Factors Associated With Reoperation After Minimally Invasive Tubular Transforaminal Lumbar Interbody Fusion: A Cohort Study of 756 Patients

Abstract

Study Design

Retrospective Cohort Study using Nested Matched Case-Control Analysis.

Objectives

High-volume data to describe the complication profile and factors associated with reoperation after minimally invasive tubular transforaminal lumbar interbody fusions (TLIFs) in the short-term (<30 days) and long-term (>30 days).

Methods

All tubular TLIFs (2011-2024) were retrospectively reviewed, performed by eight neurosurgeons at our single centre. Two matched controls per reoperation case were manually selected. Each control had undergone the same primary procedure within six months of the index case, at the same levels and, if possible, by the same surgeon. Variables included demographics, comorbidities, frailty indices, private/public status, workplace injury, preoperative antithrombotics, preoperative laboratory results, the symptomatic indications, surgical factors, and clinical outcomes. Univariate logistic regression identified candidate variables for multivariate analysis.

Results

From 756 patients, there was a 6.6% reoperation rate (n=50), with mean follow-up of 1.5 (±1.7) years. Short-term reoperation was predominantly for cage migration and multivariate analysis revealed no independent predictors. Long-term reoperation was mainly for cage migration and was associated with private-funding, WorkCover status, osteoporosis/osteopenia, and previous lumbar surgery. Combined timeframe reoperations were associated with WorkCover status, osteoporosis/osteopenia, diabetes, and previous lumbar surgery.

Patients who had reoperations fared worse than their matched controls. Additionally, a worse patient outcome was associated with diabetes, a higher Modified Charlson Comorbidity Index, and previous lumbar surgery.

Conclusion

This retrospective case-control series reports our complication profile and factors associated with reoperation and poor patient outcomes after tubular TLIFs. Reoperation was associated with osteoporosis/osteopenia, diabetes, private-funding, WorkCover status, and previous lumbar surgery at another spinal level.

Introduction

With increasing lumbar fusion rates, surgical techniques have evolved accordingly to optimise patient outcomes, namely minimally invasive spinal (MIS) approaches.^1-3 Tubular retractor systems offer smaller incisions with reduced iatrogenic tissue damage, decreased estimated blood loss, shorter lengths of stay, less postoperative pain, and faster recovery times. ⁴ One of the most popular lumbar fusion techniques is the transforaminal lumbar interbody fusion (TLIF), which allows for a posterior and safe MIS approach with good long-term outcomes.^3,5

Nevertheless, reoperation remains an issue, with 2.7-14.5% of all tubular TLIF patients requiring a return to theatre.^6-8 Such reoperations have been associated with higher complication rates, lower patient satisfaction, worse long-term outcomes, and increased cost burden.^9,10 This reflects an important part of the surgical complication profile, including nerve root injury or continued compression, infection, haematoma, and implant failure.^8,11

Previous studies have examined MIS TLIF reoperation rates, however, these often lack breadth in potential risk factors, have small sample sizes, and limited timeframes.^7-9,11 Population studies have attempted to rectify this, however, remain nonspecific for translation into clinical practice with inclusion of multiple fusion or approach techniques, and assess a small amount of clinically-relevant risk factors.^6,12-14 The present study aims to fill this gap by considering a large variety of factors and their implication on reoperation status in both the short and long-term after tubular MIS TLIFs.

The primary research question was to determine the incidence and indication profile of reoperation after elective tubular MIS TLIF, and which factors were associated with reoperation. We hypothesised that reoperation may be associated with increased age, metabolic comorbidity, and prior lumbar surgery. Secondary aims were to compare short- (<30 days) and long-term (>30 days) reoperation profiles, as well as to assess factors associated with poor clinical outcomes.

Methods

Study Design and Patient Cohort

A retrospective single-cohort study with nested matched case-control analysis was conducted for all tubular MIS TLIF procedures performed by eight neurosurgeons in a single institution from 2011 to February 2024. TLIFs were indicated for elective degenerative conditions per clinician expertise, including intractable lower back pain and radiculopathy with neural compression or spondylolisthesis after failed conservative management. All emergent, tumour, and trauma cases were excluded.

A case of reoperation was defined as any patient requiring any subsequent procedure at the same spinal level. The patient’s first TLIF at our centre was used as the index case. For each case, two controls were manually selected from the full tubular MIS TLIF cohort. Eligible controls had undergone the same primary procedure within six months of the index case, at the same level/s and, if possible, by the same surgeon. When two or more eligible controls were available, selection was prioritised in the order of closest date of surgery. This matching was blinded to outcomes, based purely on the criteria, and conducted before data-extraction commenced. Short-term reoperation was defined as within 30 days and long-term thereafter. Ethics approval was formally obtained from the Mater Ethics Committee and Mater Research Governance Committee (MSSA/MRGO/104172). Patients consented to the procedure, and data were collected retrospectively. This study was reported in accordance with the STROBE guidelines.

Patient records and operation notes were assessed until April 2024. Baseline parameters, including demographics and comorbidities were recorded on admission. Osteoporosis was defined per WHO criteria with a bone mineral density T-score ≤-2.5 SD below the young adult mean, with a T-score ≤-1 defining osteopenia. ¹⁵ Two validated frailty indices were included: the 5-factor modified Frailty Index (mFI-5) and the modified Charlson Comorbidity Index (mCCI).^16,17 The mCCI was modified to exclude age, as this was assessed independently, and has been validated as a survival prediction tool. ¹⁸ The healthcare sectors of private/public-funding were assessed, alongside WorkCover status (whereby patients receive financial compensation for work-related injuries). Preoperative sensory changes included any numbness or paraesthesia, and motor changes included weakness consistent with the spinal level of the pathology. Operative duration was defined as from the start of general anaesthesia until the patient left theatre. Clinical outcomes were assessed per the MacNab criteria, which is well-validated for use in spinal surgery follow-up, and was utilised in this study as it was consistently available across the study period. ¹⁹ This classifies outcomes as excellent, good, fair, or poor based on residual pain, functional limitation, persistent radicular symptoms, and classically by the need for further operative intervention. We adapted this last point of the criteria (Table 1) so that the index reoperation for cases did not automatically class them as ‘poor’, which would have introduced circularity as reoperation itself was the primary study endpoint. However, any subsequent reoperations were counted, thus allowing the reoperations themselves and their subsequent effects to be appropriately examined.

Table 1.

