Asian Americans are the fastest-growing ethnic group aged ≥65 years in the United States, with 85% of this group being non–US born [1-4]. Despite their rapid population growth, Asian American older adults remain under-studied and under-served, attributed in part to their heterogeneity, encompassing >40 ethnic subgroups characterized by diverse languages, cultural backgrounds, and immigration histories [1,2,5-7]. Asian American older adults are less likely to use information and communications technologies (ICTs) such as smartphones, tablets, computers, the internet, social media, or other applications as compared to White older Americans [8,9]. Intersectional factors such as socioeconomic status and English proficiency play a significant role in diminishing the likelihood of ICT use among this population [8-10]. According to data from the California Health Interview Survey (CHIS), non–US born Asian American older adults are more inclined to rate their health as fair or poor compared to individuals from other racial and ethnic backgrounds [11]. While there is a significant association between self-rated health and ICT use among older adults [12-15], this association has not been explored among Asian American older adults.

Self-rated health is a subjective measure that encompasses multiple dimensions of an individual’s well-being, including physical, mental, social, and functional aspects. Self-rated health is influenced by personal and cultural beliefs as well as health behaviors [16]. Self-rated health has an independent effect on health outcomes even when adjusting for other measures of medical, physical, cognitive, emotional, and social health [17]. Among Asian American individuals, self-rated health is influenced by factors such as gender; age; and level of acculturation, including English proficiency [18].

Poor self-rated health among Asian American older adults is also closely linked to mental well-being. Analysis of survey data from a large sample of Korean American and Chinese American older adults collected from 2011 to 2019 uncovered significant and positive associations among poor self-rated health, depression, number of chronic diseases diagnosed, and functional disabilities [19]. Furthermore, CHIS data revealed that Asian American older adults with limited English proficiency (LEP) were 3 times more likely to have a perceived mental health need as compared to their English-speaking counterparts [20]. According to the National Asian Pacific Center on Aging, >60% of Asian American individuals aged ≥65 years have LEP, and these individuals are more likely to rate their health as poor [1,20]. Among Asian American older adults, poor self-rated health is attributed to a variety of factors, including low educational attainment, low health literacy, racism and discrimination, disparities in quality of care accumulated over time, and delayed use of health care [11]. Self-rated health among Asian American older adults may be further impacted by challenges such as living far from relatives and experiencing cultural differences and limited social networks in their new environment [1,21,22].

Asian American older adults have higher poverty rates (9.3%) than the general older adult population in the United States (8.9%) [23]. Low-income older adults are more likely to rate their health as poor and report lower ICT use [24,25]. In an analysis of the CHIS, the combination of Asian ethnicity and low income had an interactive, negative effect on ICT use [10]. In fact, Asian American older adults in the lowest income category were 14 times less likely to use the internet for health information compared to non-Hispanic White older adults in the highest income category [10]. More research is needed to understand the factors that impact ICT adoption among low-income, ethnic minority older adults to inform interventions to address this digital divide.

The technology acceptance model (TAM) is a widely accepted framework used to understand the adoption of new technologies [26]. The model suggests that attitudes toward technology, including perceived usefulness (PU) and perceived ease of use (PEOU), significantly influence an individual’s intention to use and actual use of technology.

Nayak et al [14] and Chen and Chan [15] each independently adapted the TAM, adding self-rated health as a covariate predicting ICT use among older adults. In a study of adults aged ≥60 years in the United Kingdom, poor self-rated health was significantly, positively associated with hours per week spent using the internet when adjusting for attitudes toward technology, including PU and PEOU, and demographics [14]. Similarly, in a study of adults aged ≥55 years in Hong Kong, poor self-rated health was positively associated with use of gerontechnologies, a broad term encompassing ICTs, housing and daily living technology, health technology, and education and recreation technologies, when adjusting for attitudes toward technology and demographics. In this study, other health variables were incorporated as covariates, including number of health conditions, physical functioning, cognitive ability, social relationships, and attitude toward life and satisfaction [15].

