Outcomes of nonagenarians undergoing hip fracture surgery: a multicentre observational study across Asia-Oceana

Abstract

Background

Nonagenarians with hip fracture represent a rapidly growing, highly vulnerable subgroup, yet comparative data across diverse health‑care systems in Asia‑Oceania are scarce. We aimed to describe mortality, postoperative complications, and perioperative care patterns in nonagenarians undergoing hip fracture surgery in Australia and Malaysia, and to explore associations between key perioperative factors and mortality.

Methods

We conducted a binational, retrospective, multicentre observational study of consecutive nonagenarians (aged ≥ 90 to < 100 years) who underwent surgery for unilateral femoral neck, intertrochanteric, or subtrochanteric fractures at four tertiary training hospitals (one in Australia, three in Malaysia) between January 2016 and January 2021. Patient demographics, comorbidities, frailty, surgical and anaesthetic details, and postoperative complications were extracted from electronic records; complications were graded using the Clavien–Dindo classification. Primary outcomes were in‑hospital, 30‑day, and 1‑year mortality; secondary outcomes were number and severity of postoperative complications, length of hospital stay, and 30‑day readmission. Statistical analyses focused on descriptive comparisons and cohort-stratified exploratory analyses to explore associations between perioperative factors and mortality.

Results

Among 510 patients, 394 (77.3%) were treated in Australia and 116 (22.7%) in Malaysia; median age was 92.0 years (interquartile range 91.0–94.0). Baseline profiles and perioperative care differed substantially: Malaysian patients more often had hypertension, diabetes, and chronic kidney disease, whereas Australian patients more frequently had congestive heart failure and dementia, and underwent emergency surgery within 48 h. In‑hospital and 30‑day mortality were similar between Australia and Malaysia (6.3% vs. 6.9% and 8.6% vs. 10.3%, respectively), but 1‑year mortality was higher in Malaysia (36.0% vs. 14.7%; P < 0.001). Overall, 86.0% of patients experienced at least one postoperative complication, with higher rates in Australia than Malaysia (98.3% vs. 49.1%), while severe complications (Clavien–Dindo grade ≥ III) were comparable (15.0% vs. 15.5%). Median length of stay was similar (8 days in both cohorts), but 30‑day readmission was more frequent in Malaysia (28.7% vs. 2.7%; P < 0.001). In cohort‑stratified models, factors associated with mortality differed by country and time point; no single perioperative factor was consistently associated with mortality across settings.

Conclusions

In this binational cohort of nonagenarians undergoing hip fracture surgery, short‑term mortality was comparable across four tertiary hospitals in Australia and Malaysia, whereas 1‑year mortality and early readmission were substantially higher in Malaysia. These findings suggest that system‑level factors, including delayed access to surgery and limited coordinated post‑acute care, may be important determinants of longer‑term outcomes in this very old population, and highlight the need for context‑specific strategies to optimise hip fracture care in resource‑diverse settings.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12877-026-07157-8.

Background

Femoral neck fractures represent a significant public health concern throughout the Asia-Oceanian region, where the incidence of hip fractures is consistently high and the burden on healthcare systems substantial [1]. Despite the frequency of these fractures and the well-recognized impact on morbidity and mortality, there has been limited formal exploration of outcomes between countries in this region. Recent systematic reviews indicate that the crude incidence of hip fractures among individuals aged 50 years or older ranges from 89 to 341 per 100,000 people, with the highest rates observed in Australia, Taiwan, and Japan. This variation underscores both the widespread prevalence of hip fractures and the need for more comprehensive international descriptive data to inform policy and clinical practice.

In Australia, approximately 16,500 hospitalizations per year for osteoporotic hip fractures occur among Australians > 40 years old [2]. Patients with hip fractures are usually older and frail, have multiple medical comorbidities, and are at high risk of perioperative complications [3]. Similarly, in Malaysia, the overall incidence of fractures of the neck of the femur is 90 per 100,000 population among individuals 50 years or older [4]. In both countries, hip fracture rates are elevated among women (women, 140 per 100,000 population vs. men, 65 per 100,000 population) and older people (≥ 75 years, 495 per 100,000 population vs. 50–54 years, 10 per 100,000 population) [2, 4]. These patterns highlight shared demographic and clinical risk factors across the region and emphasize the importance of targeted interventions for vulnerable populations.

