A Carbon Monoxide Breathalyzer is Effective at Identifying Smokers Before Total Joint Arthroplasty Authors: Jacquelina Achkar, Michael Tanzer, Nawaf Alamiri, Sean Gilman, Claudine Després, Adam Hart Published: 2026 ABSTRACT BackgroundSmoking is a major risk factor for delayed wound healing and prosthetic joint infections, yet it is poorly identified and modified before surgery. Patients often omit or downplay their consumption, and objective methods of screening for smoking are lacking. Carbon monoxide (CO) breath testing is a validated, noninvasive tool to detect recent smoking but has yet to be used to screen patients planned for total joint arthroplasty (TJA). This study compares self-reported smoking status to a CO breathalyzer test in patients listed for TJA.MethodsWe conducted a cross-sectional study of patients with end-stage arthritis listed for TJA surgery at a single academic center. All participants completed a standardized smoking questionnaire and then underwent CO breath testing using a validated handheld breathalyzer. CO levels were categorized as low (0-6 ppm), borderline (7-9 ppm), or high (≥10 ppm). The primary outcome was the concordance between self-reported smoking status and CO levels.ResultsA total of 153 patients were enrolled, of whom 29 self-identified as active smokers, 14 as exposed nonsmokers (ENS), and 110 as nonsmokers. Smokers exhibited the highest mean CO levels (13.6 ± 7.5 ppm), followed by ENS patients (9.0 ± 7.4 ppm), while nonsmokers had significantly lower levels (1.6 ± 1.6 ppm; P < .001). In the ENS group, 42.9% had high CO levels, indicating possible discordance between self-report and biochemical testing.ConclusionsCO breath testing is a low-cost, noninvasive tool that successfully differentiated between self-reported active smokers and nonsmokers among patients who are planned to undergo a TJA. Some patients who identified as nonsmokers with an unclear exposure history were found to have high CO levels upon testing. Integrating CO screening into preoperative assessment may enhance identification of recent smoke exposure and support targeted counseling prior to TJA. Introduction Over 7 million people across the United States live with a total hip or knee arthroplasty [1]. As the incidence of these highly successful surgeries continues to rise, there is growing urgency to address prosthetic joint infections (PJIs), which are now the primary cause for revision across multiple national registries [2,3]. Treatment of PJI is highly morbid to patients and a growing burden to the healthcare system [4]. Therefore, strategies to prevent PJI are of paramount interest and include rigorous preoperative optimization of modifiable patient-specific risk factors [5]. Smoking is still relatively prevalent among patients presenting for arthroplasty and is one of the most important but also one of the least studied risk factors for PJI [6]. Smoking significantly impacts the outcome of total joint arthroplasty (TJA) surgery [7]. Cigarette smoke contains toxins that inhibit the body’s natural healing processes by increasing inflammation. Tobacco use reduces blood and oxygen flow to the surgical site, slows the healing process, and increases the risk of complications such as infection, wound dehiscence, and aseptic loosening [[8], [9], [10], [11]]. Individuals who actively smoke cigarettes are at a 1.5 to 3.2 times higher risk of experiencing wound-related complications after TJA [12,13]. In addition to PJI, smokers have a greater chance of suffering medical complications following TJA, such as deep vein thrombosis, pulmonary embolism, pneumonia, hospital readmission, and even mortality [14,15]. Smoking also interferes with the effectiveness of pain medications, as it can reduce perfusion to the surgical site. Arthroplasty patients who smoke tobacco exhibit inadequate pain management, requiring significantly increased amounts of narcotic pain medicine, compared to nonsmokers [16]. Most importantly, the deleterious effects of smoking are modifiable, and patients can vastly improve their perioperative outcomes by quitting smoking before their surgery [10]. The truthfulness of patients self-reporting their smoking status before undergoing TJA is complex due to social stigma and perceived consequences, such as their noneligibility for surgery if they admit to smoking [17]. Many smokers feel ashamed or embarrassed about their addiction and may be reluctant to report smoking (or the extent of their smoking) to their surgeon. Meanwhile, timely identification of smokers