Statin

Statin Use During Ischemic Stroke Hospitalization Is
Strongly Associated With Improved Poststroke Survival

Alexander C. Flint, MD, PhD; Hooman Kamel, MD; Babak B. Navi, MD; Vivek A. Rao, MD;
Bonnie S. Faigeles, MPH, MS; Carol Conell, PhD; Jeff G. Klingman, MD; Stephen Sidney, MD, MPH;
Nancy K. Hills, PhD; Michael Sorel, MPH; Sean P. Cullen, MD; S. Claiborne Johnston, MD, PhD
Background and PurposeStatins reduce infarct size in animal models of stroke and have been hypothesized to improve
clinical outcomes after ischemic stroke. We examined the relationship between statin use before and during stroke
hospitalization and poststroke survival.
MethodsWe analyzed records from 12 689 patients admitted with ischemic stroke to any of 17 hospitals in a large
integrated healthcare delivery system between January 2000 and December 2007. We used multivariable survival
analysis and grouped-treatment analysis, an instrumental variable method that uses treatment differences between
facilities to avoid individual patient-level confounding.
ResultsStatin use before ischemic stroke hospitalization was associated with improved survival (hazard ratio, 0.85; 95%
CI, 0.79 0.93; P0.001), and use before and during hospitalization was associated with better rates of survival (hazard
ratio, 0.59; 95% CI, 0.53 0.65; P0.001). Patients taking a statin before their stroke who underwent statin withdrawal
in the hospital had a substantially greater risk of death (hazard ratio, 2.5; 95% CI, 2.12.9; P0.001). The benefit was
greater for high-dose (60 mg/day) statin use (hazard ratio, 0.43; 95% CI, 0.34 0.53; P0.001) than for lower dose
(60 mg/day) statin use (hazard ratio, 0.60; 95% CI, 0.54 0.67; P0.001; test for trend P0.001), and earlier
treatment in-hospital further improved survival. Grouped-treatment analysis showed that the association between statin
use and survival cannot be explained by patient-level confounding.
ConclusionsStatin use early in stroke hospitalization is strongly associated with improved poststroke survival, and statin
withdrawal in the hospital, even for a brief period, is associated with worsened survival. (Stroke. 2012;43:147-154.)
Key Words: brain ischemia outcomes statins stroke management survival analysis
utpatient statin use reduces the risk of having a first or

recurrent ischemic stroke.13 Smaller observational studies have suggested that statin use around the time of stroke
hospitalization may reduce mortality and improve functional
outcomes.4 7 In one small randomized controlled trial and an
observational study, statin withdrawal at the time of ischemic
stroke appeared to worsen poststroke outcomes.8,9 Statins
have several mechanisms of action distinct from their lipidlowering effects that may explain a potentially protective role
in the setting of an acute ischemic stroke.10,11 Thus, statins
may have two parallel actions relevant to ischemic stroke; on
the one hand, they reduce the risk of a stroke occurring, and
on the other, they may improve the outcome after a stroke
should one occur. However, the potential association between
statin use at the time of stroke and improved poststroke
outcomes remains a subject of debate.
As a result, the question of when to start statin therapy for
a patient with an acute ischemic stroke remains unanswered.
Current guidelines recommend statin therapy for secondary

stroke prevention but do not address when statin therapy
should be started during hospital care of the patient with acute
stroke.12,13
Given the uncertainty and clinical implications surrounding this question, we examined the relationship between statin
use before and during stroke hospitalization and long-term
poststroke survival in a large cohort of patients hospitalized
with ischemic stroke in an integrated healthcare delivery
system.
Methods
Study Design
We identified all patients admitted to any of 17 hospitals operated by
Kaiser Permanente Northern California (KPNC) between January
2000 and December 2007 with a primary discharge diagnosis of
ischemic stroke (International Classification of Diseases, 9th Revision, Clinical Modification codes 433.01, 433.11, 433.21, 433.31,
Received June 1, 2011; accepted August 26, 2011.

From the Department of Neuroscience (A.C.F., V.A.R., B.S.F., S.P.C.), Kaiser Permanente, Redwood City, Redwood City, CA; the Department of
Neurology (J.G.K.), Kaiser Permanente, Walnut Creek, Walnut Creek, CA; the Division of Research, Kaiser Permanente (C.C., S.S., M.S., S.C.J.),
Oakland, CA; and the Department of Neurology (H.K., B.B.N., N.K.H., S.C.J.), University of California, San Francisco, San Francisco, CA.
The online-only Data Supplement is available at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.111.627729/-/DC1.
Correspondence to Alexander C. Flint, MD, PhD, Kaiser Permanente, Redwood City, Department of Neuroscience, 1150 Veterans Boulevard,
Redwood City, CA 94063. E-mail alexander.c.flint@kp.org
2011 American Heart Association, Inc.
Stroke is available at http://stroke.ahajournals.org
DOI: 10.1161/STROKEAHA.111.627729
Downloaded from http://stroke.ahajournals.org/

by guest on July 20, 2016
147
148
Stroke
January 2012
433.81, 433.91, 434.01, 434.11, 434.91, and 436) and who had
neuroimaging performed during hospitalization (CT and/or MRI of
the brain). KPNC is a large integrated healthcare delivery system
with 3 million members who are similar demographically to the
overall population of Northern California.14 Because of the integrated nature of KPNC, data are available for patient demographics,
medical comorbidities, periods of hospitalization, medication prescription details, and outpatient prescription fills. The 17 hospitals in
this study are staffed by a single large physician group, The
Permanente Medical Group, with a single set of Clinical Practice
Guidelines established for the treatment of patients with ischemic
stroke. Subjects were included for analysis if they were 50 years of
age (to focus on conventional stroke etiologies) and had no prior
stroke during the search period (only the first eligible stroke in the
search period was observed for each subject). Subjects were excluded if they had not been a member of the KPNC Health Plan for
1 year before the date of ischemic stroke admission, if the stroke
occurred at the time the patient was hospitalized for another reason,
or if the subjects zip code was outside the area served by KPNC.
Subjects were excluded if prestroke outpatient statin use could not be
accurately determined according to the specific criteria defined
below.
To define prestroke statin use as accurately as possible, statin use
was considered positive if the member had (1) an active statin
prescription at the time of admission; (2) at least 1 statin prescription
filled at a Kaiser pharmacy; and (3) had sufficient supply remaining
from the number dispensed to cover the time between the prescription fill and the date of hospital admission. Patients with no statin
prescription filled during the year before admission were considered
to be statin nonusers. Patients in neither the statin use nor statin
nonuse category as defined previously were excluded from analysis.
Statin use during hospital admission was determined from inpatient
pharmacy databases. Because both outpatient and inpatient prescriptions in the Kaiser Permanente integrated care delivery system are
generated and processed within the Kaiser pharmacy computer
system, all outpatient and inpatient prescriptions are captured. No
data are presented regarding statin treatment after ischemic stroke
hospitalization.
Data were collected on date of admission, date of discharge, date
of death (if any), demographics (age, gender, and race/ethnicity),
medical comorbidities (hypertension, diabetes, atrial fibrillation,
congestive heart failure, or coronary artery disease), and history of
prior stroke (before the search period of 2000 2007). To validate our
principal inclusion criterion of hospitalization for acute ischemic
stroke, a nurse analyst manually reviewed the paper and/or electronic
medical records of 731 randomly chosen patients (5.8%) and
confirmed that all patients in this subsample had a primary diagnosis
of ischemic stroke.
Observation for each subject began with the date of admission and
continued for 1 year after the admission date. The observation period
was truncated when the subject died or was lost to follow-up due to
withdrawal from Kaiser Permanente Health Plan membership (plan
disenrollment was treated as censoring in survival analysis to ensure
100% capture of all deaths). Six hundred fifty-five cases were
censored for disenrollment (5.2%). Information on death occurrence
and date of death was obtained from the KPNC membership database
and the State of California Death Certificates database linked to
the subject according to established methods.15
Statin Treatment Definitions and

Reference Groups
The following statin treatment definitions and their corresponding
reference groups were used: beforestatin use before hospitalization
for stroke, irrespective of statin use in the hospital (compared with no
statin use before hospitalization, irrespective of statin use in the
hospital); duringstatin use in the hospital, irrespective of statin use
before hospitalization (compared with no statin use in the hospital,
irrespective of statin use before hospitalization); before &
duringstatin use both before and during hospitalization (compared
with no statin use before and during hospitalization); initiation in
hospitalpatients not taking a statin before stroke who began
treatment with a statin in the hospital (compared with no statin use

before and during hospitalization); and withdrawal in
hospitalpatients who were taking a statin before hospitalization but
who did not receive a statin in the hospital (compared with statin use
both before and during hospitalization). Note that statin use is
compared with the appropriate reference group of nonusers and statin
withdrawal is compared with the appropriate group of statin users
who were continued on the drug in-hospital (outpatient statin users
who did not undergo withdrawal). For analysis of different doses of
statin use, the reference group is the appropriate group of statin
nonusers. For analysis of timing of statin use in the hospital, the
reference group is patients not treated with a statin in the hospital.
Statistical Analysis
General
For univariable analysis, categorical data were analyzed by Fisher
exact test and continuous data were compared using Student t test.
Multivariable methods included Cox proportional hazards models,
logistic regression, and grouped-treatment analysis using generalized
estimating equations with a logit link for binary outcomes (see
subsequently for details of grouped-treatment analysis).16,17 Unless
otherwise specified, all multivariable models included the following
variables: age (polynomial in age), gender, race/ethnicity (white
non-Hispanic, black, Hispanic, Asian, or other), hospital center,
medical comorbidities (presence of hypertension, diabetes, atrial
fibrillation, congestive heart failure, or coronary artery disease), and
year of hospital discharge. For grouped-treatment analysis, hospital
center and statin use were not included (see subsequently for details).
To compare survival functions across different levels of statin doses
and across different levels for timing of in-hospital statin administration, we used the log-rank test for trend. All reported probability
values are for 2-sided tests with a prespecified of 0.05. Statistical
analysis was performed using Stata (Version 10.0; Stata Corporation,
College Station, TX).
Grouped-Treatment Analysis
Grouped-treatment analysis is an instrumental variable method18 that
can be used to avoid the influence of individual patient-level
confounding such as confounding by indication.16,17 Instrumental
variable techniques center on the use of a predictor variable that
cannot mathematically interact with unmeasured confounders at the
individual subject level, yet measure the treatment environment of
interest. In our grouped-treatment analyses, we calculated the proportion of patients treated with a statin at each hospital and then
entered this proportion into a generalized estimating equations
multivariable model as a predictor variable without individual
patient statin use as a variable in the model. For example, for analysis
of statin treatment in-hospital, the proportion treated with a statin
in-hospital is determined for each hospital center and this proportion
is used as the predictor in the grouped-treatment model. In this
fashion, all patients from a given center share the same treatment
proportion as a predictor variable, thus eliminating sources of
potential confounding that take place at the individual patient level
because individual statin use is no longer in the model. At the same
time, known potential confounders for individual patients are controlled for in the multivariable model, improving model specification
(see Supplemental Figure I; http://stroke.ahajournals.org). For
grouped-treatment analyses and their corresponding individualpatient logistic regression analyses, only facilities with 100 subjects were included to minimize error in the estimation of the
proportion treated according to hospital center; in the statin withdrawal grouped-treatment model, 1 facility with 74 patients was
dropped for this reason, but all centers were included in the other 2
grouped-treatment models. Only in-hospital treatment decisions are
addressed in the grouped-treatment models (separately examining
use in-hospital, initiation in-hospital, and withdrawal in-hospital),
because only hospital-level treatments may be analyzed with this
technique. The grouped-treatment models are built using generalized
estimating equations with a logit link with an outcome of death by 1
Downloaded from http://stroke.ahajournals.org/ by guest on July 20, 2016
Flint et al
Table 1.
Statin Use During Ischemic Stroke Hospitalization
149
Patient Characteristics According to Statin Use by Epoch