Revised Modified MacNab Criteria for Assessing Clinical Outcome ¹⁹

MacNab criteria score	Criterion
Excellent	Nil pain or mobility restrictions.
	Able to return to normal work and activities.
Good	Occasional non-radicular pain, relief of presenting symptoms.
	Able to return to modified work.
Fair	Some improved functional capacity.
	Still handicapped or unemployed.
Poor	Continued objective symptoms of root involvement or additional operative intervention required at the index level.
	Irrespective of length of postoperative follow-up.
	Reoperation required*

*for reoperation cases, the index reoperation was not automatically classified as a poor outcome, since reoperation itself was the primary endpoint of this study. Final MacNab category was instead assigned using symptoms and function after the index reoperation. Any subsequent reoperations were classified as poor.

MIS TLIF Surgical Technique

This current series only includes patients undergoing MIS TLIF before our unit adopted robot-guided pedicle screws. We have reported our unit’s standardised surgical technique over 11 years ago. ²⁰ The operation is performed under general anaesthesia, with electrodes in the lower limbs for continuous electromyography, and prone positioning on a Jackson table. A reference pin array is inserted into the posterior iliac crest contralateral to the side of the facetectomy, and 3D images acquired with the O-arm intraoperative CT scanner (Medtronic, Minnesota).

Short bilateral Wiltze incisions allow navigated manual insertion of pedicle screws, with entry at the junction between the transverse process and lateral facet joint. A navigated Jamshidi needle connected to continuous stimulation traverses the pedicle. The length and width of the screw is then calculated on the navigation software, and then the Jamshidi needle is replaced with a K-wire.

METRx tubular dilators (Medtronic) are placed using a MAST dilator directly over the facet joint, with a trajectory through the foramen and along the angle of the disc space. Monopolar cautery exposes the facet joint, and a high-speed drill is used to expose the exiting nerve root. A bone dust collector connected to the suction tube collects bone for grafting. The nerve root is directly decompressed removing soft ligamentous tissue with a Kerrison rongeur from its junction with the thecal sac, to out past the foramen laterally.

The disc space is entered via an annulotomy, disc material is removed unilaterally and end plates prepared. Trial cages are placed to ascertain optimal cage height, whilst length is based on navigation. Bone graft (autologous bone dust combined with synthetic ceramic) is packed into the disc space (typically 5mL or more), and the cage filled with graft is then inserted under direct vision with x-ray confirmation.

Expandable cages were expanded until the torque mechanism was completed. Most patients underwent unilateral TLIF, but if bilateral cages were inserted, this process was repeated on the contralateral side.

Finally, navigated pedicle screws with percutaneous towers are then inserted over the K-wires. Connecting rods and set screws are inserted and torqued off. An O-arm spin is performed to ensure good hardware placement. The percutaneous towers are disconnected and the incisions are closed with vicryl and monocryl.

Preoperative and Postoperative Care

All patients had preoperative magnetic resonance imaging. Within a week of the procedure, antithrombotics were ceased and routine blood tests taken. All patients were reviewed postoperatively, with consults at the 6-week mark and typically at 12 months. Upon symptom resolution, patients were discharged from further follow-up with instruction to re-present if symptoms reoccurred.

Data Review

Factors were assessed between cases and controls and across dichotomised MacNab outcomes (classed as excellent/good and fair/poor) with IBM SPSS Statistics version 29 (IBM Corp., Armonk, NY, USA). ²¹ Missing data were not imputed, with the extent of missing variables reported in Supplementary Table 1. A series of univariate binary logistic regressions were completed, with thresholds of 50% and P<0.2 set as the cut-off, as per standard convention for risk-factor candidate selection. These candidates were then assessed with a multivariate backward conditional logistic regression with threshold for significance set at P<0.05. Given the exploratory nature of this retrospective analysis and the correlation between multiple variables, formal correction for multiple comparisons was not applied. Accordingly, statistically-significant results should be interpreted as associations and require external validation. The cumulative incidence of reoperation was derived as 1 – Kaplan–Meier survival estimate, with censoring at last follow-up. As the matched analysis was case-control in design, logistic regression was used for associative testing, however, any time-to-event inferences were limited by variable follow-up and event number. Due to the reduced sample size in the short-term reoperative cohort, some changes to the multivariate models were required. During multivariate modelling for short-term reoperations, MFI-5 was found to be collinear with diabetes and hypertension, as it includes these two factors in its calculation. To reduce redundancy and stabilise the model, we retained MFI-5 (as the broader comorbidity index) and excluded diabetes and hypertension per statistical advice. No such issue was observed modelling the combined timeframe cohort. For MacNab outcomes in the short-term, there was quasi-separation with sensory changes and previous surgery at the same spinal level, both were required to be excluded to produce a satisfactory model.

Results

Baseline Characteristics

A total of 756 patients underwent tubular MIS TLIF, of which 50 patients (6.6%) required reoperation, and 10 of whom were within the first 30 days. Figure 1 shows cumulative reoperation, half within 189 days. Each case was matched with two samples, giving a total study size of 150 patients. The mean age was 60.1 years, with 81 females (54%) and 86 private patients (57.3%). The population on average was obese, with a mean BMI of 30.1kg/m² (SD 5.6kg/m²), and the most common level operated being L4/5.

Figure 1.

Cumulative Rates of Reoperation after TLIF

Rates and Reason for Reoperation

The most common reason for reoperation was revision of the instrumentation (62% of cases), as per Table 2. Within that category, cage migration (14 patients) was the most common indication, with eleven in the posterior direction, two lateral, and one anterior. Five out of these 14 migrated cages occurred within 30 days. No association was noted with cage type (expandable versus static). Dislodgement of a locking cap or spacer (14%) and screw loosening/misplacement (16%) was seen more after 30 days. One case required an extension of the hardware due to incorrect fusion of the wrong level. The decompression was augmented in 22% of reoperations due to residual neural compromise via canal/recess stenosis. This was followed by washouts (8% for haematoma or infection) and, in 6% of cases, exchange to anterior lumbar interbody fusion (ALIF). One of the three ALIFs was performed within 30 days as the TLIF cage had extruded into the abdomen and required retrieval and anterior fusion. Of those requiring reoperation, 18% required a second reoperation at that level, automatically classing them as MacNab poor.

Table 2.