before TJA is critical for proper counseling and implementation of multidisciplinary smoking cessation strategies to help patients quit smoking before surgery. Serum cotinine has been used to help screen for smoking and helped identify 46% of smokers who did not self-report smoking [18]; however, these tests are expensive, require a blood or urine sample, and cannot differentiate between active smoking and nicotine replacement therapies. Carbon monoxide (CO) breath testing is a simple, inexpensive, and noninvasive point-of-care test that can accurately estimate a patient’s level of CO in their blood. This validated test is quantitative and can differentiate if someone is an active smoker, was exposed to smoke within the last 48 to 72 hours, or is a nonsmoker [[19], [20], [21]]. The CO breath test is used routinely by respirologists in smoking cessation clinics [22] but has not been reported as a screening tool for TJA. The objective of this study was to use the CO breathalyzer to screen patients with end-stage arthritis who are being scheduled for TJA and to compare it with their self-reported smoking status. We hypothesized that CO breath testing will improve the identification of active smokers. Material and methods A cross-sectional study of 153 patients with end-stage arthritis listed for TJA surgery at a single academic center was conducted. All patients scheduled for a primary total hip or knee arthroplasty from April 2021 to October 2022 were recruited into the study. We excluded patients who required revision surgery and all nonelective surgeries (emergency surgeries). The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the institutional review board of the McGill University Hospital Centre (MUHC) (Approval Code: 2020-6424; Approval Date: October 14, 2020). Written informed consent was obtained from all subjects involved in the study, as required by the MUHC Research Ethics Board. Data collection Patients first completed a validated smoking questionnaire from the MUHC Smoking Cessation Clinic to determine their detailed cigarette smoking history, including the patient's current smoking status, quantity of smoking, history of smoking, exposure to secondhand smoking, and marijuana use (Appendix Figure A). Patient demographic data including sex, age, patient body mass index (BMI), and the type of surgery they were scheduled to have between a total hip arthroplasty and total knee arthroplasty was also collected and included in the analysis. Upon enrollment, a CO breath test was then performed on each patient using the MicroCO Breath CO Monitor (MD Spiro by Microdirect Inc., Lewiston, ME) (Fig. 1).Figure 1Carbon monoxide (CO) breath monitor used for smoking status assessment. Photograph of the portable CO breath analyzer device (MicroCO Breath CO Monitor, MD Spiro by Microdirect Inc., Lewiston, ME) used to measure exhaled CO levels in participants. The device provides real-time CO readings in parts per million (ppm). This photograph was taken at the McGill University Health Centre (MUHC), Montréal, Québec, Canada. Carbon monoxide (CO) breath monitor used for smoking status assessment. Photograph of the portable CO breath analyzer device (MicroCO Breath CO Monitor, MD Spiro by Microdirect Inc., Lewiston, ME) used to measure exhaled CO levels in participants. The device provides real-time CO readings in parts per million (ppm). This photograph was taken at the McGill University Health Centre (MUHC), Montréal, Québec, Canada. CO levels were categorized as low (0-6 ppm), borderline (7-9 ppm), or high (≥10 ppm) based on previous literature [19]. Patients were then separated into 3 groups based on their self-reported cigarette smoking status: the smokers, the nonsmokers, and the exposed nonsmokers (ENS), who were patients who self-reported as nonsmokers but with exposure to secondary smoking at home or marijuana use. The primary outcome was the concordance between self-reported smoking status and CO levels. Within our smokers group, we subcategorized them into 3 groups according to how many cigarettes they smoked per day: 1-5 cigarettes, 6-10 cigarettes, and >10 cigarettes smoked per day for our subgroup analysis. Data analyses All analyses were conducted using SPSS Statistics version 29 (IBM Corp., Armonk, NY). Descriptive statistics summarized patient characteristics, with continuous variables presented as means and standard deviations (SDs) and compared using one-way analysis of variance. Categorical variables were reported as counts and percentages and compared using