Periprosthetic and peri-implant femoral fractures (PPF) include periprosthetic hip fractures (periPC), periprosthetic knee fractures (periPR), as well as peri-implant fractures associated with an osteosynthesis nail (periCL). Although risk factors for PPF are widely studied, there is scarce evidence on differential risk factors for each subtype. The objective of this study is to compare 30-day and 1-year mortality rates among the three PPF subtypes and identify differential risk factors between them.
MethodologyA retrospective cohort study conducted in a tertiary hospital, including 100 patients with PPF between 2019 and 2022, with a minimum follow-up of one year. Demographic, clinical, and surgical data were collected, along with 30-day and 1-year mortality and changes in ambulation capacity. Statistical analysis included intergroup comparison tests and stratified mortality analysis.
ResultsThere were 49 periprosthetic hip fractures, 29 knee fractures, and 22 femoral nail-related fractures. Overall mortality was 12% at 30 days and 28% at one year, with no differences between subtypes. Mortality was associated with ASA>III, CCI>5, systemic complications, intraoperative transfusions, and deterioration of ambulation capacity (p<.05). Stratified analysis identified significant differential associations in periPC (age, CCI>5, ASA>III, systemic complications, postoperative ambulatory status) and periPR (ASA, surgical time, systemic and local complications, ambulatory status), with no significant factors in periCL.
ConclusionsElevated ASA index, CCI, and systemic complications are risk factors in all three types of PPF. Surgical time and local complications were associated with worse outcomes in the periPR group. Novel findings include the need for blood transfusions and deterioration of ambulatory status as modifiable risk factors associated with higher mortality.
Las fracturas femorales periprotésicas y periimplante (PPF) agrupan a fracturas periprotésicas de cadera (periPC) y de rodilla (periPR), así como fracturas periimplante asociadas a un clavo de osteosíntesis (periCL). Aunque los factores de riesgo de las PPF están ampliamente estudiados, hay escasa evidencia de factores de riesgo diferenciales para cada subtipo. El objetivo de este trabajo es comparar las tasas de mortalidad a los 30 días y al año entre los tres subtipos de PPF e identificar factores de riesgo diferenciales entre ellos.
MetodologíaEstudio de cohorte retrospectiva en un hospital terciario, que incluyó 100 pacientes con PPF entre 2019 y 2022, con seguimiento mínimo de un año. Se recogieron datos demográficos, clínicos y quirúrgicos, así como mortalidad a 30 días y a 1 año, y cambios en la capacidad deambulatoria. El análisis estadístico incluyó pruebas de comparación intergrupal y análisis estratificado de la mortalidad.
ResultadosSe registraron 49 fracturas periprotésicas de cadera, 29 de rodilla y 22 de clavo femoral. La mortalidad global fue del 12% a 30 días y del 28% al año, sin diferencias entre subtipos. La mortalidad se asoció a ASA>III, índice de comorbilidad de Charlson (CCI)>5, complicaciones sistémicas, transfusiones intraoperatorias y deterioro de la capacidad deambulatoria (p<0,05). En el análisis estratificado se identificaron asociaciones diferenciales significativas en periPC (edad, CCI>5, ASA>III complicaciones sistémicas, estatus deambulatorio posquirúrgico) y periPR (ASA, tiempo quirúrgico, complicaciones sistémicas y locales y estatus deambulatorio), sin factores significativos en periCL.
ConclusionesEl índice de ASA elevado, el CCI y las complicaciones sistémicas son factores de riesgo en los tres tipos de PPF. El tiempo quirúrgico y las complicaciones locales se asociaron con un peor resultado en el grupo periPR. Como hallazgo novedoso, la necesidad de transfusiones sanguíneas y el deterioro del estado deambulatorio son factores de riesgo modificables que asociamos con mayor mortalidad.