An Asian Federation of Osteoporosis Societies report has projected a 3.5-fold increase in the number of hip fractures, from 6,000 to almost 21,000 fractures occurring annually by 2050, which are predicted to cost more than USD125 million (MYR540 million) in annual healthcare expenditures [5]. Malaysia’s rate of increase is expected to be among the highest in countries in the Asia‑Pacific region [5, 6]. Reports in Australia have indicated mortality rates of 22%–29% within 1 year after a fractured femoral neck, and 30-day mortality rates of 8%–10% [7–9]. For example, locally available data have demonstrated a 1-year mortality rate of 24.9% at a Queensland regional centre [9]. Mortality at 1 year is 10%–20% above that expected for age- and sex-matched groups, and most of this excess mortality occurs within the first 4 months after the fracture. In Malaysia, the mortality rates after hip fracture surgery are similar: at a teaching hospital in Malaysia, 20.1% 1-year mortality after hip fracture surgery has been reported [10]. Advanced age, dependency on activities of daily living and prolonged hospitalization stays are independent predictors of mortality.

To date, there is limited literature describing outcomes of nonagenarians undergoing hip fracture surgery between different healthcare systems, including those in Australia and Malaysia. Studies in other countries have reported mortality rates of 24%–51% at 1 year after hip fracture in nonagenarians [11–22]. In a recent study in 11,184 nonagenarians in Taiwan who underwent surgery for hip fracture, the mortality rates were 29.5% at 1 year, 45.0% at 2 years and 78.1% at 5 years [19]. The cumulative incidence rates of reoperation were 7.3% at 1 year, 9.2% at 2 years and 11.6% at 5 years, whereas the readmission rates were 18.9% at 1 month and 24.1% at 3 months. Significant risk factors for death included age, male sex, trochanteric fracture and a high Charlson comorbidity index (CCI), whereas those for reoperation were age, cervical fracture and a high CCI. However, these outcomes may vary across different healthcare systems, especially within the Asia-Oceanian region, and remains incompletely understood.

We conducted an observational study on the outcomes of nonagenarians admitted with hip fractures across two Asia-Oceanian teaching hospitals. This study examined in‑hospital, 30‑day, and 1‑year mortality, as well as length of hospital stay, major postoperative complications, and readmissions.

Methods

Study design

This cross-continent, retrospective, multicentre observational study was conducted at four hospitals: one hospital in Australia, Austin Hospital in Melbourne, and three hospitals in Malaysia: University Malaya Medical Centre (UMMC), Universiti Kebangsaan Malaysia, and Hospital Raja Permaisuri Bainun. All participating sites were tertiary training hospitals providing specialist care for orthopedic, anesthetic, and perioperative care for hip fracture patients. The Austin Health Human Research Ethics Committee and the UMMC Research Ethics Committee approved this study (approval numbers: HREC Audit/21/Austin/30 and NMRR-ID24-02302-3EZ) and, given the retrospective study design, waived the requirement for participant consent. The study was done in accordance with the Ethical Principles for Medical Research Involving Human Subjects, outlined in the Helsinki Declaration of 1975 (revised 2013).

Inclusion and exclusion criteria

All nonagenarians (aged ≥ 90 to < 100 years) presenting with a unilateral hip fracture, including femoral neck, intertrochanteric or subtrochanteric femoral fractures, who underwent surgical intervention at one of four participating hospitals between January 2016 and January 2021, were screened for inclusion. All acute and referred cases were included in this study. Surgical procedures were categorized as either arthroplasty procedures, such as bipolar hemiarthroplasty or total hip arthroplasty, or internal fixation with implants such as dynamic hip screws or intramedullary nails. Patients were excluded if they were treated non-operatively, sustained iatrogenic or periprosthetic fractures, had ipsilateral femoral shaft fractures or did not meet the age criteria.

Objectives

The primary objective of this study was to describe perioperative mortality rates in nonagenarians undergoing hip fracture surgery at three defined time points: in-hospital, 30 days post-operatively, and 1 year post-operatively. Secondary objectives were to characterize the number and severity of major post-operative complications, the length of hospital stay, and the rate of readmission within 30 days of discharge. Finally, we explored the associations between selected demographic, patient, and procedural variables and mortality.

Data collection and definitions

Data extracted from the electronic medical and anesthetic records at all four centres included patient demographics, medical comorbidities, surgical interventions, anesthesia techniques and postoperative complications. CCI scores were calculated for each patient [23]. Frailty was assessed with the Canadian Study of Health and Aging Clinical Frailty Scale, with a score ≥ 5 indicating frailty [24]. Time to surgery was calculated from the time of hospital presentation to the time of anesthesia initiation. This interval was used to determine whether surgery occurred within or after 48 h of admission.

Complications were defined as any deviation from the normal postoperative course, including return to the operating room, defined as any subsequent surgery performed during the index admission as a consequence of the original hip fracture management. Complications were classified by severity with the Clavien–Dindo classification (CDC) system, a validated framework for categorizing surgical complications [25]. Mortality was recorded at three timepoints: in-hospital, at 30 days and at 1 year after surgery. Deaths occurring after discharge were verified through hospital records and follow-up with primary care physicians. Length of hospital stay was recorded as the total number of days from hospital admission to discharge. Readmission was defined as any unplanned admission within 30 days of discharge after the index admission.