chi-square tests. Missing data were handled by case-wise deletion, and only patients with complete smoking status and CO data were included in our analyses. Concordance between self-reported smoking status and CO levels was assessed descriptively. In our subgroup analysis, the correlation between the number of cigarettes smoked per day and CO levels was analyzed using Pearson correlation, excluding the self-reported nonsmokers and ENS who reported smoking zero cigarettes. A statistical significance threshold of P < .05 was applied for all statistical comparisons. To assess the diagnostic accuracy of CO breath testing in detecting active smoking, self-reported smoking status was used as the reference standard. A CO cutoff of >6 ppm was applied to indicate biochemical evidence of smoking [19]. A 2 × 2 contingency table was generated to classify participants as true positives (self-reported smokers with positive CO), false positives (all self-reported nonsmokers with positive CO), false negatives (smokers with negative CO), and true negatives (nonsmokers with negative CO). Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated accordingly. A multivariable linear regression analysis was performed to identify factors independently associated with CO levels. Smoking status, age, BMI, sex, and marijuana use were included in the model based on their clinical relevance and potential confounding effects. Regression coefficients (β coefficients) with 95% confidence intervals (CIs) were reported. Results This study included 153 patients (93 women and 60 men) who were all awaiting TJA surgery (51 total hip arthroplasty and 77 total knee arthroplasty). The mean age among our study population was 66.1 years (SD: 11.0). The mean BMI was 29.2 kg/m2 (SD: 5.4). Based on self-reported smoking status, 71.9% of patients (n = 110) self-identified as nonsmokers, 19.0% (n = 29) as active smokers, and 9.2% (n = 14) were categorized as ENS, who self-identified as nonsmokers but were exposed to secondhand smoking or used marijuana. Marijuana use was significantly different between groups (P < .001), with 27.6% of cigarette smokers and 50.0% of ENS patients reporting marijuana use, compared to 0% among nonsmokers. Smokers were younger on average with a mean age of 60.3 years (SD: 10.7), compared to nonsmokers who had a mean age of 68.1 years (SD: 10.7) and ENS patients who had a mean age of 61.9 years (SD: 8.8) (P < .001). No difference in BMI was found between our 3 groups (P = .100). There was also an overall significant difference in marijuana use among our 3 groups with increased use among our smokers and ENS compared to our nonsmokers (P < .001) (Table 1). Smokers had the highest mean CO level (13.6 ± 7.5 ppm), followed by the ENS (9.0 ± 7.4 ppm), while nonsmokers exhibited markedly lower levels (1.6 ± 1.6 ppm) (P < .001) (Fig. 1).Table 1Patient demographic data and baseline characteristics by self-reported smoking status.Patient characteristicAll patientsSmokersNonsmokersENSP valuen1532911014Women, n (%)93 (60.8)16 (55.2)72 (65.5)5 (35.7).607Age, y (mean ± SD)66.1 ± 11.060.3 ± 10.768.1 ± 10.761.9 ± 8.8<.001Type of surgeryb THA51 (33.3)13 (44.8)31 (28.1)7 (50.0) TKA77 (50.3)10 (34.5)62 (56.3)5 (35.7)BMIa, kg/m2 (mean ± SD)29.2 ± 5.427.2 ± 5.529.8 ± 5.128.7 ± 6.2.100CO level, ppm (mean ± SD)4.5 ± 6.413.6 ± 7.51.6 ± 1.69.0 ± 7.4<.001Marijuana use, n (%)15 (9.8)8 (27.6)0 (0.0)7 (50.0)<.001THA, total hip arthroplasty; TKA, total knee arthroplasty; ppm, parts per million.aBMI was available for 136 patients due to missing data. All patients had valid CO level measurements.bThe scheduled type of surgery was available for 128 patients due to missing data. All patients had valid CO measurements. Table 1Patient demographic data and baseline characteristics by self-reported smoking status.Patient characteristicAll patientsSmokersNonsmokersENSP valuen1532911014Women, n (%)93 (60.8)16 (55.2)72 (65.5)5 (35.7).607Age, y (mean ± SD)66.1 ± 11.060.3 ± 10.768.1 ± 10.761.9 ± 8.8<.001Type of surgeryb THA51 (33.3)13 (44.8)31 (28.1)7 (50.0) TKA77 (50.3)10 (34.5)62 (56.3)5 (35.7)BMIa, kg/m2 (mean ± SD)29.2 ± 5.427.2 ± 5.529.8 ± 5.128.7 ± 6.2.100CO level, ppm (mean ± SD)4.5 ± 6.413.6 ± 7.51.6 ± 1.69.0 ± 7.4<.001Marijuana use, n (%)15 (9.8)8 (27.6)0 (0.0)7 (50.0)<.001THA, total hip arthroplasty; TKA, total knee arthroplasty; ppm, parts per million.aBMI was available for 136 patients due to missing data. All patients had valid CO level measurements.bThe scheduled type of surgery was available for 128 patients due to missing