The increased prevalence of implants and arthroplasties, along with the rise in life expectancy in Western countries, has led to a higher incidence of periprosthetic and peri-implant femur fractures (PPF).1 These fractures represent a significant clinical and healthcare burden for trauma services: in our setting. The current incidence is estimated to be between .1% and 4%.1 Furthermore, PPFs have a significant socioeconomic impact, and projections indicate that in the coming decades there will be an exponential increase in healthcare costs directly associated with this type of fracture.1–7 PPFs occur in elderly patients, frequently with multiple comorbidities with the result that morbidity and mortality associated with these fractures is considerable.1,2
The term encompasses three types of fractures: periprosthetic hip (periPC), peri-femoral nail implant (periCL), and periprosthetic knee (periPR). Although mortality from prosthetic hip fractures is clearly linked to factors such as age and a high ASA score, there is little evidence of differential risk factors for each fracture subtype.8–12
We hypothesised that independent risk factors exist for each fracture type. The primary objective of this study was to compare 30-day and 1-year mortality rates among the three prosthetic hip fracture subtypes (periPC, periPR, and periCL) and to identify any differential risk factors among them. A secondary objective was to analyse which factors influenced the worsening of patients’ ambulatory status.
MethodologyStudy designA retrospective cohort study was conducted at a tertiary centre. It included periprosthetic fractures (PPFs) performed at this centre between 2019 and 2022, with a minimum follow-up of one year. The study protocol was approved by our institution's Ethics Committee (code #2024-0819-1).
Data collection protocolAll patients coded with periprosthetic or peri-implant fractures were identified using the clinical information system (Orion Clinics®). Patients with femur fractures around a hip prosthesis (periPC), femoral nail (periCL), or knee prosthesis (periPR) were selected. Only patients with complete medical records and available radiographs, treated entirely at our centre, were included. Cases with incomplete perioperative data, intraoperative fractures, or pathological fractures secondary to malignancy or infection were excluded. Of a total of 133 patients, 33 were excluded, resulting in a final cohort of 100 patients.
For each patient, age, sex, weight, height, and body mass index (BMI) were collected. Baseline status was assessed using the American Society of Anaesthesiologists (ASA) classification obtained from medical records and the Charlson Comorbidity Index (CCI) calculated based on coexisting pathologies.8,13
Fracture mechanisms were categorised as high-energy or low-energy trauma (from standing height). Osteoporosis was considered present if: (1) low bone density was demonstrated by densitometry (T-score <−2.5); (2) there was a history of osteoporotic fractures (distal radius, vertebral, or hip); and (3) the cortical thickness index was <−.40.8 The use of anti-osteoporosis treatment was recorded before and after the fracture, as well as referral to the rheumatology service. For peri-articular fractures, the type of fixation (cemented or uncemented) was documented.
Mortality (at 30 days and 1 year) and ambulatory capacity were collected before and after the fracture. Following the example of previous studies, ambulatory capacity was classified into three stages (from best to worst): (1) independent walking; (2) walking with the aid of a cane; and (3) inability to walk. The change between the pre- and post-surgical status was assessed. The length of hospital stay for all patients was also recorded.
Fractures were categorised according to the previous implant as: periCL (femoral nail), periPC (hip prosthesis), and periPR (knee prosthesis). Treatment modalities were categorised as conservative, osteosynthesis (plate, cerclage, or nail in the case of periPC/periPR), or implant revision (new arthroplasty or nail replacement in the case of periCL). For patients who underwent surgery, the number of days until surgery, the rate of postoperative complications, the need for intraoperative transfusion, and the number of red blood cell units used were recorded. Following previous examples,8,14 the time until surgery was dichotomised as <96h or ≥96h. Postoperative complications were also classified as local (affecting the surgical wound) or systemic.
