Background:
An elevated brain natriuretic peptide (BNP) level has been shown to be associated with mortality and cardiac events in cardiac surgery, but its utility in the prediction of morbidity and mortality in hip fracture surgery is unknown. The primary aim of this study was to determine if there is a difference in BNP level at the time of injury between patients who do and do not develop complications after hip fracture surgery. The secondary aim was to determine if there is a predictive relationship between complications associated with the initial BNP level and mortality.
Methods:
A retrospective chart review of 455 hip fractures in patients ≥60 years old that were operatively treated between February 2014 and July 2018 was performed. Patients were included if they had a BNP level within 48 hours after injury (BNPi). Specific perioperative (≤7 days), 30-day, 1-year, and 2-year postoperative complications were recorded. Wilcoxon rank-sum tests were used to determine if higher BNPi values were associated with greater morbidity. The complications associated with higher BNPi values were further analyzed to assess if they were predictive of mortality, using univariate and multivariable analyses.
Results:
Higher BNPi was significantly associated with greater morbidity at all postoperative time points and with higher mortality at 1 and 2 years postoperatively. Furthermore, several complications including cardiac failure or exacerbation and altered mental status were associated with mortality at all time points in univariate analysis and at many time points in multivariable analysis.
Conclusions:
Patients with higher BNPi levels were more likely to develop complications up to 1 year postoperatively, and several of these complications were associated with increased mortality. Future studies to determine if delaying surgery until BNP levels are normalized or lowered may help guide management, and may be useful in determining the need for further medical optimization. Future studies aimed at defining a threshold BNP value at the time of injury may also help in better managing patients preoperatively.
Level of Evidence:
Prognostic Level III. See Instructions for Authors for a complete description of levels of evidence.
Geriatric hip fractures are a common orthopaedic problem, with >1.6 million occurring each year worldwide, and the subsequent 1-year mortality rate is approximately 25%1-3. Geriatric patients with hip fractures frequently have multiple comorbidities at the time of injury, and numerous articles have reported on risk stratification tools to predict morbidity and mortality after hip fractures in this population2,4-6. In addition to dementia and prosthesis-related complications, cardiac complications including congestive heart failure, myocardial infarction (MI), and cardiac arrest are the main causes of postoperative morbidity and mortality2,6,7. Thus, reducing the risk of complications involves balancing the benefits of preoperative medical optimization against the increase in risk with increasing time to surgery8.
Prior studies indicate that short and long-term complication risks increase with patient comorbidities1,2,4-6,9,10. Risk stratification calculators (e.g., Model for End-Stage Liver Disease [MELD], American Society of Anesthesiologists [ASA] class, Ranson criteria) have been developed to predict morbidity for a number of other conditions or for surgery in general. Developing preoperative tools to help anticipate morbidity and mortality after hip fracture could be used to encourage quality improvement, enhance patient counseling and informed consent, and aid in identifying high-risk patients who require medical optimization prior to surgery. Previous research has examined specific laboratory markers in patients with a hip fracture. Seyedi et al. found that high blood urea nitrogen (BUN) and creatinine (Cr) levels in plasma at the time of the hip fracture were associated with 3-month mortality rates that were 3 and 2.5 times higher, respectively11. Roche et al. demonstrated that one of the most common postoperative complications associated with increased mortality was heart failure7. Lauland et al. found that low hemoglobin levels, low total lymphocyte count (TLC), low albumin levels, albumin/TLC ratio, high Cr levels, and high parathyroid hormone levels were prognostic markers of 2-year mortality12,13.
One of the biomarkers that has received attention is brain natriuretic peptide (BNP), which is produced by ventricular myocytes in response to cardiac stress. BNP acts with the renin-angiotensin-aldosterone system to modulate blood volume and with the sympathetic nervous system to modulate sympathetic responses and vasoconstriction14. It has shown utility as a diagnostic tool in the evaluation of acute dyspnea and ventricular dysfunction, and as a prognosticator for the risk of future cardiac events. A study indicated that a blood value of <100 pg/mL made a diagnosis of congestive cardiac failure unlikely, whereas a value of >500 pg/mL made it highly likely14. BNP has been advocated as a sensitive marker of ventricular dysfunction, with the concentration of BNP progressively increasing from Grade-I to Grade-IV cardiac failure according to the New York Heart Association (NYHA) classification. Every 50-pg/mL increase in the plasma BNP concentration in elderly subjects with no history of cardiac disease was found to be associated with a 1.6-fold greater risk of cardiac events and 1.4-fold greater risk of total mortality over a 2-year follow-up period15. Analysis of the Framingham Offspring Study Database found that BNP values of >20 pg/mL were associated with an increase of >60% in the long-term risk of mortality15.
Preoperative serum BNP levels have been determined to be a predictor of mortality in patients undergoing valve or coronary artery bypass surgery16. Additionally, several studies have found associations between elevated BNP levels and both mortality and cardiac events2,15,17-20. Although the popular focus on the clinical utility of BNP values has been centered around cardiac events, current published practice guidelines regarding elevated levels are also available for other conditions including acute and chronic renal failure, hypertension, and pulmonary diseases21-23. This very broad utility of the BNP level is suggestive of its potential use as an indicator for short and/or long-term morbidity and mortality after a hip fracture. Studies have shown that patients with lower postoperative BNP levels have a lower risk of a cardiac complication in noncardiac surgeries20. The ability to identify modifiable risk factors for poor outcomes would be beneficial for surgical planning and preoperative medical optimization, and the BNP level could potentially be used to provide patients and their families with prognostic information24. Although it is currently unknown whether the BNP level is a modifiable risk factor, hip fracture outcomes are significantly worse if surgery is performed >48 hours after injury, which makes determining whether BNP levels at the time of injury can be modified in a reasonable amount of time a topic of interest. BNP is an acute-phase reactant whose level is correlated with the C-reactive protein (CRP) level, and its short half-life of approximately 20 minutes suggests that it may have utility for short-term comparisons between BNP levels at successive time points25,26.
