CLINICAL RESEARCH STUDY

Smoking Cessation Reduces Postoperative Complications:

A Systematic Review and Meta-analysis
Edward Mills, PhD, MSc,a,b Oghenowede Eyawo, MPH,b Ian Lockhart, DLitt et Phil,c Steven Kelly, MSc,c
Ping Wu, MBBS, MSc,a Jon O. Ebbert, MD, MScd
a
Department of Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, Ontario, Canada; bFaculty of Health
Sciences, University of Ottawa, Ottawa, Canada; cOutcomes Research and Evidence-Based Medicine, Pfizer Ltd, Walton on the Hill,
United Kingdom; dMayo School of Medicine, Mayo Clinic, Rochester, Minn.

ABSTRACT

OBJECTIVE: We aimed to review randomized trials and observational evidence to establish the effect of
preoperative smoking cessation on postoperative complications and to determine if there is an optimal
cessation period before surgery.
METHODS: We conducted a systematic review of all randomized trials evaluating the effect of smoking
cessation on postoperative complications and all observational studies evaluating the risk of complications
among past smokers compared with current smokers. We searched independently, in duplicate, 10
electronic databases and the bibliographies of relevant reviews. We conducted a meta-analysis of random-
ized trials using a random effects model and performed a meta-regression to examine the impact of time,
in weeks, on the magnitude of effect. For observational studies, we pooled proportions of past smokers in
comparison with current smokers.
RESULTS: We included 6 randomized trials and 15 observational studies. We pooled the 6 randomized trials
and demonstrated a relative risk reduction of 41% (95% confidence interval [CI], 15-59, P ⫽ .01) for
prevention of postoperative complications. We found that each week of cessation increases the magnitude
of effect by 19%. Trials of at least 4 weeks’ smoking cessation had a significantly larger treatment effect
than shorter trials (P ⫽ .04). Observational studies demonstrated important effects of smoking cessation
on decreasing total complications (relative risk [RR] 0.76, 95% CI, 0.69-0.84, P ⬍ .0001, I2 ⫽ 15%). This
also was observed for reduced wound healing complications (RR 0.73, 95% CI, 0.61-0.87, P ⫽ .0006, I2 ⫽
0%) and pulmonary complications (RR 0.81, 95% CI, 0.70-0.93, P ⫽ .003, I2 ⫽ 7%). Observational
studies examining duration of cessation demonstrated that longer periods of cessation, compared with
shorter periods, had an average reduction in total complications of 20% (RR 0.80, 95% CI, 3-33, P ⫽ .02,
I2 ⫽ 68%).
CONCLUSION: Longer periods of smoking cessation decrease the incidence of postoperative complications.
© 2011 Elsevier Inc. All rights reserved. • The American Journal of Medicine (2011) 124, 144-154

KEYWORDS: Meta-analysis; Perioperative; Smoking cessation; Systematic review

Tobacco smoking remains the leading cause of preventable the prevalence of smoking, considerable efforts have been
death in the world.1 Smoking cessation is associated with directed toward developing interventions to assist smokers
important benefits at individual and societal levels. Given in quitting.2 The role of smoking cessation benefits within

Funding: Development of this manuscript was sponsored by Pfizer Lockhart and Steven Kelly are employees of Pfizer Ltd. Oghenowede has
Ltd, Walton Oaks, Walton-On-The-Hill, Surrey, KT20 7NS, United King- consulted to Pfizer Ltd in the past. Jon Ebbert has no conflicts of interest.
dom. Edward Mills, Oghenowede Eyawo, and Ping Wu were paid con- Authorship: All authors had access to the data and played a role in
sultants to Pfizer in connection with the development of this manuscript. writing this manuscript.
Jon Ebbert received no compensation. Edward Mills is supported by a Requests for reprints should be addressed to Edward Mills, PhD, MSc,
Canada Research Chair. Department of Clinical Epidemiology and Biostatistics, McMaster Univer-
Conflicts of Interest: Edward Mills has consulted to Pfizer Ltd and Merck sity, 1200 Main Street West, Hamilton, Ontario, Canada L8N 3Z5.
Shire Dohme in the past. Ping Wu has consulted to Pfizer Ltd in the past. Ian E-mail address: Edward.mills@uottawa.ca

0002-9343/$ -see front matter © 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.amjmed.2010.09.013
Mills et al Smoking Cessation Reduces Perioperative Complications 145

specific disease classifications is becoming increasingly Search Strategy

clear, with, for example, decreased morbidity and mortality In consultation with a medical librarian, we established a
among recent former smokers compared with continuing search strategy. We searched independently, in duplicate,
smokers with cardiovascular diseases.3 the following 10 databases (from inception to September
There has been a recent emerging body of evidence 2009): MEDLINE, EMBASE, Cochrane CENTRAL,
showing the benefit of preopera- AMED, CINAHL, TOXNET, De-
tive and long-term postoperative velopment and Reproductive Tox-
smoking cessation.4,5 In the icology, Hazardous Substances
CLINICAL SIGNIFICANCE
United States, an estimated 53.3 Databank, PsycINFO, and Web of
million surgical and nonsurgical ● Smoking cessation before any type of Science. Given that observational
procedures are performed annua- surgery reduced risks of complications, studies are poorly indexed in
lly.6 Approximately 8 to 10 million including wound healing and pulmonary many databases, we also searched
procedures requiring surgery and databases that include the full text
complications.
anesthesia are performed on ciga- of journals (ScienceDirect, and In-
rette smokers.7,8 Smokers may be ● Longer periods of smoking cessation be- genta, including articles in full
at a greater risk of postoperative fore surgery demonstrated a signifi- text from ⬃1700 journals since
complications than nonsmokers, cantly larger reduction in complica- 1993).12 In addition, we searched
including delayed wound healing, tions. the bibliographies of published
pulmonary complications, and relevant reviews and health tech-
mortality.9 Preoperative smoking ● Meta-analysis of randomized trials
nology assessments.4,13-15 Searches
cessation seems to offer important showed a relative risk reduction of 41% were not limited by language, sex,
benefits in reducing complica- (95% confidence interval, 15-59, P ⫽ or age.
tions, although the duration of ces- .01) for prevention of postoperative
sation required and the success of complications. This was supported with Study Selection
different interventions are not un- observational data examining total Two investigators working inde-
derstood.10 Clinical trials have complications, wound healing, and pul- pendently, in duplicate, scanned
evaluated smoking-cessation in- monary complications. all abstracts and obtained the full-
terventions at varying times be- text reports of records indicating
fore surgery and found clinically that the study is an original re-
meaningful reductions in compli- search article evaluating the im-
1-3 3
cations. Trials also have evaluated both longer- and pact of smoking cessation on the outcomes of interest. After
4,11
shorter-term cessation, including as late as the day of obtaining full reports of candidate studies (in full peer-
surgery.4 We aimed to determine the strength of evidence
reviewed publication or media article), the same reviewers
supporting the role of smoking cessation and the duration of
independently assessed eligibility from full-text papers.
cessation required in preventing postoperative complica-
tions by undertaking a systematic review and meta-analysis
of observational studies and randomized trials.
Data Collection
Two reviewers conducted data extraction independently us-
MATERIALS AND METHODS ing a standardized pre-piloted form (available from the
Eligibility Criteria authors on request). Reviewers collected information about
We included both observational studies and randomized the smoking intervention tested, the population studied (age,
trials that evaluated the incidence of postoperative compli- sex, underlying conditions, and types of surgery), the treat-
cations among populations who achieved smoking absti- ment dosages and dosing schedules, the relative effective-
nence at a defined time point before surgery. We included ness of treatments begun at any time points before surgery,
any classification of cessation and any surgical populations. the specific measurement of abstinence (sustained or point-
Studies had to report the proportion of clinical complica- prevalence), and the biochemical confirmation of smoking
tions of past smokers and current smokers. status. Study evaluation assessed risk of bias per a modified
Cochrane risk-of-bias tool and included general method-
Study End Points ological quality features specific to randomized trials, in-
Our primary study outcome was postoperative complica- cluding allocation concealment, sequence generation, blind-
tions defined as total complications. Secondary outcomes ing status, intention-to-treat, sources of funding, and
included any wound healing complications, pulmonary or appropriate descriptions of loss to follow-up.16 In addition,
respiratory complications, all-cause mortality, and all-cause for nonrandomized evaluations, we applied the Newcastle-
length of hospital stay. When a primary complication was Ottawa Scale to evaluate the reporting rigor of observational
distinct and total complications were not reported (eg, studies.17 The Newcastle-Ottawa Scale is a scoring check-
wound healing), we did not combine them as a composite. list addressing issues of design and implementation of a
146 The American Journal of Medicine, Vol 124, No 2, February 2011

