Future Burden of Ischemic Stroke in Australia: Impact on Health Outcomes between 2019 and 2038

Abstract

Background

Projections of the future burden of ischemic stroke (IS) has not been extensively reported for the Australian population; the availability of such data would assist in health policy planning, clinical guideline updates, and public health.

Methods

First, we estimated the lifetime risk of IS (from age 40 to 100 years) using a multistate life table model. Second, a dynamic multistate model was constructed to project the burden of IS for the whole Australian population aged between 40 and 100 years over a 20-year period (2019–2038). Data for the study were primarily sourced from a large, representative Victorian linked dataset based on the Victorian Admitted Episode Dataset and National Death Index. The model projected prevalent and incident cases of nonfatal IS, fatal IS, and years of life lived (YLL) with and without IS. The YLL outcome was discounted by 5% annually; we varied the discounting rate in scenario analyses.

Results

The lifetime risk of IS from age 40 years was estimated as 15.5% for males and 14.0% for females in 2018. From 2019 to 2038, 644,208 Australians were projected to develop incident IS (564,922 nonfatal and 79,287 fatal). By 2038, the model projected there would be 358,534 people with prevalent IS, 35,554 people with incident nonfatal IS and 5,338 people with fatal IS, a 14.2% (44,535), 72.9% (14,988), and 106.3% (2,751) increase compared to 2019 estimations, respectively. Projected YLL (with a 5% discount rate) accrued by the Australian population were 174,782,672 (84,251,360 in males and 90,531,312 in females), with 4,053,794 YLL among people with IS (2,320,513 in males, 1,733,281 in females).

Conclusion

The burden of IS was projected to increase between 2019 and 2038 in Australia. The outcomes of the model provide important information for decision-makers to design strategies to reduce stroke burden.

Introduction

Stroke is one of the leading causes of disease burden worldwide, accounting for 11.6% of deaths globally in 2019 [1]. Further, as life expectancy increases, more people are expected to survive to ages where stroke risk is highest [2–4]. Indeed, a European study projected an overall 3% increase in incident stroke (40,000 additional cases), a 27% increase in prevalent stroke (2.6 million additional prevalent cases), and a 17% decline in mortality associated with stroke (80,000 fewer deaths) in Europe by 2047 when compared with 2017 estimates [5]. In the USA, prevalent cases of stroke are expected to increase by nearly 4% (3.4 million additional people) in 2030 relative to 2012 [6]. In Australia, there were approximately 450,000 prevalent cases of stroke, 68,000 hospitalizations (0.6% of total hospitalizations) and 8,700 deaths (5.4% of total deaths) in 2020 [7, 8]. However, it is unclear how the burden in Australia is expected to change in the future.

Therefore, we have developed a model to estimate the lifetime risk of developing ischemic stroke (IS) [9] and projected the incidence, prevalence, and years of life lived (YLL) with and without IS. The findings of this study will provide crucial information about the expected impact of IS in the future to relevant stakeholders, facilitating informed decision making, and prioritization of strategies for disease prevention and management.

Methods

Model Overview

A dynamic multistate model (shown in Fig. 1) was designed to project the burden of IS for the Australian population aged 40–100 years from 2019 to 2038 in yearly cycles. Model outcomes included the prevalence of IS, the incidence of IS (fatal and nonfatal), and YLL with and without IS (discounted at 5% per year). Discounting was applied to consider society’s preference in consuming a product or benefit (in this example: YLL) at present rather than delaying the same consumption until sometime in the future [10, 11]. Hence, present benefits were given more value than future benefits using a technique known as discounting [10, 11]. The formula depicted below was applied to calculate the discounted YLL [11]. A static version of this multistate life table model was also used to estimate the lifetime risk of developing IS for the population of Victoria, Australia, in 2018.

$$Presentvalue=\frac{Futurevalue}{{\left(1+r\right)}^t}$$

where r is the discount rate (e.g., 5%, 3%, or 0%) and t is the number of years into the future the future value (i.e., YLL) occurs.

Fig. 1.

Structure of the multistate life table model and associated transition probabilities. IS, ischemic stroke; ℓ, 1-year period; λ, ischemic stroke incidence rate; µ_other, mortality rate for people without ischemic stroke (i.e., other mortality); µ_post-IS, post-IS all-cause mortality rate; p_fatal, proportion of fatal ischemic stroke; ratesum, λ+ µ_other; tpsum, 1-exp(-ratesum* ℓ); ratesum, the sum of incidence rate of ischemic stroke and mortality rate for people without ischemic stroke (i.e., other mortality); tpsum, transition probability of ratesum.

