2. We attempted to provide further information on hy- tween 2 groups, Student t test and Mann-Whitney U test were
pertension-associated ICH of the people aged 680 years, performed in distributions of parametric and nonparametric
variables, respectively. Pearson correlation was used in continu-
regarding their blood pressure profile, common risk fac- ous variables. To find independent predictors of inhospital mor-
tors, hemorrhage topography and outcome. Our hypoth- tality, multivariate analyses were constructed by forward stepwise
esis was that ICH of the very elderly people is not neces- logistic regression, for all patients and for the very elderly and the
sarily equal to that seen at younger ages, in the context of younger controls separately. Adjusted odds ratios with the respec-
established hypertension. tive 95% confidence intervals are provided. The fitness of the
models was evaluated by the Hosmer-Lemeshow goodness-of-fit
test, which was considered reliable if p 1 0.2. All p values are 2-
sided and considered significant when p ! 0.05. SPSS version 13.0
Methods for Windows (SPSS Inc., Chicago, Ill., USA) was used in all calcu-
lations, as well as in the process of randomization and selection
From March 1999 to September 2003, unselected consecutive of the control group.
patients with acute symptomatic ICH were prospectively included
in the stroke registry database of the Hospital Civil de Guadala-
jara Fray Antonio Alcalde (Guadalajara, Jalisco, Mexico) [8, 13].
A standardized, structured questionnaire was used to collect data Results
from the caregiver or patient if possible regarding demography,
relevant antecedents and current event. A total of 83 consecutive We studied 224 hypertensive persons with ICH, 56 el-
patients with primary ICH, aged 680 years, were studied. Of the
83 very elderly patients registered, we analyzed 56 (67%) persons derly patients aged 680 years and 168 gender-matched
with a medical history of hypertension. For the purposes of this controls aged !80 years (table 1). Twenty patients (36%)
study, ‘hypertensive ICH’ was defined as the case of a brain pa- were 185 years old. Fewer cases with obesity (34 vs. 49%,
renchymal hematoma associated with established hypertension, p = 0.05) and diabetes mellitus (12 vs. 24%, p = 0.06) were
in the absence of other evident etiologies, like head trauma, anti- observed among the elders than in younger controls;
coagulant use and other causes. Cases presenting with a first ICH
of lobar location were not ruled out for definite diagnosis of amy- however, these differences did not reach statistical sig-
loid angiopathy by means of cerebral biopsy. To identify possible nificance. All blood pressure measures were significantly
age-related clinical differences in ICH location and outcome, 168 lower among the very elderly patients than in controls (in
gender-matched persons with hypertension-associated ICH, aged all, p ! 0.01), and even though the ICH volume was not
!80 years, were randomly selected by a factor of 3:1 from the same different between groups, extension of the hematoma
database. The internal committee of ethics of our center approved
this study. Informed consent was obtained from all patients or into the ventricular system occurred more frequently
their closest relatives. among the patients aged 680 years than in their coun-
In all cases included, ICH was defined as a sudden focal neu- terparts (p = 0.02). Also notably, history of previous ICH
rological deficit with intraparenchymal hemorrhage seen on head was not different between both groups (table 1). We could
computed tomography (CT). Hypertensive status was defined as not find any correlation between hematoma size and
the patient history of hypertension with or without medication.
Blood pressure readings (in millimeters Hg) obtained at presenta- blood pressure readings for any group; nevertheless, ICH
tion to the Emergency Department were considered for analyses volume inversely correlated with Glasgow coma scale
to evaluate the discrete point of evolution of the ICH in which the (GCS) score at hospital presentation in both the very el-
first clinical evaluations and decisions were performed. Mean ar- derly (r = –0.652, p ! 0.001) and younger individuals (r =
terial pressure (MAP) was calculated from the systolic and dia- –0.457, p ! 0.001), as could be expected.
stolic blood pressure (SBP and DBP, respectively) measurements
on admittance to the Emergency Department as follows: MAP = Overall, the ICH topography did not differ between
DBP + 0.412 (SBP – DBP) [14]. This formula corrects for the spu- the 2 study groups; however, thalamic hemorrhages were
rious variance of calculated MAP seen in individuals with hyper- more frequent among the very elderly patients than in
tension, when estimated with other formulas [15]. ICH volume (in controls (fig. 1). Duration of the hospital stay had a me-
cubic centimeters) was calculated by analysis of CT scans accord- dian of 9 days (8 vs. 10 days in the very elderly patients
ing to the ABC/2 method, in which A is the greatest diameter on
the largest hemorrhage slice, B is the diameter perpendicular to and controls, respectively; p = 0.35), and the inhospital
A, and C is the number of axial slices with hemorrhage multiplied mortality rate was 48% (57 vs. 45% in the very elderly pa-
by the slice thickness [14]. Obesity was defined as a body mass tients and controls, respectively; p = 0.10). A nonneuro-
index 630, smoking habit as the consumption of an average of logical cause of death (mainly hospital-acquired pneu-
65 cigarettes for at least 2 days per week during 612 months and monia) occurred in 50% of the cases (53 vs. 49% in the
alcoholism as an average of 12 alcoholic drinks per day.
