Stroke Following Percutaneous Coronary Intervention
In one of the largest population-based case-cohort analyses published to date, we have shown that the incidence of ischaemic stroke complications in contemporary PCI practice is three-fold greater than that of haemorrhagic stroke. Ischaemic stroke complications have increased in incidence over time, whilst haemorrhagic stroke complications have decreased, driven through temporal changes in clinical, procedural, drug-treatment, and demographic factors. Furthermore, we demonstrate that both ischaemic and haemorrhagic strokes are rare, but have devastating complications with high 30-day mortality and in-hospital MACE rates, with ischaemic stroke independently associated with a five-fold increased odds and haemorrhagic stroke with a 14-fold increase in odds of 30-day mortality post PCI.
Our data show an incident in-hospital stroke rate of 0.13% in our unselected national cohort of patients undergoing PCI, with a 3-fold greater incidence of ischaemic strokes compared with haemorrhagic strokes. Studies such as the SYNTAX study have reported incident stroke rates of 0.2% at 30-days with the ACUITY trial reporting rates of up to 0.3%, whilst national registries such as the NCDR and international registries such as the EuroHeart survey have reported stroke rates between 0.2 and 0.4% respectively. Other, single centre, registry studies have reported rates of between 0.15 and 0.4%. Incident stroke rates will depend on the clinical characteristics of the cohort studied, with a two-fold greater incidence reported in patients undergoing PCI for ACS indications compared with the elective setting in the EuroHeart survey.
To the best of our knowledge, our analysis is the first to systematically study haemorrhagic and ischaemic stroke complications separately following PCI from a large-scale national perspective, with such data either not being available for analysis in previous reports derived from the NCDR or not presented in the EuroHeart survey. In a single-centre study of Hoffman et al. undertaken at the Mayo clinic that reported incident stroke complications of 0.37%, 90% of all stroke complications were ischaemic in origin whereas 7% were reported as haemorrhagic strokes, although predictors, outcomes and changes in incidence over time were not studied separately for each sub-type of stroke.
We have observed that incident stroke rates have increased over a 6-year timeframe in England and Wales, which has been driven by an increase in incident ischaemic stroke complications, whereas haemorrhagic stroke rates have decreased over a similar time frame. Whilst changes in incident stroke have not been previously reported from a national perspective, single-centre studies report varying results with both significant and non-significant decreases and non-significant increases in stroke rates over time. In the Swedish heart intensive care admission registry, haemorrhagic strokes were reported to decrease by over 50% over a 10-year period in acute myocardial infarction, driven largely through changes in thrombolysis practice, although in those patients who underwent primary PCI, rates of haemorrhagic strokes remained relatively stable over time.
We have identified a number of independent predictors of ischaemic stroke complications such as female gender, increasing age, prevalent valvular heart disease, a previous history of stroke, the use of circulatory support devices, and PCI in the setting of both NSTEMI and STEMI. Similarly, previous analyses have also identified female gender, previous stroke history, age, circulatory support devices, thrombus containing lesions, and PCI for ACS indications as independent predictors of stroke complications. Our analysis suggests that the presence of valvular heart disease is amongst the strongest predictors of ischaemic stroke complications. Similarly, the smaller euro heart registry consisting of 46 888 patients has documented that prevalent valvular heart disease is associated with a two-fold increase in stroke risk although this did not reach statistical significance. The BCIS dataset does not differentiate between native valvular heart disease and its severity, and prosthetic valves that might increase risk of thrombotic embolization and ischaemic stroke complications, particularly if anticoagulants have been discontinued for the PCI procedure. In addition, valvular heart disease such as aortic stenosis and mitral valve disease are well-described independent predictors of stroke risk. Finally, the presence of valvular heart disease may be a surrogate for prevalent atrial fibrillation that is unrecorded in the BCIS dataset that would in itself increase stroke risk.
Our analysis suggests that the observed increase in incident ischaemic stroke complications over time may relate to temporal changes in patient demographics and PCI practice. The average age of the cohort has increased over time, along with the prevalence of valvular heart disease, history of stroke, ACS indications for PCI, and increases in circulatory support device use over time that would all contribute to the increases in stroke rates reported.
