The COVID-19 pandemic affected healthcare utilisation worldwide.1 Although healthcare services generally remained open during the pandemic, lockdowns and stay-at-home orders resulted in disruptions to usual healthcare, with decreases in admissions to hospitals, imaging evaluations and emergency department presentations, and routine primary care attendance in a number of countries.2–5 An Australian Institute of Health and Welfare (AIHW) report showed that elective admissions decreased by 8.3% in 2019–20 compared to 2018–19 in Australia because of restrictions on elective surgery in early 2020.6 The AIHW reported the average daily presentations to emergency departments decreased by 26% in a single month (April 2020); although average daily presentations steadily increased in May and June 2020, they remained 8.6% lower than the corresponding period in 2018–19.7 Additionally, all hospitalisations (public and private) decreased by 2.1% in 2021–22 compared to 2020–21.7
A systematic review of healthcare access during the COVID-19 pandemic8 found that consumer sentiments influenced reductions in access to healthcare services.9,10 Two Australian studies conducted among adults in 2020 found that despite believing they needed an in-person appointment, some people decided to cancel or postpone their face-to-face health appointments.10,11 This decision was driven by the fear of COVID-19 infection and hesitation related to the widespread uncertainty surrounding potential changes to travel restrictions and public health messaging. Additionally, during 2020–22 in Australia, there was mandatory isolation for people with the COVID-19 infection, and their close contacts had to quarantine, impeding their ability to access face-to-face healthcare.12
Ensuring equitable healthcare access during a pandemic is challenged by pre-existing inequities in access experienced by certain communities, such as those from rural and remote areas,10,13 and people who speak a language other than English.14 Medical claims data from 2020–21 on the number of attendances per person for non-hospital, non-referred healthcare services such as general practitioner (GP) visits, showed that people living in remote (4.2 attendances per person) and very remote areas (3.4 attendances per person) had lower attendances compared to people living in metropolitan areas (6.6 attendances per person).13 Also, a scoping review focusing on culturally and linguistically diverse (CALD) populations in Australia identified challenges such as poor health literacy, multimorbidity and a diminishing ‘health migrant’ effect that influenced access to health services.14
Most studies documenting the impact of COVID-19 on healthcare utilisation in Australia during the pandemic have focused on access (or lack of it) to specific health services (eg emergency healthcare utilisation, dental care) among the general population15,16 or conducted qualitative interviews only.10 This study focused on populations at higher risk of COVID-19 infection, severity or unintended consequences of COVID-19 restrictions, providing critical data to examine changes experienced in healthcare access or utilisation across all settings from 2020 to 2021, concerns people had about access to their usual healthcare, the association between sociodemographic factors and difficulty accessing healthcare. Also, the study was conducted in the State of Victoria, which experienced the most stringent and prolonged COVID-19 restrictions in Australia.
Methods
Study design
The Optimise study (Optimise) is a longitudinal cohort study, following participants aged 18 years or older living in Victoria, Australia, who were recruited between September 2020 and December 2021. Optimise was designed to provide in-depth insights into priority populations considered at risk of contracting COVID-19, developing severe COVID-19 or experiencing adverse consequences of the restrictions. Recruitment was conducted through paid and unpaid social media advertisements and flyers promoted through community and industry groups, community-based organisations, and social and professional networks. Further, bilingual research staff were recruited to engage with people who spoke Mandarin, Arabic and Dinka, and a written study advertisement was produced in Mandarin and Arabic. Participants could complete surveys in English or in these languages. A full description of Optimise methods can be found elsewhere.17–20 Ethics approval for Optimise was provided by The Alfred Human Research Ethics Committee (Approval Number 333/20). At recruitment, Optimise participants completed a baseline survey providing sociodemographic information, information about life before the COVID-19 pandemic and changes experienced since the pandemic. The analysis for this study was cross-sectional, using data from participants’ baseline surveys.
