REF-1470 · Xu et al., 2021 · EClinicalMedicine · Paper · PubMed · Free full text

Does smoking affect fracture healing?

Non-union odds were 2.5× higher in current or former smokers (19 studies, mostly surgical patients); delayed union and time to union showed no clear difference. Observational, not athletes.

Main caveat: Mostly non-randomised by the nature of the question: RCTs, observational studies and case series were eligible, three case series (89 patients) entered the…
fairly strong evidenceMeasured on non-athletes and recreational people

Xu and colleagues asked how smoking, preoperative cessation time, nicotine replacement, vaping and alcohol relate to healing after non-pathological fractures in adults, and how cessation, nicotine replacement and vaping relate to wound healing after any surgery. Of 122 studies, 71 entered meta-analyses.

Non-union (no healing beyond 6 months in the authors' definition) was more common in smokers: OR 2.50 (1.73-3.61). Delayed union (OR 1.35, 0.51-3.59; 3 studies, 117 patients) and time to union (mean difference 0.52 weeks, −1.29 to 2.33; 2 studies, 48 patients) were not significantly different.

Deep surgical site infection (OR 2.04) and undefined infection (OR 3.11) were higher in smokers. Stopping at least four weeks before surgery was associated with less wound infection than continuing (OR 0.37, 0.16-0.89; 6 studies, 914 patients), and at six weeks (OR 0.22). No nicotine-replacement or vaping study could be pooled.

For alcohol, non-union was not different (OR 0.97, 0.40-2.38; 3 studies, I² 77%), and one small study found longer time to union (+12.85 weeks). Our reading: a solid association for the hardest bone outcome and for deep infection, in surgical trauma patients. It does not show that fractures heal "slower" on average, and the cessation result is about wounds.

The authors' statement that stopping is beneficial for non-union goes further than their data on cessation.

Protocol

Search: MEDLINE, Embase, Cochrane CENTRAL, CINAHL, AMED to 9 August 2021; protocol registered, deviations listed.
Population: adults with non-pathological fractures for smoking and alcohol; any surgery for cessation, nicotine replacement and vaping.
Definitions: smokers include current or former smokers; non-smokers never smoked or under 100 cigarettes; delayed union 3-6 months; non-union over 6 months (changed from 9 months in the protocol).
Outcomes: delayed union, non-union, time to union; infection subtypes, malunion, normal union; for cessation, wound complications.
Funding and interests (as printed): China Scholarship Council (no. 201809120013); the competing-interest section lists none in the record ("None.").

Key Results

Outcome, smokers vs non-smokersOR or MD (95% CI)Studies, patientsI²
Non-unionOR 2.50 (1.73-3.61)19, 472664%
Delayed unionOR 1.35 (0.51-3.59)3, 1170%
Time to union (weeks)MD 0.52 (−1.29 to 2.33)2, 480%
Normal unionOR 1.60 (0.78-3.26)5, 4350%
Deep surgical site infectionOR 2.04 (1.68-2.48)17, 143650%
Undefined surgical site infectionOR 3.11 (2.09-4.63)18, 922666%
Superficial surgical site infectionOR 1.27 (0.73-2.21)4, 10590%
Stopping before surgery (any surgery)OR (95% CI)Studies, patientsI²
Wound infection, ≥4 weeks vs continued smoking0.37 (0.16-0.89)6, 91449%
Wound infection, ≥6 weeks vs continued smoking0.22 (0.06-0.83)as printed47%
Wound infection, ≥4 weeks vs non-smokers1.19 (0.43-3.32)79%
Impaired wound healing, ≥4 weeks vs non-smokers1.93 (1.17-3.20)2, 2930%
Alcohol: non-union OR 0.97 (0.40-2.38), 3 studies, 1607 people, I² 77%; time to union +12.85 weeks (1.23-24.47) in one study of 29; the authors call the evidence insufficient.
Nicotine replacement and vaping: no study could be included in the meta-analysis.
Numbers that do not add up: the Discussion says the review "confirms that smoking cessation is beneficial for nonunion and DSSI", while the same Discussion says cessation of four weeks or more had no effect on bone healing.

