Abstract
Background
The ubiquity of hip fractures pose a substantial burden on public health services worldwide. There is widespread geographical variation in mortality rates and length of stay after hip fractures. The current study investigates both the predictors of; (1) one-year mortality and (2) length of hospital stay (LOS) in adults aged 60 years or older. We aim to identify the risk factors and quantify the extent of influence they have on both outcomes.
Methodology
A retrospective multi-center cohort study identified consecutively documented hip fractures between January 2013 and September 2018. A multivariate regression analysis of 603 patients was performed to determine independent factors affecting mortality and total LOS.
Results
The study sample included 603 patients with a total one-year mortality rate of 20.6% (n = 124). Predictors of mortality included; longer LOS, increasing age, inability to return to baseline mobility and comorbid burden. The mean overall LOS was 15.1 days, and 22.6 days in the mortality group. Predictors of increased LOS included; previous hip fractures, comorbid burden; diabetic, cerebrovascular disease and smokers. Return to baseline mobility status was associated with reduced LOS.
Conclusion
Patients with a longer length of stay, inability to return to baseline mobility status, higher ASA scores, previous hip fractures and longer time to surgery had a higher mortality rate. Determinants of a longer LOS include; increased time to surgery, impeded postoperative mobility status, fixation rather than joint replacement and comorbid burden. A multifaceted approach to preoperative optimization and postoperative recovery is crucial in order to address all possible modifiable factors.
Keywords: Hip fractures, Mortality, Length of stay, Elderly, Trauma, Predictors
1. Introduction
The ubiquity of hip fractures pose a substantial burden on public health services worldwide. The consequences of these fractures are perpetuated far beyond the isolated osseous discontinuity, often leading to morbid complications.1,2 Despite advancements in preoperative and perioperative care, high worldwide one-year mortality rates persist (10–35%), particularly among the elderly.2,3 Several resources have studied the risk factors for mortality, each focusing on recognizing various modifiable factors that might be targeted to reduce mortality rates.4, 5, 6, 7
Interestingly, there are marked distinctions in fractures rates among different ethnic groups and geographical locations that may be attributed to both environmental and patient factors.8,9 In addition, as life expectancy increases, the incidence of hip fractures similarly increases, with projections denoting an incidence that doubles every 10 years of life after the age of 50.1,10
Strategies to prevent mortality from hip fractures range from; initial prevention of hip fractures, to prompt inpatient medical and surgical management and post-discharge rehabilitation. Length of stay (LOS) has recently garnered interest, with conflicting results reporting an increase or decrease in mortality as length of stay increases. Further, there is a considerable variation in length of stay worldwide.11, 12, 13 Essentially, the principle approach currently adopted consists of identifying risk factors associated with major morbidity and mortality in order to minimize burden of disease and improve end-outcomes.
Data from several resources have highlighted some factors that predispose to higher mortality rates, however clear indicators and their mechanism of influence have not been confirmed.10 This is especially true for length of stay, where data was associated with both a higher and lower mortality rate with increased LOS.
The current study investigates both the predictors of; (1) one-year mortality and (2) length of stay (LOS) in adults aged 60 years or older. We aim to identify the risk factors and quantify the extent of influence they have on both outcomes. This study is reported in line with the STROCSS criteria.14
2. Methods
2.1. Study design and patient population
A total of 876 consecutive patients with documented hip fractures were retrospectively identified from two general hospitals and one tertiary orthopedic center between January 2013 and September 2018. Patients with incomplete data (n = 241), patients with bilateral simultaneous hip fractures (n = 6) and patients that died before discharge (n = 26) were excluded from the analysis. Leaving a total of 603 patients for the final analysis. Only patients that were discharged to home were included. Ethical approval was obtained from the local ethical committee (UID 2186)
The primary outcome measurements were one year mortality and total length of stay. Patients were further evaluated according to age, sex, mobility status, fracture type, mechanism of injury, American Society of Anesthesiology (ASA) score, time to surgery, surgical duration, intervention performed, previous hip fractures, return to baseline mobility and length of stay. Further, patients were assessed for medical comorbidities including; anemia, cerebrovascular disease, malignancy, anticoagulation, hypertension, diabetes mellitus, respiratory disease, cardiovascular disease, smoking status and cognitive dysfunction.
