Survey Design and Sample Size
The study analyzed data from the Health Information National Trends Survey (HINTS), a probability-based and nationally representative survey of the U.S. adult (age 18+) noninstitutionalized population. The National Cancer Institute conducts the Hints survey to gather data on Americans' knowledge, attitudes, and health-related behaviors. This study analyzed secondary data obtained from four iterations (HINTS 4 Cycle 1: 2011, HINTS 5 Cycle 1: 2017, HINTS 5 Cycle 4: 2020, and HINTS 6 Cycle 1: 2022). The study focused on only individuals belonging to the minority races comprising Hispanic, Non-Hispanic Black or African American (non-Hispanic BAA), Non-Hispanic American Indian or Alaskan Native (non-Hispanic AIAN), Non-Hispanic Asian, Non-Hispanic Native Hawaiian or Other Pacific Islander (non-Hispanic NHOPI), and Non-Hispanic Multiple Races. The final sample included 1,067 participants in HINTS 4 Cycle 1, 966 in HINTS 5 Cycle 1, 1222 in HINTS 5 Cycle 4, and 2246. The table 1 below presents a breakdown of total sample for each survey cycle, along with the sample size for the study.
Table 1: Number of participants in the various surveys and corresponding study sample sizes.
Survey
|
Year
|
All Samples
|
Study Samples
|
HINTS 4 Cycle 1
|
2011
|
3,265
|
1,067
|
HINTS 5 Cycle 1
|
2017
|
2,777
|
966
|
HINTS 5 Cycle 4
|
2020
|
3,319
|
1,222
|
HINTS 6 Cycle 1
|
2022
|
5,437
|
2,246
|
Measurements
The study employed a set of measurements to assess various dimensions of interest. Patient-centered communication (PCC) was operationalized through seven questions, drawing on the conceptual framework outlined by Epstein and Street [15] and derived from prior investigations [27]. Respondents were prompted to rate, on a four-point continuum (1 = always, 4 = never), the frequency of their interactions with healthcare professionals over the preceding 12 months in relation to specific aspects: how often did your doctors, nurses, or other health care professionals… do the following? (1) I give you the chance to ask all the health-related questions you had; (2) I give the attention you needed to your feelings and emotions; (3) I involve you in decisions about your health care as much as you wanted; (4) I make sure you understood the things you needed to do to take care of your health; (5) I explain things in a way you could understand; (6) I spend enough time with you; and (7) I help you deal with feelings of uncertainty about your health and health care. We subjected the responses to these questions to reverse coding and then averaged them to construct a composite score. Higher values indicated a heightened degree of patient-centered communication (2011: M = 3.28, SD =.72; 2017: M = 3.38, SD =.68; 2020: M = 3.39, SD =.68; 2022: M = 3.26, SD =.72).
Health competence was evaluated by asking participants to indicate the extent to which they were confident about their ability to take good care of their health [28]. Responses were recorded on a five-point spectrum (1 = completely confident, 5 = not confident at all) and subsequently reverse coded, where an elevated score signified an augmented level of health competence (2011: M = 3.82, SD =.93, 2017: M = 3.86, SD =.85, 2020: M = 3.84, SD =.86, 2022: M = 3.88, SD =.90).
General health was measured through a single item, prompting respondents to subjectively rate their perceived general health [29]. Five grades were provided: excellent, very good, good, fair, and poor. A higher rating was indicative of superior health (2011: M = 3.25, SD =.97; M = 3.24, SD =.94; 2020: M = 3.23, SD =.97; 2022: M = 3.18, SD =.91).
Mental health was gauged using four queries derived from antecedent scholarly work [30]. Participants were prompted to reflect on their experiences over the past 12 weeks, specifically in relation to (1) having little interest or pleasure in doing things; (2) feeling down, depressed, or hopeless; (3) feeling nervous, anxious, or on edge; and (4) not being able to stop or control worrying. Responses were scored on a four-point scale ranging from 1 = nearly every day to 4 = never. Subsequently, an average was computed, with a higher value denoting better mental health (2011: M = 3.35, SD =.82; 2017: M = 3.51, SD =.73; 2020: M = 3.47, SD =.74; 2022: M = 3.42, SD =.75).
The demographic profile included respondent age, sex (1 = male, 0 = female), educational attainment (ranging from 1 = less than 8 years to 7 = postgraduate), annual household income (1 = $0 to $9,999, 9 = 200,000 or more), and racial background (1 = non-Hispanic white, 0 = others).