Rates and Indications for Reoperation

​	Short-term reoperations (<30d)	Long-term reoperations (>30d)	Combined timeframes
Required reoperation, n (%)	10 (1.3%)	40 (5.3%)	50 (6.6%)
Mean time to reoperation, days (SD)	8.4 (±5.9)	450.9 (±502.5)	362.4 (±482.6)
Reoperation type:
Revision of instrumentation, n (%)	6 (60%)	25 (62.5%)	31 (62%)
Decompression, n (%)	1 (10%)	10 (25%)	11 (22%)
Washout (for haematoma/infections), n (%)	1 (10%)	3 (7.5%)	4 (8%)
ALIF, n (%)	1 (10%)	2 (5%)	3 (6%)
Discectomy, n (%)	1 (10%)	0 (0%)	1 (2%)
Reason for Reoperation:
Cage migration, n (%)	5 (50%)	9 (22.5%)	14 (28%)
Residual neural compromise, n (%)	1 (10%)	10 (25%)	11 (22%)
Screw loosening/misplacement, n (%)	0 (0%)	8 (20%)	8 (16%)
Dislodged locking cap/spacer, n (%)	2 (20%)	5 (12.5%)	7 (14%)
Cage subsidence, n (%)	0 (0%)	3 (7.5%)	3 (6%)
Haematoma, n (%)	1 (10%)	2 (5%)	3 (6%)
New anterolisthesis, n (%)	1 (10%)	1 (2.5%)	2 (4%)
Incorrect level of fusion, n (%)	0 (0%)	1 (2.5%)	1 (2%)
Infection, n (%)	0 (0%)	1 (2.5%)	1 (2%)
2nd reoperation required, n (%)	3 (30%)	6 (15%)	9 (18%)

Factors Associated With Reoperation

Table 3 summarises the variables which acted as candidates for multivariate analyses in Table 4. Given the small number of short-term reoperations (n=10), this analysis was underpowered and should be considered only as hypothesis-generating. Long-term reoperations were significantly associated with being private (OR 2.44, 95% CI [1.03-5.78], P=0.04), WorkCover (OR 15.28, 95% CI [1.58-147.4], P=0.018), osteoporosis (OR 3.42, 95% CI [1.05-11.16], P=0.04), and previous surgery at a different level (OR 3.6, 95% CI [1.15-11.35], P=0.03). Across all timeframes, factors associated with reoperation were WorkCover (OR 7.52, 95% CI [1.59-35.79], P=0.01), diabetes (OR 3.49, 95% CI [1.25-9.76], P=0.02), osteoporosis (OR 3.2, 95% CI [1.2-8.8], P=0.02), and previous surgery at a different level (OR 3.94, 95% CI [1.44-10.77], P=0.008).

Table 3.

Baseline Demographics and Factors for Reoperation

​		Short-term reoperations (<30d)			Long-term reoperations (>30d)			Combined timeframes
		Cases	Controls	P-value	Cases	Controls	P-value	Cases	Controls	P-value
Baseline Character-istics	Females, n (%)	6 (60%)	9 (45%)	0.441	20 (50%)	46 (57.5%)	0.437	26 (52%)	55 (55%)	0.728
	Mean Age, years (SD)	66.1 (±6.4)	59 (±14.2)	0.153	61.9 (±12.5)	58.8 (±13.8)	0.228	62.7 (±11.6)	58.8 (±13.8)	0.088
	BMI, kg/m2 (SD)	33.1 (±6.0)	31.5 (±6.0)	0.428	30 (±4.3)	29.7 (±6.4)	0.788	30.6 (±4.8)	29.8 (±6.0)	0.404
	Levels:	​		1	​		1	​		1
	L3/4, n (%)	0 (0%)	0 (0%)		1 (2.5%)	2 (2.5%)		1 (2%)	2 (2%)
	L4/5, n (%)	4 (40%)	8 (40%)		20 (50%)	40 (50%)		24 (48%)	48 (48%)
	L5/S1, n (%)	5 (50%)	10 (50%)		18 (45%)	36 (45%)		23 (46%)	46 (46%)
	Multi: L2-4, n (%)	1 (10%)	2 (10%)		0 (0%)	0 (0%)		1 (2%)	2 (2%)
	Multi: L4-S1, n (%)	0 (0%)	0 (0%)		1 (2.5%)	2 (2.5%)		1 (2%)	2 (2%)
	Privately-insured, n (%)	7 (70%)	12 (60%)	0.593	28 (70%)	39 (48.8%)	0.029	35 (70%)	51 (51%)	0.028
	WorkCover Status, n (%)	1 (10%)	2 (10%)	1	5 (12.5%)	1 (1.3%)	0.03	6 (12%)	3 (3%)	0.042
Comorb-idities	Diabetes, n (%)	6 (60%)	1 (5%)	0.006	7 (17.5%)	10 (12.5%)	0.461	13 (26%)	11 (11%)	0.021
	Hypertension, n (%)	6 (60%)	6 (30%)	0.122	19 (47.5%)	30 (37.5%)	0.347	25 (50%)	36 (36%)	0.162
	Osteoporosis/osteopenia*, n (%)	3 (30%)	4 (20%)	0.544	9 (22.5%)	6 (7.5%)	0.028	12 (24%)	10 (10%)	0.026
	Depression/anxiety, n (%)	3 (30%)	8 (40%)	0.593	13 (32.5%)	21 (26.3%)	0.527	16 (32%)	29 (29%)	0.706
	Smoking status, n (%)	3 (30%)	9 (45%)	0.432	16 (40%)	27 (33.8%)	0.565	19 (38%)	36 (36%)	0.811
	Alcohol consumption status, n (%)	5 (50%)	12 (60%)	0.603	29 (72.5%)	45 (56.3%)	0.118	34 (68%)	57 (57%)	0.195
Frailty Indices	Modified Charlson Comorbidity Index, score (SD)	2 (±1.7)	0.90 (±1.3)	0.079	0.78 (±1.1)	0.64 (±0.99)	0.825	1.02 (±1.33)	0.67 (±1.06)	0.093
	MFI-5, score (SD)	1.3 (±0.82)	0.40 (±0.75)	0.024	0.55 (±0.64)	0.47 (±0.75)	0.664	0.7 (±0.74)	0.42 (±0.7)	0.028
Spinal factors	Preoperative Antithrombotics, n (%)	1 (10%)	4 (20%)	0.497	9 (18.8%)	18 (22.5%)	1	10 (20%)	22 (22%)	0.778
	Presence preoperative spondylolisthesis, n (%)	6 (60%)	10 (50%)	0.606	23 (57.5%)	49 (61.3%)	0.693	29 (58%)	59 (59%)	0.639
	Grade preoperative spondylolisthesis	​		0.32	​		0.48	​		0.162
	Nil/Grade 0, n (%)	4 (40%)	10 (50%)		17 (42.5%)	31 (38.8%)		21 (42%)	41 (41%)
	Grade 1, n (%)	2 (20%)	6 (30%)		17 (42.5%)	42 (52.5%)		19 (38%)	48 (48%)
	Grade 2, n (%)	4 (40%)	4 (20%)		6 (15%)	7 (8.8%)		10 (20%)	11 (11%)
	Previous spinal surgery at that level, n (%)	4 (40%)	6 (30%)	0.585	7 (17.5%)	5 (6.25%)	0.063	11 (22%)	11 (11%)	0.078
	Previous lumbar spinal surgery at a different level, n (%)	5 (50%)	3 (15%)	0.051	10 (25%)	6 (7.5%)	0.035	15 (30%)	9 (9%)	0.005
Blood results	Platelet count, 109/L (SD)	275 (±86.8)	259 (±57.4)	0.536	262.3 (±64.3)	264.2 (±71.6)	0.884	260.7 (±62.8)	263.2 (±68.8)	0.834
	Haemoglobin levels, g/L (SD)	135.6 (±9.0)	143.9 (±16.5)	0.155	141.7 (±15.5)	138.1 (±14.9)	0.217	139.6 (±13.3)	139.2 (±15.4)	0.887
	Cholesterol levels, mmol/L (SD)	5.2 (±1.4)	5.3 (±1.3)	0.595	6.5 (±4.2)	7.1 (±4.7)	0.465	5.41 (±2.2)	5.3 (±1.2)	0.634
	Random glucose levels, mmol/L (SD)	8.5 (±4.4)	5.4 (±0.8)	0.046	6.2 (±2.7)	6.4 (±3.0)	0.687	6.4 (±2.7)	6.0 (±2.1)	0.288
Preop symptoms	Bilateral radiculopathy, n (%)	3 (30%)	4 (20%)	0.544	12 (30%)	24 (30%)	1	15 (30%)	28 (28%)	0.798
	Motor symptoms, n (%)	3 (30%)	0 (0%)	0.998	7 (17.5%)	8 (10%)	0.724	10 (20%)	8 (8%)	0.196
	Sensory changes, n (%)	2 (20%)	2 (10%)	0.455	16 (40%)	24 (30%)	0.592	18 (36%)	26 (26%)	0.206
Intra-operative consider-ations	Side of approach	​		0.793	​		0.681	​		0.627
	Right, n (%)	4 (40%)	10 (50%)		17 (42.5%)	38 (47.5%)		21 (42%)	48 (48%)
	Left, n (%)	4 (40%)	8 (40%)		17 (42.5%)	32 (40%)		21 (42%)	40 (40%)
	Bilateral, n (%)	2 (20%)	2 (10%)		6 (15%)	10 (12.5%)		8 (16%)	12 (12%)
	Expandable cage, n (%)	3 (30%)	9 (45%)	0.432	14 (35%)	36 (45%)	0.296	17 (34%)	45 (45%)	0.199
	Operative time, minutes (SD)	232.8 (±59.5)	229.1 (±56.5)	0.864	238 (±65.6)	225.4 (±58.0)	0.286	245.6 (±69.3)	232.7 (±60.7)	0.246
	Dural Tears, n (%)	0 (0%)	0 (0%)	-	0 (0%)	3 (3.8%)	0.998	0 (0%)	3 (3%)	0.998
Post-operative consider-ations	Use of postoperative antibiotics, n (%)	9 (90%)	20 (100%)	0.998	36 (90%)	67 (83.8%)	0.359	45 (90%)	87 (87%)	0.595