While Nayak et al [14] and Chen and Chan [15] reported an inverse relationship between self-rated health and technology use, other studies have found them to be positively associated. Werner et al [12] conducted a cross-sectional survey analysis of ethnically and income-diverse adults aged ≥60 years in the United States, demonstrating that better ratings of general physical health and role-related emotional health were independently and positively associated with email and computer use, respectively, when adjusting for demographic factors. In another cross-sectional survey analysis engaging adults aged ≥80 years, Sims et al [13] found that participants with better self-rated health were more likely to report using a greater variety of ICT devices and applications.

Overall, studies support a significant association between self-rated health and ICT use among older adults. However, some have found this relationship to be positive [12,13], whereas others have found it to be negative [14,15]. This association has not been explored specifically among older Asian American individuals, who are more likely to rate their health as fair or poor as compared to older adults of other ethnicities [11].

Beyond self-rated health, cognitive ability is another important predictor of ICT use. When adjusting for self-rated health and demographic factors, cognitive ability was significantly, positively associated with gerontechnology use and ICT use [15]. This relationship has been confirmed in multiple studies. For example, in a cross-sectional study of adults aged ≥65 years, subjective cognitive decline was significantly negatively associated with frequency of ICT use even when adjusting for age, educational attainment, and depression [27]. In addition, in an analysis of the National Health and Aging Trends Study, self-reported memory problems were associated with decreased email, SMS text message, and internet use even when adjusting for age, sex, race and ethnicity, educational attainment, marital status, multimorbidity, and self-rated health [28]. In this study, we examine the influence of self-rated health on ICT acceptance among Asian American older adults adjusting for subjective cognitive decline.

This cross-sectional study examined the relationships among self-rated health, attitudinal factors (PU and PEOU), and ICT use among low-income, Asian American older adults.

In this study, we tested three hypotheses:

In examining hypotheses 1 and 2, we investigated PU and PEOU as potential mediators of the relationship between self-rated health and ICT use. Hypothesis 3 examines self-rated health and subjective cognitive decline as potential moderating or effect-modifying factors among PU; PEOU; demographic factors; and our dependent variable, ICT use. The exploration of effect modification is novel, but we felt that it was necessary to apply this model to an older adult population. This research builds on previous work [29].

This study was a secondary data analysis of the Lighthouse for Older Adults program. The Lighthouse for Older Adults program provided high-speed broadband access, ICT devices, and a series of digital literacy trainings to residents living in 8 affordable housing communities for older adults located across California. All residents (N=1050) were invited to participate. To recruit participants, on-site staff distributed flyers about the Lighthouse program, set up informational tables in high traffic areas of the community, and held social events where residents could test out a variety of ICT devices to inform device selection for the project. All Lighthouse participants were required to complete pre- and postsurveys as well as attend at least one digital literacy training course. All participants were of low income; to qualify for affordable housing for older adults, residents’ incomes must be <80% of the local area median income. The surveys were evidence based, and pilot-tested with residents and staff in 5 languages. The surveys were translated and then back translated by professional translators. The final surveys were self-administered in the residents’ preferred language (eg, Korean, Chinese, Vietnamese, Japanese, Hmong, and English), with staff available to assist as needed.

This study is a cross-sectional survey analysis of the presurveys from the Lighthouse program. Participants were included in this study if they were aged ≥62 years and identified as being of Asian ethnicity. We excluded participants who were missing survey responses to items incorporated into the outcome variable (ICT use). The final dataset included 392 participants.

On the basis of the Human Research Protection-210 Determination Request, the University of California, Davis, Institutional Review Board determined that this research was exempt as it did not directly involve human participants and used deidentified secondary data (1938286-1). The original informed consent allowed for secondary analysis without additional consent, and the survey results were deidentified to maintain the privacy and confidentiality of the participants. As compensation for participating in the Lighthouse program, participants received a free ICT device along with a series of in-language digital literacy training classes. A description of Lighthouse program data collection and measures has been published previously [29].

First, we computed descriptive statistics for demographics, ICT use, self-rated health, and subjective cognitive decline. Next, we used Pearson correlation analysis to examine relationships among all dependent and independent variables.

