Article Text
PDF
Abstract
Objectives This study aimed to examine the risk of mortality following incident and subsequent osteoporotic fractures, the effect of different fracture type combinations, and the mediating role of postfracture morbidity in a Danish population.
Methods We used the National Patient Registry to identify patients ≥60 years with incident major osteoporotic fracture of the hip, vertebrae, wrist or humerus between 2013 and 2018, and controls matched 1:10 on age and sex. Possible mediators were identified using International Classification of Diseases, 10th Revision codes registered in the 6 months following index fracture. HRs were estimated using Cox regression analyses with 95% CIs. The effect of possible mediators was estimated using mediation analyses.
Results The study included 106 303 patients and 1 062 988 controls. Mortality following index fracture was highest in the month following hip fractures (HR 10.98 (95% CI 10.23 to 11.79) in women and HR 16.40 (95% CI 15.00 to 17.93) in men). Subsequent hip fractures resulted in the highest HRs for all fracture type combinations. In women, the highest HR was observed in patients with index wrist/subsequent hip fractures (HR 2.43 (95% CI 2.12 to 2.78)). In men, the highest HR was observed in patients with index humerus/subsequent hip fractures (HR 2.69 (95% CI 2.04 to 3.54)). Pneumonia mediated the largest proportion of mortality, but dehydration, urinary tract infection and sepsis were also important factors.
Conclusions The highest mortality risk was found in the month immediately following both index and subsequent fracture. The combination of index and subsequent fractures at different skeletal sites had a substantial impact on the risk of mortality. Postfracture morbidities were found mediate the association.
- Osteoporosis
- Epidemiology
- Outcome Assessment, Health Care
Data availability statement
Data may be obtained from a third party and are not publicly available. Data are accessible through the Danish Health Data Authority.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Excess mortality following osteoporotic fracture is well examined, but the impact of index and subsequent fracture type combinations and mediating postfracture comorbidities is unclear.
WHAT THIS STUDY ADDS
This study shows that the combination of index and subsequent fractures at various skeletal sites impact excess mortality, and that mortality is mediated by comorbidities occurring in the 6 months immediately following fracture.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
The findings of this study may be of value for clinical guidelines for osteoporotic fracture patients.
Introduction
The association between osteoporotic fractures and excess mortality is well established, and while the link has been most thoroughly investigated in hip fractures,1 it has also been observed in other fracture types.2–4 Patients with a previous fracture have a higher risk of subsequent fracture,5 6 with studies also observing close to a twofold increase in mortality risk among these patients.3 5–7 The impact on mortality of different fracture type combinations have been examined to a much lesser extent8 but may be an important component in understanding this association.
Additionally, comorbidities further complicate mortality following fracture. More than 40% of fracture patients exhibit two or more comorbidities around the time of fracture.9 10 It has been demonstrated that fracture patients with comorbidities such as chronic obstructive pulmonary disease, cardiovascular or respiratory disease have a higher risk of mortality than those without.11–15 Further, previous studies have concluded that comorbidities contributed substantially to excess mortality in hip fracture patients16 and vertebral fracture patients,17 respectively. However, studies examining perifracture or postfracture comorbidities have largely been limited to hip fractures.12 14 18 As such, it is less clear how much of the observed increase in mortality risk is attributable to fracture complications and how much to the fracture itself, especially for fracture sites other than hip. To our knowledge, no studies have examined the mediating role of postfracture comorbidities and complications in a larger population.
The aim of this study was to examine the risk of mortality following incident and subsequent osteoporotic fracture, as well as the association of different fracture type combinations and the ensuing mortality in the Danish population, while also considering the mediating role of postfracture morbidities.
Methods
Study design and data sources
The study was designed as a matched nationwide register-based cohort study, based on the Danish national registers covering all citizens living in Denmark on 1 January 2013.
