Surgical urgency and outcomes in nonagenarian and centenarian patients admitted to the ICU after hip replacement: a binational multicenter cohort study Authors: Je Min Suh, Dong-Kyu Lee, Angelina Yoon, Nattaya Raykateeraroj, Andrew Hardidge, Sophia Grobler, Grace Kwon, David Pilcher, Laurence Weinberg Published: 2026 ABSTRACT AbstractIntroductionThe global population of nonagenarians and centenarians is expected to quadruple by 2050, resulting in increasing numbers of the oldest-old undergoing procedures such as hip replacement to maintain mobility and independence. Despite this trend, evidence on how surgical urgency affects outcomes in this population remains limited.MethodsWe conducted a retrospective cohort study using data from the ANZICS Adult Patient Database (2010–2023), which captures 98% of intensive care units (ICUs) in Australia and 68% in New Zealand. The study included 1,578 nonagenarian and centenarian patients who underwent non-traumatic hip replacement and were admitted to intensive care. Outcomes examined were mortality at three time intervals (<30 days, 30–365 days and >365 days) and ICU and hospital length of stay. Analytical methods included propensity score matching, Cox regression, Kaplan-Meier curves and time-dependent area under the receiver operating characteristic curve analyses to assess the prognostic value of clinical scores and biomarkers.ResultsNon-elective operation was associated with significantly higher mortality across all time points (HR 2.87 at <30 days; 3.02 at 30–365 days; 1.84 beyond 365 days). It was also linked to a 13.1% longer ICU stay (p=0.034). Acute Physiology and Chronic Health Evaluation III and Sequential Organ Failure Assessment scores showed strong predictive performance for short-term mortality, whereas age and sex had minimal prognostic value.ConclusionAmong nonagenarians and centenarians undergoing hip replacement, non-elective procedures were consistently associated with poorer outcomes, including higher mortality and longer ICU stay. Early risk stratification using routine clinical markers, together with proactive goals-of-care discussions, may improve perioperative management in this vulnerable population.Level of evidenceLevel II Therapeutic/Care Management.Keywords: orthopedic, anesthesia, fracture, hip WHAT IS ALREADY KNOWN ON THIS TOPIC Outcomes of hip replacement operations in the oldest old are poorly defined, and existing studies rarely distinguish elective from non-elective procedures, leaving uncertainty about how surgical urgency influences prognosis in nonagenarians and centenarians. Outcomes of hip replacement operations in the oldest old are poorly defined, and existing studies rarely distinguish elective from non-elective procedures, leaving uncertainty about how surgical urgency influences prognosis in nonagenarians and centenarians. WHAT THIS STUDY ADDS In this large binational cohort, non-elective hip replacement was associated with substantially higher early and late mortality and longer intensive care unit stays than elective operations, and routinely collected variables such as lactate, urine output, Sequential Organ Failure Assessment score, Acute Physiology and Chronic Health Evaluation III score, and Glasgow Coma Scale provided strong short-term prognostic value. In this large binational cohort, non-elective hip replacement was associated with substantially higher early and late mortality and longer intensive care unit stays than elective operations, and routinely collected variables such as lactate, urine output, Sequential Organ Failure Assessment score, Acute Physiology and Chronic Health Evaluation III score, and Glasgow Coma Scale provided strong short-term prognostic value. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY These findings highlight surgical urgency as a key determinant of outcome and support the need for structured perioperative assessment, early multidisciplinary involvement, and clearer goals-of-care discussions for the oldest surgical patients. These findings highlight surgical urgency as a key determinant of outcome and support the need for structured perioperative assessment, early multidisciplinary involvement, and clearer goals-of-care discussions for the oldest surgical patients. Introduction The global population of nonagenarians and centenarians is expanding rapidly, with projections estimating a fourfold increase by 2050 due to increased longevity and improvements in preventive health and chronic disease management.1,3 Concomitantly, surgical interventions in this demographic are becoming increasingly common, particularly orthopedic operations such as elective total hip arthroplasty.4 5 As functional preservation and quality of life gain prominence among older adults, hip replacement operations are now increasingly offered to patients in their 9th and 10th decades of life.6 Despite these trends, a critical knowledge gap persists regarding the perioperative outcomes of hip replacement operations, specifically in nonagenarians and centenarians. Existing literature predominantly focuses on octogenarians or fails to distinguish between elective and non-elective procedures, thus overlooking the prognostic implications of surgical urgency in a cohort characterized by marked frailty.7 