Complication Rates and Survival of Nonagenarians after Hip Hemiarthroplasty versus Proximal Femoral Nail Antirotation for Intertrochanteric Fractures: A 15‐Year Retrospective Cohort Study of 113 Cases

Abstract

Objective

Intertrochanteric fracture is a very common but serious type of hip fracture in nonagenarians. The surgical treatment remains a significant challenge for orthopedists. The objective of this study was to investigate postoperative complications and survival outcomes compared between bipolar hemiarthroplasty (HA) and proximal femoral nail anti‐rotation (PFNA) in nonagenarians with intertrochanteric fractures, and to evaluate the efficacy and safety of the two surgical procedures in this patient population.

Methods

A total of 113 consecutive nonagenarians who underwent bipolar HA or PFNA for the treatment of intertrochanteric fractures from January 2006 to August 2021 were retrospectively studied in the current paper. There were 34 males and 79 females, with a mean age of 92.2 years (range 90–101 years) at the time of operation. The average duration of follow‐up was 29.7 months (range 1–120 months). The full cohort was divided into bipolar HA (77 cases) and PFNA (36 cases) groups. Damage control orthopedics was used to determine the optimal surgery time and assist in perioperative management. A restrictive blood transfusion strategy was employed, along with appropriate adjustments under multidisciplinary assessment, throughout the perioperative period. Perioperative clinical information and prognostic data were analyzed. Kaplan–Meier survival curves were used for survival analysis, and landmark analysis divided the entire follow‐up period into 1–12 months (short‐term), 13–42 months (medium‐term) and 43–120 months (long‐term) according to the configurations of Kaplan–Meier survival curves.

Results

Both groups had similar general variables except for the proportion of high adjusted Charlson comorbidity index (aCCI) (≥6 points) (6.5% in bipolar HA group and 22.2% in PFNA group, p = 0.024). Intraoperative blood loss and transfusion requirements were greater, and the intraoperative transfusion rates were higher in the bipolar HA group compared to the PFNA group (all p < 0.05). The complications rates, 1‐ to 60‐month cumulative all‐cause mortality, postoperative optimal Harris hip score (HHS), and Barthel index (BI) presented no significant difference between the two groups (all p > 0.05). Both groups had similar overall survival curves (p = 0.37). However, landmark analysis revealed that bipolar HA group exhibited higher survival rates in medium‐term (p = 0.01), while similar survival rates were observed in the short‐ and long‐term post‐operation periods (both p > 0.05). Cox regression with survival‐time‐dependent covariate calculated the hazard ratio (HR) of bipolar HA was 0.41 in medium‐term (p = 0.039).

Conclusion

Bipolar HA is equally effective and reliable as PFNA for treating intertrochanteric fractures in nonagenarians. Despite resulting in more intraoperative blood loss and transfusions, bipolar HA therapy is associated with a higher medium‐term survival rate compared to PFNA treatment. The application of damage control orthopedics and precise perioperative patient blood management could contribute to the positive clinical outcomes observed in this patient population.

Both bipolar hemiarthroplasty (HA) and proximal femoral nail anti‐rotation (PFNA) therapies are effective and reliable for the nonagenarians with intertrochanteric fractures. Bipolar HA therapy is associated with a higher medium‐term (13‐ to 42‐month) survival rate than PFNA treatment. The short‐term (1‐ to 12‐month) and long‐term (43‐ to 120‐month) survival rates are similar between two groups. The application of damage control orthopedics and precise perioperative patient blood management could contribute to the positive clinical outcomes observed in this patient population.

Introduction

Intertrochanteric and femoral neck fractures are the majority of hip fractures, ¹ the overall incidence of which increases exponentially with advancing age. ² As China's population continues to age at an accelerated pace, the number of individuals aged 90 years and older has risen from 1.98 million in 2010 to 4.59 million in 2020. ³ , ⁴ This portends that the number of hip fractures in nonagenarians will increase sharply in the coming years. However, the two fracture types are not only epidemiologically different but also impact differently on the elderly. Intertrochanteric fractures are slightly more prevalent compared to femoral neck fractures, ⁵ , ⁶ , ⁷ and are typically more characteristically osteoporotic. ⁸ Patients with intertrochanteric fractures tend to be older and more debilitated, often accompanied by higher mortality and disability rates. ⁸ , ⁹ , ¹⁰ Timely surgery for them remains the mainstay of treatment. ¹ However, for the nonagenarians with intertrochanteric fractures, poor health status and low bone mass are two intractable challenges in making operative proposals and perioperative management. Cautious decision making for these relatively lower functioning patients is indispensable.

