Response to Letter to the Editor Regarding “New-Onset Psychiatric Disorders After Lumbar Fusion: Predictors,Timing,and Risk Stratification”

Dear Editor,

We are grateful to the author for their thoughtful and insightful Letter to the Editor regarding our study on new-onset psychiatric disorders following lumbar fusion. ¹ Their commentary highlights several important themes that meaningfully advance the conversation around postoperative psychiatric vulnerability. We appreciate the opportunity to elaborate on key aspects of our work and to contextualize the findings within the limitations inherent to large administrative datasets.

The author raises an important point regarding the definition of opioid exposure in our analysis. Because the Nationwide Readmissions Database (NRD) relies on ICD-10-CM codes without providing granular prescribing information, opioid “use,” “dependence,” and “abuse” cannot be distinguished with precision. We agree that this constraint introduces the potential for misclassification. It is worth noting, however, that such misclassification is likely nondifferential and may bias results toward the null rather than create spurious associations. The strong and directionally consistent relationships observed between perioperative opioid exposure and psychiatric outcomes echo well-established trends in spine and behavioral health literature, supporting the clinical relevance of our findings despite the coding limitations.

We also appreciate the author’s observation about right-censoring that may occur due to the calendar-year structure of the NRD. Surgeries occurring late in the year inherently have shorter follow-up windows, which may underestimate total 12-month incidence. Importantly, the primary aim of our study was to characterize early psychiatric transitions, particularly within the first 90 days, where the majority of diagnoses clustered. Because this early postoperative window is well captured within the NRD structure, the central temporal conclusions remain robust. Nonetheless, we agree that future longitudinal datasets extending beyond calendar-year boundaries would allow for more complete evaluation of long-term risk.

The author correctly notes that our outcomes were derived from inpatient diagnostic coding, which does not include outpatient psychiatric evaluations, psychotherapy encounters, or medication management visits. As such, our reported incidence likely underestimates the true burden of psychiatric morbidity following lumbar fusion. This is especially relevant for conditions such as depression and anxiety, which are frequently diagnosed and managed in outpatient settings. Conversely, rare outcomes such as homicidal ideation may be more likely to appear in inpatient encounters. These features reflect the strengths and constraints of inpatient administrative databases and underscore the value of complementary prospective and outpatient-inclusive studies.

We also appreciate the request for clarification regarding our use of the term “screened.” In the context of the NRD, psychiatric conditions are captured only when documented by providers during an inpatient encounter and do not reflect formal systematic screening using standardized tools such as PHQ-9, GAD-7, or HADS. We acknowledge that the NRD does not specify the method of diagnostic identification, and we agree that distinguishing between diagnosed psychiatric conditions and systematically screened conditions is an important conceptual distinction. Our use of the term “screened” was intended to indicate the timing of diagnostic recognition rather than formal instrument-based screening, which the database does not record.

Regarding the analytical approach, we recognize the author’s point that bivariate associations do not account for potential confounding structures. Our goal was to provide a population-level overview of incidence, early timing, and broad demographic and clinical predictors using a descriptive risk-identification framework. While more sophisticated methods, such as multivariable modeling, time-to-event analyses, or mediation analyses, would offer additional nuance, our focus in the present study was on characterizing temporal clustering and identifying practical, clinically interpretable risk signals. We concur that future work employing more advanced statistical techniques would be valuable for further elucidating causal pathways and refining risk stratification.

The author also highlights several operative and post-acute care variables, such as levels fused, surgical approach, and discharge supports, that were not available within the NRD. We agree that such variables may meaningfully influence psychiatric trajectories and represent important opportunities for future investigation. While discharge disposition was available in the database, the primary emphasis of our analysis was on demographic and modifiable clinical predictors most closely tied to early postoperative vulnerability. The observation that discharge disposition may offer additional predictive value is well taken and complements the broader conversation about identifying patients who may benefit from enhanced perioperative mental health support.

We appreciate the forward-looking suggestions provided in the letter, including the potential for time-to-event modeling with competing risks, target trial emulation of early screening vs usual care, mediation analyses exploring the relationship between pain, opioid exposure, and psychiatric outcomes, and fairness auditing of predictive tools across demographic groups. Likewise, the proposal to test structured three-point mental health screening protocols within stepped-wedge ERAS trials aligns directly with the clinical motivation behind our study. Incorporating validated tools such as PHQ-9, GAD-7, and HADS, as well as deploying care navigators for socially vulnerable or high-risk groups, represents a promising strategy for improving equity and recovery after lumbar fusion.

We thank the author once again for their careful reading and constructive input. Their commentary reinforces the clinical significance of psychiatric morbidity in spine surgery and highlights several promising avenues for advancing both research methodology and perioperative mental health care. We share their commitment to developing equitable, risk-stratified, and patient-centered pathways that integrate psychological well-being with surgical recovery. We are grateful to the Editor for the opportunity to respond and to continue this important dialogue.

Sincerely,

Ballatori et al