Values in bold depict statistically-significant findings.

*There were 7 cases of osteopenia in the >30d cohort, 3 cases and 4 controls.

Table 4.

Final Model of Multivariate Logistic Regressions for Preoperative Factors for Reoperation

​	Short-term reoperations (<30d)			Long-term reoperations (>30d)			Combined timeframes
	OR	95% CI	P-value	OR	95% CI	P-value	OR	95% CI	P-value
Private	-	-	-	2.441	1.031 – 5.781	0.042	-	-	-
WorkCover status	-	-	-	15.28	1.584 – 147.395	0.018	7.522	1.587 – 35.786	0.011
Alcohol consumption status	-	-	-	-	-	-	2.067	0.894 – 4.781	0.090
Diabetic	-	-	-	-	-	-	3.489	1.247 – 9.763	0.017
Osteoporosis/osteopenia	-	-	-	3.417	1.047 – 11.155	0.042	3.212	1.167 – 8.824	0.024
MFI-5 score	2.957	0.967 – 9.050	0.057	-	-	-	-	-	-
Previous lumbar spinal surgery at different level	-	-	-	3.603	1.145 – 11.345	0.028	3.941	1.442 – 10.772	0.008
Motor Symptoms	-	-	-	-	-	-	2.853	0.926 – 8.785	0.068
Random glucose levels, mmol/L	1.660	0.924 – 2.985	0.090	-	-	-	-	-	-
Model significance	χ 2 (2) = 13.321, P = 0.001			χ 2 (4) = 22.408, P < 0.001			χ 2 (6) = 31.102, P < 0.001
Fit (Hosmer and Lemeshow Test)	P = 0.883			P = 0.470			P = 0.245
Sensitivity of Model	60%			43.6%			46.9%
Specificity of Model	90%			92.2%			91.5%

Values in bold depict statistically-significant findings.

Overall, the multivariate models demonstrated acceptable discrimination with high specificity (90-92.2%) but modest sensitivity (43.6-60%) which limits use for individual-level prediction. These findings should therefore be interpreted as exploratory associations rather than a clinical risk-prediction tool without further validation.

Follow-Up Consequences

As per Supplementary Table 2, most factors were associated with the requirement of reoperation. Being postoperative factors, i.e. not predictive risk factors of reoperation, they were excluded from multivariate analysis. Patients were censored at last documented follow-up. As follow-up was variable and relatively short for some patients, late reoperation risk may be underestimated.

Patient Outcomes

The revised, modified MacNab scores at follow-up in the combined timeframe for patients who had a reoperation were 10% excellent, 28% good, 22% fair, 40% poor and for controls was 47% excellent, 33% good, 20% fair, 0% poor.

Factors Associated With Poor Patient Outcomes

Table 5 summarises the series of univariate logistic regressions from which candidates were selected for the multivariate regression. Table 6 revealed that a fair/poor MacNab outcome in the short-term reoperative cohort was statistically associated with diabetes (OR 14.14, 95% CI [1.04-192.57], P=0.047), whilst no statistically significant association was found for the long-term cohort, albeit the long-term model did not reach statistical significance (P=0.172) so it cannot be interpreted further. The combined timeframes noted worse MacNab outcomes for mCCI (OR 1.34, 95% CI [1.00-1.81], P=0.05) and previous surgery at the same level (OR 3.2, 95% CI [1.25-8.28], P=0.015).

