Osteoporos Int DOI 10.1007/s00198-016-3594-7
ORIGINAL ARTICLE
Body mass index, risk of allogeneic red blood cell transfusion, and mortality in elderly patients undergoing hip fracture surgery A. B. Pedersen 1 & D. Cronin Fenton 1 & M. Nørgaard 1 & N. R. Kristensen 1 & B. Kuno Møller 2 & C. Erikstrup 2
Received: 14 January 2016 / Accepted: 6 April 2016 # International Osteoporosis Foundation and National Osteoporosis Foundation 2016
Abstract Summary Despite improvements in preoperative and postoperative treatment, hip fracture surgery may lead to blood transfusion. Little is known about the impact of body mass index on transfusion risk and subsequent mortality. Opposite overweight and obese patients, underweight patients had increased risk of transfusion and death within 1 year of surgery. Introduction Despite improvements in preoperative and postoperative treatment of hip fracture patients, hip fracture surgery may lead to blood loss. We examined the risk of red blood cell transfusion (as an indirect measure of blood loss) and subsequent mortality by body mass index level in patients aged 65 and over undergoing hip fracture surgery. Methods This is a population-based cohort study using medical databases. We included all patients who underwent surgery for hip fracture during 2005–2013. We calculated the cumulative risk of red blood cell transfusion within 7 days of surgery treating death as a competing risk and, among transfused patients, short- (8–30 days postsurgery) and longterm mortality (31–365 days postsurgery). Results Among 56,420 patients, 47.7 % received at least one red blood cell transfusion within 7 days of surgery. In patients
Electronic supplementary material The online version of this article (doi:10.1007/s00198-016-3594-7) contains supplementary material, which is available to authorized users. * A. B. Pedersen
[email protected]
1
Department of Clinical Epidemiology, Aarhus University Hospital, Olof Palmes Alle 43-45, 8200 Aarhus N, Denmark
2
Department of Clinical Immunology, Aarhus University Hospital, Palle Juul-Jensens Boulevard 99, 8200 Aarhus N, Denmark
with normal weight, the risk was 48.8 % compared with 57.0 % in underweight patients (adjusted RR = 1.11; CI 1.08–1.15), 42.1 % in overweight patients (adjusted RR = 0.89; CI 0.86–0.91), and 42.2 % in obese patients (adjusted RR = 0.87; CI 0.84–0.91). Among transfused patients, adjusted HRs for short-term mortality were 1.52 (CI 1.34– 1.71), 0.70 (CI 0.61–0.80), and 0.58 (CI 0.43–0.77) for underweight, overweight, and obese patients, respectively, compared with normal-weight patients. The corresponding adjusted HRs for long-term mortality were 1.45 (CI 1.33–1.57), 0.80 (CI 0.74–0.86), and 0.58 (CI 0.50–0.69). Similar association between BMI and mortality was observed also among non-transfused patients. Conclusions Underweight patients had a higher risk of red blood cell transfusion and death in the first year of surgery than normal-weight patients, even when controlling for age and comorbidity. Opposite findings were seen for overweight and obese patients. Keywords Body mass index . Cohort study . Hip fracture . Mortality . Transfusion
Introduction Hip fractures are a common cause of hospital admission among the elderly [1]. In Denmark, there are approximately 7000 first time hip fracture hospitalizations per year among individuals aged 65 years and older, with a median age of 84 years [2]. Elderly patients with hip fractures often present with one or several comorbid diseases prior to the fracture [2, 3], warranting concurrent use of multiple prescription medications. Accordingly, hip fracture patients are often frail, and about 20 % are underweight at the time of diagnosis [2].
Osteoporos Int
Despite improvements in preoperative and postoperative treatment of these patients, hip fracture surgery may lead to red blood cell (RBC) transfusion. The American Academy of Orthopedic Surgeons guidelines recommend a blood transfusion threshold of no higher than 8 g/dL in asymptomatic postoperative hip fracture patients. Yet, transfusion rates in hip fracture patients vary substantially between European countries, Canada, and the USA [4–6]. Even the within-country variation is high according to data from Canada (24 to 48 %) [5], USA (36 to 95 %) [4], and Denmark (0 to 46 %) [7]. Still, few and small studies have examined factors that may predict RBC transfusion requirement in hip fracture patients [8–10], including older age, female gender, more severe type of fracture, and low preoperative hemoglobin. To our knowledge, no previous studies have examined whether body mass index (BMI) is associated with risk of RBC transfusion as a marker of blood loss in hip fracture patients. Mortality within 30 days following surgery for hip fracture is about 10 % [11]. In patients undergoing cardiac surgery [12] or hip replacement surgery [13], allogeneic blood transfusion is associated with an increased risk of serious complications and death. In various clinical settings, a link between overweight and reduced risk of complications or in-hospital mortality was observed [14–16]. To our knowledge, no studies have examined the association between BMI level and mortality in hip fracture patients who received RBC transfusion in relation to surgery. We therefore conducted a populationbased cohort study to examine the risk of RBC transfusion and subsequent mortality by BMI level in elderly hip fracture surgery patients.
