The present study was conducted with 46 patients who had suffered a hip fracture (46 accidents, 1.25% of residents, 15.4% of fall-accident victims) among 299 patients who had experienced a fall accident (371 accidents, 7.6% of residents) out of a total of 3,949 residents who had been admitted to one of five nursing homes between August 24, 2008 and January 31, 2014. The mean age of the patients was 82.7 yr (66-94 yr), and there were 28 males and 18 females. To examine differences in hip-joint-fracture risk factors, the rate of recovery of the capacity for activity after fracture and mortality rates, a nursing home residents (NHR) group was compared with community-dwelling individuals (CDI), a control group of the same number of subjects.

Fractures were divided into femoral neck fractures and intertrochanteric fractures according to location; and fracture types were subdivided using the Garden classification (5) and the AO classification (the Association for Osteosynthesis/Orthopaedic Trauma Association [AO/OTA] classification). The NHR group comprised 21 left hips with a fracture and 25 right hips with a fracture; among them all, 14 hips (30.4%) involved a femoral neck fracture and 32 (69.6%), an intertrochanteric fracture. In the subdivision by fracture type, Garden classification (5) type 3 was the most frequently occurring (10 hips) among the femoral neck fractures, while AO classification type 2 was the most frequent (22 hips) among the intertrochanteric fractures (Table 1). Radiologic measurement and evaluation for fracture classification were conducted by two orthopedic surgeons based on antero-posterior and lateral radiographs of the hip joint. Kappa coefficient tests were applied to measure the intra-observer (K1) and inter-observer (K2) reliability of the two observers; both sets of results were shown to be at reliable levels (K1=0.88, K2=0.81). Seven patients (15.2%) in the NHR group did not undergo surgery. These hips included one wherein surgery was not possible due to health condition, and six in which surgery was not performed owing to the family's refusal. In the CDI group, one patient, for health reasons, did not undergo surgery (P=0.026). The surgery waiting time in the NHR group's 39 hips in which surgical treatment was performed was not longer than one week in most hips (36 hips). Among the surgical methods, proximal femoral nailing was the most frequent, in 21 hips, and anesthesia was administered by general anesthesia in 23 hips and by spinal anesthesia in 16. The mean hospitalization period was 34.3 days (minimum 7 days, maximum 104 days); patients for whom the hospitalization period was 30 days or longer were the most common (16 hips). The surgery waiting time, surgical method, and hospitalization period within the NHR group were not significantly different than in the CDI group (Table 1).

The degree of capacity for activity before hip fracture and at final post-fracture follow-up was classified into three levels. Level 1 was defined as cases wherein the patient could perform activities alone or use walking-assistance tools (cane, walker) without being helped by others (independent ambulator); level 2 were cases in which the patient could perform activities only when helped by others (dependent ambulator); level 3 cases were those in which the patient could not conduct any activities other than moving in a wheelchair or resting in bed (non-functional ambulator). The capacity for activity and death were evaluated by retrospectively collecting information from hospital records, nursing home records, and phone calls to patients or their lineal family members. When information could not be easily obtained from patients or their lineal family members or patients could not be contacted in the process of the survey, contact information could be obtained with the help of government offices in order to determine accurately whether the patients had died or not and, in cases of death, when the death had occurred.

The period of time from the date of admission to the nursing home and the date of injury, along with the season and time of injury of each patient, were examined. The seasons of injury were divided into March through May (spring), June through August (summer), September through November (autumn), and December through February (winter); the time of injury was divided into three-hour intervals except for the period from 24:00 to 06:00, which was established as a single time period (sleeping time).

Age, weight, height, body mass index (BMI), the capacity for activity before injury, and any comorbidity possibly influencing patients' falls were examined. Comorbidity was evaluated by determining whether the fracture was accompanied by any lung disease, cerebrovascular disease, cardiovascular disorder, hypertension, diabetes, Parkinson's disease, or dementia.

Considering the typical characteristics of patients admitted to nursing homes, it was thought important to examine the state of mobility immediately before death in cases where the patient died within a short period after hip fracture. Therefore, the capacity for activity at final follow-up was examined in all cases, and patients who survived for at least one year underwent additional analysis. Changes in the capacity for activity were divided into maintenance, minor decline (one-level decline), and major decline (from independent ambulation to non-functional ambulation) so as to examine changes from before hip fracture to the time of final follow-up. The associations between age, fracture type, and comorbidity possibly affecting recovery of either capacity for activity or capacity for activity prior to hip fracture were examined.

