View Article PDF

Hip fractures in the older adult(those aged over 65 years) are common, affecting up to 18% women and 6% men in their lifetime.1 They are associated with significant mortality and burden of disease,2 even with prompt evidence-based treatment including surgery, orthogeriatric care and rehabilitation.2,3 In addition, approximately 10% of patients will experience a second hip fracture, which has further negative impacts on their mortality and quality of life.4

The older population in New Zealand is growing rapidly, particularly in Canterbury New Zealand, where people over 65 years make up 16.1% of the total population, up 9,500 people in the last five years. In New Zealand, it is estimated that about 72 out of 100 people with hip fracture are admitted from home but following the fracture only 55 out of those 72 people will return home.5 Given the impact of hip fracture on morbidity, mortality and quality of life, prevention programmes focus on treatments in two areas, falls prevention using physiotherapy intervention and secondary prevention of fractures (ie, treatment of osteoporosis) using medications. Pharmacological treatments for osteoporosis include bisphosphonates (eg, zoledronic acid, alendronate, risedronate),parathyroid hormone analogues (eg, teriparatide), the RANKL antibody denosumab6 and more recently discovered agents such as the sclerostin inhibitor romosozumab.7 Despite the range of treatments available, many patients remain under- or un-treated for osteoporosis and physicians encounter major challenges when implementing appropriate treatments.8

Following hip fracture, older patients may be cared for in a range of settings ideally with input from both geriatric and orthopaedic specialists.9 At Christchurch Hospital, patients over 65 years with fractures are routinely seen by a HealthCare of the Elderly Physician as a shared care model that helps manage the patient’s comorbidities and also establish whether inpatient rehabilitation is required before safe discharge home.10,11 Patients may undergo inpatient rehabilitation in older person’s health wards(OPH), for general rehabilitation under care of a geriatrician, the orthopaedic rehabilitation unit (ORU), a specialist orthogeriatric rehabilitation ward, also under care of a geriatrician. Generally, patients selected for orthogeriatric rehabilitation are done so on the basis that they have fewer co-existing active medical problems and their orthopaedic problem is their primary rehabilitation issue. Alternatively, patients may be discharged directly to their own home, ideally with nurse-led fracture liaison follow-up with the Fracture Liaison Service (FLS) to ensure they receive adequate secondary fracture prevention.12,13 The FLS nurse currently uses the Australia New Zealand Hip Fracture Registry5 to identify patients undertreated for osteoporosis, the nurse and organises bone density testing and/or general practitioner-led osteoporosis treatment, eg, bisphosphonates.14 Choice of rehabilitation location depends on a range of factors including the patient’s pre-morbid function, their medical comorbidities, cognitive function and social factors. There is an established benefit of both orthogeriatric and general geriatric rehabilitation for patients with hip fractures and other acuteconditions.15 Patients who attended both general geriatric and orthopaedic rehabilitation had lower mortality rates, greater function and were less likely to be admitted to nursing homes.15 This effect was seen both at discharge and at follow-up.

The main aim of the current study was to retrospectively audit patient characteristics, surgical management practices, rehabilitation practices and outcomes, osteoporosis treatment prescription and mortality for patients with hip fractures admitted to Christchurch Hospital. In particular, we aimed to compare outcomes for four groups: 1) publicly funded aged residential care (ARC)residents discharged directly from acute orthopaedics, 2) non-ARC residents discharged directly from acute orthopaedics, 3) patients discharged after orthogeriatric rehabilitation and 4) patients discharged after general geriatric rehabilitation. As a secondary aim we also looked at the differences between orthogeriatric and general geriatric rehabilitation groups in respect to length of stay, comorbidities, functional status and treatment for osteoporosis.

Method

Data acquisition

This was a retrospective cohort study including all consecutive patients aged 65 and over who presented to Christchurch Hospital between 1 May 2017 and 30 April 2018 with hip fracture. Patient data was extracted from multiple data sets such as the Administrative Hospital data, Marriage, Birth and Death Data (Department of Internal Affairs New Zealand), Client Claim Payment System (CCPS) (Ministry of Health) and Australian Rehabilitation Outcome Centre (AROC) data. Multiple querying tools were used to validate and identify the cohort from the hospital patient management system, coding and data warehouse, particularly signals from noise® (sfn). This is proprietary software from Lightfoot Solutions which uses the principles of Statistical Process Control. Canterbury District Health Board customised the sfn tool with tables that enable people to explore data through a process flow perspective.

Initially, the data extracted from sfn identified patients admitted to Christchurch Hospital with hip fracture based on International Classification of Diseases, Tenth Revision (ICD-10)codes. Then the software was used to determine the discharge location for these same patients. The discharge location determined the cohort selection. Cohort 1(ARC residents) and 2 (not ARC residents) were discharged directly from hospital, cohort 3 were discharged to orthopaedic rehabilitation (ORU) and cohort 4 to general geriatric rehabilitation.

The following ICD-10 codes were used to identify the cohort with hip fracture, based on the primary diagnosis (ie, their primary reason for hospital admission): S72.00 Fracture of neck of femur, part unspecified S72.01 Fracture of intracapsular section of femur, S72.03 Fracture of subcapital section of femur, S72.04 Fracture of mid-cervical section of femur, S72.05 Fracture of base of neck of femur, S72.08 Fracture of other parts of neck of femur, S72.10 Fracture of trochanteric section of femur, unspecified; S72.11 Fracture of intertrochanteric section of femur, S72.2 Subtrochanteric fracture. These patients had an ICD-10 primary procedure code:47519-00 Internal fixation of fracture of trochanteric or subcapital femur,47522-00 Hemiarthroplasty of femur, 49315-00 Partial arthroplasty of hip,49318-00 Total arthroplasty of hip, unilateral) or other treatment code (to include patients managed conservatively). These were then grouped into two groups: neck of femur and trochanteric fractures. Patients with periprosthetic and pathological fractures were excluded.

Variables

Data on a range of dependent variables were identified and recorded for the included patients; including age, National Hospital Identifier (NHI), sex, ICD-10 code, primary surgical procedure, total LOS, comorbidities, mortality, location of domicile on admission (ARC funding data and Australasian Rehabilitation Outcomes Centre[AROC]), location of domicile on discharge (ARC funding data and AROC) and change in Functional Independence Measure (FIM) score during rehabilitation(AROC data). To quantify patient comorbidity, we used a modified Charlson Comorbidity index (CCI), excluding age. Although designed for patients underage 65, the CCI can predict short- and long-term mortality in acutely ill hospitalised older adults16 and was chosen for this study so data could be easily collected from available sources. Comorbidity data was collected based on all coded diagnoses, rather than just new diagnosis codes. Prescribed medications were not available in any coded data-sets and so this data was manually extracted from patient medication and reconciliation history linked with each patient’s National Health Identifier by the first, third, fourth and fifth authors. Referral information from the Fracture Liaison Service was obtained.

Statistical analysis

Statistical analysis was performed using SAS 9.4. Chi square test was used for comparison of categorical variables except when cell sizes were small, then Fisher’s exact test was used. Normality assumptions were found to be violated for continuous variables. Hencenon-parametric Mann-Whitney test was used for comparison between two independent groups and Kruskal-Wallis test for more than two groups. For both categorical and continuous variables, if p-value for overall comparison was significant, post-hoc multiple testing with Bonferroni correction was performed to ascertain which pairwise differences were significant. To analyse the possible factors associated with mortality, a Cox proportional hazard model was developed. A stepwise selection method was used with an entry level significance of 0.2 and significance level of 0.1 for variables to stay in the model.

Results

Demographic characteristics

Over the one-year study period, 415patients over age 65 were admitted to the acute orthopaedic service with hip fracture, with over 421 episodes of care (ie, some patients presented twice).Of these patients, 67% (n=278) were referred for inpatient rehabilitation. The remaining patients were discharged directly from the acute orthopaedic ward. A higher proportion of patients who went to geriatric rehabilitation (OPH) were admitted from residential care, compared to those who went to orthogeriatric rehabilitation (ORU, see Table 1). In all groups, the majority of patients were women, and most were of European ethnicity. Demographic characteristics for each of the patient groups are detailed in Table 1. Median age for non-ARC-residents discharged from acute orthopaedics was significantly younger than patients in the other groups and approximately 20% of the patients in this group lived outside the Canterbury region.

Table 1: Demographic characteristics for patients in each group (n=415), including age, gender, ethnicity, domicile, aged residential care status, discharge status and mortality. Variables with significant between group differences are highlighted in bold.

Fracture type, wait for theatre and surgical management and length of stay

For each cohort, patients were more likely to present with a neck of femur than a trochanteric fracture (see Table2). As seen in Table 2, total arthroplasty was more likely to be performed in the non-ARC residents discharged from orthopaedics and the ORU patients. The ‘Other’ category included non-operative management, eg, allied health intervention, and other surgical procedures (potentially related to additional presenting problems).

Of patients who proceeded for surgery (n=395), over half of patients were operated on within 24 hours (n=210,53%) and most patients (n=324, 82%) were operated on within 48 hours of presenting. There were no statistically significant differences between groups in wait times for theatre (p=0.39) (see Table 3). Patients discharged from general geriatric rehabilitation had a longer length of stay than other patients (p<0.0001, see Table 3). No clear disparity in LOS was found between groups discharged directly from acute orthopaedics.

Table 2: Fracture types and primary procedures for the four patient groups admitted with hip fracture.

Note: n=421 episodes of care with some patients presenting for more than one episode of care, hence differing n-values for each patient groups when compared to total patient numbers (eg, in Table 1). For primary procedures there are missing data for four patients, so group totals are n=45, n=93, n=185 and n=94 respectively.

Previous and subsequent fractures

Rates of fracture prior to hip fracture and in the 180 days following hip fracture were also analysed. There were no between-group differences in these rates (see Table 3). Approximately 25% of patients (n=105) had a hospital presentation for a fracture prior to their hip fracture admission and 5% of patients (n=21) presented for treatment of a fracture after their hip fracture (noting variable follow-up times). A number of patients (n=181, 43%) had also had at least one hospital admission within the 365 days prior to their hip fracture.

Table 3: Total length of stay (including both acute and rehabilitation admissions, if applicable) for the four groups in days (n=421 episodes of care) with wait for operation times, prior and subsequent fractures and re-admission.

Note: n=421 episodes of care with some patients presenting for more than one episode of care, hence differing n-values for each patient groups when compared to total patient numbers (eg, in Table 1).
*Wait for operation calculated on total participant numbers of n=40, n=80, n=168 and n=88 for each group respectively, given some patients did not proceed for theatre. Percentages computed on valid n only.

Mortality and survival analysis

Patients admitted to orthogeriatric rehabilitation were less likely to die within 30 days than other patients(p<0.0001, see Table 4). No clear mortality difference was found between the other three groups. ARC residents discharged from acute orthopaedics were more likely to die at six months post-presentation than patients in the other three groups. Patients transferred to general OPH had a higher mortality at six months compared with the ORU (p<0.0001). No clear mortality difference was found between the other groups. Figure 1 show the survival probability in the four groups over 20 months. The overall mortality in all groups at 180 days was 27% and 10% at 30 days. In-hospital mortality (ie, patients who died while admitted for their hip fracture) was 5% across all groups.

No difference was found in mortality and LOS for patients with fracture neck of femur compared with trochanteric fracture. Patients with total arthroplasty have a significantly shorter length of stay and reduced mortality compared to the two other groups (see Table 5).No difference in LOS or mortality was found between patients having hemiarthroplasty or internal fixation.

To further assess factors associated with mortality, a cox proportional hazard model was developed (see Table 6 and Figure 1). Based on this model, risk factors for overall mortality were: ARC residents, a higher Charlson index and older age. The length of stay and time to wait for operation were not correlated with overall mortality.

Table 4: Mortality data at 30 days and 180 days for each of the four groups (n=415 patients).

*Percentages shown reflect proportion of total patient population.

Table 5: Mortality and LOS in relation to operative procedure used.

Table 6: Cox proportional hazard model.

Figure 1: 20-month predicted survival plot for the four groups.

