Risk factors for extended length of stay following hip fracture surgery: A retrospective cohort study of patients aged 90 years and over in China

Hip fractures are typical fractures in geriatric patients and are associated with a high risk of postoperative complications and extended length of stay (LOS) compared with other osteoporotic fractures, particularly among patients over age 90. We aimed to identify the risk factors for extended LOS (more than 14 days) in patients older than 90 years. A single-centre retrospective cohort study was performed in 50 patients (over age 90) who experienced hip fracture between June 1, 2013, and August 31, 2017. Their medical records were retrospectively reviewed from the hospital’s electronic database. The baseline characteristics of the patients, intraoperative data and postoperative complications were collected for the assessment of potential risk factors. Univariate analysis and multivariate binary logistic regression analysis were performed to determine potential risk factors for extended LOS.

confidence interval (CI)=1.02-1.41, P = 0.029. Neither the LOS nor postoperative pneumonia was found to be associated with the anaesthesia technique.

Conclusions
A lower mean diastolic blood pressure during surgery and the occurrence of postoperative pneumonia may extend the LOS in patients aged 90 years and over who undergo hip fracture surgery. The use of regional anaesthesia was not associated with a shorter LOS. Among all types of osteoporotic fractures in the People's Republic of China, hip fractures are associated with the highest costs. The majority of the cost is associated with direct medical care because the average total LOS following hip fracture surgery is approximately 12-24 days [3]. Recently, some studies [4][5][6][7] have indicated that general anaesthesia and acute postoperative complications, such as the occurrence of postoperative pneumonia, may extend the LOS. However, a recent meta-analysis showed that there was a small statistically significant difference in the LOS favouring regional anaesthesia that was unlikely to be clinically significant. The study also showed that there was no significant difference in the rates of pneumonia in patients receiving either regional or general anaesthesia [8]. As a result, the objective of this study was to identify risk factors for extended LOS (more than 14 days) in fragile hip fracture patients over the age of 90 years.

Hospital setting and data sources
This study was conducted at the Peking Union Medical College Hospital (PUMCH), a 2800bed hospital located in China. A hospital-based retrospective cohort study was conducted using the hospital's electronic database in Beijing, China. PUMCH residents had their medically necessary health care service and physician and hospital information as well as their demographic characteristics recorded in the database. The data were collected on a routine basis, and the analysis was carried out retrospectively. Therefore, no informed consent was required, as it was waived by the Institutional Review Board of PUMCH. Ethics approval was obtained from the Institutional Review Board of PUMCH.

The inclusion/exclusion of patients and data collection
Patients aged 90 years and over who underwent hip fracture surgery at PUMCH between June 1, 2013, and August 31, 2017, were identified from the hospital's electronic database. Non-PUMCH residents and patients who were transferred to other hospitals postoperatively were excluded due to missing data. Other exclusion criteria were multiple traumas and malignant tumour-associated fractures.
The following data were collected from the hospital information system (HIS) and anaesthesia information management system in the operating room (AIMSOR). The following baseline characteristics of the patients were collected: age, sex, body mass index (BMI), smoking and alcohol histories, pre-existing comorbidities, the American Society of Anesthesiologists physical status classification (ASA status), preoperative nutritional status [albumin level and prognostic nutrition index (PNI)], and baseline laboratory data. Intraoperative data included the anaesthetic and surgical procedures used (type and duration, respectively), whether the procedure was an emergency surgery, and the intraoperative haemodynamic and fluid management data. The blood pressure and heart rate data were collected every 5 minutes during the surgery.

Study outcomes
The primary outcome in this study was the extended LOS. The secondary outcomes included the incidence of postoperative complications and in-hospital mortality. The LOS was calculated according to the admission and discharge dates on the homepage of the medical records. The extended LOS was defined as an LOS exceeding 14 days.
Complications, including any medical or surgical adverse events that occurred after surgery, such as postoperative pneumonia, myocardial injury, and delirium, were

Statistical analyses
Continuous variables are expressed as the means and standard deviations, and categorical variables are expressed as the frequencies and proportions. All patients were divided into two groups: (1) patients who stayed in the hospital for more than 14 days and (2) patients who stayed in the hospital for no more than 14 days. First, we performed a univariate analysis to assess the associations between potential risk factors and extended LOS. We compared the basic and clinical characteristics between the groups using t tests (for means), Wilcoxon rank-sum tests (for medians) and χ 2 tests (for percentages).
Furthermore, we applied binary logistic regression to investigate potential risk factors for extended LOS, which included variables with a p-value<0.1 in the univariate analyses. For significant risk factors found in univariate analysis that correlated with each other, for example, HAI and pneumonia, we selected one clinically important risk factor for inclusion in the model to avoid an unstable estimation of the regression model. For secondary outcomes, we also used logistic regression to explore the associations between the patients' characteristics (basic and clinical) and postoperative complications. All statistical analyses were conducted using SAS 9.4 (SAS Institute, Inc., Cary, NC, United States), and a 2-sided p-value <0.05 was considered statistically significant.