Model Structure

The model was constructed with four health states: “alive no-IS,” “alive with new-IS,” “alive after surviving IS (i.e., post-IS),” and “death.” The death health state could either be due to causes not related to IS (i.e., “Death other”), due to IS (“fatal-IS”), or death after surviving IS (“Death Post-IS”). From the “alive no-IS” health state, people could move to either “alive with new-IS” health state, “Dead other” health state (absorbing state), or remain in the “alive no-IS” health state. The “alive new-IS” health state was transient only – people moved out of this health state within the same cycle depending on whether the IS was fatal or non-fatal. For nonfatal IS, people moved immediately to the “post-IS” health state; for fatal IS, people moved immediately to the “fatal IS” health state (absorbing state). Once people reached the “post-IS” health state, they could either die and move to the “Death Post-IS” health state (absorbing state) or remain in the “post-IS” health state.

Population

The dynamic model population was based on the Australian population in 2018 (see online suppl. Table s1; for all online suppl. material, see https://doi.org/10.1159/000538800). Age- and sex-specific Australian population data were derived from the Australian Bureau of Statistics (ABS) [12]. This population was then aged in yearly cycles from 2019 to 2038. Migrants were added to the model at the beginning of each cycle [13], as were people who reached the model age range (i.e., the cohort aged from 39 to 40 years each year).

Data Sources

Data for the model transition probabilities were primarily sourced from a Victorian linked dataset, which has been previously described [14, 15]. Briefly, this linked dataset was based on the Victorian Admitted Episode Dataset (VAED), which records all hospital admissions in Victoria. The VAED was linked to the National Death Index (NDI), which records all deaths that occur in Australia. The dataset contained all admissions for IS (defined by the presence of ICD-10 codes I63–I64 as the primary diagnosis for an admission) among people aged ≥40 years in the state of Victoria between July 1, 2012, and June 30, 2017 (with follow-up data until June 30, 2018). In addition to the linked dataset, age-group and sex-specific prevalence of IS and all-cause mortality data in the general population were sourced from the Australian Institute of Health and Welfare (AIHW) [16] and ABS [17], respectively, for the year 2018 (online suppl. Table s2).

Estimating Transition Probabilities in the Model

Stroke prevalence data were only available in broad age groups; thus, we estimated the single-year age- and sex-specific prevalence of stroke in 2018 using beta regression [18, 19]. The model outcome was the prevalence of stroke (expressed as a proportion), the predictor a log-linear effect of age (where age was the mid-point of each age group available), and used a log link function. The predicted prevalence of stroke was then multiplied by 0.87, assuming 87% of strokes were ischemic [9]. Then, these values were used to derive the prevalence of IS in 2018 for the initial model population. Alternative sources of data for IS were implemented in the scenario analysis.

All-cause mortality, incidence of IS, and post-IS mortality data were also only available in age groups; therefore, we estimated the single-year age- and sex-specific rate of each of these outcomes for their respective data in 2018 using Poisson regression [20]. A separate model was fit for each outcome, the predictor for all models was log-linear effect of age (where age was the mid-point of each age group available), used a log link function, and the offset (all log) was general population of Victoria (for all-cause mortality), people without IS (for incidence of IS) and person-years of follow-up of people who survived IS (for estimation of post-IS mortality).

To estimate the proportion of nonfatal IS and fatal IS by single-year age and sex, a beta regression model was fitted [18, 19]. The model included the mid-point of each age group as a log-linear predictor and used a log link function. The estimated values derived from these models were then used to calculate the number of people who did not survive IS.

The number of “other deaths” (i.e., deaths in people who had not experienced IS, which was used to estimate the mortality rate in people without IS) was calculated using the estimated incidence rate of IS, all-cause mortality rate, post-IS mortality rate and proportion of fatal IS from the above models for each sex and single-year age, using the formula depicted below [21, 22].

$${\text{Death}}_{other}={\mu }_{pop}*\left(Po{p}_{IS}+Po{p}_{without-IS}\right)-\left({\mu }_{post-IS}*Po{p}_{IS}+p_{fatal-Is}*\lambda *Po{p}_{without-IS}\right)$$

where Death_other is the number of deaths without developing IS, µ_pop: the estimated all-cause mortality rate, Pop_IS: people with IS; Pop_without-IS: people without IS; µ_post-IS: estimated post-IS mortality rate; p_fatal-IS: estimated proportion of fatal IS; λ: estimated incidence rate of IS.