Table 7: Modified Charlson Comorbidity Index (CCI) for each group and comorbidities. Variables with significant between group differences are highlighted in bold.

Comorbidities and Modified Charlson Index

The Charlson Comorbidity Index (CCI)for each group is shown in Table 7. The index was calculated with excluding patient age, given all patients were over age 65. Non-ARC residents discharged directly from acute orthopaedics had a significantly lower Charlson index than other patients. No clear disparity was found between the other three cohorts. In terms of specific comorbidities, the ARC residents discharged directly from acute orthopaedics and the general geriatric rehabilitation group had significantly higher rates of dementia than the other two groups. There were significantly more patients with congestive heart failure in the ORU group(32%) compared to the non-ARC residents discharged from acute orthopaedics(15%). The non-ARC residents discharged from acute orthopaedics were less likely to have dementia and diabetes complications compared to other groups. Patients admitted to general geriatric rehabilitation were more likely to have renal disease dementia, diabetic complications and congestive heart failure than non-ARC residents discharged from acute orthopaedics. There were otherwise no differences found between groups.

Osteoporosis treatment and follow-up via Fracture Liaison Service for secondary fracture prevention

Data on osteoporosis treatment for each group at time of discharge are presented in Table 8. Due to the diversity of medication regimes prescribed, these were categorised by the first author (SW) into five categories (see Table 8). Appropriate treatment included bisphosphonates (alendronate, zoledronic acid, risedronate), teriparatide ordenosumab, all with colecalciferol 1.25mg monthly (50,000 IU of Vitamin D). Colecalciferol alone or no treatment was considered suboptimal. ARC residents discharged from acute orthopaedics were more likely to have suboptimal treatment with no documentation than any other group. Patients discharged from orthogeriatric rehabilitation were most likely to have appropriate treatment with osteoporosis medications, further investigation/referral to FLS or treatment actively with-held due to comorbidity (88% patients, p<000.1).Rates of consideration or treatment for osteoporosis were 62% in general geriatric rehabilitation group and 46% in non-ARC residents discharged from acute orthopaedics.

Fracture Liaison Service (FLS)attendance was analysed for Canterbury residents only, as residents from other areas of New Zealand or overseas are not usually seen by the FLS in Canterbury. As protocols existed for inpatient treatment for osteoporosis, FLS referral was not activated for many patients undergoing inpatient rehabilitation. The FLS also does not currently follow-up patients already on appropriate bisphosphonate or other osteoporosis treatment. Rates of FLS referral were low in ARC residents discharged from acute orthopaedics (11%). Referral rates were higher in the non-ARC residents discharged from acute orthopaedics (p<0.0001), of whom 41% (31 patients) were seen by the FLS. It is important to note that 5% of patients in this group also died in hospital so were not appropriate for referral.

Table 8: Osteoporosis treatment for each patient group.

*Appropriate treatment was with bisphosphonate, denosumab or teriparatide and vitamin D, or with vitamin D alone if patient was proven not to have osteoporosis on DEXA scan.
**Suboptimal treatment due to comorbidities were patients not considered for osteoporosis treatment beyond vitamin D due to renal failure, drug intolerances, previous atypical femoral fracture etc.
***Suboptimal treatment was considered to be no vitamin D or vitamin D alone without evidence of further investigation for underlying osteoporosis.

Level of care and functional independence measure scores

For patients discharged directly from acute orthopaedics, data was only available for their ARC funding but not level of care within this funding. For the non-ARC residents discharged directly from Christchurch, 19 patients were transferred to other hospitals in New Zealand and 11 patients died during their hospital admission. Of those who were discharged directly “home”, two went to ARC, the remainder apparently went to private residence.

However, for patients admitted to rehabilitation, AROC data is collected on admission and discharge which provided more detailed information on admission and discharge location of residence (see Table 9). Most patients (83%) from the orthogeriatric group were in their own private residence before admission. On discharge, 70% of patients in the orthogeriatric group remained in their own private residence, but 6% went to rest home-level care and 7% went to hospital-level care. Approximately 10% of the population were admitted from rest home-level care but ended up in hospital-level care on discharge. In contrast, only 59% of the patients in the general geriatric rehabilitation group were admitted from private residence. On discharge, 43% of the patients in this group remained in their own residence, and 17% of patients had moved into rest home and hospital-level care (Table 9).

Table 9: Pre- and post-discharge location of residence for Cohort 3 and 4, based on AROC data.

Functional independence measures are also collected for patients on admission and discharge from rehabilitation (see Figure 2) and were available via AROC. Both FIM cognitive and motor scores were significantly higher in the orthogeriatric rehabilitation group on admission and discharge. There were no significant differences between the two groups in the improvement of scores on discharge. This data correlates well with the ANZHFR data set.5

Figure 2: Change in functional independence measures of patients in the two rehabilitation groups.

Discussion

This study presents data on hip fracture management and rehabilitation in older adults at a single centre in metropolitan New Zealand. It followed on from previous work at Christchurch Hospital which utilised sfn® in implementing and evaluating the Fast Track Hip Fracture pathway that remains in use.3 Christchurch Orthopaedic Department admits approximately 5,500 patients per year in the busiest trauma centre in New Zealand and of those over 400 are hip fractures. Despite the availability of the ANZHFR,5which collects data on hip fracture for patients around Australia and New Zealand, we chose to analyse data using sfn due to more flexibility and efficiency in comparing multiple variables, including some that are not collected by the ANZHFR (eg, comorbidities, previous hospital admissions, factors contributing to mortality). Direct comparison with the ANZHFR was not within our research aims and was not explored. The administrative data allowed us to explore all the characteristics of the heterogenous patient groups admitted following hip fracture, to identify possible contributors to the clinical outcomes for these patients and to aid future targeted service delivery.

Fracture type and surgical intervention

The fracture types and surgical procedures have not altered from our previous work.3 Time to surgery has decreased with no negative impact on mortality. There was no difference in mortality or LOS between fracture types, but mortality and LOS was less in THA reflecting appropriate operative procedure. This observation is consistent with guidelines for the operative management of hip fractures.5

Patients discharged directly home or to a care facility

Based on anecdotal experience, we anticipated the need to separate the patients discharged directly from their acute orthopaedic admission based on their pre-morbid level of care (ie, from private residence versus residential care) due to significant within-group heterogeneity. Our study provided detailed information on patients discharged directly from the acute orthopaedic ward in two groups: pre-morbid ARC residents and non-ARC residents. Non-ARC residents discharged from acute orthopaedics were more likely to be younger, less comorbid, and a significant proportion (20%)were from outside Canterbury. However, their younger age and less comorbidity were not reflected in a mortality benefit for these patients. The non-ARC residents discharged directly from hospital are the key group targeted by the Fracture Liaison service, and referral rates were suboptimal at around 40%

On the other hand, patients who were ARC residents and were discharged from acute orthopaedics were more comorbid and had higher rates of mortality at six months compared to the other groups. They had similar rates of dementia when compared to patients transferred to general geriatric rehabilitation. Patients in this ARC resident group were also more likely to have suboptimal osteoporosis treatment with no clinical documentation regarding clinical contraindication and very low rates of referral to the Fracture Liaison service.

Patients discharged following inpatient rehabilitation

Outcomes of this study are consistent with other literature on orthogeriatric or general geriatric rehabilitation for patients with hip fracture.15 The patients in our rehabilitation groups were not randomised, so a patient’s journey following hip fracture will be largely influenced by selection bias. However, patients in our study who had orthogeriatric or geriatric rehabilitation had a longer length of stay but higher rates of treatment with osteoporosis medications. Mortality in the orthogeriatric rehabilitation group was significantly lower than the other groups at 30 days, and both the rehabilitation groups had lower mortality than the ARC residents discharged from acute orthopaedics at the 180-day follow-up.

Osteoporosis treatment

High treatment rates in older patients with hip fracture may lead to a 40% reduction in major bone re-fractures.17 In this study, treatment for osteoporosis was variable for the four groups. FLS was generally not involved in the orthogeriatric rehabilitation group or the general rehabilitation group because protocols for osteoporosis treatment exist as part of the rehabilitation process. Treatment or consideration for treatment was 84%in the orthogeriatric group and 64% in the general group. However, in those discharged back to ARC facility or to another hospital or home were low at 15%and 46% respectively. These treatment rates compare favourably with other published data.18 These results are helpful because they identify where FLS should be involved. Also, accurate recording of inpatient treatment modality was identified as an issue; in-patient treatment was not captured on a database and needed manual extraction, which has been previously described.19 This may explain some of the low percentage figures from the ANZHFR and previous work.20 Similarly, to previous research,21 this work shows that inpatient treatment for osteoporosis seems to be the most effective way of ensuring initiation of treatment. These results suggest the need for more proactive FLS involvement for patients who do not have inpatient rehabilitation, and has led the hospital to introduce a mandatory field in the discharge template on osteoporosis treatment.

Mortality

This study also sought to determine local mortality rates and examine some of the factors possibly contributing to or predicting mortality. Thirty-day mortality rates reported in this study were higher than the most recent (2018) Annual Report of the ANZHFR5 for Christchurch, possibly due to differences in data source. Mortality rates vary significantly between studies, but 30-day rates in our study were largely comparable to the existingliterature;22 however, mortality rates at 180days were similar to rates at one year reported in some studies.23 The cox proportional hazard model showed that pre-morbid ARC residency was the strongest risk factor for mortality at 180 days, with patients of older age and with more comorbidities (based on Charlson Comorbidity Index) also at increased risk for death. Interestingly, total arthroplasty was also associated with improved mortality and this probably reflects selection of a healthier patient group for this procedure.

Limitations

This study primarily utilised data coded from medical records and explored variables through sfn and multiple dataset analysis. This means that the quality of the data is dependent on the way data was documented in patient records and then coded by independent clinical coders. Other data, eg, from the Department of Internal Affairs around death is likely to be accurate but released with sometime lag, hence why only 180-day mortality was included in the current study. AROC collection is also limited, as it only collects two points of outcome at admission and discharge, being cofounding by length of stay, with potential internal bias on analysis of rehabilitation improvement.

Also, regarding osteoporosis treatment data, we had access to hospital discharge summaries and electronic prescribing software which allowed us to see if treatment was prescribed in hospital, but we did not have access to longer-term data to determine if patients continued to use these medications in the community.

Finally, this was not a randomised treatment trial and as such, confounding factors in analysing outcomes could be significant. However, the data provides useful information around clinical management of this group of patients that may still alter management, eg, promote further consideration of osteoporosis treatment in certain groups of patients and greater uptake of inpatient rehabilitation.

Implications

This study shows that tools such as sfn, that easily explore multiple variables and cohorts across various data have utility in evaluating acute hospital pathways,3 but also in evaluating a patient’s whole hospital journey, including the full acute hospital stay and subsequent rehabilitation. Although this paper gives similar results to previous work,3,10 it is different as hip fracture patients are analysed according to their discharge destination. This enabled comparisons of various outcome across these four groups. It is pleasing to see the wait for operation has decreased from previous work3 to over 80% of patients having their operation within 48 hours.

Since the introduction of fast-track hip fracture management,3the length of stay in orthogeriatric rehabilitation has decreased to 14 days and to 21 days in general rehabilitation. Baseline functional status and discharge destinations were different between the two rehabilitation groups, with the orthopaedic rehabilitation group having higher baseline and post-rehabilitation functional status and a higher percentage of patients returning to private residence. Functional status on discharge for patients undergoing hip fracture rehabilitation (measured with the FIM score) is higher than the national average standard.5

Conclusions

We have demonstrated the outcomes in four groups of patients with neck of femur fractures. The time to operation and length of stay continues to decrease compared with our previous work. New data on the four post-operative groups show better outcomes (ie, return to domicile, treatment for osteoporosis and mortality) in the dedicated orthopaedic rehabilitation group. It is difficult to interpret the significance of these improved outcomes given these patients were less comorbid, specifically with less renal disease and dementia than the general rehabilitation group. This is likely to affect their functional outcomes and prescription of osteoporosis treatment. Enhancing involvement by FLS and increased prescription of osteoporosis treatment is needed for the patients discharged directly from acute orthopaedics.