The use of a cephalomedullary device is recommended for both stable and unstable intertrochanteric fractures. ¹¹ The biomechanical advantages, small incisions and the relatively intact blood supply to the femoral head contribute to the favorable clinical results. Nevertheless, osteoporosis, a condition whose prevalence increases with advancing age, can result in a sharp drop in biomechanical structural strength of the femur. This may significantly increase the risk of internal fixation failure, such as device fractures, coxa vara and nonunion. ⁶ , ¹² , ¹³ , ¹⁴ These complications can be particularly devastating for extremely elderly patients. In previous high‐quality studies, ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ the extremely old population of nonagenarians have not been analyzed separately, and the long‐term effects of more than 1 year are known to a limited extent.

The excellent results of arthroplasty have been consistently demonstrated in the treatment of femoral neck fractures by some high strength studies. ²¹ , ²² , ²³ The overall reoperation rate was reported only 0.52%–5.2%. ²⁴ , ²⁵ , ²⁶ , ²⁷ It can also facilitate early full weight‐bearing exercise and reduce the incidence of prolonged bed rest‐related complications. ⁶ But the relatively larger incision, more blood loss, higher risk of infection and dislocations require special attention during the application. ²⁸ , ²⁹ The use of arthroplasty as the primary choice for treating intertrochanteric fractures continues to be controversial. Several surgeons have investigated the application of hemiarthroplasty as a primary treatment for unstable intertrochanteric fractures and have observed some advantages in the process. ³⁰ , ³¹ , ³² Given that even stable subtypes have a failure rate ranging from 7.3% to 28.6% when treated with internal fixation, ¹⁴ , ³³ , ³⁴ , ³⁵ the extremely old patients have urgent requirements for early weight bearing and reduced tolerance for failure risk due to physiological decline. Therefore, we question whether arthroplasty could serve as an effective alternative for all types of intertrochanteric fractures in this special population.

Accordingly, the aim of this study was to investigate: (i) the postoperative complications; (ii) survival outcomes compared between bipolar hemiarthroplasty (HA) and proximal femoral nail anti‐rotation (PFNA) in nonagenarians with intertrochanteric fractures; and (iii) to evaluate the efficacy and safety of the two surgical procedures in this patient population.

Materials and Methods

Inclusion and Exclusion Criteria

The current study comprised all consecutive nonagenarians who underwent bipolar HA or PFNA for intertrochanteric fractures from January 2006 to August 2021 in Daping Hospital. Intertrochanteric fractures were diagnosed by the patients' history, symptoms, signs and radiological findings. The current study complied with the Declaration of Helsinki and was approved by the Ethics Committee of Daping Hospital of Army Medical University (2022[89]).

Inclusion criteria were: (i) patients aged 90 years or older; (ii) intertrochanteric fractures defined by the AO/OTA Fracture and Dislocation Classification; (iii) patients who received cemented bipolar HA, uncemented bipolar HA or PFNA therapy; and (iv) patients with complete clinical and follow‐up data. The exclusion criteria were: (i) intertrochanteric fractures secondary to tumoral etiology; (ii) patients with concomitant fractures or polytrauma; (iii) patients with incomplete medical records; and (iv) other surgical strategies, such as total hip arthroplasty (THA), dynamic hip screw (DHS), percutaneous compression plate system (PCCP) and so on. The AO/OTA Fracture and Dislocation Classification was cited from Buckley et al.: ³⁶ fractures in the trochanteric region of the proximal femur, including pertrochanteric fractures (31‐A1 and 31‐A2), intertrochanteric fractures with a horizontal line at the level of the lesser trochanter(reverse oblique, 31‐A3).

Clinical Information and Surgical Strategy

The eligible cases were divided into bipolar HA and PFNA groups. The general data, operation‐related information and laboratory indexes are summarized in Tables 1, 2, 3. To more thoroughly examine the fluctuations in laboratory indices, the current study focused on postoperative extreme data for analysis. The aCCI, ³⁷ physiological score of Physiological and Operative Severity Score for the enUmeration of Mortality and Morbidity (POSSUM) ³⁸ and American Society of Anesthesiologists (ASA) were used to quantify the comorbidities and health status of each patient. By utilizing the function “suv_cutpoint” in “survminer” R package (version 0.4.9), the optimal cut‐offs for aCCI and the physiological score of POSSUM were calculated as 6 and 29 respectively. We defined aCCI and POSSUM as “high” when they were at or above the cut‐offs.

TABLE 1.

