key: cord-0025370-yoftiaqi
authors: Owolabi, Mayowa O; Thrift, Amanda G; Mahal, Ajay; Ishida, Marie; Martins, Sheila; Johnson, Walter D; Pandian, Jeyaraj; Abd-Allah, Foad; Yaria, Joseph; Phan, Hoang T; Roth, Greg; Gall, Seana L; Beare, Richard; Phan, Thanh G; Mikulik, Robert; Akinyemi, Rufus O; Norrving, Bo; Brainin, Michael; Feigin, Valery L
title: Primary stroke prevention worldwide: translating evidence into action
date: 2021-10-29
journal: Lancet Public Health
DOI: 10.1016/s2468-2667(21)00230-9
sha: a32d77c212bb8159358512e2dec4d554dad5f029
doc_id: 25370
cord_uid: yoftiaqi

Stroke is the second leading cause of death and the third leading cause of disability worldwide and its burden is increasing rapidly in low-income and middle-income countries, many of which are unable to face the challenges it imposes. In this Health Policy paper on primary stroke prevention, we provide an overview of the current situation regarding primary prevention services, estimate the cost of stroke and stroke prevention, and identify deficiencies in existing guidelines and gaps in primary prevention. We also offer a set of pragmatic solutions for implementation of primary stroke prevention, with an emphasis on the role of governments and population-wide strategies, including task-shifting and sharing and health system re-engineering. Implementation of primary stroke prevention involves patients, health professionals, funders, policy makers, implementation partners, and the entire population along the life course.

Ababa, Ethiopia 135. Emergency Department. Hospital Nacional de Niños "Dr. Carlos Sáenz Herrera", CCSS, San José, Costa Rica 136. Department of Public Health, Juntendo University School of Medicine, Tokyo, Japan; National Center of Neurology and Psychiatry, Kodaira, Japan

The economic implications of a stroke can be broken down into (a) medical care and other costs that are incurred in the acute phase (e.g., hospitalisation expenses, ambulance fees, etc.); (b) costs incurred during the subsequent post-stroke care phase, that may involve rehabilitation, drug expenses, outpatient services, in some cases lasting several years; and (c) loss of income to the household of the person with stroke (and potentially production, if a working person cannot be replaced for a period of time) owing to the effect of strokes on morbidity and mortality. There may also be additional costs incurred by other parts of the economic system, including social services provided, and the excess burden (or efficiency losses) associated with the added taxes used to finance some of these consequences of stroke. The global economic impacts of stroke estimated in this supplementary section are focused on (a)-(c). In the case of (b), the cost estimates are the lifetime costs associated with the healthcare use and caregiving services to the person with stroke. In the case of (c), two main approaches are used to evaluate the economic loss. The human capital approach (see below) is used to assess the potential income loss to the household. However, this need not translate into an equivalent (social) production loss if the loss of work time due to a person being affected by stroke could be lowered by hiring a replacement. We used the "friction cost" approach (Koopmanschap et. al. 1995) to assess the potential production loss when replacement hiring occurs, as outlined in greater detail below.

We first estimate the lifetime direct medical care (including caregiving) costs associated with incident stroke cases.

Step 1-1

Country-specific estimates of the number of incident and prevalent stroke cases, and numbers by type of stroke (ischaemic or haemorrhagic) were obtained from the GBD (Global Burden of Disease) database for 2017 1 .

Step 1-2

The World Bank classification of countries by income was used to classify countries into (a) high income (HIC); (b) upper middle income (upper-MIC); (c) lower middle income (lower-MIC); and (d) low income (LIC).

We carried out an extensive search (PubMed and Google Scholar) of peer reviewed literature -but not a full systematic review -for studies on the costs of stroke. Search terms used included, stroke, economic burden, treatment costs, costs, expenditures, hospital costs, acute care, medical care, ischaemic stroke, haemorrhagic stroke, cost-effectiveness of stroke interventions, CVD, healthcare costs, etc., often in combination with individual country names. We limited our search to studies that used data from 2000 or later, and (ideally) published in the year 2005 or later. We were concerned that developments in stroke treatments over time would affect costs, and information from earlier years would not adequately reflect the direct costs of stroke at the present time. Our search led us to identify about 70 studies, of which 50 were ultimately read in full.

With two notable exceptions (Cameroon and India) most studies in low-and middle-income countries were focused on costs of hospitalisation and related services in the acute phase immediately following a stroke. In contrast, studies from high-income countries had a much larger share reporting 1-year costs following a stroke, including expenditures post-discharge. We could identify only 8 studies (all from high-income countries), that estimated lifetime healthcare costs (often including social care) following stroke, alongside 1-year costs.

Our review of the literature highlighted the very large cross-country (and sometimes within country) variations in costs during the acute phase and the 1-year period following the stroke. These variations were driven by multiple factors: patient composition (e.g., comorbidity status, or incident versus prevalent stroke cases), patients from a single (possibly high-end tertiary) hospital, inclusion (or exclusion) of social care following discharge, inclusion (exclusion) of informal care, incident versus prevalent stroke cases, stroke type, data from stroke registries and linked utilisation data versus self-reported healthcare use data from household surveys, and out of pocket spending (alternately insurer expenditures) as against all costs, whether subsidised or not. Importantly, very few of the stroke cost studies tried to address a key attribution problem: in a competing risks framework, attributing to stroke all the costs of health and social care utilisation following a stroke is likely to upwardly bias the costs of stroke. Comparability was also hindered by differences in the years for which information was available, and monetary denominations used (local currency, US$ or Euro, based on then prevailing exchange rates, or international dollars).

