Background: Dedicated artificial intelligence (AI) systems for fracture detection already exist, yet general-purpose multimodal models are increasingly accessible to clinicians despite not being developed or formally validated as medical devices. Their behavior in focused orthopedic imaging tasks remains insufficiently characterized. Purpose: To characterize how two accessible general-purpose multimodal models interpret AP pelvis radiographs with hip fractures, focusing on context dependence, overconfidence, and complementary error patterns within a surgically confirmed positive-only cohort. This was a behavioral characterization study of a fracture-positive cohort, not a diagnostic accuracy evaluation. Methods: In April 2026, we retrospectively studied 214 surgically confirmed hip fractures on AP pelvis radiographs using two general-purpose multimodal models under six prompting conditions. In runs A–D, the models were explicitly told that a hip fracture was present and were asked to classify it; in runs E–F, they were not told whether a hip fracture was present. Each image was rerun de novo in a separate chat session through vendor APIs using a fixed base prompt and no image preprocessing. We recorded hip-fracture detection, correct laterality, coarse fracture pattern, intracapsular displacement, AO/OTA grading, subtrochanteric identification, and self-reported confidence. Because the cohort contained hip fractures only, we report fracture-detection rates and classification performance within a positive-only cohort rather than full diagnostic-accuracy metrics. Results: Using the more conservative endpoint of hip-fracture detection with correct laterality, GPT-5.4 was correct in 79.0% and 86.4% of cases in runs E and F, whereas Gemini was correct in 80.4% and 93.5%, respectively. When outputs from both models were combined, this endpoint reached 89.7% in run E and 96.7% in run F, indicating complementary rather than redundant error patterns. Incorrect laterality cues markedly degraded performance, from 90.7% to 66.4% in GPT-5.4 and from 97.7% to 57.0% in Gemini. Performance remained limited for treatment-relevant subtyping, particularly AO/OTA grading and subtrochanteric identification. Both models frequently remained highly confident when wrong, and self-reported confidence did not reliably distinguish correct from incorrect outputs. Conclusions: Accessible general-purpose multimodal models showed partial capability for coarse hip-fracture interpretation, but they remained context-sensitive, unreliable for treatment-relevant subtyping, and highly confident even when incorrect. Their complementary error patterns are hypothesis-generating rather than evidence of clinical readiness. On the basis of these findings, we do not support unvalidated or uncontrolled clinical use of such models. As access to these tools expands, explicit usage boundaries, minimum performance expectations, repeated local revalidation, and sustained human oversight become increasingly necessary.

Background Rheumatoid arthritis (RA) is a systemic inflammatory disease that frequently involves the cervical spine and often necessitates surgical intervention. Although the peripheral joint burden of RA has declined in the biological era, many patients still require cervical spine surgery, and whether RA status independently influences in-hospital outcomes remains poorly defined in contemporary, nationally representative data. Methods We queried the National Inpatient Sample for all adult cervical spine surgery admissions from 2016 to 2021, identifying 164,837 unweighted cases (approximately 840,820 weighted). Patients with RA (ICD-10-CM M05–M06; n = 5082, 3.1%) were compared with non-RA controls (n = 159,755, 96.9%). Inverse probability of treatment weighting was applied to balance 27 preoperative covariates across 28 propensity model terms. Primary outcomes were in-hospital mortality, nonroutine discharge, length of stay, and total charges. Secondary outcomes included blood transfusion, blood loss anemia, pulmonary complications, sepsis/surgical site infection, venous thromboembolism, acute kidney injury, myocardial infarction/stroke, and additional cervical procedures during admission (proxy for unplanned intra-admission intervention). Results After inverse probability of treatment weighting adjustment (propensity model area under the curve = 0.702; maximum standardized mean difference = 0.085; all 40 balance metrics standardized mean difference < 0.10), the most striking finding was among RA patients undergoing occipitocervical or cervicothoracic fusion, who faced 7-fold higher odds of