We identify both a number of shared risk factors such as age, ACS indications for PCI and valvular heart disease as important independent predictors of haemorrhagic stroke complications as well as additional risk factors such as warfarin use and thrombolysis that are independent predictors for only haemorrhagic strokes. We also identify GPIIb/IIIa and haemodynamic instability such as cardiogenic shock presentation, receipt of ventilatory or circulatory support as univariate predictors of haemorrhagic stroke. Whilst age, thrombolysis and warfarin use are recognized risk factors for haemorrhagic strokes, the mechanism through which the presence of valvular heart disease independently predicts stroke is not clear. Valvular heart disease may be a surrogate for anticoagulant use, that may not have been fully captured in the BCIS dataset, both from the perspective of anticoagulant for thromboprophylaxis for atrial fibrillation (that is associated with prevalent valvular heart disease) and for prosthetic heart valves. Changes in both platelet physiology and coagulation pathways are well documented in cardiogenic shock that might contribute to the increased haemorrhagic complications observed in cardiogenic shock on univariate analysis.
It is of interest that we have identified ACS indications for PCI as independent predictors for both ischaemic and haemorrhagic stroke complications, analogous to the observation that both STEMI and NSTEMI are strong independent predictors of both stent thrombosis and major bleeding complications during PCI which may relate to increased thrombus burden in lesions treated for ACS indications, higher anti-platelet reactivity hence the requirement for more aggressive anti-platelet and anticoagulant regime use. Similarly, prevalent valvular heart disease is a strong independent predictor of both ischaemic and haemorrhagic stroke complications. As described, patients with prevalent valvular heart disease within the BCIS database are a heterogeneous cohort of patients consisting of patients with native valvular heart disease of varying severity and patients with prosthetic heart valves. These patients with prevalent valvular heart disease will have a significant prevalence of atrial fibrillation that is not captured in the BCIS dataset and a significant proportion of such patients will be anti-coagulated both for thromboprophylaxis against AF as well as for prosthetic heart valves. This heterogeneous cohort of patients will have important risk factors for both haemorrhagic and ischaemic stroke complications, hence the importance of valvular heart disease as a predictor of both haemorrhagic and ischaemic stroke complications.
It is likely that competing risks contribute overall to the observed decline in haemorrhagic stroke rates over time. Some of the independent predictors of haemorrhagic stroke have increased in prevalence over time such as warfarin use, valvular heart disease, STEMI PCI, and average age which would tend to favour an increase in haemorrhagic strokes, whilst other important independent predictors such as thrombolysis have declined over time, which reflect changes in PCI practice particularly with a move from rescue PCI, facilitated PCI or PCI following successful thrombolysis in STEMI towards primary PCI more recently. Indeed, in the Swedish heart intensive care admission registry, haemorrhagic strokes were reported to decrease by over 50% over a 10-year period in acute myocardial infarction, driven largely through changes in thrombolysis practice.
Our analysis suggests that stroke complications independently predict adverse mortality and MACE outcomes, and that haemorrhagic stroke complications are independently associated with a 10-fold increase in odds of 30-day mortality, whilst ischaemic stroke complications increase this 4-fold. Whilst most previous studies have not reported the prognostic impact of stroke complications according to whether the aetiology is ischaemic or haemorrhagic, stroke was associated with a 12-fold unadjusted increase in in-hospital mortality and a 10-fold increase in MACE in the EuroHeart registry. Similarly data derived from the ACUITY trial has reported unadjusted eight-fold increases in 30-day mortality and 2.4-fold increases in MACE events following stroke. In a patient-level pooled analysis of the seven ISAR trials, intra-cranial bleeding was associated with a 22-fold unadjusted increase risk in 1-year mortality, although these data were based on a total of nine intra-cranial bleeding events. These observations are in keeping with our reported unadjusted rates of nine-fold increase in 30-day mortality for ischaemic strokes and 44-fold increase for haemorrhagic strokes. Indeed in an analysis undertaken at the Mayo clinic in which cause of death was studied in patients undergoing PCI over a 17-year period, stroke events accounted for 6% of all mortalities during this period of time, whilst another single-centre study has suggested that neurological complications contribute to 15% of mortalities post PCI. We have also attempted to separate the prognostic impact of ischaemic strokes and TIA and shown that both TIA and stroke are independently associated with increased odds of 30-day mortality and MACE in our main analysis and propensity score matching. Interestingly, when the impact of stroke vs. TIA on 1-year all-cause mortality was separately considered in an analysis of the Acuity trial, mortality was significantly increased after the occurrence of stroke (HR 4.16, 2.13–8.13, P < 0.001) but not after TIA (HR 1.52, 0.21–10.83, P = 0.68) although this may have been limited by the statistical power of this analysis as evidenced by the wide confidence intervals. Finally, whilst we have attempted to adjust for differences in baseline covariates between patients who sustained stroke complications and those who did not, and attempted to reduce the influence of the potential confounders through the use of propensity score matching, our association between stroke complications and increased odds of mortality and MACE outcomes does not infer a causal relationship, but maybe driven by unmeasured confounders or unrecorded measures of frailty.
