Abstract
Purpose
Examine differences in trust of health information sources between the oldest old and young/middle old.
Design
Cross-sectional survey using convenience sampling.
Setting
Eleven retirement communities.
Subjects
Older adults ≥65 years (N = 353).
Measures
Self-rated trust in health information sources.
Analysis
Mann-Whitney U-test or Fisher exact test to compare trust between age groups; multinomial ordered logistic regression analyses to model trust in Internet information sources.
Results
The overall survey response rate was 26.6%. Differences in trust were identified between oldest old (n = 108) and young/middle old (n = 245) for pharmacist (p < .05), Internet (p < .001), television (p < .05), radio (p < .001), and newspaper (p < .05) sources. In the oldest old, we found associations between levels of trust in Internet sources and frequency of Internet use (β = 4.13, p < .001).
Conclusion
Understanding where differences in trust arise can inform the design of resources to support the information-seeking process. When planning widespread distribution of health information to these distinct groups, program developers need to consider these differences.
Keywords: Aged; Information Seeking Behavior; Health Promotion; Prevention Research; Manuscript format: research; Research purpose: descriptive, instrument development; Study design: nonexperimental; Outcome measure: cognitive, behavioral; Setting: local community, clinical health care; Health focus: intellectual health, medical self-care, social health; Strategy: education, culture change, behavior change; Target population: seniors; Target population circumstances: education/income level, race/ethnicity
PURPOSE
The oldest old (OO), persons 85 years or older, represent the most rapidly growing demographic group in the United States.1 This article examines trust of different health information (HI) sources in the OO and compares it with that in the young/middle old (Y/MO) population, i.e., those aged 65 to 84 years. It also examines different factors contributing to trust in HI sources, including frequency of Internet and television use, along with perceived effort to acquire and understand HI. This work informs the design of interventions that will assist the OO in attaining useful HI and making more informed health decisions, while recognizing that their unique needs and information-seeking behaviors may differ from those of their younger counterparts.
METHODS
Design and Sample
From a parent study, an anonymous cross-sectional paper-and-pencil survey of HI-seeking behavior was developed by examining which sources community-dwelling older adults most frequently use for HI, how reliable they find such sources, and the level of difficulty they experience with the information search process. The survey was distributed to 1520 adults from 11 senior housing communities. Of this number, 403 completed surveys were returned (26.6% response rate). This was a secondary analysis of data provided by respondents older than 65 years (n = 353).
Measures
Demographic variables assessed were age, gender, race, marital status, and education level. The survey included questions from the Health Information National Trends Survey (HINTS) and Krantz Health Opinion Survey (KHOS). The KHOS has a Kuder-Richardson 20 reliability of .74 and has been found to have discriminant validity against the Health Locus of Control Scale.2 The HINTS survey was developed from an underlying conceptual framework and underwent rigorous pretesting and expert review to establish reliability and validity.3 All items, except the agree/disagree KHOS questions, were scored on a four-point Likert scale, higher scores indicating greater agreement. The survey questions asked participants to rate perceived financial adequacy, frequency of access to television and Internet, difficulty and frustration in attaining HI, and trust of different HI sources (health care providers, pharmacists, friends/relatives, Internet, television, radio, newspapers, and retirement community staff). Participant’s health locus of control was assessed through the seven-item KHOS information subscale questionnaire. After checking Cronbach α for internal consistency, the economic questions (α = .939; score range = 3–12), information search questions (α = .871; score range = 4–16), and KHOS questions (α = .762; score range = 0–7) were summed to obtain aggregate scores.
Procedure
Depending on facility preferences, survey distribution methods included (1) mailbox distribution or (2) presentation followed by direct distribution to each participant. Participation was voluntary and we provided 2 to 3 weeks before collecting surveys. The university’s Institutional Review Board approved all study components.
Analysis
Statistical analyses were performed by using R Statistical Software (version 2.12.1; R Foundation for Statistical Computing). Respondents were categorized into the OO and Y/MO groups. Demographic characteristics were compared between age groups by using Mann-Whitney U-test or Fisher exact test as appropriate.
The Friedman test assessed differences in distribution of trust across groups of information sources for the OO. This was followed by post hoc analysis of information sources, using the Wilcoxon-Nemenyi-McDonald-Thompson technique.4 A topologic sort was applied on the relations identified in post hoc tests to generate a rank ordering of information sources based on level of trust.