Before Hospitalization
No Statin
No. of patients
No. of patients, lovastatin
Statin
During Hospitalization
P
No Statin
Statin
8940 (70.5%)
3749 (29.5%)
...
6395 (50.4%)
6294 (49.6%)
...
...
2715 (72.4%)
...
...
4964 (78.9%)
...
No. of patients, simvastatin
...
852 (22.7%)
...
...
1122 (17.8%)
...
No. of patients, other statin
...
182 (4.9%)
...
...
208 (3.3%)
...
75.5
73.8
0.001
76.5
73.5
0.001
Female
4922 (55.1%)
1808 (48.2%)
0.001
3595 (56.2%)
3135 (49.8%)
0.001
Hypertension
6875 (76.9%)
3304 (88.1%)
0.001
4816 (75.3%)
5363 (85.2%)
0.001
Diabetes mellitus
2313 (25.9%)
1905 (50.8%)
0.001
1769 (27.7%)
2449 (38.9%)
0.001
Mean age, y
Atrial fibrillation
2230 (24.9%)
944 (25.2%)
0.788
1771 (27.7%)
1403 (22.3%)
0.001
Coronary artery disease
1368 (15.3%)
1700 (45.4%)
0.001
1154 (18.1%)
1914 (30.4%)
0.001
Congestive heart failure
1448 (16.2%)
919 (24.5%)
0.001
1179 (18.4%)
1188 (18.9%)
739 (8.3%)
390 (10.4%)
0.001
654 (10.2%)
475 (7.6%)
Prior stroke
0.538
0.001
Columns are organized into 2 epochs: before hospitalization for stroke and during hospitalization for stroke, with each epoch divided
into nonusers of statins and statin users. Prior stroke indicates history of stroke before the study period (2000 2007). P values are
for Student t test for comparing age between statin nonusers and statin users and for Fisher exact test for comparing all other
variables between statin nonusers and statin users.
year and are compared with the corresponding individual patient

logistic regression model of death by 1 year.
The Institutional Review Board of the Kaiser Foundation Research
Institute approved this study.
Results
Patient Characteristics
Between January 2000 and December 2007, we identified
12 689 subjects meeting inclusion criteria. The median age
was 75 years (range, 50 105 years), and 53% were female.
The distribution of race/ethnicities was as follows: 62.3%
white non-Hispanic, 10.5% black, 10.4% Hispanic, 8.7%
Asian, and 8.1% other. Distribution of statin medication use
and patient demographics are presented in Table 1. Among
patients treated with a statin in hospital, 4388 of 6294 (70%)
were treated during the first 24 hours of admission, 1341 of
6294 (21%) were treated during the second 24 hours of
admission, and 565 of 6294 (9%) were treated after the
first 48 hours of admission. Statins were used by 3749 of
12 689 (30%) before admission; and 468 of 3749 (13%) of
these patients were not treated with a statin in-hospital
(statin withdrawal). Among the statin nonusers before
admission, 3013 of 8940 (34%) had a statin initiated
during the hospitalization.
Statin Use Before and During

Stroke Hospitalization
Statin use before and during stroke hospitalization was
associated with improved survival over the year after stroke
onset (Figure 1; Table 2). Because patients were cumulatively
exposed to statins across the 2 epochs of interest (before and
during stroke hospitalization), there was an increased strength
of association with poststroke survival in both univariable
and multivariable analysis (Table 2). Similar results were
obtained using lovastatin and simvastatin as predictors in
separate models (data not shown). All multivariable models
were adjusted for age, sex, medical comorbidities, year of
discharge, race/ethnicity, and hospital center (see

Methods).
Statin Initiation In-Hospital

To determine whether statin pretreatment is requisite for the
association with poststroke survival, we examined the effect
of statin initiation in hospital among patients not taking a
statin before their stroke (Figure 1C; Table 2). As shown in
Table 2, the hazard ratios for the cumulative 1-year risk of
death among patients with statin initiation in the hospital
(0.55) was similar to that seen for patients taking a statin
before and during hospitalization (0.59), suggesting that the
association between statin use and poststroke survival is
largely explained by use during stroke hospitalization.
Statin Withdrawal
Because previous reports have suggested that statin withdrawal at the time of stroke hospitalization may worsen
outcomes,8,9 we next asked whether subjects who were taking
a statin before stroke hospitalization had worsened survival if
their statin was discontinued in the hospital (Figure 1D; Table
2). As shown in Table 2, the survival rate was substantially
lower among patients who underwent statin withdrawal.
Compared with either statin use or statin nonuse, statin
withdrawal was associated with a higher mortality rate
(Figure 1D; Table 2). In multivariable Cox survival analysis
comparing patients who underwent statin withdrawal with
patients who did not receive a statin before or during stroke
hospitalization, the hazard ratio for death was 1.86 (95% CI,
1.58 2.17; P0.001).
DoseResponse Relationship
If a causal relationship indeed exists between statin use and
poststroke survival, a doseresponse relationship may be
expected. We therefore examined whether higher statin doses
are associated with better poststroke survival (Figure 2). For
statin use before stroke hospitalization, Cox regression anal-
150
Stroke
January 2012
Figure 1. Statin use before and during stroke hospitalization is associated with improved poststroke survival. A, One-year Kaplan-Meier
survival curves for statin users before hospitalization ( statin) and for statin nonusers before hospitalization ( statin). B, Survival
curves for statin users before and during hospitalization ( statin) and for statin nonusers before and during hospitalization ( statin).
C, Survival curves for patients not on a statin before stroke hospitalization but initiated on statin therapy in hospital ( statin) and for
statin nonusers before and during hospitalization ( statin). D, Survival curves for statin users before and during hospitalization ( statin), for statin nonusers before and during hospitalization ( statin), and for patients who were taking a statin before stroke hospitalization but did not receive a statin in hospital (withdrawal).
ysis modeling cumulative hazard of death over 1 year

poststroke yielded a hazard ratio of 0.89 (95% CI, 0.82 0.97;
P0.01) for low to moderate statin dose (60 mg/day) and a
hazard ratio of 0.65 (95% CI, 0.54 0.79; P0.001) for high
statin dose (60 mg/day) as compared with no statin use after
controlling for potential confounders (P0.001, log-rank test
for trend). For statin use before and during stroke hospitalization, the hazard ratios were 0.60 (95% CI, 0.54 0.67;
P0.001) for low- to moderate-dose statin use and 0.43 (95%
CI, 0.34 0.53; P0.001) for high-dose statin use (P0.001,
log-rank test for trend). Hazard ratios in both models at each
level of statin use are with reference to no statin use. To
Table 2. Raw Mortality Rates and Adjusted Cox Regression Analysis of Statin Use and
Poststroke Survival
Model
1-Y Mortality, No
Statin
1-Y Mortality,
Statin
Hazard Ratio for

Death
95% CI
Before
28.9%
25.1%
0.001
0.85
0.79 0.93
0.001
Before and during
33.8%
22.1%
0.001
0.59
0.530.65
0.001
Initiation in the hospital
33.8%
19.4%
0.001
0.55
0.500.61
0.001
Withdrawal in the hospital
46.2%
22.1%
0.001
2.5
2.12.9
0.001
Before indicates statin use before hospitalization for stroke, irrespective of statin use in the hospital (compared with no statin use
before hospitalization, irrespective of statin use in the hospital); Before and during, statin use both before and during hospitalization
(compared with no statin use before and during hospitalization); Initiation in the hospital, patients not taking a statin before stroke
who began treatment with a statin in the hospital (compared with no statin use before and during hospitalization); Withdrawal in the
hospital, patients who were taking a statin before hospitalization but who did not receive a statin in the hospital (compared with statin
use both before and during hospitalization).
Percentages for 1 y mortality are unadjusted; P values for unadjusted mortality calculated from Fisher exact test. Hazard ratios
represent cumulative 1 y hazard of death from multivariable Cox regression analysis adjusted for age, sex, medical comorbidities,
race/ethnicity, year of discharge, and hospital center.
Flint et al
151
Figure 2. Higher doses of statins are associated with greater poststroke survival. A, One-year Kaplan-Meier survival curves for highdose statin users before hospitalization (60 mg/day), for low- to medium-dose statin users before hospitalization (60 mg/day), and
for statin nonusers before hospitalization ( statin). B, Survival curves for the same 3 levels of statin use before and during
hospitalization.
control for differences in statin potency, we performed

additional survival analyses and Cox regression using patients on lovastatin and simvastatin (95.1% of the cohort)
after adjustment for the approximately 2-fold difference in
lipid-lowering potency between these 2 statins.19 In the
potency-adjusted analyses, the hazard ratios for statin use
before hospitalization were 0.75 (95% CI, 0.68 0.83;
P0.001) for low to moderate statin dose and 0.69 (95% CI,
0.59 0.80; P0.001) for high statin dose, and the hazard
ratios for statin use before and during hospitalization were
0.60 (95% CI, 0.53 0.67; P0.001) for low to moderate
statin dose and 0.43 (95% CI, 0.31 0.59; P0.001) for high
statin dose (P0.001, log-rank test for trend for both models). Similar and statistically significant results were obtained
when lovastatin and simvastatin doseresponse analyses were
performed separately (data not shown). We do not present the
relationship between the 2 dose levels examined and other
treatment paradigms (eg, statin withdrawal) because the
sample size is inadequate to do so.
for trend). The difference in hazard ratios was particularly

marked in the analysis of brief periods of statin withdrawal;
there was a hazard ratio of 0.38 (95% CI, 0.32 0.45;
P0.001) for patients continued on statin treatment on
hospital Day 1, a hazard ratio of 0.43 (95% CI, 0.34 0.53;
P0.001) for patients restarted on hospital Day 2, and a
hazard ratio of 0.66 (95% CI, 0.47 0.94; P0.019) for
patients restarted on hospital Day 3 or later (P0.001 for
log-rank test for trend).
Control for Severity

To control for the possibility that severity of illness might
confound the relationship between statin use and survival, we
next added multiple measures of severity to each of the
models presented previously. Addition of intubation, mechanical ventilation, or tracheostomy (entered as an aggregate
variable), gastrostomy tube placement, and hospital-acquired
pneumonia to all of the multivariable Cox regression models
described previously had no impact on the hazard ratios for
poststroke mortality (Supplemental Table I).
Timing of Statin Administration In-Hospital

Because the results described previously show that statin
administration in hospital (or statin withdrawal in the hospital) is strongly associated with poststroke outcome, we next
examined whether the day-by-day timing of statin administration (or withdrawal) in the hospital influences poststroke
survival. Survival curves showed that earlier treatment with a
statin in-hospital was associated with improved poststroke
survival both among patients not treated with a statin as an
outpatient before their stroke (Figure 3A) and among those
taking a statin as an outpatient before their stroke (Figure 3B).
Similar results were obtained from multivariable Cox regression modeling survival after controlling for potential confounders (Figure 3CD). A difference in hazard ratios (with
reference to statin nonusers) was seen for increasing delay of
statin initiation in hospital: there was a hazard ratio of 0.51
(95% CI, 0.45 0.58) for patients initiated on Day 1, a hazard
ratio of 0.57 (95% CI, 0.48 0.67) for patients initiated on
Day 2, and a hazard ratio of 0.68 (95% CI, 0.56 0.84) for
patients initiated on Day 3 or later (P0.001 for log-rank test
Control for Patient-Level Confounding:

Grouped-Treatment Analysis
To further control for individual-subject confounding, we
used the technique of grouped-treatment analysis to move the
level of analysis to the treating hospital center.16,17 In
grouped-treatment analysis, variation in unmeasured individual patient factors cannot influence the relationship between
statin use and outcome because individual-subject statin use
is not present in the models (see Supplemental Figure I);
instead, the predictor variable (the instrumental variable) is
the proportion of patients treated with a statin at each hospital
center. Across the 17 hospitals in our study, we observed
variation from hospital to hospital in the use of statins
in-hospital; the median center prescribed a statin to 50.3% of
stroke inpatients, and a range of inpatient statin prescription
was observed (lowest center, 37%; highest center,72.5%; SD,
10.3%). Table 3 shows the results of individual patient-level
logistic regression analysis in comparison with hospital-level
grouped-treatment analysis for statin treatment during hospi-
152
Stroke
January 2012
Figure 3. Statin administration early in stroke hospitalization is associated with greater poststroke survival. A, One-year Kaplan-Meier
survival curves for patients not taking a statin before stroke hospitalization. Survival curves based on in-hospital statin initiation are as
follows: statinnot treated with a statin; start day 1statin started Day 1; start day 2statin started Day 2; and start day 3statin
started Day 3 or later. B, One-year Kaplan-Meier survival curves for patients who were taking a statin befire stroke hospitalization. Survival curves based on in-hospital statin resumption are as follows: withdrawalnot treated with a statin in-hospital; start day 1statin
started Day 1; start day 2statin started Day 2; and start day 3statin started Day 3 or later. C, Increasing hazard of death with
delayed in-hospital statin treatment among patients not taking a statin before stroke hospitalization. Hazard ratios from multivariable
Cox regression are with reference to the hazard of death among patients not treated with a statin (hazard ratio [HR], 1.0). Error bars
indicate upper limit of 95% CI (P0.001 for each, P0.001 for log-rank test for trend). D, Increasing hazard of death with delayed
in-hospital statin treatment among patients who were taking a statin before stroke hospitalization. HRs from multivariable Cox regression are with reference to the hazard of death among patients not treated with a statin (HR, 1.0; P0.001 for Day 1 and Day 2,
P0.019 for Day 3, P0.001 for log-rank test for trend).
talization, statin initiation in-hospital, and statin withdrawal.