2.2. Statistical analysis
Statistical analysis was performed using R v 3.6.3. Counts and percentages were used to summarize the distribution of categorical variables. The mean ± standard deviation (SD) and the median/interquartile range [IQR] were used to summarize the distribution of continuous normal and non-normal variables, respectively. Chi-square test of independence was used to assess the association between various independent variables and one year mortality (yes vs. no). Binary logistic regression to construct a model that can be used to predict 1 year mortality. Model fit was assessed using AUC, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV). A Receiver operating characteristic curve (ROC) was used to demonstrate the discriminant power of the model. The area under curve (AUC) (C-index) of the model was used to quantify its predictive power with values > 0.7 indicating good discriminant ability.
Negative binomial regression was used to assess factors associated with length of stay. Negative binomial regression was used due to the skewed and over dispersed nature of the dependent variable (LOS). Hypothesis testing was performed at 5% level of significance.
3. Results
The study sample included 603 patients. Of these, 124 (20.6%) died within one year (Table 1). Results showed that higher age was associated with mortality (P < 0.001). The prevalence of fully mobile patients was lower in patients who died within one year than patients who survived past one year (59.7% vs. 78.7%, respectively, P < 0.001). Anemia, cerebrovascular accidents (CVA), anticoagulation, hypertension (HTN), cardiovascular disease (CVD), cognitive dysfunction, and diabetes mellitus (DM) were more prevalence in patients who died within one year (P < 0.05). Gender, malignancy, prevalence of respiratory diseases, mechanism of injury (MOI), fracture type, and smoking were not significantly different between groups (P > 0.05). American Society of Anesthesiology score (ASA) grades III and IV were more prevalent in patients who died within one year. Patients who died within one year were more likely to have longer time to surgery with 61.9% of the patients who died within one year spending >72 h before surgery. Surgical duration was not associated with mortality (P = 0.287). The intervention was associated with mortality with 22.6% of patients who died within one year requiring prosthesis. Return to baseline mobility was associated with better mobility (P < 0.001). Similarly, previous hip fractures was associated with higher mortality (P < 0.001). The average LOS was significantly higher in patients who died within one year (P < 0.001).
Table 1.
Descriptive statistics for the study sample.
| [ALL] | No | Yes | P Value | |
|---|---|---|---|---|
| N=603 | N=479 | N=124 | ||
| Age | 72.7 (7.68) | 71.7 (7.36) | 76.4 (7.77) | <0.001 |
| Sex: | 0.155 | |||
| Female | 270 (44.8%) | 222 (46.3%) | 48 (38.7%) | |
| Male | 333 (55.2%) | 257 (53.7%) | 76 (61.3%) | |
| Mobility Status - Pre injury: | <0.001 | |||
| Fully mobile | 451 (74.8%) | 377 (78.7%) | 74 (59.7%) | |
| Non-walker | 18 (2.99%) | 7 (1.46%) | 11 (8.87%) | |
| Walking aids | 134 (22.2%) | 95 (19.8%) | 39 (31.5%) | |
