Prediction of recovery after hip arthroplasty in elderly patients with femoral neck fractures based on decision tree model

doi:10.21203/rs.3.rs-2305388/v1

Download PDF

Research Article

Prediction of recovery after hip arthroplasty in elderly patients with femoral neck fractures based on decision tree model

https://doi.org/10.21203/rs.3.rs-2305388/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

To investigate the predictive value of the decision tree model for the recovery of femoral neck fractures after hip arthroplasty in elderly patients. A total of 206 elderly patients with femoral neck fractures who received surgeries in our hospital from January 2019 to June 2020 were recruited as subjects. Six months after the operation, they were divided into a good recovery group (Harris score ≥ 70) and a poor recovery group (Harris score < 70) according to the Harris Hip Score. General data, surgical conditions, and postoperative conditions were observed in the two groups. Python language was utilized to construct the decision tree model for postoperative recovery predictions in elderly patients with femoral neck fractures and its performance was verified. After 6 months of follow-up, 3 cases were excluded and 203 cases were finally included. Among them, 158 cases in the good recovery group accounted for 77.83% and 45 cases in the poor recovery group accounted for 22.17%. There were significant differences in age, Charlson comorbidity index, Mini-Mental State Examination score, MNA-SF, FI-CGA score, postoperative weight-bearing time, and social support rating scale score between the two groups (P < 0.05). There was no significant difference in sex and fracture site between the two groups (P > 0.05). Decision tree analysis exhibited that the MNA-SF score was an important factor affecting the postoperative recovery of hip fractures. The best parameters obtained were used for internal verification of the included subjects, and the results demonstrated that the accuracy rate of the model was 88.18%; the sensitivity was 93.33%; the specificity was 86.71%; the positive predictive value was 66.67%; the negative predictive value was 97.86%. The construction of the decision tree model can better exhibit the factors affecting the postoperative recovery of elderly patients with femoral neck fractures, and nutritional status is the most important factor affecting postoperative recovery.

femoral neck fracture

postoperative recovery

hip joint function

decision tree model

nutritional status

The hip joint is the largest ball-and-socket joint in the body. Hip fractures can cause uneven articular surfaces, severe hip pain, and impaired joint mobility. Previous studies^[1–2] have shown that 20–30% of hip fracture patients died within 1 year after injury. A survey^[3] in the United States exhibited that by 2015, more than 250,000 people were hospitalized for hip fractures, and the proportion will increase by 12.00% by 2030. Studies by Chinese scholars Liu et al.^[4] and Zha et al.^[5] have displayed that the proportion of hip fractures will increase with the development of population aging. Although surgery can restore and relieve fracture pain, correct fracture deformity, and improve hip joint function, some patients present poor postoperative recovery due to various factors. The key points are to predict the influencing factors of postoperative recovery in such patients and to take reasonable measures to prevent and treat them. Although some previous studies have analyzed the factors influencing postoperative recovery in hip fracture patients, most of them are based on Logistic regression analysis, focusing on the independent effect of a single factor, which is prone to bias when there are multicollinearity data. The problem of collinearity between independent variables and the interaction of multiple factors needs to be further addressed. The decision tree is a method for classification prediction in data mining. The decision tree model focuses more on the prediction effect of the model as a whole. The study by Liang et al.^[6] confirmed that the decision tree model can predict the risk factors of refracture in elderly patients with osteoporosis in Xinjiang Uygur Autonomous Region of China. Taken together, this study further established a postoperative recovery prediction model for elderly patients with femoral neck fractures through decision tree construction to seek relevant guidance that can predict the postoperative recovery of femoral neck fractures in elderly patients and to provide guidance and theoretical basis for clinical work.

2.1 Clinical data

206 hip fracture patients admitted to our hospital from January 2019 to June 2020 were enrolled as subjects. Inclusion criteria: (1) The diagnosis of hip fractures^[7] exhibits typical clinical manifestations such as swelling and deformity of the affected hip, local tenderness, axial percussion pain, palpable bone friction feeling, and dysfunction; femoral neck fracture confirmed by X-ray or CT scan; (2) unilateral hip fractures; (3) fracture surgery for the first time. Exclusion criteria: (1) pathological fracture; (2) fracture caused by high-energy injury; (3) combined with chronic inflammatory disease and malignant tumor; (4) combined with fractures of other parts; (5) open hip fractures. This study was approved by the hospital ethics committee. The patients and their family members signed informed consent.