8 Non-elective operations, often performed under urgent, time-constrained and physiologically unstable conditions, may confer physiological burden and mortality risks distinct from those of elective interventions.9 10 A nuanced understanding of these differences is essential to optimize perioperative care and facilitate informed consent processes, particularly in patients with limited physiological reserves. Clarifying the impact of surgical urgency in nonagenarian and centenarian patients is crucial not only for clinical risk stratification, but also for resource planning and ethical care delivery.11 In an era marked by increasing demand for intensive care resources and the potential for prolonged recovery trajectories, evidence-based guidance is needed to inform triage decisions, facilitate early goals-of-care discussions, and align treatment plans with patient preferences and expected outcomes.12 We hypothesized that non-elective hip replacement procedures in nonagenarian and centenarian patients would be associated with higher short- and long-term mortality, prolonged intensive care stays, and greater physiological derangement compared with elective procedures. This study aims to address this gap by evaluating the differential outcomes of elective and non-elective hip replacement operations among nonagenarians and centenarians admitted to intensive care using multicenter data from across Australia and New Zealand. Methods Data sources, study population, and variables De-identified patient-level data were obtained from the ANZICS Adult Patient Database, a comprehensive registry capturing demographic, clinical, and outcome data from 98% of intensive care units (ICUs) in Australia and 68% in New Zealand.13 Data collection and management were overseen by the ANZICS Center for Outcome and Resource Evaluation. This study included nonagenarian and centenarian patients who underwent hip replacement operation for non-traumatic indications, as identified using Acute Physiology and Chronic Health Evaluation (APACHE) III J diagnostic codes. Patients whose hospital admission was due to trauma, such as motor vehicle accidents or other severe injuries, were excluded. Elective arthroplasty was defined as a scheduled operation for chronic degenerative disease without an acute admission trigger. Non-elective arthroplasty was defined as an urgent, non-traumatic admission requiring operation within 24 hours for time-sensitive indications such as rapidly progressive osteonecrosis, acute implant failure without fracture, or periprosthetic infection. The extracted demographic variables included age, sex, admission source, admission type, discharge destination, and dates of ICU and hospital admission. For each patient, clinical scores such as APACHE III-J, Glasgow Coma Scale (GCS), and Sequential Organ Failure Assessment (SOFA) were recorded at the time of ICU admission. Biochemical markers were documented as peak and nadir values within the first 24 hours of the ICU stay. Relevant comorbidities, including immunosuppression; malignancy; and chronic respiratory, cardiovascular, hepatic, or renal disease, were also captured. Study outcomes The primary outcomes were the associations between surgical urgency (elective vs non-elective hip replacement) and the following1: all-cause mortality at short term (<30 days), intermediate term (30–365 days), and long term (>365 days) intervals after ICU admission and2 ICU and hospital lengths of stay. The secondary outcome was the discriminatory performance of routinely collected clinical scores (APACHE III, SOFA, and GCS) and biochemical variables (such as lactate, urine output, and creatinine) in predicting short-term mortality. Statistical analyses All statistical analyses were conducted in R software, V.4.4.3 (R Foundation for Statistical Computing, Vienna, Austria). Patients were stratified into elective and non-elective operation groups. Descriptive analyses were performed to compare baseline characteristics. Continuous variables were assessed for normality with the Shapiro-Wilk test and visual inspection of histograms and Q–Q plots. Because most continuous variables were non-normally distributed, they were summarized as medians with IQRs and compared with the Mann-Whitney U test. Categorical variables were summarized as frequencies with percentages and compared with Pearson’s χ² test or Fisher’s exact test, as appropriate. To decrease confounding and emulate randomization, we performed propensity score matching with one-to-one nearest-neighbor matching without replacement via the MatchIt package, with a caliper width of 0.1 on the logit of the propensity score. Clinically relevant covariates specified a priori, including age, sex, APACHE III score, SOFA score, and comorbidities, were included in the propensity model. Covariate balance before and after matching was assessed with standardized mean differences. Values less than 0.1 indicated acceptable balance. The results were visualized with Love plots generated with the cobalt package. Time-to-event analyses were performed with Cox proportional hazards regression models implemented with the survival package. The proportional hazards assumption was tested for each covariate with scaled Schoenfeld residuals. As global