Lindenauer PK, Pekow P, Wang K, Gutierrez B, Benjamin EM. Lipid-Lowering Therapy and In-Hospital Mortality Following Major Noncardiac Surgery. JAMA. 2004;291(17):2092-2099. doi:10.1001/jama.291.17.2092
Author Affiliations: Division of Healthcare Quality, Baystate Medical Center, Springfield, Mass (Drs Lindenauer, Pekow, and Benjamin); Department of Medicine, Tufts University School of Medicine, Boston, Mass (Drs Lindenauer and Benjamin); School of Public Health and Health Sciences, University of Massachusetts, Amherst (Dr Pekow and Mr Wang); Premier Healthcare Informatics, Premier Incorporated, Charlotte, NC (Dr Gutierrez).
Context Cardiovascular complications following major noncardiac surgery are
an important source of perioperative morbidity and mortality. Although lipid-lowering
medications are considered a key component in the primary and secondary prevention
of cardiovascular disease, their potential benefit during the perioperative
period is uncertain.
Objective To examine the association between treatment with lipid-lowering medications
and in-hospital mortality following major noncardiac surgery.
Design, Setting, and Patients A retrospective cohort study based on hospital discharge and pharmacy
records of 780 591 patients aged 18 years or older who underwent major
noncardiac surgery from January 1, 2000, to December 31, 2001, at any 1 of
329 hospitals throughout the United States. Only patients who survived through
at least the second hospital day were included. Lipid-lowering therapy was
defined as use during the first 2 hospital days. Propensity matching was used
to adjust for numerous baseline differences.
Main Outcome Measure In-hospital mortality.
Results Of the 780 591 patients, 77 082 patients (9.9%) received lipid-lowering
therapy perioperatively and 23 100 (2.96%) died during the hospitalization.
Treatment with lipid-lowering agents was associated with lower crude mortality
(2.13% vs 3.05%, P<.001). In an analysis using
matching by propensity score, 1595 patients (2.18%) treated with lipid-lowering
medications died compared with 4158 patients (3.15%) who did not receive therapy
or in whom treatment was initiated after the second day (P<.001). After adjusting for residual differences in the propensity
matched groups using conditional logistic regression, risk of mortality remained
lower among treated patients (adjusted odds ratio [OR], 0.62; 95% confidence
interval [CI], 0.58-0.67). Based on this adjusted OR, the number needed to
treat to prevent a postoperative death in the propensity matched cohort was
85 (95% CI, 77-98) and varied from 186 among patients at lowest risk to 30
among those with a revised cardiac risk index score of 4 or more. In a further
analysis using the entire study cohort and adjusting for quintile of propensity,
a significant effect of treatment persisted (adjusted OR, 0.71; 95% CI, 0.67-0.75).
Conclusions Treatment with lipid-lowering agents may reduce risk of death following
major noncardiac surgery. Clinical trials are required to confirm this observation.
Approximately 24 million surgical operations are performed each year
at hospitals across the United States1 and
it is estimated that as many as 1 million of these operations will be complicated
by a perioperative cardiovascular event.2 Perioperative
myocardial infarction is a dreaded occurrence that is associated with prolonged
hospital stay, substantial morbidity, and mortality rates as high as 25% to
40%.3,4 Among patients undergoing
major noncardiac surgery, the overall incidence of perioperative myocardial
infarction is 2% to 3%, and within high-risk populations, such as those patients
undergoing vascular surgery, rates can be as high as 34%.5,6 Although
clinical prediction instruments5,7- 9 have
improved the ability to detect patients at risk of perioperative cardiac events,
effective prevention strategies remain limited.10 The
emergence of perioperative β-blockade appears to be a major therapeutic
advance,11,12 yet rates of perioperative
cardiovascular complications among the highest-risk patients treated with β-blockers
can reach 16%.13,14 Additional
prevention modalities are therefore still needed to improve patient safety
and outcomes following surgery.
Lipid-lowering therapy is considered a cornerstone in the primary and
secondary prevention of cardiovascular disease. In addition to inhibiting
development of atherosclerotic plaques through reduction of serum cholesterol,
lipid-lowering medications are anti-inflammatory, can improve endothelial
function and coagulation, and produce a stabilizing effect on vulnerable plaques.15 These properties may be especially beneficial in
the perioperative period because the disruption of unstable plaques is believed
to be responsible for most cases of perioperative myocardial infarction.16 We therefore sought to determine whether the use
of lipid-lowering medication was associated with reduced mortality among patients
undergoing major noncardiac surgery.
