Background
Fragility fractures are a result of progressive decrease in bone density and bone strength [
1]. This presents a public health issue and has recently become a major focus of disease prevention activities for clinicians [
2,
3]. Fragility fractures are associated with a higher risk of subsequent fractures (“fracture cascade”) and increased mortality [
2,
4]. Although hip, vertebral, upper arm and forearm have been long considered the most common fracture sites, the epidemiology of the condition appears to be changing: non-hip, non-vertebral fractures now account for a substantial portion, up to 40 %, of fragility fractures, representing a non-negligible economic burden [
5,
6].
Fracture-related disability has an enormous impact on patients’ quality of life. Measured by a number of validated tools, health related quality of life (HRQoL) has been shown to consistently decrease following fragility fractures [
7‐
10]. Few studies, however, have assessed changes in HRQoL over the long term. The results of those that did demonstrated a lasting negative impact on HRQoL of hip and spine fractures in particular, and a rebound in HRQoL to pre-fracture levels in sites such as forearm or ankle [
11‐
13].
Although very informative, existing studies have important limitations. They rarely focus on the elderly population receiving HC services or residing in LTC facilities, instead focusing on all patients, although the demographic transformation in developed countries prompts researchers to explore health aspects of this population cohort in more detail [
14]. Institutionalized patients, who are hypothesized to suffer more pronounced HRQoL losses after fractures, may be underrepresented in these studies due to recruitment bias in research conducted in community-based settings [
15]. Also, the relatively small sample sizes often limits interpretation of their results [
13,
16‐
18]. The objective of this study was therefore to estimate the long-term change in HRQoL following fractures among the elderly receiving home care (HC) services or living in long-term care (LTC) facilities using linked administrative databases from Ontario, Canada.
Methods
Study overview
Several administrative datasets from the Canadian province of Ontario (population 14 million) were linked to document post-fracture changes in HRQoL among individuals receiving HC services or living in LTC. HRQoL, as evaluated with the Health Utility Index-version 2 (HUI-2) instrument, was compared 3 years post fracture with pre-fracture levels according to fracture site and residency status (LTC or HC).
Study population
In a first step, Ontarians aged 50 years and older with any acute care admission, emergency visit or same day surgery due to a fracture during fiscal years (FY) April 1, 2006 to March 31, 2011 (FY 2007/11) were identified using two databases hosted by the Canadian Institute for Health Information (CIHI): 1) Discharge Abstract Database (DAD) [
19] for acute care admissions; and 2) National Ambulatory Care Reporting System (NACRS) [
20] for emergency visits and same day surgery visits. Fracture sites were identified using the International Classification of Diseases (ICD-10-CA) fracture codes: hip, humerus, clinical vertebral, wrist (distal forearm), other sites (femur, lower leg including tibia, fibula knee and foot, lower arm including radius and ulna, ribs, shoulder, arm, sternum, clavicle, pelvis) and multiple fractures (i.e. more than 1 of the above). Fractures with high-impact trauma codes were excluded. The ICD-10-CA codes used to identify low-trauma fractures are provided in Additional file
1.
In a second step, the DAD and NACRS records for those individuals identified were linked to the Home Care Reporting System (HCRS) [
21] and the Continuing Care Reporting System (CCRS) [
22] to identify those individuals who received HC services or were living in LTC facilities during FY 2007/11, respectively. To be included in the analyses, individuals had to have pre- and 36-month post-fracture HRQoL measurements in either setting (i.e. LTC or HC). This means that individuals living at home who transferred to LTC or HC following a fracture were not included in our analysis due to the lack of HRQoL data pre fracture (see more details in next section). To document the long-term impact of fracture on HRQoL, individuals had to have at least 3 years of follow-up after a fracture. Two analytic cohorts of Ontarians experiencing a fracture were created depending on their residency status (living at home but receiving HC services or living in a LTC facility).
The primary outcome of this study was the change in HRQoL 3 years following a fracture. HRQoL was estimated using the HUI-2 which is included among other instruments/questions in the InterRai Minimum Data Set (MDS) [
23]. The MDS is a mandatory questionnaire for all Ontarians living in publicly-funded LTC facilities or for patients who receive nurse-based HC services in Ontario as prescribed by a physician. In contrast with CIHI databases which are triggered by an admission and are purely administrative (e.g. no HRQoL information available for analysis), the MDS is administered to every patient by a nurse or social worker at a minimum quarterly, or with any change in clinical condition. In addition to the HUI-2 instrument, the MDS includes more than 200 questions on clinical outcomes and resource utilization, for both individuals receiving HC services or living in LTC.