Statistical analysis

Statistical analysis was performed in R (ver. 4.4.2, R Foundation for Statistical Computing, Vienna, Austria, 2024) with compatible packages. The normality of continuous variables was evaluated with Q-Q plots, and outliers were identified according to predefined criteria: values below the 1st quartile − 1.5× IQR or above the 3rd quartile + 1.5× IQR. Any identified outliers were manually reconciled with the original medical records. Variables with missing values were identified, and missingness patterns were evaluated. The highest rate of missing data was observed for operation type (10.4%), followed by postoperative complication number and grade (9.2%), body mass index (BMI, 6.9%), length of hospital stay (6.5%), discharge destination (6.5%), and readmission within 30 days (6.5%). Given the low overall missing-data proportion, primary analyses were conducted using complete cases. No imputation was performed for the main results.

Between-cohort comparisons were conducted using the t-test or Wilcoxon rank-sum test for continuous variables and the χ² test or Fisher’s exact test for categorical variables, depending on data distribution and cell counts. Baseline differences between the Australian and Malaysian cohorts were assessed descriptively using standardized mean differences (SMDs), with values interpreted as indicators of distributional imbalance rather than as formal hypothesis tests.

Postoperative mortality was assessed at three predefined time points: in-hospital, 30-day, and 1-year mortality. Unadjusted mortality rates with corresponding P values were reported as the main results. In addition, cohort-stratified multivariable regression analyses were performed to explore factors associated with mortality within each country and at each time point. These analyses were considered exploratory and were used to describe potential heterogeneity in associations across cohorts rather than to infer causal effects.

Exploratory adjusted and weighted analyses, including propensity score–based overlap weighting and survey-weighted regression models [26–29], were prespecified as supplementary analyses to evaluate the robustness of observed crude associations. The methods and results of these analyses are reported in detail in the Supplementary Material. All statistical tests were two-sided, and a P value <0.05 was considered statistically significant.

Results

Details of the cohort

The analysis included 510 patients: 394 (77.3%) from Australia and 116 (22.7%) from Malaysia. The study flow diagram is presented in Fig. 1. Detailed demographic data are presented in Table 1. There were substantial differences in baseline comorbidity patterns and pre-operative injury characteristics between the Australian and Malaysian cohorts. The overall median age of the patients was 92.0 [91.0–94.0] years, and the Australian cohort tended to occupy the higher end of the age distribution (standardised mean difference [SMD] = 0.275). Baseline comorbidity differed substantially between cohorts. Hypertension, diabetes mellitus, and chronic renal disease were more prevalent in the Malaysian cohort, with moderate to large SMDs (0.105–0.534). In contrast, the conditions commonly associated with functional vulnerability, such as congestive heart failure and dementia, were more frequently observed in the Australian cohort (SMD = 0.219 and 0.230). This heterogeneity was also reflected in clinical scoring systems. The distribution of Charlson comorbidity indices was relatively balanced between the two cohorts (SMD = 0.109), the Australian cohort presented a moderate shift toward a higher clinical frailty score (SMD = 0.439).

Fig. 1.

STROBE flow diagram

Table 1.