Next, we excluded studies relying on patients from a single hospital, or studies that focused on patients belonging to a sub-group with high healthcare needs (the exception being studies from LMICs, given the very studies from this group), or where a distinction between incident and recurrent cases was not possible. Where a breakdown by cost-type was provided, we excluded costs of informal care from our estimates, given the considerable variability in methodology used to estimate the costs of informal care (in our calculations, costs of informal were added on later, using methods described in the supplement to the manuscript).

Treatment cost estimates available in the literature were then adjusted in two ways: (a) First, cost estimates were brought forward to 2017 at current prices in the local currency. This involved adjusting for inflation (using the GDP deflator) and for medical care inflation over and above the general inflation (to take account of technological changes that increase the cost of medical services faster than other items). The latter was assumed to be 3% per year (or 0.03) irrespective of country; (b) using purchasing power parity for 2017 to convert the stroke cost estimates into 2017 international dollars, separately (where possible) for ischaemic and haemorrhagic strokes. Tables 1 and 2 provide these estimates for high income  countries and for low-and middle-income countries, for hospitalisation costs during the acute  phase, and costs for the 1- year following stroke. These costs were separated for ischaemic and haemorrhagic strokes, where possible. Step 1-4 (Representative Regional Costs for Stroke, and Attributable Stroke Costs)

Given the limited number of stroke studies, our strategy for constructing global estimates of the healthcare costs of stroke relied on constructing "representative" costs, based on country income-classification (see Table 3 ). Among high income countries (HICs), we limited our choice to studies that sought to address the risk of double counting (owing to individuals experiencing a stroke being potentially susceptible to other health conditions in a competingrisks framework) to achieve a better attribution of stroke costs. That left us with 4 studies that used a control group (one from US, one from Finland, and two from Sweden) to achieve a better estimate of costs attributable to stroke. The US study was missing social/institutional care costs (but it did provide an estimate of the likely bias), so its stroke cost estimates were appropriately scaled up. These 4 studies were used to construct the upper and lower bounds for stroke costs in the acute phase and in the 12 months following stroke in HICs. In upper middle-and low-income countries, information was mostly missing for 1-year stroke costs. Data from a single Indian study (after excluding the costs of informal care and other adjustments) was used to scale up acute-phase hospitalisation costs to 1-year costs. With these modifications, data from China was used to construct a range of acute-phase hospitalisation and 1-year costs for that country; and data from India was used to construct a range for India. For upper-middle income countries, data from Argentina, Brazil, Malaysia, Turkey was used to derive a range, with costs at the lower end of the range comprising patients at public hospitals, and at the top end, private hospitals. For LMICs other than India, we excluded Vietnam and Pakistan that had abnormally low/high stroke costs. Data from other countries -Philippines, Cameroon, South Africa, and Nigeria -was used to construct the range of acute-phase and 1-year stroke costs. The representative regional costs are reported in Table 3 . Acute phase stroke treatment costs used for constructing direct cost estimates were based on the estimates reported in Table 3 for country income groups, China, and India. Given that the costs of stroke treatment tend to vary by type of stroke, haemorrhagic strokes were distinguished from ischaemic (and other) strokes. There are differences across countries in this proportion in the GBD database (the share of ischaemic strokes being higher in HICs and that of haemorrhagic strokes being higher in lower-MICs and LICs).

Although not depicted in Table 3 , when estimating the global direct costs of stroke, we adjusted downwards the stroke cost estimates in HMICs, LMICs, China and India by 30%, to arrive at an estimate of attributable stroke costs (in a competing risks framework). This figure was roughly the proportional change achieved in the attributable costs of stroke, in the US study by Girotra et. al. (2020) 2 that used a control group when estimating stroke costs.

It was assumed that all incident stroke cases need acute-phase treatment, reflecting the fact that the seriousness/acute nature of the condition is likely to result in the person affected (or their household) seeking treatment.

Step 1-6 Assumptions about Care Following the Acute Phase It was assumed that stroke patients that died in any given year did not receive any post-acute phase care. Available literature suggests a 1-year mortality risk for stroke of 25% in HIC. We assumed a similar mortality rate for MICs and low-income countries (LICs) given the relatively younger ages at which stroke occurs in their populations. For the remainder (i.e., number of new stroke cases (Y) MINUS the number of new stroke cases who died (D), or (Y-D)), patients were assumed to use services consistent with the full 1-year costs in their country-income group.

Step 1-7: The Costs of Informal Care Estimated costs of informal caregiving for stroke survivors are included as part of direct costs.

A key challenge that needed to be addressed in this context was the lack of evidence on hours allocated to informal caregiving for stroke in low-and middle-income countries (LMICs), although it is well known to be significant (and typically provided by women) ( 53 Australian Stroke Foundation 2020 8 ), or approximately 20 hours per stroke survivor. Because the magnitude of the provision of informal care is dependent on country health-and rehabilitation systems, which are relatively weak in LMICs, we assumed that the informal caregiver hours for informal support in LMICs are 50% higher than their counterparts in high-income countries. In another (high) scenario, we also allowed informal caregiving hours in LMICs to be double that of their counterparts in high-income countries. Although evidence is obviously needed, we believe that even this is relatively conservative, given limited access to high quality stroke services in LMICs and the poor health outcomes that will likely result.

It is likely that individuals who need informal care following stroke may have ended up needing informal care even in the absence of the stroke, owing to underlying conditions that make them susceptible to other health shocks. Joo et. al. (2014) 52 estimated that weekly incremental informal caregiving hours attributable to stroke are 8.5 hours per patient, and that the economic value of informal caregiving per stroke survivor is $8,211 per year, of which $4,356 (53%) is attributable to stroke. For our global cost estimates, it was assumed that (a) informal care in high income countries was 10 hours per week per survivor; (b) informal care in countries other than HICs was 15 hours per week per survivor (low scenario); and (c) informal care in countries other than HICs was 20 hours per week per survivor (high scenario).

