Introduction
The increasing prevalence of osteoporosis is associated with increased risk of a bone fracture [
1]. A Dutch study based on claims data from all Dutch healthcare insurers reported an annual average of 114,116 patients with a fracture was identified between 2009 and 2011, of which 32% were attributed to osteoporosis [
2]. Patients with a recent fracture after the age of 50 years have an increased risk of subsequent fractures. This risk, which is referred to as imminent subsequent fracture risk [
3], is highest immediately after the initial fracture and then declines. In addition, the majority of deaths following fractures occur within the first year; thereafter, the excess mortality gradually declines [
4]. From the perspective of caregivers, morbidity and mortality following a fracture are important clinical considerations, along with substantial loss of patients’ quality of life.
To improve secondary fracture prevention, Fracture Liaison Services (FLSs) are advocated as the most appropriate and effective approach to identify, investigate, and treat patients at risk of new fractures. The first FLS was introduced in 1999 by McLellan and colleagues in the UK [
5]; since then, awareness of initiating FLSs worldwide has increased through the Capture the Fracture (CTF) campaign in 2012 by the International Osteoporosis Foundation (IOF) [
6] and by other professional organizations [
7,
8]. Until May 2021, 644 FLSs (registered in CTF) have been implemented in 48 countries; nearly half of these are in Europe.
To capture the full burden of fractures for a society, it is essential to assess their impact on health. Insight into the loss of healthy life years can facilitate rational decision-making when allocating resources across fracture types or diseases [
9]. Health state utility value (HSUV) measures are a specified type of health-related quality of life (HRQoL) instruments that reveal the society’s preference or value for specific health states. HSUV is an essential component in economic evaluations, used to establish whether the cost of a new intervention can be justified in terms of expected health benefits [
10]; it can help decision-makers in prioritizing health interventions.
In fracture research, a large number of studies were conducted to investigate HSUV in patients not attending an FLS. For example, the large ICUROS study [
9] reported that fractures resulted in substantial HRQoL loss directly after fracture, and the HRQoL improved after 4 months but did not return to pre-fracture levels. In 2014, a meta-analysis identified 62 studies that reported HSUVs after hip, vertebral, or distal forearm fracture [
13]; the study populations were heterogeneous (e.g., pre-fracture, post-hip/vertebral/wrist fracture), and most studies had a small sample size and were limited by short follow-up periods, indicating that fracture events were associated with decrements in HSUVs which differed between fracture types.
Very few studies were conducted to investigate HRQoL in patients with fractures attending an FLS [
14,
15], especially the course of HSUV in the long term. Therefore, the main objective of this study was to investigate the course of HSUV in patients with a recent fracture presenting at an FLS in the Netherlands, as measured by two generic preference-based instruments: the EuroQol (EQ-5D-5L) and the Short Form Health Survey (SF-36), over a 3-year follow-up. In addition, considering previous studies [
9,
11,
16,
17] indicated that demographics and fracture-related characteristics such as age, previous fracture, hospitalization, and treatment initiation were significantly associated with patients’ HSUV, our secondary objective was to identify factors associated with HSUV in patients at the FLS.
Materials and methods
Subject and study procedures
This study used data from the “FX MoVie Study,” which is a 3-year prospective observational study conducted at the FLS of VieCuri Medical Center in Venlo, the Netherlands [
18]. The study protocol (registration number NL45707.072.13) was approved by an independent Medical Ethics Committee and complied with the Declaration of Helsinki. All patients gave written informed consent prior to participation and after receiving oral and written information on the study.
According to standard care, a nurse specialized in osteoporosis invited all patients aged 50 years and older, who visited the emergency department (ED) because of a recent clinical vertebral or non-vertebral fracture, to the FLS. All patients attending the FLS between October 2014 and June 2016 were screened for participation in the “FX MoVie Study.” A total of 1380 FLS attenders were screened for eligibility, of whom 990 were eligible to participate and a total of 500 patients aged between 50 and 90 years with a recent, radiologically confirmed fracture participated. We excluded non-Caucasian patients, patients with a fracture due to high-energy trauma, bone metastasis, failure of prosthesis, or osteomyelitis, and patients with cognitive impairment.
