Background
Proximal humeral fracture (PHF) is a common fracture in people aged > 60 years old [
1,
2], with PHF accounting for about 9.5% of fall-related fractures [
2]. High prevalence of osteoporosis worldwide due to an aging population [
3] has contributed to an increased incidence of PHF particularly in women [
2]. Incidence of hospitalization due to PHF in Australia increased from 26.8 per 100,000 person-years in 2008 to 45.7 per 100,000 person-years in 2017 [
1]. Treatment associated with PHF is also costly, with median hospitalization costs being US$16,447 for surgically managed patients and US$7226 for conservatively managed patients [
4]. Hospital admission and hospital length of stay (LOS) are two outcomes that quantify the amount of care a patient needs post PHF. Currently little is known about what factors are associated with hospital admission post PHF [
5,
6]. Hospital LOS has been used as a secondary outcome to evaluate outcomes following different surgical treatment methods [
4,
7‐
12], however the impact of other factors on LOS post PHF has been less explored. While increased LOS can be indicative of the complexity of a patient [
13], it may also be reflective of “inefficient hospital processes” ([
14] p12); treatment delays and poor discharge planning [
14]. Early identification of people at risk of longer LOS may assist with timely discharge planning. Another way outcomes can be evaluated is whether a person can return to their premorbid living situation following hospital discharge. Previous evidence suggests that factors such as older age [
15‐
19]; gender [
16,
18]; comorbidities [
15‐
18]; premorbid function [
18,
19]; LOS [
15,
16,
19]; concomitant fractures [
15]; acute medical complications [
15,
16] and/or surgical treatment [
15,
19,
20] may predict discharge destination, however these studies did not include premorbid living situation (i.e. the premorbid presence or lack of family or friends) as a factor. Understanding what factors may affect patient discharge destination post PHF may improve discharge planning following hospitalization, ultimately improving patient care and reducing health care costs.
There are several factors that may impact patient hospitalization, LOS and discharge destination post PHF that have previously not been explored. Orthopedic restrictions (weight-bearing restrictions and/or use of sling) following both surgical and conservative treatments can prevent the use of gait aids and thus impact on a patient’s mobility status. Similarly, a patient’s upper-limb dominance or premorbid ability to manage their personal self-care may further limit their functional independence. In addition, while LOS progressively increases with subsequent admissions due to osteoporotic fracture [
21], previous literature has not analyzed how osteoporosis influences hospital outcomes specifically post PHF. Therefore, the aims of this research are to examine in patients with PHF factors associated with (1) hospital admission; (2) LOS for patients admitted to hospital and (3) new discharge destination for patients that both present and/or are admitted to hospital.
Discussion
Discharge planning for patients presenting to hospital post PHF is a complex task. Clinicians need to consider what factors contribute to this decision making to provide patients with PHF the best care and reduce health care costs. We found that approximately one in three patients presenting with PHF required hospital admission, and one in seven were discharged to a different living situation. We also identified that patients with greater comorbidities (as indicated by the CCI) were more likely to require hospital admission, a longer LOS and new discharge destination. Conversely, compared with patients who lived home alone premorbidly, patients who were previously living at home with family/friend(s) or from an external care facility were less likely to require hospital admission, have a longer LOS or have a new discharge destination. In addition, several other clinical, fracture or premorbid factors were associated with one or more of the primary outcomes.
Our finding that a higher CCI score was associated with worse outcomes is in keeping with the limited existing literature, which found that an increased CCI score is associated with hospital admission [
5]; a longer LOS [
6] and new nursing home admission on discharge [
17]. Although the CCI has been suggested to be a valid mortality predictor tool for patients with PHF [
25,
26], there has been some criticism that condition weightings are outdated [
27]. By comparison, the Elixhauser comorbidity measure has been found to be superior to the CCI at discriminating inpatient mortality with PHFs [
28]. Nonetheless, these findings highlight the importance of considering a person’s comorbidities when planning likely outcomes and suggest that a comorbidity index should be considered for inclusion in any predictive model.
In contrast, patients who were from home with family/friend(s) or from an external care facility had a decreased likelihood of hospital admission, a shortened LOS and a reduced risk of a new discharge destination compared with patients who previously lived home alone. Of note, of the 109 (15.5%) participants requiring a new discharge destination, 49 (45%) were people who transitioned from home alone to home with family/friend(s) and only 29 (26.6%) were people discharged to an external care facility. Comparison to the literature is limited as previous studies lacked detail on how the patient’s premorbid living situation changed on discharge [
15‐
20], with most only assessing discharge to an external care facility as an outcome [
15‐
17,
20]. This supports previous research, which found patients with PHF have a significantly lower risk of residential care placement compared to other fragility fractures (using hip fractures as a reference) [
29]. Our findings likely reflect the increased care needs, either temporarily or permanently, that patients experience post PHF and highlights the importance of considering premorbid living situation in planning.
Surgical treatment (compared to patients treated conservatively) was also found to be associated with a shorter LOS. Everyone who required surgery required hospitalization, with most patients discharged quickly post-operatively (median LOS was 5 days (IQR 3, 18), one in three were discharged in two nights or less). Conversely, not every patient who was treated conservatively required hospitalization. Conservatively managed patients admitted to hospital likely required admission due to other factors, such as comorbidities and additional medical diagnoses or injuries. These issues generally do not resolve quickly thus require a longer LOS, which may have included subacute admission. Our inclusion of these settings in our study’s LOS may also explain why our study contrasts to other recent literature, which found surgical treatment was associated with longer LOS [
4,
12].