The general conditions of the 113 nonagenarians

	Bipolar HA (n = 77)	PFNA (n = 36)	t/X2	p value
Age (years)	92.1 ± 2.5	92.3 ± 2.7	−0.39	0.695
Gender (male), No. (%)	23 (29.9)	11 (30.6)	0.01	0.941
Mechanism, No. (%)
Fall	64 (83.1)	30 (83.3)	0.00	0.977
Others	13 (16.9)	6 (16.7)
Affected side, No. (%)
Left	44 (57.1)	20 (55.6)	0.03	0.874
Right	33 (42.9)	16 (44.4)
AO/OTA classification
31‐A1	23 (29.9)	14 (38.9)	2.54	0.280
31‐A2	39 (50.6)	19 (52.8)
31‐A3	15 (19.5)	3 (8.3)
Time to admission (days), No. (%)
≤3	54 (70.1)	25 (69.4)	1.00	0.606
3–7	15 (19.5)	9 (25.0)
>7	8 (10.4)	2 (5.6)
Preoperative hospital stay (h), No. (%)
≤24	34 (44.2)	14 (38.9)	2.59	0.273
24–72	20 (26.0)	6 (16.7)
>72	23 (29.9)	16 (44.4)
Pulse (beats/min), No. (%)
≤100	65 (84.4)	29 (80.6)	0.26	0.609
>100	12 (15.6)	7 (19.4)
Respiration (times/min), No. (%)
≤20	72 (93.5)	30 (83.3)	Fisher	0.101
>20	5 (6.5)	6 (16.7)
Systolic pressure (mmHg), No. (%)
<140	43 (55.8)	24 (66.7)	1.19	0.275
≥140	34 (44.2)	12 (33.3)
Cardiac function classification NYHA III & IV, No. (%)	23 (29.9)	13 (48.1)	2.95	0.086
ASA classification III & IV, No. (%)	42 (54.5)	22 (61.1)	0.43	0.512
aCCI, No. (%)
High	5 (6.5)	8 (22.2)	Fisher	0.024
Low	72 (93.5)	28 (77.8)
Physiological score, No. (%)
High	24 (31.2)	7 (19.4)	1.69	0.193
Low	53 (68.8)	29 (80.6)

Note: AO/OTA classification was cited from Buckley et al. ³⁶

Abbreviations: aCCI, age‐adjusted Charlson Comorbidity Index; ASA, American Society of Anesthesiologists; Fisher, Fisher's exact test; Physiological score, physiological score of POSSUM.

TABLE 2.

Operation‐related information

	Bipolar HA (n = 77)	PFNA (n = 36)	t/X2	p value
Anesthesia, No. (%)
Nerve block anesthesia	70 (90.9)	27 (75.0)	Fisher	0.072
Intravertebral anesthesia	6 (7.8)	7 (19.4)
General anesthesia	1 (1.3)	2 (5.6)
Operation time (min)	94.1 ± 23.6	90.7 ± 18.8	0.76	0.450
Intraoperative blood loss (ml)	322.1 ± 128.1	173.1 ± 101.5	6.13	<0.001
Intraoperative fluid infusion (ml)	983.8 ± 424.9	981.9 ± 417.5	0.02	0.983
Total blood transfusion
Volume 1 (ml)	807.1 ± 568.7	597.2 ± 529.4	1.84	0.064
Volume 2 (ml)	837.7 ± 496.1	716.7 ± 499.7	1.12	0.264
No. (%)	73 (94.8)	30 (83.3)	Fisher	0.072
Preoperative blood transfusion
Volume 1 (ml)	51.9 ± 201.1	83.3 ± 226.2	−0.74	0.459
Volume 2 (ml)	666.7 ± 350.2	600.0 ± 244.9	0.36	0.729
No. (%)	6 (7.8)	5 (13.9)	Fisher	0.323
Intraoperative blood transfusion
Volume 1 (ml)	457.1 ± 268.2	222.2 ± 223.1	4.57	<0.001
Volume 2 (ml)	510.1 ± 230.2	421.1 ± 91.8	2.56	0.012
No. (%)	69 (89.6)	19 (52.8)	19.32	<0.001
Postoperative blood transfusion
Volume 1 (ml)	273.4 ± 336.7	291.7 ± 371.4	−0.26	0.795
Volume 2 (ml)	539.7 ± 281.5	500.0 ± 363.3	0.47	0.640
No. (%)	39 (50.6)	21 (58.3)	0.58	0.446
Postoperative ICU admission
Length (day)	0.7 ± 1.3	1.2 ± 2.4	−1.09	0.278
No. (%)	27 (35.1)	15 (41.7)	0.46	0.499
Postoperative hospital stay (day)	8 ± 3.8	9.3 ± 5.1	−1.34	0.184

Abbreviations: Fisher, Fisher's exact test; Volume 1, average blood transfusion volume for all patients (including those who did not receive a transfusion); Volume 2, average blood transfusion volume for patients who received blood transfusion.