Table 2: Amount (percent) of missing values present for each variable for each year and across all years before imputation.
variable
|
2011
|
2017
|
2020
|
2022
|
Total
|
Race
|
165 (5.05%)
|
267 (9.61%)
|
299 (9.01%)
|
525 (9.66%)
|
1256 (20.05%)
|
Education
|
63 (1.93%)
|
73 (2.63%)
|
107 (3.22%)
|
295 (5.43%)
|
515 (8.22%)
|
Gender
|
83 (2.54%)
|
53 (1.91%)
|
73 (2.2%)
|
301 (5.54%)
|
433 (6.91%)
|
mh2
|
66 (2.02%)
|
88 (3.17%)
|
84 (2.53%)
|
205 (3.77%)
|
330 (5.27%)
|
mh3
|
62 (1.9%)
|
66 (2.38%)
|
74 (2.23%)
|
211 (3.88%)
|
324 (5.17%)
|
mh1
|
62 (1.9%)
|
66 (2.38%)
|
73 (2.2%)
|
199 (3.66%)
|
310 (4.95%)
|
mh4
|
61 (1.87%)
|
68 (2.45%)
|
73 (2.2%)
|
201 (3.7%)
|
308 (4.92%)
|
Age
|
47 (1.44%)
|
101 (3.64%)
|
94 (2.83%)
|
70 (1.29%)
|
235 (3.75%)
|
General Health
|
18 (0.55%)
|
32 (1.15%)
|
36 (1.08%)
|
150 (2.76%)
|
190 (3.03%)
|
Health Competence
|
20 (0.61%)
|
36 (1.3%)
|
35 (1.05%)
|
136 (2.5%)
|
186 (2.97%)
|
pcc7
|
61 (1.87%)
|
55 (1.98%)
|
50 (1.51%)
|
100 (1.84%)
|
168 (2.68%)
|
pcc6
|
54 (1.65%)
|
39 (1.4%)
|
44 (1.33%)
|
82 (1.51%)
|
166 (2.65%)
|
pcc2
|
46 (1.41%)
|
44 (1.58%)
|
46 (1.39%)
|
67 (1.23%)
|
151 (2.41%)
|
pcc3
|
40 (1.23%)
|
36 (1.3%)
|
33 (0.99%)
|
67 (1.23%)
|
140 (2.23%)
|
pcc5
|
40 (1.23%)
|
35 (1.26%)
|
28 (0.84%)
|
57 (1.05%)
|
124 (1.98%)
|
Household Income
|
62 (1.9%)
|
27 (0.97%)
|
15 (0.45%)
|
13 (0.24%)
|
109 (1.74%)
|
pcc4
|
36 (1.1%)
|
27 (0.97%)
|
18 (0.54%)
|
56 (1.03%)
|
105 (1.68%)
|
pcc1
|
31 (0.95%)
|
21 (0.76%)
|
21 (0.63%)
|
48 (0.88%)
|
96 (1.53%)
|
Data analysis
Missing Data
Table 2 provides information on the number and percentage of missing values for the variables considered for each year of data. We see a sizable number of missing values, mostly for race, education, sex, and mental health variables. These missing data are somewhat concerning to be ignored and were therefore imputed at the yearly level using the multiple imputation by chained equations (MICE) method [31]. The procedure ‘fills in’ (imputes) missing data in a dataset through an iterative series of predictive models. In each iteration, each specified variable in the dataset is imputed using the other variables in the dataset. The method is robust and very flexible in handling both continuous and categorical variables of different types, with the help of many imputation methods that are basically regression models behind the scenes. We used the random forest imputation method for our purposes.
Descriptive Statistics
This study presented summary statistics tables generated via ANOVA and chi-square tests to illustrate the distributions of the various variables of interest across the different years of data collection. P values based on ANOVA (for continuous variables such as age and PCC) and chi-square tests (for categorical/factor variables such as sex, race, and education) were included to help determine whether the observed changes over the years were statistically significant or if they could be due to random variation.