Summary

Abstract

Aim

For older patients with hip fracture, we explored patient characteristics, outcomes and osteoporosis treatments for those admitted to rehabilitation compared to those discharged directly from hospital, using data set analysis.

Method

Retrospective cohort study including all consecutive patients (65 years and over) admitted to Christchurch Hospital over one year. Outcomes were compared for patients in four groups: 1) aged residential care (ARC) residents and 2) not ARC residents discharge from acute orthopaedics, and those discharged from 3) orthogeriatric or 4) general geriatric rehabilitation. Clinical data was extracted from hospital data warehouse using signals from noise.

Results

Over 12 months, 415 patients were admitted with hip fracture. Over half (n=282) were transferred for inpatient rehabilitation. Thirty-day mortality was 10%. Mortality at 180 days was 7% in orthogeriatric rehabilitation group and 8% in general rehabilitation group. Length of stay was shorter in orthogeriatric compared with general rehabilitation (median 12.9 vs 20.4 days). Osteoporosis treatment was addressed in 88% of patients in the orthogeriatric group compared with 62% in general rehabilitation group. More patients in orthogeriatric group discharged home compared with general rehabilitation group (70% vs 43%), but functional improvements were similar. For ARC residents discharged following acute admission, length of stay was short (median 6.5 day), but overall 180-day mortality was 7%. The not-ARC resident group discharged directly from hospital was less comorbid than the other groups with lower 180-day mortality (4%). Osteoporosis treatment rates were lower in patients discharged from acute admission (15–42%).

Conclusion

Mortality was highest in ARC residents discharged from acute orthopaedics. Patients admitted for orthogeriatric rehabilitation had shorter length of stay, lower 30-day mortality, were more likely to return home and most likely to be offered osteoporosis treatment (88%), noting less underlying comorbidity and better baseline functional status in this group. This paper supports further study (eg, randomised trials) to examine the effects of targeted post-surgical rehabilitation for patients with hip fracture and explore measures to increase uptake of osteoporosis treatment.

Author Information

Samantha Warhurst, Christchurch Hospital, Canterbury District Health Board, Christchurch; Soledad Labbe-Hubbard, Planning and Funding, Canterbury District Health Board, Christchurch; Ma Yi, Biostatistician, Canterbury District Health Board, Christchurch; Jane Vella-Brincat, Pharmacy, Christchurch Hospital, Christchurch; Tonya Sammon, Ambulatory Care, Well South, Dunedin; Jane Webb, Pharmacy, Canterbury District Health Board, Christchurch; Caroline McCullough, Department of Orthopaedic Surgery and Musculoskeletal Medicine, Canterbury District Health Board, Christchurch; Gary Hooper, Department of Orthopaedic Surgery and Musculoskeletal Medicine, Canterbury District Health Board, Christchurch; Nigel Gilchrist, Orthopaedic Medicine, Canterbury District Health Board, Christchurch.

Acknowledgements

Correspondence

Dr Nigel Gilchrist, Orthopaedic Medicine, Canterbury District Health Board, Burwood Road, Christchurch.

Correspondence Email

nigel.gilchrist@cdhb.health.nz

Competing Interests

Nil.

1. Veronese N, Maggi S. Epidemiology and social costs of hip fracture. Injury, 2018. 49(8):1458–1460.

2. Bhandari M, Swiontkowski M. Management of Acute Hip Fracture. New England Journal of Medicine, 2017; 377(21):2053–2062.

3. Gilchrist N, Dalzell K, Pearson S, et al. Enhanced hip fracture management: use of statistical methods and dataset to evaluate a fractured neck of femur fast track pathway-pilot study. The New Zealand medical journal, 2017; 130(1455):91–101.

4. Sobolev B, Sheehan K J, Kuramoto L, et al. Excess mortality associated with second hip fracture. Osteoporosis International, 2015; 26(7):1903–1910.

5. Australian & New Zealand Hip Fracture Registry, ANZHFR Bi-National Annual Report of Hip Fracture Care 2018., 2018.

6. Khosla S, Hofbauer LC. Osteoporosis treatment: recent developments and ongoing challenges. The Lancet. Diabetes & endocrinology, 2017; 5(11):898–907.

7. Cosman F, Crittenden DB, Adachi JD, et al. Romosozumab Treatment in Postmenopausal Women with Osteoporosis. New England Journal of Medicine, 2016; 375(16):1532–1543.

8. Solomon DH, Johnston SS, Boytsov NN, et al, Osteoporosis medication use after hip fracture in US patients between 2002 and 2011. Journal of Bone and Mineral Research, 2014; 29(9):1929–1937.

9. Grigoryan KV, Javedan H, Rudolph JL. Ortho-geriatric care models and outcomes in hip fracture patients: a systematic review and meta-analysis. Journal of orthopaedic trauma, 2014; 28(3):e49.

10. Thwaites J, Mann F, Gilchrist N, et al. Shared care between geriatricians and orthopaedic surgeons as a model of care for older patients with hip fractures. The New Zealand Medical Journal (Online), 2005; 118(1214).

11. Elliot J, Wilkinson TJ, Hanger HC, et al. The added effectiveness of early geriatrician involvement on acute orthopaedic wards to orthogeriatric rehabilitation. The New Zealand Medical Journal, 1996; 109(1017):72–73.

12. Osteoporosis New Zealand. Osteoporosis New Zealand Strategic Plan 2017–2020. 2017 18/06/2019]; Available from: http://osteoporosis.org.nz/wp-content/uploads/ONZ-2017-Strategic-Plan-WEB.pdf

13. Osteoporosis New Zealand. Bone Plan 2020: A systematic approach to hip fracture care and prevention for New Zealand. 2012 18/06/2019]; Available from: https://osteoporosis.org.nz/wp-content/uploads/Bone-Care-2020-web.pdf

14. Kanis J, Johnell O, De Laet C, et al. A meta-analysis of previous fracture and subsequent fracture risk. Bone, 2004; 35(2):375–382.

15. Bachmann S, Finger C, Huss A, et al. Inpatient rehabilitation specifically designed for geriatric patients: systematic review and meta-analysis of randomised controlled trials. BMJ, 2010; 340:c1718.

16. Frenkel WJ, Jongerius EJ, Mandjes-van Uitert MJ, et al. Validation of the Charlson Comorbidity Index in acutely hospitalized elderly adults: a prospective cohort study. Journal of the American Geriatrics Society, 2014; 62(2):342–346.

17. Nakayama A, Major G, Holliday E, et al. Evidence of effectiveness of a fracture liaison service to reduce the re-fracture rate. Osteoporosis International, 2016. 27(3):873–879.

18. Huntjens KMB, van Geel TCM, Geusens PP, et al. Impact of guideline implementation by a fracture nurse on subsequent fractures and mortality in patients presenting with non-vertebral fractures. Injury, 2011; 42:S39–S43.

19. Klop C, Gibson-Smith D, Elders PJM, et al. Anti-osteoporosis drug prescribing after hip fracture in the UK: 2000–2010. Osteoporosis International, 2015; 26(7):1919–1928.

20. Kim D, Mackenzie D, Cutfield R. Implementation of fracture liaison service in a New Zealand public hospital: Waitemata district health board experience. The New Zealand medical journal, 2016; 129:50–55.

21. Byszewski A, Lemay G, Molnar F, et al. Closing the osteoporosis care gap in hip fracture patients: an opportunity to decrease recurrent fractures and hospital admissions. Journal of osteoporosis, 2011; 2011:404969–404969.

22. Neuburger J, Currie C, Wakeman R, et al. Increased orthogeriatrician involvement in hip fracture care and its impact on mortality in England. Age and Ageing, 2016; 46(2):187–192.

23. Mitchell R, Harvey L, Brodaty H, et al. One-year mortality after hip fracture in older individuals: the effects of delirium and dementia. Archives of Gerontology and Geriatrics, 2017; 72:135–141.

Contact diana@nzma.org.nz
for the PDF of this article

View Article PDF

Hip fractures in the older adult(those aged over 65 years) are common, affecting up to 18% women and 6% men in their lifetime.1 They are associated with significant mortality and burden of disease,2 even with prompt evidence-based treatment including surgery, orthogeriatric care and rehabilitation.2,3 In addition, approximately 10% of patients will experience a second hip fracture, which has further negative impacts on their mortality and quality of life.4

The older population in New Zealand is growing rapidly, particularly in Canterbury New Zealand, where people over 65 years make up 16.1% of the total population, up 9,500 people in the last five years. In New Zealand, it is estimated that about 72 out of 100 people with hip fracture are admitted from home but following the fracture only 55 out of those 72 people will return home.5 Given the impact of hip fracture on morbidity, mortality and quality of life, prevention programmes focus on treatments in two areas, falls prevention using physiotherapy intervention and secondary prevention of fractures (ie, treatment of osteoporosis) using medications. Pharmacological treatments for osteoporosis include bisphosphonates (eg, zoledronic acid, alendronate, risedronate),parathyroid hormone analogues (eg, teriparatide), the RANKL antibody denosumab6 and more recently discovered agents such as the sclerostin inhibitor romosozumab.7 Despite the range of treatments available, many patients remain under- or un-treated for osteoporosis and physicians encounter major challenges when implementing appropriate treatments.8

Following hip fracture, older patients may be cared for in a range of settings ideally with input from both geriatric and orthopaedic specialists.9 At Christchurch Hospital, patients over 65 years with fractures are routinely seen by a HealthCare of the Elderly Physician as a shared care model that helps manage the patient’s comorbidities and also establish whether inpatient rehabilitation is required before safe discharge home.10,11 Patients may undergo inpatient rehabilitation in older person’s health wards(OPH), for general rehabilitation under care of a geriatrician, the orthopaedic rehabilitation unit (ORU), a specialist orthogeriatric rehabilitation ward, also under care of a geriatrician. Generally, patients selected for orthogeriatric rehabilitation are done so on the basis that they have fewer co-existing active medical problems and their orthopaedic problem is their primary rehabilitation issue. Alternatively, patients may be discharged directly to their own home, ideally with nurse-led fracture liaison follow-up with the Fracture Liaison Service (FLS) to ensure they receive adequate secondary fracture prevention.12,13 The FLS nurse currently uses the Australia New Zealand Hip Fracture Registry5 to identify patients undertreated for osteoporosis, the nurse and organises bone density testing and/or general practitioner-led osteoporosis treatment, eg, bisphosphonates.14 Choice of rehabilitation location depends on a range of factors including the patient’s pre-morbid function, their medical comorbidities, cognitive function and social factors. There is an established benefit of both orthogeriatric and general geriatric rehabilitation for patients with hip fractures and other acuteconditions.15 Patients who attended both general geriatric and orthopaedic rehabilitation had lower mortality rates, greater function and were less likely to be admitted to nursing homes.15 This effect was seen both at discharge and at follow-up.

The main aim of the current study was to retrospectively audit patient characteristics, surgical management practices, rehabilitation practices and outcomes, osteoporosis treatment prescription and mortality for patients with hip fractures admitted to Christchurch Hospital. In particular, we aimed to compare outcomes for four groups: 1) publicly funded aged residential care (ARC)residents discharged directly from acute orthopaedics, 2) non-ARC residents discharged directly from acute orthopaedics, 3) patients discharged after orthogeriatric rehabilitation and 4) patients discharged after general geriatric rehabilitation. As a secondary aim we also looked at the differences between orthogeriatric and general geriatric rehabilitation groups in respect to length of stay, comorbidities, functional status and treatment for osteoporosis.

Method

Data acquisition

This was a retrospective cohort study including all consecutive patients aged 65 and over who presented to Christchurch Hospital between 1 May 2017 and 30 April 2018 with hip fracture. Patient data was extracted from multiple data sets such as the Administrative Hospital data, Marriage, Birth and Death Data (Department of Internal Affairs New Zealand), Client Claim Payment System (CCPS) (Ministry of Health) and Australian Rehabilitation Outcome Centre (AROC) data. Multiple querying tools were used to validate and identify the cohort from the hospital patient management system, coding and data warehouse, particularly signals from noise® (sfn). This is proprietary software from Lightfoot Solutions which uses the principles of Statistical Process Control. Canterbury District Health Board customised the sfn tool with tables that enable people to explore data through a process flow perspective.