| Anemia: | 131 (21.7%) | 92 (19.2%) | 39 (31.5%) | 0.005 |
| CVA: | 48 (7.96%) | 26 (5.43%) | 22 (17.7%) | <0.001 |
| Malignancy: | 40 (6.63%) | 28 (5.85%) | 12 (9.68%) | 0.185 |
| Anticoagulation: | 137 (22.7%) | 90 (18.8%) | 47 (37.9%) | <0.001 |
| HTN: | 306 (50.7%) | 223 (46.6%) | 83 (66.9%) | <0.001 |
| Respiratory Disease: | 57 (9.45%) | 40 (8.35%) | 17 (13.7%) | 0.100 |
| CVD: | 177 (29.4%) | 119 (24.8%) | 58 (46.8%) | <0.001 |
| DM: | 273 (45.3%) | 195 (40.7%) | 78 (62.9%) | <0.001 |
| Smoker: | 163 (27.0%) | 122 (25.5%) | 41 (33.1%) | 0.113 |
| Cognitive Dysfunction: | 69 (11.4%) | 42 (8.77%) | 27 (21.8%) | <0.001 |
| Fracture: | 0.346 | |||
| Femoral neck | 285 (47.3%) | 228 (47.6%) | 57 (46.0%) | |
| Intertrochanteric | 284 (47.1%) | 221 (46.1%) | 63 (50.8%) | |
| Subtrochanteric | 34 (5.64%) | 30 (6.26%) | 4 (3.23%) | |
| Mechanism of injury: | 0.442 | |||
| FFS | 499 (82.7%) | 391 (81.4%) | 108 (87.1%) | |
| FFH | 55 (9.12%) | 48 (10.0%) | 7 (5.65%) | |
| RTA | 49 (8.13%) | 40 (8.35%) | 9 (7.26%) | |
| ASA Score: | <0.001 | |||
| I | 48 (7.96%) | 46 (9.60%) | 2 (1.61%) | |
| II | 256 (42.5%) | 224 (46.8%) | 32 (25.8%) | |
| III | 220 (36.5%) | 160 (33.4%) | 60 (48.4%) | |
| IV | 79 (13.1%) | 49 (10.2%) | 30 (24.2%) | |
| Time to surgery: | <0.001 | |||
| <48 Hours | 212 (35.9%) | 190 (40.2%) | 22 (18.6%) | |
| 48-72 Hours | 135 (22.8%) | 112 (23.7%) | 23 (19.5%) | |
| >72 Hours | 244 (41.3%) | 171 (36.2%) | 73 (61.9%) | |
| Surgery time: | 0.287 | |||
| <2 Hours | 288 (47.8%) | 223 (46.6%) | 65 (52.4%) | |
| >2 Hours | 315 (52.2%) | 256 (53.4%) | 59 (47.6%) | |
| Intervention: | 0.009 | |||
| Fixation | 507 (84.1%) | 414 (86.4%) | 93 (75.0%) | |
| Non-Operative | 9 (1.49%) | 6 (1.25%) | 3 (2.42%) | |
| Prosthesis | 87 (14.4%) | 59 (12.3%) | 28 (22.6%) | |
| Previous hip fractures: | 46 (7.63%) | 22 (4.59%) | 24 (19.4%) | <0.001 |
| Length of stay | 15.1 (10.5) | 13.2 (7.37) | 22.6 (16.0) | <0.001 |
| Return to baseline mobility (Post Injury): | 246 (40.8%) | 217 (45.3%) | 29 (23.4%) | <0.001 |
| LOS; length of stay, DM; diabetes mellitus, CVA; cerebrovascular accident, ASA; American Society of Anesthesiology score, HTN; Hypertension, FFH; Fall from height, FFS; Fall from standing height; RTA; Road traffic accident; CVD; Cardiovascular disease | ||||
Results showed that adding LOS resulted in an improvement in the predictive ability of the model (R2 = 0.373). ASA scores III (OR = 2.27, P < 0.05) and IV (OR = 5.66, P < 0.001) were associated with higher mortality than ASA grades I and II (Table 2). Higher age was associated with higher mortality (OR = 1.07, P < 0.001) which indicates that 1 year increase in age was associated with a 7% increase in mortality at one year. Diabetes (OR = 1.9, P = 0.009) and CVA (OR = 2.63, P = 0.008) were associated with higher mortality at one year. Previous hip fractures (OR = 2.38, P = 0.018) was associated with higher odds of mortality at one year while returning to baseline mobility was associated with lower odds of mortality at one year (OR = 0.54, P = 0.025). Higher LOS was associated with higher mortality at one year (OR = 1.08, P < 0.001)
Table 2.
Predictors of one-year mortality. Analysis was performed using binary logistic regresion.