TABLE 3.

Laboratory indexes of the nonagenarians

	Bipolar HA (n = 77)	PFNA (n = 36)	t/X2	p value
HGB <80 g/L, No. (%)
Preoperation	13 (16.9)	6 (16.7)	0.00	0.977
Postoperation	30 (39)	16 (44.4)	0.31	0.580
HGB change (g/L)	12.5 ± 17.5	15.5 ± 22.9	0.77	0.443
HCT < 35, No. (%)
Preoperation	64 (83.1)	26 (72.2)	1.80	0.180
Postoperation	77 (100)	35 (97.2)	Fisher	0.319
HCT change	3.8 ± 5.1	4.6 ± 6.6	0.70	0.487
ALB
Preoperation <35 g/L, No. (%)	56 (72.7)	22 (61.1)	1.55	0.213
Postoperation (g/L)	23.2 ± 2.3	25.2 ± 3.5	−3.09	0.003
ALB change (g/L)	9.1 ± 5.4	7.0 ± 6.2	1.83	0.070
Na+ < 135 mmol/L, No. (%)
Preoperation	14 (18.2)	2 (5.6)	3.22	0.073
Postoperation	10 (13.0)	3 (8.3)	Fisher	0.546
Na+ change (mmol/L)	0.7 ± 11.7	1.6 ± 4.2	−0.58	0.565
K+ < 3.5 mmol/L, No. (%)
Preoperation	10 (13.0)	3 (8.3)	Fisher	0.546
Postoperation	27 (35.1)	12 (33.3)	0.03	0.857
K+ change (mmol/L)	0.3 ± 0.4	0.4 ± 0.7	−1.08	0.285
Ca2+ < 2.11 mmol/L, No. (%)
Preoperation	41 (53.2)	16 (44.4)	0.76	0.383
Postoperation	72 (93.5)	34 (94.4)	Fisher	1.000
Ca2+ change (mmol/L)	0.1 ± 0.1	0.2 ± 0.2	1.43	0.155
Crea >97 μmmol/L, No. (%)
Preoperation	22 (28.6)	10 (27.8)	0.01	0.930
Postoperation	28 (36.4)	13 (36.1)	0.00	0.979
Crea change (μmmol/L)	−8 ± 34.4	8.7 ± 70.1	−1.36	0.182
Urea>8 mmol/L, No. (%)
Preoperation	42 (54.5)	15 (41.7)	1.63	0.202
Postoperation	43 (55.8)	16 (44.4)	1.28	0.258
Urea change (mmol/L)	0.1 ± 5.3	0.9 ± 4.6	−0.80	0.427

Abbreviations: ALB, albumin; Fisher, Fisher's exact test; HCT, hematocrit; HGB, hemoglobin; Index change, preoperative value minus the corresponding postoperative value.

After admission, damage control orthopedics was used as previously described by our team to determine the optimal surgery time and assist in perioperative management. ³⁹ , ⁴⁰ The decision on surgical strategy was made synthetically by a risk–benefit assessment, the wishes of patients and their family members, and the therapeutic advice of the multidisciplinary team. Both bipolar HA and PFNA were performed by senior surgeons. Lateral decubitus position and minimally invasive posterior approach were used for bipolar HA. Preparation of the acetabulum, reaming the medullary canal of the femur, implantation of a decent size prosthesis, reduction and fixation of the trochanteric fragments were completed in order. Fig. 1 displays the preoperative and postoperative pelvic anteroposterior X‐ray images. The procedures of PFNA were performed as follows. After traction‐bed‐assisted reduction of the intertrochanteric fractures, a minimal incision was performed along the outer side of the top of the greater trochanter. Then, the insertion of a guide nail into the medullary cavity, reaming, implantation of a proper PFNA, insertion of guide nail and spiral blade into the femoral neck, fixation of the nails were consequently conducted. The guide nails were timely adjusted to the optimal direction and sited by fluoroscopy during the operation. The preoperative and postoperative pelvic anteroposterior X‐ray images are shown in Fig. 2.

FIG. 1.

A 94‐year‐old male patient with right femoral intertrochanteric fracture caused by a fall while walking at home (AO classification: 31‐A1.3) and a left bipolar prosthesis implanted three months ago due to intertrochanteric fracture. Right bipolar HA was performed. (A) Preoperative pelvic anteroposterior X‐ray image. (B) Postoperative pelvic anteroposterior X‐ray image.