and covariate-specific violations were detected, follow-up time was partitioned into three prespecified intervals (0–30 days, 31–365 days, and >365 days), and separate Cox models were fitted for each interval. After time stratification, the proportional hazards assumption was re-evaluated and was found to be satisfied for all covariates within each interval. HRs with 95% CIs are reported for both the unmatched (univariate models) and matched cohorts. Kaplan-Meier survival curves were generated and compared between groups with the log-rank test, and plots were constructed with the survminer package.14 To assess the discriminatory performance of clinical and biochemical variables within elective and non-elective groups, we constructed time-dependent receiver operating characteristic curves at multiple time points (7, 30 days) with the timeROC packages.15 The area under the receiver operating characteristic curve (AUROC) and 95% CIs were estimated via bootstrap resampling with 1000 iterations. All statistical tests were two-tailed, and p<0.05 was considered statistically significant. Missing data were minimal (<5% for all included variables) and were handled with complete-case analysis. Results Baseline characteristics A total of 1,578 nonagenarian and centenarian patients who underwent non-traumatic hip replacement operation were included, of whom 327 (20.7%) underwent non-elective procedures, and 1,251 (79.3%) underwent elective procedures. As shown in online supplemental Tables 1 and 2, before propensity score matching, patients in the non-elective group, compared with the elective group, were more frail and exhibited greater illness severity at ICU admission, as reflected by higher median APACHE III scores (64 (IQR 56–73.5) vs 58 (51–67), p<0.001) and SOFA scores (3 (2–5) vs 3 (1–4), p<0.001). The non-elective group also had lower GCS scores and higher lactate concentrations. ICU length of stay (LOS) was significantly longer in the non-elective group (29.6 (19.9–64.6) vs 24.0 (19.5–45.6) hours, p=0.003). Comorbidities such as chronic respiratory, cardiovascular, and renal diseases were more prevalent in the non-elective group. Treatment limitation orders were substantially more common in the non-elective group (37.9% vs 9.6%, p<0.001), and ICU (3.7% vs 0.7%) and hospital mortality (9.8% vs 3.2%) were also significantly higher (p<0.001 for both). After 1:1 propensity score matching, 318 matched pairs were retained for analysis (postmatching baseline characteristics in tables1 2). All demographic, physiological, and biochemical variables were well balanced between groups, and standardized mean differences were <0.1 for key covariates (visualized in figure 1). Small but statistically significant differences between the non-elective and elective groups remained in lactate (1.5 (1.1–2.2) vs 1.2 (0.9–1.63) mmol/L, respectively; p<0.001), heart rate, and serum albumin levels. Patients in the non-elective group were also less likely to be admitted from home and more likely to have treatment limitation orders (38.0% vs 12.2%, p<0.001). ICU mortality (2.8% vs 1.6%, p<0.042), hospital mortality (8.9% vs 4.4%, p<0.036), and longer-term outcomes such as 30-day (11.9% vs 4.1%, p<0.001) and 365-day mortality (31.1% vs 13.2%, p<0.001) remained significantly poorer in the non-elective group. Table 1Postmatching baseline characteristics—continuous variablesVariableNon-elective operation(n=318)Elective operation(n=318)P valueAge (years)92.6 (91.3–94.3)92.31 (91–94.2)0.250Hospital LOS (days)11.11 (7.01–18.41)10.30 (7.00–15.36)0.004ICU LOS (days)1.23 (0.83–2.65)1.03 (0.81–1.98)0.481Clinical Frailty Scale6 (5–7)5 (4–6)0.065Glasgow Coma Scale15 (14–15)15 (14–15)0.073APACHE III Score64 (55–72)63 (53–73)0.079SOFA Score3 (2–5)3.5 (2–5)0.273Survival days44.97 (13.61–553.35)310.48 (16.07–1060.05)<0.001*Lactate (mmol/L)1.5 (1.1–2.2)1.2 (0.9–1.63)<0.001*Maximum heart rate (bpm)90 (80–103)87 (76.25–100)0.003Minimum heart rate (bpm)65 (60–74)61 (55–70)0.006Maximum respiratory rate (breaths/min)23 (20–26)22 (20–25)0.748Minimum respiratory rate (breaths/min)12 (11–14)12 (10–14)0.598Maximum systolic blood pressure (mm Hg)145 (130–160)146 (135–165)0.002Minimum systolic blood pressure (mm Hg)95 (85–110)98.5 (87.25–110)0.038Maximum diastolic blood pressure (mm Hg)65.5 (56–76)65 (57–75.5)0.544Minimum diastolic blood pressure (mm Hg)45 (40–52)44 (40–50)0.867Maximum mean arterial pressure (mm Hg)93 (83–103)93 (83–105)0.520Minimum mean arterial pressure (mm Hg)63 (57–70)61 (55–70)0.572Maximum sodium (mmol/L)138 (136–140.5)138 (136–141)0.223Minimum sodium (mmol/L)137 (134–139)137 (135–139)0.175Maximum potassium (mmol/L)4.5 (4.3–4.9)4.5 (4.2–4.9)0.574Minimum potassium (mmol/L)4.2 (3.9–4.6)4.2 (3.9–4.5)0.592Maximum bicarbonate (mmol/L)24 (22–27)25 (23–27)0.181Minimum bicarbonate (mmol/L)23 (20.5–25)23 (21–26)0.159Maximum creatinine (μmol/L)93 (69.75–127)84 (66–119)0.072Minimum creatinine (μmol/L)82.5 (63–112.25)78 (61–108.5)0.702Maximum hematocrit0.32 (0.28–0.35)0.32 (0.29–0.35)0.291Minimum hematocrit0.29 (0.26–0.32)0.29 (0.26–0.32)0.991Maximum hemoglobin (g/L)104 (95–114.3)104 (94–114)0.841Minimum hemoglobin (g/L)94 (83–105)94 (84–106.5)0.438Maximum WCC (×10⁹/L)12 (9.7–15.85)11.3 (9.3–14.2)0.772Minimum WCC (×10⁹/L)10.05 (8.3–13.1)9.3 (7.4–11.6)0.241Maximum platelets (×10⁹/L)192 (153–239)198 (152.25–261)0.021Minimum platelets (×10⁹/L)169.5 (137.75–218)179 (138–229)0.774Maximum glucose (mmol/L)8.2 (6.9–9.8)8 (6.8–9.5)0.903Minimum glucose (mmol/L)6.4 (5.5–7.6)6.4 (5.6–7.45)0.535Maximum PaO₂ (mm Hg)90 (70–125)99 (74.25–136.5)0.894Maximum PaCO₂ (mm Hg)39 (34–43)40 (36–46)0.018Maximum arterial pH7.39 (7.34–7.43)7.38 (7.34–7.42)0.120Maximum urea (mmol/L)8.9 (6.43–12.43)7.8 (5.9–11.8)0.619Urine output (mL/day)1057.5 (713.75–1500)1161 (866.75–1595.75)0.015Maximum bilirubin (μmol/L)12 (8–18)10 (7–14)0.159Maximum albumin (g/L)27 (24–30)29 (25–31)<0.001*Values are presented as median (IQR). Elective and non-elective surgical patients were compared with the Wilcoxon rank-sum test for continuous variables. p<0.05 was considered statistically significant and is indicated with an asterisk (*).All physiological and biochemical values represent the maximum or minimum recorded during the ICU stay, as appropriatePaCO₂, arterial partial pressure of carbon dioxide.APACHE, Acute Physiology and Chronic Health Evaluation; ICU, intensive care unit; LOS, length of stay; PaO₂, arterial partial pressure of oxygen measured in mmHg; SOFA, Sequential Organ Failure Assessment; WCC, white blood cell count. Postmatching baseline characteristics—continuous variables Values are presented as median (IQR). Elective and non-elective surgical patients were compared with the Wilcoxon rank-sum test for continuous variables. p<0.05 was considered statistically significant and is indicated with an asterisk (*). All physiological and biochemical values represent the maximum or minimum recorded during the ICU stay, as appropriate PaCO₂, arterial partial pressure of carbon dioxide.APACHE, Acute Physiology and Chronic Health Evaluation; ICU, intensive care unit; LOS, length of stay; PaO₂, arterial partial pressure of oxygen measured in mmHg; SOFA, Sequential Organ Failure Assessment; WCC, white blood cell count. Table 2Postmatching baseline characteristics—categorical variablesVariableCategoryNon-elective operation(n=318)Elective operation(n=318)P valueCountryAustralia302 (95)313 (98.4)0.026New Zealand16 (5)5 (1.6)Hospital sourceHome/hospital in the home216 (69.9)278 (87.4)0.212Other acute hospital (not ICU/ED)30 (9.7)14 (4.4)Nursing home/chronic care/palliative care48 (15.5)18 (5.7)ICU, other hospital1 (0.3)1 (0.3)Rehabilitation4 (1.3)2 (0.6)ED, other hospital10 (3.2)5 (1.6)SexFemale198 (62.3)200 (62.9)0.199Male120 (37.7)118 (37.1)ICU sourceOT/recovery279 (87.7)317 (99.7)<0.001*Ward/coronary care/other HDU11 (3.5)0 (0)Same hospital21 (6.6)1 (0.3)Other hospital5 (1.6)0 (0)ICU, other hospital1 (0.3)0 (0)Direct ICU admission (from home or hospital in the home)1 (0.3)0 (0)Readmission3 (0.9)3 (0.9)>0.99Hospital mortality28 (8.9)14 (4.4)0.036ICU mortality9 (2.8)5 (1.6)0.042Chronic respiratory disease25 (7.9)27 (8.5)0.885Chronic cardiovascular disease73 (23)62 (19.5)0.332Chronic renal disease22 (6.9)22 (6.9)>0.99Chronic liver disease0 (0)0 (0)>0.99Immune disease1 (0.3)1 (0.3)>0.99Immunotherapy3 (0.9)3 (0.9)>0.99Hepatic failure0 (0)0 (0)>0.99Lymphoma1 (0.3)2 (0.6)>0.99Metastasis1 (0.3)1 (0.3)0.478Leukemia1 (0.3)2 (0.6)>0.99Immunosuppressed3 (0.9)3 (0.9)>0.99Insulin-dependent diabetes mellitus0 (0)2 (2)0.590Treatment limitationFull active management (without treatment limitation) treatment191 (61)274 (87.8)0.001Limitation order119 (38)38 (12.2)Palliative care of a dying patient3 (1)0 (0)DeliriumYes10 (7)12 (7.1)>0.99Acute renal failureYes15 (4.9)12 (3.9)0.681All-cause mortality151 (47.5)99 (31.1)<0.001*Hospital outcomeDied28 (8.9)14 (4.4)<0.001*Discharged home120 (38)139 (43.7)Discharged to nursing home/chronic care/palliative care85 (26.9)76 (23.9)ICU, other hospital0 (0)1 (0.3)Other acute hospital33 (10.4)24 (7.5)Rehabilitation44 (13.9)63 (19.8)Hospital in the home1 (0.3)0 (0)Other5 (1.6)1 (0.3)ICU outcomeDied9 (2.8)5 (1.6)<0.001*Survived ICU: home/ward/other302 (95)310 (97.8)Transferred to another ICU at another hospital or the same hospital7 (2.2)2 (0.6)Transferred to another hospital ward9 (2.8)5 (1.6)30-day mortality38 (11.9)13 (4.1)<0.001*365-day mortality99 (31.1)42 (13.2)<0.001*Values are presented as number (percentage).Comparisons between groups were performed with the χ² test or Fisher’s exact test, as appropriate. p<0.05 was considered statistically significant and is indicated with an asterisk (*).Elective admission indicates planned, non-emergency ICU admissions. Treatment limitations include orders such as do-not-resuscitate or limitations on escalation of care.OT, operating theater.ED, emergency department; HDU, high dependency unit; ICU, intensive care unit. Postmatching baseline characteristics—categorical variables Values are presented as number (percentage). Comparisons between groups were performed with the χ² test or Fisher’s exact test, as appropriate. p<0.05 was considered statistically significant and is indicated with an asterisk (*). Elective admission indicates planned, non-emergency ICU admissions. Treatment limitations include orders such as do-not-resuscitate or limitations on escalation of care. OT, operating theater.ED, emergency department; HDU, high dependency unit; ICU, intensive care unit. Love plot of propensity matching. APACHE, Acute