Nationwide in-hospital mortality following major fractures among hemodialysis patients and the general population: An observational cohort study

Journal Pre-proof Nationwide In-Hospital Mortality Following Major Fractures among Hemodialysis Patients and the General Population: An Observational Cohort Study Shintaro Mandai, Hidehiko Sato, Soichiro Iimori, Shotaro Naito, Haruna Tanaka, Fumiaki Ando, Koichiro Susa, Kiyoshi Isobe, Takayasu Mori, Naohiro Nomura, Eisei Sohara, Tomokazu Okado, Shinichi Uchida, Kiyohide Fushimi, Tatemitsu Rai

PII:

S8756-3282(19)30415-6

DOI:

https://doi.org/10.1016/j.bone.2019.115122

Reference:

BON 115122

To appear in: Received Date:

8 August 2019

Revised Date:

23 October 2019

Accepted Date:

24 October 2019

Please cite this article as: Mandai S, Sato H, Iimori S, Naito S, Tanaka H, Ando F, Susa K, Isobe K, Mori T, Nomura N, Sohara E, Okado T, Uchida S, Fushimi K, Rai T, Nationwide In-Hospital Mortality Following Major Fractures among Hemodialysis Patients and the General Population: An Observational Cohort Study, Bone (2019), doi: https://doi.org/10.1016/j.bone.2019.115122

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier.

Original Article

Nationwide In-Hospital Mortality Following Major Fractures among Hemodialysis Patients and the General Population: An Observational Cohort Study

of

Shintaro Mandai1, M.D., Ph.D., Hidehiko Sato1, M.D., Soichiro Iimori1, M.D., Ph.D., Shotaro Naito1, M.D., Ph.D., Haruna Tanaka1, M.D., Fumiaki Ando1, M.D., Ph.D.,

ro

Koichiro Susa1, M.D., Ph.D., Kiyoshi Isobe1, M.D., Ph.D., Takayasu Mori1, M.D.,

Ph.D., Naohiro Nomura1, M.D., Ph.D., Eisei Sohara1, M.D., Ph.D., Tomokazu Okado1,

-p

M.D., Ph.D., Shinichi Uchida1, M.D., Ph.D., Kiyohide Fushimi2, M.D., Ph.D., and

Department of Nephrology, Graduate School of Medical and Dental Sciences, Tokyo

lP

1

re

Tatemitsu Rai1, M.D., Ph.D.

Medical and Dental University, 1-5-45 Yushima, Bunkyo, Tokyo 113-8519, Japan Department of Health Policy and Informatics, Graduate School of Medical and Dental

ur na

2

Sciences, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo, Tokyo 113-8519, Japan

Jo

Corresponding Author: Tatemitsu Rai Department of Nephrology, Graduate School of Medical and Dental Sciences, Tokyo Medical and Dental University 1-5-45 Yushima, Bunkyo, Tokyo 113-8519, Japan Tel: +81-3-5803-5214; Fax: +81-3-5803-5215; E-mail: [email protected]

1

Abstract word count: 250 Text word count: 3258

Highlights 

This large retrospective population-based study investigated overall and

of

site-specific mortality after fractures in hemodialysis patients and the general

ESKD patients on hemodialysis experienced a 4.8-fold higher mortality rate

-p



ro

population.



re

after overall fractures than the general population.

Mortality after upper arm fracture was markedly high in hemodialysis patients,

Upper arm fracture was specifically associated with risk of vascular access

ur na



lP

oppositely to the lower risk in the general population.

failure in hemodialysis patients.



There is a need for clinical awareness of upper arm fracture as a fatal fracture in

Jo

ESKD patients on hemodialysis.

Abstract

Backgrounds: End-stage kidney disease (ESKD) is associated with increased risk of fracture and subsequent morbidity and mortality. However, fracture site-specific

2

mortality in ESKD patients have yet to be elucidated in comparison with the general population. Methods: In this population-based cohort derived from the Diagnosis Procedure Combination database of Japan from 2012 to 2014, we included 9,320 ESKD patients undergoing hemodialysis and 547,726 patients without ESKD who were hospitalized

of

for five major fractures, including hip (proximal femur), spine, forearm, upper arm, and leg (distal femur and proximal tibia). Overall and site-specific risks of in-hospital death

ro

were determined by logistic regression models.

Results: The age- and sex-adjusted mortality rates were 4.91% (95% confidence

-p

interval [CI], 4.46–5.37) and 1.02% (95% CI, 0.99–1.06) in the hemodialysis and

re

general population groups, respectively. The multivariate odds ratio (OR) of death in hemodialysis patients versus the general population was 2.48 (95% CI, 2.25–2.74) for

lP

overall fractures, and was particularly high for a subgroup of upper arm fracture (OR 4.82, 95% CI, 3.19–7.28). The site-specific odds of death (95% CI) among hip, spine,

ur na

forearm, upper arm, and leg (reference) fractures were 1.77 (0.98–3.18), 1.48 (0.79– 2.75), 0.19 (0.04–0.86), and 2.01 (1.01–4.01) in hemodialysis patients, and 1.28 (1.13– 1.45), 1.00 (0.88–1.14), 0.13 (0.10–0.17), and 0.83 (0.70–0.97) in the general population, respectively.