Once the number of “other deaths” was calculated, Poisson regression was used to estimate the mortality rate for people without IS in 2018, with the predictor a log-linear effect of age, people without IS the offset, and a log link function [20]. A separate model was fit for each sex.

Model Outcomes

These transition probabilities populated both the static and dynamic models described above (Fig. 1). The outcomes captured for the static model was lifetime risk of IS from ages 40 to 100 years. Outcomes from the dynamic model included projected prevalent IS, nonfatal IS, fatal IS, and YLL. To understand the impact of stroke in the population, the main outcomes such as YLL were captured for two cohorts, YLL was calculated for people without IS (YLL without IS) and for people with IS (YLL with IS), with half cycle correction.

Sensitivity and Scenario Analyses

To derive 95% uncertainty intervals (UI) for key outputs, we performed one thousand iterations of a Monte Carlo simulation. Inputs for the simulation were drawn from the distributions appropriate for each parameter (online suppl. Table s3). We also performed scenario analyses, where – (1) we varied the discount rate (0% and 3%), (2) changed the proportion of prevalent IS to 79.1% to accommodate the AIHW estimation [23], and (3) changed the duration of projection to 10 years. All the analyses were performed using Stata version 17.0 (StataCorp, USA).

Results

Lifetime Risk and Life Expectancy

The lifetime risk of developing IS was estimated to be 15.5% (95% UI: 14.9%, 16.1%) and 14.0% (95% UI: 13.4%, 14.6%) for males and females at age 40 years in 2018, respectively (shown in Fig. 2 and online suppl. Table s4).

Fig. 2.

Cumulative incidence of ischemic stroke for people aged between 40 and 100 years in Victoria, Australia, in 2018.

Prevalence and Incidence of IS from 2019 to 2038

Overall, the model projected 644,208 (95% UI: 615,723, 672,177) incident IS, comprising 564,922 (95% UI: 540,119, 589,785) nonfatal IS and 79,287 (95% UI: 75,011, 83,596) fatal IS for the Australian population aged 40–100 years between 2019 and 2038 (Tables 1, 2). By 2038, 358,534 (95% UI: 343,262, 373,590) people would be expected to live with IS in Australia compared to 313,999 (95% UI: 311,225, 316,720) people in 2019 (online suppl. Table s5–s8). The age of developing IS was different for males and females – there was higher IS in the male population between the fifth decade and the eighth decade of life while females experience more IS after the eighth decade of life (Tables 1, 2).

Table 1.

Projected prevalent and incident IS cases by age group for the Australian population aged 40–100 years over the 20-year period (2019–2038)

Projected prevalent IS cases
age group	male			female			total
	prevalent IS	LB	UB	prevalent IS	LB	UB	prevalent IS	LB	UB
40–49	344,218	338,756	349,498	230,040	226,563	233,625	574,258	565,319	583,123
50–59	506,651	492,703	519,689	326,765	317,154	336,425	833,416	809,857	856,114
60–69	795,471	773,928	815,518	533,751	517,178	550,452	1,329,222	1,291,106	1,365,970
70–79	1,023,221	995,181	1,050,971	729,921	706,432	753,170	1,753,142	1,701,613	1,804,141
80–89	810,685	782,278	838,688	660,605	637,367	684,141	1,471,290	1,419,645	1,522,829
90–100	281,422	264,195	299,841	312,554	295,368	329,744	593,976	559,563	629,585

Projected incident IS cases
age group	male			female			total
	incident IS	LB	UB	incident IS	LB	UB	incident IS	LB	UB
40–49	16,752	15,540	17,957	8,794	8,049	9,624	25,546	23,589	27,581
50–59	31,430	29,730	33,088	18,605	17,411	19,910	50,035	47,141	52,998
60–69	58,909	56,538	61,142	39,905	38,064	41,877	98,814	94,602	103,019
70–79	93,683	90,319	96,786	73,121	70,532	75,876	166,804	160,851	172,662
80–89	98,217	93,987	102,113	93,173	89,698	96,863	191,390	183,685	198,976
90–100	48,246	45,574	50,896	63,372	60,271	66,666	111,618	105,845	117,562
Total	347,238	331,594	361,446	296,970	284,129	310,731	644,208	615,723	672,177

IS, ischemic stroke; LB, lower bound of the uncertainty interval (2.5%); UB, upper bound of the uncertainty interval (97.5%).