Primary outcome
The primary outcome was healthcare access. The baseline survey asked participants ‘How easy was it for you to access the healthcare you needed in 2019?’ and ‘How easy was it for you to access the healthcare you needed in the past three months?’. The response option categories to both questions were ‘very difficult’, ‘somewhat difficult’, ‘somewhat easy’, ‘very easy’ and ‘not applicable (did not access healthcare in the past three months)’. For descriptive analysis and graphical presentation of the data, the five categories were retained. For the purposes of Chi-square tests and regression modelling, categories were combined into a binary variable of difficulty accessing healthcare in the three months prior to the baseline survey (yes/no). ‘Very difficult’ and ‘somewhat difficult’ were assigned to yes (difficulty), and ‘somewhat easy’ and ‘very easy’ were assigned to no (difficulty). ‘Not applicable’ was excluded from this variable. ‘Difficulty accessing healthcare’ was the outcome of interest. Demographic variables from the baseline survey were also included in the analysis (refer to Appendix 1; available only online).
Secondary outcomes
Four secondary outcomes were explored by descriptive analysis: (1) the number of times participants accessed healthcare services in 2019; (2) the number of times participants accessed healthcare services in the three months prior to the baseline survey; (3) changes experienced in healthcare access; and (4) concerns related to usual healthcare (Appendix 2; available online only).
Recall periods
Because the recruitment into Optimise occurred over a 16-month period (September 2020–December 2021), participants completed the baseline survey at different time points, which meant that the ‘past three months’ was relative to their own baseline survey date. Therefore, the recall period for the baseline survey responses spanned from June 2020 (three months prior to September 2020) to December 2021.
Lockdown periods
The 16-month baseline recruitment period, and therefore recall periods, include five lockdown periods. To explore the association between the key outcomes and lockdown periods, we derived a covariate (lockdown). We included lockdowns 2–6 (detailed time periods available in Appendix 3; available online only) in descriptive analysis as a categorical variable (lockdown 2, 3, 4, 4 and 5, 5, 5 and 6, 6, no lockdown) (refer to Appendix 4; available online only). Lockdown 1 (31 March 2020–12 May 2020) was not included as it was outside the recall period. These categories were then collapsed into one binary variable representing lockdown (yes/no), which was included in the logistic regression model as a model covariate to adjust for confounding.
Statistical methods
Descriptive analysis was conducted on the sociodemographic factors of the study sample, reported as frequencies. To summarise the frequency of accessing healthcare by sociodemographic factors, we reported the mean (total number of times each participant accessed healthcare divided by the total number of participants who accessed healthcare) and standard deviation (SD) of the number of times healthcare services were accessed in 2019 and in the three months prior to participant baseline survey. Descriptive statistics were generated to summarise access to healthcare, changes experienced, alternatives offered, satisfaction with telehealth and concerns.
Chi-square tests were used to test for differences in sociodemographic factors by the outcome of difficulty accessing healthcare among participants who had accessed healthcare during the COVID-19 pandemic. Exposures for the logistic regression modelling were selected from the results of Chi-square tests. Further selection of the best subset of confounders for the association between exposures and the outcome of difficulty accessing healthcare were explored by using directed acyclic graphs (DAGs). The DAGs were developed based on prior literature and theoretical understanding of sociodemographic determinants of healthcare access. Variables identified as common causes of both exposure and outcome, such as age and gender, were adjusted for in the multivariable logistic regression model. Because the DAGs also suggested each exposure required a unique covariate set (refer to Appendices 5–7; available online only), we conducted four separate multivariable logistic regression models to estimate the association between each of the four exposures (selected by the univariable analysis described above): (1) chronic illness/es; (2) household income; (3) employment status; and (4) main language spoken at home) and outcome of difficulty accessing healthcare in the three months prior to the baseline survey. Participants who had missing data for exposure, confounders or outcomes were excluded from analysis.
All statistical analyses were conducted using R version 4.1.3 (R Core Team, Vienna, Austria).21
Results
Participants
In total, 779 participants completed the baseline survey – 551 women (70.7%) and 219 men (28.1%). The mean age of participants was 44 years (SD=16.8), most lived in metropolitan Melbourne (80.6%), 19.1% were classified as high-risk workers (including aged care workers, hotel quarantine or COVID-19 border control workers, and healthcare workers), 26.8% reported having chronic illness/es and 18.1% spoke a language other than English at home (Table 1).