Jo

Conclusion: Hemodialysis patients experienced a 4.8-fold higher mortality rate after fractures than the general population. Mortality after upper arm fracture was specifically high in patients on hemodialysis, likely due to the involvement of vascular access located on the fractured arm.

Key words: fracture, end-stage kidney disease, hemodialysis, general population. 3

Introduction Osteoporosis progresses with advancing age, menopause, and declining kidney function [1,2]. Osteoporosis-associated fracture is an enormous healthcare problem given its association with functional disability and decreased life expectancy [3-5]. Fracture is

of

less likely to directly cause death, but rather results in a chain of critical comorbidities or events that lead to premature death [6-8]. Previous studies in the general population

risk, and showed increased risk at every fracture site [9,10].

ro

established the association between major sites of osteoporotic fractures and mortality

-p

Individuals with end-stage kidney disease (ESKD) are highly susceptible to

re

communicable and noncommunicable diseases, including atherosclerotic disease, glucose intolerance, sarcopenia, and metabolic bone disease [11-14]. These numerous

lP

comorbidities are responsible for unfavorable clinical outcomes despite considerable advancements in dialysis therapy over the past few decades.

ur na

In addition to loss of bone mineral density [2], bone fragility occurs in ESKD patients due to electrolyte and hormonal imbalances, malnutrition, and polypharmacy; these patients are also susceptible to falls due to muscle weakness [15-17]. A majority of the previous studies that compared the fracture risk between individuals with ESKD

Jo

and the general population focused on hip and spine fractures [13-16,18]. However, the association between fracture site and in-hospital mortality in ESKD patients has yet to be fully understood. We aimed to investigate this association in patients receiving maintenance hemodialysis with a large community-based study using a national inpatient database in Japan.

4

Materials and Methods Source of data The study population was obtained from the Diagnosis Procedure Combination (DPC) inpatient database in Japan for the years 2012–2014 [19,20]. More than 1000 hospitals, including all 82 teaching hospitals in the country, participate in the database, which

of

includes approximately half of all annual hospital admissions in Japan (nearly 7 million).

ro

There were 22,433,171 eligible hospital admissions from 2012–2014. Criteria

for entry into this study were age ≥18 years, hospitalization ≥24 h, and admission due to

-p

one of five major fractures as follows: hip (proximal femur), spine, forearm, upper arm,

re

and leg (distal femur and proximal tibia) [21,22] (Fig. 1). Participants who temporarily required or initiated renal replacement therapy, those who received peritoneal dialysis

lP

during hospitalization, and those with incomplete information on admission type were excluded. Individuals with ESKD undergoing maintenance hemodialysis were identified

ur na

based on coding of patient care procedures as follows: chronic maintenance hemodialysis with <4 hours per session, ≥4 hours and <5 hours per session, ≥5 hours per session, or chronic maintenance hemodiafiltration [19,20]. The ethics committee of the Tokyo Medical and Dental University approved

Jo

this study and waived the need for informed consent given the anonymity of the data. The study was performed in accordance with the ethical principles of the Declaration of Helsinki.

Patient characteristics As in previous studies, major fractures were defined and classified into five categories using International Classification of Diseases, 10th Revision (ICD-10) codes: hip 5

(proximal femur) (S72.0, S72.1, S72.2), spine (S12.0–S12.2, S12.7, S22.0, S22.1, S32.0–S32.2), forearm (S52), upper arm (S42.2–S42.4, S42.7), and leg (distal femur and proximal tibia) (S72.4, S82.1) [21,22]. Both traumatic and non-traumatic fractures were included. We did not include pathological fracture (M84.4) and multiple fractures (T02) in the analysis. We did not include abnormal healing of old fractures such as malunion

of

(M84.0), nonunion (M84.1), delayed union (M84.2), and sequelae of fractures (T91.1, T92.1, T93.1, T93.2). Other patient-related variables collected include age, sex, body

ro

mass index (BMI), dependence on hemodialysis, updated Charlson comorbidity index [23] excluding ‘renal disease’, admission type (emergent or elective), and year of

-p

admission.

re

Outcome

The primary outcome was occurrence of in-hospital death of any cause. Patients were

lP

followed until discharge, transfer, or death. Death within 30 days was also evaluated. Hospital length of stay was defined as the time between admission and either date of

ur na

discharge or date of death. Surgical treatment included open/arthroscopic surgery, and open/closed reduction and internal fixation in each type of fracture. Vascular access failure was defined as the requirement for any of the following interventions during hospitalization: construction or reconstruction of arteriovenous fistula or graft, insertion

Jo

of temporary or permanent dialysis catheters, or fistula/graft percutaneous transluminal angioplasty.