Hospital-level grouped-treatment analysis confirmed the primary analysis in each case, demonstrating that the relationship between statin use and improved poststroke survival
cannot be explained by confounding at the individual patient
level. In the case of the statin withdrawal grouped-treatment
model, the CIs become quite broad when compared with the
individual patient model. Although confounding cannot entirely explain the withdrawal results, this observation suggests that some degree of confounding is likely present in the
individual patient statin withdrawal models.
Discussion
We found that statin use before and during stroke hospitalization is associated with improved poststroke survival and
that statin withdrawal in the hospital is associated with
worsened poststroke survival. The highest survival rates were

associated with earlier statin treatment in-hospital and higher
doses. The worst survival rates were observed among outpatient statin users who underwent statin withdrawal in the
hospital, even for a brief period. Grouped-treatment analysis
showed that our results cannot be explained by confounding
at the individual patient level.
Our findings are consistent with smaller clinical studies.
Statin use before ischemic stroke onset has been associated
with reduced stroke severity,20 improved long-term functional outcome,5 and lower poststroke mortality.4,21 Statin
treatment after discharge from hospitalization for ischemic
stroke has been associated with reduced 10-year mortality.7 In
secondary analysis of data from the Stroke Prevention by
Aggressive Reduction in Cholesterol Levels (SPARCL) trial,
a randomized trial of high-dose atorvastatin in secondary
Flint et al
153
Table 3. Adjusted Logistic Regression Analysis and Grouped-Treatment Analysis for Hospital-Based
Statin Treatments Modeling Death by 1 Year Poststroke
Individual Patient Model (Logistic Regression)
Model
OR for Death
95% CI
Grouped Treatment Model

(Generalized Estimating Equations)
OR for Death
95% CI
During
0.48
0.43 0.53
0.001
0.37
0.23 0.60
0.001
Initiation in the hospital
0.49
0.430.56
0.001
0.42
0.250.68
0.001
Withdrawal in the hospital
3.1
2.43.8
0.001
1.472.1
0.02
10.0
During indicates statin use in the hospital, irrespective of statin use before hospitalization (compared with no statin use in the
hospital, irrespective of statin use before hospitalization); Initiation in the hospital, patients not taking a statin before stroke who began
treatment with a statin in the hospital (compared with no statin use before and during hospitalization); Withdrawal in the hospital,
patients who were taking a statin before hospitalization but who did not receive a statin in the hospital (compared with statin use both
before and during hospitalization).
ORs for individual subject analyses are from multivariable logistic regression modeling death by 1 y adjusted for age, sex, medical
comorbidities, race/ethnicity, year of discharge, and hospital center. ORs for grouped treatment analysis are from generalized
estimating equations analysis modeling death by 1 y with a logit link function adjusted for age, sex, medical comorbidities,
race/ethnicity, and year of discharge.
stroke prevention, a trend toward improved functional outcomes across the range of the outcome scale (including
mortality) was observed.22
Our findings are also consistent with laboratory data
addressing the effects of statins in acute ischemic stroke. In
addition to their lipid-lowering actions, statins have several
additional biochemical effects that are collectively referred to
as pleiotropic actions.10,23,24 The pleiotropic actions of
statins may be particularly important in stroke, because the
effects of statins on ischemic stroke prevention appear to
occur independent of cholesterol level before statin initiation,25 and cholesterol levels are not a significant risk factor
for ischemic stroke.26,27 In animal models of ischemic stroke,
statin pretreatment reduces stroke severity, but statin withdrawal abrogates this effect.28 Because statins have effects on
several biochemical pathways of relevance to the ischemic
neurovascular unit,29,30 statins may have a neuroprotective
effect during the acute phase of hospitalization for ischemic
stroke.11,23
Several aspects of our study support a potential causal
relationship between statin use and poststroke survival. First,
we removed the risk of an important source of bias in
observational studies of drug effects, individual patient-level
confounding, by using the technique of grouped-treatment
analysis to move the level of analysis away from the
individual patient. Second, we controlled for the possibility of
confounding by severity, which did not alter the association
between statin use and survival. Third, doseresponse effects
were observed in our study in 2 ways: higher daily doses of
statin use and increasing exposure to a statin across time
(before and during stroke hospitalization) were each associated with improved survival. Fourth, we found that pretreatment with a statin (use of a statin as an outpatient before the
stroke) was associated with improved outcome, an association that cannot be explained by differences in individual
patient factors related to the stroke. Fifth, we found reduced
survival among patients who underwent statin withdrawal.
Lastly, we found that initiation of statins early in hospitalization was associated with improved survival and that statin
withdrawal early in the hospital, even for a brief period, was

associated with worsened survival.
Our study also has limitations. It was observational in
design without randomization. To avoid bias in outcome
assessment, we used the outcome of all-cause mortality and
therefore we do not present information that might speak to a
possible mechanism of statin use in this setting (eg, a
selective association with specific causes of death). Some
variables of interest could not be ascertained. For example,
we do not have data on initial National Institutes of Health
Stroke Scale, which also might help clarify the mechanisms
by which statin use influences survival. We also do not have
data on long-term outcomes other than mortality such as a
functional outcome scale. The exclusive outcome of mortality
presented here raises the possibility that statin use might have
prolonged life without improving function or quality of life in
some patients. Although grouped-treatment analysis confirmed our primary findings, it should be noted that for the
statin withdrawal grouped-treatment model, the CIs are much
broader than those for the individual patient model, suggesting that indeed some individual patient confounding is taking
place in the individual patient model for statin withdrawal.
This is consistent with the notion that confounding by
indication may be a particular problem in analyzing medication withdrawal, because medication withdrawal may take
place in the context of limitation of care.
Based on the SPARCL trial,3 statins are recommended for
secondary stroke prevention.12,31 However, current guidelines
for acute care of the ischemic stroke patient do not address
statin use in the hospital setting,13 and Joint Commission
guidelines in the United States assess hospital rates of statin
use in patients with stroke only at the time of hospital
discharge.32 Because we found a strong association between
early hospital statin use and long-term survival, it seems
clinically prudent to treat patients with ischemic stroke with
a statin from the beginning of stroke hospitalization. Given
the association between statin withdrawal and worsened
survival, care should be taken to avoid interruption of statin
therapy among patients taking a statin before hospitalization.
154
Stroke
January 2012
Sources of Funding
This study was supported by the Centers for Disease Control and
Prevention and the Kaiser Permanente Community Benefits Research Fund.
Disclosures
15.
16.
None.
17.
References
1. Amarenco P, Labreuche J. Lipid management in the prevention of stroke:
review and updated meta-analysis of statins for stroke prevention. Lancet
Neurol. 2009;8:453 463.
2. ORegan C, Wu P, Arora P, Perri D, Mills EJ. Statin therapy in stroke
prevention: a meta-analysis involving 121 000 patients. Am J Med. 2008;
121:24 33.
3. The Stroke Prevention by Aggressive Reduction in Cholesterol Levels
(SPARCL) Investigators. High-dose atorvastatin after stroke or transient
ischemic attack. N Engl J Med. 2006;355:549 559.
4. Elkind MSV, Flint AC, Sciacca RR, Sacco RL. Lipid-lowering agent
use at ischemic stroke onset is associated with decreased mortality.
Neurology. 2005;65:253258.
5. Mart-Fa`bregas J, Gomis M, Arboix A, Aleu A, Pagonabarraga J, Belvs
R, et al. Favorable outcome of ischemic stroke in patients pretreated with
statins. Stroke. 2004;35:11171121.
6. Moonis M, Kane K, Schwiderski U, Sandage BW, Fisher M. HMG-CoA
reductase inhibitors improve acute ischemic stroke outcome. Stroke.
2005;36:1298 1300.
7. Milionis HJ, Giannopoulos S, Kosmidou M, Panoulas V, Manios E,
Kyritsis AP, et al. Statin therapy after first stroke reduces 10-year stroke
recurrence and improves survival. Neurology. 2009;72:1816 1822.
8. Blanco M, Nombela F, Castellanos M, Rodriguez-Yanez M, Garca-Gil
M, Leira R, et al. Statin treatment withdrawal in ischemic stroke: a
controlled randomized study. Neurology. 2007;69:904 910.
9. Colivicchi F, Bassi A, Santini M, Caltagirone C. Discontinuation of statin
therapy and clinical outcome after ischemic stroke. Stroke. 2007;38:
26522657.
10. Calabro` P, Yeh ETH. The pleiotropic effects of statins. Curr Opin
Cardiol. 2005;20:541546.
11. Cimino M, Gelosa P, Gianella A, Nobili E, Tremoli E, Sironi L. Statins:
multiple mechanisms of action in the ischemic brain. Neuroscientist.
2007;13:208 213.
12. Adams RJ, Albers G, Alberts MJ, Benavente O, Furie K, Goldstein LB,
et al. Update to the AHA/ASA recommendations for the prevention of
stroke in patients with stroke and transient ischemic attack. Stroke. 2008;
39:16471652.
13. Adams HP, del Zoppo G, Alberts MJ, Bhatt DL, Brass L, Furlan A, et al.
Guidelines for the early management of adults with ischemic stroke.
Circulation. 2007;115:e478 e534.
14. Go AS, Hylek EM, Phillips KA, Chang Y, Henault LE, Selby JV, et al.
Prevalence of diagnosed atrial fibrillation in adults: national implications
for rhythm management and stroke prevention: the AnTicoagulation and
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
Risk Factors in Atrial Fibrillation (ATRIA) Study. JAMA. 2001;285:

2370 2375.
Newman TB, Brown AN. Use of commercial record linkage software and
vital statistics to identify patient deaths. J Am Med Inform Assoc. 1997;
4:233237.
Johnston SC, Henneman T, McCulloch CE, van der Laan M. Modeling
treatment effects on binary outcomes with grouped-treatment variables
and individual covariates. Am J Epidemiol. 2002;156:753760.
Gillum LA, Johnston SC. Influence of physician specialty on outcomes
after acute ischemic stroke. J Hosp Med. 2008;3:184 192.
Newhouse JP, McClellan M. Econometrics in outcomes research: the use
of instrumental variables. Annu Rev Public Health. 1998;19:1734.
Brunton L, Lazo J, Parker K. Goodman & Gilmans The Pharmacological
Basis of Therapeutics, 11th ed. New York: McGraw-Hill Professional;
2005.
Greisenegger S, Mullner M, Tentschert S, Lang W, Lalouschek W. Effect
of pretreatment with statins on the severity of acute ischemic cerebrovascular events. J Neurol Sci. 2004;221:510.
Aslanyan S, Weir CJ, McInnes GT, Reid JL, Walters MR, Lees KR.
Statin administration prior to ischaemic stroke onset and survival:
exploratory evidence from matched treatment-control study. Eur
J Neurol. 2005;12:493 498.
Goldstein LB, Amarenco P, Zivin J, Messig M, Altafullah I, Callahan A,
et al. Statin Treatment and Stroke Outcome in the Stroke Prevention by
Aggressive Reduction in Cholesterol Levels (SPARCL) Trial. Stroke.
2009;40:3526 3531.
Endres M. Statins and stroke. J Cereb Blood Flow Metab. 2005;25:
10931110.
Liao JK, Laufs U. Pleiotropic effects of statins. Annu Rev Pharmacol
Toxicol. 2004;45:89 118.
Heart Protection Study Collaborative Group. MRC/BHF Heart Protection
Study of cholesterol lowering with simvastatin in 20 536 high-risk individuals: a randomised placebo-controlled trial. Lancet. 2002;360:722.
Lewington S, Whitlock G, Clarke R, Sherliker P, Emberson J, Halsey J,
et al. Blood cholesterol and vascular mortality by age, sex, and blood
pressure: a meta-analysis of individual data from 61 prospective studies
with 55 000 vascular deaths. Lancet. 2007;370:1829 1839.
Cholesterol, diastolic blood pressure, and stroke: 13 000 strokes in
450 000 people in 45 prospective cohorts. Prospective studies collaboration. Lancet. 1995;346:16471653.
Gertz K, Laufs U, Lindauer U, Nickenig G, Bohm M, Dirnagl U, et al.
Withdrawal of statin treatment abrogates stroke protection in mice.
Stroke. 2003;34:551557.
del Zoppo GJ. The neurovascular unit in the setting of stroke. J Intern
Med. 2010;267:156 171.
del Zoppo GL. Stroke and neurovascular protection. N Engl J Med.
354:553555.
Goldstein LB, Adams R, Alberts MJ, Appel LJ, Brass LM, Bushnell CD,
et al. Primary prevention of ischemic stroke. Stroke. 2006;37:15831633.
Joint Commission Specifications Manual for National Hospital Quality
Measures. Available at: www.jointcommission.org/specifications_manual
_for_national_hospital_inpatient_quality_measures/. Accessed January
2011.
Statin Use During Ischemic Stroke Hospitalization Is Strongly Associated With Improved
Poststroke Survival
Alexander C. Flint, Hooman Kamel, Babak B. Navi, Vivek A. Rao, Bonnie S. Faigeles, Carol
Conell, Jeff G. Klingman, Stephen Sidney, Nancy K. Hills, Michael Sorel, Sean P. Cullen and S.
Claiborne Johnston
Stroke. 2012;43:147-154; originally published online October 20, 2011;
doi: 10.1161/STROKEAHA.111.627729
Stroke is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright 2011 American Heart Association, Inc. All rights reserved.
Print ISSN: 0039-2499. Online ISSN: 1524-4628
The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://stroke.ahajournals.org/content/43/1/147
Data Supplement (unedited) at:

