This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Formative Research, is properly cited. The complete bibliographic information, a link to the original publication on https://formative.jmir.org, as well as this copyright and license information must be included.
The transmission of health information from in-person communication to web-based sources has changed over time. Patients can find, understand, and use their health information without meeting a health care provider and are able to participate more in their health care management. In recent years, the internet has emerged as the primary source of health information, although clinical providers remain the most credible source. The ease of access, anonymity, and busy schedules may be motivating factors to seek health information on the web. Social media has surfaced as a popular source of health information, as it can provide news in real time. The increase in the breadth and depth of health information available on the web has also led to a plethora of misinformation, and individuals are often unable to discern facts from fiction. Competencies in health literacy (HL) can help individuals better understand health information and enhance patient decision-making, as adequate HL is a precursor to positive health information–seeking behaviors (HISBs). Several factors such as age, sex, and socioeconomic status are known to moderate the association between HL and HISBs.
In this study, we aimed to examine the relationship between HL and HISBs in individuals living in a southern state in the United States by considering different demographic factors.
Participants aged ≥18 years were recruited using Qualtrics Research Services and stratified to match the statewide demographic characteristics of race and age. Demographics and source and frequency of health information were collected. The Health Literacy Questionnaire was used to collect self-reported HL experiences. SPSS (version 27; IBM Corp) was used for the analysis.
A total of 520 participants met the criteria and completed the survey (mean age 36.3, SD 12.79 years). The internet was cited as the most used source of health information (mean 2.41, SD 0.93). Females are more likely to seek health information from physicians than males (
Age and sex are significantly associated with HISB. Older adults may benefit from web-based resources to monitor their health conditions. Higher levels of HL are significantly associated with greater HISB. Targeted strategies to improve HISB among individuals with lower levels of HL may improve their access, understanding, and use of health information.
Health information–seeking behavior (HISB) is a complex construct that refers to the ways in which individuals seek information about health, illnesses, and health choices [
While digital health information provides a new landscape for HISB, clinical providers remain one of the most credible sources of health information, especially when it concerns major illnesses [
Health information from the internet also functions to supplement or cross-reference health information obtained elsewhere, such as from providers, especially in older adults as compared with younger adults [
While the younger generations are attuned to seeking health information on the web, the older generation maintains some habit of obtaining health information through traditionally printed materials such as books, newspapers, and magazines [
The numerous methods of acquiring health information have also led to an increase in the breadth of information available. Both accurate and inaccurate information coexist, and there is no overarching regulation to ensure the validity or reliability of information. A study assessing the validity of the search terms “vaccine safety” and “vaccine danger” found that 55% of search results contained inaccurate information within the first 2 pages [
In the context of the COVID-19 pandemic, these stressors in HISB have been exacerbated by an onslaught of new and ever-evolving information from multiple scientific and nonscientific sources. The global scale and frequent evolution of COVID-19 has led to an infodemic or an overabundance of information [
Although digital media often becomes the primary source of information during global health crises, the availability of many platforms and people’s differential access and abilities in using these media are important factors in developing successful communication strategies to mitigate the risk of misinformation [
The concept of HL has broadened over time, from a definition of understanding words and numbers in a medical context to the communication, understanding, and use of health knowledge in an interconnected manner [
With the rapid development of digital media, eHealth literacy has emerged as the use of information and communication technology to improve access to health care and health information [
The Health Literacy Questionnaire (HLQ) measures multidimensional, psychometric aspects of HL. By more than simply measuring whether an individual can read and write, the HLQ measures people’s lived experiences of HL using self-reported experiences [
Of great importance in how people obtain and use information is how organizations provide health information. Organizational HL (OHL) plays an important role in information acquisition. OHL is the extent to which an organization’s provision of health information is at a level where individuals can read, understand, and use it to make decisions about their health [
HL and HISB are inextricably linked, as the ability to seek, use, and comprehend health information requires a certain level of HL. In this way, adequate HL is a precursor to positive HISB [
Some studies show that individual characteristics, such as lower levels of income and lack of access to care associated with low HL are generally concentrated in rural areas, whereas those with higher levels of HL tend to reside in more urban areas [
Health disparities are exacerbated by macrosocial and microsocial factors, such as lack of health care access, low reading skills, health care costs, and geography. People living in the rural southern United States have higher rates of morbidity and mortality compared with their urban counterparts and those in other rural areas, with people of color experiencing higher rates of death and disease as compared with their White counterparts [
In this study, we examined the relationship between HL and HISB considering different demographic factors, such as age, sex, highest level of educational attainment, health insurance status, and county type as rural or urban. The HLQ was used to examine the lived HL experiences of individuals in a southern US state in terms of understanding, accessing, and using health information and health services as a measure of patient-reported outcomes. The purpose of this study was to examine HL and HISBs among a representative sample of adults in a southern US state by answering the following research questions:
What are the average scores on the HISB scales?