cohort or case-control, including issues of selection of par- RESULTS

ticipants, comparability of cases and controls, exposure, and
outcomes. We entered the data into an electronic database, Results of the Literature Search
such that duplicate entries existed for each study; when the We reviewed 847 abstracts that met our search criteria.
2 entries did not match, differences were resolved through During the initial title and abstract selection process, we
discussion and third-party arbitration. excluded 772 abstracts for reasons identified in Figure 1.
The search was sensitive and not initially specific. Of those
excluded, 139 abstracts were either non-human studies or
non-English abstracts, 399 abstracts did not address the
Data Analysis review topic, and a further 234 were review articles. We
To assess inter-rater reliability on inclusion of articles, we then obtained the full-text papers of the remaining 75 ab-
calculated the Phi statistic, which provides a measure of stracts, and these proceeded to the full-text review stage.
After rigorous screening of the 75 full-text articles, 30 of
interobserver agreement independent of chance.18 We cal-
them satisfied the study inclusion criteria and were included
culated the relative risk (RR) and appropriate 95% confi-
in the initial review (Phi ⬎ 0.8). We then excluded 9
dence intervals (CIs) of outcomes according to the number
observational studies because of errors in the reporting of
of complications reported in the original studies or substud-
relevant outcome data per arm27 and failing to report dif-
ies. Relative risks are typically a more clinically understood ferences between past and current smokers on the outcomes
effect measure than odds ratios but cannot be used with of interest to the review.28-35 Our review thus includes 6
some observational designs (eg, case-control studies). In randomized trials4,11,36-39 and 15 observational studies.40-54
circumstances of zero outcome events in one arm of a trial,
we planned to apply the Haldane approach and add 0.5 to
each arm, as suggested by Sheehe.19 We applied the mod-
Characteristics of Included Studies
The majority of the studies were conducted in the United
ified intent-to-treat principal, whereby a patient must re-
States.40-42,45,48,50-53 Four studies were performed in Den-
ceive at least 1 dose of the study intervention, in this case,
mark,49,55-57 2 studies were performed in Australia,36,46 3
surgery. We first pooled all randomized trials. We used the
studies were performed in Japan,43,47,54 1 study was per-
DerSimonian–Laird random effects method to pool stud-
formed in Israel,44 and 1 study was performed in Sweden.58
ies,20 which recognizes and anchors studies as a sample of Of the 21 included studies, 6 were randomized tri-
all potential studies, and incorporates an additional be- als,4,36,55-58 and the remaining were observational stud-
tween-study component to the estimate of variability.21 We ies.40-54 Of the observational studies, 10 were prospective
also calculated the I2 statistic for each analysis as a measure comparisons40-42,44-47,49,50,54 and 5 were retrospective anal-
of the proportion of the overall variation that is attributable yses of patients with complications.43,48,51-53 The included
to between-study heterogeneity.22 We performed a sensitiv- studies involved a heterogeneous array of surgical proce-
ity analysis to examine shorter-term (⬍4 weeks) and longer- dures, including lung,40,45,47,51 orthopedic/bone and abdom-
term (⬎4 weeks) effects, and to examine intensive versus inal,44,55 breast,42,48 head/neck,43 heart,52,53 and various
passive interventions, as defined by Thomsen et al.4 We elective surgeries,41,46,49,50,54,56-58 and reported on postsur-
performed a meta-regression to examine the impact of time, gical complications among study participants. Time of
in weeks, on log RR effect size. We display forest plots for study follow-up ranged from 1 day to 57 months.
our primary analysis, showing individual study effect mea-
sures with 95% CIs, and the overall DerSimonian–Laird Methodological Quality of Included Studies
pooled estimate and the regression plot with circle size Table 1 shows the risk of bias characteristics of the ran-
conferring study sample size. For observational studies re- domized trials. In general, studies were well reported with a
porting total complications, we pooled the proportions of low risk of bias. Table 2 shows the Newcastle-Ottawa Scale
past smokers and current smokers by stabilizing the vari- risk of bias for observational studies. One of the observa-
ances of the raw proportions (r/n) using a Freeman-Tukey tional studies did not clearly define the length of time that
type arcsine square root transformation and applying a ran- smokers/past smokers ceased smoking before the surgery
dom effects model.23,24 Although several methods of pool- was performed.49 Seven studies did not specify the fol-
ing proportions exist,25 the Freeman–Tukey method works low-up period.41-43,45,47,48,50
well with both fixed and random effects meta-analysis and
truncates at zero (because it is impossible to have less than Meta-analysis of Effectiveness
zero events).26 Where appropriate, we pooled using random We pooled 6 randomized trials evaluating varied smoking
effects RR. All P values are exact to ⬍ .001. We considered interventions on total complications (Tables 3 and
P ⬍ .05 as significant. Analyses were conducted using 4).4,11,36-39 We found a pooled RR of 0.59 (95% CI, 0.41-
StatsDirect (version 2.5.2, www.statsdirect.com) and Com- 0.85, P ⫽ .01, I2 ⫽ 14%; Figure 2). Our meta-regression
prehensive Meta-analysis (version 2, www.meta-analysis. analysis found that each week of cessation resulted in a
com). larger effect size (B coefficient ⫺0.191, 95% CI, ⫺0.368 to
Mills et al Smoking Cessation Reduces Perioperative Complications 147

Figure 1 Flow diagram of study selection process. RCT ⫽ randomized clinical

trial.

⫺0.014, P ⫽ .03; Figure 3). We performed a sensitivity tional studies40-44,46-49,52-54provided sufficient details of
analysis evaluating short-term studies (⬍4 weeks)4,11 (RR general complications occurring between past smokers and
0.92, 0.53-1.60, P ⫽ .78) versus longer cessation (RR 0.45, current smokers. We were able to pool data from 12 of these
95% CI, 0.30-0.68, P ⬍ .001, I2 ⫽ 0%) (P value for studies.40-44,46-48,52-54 (One study presented data only as
difference .041). We also examined the use of inten- adjusted odds and was excluded.)49 There was a statistically
sive11,37,39 (RR 0.55, 95% CI, 0.31-0.98, P ⫽ .04, I2 ⫽ significant reduction in the risk of total complications in former
61%) versus less intensive (RR 0.78, 95% CI, 0.34-1.80, smokers compared with current smokers, with an average 22%
P ⫽ .56, I2 ⫽ 0%) interventions (P value for difference .49). (95% CI, 13-34) of former smokers experiencing an event
compared with 32% (95% CI, 19-47) of current smokers (RR
Observational Studies 0.76, 95% CI, 0.69-0.84, P ⬍ .0001, I2 ⫽ 15%).
Risk of Total Postoperative Complications. Tables 5 and
6 (available online) show the characteristics and out- Risk of Pulmonary Complications. Seven studies reported
comes of the 15 observational studies. Thirteen observa- on the occurrence of pulmonary complications.41,45-47,52-54
148 The American Journal of Medicine, Vol 124, No 2, February 2011

Table 1 Risk of Bias Assessment: Randomized Clinical Trials

Report Method Report Efforts Report Blinding Refusal Rate Intent-to- Free of Lost to
First Author, of Sequence to Conceal Status of Groups of Eligible Treat Selective Follow-Up Study Funding
Year Generation Allocation or Researchers Participants Analysis Reporting ⬎ 20% Source(s)
Lindstrom ⫹ ⫹ ⫹ 39% ⫹ ⫹ ⫺ Public, industry
2008 provided drug
Moller 2002 ⫹ ⫺ ⫺ 28% ⫹ ⫹ ⫺ Public, industry
provided drug
Myles 2004 ⫹ ⫹ ⫹ ⫺ ⫹ ⫹ ⫹ Hospital, industry
provided drug
Sorensen ⫺ ⫹ ⫹ 25% ⫹ ⫹ ⫺ Public, industry
2003 provided drug
Sorensen ⫺ ⫹ ⫹ 9% ⫹ ⫹ ⫺ Public, industry
2007
Warner 2005 ⫹ _ ⫹ 4% ⫹ ⫹ ⫺ Public, industry
provided drug
⫹ ⫽ clear; ⫺ ⫽ unclear or negative.

Data were available to test for differences from all trials. We ⬍ 4 weeks cessation. The removal of this study reduced the
found differences between past (pooled incidence 15%, I2 value to 0%, but the effect size was no longer statistically
95% CI, 6-28) and current (20%, 95% CI, 8-26%) smokers, significant (P ⫽ .52).
which indicated a statistically significant decrease in pul-
monary complications (RR 0.81, 95% CI, 0.70-0.93, P ⫽ DISCUSSION
0.003, I2 ⫽ 7%) for the former. There was no statistically Our review demonstrates a clear benefit of smoking cessa-
significant difference between early and late quitters on this tion to prevent postoperative complications compared with
outcome (RR 0.88, 95% CI, 0.28-2.71, P ⫽ 0.81, I2 ⫽ continued smoking. We found that randomized trials that
94%), possibly because of low power (9%).51-53 successfully introduced a smoking cessation intervention
and attained abstinence had significantly decreased rates of
Risk of Wound-Healing Complications. Five studies pro- complications. This effect was magnified with longer dura-
vided data on wound healing.42,43,46,48,50 We found a sta- tions of cessation. This finding is in agreement with the
tistically significant reduction in wound healing complica- evidence from observational studies that smoking cessation
tions associated with former smokers compared with current reduces total postoperative complications, pulmonary com-
smokers (RR 0.73, 95% CI, 0.61-0.87, P ⫽ .0006, I2 ⫽ 0%). plications, and complications of wound healing. From our
analyses of both randomized trials and observational stud-
Length of Hospital Stay. Two studies reported on the ies, longer cessation periods provide greater reductions in
average length of hospital stay.40,50 One study reported the clinical complications.
mean duration of hospital stay as 8 days in past smokers and There are both strengths and limitations to consider in
9 days in current smokers.40 The other study found identical our analysis. Strengths include our extensive searching, data
duration of stay after nephrectomy.50 abstraction in duplicate, and inclusion of both randomized
and observational evidence that provides similar inferences.
Mortality. Only 2 studies reported on mortality.41,45 Both Our regression analysis shows that the length of time from
studies had low event rates and found no difference between smoking cessation is directly associated with the magnitude
past and current smokers (RR 1.00, 95% CI, 0.64-1.55, P ⫽ of subsequent complications. We found a larger treatment
.98). effect in randomized trials than in the observational studies.
This may be due to the smaller sample size of the pooled
Duration of Cessation Period. Seven studies provided randomized trials or may be a true therapeutic effect. The
adequate details on early versus late quitting and total com- populations examined in the randomized trials were more
plications.40,43,45,47,51-53 We found that shorter-term (⬍4 homogenous than those in the cohort studies. Limitations of
weeks) cessation compared with longer-term cessation (⬎4 our analysis are predominantly related to the heterogeneous
weeks) resulted in a relative risk decrease of 20% (RR 0.80, reporting of outcomes, inconsistent definitions of past
95% CI, 3-33, P ⫽ .02, I2 ⫽ 68%) in total complication smoking status, and differences in study designs across the
rates. The residual heterogeneity found in this estimate is observational studies. Although this has necessarily led to
from Warner et al,53 who reported only on early quitters some study exclusions from the pooled analyses because of
(⬎8 weeks cessation) versus late quitters (ⱕ8 weeks ces- lack of relevant data on the key outcomes of interest, our
sation) before surgery instead of ⬎ 4 weeks compared with observational study analyses indicated a significant decrease
 Mills et al
Table 2 Risk of Bias Tables: Observational Studiesⴱ,†
Adequate Case/ Potential for Bias Presence of Sufficient Follow- Adequacy of
First Author, Control/Group in Case/Group Controls/Comparison Comparability among Ascertainment Cessation Non-Response Assessment of Up for Outcome Follow-Up
Year Definition? Representation Group? Groups of Exposure Validation Method Rates Outcome Occurrence? (Rates)