Our aim was to broaden the understanding of the relationship between BNP levels and complications. The primary aim of this study was to determine if there is a difference in BNP levels at the time of injury (BNPi) between patients who develop complications after hip fracture surgery and those who do not. The secondary aim was to determine if there is a predictive relationship between complications associated with higher BNPi levels and mortality.
Materials and Methods
After obtaining institutional review board approval, a retrospective chart review within the electronic medical record (EMR) system of a single academic level-1 trauma center was performed for all patients ≥60 years old with operatively treated low-energy hip fractures between February 2014 and July 2018. Patients were included if they had at least 1 BNP level measured within 48 hours after injury (BNPi). International Classification of Diseases, 9th Revision (ICD-9) and 10th Revision (ICD-10) codes were queried for all types of hip fractures (see Appendix Supplements Table 1). A hip fracture was defined as a femoral neck fracture (AO31B), intertrochanteric hip fracture (AO31A1), or pertrochanteric hip fracture with subtrochanteric extension (AO31A3). Pertrochanteric hip fractures with subtrochanteric extension in this group were low-energy torsional injuries with extension from the proximal aspect of the femur to ≥5 cm distal to the lesser trochanter (Table I). Patients with a neoplastic lesion at the site of injury, ipsilateral acetabular or proximal femoral injury, or an age of <60 years were excluded.
TABLE I.
Fracture and Surgical Characteristics (N = 455)
| Fracture type | |
| Femoral neck fracture | 214 (47.0%) |
| Intertrochanteric hip fracture | 221 (48.6%) |
| Pertrochanteric hip fracture with subtrochanteric extension | 20 (4.4%) |
| Procedure | |
| Closed reduction and percutaneous pinning | 62 (13.6%) |
| Open reduction and internal fixation | 15 (3.3%) |
| Cephalomedullary nail | 229 (50.3%) |
| Hemiarthroplasty | 123 (27.0%) |
| Total hip arthroplasty | 26 (5.7%) |
A total of 1,130 patients were identified, and 455 met the eligibility requirements for the study population (Table II). Three hundred and eighteen (69.9%) of the patients were female and 137 (30.1%) were male, which is comparable with male-to-female ratios reported in the literature3,27. The median age was 82.0 years (range, 60 to 99 years), and the median body mass index (BMI) was 23.8 kg/m2 (range, 13.3 to 63.7 kg/m2). There were 221 (48.6%) intertrochanteric femoral fractures, 214 (47.0%) femoral neck fractures, and 20 (4.4%) pertrochanteric hip fractures with subtrochanteric extension (Table I). Two hundred and twenty-nine (50.3%) were treated with cephalomedullary nails; 123 (27.0%), with hemiarthroplasty; 62 (13.6%), with closed reduction and percutaneous pinning; 26 (5.7%), with total hip arthroplasty; and 15 (3.3%), with open reduction and internal fixation (Table I). Additional patient characteristics and preoperative factors are reported in Table I.
TABLE II.
Demographics and Clinical Characteristics of All Patients with Hip Fracture (N = 455)*
| Age† (yr) | 82 (60-99) |
| Sex | |
| Male | 137 (30.1%) |
| Female | 318 (69.9%) |
| BMI† (kg/m2) | 23.79 (13.32-63.70) |
| No. of comorbidities† | 3.0 (2.0-4.0) |
| CCI† | 6 (2-17) |
| ASA class | |
| 1 | 0 (0%) |
| 2 | 35 (7.7%) |
| 3 | 270 (59.3%) |
| 4 | 150 (33.0%) |
| 5 | 0 (0%) |
BMI = body mass index, CCI = Charlson Comorbidity Index, ASA = American Society of Anesthesiologists.
Non-normal data; the values are reported as the median with the range in parentheses.
Information on multiple comorbidities was collected for analysis (Table III). Cardiovascular disease was the most common (278 patients, 61.1%), followed by respiratory disease (151 patients, 30.5%), renal disease (143 patients, 31.4%), and cerebrovascular disease (139 patients, 30.5%). Complications were recorded for perioperative (≤7 days from surgery), 30-day, 1-year, and 2-year time points (Table IV).
TABLE III.
