annals of internal medicine clinical guidelines...diate-acting (atracurium, vecuronium) and...
TRANSCRIPT
Strategies To Reduce Postoperative Pulmonary Complications afterNoncardiothoracic Surgery: Systematic Review for the AmericanCollege of PhysiciansValerie A. Lawrence, MD; John E. Cornell, PhD; and Gerald W. Smetana, MD
Background: Postoperative pulmonary complications are as fre-quent and clinically important as cardiac complications in terms ofmorbidity, mortality, and length of stay. However, there has beenmuch less research and no previous systematic reviews of theevidence of interventions to prevent pulmonary complications.
Purpose: To systematically review the literature on interventions toprevent postoperative pulmonary complications after noncardiotho-racic surgery.
Data Sources: MEDLINE English-language literature search, 1 Jan-uary 1980 through 30 June 2005, plus bibliographies of retrievedpublications.
Study Selection: Randomized, controlled trials (RCTs); systematicreviews; or meta-analyses that met predefined inclusion criteria.
Data Extraction: Using standardized forms, the authors abstracteddata on study methods, quality, intervention and control groups,patient characteristics, surgery, postoperative pulmonary complica-tions, and adverse events.
Data Synthesis: The authors qualitatively synthesized, withoutmeta-analysis, evidence from eligible studies. Good evidence (2systematic reviews, 5 additional RCTs) indicates that lung expansioninterventions (for example, incentive spirometry, deep breathing
exercises, and continuous positive airway pressure) reduce pulmo-nary risk. Fair evidence suggests that selective, rather than routine,use of nasogastric tubes after abdominal surgery (2 meta-analyses)and short-acting rather than long-acting intraoperative neuromus-cular blocking agents (1 RCT) reduce risk. The evidence is conflict-ing or insufficient for preoperative smoking cessation (1 RCT), epi-dural anesthesia (2 meta-analyses), epidural analgesia (6 RCTs, 1meta-analysis), and laparoscopic (vs. open) operations (1 systematicreview, 1 meta-analysis, 2 additional RCTs), although laparoscopicoperations reduce pain and pulmonary compromise as measured byspirometry. While malnutrition is associated with increased pulmo-nary risk, routine total enteral or parenteral nutrition does notreduce risk (1 meta-analysis, 3 additional RCTs). Enteral formula-tions designed to improve immune status (immunonutrition) mayprevent postoperative pneumonia (1 meta-analysis, 1 additionalRCT).
Limitations: The overall quality of the literature was fair: Ten of 20RCTs and 6 of 11 systematic reviews were good quality.
Conclusions: Few interventions have been shown to clearly orpossibly reduce postoperative pulmonary complications.
Ann Intern Med. 2006;144:596-608. www.annals.orgFor author affiliations, see end of text.
Postoperative pulmonary complications are as commonas cardiac complications for patients undergoing non-
cardiothoracic surgery (1–6). Further, these complicationshave similar mortality rates and length of stay after electiveabdominal surgery or hip fracture repair (1, 2). In an ac-companying systematic review (7), we identify patient,procedure, and laboratory risk factors for postoperativepulmonary complications. Our current systematic reviewsynthesizes the evidence on preventive strategies and fo-cuses on atelectasis, pneumonia, and respiratory failure.While we have written the review primarily for internists,this field crosses specialty disciplines.
METHODS
Literature Search and Selection CriteriaWe performed a systematic MEDLINE English-lan-
guage literature search from 1 January 1980 to 30 June2005. The search strategy and inclusion and exclusion cri-teria are described in the accompanying review of risk fac-tors and in further detail in its Appendix, available at www.annals.org (7). The search strategy used 1) the MedicalSubject Heading (MeSH) terms preoperative care, intraop-erative care, postoperative care, intraoperative complications,
and postoperative complications as a focus of the article; 2)the MeSH text term perioperative complications as a textterm in the title or abstract; and 3) additional MeSH andtext terms for pulmonary, respiratory, or cardiopulmonaryconditions, complications, or care. In addition, we per-formed additional focused searches for preoperative chestradiography and spirometry, laparoscopic versus open ma-jor abdominal operations, general versus spinal or epiduralanesthesia, intraoperative neuromuscular blockade, postop-erative pain management, and postoperative lung expan-sion techniques. Eligible studies were randomized, con-trolled trials; systematic reviews; or meta-analyses. We
See also:
PrintRelated articles . . . . . . . . . . . . . . . . . . . . . . . . 575, 581Summary for Patients. . . . . . . . . . . . . . . . . . . . . . . I-40
Web-OnlyAppendix TablesCME quizConversion of tables into slides
Annals of Internal Medicine Clinical Guidelines
596 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 www.annals.org
Downloaded From: http://annals.org/ on 10/30/2017
excluded studies with less than 25 participants per group;studies from developing countries (because of potential dif-ferences in respiratory and intensive care technology); stud-ies that used physiologic (for example, lung volumes andflow, oximetry) rather than clinical outcome measures;studies of gastric pH manipulation; studies of complica-tions that are unique to the surgery (for example, upperairway obstruction after uvulectomy); studies of cardiopul-monary, pediatric, or organ transplantation surgery (be-cause of profoundly immunosuppressive drugs); and stud-ies that used only administrative data to identifypostoperative complications (for example, InternationalClassification of Diseases, Ninth Revision, Clinical Modi-fication [ICD-9-CM], codes) because of recent evidencethat administrative data have poor validity for postopera-tive complications (8, 9).
Assessment of Study QualityWe used the Quality of Reporting of Meta-analyses
(QUOROM) statement for reporting meta-analyses andthe U.S. Preventive Services Task Force criteria for hierar-chy of research design to assess internal validity and studyquality (good, fair, or poor) and to make conclusions aboutstrength of the evidence (10, 11).
Statistical AnalysisWe used simple means and chi-square tests to calculate
CIs and P values when they were not provided in publica-tions. We did not perform quantitative pooling becausemultiple meta-analyses were beyond the scope of a broadreview of multiple potential interventions. We reportpooled results from previous meta-analyses when applicable.
Role of the Funding SourceThe Veterans Evidence-based Research, Dissemina-
tion, and Implementation Center (VERDICT) (VeteransAffairs Health Services Research and Development, HFP98-002) provided the research librarian and administrativesupport for the study. The funding source had no role inthe design, conduct, or reporting of the study or in thedecision to submit the manuscript for publication.
RESULTS
The search and inclusion criteria identified 20 ran-domized clinical trials and 11 systematic reviews or meta-analyses (12–42). Figure 1 in the accompanying review (7)of risk factors for postoperative pulmonary complicationsdetails the search results. Appendix Tables 1 through 7(available at www.annals.org) provide detailed characteris-tics of the eligible randomized trials and systematic reviews.
Preoperative Smoking CessationIn the only trial of preoperative smoking cessation
(12), 108 older, relatively healthy men undergoing hip orknee replacement were randomly assigned to usual care orweekly meetings with a nurse for advice about smokingcessation and nicotine withdrawal plus individualized nic-otine replacement for 6 to 8 weeks before surgery until 10
days after surgery. The mean age of the men was 65 years,and 95% were American Society of Anesthesiologists(ASA) physical status class I or II. Of 56 patients in theintervention group, 36 stopped smoking and 14 reducedsmoking before surgery. Overall complications rates werelower in the intervention group (18% vs. 52%; P �0.001), primarily due to fewer wound complications andurinary tract infections. The only pulmonary outcome,postoperative ventilator support, occurred in 1 patient ineach group. Non–statistically significant trends favoredshorter mean hospital stay (11 days vs. 13 days; P � 0.41)and fewer cardiac complications (0% vs. 10%; P � 0.08)in the intervention group.
Although the trial was of good quality, several factorslimit its ability to demonstrate decreased risk for postoper-ative pulmonary complications. Pulmonary risk is inher-ently low with hip and knee replacement. Furthermore, thetiming of smoking cessation seems important. A previouscohort study showed paradoxically higher postoperativepulmonary complication rates for smokers who stopped orreduced smoking within 2 months before noncardiotho-racic surgery (43). Smoking cessation may increase short-term risk because of transiently increased mucus produc-tion due to improved mucociliary activity and reducedcoughing due to less bronchial irritation.
Anesthetic and Analgesic TechniquesAnesthetics disrupt central regulation of breathing and
result in uncoordinated neural messaging. Due to resultinghypoventilation plus positional dependence, regional atel-ectasis occurs shortly after induction. It persists postopera-tively and is compounded by ongoing disruption of respi-ratory muscles, limited respiratory excursion due to pain,and disruption of neurally mediated diaphragmatic func-tions after manipulation of abdominal viscera (43).
Neuromuscular Blockade
One good-quality trial found no difference in rates ofpostoperative pulmonary complications between interme-diate-acting (atracurium, vecuronium) and long-acting(pancuronium) neuromuscular blocking agents among 691patients undergoing elective abdominal, gynecologic, or or-thopedic surgery (13). However, the incidence of residualneuromuscular block was higher among patients receivingpancuronium (26% vs. 5%; P � 0.001). Patients with re-sidual blockade after pancuronium were 3 times morelikely to develop postoperative pulmonary complicationsthan those without residual block (17% vs. 5%; P � 0.02).In contrast, among patients receiving intermediate-actingagents, postoperative pulmonary complication rates didnot differ between those with (4%) and without (5%) pro-longed blockade. Therefore, pancuronium may directlylead to higher rates of prolonged neuromuscular blockadeand indirectly to increased pulmonary risk compared withshorter-acting agents.
Clinical GuidelinesStrategies To Reduce Postoperative Pulmonary Complications after Noncardiothoracic Surgery
www.annals.org 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 597
Downloaded From: http://annals.org/ on 10/30/2017
Anesthesia and Analgesia
Neuraxial blockade (either spinal or epidural anesthe-sia) blocks a constellation of stress responses to surgery(neuroendocrine, cytokine, and pain threshold) and mayimprove recovery and prevent complications (44). Postop-erative epidural analgesia may reduce respiratory muscledysfunction and pain-related hypoventilation. The epi-dural approach involves either a single injection or an in-fusion and can be used for both intraoperative anesthesiaand postoperative analgesia. Spinal anesthesia has a fasteronset (5 to 10 minutes vs. 15 to 20 minutes), producesdenser sensory and motor block, and is technically easierthan epidural anesthesia. However, spinal anesthesia is ad-ministered only as a single injection because of practicalconstraints of indwelling intrathecal catheters. The possiblebenefit of neuraxial blockade has generated studies of gen-eral versus neuraxial blockade anesthesia, followed by trialscomparing epidural analgesia to other modes of analgesicdelivery (for example, oral, intramuscular, intravenous, pa-tient-controlled analgesia) and, more recently, trials ofcombined epidural intraoperative anesthesia and epiduralpostoperative analgesia.
Intraoperative General Anesthesia versus Neuraxial Blockade
A recent good-quality meta-analysis combined 141 tri-als (n � 9559) comparing general anesthesia and neuraxial
blockade in patients undergoing a variety of operations(32). The authors compared patients receiving neuraxialblockade (with or without concomitant general anesthesia)with those receiving only general anesthesia. Neuraxialblockade reduced overall mortality (2% vs. 3%; odds ratio,0.70 [95% CI, 0.54 to 0.90]), pneumonia (3% vs. 5%;odds ratio, 0.61 [CI, 0.48 to 0.76]), and respiratory failure(0.5% vs. 0.8%; odds ratio, 0.41 [CI, 0.23 to 0.73]). In asubgroup analysis of trials of neuraxial blockade alone ver-sus general anesthesia alone, results were similar (odds ra-tio, 0.63 [CI, 0.46 to 0.87] for pneumonia; odds ratio,0.37 [CI, 0.11 to 1.21] for respiratory failure).
Potential sources of bias in the meta-analysis include1) clinically heterogeneous studies; 2) unusually high mor-tality rates in several trials; 3) older literature (82% of in-cluded studies were published before 1990); 4) small stud-ies (81% of included studies had � 50 patients); and 5)statistically significant benefit only for orthopedic surgeryin subgroup analyses (45–47).
A smaller good-quality systematic review identified 15randomized or quasi-randomized trials of 2162 patientsundergoing hip fracture repair (33). Postoperative pneu-monia rates were almost identical: 5.1% of 529 patientshaving neuraxial blockade and 5.5% of 567 patients havinggeneral anesthesia (odds ratio, 0.92 [CI, 0.53 to 1.59]).Twelve of the 15 trials were also included in the larger
Table 1. Randomized, Controlled Trials of Combined Intraoperative Anesthesia and Postoperative Analgesia*
Author, Year(Reference)
Type of Surgery Intervention Group Patients
Total,n
Men,n
Age, y
Norris et al.,2001 (14)
Elective abdominal aorticsurgery
Four groups: 1) intraoperative GETA � postoperative IV PCA; 2)intraoperative GETA � postoperative epidural PCEA; 3)intraoperative GETA � supplemental epidural � postoperativeIV PCA; 4) intraoperative GETA � supplemental epidural �postoperative PCEA
168 115 Mean, 68(SD, 9.5)
Rigg et al.,2002 (15)
Elective abdominal oresophageal surgery
Two groups: 1) intraoperative GETA � postoperative IV opioid;2) intraoperative GETA � epidural local anesthetic �postoperative epidural local anesthetic and additional opioidas needed
915 NR‡ NR‡
Park et al.,2001 (16)
Elective abdominal surgery(Veterans Hospitals)
Two groups: 1) intraoperative GETA � postoperative IV or IMopioid; 2) intraoperative GETA � epidural anesthesia �postoperative epidural opioid
1021 1021 Mean, 67(SD, 8.8)
Fleron et al.,2003 (17)
Elective abdominal aorticsurgery
Two groups: 1) intraoperative GETA � IV opioid �postoperative IV opioid; 2) intraoperative GETA � epiduralopioid � postoperative IV opioid
217 192 Mean, 66.5(SD, 10.5)
Mann et al.,2000 (18)
Elective major abdominalsurgery for cancer
Two groups: 1) intraoperative GETA � postoperative IV PCAwith morphine; 2) intraoperative GETA � epidural �postoperative PCEA with combined local anesthetic andsufentanil
70 38 Mean, 76.5(SD, 5.2)
Cuschieri et al.,1985 (19)
Elective cholecystectomy Three groups: 1) intraoperative GETA � postoperative IMmorphine; 2) intraoperative GETA � postoperative IVmorphine; 3) intraoperative GETA � epidural local anesthetic� postoperative epidural local anesthetic for 12 h thenmorphine as needed
75 16 52 (range,18–75)
* GETA � general endotracheal anesthesia; IM � intramuscular; IV � intravenous; NR � not reported; NS � not significant; PCA � patient-controlled analgesia; PCEA �patient-controlled epidural analgesia; PPC � postoperative pulmonary complication.† Level of evidence for all studies is I � randomized clinical trial.‡ Sex and age not given in primary publication or previous methods publication.
Clinical Guidelines Strategies To Reduce Postoperative Pulmonary Complications after Noncardiothoracic Surgery
598 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 www.annals.org
Downloaded From: http://annals.org/ on 10/30/2017
meta-analysis (32), which included 44 trials of orthopedicsurgery (n � 3617). Why the results for pneumonia differbetween the 2 meta-analyses is not clear, but importantvariables may include type or duration of procedure (hipfracture repair is inherently low risk for postoperative pul-monary complications), intraoperative fluids, and postop-erative pain and rehabilitative management.
Combined Intraoperative and Postoperative Anesthesia andAnalgesia
Table 1 summarizes 6 eligible trials that comparedvarious regimens of intraoperative anesthesia and postop-erative analgesia. In a double-blind, good-quality efficacytrial of patients undergoing abdominal aortic surgery (14),investigators randomly assigned patients to 1 of 4 com-bined anesthetic and analgesic protocols. The trial stan-dardized the entire episode of anesthesia and pain manage-ment to optimize efficacy in all 4 groups. The primaryoutcome measure was length of stay; secondary outcomesincluded postoperative pulmonary complications. Samplesizes were small (37, 38, 39, and 46 participants, respec-tively), and median length of stay (7 to 8 days for allgroups) or postoperative pulmonary complication rates didnot differ among the groups.
Strengths of the trial include the double-blind designand equally highly standardized protocols for both anesthe-
sia and analgesia (48, 49). Unequally optimized regimenscan introduce bias that systematically favors one type ofintervention. Potential weaknesses, regarding prevention ofpostoperative pulmonary complication, include length ofstay as the primary outcome measure and small sample size(50–52).
In a subsequent fair-quality effectiveness trial (15),915 patients undergoing major abdominal surgery wererandomly assigned to general anesthesia and 1) postopera-tive intravenous opioid or 2) intraoperative epidural localanesthetic plus postoperative epidural analgesia. Overall in-fections did not differ, and the authors did not report re-sults for pneumonia. Statistically significantly less pain andrespiratory failure occurred with epidural anesthesia, butonly 225 of 447 patients in the epidural group completedthe protocol.
In a subgroup analysis of high-risk patients (respira-tory insufficiency by arterial blood gas analysis, severe ob-structive or restrictive lung disease, acute respiratory failurewithin the past 2 years, or morbid obesity), rates of pneu-monia (11% vs. 12%; P � 0.71) or mechanical ventilationfor more than 24 hours (8% for both groups) did notdiffer. Respiratory failure (ventilation � 24 hours, reintu-bation, PaO2 � 50 mm Hg, or PaCO2 � 50 mm Hg onroom air) occurred significantly less often with epidural(45% vs. 29%; odds ratio, 0.5 [CI, 0.29 to 0.88]) (53).
Table 1—Continued
RandomAllocationConcealed
BlindedOutcomeAssessment
PPCs Level ofEvidence†;Study Quality
Results
Yes Yes Reintubation, prolonged intubation, pneumonia (primaryoutcome was length of stay; PPCs were secondaryoutcomes)
I; good Overall PPCs: 20% vs. 17% vs. 25% vs. 9%; P � NSReintubation: 0%–2.9%; P � 0.9Prolonged intubation: 7%–22%; P � 0.3Pneumonia: 0%–2.8%; P � 0.6
No No Respiratory failure, overall infection I; fair Infection: 47% in IV group, 43% in epidural group; P �0.26
Respiratory failure: 30% in IV group, 23% in epiduralgroup; P � 0.02
Yes No Primary outcomes: death, respiratory failure; secondaryoutcome: pneumonia
I; fair Mortality: 4% vs. 3%Respiratory failure: 10% with epidural vs. 14%; P � 0.06Pneumonia: 5% with epidural vs. 8%; P � 0.15
Yes No Pneumonia, lobar atelectasis, respiratory failure I; fair Overall PPCs: 16% with epidural, 23% with IV opioid;P � 0.32
Yes No Segmental or lobar atelectasis, pneumonia, major PPCsdefined by clinical score based on physicalexamination
I; poor Atelectasis: 23% with epidural vs. 18%; P � 0.77Major PPCs: 3% vs. 3%
Unclear No Atelectasis, pneumonia I; poor Atelectasis: 20% with epidural vs. 28% with IV and 40%with IM morphine; P � NS
Pneumonia: 4% with epidural vs. 20% with IV morphine(P � 0.20) and 24% with IM morphine (P � 0.11)
Clinical GuidelinesStrategies To Reduce Postoperative Pulmonary Complications after Noncardiothoracic Surgery
www.annals.org 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 599
Downloaded From: http://annals.org/ on 10/30/2017
In a large fair-quality trial (16), 1021 patients under-going abdominal surgery were randomly assigned to gen-eral anesthesia and 1) postoperative systemic opioid or 2)intraoperative epidural anesthesia plus postoperative epi-dural morphine. Mortality rates did not differ, and non–statistically significant trends favored the epidural approachfor pneumonia and respiratory failure.
In 1 fair-quality trial (17), 217 patients undergoingelective abdominal aortic surgery were randomly assignedto general anesthesia alone or general anesthesia plus intra-operative epidural opioid. Both groups received the samepostoperative pain management. A non–statistically signif-icant trend favored intraoperative epidural for overall post-operative pulmonary complications. Rates of individualtypes of postoperative pulmonary complications did notdiffer, but statistical power was low.
In a smaller poor-quality trial (18), 70 elderly patientsundergoing major abdominal surgery were randomly as-signed to general anesthesia and 1) postoperative patient-controlled morphine or 2) intraoperative neuraxial blockadeplus postoperative patient-controlled epidural analgesia.Rates of atelectasis or major pulmonary complications didnot differ. In an additional, small poor-quality trial (19), 75patients undergoing elective cholecystectomy were ran-domly assigned to general anesthesia and 1) postoperativeintramuscular morphine, 2) continuous intravenous mor-phine, or 3) intraoperative epidural local anesthesia plus post-operative epidural local anesthetic for 12 hours. Postoperativepneumonia occurred less often with epidural (4%) than witheither intramuscular morphine (24%; P � 0.05) or intravenousmorphine (20%; P � 0.11).