The HUI-2 is a validated instrument to measure HRQoL and it has been used in hundreds of studies and clinical settings [
24]. The HUI-2 measures HRQoL in six dimensions: sensation, mobility, emotion, cognition, self-care and pain. Each domain has 3–6 levels of ability/disability and the responder has to select one level for each domain. Using a scoring algorithm [
25], the individual responses to each of the 6 domains are transformed into a utility score where 0 represents death and 1 the best imaginable health state [
24]. As a reference, on a 0 (death) to 1 (full health) scale, the HUI-2 utility score was 0.61 for Canadians aged 80–84 years and living in the community [
26]. The minimal clinically important difference (MCID) in the HUI-2 utility score is 0.03 while the MCID associated with each of the 6 domains of the HUI-2 is 0.05 [
24]. In our study, the HUI-2 utility score was calculated for the 1-year period preceding the fracture and at post-facture periods of 1, 3, 6 months, and every 6 months thereafter for up to 3 years. Changes in HUI-2 utility scores compared to pre-fracture level, also known as disutilities, were calculated 36 months post fracture to document the impact of fracture in HRQoL.
Statistical analyses
Continuous variables (e.g. age) were summarized using mean values and standard deviations (SDs) while discrete variables (e.g. presence of comorbidities) were represented using percentages. Student t-tests and chi-square tests were used to compare study population characteristics or fracture types between the LTC and HC cohorts.
Paired t-tests were used to compare assessments across time for the HUI-2. In addition, regression analyses were conducted to explain the disutility associated with fracture at 3 years post fracture (i.e. 3-year HUI-2 utility score minus pre-fracture HUI-2 utility score). Covariates included age, sex, fracture type, residency status (HC or LTC), and comorbidities. Comorbidities were derived from the MDS questionnaire and were represented in the analyses as a binary variable for seven categories: endocrine/metabolism (diabetes, hypo/hyper-thyroidism), heart/circulation (arrhythmia, congestive heart failure, hypo/hyper-tension, peripheral vascular disease), musculoskeletal (arthritis, amputation, osteoporosis, history of fracture), neurological (Alzheimer’s Disease, dementia, stroke, Parkinson’s Disease), psychiatric/mood (anxiety disorder, depression, bipolar disorder, schizophrenia), pulmonary (asthma, emphysema, chronic obstructive pulmonary disease), and sensory (cataracts, retinopathy, glaucoma, macular degeneration). In these analyses, generalized mixed effects models were used to account for skewness and heterogeneity. In addition to analyzing the combined HC and LTC cohorts, regression analyses were also conducted for the following four populations: 1) HC cohort only; 2) LTC cohort only; 3) individuals with neurological conditions; and 4) individuals without neurological conditions. To explore the impact of censoring in the data, an analysis was conducted to compare the decline in HRQoL following a fracture between those with 3 years of data and those who died before 3 years. We also explored the decline in HRQoL for different types of hip fractures including neck of the femur (ICD-10-CA S72.0), pertrochanteric (ICD-10-CA S72.1) and subtrochanteric (ICD-10-CA S72.2) fractures.
Discussion
To our knowledge, this is the first time that HRQoL up to 3 years after a fracture has been documented in a large contemporaneous elderly population living in LTC or receiving HC services prior to experiencing a fracture. This is also the largest Canadian study of HRQoL for low-trauma fractures to date. The results of this study are important as they show, in contrast to other studies [
11‐
13], that despite a rebound one month following fracture, there was a long-term permanent decrease in HRQoL for all types of fractures i.e., humerus, vertebral, wrist, multiple and other.
It is somewhat difficult to compare our findings with results from most previous studies [
7,
9‐
11,
27‐
32] due to the differences in instruments used to measure HRQoL (e.g. EQ-5D, SF-36), study demographics (e.g. younger population) or settings (e.g. community). However, there are several Canadian studies of particular interest as they have used the HUI-2 instrument to measure HRQoL in similar settings. First, the baseline pre-fracture HUI-2 utility scores for our HC study population (e.g. 0.540) were consistent with the HUI-2 utility score reported in a study of Canadian and American older frail adults receiving HC services (0.49) [
33]. Another Canadian study reported that the HUI-2 utility score for Canadians aged 80 to 84 and receiving HC services was 0.580 compared to 0.540 for our overall HC [
25]. The same study also reported that the HUI-2 utility scores for Canadians aged 80 to 84 living in an institution was 0.33 (compared to 0.39 for our LTC cohort).
A Canadian study of particular interest is the Canadian Multicentre Osteoporosis Study (CaMos) which also used the HUI-2 instrument to document the 5-year impact of fracture on HRQoL. Out of a cohort of approximately 6600 non-institutionalized individuals (mean age: 66 years; 83 % women), 726 experienced a fracture [
12]. When compared to individuals who did not experienced a fracture, the 5-year disutilities associated with a hip fracture among women and men were −0.12 and −0.09, respectively (
p < 0.05), which is very similar to our findings when comparing pre- and post-fracture HUI-2 utility scores. This is an important result as it has been argued that HRQoL losses in institutionalized adults may be lower compared to community dwelling individuals due to their older age and worse baseline health status [
15]. CaMos findings also indicated a return to pre-fracture HRQoL for non-hip non-vertebral fractures, which was also reported in other studies at 1 year [
11], 2 years [
16] or 7 years post fracture [
13]. In both the CaMos study and our study, mobility was the HRQoL domain that was most impacted by a fracture.