2.2 Therapeutic schedule

After admission, all patients were operated on by the same group of doctors according to the results of preoperative X-ray and CT imaging examinations.

2.3 Data collection

We collected baseline data (age, sex, body mass index, fracture site, Charlson comorbidity index, cognitive function, and self-care ability), surgical conditions, laboratory tests (preoperative hemoglobin, albumin, C-reactive protein, anesthesia method, operation time, surgical method, surgical approach, intraoperative blood loss, and intraoperative blood transfusion volume) and postoperative conditions (postoperative weight-bearing time, walking aids, and social support scale) of patients of the two groups.

(1) Preoperative cognitive function: The Mini-Mental State Examination (MMSE) ^[8] with a maximum score of 30 was employed to assess orientation, registration, attention, calculation, recall, and language. The higher the score, the better the cognitive function is. 27–30 points are considered normal cognitive function. (2) Preoperative self-care ability: The activities of daily living (ADL) score ^[9] was used with a total score of 56 points. The self-care ability was evaluated from two aspects: physical self-care ability and instrumental ADL. The low scores indicate low self-care ability. (3) Comorbidity index: According to the existing and underlying diseases, the scoring criteria and principles of the Charlson comorbidity index are strictly followed ^[10], and the score is based on the latest International Classification of Diseases.(4) Nutritional status: The nutritional status was assessed by the Mini Nutritional Assessment Short Form (MNA-SF) ^[11], with a total score of 14 points. High scores indicate good nutritional status. (5) Frailty index: The frailty index based on a comprehensive geriatric assessment (FI-CGA) was utilized to evaluate the frailty ^[12], with a total score of 55 points. The higher the score, the more frailty is. (6) Social support rating scale: Social support rating scale (SSRS) ^[13] contains 14 items, with a total score of 64 points. The higher the score, the higher the degree of social support is.

2.4 Model construction

The construction method of decision tree model: Python 3.7 (windows 64) and sklearn.tree decision tree package were utilized for analysis. The steps of the regression decision tree algorithm are as follows. (1) Feature selection: The clinical data included in the univariate analysis were used as model features and the Gini coefficient was used for feature selection. (2) Decision tree generation: During tree generation, among all possible features and all their possible cut-off points, the feature with the smallest Gini coefficient and its corresponding cut-off point are selected as the optimal feature and optimal cut-off point. (3) Pruning: analyzing the decision boundaries of different parameters, and obtaining the optimal settings of pruning parameters such as min_samples_split, min_samples_leaf, max_leaf_nodes, max_features, and max_depth.

2.5 Follow-up

After discharge, the patients were followed up for 6 months. Regular review and health education were conducted in outpatient clinics every month, and the rest were followed up by WeChat or telephone to understand the outcomes of patients and provide rehabilitation guidance. The functional recovery was evaluated 6 months after the operation. They were assigned to two groups according to the Harris hip scores (using Harris hip scale ^[14]: with ten items covering four domains; the highest possible score is 100). Harris hip score ≥ 70 was considered a good recovery group; Harris hip score < 70 was considered a poor recovery group.

2.6 Statistical analysis

Data were analyzed using SPSS 22.0 statistical software. Measurement data were expressed as (mean ± SD), and two independent samples t-test was applied for comparison between groups. Count data were expressed as [(n)%], and χ² test was utilized to compare differences between groups. Python 3.7 (windows 64) and sklearn.tree decision tree model were used for analysis. A value of P < 0.05 was considered statistically significant.

3.1 Follow-up results

By the final follow-up on December 31, 2020, three cases lost to follow-up were excluded and 203 cases were included, with 158 cases (77.83%) in the good recovery group and 45 cases (22.17%) in the poor recovery group.

3.2 Comparison of baseline data of patients between the two groups

Significant differences in age, time from trauma to operation, Charlson comorbidity index, MMSE score, ADL score, MNA-SF, and FI-CGA score were detected between the two groups (P < 0.05). There were no significant differences in sex, fracture site, and body mass index between the two groups (P > 0.05), as shown in Table 1.

