Abstract
Objective. Studies have suggested higher rates of perioperative and postoperative complications in smokers compared to nonsmokers. The objective of this systematic review was to assess the association of smoking and postoperative outcomes following total hip arthroplasty (THA) or total knee arthroplasty (TKA).
Methods. A search of 6 databases (The Cochrane Library, Scopus, Proquest Dissertation abstracts, CINAHL, Ovid Medline, and Embase) was performed by a Cochrane librarian. All titles and abstracts were screened by 2 independent reviewers with expertise in performing systematic reviews. Studies were included if they were fully published reports that included smoking and any perioperative or postoperative clinical outcome in patients with TKA or THA.
Results. A total of 21 studies were included for the review, of which 6 provided multivariable-adjusted analyses, 14 univariate analyses, and one statistical modeling. For most outcomes, results from 1–2 studies could be pooled. Current smokers were significantly more likely to have any postoperative complication (risk ratio 1.24, 95% CI 1.01 to 1.54) and death (risk ratio 1.63, 95% CI 1.06 to 2.51) compared to nonsmokers. Former smokers were significantly more likely to have any post-operative complication (risk ratio 1.32, 95% CI 1.05 to 1.66) and death (risk ratio 1.69, 95% CI 1.08 to 2.64) compared to nonsmokers.
Conclusion. This systematic review found that smoking is associated with significantly higher risk of postoperative complication and mortality following TKA or THA. Studies examining longterm consequences of smoking on implant survival and complications are needed. Smoking cessation may improve outcomes after THA or TKA.
Total knee and total hip arthroplasty (TKA and THA) are very effective surgical treatment options for patients with refractory joint disease/arthritis that have failed to respond to conservative medical treatment. These procedures are performed mostly as elective procedures and are associated with reduction in pain and improvement in quality of life and lower extremity function1,2. Several patient-related factors such as age, sex, and socioeconomic status influence the outcome of TKA and THA3,4,5. In addition, modifiable factors such as smoking, medical comorbidity, and obesity may also affect these outcomes5,6,7,8,9. Identification of modifiable risk factors such as smoking is important, since presurgical smoking cessation interventions can improve outcomes of TKA and THA. Smoking has been shown to be a risk factor for non-union after spinal surgery10 and delayed bone healing after hemicallotasis, an orthopedic procedure to correct knee deformity11. However, the association of smoking with risk of complications after TKA or THA has been investigated in a few studies.
The purpose of this study was to perform a systematic review of studies of smoking status and outcomes following TKA or THA to assess whether smoking was significantly associated with any complications following arthroplasty, and if so, which complications.
MATERIALS AND METHODS
An experienced Cochrane librarian (L. Falzon) searched 6 databases in March 2010 using the key terms “knee hip arthroplasty/replacement,” “hip arthroplasty/replacement,” and “smoking” or “tobacco” (Appendix). All databases were searched from inception to March 2010 and included The Cochrane Library, Wiley InterScience (www.thecochranelibrary.com), to include the Cochrane Central Register of Controlled Trials (CENTRAL), Health Technology Assessment Database (HTA), and Database of Abstracts of Reviews of Effects (DARE), Scopus, Proquest Dissertation abstracts, CINAHL (via EBSCOHost), Ovid Medline, and Embase. Inclusion of these standard databases is recommended by the Cochrane Handbook of Systematic Reviews12 and is similar to previous Cochrane systematic reviews13,14; these databases were chosen by an experienced Cochrane librarian (LF).
All titles and abstracts were screened by 2 independent trained reviewers (J. Singh, A. Bharat) with expertise in performing systematic reviews15,16,17. Studies were included if they were fully published reports that included smoking and any perioperative or postoperative clinical outcome in patients with either TKA or THA. Studies were excluded if they were abstracts, reviews, or editorials, or did not provide clinical outcomes data. Since the study aimed at assessing smoking as a risk factor for poor postsurgical outcomes, studies of interventions targeting smoking were not included. A Cochrane systematic review of efficacy of smoking cessation interventions has been published18 that included randomized studies of smoking cessation. The Cochrane review combined multiple studies across several surgeries including arthroplasty and provided estimates of benefit for preoperative smoking cessation programs.