Data analyses Patient demographics and characteristics are presented as numbers and percentages or medians with interquartile range (IQR). Comparisons of baseline characteristics 6

between the groups were performed using t test for continuous variables and χ2 test for categorical variables. After fitting regression models, we calculated age- and sex-adjusted mortality by estimating predicted probabilities of death for each patient fixing age at a mean value of the whole cohort that is known as the marginal standardization form of predictive margins [24]. Logistic regression was used to

of

evaluate the relationship between hemodialysis dependence and overall or site-specific fracture mortality risk; we adjusted for potential confounding variables including age

ro

(≤64, 65–74, or ≥75 years), sex, BMI (≤18.4, 18.5–24.9, ≥25.0 kg/m2, or unknown),

-p

Charlson comorbidity index (0, 1-2, or ≥3), admission type (emergent or elective), and year of admission. We performed subgroup analyses and used generalized linear models

re

to assess interaction effects in logistic regressions. To validate robustness of the results obtained from logistic regressions, we performed probit and complementary log-log

lP

regressions. To clarify the relationship between fracture sites and risk of vascular access failure, logistic regression was used and adjusted for age, sex, BMI, Charlson

ur na

comorbidity index, admission type, and year. Multivariable linear regression models were used to determine the association of fracture sites with hospital length of stay with adjustment for age, sex, BMI, Charlson comorbidity index, admission type, and year. Statistical analyses were performed using Stata version 15.0 software (Stata Corp.,

Jo

College Station, TX, USA). P values < 0.05 were considered statistically significant.

Results

Patient characteristics Table 1 shows the demographics and characteristics of the study cohort, which included 9,320 (1.7%) ESKD patients receiving hemodialysis and 547,726 (98.3%) patients 7

without ESKD. Patients with ESKD on hemodialysis were slightly younger. The proportion of females was approximately 50% in the hemodialysis group and 74% in the general population group. The BMI was relatively low in both groups compared to that in the whole Japanese population [25]. Hemodialysis patients had more comorbidities, including diabetes mellitus and cardiovascular disease, resulting in a

of

higher mean Charlson comorbidity index. The leading fracture type among the hemodialysis and general population groups was hip fracture (65% and 49%,

ro

respectively), followed by spine fracture (19% and 22%, respectively). Forearm, upper

arm, and leg fractures were approximately 5% in the hemodialysis group, while forearm

-p

fracture was more common in the general population group.

re

Overall mortality after admission due to fractures

There were 468 and 8,501 all-cause in-hospital deaths in the hemodialysis and general

lP

population groups, respectively. Mortality rates of the two populations were 38.4 per 100 person-years (PY) confidence interval [CI], 35.1–42.0) and 18.0 (95% CI, 17.6–

ur na

18.3), respectively. The age- and sex-adjusted mortality rates of the two populations were 4.91% (95% confidence interval [CI], 4.46–5.37) and 1.02% (95% CI, 0.99–1.06), respectively, revealing a 4.8 times higher risk of death in the hemodialysis group (Figure 2). The mortality rate increased as patient age increased, especially in males,

Jo

and male patients aged ≥75 years were more likely to die. Hemodialysis patients were more likely to die, independent of age and sex. Notably, approximately ≥5% of the elderly in the hemodialysis group died during hospitalization after fractures. The median hospital length of stay was longer in hemodialysis patients than the general population (34 days [IQR, 22–56 days] versus 24 days [IQR, 13–40 days]), varying across the types of fractures (Fig.S1). In a multivariable linear regression 8

analysis, forearm fracture was associated with the shortest length of stay, while leg fracture was associated with the longest stay both in the hemodialysis and general population groups (Table S1). The age- and sex-adjusted mortality rates within 30 days after admission were 1.67% (95% CI, 1.40-1.94) and 0.43% (95% CI, 0.41-0.46) in hemodialysis patients and the general population, respectively. These rates were about a

of

half or less than in-hospital morality rates (Fig.2B), suggesting that fracture patients were more likely to die later than 30 days after admission independent of ESKD.

ro

In the logistic regression analysis, hemodialysis-dependent ESKD was

associated with a 2.48-fold higher odds of death versus the general population, after

-p

adjusting for major confounders (Table 2). Older age, male sex, lower BMI, higher

re

Charlson comorbidity index score, and emergent admission were associated with a greater risk of death. Hip fracture was associated with the highest mortality risk in the

lP

entire cohort.