Are demographics (sex, age, education level, and county) related to the HISB scales?
Are HISB scales predictive of HL outcomes (have sufficient information [HSI], critical appraisal [CA], FHI, and understanding health information [UHI])?
Are there distinct clusters of participants based on HISB responses? Do these clusters differ by HL outcomes and participant demographics?
Participants who lived in Georgia and were aged ≥18 years were recruited using Qualtrics Research Services and stratified to match statewide demographic characteristics of geography and race (Explore Census Data [
Demographic information on age, sex, race, highest level of educational attainment, health insurance status, and zip code was also collected. We assessed the frequency of sources of health information (from a lot to none) for printed materials, the internet, social media, physicians, and family and friends. We used the HLQ to collect different aspects of lived HL experiences related to accessing, understanding, and using health information: HSI, CA of health information, FHI, and UHI. All scales contain 4 to 6 items scored on a Likert-type scale; HSI and CA scales 1 to 4 have four response options (strongly disagree, disagree, agree, and strongly agree), and FHI and UHI scales 5 to 9 have five response options (cannot do, very difficult, quite difficult, easy, and very easy).
We used SPSS (version 27; IBM Corp, 2020) for the analysis. Descriptive statistics included means, SDs, frequencies, and chi-square calculations. A 2-step cluster analysis was performed using the 4 HL scales.
The study was approved by the institutional review board of Georgia State University under the approval number H21522.
Out of those who responded to the survey, 57.4% (520/905) met all the criteria and completed the survey.
Demographic data for overall sample (n=520).
Variable | Participants, n (%) | |
Female | 371 (71.2) | |
|
||
|
Black or African American | 167 (32.1) |
|
White | 301 (58) |
|
Asian | 28 (5.4) |
|
Hispanic | 13 (2.5) |
|
Other | 12 (2) |
|
||
|
High school diploma or less | 160 (30.7) |
|
Some college education | 177 (34) |
|
College degree | 184 (35.3) |
Has health insurance | 378 (72.6) | |
|
||
|
Urban county | 264 (50.8) |
|
Rural county | 256 (49.2) |
To address the first research question, “What are the average scores on the 5 health information–seeking behavior (HISB) scales?” means, SDs, and range values of the HISB scales are presented in
To address the second research question, “Are demographics (sex, age, educational level, and county) related to the HISB scales?” Spearman rank correlations between demographics (sex, age, educational level, and county) and the 5 HISB scales are shown in
To address the third research question, “Are HISB scales predictive of HL outcomes (HSI, CA, FHI, and UHI)?” a series of multiple regression analyses were conducted using SPSS. We ran 4 regression analyses, which included all HISB scales predicting each HL outcome (HSI, CA, FHI, and UHI;
To address the fourth research question, “Are there distinct clusters of participants based on HISB responses? Do these clusters differ by HL outcomes and participant demographics?” we conducted a 2-step cluster analysis using SPSS with the 5 HISB scales (all scored 1-4). The results indicated that there were 2 distinct HISB clusters based on the 5 scales (
Descriptives of health information–seeking behavior (HISB).
Variable | Value, mean (SD; range) |
HISB printed materials | 2.41 (0.93; 1-4) |
HISB internet | 3.28 (0.81; 1-4) |
HISB social media | 2.31 (1.00; 1-4) |
HISB physicians | 3.19 (0.82; 1-4) |
HISB family and friends | 2.78 (0.82; 1-4) |
Spearman rank correlations among the health information–seeking behavior (HISB) scales and demographicsa.