Postoperative Analgesic Technique
A fair-quality meta-analysis examined the evidence for3 epidural techniques: intercostal nerve block, systemicopioids, and wound infiltration with local anesthetic (34).The number of trials that compared any 2 strategies variedfrom 2 to 11. Compared with systemic opioids, epiduralopioids reduced atelectasis (relative risk, 0.53 [CI, 0.33 to0.85]; 11 studies) but not pneumonia (relative risk, 0.53[CI, 0.18 to 1.53]; 5 studies). Compared with systemicopioids, epidural local anesthetic reduced “pulmonary in-fection” (relative risk, 0.36 [CI, 0.21 to 0.65]; 5 studies)but not atelectasis (relative risk, 0.74 [CI, 0.50 to 1.11]; 4studies). However, the authors pooled studies of both on-demand (that is, as requested) and patient-controlled intra-venous analgesia, which could bias the results of the meta-analysis in favor of epidural analgesia.
In contrast, a good-quality meta-analysis identified 32trials (n � 1029) of patient-controlled opioid analgesiaversus the same drug given intravenously, intramuscularly,or subcutaneously (35). Opioid consumption, pain scores,length of stay, or adverse effects did not differ. In the 2trials reporting postoperative pulmonary complications,fewer complications occurred in the patient-controlled an-
algesia group (odds ratio, 0.93 [CI, 0.86 to 0.99]; numberneeded to treat, 15 [CI, 8 to 98]).
Summary
Evidence from 1 good-quality trial suggests that shorter-acting neuromuscular blocking drugs may prevent postop-erative pulmonary complications. Intraoperative neuraxialblockade, either alone or in combination with general an-esthesia, may prevent postoperative pulmonary complica-tions, but the evidence is conflicting. Several meta-analyses(which included small unblinded studies) suggest that epi-dural anesthesia may reduce pulmonary risk, but recentlarge randomized trials do not confirm benefit. Random-ized trials of combined intraoperative and postoperativeanesthetic or analgesic regimens do not clearly indicate thata combined epidural approach prevents postoperative pul-monary complications. Two meta-analyses of postoperativeanalgesic regimens suggest that part of the variability maybe due to on-demand analgesia (intravenous, intramuscu-lar, or subcutaneous) versus patient-controlled analgesia(intravenous or epidural). Postoperative epidural and pa-tient-controlled intravenous analgesia both seem superiorto on-demand delivery of opioids in preventing postoper-ative pulmonary complications. Epidural analgesia mayfurther reduce postoperative pulmonary complications.More good-quality efficacy trials with standardized optimalregimens for all groups and sufficient size to examine pul-monary complication rates are needed (14). The risk forepidural bleeding due to postoperative epidural catheters inpatients receiving heparin (especially low-molecular-weightheparin) makes timing of catheter placement importantand may influence decisions about modalities for pain con-trol and thromboembolism prophylaxis (54–56).
Laparoscopic versus Open ProceduresOur search identified many trials comparing laparo-
scopic and open procedures, but few reported postopera-tive pulmonary complication rates. Those that did focusedon cholecystectomy and colorectal surgery. Downs and col-leagues (36) performed a good-quality systematic reviewthrough March 1995 of open and laparoscopic cholecys-tectomy. They identified 18 trials (n � 1645) that werepublished with sufficient detail to judge methodologicquality. Twelve trials had at least 40 patients per studygroup, and the largest study had 150 participants pergroup. Since the authors did not quantitatively pool databecause of clinical heterogeneity and methodologic prob-lems, we examined the studies individually. None met thecriteria for inclusion in our review (�25 participants pergroup [8 studies] or no clinical postoperative pulmonarycomplications reported [10 studies]).
Among the 4 highest-quality trials reporting spiromet-ric outcomes, unblinded outcome assessment found statis-tically significantly greater compromise in FVC and FEV1
at 24 hours and 48 hours postoperatively with open cho-lecystectomy. In 1 study that followed patients until pul-
Clinical Guidelines Strategies To Reduce Postoperative Pulmonary Complications after Noncardiothoracic Surgery
600 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 www.annals.org
Downloaded From: http://annals.org/ on 10/30/2017
monary function recovered to within 10% to 15% of pre-operative levels, pulmonary function recovered 4 to 10days earlier with laparoscopic cholecystectomy. Only 1very small (n � 40) blinded trial assessed whether reducedpulmonary dysfunction translated into clinically importantdifferences in postoperative pulmonary complication rates.On postoperative chest radiography, atelectasis occurredsignificantly less often with laparoscopic operations (40%vs. 90%; P � 0.001).
We identified 1 subsequent, poor-quality trial of lapa-roscopic versus open cholecystectomy (20). Among 82 pa-tients, the frequency and severity of atelectasis, assessed byradiologists who were blinded to type of procedure, weresignificantly less among patients randomly assigned tolaparoscopic cholecystectomy (frequency, 29% vs. 63%[P � 0.05]; severity, chi-square for trend P � 0.05). How-ever, the analysis was not intention-to-treat: Patients whoconverted from laparoscopic to open operations were ex-cluded from analysis.
Table 2 summarizes the results of a good-qualitymeta-analysis of laparoscopic versus open resection of colo-rectal cancer (37). Overall mortality did not differ. Riskwas consistently less with laparoscopic operations for sev-eral complications but was not statistically significantly lesswith the more conservative statistical approach of random-effects modeling. The reduced risk for overall complica-tions was primarily due to fewer wound complications,especially wound infection. Regarding pulmonary compli-cations, a non–statistically significant trend favored laparo-scopic resection. Three studies that evaluated postoperativepulmonary complications reported that respiratory recovery(defined by spirometry) was statistically significantly faster.Nine studies reporting data confirmed shorter length of hos-pital stay (mean, 21% shorter [range, 14% to 38%]) afterlaparoscopic operations (37).
We identified 1 additional good-quality trial of lapa-roscopic versus open colorectal resection in 384 patients;269 were in an earlier publication that was included in theprevious meta-analysis discussed (21). Again, a nonsignifi-cant trend favored lower rates of pneumonia after laparo-scopic operations (3 of 190 [1.8%] patients and 6 of 194[3.5%] patients; P � 0.52) (21).
Two large retrospective cohort studies highlight theproblems with using designs other than a randomized trialand ICD-9-CM codes to identify postoperative pulmonarycomplications to compare open and laparoscopic opera-tions (57, 58). Atelectasis (the only postoperative pulmo-nary complication studied) occurred significantly less oftenwith laparoscopic (n � 19 662) compared with open (n �23 771) cholecystectomy (4% vs. 2%; P � 0.001) (57).Overall postoperative pulmonary complications were sig-nificantly less frequent after laparoscopic (n � 709) com-pared with open (n � 17 735) sigmoid resection (2.5% vs.6%; P � 0.001) (58). The results of these 2 studies may beunreliable because of lack of systematic prospective surveil-lance. Clinicians may have been biased to order more post-
operative chest radiographs (and therefore identify moreatelectasis) after open procedures. Furthermore, in a recentstudy comparing discharge ICD-9-CM codes and system-atic chart review for complications (8), specificity of ICD-9codes was high but sensitivity was low: 35% (CI, 30% to41%) for all complications and 32% (CI, 19% to 45%) forall infectious complications.
In summary, while supported by spirometric, postop-erative pain, and length of stay data, whether laparoscopicprocedures reduce the risk for clinically important pulmo-nary complications is not clear. The literature did not sys-tematically assess or report pulmonary complications, andmost studies did not have sufficient statistical power todetect differences in postoperative pulmonary complicationrates.
Nasogastric Decompression after Abdominal SurgerySelective use of nasogastric decompression, or tubes,
refers to use only for postoperative nausea or vomiting,inability to tolerate oral intake, or symptomatic abdominaldistension. Routine decompression (that is, standard useuntil bowel function returns) has been thought to speedbowel recovery and decrease risk for aspiration. We iden-tified 2 meta-analyses of studies of routine versus selectivenasogastric decompression (38, 39, 59). One or both meta-analyses included all the trials that we identified.
The first meta-analysis was of good methodologicquality up to quantitative analyses, which pooled data fromrandomized trials, nonrandomized trials, “uncontrolled”trials, and case–control studies (38). For the overall groupof 26 studies (which comprised 15 RCTs, 3 nonrandom-ized trials, and 8 case–control studies [n � 3964]), pa-tients receiving selective decompression had significantlylower rates of pneumonia (odds ratio, 0.49; P � 0.001)and atelectasis (odds ratio, 0.46; P � 0.001) and shortertime to oral intake (3.5 days vs. 4.6 days; P � 0.04). As-piration rates (odds ratio, 0.61; P � 0.88) did not differ.Selective decompression did not statistically significantlyincrease nausea, vomiting, or abdominal distension. For 20higher-quality studies (15 trials and 5 case–control stud-ies), patients receiving selective decompression also had
Table 2. Summary of Results of Meta-Analysis of 12Randomized, Controlled Trials for Laparoscopic OperationsRelative to Open Operations for Colorectal Cancer*
Result Odds Ratio (95% CI)
Fixed-Effects Model Random-Effects Model
Mortality Data not given 0.85 (0.33–2.21)Overall morbidity 0.62 (0.45–0.85) 0.68 (0.38–1.24)Overall complications 0.60 (0.45–0.79) 0.62 (0.38–1.03)Wound complications 0.47 (0.28–0.78) 0.47 (0.28–0.78)Wound infection 0.47 (0.28–0.80) 0.47 (0.28–0.80)Respiratory complications 0.65 (0.28–1.49) 0.65 (0.28–1.49)
* Data from Abraham et al. (37).
Clinical GuidelinesStrategies To Reduce Postoperative Pulmonary Complications after Noncardiothoracic Surgery
www.annals.org 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 601
Downloaded From: http://annals.org/ on 10/30/2017
lower rates of pneumonia (odds ratio, 0.59; P � 0.01) andatelectasis (odds ratio, 0.52; P � 0.002), a trend towardshorter time to oral intake (3.5 days vs. 4.5 days; P �0.07), no difference in aspiration rates, and significantlyhigher rates of vomiting (odds ratio, 1.45; P � 0.005) andabdominal distension (odds ratio, 1.34; P � 0.02)
The recent meta-analysis was of good quality, and itidentified 28 eligible trials (n � 4194) of routine versusselective nasogastric decompression after open laparotomy(39, 59). It included 15 of the 17 RCTs in the previousmeta-analysis. Of 19 trials (n � 2892) reporting postoper-ative pulmonary complication, 18 included only electiveoperations (39, 59). Patients who were randomly assignedto selective decompression had fewer postoperative pulmo-nary complications, and the benefit approached statistical
significance (relative benefit increase, 1.35 [CI, 0.98 to1.86]; P � 0.07; calculated relative risk reduction, 0.74[CI, 0.54 to 1.02]). Selective decompression also resultedin earlier bowel recovery (8 studies; n � 862; 0.46 day [CI,0.28 day to 0.64 day]; P � 0.001).
In summary, the evidence suggests that selective naso-gastric decompression (that is, for specific indicationsrather than routine decompression) improves return of bowelfunction and may reduce the incidence of postoperative pul-monary complications after elective abdominal operations.
Lung Expansion ModalitiesDecreased lung volumes and atelectasis due to surgery-
related shallow breathing, bed rest, diaphragmatic dysfunc-tion, pain, and impaired mucociliary clearance may be the
Table 3. Meta-Analyses and Randomized, Controlled Trials of Lung Expansion Interventions To Prevent Postoperative PulmonaryComplications*
Author, Year (Reference) Type of Surgery Intervention StudiesIdentified/EligibleRCTs, n/n
RandomAllocationConcealed
BlindedOutcomeAssessment
Systematic reviews ormeta-analyses
Thomas and McIntosh,1994 (40)
Any upperabdominalsurgery
IS, IPPB, DBEs 116/14 NA NA
Overend et al.,2001 (41)
Upper and lowerabdominalsurgery
IS, IPPB, DBEs, CPAP, PEP, orCPAP with PT
85/4 NA NA
Subsequent RCTsChumillas et al.,
1998 (22)Elective upper
abdominalsurgery
DBEs vs. no prophylaxis Unclear Unclear
Fagevik Olsen et al.,1997 (23)
Elective openabdominalsurgery
Low risk: DBEs vs. noprophylaxis; high risk: DBEsplus PEP vs. no prophylaxis‡
Unclear Unclear
Hall et al., 1991 (24) Abdominal surgery IS, chest PT (control) Yes Yes
Hall et al., 1996 (25) Abdominal surgery Low risk: IS vs. DBEs; high risk:IS vs. IS plus chest PT§
Yes Yes
Bohner et al., 2002 (26) Electiveintra-abdominalvascular surgery
Nasal CPAP for 12 h aftersurgery vs. O2 by nasalcannula to keep saturation� 95%
Yes Unclear
* CDC � Centers for Disease Control and Prevention; CPAP � continuous positive airway pressure; DBE � deep breathing exercise; FiO2 � fraction of inspired oxygen;IPPB � intermittent positive-pressure breathing; IQR � interquartile range; IS � incentive spirometry; NA � data not available; OR � odds ratio; PEP � positiverespiratory pressure throughout expiration; PPC � postoperative pulmonary complication; PT � physical therapy; RCT � randomized, controlled trial.† Level of evidence for all studies is I � randomized controlled trial.‡ High risk � age � 50 y and current smoker or body mass index � 30 kg/m2 or lung disease requiring daily medication.§ High risk � American Society of Anesthesiologists class � I or age � 60 y.
Clinical Guidelines Strategies To Reduce Postoperative Pulmonary Complications after Noncardiothoracic Surgery
602 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 www.annals.org
Downloaded From: http://annals.org/ on 10/30/2017
first events in a cascade leading to postoperative pulmonarycomplication. However, the evidence on prophylactic lungexpansion is limited by variable techniques, inconsistentdefinitions of postoperative pulmonary complications, andpoor-quality trials. Techniques include incentive spirome-try, deep breathing exercises, chest physical therapy (whichmay include variable combinations of deep breathing,cough, postural drainage, percussion and vibration, suc-tioning, and ambulation), intermittent positive-pressurebreathing, and continuous positive airway pressure. Table3 summarizes the evidence.
The first of 2 poor-quality systematic reviews focusedon upper abdominal surgery and identified 14 randomizedtrials (sample size, 17 to 200 participants) (40). Across alllung expansion modalities, a trend favored fewer postoper-
ative pulmonary complications compared with controls(odds ratio, 0.85 [CI, 0.59 to 1.2)], but heterogeneity wasstatistically significant. In 2 studies, postoperative pulmo-nary complications occurred less often in patients receivingincentive spirometry compared with control (odds ratio,0.44 [CI, 0.18 to 0.99]). In 4 studies, postoperative pul-monary complications occurred less often in patients whowere randomly assigned to deep breathing exercises (oddsratio, 0.43 [CI, 0.27 to 0.63]), but heterogeneity was againstatistically significant. Across other studies, no single mo-dality was clearly superior. The second systematic reviewidentified 4 randomized trials of patients undergoing ab-dominal surgery (41). The authors did not report raw orpooled postoperative pulmonary complication rates. In theonly trial in our systematic review that met our sample size
Table 3—Continued
Participants PPCs/Outcomes Level ofEvidence†;StudyQuality
Results
Total,n
Men,n
Age, y
1337 NA NA Atelectasis or infiltrate on chestradiograph
I; poor IS vs. no treatment: 2 studies (n � unclear)(OR, 0.44 [95% CI, 0.18–0.99])IS vs. IPPB: 3 studies (n � unclear)(OR, 0.73 [CI, 0.39–1.36])IS vs. DBE: 4 studies (n � unclear)(OR, 0.91 [CI, 0.57–1.4])IPPB vs. DBE: 2 studies (n � unclear)(OR, 0.94 [CI, 0.28–3.17])
NA NA NA No data except wide variabilitynoted
I; poor No quantitative pooling due to clinical heterogeneity; noquantitative results for individual studies reported. In 3 studies(all with � 25 participants per group), IS was no better thanDBEs or no treatment and was inferior to CPAP or PEP. In afourth study (41–44 participants per group), IS, DBEs, and IPPB(PPC rates of 21%, 22%, and 22%, respectively) were equallysuperior to no prophylaxis (PPC rate 48%; P � 0.05 for allcomparisons).
81 35 Mean, 64.1(range, 18–84)
Bronchitis, atelectasis, pneumonia I; poor DBEs: fewer abnormalities on chest radiograph (15% vs. 39%; P �0.02) and trend toward fewer PPCs (8% vs. 20%; P � 0.11)
368; 79(20%) werehigh-risk‡
158 Mean, 53.4(range, 19–92)
Pneumonia I; poor DBEs: lower rate of overall pneumonia (0.6% vs. 7%; P � 0.05)
876 430 Median, 55 (IQR,32–72)
Clinical examination of collapse orconsolidation plus abnormal chestradiograph or positive sputum“microbiology”
I; poor No difference between IS and chest PT in rates of PPCs (16% vs.15%) and abnormal chest radiograph (22% both groups)
456 209 Low-risk: median,36 (IQR,29–44); highrisk: median,68 (IQR,58–76)
Clinical examination of collapse orconsolidation plus abnormal chestradiograph or positive sputum“microbiology”
I; poor No difference in rate of PPCs: low risk: 8% vs. 11% (P � 0.50);high risk: 19% vs. 13% (P � 0.18)
204 166 Mean, 64 (SD11.8)
Pneumonia per CDC criteria, severehypoxemia (PaO2 � 70 mm Hgat FiO2 � 0.70)
I; good Nasal CPAP: lower rate of severe hypoxemia (5% vs. 16%; P �0.01); trends toward lower rate of pneumonia (2% vs. 5%; P �0.45) and reintubation (1% vs. 5%; P � 0.21)
Clinical GuidelinesStrategies To Reduce Postoperative Pulmonary Complications after Noncardiothoracic Surgery
www.annals.org 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 603
Downloaded From: http://annals.org/ on 10/30/2017
criteria, incentive spirometry, deep breathing exercises, andintermittent positive-pressure breathing equally preventedpostoperative pulmonary complications compared with nointervention.
We identified 5 other trials in patients undergoing ma-jor abdominal surgery. The first 4 trials were of poor qual-ity. Two trials compared chest physiotherapy with no pro-phylaxis (22, 23). In the first study, patients who wererandomly assigned to chest expansion, maximum inspira-tion exercises, cough, and early ambulation had fewer ab-normalities on postoperative chest radiography and a non–statistically significant trend toward fewer postoperativepulmonary complications (22). In the second trial, patientswho were randomly assigned to cough and deep breathingexercises had significantly lower rates of pneumonia (0.6%vs. 7%; P � 0.05) (23).
The third trial compared “conventional chest physio-therapy” with incentive spirometry in 876 patients under-going abdominal surgery and found no difference in ratesof overall postoperative pulmonary complication, abnor-mal postoperative chest radiography, or PaO2 less than 60mm Hg (24). The fourth poor-quality study compared 1)incentive spirometry and deep breathing exercises in 155low-risk patients and 2) incentive spirometry versus incen-tive spirometry plus chest physiotherapy in 301 high-riskpatients (ASA class � I or age � 60 years) undergoingabdominal surgery (25). Among high- or low-risk patients,postoperative pulmonary complication rates did not differwith any intervention.
In the fifth and only good-quality trial (26), 204 pa-tients undergoing intra-abdominal vascular surgery wererandomly assigned to supplemental oxygen to maintain ar-terial oxygen saturation greater than 95% or to nasal con-tinuous positive airway pressure for 12 hours after surgery.Severe hypoxemia (PaO2 � 70 mm Hg at fraction of in-spired oxygen � 0.70%) occurred statistically significantlyless often (5% vs. 16%; P � 0.01), and non–statisticallysignificant trends favored less pneumonia and reintubationwith continuous positive airway pressure.
For patients having abdominal surgery, the evidencesuggests that any type of lung expansion intervention isbetter than no prophylaxis. No modality seems superior,and combined modalities do not seem to provide addi-tional risk reduction. Incentive spirometry may be the leastlabor-intensive, while continuous positive airway pressuremay be particularly beneficial for patients who cannot par-ticipate in incentive spirometry or deep breathing exercises.
Nutritional SupportTotal Parenteral or Total Enteral Hyperalimentation
A fair-quality meta-analysis of 14 randomized or qua-si-randomized trials of total parenteral nutrition (TPN)versus no TPN through August 1986 concluded that rou-tine TPN in major surgery was not beneficial, except per-haps for severe malnutrition or for extended periods (10days to 14 days) of inadequate enteral nutrition (60). The
meta-analysis did not report results specific to pulmonarycomplications and was therefore ineligible for our review.