data. All patients had valid CO measurements. Patient demographic data and baseline characteristics by self-reported smoking status. THA, total hip arthroplasty; TKA, total knee arthroplasty; ppm, parts per million. BMI was available for 136 patients due to missing data. All patients had valid CO level measurements. The scheduled type of surgery was available for 128 patients due to missing data. All patients had valid CO measurements. Among self-reported smokers (n = 29), 76% of the group exhibited high CO levels (≥10 ppm). Only 14% had low CO levels, and 10% fell within the borderline range. In contrast, all self-reported nonsmokers (n = 110) demonstrated low CO levels, with 0% in the borderline or high categories. The ENS group (n = 14) showed a mixed distribution with 43% having low CO levels, 14% with borderline levels, and 43% having high CO levels. A subgroup analysis among the self-reported smokers group was performed to assess the relationship between the number of cigarettes smoked per day and CO level. A positive correlation was observed between the number of cigarettes smoked per day and CO levels (r = 0.57 and P < .001). Smokers who reported smoking >10 cigarettes per day exhibited the highest mean CO levels, whereas those smoking 1-5 cigarettes per day had lower CO levels (Fig. 3).Figure 2Mean carbon monoxide (CO) levels (in ppm) among the smokers, the nonsmokers, and the exposed nonsmokers (ENS) groups.Figure 3Distribution (in %) of carbon monoxide levels among each self-reported smoking group.Figure 4Mean CO level in correlation with the number of cigarettes smoked per day in the smokers group. Mean carbon monoxide (CO) levels (in ppm) among the smokers, the nonsmokers, and the exposed nonsmokers (ENS) groups. Distribution (in %) of carbon monoxide levels among each self-reported smoking group. Mean CO level in correlation with the number of cigarettes smoked per day in the smokers group. In the smokers group (n = 29), 25 of them (86.2%) blew a positive CO level (>6 ppm), while 4 (13.8%) had a negative result. All the self-reported nonsmokers (n = 110) had negative CO levels. Among the ENS group (n = 14), all blew positive CO values. The sensitivity of CO breath testing was calculated as 86.2% (95% CI: 69.4-94.5), the specificity as 88.7% (95% CI: 81.9-93.2), the PPV as 64.1% (95% CI: 48.4-77.3), and the NPV as 96.5% (95% CI: 91.3-98.6) (Table 2).Table 2Contingency table of self-reported smoking status and CO breath test results.Patient characteristicCO positive (>6 ppm)CO negative (≤6 ppm)TotalSelf-reported smokers (%)25 (86.2)4 (13.8%)29Self-reported nonsmokers (incl. ENS) (%)14 (11.3)110 (88.7)124Total39114153 Table 2Contingency table of self-reported smoking status and CO breath test results.Patient characteristicCO positive (>6 ppm)CO negative (≤6 ppm)TotalSelf-reported smokers (%)25 (86.2)4 (13.8%)29Self-reported nonsmokers (incl. ENS) (%)14 (11.3)110 (88.7)124Total39114153 Contingency table of self-reported smoking status and CO breath test results. In multivariable linear regression analysis, smoking status remained independently associated with higher CO levels after adjustment for age, BMI, sex, and marijuana use (β = 3.87, 95% CI: 1.55-6.19, P = .001). Marijuana use was not independently associated with CO levels after adjustment (β = 2.60, 95% CI: −1.71 to 6.90, P = .234). Similarly, age (β = −0.06, 95% CI: −0.18 to 0.06, P = .351), BMI (β = −0.18, 95% CI: −0.43 to 0.06, P = .139), and sex (β = −1.16, 95% CI: −4.10 to 1.79, P = .436) were not independently associated with CO levels (Table 3).Table 3Multivariable linear regression analysis of factors associated with carbon monoxide (CO) levels.Patient characteristicβ coefficient (B)95% CIP valueSex (male vs female)a0.03−2.08 - 2.13.981Age (per y)−0.06−0.15 - 0.04.224BMI (per kg/m2)−0.17−0.36 - 0.03.088Smoking status (smoker vs nonsmoker)a4.122.14 - 6.10<.001Marijuana use (yes vs no)a2.60−1.71 - 6.90.234aReference categories: female for sex, nonsmoker for smoking status, no for Marijuana use. Table 3Multivariable linear regression analysis of factors associated with carbon monoxide (CO) levels.Patient characteristicβ coefficient (B)95% CIP valueSex (male vs female)a0.03−2.08 - 2.13.981Age (per y)−0.06−0.15 - 0.04.224BMI (per kg/m2)−0.17−0.36 - 0.03.088Smoking status (smoker vs nonsmoker)a4.122.14 - 6.10<.001Marijuana use (yes vs no)a2.60−1.71 - 6.90.234aReference categories: female for sex, nonsmoker for smoking status, no for Marijuana use. Multivariable linear regression analysis of factors associated with