Demographic and preoperative characteristics

Variable		Overall (N = 510)	By nationality
			Australia (N = 394)	Malaysia (N = 116)	P value	Standardized difference
Demographic characteristics and clinical history
Sex (male)		128 (25.1%)	107 (27.2%)	21 (18.1%)	0.064	0.091
Age (years)		92.0 [91.0–94.0]	92.5 [91.0–94.0]	92.0 [91.0–93.0]	0.025	0.275
Weight (kg)		56.0 [50.0–65.0]	58.0 [50.9–66.7]	50.0 [45.0–60.0]	< 0.001*	0.746
Height (cm)		160.0 [152.0–165.0]	160.0 [152.8–168.0]	155.0 [150.0–160.0]	< 0.001*	0.636
Body mass index (kg/m2)		22.2 [19.8–25.4]	22.7 [19.9–25.9]	20.9 [19.5–23.4]	< 0.001*	0.416
Clinical frailty scale > 6		304 (59.7%)	274 (69.7%)	30 (25.9%)	< 0.001*	0.439
Residence†	Home	467 (91.6%)	366 (92.9%)	101 (87.1%)	0.073	-
	Nursing home	43 (8.4%)	28 (7.1%)	15 (12.9%)		0.058
Fall history	First fall	486 (95.3%)	389 (98.7%)	97 (83.6%)	< 0.001*	-
	Recurrent	24 (4.7%)	5 (1.3%)	19 (16.4%)		0.151
Preoperative medical history
Hypertension		137 (26.9%)	58 (14.7%)	79 (68.1%)	< 0.001*	0.534
Diabetes mellitus		95 (18.6%)	64 (16.2%)	31 (26.7%)	0.016*	0.105
Chronic renal disease		151 (29.6%)	99 (25.1%)	52 (44.8%)	< 0.001*	0.197
Congestive heart failure		104 (20.4%)	101 (25.6%)	3 (2.6%)	< 0.001*	0.230
Chronic obstructive lung disease		43 (8.4%)	39 (9.9%)	4 (3.4%)	0.045*	0.065
Cerebrovascular event/transient ischemic attack		82 (16.1%)	69 (17.5%)	13 (11.2%)	0.139	0.063
Dementia		170 (33.3%)	151 (38.3%)	19 (16.4%)	< 0.001*	0.219
Charlson comorbidity index	Moderate	229 (44.9%)	136 (34.5%)	93 (80.2%)	< 0.001*
	High	281 (55.1%)	258 (65.5%)	23 (19.8%)		0.109
Preoperative laboratory findings
Hemoglobin concentration (g/dL)
Normal: ≥13		124 (24.3%)	108 (27.4%)	16 (13.8%)	0.002*	0.136
Mild anemia: 10–12.9		277 (54.3%)	213 (54.1%)	64 (55.2%)		0.011
Moderate anemia: 8–9.9		95 (18.6%)	62 (15.7%)	33 (28.4%)		0.127
Severe anemia: 6.5–7.9		14 (2.7%)	11 (2.8%)	3 (2.6%)		0.002
Glomerular filtration rate (mL/min/1.73 m²)
Stage 1: >90		256 (50.2%)	248 (62.9%)	8 (6.9%)	< 0.001*	0.560
Stage 2: 60–89		117 (22.9%)	68 (17.3%)	49 (42.2%)		0.250
Stage 3a: 45–59		73 (14.3%)	45 (11.4%)	28 (24.1%)		0.127
Stage 3b: 30–44		49 (9.6%)	25 (6.3%)	24 (20.7%)		0.143
Stage 4: 15–29		12 (2.4%)	6 (1.5%)	6 (5.2%)		0.036
Stage 5: <15		3 (0.6%)	2 (0.5%)	1 (0.9%)		0.004
Albumin (g/L)		31.9 ± 5.1	32.1 ± 5.0	31.4 ± 5.5	0.249	0.131

Data are presented as mean ± SD, median [IQR], or number (percentile). *: P < 0.05

Perioperative management also differed substantially between cohorts (Table 2). A greater proportion of surgeries were performed as emergency procedures and within 48 h of admission in the Australian cohort, and weekend procedures were also more commonly observed. In contrast, operative duration tended to be shorter in the Malaysian cohort. The magnitude of these distributional differences was quantified by moderate to large SMDs for emergency surgery (0.387), surgery within 48 h (0.758), weekend surgery (0.302), and operative duration (1.106).

Table 2.

Surgery and anesthesia

Variable		Overall (N = 510)	By nationality
			Australia (N = 394)	Malaysia (N = 116)	P value	Standardized difference
Emergency operation		236 (46.3%)	217 (55.1%)	19 (16.4%)	< 0.001*	0.387
Surgery within 48 h of admission		334 (65.5%)	326 (82.7%)	8 (6.9%)	< 0.001*	0.758
Surgery performed on weekend		132 (25.9%)	129 (32.7%)	3 (2.6%)	< 0.001*	0.302
Operation time (min)		115.6 ± 48.2	126.0 ± 46.7	80.5 ± 34.6	< 0.001*	1.106
Mode of anesthesia†	General anesthesia with or without a block	318 (62.4%)	290 (73.6%)	28 (24.1%)	< 0.001*	-
	Neuraxial or block only	192 (37.6%)	104 (26.4%)	88 (75.9%)		0.495
Type of operation	Intramedullary nail	77 (16.8%)	1 (0.3%)	76 (65.5%)		0.652
	Dynamic hip screw	93 (20.4%)	88 (25.8%)	5 (4.3%)		0.215
	Arthroplasty	287 (62.8%)	252 (73.9%)	35 (30.2%)		0.437
Intraoperative hypotension episode		113 (22.2%)	37 (9.4%)	76 (65.5%)	< 0.001*	0.561
Number of patients receiving allogeneic RBC transfusion		37 (7.3%)	28 (7.1%)	9 (7.8%)	0.973	0.007
Allogeneic RBC transfusion (units)		1 (1–1) [1:3]	1 (1–1) [1:3]	1 (1–1) [1:1]	0.2501	0.021

Data are presented as mean ± SD, median [IQR], or number (percentile). *: P < 0.05

Anesthetic and surgical approaches also demonstrated heterogeneity in their distribution. General anesthesia was more frequently used in the Australian cohort, whereas neuraxial or regional techniques predominated in the Malaysian cohort. This distributional difference was reflected by a large standardized mean difference for anesthetic technique (SMD = 0.495). Episodes of intraoperative hypotension were more frequently observed in the Malaysian cohort, and this distributional difference was reflected by a large standardized mean difference (SMD = 0.561). Surgical procedures also varied across cohorts, with arthroplasty predominating in Australia and internal fixation techniques more frequently used in Malaysia, reflected in moderate standardized mean differences (SMDs approximately 0.21 to 0.44).