Pearson 2 and Fisher exact tests were used to assess propor- very elderly patients and controls, respectively; p = 0.72).
tions in nominal variables for bivariate and homogeneity (when To clarify whether the variables resulting in differences
62 variables) analyses. To compare quantitative variables be- between the 2 study groups impact on in-hospital mortal-
Hypertensive ICH in Octogenarians Cerebrovasc Dis 2007;24:196–201 197
3. 50 p = 0.04 Homogeneity, p = 0.27
40
Frequency (%)
30
20
10
0
Brain stem Cerebellum Thalamus Putamen Caudate Lobes
Fig. 1. ICH location in patients aged 680
ICH location
years (black bars) and younger persons
(white bars).
Table 1. Risk factors and clinical characteristics of the very elderly patients and gender-matched controls with
hypertensive ICH
Variables All patients Persons aged Persons aged p value
≥80 years <80 years
(n = 224) (n = 56) (n = 168)
Median age, years 57 [18–99] 83 [80–99] 63 [27–79] <0.001
Risk factors
Obesity 101 (45) 19 (34) 82 (49) 0.05
Alcoholism 24 (11) 6 (11) 18 (11) 1
Current smoking habit 41 (18) 9 (16) 32 (19) 0.62
Diabetes mellitus 48 (21) 7 (12) 41 (24) 0.06
Cardiac events 5 (2) 2 (4) 3 (2) 0.43
Previous ICH 32 (14) 9 (16) 23 (14) 0.66
Blood pressure profile at hospital admission
Mean SBP 8 SD, mm Hg 173.8833.9 158.5827.6 178.7834.4 <0.001
Mean DBP 8 SD, mm Hg 101.7817.6 95.2818.1 103.8817.0 0.006
Mean MAP 8 SD, mm Hg 131.4822.5 121.3820.1 134.7822.4 <0.001
Median GCS score at hospital admission 11 [3–15] 11 [4–15] 12 [3–15] 0.20
Median duration of hospitalization, days 9 [1–82] 8 [1–82] 10 [1–70] 0.35
ICH characteristics
Median volume, cm3 20 [1–160] 15.5 [1–160] 20 [2–160] 0.58
Irruption into ventricles 127 (57) 39 (70) 88 (52) 0.02
Surgical treatment 96 (43) 24 (43) 72 (43) 1
Figures in parentheses are percentages and values in square brackets represent ranges. GCS = Glasgow coma
scale; SD = standard deviation. p value for differences between patients aged ≥80 years and younger persons;
Pearson 2, Fisher exact test, Student t test (for means) or Mann-Whitney U test (for medians), as appropriate.
Surgical treatment: hematoma evacuation with or without ventriculostomy.
198 Cerebrovasc Dis 2007;24:196–201 Chiquete et al.
4. Table 2. Inhospital mortality in relation to differences in the blood pressure profile and ICH topography, in the
very elderly patients and gender-matched controls with hypertensive ICH
Variables All patients Persons aged Persons aged p value
≥80 years <80 years
Inhospital mortality 107 (48) 32 (57) 75 (45) 0.10
Inhospital mortality by ICH location1
Brain stem 9 (8) 0 (0) 9 (12) 0.04
Cerebellum 10 (9) 4 (12) 6 (8) 0.46
Thalamus 28 (26) 10 (31) 18 (24) 0.43
Putamen 17 (16) 6 (19) 11 (15) 0.60
Caudate 2 (2) 1 (3) 1 (1) 0.53
Lobes 14 (13) 7 (23) 7 (9) 0.08
Inhospital mortality and irruption to ventricles2
Ventricular extension present 79 (74) 27 (84) 52 (69) 0.10
Inhospital mortality and blood pressure3
Mean SBP 8 SD, mm Hg 177.7836.9 157.7 (32.1) 185.6 (35.9) 0.001
Mean DBP 8 SD, mm Hg 101.6818.7 91.7 (20.1) 105.5 (16.7) 0.004
Mean MAP 8 SD, mm Hg 132.9824.8 118.9 (23.2) 138.5 (23.4) 0.001
Figures in parentheses are percentages. SD = Standard deviation. p value for differences between patients
aged ≥80 years and younger persons; Pearson 2, Fisher exact test or Student t test, as appropriate.