| Table 1. Description of the sociodemographic factors of Optimise study participants, in Victoria, Australia (N=779) |
| Sociodemographic factors |
n |
% |
| Age (years) |
Mean=44.0
Median=42.9 |
SD=16.8
IQR=29–58 |
| Missing |
4 |
|
| Age group (years) |
|
|
| 18–24 |
137 |
17.6 |
| 25–34 |
155 |
19.9 |
| 35–44 |
121 |
15.5 |
| 45–54 |
131 |
16.8 |
| 55–64 |
127 |
16.3 |
| 65+ |
104 |
13.4 |
| Missing |
4 |
0.5 |
| Gender |
|
|
| Man |
219 |
28.1 |
| Woman |
551 |
70.7 |
| OtherA |
6 |
0.8 |
| Missing |
3 |
0.4 |
| Chronic illness |
|
|
| Yes |
209 |
26.8 |
| No |
558 |
71.6 |
| Missing |
12 |
1.5 |
| High-risk workerB |
|
|
| Yes |
149 |
19.1 |
| No |
629 |
80.7 |
| Missing |
1 |
0.1 |
| Education |
|
|
| High school and less |
143 |
18.4 |
| Tertiary education – TAFE/trade certificate |
126 |
16.2 |
| Tertiary education – undergraduate |
290 |
37.2 |
| Tertiary education – postgraduate |
213 |
27.3 |
| Missing |
7 |
0.9 |
| Residential location |
|
|
| Metro |
628 |
80.6 |
| Region |
144 |
18.5 |
| Missing |
7 |
0.9 |
| Employment status |
|
|
| Full time |
222 |
28.5 |
| Part time |
140 |
18.0 |
| Casual |
129 |
16.6 |
| Self-employed |
34 |
4.4 |
| Full-time carer/home responsibilities |
14 |
1.8 |
| Retired |
107 |
13.7 |
| Not employed |
105 |
13.5 |
| Missing |
28 |
3.6 |
| Household income (per year) |
|
|
| $1–$49,999 |
218 |
28.0 |
| $50,000–$99,999 |
191 |
24.5 |
| $100,000–$149,999 |
106 |
13.6 |
| $150,000 or more |
138 |
17.7 |
| No income |
20 |
2.6 |
| Missing |
106 |
13.6 |
| Main language spoken at home |
|
|
| English |
638 |
81.9 |
| LOTE |
141 |
18.1 |
ANon-binary/genderfluid, trans men and trans women were grouped together because of small numbers.
BHigh-risk workers include aged care workers, hotel quarantine or COVID-19 border control workers, and healthcare workers.
IQR, interquartile range; LOTE, language other than English; SD, standard deviation; TAFE, Technical and Further Education. |
Healthcare access
Of the 779 participants at baseline, 699 (89.7%) reported accessing healthcare services throughout 2019, on average 8.3 times (SD=10.7). A total of 766 (98.3%) participants reported accessing healthcare services in the three months prior to their baseline survey (recall period spanning June 2020 to December 2021), and did so on average 3.3 times (SD=4.2) across the three-month recall period (Appendix 8; available online only). Comparing healthcare access across two time periods (2019 compared to the three months prior to their baseline survey), the proportion of participants who reported it was ‘very easy’ to access healthcare reduced from 67.7% to 37.2%, whereas the proportion reporting it was ‘somewhat difficult’ and ‘very difficult’ increased from 3.7% to 16.6% (Table 2, Appendix 9; available online only). Participants whose 3-month recall period included lockdowns 4 and 5, lockdowns 5 and 6 and lockdown 6 were more likely to report difficulty in accessing healthcare than those whose 3-month recall period included other lockdown periods or were outside of lockdown (Figure 1).