Primary outcome: postoperative mortality

In-hospital and 30-day mortality rates were comparable between cohorts. In-hospital mortality was 6.3% in the Australian cohort and 6.9% in the Malaysian cohort (P > 0.99), while 30-day mortality was 8.6% and 10.3%, respectively (P = 0.702). However, 1-year mortality was significantly higher in the Malaysian cohort compared with the Australian cohort (36.0% vs. 14.7%, P < 0.001, Fig. 2). In cohort-stratified multivariate analysis, the factors associated with postoperative mortality were diverse across time points and between cohorts (Table 3). For in-hospital mortality, preoperative albumin and male sex were independently associated with mortality in the Malaysian cohort. At 30-day mortality, male sex was independently associated with mortality in the Australian cohort, preoperative albumin remained significantly associated with those in the Malaysian cohort. For 1-year mortality, arthroplasty surgery was independently associated with mortality in the Australian cohort, while severe frailty was included in the final model for the Malaysian cohort but did not reach statistical significance. Overall, the set of variables retained in the final models and their statistical associations differed by country and by mortality time point. In exploratory adjusted and weighted analyses, the difference in observed mortality was attenuated and did not reach statistical significance (Supplementary Material Table 1).

Fig. 2.

Postoperative Mortality at In-Hospital, 30-Day, and 1-Year Time Points in Nonagenarians Undergoing Hip Fracture Surgery in Australia and Malaysia. *: P<0.05

Table 3.

Factors associated with the postoperative mortality: multivariate nationality-stratified firth regression analysis

Variable	In-hospital mortality				30-day mortality				1-year mortality
	Australia cohort		Malaysia cohort		Australia cohort		Malaysia cohort		Australia cohort		Malaysia cohort
	Adjusted OR (95% CI)	P value	Adjusted OR (95% CI)	P value	Adjusted OR (95% CI)	P value	Adjusted OR (95% CI)	P value	Adjusted OR (95% CI)	P value	Adjusted OR (95% CI)	P value
Preoperative albumin	0.93 (0.85–1.01)	0.099	0.85 (0.71–0.99)	0.042*	-	-	0.87 (0.76–0.99)	0.033*	0.94 (0.87–1.01)	0.090	0.94 (0.86–1.02)	0.159
Sex (Male)	-	-	6.58 (1.38–39.21)	0.018*	2.31 (1.12–4.67)	0.023*	3.38 (0.87–13.11)	0.078	-	-	1.97 (0.72–5.50)	0.187
Charlson comorbidity index	-	-	1.44 (0.73–2.66)	0.270	-	-	-	-	-	-	-	-
Intraoperative transfusion	-	-	2.18 (0.17–17.51)	0.508	-	-	-	-	-	-	2.79 (0.66–13.32)	0.162
Clinical frailty score, 6 or more (severe)	-	-	-	-	-	-	-	-	-	-	2.35 (0.94–6.00)	0.069
Surgery: Arthroplasty	-	-	-	-	-	-	-	-	3.14 (1.21–10.18)	0.017*	-	-
Preoperative anemia	-	-	-	-	-	-	-	-	1.17 (0.50–3.03)	0.727	-	-
Intraoperative hypotension	-	-	-	-	-	-	-	-	2.93 (0.89–8.80)	0.075	-	-

Variables included as a rule of P < 0.100 from the univariate analysis (Supplementary Material)

Postoperative complications and length of hospital stay

Overall, 86.0% of patients experienced postoperative complications: 98.3% in the Australian cohort and 49.1% in the Malaysian cohort (P < 0.001) (see Table 4). The median number of postoperative complications was also higher in the Australian cohort than in the Malaysian cohort (3 [IQR, 2–4] vs. 0 [IQR, 0–2], P < 0.001). Severe complications (CDC grade III or greater) were comparable between cohorts (15.0% in Australian and 15.5% in Malaysian cohorts; P > 0.99). Similar patterns were observed in explorative adjusted and weighted analyses (Supplementary Material).

Table 4.

