Abstract
Introduction
Evidence regarding the prognostic significance of thrombocytopenia in spontaneous intracerebral haemorrhage (ICH) remains limited and inconsistent. We investigated its associations with haematoma expansion (HE), functional outcomes and mortality.
Patients and methods
We retrospectively analysed prospectively collected data from consecutive patients with spontaneous ICH in a single-centre registry between August 2022 and March 2025. Thrombocytopenia was defined as an admission platelet count <150 × 109/L. HE was defined as a haematoma-volume increase >6 mL or >33% on follow-up noncontrast computed tomography. Poor functional outcome was defined as a modified Rankin Scale (mRS) score of 3–6 at 3 months. Associations were assessed using multivariable logistic regression, restricted cubic splines, ordinal regression and sensitivity analyses.
Results
Among 496 patients, 106 (21.4%) had thrombocytopenia; 368 were included in the HE analysis. After adjustment, thrombocytopenia showed no statistically significant association with HE (adjusted OR 0.97, 95% CI, 0.52–1.76; P = .920), 3-month poor functional outcome (adjusted OR 0.88, 95% CI, 0.46–1.66; P = .693), or 3-month mortality (adjusted OR 1.43, 95% CI, 0.70–2.89; P = .324). Restricted cubic spline analysis showed no statistically significant overall association and the test for nonlinearity was not statistically significant. Findings were similar across platelet-count categories, continuous haematoma-volume changes, ordinal mRS analyses and sensitivity analyses addressing imaging timing, early surgery, acute treatments and prior anticoagulant use.
Discussion
Admission thrombocytopenia showed no statistically significant association with HE, poor functional outcome, or mortality after ICH. Because few patients had platelet counts <100 × 109/L, these findings mainly apply to mild thrombocytopenia.
Keywords: cohort study, clinical outcomes, haematoma expansion, intracerebral haemorrhage, thrombocytopenia
Graphical abstract
Graphical Abstract.
Introduction
Platelets play a central role in primary haemostasis by facilitating thrombus formation and stabilisation at sites of vascular injury.1 A reduced platelet count may impair this process and theoretically increase the risk of ongoing bleeding.2 Thrombocytopenia, commonly defined as a platelet count <150 × 109/L, has been associated with an elevated risk of bleeding events and in severe cases, may lead to intracranial haemorrhage.3–5 In spontaneous intracerebral haemorrhage (ICH), haematoma expansion (HE) is a major determinant of early neurological deterioration and poor clinical outcomes.6,7 However, the relationship between thrombocytopenia and HE remains incompletely understood.
Evidence regarding the prognostic significance of thrombocytopenia in spontaneous ICH remains limited and inconsistent.8,9 The PATCH trial reported worse outcomes following platelet transfusion in patients with antiplatelet-associated ICH10; however, these findings primarily concern platelet dysfunction and transfusion therapy rather than the prognostic significance of reduced platelet count itself.10 More directly relevant observational studies have yielded conflicting findings.8,9 A large multicentre cohort found no significant association between thrombocytopenia and HE or clinical outcomes,8 whereas a more recent cohort identified thrombocytopenia as an independent predictor of HE and mortality.9 These discrepancies underscore the need for further investigation to clarify the prognostic relevance of reduced platelet counts in this population.8,9 Interpretation of prior observational studies may also have been influenced by differences in haemostatic context and treatment exposures, particularly prior anticoagulant use and platelet transfusion.8,9 Our cohort, in which no patient received platelet transfusion and prior anticoagulant use was uncommon, therefore provides a complementary setting in which the associations of platelet count with HE and clinical outcomes can be evaluated with less influence from these factors.
Therefore, the present study aimed to evaluate the associations of thrombocytopenia with HE, functional outcomes and mortality in patients with spontaneous ICH, thereby further characterising the prognostic significance of reduced platelet counts in this population.
Patients and methods
Study design and participants
This study was a retrospective analysis of prospectively collected clinical and imaging data from a single-centre ICH registry. Consecutive patients with spontaneous ICH admitted to the Second Affiliated Hospital of Anhui Medical University between August 2022 and March 2025 were included. The inclusion criteria were: (1) age ≥18 years; (2) primary spontaneous ICH confirmed by baseline noncontrast computed tomography (NCCT); and (3) baseline NCCT performed within 48 hours of symptom onset. The exclusion criteria were: (1) missing admission platelet count data; (2) secondary ICH, including haemorrhage due to vascular malformations or tumours and haemorrhagic transformation of ischaemic stroke; (3) primary intraventricular haemorrhage (IVH); and (4) loss to 3-month follow-up. For the HE analysis, patients who did not undergo follow-up NCCT within 72 hours of baseline NCCT or who underwent surgical intervention before follow-up imaging were additionally excluded. This study was approved by the Ethics Committee of the Second Affiliated Hospital of Anhui Medical University and conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants or their legally authorised representatives. This study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement.11
Exposure
Admission platelet count was obtained from the first routine blood sample collected after hospital arrival as part of the initial emergency laboratory assessment. Blood sampling was generally performed shortly after arrival during the initial evaluation; however, the exact interval from hospital arrival to blood sampling was not recorded as a standardised study variable. Unexpectedly low platelet counts were reassessed when clinically indicated. The primary exposure of interest was thrombocytopenia on admission, defined as a platelet count <150 × 109/L.3,4 Platelet count was additionally evaluated as a continuous exposure and according to predefined categories (≥150, 100–149 and 50–99 × 109/L). A stricter definition of thrombocytopenia, using a platelet count <100 × 109/L, was evaluated in sensitivity analyses. Platelet transfusion was not routinely administered solely on the basis of a platelet count <150 × 109/L at our centre and no patient in the cohort received platelet transfusion.
Clinical and imaging assessment
Clinical variables included age, sex, history of hypertension, diabetes mellitus, previous ICH, previous ischaemic stroke, current smoking, alcohol overuse, prior antiplatelet therapy, prior anticoagulant therapy and liver disease. Laboratory parameters included admission platelet count, haemoglobin level and international normalised ratio (INR). We also recorded the time from symptom onset to baseline NCCT, admission systolic and diastolic blood pressure, Glasgow Coma Scale (GCS) score and National Institutes of Health Stroke Scale (NIHSS) score.