Mann-Whitney U-tests compared rated trust of information sources between age groups. Multinomial ordered logistic regression modeled dependent variable of trust in Internet information source, and independent variable of Internet access controlled for covariates of sex, marital status, education, economic score, television access, information search score, and KHOS score.
RESULTS
Average age of respondents was 89.5 years (SD: 3.76 years) for OO and 74.7 years (SD: 6.07 years) for Y/MO. Compared to the Y/MO, the OO were more likely to be white (OO: 96.3%; Y/MO: 83.3%), widowed (OO: 65.7%; Y/MO: 33.9%), and facing economic difficulties (OO: 8.7; Y/MO: 6.7). In addition, the OO accessed Internet computer (mean = 2.0, SD = 1.2) less frequently than the Y/MO (mean = 2.7, SD = 1.3). The OO were also less inclined to proactively seek HI as measured by the KHOS score (mean = 3.3, SD = 2.1) than their younger counterparts (mean = 4.8, SD = 2.0). No differences in gender, education, frequency of access to television, and perceived difficulty of information search were found across the two age groups.
The OO placed increasing levels of trust in different HI sources, ranked as follows: (Internet, radio, television) < (newspaper) < (friends/relatives, retirement community staff) < (health care providers, pharmacists). The OO had a lower level of trust than the Y/MO for Internet (p < .001), television (p < .05), and radio sources (p < .001), while they rated pharmacists (p < .05) and newspapers (p < .05) with greater trust (see Table).
Table.
Distribution of Responses to Trust in Information Sources†
| Trust in Information Source, % | |||||
|---|---|---|---|---|---|
| Not at All | A Little | Some | A Lot | Missing | |
| HCP | |||||
| Young/middle old | 0.4 | 5.3 | 33.1 | 55.5 | 5.7 |
| Oldest old | 0.0 | 2.8 | 28.7 | 64.8 | 3.7 |
| Pharmacist (p < .05) | |||||
| Young/middle old | 0.8 | 6.5 | 38.4 | 49.0 | 5.3 |
| Oldest old | 0.9 | 8.3 | 35.2 | 50.0 | 5.6 |
| Friends and relatives | |||||
| Older adults | 3.7 | 26.5 | 41.2 | 22.0 | 6.5 |
| Oldest old | 6.5 | 30.6 | 36.1 | 22.2 | 4.6 |
| Internet (p < .001) | |||||
| Young/middle old | 22.9 | 22.4 | 38.0 | 6.1 | 10.6 |
| Oldest old | 39.8 | 18.5 | 16.7 | 3.7 | 21.3 |
| Television (p < .05) | |||||
| Young/middle old | 20.4 | 38.8 | 29.0 | 7.8 | 4.1 |
| Oldest old | 37.0 | 26.9 | 26.9 | 4.6 | 4.6 |
| Radio (p < .001) | |||||
| Young/middle old | 29.4 | 34.3 | 22.0 | 4.1 | 10.2 |
| Oldest old | 50.9 | 23.1 | 14.8 | 0.0 | 11.1 |
| Newspaper (p < .05) | |||||
| Young/middle old | 22.9 | 33.9 | 31.0 | 7.3 | 4.9 |
| Oldest old | 20.4 | 37.0 | 28.7 | 7.4 | 6.5 |
| Retirement community | |||||
| Young/middle old | 11.8 | 29.0 | 39.2 | 13.5 | 6.5 |
| Oldest old | 9.3 | 23.1 | 42.6 | 17.6 | 7.4 |
Comparisons of distributions were made across age groups by using Mann-Whitney U-tests with different significance levels (sample size n = 353). HCP refers to health care provider.
p < .05.
p < .01.
p < .001.
An association (p < .001) between access to computer with Internet and trust in Internet health sources was found, adjusting for covariates. For every one-unit increase in frequency of Internet computer access, an individual had a 4.13 times greater odds (95% confidence interval: 2.02–8.44) of having a higher level of trust in the Internet, all other covariates being constant. The aforementioned other covariates were unrelated to trust in Internet sources.
DISCUSSION
For the OO, Internet, radio, and television are on the lowest tier of trust. Often, participants are not able to actively engage with these information sources, nor are they able to ask questions for clarification. These sources can also vary in accreditation, making it difficult to assess their validity for critical matters of HI. Research examining trust and perceived value of Internet sources cites challenges in comprehension of HI and usability of Internet sources as potential detractors towards increased trust.5 These results highlight the need to better design Internet HI tailored towards the OO population.