Materials and methods Setting We conducted this study in Denmark (a population of 5.6 million persons) using prospectively collected data from population-based medical databases. The Danish National Health Service provides tax-supported healthcare to all Danish residents, guaranteeing free medical care for emergency and general hospital admissions and for outpatient clinic visits [17]. Study population We used the Danish Hip Fracture Database (DHFD) [2] to identify all patients who underwent surgery for hip fracture. The DHFD is a nationwide population-based clinical quality database including all patients with femoral neck, pertrochanteric, or subtrochanteric fractures surgically treated in Denmark since 2003 (please see Appendix 1 for diagnosis codes). All orthopedic departments in Denmark provide data
to the DHFD, and reporting is mandatory [18]. Preoperative and perioperative data are collected prospectively upon hospital admission by the staff caring for patients using a standardized registration form. Detailed definitions of data elements are provided to ensure uniform registration of data across departments. We included first time hip fracture patients aged 65 years or older who underwent primary hip replacement or open reduction and internal fixation of the hip fracture (please see Appendix 1 for surgical procedure codes). Study period In order to have available information on drugs prescribed within 90 days before hip fracture, our study was restricted to patients registered in the DHFD after January 1, 2005, and through December 31, 2013, when prescription data was available. Data sources The Danish National Patient Registry (DNPR) has registered data on all hospital admissions since 1977 and all hospital outpatient and emergency visits since 1995, including dates of admission and discharge and up to 20 discharge diagnoses recorded according to the International Classification of Diseases (eighth edition until the end of 1993 and tenth edition thereafter) [19]. The Danish Civil Registration System (CRS) was initiated in 1968. All Danish residents are at birth or immigration assigned a unique ten-digit personal identification number encoding age, gender, and date of birth. This permits unambiguous linkage between all Danish medical databases [17]. The Danish Transfusion Database (DTDB) is a clinical quality database monitoring the use of blood components on a national level since 2000. The database retains data on all blood transfusions administered at all Danish hospitals during the study period and includes information on the civil registration number of the patient receiving the specific blood component, types and number of blood components administered to the patient, date of delivery of the blood component from the blood bank, and clinical biochemical data such as hemoglobin concentrations measured preoperatively [7]. The Danish National Health Service Prescription Database (DNHSPD) [20] has maintained information on all prescriptions for reimbursed drugs dispensed by community pharmacies in Denmark since 2004, recorded according to the Anatomical Therapeutic Chemical classification system (ATC codes). This database was established by the Danish Regions and is hosted by Aarhus University. Additional variables in the DNHSPD include the name of the drug, package identifier (permitting identification of brand, quantity, and formulation of the drug), date of refill, code identifying the
Osteoporos Int
prescribing physician, and code identifying the dispensing pharmacy. Hospital dispensaries do not report to the database.
Body mass index Information on body weight and height at the time of hip fracture was obtained from the DHFD to calculate BMI as the weight in kilograms (kg) divided by the square of height in meters (m). Patients were categorized as underweight if BMI was <18.5 kg/m2, normal weight if BMI was ≤18.5– 24.9 kg/m2, overweight if BMI was 25–29.9 kg/m2, and obese if BMI was ≥30 kg/m2. Approximately 21.7 % of patients did not have information on BMI status, and this information was imputed as described in the Statistics section. Coexistent diseases We used the DNPR to obtain a complete hospital history for all patients included in the study cohort through the 10 years preceding hip fracture surgery. As a measure of comorbidity, we computed a Charlson comorbidity index (CCI) score [21] for each patient. We defined three comorbidity levels: a score of 0 (low), given to patients with no previous record of diseases included in the CCI; a score of 1–2 (medium); and a score of 3 or more (high). Only diagnosis codes registered in relation to hospitalization for these diseases were included in CCI; thus, outpatient visits and emergency visits were left out due to expected lower validity of these diagnosis codes. RBC transfusion We used the DTDB to obtain information on all allogeneic RBC transfusions administered to patients from the date of surgery for hip fracture up to 7 days after surgery. We chose the 7-day window in order to capture the RBC transfusions associated with the hip fracture and surgery rather than transfusions due to other factors such as comorbidity. Patients were categorized, as having received either none or at least one unit of RBCs within 7 days of surgery (including the day of surgery). We further classified patients according to the number of RBC transfusions received within 7 days of surgery. Mortality The Danish Civil Registration System provided information on deaths. This national registry has maintained records on vital status, migration, and residence for the entire Danish population since 1968 [17]. We studied short-term mortality (8–30 days postsurgery) and subsequent long-term mortality (31–365 days postsurgery) in patients who received RBC transfusion within 7 days of surgery.
Covariates We obtained information on gender and age (in categories: 65–74, 75–84, and 85 years or older) from the DHFD. In addition, information on type of primary hip fracture diagnosis (femoral neck, pertrochanteric, or subtrochanteric fractures), type of surgery (primary hip replacement or open reduction and internal fixation), and surgery delay (<24 h, 25– 35, and ≥36 h) was obtained from the DHFD. We obtained information from the DTDB on preoperative hemoglobin concentrations measured within 1 month prior to or on the date of hip fracture surgery (however, the majority of patients had their hemoglobin measured 2 to 3 days prior to or on the date of hip fracture surgery) and postoperative hemoglobin concentration measured from the date of surgery and up to 7 days, thereafter classified according to the World Health Organization standard reference interval [22]. From the DNHSPD, we collected information on prescriptions for the following classes of drugs in the period of 90 days prior to surgery as possible confounders of the association between BMI and transfusion and/or mortality: non-steroidal anti-inflammatory drugs, antidepressants (selective serotonin reuptake inhibitors (SSRIs)), corticosteroids, oral anticoagulation therapy, and statins (please see Appendix 1 for ATC codes).
Statistics Descriptive analyses of the study population were performed, tabulating the number of patients in total and the number and percent of patients’ characteristics according to transfusion receipt. Among all included hip fracture patients, we computed the cumulative incidence of RBC transfusion within 7 days of surgery with 95 % confidence intervals (CIs) for each BMI group, considering death as a competing risk. We used a log-binomial model to estimate relative risks (RRs) for RBC transfusion and corresponding 95 % CIs, both crude and adjusted for age, gender, CCI, type of surgery, surgery delay, year of surgery, and prescription medication. We calculated RRs for RBC transfusion comparing four BMI groups using patients with normal weight as reference. Since hospitalrelated factors, including the experience of the surgeon and the decision-making process, which can be governed more by hospital tradition for transfusion than clinical indication, may affect the need for transfusion, we adjusted for a cluster effect by departments. In order to study any potential differences in the effect of BMI on RBC transfusion risk in subgroup of hip fracture patients, we stratified the analyses by preoperative hemoglobin level. We calculated the average number of RBC transfusions by BMI level.
