Deep Vein Thrombosis and Pulmonary Embolism in Surgical Practice — Wells-Gated Diagnosis, Caprini-Matched Prophylaxis, DOAC-Era Treatment, and the Filter-Reperfusion Restraint Rules

A 58-year-old woman is day 6 after open colectomy for cancer with a hot swollen left calf; a 66-year-old man is day 3 after hip arthroplasty on aspirin step-down with sudden pleuritic pain and tachycardia; and a 34-year-old anticoagulated woman with a prior unprovoked proximal DVT asks how long she must stay on tablets. One needs Wells-gated duplex, one needs a PE algorithm with a retrieval-dated filter question if anticoagulation is contraindicated, and one needs the Kearon duration arithmetic. The viva rewards the candidate who gates before scanning, stratifies prophylaxis by Caprini, extends cover after cancer surgery, chooses a named DOAC regimen, sets duration by provoked-versus-unprovoked status, and restrains filters and lysis with their trial prices — with every number taken from the papers named beside it.^{[1][7][8][17][27]}

Overview & Definition — provoked surgical disease recurs less, unprovoked disease recurs for years

Venous thromboembolism after surgery is provoked disease with a kinder recurrence curve than unprovoked events: over 8 years after a first symptomatic DVT, recurrence reaches 17.5% at 2 years, 24.6% at 5 years and 30.3% at 8 years — but surgery as the trigger halves recurrence risk with a hazard ratio of 0.36, against cancer at 1.72 and impaired coagulation inhibition at 1.44.^[19] The post-thrombotic syndrome shadows one-third: 22.8% at 2 years, 28.0% at 5 years and 29.1% at 8 years, and ipsilateral recurrent thrombosis multiplies PTS risk 6.4-fold — so every duration decision is also a PTS decision.^[19] PTS is graded by the Villalta scale, developed by Prandoni and introduced as a 1994 abstract as the disease-specific diagnostic and severity questionnaire, correlating 0.609 with the venous clinical severity score, 0.556 with the C of CEAP and 0.499 against the venous filling index benchmark.^[25] The strategic arc fits one sentence: gate every suspicion by probability, image proximally with duplex and centrally with CT, match prophylaxis to Caprini risk, treat acutely with a named DOAC regimen, extend cover for unprovoked disease, and restrain filters and lysis to their trial indications.^{[1][7][13][17][27]}

Classification — proximal versus distal, provoked versus unprovoked, PE risk bands

• Proximal versus distal DVT. Isolated distal (infrapopliteal) thrombosis behaves better: among 33897 registry patients, 17.5% have isolated distal disease with 90-day mortality odds of 0.47 versus proximal, a 1-year deterioration hazard of 0.83, and PTS signs in 47.6 versus 60.5% — a less ominous but not benign entity whose optimal long-term management still awaits randomised trials.^[36]
• Provoked versus unprovoked. Surgery and recent trauma or fracture protect against recurrence (hazard ratios 0.36 and 0.51), while cancer and impaired inhibition predict it — the split that decides duration.^[19] First unprovoked events stopped at 3 months recur at 27.4% per patient-year on placebo against 1.3% on warfarin, the 95% reduction that ended the trial early.^[17]
• PE risk bands. The reperfusion trials enrol normotensive patients with right-ventricular dysfunction plus troponin positivity (intermediate risk) and separately the shocked or arresting massive embolus — bands that decide lysis, catheter and surgical candidacy.^[30][33]
• Effort thrombosis. Paget-Schroetter axillosubclavian thrombosis sits outside the leg trials: 60 reports with 2653 patients rank symptom-free outcomes from 0.54 on anticoagulation alone to 0.96 on lysis plus first-rib resection.^[35]

Each label answers a different viva question — location decides imaging and prognosis, provocation decides duration, PE band decides reperfusion, and effort thrombosis decides the thoracic-outlet operation.^{[36][19][30][35]}

