age?b 118 37.3 50.0 0.008

Age <40 23 17.4 34.8 0.219

Age 40–49 26 23.1 38.5 0.219

Age 50 and older 70 48.6 58.6 0.143

What is the goal of a mammogram?b 118 27.1 25.4 0.815

What should be the prostate exam frequency at their age?c 78 28.2 30.8 0.791

Age <40 18 16.7 11.1 1.000

Age 40–49 14 0 7.1 –

Age 50 and older 47 42.6 46.8 0.754

What are the signs of skin cancer? 197 12.7 20.3 0.008

Which types of cancer can smoking cause? 197 15.7 34.0 <0.001

What are the current exercise duration recommendations? 196 26.5 36.7 0.003

What should be the fecal occult blood test frequency for their age? 195 20.5 31.8 <0.001

Age <50 years 80 8.8% 11.3 0.687

Age 50 and older 115 28.7 46.1 0.001

What should be the sigmoidoscopy frequency at their age? 191 13.6 14.7 0.832

Age <50 years 79 12.7 16.5 0.549

Age 50 and older 112 14.3 13.4 1.000

What types of cancer does a digital rectal exam detect? 192 23.4 31.8 0.005

What is the greatest cause of cancer in America? 193 40.9 51.8 0.003

What is the greatest cause of heart attacks in America? 194 15.5 27.3 <0.001

aP value computed for the pre-post comparison using McNemar’s testb Asked only of femalesc Asked only of males

332 J Canc Educ (2012) 27:327–337

evaluations are shown in Table 2 both overall and by theintervention arm. When analyzed separately by treatmentand control groups, although both groups had significantimprovement at the initial post-test (p≤0.001), neitherexhibited a significant difference from pretest at the 1 and6 month follow-up. This may be due to the significantdropouts at the later time periods, which reduced the powerto detect differences. There was no significant differenceamong the intervention arms; in neither was the time–groupinteraction term significant, indicating lack of any apprecia-ble difference in the change patterns between the two arms.As a group, subjects did improve on the post-test (p≤0.001),and maintained that improvement at 1 month (p≤0.01) and6 months (p≤0.05).

Among the between-subjects predictors used in the linearmixed model, only having a hearing spouse (p<0.001), andhaving a greater number of sources of healthcare informa-tion (p00.001) were significantly positively associated withoverall increased percentage of correct responses (Table 5).African-Americans had a lower percentage of correctresponses (p00.06). Not associated with test scores wereage, gender, marital status, use of computers, physiciancommunication, and language use (ASL, English, SignedEnglish, or SCL) in various situations (e.g., with physician,at home).

Discussion

Despite the intuitive concept that adding an ASL interpreterand low-literacy captions on an existing English-speakinginstructional video would improve d/Deaf persons’ under-standing of current cancer-preventive recommendations, wefound no significant difference of adding these communica-tion enhancers. In other words, subjects who saw the videothat included an ASL interpreter and low-literacy captionsexplaining the concepts (experimental group) did no betterthan those who only saw the graphics and the voiceover,

Table 4 Pre–post difference in percentage of correct responses

Percentage point improvement(mean±SD; N0196)

P value

Demographic variables

Age

≥50 Years 7.9±18.2 0.507<50 Years 9.5±15.3

Gender

Male 7.0±17.4 0.310Female 9.6±16.8

Ethnic background

Caucasian (N0150) 10.3±16.5 0.012African-American(N034)

0.9±16.1

All others (N013) 8.9±21.6

Employment status

Employed 9.8±17.5 0.363Not employed 7.6±16.7

College degree

Has degree 11.4±17.1 0.226No degree 7.8±17.0

Marital status

Married 11.2±16.5 0.010Not married 4.9±17.2

Income

≤$25,000 8.6±16.1 0.895>$25,000 9.0±20.2

Smoking ever

Yes 10.9±20.1 0.332No 8.0±16.2

Smoking is bad

Yes 9.0±17.7 0.298No 5.4±12.2

Computer use ever

Used computers 10.3±18.1 0.019No computer use 4.8±13.9

Computer use now

Uses computers 9.4±18.2 0.213Does not use computers 6.4±13.7

Hearing loss variables

Spouse hearing status

Deaf 9.2+16.7 0.008Hard of hearing 17.4±15.4

Hearing 17.9±13.7

No spouse 4.9±17.2

Member Deaf community

Yes 8.5±17.6 0.944No 8.7±15.0

Communication with doctor

Very good 6.0±15.3 0.060Good 6.2±17.6

Slightly good 13.9±16.3

Not good 14.3±20.3

Poor 14.6±13.8

Table 4 (continued)