Demographic characteristics of the sample. Table 3 presents a summary of the participants’ sociodemographic characteristics. The average age ranged roughly between 52 and 55 years but varied significantly across the years (p < 0.001), with a steady increase from 2011 to 2020 and a clear downward trend from 2020 to 2022. The distribution of sex at birth did not exhibit a significant change over the years (p = 0.094). In 2011, 64% of the individuals were identified as female at birth; this percentage fluctuated within a narrow range, reaching 66% in 2022, while male participation was consistently lower than female participation (ranging from 34% to 38%). The distribution of minority races significantly changed over the years (p < 0.001). Among the minority races, Hispanics and non-Hispanic Asians dominated, with participation rates ranging from 33% to 47%. While the percentage of people identified as Hispanic increased from 33% in 2011 to 41% in 2022, the percentage of people identified as Asian decreased from 47% in 2011 to 39% in 2022. However, other minority race categories have relatively stable participation rates.
Educational attainment significantly changed over the years (p < 0.001). The proportion of individuals with a college degree or higher has increased from 34% in 2011 to 40% in 2022. Moreover, the percentage of individuals with less than a high school education decreased from 16% in 2011 to 9.7% in 2022. The distribution of household income exhibited a significant change over the years (p < 0.001). The percentage of individuals with incomes less than $20,000 has decreased from 32% in 2011 to 26% in 2022. On the other hand, the percentage of individuals in the middle-income brackets ($20,000-$49,999 and $50,000-$99,999) showed relatively small changes, while the percentage in the highest income bracket (> $99,999) decreased slightly from 2011 to 2022.
Table 3: Demographic characteristics of samples from the minority population.
Characteristic
|
2011
|
2017
|
2020
|
2022
|
p value
|
|
Age
|
52.04 (15.93, [51.08, 53.00])
|
54.32 (16.07, [53.31, 55.34])
|
54.80 (16.47, [53.87, 55.72])
|
52.68 (17.10, [51.97, 53.39])
|
<0.001
|
|
Gender at birth
|
|
|
|
|
0.094
|
|
Female
|
679 (64%)
|
624 (65%)
|
757 (62%)
|
1,485 (66%)
|
|
|
Male
|
388 (36%)
|
342 (35%)
|
465 (38%)
|
761 (34%)
|
|
|
Minority race
|
|
|
|
|
<0.001
|
|
Hispanic
|
352 (33%)
|
355 (37%)
|
511 (42%)
|
931 (41%)
|
|
|
Non-Hispanic Black or African American
|
17 (1.6%)
|
7 (0.7%)
|
13 (1.1%)
|
27 (1.2%)
|
|
|
Non-Hispanic AIAN
|
131 (12%)
|
110 (11%)
|
135 (11%)
|
259 (12%)
|
|
|
Non-Hispanic Asian
|
501 (47%)
|
391 (40%)
|
458 (37%)
|
875 (39%)
|
|
|
Non-Hispanic NHOPI
|
61 (5.7%)
|
100 (10%)
|
96 (7.9%)
|
136 (6.1%)
|
|
|
Non-Hispanic Multiple Races
|
5 (0.5%)
|
3 (0.3%)
|
9 (0.7%)
|
18 (0.8%)
|
|
|
Education
|
|
|
|
|
<0.001
|
|
Less than High School
|
166 (16%)
|
105 (11%)
|
141 (12%)
|
218 (9.7%)
|
|
|
High School Graduate
|
221 (21%)
|
186 (19%)
|
257 (21%)
|
411 (18%)
|
|
|
Some College
|
318 (30%)
|
292 (30%)
|
354 (29%)
|
717 (32%)
|
|
|
College Graduate and Higher
|
362 (34%)
|
383 (40%)
|
470 (38%)
|
900 (40%)
|
|
|
Household income
|
|
|
|
|
<0.001
|
|
< $20,000
|
343 (32%)
|
256 (27%)
|
289 (24%)
|
583 (26%)
|
|
|
$20,000-$49,999
|
353 (33%)
|
287 (30%)
|
374 (31%)
|
628 (28%)
|
|
|
$50,000-$99,999
|
253 (24%)
|
271 (28%)
|
314 (26%)
|
600 (27%)
|
|
|
> $99,999
|
118 (11%)
|
152 (16%)
|
245 (20%)
|
435 (19%)
|
|
|
1Mean (SD, [95% CI]) for continuous; n (%) for categorical
|
2One-way ANOVA; Pearson's Chi-squared test
|
The study presents descriptive statistics of PCC, health competence, and health outcomes, encompassing both general health and mental health. The results in Table 4 reveal significant trends in PCC (p < 0.001) and mental health (p < 0.001) perceptions over the years, with PCC values declining and mental health values improving. However, health competence and general health perceptions seem to have remained relatively stable during the same time frame, as evidenced by the high p values of 0.3 and 0.12, respectively. A graphical representation of the changes or trends observed in the above measures of interest over the four iterations of the survey under discussion is also presented in Figures 1, 2, and 3 in the Appendix.