Baseline imaging characteristics included haematoma volume, haematoma location and the presence of IVH. Haematoma volumes were measured using the ABC/2 method12 by a single trained investigator who was blinded to platelet count and clinical outcomes. The ICH score was also calculated.13 Information on whether computed tomography angiography (CTA) was performed during the acute hospitalisation and on CTA spot-sign status was collected when available. Spot-sign prevalence was evaluated among patients who underwent CTA and had an evaluable spot-sign assessment.
Acute management variables included acute blood-pressure lowering therapy, anticoagulant reversal therapy and tranexamic acid treatment.
Outcomes
The primary outcome was HE, defined as an absolute increase in haematoma volume >6 mL or a relative increase >33% from baseline to follow-up NCCT performed within 72 hours of baseline imaging.14 Absolute haematoma-volume change was calculated as follow-up haematoma volume minus baseline haematoma volume, whereas relative haematoma-volume change was calculated as (follow-up haematoma volume − baseline haematoma volume)/baseline haematoma volume × 100%.
Secondary outcomes were poor functional outcome and mortality at 3 months. Functional status was assessed using the modified Rankin Scale (mRS), with an mRS score of 3–6 defined as poor functional outcome.15 The full distribution of mRS scores was also evaluated as an ordinal outcome. Poor functional outcome, mortality and the full mRS distribution at 6 months were assessed as supplementary outcomes among patients with available 6-month follow-up data.
Statistical analysis
Continuous variables were summarised as medians with interquartile ranges (IQRs) and categorical variables as frequencies and percentages. Between-group comparisons were performed using the Mann–Whitney U test for continuous variables and Pearson’s chi-squared test or Fisher’s exact test for categorical variables, as appropriate.
Univariable and multivariable logistic regression models were used for the main analyses of the associations of thrombocytopenia with HE, 3-month poor functional outcome and 3-month mortality. Covariates were selected a priori based on established predictors and clinical relevance. The multivariable model for HE was adjusted for prior antiplatelet therapy, time from symptom onset to baseline NCCT, baseline haematoma volume, the presence of baseline IVH, admission systolic blood pressure, haematoma location and baseline NIHSS score.16,17 Models for 3-month poor functional outcome and mortality were adjusted for age, baseline NIHSS score, prior antiplatelet therapy, the presence of baseline IVH and baseline haematoma volume.13,18 Results are reported as odds ratios (ORs) with 95% CIs.
Additional analyses were performed to characterise the associations of platelet count with haematoma growth and clinical outcomes more comprehensively. Admission platelet count was modelled as a continuous variable using restricted cubic splines with 3 knots located at the 10th, 50th and 90th percentiles, corresponding to platelet counts of 124.0, 192.5 and 276.3 × 109/L, respectively, to assess potential nonlinear associations with HE. A platelet count of 150 × 109/L was used as the reference value and the spline model was adjusted using the same covariates as the main HE model. Tests for the overall association and nonlinearity were performed. Platelet count was also categorised as ≥150, 100–149 and 50–99 × 109/L, with associations with HE evaluated using the same adjustment set; the overall association across platelet-count categories was assessed using a likelihood-ratio test.
The full distribution of mRS scores at 3 months and, as a supplementary analysis, at 6 months was evaluated using ordinal logistic regression. The proportional-odds assumption was assessed for the adjusted models. Because baseline NIHSS score and baseline haematoma volume showed evidence of non-proportional odds, partial proportional-odds models were fitted, allowing these covariates to have non-proportional effects. The adjusted ordinal models included age, baseline NIHSS score, prior antiplatelet therapy, the presence of baseline IVH and baseline haematoma volume. Absolute and relative haematoma-volume changes were also analysed as continuous outcomes. Because both measures were markedly skewed, adjusted between-group differences were estimated using median regression (quantile regression at τ = 0.50) with bootstrap standard errors based on 1000 resamples. The same covariates as in the main HE model were used for adjustment. Baseline characteristics were also compared between patients included in and excluded from the HE analysis.
Several sensitivity analyses were performed to evaluate the consistency of the main findings under alternative analytical specifications. The HE analysis was repeated after restricting baseline NCCT to within 24 hours of symptom onset. Follow-up imaging timing was examined in 2 complementary analyses: 1 was restricted to patients undergoing follow-up NCCT within 24 hours of baseline imaging and the other additionally included the interval from baseline to follow-up NCCT as a continuous covariate in the primary HE model. An exploratory analysis was also performed among patients otherwise eligible for the HE analysis before exclusion for surgery, using a composite endpoint of conventional HE or surgery before follow-up NCCT. The HE model was further adjusted for acute blood-pressure lowering therapy, anticoagulant reversal therapy and tranexamic acid treatment. The main analyses of HE, 3-month poor functional outcome and 3-month mortality were repeated after excluding patients with prior anticoagulant use. The 3-month poor functional outcome analysis was repeated using a more stringent definition of poor functional outcome (mRS score of 4–6) and the main analyses of HE, 3-month poor functional outcome and 3-month mortality were repeated using a stricter thrombocytopenia threshold of <100 × 109/L. Unless otherwise specified, HE sensitivity analyses used the same covariate adjustment set as the main HE model, whereas functional-outcome and mortality analyses used the corresponding clinical-outcome adjustment set.
Clinical outcomes at 6 months were evaluated as supplementary outcomes among patients with available follow-up data. Poor functional outcome and mortality at 6 months were analysed using the same covariate adjustment strategy as the corresponding 3-month outcomes.
Missingness of baseline covariates was minimal; therefore, analyses were performed using complete cases for the variables required in each model, without imputation. Patients without available 6-month outcome data were excluded from the corresponding 6-month analyses and missing outcome data were not imputed.
All statistical analyses were performed using R software (version 4.4.1). All tests were two-sided and a P value <.05 was considered statistically significant.