Table 2.

Projected nonfatal and fatal IS cases by age group for the Australian population aged 40–100 years over the 20-year period (2019–2038)

Projected nonfatal IS
age group	male			female			total
	nonfatal IS	LB	UB	nonfatal IS	LB	UB	nonfatal IS	LB	UB
40–49	16,346	15,144	17,535	8,568	7,841	9,374	24,914	22,985	26,909
50–59	30,218	28,508	31,877	17,827	16,659	19,088	48,045	45,167	50,965
60–69	55,332	52,815	57,727	37,208	35,288	39,218	92,540	88,103	96,945
70–79	84,906	80,953	88,566	65,334	62,237	68,462	150,240	143,190	157,028
80–89	84,432	79,697	88,807	78,028	73,848	82,233	162,460	153,545	171,040
90–100	38,461	35,785	41,166	48,262	44,902	51,672	86,723	80,687	92,838
Total	309,695	295,616	322,991	255,227	244,503	266,794	564,922	540,119	589,785

Projected fatal IS
age group	male			female			total
	fatal IS	LB	UB	fatal IS	LB	UB	fatal IS	LB	UB
40–49	406	215	658	225	122	365	631	337	1,023
50–59	1,212	736	1,805	779	482	1,146	1,991	1,218	2,951
60–69	3,578	2,472	4,934	2,697	1,893	3,653	6,275	4,365	8,587
70–79	8,777	6,497	11,329	7,787	5,895	9,838	16,564	12,392	21,167
80–89	13,785	10,863	16,983	15,145	12,205	18,330	28,930	23,068	35,313
90–100	9,785	8,052	11,644	15,110	12,780	17,687	24,895	20,832	29,331
Total	37,543	35,592	39,612	41,744	39,419	43,984	79,287	75,011	83,596

IS, ischemic stroke; LB, lower bound of the uncertainty interval (2.5%); UB, upper bound of the uncertainty interval (97.5%).

Years of Life Lived with and without IS from 2019 to 2038

Over the 20-year period, the model projected the total YLL accrued by the Australian population would be 174,782,672 (95% UI: 174,500,992, 175,045,144). Of these, 170,728,880 (95% UI: 170,452,920, 171,015,752) YLL would be accrued by people without IS and the remaining 4,053,794 (95% UI: 3,944,326, 4,161,150) YLL would be accrued by people with IS (Table 3). While females were projected to accrue a greater total number of YLL, males were projected to accrue more YLL with IS (Table 3; online suppl. Tables s9–s11).

Table 3.

Projected YLL, overall and with and without IS, by age group for the Australian population aged 40–100 years over the 20-year period (2019–2038)