Table 4: Summary statistics of PCC, health competence, mental health, and general health.
Characteristic
|
2011
|
2017
|
2020
|
2022
|
p
|
pcc1
|
3.41 (0.80, [3.36, 3.46])
|
3.52 (0.74, [3.47, 3.57])
|
3.51 (0.73, [3.47, 3.55])
|
3.40 (0.76, [3.37, 3.43])
|
<0.001
|
pcc2
|
3.16 (0.92, [3.11, 3.22])
|
3.25 (0.92, [3.19, 3.31])
|
3.27 (0.89, [3.22, 3.32])
|
3.16 (0.88, [3.12, 3.19])
|
<0.001
|
pcc3
|
3.25 (0.88, [3.20, 3.31])
|
3.34 (0.84, [3.28, 3.39])
|
3.40 (0.82, [3.35, 3.44])
|
3.27 (0.83, [3.24, 3.30])
|
<0.001
|
pcc4
|
3.46 (0.76, [3.42, 3.51])
|
3.54 (0.69, [3.49, 3.58])
|
3.55 (0.70, [3.51, 3.59])
|
3.40 (0.78, [3.37, 3.43])
|
<0.001
|
pcc5
|
3.47 (0.73, [3.42, 3.51])
|
3.57 (0.66, [3.53, 3.61])
|
3.56 (0.69, [3.52, 3.60])
|
3.43 (0.75, [3.40, 3.46])
|
<0.001
|
pcc6
|
3.15 (0.91, [3.09, 3.20])
|
3.26 (0.86, [3.21, 3.32])
|
3.24 (0.88, [3.19, 3.29])
|
3.07 (0.91, [3.03, 3.11])
|
<0.001
|
pcc7
|
3.09 (0.96, [3.03, 3.15])
|
3.19 (0.92, [3.13, 3.25])
|
3.21 (0.92, [3.15, 3.26])
|
3.07 (0.93, [3.03, 3.10])
|
<0.001
|
PCC
|
3.28 (0.72, [3.24, 3.33])
|
3.38 (0.68, [3.34, 3.42])
|
3.39 (0.68, [3.35, 3.43])
|
3.26 (0.72, [3.23, 3.29])
|
<0.001
|
mh1
|
3.29 (0.97, [3.23, 3.34])
|
3.40 (0.93, [3.35, 3.46])
|
3.43 (0.92, [3.38, 3.49])
|
3.34 (0.96, [3.30, 3.38])
|
<0.001
|
mh2
|
3.39 (0.91, [3.33, 3.44])
|
3.56 (0.79, [3.51, 3.61])
|
3.58 (0.76, [3.54, 3.63])
|
3.52 (0.79, [3.49, 3.56])
|
<0.001
|
mh3
|
3.40 (0.88, [3.34, 3.45])
|
3.52 (0.82, [3.47, 3.57])
|
3.44 (0.86, [3.39, 3.49])
|
3.42 (0.86, [3.38, 3.45])
|
0.007
|
mh4
|
3.32 (0.99, [3.26, 3.38])
|
3.54 (0.83, [3.49, 3.60])
|
3.43 (0.92, [3.38, 3.49])
|
3.42 (0.90, [3.38, 3.46])
|
<0.001
|
Mental health
|
3.35 (0.82, [3.30, 3.40])
|
3.51 (0.73, [3.46, 3.55])
|
3.47 (0.74, [3.43, 3.51])
|
3.42 (0.75, [3.39, 3.45])
|
<0.001
|
Health competence
|
3.82 (0.93, [3.76, 3.88])
|
3.86 (0.85, [3.81, 3.91])
|
3.84 (0.86, [3.79, 3.88])
|
3.88 (0.90, [3.84, 3.91])
|
0.3
|
General health
|
3.25 (0.97, [3.19, 3.31])
|
3.24 (0.94, [3.18, 3.30])
|
3.23 (0.97, [3.18, 3.29])
|
3.18 (0.91, [3.14, 3.22])
|
0.12
|
- Mean (SD, [95% CI])
- One-way ANOVA