| 12 Month Mortality | |||
|---|---|---|---|
| Predictors | Odds Ratios | CI | P Value |
| Intercept | 0.00 | 0.00–0.01 | <0.001 |
| Age [1 year increase] | 1.07 | 1.04–1.11 | <0.001 |
| DM [Yes vs. No] | 1.90 | 1.18–3.08 | 0.009 |
| CVA [Yes vs. No] | 2.63 | 1.29–5.35 | 0.008 |
| Time to surgery [> 72 h vs. less] | 1.22 | 0.75–1.98 | 0.043 |
| Return to baseline mobility [Yes vs. No] | 0.54 | 0.32–0.93 | 0.025 |
| Previous hip fractures [Yes vs. No] | 2.38 | 1.16–4.88 | 0.018 |
| LOS | 1.08 | 1.05–1.11 | <0.001 |
| ASA = I or II | Ref | ||
| ASA = III | 2.27 | 1.33–3.89 | 0.003 |
| ASA = IV | 5.66 | 2.86–11.21 | <0.001 |
| R2 | 0.373 | ||
| Analysis was performed using binary logistic regression LOS; length of stay, DM; diabetes mellitus, CVA; cerebrovascular accident, ASA; American Society of Anesthesiology score | |||
Negative binomial regression was used to assess factors associated with LOS. Variables were eliminated in a backward fashion till the final model was reached (Table 3). The model was validated using five-fold cross-validation. The root mean squared error for the final model was 9.94 days and the final model explained 36.7% of the variability in the dependent variable.
Table 3.
Predictors of length of stay (LOS) using a negative binomial regression analysis.
| Length Of Stay | |||
|---|---|---|---|
| Predictors | Incidence Rate Ratios | CI | P value |
| (Intercept) | 11.77 | 10.66–13.02 | <0.001 |
| DM [Yes vs. No] | 1.13 | 1.04–1.24 | 0.004 |
| Fracture | |||
| Femoral neck | Reference | ||
| Intertrochanteric | 1.13 | 1.03–1.25 | 0.008 |
| Subtrochanteric | 1.15 | 0.96–1.39 | 0.147 |
| Intervention | |||
| Fixation | Reference | ||
| Non-Operative | 1.96 | 1.43–2.75 | <0.001 |
| Prosthesis | 1.00 | 0.88–1.14 | 0.987 |
| Previous hip fractures [Yes vs. No] | 1.32 | 1.14–1.55 | <0.001 |
| Return to baseline mobility [Yes vs. No] | 0.74 | 0.68–0.81 | <0.001 |
| Smoker [Yes vs. No] | 1.18 | 1.07–1.29 | 0.001 |
| Time to surgery [> 72 h vs. ≤ 72 h] | 1.37 | 1.26–1.50 | <0.001 |
Diabetes was associated with higher LOS (Incidence rate ratio (IRR) = 1.13, P = 0.004) which indicates that the expected LOS in diabetic patients is higher by 13% than non-diabetic patients. Fracture type was also associated with LOS with higher expected LOS observed in patients with intertrochanteric (IRR = 1.13, P = 0.008) fracture than patients with femoral neck injury. The intervention showed a statistically significant association with higher LOS in non-operative patients than patients requiring fixation (IRR = 1.96, P < 0.001). Previous hip fractures were associated with higher LOS (IRR = 1.32, P < 0.001) while return to baseline mobility was associated with lower LOS (IRR = 0.74, P < 0.001). Smoking was associated with higher LOS (IRR = 1.18, P = 0.001) and higher time to surgery (>72 h) was associated with a 37% increase in the expected LOS (IRR = 1.37, P < 0.001). None of the remaining factors showed a statistically significant association with LOS.
4. Discussion
Despite medical and surgical advancements, the morbidity and mortality accompanying hip fractures has plateaued.15,16 In our dataset, we evaluated several factors that correlate with accumulative rates of morality and length of stay. In order to produce accurate predictors, our analysis was adjusted to diminish the effects of confounding factors. The latter, a common manifestation of similar studies analyzing numerous factors.
In our cohort, the overall incidence of mortality was 20.6%, within the range of previously described mortality rates.2,3,6 With regards to patient factors, several risk factors were implicated in greater mortality rates; increasing age, greater ASA scores, cognitive dysfunction, CVA and DM. These results are expectedly consistent with data from previous resources.4,17 Frailty and the plausibility of encumbrance by comorbidities appear to be culminating factors of mortality in this age group.