Initial BNP Levels in Patients with and without Specific Preoperative Comorbidities*
| Comorbidity | Incidence (N = 455) | Mean BNPinitial (SD) (pg/mL) | P Value |
|---|---|---|---|
| Cardiovascular disease | <0.0001 | ||
| No | 177 (38.9%) | 101.9 (92) | |
| Yes | 278 (61.1%) | 298.3 (543) | |
| Cerebrovascular disease | 0.0090 | ||
| No | 316 (69.5%) | 171.3 (275) | |
| Yes | 139 (30.6%) | 336.9 (664) | |
| Respiratory disease | 0.0117 | ||
| No | 304 (66.8%) | 182.0 (325) | |
| Yes | 151 (33.2%) | 302.2 (599) | |
| Renal disease | <0.0001 | ||
| No | 312 (68.6%) | 143.1 (151) | |
| Yes | 143 (31.4%) | 393.9 (722) | |
| Diabetes mellitus | 0.2849 | ||
| No | 343 (75.4%) | 198.9 (368) | |
| Yes | 112 (24.6%) | 292.3 (602) | |
| Rheumatoid arthritis | 0.0320 | ||
| No | 446 (98.0%) | 213.8 (416) | |
| Yes | 9 (2.0%) | 622.3 (1,051) | |
| Cognitive impairment | 0.3088 | ||
| No | 315 (69.2%) | 240.8 (473) | |
| Yes | 140 (30.8%) | 179.4 (346) | |
| Parkinson disease | 0.2753 | ||
| No | 431 (94.7%) | 224.6 (447) | |
| Yes | 24 (5.3%) | 173.6 (240) | |
| Active malignancy | 0.2496 | ||
| No | 437 (96.0%) | 224.9 (445) | |
| Yes | 18 (4.0%) | 149.9 (213) | |
| Coagulopathy | <0.0001 | ||
| No | 379 (83.3%) | 186.5 (357) | |
| Yes | 76 (16.7%) | 398.6 (695) | |
| Osteoporosis | 0.6832 | ||
| No | 304 (66.8%) | 222.2 (437) | |
| Yes | 151 (33.2%) | 221.4 (443) | |
| Tobacco use | 0.7225 | ||
| No | 405 (89.0%) | 223.1 (439) | |
| Yes | 50 (11.0%) | 212.3 (439) |
SD = standard deviation.
TABLE IV.
Initial BNP Levels in Patients with and without Development of Specific Perioperative, 30-Day, 1-Year, and 2-Year Complications*
| Complication | Perioperative, ≤7 Days | 30 Days, N = 408 | 1 Year, N = 326 | 2 Years, N = 192 | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Incidence (N = 455) | Mean BNPinitial (SD) (pg/mL) | P Value | Incidence | Mean BNPinitial (SD) (pg/mL) | P Value | Incidence | Mean BNPinitial (SD) (pg/mL) | P Value | Incidence | Mean BNPinitial (SD) (pg/mL) | P Value | |
| Any complication | 0.0004 | |||||||||||
| No | 152 (33.4%) | 136.3 (153) | 0.0002 | 157 (34.5%) | 140.1 (254) | 0.0002 | 110 (24.2%) | 135.1 (292) | <0.0001 | 97 (21.3%) | 142.3 (310) | |
| Yes | 303 (66.6%) | 264.8 (521) | 298 (65.5%) | 265.0 (504) | 345 (75.8%) | 249.6 (473) | 358 (78.7%) | 243.4 (465) | ||||
| Respiratory infection | 0.0065 | |||||||||||
| No | 413 (90.8%) | 218.8 (436) | 0.2833 | 388 (85.3%) | 209.3 (417) | 0.0133 | 353 (77.6%) | 203.3 (420) | 0.0015 | 332 (73.0%) | 206.4 (431) | |
| Yes | 42 (9.2%) | 252.4 (465) | 67 (14.7%) | 294.7 (544) | 102 (22.4%) | 286.3 (494) | 123 (27.0%) | 263.8 (457) | ||||
| Cardiac failure or exacerbation | <0.0001 | |||||||||||
| No | 364 (80.0%) | 186.9 (356) | <0.0001 | 351 (77.1%) | 173.0 (314) | <0.0001 | 320 (70.3%) | 163.2 (303) | <0.0001 | 307 (67.5%) | 163.1 (308) | |
| Yes | 91 (20.0%) | 361.8 (660) | 104 (22.9%) | 386.9 (691) | 135 (29.7%) | 361.1 (637) | 148 (32.5%) | 343.8 (612) | ||||
| Myocardial infarction | 0.0139 | |||||||||||
| No | 439 (96.5%) | 196.1 (365) | 0.0015 | 435 (95.6%) | 189.3 (334) | 0.0007 | 425 (93.4%) | 183.8 (322) | 0.0012 | 417 (91.6%) | 185.3 (324) | |
| Yes | 16 (3.5%) | 929.9 (1,173) | 20 (4.4%) | 931.4 (1,224) | 30 (6.6%) | 760.9 (1,085) | 38 (8.4%) | 622.9 (998) | ||||
| Cerebrovascular accident | 0.0400 | |||||||||||
| No | 450 (98.9%) | 222.3 (441) | 0.7157 | 448 (98.5%) | 222.7 (442) | 0.7108 | 435 (95.6%) | 222.5 (447) | 0.0652 | 429 (94.3%) | 222.2 (450) | |
| Yes | 5 (1.1%) | 186.6 (168) | 7 (1.5%) | 170.3 (143) | 20 (4.4%) | 208.3 (158) | 26 (5.7%) | 216.5 (176) | ||||
| DVT/PE | 0.4536 | |||||||||||
| No | 445 (97.8%) | 222.8 (443) | 0.3702 | 440 (96.7%) | 220.2 (441) | 0.1170 | 433 (95.2%) | 221.3 (444) | 0.1582 | 430 (94.5%) | 222.5 (446) | |
| Yes | 10 (2.2%) | 179.4 (111) | 15 (3.3%) | 272.5 (367) | 22 (4.8%) | 233.6 (311) | 25 (5.5%) | 212.1 (297) | ||||
| UTI | 0.0726 | |||||||||||
| No | 344 (75.6%) | 223.5 (457) | 0.9609 | 316 (69.5%) | 211.9 (433) | 0.5266 | 276 (60.7%) | 208.9 (455) | 0.0575 | 264 (58.0%) | 211.1 (464) | |
| Yes | 111 (24.4%) | 217.0 (377) | 139 (30.5%) | 244.6 (451) | 179 (39.3%) | 241.9 (412) | 191 (42.0%) | 236.8 (401) | ||||
| Decubitus ulcer | 0.0735 | |||||||||||
| No | 444 (97.6%) | 214.6 (417) | 0.0604 | 438 (96.3%) | 212.8 (415) | 0.1321 | 425 (93.4%) | 212.4 (420) | 0.0155 | 422 (92.7%) | 213.5 (422) | |