Variable | Sex | Age (years) | Education level | County | HISB print material | HISB internet | HISB social media | HISB physicians | HISB family and friends | |
|
||||||||||
|
|
1 | −0.038 | 0.018 | 0.081 | −0.025 | 0.075 | 0.019 | 0.121 | 0.060 |
|
—b | .40 | .69 | .07 | .57 | .09 | .67 | .006 | .18 | |
|
||||||||||
|
|
— | 1 | 0.051 | −0.028 | −0.048 | −0.108 | −0.225 | 0.053 | −0.090 |
|
— | — | .25 | .54 | .29 | .02 | <.001 | .24 | .045 | |
|
||||||||||
|
|
— | — | 1 | −0.085 | −0.007 | 0.033 | 0.057 | 0.045 | 0.023 |
|
— | — | — | .05 | .88 | .46 | .19 | .31 | .61 | |
|
||||||||||
|
|
— | — | — | 1 | −0.051 | −0.062 | 0.010 | −0.048 | 0.053 |
|
— | — | — | — | .25 | .16 | .83 | .28 | .23 | |
|
||||||||||
|
|
— | — | — | — | 1 | 0.100 | 0.331 | 0.118 | 0.221 |
|
— | — | — | — | — | .02 | <.001 | .007 | <.001 | |
|
||||||||||
|
|
— | — | — | — | — | 1 | 0.296 | 0.203 | 0.122 |
|
— | — | — | — | — | — | <.001 | <.001 | .005 | |
|
||||||||||
|
|
— | — | — | — | — | — | 1 | 0.095 | 0.330 |
|
— | — | — | — | — | — | — | .03 | <.001 | |
|
||||||||||
|
|
— | — | — | — | — | — | — | 1 | 0.274 |
|
— | — | — | — | — | — | — | — | <.001 | |
|
||||||||||
|
|
— | — | — | — | — | — | — | — | 1 |
|
— | — | — | — | — | — | — | — | — |
aThe sample size ranges from 497 to 520. To interpret the direction of the correlations for dichotomous demographic variables, being female, some college or more, and rural county were all coded higher.
bNot applicable.
Health information–seeking behavior scales predicting have sufficient informationa.
Predictor | Unique |
Coefficient (SE) | ||
Printed materials_D1b | 0.056 | −0.423 (0.070) | 6.02 (508) | <.001 |
Printed materials_D2c | 0.046 | −0.381 (0.069) | 5.50 (508) | <.001 |
Internet_D1 | 0.013 | −0.130 (0.045) | 2.89 (508) | .004 |
Internet_D2 | N/Ad | −0.078 (0.044) | 1.77 (508) | .08 |
Social media_D1 | N/A | −0.047 (0.068) | 0.69 (508) | .49 |
Social media_D2 | N/A | −0.029 (0.065) | 0.44 (508) | .66 |
Doctor_D1 | 0.064 | −0.294 (0.045) | 6.47 (508) | <.001 |
Doctor_D2 | 0.011 | −0.118 (0.044) | 2.68 (508) | .008 |
Family and friends_D1 | N/A | −0.056 (0.060) | 0.93 (508) | .36 |
Family and friends_D2 | N/A | −0.037 (0.058) | 0.64 (508) | .52 |
aThese are standardized coefficients. Total
bD1: dummy code representing “none or little.”
cD2: dummy code representing “some.”
dN/A: not applicable.
Health information–seeking behavior scales predicting critical appraisala.
Predictor | Unique |
Coefficient (SE) | ||
Printed material_D1b | 0.072 | −0.488 (0.067) | 7.31 (508) | <.001 |
Printed material_D2c | 0.037 | −0.344 (0.066) | 5.23 (508) | <.001 |
Internet_D1 | 0.043 | −0.239 (0.042) | 5.65 (508) | <.001 |
Internet_D2 | 0.019 | −0.156 (0.042) | 3.74 (508) | <.001 |
Social media_D1 | N/Ad | 0.012 (0.064) | 0.19 (508) | .85 |
Social media_D2 | N/A | 0.033 (0.061) | 0.537 (508) | .59 |
Doctor_D1 | 0.070 | −0.307 (0.043) | 7.18 (508) | <.001 |
Doctor_D2 | 0.014 | −0.135 (0.041) | 3.25 (508) | <.001 |
Family and friends_D1 | N/A | −0.057 (0.056) | 1.00 (508) | .32 |
Family and friends_D2 | N/A | −0.015 (0.055) | 0.27 (508) | .79 |
aThese are standardized coefficients. Total
bD1: dummy code representing “none or little.”
cD2: dummy code representing “some.”
dN/A: not applicable.
Health information–seeking behavior scales predicting finding health informationa.