Each study was reviewed and pertinent data were extracted using a standardized data collection form. Study characteristics were extracted, including the author, year of publication, number of patients, number of joints, patient demographics (age, race, sex, body mass index), followup duration, type of arthroplasty (hip vs knee vs both), type of study (cross-sectional, cohort), setting (single vs multicenter), population, and definition of smoking exposure (duration; pack-years, defined as number of packs per day times number of years smoked).
Each included study was assessed for quality using the Newcastle-Ottawa scale for assessment of observational studies19. This scale is designed for quality assessment of observational studies, with separate scales for case-control and cohort studies. Four points could be assigned for selection of appropriate cases/controls, 2 points for comparability of cases and controls, and 3 points for adequacy of exposure assessment. The score can range from 0 to 9, 9 representing the best quality score.
We planned to examine the studies for TKA and THA combined, since there was no reason to believe that postoperative complication risk would differ by the type of joint undergoing replacement. We had planned to analyze TKA and THA separately in subgroup analyses, in case significant heterogeneity was evident; however, due to lack of obvious heterogeneity and small number of studies available for quantitative analyses, we analyzed the groups together. We compared the number of patients with complications in each group and calculated the relative risk and 95% confidence interval (CI) using the Mantel-Haenszel method. A p value < 0.05 was considered statistically significant, which is similar to relative risk excluding unity.
We also calculated the number needed to harm (NNH) related to smoking as a risk factor. This represents the number of patients exposed to the risk (current smoking, former smoking) that leads to one extra patient with an unfavorable outcome (for example, death). This was calculated by obtaining the inverse of the absolute risk difference between current smokers and non-smokers and between former smokers and nonsmokers, using the Cates calculator Visual Rx20. Absolute risk difference was defined as the difference between risk in the treatment group and risk in the control group. The 95% CI for NNH was calculated by taking the inverse of the 95% CI of the absolute risk difference, where the lower 95% CI for absolute risk reduction becomes the higher 95% CI for NNH and vice versa.
RESULTS
Of the 516 titles and abstracts, 45 qualified for the full text review after duplicate independent review (Figure 1). Of these, 24 studies were excluded: 7 were reviews, commentaries, or editorials21,22,23,24,25,26,27; 8 did not contain outcomes of interest or did not have smoking as a predictor28,29,30,31,32,33,34,35; 1 was related to economic outcome36; 1 was a qualitative study related to a patient survey37; 1 thesis was not available for review38; 4 were related to smoking cessation39,40,41,42; 1 was a duplicate article43; and 1 was related to arthroscopy44. A total of 21 studies from the original search met the inclusion and exclusion criteria and data were extracted from these. Of the 21 included studies45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65, 13 included patients with THA45,49,50,51,52,53,54,56,59,60,61,62,63, 4 studies included TKA47,55,58,64, 3 included patients with THA or TKA46,48,57, and 1 was a statistical modeling exercise65. No additional studies were identified from the reference lists of included studies.
Selection of reports for study.
Overview of epidemiology studies
Table 1 summarizes demographic characteristics of patient populations in the 21 observational studies45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65, of which one was a statistical modeling study65 (which provided no usable data). Most studies compared smokers to nonsmokers, with few studies categorizing patients into former, current, and nonsmokers. Association of pack-years (cumulative smoking exposure) with outcomes was reported in a few studies.
Patient characteristics of studies included for review.
Table 2 provides a summary of the main findings from the study. We present data from studies that provided multivariable analyses45,46,47,48,49,50,51 followed by those that performed univariate analyses52,53,54,55,56,57,58,59,60,61,62,63,64. The quality score for each study is based on the Newcastle-Ottawa scale (range 0–9) provided in Table 2. In general, most studies with multivariable-adjusted analyses scored higher than those with only univariate analyses, although a wide range in scores was noted. This is followed by pooled estimates from these studies with calculation of number needed to harm (Table 3).
Findings from observational studies.
Unadjusted postoperative outcomes following total hip arthroplasty comparing current and former smokers to nonsmokers (reference category for both groups).