Initially, the data extracted from sfn identified patients admitted to Christchurch Hospital with hip fracture based on International Classification of Diseases, Tenth Revision (ICD-10)codes. Then the software was used to determine the discharge location for these same patients. The discharge location determined the cohort selection. Cohort 1(ARC residents) and 2 (not ARC residents) were discharged directly from hospital, cohort 3 were discharged to orthopaedic rehabilitation (ORU) and cohort 4 to general geriatric rehabilitation.

The following ICD-10 codes were used to identify the cohort with hip fracture, based on the primary diagnosis (ie, their primary reason for hospital admission): S72.00 Fracture of neck of femur, part unspecified S72.01 Fracture of intracapsular section of femur, S72.03 Fracture of subcapital section of femur, S72.04 Fracture of mid-cervical section of femur, S72.05 Fracture of base of neck of femur, S72.08 Fracture of other parts of neck of femur, S72.10 Fracture of trochanteric section of femur, unspecified; S72.11 Fracture of intertrochanteric section of femur, S72.2 Subtrochanteric fracture. These patients had an ICD-10 primary procedure code:47519-00 Internal fixation of fracture of trochanteric or subcapital femur,47522-00 Hemiarthroplasty of femur, 49315-00 Partial arthroplasty of hip,49318-00 Total arthroplasty of hip, unilateral) or other treatment code (to include patients managed conservatively). These were then grouped into two groups: neck of femur and trochanteric fractures. Patients with periprosthetic and pathological fractures were excluded.

Variables

Data on a range of dependent variables were identified and recorded for the included patients; including age, National Hospital Identifier (NHI), sex, ICD-10 code, primary surgical procedure, total LOS, comorbidities, mortality, location of domicile on admission (ARC funding data and Australasian Rehabilitation Outcomes Centre[AROC]), location of domicile on discharge (ARC funding data and AROC) and change in Functional Independence Measure (FIM) score during rehabilitation(AROC data). To quantify patient comorbidity, we used a modified Charlson Comorbidity index (CCI), excluding age. Although designed for patients underage 65, the CCI can predict short- and long-term mortality in acutely ill hospitalised older adults16 and was chosen for this study so data could be easily collected from available sources. Comorbidity data was collected based on all coded diagnoses, rather than just new diagnosis codes. Prescribed medications were not available in any coded data-sets and so this data was manually extracted from patient medication and reconciliation history linked with each patient’s National Health Identifier by the first, third, fourth and fifth authors. Referral information from the Fracture Liaison Service was obtained.

Statistical analysis

Statistical analysis was performed using SAS 9.4. Chi square test was used for comparison of categorical variables except when cell sizes were small, then Fisher’s exact test was used. Normality assumptions were found to be violated for continuous variables. Hencenon-parametric Mann-Whitney test was used for comparison between two independent groups and Kruskal-Wallis test for more than two groups. For both categorical and continuous variables, if p-value for overall comparison was significant, post-hoc multiple testing with Bonferroni correction was performed to ascertain which pairwise differences were significant. To analyse the possible factors associated with mortality, a Cox proportional hazard model was developed. A stepwise selection method was used with an entry level significance of 0.2 and significance level of 0.1 for variables to stay in the model.

Results

Demographic characteristics

Over the one-year study period, 415patients over age 65 were admitted to the acute orthopaedic service with hip fracture, with over 421 episodes of care (ie, some patients presented twice).Of these patients, 67% (n=278) were referred for inpatient rehabilitation. The remaining patients were discharged directly from the acute orthopaedic ward. A higher proportion of patients who went to geriatric rehabilitation (OPH) were admitted from residential care, compared to those who went to orthogeriatric rehabilitation (ORU, see Table 1). In all groups, the majority of patients were women, and most were of European ethnicity. Demographic characteristics for each of the patient groups are detailed in Table 1. Median age for non-ARC-residents discharged from acute orthopaedics was significantly younger than patients in the other groups and approximately 20% of the patients in this group lived outside the Canterbury region.

Table 1: Demographic characteristics for patients in each group (n=415), including age, gender, ethnicity, domicile, aged residential care status, discharge status and mortality. Variables with significant between group differences are highlighted in bold.

Fracture type, wait for theatre and surgical management and length of stay

For each cohort, patients were more likely to present with a neck of femur than a trochanteric fracture (see Table2). As seen in Table 2, total arthroplasty was more likely to be performed in the non-ARC residents discharged from orthopaedics and the ORU patients. The ‘Other’ category included non-operative management, eg, allied health intervention, and other surgical procedures (potentially related to additional presenting problems).

Of patients who proceeded for surgery (n=395), over half of patients were operated on within 24 hours (n=210,53%) and most patients (n=324, 82%) were operated on within 48 hours of presenting. There were no statistically significant differences between groups in wait times for theatre (p=0.39) (see Table 3). Patients discharged from general geriatric rehabilitation had a longer length of stay than other patients (p<0.0001, see Table 3). No clear disparity in LOS was found between groups discharged directly from acute orthopaedics.

Table 2: Fracture types and primary procedures for the four patient groups admitted with hip fracture.

Note: n=421 episodes of care with some patients presenting for more than one episode of care, hence differing n-values for each patient groups when compared to total patient numbers (eg, in Table 1). For primary procedures there are missing data for four patients, so group totals are n=45, n=93, n=185 and n=94 respectively.

Previous and subsequent fractures

Rates of fracture prior to hip fracture and in the 180 days following hip fracture were also analysed. There were no between-group differences in these rates (see Table 3). Approximately 25% of patients (n=105) had a hospital presentation for a fracture prior to their hip fracture admission and 5% of patients (n=21) presented for treatment of a fracture after their hip fracture (noting variable follow-up times). A number of patients (n=181, 43%) had also had at least one hospital admission within the 365 days prior to their hip fracture.

Table 3: Total length of stay (including both acute and rehabilitation admissions, if applicable) for the four groups in days (n=421 episodes of care) with wait for operation times, prior and subsequent fractures and re-admission.

Note: n=421 episodes of care with some patients presenting for more than one episode of care, hence differing n-values for each patient groups when compared to total patient numbers (eg, in Table 1).
*Wait for operation calculated on total participant numbers of n=40, n=80, n=168 and n=88 for each group respectively, given some patients did not proceed for theatre. Percentages computed on valid n only.

Mortality and survival analysis

Patients admitted to orthogeriatric rehabilitation were less likely to die within 30 days than other patients(p<0.0001, see Table 4). No clear mortality difference was found between the other three groups. ARC residents discharged from acute orthopaedics were more likely to die at six months post-presentation than patients in the other three groups. Patients transferred to general OPH had a higher mortality at six months compared with the ORU (p<0.0001). No clear mortality difference was found between the other groups. Figure 1 show the survival probability in the four groups over 20 months. The overall mortality in all groups at 180 days was 27% and 10% at 30 days. In-hospital mortality (ie, patients who died while admitted for their hip fracture) was 5% across all groups.

No difference was found in mortality and LOS for patients with fracture neck of femur compared with trochanteric fracture. Patients with total arthroplasty have a significantly shorter length of stay and reduced mortality compared to the two other groups (see Table 5).No difference in LOS or mortality was found between patients having hemiarthroplasty or internal fixation.

To further assess factors associated with mortality, a cox proportional hazard model was developed (see Table 6 and Figure 1). Based on this model, risk factors for overall mortality were: ARC residents, a higher Charlson index and older age. The length of stay and time to wait for operation were not correlated with overall mortality.

Table 4: Mortality data at 30 days and 180 days for each of the four groups (n=415 patients).

*Percentages shown reflect proportion of total patient population.

Table 5: Mortality and LOS in relation to operative procedure used.

Table 6: Cox proportional hazard model.

Figure 1: 20-month predicted survival plot for the four groups.

Table 7: Modified Charlson Comorbidity Index (CCI) for each group and comorbidities. Variables with significant between group differences are highlighted in bold.

Comorbidities and Modified Charlson Index

The Charlson Comorbidity Index (CCI)for each group is shown in Table 7. The index was calculated with excluding patient age, given all patients were over age 65. Non-ARC residents discharged directly from acute orthopaedics had a significantly lower Charlson index than other patients. No clear disparity was found between the other three cohorts. In terms of specific comorbidities, the ARC residents discharged directly from acute orthopaedics and the general geriatric rehabilitation group had significantly higher rates of dementia than the other two groups. There were significantly more patients with congestive heart failure in the ORU group(32%) compared to the non-ARC residents discharged from acute orthopaedics(15%). The non-ARC residents discharged from acute orthopaedics were less likely to have dementia and diabetes complications compared to other groups. Patients admitted to general geriatric rehabilitation were more likely to have renal disease dementia, diabetic complications and congestive heart failure than non-ARC residents discharged from acute orthopaedics. There were otherwise no differences found between groups.

Osteoporosis treatment and follow-up via Fracture Liaison Service for secondary fracture prevention

Data on osteoporosis treatment for each group at time of discharge are presented in Table 8. Due to the diversity of medication regimes prescribed, these were categorised by the first author (SW) into five categories (see Table 8). Appropriate treatment included bisphosphonates (alendronate, zoledronic acid, risedronate), teriparatide ordenosumab, all with colecalciferol 1.25mg monthly (50,000 IU of Vitamin D). Colecalciferol alone or no treatment was considered suboptimal. ARC residents discharged from acute orthopaedics were more likely to have suboptimal treatment with no documentation than any other group. Patients discharged from orthogeriatric rehabilitation were most likely to have appropriate treatment with osteoporosis medications, further investigation/referral to FLS or treatment actively with-held due to comorbidity (88% patients, p<000.1).Rates of consideration or treatment for osteoporosis were 62% in general geriatric rehabilitation group and 46% in non-ARC residents discharged from acute orthopaedics.

Fracture Liaison Service (FLS)attendance was analysed for Canterbury residents only, as residents from other areas of New Zealand or overseas are not usually seen by the FLS in Canterbury. As protocols existed for inpatient treatment for osteoporosis, FLS referral was not activated for many patients undergoing inpatient rehabilitation. The FLS also does not currently follow-up patients already on appropriate bisphosphonate or other osteoporosis treatment. Rates of FLS referral were low in ARC residents discharged from acute orthopaedics (11%). Referral rates were higher in the non-ARC residents discharged from acute orthopaedics (p<0.0001), of whom 41% (31 patients) were seen by the FLS. It is important to note that 5% of patients in this group also died in hospital so were not appropriate for referral.

Table 8: Osteoporosis treatment for each patient group.

*Appropriate treatment was with bisphosphonate, denosumab or teriparatide and vitamin D, or with vitamin D alone if patient was proven not to have osteoporosis on DEXA scan.
**Suboptimal treatment due to comorbidities were patients not considered for osteoporosis treatment beyond vitamin D due to renal failure, drug intolerances, previous atypical femoral fracture etc.
***Suboptimal treatment was considered to be no vitamin D or vitamin D alone without evidence of further investigation for underlying osteoporosis.

Level of care and functional independence measure scores

For patients discharged directly from acute orthopaedics, data was only available for their ARC funding but not level of care within this funding. For the non-ARC residents discharged directly from Christchurch, 19 patients were transferred to other hospitals in New Zealand and 11 patients died during their hospital admission. Of those who were discharged directly “home”, two went to ARC, the remainder apparently went to private residence.

However, for patients admitted to rehabilitation, AROC data is collected on admission and discharge which provided more detailed information on admission and discharge location of residence (see Table 9). Most patients (83%) from the orthogeriatric group were in their own private residence before admission. On discharge, 70% of patients in the orthogeriatric group remained in their own private residence, but 6% went to rest home-level care and 7% went to hospital-level care. Approximately 10% of the population were admitted from rest home-level care but ended up in hospital-level care on discharge. In contrast, only 59% of the patients in the general geriatric rehabilitation group were admitted from private residence. On discharge, 43% of the patients in this group remained in their own residence, and 17% of patients had moved into rest home and hospital-level care (Table 9).