| Yes | 11 (2.4%) | 514.8 (963) | 17 (3.7%) | 456.2 (833) | 30 (6.6%) | 355.8 (639) | 33 (7.3%) | 328.6 (614) | ||||
| Superficial infection | 0.5391 | |||||||||||
| No | 455 (100.0%) | 221.9 (438) | † | 450 (98.9%) | 223.1 (441) | 0.6333 | 445 (97.8%) | 222.9 (442) | 0.9622 | 442 (97.1%) | 224.0 (444) | |
| Yes | 0 (0.0%) | 5 (1.1%) | 109.9 (77) | 10 (2.2%) | 176.1 (218) | 13 (2.9%) | 151.5 (195) | |||||
| Deep infection | 0.0607 | |||||||||||
| No | 455 (100.0%) | 221.9 (438) | † | 453 (99.6%) | 222.6 (439) | 0.2532 | 447 (98.2%) | 224.2 (442) | 0.1452 | 446 (98.0%) | 224.6 (442) | |
| Yes | 0 (0.0%) | 2 (0.4%) | 62.5 (56) | 8 (1.8%) | 93.3 (86) | 9 (2.0%) | 85.7 (84) | |||||
| Altered mental status/delirium | <0.0001 | |||||||||||
| No | 337 (74.1%) | 189.8 (374) | 0.0013 | 310 (68.1%) | 184.4 (377) | 0.0003 | 278 (61.1%) | 180.5 (392) | <0.0001 | 270 (59.3%) | 179.8 (397) | |
| Yes | 118 (25.9%) | 313.6 (577) | 145 (31.9%) | 302.0 (540) | 177 (38.9%) | 286.9 (498) | 185 (40.7%) | 283.3 (488) | ||||
| Transfusion | ||||||||||||
| No | 314 (69.0%) | 201.4 (399) | 0.0103 | |||||||||
| Yes | 141 (31.0%) | 267.5 (514) | ||||||||||
SD = standard deviation, DVT = deep vein thrombosis, PE = pulmonary embolus, UTI = urinary tract infection.
Too few occurrences.
Statistical Analysis
The characteristics of the sample were summarized using descriptive statistics; frequencies and percentages were used to describe categorical variables, and either means and standard deviations or medians and interquartile ranges (IQRs) were used (as appropriate) to describe continuous variables. Wilcoxon rank-sum tests were used to assess differences in BNPi between patients with and without specific complications as well as between those who died and those who survived. Univariate logistic regression models were used to compare mortality as an outcome with perioperative and postoperative complications that had significant associations with increased BNPi. Multivariable logistic regressions were performed with a Firth bias correction and profile-likelihood confidence intervals to account for low frequencies of the outcome. A p value of <0.05 was considered significant. All analyses were performed using SAS 9.4 (SAS Institute).
Source of Funding
No external funding was received for this study.
Results
The median BNPi was 110.0 pg/mL (IQR, 56.5 to 221.0 pg/mL), and the mean BNPi was 221 ± 438 pg/mL. BNPi was found to be significantly correlated with the ASA class and Charlson Comorbidity Index (CCI); the Spearman correlation coefficients were 0.2786 (p < 0.0001) and 0.3111 (p < 0.0001), respectively. Patients with a history of cardiovascular (p < 0.0001), cerebrovascular (p = 0.0090), respiratory (p = 0.0117), and renal (p <0.0001) diseases and with rheumatoid arthritis (p = 0.032) and coagulopathy (p < 0.0001) had significantly higher mean BNPi than patients without these conditions (Table III). A higher BNPi was significantly associated with a higher risk of developing a complication at all time points (Table IV). For the perioperative time point, a higher BNPi was significantly associated with higher risks of cardiac failure or exacerbation (CE) (p < 0.0001), altered mental status or delirium (AMS) (p = 0.0013), MI (p = 0.0015), and a need for transfusion (p = 0.0103). For the 30-day time point, a higher BNPi was significantly associated with higher risks of respiratory infection (p = 0.0133), CE (p < 0.0001), MI (p = 0.0007), and AMS (p = 0.0003). For the 1-year time point, a higher BNPi was significantly associated with higher risks of respiratory infection (p = 0.0015), CE (p < 0.0001), MI (p = 0.0012), decubitus ulcers (p = 0.0155), and AMS (p < 0.0001). For the 2-year postoperative time point, a higher BNPi was significantly associated with higher risks of respiratory infection (p = 0.0065), CE (p < 0.0001), MI (p = 0.0139), and AMS (p < 0.0001).
Patients with a higher mean BNPi had a significantly higher risk of mortality at 1 and 2 years postoperatively. The risk of mortality was higher in patients with a higher mean BNPi at the perioperative or 30-day postoperative time point, but these differences did not reach significance (Table V). The mean BNPi was 185.0 pg/mL in patients who remained alive at 1 year postoperatively compared with 435.6 in those who had died. The mean BNPi was185.2 pg/mL in patients who remained alive at 2 years postoperatively compared with 384.1 pg/mL in those who had died.
TABLE V.