The valuation of time allocated to informal care has been done in a variety of forms in the literature, including in terms of opportunity cost of time (based on the work status of the individual providing caregiving), with unemployed individuals valued at zero, or at some low cost of leisure foregone; replacement cost (i.e., the hourly earnings of similar jobs in the economy), and so forth. The opportunity cost perspective that emphasises employment will undervalue women caregivers' contributions in LMICs given the low work participation rates of women often observed in LMICs. Moreover, work on family farms or other selfemployment activities which are commonly prevalent do not provide readily calculated measures of income from work. Our preference was for a replacement cost approach. But with limited data available, we decided to use one-half of GDP per adult worker as the annual full-time cost of caregiving.

Most stroke cost studies, even in high-income countries, tend to emphasise 1-year costs. To measure lifetime costs of an incident single stroke case, we first obtained information on the relative magnitude of lifetime stroke costs to 1-year stroke costs among incident survivors for the countries for which data was available. Analysis of stroke costs based on longitudinal data from Australia, Belgium, Germany, Finland, France, Sweden, and the United Kingdom suggests that the ratio of lifetime costs of stroke per incident (first year) survivor ranges from roughly 2.1 to 4.0, and this was the range we used for our sensitivity analysis, with a base case as the mid-point of the range (Chevreul et. al. 2013 ; 23 Second, we scaled up the 1-year cost estimates for countries by a factor of 2.1, 4.0 and the mid-point (3.05), under alternative scenarios. We acknowledge that these numbers could be even higher potentially in low-and middle-income countries with rapidly evolving health systems that allow for improved technologies and growing public insurance that could lower the patient out of pocket costs of health service use (but not overall resource use), although no data are available on this subject. This remains a limitation of the methodology adopted in this supplement.

Step 1-9: Estimates of the Global Cost of Treating Stroke.

This step essentially involved multiplying the incident stroke cases with estimates of (relevant) unit costs for acute stroke care and post-stroke care (and adjusting for lifetime stroke costs. This provided, for each income-region and globally, direct costs of stroke at the lower and the upper end. An estimate of the direct costs of stroke based on the mid-point of all the ranges of acute-phase and 1-year costs, informal caregiving time, and the ratio of lifetime to one-year costs of stroke was also constructed, and this provides the base case estimates reported in Table 4 , in addition to the lower-and upper-bound estimates. Next, we turn to the methodology for the estimates of income (production) losses.

Two approaches were used. First, income losses were estimated for incident stroke patients during the reference year (2017). Note that this is not the same as production losses to the economy or employers, which is better captured by other approaches, such as the friction cost method described below. However, income losses are likely to be more effective at capturing economic well-being of affected households and implications for equity relative to methods based on production losses.

Income losses for stroke patients were estimated by: (a) evaluating the number of years of working life lost due to stroke; and (b) the earnings losses per year during the working life lost. A discount rate for future income losses of 5% was assumed (this is typically around the mid-point of the discount rates used in the economic literature on evaluation, which range from 3%-10%). Notice that current real rates of interest (often used to guide choice of discount rates) are close to zero globally. Using a discount rate of 5% would make the estimates of income losses reported here, conservative. The annual rate of growth of real wage in the future was taken to be the same as the annual rate of growth of real GDP per working age person (15-64) in the period 2010 to 2017. Log-linear regression models were run to estimate these rates separately for each country, using annual data from the World Development Indicators database of the World Bank.

It was assumed that people retire at 65 years in all countries. Data on the incidence and prevalence of strokes for the age-group 15-64 years was obtained from the GBD database. These numbers, available in 5-yearly intervals (15-19, 20-24, …etc.), were used to first estimate the average age of an individual below 65 who had a stroke in 2017.

If the number of strokes in 5-yearly groups in the age-range 15-64 is denoted by " ! , " " , " 2 !%&'( = " ! + " " + " # + ⋯ + " !$ This calculation assumes that stroke patients are uniformly distributed within each age interval and was used to estimate Mean Age at stroke for the age group 15-64 years for each country in the GBD dataset.

Step This formula yields a lower bound to the expected years of survival because some individuals will drop out of this age group over time by moving to 65+, and not because they died. This downwardly biases the estimate of the expected years of life following a stroke (and hence the income loss upwards) but the bias to income losses will be small due to discounting, as the move to the 65+ age category will occur several years down the road.

For purposes of estimation of working years lost, it was assumed that all strokes in the age group 15-64 occurred at the mid-point of 2017 in each country and at the mean age E for the age group 15-64 in that country. Step

To illustrate the method used, assume that the wage per person in 2017 is W and the growth rate of wage (in 2017 prices) is >. Assume that the discount rate is R. It is again worth underlining that income losses cannot be taken to imply productivity losses or national income losses in any strict sense of the term, but there are equity implications, since some lose out and others gain. Previous theoretical work has also argued that household economic losses can be interpreted as a lower bound for the value of a statistical life.

Step 2-5 Method used to calculate (expected pre-stroke wage) I , and (expected post-stroke

Even if a person is of working age, they may not have a job. To account for this, the estimates used here relied on the notion of 'expected wage', and the assumption that the likelihood of a person in the age-group 15-64 holding a job was equal across ages and gender and given by the ratio of workers (or employed) in the 15-64 age group to the working age (15-64) population. This number was obtained from the World Development Indicators database (World Bank 2021).