To test hypotheses 1 and 2, we conducted linear regression using self-rated health as the independent variable. To test hypothesis 1, ICT use was examined as the dependent variable; to test hypothesis 2, PU and PEOU were examined as dependent variables. We adjusted for age, gender, educational attainment, English proficiency, ethnicity, and subjective cognitive decline in all models.

To test hypothesis 3, we conducted hierarchical multiple regression using stepwise blocks [33]. Demographic variables, PU, and PEOU were added in the first step (model 1). Health variables (self-rated health and subjective cognitive decline) were added in the second step (model 2). Finally, in model 3, interaction terms were included in a forward stepwise regression. A total of 12 interaction terms were tested, including all pairwise combinations of health variables (self-rated health and subjective cognitive decline) with demographic and attitudinal variables (age, gender, educational attainment, English proficiency, PU, and PEOU). At each step, goodness of fit (adjusted R²) was assessed, that is, the percentage of variability in the dependent variable that could be accounted for by the predictors. With each new set of terms, the change in R² was calculated to quantify the change in the predictive power. In case interactions were found, contrast statements were used to estimate contrasts of interest.

To further examine the separate and joint association between age and subjective cognitive decline and ICT use, we examined scatterplots with locally estimated scatterplot smoothing lines with age as the predictor. In total, 2 scatterplots were produced, including one with a single scatterplot of data from all participants and another set stratified by presence of subjective cognitive decline. From there, we determined that the use of 3 major age categories captured the relationships well (ages of 62-74, 75-84, and ≥85 years). This led to us updating our analysis of hypothesis 3 and the associated contrast statements to estimate differences in ICT use among the age categories with and without subjective cognitive decline. For all analyses, the α level for testing significance was set to .05. All data analyses were conducted using SPSS Statistics (version 28; IBM Corp).

Participant demographics are described in Table 2. For the sample of 392 participants, the average age was 79.1 (SD 6.95) years.

In examining the 2 health variables, most participants (270/392, 68.9%) reported fair or poor self-rated health, and 27.9% (106/380) of the participants reported subjective cognitive decline. A total of 3.1% (12/392) of the participants declined to respond to the survey question about subjective cognitive decline. The prevalence of fair or poor self-rated health and subjective cognitive decline is detailed by demographic group in Table 3. Next, we used correlation analysis to further examine relationships among ICT use, PU, PEOU, health, and demographic characteristics (Table 4).

Self-rated health was not a significant predictor of ICT use when adjusting for age, gender, educational attainment, English proficiency, ethnicity, and subjective cognitive decline; the standardized coefficient of self-rated health was 0.087 (95% CI –0.131 to 0.978; P=.13; Table 5). Therefore, our findings do not support hypothesis 1 (self-rated health will be significantly, positively associated with ICT use).

Using multiple regression, we determined that self-rated health was not a significant predictor of PU (β=.106, 95% CI −0.022 to 0.274; P=.10) or PEOU (β=.062, 95% CI −0.073 to 0.231; P=.31) when adjusting for age, gender, educational attainment, English proficiency, ethnicity, and subjective cognitive decline (Table 6). Our findings do not support hypothesis 2 (self-rated health will be significantly, positively associated with PU and PEOU).

The results of the hierarchical multiple regression are shown in Table 7. In model 1, we found that PU and PEOU were significant predictors of ICT use (PU: β=.560, 95% CI 0.161-0.960; P=.01; and PEOU: β=.888, 95% CI 0.506-1.270; P<.001). In model 2, adding self-rated health and subjective cognitive decline did not significantly improve model fit for ICT use (ΔR²=0.005; P=.15). In model 3, of the 12 interaction terms that we examined, the interaction between subjective cognitive decline and age was the only one that was statistically significant. Therefore, contrast statements were used to estimate contrasts of interest. Subsequently, we updated the single age variable to reflect 3 different age groups (62 to 74 years, 75 to 84 years, and ≥85 years) and reran the hierarchical multiple regression. In model 3, the interaction terms of subjective cognitive decline and age significantly improved model fit for ICT use (ΔR²=0.011; P=.04). Therefore, our findings support hypothesis 3 (adding self-rated health and subjective cognitive decline as main effects or interaction terms to the TAM will significantly improve model fit), and we reject the null hypothesis. Figure 4 depicts our final supported model of ICT acceptance.