| Table 2. Summary of reported healthcare access, changes to, alternatives and concerns about healthcare access (N=779) |
| |
In 2019 |
In the 3 months prior to baseline (June 2020– December 2021) |
| Healthcare access |
|
|
| Very difficult |
5 (0.6) |
28 (3.6) |
| Somewhat difficult |
24 (3.1) |
101 (13.0) |
| Somewhat easy |
182 (23.4) |
230 (29.5) |
| Very easy |
527 (67.7) |
290 (37.2) |
| Not applicable (did not access healthcare) |
29 (3.7) |
122 (15.7) |
| Missing |
12 (1.5) |
8 (1.0) |
| Access or utilisation of healthcare changed (n=657A) |
| Yes |
– |
351 (53.4) |
| No |
– |
279 (42.5) |
| Do not know |
– |
27 (4.1) |
| Changes experienced (N=351B) |
|
|
| My face-to-face appointments were rescheduled to telehealth appointments |
– |
303 (86.3) |
| My appointment(s) were cancelled/postponed by the clinic |
– |
105 (29.9) |
| I could not access my usual healthcare |
– |
46 (13.1) |
| I ran out of medication |
– |
32 (9.1) |
| I could not contact my usual healthcare provider |
– |
29 (8.3) |
| My elective surgery/procedure was delayed or cancelled |
– |
28 (8.0) |
| Other |
– |
38 (10.8) |
| Offered alternatives to usual appointments or services (N=351C) |
| Yes |
– |
284 (80.9) |
| No |
– |
55 (15.7) |
| Do not know |
– |
12 (3.4) |
| Alternatives offered (N=284D) |
|
|
| Telehealth consultation(s) |
– |
277 (97.5) |
| Online prescription services |
– |
103 (36.3) |
| Repeated prescriptions/multi-month medication |
– |
39 (13.7) |
| In-home services |
– |
3 (1.1) |
| Other |
– |
4 (1.4) |
| Satisfaction with telehealth (N=277E) |
|
|
| Very satisfied |
– |
92 (33.2) |
| Somewhat satisfied |
– |
115 (41.5) |
| Neutral |
– |
32 (11.6) |
| Somewhat dissatisfied |
– |
33 (11.9) |
| Very dissatisfied |
– |
5 (1.8) |
| Concerns |
|
|
| I delayed or avoided seeking care |
– |
190 (24.4) |
| I felt anxious about attending because of COVID-19 |
– |
187 (24.0) |
| I could not afford to access my usual healthcare |
– |
41 (5.3) |
| I could not access telehealth consultation(s) |
– |
19 (2.4) |
| No concerns in relation to my usual healthcare |
– |
347 (44.5) |
| Other |
– |
11 (1.4) |
Data are presented as n (%). –, not applicable (ie the question was not asked).
AThis question was not shown for participants who chose ‘Not applicable (did not access healthcare)’ for healthcare access.
BThis question only showed to participants who chose ‘Yes’ to ‘Access or utilisation of healthcare changed’.
CThis question only showed to participants who chose ‘Yes’ to ‘Access or utilisation of healthcare changed’.
DThis question only showed to participants who chose ‘Yes’ to ‘Offered alternatives to usual appointments or services’.
EThis question only showed to participants who used telehealth consultation(s) as alternatives. |
Figure 1. Reported difficulty accessing healthcare during the COVID-19 pandemic (in the 3 months prior to baseline completion),A–C Victoria, Australia (N=779).
A Lockdown 2: 9 July 2020–27 October 2020; Lockdown 3: 13 February 2021–17 February 2021; Lockdown 4: 29 May 2021–10 June 2021; Lockdown 5: 16 July 2021–27 July 2021; Lockdown 6: 5 August 2021–21 October 2021.
BNot applicable: did not access healthcare in the past three months.
CThree-month recall period: June 2020 to December 2021.
Changes experienced in healthcare access
Among 657 participants who accessed healthcare during the COVID-19 pandemic (recall period spanning June 2020 to December 2021), 53.4% experienced changes in access or utilisation of healthcare in that period (Table 2 and Appendix 9). Participants whose recall period included lockdowns 5 and 6 were more likely to report changes in healthcare access or utilisation than those whose recall period was outside of lockdown, and the proportion among participants whose recall period included other lockdown periods was lower again (Figure 2). Among participants who experienced changes in healthcare access or utilisation during the COVID-19 pandemic, the most frequent two changes reported were ‘My face-to-face appointments were rescheduled to telehealth appointments’ (86.3%) and ‘My appointment(s) were cancelled/postponed by the clinic’ (29.9%). A higher proportion of participants whose recall period was during lockdowns 2, 4 and 5, 5 and 6, and 6 reported ‘I couldn’t access my usual healthcare’ than those whose recall period was outside of lockdown (Figure 3). Furthermore, among 351 participants who experienced changes in access or utilisation of healthcare during the COVID-19 pandemic, 284 (80.9%) were offered alternatives to their usual appointments or services. Almost all (97.5%) of these participants chose telehealth as their alternative, and their satisfaction rate was high (33.2% ‘very satisfied’ and 41.5% ‘somewhat satisfied’).
All participants completed questions about concerns related to healthcare. Nearly one-quarter of the 779 participants reported ‘I delayed or avoided seeking care’ and ‘I felt anxious about attending due to COVID-19’ (Table 2).