Table 5.

Factors Associated with Fair/Poor Revised Modified MacNab Outcomes Compared to Excellent/Good

​		Short-term reoperations (<30d)			Long-term reoperations (>30d)			Combined timeframes
		Excellent/Good	Fair/Poor	P-value	Excellent/Good	Fair/Poor	P-value	Excellent/Good	Fair/Poor	P-value
Baseline Character-istics	Females, n (%)	7 (41.2%)	8 (61.5%)	0.273	46 (56.8%)	20 (51.3%)	0.57	53 (54.1%)	28 (53.8%)	0.199
	Mean Age, years (SD)	61.6 (±15.0)	61.1 (±8.7)	0.91	59.7 (±14.1)	60.2 (±11.9)	0.848	60 (±14.2)	60.4 (±11)	0.861
	BMI, kg/m2 (SD)	30.85 (±5.9)	33.2 (±6)	0.291	29.8 (±4.8)	29.5 (±5.8)	0.985	29.8 (±5.8)	30.7 (±5.3)	0.353
	Levels:	​		0.448	​		0.983	​		0.812
	L3/4, n (%)	0 (0%)	0 (0%)		2 (2.5%)	1 (2.6%)		2 (2%)	1 (1.9%)
	L4/5, n (%)	5 (29.4%)	7 (53.8%)		41 (50.6%)	19 (48.7%)		46 (46.9%)	26 (50%)
	L5/S1, n (%)	10 (58.8%)	5 (38.5%)		37 (45.7%)	17 (43.6%)		47 (48%)	22 (42.3%)
	Multi: L2-4, n (%)	2 (11.8%)	1 (7.7%)	​	0 (0%)	0 (0%)		2 (2%)	1 (1.9%)
	Multi: L4-S1, n (%)	0 (0%)	0 (0%)	​	1 (1.2%)	2 (5.1%)		1 (1%)	2 (3.8%)
	Privately-insured, n (%)	10 (58.8%)	9 (69.2%)	0.559	44 (54.3%)	23 (59%)	0.631	54 (55.1%)	32 (61.5%)	0.449
	WorkCover status, n (%)	2 (11.8%)	1 (7.7%)	0.714	3 (17.6%)	3 (23.1%)	0.358	5 (5.2%)	4 (7.7%)	0.528
Comorbid-ities	Diabetes, n (%)	1 (5.9%)	6 (46.2%)	0.025	11 (13.6%)	6 (15.4%)	0.791	12 (12.2%)	12 (23.1%)	0.09
	Hypertension, n (%)	7 (41.2%)	5 (38.5%)	0.88	31 (38.3%)	18 (46.2%)	0.411	38 (38.8%)	23 (44.2%)	0.687
	Osteoporosis/osteopenia*, n (%)	4 (23.5%)	3 (23.1%)	0.977	11 (13.6%)	4 (10.3%)	0.398	15 (15.3%)	7 (13.5%)	0.761
	Depression/anxiety, n (%)	5 (29.4%)	6 (46.2%)	0.349	22 (27.2%)	12 (30.8%)	0.681	27 (27.6%)	18 (34.6%)	0.37
	Smoking status, n (%)	6 (35.3%)	6 (46.2%)	0.548	26 (32.1%)	17 (43.6%)	0.221	32 (32.7%)	23 (44.2%)	0.163
	Alcohol consumption status, n (%)	10 (58.8%)	7 (53.8%)	0.785	48 (59.3%)	26 (66.7%)	0.435	58 (59.2%)	33 (63.5%)	0.61
Frailty Indices	Modified Charlson Comorbidity Index, score (SD)	0.94 (±1.25)	1.69 (±1.75)	0.188	0.58 (±1.05)	0.85 (±1.01)	0.497	0.65 (±1.1)	1.06 (±1.27)	0.045
	MFI-5, score (SD)	0.47 (±0.51)	1.00 (±1.15)	0.118	0.42 (±0.69)	0.56 (±0.64)	0.568	0.43 (±0.66)	0.67 (±0.81)	0.052
Spinal factors	Preoperative Antithrombotics, n (%)	4 (23.5%)	1 (7.7%)	0.271	15 (18.5%)	12 (30.8%)	0.136	19 (19.4%)	13 (25%)	0.426
	Presence preoperative spondylolisthesis, n (%)	7 (41.2%)	9 (69.2%)	0.133	53 (65.4%)	19 (48.7%)	0.082	60 (61.2%)	28 (53.8%)	0.118
	Grade preoperative spondylolisthesis	​		0.32	​		0.772	​		0.757
	Nil/Grade 0, n (%)	10 (58.8%)	4 (30.8%)		28 (34.6%)	20 (51.3%)		38 (38.8%)	24 (46.2%)
	Grade 1, n (%)	3 (17.6%)	5 (38.5%)		46 (56.8%)	13 (33.3%)		49 (50%)	18 (34.6%)
	Grade 2, n (%)	4 (23.5%)	4 (30.8%)		7 (8.6%)	6 (15.4%)		11 (11.2%)	10 (19.2%)
	Previous spinal surgery at that level, n (%)	3 (17.6%)	7 (53.8%)	0.045	6 (7.4%)	6 (15.4%)	0.181	9 (9.2%)	13 (25%)	0.012
	Previous lumbar spinal surgery at a different level, n (%)	2 (11.8%)	6 (46.2%)	0.047	9 (11.1%)	7 (17.9%)	0.531	11 (11.2%)	13 (25%)	0.075
Blood results	Platelet count, 109/L (SD)	272.5 (±60.2)	253.5 (±77.4)	0.444	261.3 (±60.2)	262.2 (±70.1)	0.801	264 (±68.3)	259.4 (±64.2)	0.685
	Haemoglobin levels, g/L (SD)	146.4 (±12.3)	134.3 (±15.4)	0.039	139.9 (±11.7)	138.4 (±16.0)	0.764	139.8 (±15.6)	138.5 (±12.8)	0.596
	Cholesterol levels, mmol/L (SD)	5.5 (±1.4)	5.1 (±1.2)	0.267	5.8 (±2.4)	5.1 (±1.1)	0.744	5.2 (±1.2)	5.6 (±2.1)	0.201
	Random glucose levels, mmol/L (SD)	5.8 (±1.7)	7.3 (±4.0)	0.209	6.4 (±2.5)	5.8 (±2.0)	0.346	5.8 (±1.9)	6.6 (±2.9)	0.067
Pre-operative symptoms	Bilateral radiculopathy, n (%)	2 (11.8%)	5 (38.5%)	0.102	24 (29.6%)	12 (30.8%)	0.898	26 (26.5%)	17 (32.7%)	0.428
	Motor symptoms, n (%)	1 (5.9%)	2 (15.4%)	0.406	8 (9.9%)	7 (17.9%)	0.66	9 (9.2%)	9 (17.3%)	0.507
	Sensory changes, n (%)	1 (5.9%)	3 (23.1%)	0.2	26 (32.1%)	14 (35.9%)	0.903	27 (27.6%)	17 (32.7%)	0.511
Intra-operative consider-ations	Side of approach	​		0.394	​		0.708	​		0.928
	Right, n (%)	7 (41.2%)	7 (53.8%)		40 (49.4%)	15 (38.5%)		47 (48%)	22 (42.3%)
	Left, n (%)	8 (47.1%)	4 (30.8%)		34 (42.0%)	15 (38.5%)		42 (42.9%)	19 (36.5%)
	Bilateral, n (%)	2 (11.8%)	2 (15.4%)		7 (8.6%)	9 (23.1%)		9 (9.2%)	11 (21.2%)
	Expandable cage, n (%)	7 (41.2%)	5 (38.5%)	0.88	39 (48.2%)	11 (28.2%)	0.04	46 (46.9%)	16 (30.8%)	0.057
	Operative time, minutes (SD)	214.8 (±63.7)	250.7 (±38.8)	0.093	222.9 (±53.7)	237 (±61.4)	0.358	229.1 (±61.9)	251.8 (±65.1)	0.042
	Dural Tears, n (%)	0 (0%)	0 (0%)	-	2 (2.5%)	1 (2.6%)	0.975	2 (2%)	1 (1.9%)	0.961
Post-op consider-ations	Use of postoperative antibiotics, n (%)	17 (100%)	12 (92.3%)	0.998	68 (84%)	35 (89.7%)	0.398	85 (86.7%)	47 (90.4%)	0.514
Follow up	Required reoperation, n (%)	3 (17.6%)	7 (53.8%)	0.045	16 (19.8%)	24 (61.5%)	<0.001	19 (19.4%)	31 (59.6%)	<0.001
	Length of stay, days (SD)	5.4 (±5.3)	7.3 (±7.3)	0.4	3.7 (±1.5)	3.9 (±1.4)	0.793	4.0 (±2.7)	4.7 (±4)	0.209
	Time until follow-up, days (SD)	193.6 (±151.4)	483.7 (±368.3)	0.027	378.4 (±405.2)	1077 (±839.5)	<0.001	346.3 (±379.6)	928.9 (±790.2)	<0.001
	Persistent pain, n (%)	4 (23.5%)	13 (100%)	0.998	21 (25.9%)	37 (94.9%)	<0.001	25 (25.5%)	50 (96.2%)	<0.001
	Persistent sensory changes, n (%)	0 (0%)	4 (30.8%)	0.998	7 (8.6%)	12 (30.8%)	0.003	7 (7.1%)	16 (30.8%)	<0.001
	Persistent motor changes, n (%)	0 (0%)	0 (0%)	-	1 (1.2%)	3 (7.7%)	0.106	1 (1%)	3 (5.8%)	0.127