Table 1

Comparison of baseline data between the two groups [(n),
±s]
Baseline data		Good recovery group (n = 158)	Poor recovery group (n = 45)	t/χ²	P
Sex [n(%)]				3.316	0.069
	Male	57(36.08)	23(51.11)
	Female	101(63.92)	22(48.89)
Age (year)		68.08 ± 8.59	72.52 ± 9.86	2.739	0.008
Body mass index (kg/m²)		23.31 ± 1.54	23.04 ± 1.49	1.183	0.241
Fracture site [n(%)]				1.104	0.294
Left		60(37.97)	21(46.67)
Right		98(62.03)	24(53.33)
Time from trauma to operation (h)		46.77 ± 5.98	49.36 ± 6.39	2.432	0.018
Charlson comorbidity index		2.01 ± 0.26	2.14 ± 0.37	2.207	0.031
Preoperative MMSE score (point)		27.29 ± 1.01	26.86 ± 1.29	2.063	0.043
Preoperative ADL score (point)		19.58 ± 2.51	20.71 ± 3.14	2.220	0.030
MNA-SF score (point)		12.01 ± 0.59	10.98 ± 1.52	4.451	< 0.001
FI-CGA score (point)		25.44 ± 3.57	28.24 ± 4.03	4.214	< 0.001

3.3 Comparison of surgical conditions and laboratory examinations between the two groups

There were significant differences in preoperative fasting blood glucose, preoperative hemoglobin, preoperative albumin, operation time, intraoperative blood loss, and intraoperative blood transfusion volume between the two groups (P < 0.05). There was no significant difference in C-reactive protein, anesthesia method, anesthesia classification, surgical approach, and surgical method (P > 0.05), as shown in Table 2.

Table 2

Comparison of surgical conditions and laboratory examinations between the two groups
Surgical conditions and laboratory examinations	Good recovery group (n = 158)	Poor recovery group (n = 45)	t/χ²	P
Laboratory examinations
Preoperative C-reactive protein (mg/L)	25.24 ± 5.27	27.02 ± 6.37	1.715	0.091
Preoperative fasting blood glucose	6.09 ± 1.01	6.49 ± 1.15	2.113	0.039
Preoperative hemoglobin (g/L)	127.86 ± 15.75	121.52 ± 16.72	2.273	0.026
Preoperative albumin (g/L)	31.02 ± 3.11	30.46 ± 3.15	2.209	0.028
Surgical conditions
Anesthesia method [n(%)]			0.021	0.885
General anesthesia	26(16.46)	7(15.56)
Intraspinal anesthesia	132(83.54)	38(84.44)
Anesthesia classification [n(%)]			0.078	0.781
Grade Ⅰ	20(12.66)	5(11.11)
Grade Ⅱ	138(87.34)	40(88.89)
Surgical method [n(%)]			2.212	0.136
Femoral head arthroplasty	65(41.14)	13(28.89)
Total hip arthroplasty	93(58.86)	32(71.11)
Surgical approach [n(%)]			0.030	0.863
Posterolateral approach	69(43.67)	19(42.22)
Anterolateral approach	89(56.33)	26(57.78)
Operation time (h)	1.25 ± 0.12	1.30 ± 0.15	2.056	0.044
Intraoperative blood loss (mL)	341.82 ± 98.24	432.45 ± 99.87	5.390	< 0.001
Intraoperative blood transfusion volume(mL)	512.26 ± 101.27	621.27 ± 110.26	5.955	< 0.001

3.4 Comparison of postoperative conditions between the two groups

The time of starting weight-bearing after the operation, postoperative complications, walking aids, and SSRS scores were significantly different between the two groups (P < 0.05), as shown in Table 3.

Table 3

Comparison of postoperative conditions between the two groups [(n),
±s]
Postoperative conditions	Good recovery group (n = 158)	Poor recovery group (n = 45)	t/χ²	P
Time of starting weight-bearing after operation (d)	5.35 ± 1.52	6.98 ± 1.68	5.861	< 0.001
Postoperative complications [cases(%)]			4.855	0.028
Yes	40(25.32)	19(42.22)
No	118(74.68)	26(57.78)
Walking aids [n(%)]			11.960	< 0.001
Yes	108(68.35)	18(40.00)
No	50(31.65)	27(60.00)
SSRS score (point)	39.89 ± 7.89	31.75 ± 5.64	7.758	< 0.001

3.5 Decision tree model for postoperative recovery of elderly patients with femoral neck fractures

By constructing decision tree models with different layers for analysis, the results exhibited that the five-layer model had the highest accuracy, as shown in Fig. 1. The constructed models included six features, including MNA-SF score, SSRS score, time of starting weight-bearing after the operation, operation time, Charlson comorbidity index, and walking aids, as shown in Fig. 2.