A retrospective cohort study was performed using data from 329 hospitals
throughout the United States that participated in Perspective, a database
developed for quality and utilization benchmarking by Premier Incorporated,
Charlotte, NC. In addition to the data elements available in the standard
hospital discharge file, the Perspective database contains a date-stamped
log of all billed items, including medications, laboratory, diagnostic, and
therapeutic services, at the individual patient level.
Patients were included in our analysis if they were 18 years or older,
underwent major noncardiac surgery between January 1, 2000, and December 31,
2001, and survived beyond the second hospital day. Surgical procedures were
categorized by using a health information system, all patient refined-diagnosis
related groups (APR-DRG, version 15.0, 3M Corp, Minneapolis, Minn) and were
considered major if the median length of stay for patients in the diagnosis
related groups was 3 days or more.5 Patients
undergoing obstetrical procedures were excluded. Permission to perform the
study was obtained and informed consent was waived by the institutional review
board at Baystate Medical Center, Springfield, Mass.
For each case we noted type of surgery, whether the admission was elective,
urgent, or emergent, and at which hospital the operation was performed. In
addition to age, sex, race, and insurance status, we recorded the presence
of the following comorbidities: ischemic heart disease, congestive heart failure,
cerebrovascular disease, diabetes mellitus, renal insufficiency, chronic obstructive
pulmonary disease, hypertension, and hyperlipidemia. Comorbidities were established
using a combination of International Classification of Diseases,
Ninth Revision, Clinical Modification (ICD-9-CM)
secondary diagnoses. Patients with diabetes mellitus were identified on the
basis of either a secondary diagnosis of diabetes mellitus or treatment with
an oral hypoglycemic agent during the hospitalization. Perioperative administration
of β-blockers, angiotensin-converting enzyme inhibitors, angiotensin
receptor blockers, calcium channel blockers, antiplatelet agents, loop diuretics,
thiazide diuretics, antiarrhythmics, dopamine, and dobutamine were assessed
using pharmacy records. Prophylactic antibiotic administration and use of
pharmacologic and mechanical measures for the prevention of venous thromboembolism
were obtained similarly. In-hospital mortality, length of stay, and actual
costs were obtained from the Perspective discharge file. In addition to the
information related to the admission, we noted each hospital's bed size, teaching
status, geographic region, and whether it was located in an urban or rural
To examine treatment with lipid-lowering agents, we used pharmacy records
to identify whether a lipid-lowering medication was administered at any time
during the hospitalization, and if so, the date the medication was first administered.
Because lipid-lowering agents are frequently initiated for secondary prevention
soon after a postoperative myocardial infarction, we grouped patients in whom
the drugs were first administered on day 3 or later with those patients in
whom the medications were never used. Patients treated on the first or second
hospital day were categorized as having received either statin-based or nonstatin-based
therapy. Those patients who received combination therapy that included a statin
were analyzed in the statin group.
Based on an article by Lee et al,5 we
calculated a revised cardiac risk index score for each patient, assigning
1 point for each of the following risk factors: high-risk surgery, ischemic
heart disease, congestive heart failure, cerebrovascular disease, renal insufficiency,
and diabetes mellitus. The category labelled high-risk surgery included all
intrathoracic, intraperitoneal, and suprainguinal vascular procedures. Summary
statistics for the overall sample were constructed by using frequencies and
proportions for categorical data and mean, medians, and interquartile ranges
for continuous variables. We compared patients who received perioperative
treatment with lipid-lowering agents with those patients in whom the initiation
of lipid-lowering therapy was delayed beyond the second hospital day, or in
whom the drugs were never used. χ2 and z tests were used to assess the relationship between treatment with
lipid-lowering agents and in-hospital mortality and any potential confounders.
We created a nonparsimonious logistic regression model to derive a propensity
score for treatment with lipid-lowering therapy that included all patient
and hospital characteristics as well as selected interaction terms. Patients
missing any data elements were excluded from multivariable analysis. We matched
each patient in the treated group with up to 2 in the nontreated or late-treated
group based on propensity score. The matched cohort was then evaluated for
differences between treatment groups for each of the potential confounding
factors. Conditional logistic regression was used to assess the effect of
lipid-lowering therapy on in-hospital mortality, adjusting for covariates
unbalanced between groups (P<.01).17 In
addition, using the complete cohort, we constructed a logistic regression
model for in-hospital mortality using all available covariates, as well as
a model adjusting for quintile of propensity score and covariates unbalanced
across quintiles with P<.01. Interactions between
lipid-lowering therapy and unbalanced covariates were also evaluated for each
model and retained if P<.05. The Hosmer-Lemeshow
goodness-of-fit test and the area under the curve were used to assess model
To explore differences among specific lipid-lowering therapies, we compared
mortality among statin users and nonstatin users with the nontreated or late-treated
group, estimating the unadjusted odds ratios (ORs) with 95% confidence intervals
(CIs). All analyses were performed using SAS version 8.2 (SAS Institute Inc,
A total of 780 591 patients aged 18 years or older underwent major
noncardiac surgery during the study period and were included in the analysis.