Clinical Presentation — the swollen leg, the breathless chest, the painful arm

The suspected leg presents for Wells scoring with duplex confirmation: in the derivation cohort 16.0% of 593 suspected outpatients have DVT, and probability gating plus proximal ultrasound keeps venography to 5.6% of patients.^[1] The suspected chest presents with dyspnoea or chest pain — the chief complaint in 85% of 8138 PERC-evaluated outpatients — and enters the PE algorithm by gestalt plus rule-out criteria.^[5] Post-thrombotic legs declare through the Ginsberg definition used in the stocking trial — leg pain and swelling of at least 1 month duration — the endpoint routine stockings fail to move.^[24] The high-D-dimer chest deserves respect whatever the score: prevalence quadruples above 4000 ng per mL against 500–1000, and above 2000 with an unlikely score the prevalence is 36% — a likely-group number.^[6]

Differential Diagnosis — location first, provocation second, recurrence versus sequela last

Split proximal from distal disease before choosing intensity: distal carries lower 90-day mortality and less PTS but still deteriorates, so it earns surveillance rather than complacency.^[36] Split provoked surgical thrombosis (duration-limited cover, recurrence halved) from unprovoked first events (extended cover, 27% per patient-year recurrence off treatment) before choosing a stop date.^[19][17] Split fresh ipsilateral recurrence (which multiplies PTS risk 6.4-fold and restarts the clock) from established post-thrombotic syndrome (Villalta-graded chronic sequela) before re-imaging the leg.^[19][25] Split acute PE with right-ventricular dysfunction (reperfusion candidate) from PE without strain (anticoagulation alone) from the shocked massive embolus (surgical or systemic lysis candidate) before choosing the catheter or the knife.^[30][33]

Name the provocation aloud before ordering the scan — the viva rewards the candidate who separates what the trigger decides from what only anatomy and physiology decide.^[19][17]

Clinical & Bedside Assessment — Wells both ways, PERC, Caprini, D-dimer honesty

Score the leg by Wells: low, moderate and high pretest probability carry 3%, 17% and 75% DVT prevalence, and only 0.6% of 501 ruled-out patients suffer events at 3 months — the gate that rations venography and serial scans.^[1] Score the chest by Wells: low, moderate and high carry 1.3%, 16.2% and 37.5% PE prevalence, and low probability with a negative D-dimer reaches 99.5% negative predictive value with 0.1% events when the protocol is followed.^[2] Apply PERC where gestalt is low: one-fifth of outpatients qualify, the false-negative rate is 1.0% with sensitivity 97.4% against specificity 21.9%, and the post-test probability falls below 2%.^[5] Stratify every surgical admission by Caprini: 30-day VTE runs 1.94% in the highest band (over half of all surgical inpatients), 0.97% high, 0.70% moderate and 0% low — with pregnancy or postpartum at odds 8.3, sepsis 4.0, malignancy 2.3, prior VTE 2.1 and central access 1.8.^[7] Read the D-dimer quantitatively, not binarily: overall PE prevalence 21% in the high-risk cohort, quadrupling above 4000, and very high prevalence above 4000 independent of score.^[6]

Investigations — duplex for the leg, CT for the chest, restraint on repeats

Image the leg with duplex: pooled sensitivity 94.2% for proximal disease with specificity 93.8% — duplex technique reaches 96.5% sensitivity with 94.0% specificity, compression alone 93.8% with 97.8% specificity — but distal sensitivity collapses to 63.5% overall and 56.8% with compression alone, so a negative distal scan never carries proximal authority.^[3] Repeat scanning earns restraint: the positive yield of repeat ultrasound is 1.3% with 89% venography-confirmed, on limited data with substantial heterogeneity behind every estimate.^[3] Image the chest by the Christopher sequence: unlikely probability with a normal D-dimer excludes in 32% of 3306 patients with 0.5% 3-month VTE off treatment; CT finds embolism in 20.4% and CT-negative patients suffer 1.3% VTE at 3 months — with a management decision reached in 97.9%.^[4] Remember what CT replaced: ventilation-perfusion lung scanning was the diagnostic standard whose limitations motivated the bedside model-plus-D-dimer strategy.^[2]

Management — Prophylaxis: match modality to Caprini risk, then extend where trials did