To clarify the effects of hemodialysis dependence on mortality after fractures,

ur na

we performed a subgroup analysis. As shown in Figure 3, the interaction between hemodialysis dependence and several subgroup variables was significant, and the effect of hemodialysis dependence was greater in people aged <75 years, females, and patients with fewer comorbidities. With respect to fracture sites, upper arm fracture was

Jo

specifically associated with a markedly greater impact of hemodialysis dependence on mortality.

Site-specific fracture mortality risk after fractures Mortality rates among hip, spine, forearm, upper arm, and leg fractures were 42.1, 35.3, 7.3, 42.4, and 15.8 per 100 PY in hemodialysis patients, and 22.0, 15.9, 2.5, 12.4, and 7.9 per 100 PY in the general population, respectively. Hip and spine fractures were 9

associated with a greater risk of death compared with leg fracture in the univariate analyses of both hemodialysis and general population patients (Table 3). The greater risk of death for hip fracture remained significant after adjustment for relevant covariates. In both the hemodialysis and general population groups, forearm fracture was associated with very low risk of death before and after adjustment for confounders.

of

Upper arm fracture also showed a lower mortality than leg fracture in the general population. However, upper arm fracture was contrarily associated with a two-fold

ro

higher odds of death after adjustment for confounders in hemodialysis patients, showing the greatest risk among fracture sites. To assess robustness of these findings, we also

-p

performed probit and complementary log-log regressions, yielding similar results

re

(Supplementary Table S2 and S3). We also performed the analysis including only the initial admission cases for individual patients (n = 517,725), excluding the readmission

lP

cases of the second or later fractures (n = 39,321). The results showed the similar relationship between the fracture sites and outcome (Table S4). The association between

ur na

fracture sites and 30-day mortality was not statistically significant in the hemodialysis group (Table S5) presumably due to the smaller number of outcomes, although each effect size was very similar with that in in-hospital mortality (Table 3). Subgroup analyses of fracture patients who received surgical or conservative therapy

Jo

To verify if requirement of an operation modifies the association between fracture sites and mortality, we analyzed the outcome in subgroups of patient who received surgical or non-surgical treatment after the respective types of fracture. The prevalence of patients who received surgical treatment after hip, forearm, upper arm, and leg fractures were 47, 69, 58, and 54% in ESKD, and 55, 71, 70, and 54% in the general population, respectively. As shown in Table S6, surgical treatment 10

was associated with lower mortality independent of fracture sites in the both groups. As shown in Table S6, the association of fracture sites and mortality was very similar to the results seen in Table 3, independent of surgical or conservative therapy. The risk of death after upper arm fracture was particularly high in hemodialysis patients who did not receive surgical treatment (Table S6).

of

Fracture sites and risk for vascular access failure To further estimate the underlying factors that contribute to mortality risk for upper arm

ro

fracture among hemodialysis patients receiving hemodialysis, we compared the

association between fracture sites and risk of vascular access failure using the logistic

-p

regression model. As shown in Table 4, upper arm fracture was selectively associated

re

with >2-times higher odds of vascular access failure both in univariate and multivariate analyses. Moreover, similar results were obtained with probit and complementary

ur na

Discussion

lP

log-log regressions (Supplementary Table S7 and S8).

This study is a large epidemiologic cohort of hospital admissions due to various types of fractures based on the Japanese DPC database. The primary finding of this study was that patients with ESKD on hemodialysis are at considerably higher risk of death after

Jo

admission for all types of major fractures compared with the general population. Secondarily, the effect of dialysis-dependence on mortality after fractures was particularly prominent for upper arm fracture. The risk of death after upper arm fracture was relatively low in the general population. However, this facture site was associated to the greatest risk of death in ESKD patients on hemodialysis, even after adjusting for potential confounders. 11

Mortality after fracture is usually higher in men and strongly associated with increasing age [26]. This was also seen among both individuals with and without ESKD in our study. In this study, 1.7% of the study participants were hemodialysis-dependent. This prevalence among the entire population-based cohort was consistent with a 4– 5-fold higher risk of fractures for EKSD patients compared with the general population

of

shown in previous studies [13,14], given that approximately 0.3% of the population is dialysis-dependent in Japan [27]. We found that the age- and sex-adjusted mortality

ro

ratio after overall fractures was 4.8 times higher in hemodialysis patients compared to the general population. In the previous long-term observational study of Japanese