FIG. 2.

A 90‐year‐old male patient with left femoral intertrochanteric fracture caused by a fall (AO classification: 31‐A1.2). PFNA was performed. (A) Preoperative pelvic anteroposterior X‐ray image. (B) Postoperative pelvic anteroposterior X‐ray image.

A restrictive transfusion triggered by hemoglobin 80 g/L was adopted for the patients, along with appropriate adjustments under multidisciplinary assessment, throughout the perioperative period.

Prognostic Data

The prognostic information was obtained from electronic medical records, periodical check‐ups (clinical and radiological evaluation) and telephone follow‐ups. The follow‐up ended on October 31, 2022. The complications included pulmonary infection, respiratory failure, myocardial infarct, heart failure, cardiac arrhythmias, cerebral infarction, cerebral hemorrhage, renal failure, urinary tract infection, diarrhea and constipation, deep venous thrombosis (DVT) and incision infection. All of them are summarized and classified in Table 4. The cumulative all‐cause mortality was estimated at seven distinct time points during 60 months after operation. We adopted the optimal level of Harris hip score (HHS) and Barthel index (BI) in basic activities of daily living (ADL) score during the follow‐up period to evaluate the postoperative hip function and the functional independence. The Kaplan–Meier approach was constructed for survival analysis. Based on the configurations of the two survival curves, landmark analysis was used to divide the entire follow‐up period into 1–12, 13–42 and 43–120 months, which were referred to as “short‐term,” “medium‐term,” and “long‐term” in the current study.

TABLE 4.

Postoperative complications and improvements in function

	Bipolar HA (n = 77)	PFNA (n = 36)	t/X2	p value
Respiratory complications, No. (%)	7 (9.1)	6 (16.7)	Fisher	0.342
Cardiac complications, No. (%)	10 (13)	6 (16.7)	0.27	0.601
Cerebral complications, No. (%)	1 (1.3)	1 (2.8)	Fisher	0.538
Urinary complications, No. (%)	3 (3.9)	1 (2.8)	Fisher	1.000
Digestive complications, No. (%)	1 (1.3)	1 (2.8)	Fisher	0.538
Deep vein thrombosis, No. (%)	1 (1.3)	2 (5.6)	Fisher	0.238
Incision complications, No. (%)	2 (2.6)	1 (2.8)	Fisher	1.000
MODS, No. (%)	5 (6.5)	2 (5.6)	Fisher	1.000
Cumulative deaths in 1 M, No. (%)	11 (14.3)	5 (13.9)	0.00	0.955
Cumulative deaths in 6 M, No. (%)	16 (20.8)	6 (16.7)	0.26	0.607
Cumulative deaths in 12 M, No. (%)	24 (31.2)	12 (33.3)	0.05	0.818
Cumulative deaths in 24 M, No. (%)	30 (39.0)	19 (52.8)	1.39	0.239
Cumulative deaths in 36 M, No. (%)	33 (42.9)	22 (61.1)	2.58	0.108
Cumulative deaths in 48 M, No. (%)	40 (51.9)	22 (61.1)	0.50	0.478
Cumulative deaths in 60 M, No. (%)	44 (57.1)	25 (69.4)	1.09	0.297
Barthel index	69.2 ± 13.2*	66.3 ± 13.6 †	1.05	0.298
Barthel index >60 No. (%)	49 (69.0)*	23 (67.6) †	0.02	0.888
Harris hip score	72.5 ± 7.4*	69.6 ± 7.9 †	1.84	0.069
Harris hip score >70 No. (%)	41 (57.7)*	16 (47.1) †	1.06	0.304

^*6 in‐hospital deaths were censored.

^† 2 in‐hospital deaths were censored.

Abbreviations: Cumulative deaths, cumulative deaths in postoperative 1 month, 3 months, 6 months and 12 months; Fisher, Fisher's exact test; M, month(s); MODS, multiple organ dysfunction syndrome.

Statistical Analysis

R software version 4.2.1 (R Foundation, Vienna, Austria) was used to conduct all statistical analyses. Statistical significance was assumed when p < 0.05. Categorical variables were described as count and percentages (%), while continuous variables were described as mean ± sd (⁻ x ± s). Differences between categorical variables were analyzed by the chi‐square test or Fisher's exact test, while differences among continuous variables were analyzed by t‐test. The rate of missing values in laboratory indexes was 2.3% (less than 10.0%). All of the missing values were imputed by multiple imputation methods using the “mice” R package (version 3.14.0). Time‐dependent Cox regression was used to analyze the covariate of operation type by using “survival” (version 3.3–1) and “survminer” (version 0.4.9) R package. We estimated differences in survival curves between the two groups by log rank test.