carbon monoxide (CO) levels. Reference categories: female for sex, nonsmoker for smoking status, no for Marijuana use. Discussion This cross-sectional study found that the CO breath test is a simple, objective, and inexpensive tool that can be used to identify active tobacco consumption in patients awaiting TJA surgery. To our knowledge, this is the first study that used CO breath testing preoperatively to screen patients who are awaiting TJA and compare it to a self-reported smoking questionnaire. In our cohort of patients, there was a statistically significant difference (P < .05) in CO levels between our 3 groups based on their self-reported smoking status. Our smokers group had the highest mean CO levels at 13.6 ± 7.5 ppm with over 75% of them who had high CO levels (>10 ppm). The 14% of the smokers who had low CO levels (0-6 ppm) may be explained by the fact that these participants, who self-identified as smokers, were actively attempting to quit at the time of enrollment and had not smoked in the 24 to 48 hours prior to taking the breath test. Within this same group, we also found a positive correlation between the numbers of cigarettes smoked and the CO level on the breathalyzer. These results match results found in the literature, with higher CO levels blown by heavier smokers [23,24]. Additionally, our self-identified nonsmokers group had the lowest mean CO levels at 1.6 ± 1.6 ppm, with 100% of these patients having CO levels in the low category. Interestingly, there were mixed results in the ENS group, who self-identified as nonsmokers. These patients had a mean CO level of 9.0 ± 7.4 ppm, and more than half of them had high or borderline levels of CO on the breath test (>7 ppm). These results indicate that these patients have likely smoked within 24 hours of their breathalyzer test [19,21] and underscore the limitations of relying solely on self-reporting. Marijuana use, captured as a separate variable, may have contributed to CO elevation in this group. Several studies have shown that both exclusive marijuana smoking and co-use with tobacco can result in increased exhaled CO levels, sometimes to a degree similar to that of tobacco smoking alone [[25], [26], [27]]. While self-reporting remains the standard in clinical practice, previous studies have shown that patients occasionally misrepresent their actual tobacco usage to their physicians. A systematic review by Gorber et al. [28] found that overall, patients tend to underestimate their smoking habits when comparing self-reporting to an objective biologic measure such as cotinine levels. Moreover, Attebring et al. [29] compared CO breath test levels to self-reported smoking habits in patients with coronary artery disease. They found that while most patients were truthful about their smoking habits, there remained patients who seemed to conceal or underestimate their true consumption to their physicians. In our cohort, CO breath testing demonstrated high sensitivity (86.2%) and specificity (88.7%) when compared against self-reported smoking status. The high NPV (96.5%) suggests that a negative CO breath test correlates with self-reported nonsmoking status. The moderate PPV (64.1%) can be attributed to the inclusion of the ENS in our calculations, all of whom had positive CO levels despite denying active smoking. Their self-reporting as nonsmokers artificially inflated the number of false positives, thereby decreasing the PPV of CO breath testing in our cohort. Smoking is a modifiable risk factor that increases the likelihood of postoperative complications including poor wound healing, infection, and delayed recovery [6,12]. CO breath testing provides a quick, inexpensive and noninvasive alternative to biological markers such as cotinine or thiocyanate, which require blood or urine samples for testing [17,30]. The clinical implications of this study’s findings are significant, since CO breath testing can be used as a point-of-care screening tool to identify patients requiring targeted counseling or intervention prior to surgery. Its integration could facilitate timely counseling, allow for smoking cessation interventions, and eventually help optimize perioperative outcomes by identifying modifiable risk factors more effectively than patient report alone. The ENS group in this study included individuals exposed to secondhand tobacco smoke as well as those reporting marijuana use, which represent biologically distinct exposures. Marijuana smoking has been shown to elevate exhaled CO levels and may therefore confound