Figure 2. Proportion of participants experiencing changes in healthcare access or utilisation during the COVID-19 pandemic (in the 3 months prior to baseline completionA,B), in Victoria, Australia (N=630C).
ALockdown 2: 9 July 2020–27 October 2020; Lockdown 3: 13 February 2021–17 February 2021; Lockdown 4: 29 May 2021–10 June 2021; Lockdown 5: 16 July 2021–27 July 2021; Lockdown 6: 5 August 2021–21 October 2021.
BThree-month recall period: June 2020 to December 2021.
COf 779 participants, 149 chose ‘don’t know’. They were excluded from this figure.
Figure 3. Changes experienced in healthcare access or utilisation during the COVID-19 pandemic (in the 3 months prior to baseline completion),A,B in Victoria, Australia (N=351C).
ALockdown 2: 9 July 2020–27 October 2020; Lockdown 3: 13 February 2021–17 February 2021; Lockdown 4: 29 May 2021–10 June 2021; Lockdown 5: 16 July 2021–27 July 2021; Lockdown 6: 5 August 2021–21 October 2021.
BThree-month recall period: June 2020 to December 2021.
CThis question was only shown to participants who chose ‘Yes’ to ‘Access or utilisation of healthcare changed’.
Univariate analysis
Chi-square tests identified significant associations between chronic illness/es, employment status, household income and main language spoken at home, and the outcome of difficulty in accessing healthcare. Table 3 shows that higher proportions of people with chronic illness/es or lower income or who spoke a language other than English at home had difficulty accessing healthcare than other participants. Higher proportions of people who were full-time carers/had home responsibilities (54.5%), were not employed (29.5%), had part-time jobs (23.9%) or casual jobs (20.4%) had more difficulty accessing healthcare services than other employment types.
| Table 3. Univariate analysis for sociodemographic factors related to difficulty accessing healthcare in the 3 months prior to baseline surveyA (Optimise, Victoria, Australia, 2020–21, N=649)B |
| Sociodemographic factors |
Total |
Difficulty accessing healthcare |
| Yes, n=129 (19.9%) |
No, n=520 (80.1%) |
P valueC |
| Age-group (years) |
|
|
|
0.39 |
| 18–24 |
110 |
27 (24.5) |
83 (75.5) |
|
| 25–34 |
121 |
22 (18.2) |
99 (81.8) |
|
| 35–44 |
100 |
25 (25.0) |
75 (75.0) |
|
| 45–54 |
105 |
17 (16.2) |
88 (83.8) |
|
| 55–64 |
110 |
22 (20.0) |
88 (80.0) |
|
| 65+ |
100 |
16 (16.0) |
84 (84.0) |
|
| Gender |
|
|
|
0.69 |
| Man |
162 |
36 (22.2) |
126 (77.8) |
|
| Woman |
479 |
92 (19.2) |
387 (80.8) |
|
| OtherD |
6 |
1 (16.7) |
5 (83.3) |
|
| Chronic illness |
|
|
|
0.007 |
| Yes |
202 |
54 (26.7) |
148 (73.3) |
|
| No |
436 |
75 (17.2) |
361 (82.8) |
|
| High-risk workerE |
|
|
|
0.46 |
| Yes |
128 |
29 (22.7) |
99 (77.3) |
|
| No |
520 |
100 (19.2) |
420 (80.8) |
|
| Education |
|
|
|
0.41 |
| High school and less |
120 |
21 (17.5) |
99 (82.5) |
|
| Tertiary education – TAFE/trade certificate |
105 |
25 (23.8) |
80 (76.2) |
|
| Tertiary education – undergraduate |
235 |
49 (20.9) |
186 (79.1) |
|
| Tertiary education – postgraduate |
182 |
30 (16.5) |
152 (83.5) |
|
| Residential location |
|
|
|
1 |
| Metro |
514 |
102 (19.8) |
412 (80.2) |
|
| Region |
128 |
26 (20.3) |
102 (79.7) |
|
| Employment status |
|
|
|
<0.001 |
| Full time |
173 |
18 (10.4) |
155 (89.6) |
|
| Part time |
117 |
28 (23.9) |
89 (76.1) |
|
| Casual |
108 |
22 (20.4) |
86 (79.6) |
|
| Self-employed |
29 |
5 (17.2) |
24 (82.8) |
|
| Full-time carer/home responsibilities |
11 |
6 (54.5) |
5 (45.5) |
|
| Retired |
98 |
14 (14.3) |
84 (85.7) |
|
| Not employed |
88 |
26 (29.5) |
62 (70.5) |
|
| Household income (per year) |
|
|
|
0.018 |
| $1–$49,999 |
182 |
47 (25.8) |
135 (74.2) |
|
| $50,000–$99,999 |
151 |
25 (16.6) |
126 (83.4) |
|
| $100,000–$149,999 |
95 |
15 (15.8) |
80 (84.2) |
|
| $150,000 or more |
113 |
14 (12.4) |
99 (87.6) |
|
| No income |
15 |
5 (33.3) |
10 (66.7) |
|
| Main language spoken at home |
|
|
|
<0.001 |
| English |
542 |
91 (16.8) |
451 (83.2) |
|
| LOTE |
107 |
38 (35.5) |
69 (64.5) |
|
Data are presented as n (%).