Result

General Information

During the past 15 years, a total of 123 nonagenarians with intertrochanteric fractures underwent surgical treatment in our hospital. After verifying exclusion criteria, 10 patients were excluded (five underwent THA, three underwent DHS or PCCP, two were lost to follow‐up) and 113 remained. The eligible patients included 34 males and 79 females, with a mean age of 92.2 years (range 90–101 years) at the time of operation. The bipolar HA patients used 56 cemented prostheses and 21uncemented prostheses. The demographic characters, vital signs and AO/OTA classification were similar between bipolar HA and PFNA groups (all p > 0.05, Table 1). The PFNA group demonstrated a higher proportion of high aCCI compared to the bipolar HA group (p = 0.024). No significant differences were observed in distributions of ASA classification, cardiac function classification and POSSUM physiological score between the two groups (all p > 0.05, Table 1).

Operation‐related Information and Laboratory Indexes

The operation‐related information is listed in Table 2. Intraoperative blood loss in the bipolar HA group was more than that in the PFNA group (p < 0.001). Consequently, both the blood transfusion volume and the transfusion rate during operation were higher in the bipolar HA group compared to the PFNA group (all p < 0.001). There was no significant difference in preoperative and postoperative blood transfusion, anesthesia, operation time, intraoperative fluid infusion, postoperative ICU admission and postoperative hospital stay between the two groups (all p > 0.05).

Table 3 displays the laboratory indexes. The preoperative laboratory indexes were similar between the two groups (all p > 0.05). No significant difference was found in postoperative laboratory indexes between the two groups (all p > 0.05), except for the postoperative albumin (23.2 ± 2.3g/L in bipolar HA group and 25.2 ± 3.5 g/L in PFNA group, p = 0.003). Additionally, there was no significant difference in the changes of them between the two groups (all p > 0.05).

Postoperative Complications and Improvements in Function

The incidences of postoperative complications including respiratory, cardiac, cerebral, urinary, digestive complications, incision complications, deep vein thrombosis and multiple organ dysfunction syndrome (MODS) were similar between bipolar HA and PFNA groups (all p > 0.05, Table 4). No dislocation of the affected hip, periprosthetic femoral fracture, or periprosthetic joint infection was observed in this retrospective cohort study. The cumulative deaths in postoperative 1–60 months, as well as HHS and BI demonstrated no significant difference between the two groups (all p > 0.05, Table 4).

Survival Analysis

The average duration of follow‐up was 29.7(±26.4) months (range, 1–120 months). The median survival for all of the patients was 36 months (Fig. 3). Meanwhile, the median survival in bipolar HA and PFNA groups was 45 and 22 months, respectively. Kaplan–Meier analysis and log rank tests showed no significant difference in the survival curves between the two groups (p = 0.37, Fig. 4). Landmark analysis revealed that the patients in bipolar HA group had higher medium‐term (13‐ to 42‐month) survival rates than patients in PFNA group (p = 0.01), whereas there was no significant difference in short‐term (1‐ to 12‐month) and long‐term (43‐ to 120‐month) survival rates between the two groups (p = 0.20 and p = 0.60, respectively) (Fig. 5). Time‐dependent Cox regression also confirmed that the type of operation exerted significantly distinct effects on survival during the medium‐term period, with the HR of bipolar HA being 0.41 (p = 0.039, Table 5).

FIG. 3.

The overall survival curve. The average duration of follow‐up was 29.7(±26.4) months (range 1–120 months), the median survival of the patients (n = 113) was 36 months.

FIG. 4.

Kaplan–Meier curves of bipolar HA and PFNA groups. The median survival of the patients in bipolar HA and PFNA groups was 45 and 22 months respectively. Kaplan–Meier analysis and log rank test showed no significant difference in the overall survival curves between two groups (p = 0.37). But intersections appeared in the survival curves.

FIG. 5.

Landmark analysis. Bipolar HA group had the higher medium‐term (13‐ to 42‐month) survival rates than PFNA group (p = 0.01). Short‐term (1‐ to 12‐month) and long‐term (43‐ to 120‐month) survival rates were similar between two groups (p = 0.20 and p = 0.60, respectively).

TABLE 5.

Cox regression with survival‐time‐dependent covariate

	coef	se (coef)	HR	z	p
PFNA	‐	‐	1	‐	‐
Bipolar HA
1–12 M	−0.05	0.35	0.95	−0.14	0.891
13–42 M	−0.90	0.44	0.41	−2.06	0.039
43–108 M	0.23	0.47	1.26	0.49	0.627

Abbreviations: HA, hemiarthroplasty; HR, Hazard Ratio; M, month(s); PFNA, proximal femoral nail anti‐rotation.