Results
A total of 496 patients with spontaneous ICH were included, of whom 106 (21.4%) had thrombocytopenia and 390 (78.6%) did not (Figure 1). Among patients with thrombocytopenia, 90 had platelet counts of 100–149 × 109/L and 16 had counts of 50–99 × 109/L; no patient had a platelet count <50 × 109/L. Compared with patients without thrombocytopenia, those with thrombocytopenia were older (67 vs 59 years, P < .001), more frequently had diabetes mellitus (30.2% vs 21.0%, P = .047) and prior anticoagulant therapy (3.8% vs 0.8%, P = .041) and had lower haemoglobin levels (132 vs 141 g/L, P < .001; Table 1). Among the 7 patients with prior anticoagulant use, 5 were receiving rivaroxaban, 1 warfarin and 1 apixaban. Missingness of key baseline covariates was minimal (Table S2).
Figure 1.
Flowchart of participant selection for the overall cohort and haematoma expansion analysis. Note: HE = haematoma expansion; ICH = intracerebral haemorrhage; IVH = intraventricular haemorrhage; NCCT = noncontrast computed tomography.
Table 1.
Clinical characteristics, imaging features and acute management of patients with ICH with and without thrombocytopenia.
| Total (n = 496) | With thrombocytopenia (n = 106) | Without thrombocytopenia (n = 390) | P value | |
|---|---|---|---|---|
| Demographics and medical history | ||||
| Age, y | 60 (52–71) | 67 (57–74) | 59 (50–70) | <.001 |
| Sex, male | 341 (68.8) | 73 (68.9) | 268 (68.7) | .976 |
| Hypertension | 412 (83.1) | 89 (84.0) | 323 (82.8) | .781 |
| Diabetes | 114 (23.0) | 32 (30.2) | 82 (21.0) | .047 |
| Previous ischaemic stroke | 95 (19.2) | 23 (21.7) | 72 (18.5) | .453 |
| Previous ICH | 41 (8.3) | 9 (8.5) | 32 (8.2) | .925 |
| Current smoking | 136 (27.4) | 22 (20.8) | 114 (29.2) | .083 |
| Alcohol overuse | 174 (35.1) | 35 (33.0) | 139 (35.6) | .616 |
| Previous antiplatelet therapy | 62 (12.5) | 15 (14.2) | 47 (12.1) | .562 |
| Previous anticoagulant therapy | 7 (1.4) | 4 (3.8) | 3 (0.8) | .041 |
| Warfarin | 1 (0.2) | 1 (0.9) | 0 (0.0) | |
| Apixaban | 1 (0.2) | 1 (0.9) | 0 (0.0) | |
| Rivaroxaban | 5 (1.0) | 2 (1.9) | 3 (0.8) | |
| Liver disease | 25 (5.0) | 9 (8.5) | 16 (4.1) | .067 |
| Admission characteristics | ||||
| Time from symptom onset to baseline NCCT, hour | 4 (2–13) | 5 (2–13) | 4 (2–12) | .346 |
| Admission SBP, mm Hg | 172 (152–194) | 170 (150–198) | 174 (154–194) | .518 |
| Admission DBP, mm Hg | 100 (89–114) | 97 (86–112) | 101 (90–115) | .191 |
| GCS score | 14 (9–15) | 14 (8–15) | 14 (9–15) | .213 |
| Baseline NIHSS | 11 (4–20) | 12 (4–23) | 10 (4–19) | .315 |
| Laboratory characteristics | ||||
| Platelet count, ×109/L | 195 (157–236) | 124 (111–142) | 209 (183–248) | <.001 |
| INR | 0.95 (0.90–0.99) | 0.96 (0.90–1.01) | 0.94 (0.90–0.99) | .172 |
| Haemoglobin, g/L | 139 (127–150) | 132 (120–143) | 141 (128–152) | <.001 |
| Imaging characteristics | ||||
| Haematoma location | .104 | |||
| Deep | 433 (87.3) | 99 (93.4) | 334 (85.6) | |
| Lobar | 26 (5.2) | 3 (2.8) | 23 (5.9) | |
| Brainstem | 37 (7.5) | 4 (3.8) | 33 (8.5) | |
| Baseline haematoma volume, mL | 13 (5–35) | 15 (6–45) | 12 (5–32) | .192 |
| Presence of baseline IVH | 203 (40.9) | 50 (47.2) | 153 (39.2) | .140 |
| ICH score | 1 (0–2) | 1 (0–2) | 1 (0–3) | .279 |
| CTA spot sign | 14/298 (4.7) | 5/58 (8.6) | 9/240 (3.8) | .158 |
| Acute management | ||||
| Acute blood-pressure lowering therapy | 458 (92.3) | 99 (93.4) | 359 (92.1) | .644 |
| Anticoagulant reversal therapy | 7 (1.4) | 4 (3.8) | 3 (0.8) | .041 |
| Tranexamic acid treatment | 63 (12.7) | 15 (14.2) | 48 (12.3) | .613 |
Values are presented as n (%) or median (interquartile range), as appropriate.
Computed tomography angiography spot-sign prevalence was calculated among patients who underwent CTA and had an evaluable spot-sign assessment.
Abbreviations: CTA = computed tomography angiography; DBP = diastolic blood pressure; GCS = Glasgow Coma Scale; ICH = intracerebral haemorrhage; INR = international normalised ratio; IVH = intraventricular haemorrhage; NCCT = noncontrast computed tomography; NIHSS = National Institutes of Health Stroke Scale; SBP = systolic blood pressure.
Among the 496 patients, 368 were included in the HE analysis; 60 patients did not undergo follow-up NCCT within 72 hours of baseline imaging and 68 underwent surgery before follow-up imaging. Compared with patients included in the HE analysis, excluded patients had greater neurological and radiological severity, including higher baseline NIHSS scores (23 vs 9, P < .001), larger baseline haematoma volumes (40 vs 10 mL, P < .001) and more frequent baseline IVH (54.7% vs 36.1%, P < .001), whereas the prevalence of thrombocytopenia was similar (21.1% vs 21.5%, P = .929; Table S1).