Observational studies assessing smoking as a risk factor; multivariable-adjusted analyses
This section summarizes studies that provided multivariable-adjusted analyses. Sadr Azodi, et al studied 3309 patients with primary THA using the Swedish Inpatient Register45. It was found that 373 of the 3309 patients (11%) developed one or more postoperative complications. Smoking was significantly associated with increase in risk of systemic complications. In multivariable-adjusted analyses, former and current smokers had significantly higher multivariable-adjusted odds of systemic complications compared to nonsmokers, with odds ratios of 1.32 (95% CI 1.04, 1.97) and 1.56 (95% CI 1.14, 2.14), respectively. Higher number of pack-years was significantly associated with systemic complications. Patients who smoked > 40 pack-years were significantly more likely than nonsmokers to have systemic complications (OR 2.21, 95% CI 1.28, 3.82). Those smoking 0–19.9 pack-years were not significantly associated with systemic complications compared to nonsmokers (OR 1.35, 95% CI 0.99, 1.84).
Lavernia, et al compared hospital charges following primary or revision THA or TKA between former, current, and nonsmokers46. Smoking was a significant predictor for charges (p = 0.03), operative time (p = 0.01), and anesthesia time (p = 0.003). Fisher, et al found that smoking was protective and associated with lower risk of painful/stiff knee in 71 patients who underwent TKA47. Moller, et al performed an observational study of 811 patients with THA or TKA48. Analyses were adjusted for age, sex, body mass index (BMI), ASA class, comorbidity, surgery type, duration of anesthesia, duration of surgery, and alcohol use. Smoking was significantly associated with higher risk of any complication (OR 3.2, 95% CI 1.8 to 6.0), wound complication (OR 8.5, 95% CI 1.6 to 47), and admission to intensive care unit (OR 2.4, 95% CI 1.4 to 3.8).
Espehaug, et al studied 536 patients who had undergone a primary and revision THA (cases) and 1092 patients who had undergone primary THA only (controls)49. Smoking was categorized as current, former, and nonsmoker and pack-years were recorded. Overall, neither former nor current smoking was associated with increased risk of revision surgery. Former heavy smokers (≥ 12 pack-years) had significantly higher risk of revision (OR 2.6, 95% CI 1.5 to 4.4). These associations did not change after adjustment for alcohol intake, occupation, weight, and height. Sadr Azodi, et al studied 2106 patients who underwent THA using the Swedish Inpatient Register50. In multivariable analyses that adjusted for age, calendar period, BMI, and fixation, there was no significant association of smoking status or pack-years of smoking and risk of implant dislocation up to 3 years after the primary THA. Meldrum, et al compared outcomes in 147 patients undergoing THA51. In multivariable models adjusted for age, sex, BMI, diagnosis, stem fixation, and alcohol use, smoking had a nonsignificant borderline association with higher revision rates (hazard rate 4.5; p = 0.006).
Observational studies assessing smoking as a risk factor; univariate analyses
All remaining studies used univariate analyses, i.e., associations of smoking with the outcomes were not adjusted for important confounders. Bischoff-Ferrari, et al studied the predictors of poor functional status, defined as WOMAC (Western Ontario and McMaster Universities Osteoarthritis Index) score < 50, in a sample of Medicare recipients 3 years after THA52. In univariate analyses, current smoking was not associated with poor functional status (OR 0.8, 95% CI 0.3–2.0)52. Anderson, et al examined the predictors of postoperative complications including venous thromboembolism (VTE) in patients undergoing THA53. Smoking was not associated with risk of VTE. Beksac, et al studied the risk of VTE in 1947 patients undergoing elective THA at a single medical center54; smoking was not a significant risk factor for loosening in univariable or multivariable models (p = 0.1 p = 0.3, respectively). Sharrock, et al found no association of smoking and deep venous thrombosis in 458 patients who underwent cemented TKA55. Horne, et al found that smoking was not associated with risk of infection after TKA in their study of 40 patients56. Malinzak and Ritter examined the infection rate in 17,561 total joint replacements57; smoking was not a significant correlate of postoperative wound infections.