Smoking Cessation Reduces Perioperative Complications

Barrera 2005 Yes Yes Yes PQ were significantly older Self-report Self-report via Not applicable Independent Yes ⬎80%
than NS. questionnaire
Bluman 1998 Yes Yes Yes CS were significantly Self-report Self-report via Not applicable Independent (medical Yes ⬎80%
younger than PS or NS. questionnaire records)
Chang 2000 Yes Yes Yes No significant difference Not specified Medical records Not applicable Independent (medical Unclear ⬎80%
among groups. records)
Goodwin 2005 Yes Yes Yes No major significant Self-report Self-report Not applicable Independent (medical Yes ⬎80%
demographic records)
differences except for
the higher proportion
of NS with stage 0
disease.
Kuri 2005 Yes Yes Yes NS were more likely to be Self-report Comparison of 3 Not applicable Independent (medical Unclear Unclear
female, and LQ were self-reports records),
significantly younger self-reports
than EQ.
Levin 2004 Yes Yes Yes; groups were No significant difference Medical Medical records Not applicable Independent (medical Yes ⬎80%
undefined by among groups in terms records records)
their percentages of age and gender.
relative to sample
size
Mason 2009 Yes Yes Yes NS were more likely to be Medical Medical records Not applicable Independent (medical Unclear ⬎80%
female. records records)
Myles 2002 Yes Yes Yes CS were younger than PS Self-report Exhaled air CO Not applicable Independent (medical No ⬎80%
or NS. records), self-
reports
Nakagawa Yes Yes Yes Similar demographics Self-report Medical records, Not applicable Independent (medical Unclear ⬎80%
2001 among groups. Self-report records), self-
report
Padubidri Yes Yes Yes TRAM flap surgery was Medical Medical records Not applicable Independent (medical Unclear ⬎80%
2001 used more in NS, and records records)
tissue expander method
was used more often in
smokers.
Sorensen Yes Yes Yes Baseline demographics Self-report Medical records, Not applicable Independent (medical Yes ⬎80%
2005 among groups were validated by a records)
unreported. second self-
report during
follow-up
Taber 2009 Yes Yes Yes CS had 1 y less of Self-report Self-report Not applicable Independent (medical Unclear ⬎80%
education and more records)
alcohol use than other
groups.
Vaporciyan Yes Yes Yes Groups differed Medical Medical records Not applicable Independent (medical Yes ⬎80%
2002 significantly in age, records records)
sex, and disease type.

149
150 The American Journal of Medicine, Vol 124, No 2, February 2011

in complications for total, pulmonary, and wound healing

Adequacy of
Follow-Up
complications. However, other analyses, such as those ex-

(Rates)

⬎80%
⬎80%
⬎80%
amining hospital stay and mortality, may be affected by low
power. It is possible that if more trials reported specific
outcomes, the results would be more precise.25
Sufficient Follow-
Up for Outcome

Ours is not the first systematic review to examine smok-

Occurrence?

ing cessation with postsurgical outcomes. A meta-analysis

CO ⫽ carbon monoxide; CS ⫽ current smoker; EQ ⫽ early quitters; LQ ⫽ late quitters; NS ⫽ nonsmoker; PS ⫽ past smoker; TRAM ⫽ transverse rectus abdominis myocutaneous.
by Thomsen et al4 examined complication rates in 6 ran-
Yes

Yes
Yes

domized trials and came to a similar effect estimate as that

reported in our review. They also examined smoking ces-
Independent (medical

Independent (medical

sation and demonstrated sustained cessation using a narra-

Assessment of

tive review. Another narrative review by Theadom and

Independent
records)
records)

Cropley14 included 12 observational studies examining

Outcome

postsurgical complications using inclusion criteria similar to

ours. They did not pool studies but concluded that longer
periods of cessation seem to be associated with decreased
Not applicable
Not applicable

Not applicable
Non-Response

rates of complications. Our analysis includes 3 additional

studies and pooled total complications to demonstrate that
Rates

smoking cessation in observational studies is associated

with a statistically significant reduction in total postsurgical
Validation Method

complications, in accordance with the pooled randomized

Urinary cotinine
Medical records

trials evidence, and significant reductions in pulmonary and

analysis

Self-report
Cessation

wound-healing complications. Furthermore, our analysis

demonstrated that increasing cessation periods results in a
clinically important and statistically significant reduction in
Ascertainment

complications. The 2004 Surgeon General’s Report on

Self-reports,
of Exposure

Self-report,

Self-report
medical

medical
records

records

smoking concluded that there was sufficient evidence to

infer a causal relationship between smoking and increased
risks for adverse surgical outcomes related to wound heal-
ing and respiratory complications. However, the report re-
CS were younger than PS

lied on observational studies, did not examine the magni-

Baseline demographics

Baseline demographics
among groups were

among groups were

Comparability among

tude of risks or time periods of increased risk, and reported

findings as a narrative review.59
unreported.

unreported.

A proof of concept randomized study by Sorensen et al60

or NS.
Groups

Note: The Newcastle-Ottawa Scale was used to assess observational studies.

deliberately randomized smokers to receive an incision or

not after several weeks of smoking cessation therapy. In an
effort to determine the duration of cessation required, study
Controls/Comparison

subjects were evaluated at 1, 4, 8, and 12 weeks post-

*Definition requires classification according to studied groups.

cessation, with incisions administered at each evaluation

Presence of

period. The largest and most sustained treatment effects

Group?

were observed with 4 weeks cessation.

Yes
Yes

Yes

The trials included in our randomized trials meta-analy-

sis provided a variety of interventions that we classified as
Potential for Bias

intensive or less intensive, ranging from the use of bupro-

Representation
in Case/Group

pion therapy and intensive counseling (intensive) to nicotine

replacement therapy patch and brief counseling services
Yes
Yes

Yes

(less intensive), as suggested by Thomsen et al.4 More

intensive interventions for cessation of smoking before sur-
Adequate Case/
Control/Group

gery are therefore relevant for postoperative risk reduction.

Definition?

Pharmacotherapies available to surgery populations include

Continued

nicotine replacement therapy, bupropion, and varenicline. A

Yes
Yes

Yes

recent meta-analysis we conducted examined the short-term

†Versus NS.

(⬃4 weeks) cessation rates of these pharmacotherapies. Our

First Author,

Warner 1984

Warner 1989

study demonstrated consistent benefits from each interven-

Table 2

Yamashita

tion with varenicline, bupropion, and nicotine replacement

2004
Year

therapy offering a cascade of relative benefit. The benefits

 Mills et al
Table 3 Characteristics of Included Randomized Trials
Type Cessation Cessation Follow-up
First Author, of Sample Male Period Defined Validation Pre-surgery Smoking Cessation Intervention Period

Smoking Cessation Reduces Perioperative Complications

Year Study Types of Surgery Intervention and Control Interventions Size (%) Mean Age (y) Clearly? Method Cessation Period (and Period Pre-surgery) Postsurgery

