The growing burden of stroke has led to an increase in healthcare expenditure. In addition to the high hospitalisation rates, long hospital stays, and need for rehabilitation associated with stroke,1 the disease continues to be the leading cause of long-term disability.2 Advances in acute ischaemic stroke care have increased survival rates, but this has also resulted in greater numbers of survivors presenting sequelae and dependence.

Several studies have evaluated the cost of stroke, with results varying greatly according to context, perspectives, study design, and data sources.3–5 However, no study has quantified the cost of stroke using patient data in our setting, a public, free-to-use healthcare system with universal coverage. Furthermore, few studies have compared the costs associated with stroke management against healthcare costs before stroke to calculate the incremental costs potentially attributable to stroke.

The information technology systems of the Catalan Health Service gather data on healthcare resource use from different sources, providing population-based data from all healthcare system users. With the system, healthcare resource use can be quantified and the direct costs to the healthcare system can be calculated.6,7

The purpose of this study is to quantify and describe healthcare expenditure (excluding costs associated with emergency medical transport and orthotic/prosthetic material) associated with ischaemic stroke among users of the Catalan Health Service. We quantified expenditure using a bottom-up approach based on resource use by all individuals presenting stroke between 2012 and 2016, conducting retrospective, longitudinal follow-up from one year before stroke to 3 years after the episode.

We identified a total of 36 044 patients older than 18 years who were admitted due to ischaemic stroke and had no history of stroke in the previous 7 years (Fig. 1). Table 1 shows the baseline characteristics of the study population. Women accounted for 47.4% of the sample; mean age was 71.4 years (SD: 13) in men and 78.4 (12.6) years in women. Fig. 2 shows the age and sex distribution of the sample. The in-hospital mortality rate was 9%; 32 850 patients were discharged; and 74% of patients survived the first year, 68% survived the second year, and 63% survived the third year (Fig. 3).

Figure 1.

Patient flow chart.

(0.22MB).

Figure 2.

Population pyramid.

(0.33MB).

Figure 3.

Survival and outcomes over the 3 years after stroke in our patient population.

(0.37MB).

Table 2 shows the utilisation rates for different healthcare resources, by patient-year. Mean hospitalisation time was 2.3 days per patient in the year before stroke, 17.1 days in the first year after stroke, and 3.1 and 2.8 days in the second and third years after stroke, respectively. Mean time spent at a long-stay hospital was 2.2 days per patient in the year before stroke, 35.5 days in the first year after stroke, and 4.6 and 3.9 days in the second and third years after stroke, respectively.

Online Supplementary material Figs. I-VII show the progression of the monthly use rates of the most relevant resources; a change in resource use can be observed in the months following stroke. The use of some resources increases in the months before stroke. For example, primary care consultations (Fig. I) increased from 120 to 150 per 100 patients/month before admission due to stroke, hospital admissions (Fig. II) increased from 2.5 to 5 per 100 patients/month, and emergency department visits (Fig. IV) increased from 7.5 to 12 per 100 patients/month.

Taking the year before stroke as a reference, significant increases (rate ratios > 1) can be observed in the utilisation rate of nearly all healthcare resources (Table 2), with the exception of use of non-steroidal anti-inflammatory drugs, for which utilisation rates decreased.

Table 3 presents the expenditure associated with the healthcare resources used. Resource use in the year before stroke was associated with a cost of €3230 per patient. In the first year after stroke, healthcare expenditure amounted to €11 060 per patient-year: €4434 corresponded to the costs associated with hospitalisation due to acute stroke plus any subsequent hospitalisations during the first year, and €2289 corresponded to the costs associated with institutionalisation in long-stay hospitals. During the second and third years, total expenses amounted to €4104 and €3878, respectively. Fig. 4 shows monthly healthcare expenditure per patient on different healthcare resources during the study period.

Figure 4.

Mean monthly expenditure per patient during the year before stroke and the 3 years after stroke, by healthcare resource.

(0.28MB).

Taking the year prior to stroke as a reference, healthcare expenditure was 3.5 times greater the first year after stroke, and 28% and 20% higher in the second and third years, respectively. Fig. 5 shows the distribution of incremental costs by resource category (hospital care, primary care, long-stay hospital care or rehabilitation, and pharmacological treatment).

Figure 5.

Difference between the mean annual expenditure in each of the first 3 years following stroke and in the year before stroke, and percentage of expenditure attributed to each type of resource.

(0.26MB).

This study estimated the healthcare expenditure associated with management of patients with ischaemic stroke and attended under the Catalan Health Service, and quantified the proportion of the cost that is attributable to stroke. Average expenditure per patient was €11 060 the first year (€7829 more than in the year before stroke). Healthcare expenditure decreased in the second and third years, stabilising at approximately €340 per month after 18 months (€100 more than before stroke).

Increases in expenditure mainly result from the initial hospitalisation (during the first year) and additional hospital admissions (in subsequent years). During the second and third years after stroke, expenditure associated with hospitalisation decreases considerably; the costs of consultations (outpatient and primary care consultations) and pharmacological treatment do not decrease during this period, however. The percentage of total expenditure attributed to long-stay hospital care and rehabilitation is greater in the first year after stroke, but continues to be a significant component of healthcare expenditure in subsequent years. In the second and third years, a considerable percentage of the incremental costs is attributable to pharmacological treatment: in the third year after stroke, annual drug expenditure per patient is €257 higher than before stroke.