Percentage point improvement(mean±SD; N0196)

P value

Age lost hearing

≤1 Year 8.9±16.3 0.734>1 Year 8.1±18.0

Reason lost hearing

Meningitis 3.5±17.5 0.130Family history 16.4±16.8

Rubella 8.1±16.3

Birth complications 9.9±19.9

All others 7.8±16.2

J Canc Educ (2012) 27:327–337 333

which they didn’t hear (control group). This was quiteunexpected. The experimental subjects clearly were en-thused about the presence of the interpreter and captionson the video, and commented to us on multiple occasionsthat these additions helped them understand the presenta-tion. The finding that there was no difference between thegroups underscores the complexity of providing healthcareinformation for the D&HH population, and for Deaf personsin particular. The latter are clearly unique as a result of theirlanguage use and relative isolation from society at large.

We did find an increased long-term improvement inknowledge of cancer-related preventive measures by ouroverall subject population as well as a trend in this direction(though not statistically significant) for subjects in each ofthe control and experimental groups. Others have found asimilar positive impact using American Sign Language toprovide health information; these have used different meth-ods of conveying the information such as videos, Power-point lectures, and one-on-one discussions [28, 31–34].However, only one other study that we are aware of had arandomized format with a control group. That study had anunusually high level of healthcare knowledge at baseline[32], with responses that were over 80% and even over 90%correct to most survey questions, much higher than thatfound in other studies of d/Deaf populations, regardless ofthe area of healthcare being studied [2, 4, 14, 27, 35]. Theauthors of that study acknowledge that their population maynot have been typical of d/Deaf populations in general [32].Our demographics and knowledge levels are consistent withthose of other studies.

Of interest, in one of the other studies, the authors notedthat their subjects said that there was more information thanthe subjects could absorb at one time. It was suggested thatwatching the video more than once would allow the D&HHpersons to gain a greater improvement in knowledge [34].That might have been the case with our subjects too, thoughwe did not specifically ask about this, and might partlyexplain why there was no difference between our controland experimental groups.

The greater improvement among non-African-Americansand among those with multiple sources of healthcare infor-mation was not surprising. The poorer knowledge base ofAfrican-Americans has been documented in other studies[9], and having multiple sources of information suggests abigger “base” upon which to add newly acquired healthinformation.

The finding of improved scores in persons whose spousewas hearing was interesting. It might be explained by thefact that healthcare information in our society is often com-municated via newspapers and television. Since most Deafpersons have poor English literacy (many deaf persons alsohave this), with an average reading level being at the sixth-grade level [16, 17, 20], a hearing spouse could provideaccess to the world of English-based information. In otherwords, the hearing member of the couple, who easilyreceives the information, transmits this to the d/Deaf spouse.This could result in subjects with hearing spouses beingaccustomed to processing this type of information and thusbetter able to assimilate the information we presented.

We previously reported that persons who lost their hear-ing due to meningitis were less likely to use computers andhypothesized that this may be due to brain injury occurringat a relatively older age compared to other causes of pro-found hearing loss [38]. In this study, meningitis was notstatistically associated with lesser improvement. We shouldnote, however, that those who lost their hearing due tomeningitis did have smaller improvements from the pretestto the post-test scores, though this was not statisticallysignificant—possibly because our population size was notlarge enough to detect such a difference. It remains plausiblethat people who lose their hearing due to meningitis are arelatively high-risk group of d/Deaf persons for poorerhealthcare knowledge as well as other intellectual pursuits(e.g., computer use); further studies are needed to evaluatethis.

Despite finding no benefit of an ASL interpreter andcaptions on knowledge retention, we do not believe thiseliminates communication barriers as the reason for theoverall poor health knowledge (even post-test scores werelow, with almost all questions having less than 50% correctanswers) and altered physician utilization habits of thispopulation. The video used was not created specifically forD&HH people; it was an amended version of a video creat-ed for English speaking, hearing people. The graphics andtext in the original video for hearing people may havecompeted for attention with the ASL interpreter and cap-tions added for D&HH people. Also, the graphics may haveconfused the message. For example, one African-Americansubject asked us if “those girls” were all right. When ques-tioned further, the subject clarified, “those African-American girls in the picture.” The video graphics hadincluded pictures of African-Americans as well as

Table 5 Significant between-subject predictors of post-testscores

B Standard error P value

Has a hearing spouse vs. other (including nonmarried) 0.213 0.039 <0.001

Non-African Americans vs. African-Americans 0.055 0.029 0.06

Increasing number of sources for healthcare information 0.019 0.006 0.001

334 J Canc Educ (2012) 27:327–337

Caucasians, but were not meant to illustrate that one minor-ity group was particularly prone to a specific illness. Otherfactors could be involved as well in the lack of comparativeimprovement in the experimental group, including the pos-sibility mentioned earlier that there was too much informa-tion for the subjects to absorb at one sitting. Moreover, somesubjects were so impressed that someone had developed avideo with an interpreter and captions that they seemedmore focused on evaluating the interpreter’s performancethan on the information being communicated.