-p

dialysis patients, the age-standardized mortality ratio for all-cause mortality was

re

reported to be 4.6 compared to the general population, suggesting the higher background mortality in this population [28]. However, in the studies specified for

lP

hospitalized dialysis patients, the in-hospital mortality ratios were shown to be 1.8 and 1.2 in dialysis patients admitted to intensive care units or admitted after stroke,

ur na

respectively [29.30]. Thus, the impact of dialysis dependence on risk of death may be substantially high in all fracture sites. Fractures of the upper extremities are conventionally recognized as less severe,

having a lower risk of functional morbidity and mortality than spine and lower

Jo

extremity fractures [10,31]. In this study, we focused on the fact that mortality data is lacking with respect to fractures other than in the hip and spine in ESKD [14,15,20,21], despite the well-known bone fragility in this population. Moreover, we differentiated forearm from upper arm fracture in the analyses, given the more severe and longer duration of symptoms and higher mortality following upper arm fracture [10,32]. As a result, we found that upper arm fracture is associated to a markedly higher mortality risk 12

in hemodialysis patients than in the general population and should be considered potentially fatal, similar to hip or spine fracture. The mechanism underlying the increase in mortality risk after upper arm fracture in hemodialysis patients remains unknown; however, a possible explanation would be vascular access. In Japan, 97% of ESKD patients are currently receiving

of

hemodialysis, and 99% are catheter-independent, so vascular patency is essential; most use a native arteriovenous fistula (89.7%), synthetic arteriovenous graft (7.1%),

ro

superficialization (1.8%), or direct arterial puncture (0.1%) [33]. Patients who fracture the limb with vascular access are more likely to be hospitalized. When upper arm

-p

fracture occurs in hemodialysis patients who have vascular access in their upper limb,

re

there is a 50% probability that the patients will have the fracture on the same limb with vascular access. Thus, ≥50% of the patients with upper arm fracture are likely to have

lP

the fracture on the limb with vascular access, although our database lacks the information concerning the side of the fracture and vascular access in the upper limb.

ur na

The mortality risk after upper arm fracture was particularly high in hemodialysis patients who did not receive an operation. Upper arm fracture in the limb with vascular access may prohibit appropriate surgical treatment before occurrence of subsequent comorbidities or premature death. Similarly, hip or leg fracture on the same side of a

Jo

synthetic graft in a lower limb presumably affects the patient’s outcome. However, lack of information about the location of the vascular access did not enable the analysis and validation of this speculation. A strength of our study is the use of a well-validated, large-scale, and nationally representative sample of the general population of Japan [19,20]. The validity of diagnoses and procedure records in the Japanese DPC database has been well verified 13

[34]. However, there are some limitations as well. First, our study lacked laboratory, bone mineral density, dialysis vintage, and mortality cause data. This, as well as its observational design, does not allow us to establish causality. Second, we were unable to analyze long-term follow-up outcomes after discharge because of the nature of the inpatient datasets. Thus, upper arm fracture does not necessarily imply long-term

of

unfavorable outcomes in hemodialysis patients; further investigation in a longitudinal study is needed. Third, this study focused on admission cases, and the number of

ro

fractures is under-represented. There is possibly selection bias for patients with

relatively severe fracture who require treatment under hospitalization. Thus, their

-p

outcomes may be different from those in outpatient cases. Finally, our findings may not

re

be generalizable to hemodialysis patients in other countries because of different prevalence rates of vascular access modalities [35]. The population of peritoneal

lP

dialysis patients unrelated to vascular access in our country is quite small. There were 10 overall deaths among 175 patients (47.0 per 100 PY) during 2012-2014 in this

ur na

database. Each number for hip, spine, forearm, upper arm, and leg fractures was only 6, 3, 0, 1, and 0, respectively. Further investigation of the association between upper arm fracture and mortality in a large cohort of these patients is needed. In conclusion, this study highlights the prognostic impact of hemodialysis

Jo

dependence on mortality after numerous types of fracture and revealed that individuals on maintenance hemodialysis had a 4.8 times higher in-hospital mortality than the general population. Upper arm fracture particularly resulted in an unfavorable outcome in hemodialysis patients, but not in the general population.

Declaration of Competing Interest 14

We declare no competing interests.

Acknowledgments We would like to thank all study participants. We thank Dr. Ryohei Takada (Department of Department of Orthopaedic Surgery, Tokyo Medical and Dental

of

University) for his helpful discussions. This study is supported, in part, by Grants-in-Aid for Research on Policy Planning and Evaluation from the Ministry of

Jo

ur na

lP

re

-p

ro

Health, Labour and Welfare, Japan (H28-Seisaku-Shitei-009, H29-Seisaku-Shitei-009).

15

References 1. Riggs, B.L., Wahner, H.W., Dunn, W.L., et al. Differential changes in bone mineral density of the appendicular and axial skeleton with aging: relationship to spinal osteoporosis. J Clin Invest. 67 (1981) 328–335. 2. Rix, M., Andreassen, H., Eskildsen, P., et al. Bone mineral density and biochemical

of

markers of bone turnover in patients with predialysis chronic renal failure. Kidney Int. 56 (1999) 1084–1093.

ro

3. Johnell, O., Kanis, J.A. An estimate of the worldwide prevalence and disability associated with osteoporotic fractures. Osteoporos Int. 17 (2006) 1726–1733.