Postoperative complications and outcomes

Variable		Overall (N = 510)	By nationality
			Australia (N = 394)	Malaysia (N = 116)	P value
Postoperative complications (type)
Hypotension episode		32 (6.3%)	30 (7.6%)	2 (1.7%)	0.037*
Vasopressor support		29 (5.7%)	29 (7.4%)	0 (0.0%)	0.005*
ICU admission		27 (5.3%)	24 (6.1%)	3 (2.6%)	0.211
Return to operating theatre		18 (3.5%)	17 (4.3%)	1 (0.9%)	0.140
Cerebrovascular event		5 (1.1%)	2 (0.6%)	3 (2.6%)	0.196
Delirium		149 (32.2%)	136 (39.2%)	13 (11.2%)	< 0.001*
Myocardial infarction or heart failure		72 (15.6%)	62 (17.9%)	10 (8.6%)	0.026*
Pneumonia		78 (16.8%)	58 (16.7%)	20 (17.2%)	> 0.99
Surgical site infection		13 (2.8%)	4 (1.2%)	9 (7.8%)	0.001*
Acute renal insufficiency		86 (18.6%)	65 (18.7%)	21 (18.1%)	0.99
Deep vein thrombosis		5 (1.1%)	5 (1.4%)	0 (0.0%)	0.435
Pulmonary embolism		8 (1.7%)	6 (1.7%)	2 (1.7%)	> 0.99
Pressure injury		4 (0.9%)	0 (0.0%)	4 (3.4%)	0.004*
Atrial fibrillation		48 (10.4%)	44 (12.7%)	4 (3.4%)	0.008*
Urinary tract infection		55 (11.9%)	42 (12.1%)	13 (11.2%)	0.926
Other complications		8 (1.7%)	1 (0.3%)	7 (6.0%)	< 0.001*
Postoperative complications
Any complication		398 (86.0%)	341 (98.3%)	57 (49.1%)	< 0.001*
Number of complications		2 (1–4) [0:12]	3 (2–4) [0:12]	0 (0–2) [0:5]	< 0.001*
Severity of complications (CDC)	No complication, CDC I, II	393 (84.9%)	295 (85.0%)	98 (84.5%)	> 0.99
	CDC III, IV, V	70 (15.1%)	52 (15.0%)	18 (15.5%)
Postoperative Outcomes
Length of hospital stay (days)		8 (6–11) [2:46]	8 (6–11) [2:36]	8 (6–11) [2:46]	0.538
Discharge destination	Own home	411 (86.2%)	331 (89.7%)	80 (74.1%)	< 0.001*
	Nursing home or rehabilitation centre	66 (13.8%)	38 (10.3%)	28 (25.9%)
Readmission within 30 days		41 (8.6%)	10 (2.7%)	31 (28.7%)	< 0.001*

Data are presented as mean ± SD, median [IQR], or number (percentile). *: P < 0.05

Specific complications that were significantly higher in the Australian cohort included delirium and atrial fibrillation. Length of hospital stay was comparable between the cohorts, with a median of 8 days. Readmission within 30 days was more common in the Malaysian cohort than in the Australian cohort (28.7% vs. 2.7%, P < 0.001). However, these differences were attenuated and did not remain statistically significant in exploratory adjusted and weighted analyses (Supplementary Material).

Discussion

Key findings

This binational observational cohort study describes postoperative outcomes of nonagenarians undergoing hip fracture surgery in Malaysia and Australia and reveals substantial heterogeneity in baseline characteristics, perioperative management, and postoperative course between the two cohorts. The results demonstrated comparable in-hospital and 30-day mortality rates, whereas 1-year mortality was higher in the Malaysian cohort. The factors associated with postoperative mortality differed across time points and between cohorts, with no single variable consistently associated with mortality in both cohorts. In cohort-stratified exploratory analyses, surgery after 48 h, lower preoperative albumin, and severe frailty were variably associated with short- and long-term mortality; however, the patterns differed across countries. Postoperative complications were more frequently observed in the Australian cohort, although the incidence of severe complications was similar, and the length of hospital stay did not differ between cohorts. Thirty‑day readmission was more common in Malaysian patients than in Australian patients. Exploratory, adjusted, and weighted analyses attenuated several of the observed mortality differences, suggesting that these findings should be interpreted cautiously and in the context of marked differences in patient pool, perioperative management, and post-discharge care across healthcare systems. Together, these results underscore the complexity of cross-national comparisons in nonagenarian surgical populations and emphasize the need for context-specific interpretation rather than direct causal inference.

Relationship to the literature

Recent research has indicated variability in mortality rates among nonagenarians undergoing hip fracture surgery, influenced by factors including surgical timing, pre-existing comorbidities and post-operative care [7, 30, 31]. In our cohort, the in‑hospital, 30‑day, and 1‑year mortality rates in both the Australian and Malaysian groups were within the published range. However, strikingly, the unadjusted 1-year mortality rate in the Malaysian cohort was approximately three times higher than that in the Australian cohort, despite similar in-hospital and 30-day mortality rates. This discrepancy highlights the potential roles of factors beyond the immediate postoperative period in determining longer-term outcomes. In-hospital and 30-day mortality are typically associated with surgical or anesthetic complications and limited preoperative physiological reserve [30, 32, 33], whereas longer-term mortality is more often shaped by system-level factors such as delays to surgery and limited access to coordinated post-acute care [34, 35].

