Letter to the Editor Regarding “High Modified 5 Factor Frailty Index is Associated With Worse PROMs and Decreased Return to Activities After 1 or 2 Level MI-TLIF”

To the Editor,

Frailty is gathering great interest in spine surgery. ¹ We read with great appreciation Subramanian et al.’s timely contribution. They demonstrate that higher mFI-5 scores are associated with worse Patient Reported Outcomes (PROMs), lower fusion rates, and reduced return-to-activity after Minimally Invasive Transforaminal Lumbar Interbody Fusion (MI-TLIF). ² The manuscript addresses an important clinical question and uses a prospectively maintained multi-surgeon registry to generate clinically actionable findings. However, we wish to highlight a few important methodological limitations.

The severe-frailty group (mFI-5 ≥2) includes only 31 patients, which risks unstable estimates and potential influence from outliers. Moreover, only 226/392 patients had 1-year CT for fusion assessment, creating possible verification or selection bias.

Key variables that plausibly influence both PROMs and fusion including bone mineral density, smoking status, chronic steroid use, preoperative opioid dose, graft/biologic use (rhBMP or other adjuncts), and granular radiographic degeneration were not reported or adjusted for. Surgeon- and technique-level clustering (including robotics/navigation use and graft choice) was not modeled and may underestimate variance.

Analytically, the return-to-activity outcomes may be better characterized using time-to-event methods with censoring and competing risks rather than only median days and proportions. A formal mediation analysis could test whether longer operative time and higher estimated blood loss explain part of the frailty–outcome relationship. Interobserver reliability and a standardized CT fusion definition should be provided to strengthen the striking difference in pseudarthrosis rates.

We also propose re-analyzing mFI-5 as a continuous or ordinal predictor with spline testing for non-linearity and report per-point effect sizes. Moreover, we suggest running multivariable mixed-effects models adjusted for age, BMI, smoking, preoperative opioid dose, BMD (if available), number of levels, graft/biologic use, and ASA/CCI, performing sensitivity analyses that exclude the mFI-5 functional-dependence item to quantify circularity with baseline PROMs, comparing baseline characteristics of patients who did and did not undergo 1-year CT to assess selection bias; and, if differences exist, applying inverse-probability weighting or multiple imputation to estimate fusion rates across the full cohort.

Beyond reanalysis, the findings point to pragmatic translational steps. The signal that frail patients have similar magnitude of improvement but slower or incomplete recovery suggests they may benefit from targeted pre-habilitation, from perioperative care pathways tailored to minimize anesthesia time and blood loss, and to consider biologic augmentation in those with combined frailty and low BMD. ³

More nuanced frailty indices such as the Risk Analysis Index, Edmonton Frailty Scale, or indices incorporating objective physiologic measures and better capture multidimensional vulnerability than simple comorbidity counts and will provide more accurately discriminate risk, inform targeted optimization strategies, and improve calibration for outcomes such as prolonged recovery as demonstrated in the literature.^4,5

In summary, Subramanian et al. provide an important, practice-relevant contribution. Addressing CT ascertainment bias, richer confounder adjustment, continuous frailty modelling, surgeon-level clustering, and time-to-event analyses may strengthen causal inference and clinical utility.