4th November 2011, Volume 124 Number 1345

James Irwin, Peter Wright, Paul Reeve

Intracerebral haemorrhage (ICH) is a significant cause of morbidity and mortality. Published community based incidence figures vary between 20–31 per 100,000 per year.1,2 These patients make up 10–15% of all acute stroke events in Western populations.3 Presenting symptoms include headache, focal neurological deficit, seizures and a decreased level of consciousness. Previous analyses have demonstrated that between 30–50% die within 30 days of their bleed.1,2,4

Survivors are often left dependent on others for care, and represent a significant drain on hospital and community resources.

Epidemiological studies have yielded uncertain conclusions regarding temporal trends in incidence of ICH.1 Four population based stroke incidence studies in the past 30 years have analysed temporal trends in ICH incidence.5–8 Crude incidence decreased in Perth (Australia), while remaining stable in Dijon (France), Oxford (England) and in Finland. Anecdotally we felt incidence had been increasing in our hospital, and this hospital based retrospective analysis was performed to determine if this was true. The data was also analysed to identify parameters which correlated with any identified change in incidence, and parameters which correlated with 30-day survival.

The Waikato region of New Zealand has a population of approximately 375,000, and is serviced by Waikato Hospital in Hamilton and four smaller hospitals in peripheral centres. CT scanning is available in Waikato Hospital, and in one of the peripheral hospitals (Thames Hospital). All data regarding admission to these two hospitals and all radiology are recorded on the Waikato Hospital database.


An electronic search was performed on all hospital admissions entered on the IPM database of the Waikato District Health Board, for the 10 years between 1 December 1998 and 31 November 2008. Adults (age 15 or older) with an ICD-10 coded discharge diagnosis9 of I61.0–I61.9 or I62.9 (all forms of intracerebral haemorrhage, isolated intraventricular haemorrhage and intracranial haemorrhage not specified) were captured. A search was also performed on the emergency department database for all patients who died in the emergency department during the study period.
All electronic CT and MRI reports of brain scans were reviewed for all patients. Patients with a neuroimaging report which documented a primary subarachnoid haemorrhage, subdural haemorrhage, extradural haemorrhage, traumatic intracranial haemorrhage, intracranial bleed associated with a neurosurgical procedure, an underlying malignancy or an arteriovenous malformation were excluded. Only patients with a spontaneous intracerebral haemorrhage or an isolated intraventricular haemorrhage were included in the dataset. Included patients had their age, sex, ethnicity, date of admission, date of discharge and discharge status (alive or dead) retrieved from the database.
The location of the bleed was recorded as being either lobar (in a cerebral hemisphere), deep (basal ganglia or brainstem), cerebellar or isolated intraventricular. If blood was identified in more than one location, the location where the reporting radiologist felt the primary bleed had occurred was recorded as the location of the bleed. Extension of bleed into the ventricular system was noted.
Evidence of warfarin use was sought in the electronic clinical record. The INR on admission, if performed, was recorded. Patients with an INR ≥1.5 had their paper clinical record retrieved and warfarin use was documented.
The data was analysed using R.10 Comparison of ICH rates between different time periods was performed using the rateratio.test package in R, assuming a Poisson distribution of ICH events. The trend in incidence was analysed comparing rates in 1999–2001 with 2006–2008. For each period weighted estimates of Waikato regional population counts were made from 1996, 2001 and 2006 census data.11 Association between predictor variables and 30-day mortality was analysed using logistic regression using the glm() function (stats package, version 2.11.0) in R.
Possible variables for the regression model were selected using screening univariate analysis—for categorical variables using a chi squared test, and for continuous variables using univariate logistic regression analysis. Variables significantly correlated with 30-day mortality (p value<0.10) were added sequentially into a logistic regression model, and those with significant independent correlation with 30-day mortality were retained.
The data was analysed to confirm the absence of significant outlier values, and interactions between variables were searched for and included in the final model if significant. A covariance matrix was generated to identify significant colinearity between predictor variables. Odds ratios were calculated as the exponential of the β value of each predictor variable, and as eβ(main term 1) × eβ(main term 2) × eβ(interaction term) for interaction terms. All p values calculated were 2-tailed, and a value of <0.05 was considered to be statistically significant.
Statistical information regarding age, ethnicity and sex stratified population counts in the Waikato region were obtained from Statistics New Zealand, using 1996, 2001 and 2006 Census data.11


653 episodes of ICH were recorded during the study period. 20 were recurrent episodes in patients who had already suffered an ICH episode during the study period, and 12 episodes were multifocal. The mean incidence of ICH was 17.4 per 100,000 persons per year. There was no significant difference in incidence between sexes.