Discussion

The efficacy and safety differences between bipolar HA and PFNA in nonagenarians with intertrochanteric fractures over a long‐term period are still not well understood. The controversy persists regarding bipolar HA as the primary treatment of choice for this patient population. In the current study, 1‐year all‐cause mortality was 31.9%, which was deemed acceptable and did not increase significantly in relation to extreme age and the physiological decline. This is in comparison to the previous study's findings of 29.3% in the elderly group ⁴¹ and 25.9% in the overall age group. ⁴² We observed that nonagenarians in the PFNA group exhibited a higher aCCI index, while the patients undergoing bipolar HA experienced more intraoperative blood loss and required more transfusions. However, upon in‐depth analysis, no significant difference was found in the complications between the two groups, and the postoperative hip function also improved to a similar extent in both groups. Survival analysis and landmark analysis indicated the patients receiving bipolar HA therapy exhibited higher medium‐term (13‐ to 42‐month) survival rates than those undergoing PFNA treatment. Damage control orthopedics and adequate transfusion strategy may help decrease the potential risk of adverse events.

Perioperative Assessment and Management

Our team introduced damage control orthopedics for the perioperative management of the high‐risk patients, aiming to minimize negative sequelae following operation, ³⁹ , ⁴⁰ which are affected by various factors. The wait time for surgery has been one of the most discussed issues. Although early surgery within 24–48 h of admission is associated with better outcomes, ⁴³ a more considered strategy should be adopted for the special populations like nonagenarians and the patients with severe comorbidities. POSSUM, aCCI, ASA are our preferred instruments for preoperative risk stratification. In our experience, patients whose predictive morbidity and mortality from POSSUM were below 60% and 20% would receive operation within 24–48 h. If the values rise above 60% and 20%, the operation would be postponed for 1–3 days to optimize the coexisting medical conditions and increase the patient's physiological tolerance for surgical invasion. Dynamic risk assessment and medical treatment are jointly used for the determination of optimal surgery time. Based on our past experience, we consider damage control orthopedics to be both safe and effective for elderly patients with comorbidities.

Perioperative Patient Blood Management

In general, arthroplasty is associated with a relatively greater severity of surgical trauma and increased blood loss, even when performed by proficient surgeons, compared to PFNA. Anemia caused by blood loss and bleeding is associated with postoperative mortality and morbidity in the elderly. ⁷ , ⁴⁴ Preventing or timely correcting anemia is vital for this patient population. In the present study, the blood loss, blood transfusion, and transfusion rates during the surgical procedure in the bipolar HA group were significantly higher than those in the PFNA group. This finding is consistent with the literature reported. ⁴⁵ Furthermore, we found that postoperative HGB and HCT in bipolar HA group did not show a significant decrease compared with PFNA group. This might benefit from the adequate transfusion strategy. Several studies reported that blood transfusion might increase postoperative morbidity and mortality. ⁴⁶ , ⁴⁷ But another viewpoint is perioperative blood transfusion is not closely associated with postoperative results. ⁴⁸ , ⁴⁹ Hopewellet al. ⁵⁰ believed that the varying conclusions could be attributed to potential confounders and unadjusted factors.

A less‐biased perspective is that maintaining hemoglobin, optimizing hemostasis, and minimizing blood loss should improve care processes and patient outcomes. ⁵¹ In our clinical practice, we adhere to a restrictive transfusion strategy, setting a threshold of hemoglobin 80g/L. For frail patients or those with ischaemia heart disease, the threshold was up‐regulated to hemoglobin 90g/L or 100g/L after multidisciplinary consultation. In consideration of some patients with anemia before operation, dynamic multidisciplinary assessment should run through the perioperative period. Meanwhile, the improvements in surgical technology, diathermy dissection, meticulous hemostasis and application of tranexamic acid can effectively reduce the operation duration, bleeding and blood loss. ⁵¹ In addition, the significant difference in postoperative albumin between the two groups warns us that more attention should be paid to the correction of hypoproteinemia in parallel with restoring and maintaining adequate fluid and electrolyte balance. Human serum albumin proved to be both safe and effective for use in resuscitation. ⁵²

Postoperative Complications, Functional Results and Survival Analysis

The postoperative complication rates and functional results showed no significant difference between the two groups, which was consistent with previous literature. ³¹ , ⁴⁵ , ⁵³ However, respiratory and cardiac complications constituted the largest percentage of complications in this study, emphasizing the need for comprehensive and interdisciplinary management during the perioperative period. Upon analyzing the postoperative optimal BI and HHS, we observed that nearly two thirds of the discharged patients got out of “severe dependence” (BI > 60) and half of them got a “fair” or much better hip function (HHS > 70). Emphasis should be placed on functional recovery, activities of daily living, and osteoporosis treatment by the multidisciplinary team throughout the hospitalization and post‐discharge periods.