The lesser health knowledge of this population could alsobe due to cultural and attitudinal beliefs [2, 3]. Minoritypopulations differ significantly from society at large withrespect to having unique beliefs about health and health care[7, 8, 30]. Deaf persons, a large percentage of our studypopulation, clearly have their own cultural attitudes abouthealth care [36]. Hearing physicians may deliver news inways that antagonize Deaf patients, leaving the patient feel-ing that the physician is withholding information and un-trusting of the physician [1]. The relative impact of thesebeliefs and attitudes compared with communication barriersremains to be determined.

Finally, the uniqueness of D&HH persons overall, asverified in other reports [1, 5, 6, 39], was confirmed in thisstudy, such as the low smoking prevalence and low familyincome. These are likely the result of linguistic subtletiesand cultural attitudes as well as isolation from the massmedia and lend additional support to the concept that edu-cational and healthcare interventions must be linguisticallyand culturally appropriate if they are to be effective. Perhapsvideos that are created specifically for d/Deaf people, usinga combination of ASL signs and features (including roleplay), linguistically appropriate captions, a different way ofconveying the information, and/or appropriate graphicswould be more effective than the video used in this study.

There are several limitations to our study. First, ourparticipants were all from one state and from relativelyurban areas. Furthermore, our response rate was only 20%.Though this low response rate is typical of studies of Deafpopulations [36], our findings may not apply to all D&HHpersons in Michigan or those in other settings (though oursmall sample of nonrespondents suggest this is not the case).Second, our population was predominantly Caucasian; indi-viduals from other ethnic groups may have differentresponses to this video program. Third, interactive videoshave inherent barriers, such as the inability to present ques-tions and answers simultaneously (as one can with writtenEnglish), difficulty translating some English words intoASL signs, and longer times to complete surveys [29, 31,35]. The effect of these on our outcomes is unclear. More-over, participants in this study may be individuals who weremore comfortable with or more amenable to using com-puters, compared with those who declined participation.

In summary, the provision of an ASL interpreter andcaptions to a computer-based video cancer-preventioneducation program developed for the general publicdid not increase the understanding of cancer preventionrecommendations for a group of mainly Deaf personscompared to a control group that did not have theseculturally appropriate adaptations. Reasons for this areunclear and further studies are needed to better deter-mine how to best provide healthcare information for thispopulation.

Acknowledgment The research was supported by NIH grant#R25DC004604.

Appendix Cancer Screening Survey(Pretest and Post-test)

1. How often should you get a Pap smear at your age?

– I don’t know– Every year– Every 1–3 years, if sexually active, depending on

doctor’s evaluation– Every 1–2 years, depending on doctor’s evaluation– All of the above– None of the above

2. How often should you get a mammogram at your age?

– I don’t know– Not recommended at my age– Every year– Every 3 years– Every 1–2 years, depending on doctor’s evaluation– All of the above– None of the above

3. What is the goal of a mammogram?

– I don’t know– To find breast cancer after the woman could feel

the cancer– To find breast cancer after the woman had breast

problems– To find breast cancer when it is very small– To find breast cancer when it is very big– All of the above– None of the above

4. How often should you get a prostate exam at your age?

– I don’t know– Not recommended at my age– Depending on how much risk for me, my doctor

will recommend– Every year

J Canc Educ (2012) 27:327–337 335

– Every 1–2 years, depending on doctor’s evaluation– All of the above– None of the above

5. What type of cancer testing involves the lettersA (Asymmetry)B (Border)C (Color)D (Diameter)

– I don’t know– Lung– Stomach– Skin– Colon– All of the above– None of the above

6. Which of the following problems can smoking cause?

– I don’t know– Neck cancer– Colon cancer– Lung cancer– Bladder cancer– All of the above– None of the above