Statistical analysisStatistical analysis was performed using IBM SPSS Statistics 25.0 (IBM, Chicago, IL, USA). Quantitative variables were expressed as mean and standard deviation. ANOVA was used to analyse differences between the three groups (periCL, periPC, and periPR). If statistically significant differences were found, post-hoc comparisons were performed using the Bonferroni correction. Categorical variables were analysed using the chi-square test, and in comparisons with more than two categories, corrected standardised residuals were evaluated.
To perform the stratified analysis of one-year mortality between groups, the Kolmogorov–Smirnov normality test was initially performed. For variables that followed a normal distribution, independent t-tests were performed to analyse the relationship between the event and the variable. For variables that did not follow a normal distribution, the Mann–Whitney U test was used as a non-parametric test. Logistic regression models were constructed to identify the combined effect of risk factors. A p-value <.05 was considered statistically significant.
ResultsThe final cohort consisted of 100 patients, with a mean age of 82.65±10.36 years. Sixty-four percent of the sample were women, with statistically significant differences between the groups (p-value=.002). Standardised residuals (women=−3.5; men=3.5) showed that the most marked differences were observed in the periCL group compared to the other groups.
According to fracture type: 49 were periPC, 29 periPR, and 22 periCL. The mean BMI was 27.63±5.34, with a significant difference between the periPR group (31.27kg/m2) and the other groups (p=.004).
Regarding comorbidity, 82% presented with a CCI≥5. Among the 83 patients who underwent surgical treatment, the mean ASA score was III, with 63 patients (75.9%) classified as ASA≥III.
Most fractures (90%) were due to low-energy trauma. Pre-existing osteoporosis was identified in 66% of the patients; 20% received treatment before the fracture and 25% began treatment afterward. Seven percent of the patients were referred to rheumatology. Table 1 provides complete sociodemographic data for the population and treatment groups.
Description of the population and the treatment groups.
| Totaln=100 (100%) | periCLn=22 | periPCn=49 | periPRn=29 | p | |
|---|---|---|---|---|---|
| Gender | .02* | ||||
| Women | 64 (64.0) | 18 (81.8) | 23 (46.9) | 23 (79.3) | |
| Men | 36 (36.0) | 4 (18.2) | 26 (53.1) | 6 (20.7) | |
| Age (years)a | 80.8±10.4 | 76.81±5.1 | 86.7±8.32 | 80.9±8.5 | .089 |
| BMIa | 27.3±5.3 | 25.23±6.4 | 26.8±4.2 | 31.3±4.1 | .004* |
| ICCa | 5.9±2.5 | 7.33±3.0 | 5.5±2.5 | 5.6±2.7 | .118 |
| ICC | .649 | ||||
| 0–1 | 2 (2.0) | 0 (0.0) | 2 (4.1) | 0 (0.0) | |
| 2 | 5 (5.0) | 1 (4.5) | 2 (4.1) | 2 (6.9) | |
| ≥3 | 93 (93.0) | 21 (95.5) | 45 (91.8) | 27 (93.1) | |
| ASA classification | .91 | ||||
| II | 20 (24.1) | 5 (25.0) | 10 (27.0) | 5 (19.2) | |
| III | 51 (61.4) | 13 (65.0) | 20 (54.1) | 18 (69.2) | |
| IV | 12 (14.5) | 2 (10.0) | 7 (18.9) | 3 (11.5) | |
| Aetiology | 1.00 | ||||
| Low energy | 10 (10.0) | 3 (13.6) | 7 (14.3) | 4 (13.8) | |
| High energy | 90 (90.0) | 19 (86.4) | 42 (85.7) | 25 (86.2) | |
| Cementation | |||||
| Uncemented | 32 (65.3) | ||||
| Cemented | 17 (34.7) | ||||
| Hospital stay (days) | 11.2±8.3 | 11.77±8.60 | 11.8±8.6 | 10.44±6.51 | .78 |
| Treatment | .86 | ||||
| Osteosynthesis | 67 (67.0) | 18 (81.8) | 25 (51.02) | 24 (82.8) | |
| Review | 12 (12.0) | 1 (4.8) | 10 (20.40) | 1 (3.4) | |
| Conservative | 21 (21.0) | 3 (13.6) | 14 (28.57) | 4 (13.8) | |