http://stroke.ahajournals.org/content/suppl/2011/10/24/STROKEAHA.111.627729.DC1.html
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published
in Stroke can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.
Once the online version of the published article for which permission is being requested is located, click
Request Permissions in the middle column of the Web page under Services. Further information about this
process is available in the Permissions and Rights Question and Answer document.
Reprints: Information about reprints can be found online at:
http://www.lww.com/reprints
Subscriptions: Information about subscribing to Stroke is online at:
http://stroke.ahajournals.org//subscriptions/
Supplementary Table: Cox regression analysis of statin use and post-stroke survival, controlling for severity measures.
Base models
Models controlling for severity measures*
Model
Hazard ratio
for death
95% C.I.
Hazard ratio
for death
95% C.I.
Before
0.85
0.79 0.93
< 0.001
0.84
0.77 0.91
< 0.001
Before & during
0.59
0.53 0.65
< 0.001
0.57
0.51 0.63
< 0.001
Initiation in hospital
0.55
0.50 0.61
< 0.001
0.54
0.49 0.61
< 0.001
Withdrawal in hospital
2.5
2.1 2.9
< 0.001
2.4
2.0 2.8
< 0.001
Before & during (<60 mg/d)
0.60
0.54 0.67
< 0.001
0.60
0.53 0.68
< 0.001
Before & during (>60 mg/d)
0.43
0.34 0.53
< 0.001
0.41
0.33 0.52
<0.001
Hospital Start Day 1
0.51
0.47 0.56
< 0.001
0.52
0.47 0.56
<0.001
0.57
0.50 0.65
< 0.001
0.56
0.49 0.63
<0.001
Hospital Start Day 3+
0.72
0.61 0.86
< 0.001
0.68
0.57 0.80
<0.001
Before = statin use prior to hospitalization for stroke, irrespective of statin use in hospital (compared to no statin use prior to hospitalization,
irrespective of statin use in hospital).
Before & during = statin use both before and during hospitalization (compared to no statin use before and during hospitalization).
Initiation in hospital = patients not taking a statin prior to stroke who began treatment with a statin in hospital (compared to no statin use before and
during hospitalization).
Withdrawal in hospital = patients who were taking a statin prior to hospitalization, but who did not receive a statin in hospital (compared to statin use
both before and during hospitalization).
Before & during (<60 mg/d) = statin use at a usual dose level (<60 mg per day) both before and during hospitalization (compared to no statin use
before and during hospitalization). Before & during (>60 mg/d) = statin use at a high dose level (>60 mg per day) both before and during
hospitalization (compared to no statin use before and during hospitalization).
Hospital Start = commencement of in-hospital statin use on the first hospital day (Day 1), on the second hospital day (Day 2), or on the third hospital
day or later (Day 3+) (compared to no statin use during hospitalization).
* Hazard ratios for the base models at left are from multivariable Cox regression modeling survival over 1 year, adjusted for age, sex, medical
comorbidities, race/ethnicity, year of discharge, and hospital center. Hazard ratios for the models controlling for severity measures are from
multivariable Cox regression modeling survival over 1 year, adjusted for age, sex, medical comorbidities, race/ethnicity, year of discharge, hospital
center, (intubation, mechanical ventilation, or tracheostomy), gastrostomy, and hospital acquired pneumonia. 95% C.I. = 95% confidence intervals.
Supplementary Table: Cox regression analysis of statin use and post-stroke survival, controlling for severity measures.
Base models
Models controlling for severity measures*
Model
Hazard ratio
for death
95% C.I.
Hazard ratio
for death
95% C.I.
Before
0.85
0.79 0.93
< 0.001
0.84
0.77 0.91
< 0.001
Before & during
0.59
0.53 0.65
< 0.001
0.57
0.51 0.63
< 0.001
Initiation in hospital
0.55
0.50 0.61
< 0.001
0.54
0.49 0.61
< 0.001
Withdrawal in hospital
2.5
2.1 2.9
< 0.001
2.4
2.0 2.8
< 0.001
Before & during (<60 mg/d)
0.60
0.54 0.67
< 0.001
0.60
0.53 0.68
< 0.001
Before & during (>60 mg/d)
0.43
0.34 0.53
< 0.001
0.41
0.33 0.52
<0.001
0.51
0.47 0.56
< 0.001
0.52
0.47 0.56
<0.001
0.57
0.50 0.65
< 0.001
0.56
0.49 0.63
<0.001
Hospital Start Day 3+
0.72
0.61 0.86
< 0.001
0.68
0.57 0.80
<0.001
Before = statin use prior to hospitalization for stroke, irrespective of statin use in hospital (compared to no statin use prior to hospitalization,
irrespective of statin use in hospital).
Before & during = statin use both before and during hospitalization (compared to no statin use before and during hospitalization).
Initiation in hospital = patients not taking a statin prior to stroke who began treatment with a statin in hospital (compared to no statin use before and
during hospitalization).
Withdrawal in hospital = patients who were taking a statin prior to hospitalization, but who did not receive a statin in hospital (compared to statin use
both before and during hospitalization).
Before & during (<60 mg/d) = statin use at a usual dose level (<60 mg per day) both before and during hospitalization (compared to no statin use
before and during hospitalization). Before & during (>60 mg/d) = statin use at a high dose level (>60 mg per day) both before and during
hospitalization (compared to no statin use before and during hospitalization).
Hospital Start = commencement of in-hospital statin use on the first hospital day (Day 1), on the second hospital day (Day 2), or on the third hospital
day or later (Day 3+) (compared to no statin use during hospitalization).
* Hazard ratios for the base models at left are from multivariable Cox regression modeling survival over 1 year, adjusted for age, sex, medical
comorbidities, race/ethnicity, year of discharge, and hospital center. Hazard ratios for the models controlling for severity measures are from
multivariable Cox regression modeling survival over 1 year, adjusted for age, sex, medical comorbidities, race/ethnicity, year of discharge, hospital
center, (intubation, mechanical ventilation, or tracheostomy), gastrostomy, and hospital acquired pneumonia. 95% C.I. = 95% confidence intervals.
Abstract
35
Abstract
Statin Use During Ischemic Stroke Hospitalization Is Strongly Associated With Improved
Poststroke Survival
Alexander C. Flint, MD, PhD1; Hooman Kamel, MD4; Babak B. Navi, MD4; Vivek A. Rao, MD1; Bonnie S.
Faigeles, MPH, MS1; Carol Conell, PhD3; Jeff G. Klingman, MD2; Stephen Sidney, MD, MPH3; Nancy K.
Hills, PhD4; Michael Sorel, MPH3; Sean P. Cullen, MD1; S. Claiborne Johnston, MD, PhD3,4
1
Department of Neuroscience, Kaiser Permanente, Redwood City, Redwood City, CA; 2 Department of Neurology, Kaiser Permanente, Walnut
Creek, Walnut Creek, CA; 3 Division of Research, Kaiser Permanente, Oakland, CA; and 4 Department of Neurology, University of California, San
Francisco, San Francisco, CA.
2000 1 2007 12
17
12,689
0.8595 CI0.79
0.93p 0.001
0.59
95 CI0.53 0.65p 0.001
2.595 CI
2.1 2.9p 0.001
60 mg/ 0.6095 CI
0.54 0.67
p 0.001 60 mg/
0.4395 CI0.34 0.53p 0.001
p 0.001
Stroke 2012; 43: 147-154
1.0
1.0
0.9
0.9
0.8
60 mg/
0.7
60 mg/
60 mg/
0.8
60 mg/
0.7
0.6
0.6
0.5
0.5
0
60 mg/
60 mg/
655
3,094
8,940
90
180
270
569
2,481
6,975
539
2,353
6,493
520
2,238
6,170
360
500
2,150
5,885
60 mg/
60 mg/
549
2,154
5,911
90
180
270
360
493
1,812
4,413
467
1,722
4,079
452
1,632
3,845
433
1,571
3,651
A 60 mg/ 60
2 mg/ 1 Kaplan-Meier B
stroke7-1.indb 35
12.6.22 1:56:08 PM
Full Article
Stroke
Vol. 5, No. 2
Stroke
Vol. 5, No. 2
Stroke
Vol. 5, No. 2
Stroke
Vol. 5, No. 2
El uso de estatinas durante la hospitalizacin por ictus

isqumico se asocia intensamente a una mejora de la
supervivencia post-ictus
Alexander C. Flint, MD, PhD; Hooman Kamel, MD; Babak B. Navi, MD; Vivek A. Rao, MD;
Bonnie S. Faigeles, MPH, MS; Carol Conell, PhD; Jeff G. Klingman, MD; Stephen Sidney, MD, MPH;
Nancy K. Hills, PhD; Michael Sorel, MPH; Sean P. Cullen, MD; S. Claiborne Johnston, MD, PhD
Antecedentes y objetivoLas estatinas reducen el tamao del infarto en modelos animales del ictus y se ha planteado la hiptesis de que mejoran los resultados clnicos tras un ictus isqumico. Hemos examinado la relacin del uso de estatinas
antes de la hospitalizacin por ictus y durante ella con la supervivencia post-ictus.
MtodosAnalizamos los registros de 12.689 pacientes ingresados por ictus isqumico en alguno de los 17 hospitales de un
amplio sistema integrado de prestacin de asistencia sanitaria entre enero de 2000 y diciembre de 2007. Utilizamos un anlisis
de supervivencia multivariable y un anlisis de tratamiento agrupado, que es un mtodo de variables instrumentales que utiliza
las diferencias de tratamiento entre distintos centros para evitar los factores de confusin a nivel de pacientes individuales.
ResultadosEl uso de estatinas antes de la hospitalizacin por ictus isqumico se asoci a una mejora de la supervivencia
(razn de riesgos, 0,85; IC del 95%, 0,790,93; p < 0,001), y su empleo antes de la hospitalizacin y durante ella se
asoci a una mejor tasa de supervivencia (razn de riesgos, 0,59; IC del 95%, 0,530,65; p < 0,001). Los pacientes que
estaban tomando una estatina antes de sufrir el ictus y en los que se suspendi esta medicacin en el hospital presentaron
un riesgo de mortalidad sustancialmente superior (razn de riesgos, 2,5; IC del 95%, 2,12,9; p < 0,001). El efecto favorable fue mayor con el empleo de estatinas en dosis altas (> 60 mg/da) (razn de riesgos, 0,43; IC del 95%, 0,340,53;
p < 0,001) en comparacin con el uso de estatinas en dosis ms bajas (< 60 mg/da) (razn de riesgos, 0,60; IC del 95%,
0,54 0,67; p < 0,001; prueba de tendencia p < 0,001), y un tratamiento ms temprano en el hospital mejor en mayor
medida la supervivencia. El anlisis de tratamiento agrupado mostr que la asociacin entre el uso de estatinas y la supervivencia no poda explicarse por factores de confusin a nivel de paciente.
ConclusionesEl uso temprano de estatinas en la hospitalizacin por ictus se asocia de forma intensa con la mejora de la
supervivencia post-ictus, y la suspensin de la medicacin de estatinas en el hospital, aunque sea durante un periodo de
tiempo breve, se asocia a un empeoramiento de la supervivencia. (Traducido del ingls: Statin Use During Ischemic
Stroke Hospitalization Is Strongly Associated With Improved Poststroke Survival. Stroke. 2012;43:147-154.)
Palabras clave: brain ischemia n outcomes n statins n stroke management n survival analysis
tienen varios mecanismos de accin que difieren de sus efectos hipolipemiantes y que pueden explicar un posible papel
protector en el contexto de un ictus isqumico agudo10,11. As,
las estatinas pueden ejercer dos acciones paralelas de inters
en el ictus isqumico; por un lado, reducen el riesgo de aparicin del ictus, y por otro pueden mejorar la evolucin clnica
tras el ictus en el caso de que este se produzca. Sin embargo,
la posible asociacin entre el uso de estatinas en el momento
de sufrir el ictus y la mejora de la evolucin tras este contina
siendo motivo de controversia.
l empleo ambulatorio de estatinas reduce el riesgo de sufrir un primer ictus isqumico o un ictus isqumico recurrente13. En estudios observacionales ms pequeos se ha
sugerido que el uso de estatinas aproximadamente en el momento en el que se produce la hospitalizacin por ictus puede
reducir la mortalidad y mejorar los resultados funcionales47.
En un pequeo ensayo controlado y aleatorizado, y en un estudio observacional, la suspensin del tratamiento con estatinas
en el momento en el que se produjo un ictus isqumico pareci empeorar los resultados clnicos post-ictus8,9. Las estatinas
Recibido el 1 de junio de 2011; aceptado el 26 de agosto de 2011.