Fit stockings where appropriate: 20 trials with 2853 analytic units give DVT 9 versus 21% (odds ratio 0.35, high-quality evidence), proximal DVT 1 versus 5% (odds ratio 0.26), and PE 2 versus 5% (odds ratio 0.38, low-quality) — with the surgical subgroup at odds 0.35.^[10] Add pumps plus drugs in the highest risk: combined IPC with pharmacologic cover cuts DVT from 4.10 to 2.19% (odds ratio 0.52) but adding drugs to IPC raises any bleeding from 0.66 to 4.0% (odds ratio 5.04) with major bleeding 0.1 to 1.5% (odds ratio 6.81) — while adding IPC to drugs cuts symptomatic PE from 2.92 to 1.20% (odds ratio 0.39) with no extra bleeding signal.^[11] Extend cover after cancer surgery: 4 weeks of enoxaparin against 1 week gives venographic VTE 12.0 versus 4.8% persisting 13.8 versus 5.5% at 3 months, with no bleeding or complication difference.^[8] Step down after arthroplasty: 5 days of rivaroxaban then aspirin 81 mg daily matches continued rivaroxaban — VTE 0.64 versus 0.70% with major bleeding 0.47 versus 0.29% — for 9 more days after knees and 30 after hips.^[9]

The resolution is the viva sentence: every surgical admission gets Caprini-stratified mechanical cover, pharmacologic cover where bleeding allows, extended cover after cancer surgery, and an aspirin step-down conversation after arthroplasty.^{[7][10][8][9]}

Management — Perioperative anticoagulation: PAUSE without bridging

Interrupt DOACs by clock, not by test: omit 1 day before low-bleed procedures and 2 days before high-bleed procedures, resuming at 1 day and 2–3 days respectively — a pharmacokinetic strategy built on drug properties, bleed risk and creatinine clearance with no heparin bridging and no coagulation-function testing.^[12] Quote the safety that justifies it: 3007 atrial-fibrillation patients with 30-day major bleeding 1.35% on apixaban, 0.90% on dabigatran and 1.85% on rivaroxaban, against arterial thromboembolism 0.16%, 0.60% and 0.37% — low on both sides without bridging.^[12]

Management — Acute anticoagulation: the four DOAC trials with regimens

Give EINSTEIN-DVT exactly: 3449 acute symptomatic DVT with rivaroxaban 15 mg twice daily for 3 weeks then 20 mg daily against enoxaparin plus vitamin-K antagonist — 2.1 versus 3.0% (hazard ratio 0.68) — with the continued-treatment companion giving 1.3 versus 7.1% (hazard ratio 0.18) at a 0.7% major-bleeding price.^[13] Give EINSTEIN-PE exactly: 4832 acute symptomatic PE on the same single-drug regimen — 2.1 versus 1.8% (hazard ratio 1.12) — with major bleeding halved at 1.1 versus 2.2% (hazard ratio 0.49).^[14] Give AMPLIFY exactly: 5395 acute VTE with apixaban 10 mg twice daily for 7 days then 5 mg twice daily against enoxaparin-warfarin — 2.3 versus 2.7% (relative risk 0.84) — with major bleeding 0.6 versus 1.8% (relative risk 0.31).^[15] Give RE-COVER exactly: parenteral lead-in for a median 9 days then dabigatran 150 mg twice daily against INR 2.0–3.0 warfarin — 2.4 versus 2.1% (hazard ratio 1.10) — with any bleeding 16.1 versus 21.9%.^[16]

The choosing resolution: single-drug rivaroxaban or apixaban suits the uncomplicated leg or embolus without a lead-in, dabigatran follows parenteral cover, and every regimen decision sits beside its bleeding price.^[13][15][16]

Management — Duration: unprovoked extends, intensity stays conventional

Extend cover for first unprovoked events: placebo after 3 months recurs at 27.4% per patient-year against 1.3% on warfarin — a 95% reduction that stopped the 162-patient trial early — priced at 3.8 versus 0% major bleeding per patient-year.^[17] Keep intensity conventional: 738 unprovoked patients on INR 1.5–1.9 versus 2.0–3.0 recur at 1.9 versus 0.7 per 100 patient-years (hazard ratio 2.8) with major bleeding 1.1 versus 0.9 (hazard ratio 1.2, no difference) — low intensity loses efficacy without gaining safety.^[18] State the duration rule: provoked surgical events carry the 0.36 protection factor and stop on schedule, unprovoked first events extend, and cancer or impaired inhibition push toward indefinite cover.^[19][17]

Management — Clot removal for DVT: ATTRACT ties, CaVenT selects, CAVA confirms restraint