Cates, et al examined the factors predictive of results of closed manipulation following TKA58. Smoking was not associated with any improvement in 1-year flexion or 1-year extension following closed manipulation. Grosflam, et al found that smoking status was not a significant predictor of intraoperative blood loss in 295 patients undergoing primary elective THA59. Inoue, et al examined 151 THA in 130 patients60. There was no significant association of smoking and implant loosening (OR 0.30, 95% CI 0.06 to 1.52). Nixon, et al surveyed 127 patients with cemented THA regarding smoking status; active smoking was not associated with loosening (OR 5.5, 95% CI 0.8, 38)61. Khan, et al studied 1767 patients undergoing THA followed up to 5 years; smoking was not associated with deep infection (OR 3.4, 95% CI 0.7 to 17.2), superficial infection (OR 1.0, 95% CI 0.5 to 1.8), and revision (OR 0.8, 95% CI 0.1 to 2.6)62. Malik, et al found no association of smoking and aseptic loosening in 224 patients who underwent cemented THA63.
In a retrospective single-center study by Peersman, et al, 113 infections were identified in 6489 TKA64. Smoking was reported to be significantly associated with the risk of infection following TKA (p = 0.01)64.
Pooled unadjusted estimates of risk and number needed to harm (NNH)
Table 3 provides the overall unadjusted risk ratios for 5 main postoperative outcomes comparing current and former smokers to nonsmokers. Compared to nonsmokers, the current smokers had a 24% higher risk of any post-operative complication and former smokers a 32% higher risk. Similarly, the risk of death at 3-year followup was 63% higher in current smokers and 69% higher in former smokers, with data from one study. The NNH for various complications is shown in Table 3.
Knowledge gap analysis
We found that several studies of small sample size have assessed the association of smoking with complications after TKA or THA. Most studies performed univariate analyses, and therefore residual confounding due to other important factors is an important limitation of these studies. Very few studies had longterm followup beyond 5–10 years to assess risk of revision surgery after TKA/THA. A majority of studies did not explicitly define the complication. We noted that very few studies investigated the association of duration of smoking exposure and the current amount of smoking with various postoperative complications. Several complications such as gastrointestinal complications, admission to the intensive care unit, reintubation, poor function and pain, and quality of life outcomes after arthroplasty have not been studied well. Healthcare costs and resource utilization outcomes were also rarely studied as related to smoking.
DISCUSSION
We found that, compared to nonsmokers, current smokers and former smokers had 24% and 32% higher risk, respectively, of any postoperative complication after TKA or THA. Similarly, the risk of death was 62% higher in current smokers and 69% higher in former smokers 3 years after TKA/THA, compared to nonsmokers. These (unadjusted) estimates should serve as guides to healthcare providers in their discussions with patients undergoing THA or TKA regarding smoking-related risks. In conjunction with findings from a recent Cochrane systematic review of efficacy of intensive preoperative smoking cessation interventions in reducing any complication and surgical site infections18, this observation has significant implications. Patients can be informed that their risk of postoperative complications is significantly increased by smoking and that quitting smoking can significantly reduce risk of these preventable complications. The benefits of smoking cessation are particularly greater in those with an intensive program (starting 4–8 weeks preoperative with at least one face-to-face interaction weekly), compared to shorter less intensive programs18. Clinicians have viewed the perioperative period as a “window of opportunity” to motivate patients to quit smoking66. Since > 95% of joint arthroplasty surgeries are elective, the preoperative period provides an opportunity for discussion with the patient and enough time in most cases to implement an intensive tobacco cessation program. Of course, the short- and longterm benefits of smoking cessation on cardiac, pulmonary, and overall health will then be realized, if patients can remain off tobacco after their joint replacement surgery.
Implications for clinical practice
The increased risk for mortality among current and former smokers is particularly impressive, considering the elective aspect of TKA/THA. For all candidates for TKA/THA, the postoperative mortality is extremely low, < 1% at 90 days67. Since smokers have a significantly increased risk of mortality, this risk merits a discussion with all smokers planning to undergo elective knee or hip arthroplasty. There are currently no data that have shown differences in perioperative mortality in patients quitting smoking preoperatively; however, such a study may require a high number of patients to detect differences in mortality, since the baseline risk of mortality is low in all candidates. Even in the absence of such data, with the current knowledge of increased risk of mortality with smoking and all other negative health effects of smoking, it seems prudent to discuss smoking cessation interventions with patients prior to joint arthroplasty. The NNH of 34–50 for current and former smokers for immediate postoperative complication and 34 for death at 3 years can be interpreted by comparing to NNH from other studies. For example, using the data from the Women’s Health Initiative68,69, the NNH with hormone replacement therapy was 1250 for stroke and 1250 for pulmonary embolism after 1 year of treatment, and 238 for breast cancer after 5.2 years of treatment. Thus smoking in the perioperative period is a significant risk factor for poor outcomes and is a potential target for intervention.