3.6 Evaluation of the efficacy of the decision tree model for postoperative recovery in elderly patients with femoral neck fractures

The best parameters obtained were used for internal validation of the included patients, and the results displayed that the model had an accuracy of 88.18%, a sensitivity of 93.33%, a specificity of 86.71%, a positive predictive value of 66.67%, and a negative predictive value of 97.86%.

The continuous acceleration of the aging process has made hip fracture among the top ten causes of disability worldwide[15], and hip fracture has become a global public health problem. Although the progress in surgical treatment is remarkable, a series of problems can affect postoperative recoveries, such as difficult predicted complications, surgical site infections, and refractures. In the study by Mingli et al.[16], 18.7% still had poor recovery after surgical treatment. In the study by Xia et al.^[17], the excellent and good rate of recovery 1 year after surgery was only 77.11% (64/90). In this study, the good recovery rate after femoral neck fracture was 77.83%, which is consistent with the data of the above scholars, indicating that the postoperative outcome of hip fracture is relatively severe. The factors affecting the outcome of patients with femoral neck fracture should be paid attention to and identified, and the rate of good recovery should be improved after femoral neck fracture.

Nutritional status is both an inducement and a prognostic factor for femoral neck fractures. Malnutrition can increase the risk of falls, diminish bone mass, and increase the risk of refracture[18]. During the healing of femoral neck fractures, multiple pathophysiological changes are strongly associated with nutritional status. Preoperative nutritional reserve directly affects immune function and healing ability and has a direct impact on fracture recovery. This study exhibited that the MNA-SF score of the good recovery group after hip fracture was higher than that of the poor recovery group, which also indicated that preoperative nutritional status could affect the postoperative recovery of hip fracture. It is suggested that perioperative nutritional management of patients should be done well in the clinic, attention should be paid to the nutritional status before the operation, adequate nutritional assessment should be carried out, and nutritional support should be actively provided.

In addition to nutritional status, postoperative activity also affects the recovery after fractures. Although there are good reasons to apply immobilization measures in the early stage of fracture surgery, neglect may lead to secondary dysfunction and affect postoperative recovery [19–20]. Biomechanical and animal studies have shown that the stress of early adaptation to the fracture site has a positive effect on postoperative recovery after fractures. In the present study, the time of starting weight-bearing in the good recovery group after hip fracture surgery was shorter than that in the poor recovery group, indicating that weight-bearing after surgery could improve the postoperative recovery of hip fracture. Because the postoperative healing of fractures is affected by active factors such as blood supply, mechanical stress stimulation, and the degree of injury, mechanical stimulation is the main factor. On the one hand, it can stimulate osteoblast proliferation, differentiation, and secretion of extracellular matrix. On the other hand, by acting on the bone microenvironment, it can promote the release of local growth factors, which is conducive to angiogenesis at the fracture end. Therefore, early postoperative weight-bearing can create favorable conditions for the early recovery of fractures.

The development of the biopsychosocial medical model makes modern medical work no longer limited to diseases, but to recognize the relationship between human health and diseases from a holistic and systematic perspective, gradually paying attention to the psychology, behavior, and sociality of patients. Due to the behavioral alterations induced by hip fractures, these patients are worried about their future adaptability. Furthermore, when the body is injured, the spirit and mind are also severely hit beyond the stress threshold, which affects patient’s social role and is not conducive to postoperative recovery of fractures. The studies by Smith et al.[21]and Gilboa et al.[22] exhibit that social support is one of the most potential resources for hip fracture patients in the process of coping with fracture and treatment. Effective social support can enable patients to overcome negative coping emotions. Koivunen et al.[23] found that social support for independent living after fracture in elderly patients is a predictor of mortality. Sale et al.[24] concluded that physical and emotional support provided by nursing staff can improve the outcome of patients with insufficiency fractures. In this study, the SSRS score of the good recovery group after femoral neck fracture was higher than that of the poor recovery group, confirming the importance of social support. Social support is the psychological and material support or assistance that an individual receives from society, family, and other aspects under stress. As an important resource that can be used in nursing work, it has a positive relationship with human health and has both a stimulatory effect and a direct independent protective effect. Therefore, nursing workers should establish a sound social support system to help patients effectively use the social support network, and help them actively seek more effective and multi-faceted support and help, which can mobilize patients to consciously adjust their psychological pressure and actively participate in treatment and nursing with a positive attitude to speed up recovery.