Department of Neuroscience (A.C.F., V.A.R., B.S.F., S.P.C.), Kaiser Permanente, Redwood City, Redwood City, CA; Department of Neurology
(J.G.K.), Kaiser Permanente, Walnut Creek, Walnut Creek, CA; Division of Research, Kaiser Permanente (C.C., S.S., M.S., S.C.J.), Oakland, CA; y Department of Neurology (H.K., B.B.N., N.K.H., S.C.J.), University of California, San Francisco, San Francisco, CA.
El suplemento de datos de este artculo, disponible solamente online, puede consultarse en http://stroke.ahajournals.org/lookup/suppl/
doi:10.1161/STROKEAHA.111.627729/-/DC1.
Remitir la correspondencia a Alexander C. Flint, MD, PhD, Kaiser Permanente, Redwood City, Department of Neuroscience, 1150 Veterans Boulevard, Redwood City, CA 94063. Correo electrnico alexander.c.flint@kp.org
2011 American Heart Association, Inc.
Puede accederse a Stroke en http://stroke.ahajournals.org
DOI: 10.1161/STROKEAHA.111.627729
18
Flint y cols. El uso de estatinas durante la hospitalizacin 19

En consecuencia, la cuestin de cundo iniciar un tratamiento con estatinas en un paciente que presenta un ictus isqumico agudo contina sin estar resuelta.
Las guas actuales recomiendan el tratamiento con estatinas para la prevencin secundaria del ictus, pero no abordan
la cuestin de cundo debe iniciarse el tratamiento con estos
frmacos durante la asistencia hospitalaria del paciente con
ictus agudo12,13.
Dada la incertidumbre existente y las implicaciones clnicas derivadas de esta cuestin, hemos examinado la relacin
del uso de estatinas antes de la hospitalizacin por ictus y durante ella con la supervivencia a largo plazo post-ictus en una
amplia cohorte de pacientes hospitalizados por ictus isqumico dentro de un sistema integrado de prestacin de asistencia
sanitaria.
Mtodos
Diseo del estudio
Identificamos a todos los pacientes hospitalizados en cualquiera
de los 17 hospitales gestionados por Kaiser Permanente Northern California (KPNC) entre enero de 2000 y diciembre de
2007 con un diagnstico principal de alta de ictus isqumico
(Clasificacin Internacional de Enfermedades, 9 Revisin,
Modificacin Clnica; cdigos 433.01, 433.11, 433.21, 433.31,
433.81, 433.91, 434.01, 434.11, 434.91 y 436) y que disponan
de exploraciones de neuroimagen obtenidas durante la hospitalizacin (TC y/o RM craneales). La KPNC es un gran sistema
integrado de prestacin de asistencia sanitaria, que cuenta con
ms de 3 millones de afiliados, que demogrficamente son similares a la poblacin global del Norte de California14. Dado el
carcter integrado de la KPNC, se dispone de datos respecto a
las caractersticas demogrficas de los pacientes, las comorbilidades mdicas, los periodos de hospitalizacin, informacin
sobre prescripcin de medicacin e informacin sobre dispensacin y renovacin de la prescripcin en los pacientes ambulatorios. En los 17 hospitales de este estudio prestan asistencia
los mdicos de un mismo grupo amplio, el Permanente Medical
Group, que utiliza un mismo conjunto de guas de prctica clnica establecidas para el tratamiento de los pacientes con ictus
isqumico. Se incluy a los pacientes en este anlisis si tenan
una edad > 50 aos (con objeto de centrarse en las etiologas
convencionales del ictus) y no haban sufrido ningn ictus previo durante el periodo de la bsqueda (solamente se consider
el primer ictus elegible observado en el periodo de bsqueda
para cada individuo). Se excluy a los individuos que no haban
estado afiliados al Plan de Salud de KPNC durante 1 ao antes
de la fecha de ingreso por el ictus isqumico, a aquellos en los
que el ictus se haba producido cuando el paciente estaba hospitalizado por otra razn, y a los que tenan un cdigo postal de
su lugar de residencia que corresponda a un lugar situado fuera
del rea atendida por la KPNC. Se excluy a los individuos en
los que no fue posible determinar con exactitud el uso ambulatorio de estatinas previo al ictus, determinado con los criterios
especficos que se definen a continuacin.
Con objeto de definir el uso de estatinas previo al ictus de la
forma ms exacta posible, el uso de estatinas se consider positivo si el individuo afiliado tena (1) una prescripcin activa de una
estatina en el momento del ingreso; (2) al menos una prescripcin
de estatina dispensada en una farmacia de la Kaiser; y (3) dispona
de una cantidad restante de la medicacin dispensada suficiente
para cubrir el periodo de tiempo entre la dispensacin y la fecha
de ingreso en el hospital. Se consider que los pacientes a los que
no se haba dispensado la prescripcin de estatinas durante el ao

anterior al ingreso no estaban tomando esos frmacos. Los pacientes que, segn estos criterios, no correspondan al grupo de uso de
estatinas ni al grupo de ausencia de uso de estatinas fueron excluidos del anlisis. El uso de estatinas durante el ingreso hospitalario
se determin a partir de las bases de datos de farmacia para los
pacientes hospitalizados. Dado que las prescripciones tanto ambulatorias como intrahospitalarias del sistema de asistencia integrado Kaiser Permanente son generadas y procesadas dentro del
sistema informtico de farmacia de la Kaiser, se capturan todas las
prescripciones ambulatorias y hospitalarias. No se presentan datos
relativos al tratamiento con estatinas despus de la hospitalizacin
por ictus isqumico.
Se obtuvieron datos sobre fecha de ingreso, fecha de alta, fecha de muerte (si la haba), caractersticas demogrficas (edad,
sexo y raza/origen tnico), comorbilidades mdicas (hipertensin
arterial, diabetes, fibrilacin auricular, insuficiencia cardiaca
congestiva o enfermedad coronaria), y antecedentes de ictus previo (antes del periodo de bsqueda de 20002007). Con objeto de
validar nuestro criterio de inclusin principal de hospitalizacin
por ictus isqumico agudo, una enfermera analista revis manualmente las historias clnicas en papel y/o electrnicas de 731 pacientes elegidos de forma aleatoria (5,8%) y confirm que todos
los pacientes de esta submuestra tenan un diagnstico principal
de ictus isqumico.
La observacin se inici para cada individuo en la fecha de ingreso y continu durante 1 ao tras dicha fecha. El periodo de observacin se interrumpi cuando se produca la muerte del paciente
o la prdida del seguimiento a causa de la baja en el Plan de Salud
de Kaiser Permanente (la baja en la afiliacin al plan se trat como
una censura para el anlisis de supervivencia, con objeto de capturar el 100% del total de muertes). Se censuraron 655 casos por
baja en la afiliacin (5,2%). Se obtuvo informacin sobre la muerte
y la fecha en que esta se produjo a partir de la base de datos de
afiliacin a la KPNC y a partir de la base de datos de certificados
de defuncin del Estado de California (State of California Death
Certificates) vinculada al sujeto de estudio, segn mtodos ya establecidos15.
Definiciones del tratamiento con estatinas y grupos

de referencia
Se utilizaron las siguientes definiciones del tratamiento con estatinas
y sus correspondientes grupos de referencia: antes = uso de estatinas
antes de la hospitalizacin por ictus, con independencia del uso de
estatinas en el hospital (en comparacin con la ausencia de uso
de estatinas antes de la hospitalizacin, con independencia del uso de
estatinas en el hospital); durante = uso de estatinas en el hospital,
con independencia del uso de estatinas antes de la hospitalizacin
(en comparacin con la ausencia de uso de estatinas en el hospital,
con independencia del uso de estatinas antes de la hospitalizacin);
antes y durante = uso de estatinas tanto antes de la hospitalizacin
como durante ella (en comparacin con la ausencia de uso de estatinas antes de la hospitalizacin y durante ella); inicio en el hospital
= pacientes que no estaban tomando estatinas antes del ictus e iniciaron el tratamiento con una estatina en el hospital (en comparacin con la ausencia de uso de estatinas antes de la hospitalizacin
y durante ella); y suspensin en el hospital = pacientes que estaban
tomando una estatina antes de la hospitalizacin pero que no recibieron esta medicacin en el hospital (en comparacin con la ausencia
de uso de estatinas tanto antes de la hospitalizacin como durante
ella). Obsrvese que el uso de estatinas se compara con el grupo de
referencia apropiado de pacientes sin uso de estatinas y con suspensin de estas se compara con el grupo de referencia apropiado de uso
de estatinas continuado en el hospital (pacientes con uso ambulatorio de estatinas en los que no se suspendi la administracin). Para
20 Stroke Mayo 2012
Tabla 1. Caractersticas de los pacientes segn el uso de estatinas por periodo

Antes de la hospitalizacin
Nmero de pacientes
Durante la hospitalizacin
Sin uso de
estatinas
Estatina
Sin uso de
estatinas
Estatina
8.940 (70,5%)
3.749 (29,5%)
...
6.395 (50,4%)
6.294 (49,6%)
...
Nmero de pacientes, lovastatina
...
2.715 (72,4%)
...
...
4.964 (78,9%)
...
Nmero de pacientes, simvastatina
...
852 (22,7%)
...
...
1.122 (17,8%)
...
Nmero de pacientes, otra estatina
...
182 (4,9%)
...
...
208 (3,3%)
73,8
0,001
76,5
Mujeres
4.922 (55,1%)
1.808 (48,2%)
0,001
3.595 (56,2%)
3.135 (49,8%)
0,001
Hipertensin arterial
6.875 (76,9%)
3.304 (88,1%)
0,001
4.816 (75,3%)
5.363 (85,2%)
0,001
Diabetes mellitus
2.313 (25,9%)
1.905 (50,8%)
0,001
1.769 (27,7%)
2.449 (38,9%)
0,001
Fibrilacin auricular
2.230 (24,9%)
944 (25,2%)
0,788
1.771 (27,7%)
1.403 (22,3%)
0,001
Enfermedad coronaria
1.368 (15,3%)
1.700 (45,4%)
0,001
1.154 (18,1%)
1.914 (30,4%)
0,001
Insuficiencia cardiaca congestiva 1.448 (16,2%)
919 (24,5%)
0,001
1.179 (18,4%)
1.188 (18,9%)
0,538
Ictus previo
390 (10,4%)
0,001
654 (10,2%)
475 (7,6%)
0,001
739 (8,3%)
73,5
...
75,5
Media de edad, aos
0,001
Las columnas estn organizadas en 2 periodos: antes de la hospitalizacin por ictus y durante la hospitalizacin por ictus, y cada uno
de ellos se divide en pacientes no tratados o tratados con estatinas. Ictus previo indica antecedentes de ictus antes del intervalo de estudio (20002007). Los valores de p corresponden a la prueba de t de Student para la comparacin de la edad entre los pacientes no
tratados y los tratados con estatinas, y a la prueba exacta de Fisher para la comparacin de todas las dems variables entre pacientes
tratados y no tratados con estatinas.
el anlisis de las diferentes dosis en el uso de estatinas, el grupo de

referencia es el grupo apropiado de individuos sin uso de estatinas.
Para el anlisis del momento de uso de las estatinas en el hospital, el
grupo de referencia es el de los pacientes no tratados con una estatina en el hospital.
Anlisis estadstico
General
Para el anlisis univariable, los datos discretos se analizaron con
la prueba exacta de Fisher y los datos continuos se compararon
con la prueba de t de Student. Los mtodos multivariables incluyeron modelos de riesgos proporcionales de Cox, regresin
logstica y anlisis de tratamiento agrupado, con el empleo de
ecuaciones de estimacin generalizada con un vnculo logit para los parmetros binarios (vase ms adelante una informacin
detallada sobre el anlisis de tratamiento agrupado)16,17. Salvo
que se especifique lo contrario, todos los modelos multivariables
incluyeron las siguientes variables: edad (polinmica en edad),
sexo, raza/origen tnico (blancos no hispanos, negros, hispanos,
asiticos u otros), centro hospitalario, comorbilidades mdicas
(presencia de hipertensin arterial, diabetes, fibrilacin auricular, insuficiencia cardiaca congestiva o enfermedad coronaria), y
ao de alta del hospital. Para el anlisis de tratamiento agrupado,
no se incluy el centro hospitalario y el uso de estatinas (vase
una informacin ms detallada ms adelante). Para comparar las
funciones de supervivencia en diferentes niveles de dosis de estatinas y en distintos niveles del momento de administracin de
la estatina en el hospital, utilizamos la prueba de log-rank para la
tendencia. Todos los valores de probabilidad presentados corresponden a pruebas bilaterales, con un valor de preespecificado
de 0,05. El anlisis estadstico se llev a cabo con el empleo del
programa Stata (Versin 10.0; Stata Corporation, College Station, TX).
Anlisis de tratamiento agrupado