In Denmark, administrative data on all individuals are stored in registries and all permanent residents of Denmark have unique identification numbers; thus individual-level linkage of data across registries is possible.19 We obtained data from the Civil Registration System (CRS) and the National Patient Registry (NPR), provided by The Danish Health Data Authority.
The CRS was established in 1968 and maintains complete records of births, deaths and emigration status of all residents in Denmark.20 The CRS was used to identify individuals for inclusion in the study, to obtain data on demographic characteristics, deaths and emigration.
The NPR contains data on all inpatient contacts from public hospitals in Denmark since 1977, with outpatient and emergency department contacts included from 1995 and private hospitals from 2003.21 The NPR was used to retrieve diagnostic codes (International Classification of Diseases, 10th Revision (ICD-10)) for hospital contacts from 2003 to 2018, specifically identifying those representing osteoporotic fractures and selected comorbidity variables (both Charlson Comorbidity Index (CCI) and mediators) during the lookback and study periods (figure 1). In addition to ICD-10 codes, the NPR was used to retrieve the corresponding admission dates to define the date of occurrence for all fractures included in the study as well as admission dates for the aforementioned comorbidities.
Diagram of study period, study phases and data sources. MOF, major osteoporotic fracture.
The study period started on 1 January 2013 with inclusion of data on fracture patients from the NPR until 31 December 2018. The risk of mortality was followed up until 31 December 2018 (figure 1).
Osteoporotic fractures population
The osteoporotic fracture patients included all Danish citizens sustaining a major osteoporotic fracture (MOF) between 2013 and 2018, aged ≥60 years on the date of their first fracture in the inclusion period (defined as index fracture and index date). For results to be informative on the scenario of an incident osteoporotic fracture, patients were excluded from the study population if they had sustained an MOF within 3 years prior to the index fracture. While this exclusion period is largely arbitrary, it was chosen because it matched the period used by Chen et al.8 Other scenarios were explored as sensitivity analyses.
Fracture events
MOF was defined as a hip, clinical vertebral, humerus or wrist fracture, identified as a primary or secondary diagnosis code in the NPR. ICD-10 coding for MOFs followed the ICD-10 coding applied in the Fracture Risk Evaluation Model.22 23 Included ICD-10 codes were: S720, S721, S722 (hip), S320, T08, S220, S221, S120, S121, S122 (clinical vertebral), S525, S526 (wrist), S422 and S423 (humerus). For a hip fracture to be included in the study, both an ICD-10 code for a hip fracture and a relevant surgical code (KNFB* or KNFJ4-9) must have been registered a maximum of 7 days apart. Surgical codes for defining hip fractures were chosen as suggested by Hjelholt et al.24
Subsequent fracture
Subsequent fractures followed the same definition as MOFs and were recorded as the first fracture after index fracture. For subsequent fractures at the same skeletal site as the index fracture, a grace period of 6 months was inserted between the admission date of the index fracture and of the admission date of the subsequent fracture, for the fracture to count as subsequent. This was done to ensure that the two fractures were separate incidents, as a fracture registration in the NPR<6 months of another fracture is likely associated with the first fracture, for example, as a check-up consultation. The same interval has been applied previously by Driessen et al.25
For subsequent fractures at a different skeletal site than the index fracture, it was restricted that the registered admission date had to occur after the index date.
Controls
Fracture patients were matched 1:10 by sex and year of birth with controls extracted from the CRS on the index date. Controls were assigned the index date of their matched fracture patient. Controls had to be alive at the index date of their respective matched fracture patient. Controls were sampled with replacement and changed status to cases if they sustained a fracture during the study period.
Outcome and definition of study parameters
The main outcome was mortality after the index date identified during the period 1 January 2013 to 31 December 2018 in the CRS. As a secondary outcome, we assessed mortality after cases experienced any subsequent fractures during the period 1 January 2013 to 31 December 2018.