HE occurred in 20 of 79 patients (25.3%) with thrombocytopenia and 76 of 289 patients (26.3%) without thrombocytopenia (P = .860; Table 2). At 3 months, the distribution of mRS scores differed between patients with and without thrombocytopenia in the unadjusted comparison (P = .027). Poor functional outcome occurred in 62.3% and 54.9% of patients with and without thrombocytopenia, respectively (P = .173), whereas 3-month mortality was higher among patients with thrombocytopenia (23.6% vs 15.1%, P = .040; Table 2).
Table 2.
Haematoma expansion and clinical outcomes in patients with and without thrombocytopenia.
| Variables | Total (n = 496) | With thrombocytopenia (n = 106) | Without thrombocytopenia (n = 390) | P value |
|---|---|---|---|---|
| Haematoma expansion | 96/368 (26.1) | 20/79 (25.3) | 76/289 (26.3) | .860 |
| mRS at 3 months | .027 | |||
| 0 | 56 (11.3) | 6 (5.7) | 50 (12.8) | |
| 1 | 79 (15.9) | 16 (15.1) | 63 (16.2) | |
| 2 | 81 (16.3) | 18 (17.0) | 63 (16.2) | |
| 3 | 66 (13.3) | 13 (12.3) | 53 (13.6) | |
| 4 | 88 (17.7) | 19 (17.9) | 69 (17.7) | |
| 5 | 42 (8.5) | 9 (8.5) | 33 (8.5) | |
| 6 | 84 (16.9) | 25 (23.6) | 59 (15.1) | |
| Poor functional outcome | 280 (56.5) | 66 (62.3) | 214 (54.9) | .173 |
| 3-month mortality | 84 (16.9) | 25 (23.6) | 59 (15.1) | .040 |
Abbreviations: HE = haematoma expansion; ICH = intracerebral haemorrhage; mRS = modified Rankin Scale; NCCT = noncontrast computed tomography.
Poor functional outcome was defined as an mRS score of 3–6 at 3 months.
The HE analysis included 368 patients who underwent follow-up NCCT within 72 hours of baseline NCCT and had not undergone surgical intervention before follow-up imaging.
Values are presented as n (%) or n/N (%), as appropriate.
In logistic regression analyses, no statistically significant association was detected between thrombocytopenia and HE in either the univariable model (OR 0.95, 95% CI, 0.53–1.66; P = 0.860) or the multivariable model (adjusted OR 0.97, 95% CI, 0.52–1.76; P = .920; Table 3). For 3-month clinical outcomes, no statistically significant association was detected between thrombocytopenia and poor functional outcome in the univariable analysis (OR 1.36, 95% CI, 0.88–2.12; P = .174), whereas it was associated with higher mortality (OR 1.73, 95% CI, 1.01–2.91; P = .041). After multivariable adjustment, no statistically significant independent association was detected between thrombocytopenia and either poor functional outcome (adjusted OR 0.88, 95% CI, 0.46–1.66; P = .693) or mortality (adjusted OR 1.43, 95% CI, 0.70–2.89; P = .324; Table 3).
Table 3.
Associations of thrombocytopenia with haematoma expansion and 3-month clinical outcomes in patients with ICH.
| Outcome | Platelet count <150 × 10 9 /L | Platelet count ≥150 × 10 9 /L | Odds ratio (95% CI) | P value |
|---|---|---|---|---|
| Haematoma expansion | ||||
| Unadjusted | 20/79 (25.3) | 76/289 (26.3) | 0.95 (0.53–1.66) | .860 |
| Adjusted | 0.97 (0.52–1.76) | .920 | ||
| Poor functional outcome | ||||
| Unadjusted | 66/106 (62.3) | 214/390 (54.9) | 1.36 (0.88–2.12) | .174 |
| Adjusted | 0.88 (0.46–1.66) | .693 | ||
| 3-month mortality | ||||
| Unadjusted | 25/106 (23.6) | 59/390 (15.1) | 1.73 (1.01–2.91) | .041 |
| Adjusted | 1.43 (0.70–2.89) | .324 |
The haematoma expansion model was adjusted for prior antiplatelet therapy, time from symptom onset to baseline NCCT, baseline haematoma volume, the presence of baseline IVH, admission systolic blood pressure, haematoma location and baseline NIHSS score.
Models for 3-month poor functional outcome and mortality were adjusted for age, baseline NIHSS score, prior antiplatelet therapy, the presence of baseline IVH and baseline haematoma volume. Poor functional outcome was defined as an mRS score of 3–6 at 3 months.
Abbreviations: CI = confidence interval; HE = haematoma expansion; ICH = intracerebral haemorrhage; IVH = intraventricular haemorrhage; mRS = modified Rankin Scale; NCCT = noncontrast computed tomography; NIHSS = National Institutes of Health Stroke Scale; OR = odds ratio.
Additional analyses further characterised the relationship between platelet count and haematoma growth. Restricted cubic spline analysis showed no statistically significant overall association between admission platelet count and HE (P for overall association = .738) and the test for nonlinearity was not statistically significant (P for nonlinearity = .965; Figure 3). When platelet count was categorised as ≥150, 100–149 and 50–99 × 109/L, no statistically significant association with HE was detected for either the 100–149 × 109/L group (adjusted OR 0.97, 95% CI, 0.50–1.83; P = .938) or the 50–99 × 109/L group (adjusted OR 0.94, 95% CI, 0.19–3.71; P = .930) compared with the ≥150 × 109/L group. The overall association across platelet-count categories was not statistically significant (P = .994; Table S3). When absolute and relative haematoma-volume changes were analysed as continuous outcomes, the adjusted median difference was 0.22 mL (95% CI, −0.37 to 0.82; P = .466) for absolute haematoma-volume change and 1.64 percentage points (95% CI, −6.07 to 9.35; P = .677) for relative haematoma-volume change (Table S4).
Figure 3.
Association between admission platelet count and haematoma expansion using restricted cubic spline analysis. The solid line represents the adjusted odds ratio and the shaded area represents the 95% CI, with a platelet count of 150 × 109/L used as the reference. The model was adjusted for prior antiplatelet therapy, time from symptom onset to baseline NCCT, baseline haematoma volume, the presence of baseline IVH, admission systolic blood pressure, haematoma location and baseline NIHSS score. P for overall association = .738; P for nonlinearity = .965. Note: CI = confidence interval; IVH = intraventricular haemorrhage; NCCT = noncontrast computed tomography; NIHSS = National Institutes of Health Stroke Scale.