The crossed survival curves phenomenon depicted in Fig. 4 is a noteworthy aspect. Although there was no significant difference between the two survival curves in Fig. 4, further analysis indicated that bipolar HA was associated with the higher medium‐term (13‐ to 42‐month) survival rates (Fig. 5). This appears to be determined by three factors as follows. First, survival was affected by the comorbidities. In the current study, nonagenarians undergoing PFNA tended to have more or more severe comorbidities. The percentages of high aCCI cases in PFNA and bipolar HA groups were 22.2% versus 6.5%. Charlson et al. ⁵⁴ found that increasing comorbidities were associated with 90‐day, 1‐year and even 5‐year mortality. The Charlson comorbidity index tends to increase over time. A higher baseline index is associated with a higher mortality rate, which is further aggravated by a subsequent increase in the Charlson comorbidity index over a ten‐year period. ⁵⁴ , ⁵⁵ Despite the two groups had the similar cumulative all‐cause mortality in different periods, there were 12 and 14 deaths in bipolar HA and PFNA groups from 12 to 40 months after operation. Thus, deaths in PFNA group accounted for a higher proportion during that period (12/50 vs. 14/29, p = 0.049). However, we must be cautious about this perspective because statistical evidence based on large‐scale data is lacking.

Second, patients of advanced age are especially vulnerable to unstable intertrochanteric fractures. ⁵⁶ Osteoporosis also increases the tendency of loosening and displacement of the internal fixation. ⁵⁷ Poor reduction before the operation, suboptimal fixation location, and femoral head cutting are usually caused by unstable fractures and severe osteoporosis. ⁶ For the PFNA patients, hip function and activities of daily living will be impaired by every mentioned failure type above. By contrast, evidence supported that cemented hemiarthroplasty was associated with a high postoperative health‐related quality of life. ⁵⁸

Finally, the movement deficits may persist for several months during the rehabilitative period after operation. ¹ The weak strength of upper limb made it difficult for the patients to walk effectively with the assistance of walking aid. The fear of falls and prostheses loosening also prolonged the bed rest time which was closely associated with bed‐related complications. From our observations, this phenomenon is more evident in elderly patients undergoing PFNA compared to those receiving arthroplasty. The latter patient population tends to resume weight‐bearing activities more quickly than patients undergoing PFNA. ⁵³ , ⁵⁹ Over time, the different consequences of these phenomena will be concentrated in a certain period after surgery, like 12–40 months after operation in the current study.

Strengths and Limitations

To our knowledge, this is the first study to analyze the effects compared between bipolar HA and PFNA in nonagenarians with intertrochanteric fractures. The current study lasted for 15 years, which is the longest follow‐up period reported in nonagenarians to date, allowing us to objectively observe the complications and survival rates in a long term. However, there were some limitations in the current study. First, it was a retrospective study at a single center, and the sample size was relatively small, the internal and external validity was limited. Second, the nonagenarians' compliance with periodical visits to the hospital clinic after discharge has significantly declined, and the time span of the current study is great, some postoperative information especially radiographic images of the affected hip was not obtained. Postoperative fracture healing, loosening of intramedullary nails or helical blades, and hip prosthesis loosening cannot be evaluated objectively. Third, we did not analyze all of the potential factors that may affect the prognosis and the weight of each surgical method.

Conclusions

Our study demonstrates that bipolar HA is just as effective and reliable as PFNA for treating intertrochanteric fractures in nonagenarians. Although bipolar HA therapy results in more intraoperative blood loss and transfusions, it is associated with a higher medium‐term survival rate compared to PFNA treatment. The implementation of damage control orthopedics and precise perioperative patient blood management may contribute to the positive clinical outcomes observed in this patient population.

Author Contributions

Yan Xiong and Ziming Wang proposed the conception, designed the study and revised the manuscript critically. Xingchen Lu analyzed the data, interpreted the statistic results. Xingchen Lu and Wenlong Gou drafted and revisied the manuscript. Xingchen Lu, Wenlong Gou, Siyu Wu and Yu Wang conducted follow up observations, collected and assembled the clinical data.

Ethics Statement

The current study complied with the Declaration of Helsinki and was approved by the Ethics Committee of Daping Hospital of Army Medical University (2022[89]).