Association of Initial BNP Level with Mortality*
| Time Frame | Incidence (N = 192) | Mean BNPinitial (SD) (pg/mL) | Median BNPinitial (Range) (pg/mL) | P Value |
|---|---|---|---|---|
| Perioperative, ≤7 days | 0.9505 | |||
| No | 442 (97.1%) | 221.8 (442) | 110.5 (0-3,832) | |
| Yes | 13 (2.9%) | 225.1 (289) | 82.0 (25-1,074) | |
| 30 days | 0.5649 | |||
| No | 427 (93.8%) | 206.1 (393) | 111.0 (0-3,832) | |
| Yes | 28 (6.2%) | 463.1 (853) | 102.9 (17-3,474) | |
| 1 year | 0.0043 | |||
| No | 388 (85.3%) | 185.0 (324) | 104.0 (0-3,362) | |
| Yes | 67 (14.7%) | 435.6 (807) | 181.0 (17-3,832) | |
| 2 years | 0.0029 | |||
| No | 371 (81.5%) | 185.2 (330) | 104.0 (0-3,362) | |
| Yes | 84 (18.5%) | 384.1 (731) | 158.0 (15-3,832) |
The total number of deaths in the sample population was 106. The mean time from surgery to death was 376.7 days, with a standard deviation (SD) of 471 days and a range of 0 to 2,429 days (6.7 years).
Univariate regression analysis found that patients had a significantly higher risk of mortality if they developed CE (odds ratio [OR] = 9.9, p = 0.0002), MI (OR = 5.6, p = 0.0354), a need for transfusion (OR = 3.7, p = 0.0235), or AMS (OR = 3.5, p = 0.0279) in the perioperative period (Table VI). For the 30-day postoperative time point, a higher risk of mortality was associated with respiratory infection (OR = 6.0, p < 0.0001), CE (OR = 6.0, p < 0.0001), MI (OR = 4.3, p = 0.0149), and AMS (OR = 5.0, p = 0.0001). For the 1-year time point, higher mortality was associated with respiratory infection (OR = 3.0, p < 0.0001), CE (OR = 2.9, p < 0.0001), MI (OR = 2.7, p = 0.0183), decubitus ulcers (OR = 2.7, p = 0.0183), and AMS (OR = 4.0, p < 0.0001). For the 2-year time point, mortality was associated with respiratory infection (OR = 2.3, p = 0.0010), CE (OR = 2.2, p = 0.0013), cerebrovascular accident (CVA) (OR = 2.5, p = 0.0338), and AMS (OR = 3.3, p < 0.0001).
TABLE VI.
Univariate Regression Analysis with Mortality as the Outcome*
| Complication | Perioperative, ≤7 Days | 30 Days | 1 Year | 2 Years | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| OR | 95% CI | P Value | OR | 95% CI | P Value | OR | 95% CI | P Value | OR | 95% CI | P Value | |
| Respiratory infection | † | 6.0 | 2.7-13.3 | <0.0001 | 3.0 | 1.8-5.3 | <0.0001 | 2.3 | 1.4-3.8 | 0.0010 | ||
| Cardiac failure or exacerbation | 9.9 | 3.0-32.9 | 0.0002 | 6.0 | 2.7-13.4 | <0.0001 | 2.9 | 1.7-5.0 | <0.0001 | 2.2 | 1.4-3.6 | 0.0013 |
| Myocardial infarction | 5.6 | 1.1-27.5 | 0.0354 | 4.3 | 1.3-13.8 | 0.0149 | 2.7 | 1.2-6.2 | 0.0183 | 1.4 | 0.6-3.1 | 0.3877 |
| Cerebrovascular accident | § | § | § | 2.5 | 1.1-5.8 | 0.0338 | ||||||
| DVT/PE | § | § | § | § | ||||||||
| UTI | † | † | † | † | ||||||||
| Decubitus ulcer | † | † | 2.7 | 1.2-6.2 | 0.0183 | † | ||||||
| Superficial infection | ‡ | † | † | § | ||||||||
| Deep infection | ‡ | † | † | § | ||||||||
| Altered mental status/delirium | 3.5 | 1.1-10.6 | 0.0279 | 5.0 | 2.2-11.5 | 0.0001 | 4.0 | 2.3-6.9 | <0.0001 | 3.3 | 2.0-5.4 | <0.0001 |
| Transfusion | 3.7 | 1.2-11.6 | 0.0235 | ‡ | ‡ | ‡ | ||||||
OR = odds ratio, CI = confidence interval, DVT = deep vein thrombosis, PE = pulmonary embolus, UTI = urinary tract infection.
Not estimable.
No occurrences.
Too few occurrences.
Multivariable regression analysis found that patients had a significantly higher risk of mortality if they developed CE (OR = 6.6, p = 0.0014) during the perioperative period (Table VII). For the 30-day postoperative time point, mortality was associated with respiratory infection (OR = 3.3, p = 0.0047), CE (OR = 3.6, p = 0.0032), and AMS (OR = 4.0, p = 0.0010). For the 1-year time point, mortality was associated with respiratory infection (OR = 2.1, p = 0.0103) and AMS (OR = 2.8, p = 0.0005). For the 2-year time point, mortality was associated with AMS (OR = 2.4, p = 0.0010).
TABLE VII.