Physiology and Chronic Health Evaluation; SOFA, Sequential Organ Failure Assessment. Short- and long-term mortality Non-elective operation was consistently associated with higher mortality across all time intervals (table 3 and figure 2). In the unmatched cohort, patients undergoing non-elective operation, compared with elective operation, had a 3.5-fold higher risk of death within 30 days (HR 3.51, 95% CI 2.34 to 5.27, p<0.001), a 3.4-fold higher risk between 30–365 days (HR 3.39, 95% CI 2.45 to 4.69, p<0.001), and nearly twice the risk beyond 1 year (HR 1.94, 95% CI 1.44 to 2.62, p<0.001). Age was also significantly associated with mortality: every additional year of age was associated with an 8% higher risk of death within 30 days (HR 1.08, 95% CI 1.01 to 1.17), a 16% higher risk from 30 to 365 days (HR 1.16, 95% CI 1.10 to 1.22), and a 14% higher risk beyond 1 year (HR 1.14, 95% CI 1.09 to 1.20). Each unit increase in the APACHE III score was associated with a 5%, 3%, and 2% increase in mortality risk across the respective time intervals (all p<0.001). The SOFA score was similarly associated with a 32% higher risk within 30 days, 27% higher risk during 30–365 days, and 11% higher risk beyond 1 year (all p<0.001). Male sex was associated with elevated mortality only in the <30-day and >365-day intervals. Table 3Time-dependent Cox proportional hazards regression for mortality before and after propensity score matchingTerm<30 days HR (95% CI)<30 days p value*30–365 days HR (95% CI)30–365 days p value*>365 days HR (95% CI)>365 days p value*PrematchingNon-elective operation3.510 (2.34 to 5.27)<0.001*3.390 (2.45 to 4.69)<0.001*1.940 (1.44 to 2.62)<0.001*Age1.080 (1.01 to 1.17)0.0301.160 (1.10 to 1.22)<0.001*1.140 (1.09 to 1.20)<0.001*APACHE III Score1.050 (1.04 to 1.06)<0.001*1.030 (1.03 to 1.04)<0.001*1.020 (1.01 to 1.03)<0.001*Sex (male)1.560 (1.04 to 2.35)0.0321.310 (0.949 to 1.81)0.1011.450 (1.14 to 1.85)0.003SOFA Score1.320 (1.22 to 1.43)<0.001*1.270 (1.18 to 1.36)<0.001*1.110 (1.05 to 1.19)<0.001*PostmatchingNon-elective operation2.870 (1.56 to 5.30)0.0013.020 (1.93 to 4.73)<0.001*1.840 (1.26 to 2.69)0.001Age1.050 (0.96 to 1.16)0.2921.130 (1.05 to 1.21)0.0011.130 (1.05 to 1.21)0.001APACHE III Score1.040 (1.03 to 1.06)<0.001*1.040 (1.03 to 1.05)<0.001*1.020 (1.00 to 1.03)0.026Sex (male)1.510 (0.88 to 2.60)0.1371.430 (0.94 to 2.17)0.0941.030 (0.69 to 1.54)0.886SOFA Score1.210 (1.08 to 1.35)0.0011.250 (1.13 to 1.38)<0.001*1.150 (1.05 to 1.26)0.002HRs and 95% CIs are presented for mortality across three time intervals: <30 days, 30–365 days, and >365 days after ICU admission. Analyses were conducted with time-split Cox regression models because of violation of the proportional hazards assumption in the overall model. All included variables were clinically selected.p<0.05 was considered statistically significant and is indicated with an asterisk (*).APACHE, Acute Physiology and Chronic Health Evaluation; ICU, intensive care unit; SOFA, Sequential Organ Failure Assessment. Time-dependent Cox proportional hazards regression for mortality before and after propensity score matching HRs and 95% CIs are presented for mortality across three time intervals: <30 days, 30–365 days, and >365 days after ICU admission. Analyses were conducted with time-split Cox regression models because of violation of the proportional hazards assumption in the overall model. All included variables were clinically selected. p<0.05 was considered statistically significant and is indicated with an asterisk (*). APACHE, Acute Physiology and Chronic Health Evaluation; ICU, intensive care unit; SOFA, Sequential Organ Failure Assessment. Kaplan-Meier curve of mortality among patients after elective versus non-elective hip replacement operation. ICU, intensive care unit. In the propensity score-matched cohort, non-elective operation remained significantly associated with mortality across all intervals: we observed a 2.9-fold higher risk within 30 days (HR 2.87, 95% CI 1.56 to 5.30, p=0.001), a 3.0-fold higher risk from 30 to 365 days (HR 3.02, 95% CI 1.93 to 4.73, p<0.001), and a 1.8-fold higher risk beyond 1 year (HR 1.84, 95% CI 1.26 to 2.69, p=0.001). In this matched population, each point increase in APACHE III score was associated with a 4% increase in short-term and intermediate-term mortality, and a 2% increase in long-term mortality. The SOFA score remained independently associated with mortality, and each point increase corresponded to a 21–25% increase in risk across intervals. Age was no longer associated with short-term mortality but remained a significant predictor after 30 days. Male sex was not significantly associated with mortality in the matched analysis at any time point. ICU and hospital length of stay Non-elective operation was associated with longer ICU LOS than elective operation (table 4). In the unmatched cohort, the ICU LOS was 10.9% longer for patients in the non-elective group (LOS ratio 1.109, 95% CI 1.01 to 2.21, p=0.016), whereas hospital LOS did not differ significantly between groups (LOS ratio 1.043, 95% CI 0.95 to 1.15, p=0.379). After propensity score matching, non-elective operation remained independently associated with a 13.1% longer ICU LOS (LOS ratio 1.131, 95% CI 1.009 to 1.268, p=0.034), whereas