The majority of ICH occurred in the NZ European ethnicity group (Table 1). Rates in Māori, Pacific Island and Asian populations were lower than population prevalence would predict.

Table 1. ICH incidence stratified by ethnicity
Frequency of ICH
Waikato population by ethnicity*
Ethnicity-specific incidence (per 100,000 per year)
NZ European
Pacific Island
Not recorded


*Taken as a weighted average for 1998–2008 from 1996, 2001 and 2006 census figures.

Incidence of ICH increased during the study period (Figure 1), from 14.4 per 100,000 per year (95% CI 13.7–15.1) between 1999–2001, to 21.4 (20.6–22.2) between 2006–2008 (rate ratio 1.49, p<0.0001).

107 episodes of ICH occurred in patients taking warfarin, increasing from an average of 2.20 per 100,000 per year in 1999–2001, to 4.05 in 2006–2008 (RR 1.84, p=0.018, Figure 1).

Incidence increased with age (Table 2), and the majority of ICH occurred in a lobar location (Table 3). The mean age at presentation was 71.7 years for those of NZ European ethnicity, 62.4 for Māori, 64.0 for Pacific Island and 59.9 for Asian ethnicity.

Table 2. Age-specific incidence of ICH
Frequency of ICH
Waikato age-specific population*
Age-adjusted incidence (per 100,000 per year)
*Taken as a weighted average for 1998–2008 from 1996, 2001 and 2006 Census figures.
Figure 1. Yearly frequency of ICH (blue=total, green=warfarin-associated)
Table 3. Location of ICH
Location of ICH
Percentage of all ICH episodes
Deep (brainstem/basal ganglia/deep white matter)
Isolated intraventricular
Not known (radiology not available)

By univariate analysis, intraventricular extension of haemorrhage, warfarin use, age, lobar location of bleed and Māori ethnicity were significantly associated with 30-day mortality (Table 4).

The final logistic regressional model included intraventricular extension of haemorrhage, warfarin use, lobar location of bleed and age as independent predictor variables (Table 5). A significant interaction existed between warfarin use and intraventricular extension of bleed. For those not taking warfarin intraventricular extension of bleed was strongly predictive of 30-day mortality.

For those taking warfarin this effect was multiplied threefold, meaning warfarin users who had intraventricular extension of bleed on neuroimaging had 24 times the likelihood of death within 30 days compared with those who did not have intraventricular extension of bleed and were not taking warfarin.

Table 4. Univariate analysis of predictors of 30-day mortality
Odds Ratio
New Zealand European


Age (per 10-year increment)
Male sex
Warfarin use
Location of bleed
Basal ganglia/brainstem


Isolated intraventricular
Ventricular extension of bleed
Table 5. Factors predictive of death within 30 days – final logistic regressional model
Predictor variable
β value
Odds Ratio*
Age (per 10 year increment)
Lobar location of bleed
Warfarin use, no ventricular extension of bleed
Ventricular extension of bleed, no warfarin use
Warfarin use and ventricular extension of bleed
*Odds Ratio is equivalent to eβ.

For those taking warfarin, the median INR at the time of bleed was 2.7; 62% of patients had an INR≤3.


This analysis demonstrates a 49% temporal increase in incidence of hospital observed ICH between 1999 and 2008 in the Waikato Region. It also shows ventricular extension of bleed, warfarin use, lobar location of bleed and age to be associated with increased 30-day mortality. A lower incidence of ICH in Pacific Island, Māori and Asian populations was noted.

Community based stroke incidence studies suggest ICH incidence over the past 20 years has either remained static or decreased.1 Our data shows a significant increase in observed ICH in our region’s hospitals. We have collected a set of ICH events detected in hospital, and therefore cannot extrapolate our observed trends in incidence to the general population. However, in hospital we are clearly identifying and treating increasing numbers of patients with ICH.