In ordinal regression analysis of the 3-month mRS distribution, thrombocytopenia was associated with a shift towards higher mRS scores in the unadjusted model (common OR 1.55, 95% CI, 1.06–2.26; P = .024). This association was attenuated after multivariable adjustment (adjusted common OR 1.14, 95% CI, 0.75–1.72; P = .536; Figure 2 and Table S5).
Figure 2.
Distribution of 3-month modified Rankin scale scores in patients with and without thrombocytopenia. The dashed line indicates the boundary between mRS scores 0–2 and 3–6. The adjusted common OR from ordinal regression was 1.14 (95% CI 0.75–1.72; P = 0.536), adjusted for age, baseline NIHSS score, prior antiplatelet therapy, baseline IVH and baseline haematoma volume. Note: CI = confidence interval; IVH = intraventricular haemorrhage; mRS = modified Rankin scale; NIHSS = National Institutes of Health stroke scale; OR = odds ratio.
Sensitivity analyses accounting for imaging timing yielded similar estimates. The median time from symptom onset to baseline NCCT was 4 hours (IQR, 2–13) overall and was similar between patients with and without thrombocytopenia (5 [2–13] vs 4 [2–12] hours, P = .346). The median interval from baseline to follow-up NCCT was 19.0 hours (IQR, 9.5–27.3) and likewise did not differ significantly between groups (21 [13–28] vs 19 [7–27] hours, P = .231). When the HE analysis was restricted to patients undergoing baseline NCCT within 24 hours of symptom onset, the adjusted OR was 1.03 (95% CI, 0.55–1.92; P = .932; Table S6). When the analysis was restricted to patients undergoing follow-up NCCT within 24 hours of baseline imaging, the adjusted OR was 0.87 (95% CI, 0.41–1.77; P = .703). Additional adjustment for the interval from baseline to follow-up NCCT as a continuous covariate yielded an adjusted OR of 1.01 (95% CI, 0.54–1.84; P = .979; Table S7).
In the exploratory analysis using a composite endpoint of conventional HE or surgery before follow-up imaging, no statistically significant association was detected between thrombocytopenia and the composite endpoint (adjusted OR 0.85, 95% CI, 0.48–1.49; P = .566; Table S8). Further adjustment for acute blood-pressure lowering therapy, anticoagulant reversal therapy and tranexamic acid treatment yielded an adjusted OR of 0.94 (95% CI, 0.50–1.71; P = .832; Table S9). After excluding patients with prior anticoagulant use, no statistically significant association was detected between thrombocytopenia and HE (adjusted OR 1.00, 95% CI, 0.53–1.84; P = .993), 3-month poor functional outcome (adjusted OR 0.85, 95% CI, 0.44–1.61; P = .621), or 3-month mortality (adjusted OR 1.37, 95% CI, 0.65–2.84; P = .394; Table S10).
When poor functional outcome at 3 months was defined more stringently as an mRS score of 4–6, no statistically significant association was detected between thrombocytopenia and poor functional outcome in either the univariable analysis (OR 1.42, 95% CI, 0.92–2.19; P = .109) or the multivariable analysis (adjusted OR 0.94, 95% CI, 0.48–1.79; P = .849; Table S11).
When thrombocytopenia was defined using the stricter threshold of <100 × 109/L, HE occurred in 3 of 11 patients (27.3%) with platelet counts <100 × 109/L and 93 of 357 patients (26.1%) with platelet counts ≥100 × 109/L. No statistically significant association with HE was detected in either the univariable analysis (OR 1.06, 95% CI, 0.28–4.10; P = .928) or the multivariable analysis (adjusted OR 0.94, 95% CI, 0.22–4.01; P = .935; Table S14). For clinical outcomes, patients with platelet counts <100 × 109/L had higher observed rates of 3-month poor functional outcome (81.3% vs 55.6%) and mortality (37.5% vs 16.3%). The corresponding univariable ORs were 3.46 (95% CI, 0.97–12.28; P = .055) for poor functional outcome and 3.09 (95% CI, 1.09–8.76; P = .034) for mortality. After multivariable adjustment, the estimates remained elevated but were imprecise and did not reach statistical significance (poor functional outcome: adjusted OR 4.23, 95% CI, 0.80–22.47; P = .091; mortality: adjusted OR 2.94, 95% CI, 0.72–12.04; P = .134; Table S15).
Among the 463 patients with available 6-month follow-up data, the distribution of mRS scores differed between patients with and without thrombocytopenia in the unadjusted comparison (P = .010; Table S12). Poor functional outcome at 6 months was more frequent among patients with thrombocytopenia (65.0% vs 51.8%, P = .019), whereas 6-month mortality did not differ significantly between groups (26.0% vs 18.2%, P = .083; Table S12). In ordinal regression analysis, thrombocytopenia was associated with a shift towards higher 6-month mRS scores in the unadjusted model (common OR 1.68, 95% CI, 1.14–2.47; P = .009), whereas the association was attenuated after multivariable adjustment (adjusted common OR 1.15, 95% CI, 0.75–1.77; P = .518; Table S5 and Figure S1). Similarly, thrombocytopenia was associated with 6-month poor functional outcome in the univariable analysis (OR 1.73, 95% CI, 1.10–2.76; P = .020), whereas no statistically significant association was detected after multivariable adjustment (adjusted OR 1.23, 95% CI, 0.65–2.34; P = .523). No statistically significant association was detected between thrombocytopenia and 6-month mortality in either the univariable analysis (OR 1.58, 95% CI, 0.93–2.64; P = .084) or the multivariable analysis (adjusted OR 1.11, 95% CI, 0.54–2.21; P = .780; Table S13).