AThree-month recall period: June 2020 to December 2021.
B122 participants responded ‘Not applicable (did not access healthcare)’ and eight participants responded ‘Prefer not to say’. They were excluded from the analysis.
CThis is an overall Chi-square test for independence.
DNon-binary/genderfluid, trans men and trans women were grouped together because of small numbers.
EHigh-risk workers include aged care workers, hotel quarantine or COVID-19 border control workers, and healthcare workers.
LOTE, language other than English; TAFE, Technical and Further Education. |
Multivariable logistic regression analysis
A total of 649 participants accessed healthcare in the three months prior to their baseline survey and had complete data on all covariates; therefore, they were included in the regression analyses. Of these, 129 (19.9%) reported that it was difficult to access healthcare in the three months prior to their baseline survey (Table 3). Four separate multivariable logistic regression models estimated the associations between exposures of chronic illness, household income, employment status and main language spoken at home, and the outcome of difficulty accessing healthcare after adjusting for multiple factors (Table 4). Results from adjusted models showed that participants with chronic illness/es had over twice the odds of reporting difficulty accessing healthcare relative to participants without chronic illness/es. Additionally, people who earned $1–$49,999 per year had more than twice the odds of reporting difficulty accessing healthcare relative to people earning $150,000 or more per year. Compared to people who spoke English at home, people who spoke a language other than English had 2.69-fold (95%CI: 1.37–5.27, P=0.004) the odds of reporting difficulty accessing healthcare. Employment status was not associated with difficulty accessing healthcare after adjusting for age, gender, education, residence location, household income, chronic illness/es and lockdown.
| Table 4. Association between sociodemographic variables and difficulty accessing healthcare in the 3 months prior to baseline survey;A results of multivariable logistic regression analysis (Optimise, Victoria, Australia, N=649B) |
| Variables |
Unadjusted |
AdjustedC |
| OR (95% CI) |
P value |
OR (95% CI) |
P value |
| Model 1: chronic illness |
|
|
|
|
| No |
Ref |
|
Ref |
|
| Yes |
1.76 (1.18–2.61) |
0.006 |
2.06 (1.34–3.17) |
<0.001 |
| Model 2: household income (per year) |
|
|
|
|
| $150,000 or more |
Ref. |
|
Ref. |
|
| $100,000–$149,999 |
1.33 (0.60–2.94) |
0.48 |
1.24 (0.54–2.89) |
0.61 |
| $50,000–$99,999 |
1.40 (0.70–2.90) |
0.35 |
1.27 (0.61–2.77) |
0.53 |
| $1–$49,999 |
2.46 (1.31–4.87) |
0.007 |
2.27 (1.12–4.84) |
0.027 |
| No income |
3.54 (0.99–11.60) |
0.041 |
2.97 (0.79–10.37) |
0.09 |
| Model 3: employment status |
|
|
|
|
| Not employed |
Ref |
|
Ref |
|
| Full time |
0.28 (0.14–0.54) |
<0.001 |
0.41 (0.17–1.03) |
0.06 |
| Part time |
0.75 (0.40–1.40) |
0.37 |
0.99 (0.43–2.31) |
0.98 |
| Casual |
0.61 (0.31–1.17) |
0.14 |
0.72 (0.31–1.68) |
0.45 |
| Self-employed |
0.50 (0.15–1.36) |
0.20 |
0.47 (0.12–1.56) |
0.24 |
| Full-time carer/home responsibilities |
2.86 (0.80–10.74) |
0.11 |
3.12 (0.69–14.21) |
0.13 |
| Retired |
0.40 (0.19–0.81) |
0.013 |
0.40 (0.13–1.21) |
0.11 |
| Model 4: main language spoken at home |
|
|
|
|
| English |
Ref |
|
Ref |
|
| LOTE |
2.73 (1.72–4.29) |
<0.001 |
2.69 (1.37–5.27) |
0.004 |
AThree-month recall period: June 2020 to December 2021.