Subsequently, a good-quality multisite trial randomlyassigned 395 patients undergoing laparotomy or noncar-diac thoracotomy to perioperative TPN or no TPN (27).Overall rates of major complications (26% vs. 25%) and90-day mortality (13% vs. 11%) were similar between thegroups. Total parenteral nutrition was associated withnon–statistically significant trends toward higher rates ofpneumonia and empyema but significantly lower rates ofnoninfectious complications (5.3% vs. 42.9%; P � 0.03).
We identified 1 poor-quality meta-analysis (230 pa-tients) and 2 additional, good-quality trials of TPN versustotal enteral nutrition (TEN) (28, 29, 42). In the meta-analysis, infections were twice as common among patientsreceiving TPN (35% vs. 16%; P � 0.01) even after exclud-ing patients with catheter sepsis from analysis (29% vs.16%; P � 0.03) (42). There was a nonstatistically signifi-cant trend toward more frequent pneumonia in patientsreceiving TPN. In a trial of 241 patients, there was a non–statistically significant trend toward more postoperativepulmonary complications with TEN (7% vs. 13%; P �0.12) (28). In a second, larger trial, 317 malnourished pa-tients (�10% weight loss in previous 6 months) were ran-domly assigned to TPN or TEN (29). Rates of overallcomplications and infectious complications were statisti-cally significantly lower with TEN, but rates of pneumonia(9 of 159 patients vs. 14 of 158 patients; P � 0.39) or thecombined outcome of pneumonia and respiratory failure(13 of 159 patients vs. 20 of 158 patients; P � 0.27) didnot differ.
Immunonutrition
Immunonutrition refers to enteral feedings with addi-tional ingredients (variable combinations of arginine, �-3fatty acids, or ribonucleic acids) to enhance the immunesystem and to possibly prevent infection. A good-qualitymeta-analysis of trials found that for patients undergoingelective surgery, enteral immunonutrition had no mortalitybenefit but resulted in significantly fewer overall infectiouscomplications (relative risk, 0.53 [CI, 0.42 to 0.68]) (61).The authors did not report results for respiratory infec-tions; therefore, the study was not eligible for our review.
In a subsequent good-quality trial of enteral immu-nonutrition (30), 305 patients undergoing elective resec-tion of gastrointestinal cancer were randomly assigned toan enteral solution enriched with arginine, �-3 fatty acids,and ribonucleic acids preoperatively (5 days before surgery;n � 102) or perioperatively (5 days before surgery plusjejunal tube feeding begun within 12 hours of surgery andcontinued until oral intake was resumed; n � 101) or to acontrol group (n � 102) of postoperative intravenous glu-cose and electrolytes. Overall infection rates were signifi-cantly lower with immunonutrition (14% and 16% vs.30%; P � 0.006 and 0.02, respectively), but rates of pneu-
Clinical Guidelines Strategies To Reduce Postoperative Pulmonary Complications after Noncardiothoracic Surgery
604 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 www.annals.org
Downloaded From: http://annals.org/ on 10/30/2017
monia (3 of 102 patients and 6 of 101 patients vs. 8 of 102patients) did not differ (30).
In summary, while hypoalbuminemia and malnutri-tion increase postoperative complications, including pneu-monia, routine TPN has no benefit over either TEN or nohyperalimentation, except perhaps for patients with severemalnutrition or for long periods of inadequate oral nutri-tion. More research is needed on enteral formulations thatmay enhance immune status. Prompt resumption of oralintake after surgery is important because atrophy of intes-tinal villi occurs quickly with inadequate intake and in-creases the risk for bacterial translocation across gut mu-cosa and subsequent sepsis (62).
Intervention of No Benefit: Pulmonary ArteryCatheterization
After observational data suggested higher rates of re-spiratory failure and pneumonia in patients receiving right-heart catheterization for noncardiac surgery (63), Sandhamand colleagues (31) performed a good-quality RCT. High-risk patients (n � 1994; age � 60 years; ASA class III orIV) undergoing urgent or elective major noncardiac oper-ations were randomly assigned to usual care or treatmentguided by perioperative pulmonary artery catheter. Pulmo-nary artery catheterization did not reduce the primary out-come of in-hospital all-cause mortality (7.8% vs. 7.7%) orthe rate of postoperative pneumonia, a secondary outcome(6.7% vs. 7.3%; P � 0.70).
DISCUSSION
Recent evidence has shown that postoperative pulmo-nary and cardiac complications are equally prevalent andclinically important in morbidity, mortality, and length ofstay. However, compared with postoperative cardiac com-plications, much less research on prevention of pulmonarycomplications has been published. Table 4 summarizes the
strength of available evidence, based on our systematic re-view, on interventions to reduce the risk for postoperativepulmonary complications.
Strategies of Proven BenefitGood evidence suggests that lung expansion therapy
(for example, incentive spirometry, deep breathing exer-cises, and continuous positive airway pressure) reducespostoperative pulmonary risk after abdominal surgery.Well-designed trials are needed to clarify the magnitude ofbenefit and the comparative effectiveness of different mo-dalities.
Strategies of Probable BenefitFair evidence suggests that selective nasogastric tube
decompression after abdominal surgery reduces risk. Fairevidence also suggests that short-acting neuromuscularblocking agents result in lower rates of residual neuromus-cular blockade and may reduce risk for pulmonary compli-cations.
Strategies of Possible BenefitLaparoscopic, compared with open, abdominal opera-
tions reduce pain and pulmonary compromise as measuredby spirometry and oxygenation. However, the evidence isinsufficient to determine whether laparoscopic operationsprevent clinically important pulmonary complications.Given the benefits of laparoscopic procedures in pain con-trol and length of stay, future trials to adequately assessclinical pulmonary outcomes are unlikely.
Strategies of Unclear BenefitEvidence is insufficient to judge the potential benefit
of preoperative smoking cessation in reducing risk. Riskmay actually increase transiently after stopping or reducingsmoking within 2 months of surgery due to increased se-cretions. We need trials of preoperative smoking cessation
Table 4. Strength of the Evidence for Specific Interventions To Reduce the Risk for Postoperative Pulmonary Complications
Risk Reduction Strategy Strength of Evidence* Type of Complication Studied
Postoperative lung expansion modalities A Atelectasis, pneumonia, bronchitis, severe hypoxemiaSelective postoperative nasogastric decompression B Atelectasis, pneumonia, aspirationShort-acting neuromuscular blockade B Atelectasis, pneumoniaLaparoscopic (vs. open) operation C Spirometry, atelectasis, pneumonia, overall respiratory complicationsSmoking cessation I Postoperative ventilator supportIntraoperative neuraxial blockade I Pneumonia, postoperative hypoxia, respiratory failurePostoperative epidural analgesia I Atelectasis, pneumonia, respiratory failureImmunonutrition I Overall infectious complications, pneumonia, respiratory failureRoutine total parenteral or enteral nutrition† D Atelectasis, pneumonia, empyema, respiratory failureRight-heart catheterization D Pneumonia
* Definitions for categories of strength of evidence, modified from the U.S. Preventive Services Task Force categories (11). A � good evidence that the strategy reducespostoperative pulmonary complications and benefit outweighs harm; B � at least fair evidence that the strategy reduces postoperative pulmonary complications and benefitoutweighs harm; C � at least fair evidence that the strategy may reduce postoperative pulmonary complications, but the balance between benefit and harm is too close tojustify a general recommendation; D � at least fair evidence that the strategy does not reduce postoperative pulmonary complications or harm outweighs benefit; I � evidenceof effectiveness of the strategy to reduce postoperative pulmonary complications is conflicting, of poor quality, lacking, or insufficient or the balance between benefit and harmcannot be determined.† Evidence remains uncertain (strength of evidence I) on total parenteral or enteral nutrition for severely malnourished patients or when a protracted time of inadequatenutritional intake is anticipated.
Clinical GuidelinesStrategies To Reduce Postoperative Pulmonary Complications after Noncardiothoracic Surgery
www.annals.org 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 605
Downloaded From: http://annals.org/ on 10/30/2017
before higher-risk surgeries that adequately address optimalduration of cessation.
Evidence on intraoperative epidural anesthesia andpostoperative epidural analgesia is insufficient. More good-quality efficacy trials of sufficient size (in which all groupsreceive equally standardized and optimized regimens) areneeded to accurately examine complication rates.
Strategies of No BenefitAlthough malnutrition is associated with increased risk
for postoperative pulmonary complications, good evidenceindicates that routine total parenteral or enteral hyperali-mentation nutrition does not reduce risk, except perhapsfor patients with severe malnutrition or for those undergo-ing long periods with inadequate oral intake. Enteral for-mulations that are tailored to enhance immune status andreduce postoperative infections may be promising.
Evidence from 1 well-done randomized trial indicatesthat invasive perioperative monitoring with pulmonary ar-tery catheterization does not reduce risk of pulmonarycomplications.
LimitationsA limitation of our review is the overall quality of the
literature. Only 10 of 20 RCTs and 6 of 11 systematicreviews or meta-analyses were of good quality.
Future ResearchFuture studies of interventions to reduce postoperative
pulmonary complications should be randomized trials thatare designed to overcome methodologic problems in earlierliterature. Cohort studies using secondary analyses of ad-ministrative databases should use measures for pulmonarycomplications that are proven valid and reliable by directclinical assessment or medical chart audit. Studies shouldbe large enough to adjust for known potential risk factorsand confounding variables (as synthesized in the accompa-nying systematic review of preoperative risk stratification[7]) and should go beyond surrogate or intermediate phys-iologic or spirometric outcomes to detect clinically mean-ingful differences in clinical pulmonary complications.This is important for 2 reasons: to base patient care onclinically meaningful evidence and to determine when it isappropriate to substitute intermediate outcomes to shortenstudy timelines and reduce study cost. Researchers shoulddefine postoperative pulmonary complications a priori ac-cording to explicit criteria and, whenever possible, use out-come assessment that is masked, or blinded, to interven-tion assignment.
From the South Texas Veterans Health Care System and The Universityof Texas Health Science Center at San Antonio, San Antonio, Texas, andBeth Israel Deaconess Medical Center, Harvard Medical School, Boston,Massachusetts.
Disclosure: Members of the American Society of Anesthesiologists alsoreviewed the manuscript. Their review implies neither agreement withnor endorsement of this document.
Acknowledgments: The authors gratefully acknowledge the tremendouscontribution of medical librarian Martha R. Harris, MA, for her timeand expertise in searching the medical literature and managing the re-sulting project database. They also thank the Department of Anesthesi-ology, especially Christopher Jankowski, MD, of the Mayo Clinic, Roch-ester, Minnesota, for assistance in interpreting the anesthesiologyliterature.
Grant Support: By the Veterans Evidence-based Research, Dissemina-tion, and Implementation Center (VERDICT) (Veterans Affairs HealthServices Research and Development, HFP 98-002).
Potential Financial Conflicts of Interest: Stock ownership or options(other than mutual funds): G.W. Smetana (SafeMed Harvard Imaging);Other: G.W. Smetana (Novartis Pharma Schweiz).
Requests for Single Reprints: Valerie A. Lawrence, MD, Medicine/General Medicine, University of Texas Health Science Center at SanAntonio, 7703 Floyd Curl Drive, Mail Code 7879, San Antonio, TX78229-3900; e-mail, [email protected].
Current author addresses are available at www.annals.org.
References1. Lawrence VA, Hilsenbeck SG, Mulrow CD, Dhanda R, Sapp J, Page CP.Incidence and hospital stay for cardiac and pulmonary complications after ab-dominal surgery. J Gen Intern Med. 1995;10:671-8. [PMID: 8770719]2. Lawrence VA, Hilsenbeck SG, Noveck H, Poses RM, Carson JL. Medicalcomplications and outcomes after hip fracture repair. Arch Intern Med. 2002;162:2053-7. [PMID: 12374513]3. Thomas EJ, Goldman L, Mangione CM, Marcantonio ER, Cook EF, Lud-wig L, et al. Body mass index as a correlate of postoperative complications andresource utilization. Am J Med. 1997;102:277-83. [PMID: 9217597]4. Rosen AK, Geraci JM, Ash AS, McNiff KJ, Moskowitz MA. Postoperativeadverse events of common surgical procedures in the Medicare population. MedCare. 1992;30:753-65. [PMID: 1518309]5. Escarce JJ, Shea JA, Chen W, Qian Z, Schwartz JS. Outcomes of opencholecystectomy in the elderly: a longitudinal analysis of 21,000 cases in theprelaparoscopic era. Surgery. 1995;117:156-64. [PMID: 7846619]6. Pedersen T. Complications and death following anaesthesia. A prospectivestudy with special reference to the influence of patient-, anaesthesia-, and surgery-related risk factors. Dan Med Bull. 1994;41:319-31. [PMID: 7924461]7. Smetana GW, Lawrence VA, Cornell JE. Preoperative pulmonary risk strati-fication for noncardiothoracic surgery: systematic review for the American Col-lege of Physicians. Ann Intern Med. 2006;144:581-95.8. Romano PS, Chan BK, Schembri ME, Rainwater JA. Can administrativedata be used to compare postoperative complication rates across hospitals? MedCare. 2002;40:856-67. [PMID: 12395020]9. Romano PS, Schembri ME, Rainwater JA. Can administrative data be used toascertain clinically significant postoperative complications? Am J Med Qual.2002;17:145-54. [PMID: 12153067]10. Moher D, Cook DJ, Eastwood S, Olkin I, Rennie D, Stroup DF. Improv-ing the quality of reports of meta-analyses of randomised controlled trials: theQUOROM statement. Quality of Reporting of Meta-analyses. Lancet. 1999;354:1896-900. [PMID: 10584742]11. Harris RP, Helfand M, Woolf SH, Lohr KN, Mulrow CD, Teutsch SM, etal. Current methods of the US Preventive Services Task Force: a review of theprocess. Am J Prev Med. 2001;20:21-35. [PMID: 11306229]12. Møller AM, Villebro N, Pedersen T, Tønnesen H. Effect of preoperativesmoking intervention on postoperative complications: a randomised clinical trial.Lancet. 2002;359:114-7. [PMID: 11809253]13. Berg H, Roed J, Viby-Mogensen J, Mortensen CR, Engbaek J, SkovgaardLT, et al. Residual neuromuscular block is a risk factor for postoperative pulmo-nary complications. A prospective, randomised, and blinded study of postopera-tive pulmonary complications after atracurium, vecuronium and pancuronium.Acta Anaesthesiol Scand. 1997;41:1095-1103. [PMID: 9366929]
Clinical Guidelines Strategies To Reduce Postoperative Pulmonary Complications after Noncardiothoracic Surgery
606 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 www.annals.org
Downloaded From: http://annals.org/ on 10/30/2017
14. Norris EJ, Beattie C, Perler BA, Martinez EA, Meinert CL, Anderson GF,et al. Double-masked randomized trial comparing alternate combinations of in-traoperative anesthesia and postoperative analgesia in abdominal aortic surgery.Anesthesiology. 2001;95:1054-67. [PMID: 11684971]15. Rigg JR, Jamrozik K, Myles PS, Silbert BS, Peyton PJ, Parsons RW, et al.Epidural anaesthesia and analgesia and outcome of major surgery: a randomisedtrial. Lancet. 2002;359:1276-82. [PMID: 11965272]16. Park WY, Thompson JS, Lee KK. Effect of epidural anesthesia and analgesiaon perioperative outcome: a randomized, controlled Veterans Affairs cooperativestudy. Ann Surg. 2001;234:560-9; discussion 569-71. [PMID: 11573049]17. Fleron MH, Weiskopf RB, Bertrand M, Mouren S, Eyraud D, Godet G, etal. A comparison of intrathecal opioid and intravenous analgesia for the incidenceof cardiovascular, respiratory, and renal complications after abdominal aortic sur-gery. Anesth Analg. 2003;97:2-12. [PMID: 12818934]18. Mann C, Pouzeratte Y, Boccara G, Peccoux C, Vergne C, Brunat G, et al.Comparison of intravenous or epidural patient-controlled analgesia in the elderlyafter major abdominal surgery. Anesthesiology. 2000;92:433-41. [PMID:10691230]19. Cuschieri RJ, Morran CG, Howie JC, McArdle CS. Postoperative pain andpulmonary complications: comparison of three analgesic regimens. Br J Surg.1985;72:495-8. [PMID: 4016522]20. Karayiannakis AJ, Makri GG, Mantzioka A, Karousos D, Karatzas G.Postoperative pulmonary function after laparoscopic and open cholecystectomy.Br J Anaesth. 1996;77:448-52. [PMID: 8942326]21. Vignali A, Braga M, Zuliani W, Frasson M, Radaelli G, Di Carlo V.Laparoscopic colorectal surgery modifies risk factors for postoperative morbidity.Dis Colon Rectum. 2004;47:1686-93. [PMID: 15540300]22. Chumillas S, Ponce JL, Delgado F, Viciano V, Mateu M. Prevention ofpostoperative pulmonary complications through respiratory rehabilitation: a con-trolled clinical study. Arch Phys Med Rehabil. 1998;79:5-9. [PMID: 9440408]23. Fagevik Olsen M, Hahn I, Nordgren S, Lonroth H, Lundholm K. Ran-domized controlled trial of prophylactic chest physiotherapy in major abdominalsurgery. Br J Surg. 1997;84:1535-8. [PMID: 9393272]24. Hall JC, Tarala R, Harris J, Tapper J, Christiansen K. Incentive spirometryversus routine chest physiotherapy for prevention of pulmonary complicationsafter abdominal surgery. Lancet. 1991;337:953-6. [PMID: 1678039]25. Hall JC, Tarala RA, Tapper J, Hall JL. Prevention of respiratory complica-tions after abdominal surgery: a randomised clinical trial. BMJ. 1996;312:148-52; discussion 152-3. [PMID: 8563533]26. Bohner H, Kindgen-Milles D, Grust A, Buhl R, Lillotte WC, Muller BT,et al. Prophylactic nasal continuous positive airway pressure after major vascularsurgery: results of a prospective randomized trial. Langenbecks Arch Surg. 2002;387:21-6. [PMID: 11981680]27. Perioperative total parenteral nutrition in surgical patients. The VeteransAffairs Total Parenteral Nutrition Cooperative Study Group. N Engl J Med.1991;325:525-32. [PMID: 1906987]28. Pacelli F, Bossola M, Papa V, Malerba M, Modesti C, Sgadari A, et al.Enteral vs parenteral nutrition after major abdominal surgery: an even match.Arch Surg. 2001;136:933-6. [PMID: 11485531]29. Bozzetti F, Braga M, Gianotti L, Gavazzi C, Mariani L. Postoperativeenteral versus parenteral nutrition in malnourished patients with gastrointestinalcancer: a randomised multicentre trial. Lancet. 2001;358:1487-92. [PMID:11705560]30. Gianotti L, Braga M, Nespoli L, Radaelli G, Beneduce A, Di Carlo V. Arandomized controlled trial of preoperative oral supplementation with a special-ized diet in patients with gastrointestinal cancer. Gastroenterology. 2002;122:1763-70. [PMID: 12055582]31. Sandham JD, Hull RD, Brant RF, Knox L, Pineo GF, Doig CJ, et al. Arandomized, controlled trial of the use of pulmonary-artery catheters in high-risksurgical patients. N Engl J Med. 2003;348:5-14. [PMID: 12510037]32. Rodgers A, Walker N, Schug S, McKee A, Kehlet H, van Zundert A, et al.Reduction of postoperative mortality and morbidity with epidural or spinal an-aesthesia: results from overview of randomised trials. BMJ. 2000;321:1493.[PMID: 11118174]33. Urwin SC, Parker MJ, Griffiths R. General versus regional anaesthesia forhip fracture surgery: a meta-analysis of randomized trials. Br J Anaesth. 2000;84:450-5. [PMID: 10823094]34. Ballantyne JC, Carr DB, deFerranti S, Suarez T, Lau J, Chalmers TC, et al.The comparative effects of postoperative analgesic therapies on pulmonary out-come: cumulative meta-analyses of randomized, controlled trials. Anesth Analg.