We consider this increase may be the result of increasing use of neuroimaging to detect ICH, of increasing presentation of patients with ICH to hospital, or of truly increasing ICH incidence. The potential contribution of these three factors is considered below.

A proportion of patients suffering ICH in our region during the study period will have died before having had neuroimaging performed. Without a postmortem these patients would not have been diagnosed with ICH and would not have been included in this dataset. The proportion of ICH patients who did not receive neuroimaging may have been higher 10 years ago, when CT scanning was a scarcer resource. There has been a threefold increase in the number of brain CT scans performed on emergency department patients in Waikato and Thames hospitals over the study period (Personal Communication, Radiology Department, Waikato Hospital).

In a comparably resourced area, the Auckland Regional Community Stroke Study Group (ARCOS) recorded that the proportion of all stroke sufferers who underwent neuroimaging increased from 42% in 1991-92 to 88% in 2002-03.12 In our study increasing performance of CT scanning has probably contributed to a perceived increase in ICH incidence during the study period.

There may have been a change in referral pattern for patients suffering acute neurological events over the study period; both to hospital from the community, and for neuroimaging when in hospital. Expectations of care for patients by general practitioners (GP's) and hospital specialists are influenced by availability of tests and treatments.

An increase in availability of CT scanning might be expected to alter GP referral patterns of patients suffering an acute neurological event, and also to alter acceptable indications for in-hospital neuroimaging requests. This could equally apply to patients with transient neurological symptoms as to semi-moribund patients, who in previous times may have remained at home with an expectation of death.

We also consider that factors exist which may have led to a true increase in ICH incidence. We have observed an increase in warfarin associated ICH, are aware of an increase in age of our population, and are aware of an increase in antiplatelet medication and HMG CoA reductase inhibitor (statin) use.

Warfarin use approximately doubles the risk of ICH.13 Indications for long-term warfarin treatment include prophylaxis of cardiogenic thromboembolism in patients with atrial fibrillation or artificial heart valves, and prophylaxis of venous thromboembolism. Studies in the 1990s confirmed the benefit of treatment with warfarin to reduce the risk of stroke in patients with atrial fibrillation.14-16

The number of prescriptions for warfarin in the Waikato region has increased by a factor of 1.65 between 1999 and 2008 (Personal Communication, NZ Ministry of Health [NZMOH]). Whilst acknowledging that the temporal increase in warfarin associated ICH observed in this study is also subject to the confounding effect of changing neuroimaging practice and changing referral patterns, it may have contributed to the observed temporal increase in ICH.

There has been a 41% increase in those aged ≥75 in the Waikato region between 1996 and 2006, from 15,430 to 21,730.11 This compares to a 9% increase in total population. Statistics NZ attribute this growth to an increase in life expectancy, and an increase in birth rate during the 1930’s and 1940’s. Forty six percent of all ICH events in our analysis occurred in those over the age of 75, and the observed increase in this population represents an increase in those most at risk of ICH.

There was an increase in the use of antiplatelet and statin medication over the study period, both of which are associated with increased risk of ICH.17–19 Statin use in the Waikato region increased by a factor of 4.5 over the study period, from 20,000 prescriptions per year to 85,000 (Personal Communication – NZMOH).

Aspirin use, which is associated with a 40% increase in risk of ICH,19 increased by a factor of 3.2 in the region over the study period, from 25,000 prescriptions per year to 81,000 (Personal Communication – NZMOH). Clopidogrel and dipyridamole have been regulated over this period by PHARMAC in New Zealand restricting their use,20 and therefore are unlikely to have contributed significantly to the absolute increase in ICH observed in this study.

The average annual incidence of ICH in this study of 17.4 per 100,000 per year is comparable to published incidence figures of 20 - 30 per 100,000 per year.1,2 Ethnicity specific incidence was lower in Māori, Pacific Island and Asian populations, who were also younger at presentation in comparison to NZ Europeans. Statistics NZ data show that during the study period only 0.9% of Māori, 0.8% of Pacific Islanders and 1.0% of Asians in the Waikato region were aged 75 or older, compared to 5.9% of the European population.11 In Māori and Pacific Island populations this difference is related to a shorter life expectancy, while in the Asian population it is due to migration patterns (Asian immigrants in NZ are generally young and recently immigrated).11 Asian populations have been shown to have a higher incidence of ICH,21,22 a difference attributed to higher blood pressure.