B122 participants responded ‘Not applicable (did not access healthcare)’ and eight participants responded ‘Prefer not to say’. They were excluded from the analysis.
CFor chronic illness, the regression model adjusted for age and gender. For household income, the regression model adjusted for gender, education, chronic illness and residence location. For employment status, the regression model adjusted for age, gender, education, residence location, lockdown, chronic illness and household income. For main language spoken at home, the regression model adjusted for age, gender, chronic illness, high-risk worker, education, residential location, employment status, lockdown and household income.
CI, confidence interval; LOTE, language other than English; OR, odds ratio; Ref, reference. |
Discussion
The key finding from this study of 779 highly characterised Victorian adults was that one in five participants had difficulty accessing healthcare services during the COVID-19 pandemic, particularly during longer lockdown periods. Critically, individuals with chronic illness/es, those on a low income or who spoke a language other than English at home had greater difficultly accessing healthcare services. More than half of the participants, and more women than men, reported experiencing changes in access to their usual healthcare during the COVID-19 pandemic. This study also provided key insights into the community’s concerns about accessing healthcare, with participants reporting they intentionally delayed or avoided care.
Barriers to healthcare access
The proportion of participants reporting difficulty in accessing healthcare during the COVID-19 pandemic was four-fold that measured pre-pandemic. This finding was consistent with those of previous studies, which found that even when health facilities remained open, the COVID-19 pandemic reduced the accessibility of healthcare services.22–24 Notably, the obstacles to accessing healthcare were most pronounced during longer lockdown periods, in alignment with previous studies.23,25
Importantly, our study identified key subgroups disproportionately affected in accessing healthcare. They included individuals with chronic illness/es, those who spoke a language other than English at home, and people with financial difficulties. A previous study of Australian adults26 also identified that individuals with ongoing illness/es were more inclined to refrain from seeking healthcare services during the COVID-19 pandemic because of concerns about contracting COVID-19 and their higher risk of severe disease. For CALD communities, the findings of this study might be caused by the delayed implementation or inaccurate translation of public health messages during the COVID-19 pandemic, which was also the case in a US study.27 This study’s finding that related to lower income was aligned with a study conducted in low- and middle-income countries whereby difficulty affording care because of the economic impacts of the COVID-19 pandemic was one of the key barriers to healthcare access.28 Two Australian studies focusing on families with children29,30 also highlighted the significant financial impact of the COVID-19 pandemic on families, which could potentially affect their ability to access healthcare services. Although employment status was associated with access difficulties in unadjusted models, this relationship was attenuated after accounting for income, suggesting that financial circumstances plays a more direct role than employment itself. In the event of future COVID-19 waves or another pandemic, it is important to develop processes to ensure equitable access to healthcare, particularly during lockdowns. This might include tailored guidance for individuals known to access healthcare frequently (ie those with chronic health conditions), specific timely messaging about how to access health services translated for CALD communities,27 and a structuring of services so they can be accessed by people with financial difficulties.