Lindstrom RCT Hernia repair, IG ⫽ 47%; smokers who received an intervention 102 53 55 Yes Self-report IG: smoke-free from 4 wk Yes; individual 30 d
2008 laparoscopic to keep them smoke-free from 4 wk and CO preoperatively to 4 wk counseling,
cholecystectomy, preoperatively to 4 wk postoperatively postoperatively telephone hotline,
hip/knee prosthesis CG ⫽ 53%; smokers who did not receive any In per-protocol analysis and free nicotine
such intervention for smoking cessation pre- groups: substitution (nicotine
surgery group 1, ⱖ3 wk replacement therapy)
In per protocol analysis, smokers were preoperatively and 4 for 4 wk pre-surgery
grouped as: wk postoperatively;
1. Smokers who quit ⱖ 3 wk preoperatively and group 2, 1-2 wk
4 wk postoperatively preoperatively and 4
2. Smokers who quit 1-2 wk preoperatively and 4 wk postoperatively;
wk postoperatively group 3, cessation period
3. Smokers who continued smoking or only undefined
reduced smoking
Moller 2002 RCT Hip or knee IG ⫽ 52%; defined as smokers who received an 108 42.6 65 Yes Exhaled air CO 6-8 wk pre-surgery and Yes; information and 4 wk
replacement intervention to keep them smoke-free or help 10 d postsurgery weekly counseling for
reduce smoking by 50% from 6-8 wk 6-8 wk pre-surgery
preoperatively
CG ⫽ 48%; defined as smokers who did not
receive any such intervention for smoking
cessation pre-surgery
Sorensen RCT Colorectal surgery IG: 47%; defined as smokers who received an 57 65 65.5 (median) Yes Self-report, IG: advised to be smoke- Yes; counseling and 30 d
2003 intervention to keep them smoke-free or exhaled air free or reduced nicotine replacement
reduce their smoking 2-3 wk pre-surgery and CO; salivary smoking from 2-3 wk therapy
until sutures were removed cotinine pre-surgery
CG: 53%; defined as smokers who did not receive
any such intervention and were asked to
maintain their daily smoking habits
Sorensen RCT Elective open incisional IG: 70%; defined as smokers who received an 213 84.5 54.6 (median) Yes Self-report, IG: advised to stop Yes; 3 levels: advice 3 mo
2007 or inguinal intervention by means of an advice and exhaled air smoking at least 1 mo only, advice and
herniotomy reminder to stop smoking at least 1 mo pre- CO; salivary pre-surgery telephone reminder,
surgery and until sutures were removed cotinine advice and reminder
CG: 30%; defined as smokers who did not receive and outpatient
any such intervention nicotine substitution
demonstration
Myles 2004 RCT General, orthopedic, IG: Bupropion 7 wk before expected surgery (150 47 66 45 Yes Exhaled air CO Both groups 7 wk Bupropion (150 mg bid) 6 mo
urologic, ear, nose, mg bid) ⫹ 1 telephone counseling. pre-surgery
throat, CG: placebo ⫹ 1 telephone counseling
faciomaxillary
Warner 2005 RCT Orthopaedic, intra- IG: nicotine patch applied on day of surgery 121 48 47 Yes Exhaled air CO Both groups, day of Nicotine patch applied 1 mo
abdominal, spinal, CG: placebo patch surgery on day of surgery
genitourinary,
otorhinolaryngologic,
gynecologic, other

RCT ⫽ randomized clinical trial; IG ⫽ intervention group; CG ⫽ control group; CO ⫽ carbon monoxide.

151
152 The American Journal of Medicine, Vol 124, No 2, February 2011

Table 4 Effects of Randomized Trials on Perioperative Complications

Relative Risk (95% CI)
First Author, Intensity of Preoperative for Perioperative
Year Smoking Intervention Definition of Perioperative Complication Complications
Lindstrom 2008 Intensive Events causing additional medical or surgical treatment 0.51 (0.27-0.97)
or investigation, prolonged hospital stay, unscheduled 0.49 (0.20-1.16)
postoperative checkups within 30 d
Any wound complication
Moller 2002 Intensive Death or postoperative morbidity requiring treatment 0.34 (0.19-0.64)
within 30 d 0.17 (0.05-0.56)
Wound healing complications
Myles 2004 Less intensive Postoperative wound infections 0.82 (0.06-11.33)
Sorensen 2003 Intensive Adverse events within 30 d requiring medical or surgical 0.94 (0.51-1.73)
intervention
Sorensen 2007 Less intense Postoperative wound infection with swollen or infected 0.71 (0.21-2.41)
wound or medical intervention required at suture
removal
Warner 2005 Less intensive Serious postoperative adverse events 0.86 (0.24-3.03)
CI ⫽ confidence interval.

of these short-term cessation interventions also seem to smoking-cessation intervention before surgery, and assum-
extend into longer periods of cessation (⬍6 months).2,26 ing a 25% cessation rate, this could result in 2 million (95%
CI, 1,769,600-2,248,800) complications avoided, resulting
CONCLUSIONS in large savings for both patients and health services. The
In the United States, approximately 8 to 10 million proce- review finding that each additional week of smoking cessa-
dures requiring surgery and anesthesia are performed on tion has a significant impact on the reduction of postoper-
cigarette smokers.8 To interpret the possible impact of ative complications highlights the importance of designing
smoking cessation on population-wide complications, we an appropriate secondary care smoking-cessation service. A
conservatively estimate that, if all patients were offered a service designed around early assessment of the smoking

Figure 2 Forst plot of randomized clinical trials on total complications.

Mills et al Smoking Cessation Reduces Perioperative Complications 153

Figure 3 Meta-regression plot, effect of time of cessation on complications.

status of surgery patients and rapid referral to a smoking- 14. Theadom A, Cropley M. Effects of preoperative smoking cessation on
cessation program could maximize the cessation period be- the incidence and risk of intraoperative and postoperative complica-
tions in adult smokers: a systematic review. Tob Control. 2006;15:
fore surgery, resulting in greater reductions in postoperative
352-358.
complications in the secondary care setting. 15. Cropley M, Theadom A, Pravettoni G, Webb G. The effectiveness of
smoking cessation interventions prior to surgery: a systematic review.
References Nicotine Tob Res. 2008;10:407-412.
1. Peto R, Lopez AD, Boreham J, Thun M, Heath C Jr, Doll R. Mortality 16. Lundh A, Gotzsche PC. Recommendations by Cochrane Review
from smoking worldwide. Br Med Bull. 1996;52:12-21. Groups for assessment of the risk of bias in studies. BMC Med Res
2. Wu P, Wilson K, Dimoulas P, Mills EJ. Effectiveness of smoking Methodol. 2008;8:22.
cessation therapies: a systematic review and meta-analysis. BMC Pub- 17. Wells GA, Shea B, O’Connell D, et al. The Newcastle-Ottawa
lic Health. 2006;6:300. Scale (NOS) for assessing the quality of nonrandomised studies in
3. Wilson K, Gibson N, Willan A, Cook D. Effect of smoking cessation meta-analyses. Available at: http://www.ohri.ca/programs/clinical_
on mortality after myocardial infarction: meta-analysis of cohort stud- epidemiology/oxford.htm. Accessed December 15, 2009.
ies. Arch Intern Med. 2000;160:939-944. 18. Meade MO, Guyatt GH, Cook RJ, et al. Agreement between alterna-
4. Thomsen T, Tonnesen H, Moller AM. Effect of preoperative smoking tive classifications of acute respiratory distress syndrome. Am J Respir
cessation interventions on postoperative complications and smoking Crit Care Med. 2001;163:490-493.
cessation. Br J Surg. 2009;96:451-461. 19. Sheehe PR. Combination of log relative risk in retrospective studies of
5. Moller A, Tonnesen H. Risk reduction: perioperative smoking inter- disease. Am J Public Health Nations Health. 1966;56:1745-1750.
vention. Best Pract Res Clin Anaesthesiol. 2006;20:237-248. 20. Fleiss JL. The statistical basis of meta-analysis. Stat Methods Med Res.
6. Warner DO, Patten CA, Ames SC, Offord KP, Schroeder DR. Effect 1993;2:121-145.
of nicotine replacement therapy on stress and smoking behavior in 21. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin
surgical patients. Anesthesiology. 2005;102:1138-1146. Trials. 1986;7:177-188.
7. Graham-Garcia J, Heath J. Urgent smoking cessation interventions: 22. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-
enhancing the health status of CABG patients. Crit Care Update.
analysis. Stat Med. 2002;21:1539-1558.
2000;Suppl:19-23.
23. Paoletti P, Fornai E, Maggiorelli F, et al. Importance of baseline
8. Abidi NA, Dhawan S, Gruen GS, Vogt MT, Conti SF. Wound-healing
cotinine plasma values in smoking cessation: results from a double-
risk factors after open reduction and internal fixation of calcaneal
blind study with nicotine patch. Eur Respir J. 1996;9:643-651.
fractures. Foot Ankle Int. 1998;19:856-861.
24. Campbell IA, Prescott RJ, Tjeder-Burton SM. Smoking cessation in
9. Hall MJ, Lawrence L. Ambulatory surgery in the United States, 1996.
hospital patients given repeated advice plus nicotine or placebo chew-
Adv Data. 1998;300:1-16.
10. Russell MA, Stapleton JA, Feyerabend C, et al. Targeting heavy ing gum. Respir Med. 1991;85:155-157.
smokers in general practice: randomised controlled trial of transdermal 25. Wetterslev J, Thorlund K, Brok J, Gluud C. Trial sequential analysis
nicotine patches. BMJ. 1993;306:1308-1312. may establish when firm evidence is reached in cumulative meta-
11. Sorensen LT, Jorgensen T. Short-term pre-operative smoking cessation analysis. J Clin Epidemiol. 2008;61:64-75.
intervention does not affect postoperative complications in colorectal 26. Mills EJ, Wu P, Spurden D, Ebbert JO, Wilson K. Efficacy of phar-
surgery: a randomized clinical trial. Colorectal Dis. 2003;5:347-352. macotherapies for short-term smoking abstinance: a systematic review
12. Kuper H, Nicholson A, Hemingway H. Searching for observational and meta-analysis. Harm Reduct J. 2009;6:25.
studies: what does citation tracking add to PubMed? A case study in 27. Lavernia CJ, Sierra RJ, Gomez-Marin O. Smoking and joint replace-
depression and coronary heart disease. BMC Med Res Methodol. 2006; ment: resource consumption and short-term outcome. Clin Orthop
6:4. Relat Res. 1999:172-180.
13. NICE: TA39 Smoking cessation - bupropion and nicotine replacement 28. Goodwin SJ, McCarthy CM, Pusic AL, et al. Complications in smok-
therapy: Guidance. Issue Date: March 2002 Review Date: March 2005. ers after postmastectomy tissue expander/implant breast reconstruc-
Available at: http://www.nice.org.uk/TA39. Accessed January 5, 2010. tion. Ann Plast Surg. 2005;55:16-20.
154 The American Journal of Medicine, Vol 124, No 2, February 2011