Whilst the absolute 30-day mortality rates for ischaemic and haemorrhagic strokes of 16 and 48%, respectively, are striking, even those patients who survive would have significant life-changing disabilities, for example in the SYNTAX study, close to 60% of patients who experienced a stroke complication and survived were left with residual neurological deficits post hospital discharge. Similarly, in the single-centre analysis of Dukkipati et al., 70% of patients who survived the stroke were left with significant neurological deficit with skilled home care needed in 26% of patients, nursing home or assisted living in 9%, and inpatient rehabilitation in 22%. Our group has previously reported in an analysis of over 3000 community patients admitted with stroke in England, similar 30-day mortality rates to those reported here, with 18% 30-day mortality for ischaemic strokes and 42% mortality for haemorrhagic strokes, suggesting that the adverse outcomes associated with stroke complications following PCI are similar to those observed in the community. Whilst community stroke registries have suggested that haemorrhagic strokes account for ~15% of all strokes in the UK, haemorrhagic strokes account for 25% in our analysis, which may relate to the potent drug therapies used in PCI that predispose to haemorrhagic complications.
Our study has several strengths. The BCIS dataset includes >95% of all PCI procedures performed in the UK which therefore reflects a national, real-world experience that includes high-risk patients encountered in daily interventional practice who are often excluded from randomized controlled trials. In our current analysis, we have studied neurovascular complications in over 400 000 patients and as such this analysis represents the one of the largest analyses of changes in incident neurovascular complications over time and due to its size is able to: capture relatively rare complications of PCI; and for the first time from a national perspective, systematically study changes, predictors and outcomes of both ischaemic and haemorrhagic stroke complications.
There are some limitations in this study, which are worth highlighting. Our dataset does not capture the timing of the stroke complications. Furthermore, the BCIS dataset does not provide information regarding the neurological deficit sustained following a stroke. In the study of Dukkipati et al. the three most common neurological manifestations were motor (35%), speech (33%), and mental state changes (32%), whilst in the study of Hoffman the most frequently occurring clinical features were limb motor weakness (49%), speech impairment (39%), visual disturbance (20%), and facial droop (18%), and unresponsiveness was present in 12%. Similar to previous large national and multi-national studies, BCIS does not record prevalent atrial fibrillation (AF), which is a similar limitation encountered in both analysis derived from the EuroHeart Survey and NCDR dataset. Nevertheless, whilst AF contributes to about 15% of ischaemic strokes and is hence one of the most important predictors stroke in the community setting, the role of AF as a risk factor for stroke following PCI has not been widely studied. Our analysis suggests that valvular heart disease is one of the most powerful predictors of ischaemic stroke events that may be a surrogate marker for incident AF. Interestingly in one analysis where prevalent AF data was recorded, whilst AF was more common in patients who sustained a stroke complication following PCI, prevalent AF did not independently increase risk of stroke in this setting.
Thirdly, diagnosis of stroke is self-reported by individual operators with no external validation hence there is the potential for under-reporting of neurological events, although our reported incident stroke rates are similar in magnitude to those reported in the national NCDR and the SCAAR datasets derived from USA and Sweden respectively.
Finally, our analysis reports in-hospital stroke events that may represent periods of only a few hours of time for day case procedures or a few days for other inpatient cases. Previous findings from single-centre data suggest that 60% of strokes occur within 24 h and 20% between 24 and 48 h with the remaining events occurring after 48 h that may explain differences in reported stroke rates in our analysis and those reported in randomized trials that report stroke events up to 30-days post index procedure.
In conclusion, we have shown that the incidence of ischaemic stroke complications in contemporary PCI practice is three-fold greater than that of haemorrhagic stroke, with ischaemic stroke complications increasing in incidence over time, whilst haemorrhagic stroke complications have decreased, and that these changes in incident rates have been driven through changes in clinical, procedural, drug-treatment, and demographic factors over time. Furthermore, we confirm that both ischaemic and haemorrhagic strokes are rare but devastating complications with high 30-day mortality and in-hospital MACE rates.