The ARCOS stroke group demonstrated in a population based study a crude ICH incidence in Māori, Pacific Island and Asian peoples similar to that of NZ Europeans.21 Our hospital based data demonstrated a lower crude incidence rate of ICH in these ethnic groups. The difference between our data and the ARCOS data is unlikely to be the result of younger age in Māori, Pacific Island and Asian ethnic groups, who were younger in both study populations.

We feel it is most likely to be due to the confounding effect of health treatment and health seeking patterns, which probably vary by ethnicity. Cultural, financial and language barriers alter access to healthcare, and disparities in health between Māori and non-Māori in NZ are known to be partly attributable to a difference in healthcare access.23 The reduced ICH incidence we observed in Māori, Pacific Island and Asian ethnic groups may be a consequence of reduced healthcare access.

Intraventricular extension of bleed on neuroimaging, warfarin use, age and lobar bleed location were found to correlate with 30-day mortality. Ventricular extension is associated with more severe brain injury, and are also carries risk of causing obstructive hydrocephalus. Previous analyses have documented an increase in risk of mortality associated with intraventricular blood,24 and in this analysis it was a strong predictor of 30-day mortality.

An interesting finding of this study was the interaction between warfarin use and intraventricular extension of blood noted on neuroimaging. Warfarin use increases risk of death in ICH by increasing haematoma volume and duration of bleeding.25 In our study the presence of intraventricular blood increased the likelihood of death within 30 days for all patients.

For warfarin users this effect was multiplied by a factor of three, meaning the presence of intraventicular blood increased the likelihood of death within 30 days for warfarin users by a factor of 24, in comparison to those not taking warfarin and without intraventricular extension of their bleed. For those without intraventricular extension of bleed, warfarin use did not confer an increased likelihood of death within 30 days.

The prolongation of bleeding time effected by warfarin use appears to have had an especially marked detrimental effect for patients whose bleed had extended into their ventricles. A previous analysis has shown intraventricular extension of bleed to be predictive of 30-day mortality in warfarin users,26 but the relative risk in comparison to non warfarin users has not been previously described.

As patients age the risk of dying due to ICH is known to increase,24 and our analysis conforms to this finding. As people age there is a trend for them to have more medical co-morbidity, and less physiological reserve to overcome acute illness.

Brainstem location of bleed is associated with higher mortality in ICH, followed by basal ganglia then lobar location.24,27 There is less space for haematoma expansion below the tentorium, and infratentorial bleeds may quickly cause a pressure effect on crucial brainstem structures. However, in our cohort lobar location of bleed was independently associated with increased 30-day mortality.

The retrospective nature of our study means the data was subject to a number of possible confounding factors. We acknowledge that as we have relied on ICD coding at discharge there is a risk of case ascertainment bias. To further investigate this we ran an additional query on the hospital database for ‘cerebral infarction’, ‘ICH’ and ‘stroke, not specified as haemorrhage or infarction’ as diagnoses, considering the proportion of ‘unspecified’ strokes as a surrogate marker of coding accuracy.

Unfortunately the data from before 2003 was not accessible but from 2003 to 2010 there was no appreciable change in the proportion of stroke events for which neither infarction or bleed was coded as the diagnosis. This would suggest coding practice has not significantly changed over the study period, and so is unlikely to have influenced observed ICH incidence. Secondly, obtaining radiological information from the clinical radiology report may have resulted in inaccurate haematoma identification and localization, and may have introduced interobserver variability.

We also did not include recognized factors predictive of 30-day mortality in ICH – haematoma volume and Glasgow Coma Score (GCS) on presentation.24 Our identified predictor variables may have been correlated with these factors, meaning their association with 30-day mortality may not have been truly independent. Collection of bleed volume and GCS data would have required review of radiology films and paper clinical notes for each patient which was beyond the resources of our study group. However, three of the four predictor variables we have identified have previously been described as predictive of 30-day mortality in ICH.24,28

Although warfarin use has been described previously as a predictor of 30-day mortality in ICH, we are not aware of a description of the observed interaction between warfarin use and intraventricular extension. We believe there is a plausible biological explanation for why this interaction might exist and that it is likely to be a true predictor of 30-day mortality in patients suffering ICH.