29. Brooks-Brunn JA. Predictors of postoperative pulmonary complica- 46. Myles PS, Iacono GA, Hunt JO, et al. Risk of respiratory complica-
tions following abdominal surgery. Chest. 1997;111:564-571. tions and wound infection in patients undergoing ambulatory surgery:
30. Dixon AJ, Dixon MP, Dixon JB, Del Mar CB. Prospective study of smokers versus nonsmokers. Anesthesiology. 2002;97:842-847.
skin surgery in smokers vs nonsmokers. Br J Dermatol. 2009;160:365- 47. Nakagawa M, Tanaka H, Tsukuma H, Kishi Y. Relationship between
367. the duration of the preoperative smoke-free period and the incidence of
31. Glassman SD, Anagnost SC, Parker A, Burke D, Johnson JR, Dimar postoperative pulmonary complications after pulmonary surgery.
JR. The effect of cigarette smoking and smoking cessation on spinal Chest. 2001;120:705-710.
fusion. Spine. 2000;25:2608-2615. 48. Padubidri AN, Yetman R, Browne E, et al. Complications of postmas-
32. Moore S, Mills BB, Moore RD, Miklos JR, Mattox TF. Perisurgical tectomy breast reconstructions in smokers, ex-smokers, and nonsmok-
smoking cessation and reduction of postoperative complications. Am J ers. Plast Reconstr Surg. 2001;107:342-349.
Obstet Gynecol. 2005;192:1718-1721. 49. Sorensen LT, Hemmingsen UB, Kirkeby LT, Kallehave F, Jorgensen
33. Selber JC, Kurichi JE, Vega SJ, Sonnad SS, Serletti JM. Risk factors LN. Smoking is a risk factor for incisional hernia. Arch Surg. 2005;
and complications in free TRAM flap breast reconstruction. Ann Plast 140:119-123.
Surg. 2006;56:492-497. 50. Taber DJ, Ashcraft E, Cattanach LA, et al. No difference between
34. Williams G, Daly M, Proude EM, et al. The influence of alcohol and smokers, former smokers, or nonsmokers in the operative outcomes of
tobacco use in orthopaedic inpatients on complications of surgery. laparoscopic donor nephrectomies. Surg Laparosc Endosc Percutan
Drug Alcohol Rev. 2008;27:55-64. Tech. 2009;19:153-156.
35. Woehlck HJ, Connolly LA, Cinquegrani MP, Dunning MB, Hoffmann 51. Vaporciyan AA, Merriman KW, Ece F, et al. Incidence of major
RG. Acute smoking increases ST depression in humans during general pulmonary morbidity after pneumonectomy: association with timing
anesthesia. Anesthes Analg. 1999;89:856-860. of smoking cessation. Ann Thorac Surg. 2002;73:420-426.
52. Warner MA, Divertie MB, Tinker JH. Preoperative cessation of smok-
36. Myles PS, Leslie K, Angliss M, Mezzavia P, Lee L. Effectiveness of
ing and pulmonary complications in coronary artery bypass patients.
bupropion as an aid to stopping smoking before elective surgery: a
Anesthesiology. 1984;60:380-383.
randomised controlled trial. Anaesthesia. 2004;59:1053-1058.
53. Warner MA, Offord KP, Warner ME, Lennon RL, Conover MA,
37. Lindstrom D, Sadr Azodi O, Wladis A, et al. Effects of a perioperative
Jansson-Schumacher U. Role of preoperative cessation of smoking and
smoking cessation intervention on postoperative complications: a ran-
other factors in postoperative pulmonary complications: a blinded
domized trial. Ann Surg. 2008;248:739-745.
prospective study of coronary artery bypass patients. Mayo Clinic
38. Sorensen LT, Hemmingsen U, Jorgensen T. Strategies of smoking
Proc. 1989;64:609-616.
cessation intervention before hernia surgery— effect on perioperative
54. Yamashita S, Yamaguchi H, Sakaguchi M, et al. Effect of smoking on
smoking behavior. Hernia. 2007;11:327-333.
intraoperative sputum and postoperative pulmonary complication in
39. Moller AM, Villebro N, Pedersen T, Tonnesen H. Effect of preoper-
minor surgical patients. Respir Med. 2004;98:760-766.
ative smoking intervention on postoperative complications: a random-
55. Møller AM, Villebro N, Pedersen T, Tønnesen H. Effect of preoper-
ised clinical trial. Lancet. 2002;359:114-117. ative smoking intervention on postoperative complications: a random-
40. Barrera R, Shi W, Amar D, et al. Smoking and timing of cessation: ised clinical trial. Lancet. 2002;359:114-117.
impact on pulmonary complications after thoracotomy. Chest. 2005; 56. Sorensen LT, Hemmingsen U, Jorgensen T. Strategies of smoking
127:1977-1983. cessation intervention before hernia surgery— effect on perioperative
41. Bluman LG, Mosca L, Newman N, Simon DG. Preoperative smoking smoking behavior. Hernia. 2007;11:327-333.
habits and postoperative pulmonary complications. Chest. 1998;113: 57. Sorensen LT, Jorgensen T. Short-term pre-operative smoking cessa-
883-889. tion intervention does not affect postoperative complications in colo-
42. Chang DW, Reece GP, Wang B, et al. Effect of smoking on compli- rectal surgery: a randomized clinical trial. Colorectal Dis. 2003;5:347-
cations in patients undergoing free TRAM flap breast reconstruction. 352.
Plast Reconstr Surg. 2000;105:2374-2380. 58. Lindström D, Azodi OS, Wladis A, et al. Effects of a perioperative
43. Kuri M, Nakagawa M, Tanaka H, Hasuo S, Kishi Y. Determination of smoking cessation intervention on postoperative complications: a ran-
the duration of preoperative smoking cessation to improve wound domized trial. Ann Surg. 2008;248:739-745.
healing after head and neck surgery. Anesthesiology. 2005;102:892- 59. The Health Consequences of Smoking: A Report of the Surgeon
896. General. [Atlanta, Ga.]: Dept. of Health and Human Services, Centers
44. Levin L, Herzberg R, Dolev E, Schawartz-Arad D. Smoking and for Disease Control and Prevention, National Center for Chronic
complications of onlay bone grafts and sinus lift operations. Int J Oral Disease Prevention and Health Promotion, Office on Smoking and
Maxillofac Implants. 2004;19:369-373. Health; Washington, DC: For sale by the Supt. of Docs., U.S. G.P.O.;
45. Mason DP, Subramanian S, Nowicki ER, et al. Impact of smoking 2004.
cessation before resection of lung cancer: a Society of Thoracic Sur- 60. Sorensen LT, Karlsmark T, Gottrup F. Abstinence from smoking
geons General Thoracic Surgery Database study. Ann Thorac Surg. reduces incisional wound infection: a randomized controlled trial. Ann
2009;88:362-371. Surg. 2003;238:1-5.
 Mills et al
Table 5 Characteristics of Included Observational Studies*,†
Cessation Cessation

Smoking Cessation Reduces Perioperative Complications

Period Intervention
Type of Studied Group According to Smoking Sample Male Mean Age Defined Cessation Pre-surgery Smoking (and Period Follow-up Period
First Author Study Primary Surgery Type Status Size (%) (y) Clearly?* Validation Method Cessation Period Pre-surgery) Postsurgery