All participants received a detailed questionnaire for evaluating clinical risk factors for fractures, such as medical history, medication, previous fractures, and calcium and vitamin D intake, and were scheduled for dual X-ray absorptiometry (DXA) measurement, vertebral fracture assessment (VFA), and a blood test. In addition to the questionnaire for evaluating risk factors, HRQoL questionnaires (EQ-5D-5L and SF-36) were filled out at 3, 6, 12, 24, and 36 months after inclusion. Of note, the patients’ first visits at the FLS were scheduled 3–4 months after their fracture, so HRQoL data immediately after their fracture were not available. Three and 6 months after inclusion, patients received the HRQoL questionnaires and a fall diary. Furthermore, they received a telephone call from the research assistant to verify whether they sustained a fall or a subsequent fracture and to complete the questionnaires in case of missing data. At 12, 24, and 36 months after inclusion, patients came to the hospital for a study visit and the questionnaires were repeated.
Bone mineral density (BMD) in the left or right hip and the lumbar spine was determined using DXA with the Hologic QDR 4500 (Hologic, Bedford, MA, USA). Diagnosis of osteoporosis was based on the World Health Organization criteria for BMD [
19] according to the lowest value of T-score in femoral neck, total hip, or lumbar spine: osteoporosis as a T-score of − 2.5 or less, osteopenia as a T-score between − 2.5 and − 1.0, and normal BMD as a T-score of − 1.0 or higher.
Assessment of vertebral fractures (VFs) was performed via vertebral fracture assessment (VFA). VFs were graded according to the grading of Genant et al. [
20] as mild (grade 1, 20–24% reduction in vertebral body height at the anterior, mid, or posterior location), moderate (grade 2, 25–39% reduction), or severe (grade 3, ≥ 40% reduction), respectively.
If laboratory results were abnormal, additional investigations were performed for detailed evaluation of newly diagnosed contributors to secondary osteoporosis or other metabolic bone disorders and treatment was initiated when necessary.
Demographics and disease-related characteristics
The socio-demographics included age (years at time of fracture), gender, and body mass index (BMI). Baseline fracture-related characteristics were collected through (1) questionnaire (and further verified during FLS visit): smoking, fracture site, previous fracture, previous falls (last year), parental hip fracture, use of a walking aid, visual and hearing impairment, previous treatment with anti-osteoporosis medication (AOM), and medical history (which is classified based on International Classification of Diseases version 10), and (2) laboratory tests: BMD, prevalent VFs, secondary osteoporosis, and vitamin D deficiency. In addition, the specific times of new falls and subsequent fractures were recorded for each patient during 3-year follow-up, and we assumed that none of the patients had a fall or a subsequent fracture between their baseline fracture (i.e., the fracture for which they were invited to attend the FLS) and the time they attended the FLS.
Fracture classification
Patients’ index fractures were recorded in electronic Case Report Forms. For purpose of analyses, these fractures were grouped into ten categories according to their location as clavicle/scapula, humerus, radius/ulna, hand/foot, vertebra, rib/sternum, pelvis, femur, tibia/fibula/patella, and multiple fractures (if patients had more than one index fracture). In addition, based on visual inspection, patients with femoral, vertebral, or multiple fractures had a strikingly lower baseline HSUV in comparison with patients with other fractures; the pre-defined ten categories were further divided into two groups (femoral/vertebral/multiple fractures vs. other fractures) to investigate between-group differences.