Predictor | Unique |
Coefficient (SE) | ||
Printed material_D1b | 0.018 | −0.238 (0.068) | 3.48 (509) | <.001 |
Printed material_D2c | N/Ad | −0.125 (0.067) | 1.85 (509) | .06 |
Internet_D1 | 0.022 | −0.169 (0.044) | 3.87 (509) | <.001 |
Internet_D2 | N/A | −0.067 (0.043) | 1.56 (509) | .12 |
Social media_D1 | N/A | 0.117 (0.066) | 1.77 (509) | .08 |
Social media_D2 | N/A | −0.106 (0.063) | 1.68 (509) | .09 |
Doctor_D1 | 0.099 | −0.365 (0.044) | 8.26 (509) | <.001 |
Doctor_D2 | 0.057 | −0.269 (0.043) | 6.27 (509) | <.001 |
Family and friends_D1 | N/A | −0.044 (0.058) | 0.75 (509) | .45 |
Family and friends_D2 | N/A | −0.038 (0.057) | 0.68 (509) | .50 |
aThese are standardized coefficients. Total
bD1: dummy code representing “none or little.”
cD2: dummy code representing “some.”
dN/A: not applicable.
Health information–seeking behavior scales predicting understanding health informationa.
Predictor | Unique |
Coefficient (SE) | ||
Printed material_D1b | 0.020 | −0.253 (0.069) | 3.69 (509) | <.001 |
Printed material_D2c | 0.016 | −0.225 (0.068) | 3.32 (509) | .001 |
Internet_D1 | 0.011 | −0.121 (0.044) | 2.76 (509) | .006 |
Internet_D2 | N/Ad | −0.031 (0.043) | 0.71 (509) | .48 |
Social media_D1 | N/A | −0.086 (0.066) | 1.30 (509) | .19 |
Social media_D2 | N/A | −0.086 (0.064) | 1.36 (509) | .18 |
Doctor_D1 | 0.114 | −0.391 (0.044) | 8.81 (509) | <.001 |
Doctor_D2 | 0.073 | −0.304 (0.043) | 7.06 (509) | <.001 |
Family and friends_D1 | N/A | −0.036 (0.059) | 0.61 (509) | .54 |
Family and friends_D2 | N/A | −0.029 (0.057) | 0.51 (509) | .61 |
aThese are standardized coefficients. Total
bD1: dummy code representing “none or little.”
cD2: dummy code representing “some.”
dN/A: not applicable.
Two clusters based on 5 health information–seeking behaviors (HISBs).
Our study highlights several important links between HISB and HL: (1) as age increases, people are less likely to seek health information from the internet and social media; (2) seeking health information from social media is not predictive of HL outcomes and is the least-used source of health information for people with high and low HL levels; and (3) people with high HL consistently exhibit more HISBs across multiple sources than those with low HL.
Although internet use has significantly increased in the past few years, disparities remain owing to age, gender, race, and socioeconomic status, which may persist in the digital gap between generations and among populations [
Social media allows users to quickly create and share content and participate in broad information sharing and consumption; different theoretical models propose that individuals are looking for action-oriented information, assessment of risk perception and responses, and more broadly, general information gathering [
Using cluster analysis, we were able to ascertain a high HISB and a low HISB cluster (39% and 61% of the sample, respectively). The high HISB cluster used all 5 sources of health information significantly more than the low HISB cluster in all HISB categories, and social media was used the least by both clusters. The high and low clusters were not differentiated by sex, educational level, county, or age. Interestingly, both clusters used social media the least as a health information source. Wang et al [
The high HISB cluster exhibited higher HL across all 4 scales (HSI, CA of health information, FHI, and UHI). This is consistent with prior studies that indicate that having higher HL may influence a preference for information seeking over and above demographic variables [
While this study sample mirrored the demographics of the state, we were only able to reach individuals who have computer access. Thus, we have reported findings only for individuals who have digital access and at least a minimum of digital literacy skills. As the recruitment was performed using web-based channels, sampling bias is a potential limitation of this study, as those who had difficulties in using these channels could be excluded from recruitment. Another limitation is that we were only able to survey participants in 1 southern US state. We stratified the sample to match the statewide demographic characteristics of geography and race but learned after data collection that sex and age distributions are largely skewed. Future studies should construct more complex stratification to account for this skewness in the data. Although we believe the findings are generalizable among Georgia residents, they may not be generalizable across other states.
Age and sex were significantly associated with HISBs. As older adults are more likely to use health services, they may benefit from having web-based resources to update them on their health status in real time and to provide accessible social support networks. Thus, there is a need to improve HISB skills of and interventions for older adults. Higher levels of HL are associated with greater HISB. Those with lower levels of HL may benefit from targeted strategies to improve their understanding of health information and how to access, understand, and use it, as greater understanding of health information is associated with healthier clinical outcomes. Further studies are needed, specifically those focused on HL, urbanicity, and access to health information.
critical appraisal
finding health information
health information–seeking behavior
health literacy
Health Literacy Questionnaire
have sufficient information
organizational health literacy
understanding health information
None declared.