Robotic-Assisted Pedicle Screw Placement in Spine Surgery

Accurate screw placement is defined as a pedicle screw that remains entirely within the cortical boundary of the pedicle, or with up to 2 mm breach of the pedicle wall (corresponds to Gertzbein-Robbins Grade A and B), assessed on a computed tomography. To evaluate screw placement accuracy, we will define a binary per-patient outcome: whether at least one screw is not accurately placed. The planned sample size for this substudy is 75 patients in the robotic-assisted surgery group, and 426 patients in the non-robotic-assisted group, for a total of 501 patients. For this outcome, previously treated cases may be included as specified in the approved protocol.

To evaluate the learning curve associated with robotic-assisted surgery, we will model operative time as a function of case number. Operative time is expected to decrease with experience, then plateau. We will use mixed-effects piece-wise linear regression with the surgeon as a random effect to estimate the case number at which operative time levels off. The sample size in this substudy is 100 robotic-assisted spine surgeries in each of the 3 centers currently using robotic-assisted spine surgery (Umeå, Örebro, Uppsala), in total 300 surgeries. For this outcome, previously treated cases may be included as specified in the approved protocol.

The primary outcome is the occurrence of reoperations of any cause within 2 years from the index surgery. The expected reoperation rate in the non-robotic group is likely to be higher than 3-5% over a 2-year period, and may in the robotic-assisted group be 1% or less. A difference of 2 percentage points or greater is considered clinically meaningful. Power analyses (80% power, alpha 0.05) have been made in two scenarios, with reoperation rates of 3% vs 1%, or 5% vs 3%. We estimate a ratio of 1:6 for robotic-assisted surgery vs navigation-assisted or free hand surgeries. Comparing 3% and 1% reoperation rates, 420 robotic-assisted surgery patients and 2,520 non-robotic surgery patients are needed (yielding a total sample size of 2,940 patients). In case reoperation rates are 5% and 3%, 866 robotic-assisted surgery patients and 5,196 non-robotic surgery patients are needed (yielding a total sample size of 6,062 patients).

We will evaluate screw placement precision as the deviation between preoperative planning and actual placement in robotic-assisted surgery cases. For each screw, linear deviation (at the tip and tail) perpendicular from the planned position, to the actual position, and angular deviation between the planned position and the actual position will be measured. This comparison will be made on an overlay of the preoperative planning made on a preoperative computed tomography, and compared with the computed tomography performed intra- or postoperative for assessment of pedicle screw position

Inhospital stay (days) for the index surgery

Time to return to work from the enrolment, which is the time for the index surgery

Health related quality of life measured with EQ-VAS (ranges from 0 (worst) to 100 (best))

Quality of life assessed with EQ-5D, converted into an index, ranging from around 0 (worst) to 1 (best)

PASS EQ-VAS is for elective patients 75 or more, and for non-elective patients 65 or more. EQ-VAS ranges from 0 (worst) to 100 (best)

Number (proportion) of thromboembolic events and cardiovascular complications will be assessed as secondary safety outcomes.

Mortality will be assessed as a secondary safety outcome.