Table 9: Pre- and post-discharge location of residence for Cohort 3 and 4, based on AROC data.

Functional independence measures are also collected for patients on admission and discharge from rehabilitation (see Figure 2) and were available via AROC. Both FIM cognitive and motor scores were significantly higher in the orthogeriatric rehabilitation group on admission and discharge. There were no significant differences between the two groups in the improvement of scores on discharge. This data correlates well with the ANZHFR data set.5

Figure 2: Change in functional independence measures of patients in the two rehabilitation groups.

Discussion

This study presents data on hip fracture management and rehabilitation in older adults at a single centre in metropolitan New Zealand. It followed on from previous work at Christchurch Hospital which utilised sfn® in implementing and evaluating the Fast Track Hip Fracture pathway that remains in use.3 Christchurch Orthopaedic Department admits approximately 5,500 patients per year in the busiest trauma centre in New Zealand and of those over 400 are hip fractures. Despite the availability of the ANZHFR,5which collects data on hip fracture for patients around Australia and New Zealand, we chose to analyse data using sfn due to more flexibility and efficiency in comparing multiple variables, including some that are not collected by the ANZHFR (eg, comorbidities, previous hospital admissions, factors contributing to mortality). Direct comparison with the ANZHFR was not within our research aims and was not explored. The administrative data allowed us to explore all the characteristics of the heterogenous patient groups admitted following hip fracture, to identify possible contributors to the clinical outcomes for these patients and to aid future targeted service delivery.

Fracture type and surgical intervention

The fracture types and surgical procedures have not altered from our previous work.3 Time to surgery has decreased with no negative impact on mortality. There was no difference in mortality or LOS between fracture types, but mortality and LOS was less in THA reflecting appropriate operative procedure. This observation is consistent with guidelines for the operative management of hip fractures.5

Patients discharged directly home or to a care facility

Based on anecdotal experience, we anticipated the need to separate the patients discharged directly from their acute orthopaedic admission based on their pre-morbid level of care (ie, from private residence versus residential care) due to significant within-group heterogeneity. Our study provided detailed information on patients discharged directly from the acute orthopaedic ward in two groups: pre-morbid ARC residents and non-ARC residents. Non-ARC residents discharged from acute orthopaedics were more likely to be younger, less comorbid, and a significant proportion (20%)were from outside Canterbury. However, their younger age and less comorbidity were not reflected in a mortality benefit for these patients. The non-ARC residents discharged directly from hospital are the key group targeted by the Fracture Liaison service, and referral rates were suboptimal at around 40%

On the other hand, patients who were ARC residents and were discharged from acute orthopaedics were more comorbid and had higher rates of mortality at six months compared to the other groups. They had similar rates of dementia when compared to patients transferred to general geriatric rehabilitation. Patients in this ARC resident group were also more likely to have suboptimal osteoporosis treatment with no clinical documentation regarding clinical contraindication and very low rates of referral to the Fracture Liaison service.

Patients discharged following inpatient rehabilitation

Outcomes of this study are consistent with other literature on orthogeriatric or general geriatric rehabilitation for patients with hip fracture.15 The patients in our rehabilitation groups were not randomised, so a patient’s journey following hip fracture will be largely influenced by selection bias. However, patients in our study who had orthogeriatric or geriatric rehabilitation had a longer length of stay but higher rates of treatment with osteoporosis medications. Mortality in the orthogeriatric rehabilitation group was significantly lower than the other groups at 30 days, and both the rehabilitation groups had lower mortality than the ARC residents discharged from acute orthopaedics at the 180-day follow-up.

Osteoporosis treatment

High treatment rates in older patients with hip fracture may lead to a 40% reduction in major bone re-fractures.17 In this study, treatment for osteoporosis was variable for the four groups. FLS was generally not involved in the orthogeriatric rehabilitation group or the general rehabilitation group because protocols for osteoporosis treatment exist as part of the rehabilitation process. Treatment or consideration for treatment was 84%in the orthogeriatric group and 64% in the general group. However, in those discharged back to ARC facility or to another hospital or home were low at 15%and 46% respectively. These treatment rates compare favourably with other published data.18 These results are helpful because they identify where FLS should be involved. Also, accurate recording of inpatient treatment modality was identified as an issue; in-patient treatment was not captured on a database and needed manual extraction, which has been previously described.19 This may explain some of the low percentage figures from the ANZHFR and previous work.20 Similarly, to previous research,21 this work shows that inpatient treatment for osteoporosis seems to be the most effective way of ensuring initiation of treatment. These results suggest the need for more proactive FLS involvement for patients who do not have inpatient rehabilitation, and has led the hospital to introduce a mandatory field in the discharge template on osteoporosis treatment.

Mortality

This study also sought to determine local mortality rates and examine some of the factors possibly contributing to or predicting mortality. Thirty-day mortality rates reported in this study were higher than the most recent (2018) Annual Report of the ANZHFR5 for Christchurch, possibly due to differences in data source. Mortality rates vary significantly between studies, but 30-day rates in our study were largely comparable to the existingliterature;22 however, mortality rates at 180days were similar to rates at one year reported in some studies.23 The cox proportional hazard model showed that pre-morbid ARC residency was the strongest risk factor for mortality at 180 days, with patients of older age and with more comorbidities (based on Charlson Comorbidity Index) also at increased risk for death. Interestingly, total arthroplasty was also associated with improved mortality and this probably reflects selection of a healthier patient group for this procedure.

Limitations

This study primarily utilised data coded from medical records and explored variables through sfn and multiple dataset analysis. This means that the quality of the data is dependent on the way data was documented in patient records and then coded by independent clinical coders. Other data, eg, from the Department of Internal Affairs around death is likely to be accurate but released with sometime lag, hence why only 180-day mortality was included in the current study. AROC collection is also limited, as it only collects two points of outcome at admission and discharge, being cofounding by length of stay, with potential internal bias on analysis of rehabilitation improvement.

Also, regarding osteoporosis treatment data, we had access to hospital discharge summaries and electronic prescribing software which allowed us to see if treatment was prescribed in hospital, but we did not have access to longer-term data to determine if patients continued to use these medications in the community.

Finally, this was not a randomised treatment trial and as such, confounding factors in analysing outcomes could be significant. However, the data provides useful information around clinical management of this group of patients that may still alter management, eg, promote further consideration of osteoporosis treatment in certain groups of patients and greater uptake of inpatient rehabilitation.

Implications

This study shows that tools such as sfn, that easily explore multiple variables and cohorts across various data have utility in evaluating acute hospital pathways,3 but also in evaluating a patient’s whole hospital journey, including the full acute hospital stay and subsequent rehabilitation. Although this paper gives similar results to previous work,3,10 it is different as hip fracture patients are analysed according to their discharge destination. This enabled comparisons of various outcome across these four groups. It is pleasing to see the wait for operation has decreased from previous work3 to over 80% of patients having their operation within 48 hours.

Since the introduction of fast-track hip fracture management,3the length of stay in orthogeriatric rehabilitation has decreased to 14 days and to 21 days in general rehabilitation. Baseline functional status and discharge destinations were different between the two rehabilitation groups, with the orthopaedic rehabilitation group having higher baseline and post-rehabilitation functional status and a higher percentage of patients returning to private residence. Functional status on discharge for patients undergoing hip fracture rehabilitation (measured with the FIM score) is higher than the national average standard.5

Conclusions

We have demonstrated the outcomes in four groups of patients with neck of femur fractures. The time to operation and length of stay continues to decrease compared with our previous work. New data on the four post-operative groups show better outcomes (ie, return to domicile, treatment for osteoporosis and mortality) in the dedicated orthopaedic rehabilitation group. It is difficult to interpret the significance of these improved outcomes given these patients were less comorbid, specifically with less renal disease and dementia than the general rehabilitation group. This is likely to affect their functional outcomes and prescription of osteoporosis treatment. Enhancing involvement by FLS and increased prescription of osteoporosis treatment is needed for the patients discharged directly from acute orthopaedics.

Summary

Abstract

Aim

For older patients with hip fracture, we explored patient characteristics, outcomes and osteoporosis treatments for those admitted to rehabilitation compared to those discharged directly from hospital, using data set analysis.

Method

Retrospective cohort study including all consecutive patients (65 years and over) admitted to Christchurch Hospital over one year. Outcomes were compared for patients in four groups: 1) aged residential care (ARC) residents and 2) not ARC residents discharge from acute orthopaedics, and those discharged from 3) orthogeriatric or 4) general geriatric rehabilitation. Clinical data was extracted from hospital data warehouse using signals from noise.

Results

Over 12 months, 415 patients were admitted with hip fracture. Over half (n=282) were transferred for inpatient rehabilitation. Thirty-day mortality was 10%. Mortality at 180 days was 7% in orthogeriatric rehabilitation group and 8% in general rehabilitation group. Length of stay was shorter in orthogeriatric compared with general rehabilitation (median 12.9 vs 20.4 days). Osteoporosis treatment was addressed in 88% of patients in the orthogeriatric group compared with 62% in general rehabilitation group. More patients in orthogeriatric group discharged home compared with general rehabilitation group (70% vs 43%), but functional improvements were similar. For ARC residents discharged following acute admission, length of stay was short (median 6.5 day), but overall 180-day mortality was 7%. The not-ARC resident group discharged directly from hospital was less comorbid than the other groups with lower 180-day mortality (4%). Osteoporosis treatment rates were lower in patients discharged from acute admission (15–42%).

Conclusion

Mortality was highest in ARC residents discharged from acute orthopaedics. Patients admitted for orthogeriatric rehabilitation had shorter length of stay, lower 30-day mortality, were more likely to return home and most likely to be offered osteoporosis treatment (88%), noting less underlying comorbidity and better baseline functional status in this group. This paper supports further study (eg, randomised trials) to examine the effects of targeted post-surgical rehabilitation for patients with hip fracture and explore measures to increase uptake of osteoporosis treatment.

Author Information

Samantha Warhurst, Christchurch Hospital, Canterbury District Health Board, Christchurch; Soledad Labbe-Hubbard, Planning and Funding, Canterbury District Health Board, Christchurch; Ma Yi, Biostatistician, Canterbury District Health Board, Christchurch; Jane Vella-Brincat, Pharmacy, Christchurch Hospital, Christchurch; Tonya Sammon, Ambulatory Care, Well South, Dunedin; Jane Webb, Pharmacy, Canterbury District Health Board, Christchurch; Caroline McCullough, Department of Orthopaedic Surgery and Musculoskeletal Medicine, Canterbury District Health Board, Christchurch; Gary Hooper, Department of Orthopaedic Surgery and Musculoskeletal Medicine, Canterbury District Health Board, Christchurch; Nigel Gilchrist, Orthopaedic Medicine, Canterbury District Health Board, Christchurch.

Acknowledgements

Correspondence

Dr Nigel Gilchrist, Orthopaedic Medicine, Canterbury District Health Board, Burwood Road, Christchurch.

Correspondence Email

nigel.gilchrist@cdhb.health.nz

Competing Interests

Nil.

1. Veronese N, Maggi S. Epidemiology and social costs of hip fracture. Injury, 2018. 49(8):1458–1460.

2. Bhandari M, Swiontkowski M. Management of Acute Hip Fracture. New England Journal of Medicine, 2017; 377(21):2053–2062.

3. Gilchrist N, Dalzell K, Pearson S, et al. Enhanced hip fracture management: use of statistical methods and dataset to evaluate a fractured neck of femur fast track pathway-pilot study. The New Zealand medical journal, 2017; 130(1455):91–101.

4. Sobolev B, Sheehan K J, Kuramoto L, et al. Excess mortality associated with second hip fracture. Osteoporosis International, 2015; 26(7):1903–1910.