Osteoporos Int
Among hip fracture patients who received RBC transfusion within seven postoperative days, we used the Kaplan-Meier method to compute the cumulative short-term mortality risk (time to death from 8 to 30 days after surgery, given that the patient was alive on day 8) and long-term mortality risk (time to death from 31 to 365 days after surgery, given that the patient survived the short-term period). We used Cox proportional hazards regression to estimate hazard ratios (HRs) for death and corresponding 95 % CIs, both crude and adjusted for included covariates, comparing the four BMI groups among transfused and non-transfused patients. We estimated RRs and HRs first using complete case analysis, thus excluding patients with missing information on BMI and surgery delay. Afterward, we used multiple imputation method to impute the missing data regarding BMI and surgery delay (using all available information from the patients presented in Table 1, including also outcome data), which generated five imputed datasets. All presented adjusted RRs and HRs were calculated as the geometric mean of the RRs or HRs of the five datasets, with the corresponding confidence intervals corrected for between- and withinimputation variation [23].
characteristics from the patients with information on BMI. Risk of RBC transfusion by BMI level Overall, 47.7 % (n = 26,890) of patients received at least one postoperative RBC transfusion within 7 days. The risk of RBC transfusion within 7 days of surgery by BMI level is presented in Table 2. The adjusted RRs for RBC transfusion were 1.11 (CI 1.08–1.15) for underweight patients, 0.89 (CI 0.86–0.91) for overweight patients, and 0.87 (CI 0.84–0.91) for obese patients, compared with normal-weight patients (Table 2). Basing the analysis on imputed BMIs instead of complete case did not change the estimates (Table 2). The average number of transfusions within 7 days of surgery among patients receiving at least 1 RBC was 2.63 (range 1– 30) for underweight patients, 2.66 (range 1–15) for normalweight patients, 2.73 (range 1–32) for overweight patients, and 2.88 (range 1–40) for obese patients. Although the risk of RBC transfusion increased with increasing CCI level and age, the RR estimates for underweight, overweight, and obese patients compared with normal-weight patients had similar patterns when stratified by CCI level (Table 3) and three age groups (Table 4). Preoperative hemoglobin and risk of RBC transfusion by BMI level
Results Patient characteristics Among the 56,420 hip fracture patients included in the analyses of RBC transfusion risk, 8.9 % of patients were underweight, 45.4 % were normal weight, 18.8 % were overweight, and 5.1 % were obese patients. Table 1 presents baseline characteristics of the hip fracture patients by BMI level (please see Appendix 2 for distribution of 18 comorbid diseases by BMI level). Briefly, compared with normal-weight patients, underweight patients were similar in age, were more likely to be female, to have slightly more comorbid conditions before surgery, to have complicated pertrochanter/ subtrochanter hip fracture and internal fixation of fracture, and to have received more steroids and less statins and NSAIDs during the 90 days prior to surgery. Underweight patients were also more likely to have low preoperative hemoglobin than normal-weight patients. Compared with normal-weight patients, overweight and obese patients were younger, more likely to be men, to have more severe comorbid conditions, to receive NSAIDs and statins, but less likely to have low preoperative hemoglobin. Patients without information on BMI level did not differ substantially in their
Approximately 41 % of patients had missing information on preoperative hemoglobin level irrespective of BMI (Table 1). Despite low preoperative hemoglobin, 37.3 % did not receive RBC transfusion. In contrast, 26.5 % of patients with normal preoperative hemoglobin received RBC transfusion within 7 days of surgery. Approximately 61.9 % of underweight patients had low preoperative hemoglobin, compared with 56.1, 47.8, and 47.7 % of normal-weight, overweight, and obese patients, respectively (Table 1). Among those with low preoperative hemoglobin, risk of RBC transfusion within 7 days of surgery was 69.2 % (CI 67.0–71.3 %) for underweight patients, 63.8 % (CI 62.7– 64.8 %) for normal-weight patients, 58.8 % (CI 57.1– 60.6 %) for overweight patients, and 62.3 % (CI 58.9– 65.5 %) for obese patients. This corresponded to adjusted RRs for RBC transfusion of 1.03 (CI 1.00–1.07) for underweight patients, 0.96 (CI 0.93–0.99) for overweight patients, and 0.96 (CI 0.91–1.00) for obese patients, compared with normal-weight patients. Among those with normal preoperative hemoglobin, risk of RBC transfusion within 7 days of surgery was 32.3 % (CI 29.9–35.1 %) for underweight patients, 27.6 % (CI 26.5– 28.7 %) for normal-weight patients, 24.4 % (CI 22.9– 25.9 %) for overweight patients, and 23.9 % (CI 21.1– 26.7 %) for obese patients, corresponding to adjusted RRs
Age at the time of surgery (years) 65–74 10,527 (18.7 %) 75–84 22,164 (39.3 %) 85+ 23,729 (42.1 %) Gender, female 40,568 (71.9 %) Charlson Comorbidity Index (CCI) Low (score 0) 27,659 (49.0 %) Medium (scores 1–2) 20,871 (36.9 %) High (≥3) 7890 (14.0 %) Hip fracture type Fracture of femoral neck 29,323 (52.0 %) Pertrochanter and subtrochanter fractures 27,097 (48.0 %) Type of surgery Osteosyntheses 39,716 (70.4 %) Total and hemi hip arthroplasty 16,704 (29.6 %) Surgery delay (h)b No information 179 (0.3 %) <24 32,416 (57.5 %) 25–36 10,662 (18.9 %) >36 13,163 (23.3 %) Preoperative hemoglobin concentration within 1 month prior to surgeryc No information 23,733 Male ≤8.1 mmol/L or 13.0 g/dL 18,014 (55.1 %) Female ≤7.4 mmol/L or 11.0 g/dL Male >8.1 mmol/L or 13.0 g/dL 14,673 (44.9 %) Female >7.4 mmol/L or 11.0 g/dL d Postoperative hemoglobin concentration No information 34,407 (60.9 %) Male ≤8.1 mmol/L or 13.0 g/dL 18,819 (85.5 %) Female ≤7.4 mmol/L or 11.0 g/dL Male >8.1 mmol/L or 13.0 g/dL 3194 (15.5 %) Female >7 mmol/L or 11.0 g/dL Year of surgery 2005 6134 (10.9 %) 2006 6244 (11.1 %) 2007 6377 (11.3 %) 2008 6677 (11.8 %) 2009 6248 (11.1 %) 2010 6608 (11.7 %) 2011 6465 (11.5 %) 2012 6170 (10.9 %) 2013 5497 (9.7 %) Redeemed prescriptions for medications within 90 days of hip fracture Non-steroidal anti-inflammatory drugs 21,951 (38.9 %)