In general terms, 40% of the incremental expenditure potentially attributable to stroke corresponds to drug dispensing, 26% to hospitalisation, and 24% to long-stay hospital care and rehabilitation. The increased use of hospital resources after stroke has already been studied in other settings.9,10 Although an analysis of the causes of readmission is beyond the scope of this study, stroke recurrence and other cardiovascular events are known to account for a considerable proportion of readmissions.10,11 The cost associated with stroke the first year after the event is addressed in great detail in the CONOCES study,3 a prospective study collecting data from 16 Spanish hospitals for the year 2012. Several other studies have analysed the costs of stroke, with estimates varying greatly (€5337-€27 711) depending on the context, perspective, study design, and data sources.3–5 The CONOCES study3 estimated the average direct healthcare cost at €8491 for the first year after stroke. This figure is similar to our estimate, demonstrating that administrative databases constitute a reliable source of information for estimating resource use and healthcare expenditure.

In addition to the well-known age-related increase in healthcare resource use, expenditure on these patients increases substantially after stroke. Furthermore, although resource use decreases after the acute and subacute phases, expenditure on some resources is far from reaching baseline (pre-stroke) levels. These include the costs associated with hospital care, specialised care (outpatient consultations), pharmacological treatment (particularly antithrombotics and other drugs for secondary prevention of cardiovascular disease, and, interestingly, use of analgesics and antidepressants, probably as a result of disability), and resources for patients with disability (long-stay hospitals, rehabilitation, non-emergency medical transport). Therefore, while average expenditure stabilises within 2 years of stroke, annual expenditure in the third year continues to be substantially greater than before stroke.

Patients with stroke are usually individuals of advanced age with multiple comorbidities, which means that many will already be using a considerable amount of healthcare resources prior to stroke. In fact, healthcare expenditure before stroke in our sample amounted to €3130 per patient; this is 3 times greater than the average expenditure per patient in 2014 (€990) and comparable to the average expenditure for patients older than 80 years.8 The value of comparing healthcare expenditure before and after stroke lies in the ability to estimate the costs attributable to stroke.

Although it is beyond the scope of the present study, a detailed analysis of the reasons for hospitalisation before stroke (which may include other manifestations of cardiovascular disease and episodes of atrial fibrillation) may enable the identification of patients at greater risk of stroke, which would in turn help in developing follow-up protocols for certain patient profiles to prevent or delay ischaemic stroke. This approach may be adopted by future studies using clinical and administrative databases.

Unlike similar studies analysing healthcare costs before and after stroke,9,11,12 our study also analysed non-hospital healthcare costs, which represent a considerable percentage of overall healthcare expenditure.4 In any case, our study did not analyse the social costs of disability, assumed by the healthcare system, and the costs assumed by patients and their families (informal care costs,13 direct costs associated with home adaptations, and indirect costs associated with patients’ or caregivers’ loss of productivity at work), which are believed to be even greater than healthcare costs.3

Population databases provide a large quantity of information relevant to the healthcare system (quality indicators), enabling the generation of new hypotheses for more detailed studies of clinical interest. For example, we were surprised to observe an increase in the use of certain resources in the months prior to stroke (hospitalisation, emergency department visits, and primary care consultations), which may have been associated with minor strokes or transient ischaemic attacks before the stroke that motivated the initial hospitalisation.

Population databases also enable the analysis of a complete patient cohort; the MUSSCAT registry provides exhaustive data on healthcare contacts and resource use, with the exception of emergency medical transport and orthotic/prosthetic material; these resources are relevant for patients with stroke, and exclusion of these data may have led to an underestimation of healthcare expenditure. Our cost estimates may be comparable to the actual healthcare system expenditure, since most resources are linked to a billing record, with the exception of hospitalisation and primary care consultations; the costs associated with these resources are estimated based on diagnosis-related groups14 and type of consultation (type of healthcare professional, home/outpatient consultation), respectively. As mentioned previously, healthcare expenditure in the first year after stroke in our study is consistent with estimates from a previous study conducted in our setting3; furthermore, the mortality rate observed in our sample is similar to those reported by specific studies,15,16 which supports the validity of administrative data for these purposes.

However, the use of administrative data is not without limitations. Although registries undergo regular quality controls, the presence of coding errors cannot be ruled out. Furthermore, administrative databases do not gather data on stroke severity or sequelae; resource use as a function of these variables cannot therefore be analysed. Future studies may address this limitation by combining administrative data with clinical data from the stroke registries implemented in Catalonia in recent years15–17; such registries may also be expanded to cover the whole Spanish population with the development of initiatives similar to the Sentinel Stroke National Audit Programme, which gathers data from England, Wales, and Northern Ireland. Our results cannot be extrapolated to other settings since resource costs may vary considerably between healthcare systems,18,19 potentially resulting in substantial differences in the costs associated with ischaemic stroke between countries (ranging from $2822 in Eastern European countries to $22 377 in the United Kingdom), or even between regions within the same country ($7309 vs $146 149 in the United States).4 However, data on resource use, the magnitude of incremental costs, and the percentage of these costs attributable to the use of each resource may be extrapolated to other countries with similar healthcare systems (public, free to use, and universal).

In conclusion, increased healthcare expenditure after ischaemic stroke is mainly explained by the initial need for hospitalisation, with costs decreasing a year after stroke but remaining higher than baseline levels. Information from population databases may be valuable for generating hypotheses for clinical studies and for improving the organisation of stroke care.

A. Ribera participated in study conception and design and in data interpretation, and drafted the first version of the manuscript.

E. Vela and M. Clèries participated in study conception and design and in data analysis and interpretation, and critically reviewed the manuscript.

A. García-Altés and S. Abilleira participated in study conception and design and in data interpretation, and critically reviewed the manuscript.

All authors approved the final version of the manuscript and take full responsibility for the accuracy and integrity of the data presented in the study.

This study has received no funding.

None.