1998;86:598-612. [PMID: 9495424]35. Walder B, Schafer M, Henzi I, Tramer MR. Efficacy and safety of patient-controlled opioid analgesia for acute postoperative pain. A quantitative systematicreview. Acta Anaesthesiol Scand. 2001;45:795-804. [PMID: 11472277]36. Downs SH, Black NA, Devlin HB, Royston CM, Russel RC. Systematicreview of the effectiveness and safety of laparoscopic cholecystectomy. Ann R CollSurg Engl. 1996;78:241-323.37. Abraham NS, Young JM, Solomon MJ. Meta-analysis of short-term out-comes after laparoscopic resection for colorectal cancer. Br J Surg. 2004;91:1111-24. [PMID: 15449261]38. Cheatham ML, Chapman WC, Key SP, Sawyers JL. A meta-analysis ofselective versus routine nasogastric decompression after elective laparotomy. AnnSurg. 1995;221:469-76; discussion 476-8. [PMID: 7748028]39. Nelson R, Tse B, Edwards S. Systematic review of prophylactic nasogastricdecompression after abdominal operations. Br J Surg. 2005;92:673-80. [PMID:15912492]40. Thomas JA, McIntosh JM. Are incentive spirometry, intermittent positivepressure breathing, and deep breathing exercises effective in the prevention ofpostoperative pulmonary complications after upper abdominal surgery? A system-atic overview and meta-analysis. Phys Ther. 1994;74:3-10; discussion 10-6.[PMID: 8265725]41. Overend TJ, Anderson CM, Lucy SD, Bhatia C, Jonsson BI, TimmermansC. The effect of incentive spirometry on postoperative pulmonary complications:a systematic review. Chest. 2001;120:971-8. [PMID: 11555536]42. Moore FA, Feliciano DV, Andrassy RJ, McArdle AH, Booth FV, Morgen-stein-Wagner TB, et al. Early enteral feeding, compared with parenteral, reducespostoperative septic complications. The results of a meta-analysis. Ann Surg.1992;216:172-83. [PMID: 1386982]43. Bluman LG, Mosca L, Newman N, Simon DG. Preoperative smokinghabits and postoperative pulmonary complications. Chest. 1998;113:883-9.[PMID: 9554620]44. Warner DO. Preventing postoperative pulmonary complications: the role ofthe anesthesiologist. Anesthesiology. 2000;92:1467-72. [PMID: 10781293]45. McCulloch TJ, Loadsman JA. Reduction of postoperative mortality andmorbidity. Little information was given on inclusion criteria [Letter]. BMJ. 2001;322:1182; author reply 1182-3. [PMID: 11379585]46. Pronovost PJ. Review: epidural or spinal anesthesia reduces postoperativemortality and morbidity. ACP J Club. 2001;135:1.47. Higham H, Mishra P, Foex P. Reduction of postoperative mortality andmorbidity. Research into modern anaesthesia techniques and perioperative med-icine is needed [Letter]. BMJ. 2001;322:1182-3. [PMID: 11379584]48. Karanikolas M, Kalauokalani D, Swarm R. Epidural anesthesia and analge-sia: is there really no benefit? [Letter]. Anesthesiology. 2002;97:1027; authorreply 1029-31. [PMID: 12357182]49. Norris EJ, Beattie C. In reply. Anesthesiology. 2002;97:1029-31.50. Liu SS. An intensive, structured clinical trial can markedly reduce length ofstay after abdominal aortic surgery [Letter]. Anesthesiology. 2002;97:1025; au-thor reply 1029-31. [PMID: 12357178]51. Amar D. Regional techniques and length of hospital stay after abdominalaortic surgery [Letter]. Anesthesiology. 2002;97:1029; author reply 1029-31.[PMID: 12357185]52. Todd MM. Clinical research manuscripts in Anesthesiology [Editorial]. An-esthesiology. 2001;95:1051-3. [PMID: 11684970]53. Peyton PJ, Myles PS, Silbert BS, Rigg JA, Jamrozik K, Parsons R. Periop-erative epidural analgesia and outcome after major abdominal surgery in high-riskpatients. Anesth Analg. 2003;96:548-54. [PMID: 12538211]54. Litz RJ, Hubler M, Koch T, Albrecht DM. Spinal-epidural hematomafollowing epidural anesthesia in the presence of antiplatelet and heparin therapy.Anesthesiology. 2001;95:1031-3. [PMID: 11605904]55. Porterfield WR, Wu CL. Epidural hematoma in an ambulatory surgicalpatient. J Clin Anesth. 1997;9:74-7. [PMID: 9051551]56. Horlocker TT. Low molecular weight heparin and neuraxial anesthesia.Thromb Res. 2001;101:V141-54. [PMID: 11342094]57. Zacks SL, Sandler RS, Rutledge R, Brown RS Jr. A population-based cohortstudy comparing laparoscopic cholecystectomy and open cholecystectomy. Am JGastroenterol. 2002;97:334-40. [PMID: 11866270]58. Guller U, Jain N, Hervey S, Purves H, Pietrobon R. Laparoscopic vs opencolectomy: outcomes comparison based on large nationwide databases. ArchSurg. 2003;138:1179-86. [PMID: 14609864]59. Nelson R, Edwards S, Tse B. Prophylactic nasogastric decompression after
Clinical GuidelinesStrategies To Reduce Postoperative Pulmonary Complications after Noncardiothoracic Surgery
www.annals.org 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 607
Downloaded From: http://annals.org/ on 10/30/2017
abdominal surgery. Cochrane Database Syst Rev. 2005:CD004929. [PMID:15674971]60. Detsky AS, Baker JP, O’Rourke K, Goel V. Perioperative parenteral nutri-tion: a meta-analysis. Ann Intern Med. 1987;107:195-203. [PMID: 3111322]61. Heyland DK, Novak F, Drover JW, Jain M, Su X, Suchner U. Shouldimmunonutrition become routine in critically ill patients? A systematic review ofthe evidence. JAMA. 2001;286:944-53. [PMID: 11509059]
62. Wilmore DW, Smith RJ, O’Dwyer ST, Jacobs DO, Ziegler TR, WangXD. The gut: a central organ after surgical stress. Surgery. 1988;104:917-23.[PMID: 3055397]63. Polanczyk CA, Rohde LE, Goldman L, Cook EF, Thomas EJ, MarcantonioER, et al. Right heart catheterization and cardiac complications in patients un-dergoing noncardiac surgery: an observational study. JAMA. 2001;286:309-14.[PMID: 11466096]
Clinical Guidelines Strategies To Reduce Postoperative Pulmonary Complications after Noncardiothoracic Surgery
608 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 www.annals.org
Downloaded From: http://annals.org/ on 10/30/2017
Current Author Addresses: Drs. Lawrence and Cornell: Medicine/Gen-eral Medicine, University of Texas Health Science Center at San Anto-nio, 7703 Floyd Curl Drive, Mail Code 7879, San Antonio, TX 78229-3900.
Dr. Smetana: Division of General Medicine and Primary Care, BethIsrael Deaconess Medical Center, 330 Brookline Avenue, Boston, MA02215.
Annals of Internal Medicine
www.annals.org 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 W-123
Downloaded From: http://annals.org/ on 10/30/2017
App
endi
xT
able
1.A
bstr
acte
dD
ata
for
Elig
ible
Ran
dom
ized
Tria
ls*
Aut
hor,
Yea
r(R
efer
ence
)In
terv
enti
onIn
terv
enti
onD
escr
ipti
onC
ontr
olD
escr
ipti
onTy
peof
Surg
ery
Incl
usio
nC
rite
ria
Excl
usio
nC
rite
ria
Pati
ents
Blin
ded
Out
com
eA
sses
smen
tB
linde
d
Mul
tice
nter
Ran
dom
izat
ion
Allo
cati
onC
once
alm
ent
ITT
Ana
lysi
s
Møl
ler
etal
.,20
02(1
2)Pr
eope
rativ
esm
okin
gce
ssat
ion
Wee
kly
mee
ting
with
proj
ect
nurs
e;es
timat
eof
nico
tine
depe
nden
ce;
free
,ta
ilore
dni
cotin
ere
plac
emen
t;sm
okin
gst
atus
mea
sure
dby
carb
onm
onox
ide
inex
pire
dai
r;ad
vice
abou
tce
ssat
ion
orre
duct
ion,
bene
fits,
side
effe
cts,
orm
anag
emen
tof
with
draw
alsy
mpt
oms
and
wei
ght
Stan
dard
care
:�li
ttle
orno
info
rmat
ion
abou
tris
kof
smok
ing
orce
ssat
ion
coun
selli
ng�
Elec
tive
knee
orhi
pre
plac
emen
tD
aily
smok
ers
Wee
kly
alco
holi
ntak
e�
35un
itsN
oY
esY
es:
3un
iver
sity
-af
filia
ted
hosp
itals
,C
open
hage
n,D
enm
ark
Ade
quat
eY
es
Berg
etal
.,19
97(1
3)Lo
ng-a
ctin
gvs
.in
term
edia
te-
actin
gne
urom
uscu
lar
bloc
kade
Panc
uron
ium
Atr
acur
ium
orve
curo
nium
Elec
tive
maj
orlo
wer
extr
emity
,gy
neco
logi
c,br
east
,or
abdo
min
alsu
rger
y
�Adu
lts,�
gene
ral
anes
thes
ia,
ASA
clas
sI–
III
Expe
cted
dura
tion
ofan
esth
esia
�60
min
,�
30%
idea
lbod
yw
eigh
t,ne
urom
uscu
lar
dise
ase,
preo
pera
tive
drug
saf
fect
ing
neur
omus
cula
rtr
ansm
issi
on,
rena
lor
liver
�insu
ffic
ienc
y,�
incr
ease
dpr
eope
rativ
ecr
eatin
ine
leve
l
No
data
Yes
Yes
:2
univ
ersi
ty-
affil
iate
dho
spita
ls,
Cop
enha
gen,
Den
mar
k
Ade
quat
eY
es
Nor
riset
al.,
2001
(14)
Gen
eral
anes
thes
iavs
.ge
nera
l�re
gion
alan
esth
esia
and
IVPC
Avs
.ep
idur
alPC
A
1)G
ener
al�
regi
onal
with
IVPC
A;
2)ge
nera
l�re
gion
alw
ithep
idur
alPC
A
3)G
ener
alw
ithIV
PCA
;4)
gene
ralw
ithep
idur
alPC
A
Elec
tive
abdo
min
alao
rtic
surg
ery
Elec
tive
abdo
min
alao
rtic
surg
ery
Cro
ss-c
lam
ping
ofth
orac
icao
rta,
cont
rain
dica
tion
toep
idur
al,
maj
orop
erat
ion
inth
epr
evio
us14
d,op
ioid
depe
nden
ce
Yes
Yes
No
Yes
Yes
Rig
get
al.,
2002
(15)
Intr
aope
rativ
e�
post
oper
ativ
eep
idur
alan
alge
sia
Intr
aope
rativ
ege
nera
lan
esth
esia
and
epid
ural
loca
lane
sthe
tic,
then
post
oper
ativ
eep
idur
allo
cala
nest
hetic
for
72h
with
as-n
eede
dop
ioid
(fen
tany
lor
peth
idin
e)
Intr
aope
rativ
ege
nera
lan
esth
esia
,th
enpo
stop
erat
ive
IVop
ioid
and
as-n
eede
dN
SAID
,op
ioid
,an
dpa
race
tam
ol
Elec
tive
maj
orno
nlap
aros
copi
cab
dom
inal
oper
atio
nsor
esop
hage
ctom
y
Age
�18
ySu
rger
y�
12h
afte
rad
mis
sion
,co
ntra
indi
catio
nto
epid
ural
bloc
k
No
No
Yes
:H
ospi
tals
inA
ustr
alia
,Ea
stA
sia,
and
Mid
dle
East
Yes
Yes
Park
etal
.,20
01(1
6)In
trao
pera
tive
epid
ural
anes
thes
ia�
post
oper
ativ
eep
idur
alan
alge
sia
Intr
aope
rativ
e�li
ght�
gene
ral
anes
thes
ia�
epid
ural
bupi
vaca
ine,
then
post
oper
ativ
eep
idur
alm
orph
ine
and
as-n
eede
dIV
opio
id
Intr
aope
rativ
ege
nera
lan
esth
esia
,th
enpo
stop
erat
ive
IVor
IMop
ioid
Elec
tive
aort
ic,
gast
ric,
bilia
ry,
orco
lon
oper
atio
ns
Vet
eran
s,ag
e�
21y
Age
�21
y;w
omen
;co
nfus
ion;
ASA
clas
sI,
II,or
V;
surg
ery
�12
haf
ter
adm
issi
on;
myo
card
iali
nfar
ctio
n�
6m
o;pr
evio
usab
dom
inal
surg
ery
�3
mo;
prev
ious
aort
icpr
oced
ure;
trac
heos
tom
yor
endo
trac
heal
tube
;ch
emo-
orim
mun
osup
pres
sion
ther
apy
othe
rth
anco
rtic
oste
roid
s;st
udy
drug
hype
rsen
sitiv
ity;
cont
rain
dica
tion
toep
idur
al;
phys
icia
nre
fusa
l;en
rolle
din
othe
rV
eter
ans
Aff
airs
stud
y
No
Yes
Yes
:15
Vet
eran
sA
ffai
rsho
spita
lsY
esY
es
Fler
onet
al.,
2003
(17)
Intr
aope
rativ
eep
idur
alop
ioid
Intr
aope
rativ
ege
nera
lan
esth
esia
and
epid
ural
sufe
ntan
ilan
dm
orph
ine
Intr
aope
rativ
ege
nera
lan
esth
esia
and
IVsu
fent
anil
Elec
tive
abdo
min
alao
rtic
surg
ery
Con
secu
tive
patie
nts
for
elec
tive
abdo
min
alao
rtic
surg
ery
Con
trai
ndic
atio
nto
epid
ural
No
No
No
Yes
No
Man
net
al.,
2000
(18)
Gen
eral
anes
thes
ia�
epid
ural
anes
thes
iath
enPC
EA
Intr
aope
rativ
ege
nera
lan
esth
esia
and
epid
ural
loca
lane
sthe
ticor
opio
id,
then
post
oper
ativ
ePC
EA�
opio
id
Intr
aope
rativ
ege
nera
lan
esth
esia
,th
enpo
stop
erat
ive
patie
nt-c
ontr
olle
dIV
opio
id
Elec
tive
maj
orab
dom
inal
surg
ery
for
mal
igna
ntco
nditi
on
Age
�70
y,A
SAcl
ass
Ior
II,no
rmal
preo
pera
tive
men
tals
tatu
s,no
seve
rem
alnu
triti
onor
cere
brov
ascu
lar
insu
ffic
ienc
y,no
cont
rain
dica
tion
toep
idur
al
No
addi
tiona
ldat
aN
oY
esfo
rch
est
radi
ogra
phy,
uncl
ear
for
othe
rou
tcom
es
No
Yes
No
Con
tinue
don
follo
win
gpa
ge
W-124 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 www.annals.org
Downloaded From: http://annals.org/ on 10/30/2017
App
endi
xT
able
1—C
onti
nued
Aut
hor,
Yea
r(R
efer
ence
)In
terv
enti
onIn
terv
enti
onD
escr
ipti
onC
ontr
olD
escr
ipti
onTy
peof
Surg
ery
Incl
usio
nC
rite
ria
Excl
usio
nC
rite
ria
Pati
ents
Blin
ded
Out
com
eA
sses
smen
tB
linde
d
Mul
tice
nter
Ran
dom
izat
ion
Allo
cati
onC
once
alm
ent
ITT
Ana
lysi
s
Cus
chie
riet
al.,
1985
(19)
Gen
eral
anes
thes
ia�
intr
aope
rativ
ean
dpo
stop
erat
ive
loca
lan
esth
etic
Intr
aope
rativ
ege
nera
lan
esth
esia
and
loca
lan
esth
etic
,th
enpo
stop
erat
ive
epid
ural
loca
lane
sthe
ticfo
r12
h,th
enas
-nee
ded
IMm
orph
ine
Intr
aope
rativ
ege
nera
lan
esth
esia
,th
enas
-nee
ded
IVco
ntin
uous
mor
phin
eor
inte
rmitt
ent
IMm
orph
ine
Elec
tive
chol
ecys
tect
omy
Age
�75
yN
oad
ditio
nali
nfor
mat
ion
No
Unc
lear
No
Unc
lear
Yes
Kar
ayia
nnak
iset
al.,
1996
(20)
LCLC
with
4-tr
ocar
tech
niqu
eO
Cw
ithtr
ansv
erse
right
subc
osta
linc
isio
nEl
ectiv
ech
olec
yste
ctom
yA
llpa
tient
sw
ithsy
mpt
omat
icch
olel
ithia
sis
Acu
tech
olec
ystit
is,
chol
edoc
holit
hias
is,
age
�65
y,bo
dym
ass
inde
x�
29kg
/m2 ,
hist
ory
ofpu
lmon
ary
dise
ase,
�10
ciga
rett
es/d
No
Yes
for
ches
tra
diog
raph
yNo
Yes
No
Vig
nali
etal
.,20
04(2
1)LC
RLC
RO
CR
Elec
tive
colo
rect
alre
sect
ion
Age
�18
yan
dca
ndid
ate
for
lapa
rosc
opic
rese
ctio
n
Can
cer
infil
trat
ing
adja
cent
orga
nsby
imag
ing,
NY
HA
clas
s�
3,ar
teria
lPO
2�
70m
mH
g,liv
erdy
sfun
ctio
n(C
hild
–Pug
hcl
ass
C),
�ong
oing
infe
ctio
n,�
neut
roph
ilco
unt
�2
�10
9ce
lls/L
No
Yes
No
Yes
Yes
Chu
mill
aset
al.,
1998
(22)
�Res
pira
tory
reha
bilit
atio
n�w
ithbr
eath
ing
exer
cise
s
Inst
ruct
ion
abou
tfo
rced
expi
ratio
nte
chni
que
and
coug
h,ch
est
expa
nsio
nex
erci
ses
and
diap
hrag
mm
obili
zatio
n,m
axim
umin
spira
tion
for
3–5
s,an
dea
rlyam
bula
tion
afte
rsu
rger
y;ex
erci
ses
wer
edo
nefo
r10
–15
min
QID
befo
resu
rger
yan
dfo
r10
min
ever
y2
hon
post
oper
ativ
eda
ys1
and
2
No
desc
riptio
nEl
ectiv
esu
prau
mbi
lical
lapa
roto
my
Age
�15
yN
otst
ated
apr
iori
No
Unc
lear
No
Unc
lear
No
Fage
vik
Ols
enet
al.,
1997
(23)
Che
stph
ysio
ther
apy
Preo
pera
tive
trai
ning
inD
BEs
with
purs
edlip
s,hu
ffin
g,co
ughi
ngev
ery
tent
hbr
eath
hour
lyaf
ter
surg
ery;
impo
rtan
ceof
posi
tion
chan
gein
bed
and
early
mob
iliza
tion;
high
-ris
kpa
tient
sw
ere
also
give
nPE
P;po
stop
erat
ive
ther
apy
was
adju
sted
per
phys
ioth
erap
ist
orph
ysic
ian
�acc
ordi
ngto
pulm
onar
yst
atus
�;du
ratio
nof
trea
tmen
tw
as10
–15
min
befo
rean
d15
–20
min
afte
rsu
rger
y
No
trai
ning
befo
resu
rger
y,no
ther
apy
afte
rsu
rger
yun
less
pulm
onar
yco
mpl
icat
ions
occu
rred
,th
enpa
tient
sgi
ven
phys
ioth
erap
yw
ithPE
P
Elec
tive
open
abdo
min
alsu
rger
y
No
data
No
data
No
No
data
No
No:
�[T]
oav
oid
patie
ntin
terf
eren
ce,
clus
ter
rand
omiz
atio
nw
aspe
rfor
med
inal
tern
ate
mon
ths�
Unc
lear
Hal
let
al.,
1991
(24)
ISvs
.ch
est
phys
ioth
erap
yIS
devi
ceon
beds
ide
tabl
e,pa
tient
inst
ruct
edin
use;
atte
ndin
gph
ysic
ians
’re
spon
sibi
lity
topr
omot
eits
use
perio
pera
tivel
y,pr
efer
ably
for
5m
inin
each
wak
ing
hour
Trea
tmen
tac
cord
ing
toth
ecl
inic
alju
dgm
ent
ofat
tend
ing
phys
icia
nsan
dph
ysio
ther
apis
ts
Lapa
roto
my
�with
man
ipul
atio
nof
visc
era�
Lapa
roto
my
�with
man
ipul
atio
nof
visc
era�
Age
�14
y,pr
eope
rativ
epu
lmon
ary
com
plic
atio
npe
rou
tcom
ede
finiti
ons
No
Yes
No
Yes
Yes
Hal
let
al.,
1996
(25)
Low
-ris
kpa
tient
s:1)
ISvs
.2)
DBE
s;hi
gh-r
isk
patie
nts:
3)IS
vs.