Barrera 2005 Cohort Lung tumor NS ⫽ 21% 300 48 64 ⫾ 12 Yes Self-report via EQ quit smoking for No 30 d
resection/thoracotomy PS ⫽ questionnaire ⬎ 2 mo pre-surgery
EQ: defined as those who quit smoking LQ quit smoking for
⬎ 2 mo pre-surgery ⫽ 62% ⱕ 2 mo, but ⬎ 1
LQ: defined as those who quit smoking wk pre-surgery
ⱕ 2 mo, but ⬎ 1 wk pre-surgery ⫽
13%
CS ⫽ 4%
Bluman 1998 Cohort General, orthopedic, NS ⫽ 20% 410 97 59.5 Yes Self-report via PS: ⬎2 wk pre-surgery No NR
urologic or PS ⫽ 46%; defined as those who questionnaire CS: ⱕ2 wk pre-surgery
cardiovascular elective smoked ⬎ 2 wk pre-surgery
surgery CS ⫽ 34%; defined as those who
smoked ⱕ 2 wk pre-surgery
Chang 2000 Cohort Breast reconstruction with NS: 67% 718 NR Not specified Yes Medical records PS: quit smoking at No NR
TRAM flaps PS: defined as those who quit smoking least 4 wk pre-
at least 4 wk pre-surgery ⫽ 21% surgery
CS ⫽ 13%
Goodwin Cohort Tissue expander/implant NS ⫽ 74% 515 0 47 Yes Self-report PS ⫽ stopped smoking No 20 mo
2005 breast reconstruction PS ⫽ 15%; defined as those who quit ⬎ 4 wk pre-surgery
smoking ⬎ 4 wk pre-surgery CS ⫽ continued or
CS ⫽ 11%; defined as having stopped smoking
continued or stopped smoking ⬍ 4 ⬍ 4 wk pre-surgery
wk pre-surgery
Kuri 2005 Cohort Reconstructive head and NS ⫽ 21% 188 79.8 59 Yes Comparison of 3 LQ: smoking within No NR
neck surgery PS ⫽ self-reports 8-21 d pre-surgery
LQ ⫽ 18%; defined as smoking within IQ: smoking within
8-21 d pre-surgery 22-42 d pre-surgery
IQ ⫽ 11%; defined as smoking within EQ: smoking within
22-42 d pre-surgery 43 d or longer pre-
EQ ⫽ 35%; defined as smoking within surgery
ⱖ 43 d pre-surgery CS ⫽ smoking within
CS ⫽ 15%; defined as smoking within 7 d pre-surgery
7 d pre-surgery
Levin 2004 Cohort Onlay bone graft, sinus lift PS: patients who quit smoking for ⱖ 6 128 33.6 NR Yes Medical records PS: quit smoking for No ⱖ6 mo
mo pre-surgery ⱖ 6 mo pre-surgery postsurgery
MS: CS smoking ⬍ 10 cigarettes/d and
smoking history ⬍ 10 y
HS: CS smoking ⬎ 10 cigarettes/d and
smoking history ⬎ 10 y
Groups were undefined by their
percentages relative to sample size

154.e1
 154.e2
Table 5 Continued
Cessation Cessation
Period Intervention
Type of Studied Group According to Smoking Sample Male Mean Age Defined Cessation Pre-surgery Smoking (and Period Follow-up Period
First Author Study Primary Surgery Type Status Size (%) (y) Clearly?* Validation Method Cessation Period Pre-surgery) Postsurgery

Mason 2009 Cohort Lung resections NS ⫽ 21%; defined as never smokers 7990 48.3 66 Yes Medical records PS ⫽ LQ: quit smoking No NR
or those who smoked ⬍ 100 for ⬎ 14 d to 1 mo
cigarettes in their lifetime pre-surgery
PS: LQ ⫽ 5.1%; defined as those who IQ: quit smoking 1-12
quit smoking for ⬎ 14 d to 1 mo mo pre-surgery
pre-surgery EQ: quit smoking
IQ ⫽ 12%; defined as those who quit ⬎ 12 mo pre-
1-12 mo pre-surgery surgery
EQ ⫽ 50%; defined as those who quit CS ⫽ smoking within
smoking ⬎ 12 mo pre-surgery 14 d pre-surgery
CS ⫽ 20%; defined as smoking within
14 d pre-surgery
Myles 2002 Cohort Ambulatory surgery NS ⫽ 35%; defined as never smokers 489 38 39 Yes Self-report and CO PS ⫽ quit smoking for No 7d
with a COexp of ⱕ 10 ppm analysis ⬎ 28 d pre-surgery
PS ⫽ 24%; defined as those who quit
smoking for ⬎ 28 d pre-surgery
CS ⫽ 41%; defined as CS or with COexp
of ⬎ 10 ppm
Nakagawa Cohort Pulmonary surgery NS ⫽ 41% 288 58 61.5 Yes Medical records, PS ⫽ No NR

The American Journal of Medicine, Vol 124, No 2, February 2011

2001 PS ⫽ self-report RS ⫽ 2-4 wk pre-
RS ⫽ 5%; defined as those who quit surgery
smoking 2-4 wk pre-surgery ES ⫽ ⬎4 wk pre-
ES ⫽ 42%; defined as those who quit surgery
smoking for ⬎ 4 wk pre-surgery CS ⫽ within 2 wk pre-
CS ⫽ 13%; defined as those who surgery
smoked within 2 wk pre-surgery
Padubidri Cohort Postmastectomy breast NS ⫽ 69%; defined as those who had 748 0 49.2 Yes Medical records PS: quit smoking ⱖ 3 Yes; quitting NR
2001 reconstruction never smoked wk pre-surgery smoking
PS ⫽ 10%; defined as previous was
smokers who quit smoking ⱖ 3 wk advised
pre-surgery during
CS ⫽ 21%; defined as those who were enrolment
still smoking at the time of surgery at clinic
Sorensen Cohort Elective and emergency NS ⫽ 29% 310 45.5 64.6 (median) No Medical records, Not specified No 33-57 mo
2005 laparotomies PS ⫽ 29%; defined as those who validated by a
previously smoked but quit second self-
sometime pre-surgery report during
CS ⫽ 42%; defined as those who follow-up
smoked at time of surgery and at
follow-up
Taber 2009 Cohort Laparoscopic donor NS ⫽ 63%; defined as never smokers 221 45 39 Yes Self-report PS: quit at least 2 wk No NR
nephrectomy PS ⫽ 18%; defined as those who pre-surgery
previously smoked but quit at least
2 wk pre-surgery
CS ⫽ 19%; defined as those who
smoked up to the surgery day
 Mills et al
Table 5 Continued
Cessation Cessation

Smoking Cessation Reduces Perioperative Complications

Period Intervention
Type of Studied Group According to Smoking Sample Male Mean Age Defined Cessation Pre-surgery Smoking (and Period Follow-up Period
First Author Study Primary Surgery Type Status Size (%) (y) Clearly?* Validation Method Cessation Period Pre-surgery) Postsurgery

Vaporciyan Cohort Pneumonectomy NS ⫽ 13% 257 69.6 60 ⫾ Yes Medical records LQ: quit ⬍ 1 mo pre- No 30 d
2002 S⫽ 10 (median) surgery
LQ ⫽ 23%; defined as persons who EQ: quit ⱖ 1 mo pre-
quit smoking ⬍ 1 mo pre-surgery surgery
EQ ⫽ 63%; defined as those who quit
smoking ⱖ 1 mo pre-surgery
Warner 1984 Cohort Coronary artery bypass NS ⫽ 9% 500 77 58.2 Yes Medical records Group 1: smokers who No 30 d
grafting CS ⫽ 25%; defined as those who never stopped ⬍ 2 wk
stopped smoking pre-surgery pre-surgery
Stopped smoking ⬍ 2 wk Group 2: smokers who
pre-surgery ⫽ 17% stopped 2-4 wk pre-
Stopped smoking 2-4 wk pre-surgery ⫽ surgery
9% Group 3: smokers who
Stopped smoking 4-8 wk pre-surgery ⫽ stopped 4-8 wk pre-
6% surgery
Stopped smoking ⬎ 8 wk Group 4: smokers who
pre-surgery ⫽ 35% stopped ⬎ 8 wk
pre-surgery
Warner 1989 Cohort Coronary artery bypass NS ⫽ 22%; defined as those who had 192 83 64 Yes Urinary cotinine Among PS, No 1 y (via
grafting never smoked and assigned a analysis LQ: quit ⱕ 8 wk pre- correspondence)
smoke-free day of 150 surgery
PS ⫽ 69%; defined as previous EQ: quit ⬎ 8 wk pre-
smokers who quit smoking surgery
sometime in the past, and this
group includes LQ ⫽ 11%, defined
as those who quit ⱕ 8 wk pre-
surgery, EQ ⫽ 58%, defined as
those who quit ⬎ 8 wk pre-surgery,
and CS ⫽ 9%, defined as those with
cotinine levels ⬎ 0.5 ␮g/mL and
assigned a smoke-free day of ⫺1
Yamashita Cohort Elective minor surgeries NS ⫽ 48%; defined as those who had 1008 52.6 53 Yes Self-report via PS: quit smoking ⬍ No 30 d or until
2004 never smoked interview 2 d pre-surgery discharge
PS ⫽ 37%; defined as smokers who CS: quit smoking ⬎
quit smoking ⬍ 2 d pre-surgery 1 d pre-surgery
CS ⫽ 16%; defined as smokers who
smoked until 1 d pre-surgery

CO ⫽ carbon monoxide; CS ⫽ current smoker; EQ ⫽ early quitters; ES ⫽ ex-smoker; HS ⫽ heavy smoker; IQ ⫽ intermediate quitters; LQ ⫽ late quitters; MS ⫽ mild smoker; NR ⫽ not reported; NS ⫽
nonsmoker; ppm ⫽ packs per month; PS ⫽ past smoker; RR ⫽ relative risk; RS ⫽ recent smoker; TRAM ⫽ transverse rectus abdominis myocutaneous; ␤ ⫽ group numbers presented do not sum up to total N.
*Definition requires classification according to studied groups.
†Versus NS.