Covariates
Age was calculated on index date and categorised into 5 year age groups (60–64, 65–69, 70–74, 75–79 and ≥80 years). The CCI was used to classify comorbid conditions in the study population recorded within the 10-year period prior to the index date (figure 1). The 17 comorbidities in the CCI were ICD-10-coded as suggested by Quan et al.26 Further, each CCI comorbidity was defined as a separate dichotomous variable as suggested by Möller et al.27
Mediators (postfracture comorbidity)
Possible mediators were identified among all comorbidities registered in the 6 months following the index date using primary or secondary ICD-10 codes from the NPR (figure 1). ICD-10 codes detailing comorbidities which were unlikely to occur as a result of fracture (eg, H295 Unspecified age-related cataracts) were excluded and similar ICD-10 codes were combined into categories (eg, I500, I501 and I509 categorised as ‘Heart failure’) following discussion among the authors. Comorbidities which occurred in <0.4% of the fracture population (either as single ICD-codes or categories) were excluded, and the remaining comorbidities were rated by 10 clinicians on a 5-point scale according to how likely the condition was to occur as a result of fracture. Comorbidities which were scored as either ‘somewhat likely’ or ‘very likely’ by at least 70% of respondents were included in the mediation analyses.
Statistical analyses
Baseline characteristics of the study population were presented as frequencies for categorical variables. Continuous variables were summarised as medians with quartiles.
The HRs for death were estimated by performing crude and adjusted Cox regression analyses with 95% CIs. We adjusted analyses for age and separate dichotomous CCI comorbidities.27 All analyses were stratified by sex. Survival time was calculated from index date to time of censoring or death. Controls who during follow-up met the MOF criteria were included as a case on the date of their index fracture. The study population was censored on emigration or at the end of the study period, 31 December 2018. Survival time was split into appropriate intervals to accommodate the non-proportional hazards. In the analyses of subsequent fracture and mortality, we included subsequent fracture and age as time-dependent variables, changing the values of the variables at the date of the subsequent fracture.
According to a counterfactual approach, mediation analyses were performed as suggested by VanderWeele.28 The HRs adjusted for potential confounders correspond to the total effect of fracture on the risk of death, which can be decomposed into a controlled direct effect (CDE) and the mediated effect. The CDE captures the influence of fracture on the risk of death, if removing the effect of the possible mediator. The CDE was obtained by adjustment for both potential confounders and the mediating factors—the latter were treated as time-dependent variables. Additionally, the proportion eliminated (PE) was estimated.28 PE describes the proportion of effect which might be removed by eliminating the mediator from the pathway. The 95% CIs for CDE and PE were estimated by bootstrapping using 100 replicates. Mediators resulting in a PE≥15% were reported in the main results. Full mediation analysis was reported in online supplemental materials. Individuals with pre-existing conditions of the mediator in question were excluded.
Supplemental material
Four sensitivity analyses were performed. The first sensitivity analysis excluded patients from the fracture population if they sustained an MOF in a 10-year period prior to the date of index fracture, compared with the 3-year exclusion period applied in the main analysis. The second sensitivity analysis applied a 3-month grace period for subsequent fractures within the same skeletal site as the index fracture, compared with the 6 months applied in the main analysis. The third sensitivity analysis applied a fracture hierarchy for patients who sustained multiple fractures on the index date. The hierarchy ranked fractures according to severity as follows: hip>clinical vertebral>humerus>wrist. The most severe fracture (according to the hierarchy) among the multiple fractures sustained on the index date accounted for index and subsequent fractures in analyses. The fourth sensitivity analysis stratified the population based on CCI score=0 or ≥1.
Data management and data analyses were conducted using Stata V.1729 through a remote VPN access to The Danish Health Data Authority with analysts blinded to the personal identities of the study subjects.