Discounting (5%)
age group	male			female			total
	YLL	LB	UB	YLL	LB	UB	YLL	LB	UB
Projected YLL without IS
40–49	23,624,345	23,618,901	23,630,286	24,291,366	24,287,724	24,294,775	47,915,711	47,906,625	47,925,061
50–59	20,611,251	20,597,053	20,626,802	21,588,320	21,577,556	21,598,440	42,199,571	42,174,609	42,225,242
60–69	17,676,705	17,654,297	17,701,522	19,101,545	19,081,329	19,120,948	36,778,250	36,735,626	36,822,470
70–79	12,740,220	12,709,302	12,774,919	14,290,735	14,260,379	14,320,054	27,030,955	26,969,681	27,094,973
80–89	5,968,727	5,930,566	6,011,375	7,428,561	7,390,447	7,465,098	13,397,288	13,321,013	13,476,473
90–100	1,309,598	1,282,626	1,339,546	2,097,508	2,061,443	2,131,424	3,407,106	3,344,069	3,470,970
Total	81,930,848	81,792,792	82,084,520	88,798,032	88,660,128	88,931,232	170,728,880	170,452,920	171,015,752
Projected YLL with IS
40–49	214,493	211,628	217,221	144,776	142,923	146,667	359,269	354,551	363,888
50–59	325,823	318,452	332,690	214,540	209,348	219,691	540,363	527,800	552,381
60–69	507,932	496,447	518,729	345,987	336,951	355,012	853,919	833,398	873,741
70–79	635,832	620,926	650,798	454,407	441,500	466,923	1,090,239	1,062,426	1,117,721
80–89	479,246	464,363	494,186	392,987	380,476	405,501	872,233	844,839	899,687
90–100	157,186	148,021	167,002	180,584	171,138	189,934	337,770	319,159	356,936
Total	2,320,513	2,261,371	2,378,310	1,733,281	1,682,955	1,782,840	4,053,794	3,944,326	4,161,150
Projected total YLL
40–49	23,838,838	23,833,831	23,843,712	24,436,142	24,432,846	24,438,900	48,274,980	48,266,677	48,282,612
50–59	20,937,075	20,923,648	20,950,179	21,802,860	21,792,927	21,811,292	42,739,935	42,716,575	42,761,471
60–69	18,184,637	18,162,076	18,206,473	19,447,532	19,428,536	19,464,366	37,632,169	37,590,612	37,670,839
70–79	13,376,052	13,344,252	13,408,021	14,745,142	14,715,570	14,772,272	28,121,194	28,059,822	28,180,293
80–89	6,447,973	6,407,824	6,488,128	7,821,548	7,783,592	7,856,585	14,269,521	14,191,416	14,344,713
90–100	1,466,785	1,437,332	1,496,308	2,278,092	2,241,365	2,312,259	3,744,877	3,678,697	3,808,567
Total	84,251,360	84,106,160	84,391,184	90,531,312	90,394,832	90,653,960	174,782,672	174,500,992	175,045,144

YLL, years of life lived; LB, lower bound of the uncertainty interval (2.5%); UB, upper bound of the uncertainty interval (97.5%).

Scenario Analyses

Results from the scenario analyses varying the discounting rate, the proportion of IS (of all stroke forms) and the projection period are shown in Table 4. Decreasing the discounting rate increased the total YLL by 18% when the discounting rate was 3% and 57% when the discounting rate was 0%. Changing the proportion of IS from 87% to 79.1% had a negligible effect on the total YLL (reduced the total YLL by 0.05%).

Table 4.

Results of scenario analyses for projected YLL by the Australian population aged 40–100 years

Outcome	Male			Female			Total
	estimate	LB	UB	estimate	LB	UB	estimate	LB	UB
20-year projection
Discount rate 5%
YLL w/o IS	81,930,848	81,792,792	82,084,520	88,798,032	88,660,128	88,931,232	170,728,880	170,452,920	171,015,752
YLL with IS	2,320,513	2,261,371	2,378,310	1,733,281	1,682,955	1,782,840	4,053,794	3,944,326	4,161,150
Total YLL	84,251,360	84,106,160	84,391,184	90,531,312	90,394,832	90,653,960	174,782,672	174,500,992	175,045,144
Discount rate 3%
YLL w/o IS	96,954,336	96,782,608	97,145,504	105,133,672	104,962,336	105,299,312	202,088,008	201,744,944	202,444,816
YLL with IS	2,731,027	2,657,980	2,802,101	2,033,432	1,971,926	2,094,133	4,764,459	4,629,906	4,896,234
Total YLL	99,685,368	99,505,016	99,859,520	107,167,104	106,998,032	107,319,320	206,852,472	206,503,048	207,178,840
Discount rate 0%
YLL w/o IS	128,611,952	128,367,264	128,884,384	139,570,896	139,327,136	139,806,816	268,182,848	267,694,400	268,691,200
YLL with IS	3,593,116	3,490,459	3,692,708	2,662,685	2,576,842	2,747,726	6,255,801	6,067,301	6,440,434
Total YLL	132,205,064	131,948,632	132,453,496	142,233,584	141,994,064	142,449,792	274,438,648	273,942,696	274,903,288
Proportion of IS (79.1%)a
YLL w/o IS	82,016,280	81,878,080	82,170,208	88,849,976	88,711,960	88,983,096	170,866,256	170,590,040	171,153,304
YLL with IS	2,192,133	2,134,778	2,247,746	1,637,722	1,589,301	1,685,815	3,829,855	3,724,079	3,933,561
Total YLL	84,208,416	84,065,008	84,348,384	90,487,696	90,352,032	90,609,864	174,696,112	174,417,040	174,958,248
10-year projection
Discount rate 5%
YLL w/o IS	47,493,200	47,440,248	47,552,072	51,290,880	51,237,096	51,342,248	98,784,080	98,677,344	98,894,320
YLL with IS	1,392,480	1,367,502	1,417,562	1,060,352	1,037,785	1,082,850	2,452,832	2,405,287	2,500,412
Total YLL	48,885,680	48,829,720	48,939,172	52,351,236	52,297,244	52,399,592	101,236,916	101,126,964	101,338,764

YLL, years of life lived; w/o, without; LB, lower bound of the uncertainty interval (2.5%); UB, upper bound of the uncertainty interval (97.5%).