Denote the likelihood of any person aged 15-64 years holding a job in country ′L′as M , , and that every person in that group has an equal likelihood of being employed. Then a reasonable proxy for the expected wage for an individual is I , , which depends on the share of people who work and the output per worker. For I -, it was assumed that the functional status of individuals after their stroke was less than before their stroke, noting that there are both minor and major strokes, and not all cases will completely exit from the workforce. Previous work, mainly from HICs and a small number of middle-income countries, suggests that between 40%-75% of working age stroke survivors return to employment over time. [57] [58] [59] [60] [61] As well as lower workforce participation rates compared to counterparts who did not experience a stroke, the wages of stroke survivors who return to work are also lower. 62 Based on this evidence we assume that employed individuals who first had a stroke in 2017 see a halving of their income for the rest of their working life (those who die lose their full pre-stroke income). So, the working assumption was I -, = 0.5I , .

Step

These are estimated by multiplying for each country the number of incident cases of stroke by the income loss per incident stroke case, Y , among people in the age group 15-64 years. Alternative ways to estimate income losses exist. For example, data on the national (regional) age distribution of incident stroke cases, combined with expected years of survival for each age group (in that distribution) could also be used to construct a measure of national (global) economic loss. The approach used in this note, by appealing to steady state considerations, enables a more direct method to estimate income losses.

The basic idea under the friction cost approach is that the method used to assess the income loss to an individual (human capital approach) does not work as well to assess production losses to society. This is because in a setting with unemployment, persons leaving work permanently (via death or disability) or temporarily (sick leave) can be replaced by others who may be available. From this perspective, what matters for costs is the "friction period", the time between the absence from work of the sick person (in this case the individual with stroke), and any costs related to the hiring or training of the replacement. Beyond that no productivity loss is envisaged. One exception is when an already employed person replaces the one who is sick, thereby creating another vacancy elsewhere, and so on. This suggests that a more complete friction cost approach ought to the account for the cumulative sum of friction periods resulting from a vacancy generated by death, disability, or temporary absence from work.

The major challenge in applying the friction cost approach is lack of data, especially in LMICs, related to (a) the length of the friction period -the time over which the person with stroke (mortality or disability) is replaced by another, and thus the period over which production is lost; (b) the cumulative nature of friction periods, as when people move from an employed status to the vacancy created by the person affected by stroke, thus creating a vacancy in their existing position; and (c) accounting and valuing for loss of leisure time (when an individual moves from an unemployed state to fill the vacancy generated by the person with stroke). For instance, a recent study for Malaysia estimated the costs of CVD using a friction cost approach but relied almost completely on data from European countries for (a) and (b) (Ministry of Health Malaysia 2020).

Step 3-1: Estimating the Friction Period

As noted above data on friction periods is rarely available in LMICs. Studies using European data work with a friction period of 60 days, sometimes higher. In the United States, available estimates suggest that costs of replacing workers amount to roughly 21% of annual employee salary, or equivalent to roughly 55 days (Boushey and Glynn 2012). 63 In India, a study of the Information Technology (IT) industry estimated costs of replacing workers to be approximately 10% of annual employee salary, or effectively about 26 days of work (Gochhayat and Nanda 2011), 64 reflecting the highly competitive nature of this industry and its attraction to employees. Presumably, one can assume that similar friction periods apply to small enterprises in the informal sector that hire employees on a wage basis, and in elementary occupations, such as unskilled agricultural work, where substitutes are readily available.

It is not obvious, though, that estimated costs of worker replacement in the literature, which are typically calculated for formal sector jobs, can be readily translated to all categories of informal sector work, especially self-employment, including sole proprietorship. For instance, sole proprietorship has to do with entrepreneurial skills, and these are not necessarily replaceable, and the human capital approach is likely to be a better indicator of the production loss here. Nor is it obvious, given the onerous labour laws and hiring practices in many LMICs -that all formal sector jobs operate on the same principles as the software industry -and indeed the friction periods and hiring costs will likely be much higher for government jobs, a major source of employment. But paucity of data makes accounting for these possibilities in our calculations difficult.

Here we undertake a particularly simple version of friction cost analysis. In the absence of data on cumulative friction periods arising from a vacancy, we used 60 days as the applicable friction period for high income countries, and 30 days for low-income countries (based on the Indian evidence). Because we did not include the added costs of leisure foregone by the new workers or allow for a sequence of friction periods arising from the filling of vacancies arising from stroke, our estimates are likely to constitute a lower bound to production losses from a more complete friction costs approach. This is a limitation of our analysis.

Step 3-2: The Annual Output per Worker

As in the case of the human capital approach, we assumed the annual output per worker to be GDP per member of working age group, that is I , for a given country L . The resulting estimates of productivity losses are provided in Table 4 , and provide a lower bound for production losses, if estimates based on the human capital approach are taken as an indicator of production losses as well. Table 4 summarises the main findings for global treatment expenditures incurred in the first 12 months of stroke. These point to health expenditures on incident stroke cases of about I$ 393 billion, or about 0.3%, of global GDP measured in I$, in 2017. Given that the global economy was growing annually at about 3.5% for much of the decade, healthcare costs for stroke would have accounted for almost one-tenth of annual growth in that year. Table 4 also provides a low-end and high-end range of direct cost estimates for stroke, from 0.16% of global GDP to 0.5% of global GDP. Table 4 also provides estimates of output losses using two different approaches. The lower bound estimates rely on the friction cost method which is well suited to capturing production losses. Compared to direct costs, these estimates point to limited effects on global output, approximately 0.05% of global GDP. The upper-bound estimate relies on the human capital approach and suggests output losses of almost I$976 billion in 2017 (or about 0.8% of global GDP). In settings with high levels of unemployment and underemployment, the human capital approach is likely to yield overestimates of production losses. However, it is still a useful measure of income losses to individuals affected by stroke and their households, with important welfare and equity implications.