Observed ICH has increased in incidence over the past 10 years in the hospitals of the Waikato region of NZ. Increasing availability of neuroimaging, increasing age and increasing use of warfarin were identified as potential contributing factors. Radiological evidence of intraventricular extension of bleed, the use of warfarin, lobar location of bleed and increasing age correlated with poorer survival. This data will be available for comparison with future studies to assess trends in incidence, patient characteristics and outcome in ICH.


This study investigated intracerebral haemorrhage (ICH) or sudden bleeding within the brain detected in the hospitals of the Waikato region of New Zealand between 1999 and 2008. 653 episodes of ICH were identified. Observed ICH has increased in incidence in our hospitals over the past 10 years. Increasing availability of neuroimaging, increasing numbers of elderly, and increasing warfarin (a blood thinner used in heart conditions)-associated ICH were likely contributors to this observed increase. Radiological evidence of extension of intraventricular bleed, warfarin use, lobar location of bleed, and increasing age correlated with poorer survival. This data will be available for comparison with future studies to assess trends in incidence, patient characteristics and outcome in ICH.



To determine the incidence, and any change in incidence, of spontaneous intracerebral haemorrhage (ICH) detected in the hospitals of the Waikato region of New Zealand (NZ) between 1999 and 2008. To analyse clinical and patient parameters, and to correlate these with outcome.


A retrospective analysis was performed on patients presenting to Waikato and Thames Hospitals with ICH during the study period. Radiology reports, blood tests and the electronic clinical record were reviewed for each patient.


653 episodes of ICH were identified. The average annual incidence per 100,000 per year was 17.4 (16.1–18.7, 95% confidence interval). This increased from an average of 14.4 (13.7–15.1) between 1999–2001 to 21.4 (20.6–22.2) between 2006–2008 (rate ratio 1.49, p<0.0001). 249 (38.1%) patients died within 30 days of their sentinel bleed. The presence of intraventricular extension of bleed on neuroimaging (Odds Ratio (OR) 6.18, p<0.001), warfarin use (OR 1.11, p=0.76), warfarin use and intraventricular extension of bleed (OR 23.8, p=0.014), lobar location of bleed (OR 1.88, p=0.001) and age (OR 1.16 for every 10-year increase in age, p=0.02) increased the likelihood of death within 30 days.


Observed ICH has increased in incidence in our hospitals over the past 10 years. Increasing availability of neuroimaging, increasing numbers of elderly, and increasing warfarin associated ICH were likely contributors to this observed increase. Radiological evidence of extension of intraventricular bleed, warfarin use, lobar location of bleed, and increasing age correlated with poorer survival. This data will be available for comparison with future studies to assess trends in incidence, patient characteristics and outcome in ICH.

Author Information

James Irwin, Medical Registrar; Peter Wright, Neurologist, Department of Neurology; Paul Reeve, General Physician, Department of General Medicine; Waikato Hospital, Hamilton


We thank Stephen Holmes (Clinical Audit Support Coordinator, Waikato Hospital) for his assistance.


James Irwin, Medical Registrar, Department of General Medicine, Waikato Hospital, Private Bag 3200, Hamilton 3240, New Zealand.