1
Location of hematomas in the patients who died in hospital.
2 Presence of extension to ventricles in the patients who died in hospital.
3 Mean blood pressure measures in the patients who died in hospital.
Table 3. Determinants of inhospital mortality: three binary logistic regression models
Variables Multivariate odds ratios
all patients persons aged persons aged
≥80 years <80 years
Age 1.05 (1.02–1.08) NS 1.06 (1.01–1.10)
GCS score at hospital admission 0.61 (0.52–0.75) 0.66 (0.50–0.86) 0.58 (0.49–0.70)
ICH volume 1.03 (1.01–1.05) NS 1.03 (1.01–1.05)
ICH in infratentorial location 7.47 (2.20–25.30) NS 6.28 (1.49–26.4)
Figures in parentheses are 95% confidence intervals. NS = Nonsignificant variable, hence, not appearing as
predictor after multivariate analysis. Hosmer-Lemeshow test for goodness of fit in final step of the regression
model. Only variables significantly associated with mortality are shown. All patients: 2 = 4.38, d.f. = 8, p =
0.63. Persons aged ≥80 years: 2 = 4.90, d.f. = 6, p = 0.77. Persons aged <80 years: 2 = 4.76, d.f. = 8, p = 0.78.
ity, we analyzed the ICH topography and blood pressure Multivariate analyses were performed to find indepen-
measures among the patients who died in the hospital, dent predictors of inhospital mortality, among the very el-
according to the respective age group (table 2). Except for derly patients, younger controls and all patients combined
the brain stem location of ICH and the blood pressure (table 3). Age, GCS score at hospital admission, ICH vol-
readings, no differences were found suggesting a clear ume and infratentorial location of the hematoma were in-
impact on inhospital mortality. In any case, the observed dependent factors associated with inhospital mortality in
differences favored the very elderly group (i.e. the blood all patients combined and in persons aged !80 years sepa-
pressure profile was less severe and no brain stem loca- rately. In the very elderly group, only GCS score at hospital
tions were found in persons aged 680 years). admission was inversely associated with mortality.
Hypertensive ICH in Octogenarians Cerebrovasc Dis 2007;24:196–201 199
5. Discussion tive effect on prognosis, which confirms previous findings
on the importance of variables related to stroke severity
The very elderly represent the fastest growing sub- and neurological deterioration as the main predictors of
group of the world population [1]. Hence, there is an ur- immediate unfavorable outcomes [13, 25–28]. Apparently
gent need for scientific information on common chronic- our results question whether a binomial age cutoff of 80
degenerative diseases in this particular group of people years is appropriate in distinguishing the patients with the
[12, 16, 17]. To the best of our knowledge, this is the first highest probability of dying after ICH [29]. However, the
study that examines the characteristics of hypertension- present study is not powered to prove this issue.
associated ICH in people aged 680 years. We observed The reason of the lower blood pressure readings and
several differences in hypertensive ICH occurring in very the trend towards less obesity and diabetes mellitus in the
elderly persons, as compared with their younger counter- very elderly people as compared with younger patients
parts. At admission to the emergency room, the very el- should be explored. Nonetheless, it supports the observa-
derly patients had lower blood pressure readings than the tion that persons reaching the very old age had fewer vas-
younger patients, which might be due to a distinct pattern cular risk factors earlier in midlife [17, 30] and thus reach
of hypertension of the very elderly [18], to a healthier risk an advanced old age with better quality of life. Usually
profile or to a different autonomic response to the brain this ‘old-old’ population begins with chronic comorbidi-
damage [19, 20], issues that deserve more exploration. ties late in life and spends !1 year of their lives in a
Brain hematomas located in the thalamus were more fre- medical institution [31].