Our economic analysis was designed to develop an economic case for pragmatic solutions. We estimated the global financial costs of providing care to patients with stroke immediately following admission, in the period 12 months following, lifetime costs, and associated income losses. We obtained data from the GBD study on the number of new stroke cases, and deaths from stroke in 2017, 1 and combined these figures with available estimates of the costs of treatment. Our estimates suggest that stroke-related treatment and rehabilitation costs ranged from approximately I$206 billion to I$638 billion among new stroke cases in 2017, or between 0.16% and 0.50% of global GDP measured in international dollars. The low-and highends of the range of aggregate stroke costs primarily reflect the wide range of stroke treatment costs across countries, and the considerable variation in the components included in the costs of post-stroke care reported in the literature. Roughly one-half of the stroke treatment expenditure occurred in HICs, reflecting both the higher incidence of strokes as well the considerably larger costs of treatment and rehabilitation in these regions. Another 35%-40% of the spending was in upper middle-income countries, and the rest, about 5%-10% of global treatment costs and post-stroke care spending, was in LMICs.

In addition to treatment costs, there was also a significant cost in terms of income losses due to premature death and disability among stroke affected individuals (and their households), although the adverse effects on national output are less apparent. Previous work, mainly from HICs and a small number of middle-income countries, suggests that between 40%-75% of working age stroke survivors return to employment over time. [57] [58] [59] [60] [61] As well as lower workforce participation rates compared to counterparts who did not experience a stroke, the wages of stroke survivors who return to work are also lower. 62 We estimate that the discounted lifetime economic losses to households with incident stroke cases in 2017 amount to I$976 billion globally (or 0.8% of global GDP). Roughly one-third of the global economic losses from stroke occurred in HICs and another 50% in upper middle-income countries. Income losses from stroke in LMICs and LICs accounted for only 15% of the global total, primarily because of their much lower income levels. The share of income losses accounted for between threefifths to four-fifths of the total global economic costs of stroke.

With significant treatment, rehabilitation, and other direct costs that amount to about 10% of the economic gains worldwide, interventions that can help prevent or reduce stroke incidence and/or mortality by small numbers can potentially yield very large economic returns.

The United Nations (UN) targeted Sustainable Development Goal (SDG) 3.4 estimated that a one-third reduction of premature mortality from stroke through better acute care and prevention would avert 2.4 million deaths in 2030. 65 In a broader context, investing in prevention of CVDs is integral to achieving UN defined SDG target 3.4 and to progress towards at least nine SDGs. 68 Many countries have implemented cost-effective interventions at low levels, so the potential to achieve these targets and strengthen national income by scaling up these interventions is enormous. 68 A strengthened effort across multiple sectors with effective economic tools, such as price policies, subsidised access to preventative services and insurance, is necessary. Stroke is heavily clustered in people with low socioeconomic status and is an important cause of medical impoverishment, 68 thereby exacerbating economic inequities within societies. Thus, stroke is a barrier to achieving SDG 1 (No Poverty), SDG 2 (Zero Hunger), SDG 4 (Education), SDG 5 (Gender Equality), and SDG 10 (Reduce Inequalities). Productivity gains from preventing and managing stroke will contribute to SDG 8 (Economic Growth). SDG 11 (Sustainable Cities and Communities) and SDG 12 (Responsible Consumption and Production) offer clear opportunities to reduce the burden of stroke and other NCDs and to create sustainable and healthy cities. 68

The 2019 American ACC/AHA Guideline 69 on the primary prevention of CVD emphasises the importance of promoting a healthy lifestyle throughout life, evaluation of social determinants of health to inform treatment decisions, screening of adults aged 40 to 75 for 10-year CVD risk to inform pharmacological therapy, healthy diet (low sodium, low fat, low cholesterol and low glycaemic index), adequate physical activity (150 min per week of moderate-intensity or 75 min per week vigorous-intensity physical activity), smoking cessation as well as adequate control of low-density lipoprotein (LDL) cholesterol and blood pressure. They also acknowledge the importance of a combination of population-wide preventative strategies with a comprehensive person-centred approach that addresses all of an individual's lifestyle habits and estimated risk of a future CVD event.

As stroke shares many risk factors with other major NCDs (such as ischaemic heart disease, diabetes mellitus, vascular dementia, and some types of cancer), it is logical that to be most cost-effective, primary stroke prevention strategies must be integrated into the overall strategies for prevention of all of these NCDs to achieve the suggested global targets in reducing NCD mortality and prevalence of risk factors. Only by joining forces with other interventions for NCDs prevention will stroke prevention have its full impact. [70] [71] [72] The Global Alliance for Chronic Diseases 73 is a good example of such an integrative approach. In our paper we tried to provide evidence-based pragmatic solutions on strategies for primary stroke prevention, including a critical review of existing primary prevention strategies and current guidelines, economic analysis, and identification of gaps in primary stroke prevention ( Figure  2 ). On the individual level in the general population, the suggested target SBP goals are: <120 mm Hg for people younger than 50 years, <130 mm Hg for people aged 50-74 years, and <140 mm Hg for people aged 75 years and older. 74 It was estimated that healthy lifestyle modifications on the individual level can reduce stroke incidence by about 40-50%, 75,76 but the effect of such modifications on the population level (shifting the lifestyle risk profile of the community towards healthier lifestyle) could be substantially greater. 77 A similarly large effect of lifestyle modification (healthy diet, adequate physical activity, smoking cessation etc.) was suggested for reducing the risk of recurrent stroke. 78 As many lifestyle habits are set early in life, culturally appropriate education about healthy lifestyles should be incorporated into standard education curricula and started early in life, with reinforcement across the lifespan. 79 These preventative strategies should be facilitated by adequate stroke education campaigns that take into account not only cultural and subcultural differences and beliefs of people of various races and ethnicities 80 but also significant geographical differences in the lifetime risk of stroke 79 and its risk factors. 81 There is also evidence that the intensity of primary stroke prevention should not be reduced in older people. 79 Although effective primary stroke prevention must include a combination of population-wide and individual-based preventative strategies, 82 the World Stroke Organization and World Heart Federation emphasise that priority (including resource allocation) should be given to the population-wide strategies ( figure 3) . 83, 84 Given the fact that the majority of stroke burden (60-70%) across all countries in the world is associated with elevated systolic blood pressure (SBP) and unhealthy lifestyle risk factors, such as smoking, obesity, low physical activity and poor diet (including excessive salt, sugar and alcohol intake; ), 75,81,85-90 reducing exposure to these risk factors and treating hypertension should be the priority targets for both population-wide and individual-based preventative interventions for primary (and also secondary) stroke prevention. 81 Additional priorities for setting up population-specific stroke preventative strategies should be informed by the most recent estimates of the populationattributable risks, which are currently available for 204 countries. 67 