Multiple Regression Analysis with Mortality as the Outcome*
| Complication | Perioperative, ≤7 Days | 30 Days | 1 Year | 2 Years | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| OR | 95% CI | P Value | OR | 95% CI | P Value | OR | 95% CI | P Value | OR | 95% CI | P Value | |
| Respiratory infection | † | 3.336 | 1.424-7.682 | 0.0047 | 2.129 | 1.192-3.763 | 0.0103 | 1.687 | 0.994-2.842 | 0.0519 | ||
| Cardiac failure or exacerbation | 6.ov615 | 2.021-24.773 | 0.0014 | 3.560 | 1.515-8.510 | 0.0032 | 1.534 | 0.839-2.781 | 0.1619 | 1.301 | 0.748-2.245 | 0.3490 |
| Myocardial infarction | 2.925 | 0.460-13.900 | 0.2034 | 2.149 | 0.551-7.208 | 0.2432 | 1.510 | 0.597-3.601 | 0.3740 | 0.926 | 0.382-2.073 | 0.8585 |
| Cerebrovascular accident | § | § | § | 1.872 | 0.748-4.442 | 0.1730 | ||||||
| DVT/PE | § | § | § | § | ||||||||
| UTI | † | † | † | † | ||||||||
| Decubitus ulcer | † | † | 1.628 | 0.655-3.810 | 0.2837 | † | ||||||
| Superficial infection | ‡ | † | † | § | ||||||||
| Deep infection | ‡ | † | † | § | ||||||||
| Altered mental status/delirium | 2.143 | 0.670-6.899 | 0.1726 | 4.019 | 1.776-9.668 | 0.0010 | 2.769 | 1.574-4.969 | 0.0005 | 2.399 | 1.432-4.066 | 0.0010 |
| Transfusion | 2.569 | 0.830-8.501 | 0.0897 | ‡ | ‡ | ‡ | ||||||
OR = odds ratio, CI = confidence interval, DVT = deep vein thrombosis, PE = pulmonary embolus, UTI = urinary tract infection.
Not estimable.
No occurrences.
Too few occurrences.
The cut-point analysis did not reveal meaningful information, as the area under the receiver operating characteristic (ROC) curve approached 0.5 for all complications. Therefore, we were unable to determine a BNP threshold value for each comorbidity.
Discussion
High morbidity and mortality following hip fracture surgery in geriatric patients has been well documented, but standardized tools to identify which patients are most at risk lack consensus and their use is not standard practice. Our data suggest that higher BNPi values are associated with increased risk of developing certain complications up to 2 years postoperatively and the occurrence of certain complications is predictive of mortality.
The present study differs from prior investigations in its focus on BNPi as the primary variable, a wider breadth of complications analyzed, and the inclusion of only hip fractures. Zhao et al.18 analyzed 1,051 intertrochanteric hip fractures, examining multiple indicators at the time of admission. They found significant differences in chronic diseases of major organs, postoperative complications, APACHE (Acute Physiology and Chronic Health Evaluation) II scores, BNP values, and various laboratory indicators, which were risk factors for perioperative mortality in these patients.
Villacorta Junior et al.19 investigated the value of BNP in predicting cardiac complications after orthopaedic surgeries by analyzing 208 patients undergoing surgical treatment of a femoral fracture, total hip arthroplasty, or total knee arthroplasty. They analyzed the ability of BNP to predict multiple postoperative cardiac events, including cardiac-related mortality, using a multivariable logistic regression analysis. The median BNP level of 93 pg/mL (IQR, 73 to 424 pg/mL) in patients who experienced cardiac events was significantly higher than the median of 26.6 pg/mL (IQR, 13.2 to 53.1 pg/mL) in those without cardiac events (p = 0.0001). Cut-point analysis using an ROC curve identified the optimal value as 60 pg/mL. Villacorta Junior et al. concluded that BNP is an independent predictor of cardiac events after these orthopaedic surgical procedures.
Rodseth et al.20 performed a meta-analysis regarding the prognostic value of both preoperative and postoperative BNP values in patients undergoing noncardiac surgery. They included 18 studies with a total of 2,179 patients and determined that obtaining a postoperative BNP value enhanced risk stratification for the composite outcome of mortality or postoperative MI at both 30 and ≥180 days.
Pili-Floury et al.2 performed a prospective study to determine if BNP levels could identify preoperative cardiovascular disease in geriatric patients with a hip fracture. The study included 75 patients who had both a BNP measurement and transthoracic echocardiography performed within 24 hours after admission. They found that a preoperative BNP value of ≥285 pg/mL could discriminate between patients with and those without major echocardiographic abnormalities.
In the present study, increased BNPi values were associated with several complications (Table IV). These findings suggest that patients with lower BNPi levels may be less likely to develop complications after a hip fracture. Inpatient treatment approaches designed to lower BNP levels following a hip fracture may reduce the likelihood of certain perioperative and postoperative complications.
BNPi values were higher in the patients who died within 1 and 2 years after surgery than in those who survived (Table V). However, the threshold BNP level at which morbidity and mortality increase is unknown and warrants further study. We also analyzed how mortality during each of the time periods was related to each of the complications that had been found to be associated with increased BNPi values. (Tables VI and VII). The odds of mortality at all time points were greater in patients who developed CE or AMS. The odds of mortality were greater at all time points except 2 years in patients who developed a respiratory infection. Furthermore, the odds of mortality in the perioperative and 30-day periods remained greater in patients who developed CE when controlling for other variables. Additionally, the odds of mortality at all time points except perioperatively remained greater in patients who developed AMS when controlling for other variables.
It may be helpful to perform studies to determine if medical management and/or appropriate prophylaxis during the preoperative, perioperative, and postoperative periods will lessen the risk of complications and lead to improved survival following hip fracture surgery. The results of the present study do not suggest waiting longer than the standard 48-hour maximum prior to surgery in order to perform optimization, but rather using BNP as another means for assessment of the need for further medical optimization. Given the short half-life of BNP (20 minutes), medical optimization may decrease its level rapidly and thus allow it to potentially serve as a marker for when it is an appropriate time to proceed to surgery. However, that cannot be determined from the results of the present study; future studies are needed to determine if decreasing BNPi acutely will decrease the risk of complications or if the BNPi is only useful in preoperative prognostic assessment.