7. How many minutes a day, most days of the week,should you exercise?

– I don’t know– Not less than five– Not less than fifteen– Not less than thirty– Not less than forty-five– All of the above– None of the above

8. How often should you get a fecal occult blood test (testbowel movement sample for blood) for colon cancer atyour age?

– I don’t know– Not recommended at my age– Every year– Every 2 years– Every 5 years– All of the above– None of the above

9. How often should you get a sigmoidoscopy (doctorlooks inside rectum with tube) at your age?

– I don’t know– Not recommended at my age– Every year

– Every 3 years– Every 5 years– All of the above– None of the above

10. Digital Rectal Exam (doctor checks rectum with fin-gers) can find which of the following kinds of cancer?

– I don’t know– Colon– Skin– Ovary– Testicles– All of the above– None of the above

11. What is the biggest cause of cancer in America?

– I don’t know– Alcohol and drug abuse– Smoking– Too many X-rays– Not enough exercise– All of the above– None of the above

12. What is the biggest cause of heart attacks in America?

– I don’t know– Alcohol and drug abuse– Smoking– Dangerous chemicals– Not enough exercise– All of the above– None of the above

References

1. Barnett S (1999) Clinical and cultural issues in caring for deafpeople. Fam Med 31(1):17–22

2. Tamaskar P, Malia T, Stern C, Gorenflo D, Meador H, Zazove P(2000) Preventive attitudes and beliefs of deaf and hard-of-hearingindividuals. Arch Fam Med 9(6):518–525, discussion 526

3. Woodroffe T, Gorenflo DW, Meador HE, Zazove P (1998) Knowl-edge and attitudes about AIDS among deaf and hard of hearingpersons. AIDS Care 10(3):377–386

4. Margellos-Anast H, Estarziau M, Kaufman G (2006) Cardiovas-cular disease knowledge among culturally Deaf patients in Chi-cago. Prev Med 42(3):235–239

5. Zazove P, Niemann LC, Gorenflo DW, Carmack C, Mehr D,Coyne JC, Antonucci T (1993) The health status and health careutilization of deaf and hard-of-hearing persons. Arch Fam Med 2(7):745–752

6. Dolnick E (1993) Deafness as culture. Atlantic Monthly 272:37–53

7. Bailey EJ, Erwin DO, Belin P (2000) Using cultural beliefs andpatterns to improve mammography utilization among African-

336 J Canc Educ (2012) 27:327–337

American women: the Witness Project. J Natl Med Assoc 92(3):136–142

8. Becker G, Beyene Y, Newsom EM, Rodgers DV (1998) Knowl-edge and care of chronic illness in three ethnic minority groups.Fam Med 30(3):173–178

9. Green MJ, Peterson SK, Baker MW, Harper GR, Friedman LC,Rubinstein WS, Mauger DT (2004) Effect of a computer-baseddecision aid on knowledge, perceptions, and intentions aboutgenetic testing for breast cancer susceptibility: a randomized con-trolled trial. Jama 292(4):442–452

10. Guidry JJ, Matthews-Juarez P, Copeland VA (2003) Barriers tobreast cancer control for African-American women: the interde-pendence of culture and psychosocial issues. Cancer 97(1Suppl):318–323

11. Paskett ED, Tatum C, Rushing J, Michielutte R, Bell R, Foley KL,Bittoni M, Dickinson S (2004) Racial differences in knowledge,attitudes, and cancer screening practices among a triracial ruralpopulation. Cancer 101(11):2650–2659

12. Rajaram SS, Rashidi A (1998) Minority women and breast cancerscreening: the role of cultural explanatory models. Prev Med 27(5Pt 1):757–764

13. Watts T, Merrell J, Murphy F, Williams A (2004) Breast healthinformation needs of women from minority ethnic groups. J AdvNurs 47(5):526–535

14. Zazove P, Meador HE, Reed BD, Sen A, Gorenflo DW (2009)Cancer prevention knowledge of people with profound hearingloss. J Gen Intern Med 24(3):320–326

15. McEwen E, Anton-Culver H (1988) The medical communicationof deaf patients. J Fam Pract 26(3):289–291

16. Dolan NC, Ferreira MR, Davis TC, Fitzgibbon ML, Rademaker A,Liu D, Schmitt BP, Gorby N, Wolf M, Bennett CL (2004) Colo-rectal cancer screening knowledge, attitudes, and beliefs amongveterans: does literacy make a difference? J Clin Oncol 22(13):2617–2622