Mobile technology applications, also known as apps, provide a novel way to address current gaps in primary prevention. There was a massive rise in recent years in mobile data usage and it was expected to grow by more than 50% by 2020. 91 Mobile health (mHealth) has become an increasingly popular mode of healthcare delivery by providing a novel way to address current gaps in primary prevention. Advances in mobile devices coupled with highspeed internet means phones are becoming more than communications devices and are acting more as hubs for educational purposes, as well as portals that other devices can communicate through. In 2017 there were approximately 325,000 mHealth apps, with an annual download of >3.7 billion. There is widespread evidence of the feasibility and efficacy of apps for behaviour change across a range of health behaviours 92,93 and outcomes including clinically meaningful reductions in blood pressure. 94 There also are a lot of apps aimed at generic health factors such as diet and exercise that decrease users' CVD risk, despite not being explicitly marketed for this purpose. Activity trackers like Fitbit TM have huge user bases (recent filings indicate there are approximately 19 million registered users with over 9 million "active" users 95 ), and research indicates that usage of wearable devices such as these does

show an increase in physical activity. 96, 97 Similarly, diet trackers and calorie counters such as MyFitnessPal TM also show efficacy when used on a regular basis. 98 Similar generic apps are also available for blood pressure, 99 smoking cessation 100,101 and other associated risk factors. However, there is a lack of proven effective, valid, internationally endorsed, comprehensive (providing not only risk estimates and their monitoring but also a user's risk profile and specific evidence-based recommendations for controlling those risks) and free of charge apps for primary prevention of specific major non-communicable disorders. To the best of our knowledge, in primary stroke prevention there is only one app that meets most of these requirements: the Stroke Riskometer TM app.

The Stroke Riskometer TM app is a free app owned and copyrighted by Auckland University of Technology, New Zealand. 102 The Stroke Riskometer TM algorithm was derived from the Framingham Stroke Risk Score 103 prediction algorithm and enhanced to include several additional major risk factors shown to be important for stroke, largely based on the INTERSTROKE study. 86 The Stroke Riskometer has been found to be comparable in performance for stroke prediction with FSRS and QStroke. 104 The app not only calculates 5-and 10-year absolute and relative risk of stroke occurrence but also provides patient-tailored recommendations for primary stroke prevention, which are based on the internationally recognised primary stroke and CVD prevention guidelines. 69, 105 It incorporates several evidence-based tools to promote behaviour change aligned with internationally recognised stroke prevention guidelines. 69,105 These include:

1) Provision of feedback on absolute risk of stroke within the next 5 to 10 years and compares a person's relative risk with those of a person of the same age and sex without risk factors. This approach has been demonstrated to motivate behaviour change when used in conjunction with other methods. 106,107 2) Employs tailored self-management strategies including goal setting to engage the person in behaviour modification. 108 3) Includes information on stroke risk factors and warning signs aligned with the internationally relevant Face, Arm, Speech, Time (FAST) international mass media campaign. 4) Uses reminders, known as "push notifications", to prompt users to achieve their goals.

Such reminders have been shown to increase adherence to programmes. 109 This mobile app applies "motivational mass strategy for primary stroke prevention" 110 regardless of the level of stroke/CVD risk, 72, 83 which is complementary to the high CVD risk strategy recommended by the WHO. 111 It has been endorsed by the World Stroke Organization, World Federation of Neurology, World Heart Federation, European Stroke Organisation and a number of national stroke organisations, including Australian Stroke Foundation, French Neuro-Vascular Society, and Chinese Stroke Society to name a few. The Stroke Riskometer TM app, its advantages and disadvantages and use/place in primary stroke/CVD prevention were discussed at various conferences and intensively reviewed, 72, 80, 84, 102, 104, 106, 107, 110, [112] [113] [114] [115] [116] [117] [118] [119] [120] [121] [122] [123] [124] [125] [126] [127] [128] [129] and the app is recommended for the world-wide use. [130] [131] [132] At the recent Latin American summit of Health Ministers, the Stroke Riskometer TM app was recommended for implementation across all Latin American countries. 114 Being available in 19 languages (English, Bengali, Bulgarian, Chinese [Mandarin], Croatian, Czech, French, German, Greek, Hindi, Italian, Malay, Nepali, Portuguese, Brazilian-Portuguese, Russian, Spanish, Swedish, Thai) this free mobile stroke education and prevention tool is already available to 5.3 billion people in their native languages. It has already been downloaded by >200,000 people from 78 countries.