In our review of observational studies in patients with THA or TKA, a higher complication rate was noted in smokers. Current smokers and former smokers were both at higher risk than nonsmokers. We did not observe a linear trend of increased risk of postoperative complications from former smokers to current smokers in pooled unadjusted analyses, which may be more likely related to limited data (from 1 or 2 studies) and residual confounding, rather than lack of such a trend. Current smokers tend to be a younger cohort than nonsmokers and former smokers due to survival disadvantage, and confounding bias can lead to falsely lower unadjusted rates in current smokers. This is further evident by an example from one of the studies by Sadr Azodi, et al45. Compared to nonsmokers, unadjusted odds ratios for systemic complications in current and former smokers were 1.39 and 1.50, respectively (numerically higher for former smokers), whereas respective adjusted odds ratios were 1.56 and 1.32 (numerically higher for current smokers). Therefore, the unadjusted pooled risk ratios shown in Table 3 must be interpreted with some caution and with consideration of this confounding bias.
A recent Cochrane systematic review of randomized controlled trials of preoperative smoking cessation that included various surgical populations provides convincing data regarding risk reduction with smoking cessation18. An intensive 4–8 week preoperative smoking cessation program was associated with 58% risk reduction for any complication (RR 0.42. 95% CI 0.27–0.65) and with 69% risk reduction for wound complications (RR 0.31, 95% CI 0.16–0.62) compared to a control group. Based on this evidence and the fact that most knee and hip arthroplasties are elective, discussing preoperative smoking cessation with patients undergoing TKA and THA seems prudent. Although there is some evidence for it, further studies should examine whether there is a clear dose relationship with amount of smoking (pack-years) and outcomes.
The evidence for the association of smoking with postoperative outcomes was derived from several multivariable-adjusted studies45,46,47,48,48,50, with few univariate analyses43,51,52,53,54,55,56,57,58,59,60,61,62,63. Multivariable-adjusted analyses also showed association of smoking with higher resource utilization among former smokers and current smokers compared to nonsmokers, and were mostly of higher quality compared to univariate analyses. Therefore the evidence must be interpreted considering the study quality, with more confidence in evidence emerging from higher quality studies.
Implications for research
Our study highlights that several knowledge gaps exist in this area. The evidence from included studies is suggestive but not conclusive that smoking is associated with poor postsurgical outcomes in patients undergoing THA or TKA. Most studies to date had small sample sizes, and smoking exposure as a risk factor was measured differently (current, former, and never-smoker in some, and smoker vs nonsmoker in others). Few studies examined pack-years to determine dose relationships, but more data are needed. Due to lack of data, it is unknown whether smoking is associated with longterm implant-related complications. Future studies needed to address these knowledge gaps.
We performed a systematic review of the literature regarding smoking and postoperative complications following elective THA or TKA. We found that smoking is a risk factor for higher postoperative complications and postoperative mortality. The number of pack-years also seems to be related to risk of these complications. More research is needed to determine the association of smoking with other postoperative complications, including cardiac and pulmonary complications, prosthetic loosening, and infection in patients undergoing THA or TKA. Evidence from other randomized trials shows the benefit of intensive preoperative smoking cessation programs in reducing complications. Research is also needed to study the most optimal time for quitting smoking preoperatively and assess its influence on these immediate postoperative and later outcomes.
Acknowledgment
I thank Louise Falzon from the Cochrane Library for performing the search and Assem Bharat for reviewing abstracts and titles for inclusion in the study.
APPENDIX 1. Details of the search strategy
Footnotes
-
The views expressed in this article are those of the author and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the United States government.
-
Supported by National Institutes of Health (NIH) Clinical Translational Science Award 1 KL2 RR024151-01 (Mayo Clinic Center for Clinical and Translational Research).
- Accepted for publication April 4, 2011.