Correspondence Email


Competing Interests



  1. Van Asch CJ, Luitse MJ, Rinkel GJ, et al. Incidence, case fatality, and functional outcome of intracerebral haemorrhage over time, according to age, sex, and ethnic origin: a systematic review and meta-analysis. Lancet Neurol 2010;9:167-76.
  2. Sacco S, Marini C, Toni D, et al; Incidence and 10 year survival of intracerebral haemorrhage in a population based registry. Stroke 2009;40:394-9.
  3. Qureshi AI, Tuhrim S, Broderick JP, et al. Spontaneous intracerebral haemorrhage. N Engl J Med 2001;344:1450-60.
  4. Anderson CS, Chakera TM, Stewart-Wynne EG, et al; Spectrum of primary intracerebral haemorrhage in Perth, Western Australia 1989-90: incidence and outcome. J Neurol Neurosurg Psychiatry 1994;57(8):936-40.
  5. Islam MS, Anderson CS, Hankey GJ, et al. Trends in incidence and outcome of stroke in Perth, Western Australia during 1989 to 2001: the Perth community stroke study. Stroke 2008;39:776-82.
  6. Lovelock CE, Molyneux AJ, Rothwell PM. Change in incidence and aetiology of intracerebral haemorrhage in Oxfordshire, UK, between 1981 and 2006: a population based study. Lancet Neurol 2007;6:487-93.
  7. Benatru I, Rouaud O, Durier J, et al. Stable stroke incidence rates but improved case fatality in Dijon, France, from 1985 to 2004. Stroke 2006;37:1674-9.
  8. Sivenius J, Tuomilehto J, Immonen-Raiha P, et al. Continuous 15 year decrease in incidence and mortality of stroke in Finland: the FINSTROKE study. Stroke;2004:35:420-5.
  9. World Health Organization http://apps.who.int/classifications/apps/icd/icd10online
  10. R Development Core Team (2008). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org
  11. Statistics New Zealand http://www.stats.govt.nz
  12. Anderson CS, Carter KN, Hackett ML, et al. Trends in Stroke Incidence in Auckland, New Zealand, During 1981 to 2003. Stroke 2005;36;2087-93.
  13. Hart RG, Benavente O, McBride R, et al. Antithrombotic Therapy To Prevent Stroke in Patients with Atrial Fibrillation: A Meta-Analysis. Ann Intern Med 1999;131:492-501.
  14. Stroke Prevention in Atrial Fibrillation Investigators. Stroke Prevention in Atrial Fibrillation Study; Final results. Circulation 1991;84:527-39.
  15. Stroke Prevention in Atrial Fibrillation Investigators. Warfarin versus aspirin for prevention of thromboembolism in atrial fibrillation: Stroke Prevention in Atrial Fibrillation II Study. Lancet 1994;343:687-91.
  16. European Atrial Fibrillation Trial Study Group. Secondary prevention in non-rheumatic atrial fibrillation after transient ischemic attack or minor stroke. Lancet 1993;342:1255-62.
  17. Iso H, Jacobs DR, Wentworth D, et al. Serum cholesterol levels and six-year mortality from stroke in 350,977 men screened for the multiple risk factor intervention trial. N Engl J Med 1989;320:904-10.
  18. SPARCL investigators. High Dose Atorvastatin after Stroke or Transient Ischaemic Attack. N Engl J Med 2006;355:549-59.
  19. He J, Whelton PK, Vu B, et al. Aspirin and risk of hemorrhagic stroke: a meta-analysis of randomized controlled trials. JAMA 1998;280(22):1930-5.
  20. Pharmac NZ. http://www.pharmac.govt.nz
  21. Feigin V, Carter KM, Hackett ML, et al. Ethnic disparities in incidence of stroke subtypes: Auckland Regional Community Stroke Study, 2002-2003. Lancet Neurol 2006;5:130-9.
  22. Jiang B, Wang WZ, Chen H et al. Incidence and trends of stroke and its subtypes in China: Results from three large cities. Stroke 2006;37:63-5.
  23. Ellison-Loschmann L, Pearce N. Improving Access to Health Care Among New Zealand’s Māori Population. Am J Public Health. 2006 April; 96(4):612–7.
  24. Hemphill JC 3rd, Bonovich DC, Besmertis L, et al. The ICH score: a simple, reliable grading scale for intracerebral hemorrhage. Stroke 2001;32:891-7.
  25. Flaherty ML, Tao H, Haverbusch M, et al. Warfarin use leads to larger intracerebral hematomas. Neurology 2008;71:1084-9.
  26. Zubkov A, Claassen DO, Rabinstein AA. Warfarin associated intraventricular hemorrhage. Neurol Res. 2007 Oct;29(7):661-3.
  27. Rosenow F, Hojer C, Meyer-Lohmann C, et al. Spontaneous Intracerebral Haemorrhage. Prognostic factors in 896 cases. Acta Neurol Scand 1997;96:174–82.
  28. Rosand J, Eckman MH, Knudsen KA, et al; The effect of warfarin and intensity of anticoagulation on outcome of intracerebral hemorrhage. Arch Intern Med 2004;164(8):880-4.