Discussion
In one of the largest population-based case-cohort analyses published to date, we have shown that the incidence of ischaemic stroke complications in contemporary PCI practice is three-fold greater than that of haemorrhagic stroke. Ischaemic stroke complications have increased in incidence over time, whilst haemorrhagic stroke complications have decreased, driven through temporal changes in clinical, procedural, drug-treatment, and demographic factors. Furthermore, we demonstrate that both ischaemic and haemorrhagic strokes are rare, but have devastating complications with high 30-day mortality and in-hospital MACE rates, with ischaemic stroke independently associated with a five-fold increased odds and haemorrhagic stroke with a 14-fold increase in odds of 30-day mortality post PCI.
Our data show an incident in-hospital stroke rate of 0.13% in our unselected national cohort of patients undergoing PCI, with a 3-fold greater incidence of ischaemic strokes compared with haemorrhagic strokes. Studies such as the SYNTAX study have reported incident stroke rates of 0.2% at 30-days with the ACUITY trial reporting rates of up to 0.3%, whilst national registries such as the NCDR and international registries such as the EuroHeart survey have reported stroke rates between 0.2 and 0.4% respectively. Other, single centre, registry studies have reported rates of between 0.15 and 0.4%. Incident stroke rates will depend on the clinical characteristics of the cohort studied, with a two-fold greater incidence reported in patients undergoing PCI for ACS indications compared with the elective setting in the EuroHeart survey.
To the best of our knowledge, our analysis is the first to systematically study haemorrhagic and ischaemic stroke complications separately following PCI from a large-scale national perspective, with such data either not being available for analysis in previous reports derived from the NCDR or not presented in the EuroHeart survey. In a single-centre study of Hoffman et al. undertaken at the Mayo clinic that reported incident stroke complications of 0.37%, 90% of all stroke complications were ischaemic in origin whereas 7% were reported as haemorrhagic strokes, although predictors, outcomes and changes in incidence over time were not studied separately for each sub-type of stroke.
We have observed that incident stroke rates have increased over a 6-year timeframe in England and Wales, which has been driven by an increase in incident ischaemic stroke complications, whereas haemorrhagic stroke rates have decreased over a similar time frame. Whilst changes in incident stroke have not been previously reported from a national perspective, single-centre studies report varying results with both significant and non-significant decreases and non-significant increases in stroke rates over time. In the Swedish heart intensive care admission registry, haemorrhagic strokes were reported to decrease by over 50% over a 10-year period in acute myocardial infarction, driven largely through changes in thrombolysis practice, although in those patients who underwent primary PCI, rates of haemorrhagic strokes remained relatively stable over time.
We have identified a number of independent predictors of ischaemic stroke complications such as female gender, increasing age, prevalent valvular heart disease, a previous history of stroke, the use of circulatory support devices, and PCI in the setting of both NSTEMI and STEMI. Similarly, previous analyses have also identified female gender, previous stroke history, age, circulatory support devices, thrombus containing lesions, and PCI for ACS indications as independent predictors of stroke complications. Our analysis suggests that the presence of valvular heart disease is amongst the strongest predictors of ischaemic stroke complications. Similarly, the smaller euro heart registry consisting of 46 888 patients has documented that prevalent valvular heart disease is associated with a two-fold increase in stroke risk although this did not reach statistical significance. The BCIS dataset does not differentiate between native valvular heart disease and its severity, and prosthetic valves that might increase risk of thrombotic embolization and ischaemic stroke complications, particularly if anticoagulants have been discontinued for the PCI procedure. In addition, valvular heart disease such as aortic stenosis and mitral valve disease are well-described independent predictors of stroke risk. Finally, the presence of valvular heart disease may be a surrogate for prevalent atrial fibrillation that is unrecorded in the BCIS dataset that would in itself increase stroke risk.
Our analysis suggests that the observed increase in incident ischaemic stroke complications over time may relate to temporal changes in patient demographics and PCI practice. The average age of the cohort has increased over time, along with the prevalence of valvular heart disease, history of stroke, ACS indications for PCI, and increases in circulatory support device use over time that would all contribute to the increases in stroke rates reported.
