Too thin for safety, too heavy for discharge: body mass index and postoperative independence

Bald et al. provide a clear description of the non-linear association between BMI and loss of independent living after non-cardiac surgery 1, showing a U-shaped relationship across more than 73,000 procedures and identifying inflection points around 21.8 kg.m-2 and 39.3 kg.m-2. As surgeons, we welcome their use of non-home discharge as a patient-centred endpoint and their careful adjustment for comorbidity and frailty markers, which add important detail to how the ‘obesity paradox’ is understood in peri-operative care. We suggest two further analyses, both feasible within their existing dataset, which could strengthen the clinical interpretation of these findings. The primary outcome combines in-hospital mortality with discharge to a skilled nursing facility. This is in keeping with previous work, but the mechanisms, prognosis and clinical responses differ substantially between patients who die in hospital or survive but require institutional care 2. The U-shaped curve for adverse discharge appears similar to the patterns seen for ICU admission and 30-day mortality, yet the separate contributions of death and new dependency at different BMI values are not shown. From a head and neck oncology standpoint, this distinction is important clinically: very low BMI often reflects cancer-related cachexia and a trajectory towards end-of-life, whereas higher BMI more often predicts prolonged rehabilitation needs, for example after major resections with complex reconstruction. Decomposing the composite outcome, for example by presenting separate spline curves for mortality and for non-fatal discharge to facilities, or by using a multinomial model, could clarify whether extremely low BMI reflects terminal disease, while severe obesity more often captures functional dependency and care needs. In turn, this would help teams decide whether to prioritise early palliative care involvement or prehabilitation and discharge planning interventions for patients at the lower and upper ends of the BMI distribution. A second, practice-oriented question is how the empirically-derived BMI thresholds relate to thresholds that are already embedded in peri-operative pathways. The adjusted risk of adverse discharge rises significantly at BMIs 39.3 kg.m-2, while ‘underweight’ is classified using the conventional 18.5 kg.m-2 boundary. This implies that a sizeable group of patients with normal BMI between 18.5 and 22 kg.m-2 may already carry higher risk of losing independence yet would not trigger nutritional or frailty-focused screening 3. In head and neck cancer practice, many patients with longstanding dysphagia or weight loss fall into this range and can appear deceptively ‘acceptable’ by BMI alone. It would be informative if the authors could quantify, within their cohort, the proportion and adverse-discharge rate of patients in this 18.5–22 kg.m-2 BMI range, and estimate how many additional high-risk patients would be identified if peri-operative nutritional vulnerability flags were moved towards the empirical threshold, perhaps combined with the existing nutritional-risk variables. This would speak directly to whether BMI-based triggers in current pre-assessment pathways should be recalibrated and could help ensure that patients with borderline BMI, but substantial functional risk, are not overlooked. Bald et al. show that very low and very high BMIs carry a higher risk of postoperative loss of independence, whereas overweight and mild obesity may be relatively protective. Clarifying the roles of mortality and new institutionalisation, and tying the spline thresholds to simple triage thresholds, could further strengthen the clinical use of these data. It is our privilege to comment on this work, and we hope our observations help colleagues in anaesthesia, surgery and rehabilitation who are working to preserve independence after major surgery.