4)IS
�ch
est
phys
ioth
erap
y
1)an
d3)
IS:
info
rmat
ion
shee
tan
den
cour
agem
ent
tous
eIS
atle
ast
10tim
esho
urly
2)D
BEs:
seen
once
and
enco
urag
edto
take
10de
epbr
eath
sho
urly
;4)
IS�
ches
tph
ysio
ther
apy
had
phys
ioth
erap
yai
med
atpr
oduc
ing
max
imum
insp
irato
ryef
fort
atle
ast
once
daily
for
post
oper
ativ
eda
ys1–
3,th
enat
furt
her
rate
per
phys
ioth
erap
ist
Lapa
roto
my
�with
man
ipul
atio
nof
visc
era�
Lapa
roto
my
�with
man
ipul
atio
nof
visc
era�
;hi
gh-r
isk
patie
nts
defin
edas
ASA
clas
s�
Ior
age
�60
y
Lang
uage
,pu
lmon
ary
com
plic
atio
nbe
fore
surg
ery,
lack
ofco
nsen
tdu
eto
decl
ined
part
icip
atio
nor
insu
ffic
ient
time
befo
resu
rger
y
No
Yes
No
Yes
Yes
Con
tinue
don
follo
win
gpa
ge
www.annals.org 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 W-125
Downloaded From: http://annals.org/ on 10/30/2017
App
endi
xT
able
1—C
onti
nued
Aut
hor,
Yea
r(R
efer
ence
)In
terv
enti
onIn
terv
enti
onD
escr
ipti
onC
ontr
olD
escr
ipti
onTy
peof
Surg
ery
Incl
usio
nC
rite
ria
Excl
usio
nC
rite
ria
Pati
ents
Blin
ded
Out
com
eA
sses
smen
tB
linde
d
Mul
tice
nter
Ran
dom
izat
ion
Allo
cati
onC
once
alm
ent
ITT
Ana
lysi
s
Bohn
eret
al.,
2002
(26)
Nas
alC
PAP
for
12h
afte
rsu
rger
yIn
term
edia
teca
refo
rat
leas
t24
haf
ter
surg
ery
with
nasa
lCPA
Pat
10cm
H2O
for
atle
ast
12h
afte
rsu
rger
yto
keep
O2
satu
ratio
n�
95%
Inte
rmed
iate
care
for
atle
ast
24h
afte
rsu
rger
yw
ithox
ygen
byfa
cem
ask
orna
salc
annu
lato
keep
oxyg
ensa
tura
tion
�95
%
Elec
tive
mid
line
abdo
min
alva
scul
arsu
rger
y
Elec
tive
mid
line
abdo
min
alva
scul
arsu
rger
ysc
hedu
led
tobe
extu
bate
din
the
oper
atin
gro
om
Emer
genc
ysu
rger
y,re
pair
ofth
orac
oabd
omin
alan
eury
sm,
excl
usiv
ely
retr
oper
itone
alap
proa
ch
No
No
data
No
Yes
Yes
VA
TPN
Coo
pera
tive
Stud
yG
roup
,19
91(2
7)
Preo
pera
tive
TPN
TPN
toca
loric
goal
of10
00kc
al�
rest
ing
met
abol
icex
pend
iture
;op
timal
was
7–15
dbe
fore
surg
ery;
subo
ptim
alw
as�
7d;
TPN
cont
inue
d72
haf
ter
surg
ery;
oral
inta
keas
clin
ical
lyal
low
edor
tole
rate
d
No
TPN
orfo
rced
ente
ral
inta
kebe
fore
surg
ery
oraf
ter
surg
ery
for
72h,
then
TPN
oren
tera
lif
indi
cate
d
Elec
tive
lapa
roto
my
orth
orac
otom
yA
ge�
21y,
elec
tive
lapa
roto
my
orth
orac
otom
y
Dea
thex
pect
edw
ithin
90d,
TPN
inpr
evio
us15
d,ot
her
surg
ery
with
inpr
evio
us30
d,co
ntra
indi
catio
nto
dela
yin
surg
ery
for
TPN
,TP
Nco
ntra
indi
cate
d,m
ajor
conc
urre
ntill
ness
,TP
Nes
sent
ial,
wel
l-no
uris
hed
No
No
Yes
Yes
Yes
Pace
lliet
al.,
2001
(28)
TEN
vs.
TPN
Post
oper
ativ
eTE
N(d
urin
gin
duct
ion
ofTE
N,
TPN
was
adde
dto
achi
eve
the
sam
eca
loric
inta
keas
inth
eTP
Ngr
oup)
Post
oper
ativ
eTP
NM
ajor
elec
tive
abdo
min
alop
erat
ions
Age
18–8
0y,
mal
nour
ishe
d,nu
triti
onal
risk
inde
x�
90%
Emer
genc
ysu
rger
y,el
ectiv
eap
pend
ecto
my,
chol
ecys
tect
omy
orvi
scer
olys
is
No
Unc
lear
Yes
Yes
Yes
Bozz
etti
etal
.,20
01(2
9)
TEN
vs.
TPN
Post
oper
ativ
eTE
N:
jeju
nost
omy
feed
ing
tube
orna
soje
juna
lfee
ding
tube
plac
eddu
ring
surg
ery
Post
oper
ativ
eTP
NEl
ectiv
em
ajor
rese
ctio
nfo
rga
stro
inte
stin
alca
ncer
Wei
ght
loss
�10
%us
ualb
ody
wei
ght
inpr
evio
us6
mo,
hist
olog
ical
lypr
oven
canc
er
Age
�18
y,liv
erdy
sfun
ctio
n(C
hild
–Pug
hcl
ass
�2)
,se
rum
crea
tinin
ele
vel�
265.
2�
mol
/L(�
3m
g/dL
)or
hem
odia
lysi
s,N
YH
Afu
nctio
nalc
lass
�III
,hi
stor
yof
stro
ke,
preg
nanc
y,on
goin
gin
fect
ion,
prev
ious
inte
stin
alan
asto
mos
isof
the
larg
ebo
wel
with
out
adi
vert
ing
stom
a
No
No
Yes
Yes
Yes
Gia
nott
iet
al.,
2002
(30)
Ente
ral
imm
unon
utrit
ion
1)O
rali
mm
unon
utrit
ion
for
5d
befo
resu
rger
y,no
supp
lem
enta
tion
afte
rsu
rger
y;2)
oral
imm
unon
utrit
ion
for
5d
befo
resu
rger
yan
dby
jeju
nalf
eedi
ngaf
ter
surg
ery
until
oral
inta
kere
sum
ed
3)U
sual
care
,no
ente
ral
supp
lem
ent,
IV5%
gluc
ose
and
elec
trol
ytes
Elec
tive
maj
orre
sect
ion
for
gast
roin
test
inal
canc
er
His
tolo
gica
llypr
oven
canc
erW
eigh
tlo
ss�
10%
ofus
ualb
ody
wei
ght
inpa
st6
mo,
age
�18
y,liv
erdy
sfun
ctio
n(C
hild
–Pug
hcl
ass
�B)
,Pa
O2
�70
mm
Hg,
crea
tinin
ele
vel�
265.
2�
mol
/L(�
3m
g/dL
)or
hem
odia
lysi
s,N
YH
Acl
ass
�3,
Kar
nofs
ky�
60,
preg
nanc
y,on
goin
gin
fect
ion,
neoa
djuv
ant
radi
oche
mot
hera
pyor
neut
roph
ilco
unt
�2
�10
9ce
lls/L
No
Unc
lear
No
Yes
Yes
Sand
ham
etal
.,20
03(3
1)
Pulm
onar
yar
tery
cath
eter
inhi
gh-r
isk
surg
ical
patie
nts
Pulm
onar
yar
tery
cath
eter
plac
edbe
fore
surg
ery;
trea
tmen
tdi
rect
edto
apr
iori
esta
blis
hed
phys
iolo
gic
goal
san
dpr
iorit
ies
No
pulm
onar
yar
tery
cath
eter
,ce
ntra
lven
ous
cath
eter
allo
wed
Urg
ent
and
elec
tive
maj
orab
dom
inal
,th
orac
ic,
vasc
ular
,or
hip
frac
ture
surg
ery
Age
�60
y,A
SAcl
ass
IIIor
IVN
oad
ditio
nali
nclu
sion
crite
riaN
oY
esY
esY
esY
es
*A
SA�
Am
eric
anSo
ciet
yof
Ane
sthe
siol
ogis
ts;C
PAP
�co
ntin
uous
posi
tive
airw
aypr
essu
re;D
BE
�de
epbr
eath
ing
exer
cise
;IM
�in
tram
uscu
lar;
IS�
ince
ntiv
esp
irom
etry
;IT
T�
inte
ntio
n-to
-tre
at;I
V�
intr
aven
ous;
LC�
lapa
rosc
opic
chol
ecys
tect
omy;
LCR
�la
paro
scop
icco
lore
ctal
rese
ctio
n;N
SAID
�no
nste
roid
alan
ti-i
nflam
mat
ory
drug
;NY
HA
�N
ewY
ork
Hea
rtA
ssoc
iati
on;O
C�
open
chol
ecys
tect
omy;
OC
R�
open
colo
rect
alre
sect
ion;
PCA
�pa
tien
t-co
ntro
lled
anal
gesi
a;PC
EA
�pa
tien
t-co
ntro
lled
epid
ural
anal
gesi
a;PE
P�
posi
tive
expi
rato
rypr
essu
re;
QID
�fo
urti
mes
daily
;T
EN
�to
tal
ente
ral
nutr
itio
n;T
PN�
tota
lpa
rent
eral
nutr
itio
n;V
AT
PN�
Vet
eran
sA
ffai
rsT
otal
Pare
nter
alN
utri
tion
.
W-126 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 www.annals.org
Downloaded From: http://annals.org/ on 10/30/2017
App
endi
xT
able
2.A
bstr
acte
dD
ata
for
Elig
ible
Ran
dom
ized
Tria
ls,
Con
tinu
ed*
Aut
hor,
Yea
r(R
efer
ence
)Sa
mpl
ing
Stra
tegy
Com
orbi
dC
ondi
tion
sIm
port
ant
Bas
elin
eD
iffe
renc
es?
Impo
rtan
tIn
trao
pera
tive
Dif
fere
nces
?A
nest
hesi
aFo
llow
-up
Impo
rtan
tD
iffe
renc
esin
Co-
Inte
rven
tions
Rel
evan
tfo
rPu
lmon
ary
Com
plic
atio
ns?
Cro
ssov
ers
Part
icip
ant
Acc
rual
Part
icip
ant
Att
riti
on
Møl
ler
etal
.,20
02(1
2)C
onse
cutiv
eC
hron
iche
art
dise
ase,
chro
nic
obst
ruct
ive
lung
dise
ase,
diab
etes
No
Non
ere
gard
ing
perc
enta
geof
patie
nts
rece
ivin
gge
nera
lane
sthe
sia,
dura
tion
ofsu
rger
y,or
num
ber
ofhi
por
knee
oper
atio
ns
Gen
eral
orre
gion
alH
ospi
tals
tay
No
No
166
elig
ible
,46
decl
ined
,12
0ra
ndom
lyas
sign
ed4
(inte
rven
tion)
and
8(c
ontr
ol):
oper
atio
nde
laye
dor
canc
eled
Berg
etal
.,19
97(1
3)N
oda
taSm
okin
gst
atus
,pu
lmon
ary
dise
ase
No;
tren
dsto
war
dm
ore
smok
ers,
drop
erid
ol,
and
inha
led
anes
thet
icin
the
panc
uron
ium
grou
p
Sign
ifica
ntly
mor
ean
esth
etic
min
utes
,re
sidu
alne
urom
uscu
lar
bloc
kan
dm
inut
esto
extu
batio
nw
ithpa
ncur
oniu
mco
mpa
red
with
vecu
roni
umor
atra
curiu
m
Gen
eral
Post
oper
ativ
eda
y6
No
No
No
data
othe
rth
an69
3pa
tient
sel
igib
lean
den
rolle
d2
prot
ocol
viol
atio
ns
Nor
riset
al.,
2001
(14)
Con
secu
tive?
Dia
bete
s,hy
pert
ensi
on,
rena
lin
suff
icie
ncy,
card
iova
scul
aran
dpe
riphe
ralv
ascu
lar
dise
ase,
smok
ing
No
Gen
eral
anes
thes
iaas
soci
ated
with
sign
ifica
ntly
less
oper
atio
nan
dcr
oss-
clam
ping
time
com
pare
dw
ithge
nera
l�re
gion
alan
esth
esia
Post
oper
ativ
eda
y7
and
1,3,
6,an
d12
mo
for
mor
talit
y
No
No
309
eval
uate
d,62
inel
igib
le,
48de
clin
ed,
24�a
dmin
istr
ativ
eex
clus
ions
,�17
6co
nsen
ted,
7no
tra
ndom
lyas
sign
eddu
eto
faile
dep
idur
al,
8ra
ndom
lyas
sign
edto
pilo
tst
udy,
160
rand
omly
assi
gned
tore
port
edst
udy
Non
e
Rig
get
al.,
2002
(15)
Con
secu
tive?
Mor
bid
obes
ity,
diab
etes
,ch
roni
cre
nalf
ailu
re,
card
iac
failu
re,
resp
irato
ryin
suff
icie
ncy,
acut
em
yoca
rdia
lin
farc
tion,
exer
tiona
lang
ina,
myo
card
ial
isch
emia
,se
vere
liver
dise
ase
No
No
data
30d
for
mor
talit
y,co
mpl
icat
ions
;ho
spita
lsta
y?
No
data
Unc
lear
how
anal
gesi
aw
asha
ndle
dfo
r22
2pa
tient
sas
sign
edto
epid
ural
but
not
fully
adhe
rent
topr
otoc
ol
920
rand
omly
assi
gned
,32
excl
uded
afte
rra
ndom
izat
ion
32ex
clud
edaf
ter
rand
omiz
atio
n:5
ente
red
inst
udy
for
seco
ndop
erat
ion
and
27in
elig
ible
orca
ncel
edsu
rger
y
Park
etal
.,20
01(1
6)C
onse
cutiv
e?A
SAcl
ass
IIor
III,
Gol
dman
card
iac
risk
inde
x,pr
evio
usan
gina
,m
yoca
rdia
lin
farc
tion,
cong
estiv
ehe
art
failu
re,
hype
rten
sion
,ch
roni
cob
stru
ctiv
elu
ngdi
seas
e,di
abet
es,
rena
lfa
ilure
,ce
rebr
ovas
cula
rac
cide
nt,
curr
ent
smok
ing,
hist
ory
ofal
coho
lism
,al
coho
licliv
erdi
seas
e
No
No
diff
eren
ces
inop
erat
ion
seve
rity
ordu
ratio
n30
dG
roup
sre
ceiv
edsi
mila
ran
tibio
ticpr
ophy
laxi
s,bo
wel
prep
arat
ion,
and
intr
aope
rativ
em
onito
ring
Con
trol
toin
terv
entio
n:48
patie
nts;
inte
rven
tion
toco
ntro
l:32
patie
nts
2731
scre
ened
,13
60ex
clud
ed,
350
decl
ined
,10
21ra
ndom
lyas
sign
ed
26su
rger
ies
canc
eled
,11
with
draw
als
Fler
onet
al.,
2003
(17)
Con
secu
tive
Cor
onar
yar
tery
dise
ase,
hype
rten
sion
,co
nges
tive
hear
tfa
ilure
,ch
roni
cob
stru
ctiv
elu
ngdi
seas
e,di
abet
es
No
No
30d
No
Lum
bar
punc
ture
coul
dno
tbe
done
in3
patie
nts
inep
idur
algr
oup:
They
wer
esw
itche
dto
the
cont
rolg
roup
for
anal
ysis
�217
patie
nts
who
met
all
incl
usio
ncr
iteria
wer
era
ndom
ized
�
Man
net
al.,
2000
(18)
Con
secu
tive?
Cor
onar
yar
tery
dise
ase,
diab
etes
,hy
pert
ensi
on,
chro
nic
obst
ruct
ive
lung
dise
ase,
depr
essi
on
No
Patie
nts
rand
omly
assi
gned
toep
idur
alha
dsi
gnifi
cant
lyle
sssu
fent
anil
and
isof
lura
ne,
sign
ifica
ntly
mor
eep
hedr
ine,
and
sign
ifica
ntly
long
ertim
eto
extu
batio
n
Clin
ical
:da
ilyth
roug
hpo
stop
erat
ive
day
7;ch
est
radi
ogra
phy:
post
oper
ativ
eda
ys1,
3,5
No
No
data
108
eval
uate
d,4
decl
ined
,34
inel
igib
le,
70ra
ndom
lyas
sign
ed
4pa
tient
s:no
surg
ical
rese
ctio
n;2
patie
nts:
decl
ined
tous
epa
tient
-con
trol
led
anes
thes
ia
Con
tinue
don
follo
win
gpa
ge
www.annals.org 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 W-127
Downloaded From: http://annals.org/ on 10/30/2017
App
endi
xT
able
2—C
onti
nued
Aut
hor,
Yea
r(R
efer
ence
)Sa
mpl
ing
Stra
tegy
Com
orbi
dC
ondi
tion
sIm
port
ant
Bas
elin
eD
iffe
renc
es?
Impo
rtan
tIn
trao
pera
tive
Dif
fere
nces
?A
nest
hesi
aFo
llow
-up
Impo
rtan
tD
iffe
renc
esin
Co-
Inte
rven
tions
Rel
evan
tfo
rPu
lmon
ary
Com
plic
atio
ns?
Cro
ssov
ers
Part
icip
ant
Acc
rual
Part
icip
ant
Att
riti
on
Cus
chie
riet
al.,
1985
(19)
Con
secu
tive?
Wei
ght,
smok
ing,
resp
irato
rydi
seas
eN
oN
ofo
rdu
ratio
nof
anes
thes
ia,
tren
dto
war
dm
ore
intr
aope
rativ
eop
ioid
san
dlo
nger
time
tofir
stdo
seof
post
oper
ativ
ean
alge
sia
inIM
mor
phin
egr
oup
Post
oper
ativ
eda
y3
No
data
4fa
iled
epid
ural
cath
eter
plac
emen
tan
dre
ceiv
edIM
mor
phin
e
�75
patie
nts
wer
ein
clud
edin
the
stud
y,25
inea
chgr
oup�
Non
e
Kar
ayia
nnak
iset
al.,
1996
(20)
Con
secu
tive
Non
ere
port
edN
ofo
rag
e,se
x,w
eigh
t,he
ight
,or
othe
rba
selin
eda
tare
port
ed
No
for
oper
atio
nan
dan
esth
esia
time
but
tren
dto
war
dsh
orte
rop
erat
ion
time
(97
min
�19
min
vs.
108
min
�23
min
)an
dan
esth
esia
time
(116
min
�15
min
vs.
139
min
�18
min
)w
ithLC
Gen
eral
Hos
pita
lsta
yN
o3
rand
omly
assi
gned
toLC
requ
ired
conv
ersi
onto
open
oper
atio
nan
dw
ere
excl
uded
from
anal
ysis
147
elig
ible
,49
decl
ined
,98
rand
omly
assi
gned
7(2
LCs,
5O
Cs)
decl
ined
afte
rra
ndom
izat
ion;
2LC
sex
clud
eddu
eto
inco
mpl
ete
pulm
onar
yte
sts;
3LC
sex
clud
eddu
eto
conv
ersi
onto
OC
;4
OC
sex
clud
eddu
eto
com
mon
bile
duct
expl
orat
ion
Vig
nali
etal
.,20
04(2
1)C
onse
cutiv
eA
SAcl
ass,
wei
ght
loss
�10
%,
canc
erPa
tient
sha
ving
OC
Rw
ere
olde
rth
anth
ose
havi
ngLC
R(6
2y
�13
.4y
vs.
66y
�12
.2y;
P�
0.02
);no di
ffer
ence
sfo
rot
her
repo
rted
com
orbi
dco
nditi
ons
No
for
intr
aope
rativ
etr
ansf
usio
nam
ount
,tu
mor
stag
e,or
reas
onfo
rop
erat
ion;
LCR
was
asso
ciat
edw
ithlo
nger
oper
atio
ntim
e(2
21m
in�
69m
invs
.17
8m
in�
68m
in;
P�
0.00
01),
less
intr
aope
rativ
ebl
ood
loss
(177
mL
�20
0m
Lvs
.26
4m
L�
292
mL;
P�
0.01
),le
ssfr
eque
nttr
ansf
usio
n(1
7%vs
.42
%;
P�
0.00
01).
