The Robot Costs More. On Thirty-Day Outcomes, the Verdict Splits by Procedure

The robot has been arriving in American operating theatres on an argument that is easy to state and hard to prove: whatever else it costs, it should be at least as safe as laparoscopy. Two national analyses published six weeks apart now test that on thirty-day outcomes, and neither finds the advantage. What one of them finds instead is more useful than a verdict — the direction of the effect depends on which operation the robot is performing.

A million operations and a bill

Writing in Surgery at the end of August, Tran and colleagues at UCLA extracted every elective adult hospitalisation for primary metabolic and bariatric surgery from the National Readmissions Database between 2016 and 2022: an estimated 1,136,027 operations, 82.6% laparoscopic. The laparoscopic cohort was less often female and carried a lower Elixhauser comorbidity index.

After risk adjustment, laparoscopy was associated with lower odds of major adverse events (adjusted odds ratio 0.58, 95% CI 0.46–0.75), infectious complications (0.28, 0.18–0.42) and thirty-day readmission (0.85, 0.81–0.90), and with lower costs (coefficient −4.1, 95% CI −4.6 to −3.7). The authors were unusually direct: robotic surgery “was not associated with a short-term clinical advantage over laparoscopic approaches, despite higher resource utilization.” Administrative data cannot see case difficulty beyond coded comorbidities, which robot was used, or who was operating. Across a million operations, it can see the bill.

Where the register splits in two

Five weeks later, Sebastian and colleagues asked a similar question of a different registry. Using MBSAQIP data from 2020 to 2023, they matched robotic against laparoscopic cases on 22 preoperative variables — 502,645 sleeve gastrectomies and 194,300 gastric bypasses — and reported the result operation by operation.

For sleeve gastrectomy, robotic assistance was associated with higher rates of emergency department visits, blood transfusions, readmissions, venous thromboembolism, surgical site infections and postoperative bleeding. For gastric bypass, four endpoints moved the other way: fewer transfusions, fewer infections, less bleeding, fewer strictures. A pooled estimate across both would have averaged two opposite signals into a number describing nobody.

The same split appeared in adolescents this month: in an MBSAQIP analysis of patients aged 13 to 21, robotic use rose from 6.8% to 34.3% between 2015 and 2024 and every adjusted odds ratio crossed 1, as we reported then.

Why the anastomosis might be the hinge

None of these papers explains its own split, and what follows is a hypothesis. A sleeve is a stapled division with no join to construct; a bypass contains two anastomoses, and those joins are where leaks, strictures and bleeding originate. If a platform’s advantage is precision in suturing — articulated wrists, tremor filtration — it should show up where there is more joining, not in an operation that is mostly a long staple line. The NIDDK’s patient-facing description sets out that structural difference plainly.

Adoption that doesn’t follow the results

If the clinical case is this thin, the spread of robotics needs another explanation, and a Michigan study in Obesity Surgery offers an uncomfortable one. Marwaha and colleagues followed the Michigan Bariatric Surgery Collaborative from 2006 to 2025 — 87,068 procedures, 121 surgeons, 41 hospitals — asking whether the first surgeon to bring a robot to a site predicted whether colleagues followed. Of 106 peer surgeons, 50 (47.2%) eventually adopted, at a median of 1,730 days.

The striking part is the direction: sites whose index adopter was a high-volume surgeon with good outcomes saw lower peer adoption (adjusted OR 0.30, 95% CI 0.09–0.84). The authors call it crowding out, and suggest diffusion may depend less on peer example than on who gets access to the equipment.

The learning curve belongs to the room

Two single-centre series in the Journal of Robotic Surgery suggest where the early costs sit. Huang and colleagues separated surgeon console time from team docking time across their first 100 Hugo-assisted sleeve procedures in Taiwan: medians of 16.0 and 5.0 minutes, a docking breakpoint at case 33.1 (95% CI 26.2–40.0), total operating time falling from 95.0 to 77.0 minutes. Biglari and colleagues analysed 147 of their first 150 robotic bypasses in Belgium: mean console time 104.6 minutes, a breakpoint near case 17, complications in 7.5%, mean stay 2.03 days, no deaths.

Both are retrospective and unblinded, with no laparoscopic comparator — the weakest evidence here, and the only kind that shows implementation from inside a theatre: a team learning a workflow, not a surgeon learning a technique.

What thirty days cannot answer

Every outcome discussed here is a complication code or a bill. None of these datasets carries weight loss, diabetes remission, reflux or revisional surgery at five or ten years, which is where the argument about any bariatric operation is eventually settled. The Journal of the American College of Surgeons ran an invited commentary this month titled “Interpreting Short-Term Safety in Robotic Bariatric Surgery: The Importance of What Drives the Outcome” — a fair summary of where the field’s own discussion has landed.

Settling it would take a prospective multicentre cohort with protocol-driven follow-up rather than claims codes. The LABS study, funded by NIDDK, is the template.