We identify both a number of shared risk factors such as age, ACS indications for PCI and valvular heart disease as important independent predictors of haemorrhagic stroke complications as well as additional risk factors such as warfarin use and thrombolysis that are independent predictors for only haemorrhagic strokes. We also identify GPIIb/IIIa and haemodynamic instability such as cardiogenic shock presentation, receipt of ventilatory or circulatory support as univariate predictors of haemorrhagic stroke. Whilst age, thrombolysis and warfarin use are recognized risk factors for haemorrhagic strokes, the mechanism through which the presence of valvular heart disease independently predicts stroke is not clear. Valvular heart disease may be a surrogate for anticoagulant use, that may not have been fully captured in the BCIS dataset, both from the perspective of anticoagulant for thromboprophylaxis for atrial fibrillation (that is associated with prevalent valvular heart disease) and for prosthetic heart valves. Changes in both platelet physiology and coagulation pathways are well documented in cardiogenic shock that might contribute to the increased haemorrhagic complications observed in cardiogenic shock on univariate analysis.
It is of interest that we have identified ACS indications for PCI as independent predictors for both ischaemic and haemorrhagic stroke complications, analogous to the observation that both STEMI and NSTEMI are strong independent predictors of both stent thrombosis and major bleeding complications during PCI which may relate to increased thrombus burden in lesions treated for ACS indications, higher anti-platelet reactivity hence the requirement for more aggressive anti-platelet and anticoagulant regime use. Similarly, prevalent valvular heart disease is a strong independent predictor of both ischaemic and haemorrhagic stroke complications. As described, patients with prevalent valvular heart disease within the BCIS database are a heterogeneous cohort of patients consisting of patients with native valvular heart disease of varying severity and patients with prosthetic heart valves. These patients with prevalent valvular heart disease will have a significant prevalence of atrial fibrillation that is not captured in the BCIS dataset and a significant proportion of such patients will be anti-coagulated both for thromboprophylaxis against AF as well as for prosthetic heart valves. This heterogeneous cohort of patients will have important risk factors for both haemorrhagic and ischaemic stroke complications, hence the importance of valvular heart disease as a predictor of both haemorrhagic and ischaemic stroke complications.
It is likely that competing risks contribute overall to the observed decline in haemorrhagic stroke rates over time. Some of the independent predictors of haemorrhagic stroke have increased in prevalence over time such as warfarin use, valvular heart disease, STEMI PCI, and average age which would tend to favour an increase in haemorrhagic strokes, whilst other important independent predictors such as thrombolysis have declined over time, which reflect changes in PCI practice particularly with a move from rescue PCI, facilitated PCI or PCI following successful thrombolysis in STEMI towards primary PCI more recently. Indeed, in the Swedish heart intensive care admission registry, haemorrhagic strokes were reported to decrease by over 50% over a 10-year period in acute myocardial infarction, driven largely through changes in thrombolysis practice.
Our analysis suggests that stroke complications independently predict adverse mortality and MACE outcomes, and that haemorrhagic stroke complications are independently associated with a 10-fold increase in odds of 30-day mortality, whilst ischaemic stroke complications increase this 4-fold. Whilst most previous studies have not reported the prognostic impact of stroke complications according to whether the aetiology is ischaemic or haemorrhagic, stroke was associated with a 12-fold unadjusted increase in in-hospital mortality and a 10-fold increase in MACE in the EuroHeart registry. Similarly data derived from the ACUITY trial has reported unadjusted eight-fold increases in 30-day mortality and 2.4-fold increases in MACE events following stroke. In a patient-level pooled analysis of the seven ISAR trials, intra-cranial bleeding was associated with a 22-fold unadjusted increase risk in 1-year mortality, although these data were based on a total of nine intra-cranial bleeding events. These observations are in keeping with our reported unadjusted rates of nine-fold increase in 30-day mortality for ischaemic strokes and 44-fold increase for haemorrhagic strokes. Indeed in an analysis undertaken at the Mayo clinic in which cause of death was studied in patients undergoing PCI over a 17-year period, stroke events accounted for 6% of all mortalities during this period of time, whilst another single-centre study has suggested that neurological complications contribute to 15% of mortalities post PCI. We have also attempted to separate the prognostic impact of ischaemic strokes and TIA and shown that both TIA and stroke are independently associated with increased odds of 30-day mortality and MACE in our main analysis and propensity score matching. Interestingly, when the impact of stroke vs. TIA on 1-year all-cause mortality was separately considered in an analysis of the Acuity trial, mortality was significantly increased after the occurrence of stroke (HR 4.16, 2.13–8.13, P < 0.001) but not after TIA (HR 1.52, 0.21–10.83, P = 0.68) although this may have been limited by the statistical power of this analysis as evidenced by the wide confidence intervals. Finally, whilst we have attempted to adjust for differences in baseline covariates between patients who sustained stroke complications and those who did not, and attempted to reduce the influence of the potential confounders through the use of propensity score matching, our association between stroke complications and increased odds of mortality and MACE outcomes does not infer a causal relationship, but maybe driven by unmeasured confounders or unrecorded measures of frailty.