Total patients at risk, N = 56,420
4460 (17.4 %) 9836 (38.4 %) 11,343 (44.2 %) 18,630 (72.7 %) 12,986 (50.7 %) 9321 (36.3 %) 3332 (13.0 %) 13,487 (52.6 %) 12,152 (47.4 %) 17,809 (69.5 %) 7830 (30.5 %) 71 (0.3 %) 14,791 (57.7 %) 4813 (18.8 %) 5964 (23.3 %)
10,747 (41.9 %) 8350 (56.1 %) 6542 (43.9 %)
15,687 (61.2 %) 8586 (86.3 %) 1366 (13.7 %)
2336 (9.1 %) 2757 (10.8 %) 3121 (12.2 %) 3197 (12.5 %) 3007 (11.7 %) 2715 (10.6 %) 2889 (11.3 %) 2985 (11.6 %) 2632 (10.3) 9618 (37.5 %)
2325 (46.4 %) 2013 (40.2 %) 672 (13.4 %) 2326 (46.4 %) 2684 (53.6 %) 3767 (75.2 %) 1243 (24.8 %) 24 (0.5 %) 2879 (57.5 %) 963 (19.2 %) 1144 (22.8 %)
2105 (42.0 %) 1799 (61.9 %) 1106 (38.1 %)
3274 (65.0 %) 1522 (87.7 %) 214 (12.3 %)
448 (8.9 %) 549 (11.0 %) 637 (12.7 %) 659 (13.2 %) 602 (12.0 %) 519 (10.4 %) 564 (11.3 %) 534 (10.7 %) 498 (9.9 %) 1671 (33.4 %)
Normal-weight patients, N = 25,639
863 (17.2 %) 1801 (36.0 %) 2346 (46.8 %) 4347 (86.8 %)
Underweight patients, N = 5010
Baseline characteristics of 56,420 hip fracture patients according to body mass index (BMI) level
Patient characteristics, total and by BMI levela
Table 1
%) %) %) %)
4491 (42.2 %)
897 (8.4 %) 1066 (10.0 %) 1137 (10.7 %) 1287 (12.1 %) 1162 (10.9 %) 1170 (11.0 %) 1453 (13.7 %) 1268 (11.9 %) 1192 (11.2 %)
745 (16.2 %)
6027 (56.7 %) 3860 (83.8 %)
3280 (52.2 %)
4350 (40.9 %) 3002 (47.8 %)
20 (0.2 %) 6168 (58.0 %) 1956 (18.4 %) 2488 (23.4 %)
7273 (68.4 %) 3359 (31.6 %)
5797 (54.5 %) 4835 (45.5 %)
5298 (49.8 %) 3808 (35.8 %) 1526 (14.4 %)
2281 (21.4 4543 (42.7 3808 (35.8 6898 (64.9
Overweight patients, N = 10,632
1327 (45.9 %)
252 (8.7 %) 284 (9.8 %) 312 (10.8 %) 320 (11.1 %) 288 (10.0 %) 337 (11.7 %) 396 (13.7 %) 364 (12.6 %) 336 (11.6 %)
220 (17.4 %)
1627 (56.3 %) 1042 (82.6 %)
909 (52.3 %)
1150 (39.8 %) 830 (47.7 %)
6 (0.2 %) 1635 (56.6 %) 577 (19.9 %) 671 (23.2 %)
2044 (70.8 %) 845 (29.3 %)
1497 (51.8 %) 1392 (48.2 %)
1365 (47.3 %) 1022 (35.4 %) 502 (17.4 %)
817 (28.3 %) 1319 (45.7 %) 753 (26.1 %) 2046 (70.8 %)
Obese patients, N = 2889
%) %) %) %)
4844 (39.5 %)
2201 (18.0 %) 1588 (13.0 %) 1170 (9.6 %) 1214 (9.9 %) 1189 (9.7 %) 1867 (15.2 %) 1163 (9.5 %) 1019 (8.3 %) 839 (6.9 %)
649 (14.6 %)
7792 (63.6 %) 3809 (85.4 %)
2836 (41.3 %)
5381 (43.9 %) 4033 (58.7 %)
58 (0.5 %) 6943 (56.7 %) 2353 (19.2 %) 2896 (23.6 %)
8823 (72.0 %) 3427 (28.0 %)
6216 (50.7 %) 6034 (49.3 %)
5685 (46.4 %) 4707 (38.4 %) 1858 (15.2 %)
2106 (17.2 4665 (38.1 5479 (44.7 8647 (70.6
Patients with missing BMI value, N = 12,250
Osteoporos Int
for RBC transfusion of 1.10 (CI 1.04–1.16) for underweight patients, 0.89 (CI 0.83–0.95) for overweight patients, and 0.92 (CI 0.84–1.00) for obese patients, compared with normalweight patients.
Calculations were done in the same way as for preoperative hemoglobin
Mortality in transfused and non-transfused patients by their BMI level
d
Surgery delay in hours calculated from the time of admission for hip fracture to time of surgery
Percentage for no information groups was calculated out of all patients in the specific column. Percentages for two groups of preoperative hemoglobin were calculated in related to only patients with available information on preoperative hemoglobin in the specific column
Underweight if BMI < 18.5 kg/m2 , normal weight if BMI 18.5–24.9 kg/m2 , overweight if BMI 25–29.9 kg/m2 , and obese if BMI ≥ 30 kg/m2 a
b
c
2864 (23.4 %) 1040 (8.5) 289 (2.4 %) 1811 (14.8 %) 545 (19.9 %) 258 (8.9 %) 43 (1.5 %) 756 (26.2 %) 2201 (20.7 %) 859 (8.1 %) 199 (1.9 %) 2236 (21.0 %) 5060 (19.7 %) 2055 (8.0 %) 463 (1.8 %) 3914 (15.3 %) 1005 (20.1 %) 585 (811.7 %) 95 (1.9 %) 562 (11.2 %) 11,675 (20.7 %) 4797 (8.5 %) 1089 (1.9 %) 9279 (16.5 %) Antidepressants (selective serotonin reuptake inhibitors) Corticosteroids Oral anticoagulation therapy Statins
Patient characteristics, total and by BMI levela
Table 1 (continued)
Total patients at risk, N = 56,420
Underweight patients, N = 5010
Normal-weight patients, N = 25,639
Overweight patients, N = 10,632
Obese patients, N = 2889
Patients with missing BMI value, N = 12,250
Osteoporos Int
Mortality among transfused patients is presented in Fig. 1. The adjusted HRs for short-term mortality among transfused patients was 1.52 (95 % CI 1.34–1.71) for underweight transfused patients, 0.70 (95 % CI 0.61–0.80) for overweight transfused patients, and 0.58 (95 % CI 0.43–0.77) for obese transfused patients prior to surgery, compared with normal-weight transfused patients. The adjusted HRs for long-term mortality among transfused patients was 1.45 (95 % CI 1.33–1.57) for underweight transfused patients, 0.80 (95 % CI 0.74–0.86) for overweight transfused patients, and 0.58 (95 % CI 0.50–0.69) for obese transfused patients prior to surgery, compared with normal-weight transfused patients. The association between BMI level and both short- and long-term mortality among transfused patients showed the similar direction and magnitude irrespective of the CCI level. Among non-transfused patients, the adjusted HRs for short-term mortality was 1.60 (95 % CI 1.34–1.90) for underweight non-transfused patients, 0.59 (95 % CI 0.52–0.66) for overweight non-transfused patients, and 0.53 (95 % CI 0.36– 0.78) for obese non-transfused patients, compared with normal-weight non-transfused patients. Among nontransfused patients, the adjusted HRs for long-term mortality was 1.60 (95 % CI 1.46–1.75) for underweight non-transfused patients, 0.74 (95 % CI 0.74–0.86) for overweight nontransfused patients, and 0.68 (95 % CI 0.59–0.78) for obese non-transfused patients, compared with normal-weight nontransfused patients. The association between BMI level and both short- and long-term mortality among non-transfused patients showed the similar direction and magnitude irrespective of the CCI level.