5. Australian & New Zealand Hip Fracture Registry, ANZHFR Bi-National Annual Report of Hip Fracture Care 2018., 2018.

6. Khosla S, Hofbauer LC. Osteoporosis treatment: recent developments and ongoing challenges. The Lancet. Diabetes & endocrinology, 2017; 5(11):898–907.

7. Cosman F, Crittenden DB, Adachi JD, et al. Romosozumab Treatment in Postmenopausal Women with Osteoporosis. New England Journal of Medicine, 2016; 375(16):1532–1543.

8. Solomon DH, Johnston SS, Boytsov NN, et al, Osteoporosis medication use after hip fracture in US patients between 2002 and 2011. Journal of Bone and Mineral Research, 2014; 29(9):1929–1937.

9. Grigoryan KV, Javedan H, Rudolph JL. Ortho-geriatric care models and outcomes in hip fracture patients: a systematic review and meta-analysis. Journal of orthopaedic trauma, 2014; 28(3):e49.

10. Thwaites J, Mann F, Gilchrist N, et al. Shared care between geriatricians and orthopaedic surgeons as a model of care for older patients with hip fractures. The New Zealand Medical Journal (Online), 2005; 118(1214).

11. Elliot J, Wilkinson TJ, Hanger HC, et al. The added effectiveness of early geriatrician involvement on acute orthopaedic wards to orthogeriatric rehabilitation. The New Zealand Medical Journal, 1996; 109(1017):72–73.

12. Osteoporosis New Zealand. Osteoporosis New Zealand Strategic Plan 2017–2020. 2017 18/06/2019]; Available from: http://osteoporosis.org.nz/wp-content/uploads/ONZ-2017-Strategic-Plan-WEB.pdf

13. Osteoporosis New Zealand. Bone Plan 2020: A systematic approach to hip fracture care and prevention for New Zealand. 2012 18/06/2019]; Available from: https://osteoporosis.org.nz/wp-content/uploads/Bone-Care-2020-web.pdf

14. Kanis J, Johnell O, De Laet C, et al. A meta-analysis of previous fracture and subsequent fracture risk. Bone, 2004; 35(2):375–382.

15. Bachmann S, Finger C, Huss A, et al. Inpatient rehabilitation specifically designed for geriatric patients: systematic review and meta-analysis of randomised controlled trials. BMJ, 2010; 340:c1718.

16. Frenkel WJ, Jongerius EJ, Mandjes-van Uitert MJ, et al. Validation of the Charlson Comorbidity Index in acutely hospitalized elderly adults: a prospective cohort study. Journal of the American Geriatrics Society, 2014; 62(2):342–346.

17. Nakayama A, Major G, Holliday E, et al. Evidence of effectiveness of a fracture liaison service to reduce the re-fracture rate. Osteoporosis International, 2016. 27(3):873–879.

18. Huntjens KMB, van Geel TCM, Geusens PP, et al. Impact of guideline implementation by a fracture nurse on subsequent fractures and mortality in patients presenting with non-vertebral fractures. Injury, 2011; 42:S39–S43.

19. Klop C, Gibson-Smith D, Elders PJM, et al. Anti-osteoporosis drug prescribing after hip fracture in the UK: 2000–2010. Osteoporosis International, 2015; 26(7):1919–1928.

20. Kim D, Mackenzie D, Cutfield R. Implementation of fracture liaison service in a New Zealand public hospital: Waitemata district health board experience. The New Zealand medical journal, 2016; 129:50–55.

21. Byszewski A, Lemay G, Molnar F, et al. Closing the osteoporosis care gap in hip fracture patients: an opportunity to decrease recurrent fractures and hospital admissions. Journal of osteoporosis, 2011; 2011:404969–404969.

22. Neuburger J, Currie C, Wakeman R, et al. Increased orthogeriatrician involvement in hip fracture care and its impact on mortality in England. Age and Ageing, 2016; 46(2):187–192.

23. Mitchell R, Harvey L, Brodaty H, et al. One-year mortality after hip fracture in older individuals: the effects of delirium and dementia. Archives of Gerontology and Geriatrics, 2017; 72:135–141.

Contact diana@nzma.org.nz
for the PDF of this article

View Article PDF

Hip fractures in the older adult(those aged over 65 years) are common, affecting up to 18% women and 6% men in their lifetime.1 They are associated with significant mortality and burden of disease,2 even with prompt evidence-based treatment including surgery, orthogeriatric care and rehabilitation.2,3 In addition, approximately 10% of patients will experience a second hip fracture, which has further negative impacts on their mortality and quality of life.4

The older population in New Zealand is growing rapidly, particularly in Canterbury New Zealand, where people over 65 years make up 16.1% of the total population, up 9,500 people in the last five years. In New Zealand, it is estimated that about 72 out of 100 people with hip fracture are admitted from home but following the fracture only 55 out of those 72 people will return home.5 Given the impact of hip fracture on morbidity, mortality and quality of life, prevention programmes focus on treatments in two areas, falls prevention using physiotherapy intervention and secondary prevention of fractures (ie, treatment of osteoporosis) using medications. Pharmacological treatments for osteoporosis include bisphosphonates (eg, zoledronic acid, alendronate, risedronate),parathyroid hormone analogues (eg, teriparatide), the RANKL antibody denosumab6 and more recently discovered agents such as the sclerostin inhibitor romosozumab.7 Despite the range of treatments available, many patients remain under- or un-treated for osteoporosis and physicians encounter major challenges when implementing appropriate treatments.8

Following hip fracture, older patients may be cared for in a range of settings ideally with input from both geriatric and orthopaedic specialists.9 At Christchurch Hospital, patients over 65 years with fractures are routinely seen by a HealthCare of the Elderly Physician as a shared care model that helps manage the patient’s comorbidities and also establish whether inpatient rehabilitation is required before safe discharge home.10,11 Patients may undergo inpatient rehabilitation in older person’s health wards(OPH), for general rehabilitation under care of a geriatrician, the orthopaedic rehabilitation unit (ORU), a specialist orthogeriatric rehabilitation ward, also under care of a geriatrician. Generally, patients selected for orthogeriatric rehabilitation are done so on the basis that they have fewer co-existing active medical problems and their orthopaedic problem is their primary rehabilitation issue. Alternatively, patients may be discharged directly to their own home, ideally with nurse-led fracture liaison follow-up with the Fracture Liaison Service (FLS) to ensure they receive adequate secondary fracture prevention.12,13 The FLS nurse currently uses the Australia New Zealand Hip Fracture Registry5 to identify patients undertreated for osteoporosis, the nurse and organises bone density testing and/or general practitioner-led osteoporosis treatment, eg, bisphosphonates.14 Choice of rehabilitation location depends on a range of factors including the patient’s pre-morbid function, their medical comorbidities, cognitive function and social factors. There is an established benefit of both orthogeriatric and general geriatric rehabilitation for patients with hip fractures and other acuteconditions.15 Patients who attended both general geriatric and orthopaedic rehabilitation had lower mortality rates, greater function and were less likely to be admitted to nursing homes.15 This effect was seen both at discharge and at follow-up.

The main aim of the current study was to retrospectively audit patient characteristics, surgical management practices, rehabilitation practices and outcomes, osteoporosis treatment prescription and mortality for patients with hip fractures admitted to Christchurch Hospital. In particular, we aimed to compare outcomes for four groups: 1) publicly funded aged residential care (ARC)residents discharged directly from acute orthopaedics, 2) non-ARC residents discharged directly from acute orthopaedics, 3) patients discharged after orthogeriatric rehabilitation and 4) patients discharged after general geriatric rehabilitation. As a secondary aim we also looked at the differences between orthogeriatric and general geriatric rehabilitation groups in respect to length of stay, comorbidities, functional status and treatment for osteoporosis.

Method

Data acquisition

This was a retrospective cohort study including all consecutive patients aged 65 and over who presented to Christchurch Hospital between 1 May 2017 and 30 April 2018 with hip fracture. Patient data was extracted from multiple data sets such as the Administrative Hospital data, Marriage, Birth and Death Data (Department of Internal Affairs New Zealand), Client Claim Payment System (CCPS) (Ministry of Health) and Australian Rehabilitation Outcome Centre (AROC) data. Multiple querying tools were used to validate and identify the cohort from the hospital patient management system, coding and data warehouse, particularly signals from noise® (sfn). This is proprietary software from Lightfoot Solutions which uses the principles of Statistical Process Control. Canterbury District Health Board customised the sfn tool with tables that enable people to explore data through a process flow perspective.

Initially, the data extracted from sfn identified patients admitted to Christchurch Hospital with hip fracture based on International Classification of Diseases, Tenth Revision (ICD-10)codes. Then the software was used to determine the discharge location for these same patients. The discharge location determined the cohort selection. Cohort 1(ARC residents) and 2 (not ARC residents) were discharged directly from hospital, cohort 3 were discharged to orthopaedic rehabilitation (ORU) and cohort 4 to general geriatric rehabilitation.

The following ICD-10 codes were used to identify the cohort with hip fracture, based on the primary diagnosis (ie, their primary reason for hospital admission): S72.00 Fracture of neck of femur, part unspecified S72.01 Fracture of intracapsular section of femur, S72.03 Fracture of subcapital section of femur, S72.04 Fracture of mid-cervical section of femur, S72.05 Fracture of base of neck of femur, S72.08 Fracture of other parts of neck of femur, S72.10 Fracture of trochanteric section of femur, unspecified; S72.11 Fracture of intertrochanteric section of femur, S72.2 Subtrochanteric fracture. These patients had an ICD-10 primary procedure code:47519-00 Internal fixation of fracture of trochanteric or subcapital femur,47522-00 Hemiarthroplasty of femur, 49315-00 Partial arthroplasty of hip,49318-00 Total arthroplasty of hip, unilateral) or other treatment code (to include patients managed conservatively). These were then grouped into two groups: neck of femur and trochanteric fractures. Patients with periprosthetic and pathological fractures were excluded.

Variables

Data on a range of dependent variables were identified and recorded for the included patients; including age, National Hospital Identifier (NHI), sex, ICD-10 code, primary surgical procedure, total LOS, comorbidities, mortality, location of domicile on admission (ARC funding data and Australasian Rehabilitation Outcomes Centre[AROC]), location of domicile on discharge (ARC funding data and AROC) and change in Functional Independence Measure (FIM) score during rehabilitation(AROC data). To quantify patient comorbidity, we used a modified Charlson Comorbidity index (CCI), excluding age. Although designed for patients underage 65, the CCI can predict short- and long-term mortality in acutely ill hospitalised older adults16 and was chosen for this study so data could be easily collected from available sources. Comorbidity data was collected based on all coded diagnoses, rather than just new diagnosis codes. Prescribed medications were not available in any coded data-sets and so this data was manually extracted from patient medication and reconciliation history linked with each patient’s National Health Identifier by the first, third, fourth and fifth authors. Referral information from the Fracture Liaison Service was obtained.

Statistical analysis

Statistical analysis was performed using SAS 9.4. Chi square test was used for comparison of categorical variables except when cell sizes were small, then Fisher’s exact test was used. Normality assumptions were found to be violated for continuous variables. Hencenon-parametric Mann-Whitney test was used for comparison between two independent groups and Kruskal-Wallis test for more than two groups. For both categorical and continuous variables, if p-value for overall comparison was significant, post-hoc multiple testing with Bonferroni correction was performed to ascertain which pairwise differences were significant. To analyse the possible factors associated with mortality, a Cox proportional hazard model was developed. A stepwise selection method was used with an entry level significance of 0.2 and significance level of 0.1 for variables to stay in the model.

Results

Demographic characteristics

Over the one-year study period, 415patients over age 65 were admitted to the acute orthopaedic service with hip fracture, with over 421 episodes of care (ie, some patients presented twice).Of these patients, 67% (n=278) were referred for inpatient rehabilitation. The remaining patients were discharged directly from the acute orthopaedic ward. A higher proportion of patients who went to geriatric rehabilitation (OPH) were admitted from residential care, compared to those who went to orthogeriatric rehabilitation (ORU, see Table 1). In all groups, the majority of patients were women, and most were of European ethnicity. Demographic characteristics for each of the patient groups are detailed in Table 1. Median age for non-ARC-residents discharged from acute orthopaedics was significantly younger than patients in the other groups and approximately 20% of the patients in this group lived outside the Canterbury region.