This study has several limitations inherent to retrospective research. Foremost, it relies on the availability of information and its proper documentation at the time it was entered into the EMR. All retrospective studies have some degree of error that cannot be identified or removed or corrected. Additionally, obtaining a BNP level at the time of admission only recently became the standard of care at our institution, limiting the number of charts we were able to review. Because this change was gradual, the patients who had a BNP level measured at the time of injury were also potentially subject to selection bias. Although univariate and multivariable analyses were used to determine the risk of postoperative mortality associated with specific complications, confounding bias is inevitable in this population because groups with more comorbidities have increased mortality risk. Efforts were made to mitigate this issue, but the true effect is possibly unknown. Additionally, we were unable to determine a threshold BNP level above which complications tend to increase. Future studies with larger patient populations may be able to determine a cutoff value with respect to morbidity, but current literature suggests that it is in the range of 60 to 285 pg/mL2,19. Logically, the cutoff value for mortality may be slightly higher.
Although the specific BNP level that would raise concern could not be determined from the present study, our data suggest that higher BNP levels at the time of hip fracture surgery are a surrogate marker for increased morbidity in the perioperative, 30-day, and 1-year postoperative periods. Additionally, the risk of mortality significantly increased with the postoperative development of specific complications whose rates were found to be elevated in patients with higher BNPi. Given these findings, future studies are warranted to determine if delaying surgery until BNP levels are normalized or lowered outweighs the risk of delaying surgery. This may help guide management and may be useful in determining the need for further medical optimization. Future studies aimed at defining a threshold BNP value at the time of injury that puts patients at increased risk may also help to better manage patients preoperatively.
Appendix
Supporting material provided by the authors is posted with the online version of this article as a data supplement at jbjs.org (http://links.lww.com/JBJSOA/A494).
Footnotes
Investigation performed at Baylor Scott & White Memorial Hospital, Temple, Texas
Disclosure: The Disclosure of Potential Conflicts of Interest forms are provided with the online version of the article (http://links.lww.com/JBJSOA/A493).
References
- 1.Cooper C, Campion G, Melton LJ, 3rd. Hip fractures in the elderly: a world-wide projection. Osteoporos Int. 1992. Nov;2(6):285-9. [DOI] [PubMed] [Google Scholar]
- 2.Pili-Floury S, Ginet M, Saunier L, Besch G, Bartholin F, Chopard R, Boillot A, Mebazaa A, Samain E. Preoperative plasma B-type natriuretic peptide (BNP) identifies abnormal transthoracic echocardiography in elderly patients with traumatic hip fracture. Injury. 2012. Jun;43(6):811-6. [DOI] [PubMed] [Google Scholar]
- 3.Basques BA, Bohl DD, Golinvaux NS, Leslie MP, Baumgaertner MR, Grauer JN. Postoperative length of stay and 30-day readmission after geriatric hip fracture: an analysis of 8434 patients. J Orthop Trauma. 2015. Mar;29(3):e115-20. [DOI] [PubMed] [Google Scholar]
- 4.Pugely AJ, Martin CT, Gao Y, Klocke NF, Callaghan JJ, Marsh JL. A risk calculator for short-term morbidity and mortality after hip fracture surgery. J Orthop Trauma. 2014. Feb;28(2):63-9. [DOI] [PubMed] [Google Scholar]
- 5.Jiang HX, Majumdar SR, Dick DA, Moreau M, Raso J, Otto DD, Johnston DW. Development and initial validation of a risk score for predicting in-hospital and 1-year mortality in patients with hip fractures. J Bone Miner Res. 2005. Mar;20(3):494-500. [DOI] [PubMed] [Google Scholar]
- 6.Holt G, Smith R, Duncan K, Finlayson DF, Gregori A. Early mortality after surgical fixation of hip fractures in the elderly: an analysis of data from the Scottish Hip Fracture Audit. J Bone Joint Surg Br. 2008. Oct;90(10):1357-63. [DOI] [PubMed] [Google Scholar]
- 7.Roche JJW, Wenn RT, Sahota O, Moran CG. Effect of comorbidities and postoperative complications on mortality after hip fracture in elderly people: prospective observational cohort study. BMJ. 2005. Dec 10;331(7529):1374. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Petersen MB, Jørgensen HL, Hansen K, Duus BR. Factors affecting postoperative mortality of patients with displaced femoral neck fracture. Injury. 2006. Aug;37(8):705-11. [DOI] [PubMed] [Google Scholar]
- 9.Belmont PJ, Jr, Garcia EJ, Romano D, Bader JO, Nelson KJ, Schoenfeld AJ. Risk factors for complications and in-hospital mortality following hip fractures: a study using the National Trauma Data Bank. Arch Orthop Trauma Surg. 2014. May;134(5):597-604. [DOI] [PubMed] [Google Scholar]