El anlisis de tratamiento agrupado es un mtodo de variables
instrumentales 18 que puede utilizarse para evitar la influencia
de los factores de confusin a nivel del paciente individual,

como la confusin segn la indicacin 16,17. Las tcnicas de variables instrumentales se centran en el uso de una variable predictiva que no pueda interaccionar matemticamente con factores de confusin no medidos a nivel del paciente individual, y
que sin embargo mida el entorno de tratamiento de inters. En
nuestros anlisis de tratamiento agrupado, calculamos el porcentaje de pacientes tratados con una estatina en cada hospital
y luego introdujimos este porcentaje en un modelo multivariable de ecuaciones de estimacin generalizada como variable
predictiva sin utilizar el uso de estatinas del paciente individual como variable en el modelo. Por ejemplo, para el anlisis
del tratamiento con estatinas en el hospital, el porcentaje de
individuos tratados con una estatina en el hospital se determina
para cada centro hospitalario y se emplea este porcentaje como
factor predictivo en el modelo de tratamiento agrupado. De esta forma, todos los pacientes de un determinado centro tienen
en comn la misma proporcin de tratamiento como variable
predictiva, con lo que se eliminan las fuentes de posible confusin que se dan a nivel del paciente individual ya que el uso individual de estatinas deja de formar parte del modelo. Al mismo tiempo, se introduce en el modelo multivariable un control
respecto a los posibles factores de confusin conocidos para
los pacientes individuales, con lo que mejora la especificacin
del modelo (vase la Figura I del Suplemento; http://stroke.
ahajournals.org). Para los anlisis de tratamiento agrupado y
sus correspondientes anlisis de regresin logstica de pacientes individuales, solamente se incluyeron los centros con ms
de 100 participantes, con objeto de reducir al mnimo el error
en la estimacin de la proporcin de pacientes tratados segn el
centro hospitalario; en el modelo de tratamiento agrupado de
la suspensin de las estatinas, 1 centro con 74 pacientes fue
excluido por esta razn, pero se incluyeron todos los centros en
los otros 2 modelos de tratamiento agrupado. En los modelos
de tratamiento agrupado solamente se abordan las decisiones de
tratamiento en el hospital (examinando por separado el uso en
el hospital, el inicio en el hospital y la suspensin en el hospital), ya que con esta tcnica solamente pueden analizarse los
Figura 1. El uso de estatinas antes de la hospitalizacin por ictus y durante ella se asocia a una mejora de la supervivencia post-ictus. A, Curvas de supervivencia de Kaplan-Meier a un ao para los pacientes tratados con estatinas antes de la hospitalizacin (+ estatina) y para los
no tratados con estos frmacos antes de la hospitalizacin (- estatina). B, Curvas de supervivencia para los pacientes tratados con estatinas
antes de la hospitalizacin y durante ella (+ estatina) y para los pacientes no tratados con estatinas antes de la hospitalizacin y durante ella
(- estatina). C, Curvas de supervivencia para los pacientes no tratados con estatinas antes de la hospitalizacin por ictus pero que iniciaron
un tratamiento con estos frmacos en el hospital (+ estatina) y para los pacientes no tratados con estatinas antes de la hospitalizacin y
durante ella (- estatina). D, Curvas de supervivencia para los pacientes tratados con estatinas antes de la hospitalizacin y durante ella (+ estatina), para los no tratados con estatinas antes de la hospitalizacin y durante ella (- estatina), y para los pacientes que estaban en tratamiento con una estatina antes de la hospitalizacin por ictus pero que no recibieron este frmaco en el hospital (suspensin del tratamiento).
tratamientos a nivel hospitalario. Los modelos de tratamiento

agrupado se construyen con el empleo de ecuaciones de estimacin generalizada con un vnculo logit con un resultado de
muerte en 1 ao, y se comparan con los correspondientes mo-
delos de regresin logstica para pacientes individuales de la

mortalidad en 1 ao.
Este estudio fue aprobado por el consejo de revisin interno del
Kaiser Foundation Research Institute.
Tabla 2. Tasas de mortalidad brutas y anlisis de regresin de Cox ajustado para el uso de estatinas y la
supervivencia post-ictus
Modelo
Mortalidad a 1
ao, sin estatinas
Mortalidad a 1
ao, con estatina
Razn de riesgos
para la mortalidad
IC del 95%
Antes
28,9%
25,1%
0,001
0,85
0,79 0,93
0,001
Antes y durante
33,8%
22,1%
0,001
0,59
0,530,65
0,001
Inicio en el hospital
33,8%
19,4%
0,001
0,55
0,500,61
0,001
Suspensin en el hospital
46,2%
22,1%
0,001
2,5
2,12,9
0,001
Antes indica un uso de estatinas antes de la hospitalizacin por ictus, con independencia del uso de estatinas en el hospital (en comparacin con la ausencia de uso de estatinas antes de la hospitalizacin, con independencia del uso de estatinas en el hospital); antes
y durante, uso de estatinas tanto antes de la hospitalizacin como durante ella (en comparacin con la ausencia de uso de estatinas antes
de la hospitalizacin y durante ella); inicio en el hospital, pacientes no tratados con una estatina antes del ictus que iniciaron el tratamiento con una estatina en el hospital (comparacin con la ausencia de uso de estatinas antes de la hospitalizacin y durante ella); suspensin en el hospital, pacientes que estaban tomando una estatina antes de la hospitalizacin pero que no recibieron frmacos de este
tipo en el hospital (comparacin con el uso de estatinas tanto antes de la hospitalizacin como durante ella).
Los porcentajes relativos a la mortalidad a 1 ao son sin ajustar; los valores de p para la mortalidad sin ajustar se calcularon con la
prueba exacta de Fisher. Las razones de riesgos corresponden al riesgo acumulado de muerte a 1 ao derivado de un anlisis de regresin de Cox multivariable, con un ajuste para edad, sexo, comorbilidades mdicas, raza/origen tnico, ao de alta y centro hospitalario.
22 Stroke Mayo 2012
Figura 2. Las dosis ms altas de estatinas se asocian a una mayor supervivencia post-ictus. A, Curvas de supervivencia de Kaplan-Meier
a un ao para los pacientes tratados con estatinas en dosis altas antes de la hospitalizacin (60 mg/da), para los pacientes tratados con
estatinas en dosis bajas a medias antes de la hospitalizacin (<60 mg/da), y para los pacientes no tratados con estatinas antes de la
hospitalizacin (- estatina). B, Curvas de supervivencia para los mismos 3 niveles de uso de estatinas antes de la hospitalizacin y durante ella.
Resultados
Caractersticas de los pacientes
Entre enero de 2000 y diciembre de 2007, identificamos a
12.689 pacientes que cumplan los criterios de inclusin. La
mediana de edad era de 75 aos (rango, 50105 aos), y un
53% eran mujeres. La distribucin de razas/orgenes tnicos
era la siguiente: 62,3% blancos no hispanos, 10,5% negros,
10,4% hispanos, 8,7% asiticos y 8,1% otros. La distribucin
del uso de medicacin de estatinas y de las caractersticas demogrficas de los pacientes se presentan en la Tabla 1. En
los pacientes tratados con una estatina en el hospital, 4.388
de 6.294 (70%) fueron tratados durante las primeras 24 horas
de ingreso, 1.341 de 6.294 (21%) fueron tratados durante las
segundas 24 horas de ingreso, y 565 de 6.294 (9%) lo fueron
despus de las primeras 48 horas de ingreso. Utilizaban estatinas un total de 3.749 de los 12.689 pacientes (30%) antes
del ingreso; y 468 de 3.749 (13%) de estos pacientes no recibieron tratamiento con una estatina en el hospital (suspensin
de la estatina). De entre los pacientes no tratados con estatinas antes del ingreso, en 3.013 de 8.940 (34%) se inici un
tratamiento con estatinas durante la hospitalizacin.
Uso de estatinas antes de la hospitalizacin y durante ella

El uso de estatinas antes de la hospitalizacin por ictus y durante ella se asoci a una mejora de la supervivencia a lo largo del ao siguiente a la aparicin del ictus (Figura 1; Tabla
2). Dado que los pacientes estuvieron expuestos a estatinas
de forma acumulativa en los 2 periodos de inters (antes de
la hospitalizacin por ictus y durante ella), hubo un aumento
de la intensidad de la asociacin con la supervivencia postictus tanto en el anlisis univariable como en el multivariable
(Tabla 2). Se obtuvieron resultados similares con el empleo
de lovastatina y simvastatina como factores predictivos en
modelos separados (datos no presentados). En todos los modelos multivariables se introdujo un ajuste respecto a edad,
sexo, comorbilidades mdicas, ao de alta, raza/origen tnico
y centro hospitalario (vase Mtodos).
Inicio de estatina en el hospital

Para determinar si el tratamiento previo con estatinas constituye un requisito para la asociacin con la supervivencia
post-ictus, examinamos el efecto del inicio de la administracin de estatinas en el hospital en los pacientes que no estaban tomando estatinas antes de sufrir el ictus (Figura 1C;
Tabla 2). Tal como se indica en la Tabla 2, las razones de
riesgos para el riesgo acumulado de muerte a 1 ao en los pacientes con un inicio del tratamiento con estatinas en el hospital (0,55) fueron similares a las observadas en los pacientes
que tomaban estatinas antes de la hospitalizacin y durante
ella (0,59), lo cual sugiere que la asociacin entre el uso de
estatinas y la supervivencia post-ictus se explica en gran parte por el uso de los frmacos durante la hospitalizacin por
ictus.
Suspensin de las estatinas

Dado que en estudios previos se ha sugerido que la suspensin de las estatinas en el momento de la hospitalizacin por
ictus puede empeorar los resultados clnicos8,9, nos preguntamos luego si los pacientes que estaban tomando una estatina
antes de la hospitalizacin por ictus presentaban una peor supervivencia si se suspenda el uso de la estatina en el hospital
(Figura 1D; Tabla 2). Tal como se indica en la Tabla 2, la tasa
de supervivencia fue sustancialmente inferior en los pacientes en los que se suspendi la administracin de estatinas. En
comparacin con el uso de estatinas o con la ausencia de uso
de estos frmacos, la suspensin de las estatinas se asoci a
una tasa de mortalidad ms elevada (Figura 1D; Tabla 2). En
un anlisis de Cox multivariable de la supervivencia en el
que se compar a los pacientes en los que se suspendieron las
estatinas con pacientes que no recibieron estatinas antes ni
durante la hospitalizacin por ictus, la razn de riesgos para
la muerte fue de 1,86 (IC del 95%, 1,58 2,17; p < 0,001).
Relacin dosis-respuesta
Si existe realmente una relacin causal entre el uso de estatinas y la supervivencia post-ictus, cabe prever una relacin
dosis-respuesta. As pues, examinamos si las dosis ms altas
Figura 3. La administracin de estatinas temprana en la hospitalizacin por ictus se asocia a una mayor supervivencia post-ictus. A, Curvas
de supervivencia de Kaplan-Meier a un ao en pacientes no tratados con estatinas antes de la hospitalizacin por ictus. Las curvas de supervivencia basadas en el inicio del tratamiento con estatinas en el hospital son las siguientes: -estatina = no tratados con estatinas; inicio
da 1 = inicio del tratamiento con estatinas el Da 1; inicio da 2 = inicio del tratamiento con estatinas el Da 2; e inicio da 3+ = inicio del tratamiento con estatinas el Da 3 o posterior. B, Curvas de supervivencia de Kaplan-Meier a un ao en pacientes que estaban tomando una
estatina antes de la hospitalizacin por ictus. Las curvas de supervivencia basadas en la reanudacin del tratamiento con estatinas en el
hospital son las siguientes: suspensin= no tratados con estatinas en el hospital; inicio da 1 = inicio del tratamiento con estatinas el Da 1;
inicio da 2 = inicio del tratamiento con estatinas el Da 2; e inicio da 3+ = inicio del tratamiento con estatinas el Da 3 o posterior. C, Aumento
del riesgo de muerte con el retraso en el tratamiento con estatinas en el hospital en pacientes no tratados con estatinas antes de la hospitalizacin por ictus. Las razones de riesgos derivadas de la regresin de Cox multivariable toman con referencia el riesgo de muerte en los
pacientes no tratados con estatinas (razn de riesgos [HR], 1,0). Las barras de error indican el lmite superior del IC del 95% (p < 0,001 para
cada uno, p < 0,001 para la prueba de log-rank para la tendencia). D, Aumento del riesgo de muerte con el retraso en el tratamiento con
estatinas en el hospital en pacientes que estaban tomando una estatina antes de la hospitalizacin por ictus. Las HR derivadas de la regresin de Cox multivariables toman como referencia el riesgo de muerte en los pacientes no tratados con estatinas (HR, 1,0; p < 0,001 para
el Da 1 y el Da 2, p = 0,019 para el Da 3+, p < 0,001 para la prueba de log-rank para la tendencia).
de estatinas se asocian a una mejor supervivencia post-ictus