Discussion
In this retrospective analysis of prospectively collected clinical and imaging data, no statistically significant association was detected between admission thrombocytopenia and HE, poor functional outcome, or mortality after multivariable adjustment. For HE, the adjusted effect estimate was close to unity (adjusted OR 0.97, 95% CI, 0.52–1.76), although the CI indicated limited precision. The adjusted estimates for 3-month poor functional outcome (OR 0.88, 95% CI, 0.46–1.66) and mortality (OR 1.43, 95% CI, 0.70–2.89) were also imprecise, with the mortality estimate remaining above unity. The HE findings were broadly similar when platelet count was examined continuously, across different platelet-count categories and using absolute and relative haematoma-volume changes. Likewise, differences in functional outcome observed in unadjusted analyses were attenuated after adjustment for age, neurological severity and haematoma burden, including when the full distribution of mRS scores was analysed. Taken together, these complementary analyses provide a more comprehensive characterisation of the associations of predominantly mild thrombocytopenia with haematoma growth and clinical outcomes in our cohort.
Previous studies have examined platelet-related factors in ICH from different perspectives.10 The PATCH trial evaluated whether platelet transfusion could improve outcomes in patients with spontaneous ICH who were receiving antiplatelet therapy before haemorrhage onset.10 Platelet transfusion was associated with a higher risk of death or dependence and more ICH-related serious adverse events, indicating that platelet transfusion in the setting of antiplatelet-associated ICH does not necessarily improve clinical outcomes.10 However, PATCH addressed a treatment strategy in antiplatelet-associated ICH rather than the prognostic significance of reduced platelet count.10 Importantly, patients with platelet counts <100 × 109/L were excluded from the trial.10 Therefore, although PATCH highlighted the complexity of platelet-related haemostasis in ICH, it did not establish whether thrombocytopenia itself is associated with HE or clinical outcomes, leaving the prognostic significance of reduced platelet count uncertain.8,9
Evidence from observational studies examining thrombocytopenia itself has remained inconsistent.8,9 Mrochen et al. analysed 2183 patients with ICH in a large multicentre study and found that thrombocytopenia was not associated with increased HE or worse functional outcome, irrespective of prior antiplatelet therapy.8 These findings are broadly consistent with those of the present study, although the clinical populations differed substantially. More than half of the patients in that study had vitamin K antagonist-associated ICH, whereas prior anticoagulant exposure was uncommon in our cohort.8 Anticoagulant-associated haemorrhage represents a distinct haemostatic context because both the underlying anticoagulant effect and subsequent reversal therapy may influence haematoma growth.5,16 Our findings therefore provide complementary evidence from a predominantly non-anticoagulated population with spontaneous ICH.
More recently, Wilson et al. reported different findings in 1002 patients with spontaneous ICH.9 Thrombocytopenia was associated with a higher incidence of HE and was independently associated with hospital mortality. In multivariable analysis, thrombocytopenia remained associated with HE when platelet transfusion was not included in the model, with an adjusted OR of 1.77 and a 95% CI of 1.20–2.59; when platelet transfusion was included, the estimate was attenuated to 1.28 with a 95% CI of 0.82–2.00.9 The authors therefore emphasised the difficulty of separating the contributions of thrombocytopenia and platelet transfusion to HE. Notably, the crude HE rates differed substantially between their study and ours. HE occurred in 45.6% of patients with thrombocytopenia and 31.5% of those without thrombocytopenia in the study by Wilson et al., compared with 25.3% and 26.3%, respectively, in our cohort.9 Patients with thrombocytopenia in their study also had greater baseline ICH severity, including lower GCS scores, larger haematoma volumes, more frequent IVH and higher ICH scores, whereas these differences were less pronounced in our cohort.9 More pronounced thrombocytopenia was also more frequent, with platelet counts <100 × 109/L in 32.4% of thrombocytopenic patients included in their HE analysis compared with 13.9% in ours.9 In addition, 33 thrombocytopenic patients received platelet transfusion, whereas no patient in our cohort did.9 Differences in baseline haemorrhage severity, the distribution of platelet counts and platelet transfusion practices may therefore partly account for the contrasting findings.
Against these discordant findings, the present study provides complementary evidence from an independent contemporary cohort with prospectively collected clinical and imaging data. Patients were enrolled over a relatively short period from 2022 to 2025, with broadly consistent acute management.19,20 Importantly, no patient received platelet transfusion and prior anticoagulant use was rare (1.4%), providing a distinct haemostatic context with less potential confounding from these factors. This is particularly relevant given that the association reported by Wilson et al.9 was attenuated when platelet transfusion was included in the multivariable model, whereas anticoagulant-associated ICH was common in the cohort reported by Mrochen et al.8 Our cohort therefore provides a complementary setting in which the association between platelet count and haematoma growth can be examined with less influence from platelet transfusion and anticoagulant exposure. In addition, we characterised this association beyond the conventional dichotomous definition of thrombocytopenia. Restricted cubic spline analysis was used to examine the continuous relationship between platelet count and HE and to assess potential nonlinearity, although the limited number of patients at lower platelet counts reduced our ability to evaluate potential threshold effects in this range. Absolute and relative changes in haematoma volume provided complementary measures of haematoma growth without relying solely on the conventional binary HE definition. Additional analyses addressed the potential influence of imaging timing, early surgery, acute treatment and prior anticoagulant use. Taken together, these analyses provide a more detailed characterisation of the platelet count–haematoma growth relationship and extend the existing evidence to a contemporary ICH population characterised predominantly by mild thrombocytopenia.
Beyond haematoma growth, the clinical outcome findings also warrant consideration. Patients with thrombocytopenia had higher unadjusted mortality and less favourable functional outcomes, whereas these associations were attenuated after adjustment for age, neurological severity, haematoma burden and other relevant factors. A similar pattern was observed when functional outcome was assessed across the full ordinal distribution of mRS scores. Patients with thrombocytopenia were older and more frequently had diabetes, which may have contributed to their higher baseline risk and, in part, to the differences observed in unadjusted analyses. Overall, no statistically significant independent association was detected between predominantly mild thrombocytopenia and adverse clinical outcomes at 3 or 6 months. Analyses using the stricter platelet-count threshold of <100 × 109/L require particular caution. For HE, the adjusted point estimate remained close to unity (OR 0.94, 95% CI, 0.22–4.01), but the CI was very wide because only 11 patients with platelet counts <100 × 109/L were included in the HE analysis. In contrast, the adjusted point estimates were elevated for poor functional outcome (OR 4.23, 95% CI, 0.80–22.47) and mortality (OR 2.94, 95% CI, 0.72–12.04). Although highly imprecise, the elevated point estimates for poor functional outcome and mortality leave open the possibility of clinically important associations at lower platelet counts. Given the small number of patients with platelet counts <100 × 109/L, our study had limited statistical power to reliably confirm or exclude such associations in patients with more pronounced thrombocytopenia.