no significant difference was observed in hospital LOS (LOS ratio 1.065, 95% CI 0.948 to 1.195, p=0.289). Table 4Log transformed linear regression of LOSTermICU LOS ratio (95% CI)ICU LOSP valueHospital LOS ratio (95% CI)Hospital LOSP valuePrematchingNon-elective operation1.109 (1.01 to 2.206)0.0161.043 (0.95 to 1.146)0.379Age0.99 (0.976 to 1.004)0.1520.992 (0.976 to 1.007)0.289APACHE III Score1.02 (0.996 to 1.04)0.2121.006 (1.002 to 1.009)0.008SOFA Score1.058 (1.036 to 1.081)<0.001*1.023 (0.999 to 1.047)0.063Male1.042 (0.970 to 1.119)0.2621.032 (0.952 to 1.118)0.445PostmatchingNon-elective operation1.131 (1.009 to 1.268)0.0341.065 (0.948 to 1.195)0.289Age0.975 (0.953 to 0.998)0.0310.975 (0.952 to 0.998)0.031APACHE III Score1.000 (0.995 to 1.005)0.9631.004 (0.999 to 1.010)0.096SOFA Score1.105 (1.067 to 1.144)<0.001*1.010 (0.974 to 1.046)0.594Male1.033 (0.917 to 1.166)0.5901.059 (0.938 to 1.198)0.350Values are expressed as back-transformed beta coefficients and corresponding 95%CIs, from the regression analysis with log-transformed LOS. Analyses were conducted separately for ICU and hospital LOSs. Variables were selected according to clinical relevance and were assessed for multicollinearity before inclusion. Non-elective operation refers to unplanned or urgent procedures.p<0.05 was considered statistically significant and is indicated with an asterisk (*).LOS, length of stay.APACHE, Acute Physiology and Chronic Health Evaluation; ICU, intensive care unit; SOFA, Sequential Organ Failure Assessment. Log transformed linear regression of LOS Values are expressed as back-transformed beta coefficients and corresponding 95%CIs, from the regression analysis with log-transformed LOS. Analyses were conducted separately for ICU and hospital LOSs. Variables were selected according to clinical relevance and were assessed for multicollinearity before inclusion. Non-elective operation refers to unplanned or urgent procedures. p<0.05 was considered statistically significant and is indicated with an asterisk (*). LOS, length of stay.APACHE, Acute Physiology and Chronic Health Evaluation; ICU, intensive care unit; SOFA, Sequential Organ Failure Assessment. Higher SOFA scores were consistently associated with longer ICU LOS in both the unmatched (LOS ratio 1.058 per point, p<0.001) and matched cohorts (LOS ratio 1.105, p<0.001). The APACHE III score showed a small but significant association with hospital LOS prematching (p=0.008), but not with ICU LOS in the matched cohort (p=0.096). Increasing age was associated with slightly shorter ICU and hospital LOS postmatching (LOS ratio 0.975 per year, p=0.031 for both). Sex was not significantly associated with LOS in any model. Discriminatory performance of the predictors As shown in figure 3 and online supplemental Table 3, the strongest predictors of very short term (7-day) mortality in the elective operation group were lactate (AUROC 0.812, 95% CI 0.607 to 1.000), low white blood cell count (0.773, 95% CI 0.567–0.980), high heart rate (0.752, 95% CI 0.630–0.874), and high potassium (0.733, 95% CI 0.579–0.886). At 30 days, the top-ranking variables were urea (0.684, 95% CI 0.604–0.764), APACHE III score (0.672, 95% CI 0.585–0.759), low creatinine (0.670, 95% CI 0.574–0.767), high creatinine (0.668, 95% CI 0.584–0.753), and lactate (0.663, 95% CI 0.552–0.774), although the overall discriminatory performance was poor. Time-dependent AUROC curve of top differential variables. APACHE, Acute Physiology and Chronic Health Evaluation; AUROC, area under the receiver operating characteristic curve; ICU, intensive care unit; SOFA, Sequential Organ Failure Assessment. In the non-elective group, 7-day mortality was best predicted by urine output (0.795, 0.711–0.880), SOFA score (0.756, 0.625–0.888), GCS score (0.754, 0.638–0.870), lactate (0.748, 0.563–0.934), and APACHE III score (0.735, 0.588–0.882). These variables remained the top predictors at 30 days (urine output (0.771, 0.693–0.848), APACHE III (0.730, 0.636–0.825), lactate (0.706, 0.570–0.842), GCS score (0.703, 0.618–0.787), and SOFA score (0.700, 0.609–0.792)). Other predictors had poor discriminatory ability across both timepoints. The discriminatory ability was higher at 7 days than at 30 days, and was consistently stronger in patients in the non-elective than the elective group. Discussion Key findings This binational cohort study in 1,578 nonagenarian and centenarian patients undergoing hip replacement operation indicated that non-elective procedures, compared with elective procedures, were associated with significantly poorer outcomes, including greater ICU, hospital, 30-day, and 365-day mortality. After propensity score matching, non-elective operation remained independently associated with a nearly threefold elevation in early and intermediate-term mortality, and a sustained excess risk beyond 1 year. Patients in the non-elective group, compared with the elective group, also had longer ICU stays, reflecting greater resource utilization. Time-dependent AUROC analysis identified lactate, urine output, SOFA score, APACHE III score, and GCS score as the most informative predictors of short-term mortality, particularly among patients in the non-elective group. The consistently higher discriminatory