A desktop version of the Stroke Riskometer TM app -called the PreventS Ó webapp -takes only a couple of minutes to complete by a clinician. Like the Stroke Riskometer app it also provides patient-tailored recommendations and risk monitoring for primary stroke prevention and other major NCDs (e.g., ischaemic heart disease, type II diabetes mellitus, renal disease, vascular dementia and some types of cancer) for the individual patient, but unlike the Stroke Riskometer app, these recommendations can be further edited by the clinician to make it more specific and appropriate for the individual. The clinician also has the option of printing out the patient's summary (including recommendations and risks dynamic), saving the data for further monitoring of the risks progress, or sharing it with another clinician. As such, the PreventS Ó webapp is the only decision-making tool for integrative primary stroke and other major NCDs currently available for clinicians. Further details on the Stroke Riskometer TM app and PreventS-MD TM webapp can be found on the Auckland University of Technology website. 67, 79 (estimates of the stroke cost were derived from the current publication) Supplement Figure 6 . Outline of the PreventS-MD TM cloud-based platform for clinicians. The PreventS-MD TM algorithms for calculation absolute and relative risks of stroke are based on the validated and internationally endorsed Stroke Riskometer TM app. 107, 110, 133 

Regional and Country-Specific Burden of Ischaemic Stroke, Intracerebral Haemorrhage and Subarachnoid Haemorrhage: A Systematic Analysis of the Global Burden of Disease Study

A contemporary and comprehensive analysis of the costs of stroke in the United States

Initial and subsequent 3-year cost after hospitalization for first acute ischemic stroke and intracerebral hemorrhage

Costs of hospitalization for stroke patients aged 18-64 years in the United States

Short-and longer-term health-care resource utilization and costs associated with acute ischemic stroke

Cost of stroke in Canada: a 1-year prospective study

The potential health and economic impact of improving stroke care standards for Australia

The Economic Impact of Stroke in Australia 2020. Sydney: Stroke Foundation and Deloitte-Access Economics

Direct medical cost of stroke in Singapore

Factors Influencing Costs of Inpatient Ischaemic Stroke Care in Singapore

The Economic Burden of Stroke Based on South Korea's National Health Insurance Claims Database

Modified Rankin scale as a determinant of direct medical costs after stroke

Lifetime cost of ischemic stroke in Germany: results and national projections from a population-based stroke registry: the Erlangen Stroke Project

Direct costs of patients with stroke can be continuously monitored on a national level: performance, effectiveness, and Costs of Treatment episodes in Stroke (PERFECT Stroke) Database in Finland

Cost Assessment of Inpatient Care Episodes of Stroke in Romania

Societal Cost of Ischemic Stroke in Romania: Results from a Retrospective County-Level Study

Cost-Utility Analysis of Magnetic Resonance Imaging Management of Patients with Acute Ischemic Stroke in a Spanish Hospital

Economic impact of patients admitted to stroke units in Spain

Economic burden of stroke in a large county in Sweden

Four-fold increase in direct costs of stroke survivors with spasticity compared with stroke survivors without spasticity: the first year after the event

The burden of stroke in the Netherlands: estimating quality of life and costs for 1 year poststroke

A cost analysis of the first year after stroke -early triage and inpatient rehabilitation may reduce long term costs

Cost of stroke in France

The economic burden of stroke in Italy. The EcLIPSE Study: Economic Longitudinal Incidence-based Project for Stroke Evaluation

Estimated societal costs of stroke in the UK based on a discrete event simulation

Costs of hospitalization for stroke from two urban health insurance claims data in Guangzhou City, southern China

Inpatient Cost of Stroke in Beijing: A Descriptive Analysis

Difference of hospital charges for stroke inpatients between hospitals with different levels and therapeutic modes in Beijing, China

Effect of health insurance on direct hospitalisation costs for in-patients with ischaemic stroke in China

The association of hospital competition with inpatient costs of stroke: Evidence from China

Variations and determinants of hospital costs for acute stroke in China

Medical Service Utilization and Direct Medical Cost of Stroke in Urban China

Direct medical cost of stroke: findings from a tertiary hospital in Malaysia

The cost of stroke in private hospitals in Brazil: a oneyear prospective study

The cost of stroke in a public hospital in Brazil: a oneyear prospective study

Acute treatment costs of intracerebral hemorrhage and ischemic stroke in Argentina

Cost of acute ischemic and hemorrhagic stroke in Turkey

Economic Aspects of Post-Stroke Rehabilitation: A Retrospective Data at a Traditional Medicine Hospital in Vietnam

Cost of acute care for ischemic stroke in Thailand

Cost of acute stroke care at a tertiary care hospital in Karachi, Pakistan

Cost of hospitalization for stroke in a lowmiddle-income country: Findings from a public tertiary hospital in the Philippines

Cost of acute stroke care in South Africa

Cost of stroke in low and middle income countries: a systematic review

Economic burden of stroke in a rural South African setting

Estimation and determinants of direct medical costs of ischaemic heart disease, stroke and hypertensive heart disease: evidence from two major hospitals in Cameroon

Cost of stroke from a tertiary center in northwest India

Emerging Areas of Stroke Rehabilitation Research in Low-and Middle-Income Countries

Cost and burden of informal caregiving of dependent older people in a rural Indian community

Informal Caregiving: A Growing Need for Inclusion in Stroke Rehabilitation

Patients' productivity losses and informal care costs related to ischemic stroke: a French population-based study

Informal care for stroke survivors: results from the North East Melbourne Stroke Incidence Study (NEMESIS)

Cost of informal caregiving associated with stroke among the elderly in the United States

Care Received by Elderly US Stroke Survivors May Be Underestimated

Incidence and outcome of subtypes of ischaemic stroke: initial results from the north East Melbourne stroke incidence study (NEMESIS)

Long-term costs of stroke using 10-year longitudinal data from the North East Melbourne Stroke Incidence Study

Direct Costs of Patients With Stroke Can Be Continuously Monitored on a National Level

Factors predictive of return to work after stroke in patients with mild-moderate disability in India