The median age of the patients was 64 years, 55% were women, and 68% were
white (Table 1). Hypertension,
diabetes mellitus, and ischemic heart disease were the most commonly noted
comorbidities and 55% of the patients had a revised cardiac risk index score
of 1 or more. Orthopedic and abdominal operations accounted for 68% of cases,
30% of the procedures were labeled high-risk, and 50% of admissions were elective.
The median length of stay was 5 days (Table
2). A total of 23 100 study patients (2.96%) died during the
hospitalization. The majority of participating hospitals were in the south;
most were medium-sized, nonteaching, and located in urban areas (Table 3).
A total of 77 082 patients (9.9%) were treated with lipid-lowering
therapy on the first or second hospital day. Of these patients, 70 159
(91%) received a statin either alone or in combination with a nonstatin agent.
Mean rates of perioperative lipid-lowering therapy varied from 7.1% at hospitals
sized with 1 to 200 beds to 12.4% at hospitals with more than 800 beds. A
total of 31 448 patients (26%) with documented ischemic heart disease
received treatment in the early perioperative period.
Patients treated with lipid-lowering agents were older, were more often
white men, had a higher number of comorbidities, and a higher revised cardiac
risk index score than those patients who were not treated. Treated patients
were more likely to have undergone orthopedic or vascular procedures, to have
been admitted electively, and to list Medicare as their primary form of insurance.
They were more likely to be administered β-blockers and other cardiovascular
agents, and to receive measures to prevent venous thromboembolism. One thousand
six hundred forty patients (2.13%) who were treated with lipid-lowering agents
on the first or second day died compared with 21 460 patients (3.05%)
in whom treatment was delayed beyond the second day or was not administered
Seventeen patients were excluded from multivariable analysis because
of missing information. We successfully matched at least 1 nontreated patient
based on propensity score for 73 050 patients (95%) in the treated group
(76% with 2 matches and 19% with 1 match) (Table 4 and Table 5).
In this propensity matched cohort, 1595 patients (2.18%) treated with lipid-lowering
medications died compared with 4158 patients (3.15%) who did not receive therapy
or in whom treatment was initiated after the second day (P<.001). A number of covariates remained unbalanced in the matched
cohort. In a conditional logistic model using the matched cohort that adjusted
for unbalanced covariates, the perioperative administration of lipid-lowering
medications was associated with an adjusted OR of in-hospital mortality of
0.62 (95% CI, 0.58-0.67) (Table 6).
Among this group of patients, the number needed to treat was 85 (95% CI, 77-98)
and this varied with cardiac risk, ranging from 186 among the 34% of patients
with no risk factors to 30 among the 2% of patients with an index of 4 or
more (Table 7). No significant
interactions with lipid-lowering therapy were found for any of the covariates
included in the model.
Using the full study cohort and adjusting for quintile of propensity,
treatment with lipid-lowering therapy was associated with decreased mortality
among patients in all but the lowest quintile of propensity (Figure 1). In the overall sample, the quintile-adjusted OR of in-hospital
mortality was 0.71 (95% CI, 0.67-0.75) (Figure
2). With the exception of those patients least likely to receive
treatment with lipid-lowering therapy, treatment was associated with similar
mortality benefits across propensity quintiles. When adjusting for covariates
that remained unbalanced after propensity matching, the OR for mortality among
the overall sample of patients treated with lipid-lowering therapy was 0.67
(95% CI, 0.63-0.71).
In a standard multivariable logistic regression model (Table 6) that adjusted for age, sex, race, admission type, procedure
type, comorbidities, revised cardiac risk index score, other medication use,
insurance type, and hospital characteristics, the perioperative administration
of lipid-lowering medications was associated with an adjusted OR of in-hospital
mortality of 0.71 (95% CI, 0.67-0.75).
In an unadjusted analysis, the benefits of statins, administered either
alone or in combination with nonstatin agents, appeared more than when nonstatin
agents were prescribed alone. Among 70 159 statin users (91.0%), in-hospital
mortality was 2.09% and unadjusted OR was 0.68 (95% CI, 0.64-0.72) relative
to nontreated or late-treated patients. In comparison, among 6923 nonstatin
users (9.0%), mortality was 2.50% and unadjusted OR was 0.81 (95% CI, 0.70-0.95)
relative to nontreated or late-treated patients.