Whilst the absolute 30-day mortality rates for ischaemic and haemorrhagic strokes of 16 and 48%, respectively, are striking, even those patients who survive would have significant life-changing disabilities, for example in the SYNTAX study, close to 60% of patients who experienced a stroke complication and survived were left with residual neurological deficits post hospital discharge. Similarly, in the single-centre analysis of Dukkipati et al., 70% of patients who survived the stroke were left with significant neurological deficit with skilled home care needed in 26% of patients, nursing home or assisted living in 9%, and inpatient rehabilitation in 22%. Our group has previously reported in an analysis of over 3000 community patients admitted with stroke in England, similar 30-day mortality rates to those reported here, with 18% 30-day mortality for ischaemic strokes and 42% mortality for haemorrhagic strokes, suggesting that the adverse outcomes associated with stroke complications following PCI are similar to those observed in the community. Whilst community stroke registries have suggested that haemorrhagic strokes account for ~15% of all strokes in the UK, haemorrhagic strokes account for 25% in our analysis, which may relate to the potent drug therapies used in PCI that predispose to haemorrhagic complications.
Our study has several strengths. The BCIS dataset includes >95% of all PCI procedures performed in the UK which therefore reflects a national, real-world experience that includes high-risk patients encountered in daily interventional practice who are often excluded from randomized controlled trials. In our current analysis, we have studied neurovascular complications in over 400 000 patients and as such this analysis represents the one of the largest analyses of changes in incident neurovascular complications over time and due to its size is able to: capture relatively rare complications of PCI; and for the first time from a national perspective, systematically study changes, predictors and outcomes of both ischaemic and haemorrhagic stroke complications.
There are some limitations in this study, which are worth highlighting. Our dataset does not capture the timing of the stroke complications. Furthermore, the BCIS dataset does not provide information regarding the neurological deficit sustained following a stroke. In the study of Dukkipati et al. the three most common neurological manifestations were motor (35%), speech (33%), and mental state changes (32%), whilst in the study of Hoffman the most frequently occurring clinical features were limb motor weakness (49%), speech impairment (39%), visual disturbance (20%), and facial droop (18%), and unresponsiveness was present in 12%. Similar to previous large national and multi-national studies, BCIS does not record prevalent atrial fibrillation (AF), which is a similar limitation encountered in both analysis derived from the EuroHeart Survey and NCDR dataset. Nevertheless, whilst AF contributes to about 15% of ischaemic strokes and is hence one of the most important predictors stroke in the community setting, the role of AF as a risk factor for stroke following PCI has not been widely studied. Our analysis suggests that valvular heart disease is one of the most powerful predictors of ischaemic stroke events that may be a surrogate marker for incident AF. Interestingly in one analysis where prevalent AF data was recorded, whilst AF was more common in patients who sustained a stroke complication following PCI, prevalent AF did not independently increase risk of stroke in this setting.
Thirdly, diagnosis of stroke is self-reported by individual operators with no external validation hence there is the potential for under-reporting of neurological events, although our reported incident stroke rates are similar in magnitude to those reported in the national NCDR and the SCAAR datasets derived from USA and Sweden respectively.
Finally, our analysis reports in-hospital stroke events that may represent periods of only a few hours of time for day case procedures or a few days for other inpatient cases. Previous findings from single-centre data suggest that 60% of strokes occur within 24 h and 20% between 24 and 48 h with the remaining events occurring after 48 h that may explain differences in reported stroke rates in our analysis and those reported in randomized trials that report stroke events up to 30-days post index procedure.
In conclusion, we have shown that the incidence of ischaemic stroke complications in contemporary PCI practice is three-fold greater than that of haemorrhagic stroke, with ischaemic stroke complications increasing in incidence over time, whilst haemorrhagic stroke complications have decreased, and that these changes in incident rates have been driven through changes in clinical, procedural, drug-treatment, and demographic factors over time. Furthermore, we confirm that both ischaemic and haemorrhagic strokes are rare but devastating complications with high 30-day mortality and in-hospital MACE rates.
SHARE