The advent of the era of big data has made data mining and machine learning more and more extensively utilized in the medical field. The decision tree model algorithm can deal with the interaction between variables, analyze the specific form of a factor acting on each subgroup, and provide a reasonable analysis method for predictions. There are many factors affecting the postoperative recovery of hip fractures. In this study, the constructed decision tree model algorithm contained six features, including MNA-SF score, SSRS score, time of starting weight-bearing after the operation, operation time, Charlson comorbidity index, and walking aids. The five-layer model had the highest accuracy. However, FI-CGA, fasting blood glucose, hemoglobin, albumin, age, intraoperative blood loss, and blood transfusion volume were not included in the decision tree model. This is probably associated with the following factors: MNA-SF scale consists of somatometric measurement, diet evaluation, overall evaluation, and self-evaluation; compared with FI-CGA, fasting blood glucose and hemoglobin can fully reflect the nutritional status of patients; intraoperative blood loss is timely treated. Decision tree analysis revealed that the root node of the decision tree model was the MNA-SF score, which indicated that the MNA-SF score had the highest correlation with the postoperative recovery of femoral neck fractures, and was the most important influencing factor. Internal validation results of patients included in the study using the best parameters obtained demonstrated that the model had an accuracy of 88.18%, a sensitivity of 93.33%, a specificity of 86.71%, a positive predictive value of 66.67%, and a negative predictive value of 97.86%. The decision tree model intuitively presented the importance of the influencing factors such as MNA-SF score, SSRS score, and the time of starting weight-bearing after surgery on the postoperative recovery of femoral neck fractures and the interaction between variables, which is conducive to better understanding the relationship between various factors and the relationship with the postoperative recovery of femoral neck fractures, and provide a theoretical basis for the development of clinical nursing work.

In summary, preoperative nutritional status is the most important factor affecting the postoperative recovery of elderly femoral neck fractures, and the establishment of a decision tree model can better show various factors affecting postoperative recovery of femoral neck fractures in elderly patients. This study also has some limitations. The sample size is small. Only one type of fracture, femoral neck fractures, is discussed, and only the included cases are used for internal verification, which may lead to bias in the results. The amount of data will be further expanded in the future for external validation to optimize model parameters.

MMSE: Mini-Mental State Examination; ADL: Activities of daily living; MNA-SF: Mini Nutritional Assessment Short Form; FI-CGA: Frailty index based on a comprehensive geriatric assessment; SSRS: Social support rating scale.

Acknowledgements

Not applicable

Authors’ contributions

Huaping Chen: Conceptualization, Methodology, Software, Writing-review & editing. Xiao Xu: Conceptualization, Methodology, Software, Editing. Jingjing Xia: Visualization, Investigation. Huiping Sun: Software, Validation. All authors read and approved the final manuscript.

Funding

Zhejiang Province Medical and Health Science and Technology Project, No.: 2021KY864.

Availability of data and materials

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate

This study was approved by the Medical Ethics Committee of the Affiliated Hangzhou First People's Hospital. All methods were carried out in accordance with relevant guidelines and regulations. Informed consent was obtained from all participants.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Author details

Affiliated Hangzhou First People's Hospital Zhejiang University School of Medicine, Hangzhou 310006, China