Results
MOFs occurred in 1 77 639 women and 58 692 men aged ≥60 years at time of fracture between 1 January 2013 and 31 December 2018. Of these, 99 502 women and 30 526 men had been diagnosed with an MOF within 3 years prior to the index date and were excluded, resulting in 78 137 women and 28 166 men included in the fracture population. Following 1:10 matching, 781 359 women and 281 629 men were included as controls for a total study population of 1 169 291 persons (figure 2). For a few cases, it was not possible to identify 10 controls. These were matched with the maximum available number of controls.
Flow chart of inclusion and exclusion of study population. MOF, major osteoporotic fracture.
The fracture population included 106 303 individuals sustaining an MOF between 2013 and 2018, of which 73.5% were women and the median age at time of index fracture was 76.3 years (68.9; 84.5) (table 1). The most common fracture types were wrist and hip at 39 218 and 35 092 individuals, respectively, followed by 21 714 individuals sustaining a humerus fracture and 11 974 individuals sustaining a clinical vertebral fracture (table 1).
Index fractures, number and age of fracture population studied according to fracture sites (N=106 303)
Table 2 shows the baseline characteristics of the study population stratified by sex and fracture status. The fracture control populations were overall comparable on age group distributions in both sexes. Distributions varied to a greater extent in terms of CCI comorbidities, with the fracture population displaying a higher occurrence of all comorbidities (table 2).
Study population characteristics stratified by sex and case status
Table 3 shows risk of death following index fracture with time split into intervals from 1 month to ≥5 years after fracture. Overall, mortality was highest in the first month immediately following fracture before steadily declining over time. Men displayed higher HRs across fracture types compared with women. In both sexes, the highest mortality was observed in the first month following a hip fracture (HR 10.98 (95% CI 10.23 to 11.79) in women and HR 16.40 (95% CI 15.00 to 17.93) in men) (table 3).
Frequency of deaths and Hazard Ratios (HR) with 95% Confidence Intervals (CI) for each fracture site stratified by time and sex
Increased HRs were observed among individuals sustaining a subsequent fracture, compared with those only sustaining an index fracture (figure 3). Among individuals with any MOF as index fracture, the highest mortality was observed in those sustaining a subsequent hip fracture (HR 1.86 (95% CI 1.73 to 2.01) in women and HR 1.97 (95% CI 1.73 to 2.26) in men). Subsequent hip fractures resulted in the highest HRs for death across all index fracture types in both sexes. In women, the highest HR was observed in patients with an index wrist fracture and a subsequent hip fracture (HR 2.43 (95% CI 2.12 to 2.78)). In men, the highest HR was observed in patients with an index humerus fracture and a subsequent hip fracture (HR 2.69 (95% CI 2.04 to 3.54)) (figure 3). Similarly to the tendency observed following index fracture, the highest mortality was observed in the months immediately following a subsequent fracture (online supplemental table S1).
Forest plot of HR with 95% CI for mortality following subsequent fractures, by each initial fracture site and stratified by sex. HRs were adjusted for age and CCI comorbidities. Estimates are presented with 95% CI. CCI, Charlson Comorbidity Index; MOF, major osteoporotic fracture; CVF, clinical vertebral fracture.
Mediation analysis
After removing duplicates, 242 different ICD-codes were registered in the fracture population within 6 months of the index date. A total of 155 ICD-codes were excluded, as these were assessed to not occur as consequences of fracture, and thus not explain the association between MOF and mortality. The remaining 87 ICD-10 codes were combined into 41 comorbidity categories. Based on 10 clinician ratings, 13 of these comorbidity categories were included in the mediation analyses (online supplemental table S2).