Values in bold depict statistically-significant findings.

*There were 7 cases of osteopenia in the >30d cohort, 6 in the excellent/good and 1 in the fair/poor group.

Table 6.

Multivariate Logistic Regression for Risks for Fair/Poor Patient Outcomes

​	Short-term reoperations (<30d)			Long-term reoperations (>30d)			Combined timeframes
	OR	95% CI	P-value	OR	95% CI	P-value	OR	95% CI	P-value
Diabetic	14.143	1.039 – 192.571	0.047	-	-	-	-	-	-
Modified Charlson Comorbidity Index	-	-	-	-	-	-	1.344	1.000 – 1.806	0.050
Previous spinal surgery at that level	-	-	-	2.538	0.655 – 9.840	0.178	3.220	1.252 – 8.282	0.015
Bilateral radiculopathy	6.769	0.776 – 59.063	0.084	-	-	-	-	-	-
Expandable cage	-	-	-	-	-	-	-	-	-
Operative time	1.015	0.996 – 1.034	0.125	-	-	-	1	1 – 1.01	0.236
Model significance	χ 2 (3) = 11.595, P = 0.009			χ 2 (1) = 1.862, P = 0.172			χ 2 (2) = 10.126, P = 0.006
Fit (Hosmer and Lemeshow Test)	P = 0.898			P = 0.602			P = 0.277
Sensitivity of Model	66.7%			18.8%			27.5%
Specificity of Model	76.5%			91.7%			86.2%

Values in bold depict statistically-significant findings.

Overall, the significance of the models was worse than the reoperations models. Sensitivity was quite poor, ranging between 18.8-66.7% whilst specificity was strong between 76.5-91.7%.

Discussion

This high-volume single-centre cohort study reports factors associated with reoperation and poor patient outcomes after tubular MIS TLIF, with reoperations stratified into timeframes. Our overall reoperation rate was 6.6%, with mean follow-up 1.5 years (SD=1.7), within reported literature ranges (Table 7). The complication profile remains similar to our early experiences 11 years ago. ²⁰ In comparison to alternative approaches, our MIS TLIF reoperation rates are below those reported for open TLIFs, but remain above the few reported rates after endoscopic approaches.^{7,14,22-2425-30}

Table 7.