Table 1: Demographic characteristics for patients in each group (n=415), including age, gender, ethnicity, domicile, aged residential care status, discharge status and mortality. Variables with significant between group differences are highlighted in bold.

Fracture type, wait for theatre and surgical management and length of stay

For each cohort, patients were more likely to present with a neck of femur than a trochanteric fracture (see Table2). As seen in Table 2, total arthroplasty was more likely to be performed in the non-ARC residents discharged from orthopaedics and the ORU patients. The ‘Other’ category included non-operative management, eg, allied health intervention, and other surgical procedures (potentially related to additional presenting problems).

Of patients who proceeded for surgery (n=395), over half of patients were operated on within 24 hours (n=210,53%) and most patients (n=324, 82%) were operated on within 48 hours of presenting. There were no statistically significant differences between groups in wait times for theatre (p=0.39) (see Table 3). Patients discharged from general geriatric rehabilitation had a longer length of stay than other patients (p<0.0001, see Table 3). No clear disparity in LOS was found between groups discharged directly from acute orthopaedics.

Table 2: Fracture types and primary procedures for the four patient groups admitted with hip fracture.

Note: n=421 episodes of care with some patients presenting for more than one episode of care, hence differing n-values for each patient groups when compared to total patient numbers (eg, in Table 1). For primary procedures there are missing data for four patients, so group totals are n=45, n=93, n=185 and n=94 respectively.

Previous and subsequent fractures

Rates of fracture prior to hip fracture and in the 180 days following hip fracture were also analysed. There were no between-group differences in these rates (see Table 3). Approximately 25% of patients (n=105) had a hospital presentation for a fracture prior to their hip fracture admission and 5% of patients (n=21) presented for treatment of a fracture after their hip fracture (noting variable follow-up times). A number of patients (n=181, 43%) had also had at least one hospital admission within the 365 days prior to their hip fracture.

Table 3: Total length of stay (including both acute and rehabilitation admissions, if applicable) for the four groups in days (n=421 episodes of care) with wait for operation times, prior and subsequent fractures and re-admission.

Note: n=421 episodes of care with some patients presenting for more than one episode of care, hence differing n-values for each patient groups when compared to total patient numbers (eg, in Table 1).
*Wait for operation calculated on total participant numbers of n=40, n=80, n=168 and n=88 for each group respectively, given some patients did not proceed for theatre. Percentages computed on valid n only.

Mortality and survival analysis

Patients admitted to orthogeriatric rehabilitation were less likely to die within 30 days than other patients(p<0.0001, see Table 4). No clear mortality difference was found between the other three groups. ARC residents discharged from acute orthopaedics were more likely to die at six months post-presentation than patients in the other three groups. Patients transferred to general OPH had a higher mortality at six months compared with the ORU (p<0.0001). No clear mortality difference was found between the other groups. Figure 1 show the survival probability in the four groups over 20 months. The overall mortality in all groups at 180 days was 27% and 10% at 30 days. In-hospital mortality (ie, patients who died while admitted for their hip fracture) was 5% across all groups.

No difference was found in mortality and LOS for patients with fracture neck of femur compared with trochanteric fracture. Patients with total arthroplasty have a significantly shorter length of stay and reduced mortality compared to the two other groups (see Table 5).No difference in LOS or mortality was found between patients having hemiarthroplasty or internal fixation.

To further assess factors associated with mortality, a cox proportional hazard model was developed (see Table 6 and Figure 1). Based on this model, risk factors for overall mortality were: ARC residents, a higher Charlson index and older age. The length of stay and time to wait for operation were not correlated with overall mortality.

Table 4: Mortality data at 30 days and 180 days for each of the four groups (n=415 patients).

*Percentages shown reflect proportion of total patient population.

Table 5: Mortality and LOS in relation to operative procedure used.

Table 6: Cox proportional hazard model.

Figure 1: 20-month predicted survival plot for the four groups.

Table 7: Modified Charlson Comorbidity Index (CCI) for each group and comorbidities. Variables with significant between group differences are highlighted in bold.

Comorbidities and Modified Charlson Index

The Charlson Comorbidity Index (CCI)for each group is shown in Table 7. The index was calculated with excluding patient age, given all patients were over age 65. Non-ARC residents discharged directly from acute orthopaedics had a significantly lower Charlson index than other patients. No clear disparity was found between the other three cohorts. In terms of specific comorbidities, the ARC residents discharged directly from acute orthopaedics and the general geriatric rehabilitation group had significantly higher rates of dementia than the other two groups. There were significantly more patients with congestive heart failure in the ORU group(32%) compared to the non-ARC residents discharged from acute orthopaedics(15%). The non-ARC residents discharged from acute orthopaedics were less likely to have dementia and diabetes complications compared to other groups. Patients admitted to general geriatric rehabilitation were more likely to have renal disease dementia, diabetic complications and congestive heart failure than non-ARC residents discharged from acute orthopaedics. There were otherwise no differences found between groups.

Osteoporosis treatment and follow-up via Fracture Liaison Service for secondary fracture prevention

Data on osteoporosis treatment for each group at time of discharge are presented in Table 8. Due to the diversity of medication regimes prescribed, these were categorised by the first author (SW) into five categories (see Table 8). Appropriate treatment included bisphosphonates (alendronate, zoledronic acid, risedronate), teriparatide ordenosumab, all with colecalciferol 1.25mg monthly (50,000 IU of Vitamin D). Colecalciferol alone or no treatment was considered suboptimal. ARC residents discharged from acute orthopaedics were more likely to have suboptimal treatment with no documentation than any other group. Patients discharged from orthogeriatric rehabilitation were most likely to have appropriate treatment with osteoporosis medications, further investigation/referral to FLS or treatment actively with-held due to comorbidity (88% patients, p<000.1).Rates of consideration or treatment for osteoporosis were 62% in general geriatric rehabilitation group and 46% in non-ARC residents discharged from acute orthopaedics.

Fracture Liaison Service (FLS)attendance was analysed for Canterbury residents only, as residents from other areas of New Zealand or overseas are not usually seen by the FLS in Canterbury. As protocols existed for inpatient treatment for osteoporosis, FLS referral was not activated for many patients undergoing inpatient rehabilitation. The FLS also does not currently follow-up patients already on appropriate bisphosphonate or other osteoporosis treatment. Rates of FLS referral were low in ARC residents discharged from acute orthopaedics (11%). Referral rates were higher in the non-ARC residents discharged from acute orthopaedics (p<0.0001), of whom 41% (31 patients) were seen by the FLS. It is important to note that 5% of patients in this group also died in hospital so were not appropriate for referral.

Table 8: Osteoporosis treatment for each patient group.

*Appropriate treatment was with bisphosphonate, denosumab or teriparatide and vitamin D, or with vitamin D alone if patient was proven not to have osteoporosis on DEXA scan.
**Suboptimal treatment due to comorbidities were patients not considered for osteoporosis treatment beyond vitamin D due to renal failure, drug intolerances, previous atypical femoral fracture etc.
***Suboptimal treatment was considered to be no vitamin D or vitamin D alone without evidence of further investigation for underlying osteoporosis.

Level of care and functional independence measure scores

For patients discharged directly from acute orthopaedics, data was only available for their ARC funding but not level of care within this funding. For the non-ARC residents discharged directly from Christchurch, 19 patients were transferred to other hospitals in New Zealand and 11 patients died during their hospital admission. Of those who were discharged directly “home”, two went to ARC, the remainder apparently went to private residence.

However, for patients admitted to rehabilitation, AROC data is collected on admission and discharge which provided more detailed information on admission and discharge location of residence (see Table 9). Most patients (83%) from the orthogeriatric group were in their own private residence before admission. On discharge, 70% of patients in the orthogeriatric group remained in their own private residence, but 6% went to rest home-level care and 7% went to hospital-level care. Approximately 10% of the population were admitted from rest home-level care but ended up in hospital-level care on discharge. In contrast, only 59% of the patients in the general geriatric rehabilitation group were admitted from private residence. On discharge, 43% of the patients in this group remained in their own residence, and 17% of patients had moved into rest home and hospital-level care (Table 9).

Table 9: Pre- and post-discharge location of residence for Cohort 3 and 4, based on AROC data.

Functional independence measures are also collected for patients on admission and discharge from rehabilitation (see Figure 2) and were available via AROC. Both FIM cognitive and motor scores were significantly higher in the orthogeriatric rehabilitation group on admission and discharge. There were no significant differences between the two groups in the improvement of scores on discharge. This data correlates well with the ANZHFR data set.5

Figure 2: Change in functional independence measures of patients in the two rehabilitation groups.

Discussion

This study presents data on hip fracture management and rehabilitation in older adults at a single centre in metropolitan New Zealand. It followed on from previous work at Christchurch Hospital which utilised sfn® in implementing and evaluating the Fast Track Hip Fracture pathway that remains in use.3 Christchurch Orthopaedic Department admits approximately 5,500 patients per year in the busiest trauma centre in New Zealand and of those over 400 are hip fractures. Despite the availability of the ANZHFR,5which collects data on hip fracture for patients around Australia and New Zealand, we chose to analyse data using sfn due to more flexibility and efficiency in comparing multiple variables, including some that are not collected by the ANZHFR (eg, comorbidities, previous hospital admissions, factors contributing to mortality). Direct comparison with the ANZHFR was not within our research aims and was not explored. The administrative data allowed us to explore all the characteristics of the heterogenous patient groups admitted following hip fracture, to identify possible contributors to the clinical outcomes for these patients and to aid future targeted service delivery.

Fracture type and surgical intervention

The fracture types and surgical procedures have not altered from our previous work.3 Time to surgery has decreased with no negative impact on mortality. There was no difference in mortality or LOS between fracture types, but mortality and LOS was less in THA reflecting appropriate operative procedure. This observation is consistent with guidelines for the operative management of hip fractures.5

Patients discharged directly home or to a care facility

Based on anecdotal experience, we anticipated the need to separate the patients discharged directly from their acute orthopaedic admission based on their pre-morbid level of care (ie, from private residence versus residential care) due to significant within-group heterogeneity. Our study provided detailed information on patients discharged directly from the acute orthopaedic ward in two groups: pre-morbid ARC residents and non-ARC residents. Non-ARC residents discharged from acute orthopaedics were more likely to be younger, less comorbid, and a significant proportion (20%)were from outside Canterbury. However, their younger age and less comorbidity were not reflected in a mortality benefit for these patients. The non-ARC residents discharged directly from hospital are the key group targeted by the Fracture Liaison service, and referral rates were suboptimal at around 40%

On the other hand, patients who were ARC residents and were discharged from acute orthopaedics were more comorbid and had higher rates of mortality at six months compared to the other groups. They had similar rates of dementia when compared to patients transferred to general geriatric rehabilitation. Patients in this ARC resident group were also more likely to have suboptimal osteoporosis treatment with no clinical documentation regarding clinical contraindication and very low rates of referral to the Fracture Liaison service.

Patients discharged following inpatient rehabilitation

Outcomes of this study are consistent with other literature on orthogeriatric or general geriatric rehabilitation for patients with hip fracture.15 The patients in our rehabilitation groups were not randomised, so a patient’s journey following hip fracture will be largely influenced by selection bias. However, patients in our study who had orthogeriatric or geriatric rehabilitation had a longer length of stay but higher rates of treatment with osteoporosis medications. Mortality in the orthogeriatric rehabilitation group was significantly lower than the other groups at 30 days, and both the rehabilitation groups had lower mortality than the ARC residents discharged from acute orthopaedics at the 180-day follow-up.