Health state utility value (outcome)
HRQoL is expressed in the form of HSUV, which is scored on a scale that assigns a value of 1 to a state equivalent to full health and 0 to a state equivalent to death. In our study, HSUVs were calculated according to EQ-5D-5L and SF-36 data. The EQ-5D-5L quantifies health status in terms of five dimensions: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Patients score each dimension on a five-level Likert scale (no problem, slight problem, moderate problem, severe problem, and extreme problem). To translate the EQ-5D profiles to societal HSUVs, a value set based on population preferences in the Netherlands was used [
21]. The predicted values for the EQ-5D-5L range from − 0.446 to 1, where HSUV below 0 represents health states considered worse than death. The SF-6D was derived from the SF-36 health status measure (version 1, UK Programme) including 6 dimensions of health: physical functioning, social functioning, role limitations, pain, mental health, and vitality, with each dimension having four to six levels. The societal HSUVs were computed using the algorithm developed by Brazier et al. [
22]. The predicted values for the SF-6D range from + 0.291 to + 1.
Statistical analysis
For baseline characteristics, descriptives are provided as means and standard deviations (SD) for continuous variables, number, and percentage (%) for categorical variables. Comparisons between groups (different fracture sites) were conducted using the independent samples t-tests or one-way analysis of variance (ANOVA) for continuous variables, and Pearson’s chi-square test or Fisher’s exact test for categorical variables.
Multiple imputation (MI) with fully conditional specification was employed to impute missing EQ-5D and SF-6D data. The number of imputations was set to 18, given that approximately 18% of incomplete cases were identified. Patients’ missing index values were drawn at six time points, using predictive mean matching. Details of missing data and MI were reported in Supplement
1.
Linear mixed-effects regression with random intercept and slope for patients was applied to compare the mean HSUV between baseline and each time point, to explore the course of the HSUV for all sites of fracture over the 3-year follow-up, and to investigate the impact of fracture site on the course of HSUV over time by including an interaction term of time and baseline fracture group (categorical variable with ten pre-defined categories) in the model.
Additionally, given that the time period between index fracture and FLS visit varied among patients, a subgroup analysis was conducted to explore the difference in the change of HSUV. Since this time period was normally distributed, we used the mean (107 days) as a cut-off to categorize our patients into two groups (i.e., time between index fracture and FLS visit < 107 days vs. ≥ 107 days). A clustered line graph was used to visually display the difference in HSUV change over time and the mean difference (MD) was calculated between baseline and each time point. Besides, considering patients’ characteristics might influence the results, we also investigated the difference in five most relevant baseline characteristics, i.e., age, gender, BMD, index fracture (femoral/vertebral/multiple fracture, other fractures), and medical history (ICD-10 coded diseases) between two groups using previous mentioned statistical method to further validate our results.
Moreover, the impact of subsequent fracture on HSUV was also investigated. Of note, 20 patients already had a subsequent fracture before the first FLS visit and were therefore excluded for the following analyses. First, the subsequent fracture was treated as a time-varying variable to explore the overall association with HSUV, both between-subjects and within-subjects interpretation were provided. Second, for patients who had a subsequent fracture (during follow-up), the HSUVs before and after a subsequent fracture were compared for these patients. To capture the maximum impact of a subsequent fracture, the HSUV just before and immediately after the subsequent fracture was treated as pre- and post-HSUV, respectively (i.e., if one patient had a subsequent fracture at 6 months, the HSUV at 3 months was treated as pre-HSUV and the HSUV at 6 months as post-HSUV). Third, the impact of different locations of subsequent fracture was also investigated. Patients were subdivided into two groups (subsequent femoral/vertebral/multiple fractures vs. subsequent other fractures), the HSUVs before and after a corresponding subsequent fracture were compared for each subgroup. Fourth, we subsequently compared the HSUVs in the group of patients without subsequent fracture to those with a subsequent fracture and applied the median time to subsequent fracture (which was 354 days) to both groups in order to compare HSUVs in the time period before and after subsequent fracture (period 1: 0–364 days vs. period 2: 365–1095 days) in both groups. Given the HSUVs were estimated at discrete time points, and the median time to subsequent fracture was 364 days, the mean HSUV of baseline, 3 months, 6 months, and 12 months was therefore treated as pre-HSUV, and the mean HSUV of 24 months and 36 months was treated as post-HSUV for both groups. Through pre-testing, normal distribution was indicated for the difference (in mean HSUV) between period 1 and 2 for both groups, and given the HSUV of each patient was repeated measured, mean difference was therefore calculated through paired-samples T-test.