Changes experienced in healthcare utilisation
This study found that over half of participants who accessed healthcare during the COVID-19 pandemic experienced changes in healthcare access or utilisation. Changes such as appointments being cancelled/postponed, inability to access usual healthcare and elective surgery being delayed/cancelled were all consistent with COVID-19 restrictions.31 Because of the restrictions on unnecessary travel, public transport limitations and mandatory isolation/quarantine following infection/exposure, people experienced difficulty accessing their usual healthcare during the COVID-19 pandemic.31,32 At the start of the COVID-19 pandemic, non-urgent elective surgery was suspended for one month in Australia, and remained restricted for months thereafter in some regions and for various periods.33 A narrative review of Victoria’s surgical response highlighted that the number of patients on the planned surgery waiting list, as well as waiting times for surgery, doubled between March 2020 and March 2022.34
This study further indicated that among individuals who experienced changes in healthcare access or utilisation, the predominant course of action involved rescheduling appointments to telehealth. Concordant with the data presented in an AIHW report,33 23% of all Medicare services in 2020 were delivered through telehealth. Particularly in Victoria, there was a discernible escalation in the utilisation of telehealth services, with the state experiencing the highest rate of telehealth use (37.1% of Medicare services) of all states and territories. This was probably attributable to the prolonged and recurrent imposition of public health restrictions in Victoria.35 Consistent with the Australia Health Consumer Sentiment Survey,36 our participants reported high satisfaction with telehealth. It provided convenient and timely access to healthcare services that did not require physical contact37 and reduced the risk of contracting COVID-19 when travelling, especially for those living in rural or remote communities.35 It is worth noting, however, that 13.7% of participants expressed dissatisfaction with telehealth services. A previous study by Javanparast et al about the pandemic-era telehealth experiences of Australian general practice patients at high risk of poor health outcomes highlighted the obstacles these individuals faced in effectively using telehealth services.37 These challenges encompassed difficulties in articulating their health concerns and clinicians being unable to conduct physical examinations. Further, Javanparast et al highlighted that the option of telephone (as opposed to video consultation) was important for their participants because some have poor digital literacy.37 Therefore, for future healthcare preparedness, telehealth services should be extended and improved; patient triage could determine if telehealth consultation is appropriate,38 clinics’ telehealth capacity could be enhanced by upgrading their platforms and improving professionals’ skills,38 and flexible modes of telehealth could be developed to account for varying levels of comfort or access to technology.
Strengths and limitations
This study has many strengths. First, it used detailed individual-level quantitative data to assess the association between sociodemographic factors and healthcare access during the COVID-19 pandemic in Australia. Second, our study participants were recruited because they were at heightened risk of COVID-19 infection, severe COVID-19, or unintended consequences of COVID-19 restrictions. Thus, we have presented critical data on healthcare access from those members of the community most in need of healthcare. Although complete case analysis was used, there was minimal missing data; therefore, the results will have minimal selection bias.
Some limitations were encountered in this study. First, given our deliberate recruitment strategy of sampling high-risk groups, our sample is not representative of the wider population in Victoria. This might have increased the likelihood of our participants already having a higher demand for healthcare services and subsequently encountering more difficulties. However, as outlined above, this could be considered a strength. Second, over the 16-month baseline recruitment period, policies around accessing healthcare services changed, potentially affecting our outcomes. This includes the roll-out of the first COVID-19 vaccines, which began in February 2021. Although most of our data were collected before large uptake of the vaccine, and vaccines were available outside healthcare settings (ie community-based vaccinations centres, workplaces), participants reported healthcare access might have been related to accessing vaccine/s. Third, it is important to highlight that this study used self-report data, and the responses could be subject to recall bias and social desirability bias. Participants might have over- or under-estimated their access to healthcare, and this misclassification might mean we missed important relationships in the data and so the results should be interpreted with some caution. We recommend future studies collect outcome data using self-report alongside administrative data. Fourth, participants might count COVID-19 testing as accessing healthcare, which could result in an over-estimate of the frequency of healthcare access. Fifth, this study used a cross-sectional design, so did not allow for causal interpretations, and the observed associations might be subject to residual confounding. Additionally, the cell sizes of some strata of confounders were small (ie ≤10) with associated lack of precision in the confidence intervals around model estimates; therefore, caution should be used when interpreting model results. Also, the inclusion of a binary variable for ‘lockdown’ in the model might have meant some loss of information (eg it does not account for time spent in lockdown). Finally, our study was not well placed to examine geographical disparities in healthcare access because of the sample being mainly participants residing in a metropolitan area.
Conclusion
This study demonstrated that adults in Victoria, Australia, had difficulty in accessing healthcare services during the COVID-19 pandemic, especially during longer lockdown periods. Most participants experienced changes in healthcare utilisation. Individuals who had chronic illness/es or low income or spoke a language other than English at home had increased odds of reporting difficulty accessing healthcare services. Future pandemic planning should ensure healthcare is universally accessible, especially for the most vulnerable members of our community.