154.e3
 154.e4
Table 6 Effects of Smoking Cessation in Observational Studies*,†
Sample Male Complication Risk/Percentage Risk/Relative
First Author Studied Group According to Smoking Status Size (%) Risk/Odds Ratio Important Findings

Barrera 2005 NS ⫽ 21% 300 48 % Overall complications and 95% CI Overall pulmonary complications and pneumonia incidence were
PS ⫽ NS ⫽ 8% (1.24-14.38) higher among CS compared with PS and NS.
EQ: defined as those who quit smoking ⬎ 2 mo pre-surgery ⫽ 62% PS ⫽ 19% (14.1-24.46) CS had longer hospital stays than PS and NS.
LQ: defined as those who quit smoking ⱕ 2 mo but ⬎ 1 wk pre- CS ⫽ 23% (0.18-45.98) PQ (⬎2 mo quit time) had fewer complications compared with
surgery ⫽ 13% % Pneumonia and 95% CI recent quitters (ⱕ2 mo, ⬎1 wk quit time).
CS ⫽ 4% NS ⫽ 3% (⫺1.14 to 7.4)
PS ⫽ 11% (6.69-14.83)
CS ⫽ 23% (0.18-45.98)
% Atelectasis and 95% CI
NS ⫽ 0%
PS ⫽ 5% (2.09-7.77)
CS ⫽ 0% Mean hospital length of stay (d)
NS ⫽ 6
PS ⫽ 8
CS ⫽ 9
Bluman 1998 NS ⫽ 20% 410 97 % Any complications and 95% CI Postsurgical pulmonary complication risk was 6⫻ higher in CS
PS ⫽ 46%; defined as smoking ⬎ 2 wk pre-surgery NS ⫽ 5% (0.22-9.54) compared with NS (OR 5.5, CI, 1.9-16.2).
CS ⫽ 34%; defined as smoking ⱕ 2 wk pre-surgery PS ⫽ 13% (8.04-17.62) CS who reduced smoking pre-surgery were 7⫻ more likely to
CS ⫽ 22% (15.15-28.83) develop postsurgical pulmonary complications compared with
% Pulmonary infections and 95% CI those who did not reduce smoking.
NS ⫽ 0% Among those reducing cigarette smoking ⱕ 1 mo pre-surgery,

The American Journal of Medicine, Vol 124, No 2, February 2011

PS ⫽ 1% (⫺0.4 to 2.54) those who reduced closest to surgery were at greatest risk of
CS ⫽ 3% (0.1-5.58) developing complications.
% Atelectasis and 95% CI
NS ⫽ 4% (⫺0.4 to 7.72)
PS ⫽ 7% (3.72-11.26)
CS ⫽ 8% (3.37-12.23)
% Hospital readmission for pneumonia and 95% CI
NS ⫽ 1% (⫺1.16 to 3.6)
PS ⫽ 0
CS ⫽ 1% (⫺0.68 to 2.1)
% Mortality due to pulmonary disease and 95% CI
NS ⫽ 0
PS ⫽ 1%
CS ⫽ 1% (⫺0.51 to 1.57)
Chang 2000 NS: 67% 718 NR % Overall flap necrosis and 95% CI PS who quit smoking at least 4 wk pre-surgery have reduced
PS: defined as persons who quit smoking at least 4 wk NS: 22.6% (18.8-26.3) postsurgical complications compared with CS.
pre-surgery ⫽ 21% PS: 25.3% (18.4-32.3) Smokers with a smoking history of ⬎ 10 pack-y were at an
CS ⫽ 13% CS: 31.1% (21.6-40.7) increased risk of perioperative complications compared with
% Mastectomy skin flap necrosis and 95% CI those with ⬍ 10 pack-y.
NS: 9% (6.4-11.6) CS were at a higher risk of mastectomy skin flap necrosis
PS: 10% (5.2-14.8) than NS.
CS: 18.9% (10.8-27.0)
% Flap fat necrosis and 95% CI
NS: 6.5% (4.3-8.7)
PS: 8.7% (4.2-13.2)
CS: 7.8% (2.3-13.3)
 Mills et al
Smoking Cessation Reduces Perioperative Complications
Table 6 Continued
Sample Male Complication Risk/Percentage Risk/Relative
First Author Studied Group According to Smoking Status Size (%) Risk/Odds Ratio Important Findings

Goodwin 2005 NS ⫽ 74% 515 0 Comparison is between NS and smokers (CS ⫹ PS) Because there was no difference in the overall complication
PS ⫽ 15%; defined as having stopped smoking ⬎ 4 wk pre-surgery % Total complications and 95% CI rates in PS and CS (39.7% vs 36.5%), PS were therefore
CS ⫽ 11%; defined as having continued or stopped NS ⫽ 15.1% (11.6-18.7) considered as part of CS in the comparisons.
smoking ⬍ 4 wk pre-surgery CS ⫽ 37.9% (29.6-46.16) In comparison with NS, smokers were 3⫻ more likely to
% Reconstructive failure and 95% CI experience a complication postsurgery.
NS ⫽ 1.6% (0.34-2.86) A positive relationship was observed between cigarette ppd and
CS ⫽ 5.3% (1.48-9.12) overall complications (OR 1.80; 95% CI, 1.00-3.34).
% Skin flap necrosis and 95% CI A similar relationship also was seen between duration of
NS ⫽ 6.5% (4.03-8.97) smoking history and overall complications.
CS ⫽ 16.7% (10.31-23.03)
% Infection and 95% CI
NS ⫽ 2.9% (1.22-4.58)
CS ⫽ 9.1% (4.19-14.01)
Kuri 2005 NS ⫽ 21% 188 79.8 % Incidence of impaired wound healing This study suggests that a 3-wk smoke-free period pre-surgery
PS ⫽ and 95% CI can reduce the incidence of impaired wound healing among
LQ ⫽ 18%; defined as smoking within 8-21 d pre-surgery NS: 47.5% (32-63) patients undergoing reconstructive head and neck surgery.
IQ ⫽ 11%; defined as smoking within 22-42 d pre-surgery PS: The data suggest that smoking cessation for ⱖ 3 wk before
EQ ⫽ 35%; defined as smoking within ⱖ 43 d pre-surgery LQ ⫽ 67.6% (52-83) reconstructive head and neck surgery can provide benefits
CS ⫽ 15%; defined as smoking within 7 d pre-surgery IQ ⫽ 55.0% (33-77) for smokers, regardless of the level of cigarette
EQ ⫽ 59.1% (47-71) consumption.
CS: 85.7% (73-97)
Risk of impaired wound healing development (OR
and 95% CI) among study groups
NS: 0.11 (0.03-0.51)
PS:
LQ ⫽ 0.31 (0.08-1.24)
IQ ⫽ 0.17 (0.04-0.75)
EQ ⫽ 0.17 (0.05-0.60)
Levin 2004 PS: patients who quit smoking for ⱖ 6 mo pre-surgery 128 33.6 % Total complications and 95% CI (patients with Compared with PS who had quit smoking for ⬎ 6 mo, CS were
MS: CS smoking ⬍ 10 cigarettes/d and smoking history ⬍ 10 y onlay graft) more likely to experience postoperative complications after
HS: CS smoking ⬎ 10 cigarettes/d and smoking history ⬎ 10 y PS: 23.1% (11.63-34.53) onlay graft surgeries.
Groups were undefined by their percentages relative to sample size CS: 50% (21.71-78.29) Smoking cessation for ⬎ 6 mo can reduce complication risk to
% Total complications and 95% CI (patients with similar levels as those of NS.
sinus lift)
PS: 63.3% (49.77-76.77)
CS: 66.7% (49.8-83.54)

154.e5
 154.e6
Table 6 Continued
Sample Male Complication Risk/Percentage Risk/Relative
First Author Studied Group According to Smoking Status Size (%) Risk/Odds Ratio Important Findings

Mason 2009 NS ⫽ 21%; defined as never smokers or smoked ⬍ 100 cigarettes 7990 48.3 Hospital mortality rate (OR and 95% CI) Any smoking, past or current, was clearly associated with
in their lifetime NS: 0.39% increased hospital mortality and pulmonary complications.
PS ⫽ PS: Therefore, surgeons should counsel smokers that risk remains
LQ ⫽ 5.1%; defined as persons who quit smoking for ⬎ 14 d to LQ ⫽ 1.7% (4.6 [1.2-18])† elevated regardless of timing of cessation, but that quitting
1 mo pre-surgery IQ ⫽ 1.3% (2.6 [0.65-11])† holds a benefit that improves over time.
IQ ⫽ 12%; defined as persons who quit 1-12 mo pre-surgery EQ ⫽ 1.5% (2.5 [0.82-7.6])† Smoking seemed to have less of an effect on pulmonary
EQ ⫽ 50%; defined as persons who quit smoking ⬎ 12 mo pre- CS: 1.5% (3.5 [1.1-11])† complications compared with mortality. The longer the
surgery Pulmonary complications (OR and 95% CI) cessation period, the greater the risk reduction in pulmonary
CS ⫽ 20%; defined as smoking within 14 d pre-surgery NS: 2.6 complications.
PS: During this study, the authors were unable to identify an
LQ ⫽ 6.2 (1.6 [0.85-3.1])† optimal interval of smoking cessation.
IQ ⫽ 6.4 (1.5 [0.81-2.8])†
EQ ⫽ 5.8 (1.3 [0.77-2.2])†
CS ⫽ 6.9 (1.8 [1.05-3.1])†
Myles 2002 NS ⫽ 35%; defined as never smokers and with a COexp of ⱕ 10 489 38 % Any respiratory complication and 95% CI Smokers (PS ⫹ CS) had a significantly increased risk of
ppm NS: 25.9% (19.47-32.55) respiratory complication compared with NS.
PS ⫽ 24%; defined as patients who quit smoking for ⬎ 28 d pre- PS: 34.5% (25.83-43.13) Adjusted OR 1.66 (95% CI, 1.07-2.57)
surgery CS: 33.5% (26.96-40.04)
CS ⫽ 41%; defined as CS or with COexp of ⬎ 10 ppm CS vs NS: (adjusted OR 1.71, 95% CI, 1.03-2.84)
% wound infections and 95% CI
NS: 0.6% (⫺0.55 to 1.71)

