Introduction. Although we know that smoking damages health, we do not know impact of smoking on a patient's outcome following primary knee arthroplasty (KA). In the UK, clinical commissioning groups (CCGs) have the authority (& funds) to commission healthcare services for their communities. Over the past decade, an increasing number of CCGs are using smoking as a contraindication for patients with end-stage symptomatic knee arthritis being referred to a specialist for due consideration of KA without any clear evidence of the associated risks & benefits. The overall objective of this study is to compare clinical outcomes after knee arthroplasty surgery in smokers, ex-smokers & non-smokers. Methods. We obtained data from the UK Clinical Research Practice Datalink (CPRD) that contains information on over 11 million patients (7% of the UK population) registered at over 600 general practices. CPRD data was linked to Hospital Episode Statistics, hospital admissions & Patient Reported Outcome Measures (PROMs) data. We collected data on all KAs (n=64,071) performed over a 21-year period (1995 to 2016). Outcomes assessed included: local & systemic complications (at 6-months post-surgery): infections (wound, respiratory, urinary), heart attack, stroke & transient ischaemic attack, venous thromboembolism, hospital readmissions & GP visits (1-year), analgesic use (1-year), surgical revision (up to 20-years), mortality (90-days and 1-year), & 6-month change from pre-operative scores in Oxford Knee Score (OKS). Regression modelling is used to describe the association of smoking on outcomes, adjusting for confounding factors. Results. Smoking was associated with an increased risk of lower respiratory tract infections (LRTI) (4.2% smokers vs. 2.7% non-smokers) (Odds Ratio (OR) 0.76, p-value 0.017). LRTI were similar in ex-smokers & smokers at 3.9%. There was no association with any of the other 6-month complications. Pain medication use over 1-year post surgery was higher in smokers compared to non-smokers: gabapentinoids 7.4% vs. 5.2% (OR 0.74, p< 0.001), opioids 45.9% vs. 35.3% (OR 0.79, p< 0.001), NSAIDs 51.6% vs. 46.1% (OR 0.91, p = 0.044). Mortality was higher in smokers at 1-year compared to non-smokers (hazard ratio (HR) 0.53, p<0.001) & ex-smokers (HR 0.65, p = 0.037), but there was no difference observed at 90-days. There was no association of smoking on revision surgery over 20-years follow up. Smoking was associated with worse postoperative OKS being 3.1 points higher in non-smokers (p<0.001) & 3.0 points higher in ex-smokers (p<0.001). The overall change in OKS before & after surgery was 13.9 points in smokers versus 16.3 points in non-smokers (p<0.001) & 15.7 points in ex-smokers (p<0.001). Over the year following surgery, smokers were more likely to visit their GP, but there was no association with hospital readmission rates. Conclusion. This is the largest study with linked primary care & secondary care data highlighting impact of a preventable patient factor on outcome of a routinely performed planned intervention. Smokers achieved clinical meaningful improvements in patient reported pain & function (OKS) following KA, although their attained post-operative OKS was lower than in non-smokers & ex-smokers. Levels of pain medication use were notably higher in both smokers & ex-smokers. As smokers achieved good clinical outcomes following KA surgery, smoking should not be a barrier to referral for or consideration of KA. However, the study does highlight particular risks a patient is taking if he/she continues to smoke when being considered for elective knee arthroplasty. This study will help the family