Return to work after mild-to-moderate stroke: work satisfaction and predictive factors

Return to Work after a Stroke in Working Age Persons; A Six-Year Follow Up

Determinants of return to work among Nigerian stroke survivors

Long-term increased risk of unemployment after young stroke: a long-term follow-up study

Lost Productivity in Stroke Survivors: An Econometrics Analysis

There Are Significant Business Costs to Replacing Employees

Attrition leads to employee turnover costs -a study of the Indian software industry

Transforming our world: the 2030 agenda for sustainable development

Global and regional burden of stroke during 1990-2010: findings from the Global Burden of Disease Study

a systematic analysis for the Global Burden of Disease Study

The global and regional burden of stroke

ACC/AHA Guideline on the Primary Prevention of Cardiovascular Disease: Executive Summary: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines

The burden of stroke in Africa: a glance at the present and a glimpse into the future

The global burden of neurological disorders: translating evidence into policy

What Is the Best Mix of Population-Wide and High-Risk Targeted Strategies of Primary Stroke and Cardiovascular Disease Prevention?

Global Alliance for Chronic Diseases

Hypertension in 2017-what is the right target?

Primary prevention of stroke: impact of healthy lifestyle

Primary preventive potential for stroke by avoidance of major lifestyle risk factors: The European prospective investigation into cancer and nutrition-heidelberg cohort

Strategy of prevention: lessons from cardiovascular disease

Secondary stroke prevention

The GBD 2016 Lifetime Risk of Stroke Collaborators. Global, Regional, and Country-Specific Lifetime Risks of Stroke

Prevention of stroke: a strategic global imperative

Global burden of stroke and risk factors in 188 countries, during 1990-2013: a systematic analysis for the Global Burden of Disease Study

Rose's Strategy of Preventive Medicine

WSO and WHF joint position statement on population-wide prevention strategies

Global prevention of stroke and dementia: the WSO Declaration

Dominant modifiable risk factors for stroke in Ghana and Nigeria (SIREN): a case-control study

Global and regional effects of potentially modifiable risk factors associated with acute stroke in 32 countries (INTERSTROKE): a case-control study

Primary prevention of stroke

Primary prevention of stroke by a healthy lifestyle in a high-risk group

Global, regional, and national burden of stroke, 1990-2016: a systematic analysis for the Global Burden of Disease Study

Unraveling the risk factors for spontaneous intracerebral hemorrhage among West Africans

Compound annual growth rate of global units and global mobile data traffic from

Efficacy of interventions that use apps to improve diet, physical activity and sedentary behaviour: a systematic review

Can Mobile Phone Apps Influence People's Health Behavior Change? An Evidence Review

Cardiorespiratory Improvements Achieved by American College of Sports Medicine's Exercise Prescription Implemented on a Mobile App

Tons of people are buying Fitbits, but are they actually using them? The Verge

Randomized Trial of a Fitbit-Based Physical Activity Intervention for Women

Persuasive technology in the real world: a study of long-term use of activity sensing devices for fitness

Diet and Physical Activity Apps: Perceived Effectiveness by App Users

Mobile Apps for Blood Pressure Monitoring: Systematic Search in App Stores and Content Analysis

A randomized controlled trial of a smoking cessation smartphone application with a carbon monoxide checker

Smoking cessation with smartphone applications (SWAPP): study protocol for a randomized controlled trial

New Strategy to Reduce the Global Burden of Stroke

Probability of stroke: A risk profile from the Framingham Study

The Stroke Riskometer App: validation of a data collection tool and stroke risk predictor

Guidelines for the primary prevention of stroke: A statement for healthcare professionals from the American Heart Association/American Stroke Association

Effectiveness of the Stroke Riskometer App for Primary Stroke Prevention: A Pilot RCT. Final Program of the International Stroke Conference

Presentation Number LBP18

Mobile technology for primary stroke prevention: a proof-of-concept pilot randomised controlled trial -a brief report

Effectiveness of multicultural health workers in chronic disease prevention and self-management in culturally and linguistically diverse populations: A systematic literature review

The Impact of Mobile Health Interventions on Chronic Disease Outcomes in Developing Countries: A Systematic Review

Primary prevention of cardiovascular disease through population-wide motivational strategies: insights from using smartphones in stroke prevention

World Health Organization cardiovascular disease risk charts: revised models to estimate risk in 21 global regions

Декларация по глобальной первичной профилактике инсульта и деменции Всемирной организации по борьбе с инсультом

Can we stop the stroke tsunami? Mitigating the barriers, amplifying the facilitators

Priorities to reduce the burden of stroke in Latin American countries

Multi-level community interventions for primary stroke prevention: A conceptual approach by the World Stroke Organization

Anthology of stroke epidemiology in the 20th and 21st centuries: Assessing the past, the present, and envisioning the future

Digital data (mHealth and social media)

Cut stroke in half: Polypill for primary prevention in stroke

Primary stroke prevention needs overhaul

French version of the Stroke Riskometer™ App: A new tool to reduce the burden of stroke

A new approach to stroke prevention in Russia

Stroke is largely preventable across the globe: where to next? The Lancet

The power of the App: can mobile-technology save lives?

Primary stroke prevention in China -a new approach

Putting your health in your own hands: A new strategy for primary stroke prevention

the Stroke Riskometer: A new tool to measure and monitor stroke risk

Stroke Prevention Worldwide--What Could Make It Work?

Approaches to Prevention of Cardiovascular Disease

A new paradigm for primary prevention strategy in people with elevated risk of stroke

Population Impact of Potentially Modifiable Risk Factors for Stroke

WSO and WFN joint position statement on population-wide prevention strategies

The Stroke Riskometer(TM) App: validation of a data collection tool and stroke risk predictor