Finally, in a supplementary analysis, we considered all patients including
the 2378 patients who died during the first hospital days. As with the primary
analysis, use of lipid-lowering drugs was associated with a lower risk of
in-hospital mortality (adjusted OR, 0.69; 95% CI, 0.65-0.73).
In this large observational study, we found that the administration
of lipid-lowering agents during the early perioperative period was associated
with a 1% absolute reduction of hospital mortality and a 38% reduction in
the odds of in-hospital mortality among patients undergoing major noncardiac
surgery who were matched for likelihood of treatment. Our findings suggest
that lipid-lowering therapy may represent an important addition to the limited
armamentarium of the perioperative consultant.
How might lipid-lowering medications produce the observed association
in this study? It is well known that during long periods of administration
these agents inhibit the development and progression of atherosclerosis.18 Additionally, in time frames as short as 4 to 8 weeks,
statins have been shown to reduce platelet aggregation,19 improve
endothelial dependent vasodilation,20- 22 and
lower levels of C-reactive protein.23 These
local and systemic effects may reduce plaque formation and stabilize existing
plaques during periods of stress, such as are encountered around the time
of major surgery.
Because our data were limited to the inpatient setting, we are unable
to draw conclusions about how far in advance of surgery the medications might
need to be started, if in fact the observed association is causal. Unlike β-blockade,
the administration of lipid-lowering agents in the perioperative period is
not the current standard of care, and it would be unusual to initiate such
medications early during a hospital admission for a surgical procedure. The
use of lipid-lowering medications observed in this study therefore most likely
represented the continuation of a patient's outpatient regimen, but we did
not know how far in advance of surgery lipid-lowering therapy was started.
Only 1 study has previously addressed the topic of lipid-lowering therapy
in surgical patients. Poldermans et al24 performed
a case-control study among 2816 patients who underwent major vascular surgery
during a 10-year period at a single medical center in The Netherlands to examine
the association between statin use and perioperative mortality. They found
an adjusted OR for perioperative mortality among statin users of 0.22 (95%
CI, 0.10-0.47). Although similar, our findings extend this analysis to multiple
procedure types and drug classes using a multicenter cohort design.
Strengths of our study include its large size, national scope, and our
use of a previously validated risk stratification tool. We focused on an important
outcome, in-hospital mortality, and used both multivariable logistic regression
and propensity analysis to adjust for a wide variety of potential confounders,
including a large number of medications.
Our findings should nevertheless be interpreted with caution. First,
this was an observational study that relied on administrative data based on
physician documentation and coding to determine the presence of comorbidities
that were later used to compare and adjust for differences between groups.
It is possible that patients treated with lipid-lowering medications were
in fact healthier than their untreated counterparts even though they were
older and appeared to have a greater number of comorbidities. While acknowledging
this potential bias, our results persisted in a propensity matched cohort
that was balanced for many of these factors. Second, perioperative administration
of lipid-lowering medications may simply be a marker of high-quality perioperative
care in general, or more documentation of comorbidities. We observed more
use of perioperative β-blockers as well as measures to prevent venous
thromboembolism among those patients treated with lipid-lowering medications.
Although we adjusted for these known differences, it is possible that unmeasured
confounding related to physician practice remains unaccounted for in our analyses.
Third, we were unable to determine how often patients taking lipid-lowering
agents in the outpatient setting had their medications acutely discontinued
during the hospitalization and whether this may have had any adverse consequences
on their postoperative course. Moreover, it is possible that lipid-lowering
medications may have been withheld precisely because the patients were too
sick to resume treatment following surgery. To minimize confounding resulting
from the misclassification of patients who may have been too ill to resume
lipid-lowering therapy in the early postoperative period, we limited our study
to patients who survived the first 2 days and we included other orally administered
medications in our matching strategy. Fourth, we did not have access to laboratory
results, such as serum cholesterol or C-reactive protein levels, and thus
were unable to examine whether the benefits we observed were associated with
levels of either of these markers. Similarly, we did not have reliable information
concerning smoking status or left ventricular function. Lastly, our findings
are limited by the lack of information on rates of postoperative cardiovascular
complications, such as myocardial infarction or heart failure, which cannot
be reliably obtained from administrative databases.25
The use of lipid-lowering medications in the perioperative period is
associated with reduced mortality among patients undergoing major noncardiac
surgery. Clinical trials are required to confirm this observation and to determine
the optimal timing and duration of therapy.