Pertinent technical factors associated with greater mortality were longer time to surgery (>72 h), the presence of a previous hip fractures and whether the patient returned to baseline mobility. Undoubtedly, a direct focus on confronting modifiable factors should take precedence. While a time-sensitive approach to preoperative optimization has been the standard of care, practitioners must likewise endeavor to address the fundamental components of the postoperative course. This is emphasized in the greater rates of mortality among patients who did not return to their baseline mobility status. An interesting finding in our results is the higher rate of mortality in patients with a longer length of hospital stay. This finding is consistent with data from previous studies Nikkel et al.,11 yet contends with results from Yoo et al.12 However, a consideration of where patients are discharged to; home or care-facility, is necessary for an accurate appraisal.
In our dataset, the overall mean LOS was 15.1 days, and 22.6 days in the mortality group. Previous resources have reported a mean LOS ranging from 8 to more than 30 days.11, 12, 13 Yet, the geographical variation in LOS worldwide may be confounded by varying postoperative weight bearing and rehabilitation protocols. Further, LOS may include patients admitted into rehabilitation centers for lengthy periods of time. Whether longer LOS is a prognostic indicator of mortality appears to be a contentious matter.
In our cohort, length of stay was increased in patients with previous hip fractures and those that did not return to baseline mobility status, possibly reflecting the loss of independent mobility. In addition time to surgery >72 h was significantly associated with a higher LOS, findings that are in line with data by Nikkel et al.11 Findings suggest that patients treated with arthroplasty/prosthesis, have a lower LOS, and may suggest an expedited recovery of mobility status. Finally, comorbidities and smoking status was significantly associated with a higher LOS. Due to the retrospective nature of the study, it was not possible to determine whether the comorbidities directly affected LOS, or occurred as consequence of a higher risk of complications necessitating longer stay. Nevertheless, it is worth mentioning that particular comorbidities may directly hinder probabilities of returning to baseline mobility; CVA and cognitive dysfunction.
There are however, certain factors that may affect the accuracy of factors affecting length of stay and mortality after hip fractures. Firstly; there exists considerable overlap between independent variables. For instance, comparing time to surgery and length of stay as predictors of mortality, as both factors are intertwined and may serve as confounding factors. On the other hand, standard care and logistic factors may have a role, this is the case when comparing where patients are discharged to; home or a rehabilitation center. On the other hand, our data demonstrates the relationship between mortality and length of stay in a relatively large sample size. Further it analyses each factor affecting both outcomes individually.
Previous resources have produced contentious results, especially surrounding length of stay and its’ impact on mortality. Yet clearly, numerous risk factors are intertwined, and efforts to disentangle the influence of each singular factor has proven challenging. Future efforts may benefit from prospectively uncovering whether the disabling outcomes are intrinsic to the fracture, or are essentially catalyzed by it.
5. Conclusion
The determinants of mortality and length of stay in the elderly after hip fractures may be challenging to extricate due to the multifaceted nature of both fracture and patient. Although it is potentially challenging to address all risk factors in the clinical setting, patients may benefit from a careful multidisciplinary approach towards reducing the burden of all modifiable factors.
Conflict of interest and authorship conformation form
o All authors have participated in (a) conception and design, or analysis and interpretation of the data; (b) drafting the article or revising it critically for important intellectual content; and (c) approval of the final version.
o This manuscript has not been submitted to, nor is under review at, another journal or other publishing venue.
o The authors have no affiliation with any organization with a direct or indirect financial interest in the subject matter discussed in the manuscript
Funding
No source of funding was obtained for this study.
Contributor Information
Ali Lari, Email: Dr.alilari@gmail.com.
Abdullah Haidar, Email: Ajhaidar@outlook.com.
Yasmen AlRumaidhi, Email: Alrumaidhi.y@gmail.com.
Mohammad Awad, Email: Al.mo.jawad@gmail.com.
Owayed AlMutairi, Email: Dr.owayed@gmail.com.