Our study has several strengths including the large sample size and the 3-year follow-up. This was made possible by linking several administrative databases to identify our study population. While many Canadian administrative databases do not contain information on HRQoL, every Ontario resident receiving nurse-based HC services prescribed by a physician or living in LTC are required to complete the MDS questionnaire during face to face interviews with a nurse or social worker. Since the HUI-2 instrument, a validated HRQoL instrument, is part of the outcomes measured in this questionnaire, the data linkage allows the identification of individuals with fractures (through the hospitalizations or emergency visit databases) living in LTC facilities or receiving HC services. An additional strength of this study is to include all individuals receiving HC services or living in LTC facilities and who had a fracture in the largest province of Canada. In addition to the comprehensiveness in the patient selection, the HUI-2 has shown high reliability and validity [
24].
Some of the limitations of the study are related to the nature of the MDS, which is either conducted at home or in LTC facilities by nurses of social workers as part of their normal routine to monitor patient’s progress. As such, the study population excluded individuals who received care in the first few months after fracture in the acute care and non-acute institutional care setting (e.g. admissions to rehabilitation hospital and complex continuing care hospitals). Similarly, to document change in HRQoL pre and post fractures, we excluded anyone who did not receive HC services or live in LTC facilities before the fracture. Although everyone in our study had to have a HUI-2 utility score pre fracture and 3 years after a fracture, the frequency of HUI-2 measurements over time varied, as the MDS is not consistently administered (e.g. every 3 months). For example, the mean number of pre- and post-fracture HRQoL assessments was less than 3 and 6, respectively. For these reasons, we took an average of the HUI-2 utility scores over the year preceding the fracture rather than only using the closest pre-fracture HUI-2 utility score. Since we did not impute the missing data, we did not use time series techniques or repeated measures analyses when estimating losses in HRQoL over time. Rather, we explored changes in HRQoL between two time points for which we had complete data (3-years post fracture minus pre-fracture levels). Due to the recent implementation of the MDS in Ontario, we were only able to follow up patients for 3 years. Future research will allow us to follow the cohort for longer periods as the mandatory data capture of the MDS continues. Another limitation related to the MDS may be related to the capture and reporting of comorbidities, which may be underreported. It is currently unknown, for example, how the MDS reporting of comorbidities compares to claims data. We also did not have access to data on factors other than medical conditions which may have impacted HRQoL, such as level of education, or family and other social support. In addition, the study used a pre-post design to evaluate the impact of low-trauma fractures on HRQoL. As such, we were not able to evaluate the HRQoL of a comparable cohort with no fracture. It should also be noted that the HUI is a general QoL instrument which may not be responsive to small changes in clinical status compared to a disease specific QoL questionnaire. While we believe that the vast majority of the low-trauma fractures included in our analysis were fragility fractures, mechanism of injury cannot be reliably ascertained from the use of ICD-10 codes. As a reference, fragility fracture has been defined as a fracture that occurs spontaneously or following a minor trauma, such as a fall from standing height or less [
1]. For these reasons, we concentrated our analyses by identifying through ICD-10 codes all fractures that were not high-trauma fractures. Similarly, while we believe that most of our population was frail due to the high presence of comorbidities, we could not ascertain that all individuals receiving home care services were frail. Finally, while the data are from Ontario, we believe that the results are applicable to the other provinces of Canada.
Competing interests
Dr. Tarride has received grant/research support from Amgen Canada.
Dr. Leslie has received grant/research support from Amgen and has served as a speaker for Amgen, Eli Lilly, and Novartis
Dr. Morin has received grant/research support from Amgen and Merck and consultant fees from Amgen.
Dr. Adachi has received research support from Amgen and has been a consultant for Amgen, Eli Lilly, Merck, Novartis, and Actavis.
Dr. Papaioannou has received grant/research support and consultant fees from Amgen.
Dr. Bessette has received grant/research support from Amgen.
Dr. Brown has received grant/research support, consulting fees, and speakers’ bureau fees from Amgen and Eli Lilly, and grant/research support from Novartis.
Dr. Pericleous is employed by Amgen Canada, which funded this study.
Authors’ contributions
JET participated in the conception and design of the work, the interpretation of data, and drafting of the manuscript. RH participated in the conception and design of the work, conducted the statistical analysis, and participated in the interpretation of data and drafting of the manuscript. NB, WL, SNM, JA, AP, LB, JB, and LP participated in the conception and design of the work, interpretation of data, and revision of manuscript for important intellectual content. SM participated in the interpretation of data and drafting of the manuscript. All authors read and approved the final manuscript.