quent among the very elderly than in younger persons, an The present report indeed has limitations, mainly the
intriguing finding not previously reported for the hyper- sample size and cross-sectional design that may be un-
tensive very elderly patients. This observation should be derpowered to detect some differences. Thus, the find-
confirmed with a larger cohort which includes a detailed ings reported here should be interpreted cautiously. Fur-
classification based on the topographic subtypes of tha- ther studies with a larger number of observations are nec-
lamic hemorrhage. essary to confirm our results, especially longitudinal
Hematoma extension into the ventricles occurred in long-term analyses on prognosis. Of note is the fact that
70% of cases of the eldest group and in 52% of their youn- even when we studied a very old population, the frequen-
ger counterparts, which is a significant difference. This cy of lobar hemorrhages was only 12.5%, which empha-
finding suggests that the parenchymal degeneration ob- sizes the participation of the established hypertension in
served with the aging process is an important factor that the occurrence of ICH. However, it is important to under-
may contribute to the more catastrophic effect of ICH in line that other causative or permissive factors were not
the very elderly people [12, 13, 21–24]. We have hypoth- definitively excluded. Cerebral amyloid angiopathy might
esized that the aging process may imply changes in brain play a role in the population described here, especially in
plasticity, a factor that may cause a decline in the strength the cases of lobar hematomas [22], for whom cerebral bi-
of the brain tissue normally acting as a restriction wall to opsy or necropsy was not performed. Therefore, taking
the extension of the hematoma [13]. into account the level of evidence provided here, hyper-
The existing differences between the very elderly and tension should be considered as a precipitating factor of
the younger persons did not influence the immediate out- ICH, rather than a definite etiology.
come, as evaluated by inhospital mortality rate, which
contrasts with a previous report on all-cause ICH in per-
sons aged 685 years [12]. However, given the small sample Conclusions
size of our study, it is possible that this nonsignificant dif-
ference (on a statistical basis) in mortality between the Hypertensive very elderly patients presenting with
very elderly cohort and the control group is influenced by ICH have several differences from their younger counter-
random bias and is thus a clinically significant difference parts; however, these differences do not seem to impact
(absolute difference in mortality of 12%). We observed that significantly on the immediate outcome. Further re-
previous comorbidities and risk factors did not determine search is needed to clarify whether very elderly patients
an immediate fatal outcome and that independent predic- with hypertensive ICH indeed have a distinct profile in
tors of mortality were different from those resulting in dif- risk factors and a different long-term prognosis influ-
ferences between the 2 age groups. Also, further increas- enced by biological variables other than age.
ing in age in the very elderly group did not add any nega-
200 Cerebrovasc Dis 2007;24:196–201 Chiquete et al.
6. References
1 Louria DB: Extraordinary longevity: indi- 11 Longstreth WT Jr, Bernick C, Fitzpatrick A, 20 Salerno JA, Grady C, Mentis M, et al: Brain
vidual and societal issues. J Am Geriatr Soc et al: Frequency and predictors of stroke metabolic function in older men with chron-
2005;53:S317–S319. death in 5888 participants in the Cardiovas- ic essential hypertension. J Gerontol A Biol
2 Lloyd-Jones DM, Evans JC, Levy D: Hyper- cular Health Study. Neurology 2001;56:368– Sci Med Sci 1995;50:M147–M154.
tension in adults across the age spectrum: 375. 21 Hiroki M, Miyashita K, Oe H, et al: Link be-
current outcomes and control in the com- 12 Arboix A, Vall-Llosera A, Garcia-Eroles L, tween linear hyperintensity objects in cere-
munity. JAMA 2005;294:466–472. Massons J, Oliveres M, Targa C: Clinical fea- bral white matter and hypertensive intrace-
3 European Stroke Initiative Writing Com- tures and functional outcome of intracere- rebral hemorrhage. Cerebrovasc Dis 2004;
mittee; Writing Committee for the EUSI Ex- bral hemorrhage in patients aged 85 and old- 18:166–173.
ecutive Committee: Recommendations for er. J Am Geriatr Soc 2002;50:449–454. 22 Ritter MA, Droste DW, Hegedus K, et al:
the management of intracranial haemor- 13 Ruiz-Sandoval JL, Romero-Vargas S, Chi- Role of cerebral amyloid angiopathy in intra-
rhage. I. Spontaneous intracerebral haemor- quete E, et al: Hypertensive intracerebral cerebral hemorrhage in hypertensive pa-
rhage. Cerebrovasc Dis 2006;22:294–316. hemorrhage in young people: previously un- tients. Neurology 2005;64:1233–1237.
4 Brott T, Thalinger K, Hertzberg V: Hyper- noticed age-related clinical differences. 23 Meyer JS, Rogers RL, Mortel KF: Prospective
tension as a risk factor for spontaneous intra- Stroke 2006;37:2946–2950. analysis of long-term control of mild hyper-
cerebral hemorrhage. Stroke 1986; 17: 1078– 14 Kothari RU, Brott T, Broderick JP: The ABCs tension on cerebral blood flow. Stroke 1985;
1083. of measuring intracerebral hemorrhage vol- 16:985–990.