Discussion In this population-based cohort study of 56,420 hip fracture patients aged 65 years and older, underweight patients were at increased risk of receiving RBC transfusion compared with normal-weight patients, irrespective of their previous comorbidity status. Underweight transfused and non-transfused patients also had elevated mortality up to 1 year following hip fracture surgery compared with normal-weight transfused patients, even after adjusting for patient characteristics and medication use before surgery.
Osteoporos Int Table 2 Cumulative risk and adjusted relative risk for red blood cell transfusion within 7 days of surgery among 56,420 hip fracture patients according to body mass index (BMI) level BMIa
Cumulative transfusion risk (%) (95 % CI)
Complete case method
Multiple imputation method
Adjusted RRb (95 % CI)
Crude RR (95 % CI)
Adjusted RRb (95 % CI)
Normal-weight patients
48.8 (48.2–49.4)
1.0 (reference)
1.0 (reference)
1.0 (reference)
Underweight patients
57.0 (55.7–58.4)
1.12 (1.10–1.16)
1.16 (1.12–1.20)
1.11 (1.08–1.15)
Overweight patients Obese patients
42.1 (41.1–43.0) 42.2 (40.4–44.0)
0.88 (0.86–0.90) 0.89 (0.85–0.92)
0.87 (0.84–0.90) 0.85 (0.81–0.89)
0.89 (0.86–0.91) 0.87 (0.84–0.91)
a
Underweight if BMI < 18.5 kg/m2 , normal weight if BMI 18.5–24.9 kg/m2 , overweight if BMI 25–29.9 kg/m2 , and obese if BMI ≥ 30 kg/m2
b
RR: relative risk adjusted for age, gender, CCI, type of surgery, surgery delay, year of surgery, and prescription medication. All available information from the patients presented in Table 1, including also outcome data were used in order to generate imputed datasets
Comparison with previous literature Our finding of an overall risk of RBC transfusion of 47.7 % is well within the range of previously reported risks of transfusion, which have varied between 0 and 95 % [4–6]. This wide variation in transfusion risk may be partly derived from differences in defining transfusion during 72 h up to 10 days of admission. Since the published studies were based on patients operated in different study periods, the change in transfusion practice over time [13] could further contribute to the variation in transfusion rates between countries. Another possible explanation might be variation in the selection and characteristics of study populations, such as differences in the extent of
comorbid diseases before hip fracture. Unfortunately, comorbidities were not uniformly or consistently reported in previous studies. The large variation in transfusion risks may just be the results of different traditions and practice regarding transfusion irrespective of the clinical indication for transfusion. It is suggested, for example, that treatment of patients at private hospitals is related to low risk of transfusion [7]. Although preoperative hemoglobin level has been shown to be a strong predictor of perioperative RBC transfusion in hip fracture patients [8–10], approximately 41.4 % of our transfused patients had high preoperative hemoglobin. This suggests that either these patients suffered from substantial blood loss during surgery necessitating blood transfusion or that Denmark has a liberal
Table 3 Cumulative risk and adjusted relative risk for red blood cell transfusion within 7 days of surgery in 56,420 hip fracture patients by body mass index (BMI) for three Charlson comorbidity groups
BMIa
Cumulative transfusion risk (%) (95 % CI)
Charlson comorbidity index, low (score 0) Normal-weight patients 44.9 (44.1–45.8) Underweight patients 54.8 (52.7–56.8) Overweight patients 37.7 (36.4–39.0) Obese patients 39.0 (36.4–41.6) Charlson comorbidity index, medium (scores 1–2) Normal-weight patients 51.9 (50.9–52.9) Underweight patients 58.5 (56.3–60.6) Overweight patients 44.8 (43.2–46.3) Obese patients 42.3 (39.2–45.3) Charlson comorbidity index, high (score ≥3) Normal-weight patients 55.4 (53.7–57.1) Underweight patients 60.5 (56.7–64.1) Overweight patients 50.3 (47.8–52.8) Obese patients 50.7 (46.2–55.0) a b
Complete case method
Multiple imputation method
Adjusted RRb (95 % CI)
Crude RR (95 % CI)
Adjusted RRb (95 % CI)
1.0 (reference) 1.15 (1.10–1.21) 0.87 (0.84–0.91) 0.90 (0.83–0.96)
1.0 (reference) 1.20 (1.15–1.26) 0.85 (0.81–0.89) 0.84 (0.78–0.91)
1.0 (reference) 1.15 (1.10–1.20) 0.88 (0.84–0.91) 0.87 (0.81–0.94)
1.0 (reference) 1.10 (1.04–1.15) 0.88 (0.85–0.91) 0.84 (0.80–0.89)
1.0 (reference) 1.12 (1.06–1.18) 0.87 (0.84–0.90) 0.81 (0.76–0.86)
1.0 (reference) 1.09 (1.03–1.15) 0.89 (0.86–0.92) 0.84 (0.80–0.89)
1.0 (reference) 1.07 (1.00–1.15) 0.92 (0.87–0.97) 0.92 (0.86–1.01)
1.0 (reference) 1.09 (1.04–1.15) 0.91 (0.86–0.96) 0.91 (0.83–0.99)
1.0 (reference) 1.08 (1.02–1.14) 0.92 (0.87–0.97) 0.92 (0.84–1.00)
Underweight if BMI < 18.5 kg/m2 , normal weight if BMI 18.5–24.9 kg/m2 , overweight if BMI 25–29.9 kg/m2 , and obese if BMI ≥ 30 kg/m2
RR: relative risk adjusted for age, gender, CCI, type of surgery, surgery delay, year of surgery, and prescription medication. All available information from the patients presented in Table 1, including also outcome data were used in order to generate imputed datasets
Osteoporos Int Table 4 Cumulative risk and adjusted relative risk for red blood cell transfusion within 7 days of surgery in 56,420 hip fracture patients by body mass index (BMI) for three age groups Complete case method
Multiple imputation method
Cumulative transfusion risk (%) (95 % CI)
Adjusted RRa (95 % CI)
Crude RR (95 % CI)
Adjusted RRa (95 % CI)
Normal-weight patients
33.9 (32.6–35.3)
1.0 (reference)
1.0 (reference)
1.0 (reference)
Underweight patients Overweight patients
46.8 (43.4–50.1) 27.9 (26.1–29.7)