The relationship between subjective cognitive decline and ICT use varied in 3 age segments (Figure 5). In the youngest age group (62 to 74 years), ICT use increased with the presence of subjective cognitive decline, whereas in the older age groups (75 to 84 years and ≥85 years), ICT use decreased with the presence of subjective cognitive decline, more so in the oldest age category.

Using regression analysis, we determined that, among participants with subjective cognitive decline, ICT use significantly decreased in the middle and older age groups as compared to the youngest age group. However, among participants without subjective cognitive decline, the difference in use among age groups was not significant (Table 8).

Our study has several limitations. The data represent a specific population of Asian American older adults; most participants were Korean, and all participants lived in an age-limited setting (affordable housing for older adults), limiting the generalizability of our findings. However, few studies exist targeting this significant Asian population group, enhancing within-group understanding. The use of self-reported measures, including self-rated health and subjective cognitive decline, could introduce response bias.

Due to low educational attainment among study participants, validated survey measures were shortened and adapted to the population. In this study, we operationalized TAM constructs based on elements analogous to the original TAM measures. Nevertheless, the comparability of results across studies is challenging due to variations in outcomes, the operationalization of constructs, and the inclusion of diverse covariates in regression analyses.

Despite these limitations, our sample was unique in that all participants were of low income and represented various Asian ethnicities and most participants had LEP. Our findings have implications for promoting digital engagement among low-income Asian American older adults. Our findings indicate that health-related factors, particularly subjective cognitive decline, can play a role in shaping technology adoption and use patterns. Tailored interventions and support strategies should consider the age-related nuances in the relationship between cognitive decline and technology use.

Enhancing factors such as PEOU and PU through digital literacy training and user interface customization can be particularly beneficial for increasing ICT acceptance among this population [53-58]. Demonstrating culturally relevant use cases for technology, such as connecting with overseas relatives, accessing health information, or enjoying entertainment in one’s native language, can make technology more meaningful for Asian American older adults. For example, the Lighthouse for Older Adults program, which provided the data for this study, improved technology adoption in affordable housing for older adults by offering equipment, training, and support, thereby fostering a community of learners who benefit from shared experiences [59].

The study population had very low English proficiency, and most technology uses Roman characters and is English based. Addressing language barriers is crucial for promoting ICT acceptance and use among low-income, Asian American older adults. Expanding the language capabilities of voice interfaces such as Google Voice and Alexa can make these technologies more user-friendly for those who do not write, read, or speak English [60,61].

Furthermore, the costs of internet access and devices can hinder ICT use among low-income older adults. During the COVID-19 pandemic, several policies were implemented to address these financial barriers. The Accessible, Affordable Internet for All Act of 2021 allocated US $94 billion for broadband infrastructure and subsidies [62]; the Infrastructure Investment and Jobs Act of 2021 set aside US $65 billion for internet access, digital literacy, and rural expansion [63]; and the Affordable Connectivity Program offered low-income families a US $30 monthly subsidy toward their internet costs and required internet service providers to be transparent about pricing [63]. Although the Affordable Connectivity Program ended in February 2024, the Lifeline support program now offers families a smaller subsidy. New state programs are emerging, but continued advocacy is necessary to improve the affordability of digital access.

Future research could further explore the reasons behind the age-dependent effects of cognitive decline on technology use and examine potential interventions to enhance technology accessibility as well as digital literacy and technology adoption among older adults with varying health profiles. Longitudinal studies and intervention designs could provide a deeper understanding of these dynamics.

This study contributes to our understanding of the intricate relationship among self-rated health, subjective cognitive decline, and the acceptance and use of ICTs among low-income, Asian American older adults. By highlighting the age-specific effects of cognitive decline on technology adoption, our findings can inform targeted interventions to promote digital engagement and enhance the quality of life of aging populations.