Several strengths of this study should be noted. Clinical and imaging data were prospectively collected in a contemporary ICH cohort and no patient received platelet transfusion. Platelet count was evaluated using both categorical and continuous approaches, while haematoma growth was assessed using conventional HE criteria and continuous volume changes. Functional outcome was similarly evaluated using both dichotomous and ordinal approaches and additional analyses addressed potential influences of imaging timing, early surgery, acute treatment and prior anticoagulant use.
Several limitations should also be acknowledged. First, this was a retrospective analysis of a single-centre prospective ICH registry conducted in a Chinese population, with a predominance of deep ICH and relatively few lobar haemorrhages. Given known differences in ICH subtype distribution across populations, the generalizability of our findings to other ethnic and geographic populations may be limited.21 Second, the HE analysis excluded patients without follow-up NCCT within 72 hours of baseline NCCT and those who underwent surgery before follow-up imaging. These excluded patients had greater neurological and radiological severity, raising the possibility of selection bias. In addition, the relatively broad imaging windows may have influenced HE ascertainment. Although sensitivity analyses addressing imaging timing and early surgery yielded similar estimates, residual selection and timing-related bias cannot be entirely excluded. Third, platelet exposure was based on the admission platelet count alone. Serial platelet counts were not systematically available, preventing assessment of whether changes in platelet count during the subsequent 24–72 hours were associated with haematoma growth or clinical outcomes. Pseudothrombocytopenia was not systematically excluded. Fourth, platelet function was not directly assessed. Platelet count reflects platelet quantity but does not capture platelet activation, aggregation, or other aspects of haemostatic function. Previous studies have suggested that impaired platelet activity may be associated with early haematoma growth and adverse outcomes after ICH.3,22,23 Thus, our findings regarding platelet count should not be extrapolated to qualitative platelet dysfunction. Fifth, although a history of liver disease was recorded, other potential causes of thrombocytopenia, including hematologic disorders, malignancy and acute infection or sepsis, were not systematically characterised or adjudicated. Therefore, the extent to which thrombocytopenia reflected an underlying systemic disorder rather than an isolated reduction in platelet count could not be determined and residual confounding remains possible. Sixth, relatively few patients had platelet counts <100 × 109/L and none had counts <50 × 109/L, resulting in imprecise estimates and limiting conclusions regarding more pronounced thrombocytopenia. Finally, haematoma volume was estimated using the ABC/2 method rather than semiautomated volumetric analysis and 6-month analyses were restricted to patients with available follow-up data, introducing the possibility of attrition bias.
Conclusion
In this cohort of patients with spontaneous ICH, in which thrombocytopenia was predominantly mild, no statistically significant independent association was detected between admission thrombocytopenia and HE, poor functional outcome, or mortality. However, the limited number of patients with platelet counts <100 × 109/L precludes firm conclusions regarding more pronounced thrombocytopenia.
Supplementary Material
Acknowledgements
We thank all participants and healthcare professionals who contributed to this study.
Contributor Information
Zijie Wang, Department of Neurology, The Second Affiliated Hospital of Anhui Medical University, Hefei, China.
Xiao Hu, Department of Neurology, The Second Affiliated Hospital of Anhui Medical University, Hefei, China.
Min Liu, Department of Neurology, The Second Affiliated Hospital of Anhui Medical University, Hefei, China.
Jiaxin Luo, Department of Neurology, The Second Affiliated Hospital of Anhui Medical University, Hefei, China.
Zizhen Meng, Department of Neurology, The Second Affiliated Hospital of Anhui Medical University, Hefei, China.
Qi Li, Department of Neurology, The Second Affiliated Hospital of Anhui Medical University, Hefei, China.
Author contributions
Han Wang (Formal analysis [lead], Methodology [equal], Writing—original draft [lead]), Rui Cheng (Methodology [equal], Validation [equal], Writing—original draft [equal]), Sijia Liu (Methodology [equal], Validation [equal], Writing—original draft [equal]), Zijie Wang (Data curation [equal], Investigation [equal], Resources [equal]), Xiao Hu (Data curation [equal], Investigation [equal], Resources [equal]), Min Liu (Data curation [equal], Investigation [equal], Resources [equal]), Jiaxin Luo (Data curation [equal], Investigation [equal], Resources [equal]), Zizhen Meng (Data curation [equal], Investigation [equal], Resources [equal]), and Qi Li (Conceptualisation [lead], Funding acquisition [lead], Project administration [lead], Supervision [lead], Writing—review & editing [lead])
Conflicts of interest
The authors declare that there is no conflict of interest.
Funding
This study was supported by the National Natural Science Foundation of China (No. 82471368), Noncommunicable Chronic Disease-National Science and Technology Major Project (No. 2025ZD0552602), Clinical and Translational Research Project of Anhui Province (No. 202427b10020090, No. 202427b10020053) and Excellent Research and Innovation Team Project of Anhui Province (No. 2024AH010014).
Data availability
The de-identified data supporting the findings of this study are available from the corresponding author upon reasonable request.
Ethical approval
This study was conducted in accordance with the World Medical Association Declaration of Helsinki and its subsequent amendments. The study protocol was approved by the Institutional Review Boards of the Second Affiliated Hospital of Anhui Medical University (Nos. YX2025-174 and YX2024-067).
Informed consent
Written informed consent was obtained from all participants or their legally authorised representatives.
Guarantor
Qi Li is the guarantor of this study and accepts responsibility for the integrity of the work.