-p

4. Melton, L.J., 3rd. Adverse outcomes of osteoporotic fractures in the general

re

population. J Bone Miner Res. 18 (2003) 1139–1141.

5. Braithwaite, R.S., Col, N.F., Wong, J.B. Estimating hip fracture morbidity, mortality

lP

and costs. J Am Geriatr Soc. 51 (2003) 364–370.

6. Kannegaard, P.N., van der Mark, S., Eiken, P., et al. Excess mortality in men

ur na

compared with women following a hip fracture. National analysis of comedications, comorbidity and survival. Age Ageing. 39 (2010) 203–209. 7. Panula, J., Pihlajamäki, H., Mattila, V.M., et al. Mortality and cause of death in hip fracture patients aged 65 or older: a population-based study. BMC Musculoskelet

Jo

Disord. 12 (2011) 105.

8. Kanis, J.A., Oden, A., Johnell, O., et al. The components of excess mortality after hip fracture. Bone. 32 (2003) 468–473 9. Center, J.R., Nguyen, T.V., Schneider, D., et al. Mortality after all major types of osteoporotic fracture in men and women: an observational study. Lancet. 353 (1999) 878–882. 16

10. Johnell, O., Kanis, J.A., Odén, A., et al. Mortality after osteoporotic fractures. Osteoporos Int. 15 (2004) 38–42. 11. Go, A.S., Chertow, G.M., Fan, D., et al. Chronic kidney disease and the risks of death, cardiovascular events, and hospitalization. N Engl J Med. 351 (2004) 1296–1305. 12. Sarnak, M.J., Jaber, B.L. Mortality caused by sepsis in patients with end-stage renal

of

disease compared with the general population. Kidney Int. 58 (2000) 1758–1764.

patients, 1993-2010. Am J Kidney Dis. 62 (2013) 747–754.

ro

13. Arneson, T.J., Li, S., Liu, J., et al. Trends in hip fracture rates in US hemodialysis

14. Delgado, C., Frassetto, L. Hip fractures in hemodialysis patients. Am J Kidney Dis.

-p

62 (2013) 653–655.

re

15. Danese, M.D., Kim, J., Doan, Q.V., et al. PTH and the risks for hip, vertebral, and pelvic fractures among patients on dialysis. Am J Kidney Dis. 47 (2006) 149–156.

lP

16. Rudser, K.D., de Boer, I.H., Dooley, A., et al. Fracture risk after parathyroidectomy among chronic hemodialysis patients. J Am Soc Nephrol. 18 (2007) 2401–2407.

ur na

17. Iimori, S., Mori, Y., Akita, W., et al. Diagnostic usefulness of bone mineral density and biochemical markers of bone turnover in predicting fracture in CKD stage 5D patients--a single-center cohort study. Nephrol Dial Transplant. 27 (2012) 345–351. 18. Tentori, F., McCullough, K., Kilpatrick, R.D., et al. High rates of death and

Jo

hospitalization follow bone fracture among hemodialysis patients. Kidney Int. 85 (2014) 166–173.

19. Akiyama, T., Chikuda, H., Yasunaga, H., et al. Incidence and risk factors for mortality of vertebral osteomyelitis: a retrospective analysis using the Japanese diagnosis procedure combination database. BMJ Open. 3 (2013) e002412.

17

20. Mandai, S., Sato, H., Iimori, S., et al. Dialysis Case Volume Associated With In-Hospital Mortality in Maintenance Dialysis Patients. Kidney Int Rep. 3 (2017) 356– 363. 21. Bliuc, D., Nguyen, N.D., Milch, V.E., et al. Mortality risk associated with low-trauma osteoporotic fracture and subsequent fracture in men and women. JAMA.

of

301 (2009) 513–521. 22. Klop, C., Welsing, P.M., Leufkens, H.G., et al. The Epidemiology of Hip and Major

ro

Osteoporotic Fractures in a Dutch Population of Community-Dwelling Elderly:

Implications for the Dutch FRAX® Algorithm. PLoS One. 10 (2015) e0143800.

-p

23. Quan, H., Li, B., Couris, C.M., et al. Updating and validating the Charlson

re

comorbidity index and score for risk adjustment in hospital discharge abstracts using data from 6 countries. Am J Epidemiol. 173 (2011) 676–682.

lP

24. Williams, R. Using the margins command to estimate and interpret adjusted predictions and marginal effects. Stata J. 12 (2012) 308–331.

ur na

25. Hozawa. A., Hirata. T., Yatsuya. H., et al. Association between body mass index and all-cause death in Japanese population: pooled individual participant data analysis of 13 cohort studies. J Epidemiol. (2018). doi: 10.2188/jea.JE20180124. 26. Ravindrarajah, R., Hazra, N.C., Charlton, J., et al. Incidence and mortality of

Jo

fractures by frailty level over 80 years of age: cohort study using UK electronic health records. BMJ Open. 8 (2018) e018836. 27. Masakane, I., Nakai, S., Ogata, S., et al. An Overview of Regular Dialysis Treatment in Japan (As of 31 December 2013). Ther Apher Dial 19 (2015) 540–574.