Osteoporosis treatment

High treatment rates in older patients with hip fracture may lead to a 40% reduction in major bone re-fractures.17 In this study, treatment for osteoporosis was variable for the four groups. FLS was generally not involved in the orthogeriatric rehabilitation group or the general rehabilitation group because protocols for osteoporosis treatment exist as part of the rehabilitation process. Treatment or consideration for treatment was 84%in the orthogeriatric group and 64% in the general group. However, in those discharged back to ARC facility or to another hospital or home were low at 15%and 46% respectively. These treatment rates compare favourably with other published data.18 These results are helpful because they identify where FLS should be involved. Also, accurate recording of inpatient treatment modality was identified as an issue; in-patient treatment was not captured on a database and needed manual extraction, which has been previously described.19 This may explain some of the low percentage figures from the ANZHFR and previous work.20 Similarly, to previous research,21 this work shows that inpatient treatment for osteoporosis seems to be the most effective way of ensuring initiation of treatment. These results suggest the need for more proactive FLS involvement for patients who do not have inpatient rehabilitation, and has led the hospital to introduce a mandatory field in the discharge template on osteoporosis treatment.

Mortality

This study also sought to determine local mortality rates and examine some of the factors possibly contributing to or predicting mortality. Thirty-day mortality rates reported in this study were higher than the most recent (2018) Annual Report of the ANZHFR5 for Christchurch, possibly due to differences in data source. Mortality rates vary significantly between studies, but 30-day rates in our study were largely comparable to the existingliterature;22 however, mortality rates at 180days were similar to rates at one year reported in some studies.23 The cox proportional hazard model showed that pre-morbid ARC residency was the strongest risk factor for mortality at 180 days, with patients of older age and with more comorbidities (based on Charlson Comorbidity Index) also at increased risk for death. Interestingly, total arthroplasty was also associated with improved mortality and this probably reflects selection of a healthier patient group for this procedure.

Limitations

This study primarily utilised data coded from medical records and explored variables through sfn and multiple dataset analysis. This means that the quality of the data is dependent on the way data was documented in patient records and then coded by independent clinical coders. Other data, eg, from the Department of Internal Affairs around death is likely to be accurate but released with sometime lag, hence why only 180-day mortality was included in the current study. AROC collection is also limited, as it only collects two points of outcome at admission and discharge, being cofounding by length of stay, with potential internal bias on analysis of rehabilitation improvement.

Also, regarding osteoporosis treatment data, we had access to hospital discharge summaries and electronic prescribing software which allowed us to see if treatment was prescribed in hospital, but we did not have access to longer-term data to determine if patients continued to use these medications in the community.

Finally, this was not a randomised treatment trial and as such, confounding factors in analysing outcomes could be significant. However, the data provides useful information around clinical management of this group of patients that may still alter management, eg, promote further consideration of osteoporosis treatment in certain groups of patients and greater uptake of inpatient rehabilitation.

Implications

This study shows that tools such as sfn, that easily explore multiple variables and cohorts across various data have utility in evaluating acute hospital pathways,3 but also in evaluating a patient’s whole hospital journey, including the full acute hospital stay and subsequent rehabilitation. Although this paper gives similar results to previous work,3,10 it is different as hip fracture patients are analysed according to their discharge destination. This enabled comparisons of various outcome across these four groups. It is pleasing to see the wait for operation has decreased from previous work3 to over 80% of patients having their operation within 48 hours.

Since the introduction of fast-track hip fracture management,3the length of stay in orthogeriatric rehabilitation has decreased to 14 days and to 21 days in general rehabilitation. Baseline functional status and discharge destinations were different between the two rehabilitation groups, with the orthopaedic rehabilitation group having higher baseline and post-rehabilitation functional status and a higher percentage of patients returning to private residence. Functional status on discharge for patients undergoing hip fracture rehabilitation (measured with the FIM score) is higher than the national average standard.5

Conclusions

We have demonstrated the outcomes in four groups of patients with neck of femur fractures. The time to operation and length of stay continues to decrease compared with our previous work. New data on the four post-operative groups show better outcomes (ie, return to domicile, treatment for osteoporosis and mortality) in the dedicated orthopaedic rehabilitation group. It is difficult to interpret the significance of these improved outcomes given these patients were less comorbid, specifically with less renal disease and dementia than the general rehabilitation group. This is likely to affect their functional outcomes and prescription of osteoporosis treatment. Enhancing involvement by FLS and increased prescription of osteoporosis treatment is needed for the patients discharged directly from acute orthopaedics.

Summary

Abstract

Aim

For older patients with hip fracture, we explored patient characteristics, outcomes and osteoporosis treatments for those admitted to rehabilitation compared to those discharged directly from hospital, using data set analysis.

Method

Retrospective cohort study including all consecutive patients (65 years and over) admitted to Christchurch Hospital over one year. Outcomes were compared for patients in four groups: 1) aged residential care (ARC) residents and 2) not ARC residents discharge from acute orthopaedics, and those discharged from 3) orthogeriatric or 4) general geriatric rehabilitation. Clinical data was extracted from hospital data warehouse using signals from noise.

Results

Over 12 months, 415 patients were admitted with hip fracture. Over half (n=282) were transferred for inpatient rehabilitation. Thirty-day mortality was 10%. Mortality at 180 days was 7% in orthogeriatric rehabilitation group and 8% in general rehabilitation group. Length of stay was shorter in orthogeriatric compared with general rehabilitation (median 12.9 vs 20.4 days). Osteoporosis treatment was addressed in 88% of patients in the orthogeriatric group compared with 62% in general rehabilitation group. More patients in orthogeriatric group discharged home compared with general rehabilitation group (70% vs 43%), but functional improvements were similar. For ARC residents discharged following acute admission, length of stay was short (median 6.5 day), but overall 180-day mortality was 7%. The not-ARC resident group discharged directly from hospital was less comorbid than the other groups with lower 180-day mortality (4%). Osteoporosis treatment rates were lower in patients discharged from acute admission (15–42%).

Conclusion

Mortality was highest in ARC residents discharged from acute orthopaedics. Patients admitted for orthogeriatric rehabilitation had shorter length of stay, lower 30-day mortality, were more likely to return home and most likely to be offered osteoporosis treatment (88%), noting less underlying comorbidity and better baseline functional status in this group. This paper supports further study (eg, randomised trials) to examine the effects of targeted post-surgical rehabilitation for patients with hip fracture and explore measures to increase uptake of osteoporosis treatment.

Author Information

Samantha Warhurst, Christchurch Hospital, Canterbury District Health Board, Christchurch; Soledad Labbe-Hubbard, Planning and Funding, Canterbury District Health Board, Christchurch; Ma Yi, Biostatistician, Canterbury District Health Board, Christchurch; Jane Vella-Brincat, Pharmacy, Christchurch Hospital, Christchurch; Tonya Sammon, Ambulatory Care, Well South, Dunedin; Jane Webb, Pharmacy, Canterbury District Health Board, Christchurch; Caroline McCullough, Department of Orthopaedic Surgery and Musculoskeletal Medicine, Canterbury District Health Board, Christchurch; Gary Hooper, Department of Orthopaedic Surgery and Musculoskeletal Medicine, Canterbury District Health Board, Christchurch; Nigel Gilchrist, Orthopaedic Medicine, Canterbury District Health Board, Christchurch.

Acknowledgements

Correspondence

Dr Nigel Gilchrist, Orthopaedic Medicine, Canterbury District Health Board, Burwood Road, Christchurch.

Correspondence Email

nigel.gilchrist@cdhb.health.nz

Competing Interests

Nil.

1. Veronese N, Maggi S. Epidemiology and social costs of hip fracture. Injury, 2018. 49(8):1458–1460.

2. Bhandari M, Swiontkowski M. Management of Acute Hip Fracture. New England Journal of Medicine, 2017; 377(21):2053–2062.

3. Gilchrist N, Dalzell K, Pearson S, et al. Enhanced hip fracture management: use of statistical methods and dataset to evaluate a fractured neck of femur fast track pathway-pilot study. The New Zealand medical journal, 2017; 130(1455):91–101.

4. Sobolev B, Sheehan K J, Kuramoto L, et al. Excess mortality associated with second hip fracture. Osteoporosis International, 2015; 26(7):1903–1910.

5. Australian & New Zealand Hip Fracture Registry, ANZHFR Bi-National Annual Report of Hip Fracture Care 2018., 2018.

6. Khosla S, Hofbauer LC. Osteoporosis treatment: recent developments and ongoing challenges. The Lancet. Diabetes & endocrinology, 2017; 5(11):898–907.

7. Cosman F, Crittenden DB, Adachi JD, et al. Romosozumab Treatment in Postmenopausal Women with Osteoporosis. New England Journal of Medicine, 2016; 375(16):1532–1543.

8. Solomon DH, Johnston SS, Boytsov NN, et al, Osteoporosis medication use after hip fracture in US patients between 2002 and 2011. Journal of Bone and Mineral Research, 2014; 29(9):1929–1937.

9. Grigoryan KV, Javedan H, Rudolph JL. Ortho-geriatric care models and outcomes in hip fracture patients: a systematic review and meta-analysis. Journal of orthopaedic trauma, 2014; 28(3):e49.

10. Thwaites J, Mann F, Gilchrist N, et al. Shared care between geriatricians and orthopaedic surgeons as a model of care for older patients with hip fractures. The New Zealand Medical Journal (Online), 2005; 118(1214).

11. Elliot J, Wilkinson TJ, Hanger HC, et al. The added effectiveness of early geriatrician involvement on acute orthopaedic wards to orthogeriatric rehabilitation. The New Zealand Medical Journal, 1996; 109(1017):72–73.

12. Osteoporosis New Zealand. Osteoporosis New Zealand Strategic Plan 2017–2020. 2017 18/06/2019]; Available from: http://osteoporosis.org.nz/wp-content/uploads/ONZ-2017-Strategic-Plan-WEB.pdf

13. Osteoporosis New Zealand. Bone Plan 2020: A systematic approach to hip fracture care and prevention for New Zealand. 2012 18/06/2019]; Available from: https://osteoporosis.org.nz/wp-content/uploads/Bone-Care-2020-web.pdf

14. Kanis J, Johnell O, De Laet C, et al. A meta-analysis of previous fracture and subsequent fracture risk. Bone, 2004; 35(2):375–382.

15. Bachmann S, Finger C, Huss A, et al. Inpatient rehabilitation specifically designed for geriatric patients: systematic review and meta-analysis of randomised controlled trials. BMJ, 2010; 340:c1718.

16. Frenkel WJ, Jongerius EJ, Mandjes-van Uitert MJ, et al. Validation of the Charlson Comorbidity Index in acutely hospitalized elderly adults: a prospective cohort study. Journal of the American Geriatrics Society, 2014; 62(2):342–346.

17. Nakayama A, Major G, Holliday E, et al. Evidence of effectiveness of a fracture liaison service to reduce the re-fracture rate. Osteoporosis International, 2016. 27(3):873–879.

18. Huntjens KMB, van Geel TCM, Geusens PP, et al. Impact of guideline implementation by a fracture nurse on subsequent fractures and mortality in patients presenting with non-vertebral fractures. Injury, 2011; 42:S39–S43.

19. Klop C, Gibson-Smith D, Elders PJM, et al. Anti-osteoporosis drug prescribing after hip fracture in the UK: 2000–2010. Osteoporosis International, 2015; 26(7):1919–1928.

20. Kim D, Mackenzie D, Cutfield R. Implementation of fracture liaison service in a New Zealand public hospital: Waitemata district health board experience. The New Zealand medical journal, 2016; 129:50–55.

21. Byszewski A, Lemay G, Molnar F, et al. Closing the osteoporosis care gap in hip fracture patients: an opportunity to decrease recurrent fractures and hospital admissions. Journal of osteoporosis, 2011; 2011:404969–404969.

22. Neuburger J, Currie C, Wakeman R, et al. Increased orthogeriatrician involvement in hip fracture care and its impact on mortality in England. Age and Ageing, 2016; 46(2):187–192.

23. Mitchell R, Harvey L, Brodaty H, et al. One-year mortality after hip fracture in older individuals: the effects of delirium and dementia. Archives of Gerontology and Geriatrics, 2017; 72:135–141.

Contact diana@nzma.org.nz
for the PDF of this article

Subscriber Content

The full contents of this pages only available to subscribers.

LOGINSUBSCRIBE