Data and sample
Data from the 2018 US National Health Interview Survey (NHIS) public-use data files were downloaded from the CDC’s National Center for Health Statistics website, which contains a variety of health information Internet use . As an annual, nationally representative, cross-sectional household survey, NHIS provides information on the health and health care access of the civilian, non-institutionalized US population . The NHIS obtains data through a complex multistage sample design that involves the stratification and clustering of specific population subgroups. For each sampled family, a face-to-face interview is conducted with an adult family member, who answers questions about the demographic and health conditions of each family member. A detailed description of the objectives and methods of NHIS has been reported elsewhere . Since these data are provided to the public in an identifiable format, this study was exempt from review by the Institutional Review Board. In the 2018 NHIS annual data, a sample of 72,746 adult respondents between the ages 18 and 85 was obtained (to protect confidentiality among the oldest adults, all age variables were top-coded to “85 years and older” (85 +) in the HNIS database). In this study, we focused on patients aged 65 years and older with physical multi-morbidity, and we excluded respondents who had missing values of dependent or independent variables.
We defined multi-morbidity as the presence of two or more physical chronic non-communicable diseases . We used 13 non-communicable diseases to measure physical multi-morbidity, including diagnosed hypertension, high cholesterol, coronary heart disease, angina pectoris, heart condition, stroke, chronic obstructive pulmonary disease, asthma, cancer, diabetes, failing kidneys, liver condition, and arthritis. We counted the number of non-communicable diseases for each participant to identify those who had physical multi-morbidity.
Respondents were asked if they had used the Internet in the past year (Yes = 1 and No = 0).
Health information technology use
Respondents were asked if they had (1) looked up health information on the Internet, (2) filled a prescription on the Internet, (3) scheduled a medical appointment on the Internet, and (4) communicated with a health care provider by email in the past 12 months. In this study, HIT use refers to any of these 4 aspects.
In NHIS, self-rated health was obtained by asking respondents, “Would you say your health, in general, is excellent, very good, good, fair, or poor?” We assigned "excellent", "very good" and “good” to 1, "fair" to 2, and "poor" to 3.
Respondents were asked if they had Medicare, Medicaid, private health insurance, and veterans/military insurance coverage in the past 12 months (yes = 1 and no = 0 for each).
Health care use
Respondents were asked if they (1) saw/talked to a medical specialist, (2) saw/talked to a general doctor, (3) received home care from a health professional, and (4) received health care 10 or more times in the past 12 months (yes = 1 and no = 0 for each).
A set of sociodemographic factors were included: (1) gender, (2) age (65—74, 75—84, and ≥ 85 years), (3) marital Status (Currently married, Widowed, Divorced/separated, Never married), (4) region (Northeast, Midwest, South, West), (5) race (Hispanic, Non-Hispanic White, Non-Hispanic Black, Non-Hispanic Asian, Non-Hispanic All other race groups), (6) highest education (High school and below, Some college, College, and Master degree and above), (7) the percentage of family income to the federal poverty guidelines (FPG) (< 200% FPG, 200%-399% FPG, ≥ 400% FPG).
Statistical analyses were performed using the SPSS statistics 22.0 (IBM Corp. Released 2013. IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY: IBM Corp). Some covariates contained missing values, and the proportion of missing values was less than 5% . Thus, we replaced the missing data with the mean of their integrity items. Frequency and percentages were calculated to describe sociodemographic parameters and level distributions among respondents. We used logistic regression models with Internet and HIT use as dependent variables to determine factors that affect use of Internet and HIT among older adults with multi-morbidity. We report associations as unadjusted rates (uORs) and odds ratios adjusted (aORs) for gender, age, marital status, region, race, highest education, family income, health status, health insurance, and health care use. A significance level of 95% was used with α = 0.05.