The effect of delayed surgery, defined as operative interventions more than 48 h after admission, differed significantly between the national cohorts. More than 80% of the Australian cohort underwent surgery within 48 h of admission, and 32.7% were performed during the weekend. In contrast, fewer than 10% of the Malaysian cohort received surgery within 48 h, and weekend surgery was also uncommon. Prior studies have identified significant barriers to timely surgical access in Malaysia, including limited operating theatre availability, financial constraints, and staffing shortages, particularly during weekends and after-hours periods [36, 37]. Additional barriers encompass access to preoperative investigations such as echocardiography and pulmonary function testing. However, emerging evidence questions the necessity of these investigations in urgent surgical settings, as they may not meaningfully alter management decisions [36, 38]. At the three Malaysian centres in our study, these barriers remain, with a reported average surgical delay of 5 days—well beyond international benchmarks [36, 38]. In contrast, Australian health systems appear to be better equipped to facilitate timely surgical care. Weinberg et al. reported a mean time to surgery of 34–36 h in 537 nonagenarian patients [31], which may explain why delays beyond 48 h were not associated with excess early mortality in our Australian cohort. Although exact timings were not analysed, most delays in Australia likely remained close to the 48‑hour threshold, unlike in Malaysia.

In cohort-stratified analyses, delayed surgery was not independently associated with in-hospital, 30-day, or 1-year mortality in either cohort. The direction and magnitude of the estimated ORs were diverse across time points and cohorts, indicating considerable uncertainty. Although delayed surgery had little effect on 1-year mortality, its association with early mortality underscores the importance of timely operative intervention in nonagenarians with hip fractures. Several studies have shown that delays beyond 48 h are associated with increased 30-day mortality [39, 40]. These findings are further supported by international guidelines and high-quality evidence from systematic reviews, meta-analyses, and large database registries [38, 41, 42]. Thus, our findings should be interpreted as complementary to existing evidence, highlighting that the delayed surgery effect may not be uniform across age groups, or may dependent on clinical context and healthcare systems.

Weekend surgery was associated with lower early mortality in both cohorts, although these associations were not statistically significant. While some studies have described a so-called ‘weekend effect’, linking weekend operations to increased complications and mortality due to factors like fatigue or limited staffing [43, 44], others, including large-scale analyses, have challenged its significance [45, 46]. A recent meta-analysis by Kim et al. found no increase in mortality associated with after-hours surgery and advocated for prioritising hip fracture operations within 48 h, regardless of timing [47]. This aligned with our finding: all weekend surgeries were emergencies, with 77.8% performed within 48 h. These results suggest that early surgery, rather than the day of operation, is more relevant to the observed association with mortality.

We observed notable differences in the prevalence of specific comorbidities across cohorts: Malaysian patients had higher rates of hypertension and chronic kidney disease, whereas Australian patients more often had congestive heart failure. The literature remains divided on the impact of comorbidities in this age group. Some studies have reported strong associations with mortality [32, 48–51]. For instance, Porter et al. found increased mortality in hip fracture patients with chronic kidney disease [48], whereas others report weaker or inconsistent effects [14, 52]. Thus, no definitive conclusions can be drawn from our study regarding comorbidity-related risk.

Surgical and anesthetic approaches also varied significantly between cohorts. Neither anesthesia type nor surgical approach was independently associated with early mortality in our analysis. This contrasts with prior studies suggesting neuraxial anesthesia improves survival in older adults [30, 40, 53]. Chia et al. also found general anesthesia predicted mortality in Australian patients, though this may reflect selection bias, as sicker patients were more likely to receive it [7].

Our analyses suggested that the impact of arthroplasty on mortality differed by country, appearing harmful in the Australian cohort at 1-year mortality. This aligns with prior literature showing mixed results regarding surgical approach in nonagenarians, with some studies reporting no significant mortality difference between arthroplasty and internal fixation [19, 54, 55], while others, such as Ovidiu et al., identified hemiarthroplasty as a mortality risk factor [56]. The divergent patterns may reflect selection bias or context-specific differences in surgical practice and postoperative care.

Intraoperative hypotension was more frequent in Malaysian patients but was not independently associated with the mortality. Although hypotension is linked to higher postoperative mortality in older adults [57], its lack of effects here may reflect variations in anesthetic protocols or perioperative monitoring practices and warrants further investigation.

Arthroplasty was associated with higher 1‑year mortality in the Australian cohort but not in the Malaysian cohort. Arthroplasty was the predominant surgery in Australia (62.8%), compared with less than one-third in Malaysia; this finding may reflect the differences in surgical practice patterns between cohorts. In Australia, established orthogeriatric pathways, early mobilization, rehabilitation protocols, and structured follow-up may support better long-term survival [58]. Conversely, Malaysia’s post-acute care system faces challenges, including limited access to rehabilitation and healthcare variability [36, 59–61]. Studies by Chiang et al. and Backman et al. have shown that structured post-acute care is crucial in decreasing mortality in older adults after hip fracture [62, 63]. Therefore, future research should investigate the specific effects of various post-acute care elements on nonagenarian mortality to better inform system-level change, improve resource allocation, and ultimately enhance outcomes for nonagenarians in diverse healthcare settings.