blood transfusion (OR 7.41; 95% CI 3.88–14.2) and nearly 4-fold higher odds of nonroutine discharge (OR 3.84; 95% CI 3.17–4.64) compared with single-level fusion. Across the full cohort, RA was independently associated with higher odds of nonroutine discharge (OR 1.339; 95% CI 1.261–1.421; P < 0.001), longer length of stay (+0.28 days; 95% CI 0.11–0.45; P = 0.001), greater total charges (+9.1 %; P < 0.001), blood transfusion (OR 1.526; 95% CI 1.263–1.844; P < 0.001), blood loss anemia (OR 1.479; 95% CI 1.318–1.660; P < 0.001), pulmonary complications (OR 1.183; 95% CI 1.018–1.375; P = 0.028), and additional cervical procedures during the same admission (OR 1.331; 95% CI 1.073–1.651; P = 0.009). In-hospital mortality did not differ significantly (OR 0.856; 95% CI 0.500–1.464; P = 0.569). Conclusions RA is independently associated with increased in-hospital morbidity and resource utilization following cervical spine surgery. These risks are graded by procedural complexity and are most pronounced among RA patients undergoing occipitocervical or cervicothoracic fusion, who demonstrate markedly increased transfusion requirements. The magnitude and procedure-specific gradient of these risks—quantified here in contemporary, biological-era data—provide an actionable basis for preoperative risk stratification, transfusion planning, and anticipatory discharge coordination in this population. Clinical Relevance RA patients undergoing cervical spine surgery—particularly those requiring occipitocervical or cervicothoracic fusion—warrant proactive transfusion preparedness (type-and-screen and blood-product availability), preoperative anemia optimization, and discharge planning initiated at the time of surgical scheduling. These procedure-specific risk estimates allow surgeons to counsel patients and plan perioperative care in concrete rather than generic terms. Level of Evidence 3.

Background: Whether coded rheumatoid arthritis (RA) among patients undergoing elective total hip arthroplasty (THA) is associated with excess short-term inpatient morbidity in contemporary U.S. practice remains uncertain. Methods: We performed a retrospective cohort study using the National Inpatient Sample (2016–2021). Adults undergoing elective primary THA were identified using ICD-10-CM/PCS codes. The exposure cohort included admissions with coded RA in any diagnosis position. The comparator cohort included admissions with narrow primary hip osteoarthritis (OA) as the principal diagnosis and no RA. Propensity scores were estimated from demographic, socioeconomic, hospital, and comorbidity variables. Inverse probability of treatment weighting (IPTW) with survey weighting was used for the primary analysis, followed by doubly adjusted regression models that included age and all variables with persistent post-weighting imbalance greater than 0.10 (female sex, osteoporosis, chronic obstructive pulmonary disease, chronic anemia, depression, and race). Additional sensitivity analyses included stricter weight truncation, a restriction analysis, conventional multivariable modeling, and sex-stratified analyses addressing residual sex imbalance. Results: A total of 274,582 admissions met eligibility criteria; 274,539 were included in the final analytic cohort after exclusion of 43 records with missing or invalid propensity-weight inputs. The analytic cohort included 7,264 RA admissions and 267,275 OA admissions. In the primary weighted analysis, RA was associated with higher odds of acute blood loss anemia (OR 1.30, 95% CI 1.23–1.38), red blood cell transfusion (OR 1.40, 95% CI 1.23–1.60), acute kidney injury (OR 1.47, 95% CI 1.26–1.71), non-home discharge (OR 1.27, 95% CI 1.19–1.36), and a composite hematologic/renal in-hospital complication (OR 1.31, 95% CI 1.24–1.39).RA was also associated with slightly longer length of stay (+ 0.14 days, 95% CI + 0.10 to + 0.19) and modestly higher total hospital charges (+$1,018, 95% CI +$89 to +$1,947). Findings were directionally consistent across sensitivity analyses, including stricter weight truncation, restriction to admissions with principal OA plus concomitant RA, and sex-stratified analyses. Conclusions: In this contemporary national THA cohort, coded RA was associated with a small but consistent increase in in-hospital complication and resource-utilization burden compared with OA. These findings should be interpreted as adjusted associations rather than causal effects, and the modest effect sizes should not be overstated. They may provide context for perioperative risk assessment and discharge planning among patients with coded RA undergoing elective THA.