| Time to surgery (days) (n=79) | 7.04±5.29 | 7.00±8.23 | 7.59±4.13 | 6.30±3.50 | .678 |
| Time in surgery (min) (n=79)* | 146.9±56.0 | 138.5±55.1 | 159.8±65.1 | 140.9±47.7 | .39 |
| Transfusion units (n=79) | 22 (29.7) | 3 (15.8) | 13 (40.1) | 6 (26.1) | .16 |
| Red blood cell count (n=79)* | 1.1±1.1 | .4±.69 | 1.6±1.3 | .87±.64 | .02* |
| Mortality | |||||
| 30 days | 12 (12.0) | 1 (4.5) | 8 (16.3) | 3 (10.3) | .357 |
| 1 year | 28 (28.0) | 6 (27.3) | 14 (28.6) | 8 (27.6) | .992 |
ASA: American Society of Anaesthesiologists; CHI: Charlson Comorbidity Index; BMI: body mass index.
The overall mortality rate was 12% at 30 days and 28% at one year, with no statistically significant differences among the three groups (p>.05). Twelve patients died during hospitalisation, 10 of them after surgery. Postoperative complications occurred in 35 patients, 77% of which were systemic (mainly nosocomial infections and cardiovascular events). Eight patients had local complications; only 3 were serious, including nerve injury (n=2) and arterial ischaemia (n=1).
Overall 30-day mortality was statistically significantly associated with ASA classification >III (p=.038), inability to walk or use a cane prior to surgery (p=.05), and the number of intraoperative transfusions (p=.004). A non-significant trend toward higher mortality was observed in patients with ICC>5 (p=.054). In this study, time to surgery was not statistically significantly associated with 30-day mortality.
Overall one-year mortality was associated with ASA>III (p=.009), ICC>4 (p=.04), worsening ambulatory status (p=.01), need for intraoperative transfusions (p=.019), and number of units transfused (p=.028).
A stratified analysis of one-year mortality was performed. In the periPC group, it was associated with age (p=.047), ICC>5 (p=.016), ASA>III (p=.015), the presence of systemic complications (p<.001), and ambulatory status after the intervention (p=.028). In the periPR group, it was associated with ASA>III (p=.02), surgical time (p=.024), presence of systemic complications (p=.012), local complications (p=.042), and worsening of ambulatory status (p=.022). No statistically significant risk factors were identified in the periCL group.
In this study, the type of treatment used (conservative, osteosynthesis, or revision) was not statistically significantly associated with mortality in either the overall or stratified analyses.
Logistic regression did not identify any significant odds ratios for the risk factors assessed.
Functional outcomesBefore the fracture, 38 patients were ambulatory, 45 required a cane or walker, and 17 were non-ambulatory. After treatment, the percentage of non-ambulatory patients increased from 17% to 37% (p<.001) (Fig. 1). Loss of ambulation was associated with prolonged preoperative hospital stays (p=.056), higher ASA scores (p≤.001), and ICC>5 (p=.06). It was not associated with either the type of fracture or the treatment used.
Discussion
The 30-day mortality rate following a peri-implant femoral fracture reported in our study is 12%, which is higher than the 7.8% reported in 2024 by the Spanish National Hip Registry for hip fractures during the same period.15 Annual mortality after a hip fracture ranges from 20% to 40%, while in our cohort, the one-year mortality rate after a peri-implant femoral fracture was 28%.2Table 2 compares the values presented in this study with those of other previous studies.3,8–12,16–21 Unlike other studies that focus on a single fracture type (peri-PC, periPR, or peri-CL), this study presents the overall mortality rate for PPFS, stratified by group.