Gen
eral
�th
orac
icep
idur
al
30d
afte
rdi
scha
rge
with
wee
kly
offic
evi
sits
No
10ra
ndom
lyas
sign
edto
LCR
requ
ired
conv
ersi
onto
OC
R
384
rand
omly
assi
gned
Non
e
Chu
mill
aset
al.,
1998
(22)
Con
secu
tive
No
data
No
data
exce
pt�n
osi
gnifi
cant
diff
eren
ces
insa
mpl
ech
arac
teris
tics
orris
kfa
ctor
sof
both
grou
ps,
incl
udin
gpr
eope
rativ
ech
est
x-ra
y,w
ithth
eex
cept
ion
ofse
xdi
strib
utio
n�
No
data
exce
pt�n
osi
gnifi
cant
diff
eren
ces
inav
erag
eop
erat
ion
dura
tion
orty
pes
ofin
cisi
on�
Gen
eral
Post
oper
ativ
eda
y6
No
No
data
115
rand
omly
assi
gned
,34
excl
uded
(em
erge
ncy
oper
atio
n,ex
trap
ulm
onar
yco
mpl
icat
ions
,in
frau
mbi
lical
exte
nsio
nof
inci
sion
,pa
tient
coop
erat
ion
with
inte
rven
tion)
,81
�eva
luab
lepa
tient
s�
Fage
vik
Ols
enet
al.,
1997
(23)
Con
secu
tive
Ove
rwei
ght
smok
ers
with
high
-ris
kA
SAcl
ass
�No
sign
ifica
ntdi
ffer
ence
sin
back
grou
ndva
riabl
es�
No
sign
ifica
ntdi
ffer
ence
indi
strib
utio
nof
oper
atio
nty
pes
but
tren
dto
war
dm
ore
uppe
rab
dom
inal
oper
atio
nsin
cont
rolg
roup
(30%
vs.
39%
)
Gen
eral
Hos
pita
lsta
yN
oda
taN
oda
ta36
8ra
ndom
lyas
sign
ed4
nonc
ompl
etio
ns:
2co
ntro
l,2
trea
tmen
t
Hal
let
al.,
1991
(24)
Con
secu
tive
Preo
pera
tive
hosp
ital
stay
�3
d,cu
rren
tsm
oker
,ch
roni
cbr
onch
itis,
abno
rmal
ches
tra
diog
raph
,P O
2�
80m
mH
g,A
SAcl
ass
No
No
for
anes
thes
iatim
e,di
strib
utio
nam
ong
12su
rgeo
ns,
intr
aper
itone
alin
fect
ion,
type
ofin
cisi
on,
post
oper
ativ
ena
soga
stric
deco
mpr
essi
on
Unc
lear
,pr
esum
ably
allg
ener
al
Hos
pita
lsta
yN
oda
taN
oda
ta10
32sc
reen
ed;
156
excl
uded
(5�
14y
ofag
e,3
reta
rdat
ion,
8la
ngua
ge,
14de
clin
ed,
25pr
eope
rativ
epu
lmon
ary
com
plic
atio
n,10
1in
suff
icie
nttim
eto
cons
ent)
;87
6ra
ndom
lyas
sign
ed
35ra
ndom
lyas
sign
edpa
tient
sdi
dno
tha
vesu
rger
y:21
IS,
14ch
est
phys
ioth
erap
y
Hal
let
al.,
1996
(25)
Con
secu
tive
ASA
clas
s,ca
ncer
,cu
rren
tsm
oker
,ch
roni
cbr
onch
itis
No
No
for
oper
atio
ntim
e,pr
oced
ure
type
,in
trap
erito
neal
infe
ctio
n,in
cisi
onty
peor
leng
th,
reop
erat
ion,
naso
gast
ricde
com
pres
sion
,ep
idur
alan
alge
sia
No
data
,pr
esum
ably
allg
ener
al,
som
ew
ithad
ditio
nal
epid
ural
anes
thes
ia
Hos
pita
lsta
yN
oda
taN
oda
ta61
9sc
reen
ed;
143
excl
uded
(13
lang
uage
,14
decl
ined
,15
preo
pera
tive
pulm
onar
yco
mpl
icat
ion,
115
lack
ofco
nsen
t);
476
rand
omly
assi
gned
20ra
ndom
lyas
sign
edpa
tient
sdi
dno
tha
vesu
rger
y
Con
tinue
don
follo
win
gpa
ge
W-128 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 www.annals.org
Downloaded From: http://annals.org/ on 10/30/2017
App
endi
xT
able
2—C
onti
nued
Aut
hor,
Yea
r(R
efer
ence
)Sa
mpl
ing
Stra
tegy
Com
orbi
dC
ondi
tion
sIm
port
ant
Bas
elin
eD
iffe
renc
es?
Impo
rtan
tIn
trao
pera
tive
Dif
fere
nces
?A
nest
hesi
aFo
llow
-up
Impo
rtan
tD
iffe
renc
esin
Co-
Inte
rven
tions
Rel
evan
tfo
rPu
lmon
ary
Com
plic
atio
ns?
Cro
ssov
ers
Part
icip
ant
Acc
rual
Part
icip
ant
Att
riti
on
Bohn
eret
al.,
2002
(26)
Con
secu
tive
Smok
ing,
coro
nary
hear
tdi
seas
e,pu
l-m
onar
ydi
seas
e,A
SAcl
ass
No
No
for
dura
tion
ofsu
rger
y,bl
ood
loss
,cr
ysta
lloid
,tr
ansf
usio
n,au
totr
ansf
u-si
on,
hypo
tens
ion,
hype
r-te
nsio
n,ox
ygen
atio
n
No
data
,pr
esum
ably
gene
ral
Hos
pita
lsta
yN
o9
patie
nts
did
not
tole
rate
nasa
lC
PAP
237
rand
omly
assi
gned
33ex
clud
edaf
ter
rand
omiz
atio
nfo
rin
elig
ible
surg
ery
orpa
tient
coul
dno
tbe
extu
bate
din
oper
atin
gro
om:
17na
sal
CPA
P,16
cont
rol
VA
TPN
Coo
pera
tive
Stud
yG
roup
,19
91(2
7)
Con
secu
tive
Surg
ical
diag
nosi
s,nu
triti
onal
stat
us,
perc
enta
geof
usua
lbod
yw
eigh
t,se
rum
albu
min
leve
l,se
rum
prea
l-bu
min
leve
l,tr
icep
ssk
info
ld,
nutr
ition
risk
inde
xsc
ore,
subj
ectiv
egl
obal
asse
ssm
ent
No
diff
eren
ces
exce
ptlo
wer
seru
mal
bu-
min
leve
l(3
.65
�3.
6m
g/dL
vs.
3.71
�3.
7m
g/dL
;P
�0.
06†)
and
nutr
ition
alris
kin
dex
scor
e(9
2.3
�6.
4vs
.93
.8�
6.0;
P�
0.01
),m
ore
seve
rem
alnu
triti
on(1
5%vs
.9%
;P
�0.
03)
inTP
Ngr
oup
No
data
No
data
,pr
esum
ably
gene
ral
30d
afte
rsu
rger
yN
oda
taO
f19
2ra
ndom
lyas
-si
gned
toTP
N:
130
rece
ived
optim
alTP
N,
49re
ceiv
edsu
bopt
imal
TPN
,an
d13
rece
ived
noTP
N;
of20
3ra
n-do
mly
assi
gned
toco
ntro
l,3
rece
ived
TPN
3259
scre
ened
,81
1ex
clud
ed,
1497
did
not
mee
tnu
triti
oncr
iteria
,16
9no
surg
ery,
323
decl
ined
,45
9ra
ndom
lyas
-si
gned
Of
459
rand
omly
assi
gned
,64
did
not
have
surg
ery
(n�
395
stud
ypa
tient
s)
Pace
lliet
al.,
2001
(28)
Con
secu
tive
Wei
ght,
perc
enta
geof
usua
lwei
ght,
seru
mal
bum
inle
vel,
type
ofca
n-ce
r,be
nign
gast
ro-
inte
stin
aldi
seas
e
�Sim
ilar�
for
dem
ogra
phic
char
acte
ris-
tics,
nutr
i-tio
nals
tatu
s,an
dsu
rgic
aldi
agno
sis
No
for
type
ofsu
rger
y,bl
ood
loss
,in
trao
pera
tive
con-
tam
inat
ion
No
data
,pr
esum
ably
gene
ral
Hos
pita
lsta
yA
llre
ceiv
edhe
parin
subc
utan
eous
lyfo
rpr
ophy
laxi
san
dan
tibio
ticpr
ophy
laxi
s
14pa
tient
sre
ceiv
ing
TEN
conv
erte
dto
TPN
241
rand
omly
assi
gned
Non
e
Bozz
etti
etal
.,20
01(2
9)C
onse
cutiv
eH
yper
tens
ion;
hear
tva
lve
dise
ase;
dia-
bete
s;ar
rhyt
hmia
;at
hero
scle
rotic
dis-
ease
(car
diac
orpe
riphe
ral);
resp
ira-
tory
,liv
er,
and
cen-
tral
nerv
ous
syst
emdi
seas
e;ne
oadj
u-va
ntth
erap
y
No
No
for
site
ofpr
imar
ytu
mor
,ty
peof
surg
ery,
intr
aope
r-at
ive
cont
amin
atio
n,du
ra-
tion
ofsu
rger
y,bl
ood
loss
,tr
ansf
usio
n
No
data
,pr
esum
ably
gene
ral
Hos
pita
lsta
yN
oda
ta14
patie
nts
rece
ivin
gTE
Nco
nver
ted
toTP
N:
5tu
bepr
ob-
lem
s,3
diar
rhea
,5
anas
tom
otic
leak
orbl
eedi
ng,
1in
test
i-na
lobs
truc
tion
411
scre
ened
,31
7ra
ndom
lyas
sign
edN
one
Gia
nott
iet
al.,
2002
(30)
Con
secu
tive
Hyp
erte
nsio
n;he
art
valv
edi
seas
e;di
a-be
tes;
arrh
ythm
ia,
athe
rosc
lero
ticdi
s-ea
se(c
ardi
acor
perip
hera
l);re
spira
-to
ry,
liver
,an
dce
n-tr
alne
rvou
ssy
stem
dise
ase;
neoa
dju-
vant
ther
apy
No
No
for
type
ofsu
rger
y,bl
ood
loss
,or
tran
sfus
ion
No
data
,pr
esum
ably
gene
ral
30d
afte
rdi
scha
rge
No
data
exce
ptsi
mila
rbo
wel
prep
arat
ion
inal
lgro
ups
No
data
517
scre
ened
,21
2in
elig
ible
,30
5ra
ndom
lyas
sign
edN
one
Sand
ham
etal
.,20
03(3
1)C
onse
cutiv
eH
isto
ryof
angi
na,
myo
card
iali
nfar
c-tio
n,co
nges
tive
hear
tfa
ilure
,N
YH
Acl
ass,
ASA
risk
clas
s
No
No
for
type
ofsu
rger
yor
perc
enta
geof
urge
ntca
ses
No
data
Hos
pita
lsta
y,12
mo
for
mor
talit
y
No
Inte
rven
tion:
58di
dno
tre
ceiv
epl
anne
dth
erap
y,5
with
drew
cons
ent,
5pu
lmon
ary
arte
ryca
thet
erfa
iled;
cont
rol:
52di
dno
tre
ceiv
epl
anne
dth
erap
y,24
cros
sed
over
tous
eof
pul-
mon
ary
arte
ryca
thet
er
3803
scre
ened
,10
74de
clin
ed,
370
noIC
Ube
d,36
5ph
ysi-
cian
sdi
dno
tre
fer
tost
udy,
1994
rand
omly
assi
gned
Hos
pita
lsta
y,no
ne
*A
SA�
Am
eric
anSo
ciet
yof
Ane
sthe
siol
ogis
ts;C
PAP
�co
ntin
uous
posi
tive
airw
aypr
essu
re;I
CU
�in
tens
ive
care
unit
;IM
�in
tram
uscu
lar;
IS�
ince
ntiv
esp
irom
etry
;LC
�la
paro
scop
icch
olec
yste
ctom
y;LC
R�
lapa
rosc
opic
colo
rect
alre
sect
ion;
NY
HA
�N
ewY
ork
Hea
rtA
ssoc
iati
on;O
C�
open
chol
ecys
tect
omy;
OC
R�
open
colo
rect
alre
sect
ion;
TE
N�
tota
lent
eral
nutr
itio
n;T
PN�
tota
lpar
ente
raln
utri
tion
;VA
TPN
�V
eter
ans
Aff
airs
Tot
alPa
rent
eral
Nut
riti
on.
†T
oco
nver
tse
rum
albu
min
valu
esto
g/L,
mul
tipl
yby
10.
www.annals.org 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 W-129
Downloaded From: http://annals.org/ on 10/30/2017
Appendix Table 3. Abstracted Data for Eligible Randomized Trials, Continued*
Author, Year(Reference)
Lost to Follow-up orIncomplete Follow-up
Patients inIntervention Group,n
Patients inControl Group, n
Age Range,y
Mean (SD) Age,y
Men, n (%)
Møller et al.,2002 (12)
None 56 52 30–85 66 (64) 24 (22)
Berg et al.,1997 (13)
2 patients with clinical signsof pneumonia declinedchest radiography
230 pancuronium 230 vecuronium;231atracurium
24–81 53 pancuronium;54vecuronium;50 atracurium(no data givenfor SDs)
No data
Norris et al.,2001 (14)
None 1) 39 general �regional � IVPCA; 2) 46general �regional �epidural PCA
3) 37 general �IV PCA; 4) 38general �epidural PCA
1) 70 (9.5); 2)67 (10); 3)68 (9.9); 4)68 (8.4)
1) 29 (78); 2) 26 (69);3) 25 (63); 4)35 (77)
Rigg et al.,2002 (15)
None 441 447 Intervention:22–93;control:26–92
Intervention:69 (11);control:69 (11)
503 (57)
Park et al.,2001 (16)
Follow-up at 30 d notcompleted for 11patients
514 enrolled; 489completedfollow-up at 30 d
507 enrolled;495completedfollow-up at30 d
Intervention:66.5 (8.9);control:67 (8.8)
1021 (100); womenexcluded
Fleron et al.,2003 (17)
None 102 (105 randomlyassigned, epiduralcould not be donein 3 patients sothey wereassigned to thecontrol group foranalysis)
115 (3 patientswere from theinterventiongroup)
Intervention:67 (11);control:66 (10)
Intervention: 93 (89);control: 99 (88)
Mann et al.,2000 (18)
None 35 35 Epidural:76 (5.6);control:76.8 (4.7)
Epidural: 20 (57);control: 18 (51)
Cuschieri et al.,1985 (19)
None 25 25 and 25 18–75 Epidural: 51; IVmorphine: 52;IM morphine:52 (no datagiven for SDs)
Epidural: 5 (20); IVmorphine: 4 (28); IMmorphine: 7 (28)
Karayiannakis et al.,1996 (20)
None 42 40 LC: 32–79;OC:34–76
LC: 57 (range,32–79); OC:56 (range,34–76)
LC: 18 (43); OC:18 (45)
Vignali et al.,2004 (21)
None 190 194 LCR: 62 (13.4);OCR:66 (12.2)
LCR: 92 (48); OCR:108 (56)
Chumillas et al.,1998 (22)
None 40 41 18–85 64 (range,18–84)
35 (43)
Fagevik Olsen etal., 1997 (23)
None 172 192 19–92 Intervention:53.5 (17.4);control:52.9 (17.5)
Intervention: 72 (41);control: 86 (44)
Hall et al.,1991 (24)
None 431 IS 445 chestphysiotherapy
IS: IQR,32–70;chestphysiotherapy:IQR,32–72
IS: 54; chestphysiotherapy:56
IS: 221 (51); chestphysiotherapy:216 (49)
Hall et al.,1996 (25)
None 1) 79 low-risk IS; 3)152 high-risk IS
2) 76 low-riskDBE; 4) 149high-risk IS �chestphysiotherapy
1) IQR,29–44; 2)IQR,62–76; 3)IQR,29–43; 4)IQR,58–76
1) 38 (IQR,29–44); 2) 34(IQR, 29–43);3) 68 (IQR,62–76); 4) 67(IQR, 58–76)
1) Low-risk IS: 34 (43);2) high-risk IS:70 (46); 3) low-riskDBE: 34 (45); 4)high-risk IS � chestphysiotherapy:71 (48)
Bohner et al.,2002 (26)
None 99 105 Nasal CPAP:64.1 (12);control:64.5 (11)
Nasal CPAP: 84 (85);control: 82 (78)
VA TPNCooperativeStudy Group,1991 (27)
None 192 203 62.9 (9.9) 455 (99)
Pacelli et al.,2001 (28)
None 119 TEN 122 TPN TEN: 61.5 (10.8);TPN:61.6 (11.8)
TEN: 73 (61); TPN:72 (59)
Bozzetti et al.,2001 (29)
None 159 TEN 158 TPN TEN: 64.8 (11);TPN:64.1 (10)
TEN: 93 (58); TPN:92 (58)
Gianotti et al.,2002 (30)
None 1) 102; 2) 101 3) 102 1) 62.3 (12.3); 2)65.6 (11.5); 3)63.4 (11.9)
1) 50 (49); 2) 60 (59);3) 56 (55)
Sandham et al.,2003 (31)
101 patients, 5% for 6-momortality; 143 patients,7% for 12-mo mortality
997 997 Intervention:72.3 (7);control:72.6 (7)
Intervention: 716 (72);control: 702 (70)
* CPAP � continuous positive airway pressure; DBE � deep breathing exercise; IM � intramuscular; IQR � interquartile range; IS � incentive spirometry; IV �intravenous; LC � laparoscopic cholecystectomy; LCR � laparoscopic colorectal resection; OC � open cholecystectomy; OCR � open colorectal resection; PCA �patient-controlled analgesia; TEN � total enteral nutrition; TPN � total parenteral nutrition; VA TPN � Veterans Affairs Total Parenteral Nutrition.
W-130 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 www.annals.org
Downloaded From: http://annals.org/ on 10/30/2017
App
endi
xT
able
4.A
bstr
acte
dD
ata
for
Elig
ible
Ran
dom
ized
Tria
ls,
Con
tinu
ed*
Aut
hor,
Yea
r(R
efer
ence
)O
utco
mes
Out
com
eM
easu
reIn
terv
enti
onC
ontr
olP
Val
ue
Møl
ler
etal
.,20
02(1
2)O
vera
llco
mpl
icat
ions
10(1
8%)
27(5
2%)
0.00
03W
ound
com
plic
atio
ns3
(5%
)16
(31%
)0.
001
Seco
ndop
erat
ion
2(4
%)
8(1
5%)
0.07
Car
diov
ascu
lar
insu
ffic
ienc
y0
5(1
0%)
0.08
Res
pira
tory
insu
ffic
ienc
y1
(2%
)1
(2%
)0.
97N
onor
thop
edic
hosp
itald
ays
2/75
2(0
.3%
)49
/816
(6%
)0.
0001
Berg
etal
.,19
97(1
3)Po
stop
erat
ive
O2
�3
L/m
in23
(10%
)4.
8(6
%)
Panc
uron
ium
vs.
atra
curiu
m,
0.04
7;pa
ncur
oniu
mvs
.ve
curo
nium
,0.
16PP
C:
pneu
mon
iaor
atel
ecta
sis
19(8
%)
14(6
.1%
)D
iffer
ence
NS;
Pno
tgi
ven
Patie
nts
with
PPC
and
trai
n-of
-fou
rra
tio�
0.70
10/5
9(1
7%)
1/24
(4.2
%)
atra
curiu
mor
vecu
roni
um�
0.00
2,bu
tth
isP
valu
eis
for
com
paris
onof
panc
uron
ium
with
and
with
out
trai
n-of
-fou
r�
0.70
,no
ta
com
paris
onof
panc
uron
ium
vs.
vecu
roni
umor
atra
curiu
mM
edia
nen
dof
surg
ery
toex
tuba
tion
(5th
–95t
hpe
rcen
tile)
15m
in(0
–40
min
)11
min
(2–2
7m
in);
10m
in(0
–25
min
)�
0.05
Med
ian
dura
tion
ofan
esth
esia
(5th
–95t
hpe
rcen
tile)
160
min
(75–
290
min
)15
0m
in(7
0–26
5m
in);
152
min
(70–
280
min
)�
0.05
Post
oper
ativ
etr
ain-
of-f
our
ratio
�0.
7059
(26%
)13
(6%
);11
(5%
)�
0.00
1N
orris
etal
.,20
01(1
4)Le
ngth
ofst
ay1)
7d;
2)7
d3)
8d;
4)7
d0.
83H
ours
toex
tuba
tion
1)19
h;2)
16h
3)19
h;4)
13h
0.01
Hos
pita
lmor
talit
y1)
2(5
%);
2)2
(5%
)3)
3(7
%);
4)2
(4%
)0.
96C
ardi
acm
orta
lity
1)0;
2)1
(2.8
%)
3)0;
4)0
0.47
12-m
om
orta
lity
1)4
(10%
);2)
2(5
%)
3)4
(10%
);4)
2(4
.4%
)0.
64R
eint
ubat
ion
1)1/
35(3
%);
2)0/
363)
1/30
(3%
);4)
1/44
(2%
)0.