1.36 (1.24–1.50) 0.82 (0.75–0.89)
1.33 (1.22–1.44) 0.84 (0.78–0.90)
1.32 (1.21–1.43) 0.83(0.78–0.90)
30.3 (27.2–33.5)
0.87 (0.79–0.95)
0.88 (0.80–0.96)
0.86 (0.78–0.93)
Normal-weight patients
45.9 (44.9–46.9)
1.0 (reference)
1.0 (reference)
1.0 (reference)
Underweight patients Overweight patients
53.9 (51.6–56.2) 41.0 (39.5–42.4)
1.16 (1.09–1.23) 0.89 (0.86–0.93)
1.16 (1.11–1.22) 0.89 (0.85–0.94)
1.15 (1.09–1.22) 0.89 (0.85–0.93)
Obese patients
42.9 (40.2–45.6)
0.91 (0.85–0.98)
0.91 (0.85–0.97)
0.89 (0.83–0.95)
57.3 (56.3–58.2) 63.2 (61.2–65.1) 51.8 (50.2–53.4) 53.7 (50.0–57.1)
1.0 (reference) 1.07 (1.03–1.12) 0.91 (0.88–0.94) 0.92 (0.85–1.00)
1.0 (reference) 1.09 (1.05–1.14) 0.91 (0.87–0.94) 0.92 (0.84–0.99)
1.0 (reference) 1.07 (1.03–1.11) 0.91 (0.88–0.94) 0.91 (0.84–0.98)
Age at the time of surgery
Age 65–74 years
Obese patients Age 75–84 years
Age 85+ years Normal-weight patients Underweight patients Overweight patients Obese patients
Underweight if BMI < 18.5 kg/m2 , normal-weight if BMI 18.5–24.9 kg/m2 , overweight if BMI 25–29.9 kg/m2 , and obese if BMI ≥ 30 kg/m2 a
RR: relative risk adjusted for age, gender, CCI, type of surgery, surgery delay, year of surgery, and prescription medication. All available information from the patients presented in Table 1, including also outcome data were used in order to generate imputed datasets
transfusion practice [9]. A recent Cochrane meta-analysis showed that liberal transfusion practice was not associated with either reduced risk of mortality and postoperative morbidity or improved functional recovery compared with restrictive practice, but the findings should be interpreted cautiously due to heterogeneity and low-quality evidence of included studies [24]. The largest randomized study of restrictive versus liberal transfusion
Fig. 1 Cumulative 1-year mortality in 26,890 transfused patients by body mass index (BMI) level. Underweight if BMI was <18.5 kg/m2, normal weight if BMI was ≤18.5–24.9 kg/m2, overweight if BMI 25–29.9 kg/ m2, and obese if BMI ≥ 30 kg/m2
strategy including 2016 patients undergoing hip surgery reported no difference between groups in mortality or inability to walk independently after 60 days of follow-up [25]. Indeed, according to a recent Danish study including frail elderly patients from nursing homes and sheltered housing facilities, a liberal strategy may be beneficial for the recovery of activities of daily living after 1 year of follow-up, and it is not associated with increased
Osteoporos Int
risk of infection within 30 days of surgery [26]. Thus, it is still unknown which transfusion strategy provides the best balance between effectiveness and harmful effects. To our knowledge, this is the first population-based study to examine the risk of RBC transfusion in relation to BMI among elderly hip fracture patients. However, four studies have been published evaluating the association between BMI and blood loss or transfusion in elective hip arthroplasty patients [27–30]. Our findings concur with the results of two studies in hip arthroplasty patients [28, 30] showing that underweight patients have increased blood loss or transfusion risk, whereas other studies found no association [27–30]. Since approximately 30 % of hip fracture patients are treated with total or partial hip arthroplasty, these findings may not be comparable. There are several possible explanations for our findings. Underweight patients are more prone to develop anemia and nutritional deficiencies [31]—we note that low preoperative hemoglobin was most frequent among underweight patients in our study. These numbers may be misclassified as we lacked data on preoperative hemoglobin in about 40 % of patients irrespective of BMI. The difference in transfusion risk between BMI groups could be the result of local clinical guidelines with particular focus on frail underweight patients in order to secure early mobilization and active postsurgical rehabilitation. Early mobilization within 24 h is considered as one of the quality indicators for optimal treatment of hip fracture patients in Denmark, and it has been used to compare the quality of treatment at Danish orthopedic departments [2]. However, it is important to point out that underweight is not included as a quality of care indicator in the Danish national clinical guidelines for blood transfusion [32]. According to national transfusion guidelines, blood transfusion is indicated for patients with low hemoglobin but the hemoglobin threshold is higher for patients with heart disease [32] and in case of anemia symptoms. Underweight is associated with the loss of both peripheral and respiratory muscles and increased risk of respiratory disease-specific mortality [33], as well as a reduced immune response, which may increase susceptibility to infections [34]. This may explain our finding that underweight transfused and non-transfused hip fracture patients had increased mortality compared to normal-weight patients. Underweight could be a marker of any underlying chronic diseases prior to hip fracture surgery [35], which may have been undiagnosed at the time we collected information on comorbidity. Obesity has been recognized as a risk factor for a variety of diseases [36, 37]. However, our findings corroborated the Bobesity paradox^ which has been observed in different clinical settings, namely, a reduced risk of complications or in-hospital mortality among individuals who are overweight or obese compared with normal-weight patients [14–16, 38]. It is possible, however, that our findings reflect a Bhealthy survivor^ effect, i.e., persons who had very severe comorbid conditions are more