Furthermore, the identification of factors associated with the average HSUV (over 3 years) was also conducted, using the linear mixed-effect regression model. The above-mentioned demographics and disease-related characteristics (16 baseline and two longitudinally assessed variables) were considered potential factors and therefore included in the model. A backward stepwise elimination was applied to omit insignificant factors from the model.
The above-mentioned longitudinal analyses for EQ-5D HSUV were included as the main analysis, and the analyses based on SF-6D HSUV were included as sensitivity analysis. All mixed-effects regression analyses were adjusted for age, gender, and baseline BMD. All analyses were conducted using SPSS (version 26.0, IBM Statistics), and a P-value of ≤ 0.05 was considered statistically significant.
Discussion
This study, to our knowledge, is the first longitudinal study using prospective data over 3 years, with the objective of estimating the HRQoL of patients following visits to a FLS because of a recent fracture. With regard to EQ-5D HSUV, no significant change was captured over 3 years, although a small but statistically significant improvement was observed at 6 and 12 months in comparison with baseline HSUV. This short-term improvement is more likely due to natural healing of the fracture rather than the effect of attending an FLS or having a fracture risk evaluation at the FLS. When patients were stratified by baseline fracture (femoral/vertebral/multiple fractures vs. other fractures), our results remained (i.e., there was no significant overall change). We did not find any previous study investigating the long-term HRQoL of patients attending a FLS. In contrast, two studies [
14,
15] compared the HSUV of patients with a recent fracture before and after the introduction of a FLS; no significant difference between two groups was identified at the 6 or 12 months follow-up.
The primary potential reason for a non-significant change in HSUV over 3 years is that the patients included in our study attended the FLS 3–4 months after their index fracture; thus, their HRQoL might have already improved through natural fracture recovery and/or through treatment in the emergency department before attending the FLS, resulting in non-striking change after attending FLS. This was also found in the ICUROS studies [
9,
11,
12]: although substantial loss of HRQoL was captured in the short term after fracture, patients’ HRQoL was largely improved at 4 months. Except for ICUROS studies, the improvement was also identified in patients who did not attend a FLS in recent studies [
14,
15], which can also be attributable to natural healing (recovery) of the fracture. The second potential reason for lack of significant change may be selection bias; it is likely that patients with more severe fractures, older patients, or patients who were hospitalized did not attend the FLS. Furthermore, approximately half of FLS attenders did not consent to participate in this study. In this study, the average baseline HSUV was 0.81, which is a bit lower but still comparable to the HSUV of community-dwelling Dutch residents aged 65 years and older as reported by Mangen et al. [
23]. Therefore, our study included relatively healthy patients, which may have resulted in a relatively good HRQOL after the index fracture, without a significant change over time.
Additionally, the greater increase in QOL in patients with a shorter time period (< 107 days) between index fracture and FLS visit may be attributed to the earlier stage of the fracture healing process, and by the timely treatment from FLS clinic compared to patients with a longer time period to FLS visit.
Of note, although patients’ baseline HRQoL was measured 3–4 months after their index fracture, patients with femoral, vertebral, or multiple fractures still had a lower HSUV in comparison with patients with other fractures. Also, significant improvement in HRQoL takes a longer time to capture (baseline HSUV as the reference) as reported in our study. Fisher et al. [
24] assessed the timeline of functional recovery after hip fracture in seniors (aged 65 years and older) and reported that objective functional recovery (lower extremity function) was largely complete in the first 6 months, whereas subjective recovery (HRQoL) improved up to 9 months after hip fracture. In addition, we also identified that the number of baseline fractures impacted patients’ HRQoL. Patients with multiple fractures had a significantly lower HSUV in comparison with patients with only one index fracture such as clavicle/scapula, radius/ulna, hand/foot, rib/sternum fracture. This finding is supported by a previous study showing that patients with multiple clinical fractures would experience an additive effect, resulting in disability similar to a single hip or vertebral fracture [
25] which is in line with our finding that the absolute average HSUV (over 3 years) difference between femoral, vertebral, and multiple fractures was not significant.