Overall, pneumonia was the most significant mediator across fracture types (table 4). The analysis showed that the proportion of excess mortality following MOF which could be explained by pneumonia was 21.85% (95% CI 19.95% to 23.74%) for women and 29.15% (95% CI 26.97% to 31.32%) for men. This trend occurred across all fracture types. In clinical vertebral fractures, dehydration also mediated a substantial part of the association with death (PE 15.02% (95% CI 10.24% to 19.80%) in women and PE 10.17% (95% CI 5.43% to 14.92%) in men). In hip fractures, the second largest proportion of the association was mediated by urinary tract infections (UTIs) in women (PE 18.40% (95% CI 16.83% to 19.98%)) and sepsis in men (PE 15.36% (95% CI 13.31% to 17.40%)). In humerus fractures, sepsis mediated the second largest proportion in both sexes (PE 17.36% (95% CI 9.30% to 25.43%) in women and PE 15.59% (95% CI 8.03% to 23.15%) in men). In wrist fractures, the association in women was, besides pneumonia, also mediated by UTIs, dehydration and fall tendency (PE 18.53% (95% CI 6.75% to 30.32%), PE 16.86% (95% CI 3.37% to 30.35%) and PE 15.35% (95% CI 7.70% to 23.00%), respectively). For wrist fractures in men, estimates were imprecise due to wide and insignificant CIs and low case numbers (table 4). We will, therefore, not conclude on these results. The full mediation analysis is presented in online supplemental table S3.
The role of select mediators* on the association between fractures and mortality, for each fracture type stratified by sex
Sensitivity analysis
Analyses applying a 10-year exclusion period for MOFs prior to the index date displayed the same trends as the main analyses, however, with slightly inflated estimates (online supplemental tables S4–S7B file 1). Analyses using a 3-month grace period between index and subsequent fractures at the same site yielded results similar to the main analysis (data not shown). Analyses using the fracture hierarchy hip>clinical vertebral>humerus>wrist to categorise the index fracture also produced similar results compared with the main analyses (data not shown). Analysis stratifying the population on CCI score proved an excess mortality trend following fracture, with inflated estimates for CCI=0 compared with the main analysis, whereas estimates for CCI≥1 decreased compared with the main analysis (online supplemental tables S8–9).
Discussion
Main findings
The results of this nationwide register-based cohort study showed that the highest relative mortality risk was observed within the first month following an osteoporotic hip fracture in both men and women. In patients sustaining a subsequent osteoporotic fracture, the highest relative mortality risk was also observed in the first month following a subsequent hip fracture. Overall, male fracture patients displayed higher HRs than female fracture patients. In subsequent fracture patients, a subsequent hip fracture resulted in the highest HRs in both sexes, although the index fracture types differed. Pneumonia proved to be the most significant mediator, contributing to the largest proportions of excess mortality following an osteoporotic fracture, but other factors also showed important mediating effects.
Excess mortality following index and subsequent fractures
In this study, the highest mortality risk following an index fracture was observed in the first month after fracture. Overall, male fracture patients displayed higher relative mortality risk than female fracture patients. In both sexes, the highest mortality risk was observed within the first month following a hip fracture (table 3). Our findings are in accordance with previous studies, and increased short-term mortality is well established in hip fractures. A meta-analysis by Haentjens et al indicated a relative hazard of 7.95 (95% CI 6.13 to 10.30) for men and 5.75 (95% CI 4.6 to 6.69) for women in the first 3 months following hip fracture, compared with controls,30 and multiple other studies have reported high mortality immediately following fracture and decreasing over time.1 31–33 Our results indicate that the excess mortality decreases in the years following index fracture for all fracture types, but continues to be elevated compared with the control population for the duration of the study period (table 3). Existing literature indicates that excess mortality may persist for many years after the fracture occurs.1 34 However, other studies have reported that excess mortality does not persist after the first 6 months following fracture, as a large proportion of it may be explained by greater frailty or advanced age in the fracture population.35 36