Summary of literature

Study	Methods	Sample size (n)	Reoperation rates	Risks for reoperation (univariate listed, multivariate in bold)	Follow-up
MIS:
Current study	Retrospective cohort study from single centre	756 MIS	6.6% Cage migration (n=14) Residual neural compromise (n=11) Screw loosening/misplacement (n=8) Dislodged cap/spacer (n=7) Cage subsidence (n=3) Haematoma (n=3) New anterolisthesis (n=2) Incorrect fusion level (n=1) Infection (n=1)	• Private • Diabetic • Osteoporosis • MFI-5 • Previous lumbar surgery not at index level • Motor symptoms	1.5 years
Ramanathan et al. 2023 6 USA	Retrospective cohort comparing MIS and open TLIFs	135 MIS	11.1% 5 decompressions, 10 re-do fusions Unspecified reasons (n=15)	-	5 years
Liang et al. 2022 28 China	Retrospective cohort comparing to decompressions	51 MIS	5.66% Unspecified reason (n=3)	-	2 years
Chan et al. 2020 27 USA	Database search comparing MIS and open TLIF	72 MIS	2.8% “Related reoperation” (n=1) ASD (n=1)	-	2 years
Kobayashi et al. 2018 29 Japan	Prospective database search of TLIF reoperations	65 MIS	3.1% When combined with open/PLIF techniques: • ASD • Infection • Screw malposition • Haematoma • Pseudoarthrosis • Cage protrusion	When combined with open/PLIF techniques: • Multilevel • Dural tear • Diabetes • EBL • Operative time	2 years
Tian et al. 2017 30 China	Prospective comparison between computer-assisted MIS and open	30 MIS	3.3% Haematoma (n=1)	-	2 years
Wong et al. 2014 7 USA	Comparison of two case series	144 MIS	2.7% Screw malposition (n=2) Cage migration (n=1) Vascular injury (n=1)	-	46 months
Parker et al. 2014 23 USA	Prospective comparison between MIS and open	50 MIS	0%	-	90 days
Archavlis and Carvi 2013 31 Germany	Retrospective cohort comparing MIS with open	24 MIS	12.5% Recurrence (n=2) Screw malposition (n=1)	-	2 years
Lee et al. 2012 32 Singapore	Prospective observational cohort of MIS vs open	72 MIS	1.4% Screw malposition (n=1)	-	2 years
Lee et al. 2012 33 Korea	Prospective case series	86 MIS	10.5% Pseudoarthrosis (n=4) Screw malposition (n=2) Infection (n=2) Bone fragment retrieval (n=1)	-	25 months
Rouben et al. 2011 25 USA	Retrospective case series	169 MIS	14.2% Mainly due to removal of “painful screws” Followed by infection	-	49 months
Lau et al. 2011 26 USA	Prospective case series	10 MIS	20% Infection (n=1) Instrumentation failure (n=1)	-	1 year
Villavicencio et al. 2010 34 USA	Cohort study comparing MIS and open	76 MIS	17.1% Recurrence (n=5) Screw malposition (n=4) Device-related issues (n=3) Infection (n=1)	-	37.5 months
Deutsch and Musacchio, 2006 35 USA	Prospective case series	20 MIS	5% Screw malposition (n=1)	-	6 months
Open:
Ramanathan et al. 2023 6 USA	Retrospective cohort comparing MIS and open	562 open	29.2% 44 decompressions and 120 re-do fusions Unspecified reasons (n=164)	-	5 years
Wang et al. 2022 9 Korea	Retrospective cohort from single centre	821 open	8.5% Device-related issues (n=20) SSI (n=17) ASD (n=17) Haematoma (n=8) Pain (n=7) Dural tear (n=1)	• Osteoporosis • Diabetes • Older age • Multilevel • Longer operative time • EBL	>2 years
Takaoka et al. 2021 36 Japan	Retrospective cohort comparing TLIF to OLIF	78 open	2.56% Adjacent segment disease (n=2)	-	3 years
Katuch et al. 2021 37 Slovakia	Prospective cohort comparing PLIF and TLIF	119 open	1.7% Wound infection (n=1) Dural tear (n=1)	-	4.3 years
Chan et al. 2020 27 USA	Database search comparing MIS and open TLIF	225 open	14.7% “Related reoperation” (n=16) ASD (n=6) Infection (n=6) Device-related issues (n=4) Pseudoarthrosis (n=1)	-	2 years
Kobayashi et al. 2018 29 Japan	Prospective database search of TLIF reoperations	525 open 65 MIS	1.9% When combined with MIS/PLIF techniques: • ASD • Infection • Screw malposition • Haematoma • Pseudoarthrosis • Cage protrusion	When combined with MIS/PLIF techniques: • Multilevel • Dural tear • Diabetes • EBL • Operative time	2 years
Tian et al. 2017 30 China	Prospective comparison between computer-assisted MIS and open	31 open	3.3% Deep infection (n=1)	-	2 years
Wong et al. 2014 7 USA	Comparison of two case series	54 open	11.1% Infection (n=3) Screw malposition (n=2) Cage migration (n=1)	-	46 months
Parker et al. 2014 23 USA	Prospective comparison between MIS and open	50 open	6% Infection (n=2) Device-related issues (n=1)	-	90 days
Lee et al. 2012 32 Singapore	Prospective observational cohort of MIS vs open	72 open	0%	-	2 years
Dhall et al. 2008 38 USA	Retrospective cohort comparing against mini-open	21 open	9.5% Screw malposition (n=1) Cage migration (n=1)	​	34 months
Endoscopic:
Zhu et al. 2021 39 China	Systematic review and meta-analysis of endoscopic TLIF approaches	548 endoscopic	0.5% Screw loosening (n=3)	-	19.8 months
Broad Database Studies:
Hikata et al. 2024 11 Japan	Propensity-matched retrospective cohort from multicentre database	1263 unspecified posterior lumbar fusions	5.7% SSI (n=12) Device-related issues (n=11) ASD (n=8) Pseudoarthrosis (n=8) Recurrence of central stenosis (n=6) Vertebral fracture (n=5) Haematoma (n=4) Other (n=3)	• Diabetes • Preoperative ambulation status • ASA	∼2 years
Noh et al. 2022 14 Korea	Insurance database search for any lumbar surgery with 10 years follow-up	6300 (39% unspecified fusion)	13.2% Unspecified reason (n=831)	• Male • Age in 60’s • Alcohol consumption • Smoking • Haemoglobin count • CCI ≥3	10 years
Kim et al. 2019 1 Korea	Insurance database search	11706 unspecified lumbar fusions	8.1% Unspecified reason (n=855)	-	5 years
Durand et al. 2018 13 USA	American College of Surgeons database search for posterior lumbar fusion reoperations within 30 days	22151 unspecified posterior lumbar fusions	3.4% Missing reason (n=172) Infection (n=121) Device-related issue (n=65) Haematoma (n=57) Seroma (n=42) Dural tear (n=21) Wound issue (n=19) Other including recurrence (n=36)	• Disseminated cancer • Bleeding disorders • ASA 3 • BMI ≥35 • Multilevel fusion	30 days

Findings in bold reflect factors associated with reoperation after multivariate analysis in the literature.

Importantly, 42 of 50 reoperations were for revision of instrumentation or residual neural compression, corresponding to 5.5% of the total cohort. These events may reflect further procedure-specific technical factors that were not included in this study, including cage positioning, hardware placement, decompression adequacy, endplate preparation, and patient anatomical/bone-quality factors. As the cumulative incidence curve (Figure 1) suggests ongoing reoperation risk beyond the mean follow-up period, the reported long-term rate should be considered a minimum estimate rather than a definitive lifetime reoperation risk.