References
- 1. Golebiewska EM, Poole AW. Platelet secretion: from haemostasis to wound healing and beyond. Blood Rev. 2015;29:153–162. 10.1016/j.blre.2014.10.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Raghunathan S, Rayes J, Sen GA. Platelet-inspired nanomedicine in hemostasis thrombosis and thromboinflammation. J Thromb Haemost. 2022;20:1535–1549. 10.1111/jth.15734 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Ziai WC, Torbey MT, Kickler TS, Oh S, Bhardwaj A, Wityk RJ. Platelet count and function in spontaneous intracerebral hemorrhage. J Stroke Cerebrovasc Dis. 2003;12:201–206. 10.1016/S1052-3057(03)00075-2 [DOI] [PubMed] [Google Scholar]
- 4. Gauer RL, Braun MM. Thrombocytopenia. Am Fam Physician. 2012;85:612–622. [PubMed] [Google Scholar]
- 5. Greenberg SM, Ziai WC, Cordonnier C, et al. 2022 guideline for the Management of Patients with spontaneous intracerebral hemorrhage: a guideline from the American Heart Association/American Stroke Association. Stroke. 2022;53:e282–e361. 10.1161/STR.0000000000000407 [DOI] [PubMed] [Google Scholar]
- 6. Morotti A, Boulouis G, Dowlatshahi D, et al. Intracerebral haemorrhage expansion: definitions, predictors, and prevention. Lancet Neurol. 2023;22:159–171. 10.1016/S1474-4422(22)00338-6 [DOI] [PubMed] [Google Scholar]
- 7. Morotti A, Boulouis G, Nawabi J, et al. Association between hematoma expansion severity and outcome and its interaction with baseline intracerebral hemorrhage volume. Neurology. 2023;101:e1606–e1613. 10.1212/WNL.0000000000207728 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Mrochen A, Sprügel MI, Gerner ST, et al. Thrombocytopenia and clinical outcomes in intracerebral hemorrhage: a retrospective multicenter cohort study. Stroke. 2021;52:611–619. 10.1161/STROKEAHA.120.031478 [DOI] [PubMed] [Google Scholar]
- 9. Wilson M, Incontri D, Vu S, et al. Thrombocytopenia as a predictor of hematoma expansion and mortality in intracerebral hemorrhage. Int J Stroke. 2026;21:718–728. 10.1177/17474930251392707 [DOI] [PubMed] [Google Scholar]
- 10. Baharoglu MI, Cordonnier C, Al-Shahi Salman R, et al. Platelet transfusion versus standard care after acute stroke due to spontaneous cerebral haemorrhage associated with antiplatelet therapy (PATCH): a randomised, open-label, phase 3 trial. Lancet. 2016;387:2605–2613. 10.1016/S0140-6736(16)30392-0 [DOI] [PubMed] [Google Scholar]
- 11. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet. 2007;370:1453–1457. 10.1016/S0140-6736(07)61602-X [DOI] [PubMed] [Google Scholar]
- 12. Newman GC. Clarification of abc/2 rule for ICH volume. Stroke. 2007;38:862. 10.1161/01.STR.0000257309.50643.0a [DOI] [PubMed] [Google Scholar]
- 13. Hemphill JC, Bonovich DC, Besmertis L, Manley GT, Johnston SC. The ICH score: a simple, reliable grading scale for intracerebral hemorrhage. Stroke. 2001;32:891–897. 10.1161/01.str.32.4.891 [DOI] [PubMed] [Google Scholar]
- 14. Demchuk AM, Dowlatshahi D, Rodriguez-Luna D, et al. Prediction of haematoma growth and outcome in patients with intracerebral haemorrhage using the CT-angiography spot sign (PREDICT): a prospective observational study. Lancet Neurol. 2012;11:307–314. 10.1016/S1474-4422(12)70038-8 [DOI] [PubMed] [Google Scholar]
- 15. Anderson CS, Heeley E, Huang Y, et al. Rapid blood-pressure lowering in patients with acute intracerebral hemorrhage. N Engl J Med. 2013;368:2355–2365. 10.1056/NEJMoa1214609 [DOI] [PubMed] [Google Scholar]
- 16. Al-Shahi Salman R, Frantzias J, Lee RJ, et al. Absolute risk and predictors of the growth of acute spontaneous intracerebral haemorrhage: a systematic review and meta-analysis of individual patient data. Lancet Neurol. 2018;17:885–894. 10.1016/S1474-4422(18)30253-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Brouwers HB, Chang Y, Falcone GJ, et al. Predicting hematoma expansion after primary intracerebral hemorrhage. JAMA Neurol. 2014;71:158–164. 10.1001/jamaneurol.2013.5433 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Yu HH, Pan C, Tang YX, et al. Effects of prior antiplatelet therapy on the prognosis of primary intracerebral hemorrhage: a meta-analysis. Chin Med J (Engl). 2017;130:2969–2977. 10.4103/0366-6999.220302 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Li Q, Yakhkind A, Alexandrov AW, et al. Code ICH: a call to action. Stroke. 2024;55:494–505. 10.1161/STROKEAHA.123.043033 [DOI] [PubMed] [Google Scholar]
- 20. Li Q, Morotti A, Dowlatshahi D, et al. Integrated ICH care pathway across stroke service levels: time for a new model of care. Stroke. 2026;57:219–229. 10.1161/STROKEAHA.125.052829 [DOI] [PubMed] [Google Scholar]
- 21. Juan YC, Kuo JS, Chen HL, et al. Genetic risk factors for intracerebral hemorrhage in populations of east Asian ancestry. Neurol Genet. 2026;12:e200390. 10.1212/NXG.0000000000200390 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Naidech AM, Bernstein RA, Levasseur K, et al. Platelet activity and outcome after intracerebral hemorrhage. Ann Neurol. 2009;65:352–356. 10.1002/ana.21618 [DOI] [PubMed] [Google Scholar]
- 23. Naidech AM, Jovanovic B, Liebling S, et al. Reduced platelet activity is associated with early clot growth and worse 3-month outcome after intracerebral hemorrhage. Stroke. 2009;40:2398–2401. 10.1161/STROKEAHA.109.550939 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The de-identified data supporting the findings of this study are available from the corresponding author upon reasonable request.