In patients sustaining a subsequent osteoporotic fracture, the highest mortality risk was observed in the first 6 months after fracture occurrence in our study (online supplemental table S1). In general, knowledge of immediate excess mortality following subsequent fracture (within the first 6 months postfracture) is not well established. Increased long-term mortality has been observed in refracture patients in multiple studies.37–39
Fracture type combinations
In the analysis of different index and subsequent fracture type combinations, subsequent hip fracture had the highest HRs regardless of index fracture type (figure 3). Mortality risk was highest for men who sustained an index humerus fracture (HR 2.69 (95% CI 2.04 to 3.54)) and for women who sustained an index wrist fracture (HR 2.34 (95% CI 2.12 to 2.78)) prior to their subsequent hip fracture. This finding is consistent with that of Chen et al, who observed an increased mortality risk for both the combination of index humerus/subsequent hip fractures (HR 1.66 (95% CI 1.41 to 1.95)) and index wrist/subsequent hip fractures (HR 2.65 (95% CI 2.29 to 3.08)), with the wrist/hip combination resulting in the highest HR for all fracture types.8 Chen et al also reported that subsequent hip fracture produced the highest HRs in combination with nearly all index fracture types,8 which is in accordance with our findings. A possible explanation for the high HR resulting from the index wrist/subsequent hip fracture combination might be that wrist fractures alone result in a relatively limited excess mortality.40 This may cause the excess mortality of a subsequent hip fracture to appear more pronounced when compared with those who only sustain a single wrist fracture. Additionally, subsequent wrist fractures produced the lowest HRs for mortality across multiple index fracture types for both sexes in our study (figure 3), and similar findings have been reported in other studies.8 39
Mediation of mortality risk
Prior studies have demonstrated that comorbidities significantly impact the excess mortality associated with fractures,11–18 but studies examining perifracture or postfracture comorbidities have been mostly focused on hip fractures.12 14 18 Of all the comorbidities included in this study as possible mediators, pneumonia proved to be the most significant mediator across all fracture types in both sexes (table 4). The proportion of mortality risk following MOF which hypothetically could be eliminated by preventing pneumonia was 21.85% (95% CI 19.95% to 23.74%) for women and 29.15% (95% CI 26.97% to 31.32%) for men, and some of the highest estimates were observed following hip fractures in both sexes (HR 28.46% (95% CI 26.02% to 30.91%) in women and HR 37.81% (95% CI 34.79% to 40.84%) in men) (table 4). In studies conducted by Whitney et al on the effect of incident respiratory disease following fractures in different study populations, pneumonia was shown to mediate 11.3%41 and 15%42 of the association between fractures and mortality in adults ≥65 years. Jang et al and Lv et al demonstrated an increased risk of mortality in hip fracture patients with postoperative pneumonia.18 43 While the studies by Whitney et al41 42 included study populations that were different from ours, they support that incident postfracture pneumonia mediates a significant proportion of the excessive mortality in fracture patients. A recent systematic review and meta-analysis by Gao et al showed that, while factors such as age, pre-existing comorbidities, length of hospital stay, surgical delay and dependent function status has been linked to an increased risk of postoperative pneumonia in existing literature, fracture type has not (OR 0.96 (95% CI 0.69 to 1.35).44 Targeted preventive measures may help mitigate the risk of pneumonia and, by extension, excess mortality following fracture. This should be considered in postfracture management.
Sensitivity analyses
Several sensitivity analyses were conducted in the study. Some resulted in minimal changes to the risk estimates, while applying a 10-year exclusion period for MOFs prior to index fracture resulted in slightly increased risk estimated that followed the same trends as the main analysis (online supplemental table S4–S7B). The last sensitivity analysis, in which the study population was stratified as either CCI score=0 or CCI score≥1, resulted in inflated estimates for the group with no CCI comorbidities, especially among men sustaining a hip fracture (online supplemental table S8), and decreased for the group with at least one CCI comorbidity prior to index fracture (online supplemental table S9). While this pattern is not unexpected, the authors do not have a definitive explanation for the magnitude of the increase in HR observed in patients with no comorbidity registered prior to index fracture, and advise caution when interpreting these results. Overall, the results display the same trends as those observed in the main analysis.