Factors associated with both hospital admission and longer LOS included osteoporosis diagnosis, additional principle acute medical diagnosis and additional significant injuries. Other studies [
5,
6,
15,
17‐
19] have not explored the relationship between osteoporosis and outcomes; it may be an indicator of frailty which may explain the association with negative outcomes. The remaining results are in keeping with three previous studies, which found patients who had “polytrauma” ([
5], p156) were more likely to be hospitalized and patients with additional acute medical issues post-operatively [
30,
31] were associated with a longer LOS. These results conflict with one study conducted in a specialist trauma centre, which found no differences in LOS between patients with isolated PHFs and “concomitant fractures” ([
32], p102). We believe patients with these additional medical diagnoses or significant injuries are more likely to need hospitalization and a longer LOS in order to receive further medical treatment and provide more time to recover and rehabilitate.
Additionally, we found some factors were only associated with either hospital admission or longer LOS. Our findings that displaced fracture and premorbid use of a frame were associated with hospitalization is supported by Myeroff et al. [
6] who found Neer fracture classification (> 1 vs 1, i.e., displaced fracture) and premorbid use of a “cane/walker/wheelchair.” ([
6], p4) for mobility were predictors of hospital admission. These patients presenting may be more likely to need hospitalization to ensure that the relevant medical specialities can decide the best treatment methods and/or allied health can provide input for management of mobility issues if gait aid use is restricted. We also found that patients who required assistance with premorbid personal self-care were associated with a longer LOS. While it is recognised that “most” ([
33], p885) people require family or friends to assist with personal self-care following PHF injury [
33], with previous literature only reviewing outcome measures to assess this function [
34,
35], no research has investigated how premorbid personal self-care function influences hospital outcomes. People who needed assistance with personal self-care premorbidly are likely to have substantial care needs following PHF. These additional needs may necessitate a longer LOS to allow patients and/or carers to learn adaptive strategies and rehabilitate post fracture or obtain community based personal-care services.
Two factors, older age and dementia diagnosis, were found to be associated with new discharge destination. While direct comparison is difficult, this is in keeping with previous studies [
15,
17‐
19,
36], which also found age and dementia diagnosis were associated with discharge to “short-term or long-term care facility” ([
15], p1703), “new nursing home” ([
17], p1604), or non-home discharge [
18,
19,
36]. Although it appears that older age [
15,
17,
19] and dementia [
15] may be factors associated with discharge to a residential care facility post PHF, more research with clearly defined premorbid and discharge destination locations needs to explore this change directly. In addition, further research needs to explore the influence of dementia diagnosis on patient discharge destination outcomes post PHF, as this has only been analyzed in one previous study [
15] which differed in its methods of diagnosis to our study. While conflicting results in a study by Wang, Youssef and Smerdely [
17] found no association between cognitive impairment and a new nursing home admission post PHF [
17] this may be explained by the severity of the condition. It also however raises the question on whether patients with PHF and a diagnosis dementia were deemed by clinicians to have limited capacity to make functional gains, therefore required a new discharge destination to ensure new caregivers at home or in an external care facility could take on their increased care needs. Further exploration of issues surrounding access to rehabilitation and discharge planning in people with dementia is warranted.
Improved understanding on factors that are associated with hospitalization, LOS and discharge destination post PHF will assist guiding development and implementation of a tool to assist decision making and discharge planning. Many of the factors we found to be associated with our primary outcomes may also be linked to patient frailty, therefore frailty indexes may be a more feasible way to help clinicians to accurately analyze patients. The current evidence however to support the use of frailty indexes is conflicting in patients post PHF. The modified Frailty Index (mFI-5) [
37], which assesses similar comorbidities as the modified CCI (mCCI) in addition to functional status, has been found in patients post PHF surgery to be strongly associated with adverse events [
38] or complications [
39], while also predictive of readmission rates and the need for inpatient rehabilitation admission [
39] and therefore risk of a longer LOS. In contrast, Yi et al. [
38] found that both the mFI-5 and the mCCI had a limited ability to predict complications following PHF surgery. Another alternative tool includes the Score for Trauma Triage in the Geriatric and Middle-Aged (STTGMA) [
40] which assesses patient age and CCI score, in addition to level of consciousness, the mechanism of injury and its severity at presentation [
41]. The STTGMA tool may assist in “early decision making processes” ([
40], p6) as one study found patients with PHF and high risk scores had a longer LOS, increased hospital needs and had fewer discharges directly home [
40]. To date, the STTGMA has only been validated retrospectively [
41] and the one study completed in patients with PHF only had a small sample size and was limited to adults 55-years or greater (as opposed to adults 18-years or greater) [
40].
The main limitation of this study is that it was conducted retrospectively using data from a single health service. Results are likely reflective of people in a lower socioeconomic community. All data was collected based on documentation in the medical records and imaging reports utilising a specifically designed data audit tool. Missing or incorrect information may have caused errors in the data collected. Data were audited over a 54-month period. While there were no organizational changes that would have impacted on discharge planning over this time period, it is possible that local changes in clinical practice may have impacted LOS. Inspection of LOS data by year demonstrated no consistent changes in LOS. We were unable to include other confounding variables, such as ethnicity and pain, as these were not always recorded in a systematic way during the study time period. Nonetheless, the audit was conducted on a large sample of participants (n = 701) which has allowed the inclusion of multiple variables in the analysis. This will assist future researchers in the identification of factors to be included in the development of predictive tools.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.