17. Lindau ST, Tomori C, Lyons T, Langseth L, Bennett CL, Garcia P(2002) The association of health literacy with cervical cancerprevention knowledge and health behaviors in a multiethnic cohortof women. Am J Obstet Gynecol 186(5):938–943

18. Hunter JL (2005) Cervical cancer educational pamphlets: do theymiss the mark for Mexican immigrant women’s needs? CancerControl 12(Suppl 2):42–50

19. Clayman ML, Manganello JA, Viswanath K, Hesse BW, Arora NK(2010) Providing health messages to Hispanics/Latinos: under-standing the importance of language, trust in health informationsources, and media use. J Health Commun 15(Suppl 3):252–263

20. Drasgow E (1993) Bilingual bicultural deaf education: an over-view. Sign Lang Stud 80:243–266

21. Singleton JL, Morgan D, DiGello E, Wiles J, Rivers R (2004)Vocabulary use by low, moderate, and high ASL-proficient writerscompared to hearing ESL and monolingual speakers. J Deaf StudDeaf Educ 9(1):86–103

22. Echeverry DM, Dike MR, Washington C, Davidson MB (2003)The impact of using a low-literacy patient education tool on

process measures of diabetes care in a minority population. J NatlMed Assoc 95(11):1074–1081

23. Dormandy E, Tsui EY, Marteau TM (2007) Development of ameasure of informed choice suitable for use in low literacy pop-ulations. Patient Educ Couns 66(3):278–295

24. Davis RE, Armstrong DK, Dignan M, Norling GR, Redmond J(2006) Evaluation of educational materials on colorectal cancerscreening in Appalachian Kentucky. Prev Chronic Dis 3(2):A43

25. Paskett ED, Tatum C, Wilson A, Dignan M, Velez R (1996) Use ofa photoessay to teach low-income African American women aboutmammography. J Cancer Educ 11(4):216–220

26. Pander Maat H, Lentz L (2010) Improving the usability of patientinformation leaflets. Patient Educ Couns 80(1):113–119

27. McKee MM, Barnett SL, Block RC, Pearson TA (2011) Impact ofcommunication on preventive services among deaf American SignLanguage users. Am J Prev Med 41(1):75–79

28. MacKinney TG, Walters D, Bird GL, Nattinger AB (1995)Improvements in preventive care and communication for deafpatients: results of a novel primary health care program. J GenIntern Med 10(3):133–137

29. Berman BA, Eckhardt EA, Kleiger HB, Wong G, Lipton DS,Bastani R, Barkin S (2000) Developing a tobacco survey for deafyouth. Am Ann Deaf 145(3):245–255

30. Freimuth VS, Quinn SC (2004) The contributions of health com-munication to eliminating health disparities. Am J Public Health94(12):2053–2055

31. Steinberg AG, Lipton DS, Eckhardt EA, Goldstein M, Sullivan VJ(1998) The diagnostic interview schedule for deaf patients oninteractive video: a preliminary investigation. Am J Psychiatry155(11):1603–1604

32. Choe S, Lim RS, Clark K, Wang R, Branz P, Sadler GR (2009) Theimpact of cervical cancer education for deaf women using a videoeducational tool employing American sign language, open cap-tioning, and graphics. J Cancer Educ 24(1):10–15

33. Kaskowitz SR 3rd, Nakaji MC, Clark KL, Gunsauls DC, SadlerGR (2006) Bringing prostate cancer education to deaf men. CancerDetect Prev 30(5):439–448

34. Folkins A, Sadler GR, Ko C, Branz P, Marsh S, Bovee M (2005)Improving the Deaf community’s access to prostate and testicularcancer information: a survey study. BMC Publ Health 5:63

35. Goldstein MF, Eckhardt EA, Joyner-Creamer P, Berry R, ParadiseH, Cleland CM (2010) What do deaf high school students knowabout HIV? AIDS Educ Prev 22(6):523–537

36. Meador HE, Zazove P (2005) Health care interactions with deafculture. J Am Board Fam Pract 18(3):218–222

37. Centers for Disease Control and Prevention (2001) Prevalence ofhealthy lifestyle characteristics—Michigan, 1998 and 2000.MMWR Morb Mortal Wkly Rep 50(35):758–761

38. Zazove P, Meador HE, Derry HA, Gorenflo DW, Burdick SW,Saunders EW (2004) Deaf persons and computer use. Am AnnDeaf 148(5):376–384

39. Barnett S, Franks P (1999) Smoking and deaf adults: associationswith age at onset of deafness. Am Ann Deaf 144(1):44–50

J Canc Educ (2012) 27:327–337 337