(Figura 2). Para el uso de estatinas antes de la hospitalizacin
por ictus, el modelo del anlisis de regresin de Cox para el
riesgo acumulado de muerte a lo largo de 1 ao despus del
ictus indic una razn de riesgos de 0,89 (IC del 95%, 0,82
0,97; p = 0,01) para las dosis bajas a moderadas de estatinas
(<60 mg/da) y una razn de riesgos de 0,65 (IC del 95%,
0,540,79; p < 0,001) para las dosis altas de estatinas ( 60
mg/da) en comparacin con la ausencia de uso de estatinas
tras introducir un control respecto a posibles factores de confusin (p < 0,001, prueba de log-rank para la tendencia). Para
el uso de estatinas antes de la hospitalizacin por ictus y durante ella, las razones de riesgos fueron de 0,60 (IC del 95%,
0,540,67; p < 0,001) para el uso de estatinas en dosis bajas

a moderadas y de 0,43 (IC del 95%, 0,340,53; p < 0,001)
para el uso de estatinas en dosis altas (p < 0,001, prueba de
log-rank para la tendencia). Las razones de riesgos para ambos modelos a cada nivel de uso de estatinas se calculan con
referencia a la ausencia de uso de estatinas. Con objeto de introducir un control respecto a las diferencias en la potencia de
las estatinas, realizamos otros anlisis de supervivencia adicionales y una regresin de Cox incluyendo a los pacientes
tratados con lovastatina y simvastatina (95,1% de la cohorte)
tras un ajuste respecto a la diferencia de aproximadamente 2
veces en la potencia hipolipemiante de estos 2 frmacos19.
En los anlisis con un ajuste segn la potencia, las razones
24 Stroke Mayo 2012
Tabla 3. Anlisis de regresin logstica ajustada y anlisis de tratamiento agrupado para los modelos de la muerte
a 1 ao post-ictus con los tratamientos basados en estatinas en el hospital
Modelo de pacientes individuales
(regresin logstica)
Modelo de tratamiento agrupado

(ecuaciones de estimacin generalizada)
Modelo
OR para
la muerte
IC del 95%
Durante
0,48
0,43 0,53
0,001
Inicio en el hospital
0,49
0,430,56
0,001
Suspensin en el hospital
3,1
2,43,8
0,001
OR para
la muerte
IC del 95%
0,37
0,23 0,60
0,001
0,42
0,250,68
0,001
1,472,1
0,02
10,0
Durante indica un uso de estatinas en el hospital, con independencia del uso de estatinas antes de la hospitalizacin (en comparacin con la ausencia de uso de estatinas en el hospital, independientemente del uso de estatinas antes de la hospitalizacin); inicio en
el hospital, pacientes no tratados con una estatina antes del ictus que iniciaron el tratamiento con una estatina en el hospital (comparacin con la ausencia de uso de estatinas antes de la hospitalizacin y durante ella); suspensin en el hospital, pacientes que estaban
tomando una estatina antes de la hospitalizacin pero que no recibieron frmacos de este tipo en el hospital (comparacin con el uso
de estatinas tanto antes de la hospitalizacin como durante ella).
Los valores de OR para los anlisis de pacientes individuales proceden de un modelo de regresin logstica multivariable para la muerte en el plazo de 1 ao, con un ajuste respecto a edad, sexo, comorbilidades mdicas, raza/origen tnico, ao de alta y centro hospitalario. Los valores de OR para el anlisis de tratamiento agrupado proceden de un modelo de anlisis de ecuaciones de estimacin generalizada para la muerte en el plazo de 1 ao, con una funcin de vnculo logit ajustada para edad, sexo, comorbilidades mdicas,
raza/origen tnico y ao de alta.
de riesgos para el uso de estatinas antes de la hospitalizacin

fueron de 0,75 (IC del 95%, 0,680,83; p < 0,001) para las
dosis bajas a moderadas de estatinas y de 0,69 (IC del 95%,
0,590,80; p < 0,001) para las dosis altas de estatinas, y las
razones de riesgos para el uso de estatinas antes de la hospitalizacin y durante ella fueron de 0,60 (IC del 95%, 0,530,67;
p < 0,001) para las dosis bajas a moderadas de estatinas y de
0,43 (IC del 95%, 0,310,59; p < 0,001) para las dosis altas
de estatinas (p < 0,001, prueba de log-rank para la tendencia
en ambos modelos). Se obtuvieron unos resultados similares y
estadsticamente significativos al realizar los anlisis de dosisrespuesta de lovastatina y simvastatina por separado (datos no
presentados). No presentamos datos de la relacin entre los 2
niveles de dosis examinados y otros paradigmas de tratamiento (por ejemplo, suspensin de las estatinas) debido a que el
tamao muestral es insuficiente para hacerlo.
Momento de administracin de las estatinas en el

hospital
Dado que los resultados antes descritos indican que la administracin de estatinas en el hospital (o la suspensin del
tratamiento con estatinas en el hospital) se asocian claramente a la evolucin clnica tras el ictus, examinamos a
continuacin si el momento de administracin, da a da,
de la estatina (o la suspensin de esta) en el hospital influa
en la supervivencia post-ictus. Las curvas de supervivencia
pusieron de manifiesto que un tratamiento ms temprano
con una estatina en el hospital se asociaba a una mejora de
la supervivencia post-ictus, tanto en los pacientes no tratados con una estatina ambulatoriamente antes de sufrir el
ictus (Figura 3A) como en los que estaban tomando una
estatina ambulatoriamente antes del ictus (Figura 3B). Se
obtuvieron resultados similares en los modelos de regresin
de Cox multivariables para la supervivencia, tras introducir
un control respecto a posibles factores de confusin (Figura
3CD). Se observ una diferencia de razones de riesgos (tomando como referencia a los pacientes no tratados con estatinas) para el mayor tiempo hasta el inicio del tratamiento
con estatinas en el hospital: hubo una razn de riesgos de

0,51 (IC del 95%, 0,450,58) para los pacientes en los que
se inici el tratamiento el da 1, una razn de riesgos de
0,57 (IC del 95%, 0,48 0,67) para los pacientes en los que
se inici el da 2, y una razn de riesgos de 0,68 (IC del
95%, 0,560,84) para los pacientes en los que se inici el
da 3 o ms tarde (p < 0,001 para la prueba de log-rank para
la tendencia). La diferencia en las razones de riesgos fue
especialmente notable en el anlisis de los periodos breves
de suspensin de las estatinas; hubo una razn de riesgos de
0,38 (IC del 95%, 0,320,45; p < 0,001) para los pacientes
que continuaron con el tratamiento de estatinas el da 1 de
hospitalizacin, una razn de riesgos de 0,43 (IC del 95%,
0,340,53; p < 0,001) para los pacientes que reiniciaron el
tratamiento en el da 2 de hospitalizacin, y una razn de
riesgos de 0,66 (IC del 95%, 0,470,94; p < 0,019) para los
pacientes que reanudaron el tratamiento en el da 3 o posterior de hospitalizacin (p < 0,001 para la prueba de logrank para la tendencia).
Control respecto a la gravedad

Con objeto de introducir un control respecto a la posibilidad
de que la gravedad de la enfermedad pudiera introducir una
confusin en la relacin entre el uso de estatinas y la supervivencia, aadimos entonces mltiples medidas de la gravedad
a cada uno de los modelos presentados antes. La adicin de la
intubacin, la ventilacin mecnica o la traqueostoma (introducidas como una variable agregada), la colocacin de una
sonda de gastrostoma, y la neumona adquirida en el hospital a todos los modelos de regresin de Cox multivariables
que se han descrito antes no tuvo repercusin alguna en las
razones de riesgos para la mortalidad post-ictus (Tabla I del
Suplemento).
Control respecto a los factores de confusin a nivel

del paciente: anlisis de tratamiento agrupado
Con objeto de introducir un control adicional respecto a los
factores de confusin a nivel del paciente individual, utili-

zamos la tcnica del anlisis de tratamiento agrupado para
desplazar el nivel del anlisis al centro hospitalario en el que
se realizaba el tratamiento16,17. En el anlisis de tratamiento
agrupado, la variacin de los factores del paciente individuales no medidos no puede influir en la relacin entre el uso de
estatinas y la evolucin clnica, puesto que el uso individual
de estatinas del paciente no est presente en los modelos;
en su lugar, la variable predictiva (variable instrumental) es
el porcentaje de pacientes tratados con una estatina en cada
centro hospitalario. En los 17 hospitales de nuestro estudio,
observamos una diferencia de un hospital a otro en cuanto al
uso de estatinas en el hospital; la mediana de la prescripcin
de estatinas de los centros fue del 50,3% de los pacientes
hospitalizados por ictus, y el rango observado de prescripcin de estatinas en el hospital fue del 37% en el centro con
menor prescripcin al 72,5% en el centro de mayor prescripcin (DE, 10,3%). En la Tabla 3 se presentan los resultados
del anlisis de regresin logstica a nivel de pacientes individuales en comparacin con el anlisis de tratamiento agrupado a nivel de hospital para el empleo de estatinas durante la
hospitalizacin, el inicio del tratamiento con estatinas en el
hospital, y la suspensin de las estatinas. El anlisis de tratamiento agrupado a nivel hospitalario confirm el anlisis
primario en todos los casos, y demostr que la relacin entre
el uso de estatinas y la mejora de la supervivencia tras el ictus no puede explicarse por los factores de confusin a nivel
del paciente individual. En el caso del modelo de tratamiento
agrupado de la suspensin de las estatinas, los IC eran demasiado amplios al realizar la comparacin con el modelo del
paciente individual. Aunque los factores de confusin no permiten explicar por completo los resultados de la suspensin,
esta observacin sugiere que es probable un cierto grado de
confusin en los modelos de suspensin de las estatinas en
pacientes individuales.
Discusin
Observamos que el uso de estatinas antes de la hospitalizacin por ictus y durante ella se asocia a una mejora de la supervivencia post-ictus y que la suspensin de las estatinas en
el hospital se asocia a un empeoramiento de la supervivencia post-ictus. Las tasas ms altas de supervivencia fueron
las asociadas al tratamiento ms temprano con estatinas en
el hospital y las correspondientes a las dosis ms elevadas.
Las peores tasas de supervivencia fueron las observadas en
los pacientes ambulatorios tratados con estatinas en los que
se suspendi este tratamiento en el hospital, aunque fuera durante un periodo de tiempo breve. El anlisis del tratamiento
agrupado mostr que nuestros resultados no pueden explicarse por el efecto de confusin a nivel del paciente individual.
Nuestros resultados concuerdan con los de otros estudios
clnicos ms pequeos. El uso de estatinas antes del inicio
de un ictus isqumico se ha asociado a una reduccin de la
gravedad de este20, una mejora de los resultados funcionales a largo plazo5 y una menor mortalidad post-ictus4,21. El
tratamiento con estatinas despus del alta hospitalaria tras
sufrir un ictus isqumico se ha asociado a una reduccin de
la mortalidad a 10 aos7. En el anlisis secundario de los datos del ensayo Stroke Prevention by Aggressive Reduction
in Cholesterol Levels (SPARCL), un ensayo aleatorizado del
empleo de atorvastatina en dosis altas para la prevencin secundaria del ictus, se observ una tendencia a la mejora de
los resultados funcionales en todo el rango de la escala de
resultados (incluida la mortalidad)22.
Nuestros resultados concuerdan tambin con los datos de
laboratorio que abordan los efectos de las estatinas en el ictus
isqumico agudo. Adems de sus acciones hipolipemiantes,
las estatinas tienen otros efectos bioqumicos adicionales a
los que se denomina globalmente acciones pleiotrpicas
10,23,24
. Las acciones pleiotrpicas de las estatinas pueden ser
especialmente importantes en el ictus, ya que los efectos de
estos frmacos en la prevencin del ictus isqumico parecen
darse de manera independiente del nivel de colesterol existente antes de iniciar la administracin de la estatina25, y los
niveles de colesterol no son un factor de riesgo significativo
para el ictus isqumico26,27. En modelos animales del ictus
isqumico, el pretratamiento con estatinas reduce la gravedad del ictus, pero la suspensin de las estatinas anula este
efecto28. Dado que las estatinas ejercen efectos sobre varias
vas bioqumicas importantes para la unidad neurovascular
isqumica29,30, estos frmacos pueden tener un efecto neuroprotector durante la fase aguda de hospitalizacin por ictus
isqumico11,23.
Varios aspectos de nuestro estudio respaldan una posible
relacin causal entre el uso de estatinas y la supervivencia
post-ictus. En primer lugar, evitamos el riesgo de una importante fuente de sesgo en los estudios observacionales de
los efectos de los frmacos, la confusin a nivel del paciente
individual, mediante el uso de la tcnica de anlisis de tratamiento agrupado para desplazar el nivel de anlisis apartndolo del mbito del paciente individual. En segundo lugar,
controlamos la posibilidad de confusin derivada de la gravedad, que no modific la asociacin entre el uso de estatinas y
supervivencia. En tercer lugar, se observ una relacin dosisrespuesta en los efectos identificados en nuestro estudio, en
2 formas: tanto las dosis diarias ms altas de estatinas como
el aumento de exposicin a una estatina a lo largo del tiempo (antes de la hospitalizacin por ictus y durante ella) se
asociaron a un aumento de la supervivencia. En cuarto lugar,
observamos que el pretratamiento con una estatina (uso de
una estatina como tratamiento ambulatorio antes del ictus) se
asociaba a una mejora de los resultados, asociacin esta que
no puede explicarse por diferencias en factores individuales
del paciente relacionados con el ictus. En quinto lugar, observamos una reduccin de la supervivencia en los pacientes en
los que se suspendieron las estatinas. Finalmente, comprobamos que un inicio temprano del tratamiento con estatinas
durante la hospitalizacin se asociaba a una mejora de la supervivencia y que la suspensin temprana de las estatinas en
el hospital, incluso durante un periodo de tiempo breve, se
asociaba a un empeoramiento de la supervivencia.
Nuestro estudio tiene tambin limitaciones. Se utiliz un
diseo observacional, sin asignacin aleatoria. Con objeto de
evitar el sesgo en la evaluacin de los resultados, utilizamos
el parmetro de mortalidad por todas las causas y, por tanto,
no presentamos informacin que pudiera ser indicativa de un
posible mecanismo del uso de estatinas en este contexto (p.
ej., una asociacin selectiva con causas de muerte especficas). Algunas variables de inters no pudieron ser evaluadas.
26 Stroke Mayo 2012