The American Journal of Medicine, Vol 124, No 2, February 2011

PS: 2.6% (⫺0.3 to 5.48)
CS: 3.7% (1.08-6.32)
CS vs NS: (adjusted OR: 16.3, 95% CI, 1.58-1.75)
Nakagawa 2001 NS ⫽ 41% 288 58 % Overall incidence of postoperative pulmonary For patients undergoing pulmonary surgery, smoking cessation
PS ⫽ complications and 95% CI of at least 4 wk pre-surgery can lower the risk of
RS ⫽ 5%; defined as patients who quit smoking 2-4 wk pre-surgery NS: 23.9% (16.2-31.66) postoperative pulmonary complications.
ES ⫽ 42%; defined as patients who quit smoking for ⬎ 4 wk pre- PS:
surgery RS ⫽ 53.8% (26.75-80.95)
CS ⫽ 13%; defined as patients who smoked within 2 wk pre- ES ⫽ 34.7% (26.23-43.19)
surgery CS: 43.2% (27.28-59.2)
Adjusted OR and 95% CI
CS vs NS: 2.09 (0.83-5.25)
RS vs NS: 2.44 (0.67-8.89)
ES vs NS: 1.03 (0.47-2.26)
Padubidri 2001 NS ⫽ 69%; defined as persons who had never smoked 748 0 % Overall complication and 95% CI Patients who are scheduled to undergo breast reconstructions
PS ⫽ 10%; defined as previous smokers who quit smoking ⱖ 3 wk NS: 25.9% (22.14-29.7) should quit smoking for at least 3 wk pre-surgery.
pre-surgery PS: 25% (15.26-34.74) Quitting in this time frame or earlier can lower the
CS ⫽ 21%; defined as those who were still smoking at the time of CS: 39.4% (31.66-47.04) complications rate to that observed among NS.
surgery % Mastectomy flap necrosis and 95% CI
NS: 1.5% (0.49-2.61)
PS: 2.6% (⫺0.97 to 6.23)
CS: 7.7% (3.53-11.95)
Sorenson 2005 NS ⫽ 29% 310 45.5 Association of smoking status with incisional CS are 4⫻ at greater risk of incisional hernia than NS.
PS ⫽ 29%; defined as those who previously smoked but quit hernia (adjusted OR and 95% CI)
sometime pre-surgery NS: 1
CS ⫽ 42%; defined as those who smoked at time of surgery and at PS: 1.57 (0.81-3.64)
follow-up CS: 3.93 (1.82-8.49)
 Mills et al
Smoking Cessation Reduces Perioperative Complications
Table 6 Continued
Sample Male Complication Risk/Percentage Risk/Relative
First Author Studied Group According to Smoking Status Size (%) Risk/Odds Ratio Important Findings

Taber 2009 NS ⫽ 63%; defined as never smokers 221 45 Mean hospital length of stay (days ⫾SD) Among patients receiving laparoscopic donor nephrectomies,
PS ⫽ 18%; defined as those who previously smoked but quit at NS: 2 ⫾ 1 postoperative outcomes were similar among CS, PS who quit
least 2 wk pre-surgery PS: 2 ⫾ 1 at least 2 wk pre-surgery, and NS.
CS ⫽ 19%; defined as those who smoked up to the surgery day CS: 2 ⫾ 1
% Postoperative infection and 95% CI
NS: 3% (0.1-5.62)
PS: 8% (⫺0.67-16.05)
CS: 5% (⫺1.68-11.2)
Vaporciyan 2002 NS ⫽ 13% 257 69.6 % Incidence of major pulmonary events and Timing of smoking cessation was a predictor of developing
S⫽ 95% CI major pulmonary events.
LQ ⫽ 23%; defined as persons who quit smoking ⬍ 1 mo LQ: 21.7% (11.25-32.09) Smoking within 1 mo of pneumonectomy was strongly
pre-surgery EQ: 9.2% (4.76-13.64) associated with the development of major pulmonary events.
EQ ⫽ 63%; defined as persons who quit smoking ⱖ 1 mo Risk of developing major pulmonary events (OR Patients who quit smoking late (ie, smoked within 1 mo pre-
pre-surgery and 95% CI) surgery) were 2.7⫻ more likely to develop major pulmonary
LQ vs EQ ⫽ 2.70 (1.18-6.17) events compared with those who quit smoking for periods ⱖ
1 mo.
Warner 1984 NS ⫽ 9% 500 77 % Overall pulmonary problems (NS vs smokers) and Among patients undergoing coronary artery bypass grafting,
CS ⫽ 25%; defined as persons who never stopped smoking pre- 95% CI clinical benefit due to smoking cessation was observed only
surgery NS ⫽ 11.4% (1.98-20.74) in patients who quit smoking for ⬎ 8 wk pre-surgery.
Stopped smoking ⬍ 2 wk pre-surgery ⫽ 17% Smokers: 39% (34.56-43.52) Smoking cessation in this period can lower the incidence of
Stopped smoking 2-4 wk pre-surgery ⫽ 9% % Estimate of incidence of pulmonary postoperative complications to levels comparable to those
Stopped smoking 4-8 wk pre-surgery ⫽ 6% complications and 95% CI of NS.
Stopped smoking ⬎ 8 wk pre-surgery ⫽ 35% NS ⫽ 11.4% (1.98-20.74) Preoperative smoking cessation for ⬍ 8 wk does not lower
Stopped ⬎ 8 wk pre-surgery ⫽ 17% (11.45-22.55) postoperative pulmonary complications.
Stopped 4-8 wk pre-surgery ⫽ 46% (27.54-64.46)
Stopped 2-4 wk pre-surgery ⫽ 62% (47.66-76.34)
Stopped ⬍ 2 wk pre-surgery ⫽ 57% (46.56-67.72)
CS ⫽ 48.4% (39.59-57.19)
Warner 1989 NS ⫽ 22%; defined as persons who never smoked and assigned a 192 83 % Incidence of postoperative pulmonary Among patients undergoing coronary artery bypass grafting,
smoke-free day of 150 complications and 95% CI smoking cessation ⬍ 2 mo pre-surgery does not seem to
PS ⫽ 69%; defined as previous smokers who quit smoking NS: 11.9% (2.11-21.69) reduce the incidence of postoperative pulmonary
sometime in the past, and this group includes: ⬎6 mo: 11.1% complications below that observed in CS.
LQ ⫽ 11%; defined as persons who quit smoking ⱕ 8 wk EQ (⬎8 wk): 14.5% (7.88-20.94) Patients who quit smoking ⱕ 2 mo pre-surgery were ⬃4⫻ more
pre-surgery LQ (ⱕ8 wk): 57.1% (35.97-78.31) at risk of developing postoperative pulmonary complications
EQ ⫽ 58%; defined as persons who quit smoking ⬎ 8 wk CS: 33% (11.55-55.11) (57.1% vs 14.5%).
pre-surgery Smokers who quit ⬎ 6 mo pre-surgery had comparable
CS ⫽ 9%; defined as persons with cotinine levels ⬎ 0.5 ␮g/mL complication rates as NS (11.1% vs 11.9%).
and assigned a smoke-free day of ⫺1

154.e7
 154.e8
Table 6 Continued
Sample Male Complication Risk/Percentage Risk/Relative
First Author Studied Group According to Smoking Status Size (%) Risk/Odds Ratio Important Findings

Yamashita 2004 NS ⫽ 48%; defined as persons who had never smoked 1008 52.6 % Intraoperative sputum volume increase and 95% CI In minor surgical patients undergoing general anesthesia, PS
PS ⫽ 37%; defined as smokers who quit smoking ⬍ 2 d NS: 9.4% (6.77-11.99) and CS were ⬃2⫻ more likely to have an increased
pre-surgery PS: 17.9% (13.98-21.8) intraoperative sputum volume than NS.
CS ⫽ 16%; defined as smokers who smoked until 1 d pre-surgery CS: 18.2% (12.24-24.24) Smoking cessation ⱖ 2 mo pre-surgery can lower the risk of
Smoke-free period and risk of intraoperative intraoperative sputum volume increase.
sputum volume increase (RR and 95% CI)
Quit ⬍ 2 mo pre-surgery
PS/CS vs NS: 2.0 (0.9-4.3)
Quit ⬍ 2 wk pre-surgery
PS/CS vs NS: 2.4 (1.2-4.8)
% Total postoperative pulmonary complications
and 95% CI
NS: 1.7% (0.52-2.82)
PS: 1.4% (0.18-2.54)
CS: 1.9% (⫺0.23 to 4.01)

AS ⫽ acute smoker; CG ⫽ control group; CI ⫽ confidence interval; CO ⫽ carbon monoxide; CS ⫽ current smoker; EQ ⫽ early quitters; ES ⫽ ex-smoker; HS ⫽ heavy smoker; IG ⫽ intervention group; IQ ⫽ intermediate
quitters; LQ ⫽ late quitters; MS ⫽ mild smoker; N ⫽ sample size; NA ⫽ not applicable; NR ⫽ not reported; NS ⫽ nonsmoker; OR ⫽ odds ratio; ppd ⫽ packs per day; ppm ⫽ packs per month; PS ⫽ past smoker;
PQ ⫽ past quitter; RR ⫽ relative risk; RS ⫽ recent smoker; SD ⫽ standard deviation; ␤ ⫽ group numbers presented do not sum up to total N.
*Definition requires classification according to studied groups.
†Versus NS.

The American Journal of Medicine, Vol 124, No 2, February 2011