Give ATTRACT exactly: 692 acute proximal DVT with pharmacomechanical lysis plus anticoagulation — PTS 47 versus 48% (risk ratio 0.96) — with major bleeding 1.7 versus 0.3% and recurrent VTE 12 versus 8%; moderate-severe PTS falls to 18 versus 24% (risk ratio 0.73) with lower Villalta scores but no quality-of-life gain.^[20] Give CaVenT exactly: 209 first-time iliofemoral DVT with 24-month PTS 41.1 versus 55.6% (absolute reduction 14.4%, NNT 7) and patency 65.9 versus 47.4% — growing at 5 years to 43 versus 71% (absolute reduction 28%, NNT 4) with severe disease 5 versus 1% and still no quality-of-life difference — priced at 20 CDT bleeds with 3 major and 5 clinically relevant.^[21][22] Give CAVA exactly: 184 acute iliofemoral DVT with ultrasound-accelerated urokinase (250000-unit bolus then 100000 units per hour to 96 hours) — 12-month PTS 29 versus 35% (odds ratio 0.75, p equals 0.42) — with major bleeding 5% versus none including one peroneal neuropraxia.^[23]

The choosing resolution: no routine clot removal for proximal DVT — offer catheter-directed lysis only for extensive iliofemoral disease with low bleeding risk in a centre that owns the technique, counsel the bleeding price, and never promise PTS prevention.^[20][21][23]

Management — Post-thrombotic syndrome: stockings fail, Villalta grades, recurrence drives

Do not prescribe stockings to prevent PTS: 806 first proximal DVT with active versus placebo stockings give PTS 14.2 versus 12.7% (hazard ratio 1.13) — no support for routine wearing after DVT.^[24] Grade what remains with Villalta — the 1994 disease-specific scale correlating with severity scores and venous filling — and prevent PTS through recurrence prevention, since ipsilateral recurrence multiplies PTS risk 6.4-fold.^[25][19]

Management — Vena caval filters: PREPIC trades, PREPIC2 refuses, retrievals lapse

Give PREPIC exactly: 400 high-risk proximal DVT with permanent filters — day-12 PE 1.1 versus 4.8% (odds ratio 0.22) — traded for 2-year recurrent DVT 20.8 versus 11.6% (odds ratio 1.87) with no mortality difference.^[26] Give PREPIC2 exactly: 399 severe acute PE with retrievable filters plus anticoagulation — 3-month recurrent PE 3.0 versus 1.5% (relative risk 2.00, p equals 0.50) — with retrieval as planned in 153 of 164 attempts and no support for filters in anticoagulable patients.^[27] Give the abandonment honesty exactly: 220 real-world retrievable-filter patients with retrieval attempted in only 25.3% (92.7% first-attempt success at median 51 days) — over a third placed for perioperative anticoagulation stops — with 17.0% suffering a venous event.^[28]

The filter rule the examiner wants: filter only for contraindication to anticoagulation or mandated perioperative interruption, set the retrieval date at insertion, and never present a filter as treatment.^[26][27][28]

Management — PE reperfusion: decompensation prevented, bleeding priced, surgery for the blocked

Give MAPPET-3 exactly: 256 stable submassive PE with 100 mg alteplase over 2 hours plus heparin — escalation 24.6 versus 10.2% with mortality 3.4 versus 2.2% (no difference) and no fatal or cerebral bleeding.^[29] Give PEITHO exactly: 1006 intermediate-risk PE with tenecteplase — death or decompensation 2.6 versus 5.6% (odds ratio 0.44) — with day-30 deaths tied at 2.4 versus 3.2%, extracranial bleeding 6.3 versus 1.2%, and stroke 2.4 versus 0.2% with ten haemorrhagic.^[30] Give ULTIMA exactly: 59 intermediate-risk PE with 10–20 mg catheter alteplase over 15 hours — RV-to-LV fall 0.30 versus 0.03 at 24 hours — with 1 heparin-arm death, no major bleeding and no recurrent VTE at 90 days.^[31] Give CANARY exactly: 94 intermediate-high PE stopped early for COVID — primary 4.3 versus 12.8% (odds ratio 0.31, non-significant) — with 72-hour RV-to-LV above 0.9 in 27.0 versus 52.1% and one gastrointestinal major bleed.^[32] Give Leacche exactly: 47 emergency surgical embolectomies — 45% lysis-contraindicated, 10% failed medical therapy, 32% RV dysfunction, with 26% in shock and 11% in arrest — 6% operative deaths with 86% and 83% survival at 1 and 3 years.^[33]