susceptible to obesity-related diseases and complications and have died before sustaining hip fracture surgery. Thus, we see relative Bhealthy^ obese hip fracture patients in our study population compared to obese persons in the general population. Limitations Extensive efforts are made to ensure completeness of DHFD data [18] including hospital reimbursement only after registration of a surgical procedure. Due to prospective registration of data, any lack of registration of hip fracture patients in the DHFD is probably independent of BMI level. Although we adjusted for comorbidity, we lacked information on the severity of some medical conditions included in the CCI score, which may have introduced residual confounding into our analyses. Our estimates could also be affected by unmeasured confounding and a Bhealthy user effect^ since there may be unknown or unmeasured reasons why some patients are prescribed RBC transfusion or not. We used multiple imputations to handle missing data on BMI and surgery delay. We did not impute preoperative and postoperative hemoglobin values because this information was lacking in more than 40 % of patients. Due to the same reason, we did not control for preoperative hemoglobin values or blood loss during the surgery. Nevertheless, we presented estimates for RBC transfusion risk stratified on measured preoperative hemoglobin, and although pointing in the same direction, these results should be interpreted with caution. Yet, hemoglobin may also be an intermediate factor in association between BMI and RBC transfusion risk. We lacked data on lifestyle factors. Thus, we may have unmeasured confounding leading to the appearance of an association between underweight and transfusion or mortality [39], if lifestyle factors differ between BMI groups. As only the date and not the exact time of transfusion was known, it was not possible to distinguish between transfusion immediately before and after the surgical procedure was initiated. Significance and implications of the study results Since the prevalence of obesity is increasing worldwide [40], focus has been on the association between obesity and adverse events [41]. No previous study has examined the prognosis of underweight hip fracture patients, although hip fracture is one of the most common causes of hospital admissions in older people. In order to improve treatment quality and reduce hospital costs, it is important to better understand the effect of both obesity and underweight on postoperative complications, including RBC transfusion and mortality in a large population of older hip fracture patients.
Osteoporos Int
Conclusions Compared with normal-weight patients, underweight patients were at increased risk of receiving RBC transfusion and, if transfused, at increased risk of dying in the first year following hip fracture surgery. The associations were independent of comorbidity level before fracture.
13.
14.
15.
Acknowledgments The authors wish to thank the orthopedic surgeons and other healthcare professionals at all hospitals in Denmark for cooperation regarding the data registration in Danish national registries.
16.
Compliance with ethical standards The study was approved by the Danish Data Protection Agency (Jr. number 2014-41-2803). Permission for using data from the DHFD, including comorbidity and vital status data, as well as data from the DTDB was obtained from Danish Data Protection Agency (Jr. number 2014-331-0769).
17.
Conflicts of interest None.
19.
References
20.
1.
2.
3.
4.
5.
6. 7.
8.
9.
10. 11.
12.
Kanis JA, Oden A, McCloskey EV, Johansson H, Wahl DA, Cooper C (2012) A systematic review of hip fracture incidence and probability of fracture worldwide. Osteoporos Int 23:2239–2256 (2015) Danish Hip Fracture Database. Annual report 2014. Available on https://www.sundhed.dk/sundhedsfaglig/kvalitet/ kliniskekvalitetsdatabaser/akutte-sygdomme/hoftebrud-nip. Accessed 13 April 2016 Vestergaard P, Rejnmark L, Mosekilde L (2007) Increased mortality in patients with a hip fracture-effect of pre-morbid conditions and post-fracture complications. Osteoporos Int 18:1583–1593 Hasley PB, Lave JR, Hanusa BH, Arena VC, Ramsey G, Kapoor WN, Fine MJ (1995) Variation in the use of red blood cell transfusions. A study of four common medical and surgical conditions. Med Care 33:1145–1160 Hutton B, Fergusson D, Tinmouth A, McIntyre L, Kmetic A, Hebert PC (2005) Transfusion rates vary significantly amongst Canadian medical centres. Can J Anaesth 52:581–590 The Sanguis Study Group (1994) Use of blood products for elective surgery in 43 European hospitals. Transfus Med 4:251–268 (2012) Danish Transfusion Database. Annual report. Available on https:// www.sundhed.dk/sundhedsfaglig/kvalitet/kliniskekvalitetsdatabaser/ specifikke-procedurer/transfusionsdatabase/. Accessed 13 April 2016 Kadar A, Chechik O, Steinberg E, Reider E, Sternheim A (2013) Predicting the need for blood transfusion in patients with hip fractures. Int Orthop 37:693–700 Madsen CM, Jorgensen HL, Norgaard A, Riis T, Jantzen C, Pedersen OB, Duus BR, Lauritzen JB (2014) Preoperative factors associated with red blood cell transfusion in hip fracture patients. Arch Orthop Trauma Surg 134:375–382 Swain DG, Nightingale PG, Patel JV (2000) Blood transfusion requirements in femoral neck fracture. Injury 31:7–10 Nielsen KA, Jensen NC, Jensen CM, Thomsen M, Pedersen L, Johnsen SP, Ingeman A, Bartels PD, Thomsen RW (2009) Quality of care and 30 day mortality among patients with hip fractures: a nationwide cohort study. BMC Health Serv Res 9:186 Koch CG, Li L, Duncan AI, Mihaljevic T, Cosgrove DM, Loop FD, Starr NJ, Blackstone EH (2006) Morbidity and mortality risk associated with red blood cell and blood-component transfusion in isolated coronary artery bypass grafting. Crit Care Med 34:1608–1616