9Pr
olon
ged
intu
batio
n1)
6/35
(17%
);2)
6/36
(17%
)3)
8/36
(22%
);4)
3/44
(7%
)0.
27Pn
eum
onia
1)0/
35;
2)1/
36(3
%)
3)1/
36(3
%);
4)0/
440.
58R
igg
etal
.,20
02(1
5)30
-dm
orta
lity
5.2%
4.3%
0.67
Res
pira
tory
failu
re23
.3%
30.2
%0.
02C
ardi
ovas
cula
rev
ent
25.7
%24
.0%
0.61
Park
etal
.,20
01(1
6)M
orta
lity
20(4
%)
17(3
%)
0.74
Res
pira
tory
failu
re51
(10%
)71
(14%
)0.
06Pn
eum
onia
28(5
%)
40(8
%)
0.15
Maj
orca
rdio
vasc
ular
com
plic
atio
n44
(9%
)57
(11%
)0.
12Su
bgro
upan
alys
is,
abdo
min
alao
rtic
surg
ery
Mor
talit
y4/
184
(2%
)5/
190
(3%
)0.
96R
espi
rato
ryfa
ilure
26/1
84(1
4%)
53/1
90(2
8%)
�0.
01Pn
eum
onia
8/18
4(4
%)
19/1
90(1
0%)
0.06
Maj
orca
rdio
vasc
ular
com
plic
atio
n18
/184
(10%
)34
/190
(18%
)0.
03Fl
eron
etal
.,20
03(1
7)M
orta
lity
2(2
%)
7(6
%)
NS;
Pno
tgi
ven
Loba
rat
elec
tasi
s3
(3%
)9
(8%
)N
S;P
not
give
nPn
eum
onia
5(5
%)
6(5
%)
NS;
Pno
tgi
ven
Res
pira
tory
failu
re7
(7%
)9
(8%
)N
S;P
not
give
nA
nypu
lmon
ary
com
plic
atio
n17
(16%
)26
(23%
)P
calc
ulat
e�
0.32
Man
net
al.,
2000
(18)
Segm
enta
lor
loba
rat
elec
tasi
s7/
31(2
3%)
6/33
(18%
)0.
77M
oder
ate
pulm
onar
yco
mpl
icat
ion
3/31
(10%
)2/
33(6
%)
NS
Maj
orpu
lmon
ary
com
plic
atio
n1/
31(3
%)
1/33
(3%
)N
SC
usch
ieri
etal
.,19
85(1
9)A
tele
ctas
is5
(20%
)IM
:10
(40%
);IV
:7
(28%
)IM
orIV
vs.
epid
ural
�N
S;IM
vs.
epid
ural
�0.
05re
port
ed;0
.11
calc
ulat
edfr
omra
wda
ta;I
Vvs
.epi
dura
l�0.
20;I
Mvs
.epi
dura
l�0.
01;I
Vvs
.epi
dura
l�N
SC
hest
infe
ctio
n1
(4%
)IM
:6
(24%
);IV
:5
(20%
)A
llpu
lmon
ary
com
plic
atio
ns6
(24%
)IM
:16
(64%
);IV
:12
(48%
)
Con
tinue
don
follo
win
gpa
ge
www.annals.org 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 W-131
Downloaded From: http://annals.org/ on 10/30/2017
App
endi
xT
able
4—C
onti
nued
Aut
hor,
Yea
r(R
efer
ence
)O
utco
mes
Out
com
eM
easu
reIn
terv
enti
onC
ontr
olP
Val
ue
Kar
ayia
nnak
iset
al.,
1996
(20)
Hos
pita
lsta
y2.
04�
0.62
5.65
�1.
16�
0.05
Ate
lect
asis
inci
denc
e12
/42
(27%
)24
/40
(63%
)�
0.05
Ate
lect
asis
seve
rity
7m
icro
,3
foca
l,2
segm
enta
l,0
loba
r14
mic
ro,
7fo
cal,
3se
gmen
tal,
0lo
bar
�0.
05
Pneu
mon
iaN
opn
eum
onia
No
pneu
mon
iaN
otap
plic
able
Vig
nali
etal
.,20
04(2
1)R
espi
rato
rytr
act
infe
ctio
n3/
190
(1.8
%)
6/19
4(3
.5%
)0.
52C
hum
illas
etal
.,19
98(2
2)C
linic
alpu
lmon
ary
com
plic
atio
ns(a
tele
ctas
is,
bron
chiti
s,pn
eum
onia
)3
(7.5
%)
8(2
0%)
0.11
Ate
lect
asis
onch
est
radi
ogra
phy
6(1
5%)
16(3
9%)
0.01
7Fa
gevi
kO
lsen
etal
.,19
97(2
3)Pu
lmon
ary
com
plic
atio
ns(S
a O2
�92
%or
2of
tem
pera
ture
�38
.2°C
,pa
thol
ogic
ausc
ulta
tion,
atel
ecta
sis,
orpn
eum
onia
onch
est
radi
ogra
phy)
10/1
72(6
%)
52/1
92(2
7%)
�0.
001
Pneu
mon
ia1/
172
(0.6
%)
13/1
92(7
%)
�0.
05O
vera
llpu
lmon
ary
com
plic
atio
nsH
igh-
risk
patie
nts
6/40
(15%
)20
/39
(51%
)�
0.00
1Lo
w-r
isk
patie
nts
4/13
2(3
%)
32/1
53(2
1%)
�0.
001
Obe
sepa
tient
s3/
36(8
%)
27/4
8(5
6%)
�0.
001
Hal
let
al.,
1991
(24)
Clin
ical
feat
ures
ofco
nsol
idat
ion
orco
llaps
epl
uste
mpe
ratu
re�
38°C
and
abno
rmal
ches
tra
diog
raph
orsp
utum
cultu
re
IS:
68/4
71(1
6%)
Che
stph
ysio
ther
apy:
68/4
45(1
5%)
0.84
Abn
orm
alch
est
radi
ogra
phIS
:96
(22%
)C
hest
phys
ioth
erap
y:96
(22%
NS
Hal
let
al.,
1996
(25)
Clin
ical
feat
ures
ofco
nsol
idat
ion
orco
llaps
epl
uste
mpe
ratu
re�
38°C
and
abno
rmal
ches
tra
diog
raph
orsp
utum
cultu
re
1)Lo
w-r
isk
IS:
6/79
(8%
);3)
high
-ris
kIS
:29
/152
(19%
)2)
Low
-ris
kD
BE:
8/76
(11%
);4)
high
-ris
kIS
�ch
est
phys
ioth
erap
y:20
/149
(13%
)Lo
wris
k:1
vs.
3(P
�0.
53);
high
risk:
2vs
.4
(P�
0.18
)
Bohn
eret
al.,
2002
(26)
PaO
2�
70m
mH
gw
ithFi
O2
�0.
75
(5.1
%)
17(1
6.2%
)0.
012
2(2
%)
5(4
.8%
)0.
45R
eint
ubat
ion
1(1
%)
5(4
.8%
)0.
21V
ATP
NC
oope
rativ
eSt
udy
Gro
up,
1991
(27)
Pneu
mon
iaor
empy
ema
16(8
%)
9(4
%)
0.15
Res
pira
tory
failu
re�
4d
13(7
%)
11(5
%)
0.67
Ate
lect
asis
6(3
%)
8(4
%)
0.79
Tran
sien
tre
spira
tory
failu
re6
(3%
)6
(3%
)1.
0Pa
celli
etal
.,20
01(2
8)M
ajor
post
oper
ativ
eco
mpl
icat
ion
45(3
8%)
48(3
9%)
0.89
Dea
th7
(6%
)3
(2.5
%)
0.21
Maj
orin
fect
ious
com
plic
atio
n17
(14%
)14
(11%
)0.
57M
ajor
noni
nfec
tious
com
plic
atio
n28
(24%
)27
(9%
)0.
88Pn
eum
onia
10(8
%)
5(4
%)
0.19
Res
pira
tory
failu
re6
(5%
)4
(3%
)0.
54Pn
eum
onia
�re
spira
tory
failu
re16
(13%
)9
(7%
)0.
14M
inor
infe
ctio
n23
(19%
)23
(19%
)1.
0Bo
zzet
tiet
al.,
2001
(29)
Ove
rall
com
plic
atio
ns54
(34%
)78
(49%
)0.
005
Min
orco
mpl
icat
ions
40(2
5%)
57(3
6%)
0.03
5M
ajor
com
plic
atio
ns14
(9%
)21
(13%
)0.
207
Infe
ctio
usco
mpl
icat
ions
25(1
6%)
42(2
7%)
0.01
8N
onin
fect
ious
com
plic
atio
ns42
(26%
)57
(36%
)0.
064
Res
pira
tory
trac
tin
fect
ion
9(6
%)
14(9
%)
0.39
Res
pira
tory
failu
re4
(3%
)6
(4%
)0.
75R
espi
rato
ryin
fect
ion
�fa
ilure
13(8
%)
20(1
3%)
0.27
Con
tinue
don
follo
win
gpa
ge
W-132 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 www.annals.org
Downloaded From: http://annals.org/ on 10/30/2017
App
endi
xT
able
4—C
onti
nued
Aut
hor,
Yea
r(R
efer
ence
)O
utco
mes
Out
com
eM
easu
reIn
terv
enti
onC
ontr
olP
Val
ue
Gia
nott
iet
al.,
2002
(30)
Dea
th1)
1(1
%);
2)2
(2%
)3)
1(1
%)
NS
Infe
ctio
usco
mpl
icat
ions
1)14
(14%
);2)
16(1
6%)
3)31
(30%
)1
vs.
3�
0.00
6;2
vs.
3�
0.02
Non
infe
ctio
usco
mpl
icat
ions
1)30
(29%
);2)
28(2
8%)
3)36
(35%
)N
SA
nyco
mpl
icat
ion
1)36
(35%
);2)
34(3
4%)
3)49
(48%
)N
SR
espi
rato
rytr
act
infe
ctio
n1)
3(3
%);
2)6
(6%
)3)
8(8
%)
1vs
.3
�0.
21;
2vs
.3
�0.
78R
espi
rato
ryfa
ilure
1)6
(6%
);2)
9(9
%)
3)6
(6%
)1
vs.
3�
1.0;
2vs
.3
�0.
59R
espi
rato
ryin
fect
ion
orfa
ilure
1)9
(9%
);2)
15(1
5%)
3)14
(14%
)1
vs.
3�
0.38
;2
vs.
3�
1.0
Sand
ham
etal
.,20
03(3
1)In
-hos
pita
lmor
talit
y78
(7.8
%)
77(7
.7%
)0.
93Pn
eum
onia
63(6
.6%
)70
(7.3
%)
0.70
*D
BE
�de
epbr
eath
ing
exer
cise
;FiO
2�
frac
tion
ofin
spir
edox
ygen
;IM
�in
tram
uscu
lar;
IV�
intr
aven
ous;
NS
�no
tsi
gnifi
cant
;PPC
�po
stop
erat
ive
pulm
onar
yco
mpl
icat
ion;
SaO
2�
arte
rial
oxyg
ensa
tura
tion
;VA
TPN
�V
eter
ans
Aff
airs
Tot
alPa
rent
eral
Nut
riti
on.
www.annals.org 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 W-133
Downloaded From: http://annals.org/ on 10/30/2017
App
endi
xT
able
5.A
bstr
acte
dD
ata
for
Elig
ible
Ran
dom
ized
Tria
ls,
Con
tinu
ed*
Aut
hor,
Yea
r(R
efer
ence
)A
dver
seEf
fect
:In
terv
enti
onA
dver
seEf
fect
:C
ontr
olP
Val
ueC
oncl
usio
nSt
udy
Qua
lity
Com
men
t
Møl
ler
etal
.,20
02(1
2)N
oda
taN
oda
taTh
esm
okin
gce
ssat
ion
prog
ram
redu
ced
over
all
post
oper
ativ
eco
mpl
icat
ions
,pr
imar
ilydu
eto
asi
gnifi
cant
redu
ctio
nin
wou
ndco
mpl
icat
ions
,an
dno
nort
hope
dic
hosp
itald
ays
into
talh
ipan
dkn
eere
plac
emen
t.Th
era
teof
PPC
was
too
low
tosh
owan
effe
ct.
Goo
d
Berg
etal
.,19
97(1
3)N
oda
taN
oda
taIn
cide
nce
and
dura
tion
ofre
sidu
albl
ock
occu
rssi
gnifi
cant
lym
ore
ofte
nw
ithpa
ncur
oniu
m.
Res
idua
lblo
ckw
ithlo
ng-a
ctin
gpa
ncur
oniu
mis
asso
ciat
edw
ithm
ore
PPC
s;re
sidu
albl
ock
with
inte
rmed
iate
-act
ing
vecu
roni
umor
atra
curiu
mis
not
asso
ciat
edw
ithhi
gher
risk
for
PPC
s.
Goo
d
Nor
riset
al.,
2001
(14)
No
data
No
data
No
adva
ntag
eto
epid
ural
anal
gesi
aor
com
bine
dge
nera
l�re
gion
alan
esth
esia
;ve
ryfe
wPP
Cs
over
allo
ccur
red.
Goo
d
Rig
get
al.,
2002
(15)
No
data
No
data
Com
bine
din
trao
pera
tive
gene
rala
nest
hesi
aan
dep
idur
allo
cala
nest
hetic
�po
stop
erat
ive
epid
ural
loca
lane
sthe
ticw
asas
soci
ated
with
asi
gnifi
cant
lylo
wer
rate
ofre
spira
tory
failu
rebu
tno
diff
eren
cein
30-d
mor
talit
y,ca
rdia
cco
mpl
icat
ions
,or
othe
rpo
stop
erat
ive
mor
bidi
ty.
Fair
Onl
y22
5of
447
patie
nts
assi
gned
toep
idur
alw
ere
fully
adhe
rent
toth
eep
idur
alpr
otoc
ol(2
22pr
otoc
olvi
olat
ions
:no
epid
ural
cath
eter
,29
;ca
thet
erre
mov
ed�
72h
afte
rsu
rger
y,19
0;ca
thet
erre
mov
ed�
72h
afte
rsu
rger
y,3)
.Pa
rket
al.,
2001
(16)
No
data
No
data
Ove
rall,
epid
ural
anes
thes
iaan
dan
alge
sia
was
asso
ciat
edw
itha
tren
dto
war
dle
ssre
spira
tory
failu
re(P
�0.
06),
pneu
mon
ia(P
�0.
15),
and
maj
orca
rdio
vasc
ular
com
plic
atio
ns(P
�0.
18).
Inth
esu
bgro
upha
ving
abdo
min
alao
rtic
surg
ery,
epid
ural
was
asso
ciat
edw
ithsi
gnifi
cant
lyle
ssre
spira
tory
failu
re(P
�0.
01)
and
maj
orca
rdio
vasc
ular
com
plic
atio
ns(P
�0.
03)
and
atr
end
tow
ard
less
pneu
mon
ia(P
�0.
06).
Ther
ew
asno
diff
eren
cein
mor
talit
y.
Fair
Fler
onet
al.,
2003
(17)
No
data
No
data
Intr
aope
rativ
eep
idur
alop
ioid
s,co
mpa
red
with
intr
aope
rativ
eIV
opio
ids,
wer
eno
tas
soci
ated
with
redu
ced
PPC
s.Fa
ir
Man
net
al.,
2000
(18)
Post
oper
ativ
ehy
pote
nsio
n:5
patie
nts
Post
oper
ativ
ehy
pote
nsio
n:no
patie
nts
0.01
Com
bine
din
trao
pera
tive
gene
rala
ndep
idur
alan
esth
esia
and
anal
gesi
aw
ithpo
stop
erat
ive
PCEA
was
not
asso
ciat
edw
ithfe
wer
PPC
sco
mpa
red
with
gene
ral
anes
thes
iaal
one
and
post
oper
ativ
eIV
PCA
.
Poor
Low
stat
istic
alpo
wer
Seve
rehy
pote
nsio
n,sy
stol
icBP
�78
mm
Hg:
nopa
tient
sSe
vere
hypo
tens
ion,
syst
olic
BP�
78m
mH
g:no
patie
nts
NS
Mot
orbl
ocka
de:
nopa
tient
sM
otor
bloc
kade
:no
patie
nts
NS
Abs
cess
orne
urol
ogic
com
plic
atio
ndu
eto
epid
ural
:no
patie
nts
Cus
chie
riet
al.,
1985
(19)
Hyp
oten
sion
:9
patie
nts;
urin
ary
trac
tin
fect
ion:
4pa
tient
sN
oda
taIn
trao
pera
tive
and
post
oper
ativ
eep
idur
alan
esth
etic
may
redu
cePP
Cs
com
pare
dw
ithan
alge
sia
with
IMm
orph
ine.
Poor
Smal
lsam
ple
size
,no
the
lpfu
ltha
tep
idur
alse
emed
bett
erth
anIM
mor
phin
ebe
caus
eIM
mor
phin
eis
rare
lyus
edno
w.
Pva
lues
wer
eno
tre
port
edfo
rIV
vs.
epid
ural
com
paris
ons;
we
calc
ulat
edth
emfr
omth
era
wda
ta.
Kar
ayia
nnak
iset
al.,
1996
(20)
3LC
patie
nts
conv
erte
dto
OC
but
wer
eex
clud
edfr
oman
alys
isLC
was
asso
ciat
edw
itha
sign
ifica
ntly
low
erin
cide
nce
and
seve
rity
ofat
elec
tasi
sco
mpa
red
with
OC
.Po
orA
naly
sis
was
not
inte
ntio
n-to
-tre
at;
nopn
eum
onia
occu
rred
inei
ther
grou
p.V
igna
liet
al.,
2004
(21)
10LC
Rpa
tient
sco
nver
ted
toO
CR
but
wer
ein
clud
edin
inte
ntio
n-to
-tre
atan
alys
es
LCR
was
asso
ciat
edw
itha
nons
igni
fican
ttr
end
tow
ard
few
erre
spira
tory
trac
tin
fect
ions
.G
ood
Chu
mill
aset
al.,
1998
(22)
Apr
ogra
mof
perio
pera
tive
brea
thin
gex
erci
ses
that
incl
uded
forc
edex
pira
tion,
coug
h,ch
est
expa
nsio
nex
erci
ses
and
diap
hrag
mm
obili
zatio
n,m
axim
umin
spira
tion
for
3–5
s,an
dea
rlyam
bula
tion
afte
rsu
rger
yw
asas
soci
ated
with
few
erPP
Cs.
Poor
Low
stat
istic
alpo
wer
Fage
vik
Ols
enet
al.,
1997
(23)
Inte
nsiv
ech
est
phys
ioth
erap
y,co
mpa
red
with
none
,w
asas
soci
ated
with
few
erPP
Cs.
Poor
Ver
yw
eak
defin
ition
ofpu
lmon
ary
com
plic
atio
ns;
appa
rent
lyno
unifo
rmsu
rvei
llanc
epr
otoc
ol.
Hal
let
al.,
1991
(24)
No
diff
eren
cebe
twee
nIS
and
ches
tph
ysio
ther
apy
inra
teof
post
oper
ativ
epn
eum
onia
.Po
orN
oro
utin
esu
rvei
llanc
epr
otoc
olfo
ral
lpat
ient
s;ch
est
radi
ogra
phy
done
only
onpa
tient
ssu
spec
ted
ofa
pulm
onar
yco
mpl
icat
ion
(IS,
44%
;ch
est
phys
ioth
erap
y,43
%).
No
info
rmat
ion
abou
tam
ount
and
inte
nsity
ofth
erap
yac
tual
lyre
ceiv
ed.
Con
tinue
don
follo
win
gpa
ge
W-134 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 www.annals.org
Downloaded From: http://annals.org/ on 10/30/2017
App
endi
xT
able
5—C
onti
nued
Aut
hor,
Yea
r(R
efer
ence
)A
dver
seEf
fect
:In
terv
enti
onA
dver
seEf
fect
:C
ontr
olP
Val
ueC
oncl
usio
nSt
udy
Qua
lity
Com
men
t
Hal
let
al.,
1996
(25)
Inlo
w-r
isk
patie
nts,
ther
ew
asno
diff
eren
cein
rate
ofPP
Cs
betw
een
ISan
dD
BEs.
Inhi
gh-r
isk
patie
nts,
ther
ew
asno
diff
eren
cebe
twee
nIS
and
ISco
mbi
ned
with
ches
tph
ysio
ther
apy.
Poor
Both
stud
ies
byH
alla
ndco
lleag
ues
seem
toha
veid
entic
alm
etho
ds;
som
eof
the
met
hods
lang
uage
isid
entic
al,
som
eno
t.�P
rese
nce
ofcl
inic
alsi
gns
was
dete
rmin
edea
chda
yby
the
atte
ndin
gph
ysic
ian.