Current state of the literature on PPF.
| n | Mean age | ASAa | CCIa | Fracture type | Time to surgery | Mortality 30 days/1 year | |
|---|---|---|---|---|---|---|---|
| Griffiths et al. (2013) | 60 | 78 | 3 | − | periPC | 4 | 10%/− |
| Shields et al. (2014) | 113 | 83 | − | 2.8 | periPC/periPR | − | −/17.1% |
| Füchtmeier et al. (2015) | 121 | 75.5 | 3 | − | periPC | 1.3 | 1.6%/13.2% |
| Moreta et al. (2015) | 59 | 79 | 3 | − | periPC | − | −/24% |
| Drew et al. (2016) | 291 | 76 | − | 4.3 | periPC | − | −/13% |
| Gitajn et al. (2017) | 203 | 76.6 | 3 | 5.9 | periPC | 3.1 | −/13% |
| Jennison et al. (2018) | 32 | 76.5 | 3 | 0–1: 53%2–3: 31%>5: 15.6% | periPC | 4.4 | 12.5%/28.1% |
| Finlayson et al. (2019) | 189 | 79 | 3 | − | periPC | 5.8 | 2.1%/11.6% |
| Moreta et al. (2021) | 107 | 81 | 3 | 2.1 | periPC | 3.8 | 9.3%/22.4% |
| Lützner et al. (2023) | 626 | 78.8 | − | − | periPR | − | −/13.2% |
| Andrés-Peir et al. (2024) | 965 | 83.5 | − | 5.7 | periPC | 9.3 | −/16.2% |
| Prevot et al. (2024) | 25 | 84.5 | II: 44%III: 56% | 2–3: 4%4–5: 64%6–7: 24%>7: 8% | periCL | − | 20%/28% |
| Shah et al. (2024) | 129 | periPC: 74periPR: 73.8 | − | − | periPC/periPR | − | periPC: 2.1%/17%periPR: 4.7%/14% |
| Aguado et al. (2025) | 262 | 262 | 3 | 5 | periPC Vancouver C | <48h: 196>48h: 56 | 6.7%/5.7% |
| Current study | 100 | 80.85 | 3 | 5.86 | periCL/periPC/periPR | 6.58 | 12%/28% |
Our initial hypothesis proposed the existence of independent risk factors for each fracture type. Most of the identified factors were similar across the three cohorts and correspond to non-modifiable factors, such as age, ASA score>III, and ICC score>3, findings that are consistent with previous literature previa.3,8,11,17,18 In the peri-PR group, we were able to identify surgical time and local complications as being associated with a poorer outcome.
The main contribution of our study is the identification of the need for blood transfusions, the number of transfusions, and the loss of ambulatory capacity in relation to the annual mortality of PPF. Unlike the other factors, both are potentially modifiable through perioperative optimisation.
Evidence on transfusion and complications in periprosthetic fractures is limited. Griffiths et al. observed that in elderly patients with periprosthetic hip fractures, the blood transfusion rate was 50% higher in the group without complications compared to patients with complications.3 However, the work by Haughom et al. blood transfusions are linked to a higher incidence of minor complications.22 Only the group by Richard et al. associated the need for blood transfusions with mortality in the PPF.23
Based on our results and the conclusions of these authors, our hypothesis is that the need for transfusions reflects the importance of optimising the patient's condition before surgery and of surgical haemostasis. Regarding the latter, there is growing interest in antibrinolytic agents, such as tranexamic acid (TXA) and epsilon-aminocaproic acid (ɛACA), to minimise blood loss during primary hip replacement. However, their efficacy in PPF has not yet been demonstrated.24,25
Although other studies highlight the loss of ambulatory capacity after PPF8,9,16 ours is the first to link ambulatory status with mortality. Inability to walk before surgery or increased postoperative dependence were associated with overall mortality at 30 days and 1 year, as well as in the peri-PC and peri-PR groups. Loss of walking ability was associated with ASA score>III and a postoperative stay >96h, with no relation to fracture type or treatment performed.