Strengths and limitations
This study had several strengths adding value to results and conclusions. First, the large sample size (n=1 169 291) added statistical strength to results. Second, the study sample was retrieved from nationwide Danish registers that cover all citizens in Denmark regardless of gender and socioeconomic status, thus adding to the generalisability of results and minimising the risk of selection bias. Third, this study performed mediation analyses examining the mediating role of postfracture comorbidities in the association between osteoporotic fractures and excess mortality. To the best of our knowledge, no studies have performed this analysis in a larger study population, thus adding novelty to results. Lastly, this study investigated excess mortality for MOFs which includes four fracture sites (fracture of hip, clinical vertebrae, humerus or wrist) which adds to the knowledge base of the association between osteoporotic fractures and excess mortality, which has previously been examined primarily in hip fractures.1
Conversely, this study also had methodological limitations. First, the use of Danish register data eliminates the possibility to obtain information on potentially relevant confounders and covariates. These could for instance include body mass index or lifestyle factors such as level of physical activity, alcohol consumption, smoking habits, etc. The chance of unmeasured confounding should, therefore, be considered in the interpretation of results. However, our study did account for the effect of sex and age on mortality risk by stratifying analyses on sex and by adjusting for age in regression analyses. Second, the validity is unknown in the NPR of the ICD-10 coding of comorbidity variables used for mediation analyses as well as ICD-10 coding used in the outcome definition of MOF, which leaves risk of misclassification. However, the data quality of the NPR is overall estimated to be high and the register is acknowledged as a valuable data source for epidemiological research.21 Lastly, it should be acknowledged that the underlying mechanisms of MOF cannot be asserted using the ICD-10 codes available in this study. This could possibly lead to the inclusion of fractures that are non-osteoporotic by nature. However, MOF is used as the outcome measure in this study as it is a well established concept which stems from the Fracture Risk Assessment Tool where it used as the endpoint of prediction,45 and it continues to be used in research definitions of osteoporotic fractures.46–48
Conclusion
In conclusion, this study showed that mortality risk is highest in the months immediately after an osteoporotic fracture occurs, especially following hip fracture. This tendency was observed following both index and subsequent fractures in both sexes. The combination of index and subsequent fracture types was found to have a significant impact on the excess mortality, with subsequent hip fracture resulting in the highest HRs across all index fracture types in both sexes. In mediation analyses, pneumonia proved to be the most significant mediator across all fracture types and in both sexes. These findings could inform the development of clinical guidelines for the care of future osteoporotic fracture patients at risk of excess mortality.
Data availability statement
Data may be obtained from a third party and are not publicly available. Data are accessible through the Danish Health Data Authority.
Ethics statements
Patient consent for publication
Ethics approval
An ethics committee approval is not required for register-based studies according to Danish legislation. The study was approved by the Danish Data Protection Agency (jf.nr. 2008-58-0035) and The Danish Health Data Authority.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors ERC, AC and KHR were responsible for the initial study concept. TGP was responsible for the statistical analysis in collaboration with SM. All authors contributed to the design and choice of methods. MKS and BA advised on clinically relevant aspects of the study. ERC and AC drafted the manuscript and conducted the initial interpretation of the data alongside KHR and TGP. All authors critically reviewed the manuscript multiple times for important intellectual content and have approved the final version for publication. ERC and AC contributed equally to this paper. KHR is guarantor and accepts final responsibility for the finished work, and controlled the decision to publish.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests ERC, AC, TGP, SM and KHR declare no competing interests. BA reports institutional research grants and personal fees from UCB, institutional research grants and personal fees from Kyowa-Kirin, institutional research grants and personal fees from Pharmacosmos, speakers fees from Amgen and speakers fees from Gedeon-Richter. MKS reports an educational grant from UCB, institutional research grants from UCB/Amgen and Region Zealand Health Scientific Research Foundation and a personal speakers fee from UCB Nordic.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.