The choosing resolution: anticoagulate the stable embolus, weigh systemic lysis for deteriorating intermediate-risk disease against stroke and bleeding, consider catheter-directed lysis where the team owns it, and operate the massive central embolus with contraindicated or failed lysis — then counsel the long-term honesty that lysis changes no late outcome: at median 37.8 months mortality ties at 20.3 versus 18.0%, dyspnoea persists in one-third either way, and chronic thromboembolic pulmonary hypertension is confirmed in only 2.1 versus 3.2%.^{[29][30][31][33][34]}

Complications & Pitfalls — the six traps

The ungated-scan trap — imaging every swollen leg and breathless chest when Wells-plus-D-dimer excludes at 99.5% negative predictive value and PERC-negatives fall below 2% post-test probability.^[2][5] The distal-complacency trap — dismissing infrapopliteal thrombus when registry deterioration still occurs, or over-treating it as proximal disease when mortality and PTS both run lower.^[36] The routine-lysis trap — offering catheter-directed thrombolysis for every proximal DVT when ATTRACT ties on PTS at 47 versus 48% with more major bleeding.^[20] The abandoned-filter trap — placing a retrievable filter without a retrieval date when only one-quarter ever come out and PREPIC2 shows no recurrent-PE reduction in anticoagulated patients.^[27][28] The stocking-prescription trap — promising PTS prevention with stockings when SOX ties at 14.2 versus 12.7% against placebo.^[24] The lysis-survival trap — quoting decompensation prevention as survival gain when PEITHO day-30 deaths tie and long-term mortality, dyspnoea and pulmonary hypertension never separate.^[30][34]

Prognosis & Disposition — the numbers that set expectations

Special Populations — contexts that change the emphasis

The cancer-surgery patient runs on ENOXACAN II: 4 weeks of enoxaparin after open abdominal or pelvic cancer surgery halves venographic thrombosis against 1 week with no bleeding excess — extended cover is the standard, not the exception.^[8] The arthroplasty patient runs on EPCAT II: 5 days of rivaroxaban then aspirin 81 mg daily matches continued rivaroxaban for symptomatic VTE with indistinguishable bleeding.^[9] The anticoagulated patient facing elective surgery runs on PAUSE: omit by bleed-risk clock, resume by bleed-risk clock, no bridging, no testing, with major bleeding under 2% and arterial events under 1% in every DOAC cohort.^[12] The young patient with effort thrombosis runs on the Paget-Schroetter ranking: anticoagulation alone leaves 46% symptomatic while lysis plus first-rib resection leaves 96% symptom-free — refer for decompression, not reassurance.^[35] The distal-DVT leg runs on RIETE restraint: lower mortality and less PTS than proximal disease, with randomised management trials still awaited — survey, do not ignore.^[36] The unprovoked first-event patient runs on Kearon duration: extend beyond 3 months at conventional intensity, since placebo recurs at 27% per patient-year and low intensity fails without bleeding benefit.^[17][18]

Evidence, Guidelines & Regional Differences — the four stories and who led them

The diagnosis story is Wells-to-CT: the 1997 DVT model rations venography and serial scans, the 2001 PE model plus D-dimer reaches 99.5% negative predictive value, duplex proves near-definitive proximally with distal weakness, and Christopher validates the CT algorithm at 97.9% decisiveness — built when ventilation-perfusion scanning was the limited standard.^[1][2][3][4] The prophylaxis story is Caprini-to-extension: Bahl validates the score in 8216 surgical patients, Cochrane quantifies stockings and pumps, ENOXACAN II extends cancer-surgery cover to 4 weeks, EPCAT II steps arthroplasty down to aspirin, and PAUSE standardises DOAC interruption without bridging.^{[7][10][8][9][12]} The treatment story is heparin-to-DOAC-to-duration: EINSTEIN, AMPLIFY and RE-COVER fix single-drug or short-lead-in regimens with halved bleeding, while Kearon and ELATE fix extension for unprovoked disease at conventional intensity.^{[13][15][17][18]} The intervention story is enthusiasm-to-restraint: CaVenT rewards selected iliofemoral lysis while ATTRACT and CAVA tie, SOX retires PTS stockings, PREPIC and PREPIC2 retire routine filters, and PEITHO with ULTIMA, CANARY and Leacche prices every reperfusion choice.^{[21][20][24][26][30]} The set's stated honesty: the PEP aspirin trial carries no quotable abstract here so arthroplasty-aspirin numbers are EPCAT-owned; no guideline PDF bytes were fetched this session, so every threshold above is trial-owned, never guideline-numbered.^[9]