Although we did not identify an association between surgical delay (>96h) and mortality, we did find a statistically significant association between surgical delay and loss of ambulation: the latter was statistically significantly associated with higher mortality. However, the literature directly associates surgical delay with higher mortality and delayed functional recovery.3,26 Bhattacharyya et al. suggested that, in PPF, surgery should be performed within the first 48h to reduce mortality.27 In our cohort, the mean time to surgery was 7.04±5.29 days. One explanation for this delay may be that 14 patients had hospital stays longer than 10 days during 2020, when surgical activity at our centre was limited due to the COVID-19 pandemic.
The impact of the type of treatment on mortality remains debated. Bhattacharyya et al. reported higher mortality rates in patients treated with osteosynthesis (33%) than in those undergoing prosthetic revision (12%).27 However, other studies found no significant differences in mortality according to the type of treatment.8,16,17 In our study, conservative treatment was not associated with higher mortality, a greater number of systemic complications, or loss of ambulatory capacity, although in our case this could be due to an insufficient sample size.
Surprisingly, despite the high prevalence (66%) of osteoporosis, only 20% of our cohort received treatment before the fracture, and only 25% after the fracture. Less than 10% of patients were referred to the rheumatology service. This highlights the need for osteoporosis prevention strategies, recommending comprehensive fracture prevention programs to reduce recurrence and improve long-term outcomes.1
LimitationsThis study has several limitations. Its retrospective design and relatively short follow-up limit its universality. Treatment was not standardised and depended on the surgeon's decision and the patient and fracture characteristics. However, other authors point to this characteristic as an advantage, since the study better represents the daily activity of the centres. Although this study focuses on morbidity and mortality, functional scales were not used to assess recovery.
The authors of this study acknowledge that almost half of the cases are peri-PC fractures, with older patients than the other groups and predominantly ASA IV, which implies a high number of revisions of the prosthetic component (open surgery) and longer surgical times. It is logical that these patients require a greater number of blood transfusions and also have a higher mortality rate. Although our results align with the hypotheses of Haughom et al. and Richard et al.,22,23 our conclusions should be considered within the context of our cohort and should not be generalised to all series.
Finally, there are few studies that group peri-implant and periprosthetic femoral fractures, making external comparisons difficult. Also, due to the heterogeneous sample size, each subgroup (peri-PC, peri-CL, and peri-PR) has a limited sample size, resulting in low statistical power. This may have hindered the identification of true associations (Type II error) and influenced our conclusions.
ConclusionsAdvanced age, a high ICC score, ASA score>III, and the presence of systemic complications are common risk factors for PPF mortality. Most risk factors are common to the peri-peri-PC, peri-PR, and peri-CL subtypes. Surgical time and local complications were associated with a worse outcome only in the peri-PR group. Our new finding that the need for blood transfusions and loss of ambulation, both modifiable through perioperative optimisation, are associated with higher mortality in PPF. Future studies are needed to determine whether transfusions reflect insufficient preconditioning or surgical blood loss, and to evaluate the impact of rehabilitation on functional recovery and survival.
Level of evidenceLevel of evidence II.
Ethical considerationsThis study received ethical approval from the Scientific Research Ethics Committee of the Hospital Universitari i Politècnic La Fe (approval 2024-819-1) on November 20, 2024. This is a retrospective study approved by the IRB; all patient information was anonymised, and informed consent was not required. Patient data will not be shared with third parties.
Declaration of generative AI and AI-assisted technologies in the writing processDuring the preparation of this study, the authors used ChatGPT (OpenAI) to assist in the review and translation of texts. After using this tool/service, the authors reviewed and edited the content as necessary and assume full responsibility for the content of the published article.
FundingThe authors received no funding for the research, authorship and/or publication of this article.
Conflict of interestsThe authors declare no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