^aThe proportion of IS is the percentage of IS of all forms for stroke. YLL are reported for the annual 5% discount rate.

Discussion

The lifetime risk of IS from age 40 to 100 years was estimated to be 15.5% for males and 14.0% for females in Victoria in 2018 (i.e., approximately 1 in 7 people were expected to develop IS over their lifetime). Our dynamic model projected that, over the 20 years from 2019 to 2038, 0.64 million Australians aged 40 years or older would experience an incident IS and approximately 4 million (2% of total) YLL would accrue among people with IS.

The lifetime risk of IS in our study is lower than the reported lifetime risk from the Framingham study (around 1 in 6 for men and 1 in 5 for women) [24], Rotterdam study (around 1 in 5 for both men and women) [25] and a Japanese study (around 1 in 5 for both men and women) [26]. The difference in the estimation of the lifetime risk of IS partly attributable to the fact that our study mainly focused on the lifetime risk of IS, while in contrast, the previous studies primarily reported on the lifetime risk of stroke in general [24–26].

Our study presented the projected health burden of IS by age and sex stratification. In keeping with existing knowledge [27], we projected that a higher number of males would experience the burden of IS in early years of life while females experience the burden in later years of life. However, the overall fatal IS throughout the projection period was higher for females than males. Recent studies also showed that fatal stroke in females exceeded that of males [27–30]. Multiple factors could contribute to the exceeding stroke case fatality in females than males, the leading on likely being higher life expectancy [28, 31] although other female-specific risk factors that increase the risk of developing stroke could include use of menopausal hormone therapy [28, 32, 33], combined oral contraceptives [28, 34], time from menarche to menopause with elevated stroke risk in premature (<40 years) and early (40–44 years) menopause [28, 34, 35]. Pregnancy related adverse outcomes, including preeclampsia, gestational hypertension, and preterm delivery and fetal growth restriction have been associated with long-term risk of stroke [28, 36–38]. The living condition at the onset of stroke also contributes to the case fatality of stroke in female population [28, 39]. Females, due to longer life expectancy, in general tend to be widowed, unmarried, or living alone and more likely affected by disability curbing their activity of daily living compared to males at onset of stroke [28, 39]. With increasing life expectancy and aging population [31, 40], and given predominance of females in the ≥85 years age group [40], the future burden of stroke would disproportionately impact females.

Current research in biological sex differences in stroke provides better understanding of the risk factors specific to each sex and could help tailor preventive treatment and recovery strategies accordingly [28, 41]. With increasing research interest on gender (i.e., the social construct) in addition to biological sex difference in stroke, there is an opportunity to capture more information on individuals’ traits beyond biological characteristics including gender identity, expression, roles, and stereotypes [28, 42]. Furthermore, gender is also relating to individual’s economic resources and healthcare access which could affect health [28, 41]. This could unmask more knowledge to stroke burden not addressed by biological sex stratification [28, 41] and could modify sex differences in the projected IS burden in the coming decades.

The projected increase in the health burden of IS underscores the importance of primary preventive strategies. Reprioritising primary prevention strategies could have a considerable impact on the economy [43] – every US$1 invested in the prevention of stroke and cardiovascular disease is estimated to return US$10 [43]. However, priorities for high-risk population screening as a primary prevention strategy for stroke left out many people without appropriate interventions [44]. Resetting the priorities toward population-wide strategies that reduce exposure to risk factors across a lifespan of the whole population is needed [44, 45]. Integrating healthy lifestyle with interventions (e.g., polypill) targeting risk factors for stroke has been proposed as an effective way of implementing population-wide preventive strategies [45, 46]. Policies should be in place to make sure sustainable funding available for implementing population-wide strategies, training and deployment of community health workers to facilitate implementation of prevention strategies and monitoring the effectiveness of such strategies through research [44–46].

