In Australia, the COVID-19 pandemic has been well controlled compared to many other countries around the world. However, in April 2020, cases and community transmissions had been rising exponentially, and the country was placed under lockdown, including closure of schools and workplaces and restrictions on gatherings and movement. Citizens were required to stay home except for essential purposes (eg, work, essential shopping, exercise). In June 2020, cases were under control and many regulations were eased, although some restrictions remained, such as small gathering sizes, which varied from state to state. A second wave occurred in the state of Victoria shortly after this, requiring new restrictions such as mandatory masks and curfews, but our data were collected prior to this. Thus, a comparison of April and June data presents an opportunity to look at the effect of a short-term lockdown between a time of strong COVID-19 restrictions and good control (Figure 1).

Data from a national Australian survey were used to conduct retrospective case-control analyses comparing hypertension and control groups. Baseline data were collected from all states and territories in April 2020 during the COVID-19 lockdown, with a subsample followed up in June 2020 when restrictions were eased.

Ethics approval was obtained from the University of Sydney Human Research Ethics Committee (2020/212).

The survey measures and full sample results are reported elsewhere [17,18], including the Health Literacy single-item screener [19], Consumer Health Activation Index (CHAI) patient activation measure [20], and State-Trait Anxiety Inventory (STAI) [21]. Participants were asked if they had any of the following conditions: asthma, chronic obstructive pulmonary disease, high blood pressure (hypertension), cancer, heart disease, stroke, diabetes, depression, or anxiety; and whether they take any prescription medication (not specified). The single-item screener provides a brief measure of health literacy, that is, the skills needed to engage in health [19]; and the CHAI provides a measure of patient activation, that is, the extent that a person actively involves themselves in decisions to manage one’s health [20]. Risk perceptions and prevention behaviors (including vaccination intentions) were measured using Likert and categorical scales. Items pertaining to risk perception were based on items developed for an earlier US COVID-19 study [18]. The perceived seriousness of threat from COVID-19 was captured using a 10-point scale (1=”no threat at all to” 10=”very serious public health threat”). The social distancing score reflects perceived importance of social distancing. This outcome is based on 4 items, each answered using a 7-point Likert scale. The items were adapted from existing vaccine attitude instruments to instead reflect on social distancing (“social distancing is important for my family’s health,” “social distancing is important for the health of others in my community,” “when everyone else is socially distancing, I don’t need to,” “I socially distance to protect people with a weaker immune system”). Perceived seriousness was asked generally at baseline; at follow-up, participants were asked about the public health risk from COVID-19 in general, globally, and in Australia specifically, given the divergent pattern of control across countries.

Individuals with hypertension and no other comorbidities (n=466) were retrospectively matched without replacement to healthy controls (with no comorbidities; n=2251) using the calipmatch function in Stata (StataCorp) [22]. For each case, potential controls were initially identified based on age (±3 years) and exactly matching on gender, education, and health literacy adequacy (selected given observed differences as a function of these variables in COVID-19–related knowledge, attitudes, awareness, and behaviors in our baseline survey [17]). One matching control is then randomly selected for the case and removed from the list of available controls for subsequent cases. Because the search strategy for controls is greedy (ie, selecting cases for matching in random order and removing controls without replacement for subsequent case matching), some cases may be left unmatched. The initial matching run resulted in 95.7% (446/466) of cases successfully matched to a control. The constraints for matching were iteratively relaxed (eg, allowing age to vary by ±10 years; education level to differ by one category) until all remaining cases were paired to a control. The matching procedure was repeated for the follow-up sample.

Analyses were conducted using Stata/IC v16.1 (StataCorp). Pairwise comparisons of baseline demographic characteristics were undertaken to confirm the appropriateness of the matching procedure of cases to controls, and to identify potential differences in demographic characteristics between those who were invited and returned for follow-up compared to those who were not followed up. Regression models with robust error variances to account for clustering within pairs, and adjusted for matching variables (age, gender, education, and health literacy adequacy), were used to analyze outcome variables. Linear models were used for continuous outcomes (risk perceptions, STAI anxiety, perceived importance of social distancing) to estimate marginal mean differences (MMD). Generalized linear models with a modified Poisson approach [23] were used for the dichotomous outcome “not feeling stressed due to COVID-19,” generating adjusted prevalence ratios (aPR). Ordinal logistic regression models were used for ordered categorical outcomes (frequency of leaving one’s home, vaccination intentions), resulting in adjusted odds ratios (aOR). Separate models were conducted for each time point. All estimates are provided with 95% CI values. A P value of .05 was used as the threshold for statistical significance.

Data are available upon reasonable request subject to ethics approval.

To isolate the effects of hypertension, the hypertension sample included 466 people reporting only high blood pressure and no other chronic health conditions. The mean age was 54 years (SD 15.5), and the sample comprised 52% (n=243) female, 47% (n=220) male, and 1% (n=3) unspecified. The majority had a university degree (n=282, 61%) and adequate health literacy (n=427, 92%). The average patient activation score was comparable to other patient populations (mean scaled CHAI 74.9). Most were taking medications (n=359, 77%), with 45% (n=163) obtaining a refill during the lockdown, 5% (n=19) switching to a longer prescription, and only 1 person stopping their medication. As seen in Table 1, the sample descriptive characteristics were comparable between individuals with hypertension and the matched controls. There was no statistical difference across age (P=.33), gender (P>.99), education (P=.97), or health literacy adequacy (P=.63) between cases and controls. Cases who were invited and returned for follow-up were of similar age and gender but had higher levels of education (P=.02) and were more likely to have adequate health literacy (P=.009) than those who were not followed up.