18

28. Wakasugi, M., Kazama, J.J., Yamamoto, S., Kawamura, K., Narita, I. Cause-specific excess mortality among dialysis patients: comparison with the general population in Japan. Ther Apher Dial. 17 (2013) 298–304. 29. Iwagami, M., Yasunaga, H., Matsui, H., et al. Impact of end-stage renal disease on hospital outcomes among patients admitted to intensive care units: A retrospective

of

matched-pair cohort study. Nephrology (Carlton). 22 (2017) 617–623. 30. Usui, T., Hanafusa, N., Yasunaga, H., Nangaku, M. Association of dialysis with

ro

in-hospital disability progression and mortality in community-onset stroke. Nephrology (Carlton). 24 (2019) 737–743.

-p

31. Shortt, N.L., Robinson, C.M. Mortality after low-energy fractures in patients aged at

re

least 45 years old. J Orthop Trauma. 19 (2005) 396–400.

32. Clement, N.D., Duckworth, A.D., McQueen, M.M., et al. The outcome of proximal

(2014) 970–977.

lP

humeral fractures in the elderly: predictors of mortality and function. Bone Joint J. 96-B

ur na

33. Kukita, K., Ohira, S., Amano, I., et al; Vascular Access Construction and Repair for Chronic Hemodialysis Guideline Working Group, Japanese Society for Dialysis Therapy. 2011 update Japanese Society for Dialysis Therapy Guidelines of Vascular Access Construction and Repair for Chronic Hemodialysis. Ther Apher Dial. 19 Suppl

Jo

1 (2015) 1–39.

34. Yamana, H., Moriwaki, M., Horiguchi, H., et al. Validity of diagnoses, procedures, and laboratory data in Japanese administrative data. J Epidemiol. 27 (2017) 476–482. 35. Ravani, P., Quinn, R., Oliver, M., et al. Examining the Association between Hemodialysis Access Type and Mortality: The Role of Access Complications. Clin J Am Soc Nephrol. 12 (2017) 955–964. 19

20

of

ro

-p

re

lP

ur na

Jo

Tables Table 1. Characteristics of hospitalized patients due to major fractures from a national

(N = 9,320)

(N = 547,726)

77 (69–82)

80 (70–86)

≤64

1,323 (14)

94,329 (17)

65–74

2,560 (28)

89,848 (17)

≥75

5,437 (58)

363,549 (66)

Female

4,935 (53)

BMI (kg/m2)

20 (18–22) 2,456 (26)

18.5–24.9

5,320 (57)

≥25.0

969 (10)

Unknown

575 (6)

ur na

Charlson comorbidity

lP

≤18.4

<0.001

<0.001

404,995 (74)

<0.001

21 (19–24)

<0.001

112,324 (21)

<0.001

re

Age (year)

P value

of

General population

ro

Hemodialysis patients

-p

inpatient database

297,164 (54)

101,017 (18)

37,221 (7)

Cardiovascular disease

2,043 (22)

81,377 (15)

<0.001

Diabetes mellitus

5,870 (37)

84,857 (15)

<0.001

5,316 (57)

397,427 (73)

<0.001

3,363 (36)

126,231 (23)

641 (7)

24,068 (4)

7,655 (82)

414,760 (76)

<0.001

Hip

6,085 (65)

268,794 (49)

<0.001

Spine

1,775 (19)

122,241 (22)

Charlson comorbidity index 0

Jo

1–2 ≥3

Emergent admission Type of osteoporotic fracture

21

Forearm

433 (5)

86,755 (16)

Upper arm

563 (6)

41,759 (8)

Leg

464 (5)

28,177 (5)

2012

3,108 (33)

180,521 (33)

2013

2,413 (26)

151,071 (28)

2014

3,799 (41)

216,134 (39)

Year

of

ro

Data are numbers (percentiles) or medians (interquartile range).

0.001

Jo

ur na

lP

re

-p

BMI, body mass index.