Nevertheless, the indications for reoperation are consistent with those found in other studies. Posterior cage migration (0.7-6% in the literature) and general instrumentation failure remain inherent issues of the minimally invasive nature of this posterior approach, largely attributable to limited contact between the cage and vertebral body.^7-9,32 Such complications can be minimised by avoiding endplate injury, assessing disc shape to allow anterior placement of cage intraoperatively, and carefully considering cage size. ⁴⁰ Theoretically, expandable cages may prevent migrations, however, this association was not noted in our study. ⁴¹ Our infection rates were low at 0.1% (compared to 0.6-4% in the literature), alongside far fewer CSF leaks than the 1.3-8% reported in the literature, with only three intraoperative dural tears noted, all of which were managed conservatively.^8,22

Overall factors associated with reoperation were private and WorkCover status, reduced bone density, diabetes, and previous lumbar surgery not at the index level. These aforementioned factors were noted for long-term reoperation, with diabetes additionally noted in the combined timeframes. No independent short-term associations were identified, however, the short-term subgroup contained only 10 events which limits model stability and increases the risk of a type two error or overfitting.

The mechanisms remain incompletely understood. Osteoporosis is known to contribute to hardware malfunction by promoting poor implantation and cage subsidence. Additionally, it is a risk factor for further degenerative spondylolisthesis, and therefore, has already been associated with increased reoperations.^42,43 At our centre, patients at risk of osteoporosis undergo dual-energy X-ray absorptiometry screening to help guide decision-making. Previous lumbar surgery at adjacent levels likely reflects a generally worse condition of the spine, which would perpetuate adjacent segment disease, a common indication for reoperation in other papers.^6,9,11 The split of patients into private and public sectors is only relevant for some healthcare systems and is rarely reported in the literature. This association likely reflects numerous unmeasured confounding factors, including socioeconomic status, healthcare access, patient expectations, compensation status, and health literacy. ⁴⁴ Although public patient status was observed to be associated with reduced rates of reoperation, this finding should therefore be viewed as a context-specific health-system association, rather than a causal predictor. WorkCover patients receive financial compensation for missing work, with this cohort repeatedly linked with worse outcomes and a higher reoperation rate in the literature.^45,46 Finally, diabetes has been proposed to impact healing, however this may not fully account for its role in the short-term setting.^9,47 Whilst frailty indices have been linked with reoperation, due to their representation of the overall health of the patient, we did not note any such statistical association.^13,14,47

Alternative techniques such as open were noted to share diabetes and osteoporosis as associated factors, however, noted additional factors such as male sex, smoking and alcohol consumption.^8,14 Comparison with an endoscopic approach is difficult, with a recent meta-analysis by Zhu et al. noting only 3 cases of reoperation. ³⁹

The association of patients who had reoperations with a longer time until the most recent follow-up appointment likely reflects the chronic nature of TLIF complications. Thirty-two percent of these reoperations occurred after the standard follow-up of 12 months (Figure 1). However, this may also represent a bias of capturing complications that re-present, although the opposite may also be true whereby dissatisfied patients are lost to follow-up.

As a secondary aim of this study, we assessed the impact of the recorded parameters on patient outcomes. As expected, cases of reoperation skewed towards worse MacNab outcomes than their control counterparts. Across the stratified cohorts, fair/poor outcomes were associated with diabetes, the mCCI frailty index, and previous surgery at the same level. These are similar to those for reoperation, albeit with previous surgery at the same level rather than different levels, possibly reflecting a dissatisfaction with an unresolving pathology.

Limitations

Few studies have reported such a wide array of potential risk factors for MIS TLIF reoperation, however, some limitations remain. Due to the nature of case-control matching, the level and number of levels could not be assessed as potential factors. Other studies have noted multilevel procedures requiring more reoperations.^9,13 Manual control selection, although blinded to outcomes and performed before data extraction, may introduce selection bias with some potential that not all non-reoperation patients are fully represented. Although our centre has extensive follow-up and two emergency departments (public and private) which capture the majority of emergent re-presentations, there remains the possibility of an unknown portion of patients who were lost to follow-up. This, however, is minimised in the context of short-term reoperations within 30 days and high attendance rates to early reviews; nevertheless, it may introduce bias as less-satisfied patients may seek follow-up elsewhere. In terms of statistical limitations, the number of variables assessed increases the risk of type one errors from the multiple comparisons, although formal correction was not applied given the exploratory retrospective design and correlation with clinical variables. Thus, the significant findings should be noted as potential associations, with further external validation required. Although Kaplan-Meier modelling accounted for censoring at last follow-up, the matched case-control design limited formal time-to-event analysis of associated factors. Therefore, late reoperation risk may be underestimated and its associations should be interpreted in context of differential follow-up. Additionally, factors assessed could only be selected from those recorded by hospital staff, which in future will be minimised by using a prospective questionnaire. The associations between funding status, WorkCover status, and reoperation may reflect system-level confounding rather than biological risk, which requires further studies incorporating health-system factors (for example socioeconomic status, healthcare access) to better delineate these findings. Finally, the extensive dataset collected does still omit some variables of interest, including pain visual analogue scores, Oswestry Disability Index, and fusion rates (which may allow for further assessments of reoperation mechanisms, such as pseudoarthrosis or adjacent segment disease), all of which will be amended in future research through the prospective questionnaire and set follow-up protocols.

Conclusion

We present one of the largest single-centre experiences with MIS TLIFs, with an overall reoperation rate of 6.6% and a mean cohort follow-up of 1.5 years. Private status, WorkCover status, osteoporosis/osteopenia, diabetes, and previous lumbar surgery at a different level were all associated with reoperation. Worse patient outcomes were associated with similar factors, namely diabetes, a higher mCCI frailty index, and previous surgery at the same level. This study reports our complication profile and identifies associated preoperative factors that may help inform decision-making in the preoperative counselling phase after further external validation.

ORCID iD

Aaron Lerch https://orcid.org/0000-0002-4992-9764

Ethical Considerations

This research was formally approved by Mater Misericordiae human research ethics committee and governance committee – MSSA/MRGO/104172 (V3). This study was conducted in accordance with the Declaration of Helsinki. Specific consent for inclusion in this study was waived by the ethics committee due to the impracticality of retrospective consent, however, all patients consented to the procedure and the recording of their data.

Data Availability Statement

Data is available on request from the authors.*