Suffering a subsequent fracture was associated with a decrease in HSUV; this finding is supported by several previous studies [
26‐
28] reporting that subsequent fractures have a significantly negative impact on the QoL, greater loss of function, and increased mortality. However, whether the effect of FLS is larger immediately after subsequent fracture, and whether the cost-effectiveness of FLS is somewhat better, it remains unknown based on our data and recently published studies. Furthermore, to accurately estimate the effect of FLS on QoL is even difficult given QoL is determined by multiple factors such as lifestyle, aging, and comorbidities; however, it would be interesting for future studies. Also, the presence of a moderate or severe prevalent VF was associated with a lower HSUV, which is in line with recent findings of Shah et al. [
29]. Besides, patients who previously received treatment with AOM reported lower utility in comparison with their counterparts. The lower HSUV might be explained by two aspects: first, since the indication for treatment (according to the Dutch guideline) is osteoporosis and/or a moderate or severe VF, patients’ awareness of having osteoporosis and/or VFs and increased future fracture risk might result in over cautiousness and limiting daily activities; second, the potential side effects of oral AOM including bone, joint, or muscle pain, as well as nausea, difficulty swallowing, and heartburn might affect patients’ HRQoL, and the rare side effects such as osteonecrosis of the jaw and atypical femoral fracture actually scare many patients away from taking AOM, with the result that suboptimal persistence and adherence lead to an increased rate of fracture and to worse HRQoL [
30]. Moreover, patients’ HSUVs following a fracture are clearly negatively associated with high BMI since excessive body fat produces inflammatory cytokines which may stimulate bone resorption and reduce bone strength [
31]. Several recent studies reported that both low and high BMI are risk factors for fragility fracture. The study of Yan et al. [
32] investigated the relationships of BMI with HRQoL in adults 65 years and older and revealed that compared with normal-weight people, both underweight and obese older adults reported impaired QoL, particularly worse physical functioning and physical well-being. Furthermore, some previous studies identified that gender [
13,
16] and previous fracture [
9] are important factors; these were not captured by EQ-5D in our study.
In a sensitivity analysis using another HRQoL measure, namely the SF-6D, comparable results were identified and our main message remained, though there were small differences. For example, the mean SF-6D HSUV was generally lower than the EQ-5D HSUV, and only SF-6D indicated that the course of HSUV for patients with hand/foot fracture was significant. Besides, when the median time to subsequent fracture (364 days) was applied to compare the HSUVs patients without subsequent fracture to those with a subsequent fracture, a higher median EQ-5D HSUV was captured in period 2 for both groups, however, which was only observed for patients without subsequent fracture by SF-6D HSUV. These discrepancies between the two instruments might result from differences in the content of the descriptive systems and in the variation of scoring algorithms [
33]; a further head-to-head comparison will be conducted to explore the equivalency of EQ-5D-5L and SF-6D cross-sectionally, including scoring distribution, domain content, and longitudinal validity.
This study has some limitations. First, relatively healthy patients were included in this study (selection bias), so the generalizability of the findings of this study to patients with a fracture could therefore be questionable. Second, our patients attended the FLS approximately 3–4 months after their index fracture; we therefore lacked HRQoL data immediately after fracture, which might limit our capturing an overall significant change of HRQoL in the first several weeks of recovery from a fracture. Finally, 18 patients scored an item of EQ-5D twice; we calculated their two utility scores and used the average value for further analyses, which might affect the results.
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.