At baseline, the perceived seriousness of threat from COVID-19 in the hypertension group was high (mean 7.72, out of 10) but similar to controls (mean 7.66). On average, the hypertension sample believed that 7% of people who get COVID-19 would die as a result and 63% would experience only mild symptoms (asked separately). There were no statistically significant differences between the hypertension group and the matched controls at baseline. At follow-up, those with hypertension perceived a greater threat (mean 6.12) than controls (mean 5.52) when asked about Australia (MMD 0.60, 95% CI 0.05-1.15; P=.03) but not in general or globally.

At baseline, 76% (n=353) of the hypertension group had felt nervous or stressed about COVID-19 in the past week at least some of the time. On average, the mean STAI was 1.90 units higher (95% CI 0.19 to 3.61; P=.03, Cohen d=0.13) for those with hypertension (mean 40.75) than matched controls (mean 38.85). At follow-up, there was no longer a significant difference between the hypertension (mean 37.02) and control (mean 36.08) groups (MMD 0.94, 95% CI –2.57 to 4.45; P=.60, Cohen d=0.06).

At baseline, the hypertension group had a social distancing score of 6.48 out of 7, indicating strong agreement with the importance of social distancing for ones’ own health and the health of the public; this was similar to the controls (6.42 out of 7). Most people left home a few times a week (n=176, 38%) or once a day (n=154, 33%) during the lockdown. Overall, 83% (n=386) agreed they would get the influenza vaccine, and 87% (n=404) would get the COVID-19 vaccine. Compared to healthy matched controls, the hypertension group was more likely to agree that they would (or have already) received the influenza vaccine this year (aOR 1.52, 95% CI 1.10-2.11; P=.01). There were no significant differences in willingness to vaccinate for COVID-19 (if it became available), perceived importance of social distancing, or frequency of leaving one’s home. At follow-up, there was no longer a significant difference between the hypertension and control groups for influenza vaccination intention (aOR 1.90, 95% CI 0.93-3.90; P=.08), with intentions remaining high for both influenza and COVID-19 vaccination (>80% for both groups).

The main observation of this study was the significant difference in anxiety levels between hypertension only and matched control groups, with all groups reporting higher than “normal” levels. This is consistent with the Australian Bureau of Statistics’ finding that the rate of anxiety in the general population had doubled in April 2020 compared to a survey from 2017-18 [24]. Prioritizing mental health screening for more vulnerable clinical groups with higher anxiety may be warranted when local community transmission rates are high.

Overall, there were few differences between people with hypertension and healthy matched controls. No significant differences were found for COVID-19 risk perceptions or perceived importance of social distancing behaviors. This is consistent with another study, which found that 20% of people with chronic conditions did not perceive greater risk [6], but differs from other survey reports that indicate people with different chronic conditions are more likely to engage in COVID-19 prevention behaviors and perceive COVID-19 as a serious threat [18,25]. This may be due to a close resemblance between the hypertension and general populations in our study, or it may be a result of our method of matching cases to controls rather than comparing groups without such adjustment. Another Australian survey found similarly high risk perceptions, so there may also be a ceiling effect in Australia across community groups [26].

Responses to flu vaccine uptake varied across the two groups, whereby those with hypertension were more likely to intend to vaccinate compared to healthy controls. It is possible this is due to the former’s greater exposure to the health system where doctors may mention the flu vaccine each year. This difference does not appear to transfer to increased intent for COVID-19 vaccine uptake, but this may be due to a ceiling effect with high acceptance rates in Australia [27] compared to other countries such as France [28]. It should be noted that vaccine acceptance rates are changing over time as new information (and misinformation) becomes available about the various vaccines [29] now being used around the world. No COVID-19 vaccinations were available to Australians at the time of the study in 2020.

Differences in medication use were found between groups, but this was to be expected given that preventive medication is recommended for hypertension. Surprisingly, there were no differences in access difficulties or changes to medication. The Australian Bureau of Statistics reported in April 2020 that almost half (47%) of respondents with a chronic condition had used telehealth [24], including electronic prescriptions; this was not a focus of our survey but may explain why little change was detected.

The strengths of this study include a large national sample with data during and after lockdown restrictions, which enabled matched case-control analyses between participants with self-reported hypertension and healthy controls and the use of established, well-validated measures.

The sample was recruited via an online research panel and social media, and has a low proportion of culturally and linguistically diverse participants; hence, different results may be found in other populations. We are currently conducting a separate survey of these communities in their preferred language. The survey involved nonstratified sampling without targeted recruitment of specific health conditions, and only a subset were included in the longitudinal substudy. Future research could explore the influence of multimorbidity and differences between social media users and other community members, given misinformation concerns in Australia [30].

Anxiety was above normal levels for all groups during the COVID-19 lockdown. This was higher among people with hypertension, who also had higher influenza vaccination intentions but similar COVID-19 vaccination intentions. In Australia, where lockdown measures effectively reduced the spread of COVID-19 and restrictions eased relatively quickly, these differences dissipated after 2 months, but locations with prolonged restrictions may require targeted psychological screening for vulnerable groups. Despite a decrease in perceived seriousness and anxiety after 2 months of lockdown restrictions, vaccination intentions for both influenza and COVID-19 remained high (80%), which is encouraging for the future prevention of COVID-19.