18.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
Pedersen AB, Mehnert F, Overgaard S, Johnsen SP (2009) Allogeneic blood transfusion and prognosis following total hip replacement: a population-based follow up study. BMC Musculoskelet Disord 10:167 Escalante A, Haas RW, del Rincón I (2005) Paradoxical effect of body mass index on survival in rheumatoid arthritis: role of comorbidity and systemic inflammation. Arch Intern Med 165:1624–1629 Park J, Ahmadi SF, Streja E, Molnar MZ, Flegal KM, Gillen D, Kovesdy CP, Kalantar-Zadeh K (2014) Obesity paradox in endstage kidney disease patients. Prog Cardiovasc Dis 56:415–425 Vemmos K, Ntaios G, Spengos K, Savvari P, Vemmou A, Pappa T, Manios E, Georgiopoulos G, Alevizaki M (2011) Association between obesity and mortality after acute first-ever stroke: the obesitystroke paradox. Stroke 42:30–36 Schmidt M, Pedersen L, Sorensen HT (2014) The Danish Civil Registration System as a tool in epidemiology. Eur J Epidemiol 29:541–549 Mainz J, Krog BR, Bjornshave B, Bartels P (2004) Nationwide continuous quality improvement using clinical indicators: the Danish National Indicator Project. Int J Qual Health Care 16(Suppl 1):i45–i50 Andersen TF, Madsen M, Jorgensen J, Mellemkjoer L, Olsen JH (1999) The Danish National Hospital Register. A valuable source of data for modern health sciences. Dan Med Bull 46:263–268 Johannesdottir SA, Horvath-Puho E, Ehrenstein V, Schmidt M, Pedersen L, Sorensen HT (2012) Existing data sources for clinical epidemiology: the Danish National Database of Reimbursed Prescriptions. J Clin Epidemiol 4:303–313 Charlson ME, Pompei P, Ales KL, MacKenzie CR (1987) A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 40:373–383 World Health Organization (2008) Worldwide prevalence of anaemia 1993–2005 (pdf). Geneva: World Health Organization. ISBN 978-92-4-159665-7. Accessed on 8 Mar 2016 Van Buuren S (2012) Flexibile imputation of missing data. Chapman & Hall/CRC. Interdisciplinary Statistics Series, Boca Raton Brunskill SJ, Millette SL, Shokoohi A, Pulford EC, Doree C, Murphy MF, Stanworth S (2015) Red blood cell transfusion for people undergoing hip fracture surgery. Cochrane Database Syst Rev 4, CD009699 Carson JL, Terrin ML, Noveck H, Sanders DW, Chaitman BR, Rhoads GG, Nemo G, Dragert K, Beaupre L, Hildebrand K, Macaulay W, Lewis C, Cook DR, Dobbin G, Zakriya KJ, Apple FS, Horney RA, Magaziner J (2011) Liberal or restrictive transfusion in high-risk patients after hip surgery. N Engl J Med 365:2453–2462 Gregersen M, Borris LC, Damsgaard EM (2015) Blood transfusion and overall quality of life after hip fracture in frail elderly patients— the transfusion requirements in frail elderly randomized controlled trial. J Am Med Dir Assoc 16:762–766 Bowditch MG, Villar RN (1999) Do obese patients bleed more? A prospective study of blood loss at total hip replacement. Ann R Coll Surg Engl 81:198–200 Carling MS, Jeppsson A, Eriksson BI, Brisby H (2015) Transfusions and blood loss in total hip and knee arthroplasty: a prospective observational study. J Orthop Surg Res 10:48 Jibodh SR, Gurkan I, Wenz JF (2004) In-hospital outcome and resource use in hip arthroplasty: influence of body mass. Orthopedics 27:594–601 Pola E, Papaleo P, Santoliquido A, Gasparini G, Aulisa L, De SE (2004) Clinical factors associated with an increased risk of perioperative blood transfusion in nonanemic patients undergoing total hip arthroplasty. J Bone Joint Surg Am 86-A:57–61 Hausman DB, Johnson MA, Davey A, Poon LW (2011) Body mass index is associated with dietary patterns and health conditions in Georgia centenarians. J Aging Res 2011:138015
Osteoporos Int 32.
Sundhedsstyrelsen (2015) Vejledning om blodtransfusion. Vejledning nr. 9038 af 15. januar 2015. https://sundhedsstyrelsen.dk/da/nyheder/ 2015/~/media/8147748CEC6140FF9112674B5EF1307B.ashx 33. Kvamme JM, Holmen J, Wilsgaard T, Florholmen J, Midthjell K, Jacobsen BK (2012) Body mass index and mortality in elderly men and women: the Tromso and HUNT studies. J Epidemiol Community Health 66:611–617 34. Weiss A, Beloosesky Y, Boaz M, Yalov A, Kornowski R, Grossman E (2008) Body mass index is inversely related to mortality in elderly subjects. J Gen Intern Med 23:19–24 35. Mahan KL, Escott SS, Raymond JL (2011) Krauses’s food & the nutrition care process. In: Mahan KL, Escott SS, Raymond JL (eds) Krauses’s food & the nutrition care process, 13th edn. Elsevier Saunders, St.Louis 36. Henegar JR, Bigler SA, Henegar LK, Tyagi SC, Hall JE (2001) Functional and structural changes in the kidney in the early stages of obesity. J Am Soc Nephrol 12:1211–1217
37.
Hsu CY, McCulloch CE, Iribarren C, Darbinian J, Go AS (2006) Body mass index and risk for end-stage renal disease. Ann Intern Med 144:21–28 38. Barba R, Zapatero A, Losa JE, Valdes V, Todoli JA, Di MP, Monreal M (2008) Body mass index and mortality in patients with acute venous thromboembolism: findings from the RIETE registry. J Thromb Haemost 6:595–600 39. Mokdad AH, Marks JS, Stroup DF, Gerberding JL (2004) Actual causes of death in the United States, 2000. JAMA 291:1238–1245 40. Flegal KM, Carroll MD, Kit BK, Ogden CL (2012) Prevalence of obesity and trends in the distribution of body mass index among US adults, 1999–2010. JAMA 307:491–497 41. Calle EE, Thun MJ, Petrelli JM, Rodriguez C, Heath CW Jr (1999) Body-mass index and mortality in a prospective cohort of U.S. adults. N Engl J Med 341:1097–1105