Of the eight information sources, Internet, television, and radio were rated lower by the OO than by the Y/MO, while pharmacists and newspapers were rated higher (see Table). There is greater variability in trust for Internet, television, and radio sources possibly because these information sources are highly diverse, ranging in quality and content. The OO represent a distinct group from the Y/MO population; understanding where differences in trust for HI arise can better inform our design of HI systems and resources to support the information-seeking process.
Frequency of access to Internet was a strong predictor of trust in Internet as a HI source for both the OO and Y/MO. However, this association may also be confounded by difficulties accessing information through the Internet owing to age-related changes (such as vision changes, arthritis). Work by Czaja et al6 indicates that the source of Internet HI and the design features of a Web site also impact user’s perceptions of trust. It is important to maximize the utility of the Internet for older adults given the continuously growing Internet adoption by this population group and the breadth of information available online. This work highlights the need for usability and for ways to assist older adults to navigate online and be able to assess trustworthiness of sources when seeking out HI. This has implications for system designers who need to address needs, expectations, and potential functional limitations of older adult users.
Limitations
The surveys were limited to a geographic region and convenience sampling; generalizations from our results should be made with caution. Participants returned the surveys at a 26.6% response rate, comparable to that found in other similar studies.7,8 Yet, the low response rate raises concerns about potential nonresponse bias.
SO WHAT? Implications for Health Promotion Practitioners and Researchers.
What is already known on this topic?
Although health information (HI)–seeking behavior is a well-studied area of work, there has been limited research addressing issues of trust in information sources for the oldest old (OO).
What does this article add?
The OO expressed less overall trust for information channels that are indirect, such as Internet, television, or radio.
What are the implications for health promotion practice or research?
Designers must be aware of trust differences when using various mediums for widespread distribution of HI and should focus on reducing the effort to acquire and comprehend HI. The OO present unique challenges for designing HI systems; our results shed light on perceived trust of HI across different delivery modes.
Acknowledgments
This research is supported by a grant from the National Library of Medicine (2T15LM007442-11).
Contributor Information
Thai Le, Department of Biomedical Informatics and Medical Education, University of Washington School of Medicine, Seattle, Washington.
Shomir Chaudhuri, Department of Biomedical Informatics and Medical Education, University of Washington School of Medicine, Seattle, Washington.
Cathy White, Clinical Informatics and Patient-Centered Technologies, University of Washington, Seattle, Washington.
Hilaire Thompson, Department of Biobehavioral Nursing and Health Systems, University of Washington School of Nursing, Seattle, Washington.
George Demiris, Department of Biomedical Informatics and Medical Education, University of Washington School of Medicine, Seattle, Washington.
References
- 1.Campion EW. The oldest old. N Engl J Med. 1994;330(25):1819–1820. doi: 10.1056/NEJM199406233302509. [DOI] [PubMed] [Google Scholar]
- 2.Krantz DS, Baum A, Wideman MV. Assessment of preferences for self-treatment and information in health care. J Pers Soc Psychol. 1980;39(5):977–990. doi: 10.1037//0022-3514.39.5.977. [DOI] [PubMed] [Google Scholar]
- 3.Nelson DE, Kreps GL, Hesse BW, et al. The Health Information National Trends Survey (HINTS): development, design, and dissemination. J Health Commun. 2004;9(5):443–460. doi: 10.1080/10810730490504233. [DOI] [PubMed] [Google Scholar]
- 4.Eugster MJA, Hothorn T, Leisch F. Department of Statistics: Technical Reports. 2008. Exploratory and inferential analysis of benchmark experiments. [Google Scholar]
- 5.Jensen JD, King AJ, Davis LA, Guntzviller LM. Utilization of Internet technology by low-income adults: the role of health literacy, health numeracy, and computer assistance. J Aging Health. 2010;22(6):804–826. doi: 10.1177/0898264310366161. [DOI] [PubMed] [Google Scholar]
- 6.Czaja SJ, Sharit J, Nair SN, Lee CC. Older adults and Internet health information seeking. Proc Hum Fact Ergon Soc Annu Meet. 2009;53(2):126–130. [Google Scholar]
- 7.Parker L. Collecting data the e-mail way. Train Dev J. 1992;46(7):52. [Google Scholar]
- 8.Westat. Health Information National Trends Survey 4 (HINTS 4): cycle 1 methodology report. National Cancer Institute; [Accessed September 3, 2012]. Available at: http://hints.cancer.gov/docs/HINTS4_Cycle1_Methods_revised_Jun2012.pdf. Published June 2012. [Google Scholar]