Johansen A, Grose C, Havelock W. Hip fractures in the winter - Using the National Hip Fracture Database to examine seasonal variation in incidence and mortality. Injury. 2020. 51(4): 1011-1014.
Yoo JI, Kim HY, Jang S, Lee YK, Ha YC. Risk factors for subsequent hip fractures and fatality after an initial hip fracture in Korea: using nationwide claims data. Arch Osteoporos. 2020. 15(1): 165.
Lewiecki EM, Wright NC, Curtis JR, et al. Hip fracture trends in the United States, 2002 to 2015[J]. Osteoporos Int, 2018,29(3):717-722.
Liu ZM, Lü X, Liu JY, et al. Epidemiological distribution characteristics of 2342 cases of hip fracture: a single center analysis [J]. Zhongguo Zuzhi Gongcheng Yanjiu,2020,24(32):5085-5091.
Zha GC, Sun JY, Chen XY, et al. Clinical epidemiology features of the acetabular fracture in patients with different ages[J]. Zhonghua Guke Zazhi,2016,36(18):1175-1184.
Liang LL, Zheng Hui, Wang Chengwei, et al. Risk factors of re-fracture in senile osteoporosis patients in Xinjiang region based on decision tree model and logistic regression analysis [J]. Zhonghua Gu Yu Guanjie Waike Zazhi,2019,12(11):873-876,881.
Zhao YP, Chen XP. Surgery (Volume II) [M]. Beijing: People's Medical Publishing House,2015:887-888.
Jia JP. Guidelines for Diagnosis and Treatment of Dementia and Cognitive Impairment in China (2015 Edition) [M]. Beijing: People's Medical Publishing House,2016:27-29.
Wang YL, Lv X, Guo SS, et al. Design of a method for assessing the situation of self-care ability in daily life [J]. Zhonghua Wuli Yixue Yu Kangfu Zazhi,2018,40(11):840-843.
Frenkel WJ, Jongerius EJ, Mandjes-van Uitert MJ, et al. Validation of the Charlson Comorbidity Index in acutely hospitalized elderly adults: a prospective cohort study[J]. J Am Geriatr Soc, 2014,62(2):342-346.
Kaiser MJ, Bauer JM, Ramsch C, et al. Validation of the Mini Nutritional Assessment short-form (MNA-SF): a practical tool for identification of nutritional status[J]. J Nutr Health Aging, 2009,13(9):782-788.
Jones DM, Song X, Rockwood K. Operationalizing a frailty index from a standardized comprehensive geriatric assessment[J]. J Am Geriatr Soc, 2004,52(11):1929-1933.
Wang XD, Wang XL, Ma H. Mental Health Rating Scale Handbook (Updated) [J]. Beijing: Zhongguo Xinliweisheng Zazhi,1999:318-320.
Xia ZL, He B, Fan MX, et al. Reliability and Validity of Chinese Version of Oxford Hip Score [J]. Huli Xuebao,2012,19(5):23-25.
Abeygunasekara T, Lekamwasam S, Lenora J, et al. Current incidence and future projections of fragility hip fractures in Sri Lanka[J]. Arch Osteoporos, 2020,15(1):178.
Mingli F, Huiliang S, Guanglei C, et al. A clinical study on arthroplasty for failed internal fixation of hip fractures and review of literature[J]. Pak J Med Sci, 2017,33(4):798-803.
Xia CY. Analysis on postoperative recovery of joint function and influencing factors in elderly patients with fracture of the hip joint [J]. Yiyao Qianyan,2019,9(27):52-53.
Lins Vieira NF, da Silva Nascimento J, do Nascimento CQ, et al. Association between Bone Mineral Density and Nutritional Status, Body Composition and Bone Metabolism in Older Adults[J]. J Nutr Health Aging, 2021,25(1):71-76.
Consigliere P, Iliopoulos E, Ads T, et al. Early versus delayed weight bearing after surgical fixation of distal femur fractures: a non-randomized comparative study[J]. Eur J Orthop Surg Traumatol, 2019,29(8):1789-1794.
Kuru T, Olçar HA. Effects of early mobilization and weight bearing on postoperative walking ability and pain in geriatric patients operated due to hip fracture: a retrospective analysis[J]. Turk J Med Sci, 2020,50(1):117-125.
Smith TO, Dainty JR, MacGregor A. Trajectory of social isolation following hip fracture: an analysis of the English Longitudinal Study of Ageing (ELSA) cohort[J]. Age Ageing, 2018,47(1):107-112.
Gilboa Y, Maeir T, Weber A, et al. Predictors of community reintegration and quality of life after hip fracture among community-dwelling older adults[J]. Int J Rehabil Res, 2019,42(3):234-239.
Koivunen K, Sillanpää E, von Bonsdorff M, et al. Living alone vs. living with someone as a predictor of mortality after a bone fracture in older age[J]. Aging Clin Exp Res, 2020,32(9):1697-1705.
Sale J, Frankel L, Paiva J, et al. Having caregiving responsibilities affects management of fragility fractures and bone health[J]. Osteoporos Int, 2020,31(8):1565-1572.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Prediction of recovery after hip arthroplasty in elderly patients with femoral neck fractures based on decision tree model

Status:

Version 1

Abstract

Figures

1 Introduction

2 Data And Methods

2.1 Clinical data

2.2 Therapeutic schedule

2.3 Data collection

2.4 Model construction

2.5 Follow-up

2.6 Statistical analysis

3 Results

3.1 Follow-up results

3.2 Comparison of baseline data of patients between the two groups

3.3 Comparison of surgical conditions and laboratory examinations between the two groups

3.4 Comparison of postoperative conditions between the two groups

3.5 Decision tree model for postoperative recovery of elderly patients with femoral neck fractures

4 Discussion

Abbreviations

Declarations

References

Additional Declarations

Status:

Version 1