performance at 7 days than at 30 days underscored the prognostic relevance of acute physiological derangement. Relationship of the findings to the literature The findings of this large multicenter analysis make a meaningful contribution to the growing literature on surgical outcomes in the oldest populations, particularly in the context of surgical urgency. Previous studies have established an independent association between non-elective operation and elevated perioperative risk in older adults, a relationship partly attributable to the constrained window for physiological optimization, limited preoperative planning, and the acute physiological derangement often present at presentation.16,18 Consistent with these observations, Fariña-Castro et al19 reported markedly higher operative mortality among nonagenarians undergoing emergency operation compared with elective procedures (37.6% vs 4.8%), with emergency status independently associated with poorer 5-year survival (HR 1.64, 95% CI 1.05 to 2.57). Our results are consistent with these observations and extend them to specifically encompass nonagenarian and centenarian populations, whose representation in surgical outcome research remains limited despite their growing surgical volume. In line with findings from prior studies,20,22 we identified that routinely collected physiological and biochemical parameters—including lactate, urine output, SOFA score, APACHE III score, and GCS score—were strongly associated with early mortality in critically ill surgical patients. The prognostic validity of these markers has been well documented across diverse critical care cohorts. Notably, elevated serum lactate and oliguria are established indicators of global hypoperfusion and acute organ dysfunction, correlating with adverse clinical trajectories in diverse contexts.23 24 Our application of these predictors to a notably aged surgical cohort underscores their clinical pertinence and supports their incorporation into geriatric risk stratification tools. Moreover, our findings complement existing geriatric surgical guidelines advocating for early, multidisciplinary engagement, structured preoperative evaluation, and shared decision-making frameworks tailored to the goals and values of older patients.25 26 Although most of the existing literature has focused on octogenarian populations or elective procedures,7 27 our study offers novel insights into the distinct and sustained prognostic effects of non-elective operation on prognosis in nonagenarians and centenarians, thus reinforcing the need for targeted perioperative care and ICU triage protocols tailored to this expanding and clinically complex demographic. Clinical implications This large multicenter analysis provides compelling evidence that non-elective hip replacement operation in nonagenarian and centenarian patients is associated with significantly elevated short- and long-term mortality, prolonged ICU stays, and greater physiological derangement at presentation. In contrast to those in the elective group, patients in the non-elective group demonstrated markedly poorer outcomes, even after adjustment for baseline characteristics. Our findings emphasize that surgical urgency is a key determinant of prognosis, and underscore the need for systematic preoperative assessment and structured perioperative planning, particularly in emergency contexts, in which decision-making is time constrained. For practicing clinicians, this evidence provides an objective foundation that might support early, multidisciplinary discussions involving surgical, anesthetic, intensive care, geriatric, and palliative care teams. In non-elective settings, in which opportunities for shared decision-making are often limited, the ability to quantify early mortality risk with routinely available markers, such as lactate, urine output, SOFA score, APACHE III score, and GCS score, can guide clinical judgment, improve prognostic clarity, and facilitate transparent conversations with patients and families. Consequently, treatment intensity could be more deliberately aligned with patient values, non-beneficial ICU admissions could be avoided, and clearer advanced care planning could be achieved. For hospitals and health systems, our findings support the development of standardized, evidence-based pathways for the perioperative management of the oldest populations. Integrating these prognostic tools into electronic health records, triage protocols, and ICU admission frameworks might enhance risk-adjusted decision-making, optimize resource allocation, and promote more ethically sound, person-centered care. As the number of nonagenarian surgical candidates increases globally, this study provides timely and actionable data to inform care models that are both clinically effective and system-efficient. Strengths and limitations A key strength of this study is its large, binational cohort drawn from more than 200 ICUs across Australia and New Zealand, thus representing one of the most comprehensive datasets examining surgical outcomes in nonagenarian and centenarian patients to date. The inclusion of both elective and non-elective hip replacement cases enabled meaningful stratification by surgical urgency, an aspect rarely addressed in this age