�C
hest
radi
ogra
phy
was
done
only
for
susp
ecte
dco
mpl
icat
ions
.Sp
utum
was
sent
for
test
ing
�whe
nth
epa
tient
prod
uced
disc
olou
red
sput
um.�
Som
esu
rvei
llanc
efo
rPP
Cs
was
rout
ine
(clin
ical
sign
s),
som
eno
t(c
hest
radi
ogra
phy,
sput
umte
stin
g).
Ther
eis
now
ayto
tell
how
man
ypa
tient
sfr
omth
epr
evio
uspu
blic
atio
nw
ere
incl
uded
inth
ere
port
.Bo
hner
etal
.,20
02(2
6)Su
perf
icia
lnos
eul
cer:
4(4
%)
Non
e(0
%)
Nas
alC
PAP
for
12h
afte
rhi
gh-r
isk
abdo
min
alva
scul
arsu
rger
yw
asas
soci
ated
with
few
erep
isod
esof
seve
rehy
poxe
mia
(Fi O
2�
70m
mH
gw
ithFi
O2
�0.
7)an
dtr
ends
tow
ard
few
erep
isod
esof
pneu
mon
iaan
dre
spira
tory
failu
re.
Goo
d
VA
TPN
Coo
pera
tive
Stud
yG
roup
,19
91(2
7)
Maj
orin
fect
ious
com
plic
atio
ns:
27/1
92(1
4%)
Maj
orin
fect
ious
com
plic
atio
ns:
13/2
03(6
%)
0.01
Ove
rall,
ther
ew
asno
bene
fitof
TPN
for
prev
entin
gpu
lmon
ary
com
plic
atio
ns.
Ove
rall,
TPN
was
asso
ciat
edw
ithm
ore
maj
orin
fect
ious
com
plic
atio
nsan
dm
ore
cath
eter
-rel
ated
com
plic
atio
ns.
Goo
dSu
bgro
upan
alys
essu
gges
ted
mild
mal
nutr
ition
,no
bene
fit;
TPN
asso
ciat
edw
ithm
ore
infe
ctio
ns,
espe
cial
lypn
eum
onia
and
wou
ndin
fect
ion;
seve
rem
alnu
triti
on,
noin
crea
sed
infe
ctio
n,bu
tsi
gnifi
cant
lyfe
wer
noni
nfec
tious
com
plic
atio
ns.
Maj
orno
ninf
ectio
usco
mpl
icat
ions
:32
/192
(17%
)M
ajor
noni
nfec
tious
com
plic
atio
ns:
45/2
03(2
2%)
0.20
Non
infe
ctio
usca
thet
er-r
elat
edco
mpl
icat
ions
:11
/192
(6%
)N
onin
fect
ious
cath
eter
-rel
ated
com
plic
atio
ns:
2/20
3(1
%)
0.01
Pace
lliet
al.,
2001
(28)
3bl
oatin
g;4
diar
rhea
;4
tube
prob
lem
s;2
chyl
ous
fistu
la;
1bl
eed
from
jeju
nost
omy
5tr
ansi
ent
hypo
glyc
emia
;2
cath
eter
seps
is0.
11Th
ere
was
nodi
ffer
ence
betw
een
post
oper
ativ
eTE
Nan
dTP
Nin
rate
sof
over
allc
ompl
icat
ions
,pn
eum
onia
,re
spira
tory
failu
re,
com
bine
dpu
lmon
ary
com
plic
atio
ns,
orad
vers
eev
ents
ofth
e2
inte
rven
tions
.
Goo
d
Bozz
etti
etal
.,20
01(2
9)D
iste
nsio
n:23
(14%
)10
(6%
)0.
018
TEN
,co
mpa
red
with
TPN
,w
asas
soci
ated
with
few
erov
eral
l,m
inor
,an
din
fect
ious
com
plic
atio
nsan
dm
argi
nally
sign
ifica
ntly
few
erno
ninf
ectio
usco
mpl
icat
ions
but
nobe
nefit
rega
rdin
gpu
lmon
ary
com
plic
atio
ns.
TEN
,co
mpa
red
with
TPN
,w
asas
soci
ated
with
sign
ifica
ntly
mor
eab
dom
inal
dist
ensi
onan
dcr
amps
but
not
mor
edi
arrh
eaan
dvo
miti
ng.
Goo
d
Cra
mps
:21
(13%
)8
(5%
)0.
012
Dia
rrhe
a:13
(8%
)9
(6%
)0.
385
Vom
iting
:4
(3%
)3
(2%
)0.
709
Tota
l:56
(35%
)22
(14%
)�
0.00
01G
iano
ttie
tal
.,20
02(3
0)C
ram
ping
orbl
oatin
g:1)
16,
2)42
;di
arrh
ea:
1)3,
2)7;
vom
iting
:1)
1,2)
2
Cra
mpi
ngor
bloa
ting:
14;
diar
rhea
:3)
3;vo
miti
ng:
3)2
All
NS,
exce
ptP
�0.
001
for
2(p
erio
pera
tive)
vs.
1(p
reop
erat
ive)
and
3(c
ontr
ol)
Preo
pera
tive
and
perio
pera
tive
ente
rali
mm
unon
utrit
ion,
com
pare
dw
ithno
ente
raln
utrit
ion
was
asso
ciat
edw
ithfe
wer
infe
ctio
usco
mpl
icat
ions
but
nobe
nefit
inno
ninf
ectio
usco
mpl
icat
ions
,re
spira
tory
trac
tin
fect
ion,
orre
spira
tory
failu
re.
Goo
d
Sand
ham
etal
.,20
03(3
1)O
vera
ll:17
Ove
rall:
50.
016
Perio
pera
tive
pulm
onar
yar
tery
cath
eter
sin
high
-ris
ksu
rgic
alpa
tient
sdi
dno
tim
prov
ein
patie
ntm
orta
lity
orre
duce
the
rate
ofth
ese
cond
ary
outc
ome
ofpn
eum
onia
and
was
asso
ciat
edw
itha
stat
istic
ally
sign
ifica
ntly
high
erra
teof
adve
rse
cath
eter
-rel
ated
even
ts.
For
uncl
ear
reas
ons,
the
rate
ofth
ese
cond
ary
outc
ome
ofpu
lmon
ary
embo
lism
was
also
sign
ifica
ntly
high
erin
patie
nts
rece
ivin
gpu
lmon
ary
arte
ryca
thet
er(8
vs.
0;P
�0.
004)
.
Goo
d
*B
P�
bloo
dpr
essu
re;C
PAP
�co
ntin
uous
posi
tive
airw
aypr
essu
re;D
BE
�de
epbr
eath
ing
exer
cise
;FiO
2�
frac
tion
ofin
spir
edox
ygen
;IM
�in
tram
uscu
lar;
IS�
ince
ntiv
esp
irom
etry
;IV
�in
trav
enou
s;LC
�la
paro
scop
icch
olec
yste
ctom
y;LC
R�
lapa
rosc
opic
colo
rect
alre
sect
ion;
NS
�no
tsig
nific
ant;
OC
�op
ench
olec
yste
ctom
y;O
CR
�op
enco
lore
ctal
rese
ctio
n;PC
A�
pati
ent-
cont
rolle
dan
alge
sia;
PCE
A�
pati
ent-
cont
rolle
dep
idur
alan
alge
sia;
PPC
�po
stop
erat
ive
pulm
onar
yco
mpl
icat
ion;
TE
N�
tota
len
tera
lnu
trit
ion;
TPN
�to
tal
pare
nter
alnu
trit
ion;
VA
TPN
�V
eter
ans
Aff
airs
Tot
alPa
rent
eral
Nut
riti
on.
www.annals.org 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 W-135
Downloaded From: http://annals.org/ on 10/30/2017
App
endi
xT
able
6.A
bstr
acte
dD
ata
for
Elig
ible
Syst
emat
icR
evie
ws
and
Met
a-A
naly
ses*
Aut
hor,
Yea
r(R
efer
ence
)Ty
peof
Surg
ery
Inte
rven
tion
Lite
ratu
reSe
arch
Dat
esEl
ectr
onic
Dat
abas
esG
ray
Lite
ratu
reU
npub
lishe
dSt
udie
sEn
glis
h-La
ngua
geLi
tera
ture
Onl
y
Tria
lsId
enti
fied
,n
Elig
ible
Tria
ls,
n
Rod
gers
etal
.,20
00(3
2)N
ore
stric
tion
Reg
iona
lan
esth
esia
(epi
dura
lor
spin
al)
1966
–199
8C
urre
ntC
onte
nts,
EMBA
SE,
MED
LIN
E,C
ochr
ane
Libr
ary
No
data
Yes
No
158
141
Urw
inet
al.,
2000
(33)
Hip
frac
ture
repa
irR
egio
nal
anes
thes
ia(e
pidu
ralo
rsp
inal
)
Not
stat
edC
urre
ntC
onte
nts,
EMBA
SE,
MED
LIN
E,C
ochr
ane
Libr
ary
Clin
ical
Tria
lsR
egis
try,
CIN
AH
L
No
data
No
No
Not
stat
ed15
Balla
ntyn
eet
al.,
1998
(34)
No
rest
rictio
nPo
stop
erat
ive
anal
gesi
cth
erap
y
1966
–199
5M
EDLI
NE
No
data
No
No
data
195
48
Wal
der
etal
.,20
01(3
5)N
ore
stric
tion
IVPC
A19
66–2
000
EMBA
SE,
MED
LIN
E,C
ochr
ane
Libr
ary
Clin
ical
Tria
lsR
egis
try
No
data
No
No
9532
Dow
nset
al.,
1996
(36)
Cho
lecy
stec
tom
yLC
1987
–199
6EM
BASE
,M
EDLI
NE
Yes
Yes
No
Unc
lear
15to
tal;
only
1as
sess
eda
clin
ical
lyre
leva
ntpu
lmon
ary
com
plic
atio
n(a
tele
ctas
is)
Abr
aham
etal
.,20
04(3
7)R
esec
tion
ofco
lore
ctal
canc
er
Lapa
rosc
opic
rese
ctio
n19
66–2
002
EMBA
SE,
MED
LIN
E,C
ochr
ane
Libr
ary
Clin
ical
Tria
lsR
egis
try
No
data
No
Yes
6212
;un
clea
rnu
mbe
ras
sess
ing
pulm
onar
yco
mpl
icat
ions
Che
atha
met
al.,
1995
(38)
Elec
tive
lapa
roto
my
Sele
ctiv
epo
stop
erat
ive
naso
gast
ricde
com
pres
sion
No
data
,pu
blis
hed
1995
Cur
rent
Con
tent
s,M
EDLI
NE
No
data
No
Yes
1715
Nel
son
etal
.,20
05(3
9)La
paro
tom
ySe
lect
ive
post
oper
ativ
ena
soga
stric
deco
mpr
essi
on
No
data
,pu
blis
hed
2005
EMBA
SE,
MED
LIN
E,C
ochr
ane
Libr
ary
Clin
ical
Tria
lsR
egis
try
No
data
No
Unc
lear
3328
tota
l;19
repo
rtpu
lmon
ary
com
plic
atio
ns
Thom
asan
dM
cInt
osh,
1994
(40)
Upp
erab
dom
inal
surg
ery
Post
oper
ativ
elu
ngex
pans
ion
ther
apie
s
1966
–199
2M
EDLI
NE,
CIN
AH
LN
oda
taN
oY
esU
ncle
ar14
Ove
rend
etal
.,20
01(4
1)A
bdom
inal
surg
ery
Post
oper
ativ
elu
ngex
pans
ion
ther
apie
s
1966
–200
0C
urre
ntC
onte
nts,
MED
LIN
E,C
INA
HL,
Hea
lthST
AR
No
data
No
Yes
Unc
lear
26
Moo
reet
al.,
1992
(42)
Hig
h-ris
ksu
rger
y,no
rest
rictio
nsEa
rlyen
tera
lvs.
TPN
No
data
No
data
Yes
Yes
No
data
Unc
lear
8to
tal;
alla
ppea
rto
bein
dust
ry-s
pons
ored
by1
com
pany
*IV
�in
trav
enou
s;LC
�la
paro
scop
icch
olec
yste
ctom
y;PC
A�
pati
ent-
cont
rolle
dan
alge
sia;
TPN
�to
tal
pare
nter
alnu
trit
ion.
W-136 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 www.annals.org
Downloaded From: http://annals.org/ on 10/30/2017
Appendix Table 7. Abstracted Data for Eligible Systematic Reviews and Meta-Analyses, Continued*
Author, Year(Reference)
Analysis Results
Fixed- orRandom-EffectsModels
HeterogeneityAssessed
SensitivityAnalysis
SubgroupAnalysis
PublicationBiasAssessed
StudyQuality
Rodgers et al.,2000 (32)
Fixed Yes Yes Yes Yes Good Regional anesthesia, with or without generalanesthesia, was associated with lower mortalityoverall (OR, 0.70 [95% CI, 0.51–0.97]) andorthopedic surgery (results depicted in Forestplot; OR and CI not stated) but not for othersurgical subgroups. Regional anesthesia vs.general anesthesia alone was also associated withreduced mortality (OR, 0.64 [CI, 0.47–0.87]; n �5202). Regional anesthesia was associated withless pneumonia (OR, 0.61 [CI, 0.48–0.76]),respiratory depression (OR, 0.41 [CI, 0.23–0.73]),deep venous thrombosis (OR, 0.56 [CI,0.43–0.72]), and less need for transfusion (OR,0.50 [CI, 0.39–0.66]).
Urwin et al.,2000 (33)
Fixed orrandom perheterogeneity(P � 0.1)
Yes No Yes No data Good Regional, compared with general, anesthesia wasassociated with lower 30-d mortality (OR, 0.66[CI, 0.47–0.96]) and deep venous thrombosis(OR, 0.41 [CI, 0.23–0.72]) but not lower 3-, 6-,or 12-mo mortality, risk for pneumonia (OR, 0.92[CI, 0.53–1.59]), or several other medicalcomplications, including all pulmonary embolisms.Regional anesthesia was associated withsignificantly fewer fatal pulmonary embolisms(OR and CI not stated).
Ballantyne et al.,1998 (34)
Random Yes Yes Yes No data Fair Epidural opioid, compared with systemic opioid, wasassociated with less atelectasis (OR, 0.53 [CI,0.33–0.85]) but not �pulmonary infection� (OR,0.53 [CI, 0.18–1.53]) or overall PPCs (OR, 0.51[CI, 0.20–1.33]). Epidural local anesthetic,compared with systemic opioid, was associatedwith less �pulmonary infection� (OR, 0.36 [CI,0.21–0.65]) and overall PPCs (OR, 0.58 [CI,0.42–0.80]) but not atelectasis (OR, 0.74 [CI,0.50–1.11]). There were nonsignificant trendstoward fewer PPCs with epidural opioid �anesthetic compared with systemic opioid andwith intercostal nerve block compared withsystemic opioid.
Walder et al.,2001 (35)
Fixed orrandom perheterogeneity(P � 0.1)
Yes No Yes No data Good In a subgroup analysis of 2 morphine trialsreporting PPCs (n � 147), IV PCA was associatedwith lower risk (OR, 0.93 [CI, 0.86–0.99]). In aseparate trial of 60 patients, there was no benefitregarding �chest infection� (no data given).Among 689 patients, respiratory depression wasnot more frequent with PCA (OR, 1.08 [CI,0.44–2.68]).
Downs et al.,1996 (36)
Quantitativepooling notdone
No data Good LC, compared with OC, was associated with lesscompromise and faster recovery of postoperativepulmonary function. In 1 trial of 40 patients withblinded assessment of postoperative chestradiography, LC was associated with lessatelectasis (frequency, 29% vs. 63%; P � 0.05;severity, chi-square for trend, P � 0.05).
Abraham et al.,2004 (37)
Fixed orrandom perheterogeneityassessment
Yes No No No data Good LCR, compared with OCR, for cancer wasassociated with no mortality benefit, a trendtoward fewer respiratory complications (OR, 0.65[CI, 0.28–1.49]), fewer overall complications (OR,0.62 [CI, 0.38–1.03])—primarily due to fewerwound complications, primarily wound infection(OR, 0.47 [CI, 0.28–0.80])—faster recovery ofrespiratory function (PEF, 44% faster [CI,32%–67%]; FEV1, 36% faster [CI, –33% to50%]; FVC, 40% faster [CI, 0%–50%]) andshorter hospital stay (21% shorter [CI,14%–38%]).
Continued on following page
www.annals.org 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 W-137
Downloaded From: http://annals.org/ on 10/30/2017
Appendix Table 7—Continued
Author, Year(Reference)
Analysis Results
Fixed- orRandom-EffectsModels
HeterogeneityAssessed
SensitivityAnalysis
SubgroupAnalysis
PublicationBiasAssessed
StudyQuality
Cheatham et al.,1995 (38)
Quantitativepooling notdone
No Poor The meta-analysis was good quality up toquantitative pooling, which pooled RCTs,uncontrolled studies, and case–control studies,thus rendering the results unusable. For theoverall group of 26 studies (which appears tocomprise 15 RCTs, 3 nonrandomized trials, and 8case–control studies [n � 3964]), selectivedecompression was associated with lesspneumonia (RR, 0.49; P � 0.0001) andatelectasis (RR, 0.46; P � 0.001) and shorter timeto oral intake (3.5 d vs. 4.6 d; P � 0.04). Therewas no difference in aspiration rates (RR, 0.61; P� 0.88), nausea (RR, 0.98; P � 0.31), vomiting(RR, 1.19; P � 0.11), or abdominal distension(RR, 0.98; P � 0.36). For 20 higher-qualitystudies (15 RCTs plus 5 case–control studies [n �2915]), selective nasogastric decompression wasalso associated with less pneumonia (RR, 0.59;P � 0.01) and atelectasis (RR, 0.52; P � 0.002),a trend toward shorter time to oral intake (3.5 dvs. 4.5 d; P � 0.07), no difference in aspiration(RR, 0.94; P � 0.91) but more vomiting (RR,1.45; P � 0.005) and abdominal distension (RR,1.34; P � 0.02). Insufficient data were reportedfor calculating pooled effects for RCTs only andCIs.
Nelson et al.,2005 (39)
Fixed orrandom perheterogeneityassessment
Yes Yes Yes No Good Selective, compared with routine, nasogastricdecompression was associated with a trendtoward fewer PPCs (reported as relative benefitincrease of 1.35 [CI, 0.98–1.86] converted to RRreduction of 0.74 [CI, 0.54–1.02]; P � 0.07).Data were insufficient or too heterogeneous topool for nausea, vomiting, aspiration, orabdominal distension, and 15 of the 28 includedtrials were also included in the Cheatham et al.review (38).
Thomas andMcIntosh,1994 (40)
No data Yes No Yes No Poor Across all lung expansion modalities, there was atrend toward fewer PPCs compared with controls(OR, 0.85 [CI, 0.59–1.2]), but there wasunexplained significant heterogeneity. IS,compared with control (2 studies [n � 212]) wasassociated with fewer PPCs (OR, 0.44 [CI,0.18–0.99]) with no significant heterogeneity.DBEs, compared with control (4 studies [n �564]), were also associated with fewer PPCs (OR,0.43 [CI, 0.27–0.63]), but the heterogeneity testwas significant. Among studies comparingdifferent modalities, none (IS, DBEs, IPPB) wasclearly superior.
Overend et al.,2001 (41)
Quantitativepooling notdone
No Poor The authors reported no raw data on rates of PPCs.In the only trial in the review that met oursample size inclusion criteria, DBEs and IPPBreportedly equally prevented PPCs compared withno lung expansion intervention.
Moore et al.,1992 (42)
Fixed Yes No Yes No Poor Infections were twice as frequent among patientsreceiving TPN compared with those receivingearly enteral nutrition (35% vs. 16%; P � 0.01),even after excluding patients with catheter sepsisfrom analysis (29% vs. 16%; P � 0.03). Overallinfections and pneumonia were significantlyreduced in trauma patients, but power was verylow for �nontrauma� (? elective surgery) patientsfor overall infections (4/28 vs. 3/32; P � 0.70)and pneumonia (3/28 vs. 1/32; P � 0.33).
* DBE � deep breathing exercise; IPPB � intermittent positive-pressure breathing; IS � incentive spirometry; IV � intravenous; LC � laparoscopic cholecystectomy;LCR � laparoscopic colorectal resection; OC � open cholecystectomy; OCR � open colorectal resection; OR � odds ratio; PCA � patient-controlled analgesia; PEF �peak expiratory flow; PPC � postoperative pulmonary complication; RCT � randomized, controlled trial; RR � relative risk; TPN � total parenteral nutrition.
W-138 18 April 2006 Annals of Internal Medicine Volume 144 • Number 8 www.annals.org
Downloaded From: http://annals.org/ on 10/30/2017