Strengthening secondary prevention strategies is also needed to reduce the recurrence of IS [45]. A recent report from Australia showed that the proportion of people with IS discharged from hospitals on antihypertensive medications, antithrombotic medications or lipid-lowering medications were below the achievable benchmarks [47]. For instance, 77% of people with IS discharged on antihypertensive medications compared to the achievable benchmark of 94% [47].

Strengths and Limitations

Our study projects outcomes over a 20-years period (2019–2038) specifically for IS, using a large, representative dataset as the data source. Victoria state account for the second largest population in Australia [12] and projecting data sourced from the Victorian linked dataset to the whole Australian population would represent the health burden of IS. The dynamic nature of the model allows us to account temporal changes in population demography into the model while projecting these outcomes.

Lifetime risk estimates are only applicable to the entire population, and we did not have access to clinical data allowing us to stratify by important clinical characteristics. Despite this caveat, lifetime risk estimations are useful for public education and awareness campaigns as they are intuitive to comprehend [48]. For instance, presenting the lifetime risk of IS for the male population aged ≥40 years in Australia as “1 in 7” will have a better impact on awareness creation than providing other measures of risk.

Another limitation is that the post-IS mortality data were derived from a short follow-up period (up to 6 years); thus, we likely overestimated post-IS mortality rates. Our model depended on estimates from overt IS without the consideration of the additional burden of covert (silent) IS; moreover, deaths not recorded in the NDI were not considered which will encompass people who died overseas or died before being admitted to the hospital. These limitations will have led to underestimation of the true burden of IS. Furthermore, the rates and proportions estimated held constant throughout the projection time. Thus, our estimates should be considered projections, and not predictions, as the underlying rates and proportions used in the models will likely change from 2019 to 2038.

The study also did not consider socioeconomic characteristics in the model – people with greater socioeconomic disadvantage accumulate a greater burden of CVD [49]. Finally, although our study is likely generalizable to the Australian population, it is unlikely to be representative globally, as demonstrated by the regional differences in lifetime risk of stroke [50] and stroke-related mortality rates [1] in different parts of the world.

Conclusion

Our model projected the burden of IS in Australia from 2019 to 2038. The findings of this study provide a multitude benefit for various stakeholders – policy-makers may utilize the outcomes to develop and implement targeted prevention and treatment strategies for IS in Australia; clinicians may use the findings for future guideline updates related to IS; and outcomes such as lifetime risk can be used to as an effective way of disseminating the burden of IS information across the community.

Acknowledgments

The authors would like to thank the Victorian Department of Health as the source of VAED data; the Center for Victorian Data Linkage (Victorian Department of Health) for the provision of data linkage, and the Australian Institute of Health and Welfare provided access to the NDI.

Statement of Ethics

This study was approved by the Australian Institute of Health and Welfare Ethics Committee (EO2018/4/468) and Monash University Human Research Ethics Committee (14339). We have received a waiver for informed consent to access data.

Conflict of Interest Statement

The authors have stated explicitly that there are no conflicts of interest in connection with this article.

Funding Sources

T.B.A. is supported by Monash Graduate Research Scholarship and Monash International Tuition Fee Scholarship. J.I.M. and Z.A. were supported by the National Health and Medical Research Council Ideas Grants Application ID: 2012582. The funder had no input into the design of the study or decision to submit for publication. J.I. has received funding from AstraZeneca and Amgen for projects unrelated to this study.

Author Contributions

T.B.A.: conceptualization (support); formal analysis (lead); methodology (equal); software (lead); visualization (lead); writing-original draft (lead); and writing-review and editing (equal). J.I.M.: conceptualization (equal); methodology (equal); software (support); visualization (support); writing – review and editing (equal); and supervision (equal). J.I.: conceptualization (support); writing – review and editing (equal); and supervision (support). Z.A.: conceptualization (lead); methodology (lead); visualization (support); writing – review and editing (equal); supervision (lead); and resources (lead).

Funding Statement

T.B.A. is supported by Monash Graduate Research Scholarship and Monash International Tuition Fee Scholarship. J.I.M. and Z.A. were supported by the National Health and Medical Research Council Ideas Grants Application ID: 2012582. The funder had no input into the design of the study or decision to submit for publication. J.I. has received funding from AstraZeneca and Amgen for projects unrelated to this study.

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.