Those who died within one year of hip fracture were analyzed. Patients' age, BMI, comorbidity, number of accompanying diseases, whether the patient had dementia or not, fracture type, whether the patient underwent surgery or not, and changes in the capacity for activity that were considered to have possibly influenced death were examined. For comparison of mortality rates between the NHR group and the CDI group, survival analyses were conducted.

The kinds and contents of disputes occurring between patients or their family and nursing homes after the occurrence of hip fracture were examined. The kinds of disputes and resolution processes were analyzed based on public-record documents and insurance-related documents kept in the nursing homes.

A control group (CDI) was organized by selecting a group with the smallest difference in age, specifically by repeating propensity-score matching processes (matched-pair analysis performed 100 times) with patients selected based on age (at least 65 yr old) and gender from those who visited the emergency room of this hospital due to hip fracture while living at home (excluding pathological fractures and bilateral fractures). Kaplan-Meier survival analyses were conducted using death within one year as the end point as well as Student's t tests and chi-square tests for comparisons between the two groups and comparisons between the dead and survivors within each group. For the NHR group, factors that were considered to have affected the capacity for activity and death within one year were analyzed by logistic regression using SPSS 18.0 (Chicago, IL, USA). Differences wherein the P value was smaller than 0.05 were considered to be statistically significant.

The design and protocol of this retrospective study were approved by the institutional review board of Gwangju Veterans Hospital (GJVH-IRB No. 2014-6-2). All patients were informed that their medical data could be used in a scientific study and provided their consent.

The mean period of time from the date of admission to the nursing home to the date of fracture was 484.2 days (minimum 2 days, maximum 1,925 days). Fractures occurred the most frequently (14 hips; 30.4%) during the period between six months and 12 months after admission. The classification of fracture by season revealed that fractures occurred the most frequently (17 hips; 36.9%) between December and February. With regard to the time of injury, fractures occurred the most frequently (11 hips; 24%) between 24:00 and 06:00; the incidence rate was highest (22 hips; 47.9%) between 21:00 and 09:00, the time considered to correspond most closely to the sleeping period (Table 2).

Among the patients, males and females had fractures that involved 28 and 18 hips, respectively. The mean age was 82.7 yr (minimum 66 yr, maximum 94 yr), and those in their 80s were the largest in number at 31 cases. The mean BMI was 19.9 kg/m² (minimum 15.7 kg/m², maximum 32 kg/m²). The BMIs were divided into three levels for the analysis (level 1, lower than 18.5; level 2, 18.5-22.9; level 3, 23 or higher): 18 hips were at level 1, 19 at level 2, and 3 at level 3. According to the evaluation of the capacity for activity before hip fracture, dependent ambulators were the largest in number, at 22 cases. Of the total 46 cases, 27 were dementia patients; among the other accompanying diseases, hypertension was shown to be the most common (29 cases) (Table 1).

In both groups, intertrochanteric fractures occurred more frequently than other types of fractures; and there was no significant difference in fracture types between the two groups. The NHR group showed significantly lower BMIs (P=0.002) and capacity for activity before hip fracture (P=0.002) as well as significantly more cases of cerebrovascular disease, Parkinson's disease, and dementia. The cases wherein the number of accompanying internal-medicine diseases was 3 or larger (P=0.828) and those in a high-risk group with an ASA score for physical status before surgery of 3 or more (P=0.460) were larger in the NHR group, though the differences were not statistically significant (Table 1).

Among the 46 subject patients in the NHR group, the level of capacity for activity was maintained in 23 patients, dropped by one level in 19, and dropped by two levels in four. In the CDI group, the levels were maintained in 31, dropped by one level in 13, and dropped by two levels in two. The differences between the two groups were not statistically significant (P=0.226). However, the number of non-functional ambulators increased by 7 in the CDI group and by 17 in the NHR group, reflecting significant declines in the capacity for activity (P=0.000) (Table 3). Logistic regression analyses were conducted at final follow-up in order to determine and analyze the factors affecting such declines. Based on the results of the analyses, it was concluded that hypertension (P=0.075, odds ratio, 2.176) or multiple accompanying diseases (P=0.083, odds ratio, 1.687) could affect the declines, though their effects were not statistically significant (Table 4).