22

Table 2. In-hospital mortality after major fractures in the population-based cohort: overall and stratified by risk factors and year

Hemodialysis vs. general population

OR (95% CI)

P value

2.48 (2.25–2.74)

<0.001

Reference

65–74

2.13 (1.85–2.45)

<0.001

≥75

4.28 (3.78–4.85)

<0.001

0.42 (0.41–0.44)

<0.001

ro

≤64

BMI (kg per m2) 1.98 (1.89–2.08)

18.5–24.9

Reference

≥25.0

0.76 (0.70–0.82)

0

lP

Charlson comorbidity index

<0.001

re

≤18.4

-p

Sex Female vs. male

<0.001

Reference

1.78 (1.70–1.87)

<0.001

4.23 (3.98–4.50)

<0.001

1.29 (1.19–1.38)

<0.001

Hip

1.31 (1.15–1.48)

<0.001

Spine

1.02 (0.90–1.16)

0.9

Forearm

0.13 (0.10–0.17)

<0.001

Upper arm

0.87 (0.74–1.02)

0.06

Leg

Reference

ur na

1–2 ≥3

of

Age (year)

Admission type

Emergent vs. elective

Jo

Type of osteoporotic fracture

23

Year 2012

Reference

2013

0.85 (0.80–0.90)

<0.001

2014

0.93 (0.89–0.98)

0.003

comorbidity index, admission type, fracture type, and admission year.

Jo

ur na

lP

re

-p

ro

BMI, body mass index; CI, confidence interval; OR, odds ratio.

of

Multivariate logistic regression models were adjusted for age, sex, BMI, Charlson

24

Table 3. Fracture site-specific mortality after hospitalization with fractures among ESKD patients receiving hemodialysis and the general population

Multivariatea

Univariate Death/N

OR (95% CI)

P value

OR (95% CI)

P value

Hip

344/6,085

2.26 (1.26–4.05)

0.006

1.77 (0.98–3.18)

0.06

Spine

81/1,775

1.80 (0.97–3.33)

0.06

1.48 (0.79–2.75)

0.2

Forearm

2/433

0.17 (0.04–0.79)

0.023

0.19 (0.04–0.86)

0.031

Upper arm

29/563

2.05 (1.03–4.06)

0.040

2.01 (1.01–4.01)

0.046

Leg

12/464

Reference

Hip

6,023/268,794

2.41 (2.13–2.73)

<0.001

1.28 (1.13–1.45)

<0.001

Spine

1,818/122,241

1.59 (1.40–1.81)

<0.001

1.00 (0.88–1.14)

1

Forearm

63/86,755

0.08 (0.06–0.10)

<0.001

0.13 (0.10–0.17)

<0.001

Upper arm

332/41,759

0.84 (0.72–0.99)

0.041

0.83 (0.70–0.97)

0.022

Leg

265/28,177

Reference

Hemodialysis

ur na

lP

population

a

ro

-p

Reference

re

General

of

patients

Reference

Multivariate logistic regression models were adjusted for age, sex, body mass index

Jo

Charlson comorbidity index, admission type, fracture type, and admission year. CI, confidence interval; ESKD, end-stage kidney disease; OR, odds ratio

25

Table 4. Association between fracture sites and vascular access failure among ESKD patients receiving hemodialysis

Multivariateb

Univariate OR (95% CI)

P value

OR (95% CI)

P value

Hip

705/6,085

1.22 (0.89–1.68)

0.2

1.17 (0.85–1.61)

0.3

Spine

147/1,775

0.84 (0.59–1.19)

0.3

0.84 (0.59–1.19)

0.3

Forearm

48/433

1.16 (0.76–1.78)

0.5

1.28 (0.83–1.98)

0.3

Upper arm

111/563

2.29 (1.58–3.31)

<0.001

2.33 (1.60–3.39)

<0.001

Leg

45/464

Reference

ro

of

Eventa/N

-p

Reference

a

re

Vascular access failure that required construction or reconstruction of arteriovenous

fistula or graft, insertion of temporary or permanent dialysis catheters, or graft/fistula

lP

percutaneous transluminal angioplasty. bMultivariate logistic regression models were adjusted for age, sex, body mass index, Charlson comorbidity index, admission type,

ur na

fracture type, and admission year. CI, confidence interval; ESKD, end-stage kidney

Jo

disease; OR, odds ratio.

26

Figure legends

ur na

lP

re

-p

ro

of

Figure 1. Flowchart of patient selection.

Figure 2. In-hospital and 30-day mortality rates after major fractures among hemodialysis patients and the general population according to age and sex. A. Rates of

Jo

overall in-hospital death. B. Rates of death within 30 days after admission. Each circle or diamond represents a mean and the solid lines represent the corresponding 95% CI. CI, confidence interval.

27

of ro -p re lP

Figure 3. Impact of hemodialysis dependence on in-hospital mortality after fractures in

ur na

the subgroup analysis. Multivariate logistic regression models were adjusted for age, sex, body mass index, Charlson comorbidity index, admission type, fracture type, and admission year. aInteraction analysis for relationship between the effect of hemodialysis dependence on mortality and upper arm or non-upper arm fractures was assessed. Each

Jo

circle represents a point estimate of OR, and the solid lines represent the corresponding 95% CI. Dotted lines represent multiples of four in OR. CI, confidence interval; OR, odds ratio.

28

29

of

ro

-p

re

lP

ur na

Jo