5 Feldmann E, Broderick JP, Kernan WN, et al: umes. Stroke 1996; 27:1304–1305. 24 O’Sullivan C, Duggan J, Lyons S, et al: Hy-
Major risk factors for intracerebral hemor- 15 Meaney E, Alva F, Moguel R, Meaney A, Alva pertensive target-organ damage in the very
rhage in the young are modifiable. Stroke J, Webel R: Formula and nomogram for the elderly. Hypertension 2003;42:130–135.
2005;36:1881–1885. sphygmomanometric calculation of the 25 Dennis MS: Outcome after brain haemor-
6 Zia E, Pessah-Rasmussen H, Khan FA, et al: mean arterial pressure. Heart 2000;84:64. rhage. Cerebrovasc Dis 2003;16:224–229.
Risk factors for primary intracerebral hem- 16 Arboix A, Garcia-Eroles L, Massons J, Oli- 26 Rabinstein AA, Tisch SH, McClelland RL,
orrhage: a population-based nested case- veres M, Targa C: Lacunar infarcts in pa- Wijdicks EF: Cause is the main predictor of
control study. Cerebrovasc Dis 2006; 21: 18– tients aged 85 years and older. Acta Neurol outcome in patients with pontine hemor-
25. Scand 2000;101:25–29. rhage. Cerebrovasc Dis 2004;17:66–71.
7 Ruiz-Sandoval JL, Cantu C, Barinagarre- 17 Zeevi N, Chabra J, Silverman IE, Lee NS, Mc- 27 Dziewas R, Kremer M, Ludemann P, Nabavi
menteria F: Intracerebral hemorrhage in Cullough LD: Acute stroke management in DG, Drager B, Ringelstein B: The prognostic
young people: analysis of risk factors, loca- the elderly. Cerebrovasc Dis 2007; 23: 304– impact of clinical and CT parameters in pa-
tion, causes, and prognosis. Stroke 1999; 30: 308. tients with pontine hemorrhage. Cerebro-
537–541. 18 Thrift AG, McNeil JJ, Forbes A, Donnan GA; vasc Dis 2003;16:224–229.
8 Ruiz-Sandoval JL, Ortega-Alvarez L, Gar- Melbourne Risk Factor Study Group: Three 28 Ruiz-Sandoval JL, Chiquete E, Romero-Var-
cia-Navarro V, Romero-Vargas S, Gonzalez- important subgroups of hypertensive per- gas S, Padilla-Martinez JJ, Gonzalez-Corne-
Cornejo S: Intracerebral haemorrhage in a sons at greater risk of intracerebral hemor- jo S: Grading scale for prediction of outcome
referral hospital in the central-western re- rhage. Hypertension 1998;31:1223–1229. in primary intracerebral hemorrhages.
gion of Mexico. Rev Neurol 2005; 40: 656– 19 Sadoshima S, Yoshida F, Ibayashi S, Shioka- Stroke 2007;38:1641–1644.
660. wa O, Fujishima M: Upper limit of cerebral 29 Hemphill JC 3rd, Bonovich DC, Besmertis L,
9 Broderick J, Brott T, Tomsick T, Leach A: Lo- autoregulation during development of hy- Manley GT, Johnston SC: The ICH score: a
bar hemorrhage in the elderly: the undimin- pertension in spontaneously hypertensive simple, reliable grading scale for intracere-
ishing importance of hypertension. Stroke rats – effect of sympathetic denervation. bral hemorrhage. Stroke 2001;32:891–897.
1993;24:49–51. Stroke 1985;16:477–481. 30 Joe S, Baser RE, Breeden G, Neighbors HW,
10 Rothwell PM, Coull AJ, Giles MF, et al: Jackson JS: Prevalence of and risk factors for
Change in stroke incidence, mortality, case- lifetime suicide attempts among blacks in
fatality, severity, and risk factors in Oxford- the United States. JAMA 2006; 296: 2112–
shire, UK, from 1981 to 2004 (Oxford Vascu- 2123.
lar Study). Lancet 2004;363:1925–1933. 31 Vaillant GE, Mukamal K: Successful aging.
Am J Phychiatry 2001;158:839–847.
Hypertensive ICH in Octogenarians Cerebrovasc Dis 2007;24:196–201 201