Among the patients in the NHR group, 35 survived for at least one year after hip fracture, while the capacity for activity was maintained at final follow-up in 19, declined by one level in 15, and declined by two levels in one case. In the CDI group, 36 survived for at least one year after hip fracture in the CDI group, while the capacity for activity was maintained at final follow-up in 23, declined by one level in 11, and declined by two levels in two (P=0.518). The differences in these changes within the NHR group were not statistically significant. However, in the evaluation of the capacity for activity at the final follow-up, the NHR group showed significant declines (P=0.002) as well as an increase of 11 in the number of non-functional ambulators, while the CDI group showed an increase of only 6. This significant difference was attributed to the larger number of dependent ambulators prior to hip fracture in the NHR group (P=0.001) (Table 5).

The analysis of factors that could affect declines in the capacity for activity in the NHR group indicated that the capacity for activity had significant effects (P=0.041, odds ratio , 2.996) and that accompanying cerebrovascular diseases were highly likely to have effects, though this result was not significant (P=0.12, odds ratio, 3.025) (Table 6).

In the NHR group, 11 patients died within one year (23.9%), and among them, death occurred a mean of 161.1 days (minimum 21 days, maximum 320 days) after hip fracture. The cumulative number of deceased patients by period was 3 within three months, 7 within six months, and 11 within 12 months. In the CDI group, 10 (21.7%) died within one year after hip fracture; and among them, death occurred a mean of 127.4 days (minimum 32 days, maximum 266 days) after hip fracture. The cumulative number of deceased patients by period was 4 within three months, 7 within six months, and 10 within 12 months. Kaplan-Meier survival analyses of the two groups showed no significant differences between them (P=0.842) (Fig. 1) (Table 7).

According to the comparison between those who died within one year and those who survived (35 hips, 76.1%) in the NHR group, patients' age, comorbidity, and fracture type had no relationship with survival. However, although not statistically significant, whether the patient underwent surgery or not (P=0.097, odds ratio, 4.571) and a decline in the capacity for activity (P=0.07, odds ratio, 2.705) were considered highly likely to have a significant relationship. Additionally, logistic regression analyses of the risk of death within one year indicated that on the basis on an odds value of 1, the risk of death of patients with hypertension was approximately 2.6 times higher, and that of patients with cardiovascular diseases was approximately 2.8 times higher. There were three cases where the patient did not undergo surgery but survived for at least one year. Of these, two were non-functional ambulators before fracture and Garden classification type 1 and type 4 femoral neck fracture patients, respectively, and the remaining single case was an independent ambulator before fracture who became a non-functional ambulator after fracture (and had an AO classification type 2 intertrochanteric fracture) (Table 8). The three cases of postoperative infection that could have affected death within one year all occurred in the NHR group (P=0.078) (Table 7). Of these, two patients survived for at least one year after removal of artificial joints in the final surgery, though the remaining one patient who had undergone cement spacer insertion under a planned two-stage replacement died within one year.

Among the patients who had hip fractures that occurred in nursing homes, insurance companies that had contracts with nursing homes compensated patients and their family members in 32 cases (69.6%). Patients' family members raised civil complaints or filed lawsuits against nursing homes in 41 cases. The contents of civil complaints (41 cases) and lawsuits (2 cases) overlapped in the evaluation, and among them, claims for nursing expenses during hip fracture treatment (37 cases) and complaints about caregivers in nursing homes (27 cases) were the most frequent; there also were complaints about initial responses to non-functional ambulators because fractures had been overlooked for several days (3 cases). In light of the fact that the civil complaints included many indicating that after hip fracture, the patient could not be adequately treated at hospitals or at nursing homes (22 cases), the period after fracture and before being readmitted to a nursing home to start living there again (the period before nursing home readmission) was examined among patients who survived for at least one year. The period before readmission was classified into four levels: 7 days or less, 8-14 days, 15-28 days, and 28 days or more. Cases of 15-28 days were the most frequent (11 patients). Patients who survived for at least one year were divided into those with at least one-level declines ("any decline") in the capacity for activity and those with no decline, in order to determine whether the readmission period was related to a change (a decline in all cases) of any kind. Of the 35 patients who survived for at least one year, 16 showed at least one-level declines along with significantly shorter periods before readmission (P=0.015). Additionally, logistic regression analyses indicated that the period before readmission was related to declines in the capacity for activity (P=0.009, odds ratio, 7.277).