Exam Pearls — the one-liners that score

• Surgery as trigger halves 8-year recurrence (HR 0.36); unprovoked first events recur at 27% per patient-year off cover.^[19][17]
• Wells DVT 3/17/75% with 0.6% ruled-out events; Wells PE 1.3/16.2/37.5% with 99.5% NPV for low-plus-negative-D-dimer.^[1][2]
• Duplex: proximal 94.2% sensitive, distal 63.5%; compression alone distal 56.8%; repeat yield 1.3%.^[3]
• Christopher: unlikely-plus-normal-D-dimer 32% excluded at 0.5%; CT-negative 1.3% at 3 months.^[4]
• PERC: eight criteria, 20% qualify, 1.0% false-negative, sensitivity 97.4%.^[5]
• D-dimer above 4000 is very-high prevalence whatever the score; above 2000 with unlikely score gives 36%.^[6]
• Caprini highest band 1.94% of 8216 surgical patients; pregnancy 8.3, sepsis 4.0, cancer 2.3, prior VTE 2.1.^[7]
• Cancer surgery: 4 weeks enoxaparin 12.0 versus 4.8%; arthroplasty: aspirin step-down 0.64 versus 0.70%.^[8][9]
• PAUSE: omit 1/2 days, resume 1/2–3 days, no bridging — bleeding under 2%, arterial under 1%.^[12]
• DOAC regimens: rivaroxaban 15 mg twice daily 3 weeks then 20 mg daily; apixaban 10 mg twice daily 7 days then 5 mg twice daily; dabigatran 150 mg twice daily after 9-day lead-in.^[13][15][16]
• ATTRACT ties 47 versus 48%; CaVenT NNT 7 at 2 years, 4 at 5 years; CAVA ties 29 versus 35%.^[20][21][23]
• SOX: stockings 14.2 versus 12.7% — no PTS prevention; Villalta grades instead.^[24][25]
• PREPIC trades day-12 PE for 2-year DVT; PREPIC2 negative; filters retrieved in one-quarter.^[26][27][28]
• PEITHO: decompensation 2.6 versus 5.6% at stroke 2.4% with no survival gain; Leacche operates massive emboli at 6% mortality.^[30][33]

Revision summary

Postoperative VTE is provoked disease with halved recurrence, diagnosed by Wells-gated duplex and CT algorithms with PERC and quantitative D-dimer honesty. Caprini stratifies every admission for mechanical cover, cancer surgery earns 4-week enoxaparin, arthroplasty earns aspirin step-down, and PAUSE sets the DOAC clock without bridging. Acute disease takes a named DOAC regimen, unprovoked first events extend at conventional intensity, clot removal stays selective under ATTRACT-CaVenT-CAVA equipoise, stockings never prevent PTS, filters need a retrieval date, and intermediate-risk PE lysis prevents decompensation without survival gain while surgery rescues the massive embolus.^{[19][2][7][13][17][20][27][30]}

8216 NSQIP surgical inpatients with 30-day VTE 1.4%; highest 1.94%, high 0.97%, moderate 0.70%, low 0%; over half the population scores highest (PMID 19779324).^[2] 20 trials with 2853 units; DVT 9 vs 21% (OR 0.35 high-quality); proximal 1 vs 5% (OR 0.26); PE 2 vs 5% (OR 0.38 low-quality); surgical subgroup OR 0.35 (PMID 30390397).^[10] 220-patient ALN cohort with retrieval attempted in only 25.3% (92.7% first-attempt success, median 51 days); perioperative-stop indication in 37.7% with 17.0% suffering a VTE event (PMID 17218580).^[28] median 37.8 months with mortality 20.3 vs 18.0% (ns); persistent dyspnoea 36.0 vs 30.1% (ns); CTEPH 2.1 vs 3.2%; lysis changes no long-term outcome (PMID 28335835).^[34]