interpretation of CO breath testing. To further evaluate this potential confounding effect, we performed a multivariable linear regression analysis adjusting for age, BMI, sex, smoking status, and marijuana use. Smoking status remained independently associated with higher CO levels, while marijuana use was not independently associated with CO levels after adjustment. These findings suggest that elevated CO levels in our cohort are more strongly related to tobacco exposure than marijuana use alone. Nevertheless, given the heterogeneity within the ENS group, these findings should be interpreted with caution. This study has several limitations. It was conducted at a single center using a cross-sectional design, which limits generalizability and prevents assessment of long-term outcomes. Additionally, the study was not powered to detect differences in postoperative complication rates based on CO levels. The sample size within certain subgroups was relatively small, which can affect both the reliability and generalizability of subgroup analyses. Another limitation of this study is the use of self-reported smoking status as the reference standard for estimating diagnostic accuracy. Given that self-reporting is known to underestimate true tobacco exposure, the calculated sensitivity, specificity, PPV, and NPV should be interpreted with caution. These metrics likely reflect concordance with self-report rather than true diagnostic performance and may therefore overestimate the validity of CO breath testing. Future studies incorporating biochemical gold standards such as serum or urine cotinine would provide a more accurate assessment of CO breath testing performance. Despite these limitations, the observed associations suggest that CO breath testing may be a valuable adjunctive screening tool within multidisciplinary preoperative optimization pathways. Future research should aim to validate these findings across broader populations, assess longitudinal outcomes, and evaluate whether CO-guided interventions reduce complication rates and improve recovery after TJA. Over time, we envision CO breath testing as a longitudinal behavioral tool, used not only to verify smoking status before TJA but also as a guided intervention tool. Regular CO monitoring could provide immediate, objective feedback and could motivate patients awaiting TJA in their smoking cessation efforts. Conclusions CO breath testing is a simple, rapid, and noninvasive method that identifies recent smoking behavior in patients preparing for TJA. In this study, CO levels were significantly higher among self-reported smokers and positively correlated with the number of cigarettes smoked per day. Notably, patients in the ENS group, who self-identified as nonsmokers, exhibited elevated CO levels, underscoring the potential limitations of self-reported smoking histories. Given that smoking is a modifiable risk factor that has been well-documented to negatively affect postsurgical outcomes, our findings suggest that integrating CO breath testing into preoperative assessment offers an objective means to detect recent tobacco exposure, enhance patient counseling, and support targeted cessation efforts. Its potential for broader use across orthopaedic specialties and as a longitudinal tool for monitoring smoking reduction further reinforces its clinical utility in orthopaedics. Conflicts of interest A. Hart is on the speakers bureau/paid presentations for Stryker Canada; is a paid consultant for Stryker Canada; receives research support from Stryker Canada, Zimmer Biomet, and DePuy Synthes; receives other financial or material support from Irrimax; and is a board member/committee appointments for AO North America. M. Tanzer is a paid consultant for Stryker; receives other financial or material support from Zimmer/DePuy/Stryker; is a member of the medical/orthopaedic publications editorial/governing board at the Journal of Arthroplasty; and is a board member/committee appointments for Research Council for the AAOS; all other authors declare no potential conflicts of interest. For full disclosure statements refer to https://doi.org/10.1016/j.artd.2026.102089. CRediT authorship contribution statement Jacquelina Achkar: Writing – original draft, Investigation, Formal analysis. Michael Tanzer: Writing – review & editing, Conceptualization. Nawaf Alamiri: Methodology. Sean Gilman: Writing – review & editing, Conceptualization. Claudine Després: Investigation. Adam Hart: Writing – review & editing, Supervision, Data curation, Conceptualization.