group. Robust statistical methods, including propensity score matching and time-dependent Cox regression, enhanced the internal validity of the findings by decreasing confounding and addressing violations of the proportional hazards assumption. Furthermore, the use of routinely collected clinical variables strengthened the study’s translational potential, because these markers are readily accessible in most care settings and can be integrated into standard perioperative workflows. However, several limitations should be acknowledged. First, as a retrospective observational study, the analysis was subject to residual confounding and cannot establish causal relationships. Despite matching and multivariable adjustment, unmeasured factors such as frailty, baseline functional status, and cognitive impairment may have influenced both surgical selection and outcomes. Second, detailed intraoperative and anesthetic data, including anesthetic technique, surgical approach, blood loss, and intraoperative complications, were unavailable, limiting the ability to account for procedural variability. Third, because the dataset included only patients admitted to the ICU, individuals managed entirely on the ward were not captured, potentially introducing selection bias toward higher-acuity cases and limiting generalizability, particularly for lower-risk elective procedures that may not require postoperative intensive care. In addition, the registry did not reliably distinguish between specific arthroplasty techniques and outcomes could not be stratified by operative type. Moreover, the registry records all-cause rather than cause-specific mortality, and therefore the underlying mode of death could not be determined. Finally, whereas mortality provides an objective and clinically relevant outcome, the absence of long-term functional, cognitive, and quality-of-life data restricts interpretation of recovery and overall patient-centered outcomes. Future directions Future research should aim to build on these findings by prospectively validating the identified clinical predictors of mortality in nonagenarian and centenarian surgical populations across diverse healthcare settings. Studies incorporating preoperative functional status, cognitive impairment, frailty phenotypes, and patient-reported outcomes would provide a more comprehensive understanding of surgical risk and recovery trajectories in this age group. Additionally, the utility of structured decision-support tools that integrate real-time physiological and biochemical data must be evaluated to inform perioperative triage, ICU admission criteria, and treatment limitation decisions. Furthermore, prospective data collection that includes both operative and non-operative patients would enable a more comprehensive evaluation of comparative outcomes and decision-making processes in this population. Qualitative research exploring patient and family perspectives on surgical decision-making in urgent contexts might also yield valuable insights to guide shared decision-making models that are both ethically grounded and clinically responsive. Finally, interventional studies are warranted to assess whether early multidisciplinary consultation, goal-concordant care planning, or geriatric co-management can mitigate the excess mortality and resource burden observed in non-elective cases. As the oldest populations continue to expand globally, generating high-quality evidence to guide equitable, person-centered surgical care in this cohort remains a critical priority. Conclusion In this large multicenter cohort of nonagenarian and centenarian patients undergoing hip replacement operation, non-elective procedures, compared with elective procedures, were associated with substantially elevated short- and long-term mortality, prolonged ICU stays, and greater physiological derangement at presentation. Routinely available clinical markers, such as lactate, urine output, SOFA score, APACHE III score, and GCS score, provided useful prognostic information regarding early mortality, particularly in the non-elective setting. These findings highlight the prognostic importance of surgical urgency in the oldest populations, and underscore the need for structured perioperative risk assessment, multidisciplinary care planning, and early goals-of-care discussions to support patient-centered decision-making and optimize outcomes in this growing population. Supplementary material Footnotes Funding: The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors. Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Data availability free text: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Ethics approval: Ethical approval for this study was obtained from the Alfred Hospital Human Research Ethics Committee (project No. 253/24). Given the study’s retrospective design and use of de-identified data, the requirement for individual informed consent was waived. Data analysis was initiated only after ethics approval was granted. The study was conducted in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. Data availability statement Data are available upon reasonable request. References Associated Data Supplementary Materials Data Availability Statement Data are available upon reasonable request.