References
- 1.Zuckerman J.D. Hip fracture. Eng J Med. 1996;334:1519–1525. doi: 10.1056/NEJM199606063342307. [DOI] [PubMed] [Google Scholar]
- 2.Marks R. Hip fracture epidemiological trends, outcomes, and risk factors, 1970-2009. Int J Gen Med. 2010;3:1–17. Published 2010 Apr 8. [PMC free article] [PubMed] [Google Scholar]
- 3.Brauer C.A., Coca-Perraillon M., Cutler D.M., Rosen A.B. Incidence and mortality of hip fractures in the United States. JAMA. 2009;302(14):1573–1579. doi: 10.1001/jama.2009.1462. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Chang W., Lv H., Feng C., et al. Preventable risk factors of mortality after hip fracture surgery: systematic review and meta-analysis. Int J Surg. 2018 Apr;52:320–328. doi: 10.1016/j.ijsu.2018.02.061. Epub 2018 Mar 9. PMID: 29530826. [DOI] [PubMed] [Google Scholar]
- 5.Sheehan K.J., Williamson L., Alexander J., et al. Prognostic factors of functional outcome after hip fracture surgery: a systematic review. Age Ageing. September 2018;47(5):661–670. doi: 10.1093/ageing/afy057. [DOI] [PubMed] [Google Scholar]
- 6.Kanis J.A., Odén A., McCloskey E.V., et al. A systematic review of hip fracture incidence and probability of fracture worldwide. Osteoporos Int. 2012;23(9):2239–2256. doi: 10.1007/s00198-012-1964-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Pioli G., Barone A., Giusti A., et al. Predictors of mortality after hip fracture: results from 1-year follow-up. Aging Clin Exp Res. 2006 Oct;18(5):381–387. doi: 10.1007/BF03324834.PMID:17167302. [DOI] [PubMed] [Google Scholar]
- 8.Dhanwal D.K., Dennison E.M., Harvey N.C., Cooper C. Epidemiology of hip fracture: worldwide geographic variation. Indian J Orthop. 2011;45(1):15–22. doi: 10.4103/0019-5413.73656. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Nieves J.W., Bilezikian J.P., Lane J.M., et al. Fragility fractures of the hip and femur: incidence and patient characteristics. Osteoporos Int. 2010;21(3):399–408. doi: 10.1007/s00198-009-0962-6. [DOI] [PubMed] [Google Scholar]
- 10.Cooper C., Campion G., Melton L.J., 3rd Hip fractures in the elderly: a world-wide projection. Osteoporos Int. 1992 Nov;2(6):285–289. doi: 10.1007/BF01623184.PMID:1421796. [DOI] [PubMed] [Google Scholar]
- 11.Nikkel L.E., Kates S.L., Schreck M., Maceroli M., Mahmood B., Elfar J.C. Length of hospital stay after hip fracture and risk of early mortality after discharge in New York state: retrospective cohort study. BMJ. 2015 Dec 10;351:h6246. doi: 10.1136/bmj.h6246. PMID: 26655876; PMCID: PMC4674667.Pubmed Partial Page. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Yoo J., Lee J.S., Kim S., et al. Length of hospital stay after hip fracture surgery and 1-year mortality. Osteoporos Int. 2019 Jan;30(1):145–153. doi: 10.1007/s00198-018-4747-7. Epub 2018 Oct 25. PMID: 30361752. [DOI] [PubMed] [Google Scholar]
- 13.Richards T., Glendenning A., Benson D., Alexander S., Thati S. The independent patient factors that affect length of stay following hip fractures. Ann R Coll Surg Engl. 2018;100(7):556–562. doi: 10.1308/rcsann.2018.0068. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Agha R., Abdall-Razak A., Crossley E., et al. STROCSS 2019 Guideline: strengthening the reporting of cohort studies in surgery. Int J Surg. 2019;72:156–165. doi: 10.1016/j.ijsu.2019.11.002. [DOI] [PubMed] [Google Scholar]
- 15.Zhang C., Feng J., Wang S., et al. Incidence of and trends in hip fracture among adults in urban China: a nationwide retrospective cohort study. PLoS Med. 2020;17(8) doi: 10.1371/journal.pmed.1003180. Published 2020 Aug 6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Cooper C., Cole Z.A., Holroyd C.R., et al. Secular trends in the incidence of hip and other osteoporotic fractures. Osteoporos Int. 2011;22(5):1277–1288. doi: 10.1007/s00198-011-1601-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Kanis J., Oden A., Johnell O., De Laet C., Jonsson B., Oglesby A. The components of excess mortality after hip fracture. Bone. 2003;32(5):468–473. doi: 10.1016/s8756-3282(03)00061-9. [DOI] [PubMed] [Google Scholar]