Mechanisms

Measured: associations only.
Authors' interpretation: nicotine as a vasoconstrictor lowers peripheral blood flow, carbon monoxide lowers oxygen-carrying capacity, hydrogen cyanide inhibits aerobic metabolism; smoking also dampens inflammatory healing and bacterial killing. These are cited from elsewhere, not tested here.
Inference: the mechanisms would apply to bone stress injury too, but this review does not contain that population, so the base's bone-stress guidance (SYNTH-94, REF-3498) would be extended by inference only.

Practical Application

Does smoking affect fracture healing? Non-union odds were 2.5× higher in current or former smokers (19 studies, mostly surgical patients); delayed union and time to union showed no clear difference. Observational, not athletes.
Before planned surgery: stopping at least 4 weeks before was associated with less wound infection than continuing, in mixed surgery. Not from this paper; the base's own practice: this goes to the surgeon or doctor; no product or dose advice.
Bone stress injury: the review has no stress-fracture data. Not from this paper; the base's own practice: an athlete may be told smoking could matter, labelled as a transfer, with the clinician in charge (SYNTH-94).
Vapes and nicotine replacement: no pooled data for bone or wounds here; do not say they are safe or unsafe for healing on this review.
What it does not show: that smoking makes fractures slower on average (small null analyses cannot show that either way); a cause; anything in athletes.

Limitations

Mostly non-randomised by the nature of the question: RCTs, observational studies and case series were eligible, three case series (89 patients) entered the pooling, and not all studies adjusted for confounders.
Who was measured: trauma and surgical patients; mostly surgical fixation; not athletes; no stress fractures.
Former smokers count as smokers in the exposure definition, and doses are mostly unrecorded.
Small analyses for delayed union and time to union, so null results are weak.
Post-hoc changes to the protocol, stated by the authors, include the non-union window and exposure groups.
Non-pathological fractures not clearly stated in 61% of the included fracture studies by the authors' table; results unchanged on sensitivity analysis by their report.
Appendices (including risk of bias and dose analyses) were not read.
Funding and interests: China Scholarship Council; none declared.
How this entry was checked and changed
Why this level

a systematic review with meta-analysis (as REF-1388, A-, for an observational base), held below A: smoking exposure cannot be randomised (RCTs, observational studies and case series were eligible; the design split is in appendices not read), I² 64% on the main result, post-hoc changes to the protocol stated by the authors, and patients rather than athletes

Full text (checked 2026-10-30)
One hundred and twenty-two studies with 417,767 patients were eligible for the systematic review and 71 of the studies with 39,920 patients were eligible for the meta-analysis (full text)
Nineteen studies including 4726 patients: OR 2·50, 95% CI [1·73–3·61], I2=64% (full text)
Three studies including 117 patients: delayed union OR 1·35, 95% CI [0·51–3·59]; two studies including 48 patients: time to union MD 0·52 weeks, 95% CI [-1·29–2·33] (full text)
Seventeen studies with 14,365 patients: deep surgical site infection OR 2·04, 95% CI [1·68–2·48] (full text)
Six studies with 914 patients: smoking cessation for four weeks or more OR 0·37, 95% CI [0·16–0·89], I2=49%, and six weeks or more OR 0·22, 95% CI [0·06–0·83], I2=47% (full text)
No studies concerning NRT and vaping could be included in the meta-analysis (full text)
Eighty-one (79%) of the 103 studies included fractures that were treated with surgical fixation (full text)
The study also found that the cessation of smoking for four weeks or more had no effect on bone healing (full text)
Three studies including 1607 people: alcohol, nonunion OR 0·97, 95% CI [0·40–2·38], I2=77%; one study including 29 patients: MD 12·85 weeks, 95% CI [1·23–24·47] (full text)
BX was supported by the China Scholarship Council (no. 201809120013) (full text)