Study implications

The binational cohort study’s findings of comparable short-term mortality outcomes for nonagenarian hip fracture surgery between the Malaysia and Australia cohorts carries significant clinical implications. First, it demonstrates that tertiary teaching hospitals in economically diverse Asia-Oceanian settings can achieve equivalent short-term survival benchmarks when managing this high-risk population. However, the pronounced divergence in perioperative mortality predictors necessitates urgent health system interventions. In particular, surgical delay as well as the choice of operative strategy appear to effect on mortality risk in nonagenarian patients. From an international perspective, our findings highlight the need for context‑specific approaches to the care of nonagenarians with hip fracture, with timely surgical access and careful operative decision‑making, rather than direct replication of health‑care system models across countries.

Strengths and limitations

This binational study provides new data on perioperative influences on mortality in nonagenarians undergoing hip fracture surgery, comparing patients treated at four hospitals in Malaysia and Australia. As nonagenarians are a vulnerable and rapidly expanding demographic often underrepresented in surgical outcome studies, this study addresses a key evidence gap. Strengths include well‑characterized cohorts from distinct health‑care settings, detailed perioperative data, and a cautious analytic approach that prioritized descriptive and exploratory analyses to avoid overinterpretation of causal effects.

Several limitations should be acknowledged. The retrospective design limited data quality and introduced the possibility of unmeasured confounders, such as socioeconomic status, access to rehabilitation and pre-fracture functional capacity. The use of single-centre data from Australia limits generalizability to broader national practices. Although this study includes one of the largest cohorts of nonagenarians undergoing hip fracture surgery, the relatively small sample size in some strata— particularly in cohort‑stratified and exploratory analyses, which may have limited the precision of effect estimates and the ability to detect robust associations. A further limitation is the lack of information on post-discharge care, which likely contributed to the observed 1-year mortality differences. Finally, this study did not assess functional outcomes, which are particularly important in the nonagenarian population. Future studies should consider incorporating validated measures such as the World Health Organization Disability Assessment Schedule (WHODAS) to better capture patient-centred outcomes.

Conclusions

In conclusion, this binational observational study of nonagenarian hip fracture surgery outcomes across four Asia-Oceanian tertiary teaching hospitals revealed comparable short-term mortality between countries, whereas 1-year mortality and postoperative outcomes differed substantially. The factors associated with postoperative mortality varied by time point and cohort; no single preoperative variable consistently increased mortality risk across the two cohorts. These findings highlight the context-dependent nature of commonly used clinical thresholds in nonagenarian patients and emphasize the limitations of a uniform approach across heterogeneous healthcare systems. The markedly higher 1‑year mortality and 30‑day readmission rates in Malaysia suggest the influence of health‑care system‑level factors, particularly related to access to timely surgery and coordinated post-acute care. Although these factors were not directly measured, their potential impact underscores the importance of system-level reform. For international practice, our findings underscore that standardising access to prompt surgery and seven‑day specialist services, rather than mirroring entire health‑system structures, is the most actionable strategy for improving outcomes in nonagenarians globally. These findings also emphasize the need to further investigate health system factors, such as access to rehabilitation, infrastructure, and continuity of post-acute care. Optimizing hip fracture care in nonagenarians is an important and achievable step toward improving survival and quality of life in the ageing population.

Abbreviations

BMI

Body mass index

CI

Confidence intervals

CCI

Charlson comorbidity index

CDC

Clavien-Dindo classification

HREC

Health Human Research Ethics Committee

RRR

Elative risk

UMMC

University Malaya Medical Centre

Authors’ contributions

LW and IIS were responsible for conceptualization, supervision, data curation, investigation, methodology, project administration, validation, visualization, writing—original draft, and writing—review &amp; editing. KL analyzed and interpreted the patient data and was responsible for visualization, writing—original draft, and writing—review &amp; editing. SS, SNNM, NLW, AH, BW, JAS, NR were responsible for data curation, investigation, visualization, writing—original draft, and writing—review &amp; editing. All authors read and approved the final manuscript.

Funding

None.

Data availability

The datasets generated and/or analysed during the current study are not publicly available as patients have not consented to their data being shared; but all data are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The Austin Health Human Research Ethics and the University Malaya Medical Centre Research Ethics Committees approved this study (approval numbers: HREC Audit/21/Austin/30 and MREC 20245-13627).

Consent for publication

Given the retrospective study design, the ethics committee waived the requirement for participant consent.

Competing interests

The authors declare no competing interests.