Patient characteristics
During the study period, a total of 50 patients underwent surgery for the management of hip fractures (descriptive characteristics are shown in Table 1). The age of the patients ranged from 90 to 101 years (median, 92 years), and most of the patients were women.
Hypertension (n = 25, 50%) was the most common underlying disease, followed by coronary artery disease (n = 15, 30%). Most of the patients had an ASA status of 3 (n = 39) or 4 (n = 1). The preoperative nutritional status was poor among patients with anaemia, hypo-albuminemia, and a lower PNI. Regional anaesthesia (i.e., spinal or nerve block) was the most common type of anaesthesia and was used in 38 patients, followed by general anaesthesia. Artificial femoral head replacement was the most common type of surgery (n = 30, 60%), followed by open reduction and internal fixation (n = 20, 40%).
During surgery, approximately 44% of patients received a red blood cell transfusion, and the median estimated blood loss was 138 (100, 200) mL.
In this cohort, 35 patients experienced a complicated course of treatment with an inhospital mortality rate of 4% (n = 2). The most common complications were the occurrence of HAI (48%), delirium/postoperative cognitive dysfunction (28%) and MINS (26%). Among the patients with HAI, postoperative pneumonia (38%) was the most common infection, and among the patients with MINS, only myocardial injury was observed without any adverse events.

Risk factors for extended LOS
The median duration of hospital stay was 18 (11, 21) days. Thirty-three (66%) patients stayed in the hospital for more than 14 days. The univariate analyses (Table 3)  compared with the use of general anaesthesia, the use of regional anaesthesia was not associated with a shorter LOS or a lower morbidity of postoperative pneumonia.

Discussion
In the present study, we found that thirty-three (66%) patients were hospitalized for more than 14 days. An extended LOS was significantly associated with the occurrence of pneumonia after surgery (OR, 8.95) and the mean diastolic blood pressure during surgery (OR, 0.86) but not the anaesthesia technique. The mean diastolic blood pressure of patients who were hospitalized for no more than 14 days in our study was 71.7 mmHg, while it was 63.9 mmHg in patients who were hospitalized for more than 14 days. In addition, we found that compared with the use of general anaesthesia, the use of regional anaesthesia was not associated with a lower incidence of postoperative pneumonia or a shorter LOS. The results do not support an LOS benefit of regional anaesthesia compared with general anaesthesia in patients aged 90 years and over.
However, it is important to note that most of the very elderly patients in our study received regional anaesthesia (76%). This is in line with the general trend of the guidelines and literature [4][5][6][7] requirements for hip joint surgery in elderly patients. Some studies [14,15] suggest that compared with general anaesthesia, regional anaesthesia reduces the incidence of postoperative adverse outcomes. While we cannot deny that regional anaesthesia may be beneficial in older hip fracture surgery patients, the conclusions of these studies or guidelines are generally based on older patients over the age of 65. It is questionable whether these findings are applicable to elderly patients over 90 years old. Advanced age (over 90 years) itself is a major risk factor for poor prognosis.
In our study, the proportion of patients undergoing regional anaesthesia was approximately the same (74.2% vs 78.9%) in the pneumonia group as in the nonpneumonia group, with no significant difference. Although regional anaesthesia may stabilize haemodynamics in older hip surgery patients, the target blood pressure values are still of more concern.
Perioperative hypotension is associated with an increase in postoperative morbidity and extended LOS. Individualized blood pressure management strategies could reduce the risk of postoperative pneumonia among high-risk patients undergoing major surgery [16].
However, for patients over 90 years of age undergoing hip surgery, there is no recognized blood pressure index and specific threshold. Our research found that the mean diastolic blood pressure rather than the mean arterial pressure was another risk factor for extended LOS. One previous study [17] noted that in the elderly population, a diastolic blood pressure reduction below 70 mmHg should be avoided because it is associated with increased mortality. A possible explanation for this phenomenon could be an imbalance between sufficient perfusion pressure and arteriolar vasodilation, both of which are required for adequate tissue perfusion [18]. Impaired microcirculation, especially in the coronary bed, may account for increased mortality and extended LOS. Anaesthesia can be seen as another antihypertensive method, and perhaps we should keep the mean diastolic blood pressure above 70 mmHg in geriatric patients during surgery to shorten the LOS.
There are several limitations to be addressed. First, it must be mentioned that the conclusions of our study are surely limited by the comparatively small sample of patients, which only results in information on a very specific and small cohort. We could not analyse the risk factors for in-hospital mortality because of the relatively small number of events (2 deaths in 50 patients). Second, the study was only observational and occurred in one hospital unit with a short observation period, which only included the in-hospital period.
The information regarding patient symptoms relied solely on the input of information by the treating doctors into the medical database, although we expect any related errors or omissions to be evenly distributed. Third, we did not take the elapsed time from hospital arrival to surgery into consideration, but this time is regarded as an important risk factor for extended LOS and 30-day mortality. All patients in the cohort waited more than 24 hours to undergo surgery. The 2 main reasons for this delay were preoperative medical clearance and operating room access. Fourth, in the univariate and multivariate analyses, multiple comparisons and statistical tests were conducted. However, considering the small sample size in our analysis, we did not use any correction for the type I error, which may have led to false positive findings. The results from the models should be regarded as hypothesis-generating rather than as solid evidence of the risk factors for extended LOS.
In conclusion, complications after hip fracture surgery are common in geriatric patients. A lower mean diastolic blood pressure during surgery and the occurrence of postoperative pneumonia may extend the LOS. The use of regional anaesthesia compared with general anaesthesia was associated with neither a lower incidence of postoperative pneumonia nor a shorter LOS.