Por ejemplo, no disponemos de datos de evaluacin inicial
en la National Institutes of Health Stroke Scale, que podran
haber sido tiles para esclarecer los mecanismos a travs de
los cuales el uso de estatinas influye en la supervivencia.
Tampoco disponemos de datos sobre los resultados a largo
plazo distintos de la mortalidad, como una escala de valoracin funcional. El resultado exclusivo de la mortalidad que se
presenta aqu plantea la posibilidad de que el uso de estatinas
pudiera haber prolongado la vida de los pacientes sin mejorar
la funcin o la calidad de vida de algunos de ellos. Aunque
el anlisis de tratamiento agrupado confirm los resultados
primarios, debe sealarse que para el modelo de tratamiento
agrupado de la suspensin de las estatinas, los IC son mucho
ms amplios que los existentes en el modelo de pacientes individuales, lo cual sugiere que de hecho se est produciendo
realmente una cierta confusin derivada de algunos pacientes
individuales en el modelo de suspensin de las estatinas a
nivel de pacientes individuales. Esto concuerda con la idea
de que la confusin por la indicacin puede ser un problema
especialmente importante al analizar la suspensin de la medicacin, ya que ello puede producirse en el contexto de una
limitacin de la asistencia.
A la vista de los resultados del ensayo SPARCL3, se recomienda el uso de estatinas para la prevencin secundaria del
ictus12,31. Sin embargo, las guas actuales sobre la asistencia
aguda de los pacientes con ictus isqumico no abordan el uso
de estatinas en el contexto hospitalario 13, y las guas de la
Joint Commission de los Estados Unidos evalan las tasas
hospitalarias de uso de estatinas en pacientes con ictus tan
solo en el momento del alta hospitalaria32. Dado que observamos una relacin intensa entre el uso temprano de estatinas
en el hospital y la supervivencia a largo plazo, parece clnicamente prudente tratar a los pacientes con ictus isqumico utilizando una estatina desde el inicio de la hospitalizacin por
ictus. A la vista de la asociacin entre la suspensin de las
estatinas y el empeoramiento de la supervivencia, es preciso
tener precaucin para evitar la interrupcin del tratamiento
con estatinas en los pacientes que toman estos frmacos antes
de la hospitalizacin.
Fuentes de financiacin
Este estudio fue financiado por los Centers for Disease Control and Prevention y el Kaiser Permanente Community Benefits Research Fund.
Declaraciones de intereses
Ninguna.
Bibliografa
1. Amarenco P, Labreuche J. Lipid management in the prevention of stroke:
review and updated meta-analysis of statins for stroke prevention. Lancet
Neurol. 2009;8:453 463.
121:24 33.
Neurology. 2005;65:253258.
statins. Stroke. 2004;35:11171121.
2005;36:1298 1300.

121:24 33.
Neurology. 2005;65:253258.
statins. Stroke. 2004;35:11171121.
2005;36:1298 1300.
8. Blanco M, Nombela F, Castellanos M, Rodriguez-Yanez M, Garca-Gil
M, Leira R, et al. Statin treatment withdrawal in ischemic stroke: a
controlled randomized study. Neurology. 2007;69:904 910.
9. Colivicchi F, Bassi A, Santini M, Caltagirone C. Discontinuation of statin
therapy and clinical outcome after ischemic stroke. Stroke. 2007;38:
26522657.
10. Calabro` P, Yeh ETH. The pleiotropic effects of statins. Curr Opin
Cardiol. 2005;20:541546.
11. Cimino M, Gelosa P, Gianella A, Nobili E, Tremoli E, Sironi L. Statins:
multiple mechanisms of action in the ischemic brain. Neuroscientist.
2007;13:208 213.
12. Adams RJ, Albers G, Alberts MJ, Benavente O, Furie K, Goldstein LB,
et al. Update to the AHA/ASA recommendations for the prevention of
stroke in patients with stroke and transient ischemic attack. Stroke. 2008;
39:16471652.
13. Adams HP, del Zoppo G, Alberts MJ, Bhatt DL, Brass L, Furlan A, et al.
Guidelines for the early management of adults with ischemic stroke.
Circulation. 2007;115:e478 e534.
14. Go AS, Hylek EM, Phillips KA, Chang Y, Henault LE, Selby JV, et al.
Prevalence of diagnosed atrial fibrillation in adults: national implications
for rhythm management and stroke prevention: the AnTicoagulation and
Risk Factors in Atrial Fibrillation (ATRIA) Study. JAMA. 2001;285:
2370 2375.
15. Newman TB, Brown AN. Use of commercial record linkage software and
vital statistics to identify patient deaths. J Am Med Inform Assoc. 1997;
4:233237.
16. Johnston SC, Henneman T, McCulloch CE, van der Laan M. Modeling
treatment effects on binary outcomes with grouped-treatment variables
and individual covariates. Am J Epidemiol. 2002;156:753760.
17. Gillum LA, Johnston SC. Influence of physician specialty on outcomes
after acute ischemic stroke. J Hosp Med. 2008;3:184 192.
18. Newhouse JP, McClellan M. Econometrics in outcomes research: the use
of instrumental variables. Annu Rev Public Health. 1998;19:1734.
19. Brunton L, Lazo J, Parker K. Goodman & Gilmans The Pharmacological
Basis of Therapeutics, 11th ed. New York: McGraw-Hill Professional;
2005.
20. Greisenegger S, Mullner M, Tentschert S, Lang W, Lalouschek W. Effect
of pretreatment with statins on the severity of acute ischemic cerebrovascular events. J Neurol Sci. 2004;221:510.
21. Aslanyan S, Weir CJ, McInnes GT, Reid JL, Walters MR, Lees KR.
Statin administration prior to ischaemic stroke onset and survival:
exploratory evidence from matched treatment-control study. Eur
J Neurol. 2005;12:493 498.
22. Goldstein LB, Amarenco P, Zivin J, Messig M, Altafullah I, Callahan A,
et al. Statin Treatment and Stroke Outcome in the Stroke Prevention by
Aggressive Reduction in Cholesterol Levels (SPARCL) Trial. Stroke.
2009;40:3526 3531.
23. Endres M. Statins and stroke. J Cereb Blood Flow Metab. 2005;25:
10931110.
24. Liao JK, Laufs U. Pleiotropic effects of statins. Annu Rev Pharmacol
Toxicol. 2004;45:89 118.
25. Heart Protection Study Collaborative Group. MRC/BHF Heart Protection
Study of cholesterol lowering with simvastatin in 20 536 high-risk individuals: a randomised placebo-controlled trial. Lancet. 2002;360:722.
26. Lewington S, Whitlock G, Clarke R, Sherliker P, Emberson J, Halsey J,
et al. Blood cholesterol and vascular mortality by age, sex, and blood
pressure: a meta-analysis of individual data from 61 prospective studies
27. Cholesterol, diastolic blood pressure, and stroke: 13 000 strokes in
450 000 people in 45 prospective cohorts. Prospective studies collaboration. Lancet. 1995;346:16471653.
28. Gertz K, Laufs U, Lindauer U, Nickenig G, Bohm M, Dirnagl U, et al.
Stroke. 2003;34:551557.
29. del Zoppo GJ. The neurovascular unit in the setting of stroke. J Intern
Med. 2010;267:156 171.
30. del Zoppo GL. Stroke and neurovascular protection. N Engl J Med.
354:553555.
31. Goldstein LB, Adams R, Alberts MJ, Appel LJ, Brass LM, Bushnell CD,
32. Joint Commission Specifications Manual for National Hospital Quality
2011.

27. Cholesterol, diastolic blood pressure, and stroke: 13 000 strokes in
Stroke. 2003;34:551557.
450 000 people in 45 prospective cohorts. Prospective studies collabo29. del Zoppo GJ. The neurovascular unit in the setting of stroke. J Intern
ration. Lancet. 1995;346:16471653.
Med. 2010;267:156 171.
del Zoppo
GL.deStroke
and neurovascular
N Engl J Med.
Withdrawal of statin treatment abrogates stroke protection in mice.Flint y30.
cols.
El uso
estatinas
durante laprotection.
hospitalizacin
27
354:553555.
Stroke. 2003;34:551557.
29. del Zoppo GJ. The neurovascular unit in the setting of stroke. J Intern
Med. 2010;267:156 171.
30. del Zoppo GL. Stroke and neurovascular protection. N Engl J Med.
354:553555.
2011.
2011.

Statin

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Statin

Uploaded by

Copyright:

Available Formats

Statin Use During Ischemic Stroke Hospitalization Is

Strongly Associated With Improved Poststroke Survival

utpatient statin use reduces the risk of having a first or

Current guidelines recommend statin therapy for secondary

Received June 1, 2011; accepted August 26, 2011.

Downloaded from http://stroke.ahajournals.org/

Statin Treatment Definitions and

treatment with a statin in the hospital (compared with no statin use

Downloaded from http://stroke.ahajournals.org/ by guest on July 20, 2016

Statin Use During Ischemic Stroke Hospitalization

Patient Characteristics According to Statin Use by Epoch

No. of patients, simvastatin

No. of patients, other statin

Coronary artery disease

Congestive heart failure

year and are compared with the corresponding individual patient

Statin Use Before and During

discharge, race/ethnicity, and hospital center (see

Statin Initiation In-Hospital

Downloaded from http://stroke.ahajournals.org/ by guest on July 20, 2016

ysis modeling cumulative hazard of death over 1 year

Hazard Ratio for

Before and during

Initiation in the hospital

Withdrawal in the hospital

Downloaded from http://stroke.ahajournals.org/ by guest on July 20, 2016

Statin Use During Ischemic Stroke Hospitalization

control for differences in statin potency, we performed

for trend). The difference in hazard ratios was particularly

Control for Severity

Timing of Statin Administration In-Hospital

Control for Patient-Level Confounding:

Downloaded from http://stroke.ahajournals.org/ by guest on July 20, 2016

talization, statin initiation in-hospital, and statin withdrawal.

worsened poststroke survival. The highest survival rates were

Downloaded from http://stroke.ahajournals.org/ by guest on July 20, 2016

Statin Use During Ischemic Stroke Hospitalization

Grouped Treatment Model

Initiation in the hospital

Withdrawal in the hospital

withdrawal early in the hospital, even for a brief period, was

Downloaded from http://stroke.ahajournals.org/ by guest on July 20, 2016

Risk Factors in Atrial Fibrillation (ATRIA) Study. JAMA. 2001;285:

Downloaded from http://stroke.ahajournals.org/ by guest on July 20, 2016

Data Supplement (unedited) at:

Downloaded from http://stroke.ahajournals.org/ by guest on July 20, 2016

Models controlling for severity measures*

Before & during

Before & during (<60 mg/d)

Before & during (>60 mg/d)

Hospital Start Day 1

Hospital Start Day 2

Hospital Start Day 3+

Models controlling for severity measures*

Before & during

Before & during (<60 mg/d)

Before & during (>60 mg/d)

Hospital Start Day 1

Hospital Start Day 2

Hospital Start Day 3+

95 CI0.53 0.65p 0.001

Stroke 2012; 43: 147-154

El uso de estatinas durante la hospitalizacin por ictus

Recibido el 1 de junio de 2011; aceptado el 26 de agosto de 2011.