- 10.Carpintero P, Caeiro JR, Carpintero R, Morales A, Silva S, Mesa M. Complications of hip fractures: A review. World J Orthop. 2014. Sep 18;5(4):402-11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Seyedi HR, Mahdian M, Khosravi G, Bidgoli MS, Mousavi SG, Razavizadeh MR, Mahdian S, Mohammadzadeh M. Prediction of mortality in hip fracture patients: role of routine blood tests. Arch Bone Jt Surg. 2015. Jan;3(1):51-5. [PMC free article] [PubMed] [Google Scholar]
- 12.Laulund AS, Lauritzen JB, Duus BR, Mosfeldt M, Jørgensen HL. Routine blood tests as predictors of mortality in hip fracture patients. Injury. 2012. Jul;43(7):1014-20. [DOI] [PubMed] [Google Scholar]
- 13.Zhu XZ, Tao YL, Ma Z. Routine blood tests as predictors of mortality in hip fracture patients. Injury. 2013. Nov;44(11):1659. [DOI] [PubMed] [Google Scholar]
- 14.Hutfless R, Kazanegra R, Madani M, Bhalla MA, Tulua-Tata A, Chen A, Clopton P, James C, Chiu A, Maisel AS. Utility of B-type natriuretic peptide in predicting postoperative complications and outcomes in patients undergoing heart surgery. J Am Coll Cardiol. 2004. May 19;43(10):1873-9. [DOI] [PubMed] [Google Scholar]
- 15.Wang TJ, Larson MG, Levy D, Benjamin EJ, Leip EP, Omland T, Wolf PA, Vasan RS. Plasma natriuretic peptide levels and the risk of cardiovascular events and death. N Engl J Med. 2004. Feb 12;350(7):655-63. [DOI] [PubMed] [Google Scholar]
- 16.Murad Junior JA, Nakazone MA, Machado MdeN, Godoy MF. Predictors of mortality in cardiac surgery: brain natriuretic peptide type B. Rev Bras Cir Cardiovasc. 2015. Mar-Apr;30(2):182-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Sear JW, Howard-Alpe G. Preoperative plasma BNP concentrations: do they improve our care of high-risk non-cardiac surgical patients? Br J Anaesth. 2007. Aug;99(2):151-4. [DOI] [PubMed] [Google Scholar]
- 18.Zhao F, Wang X, Dou Y, Wang H, Zhang Y. Analysis of risk factors for perioperative mortality in elderly patients with intertrochanteric fracture. Eur J Orthop Surg Traumatol. 2019. Jan;29(1):59-63. [DOI] [PubMed] [Google Scholar]
- 19.Villacorta Junior H, Castro IS, Godinho M, Mattos C, Visconti R, Saud M, Novais M, Murad C, Nogueira F. B-type natriuretic peptide is predictive of postoperative events in orthopedic surgery. Arq Bras Cardiol. 2010. Dec;95(6):743-8. [DOI] [PubMed] [Google Scholar]
- 20.Rodseth RN, Biccard BM, Le Manach Y, Sessler DI, Lurati Buse GA, Thabane L, Schutt RC, Bolliger D, Cagini L, Cardinale D, Chong CP, Chu R, Cnotliwy M, Di Somma S, Fahrner R, Lim WK, Mahla E, Manikandan R, Puma F, Pyun WB, Radović M, Rajagopalan S, Suttie S, Vanniyasingam T, van Gaal WJ, Waliszek M, Devereaux PJ. The prognostic value of pre-operative and post-operative B-type natriuretic peptides in patients undergoing noncardiac surgery: B-type natriuretic peptide and N-terminal fragment of pro-B-type natriuretic peptide: a systematic review and individual patient data meta-analysis. J Am Coll Cardiol. 2014. Jan 21;63(2):170-80. [DOI] [PubMed] [Google Scholar]
- 21.Daniels LB, Maisel AS. Natriuretic peptides. J Am Coll Cardiol. 2007. Dec 18;50(25):2357-68. [DOI] [PubMed] [Google Scholar]
- 22.Maisel A, Mueller C, Adams K, Jr, Anker SD, Aspromonte N, Cleland JG, Cohen-Solal A, Dahlstrom U, DeMaria A, Di Somma S, Filippatos GS, Fonarow GC, Jourdain P, Komajda M, Liu PP, McDonagh T, McDonald K, Mebazaa A, Nieminen MS, Peacock WF, Tubaro M, Valle R, Vanderhyden M, Yancy CW, Zannad F, Braunwald E. State of the art: using natriuretic peptide levels in clinical practice. Eur J Heart Fail. 2008. Sep;10(9):824-39. [DOI] [PubMed] [Google Scholar]
- 23.Yancy CW, Jessup M, Bozkurt B, Butler J, Casey DE, Jr, Drazner MH, Fonarow GC, Geraci SA, Horwich T, Januzzi JL, Johnson MR, Kasper EK, Levy WC, Masoudi FA, McBride PE, McMurray JJ, Mitchell JE, Peterson PN, Riegel B, Sam F, Stevenson LW, Tang WH, Tsai EJ, Wilkoff BL; American College of Cardiology Foundation; American Heart Association Task Force on Practice Guidelines. 2013 ACCF/AHA guideline for the management of heart failure: a report of the American College of Cardiology Foundation/American Heart Association task force on practice guidelines. J Am Coll Cardiol. 2013. Oct 15;62(16):e147-239. [DOI] [PubMed] [Google Scholar]
- 24.Fisher A, Fisher L, Srikusalanukul W, Smith PN. Usefulness of simple biomarkers at admission as independent indicators and predictors of in-hospital mortality in older hip fracture patients. Injury. 2018. Apr;49(4):829-40. [DOI] [PubMed] [Google Scholar]
- 25.Weber M, Mitrovic V, Hamm C. B-type natriuretic peptide and N-terminal pro-B-type natriuretic peptide - Diagnostic role in stable coronary artery disease. Exp Clin Cardiol. 2006. Summer;11(2):99-101. [PMC free article] [PubMed] [Google Scholar]
- 26.Yetkin O, Hacievliyagil SS, Gunen H. Assessment of B-type natriuretic peptide in patients with pneumonia. Int J Clin Pract. 2008. Mar;62(3):488-91. [DOI] [PubMed] [Google Scholar]
- 27.Skála-Rosenbaum J, Bartoníček J, Ríha D, Waldauf P, Džupa V. Single-centre study of hip fractures in Prague, Czech Republic, 1997-2007. Int Orthop. 2011. Apr;35(4):587-93. [DOI] [PMC free article] [PubMed] [Google Scholar]