Eur J Orthop Surg Traumatol DOI 10.1007/s00590-015-1609-2
ORIGINAL ARTICLE • HIP - FRACTURES
Blood transfusion and risk of infection in frail elderly after hip fracture surgery: the TRIFE randomized controlled trial Merete Gregersen • Else Marie Damsgaard Lars Carl Borris
•
Received: 17 December 2014 / Accepted: 29 January 2015 Springer-Verlag France 2015
Abstract Objectives It is still under debate that red blood cell (RBC) transfusions might increase the risk of healthcareassociated infections after hip fracture surgery. Previously, we found that a liberal RBC transfusion strategy improved survival in nursing home residents. Our aim, therefore, was to investigate whether a more liberal RBC transfusion strategy was associated with a higher infection risk in frail elderly hip fracture patients. Design Prospective, assessor-blinded, randomized and controlled trial. Settings Orthopedic ward, Geriatric ward, and Hospitalat-home. Patients 284 consecutively hospital-admitted elderly with hip fracture from nursing homes or sheltered housing facilities were included. Intervention A restrictive RBC transfusion strategy (hemoglobin \9.7 g/dL; 6 mmol/L) compared with a liberal strategy (hemoglobin\11.3 g/dL; 7 mmol/L) administered within 30 days after surgery. Main outcome measurements Leukocytes and C-reactive protein (CRP) in repeated blood samples within 30 days, and number of all infections (pneumonia, urinary tract infection, and other infections) within 10 days. Results 88 % of the patients received a RBC transfusion. A median of 1 RBC unit (interquartile range (IQR): 1–2) was transfused for the restrictive strategy group versus 3 M. Gregersen (&) E. M. Damsgaard Department of Geriatrics, Aarhus University Hospital, ˚ rhus C, Denmark P.P. Oerumsgade 11, 8000 A e-mail: [email protected] L. C. Borris Department of Orthopaedics, Aarhus University Hospital, ˚ rhus C, Denmark Nørrebrogade 44, 8000 A
RBC units (IQR: 2–5) for the liberal group. Leukocytes and CRP measurements were similar for both groups. Rates of infection were 72 % for the restrictive group compared to 66 % for the liberal group (risk ratio 1.08; 95 % confidence interval 0.93–1.27, p value 0.29). Conclusions A more liberal RBC transfusion strategy was not associated with higher risk of infection among residents from nursing homes or sheltered housing undergoing hip fracture surgery. Keywords Infection Hip fracture Frail elderly Blood transfusion
Introduction Red blood cell (RBC) transfusion might reduce a recipient’s immune response and thereby increase the risk of infection [1]. Some studies have found that allogeneic RBC transfusions do increase the risk of postoperative bacterial infections, such as pneumonia, deep wound infection, urinary tract infection (UTI), and sepsis, compared to rates for non-transfused patients [1–5], whereas other studies have found no such association [6–9]. RBC transfusions are given to immediately correct anemia and operative hemorrhage. The World Health Organization (WHO) defines anemia as hemoglobin (Hb) concentrations \12 g/dL (7.5 mmol/L) for women, and \13 g/dL (8.1 mmol/L) for men [10]. In general, independent predictors of RBC transfusions are as follows: advanced age, dependency on activities of daily living (ADL), trauma, surgery, renal insufficiency, low body weight, presence of comorbidities, anemia at hospital admission, and cardiovascular disease (CVD) [7, 9, 11–14]. For hip fracture (HF) patients, the incidence of
123
Eur J Orthop Surg Traumatol
postoperative RBC transfusions varies between 48 and 69 % [14–16]. Postoperative infections remain a serious problem for surgical patients. The most frequent infection after HF surgery is UTI involving 26 % of elderly patients [4], most commonly women [17]. Pneumonia after HF is associated with death (18) and is the most common infection for men [17]. Previous randomized controlled trials (RCTs) and metaanalyses, with the aim to examine the association between RBC transfusion strategies and healthcare-associated infections, compare a restrictive Hb thresholds range from 7 to 9 g/dL (4.3 to 5.6 mmol/L) with a liberal thresholds range from 9 to 10 g/dL (5.6 to 6.2 mmol/L [19]. The most recent meta-analysis which includes 18 trials concludes that among hospitalized patients, a restrictive transfusion strategy is associated with a reduced risk of serious infections compared to a liberal strategy [20]. The populations span patients with low birth weight, cardiac disease, gastrointestinal bleeding, sepsis, sickle cell anemia, patients in critical care, and postpartum and orthopedic patients. However, in the frailest elderly HF patients, it seems that a restrictive transfusion strategy is fatal compared to a liberal strategy [21], and risk of infection is not examined solely in this population. Therefore, we hypothesized that for the frailest elderly HF patients, a restrictive RBC transfusion strategy compared to an even more liberal postoperative RBC transfusion strategy, than used in previous studies, is associated with infection.
Patients and methods Study design The ‘Transfusion Requirements In Frail Elderly (TRIFE) study’ of HF patients is a prospective, assessor-blinded, randomized and controlled trial. Patients were enrolled consecutively from an Orthopedic ward. Eligible patients (aged 65 ? years) were admitted for surgical repair of unilateral HFs with subsequent Hb levels \11.3 g/dL (7 mmol/L) during the first six postoperative days, with a home-address of nursing homes or sheltered housing facilities. Fractures were categorized as three subtypes (International Code of Disease (ICD 10) according to fracture line location: (1) femoral neck (ICD-10 S72.0); (2) intertrochanteric (ICD-10 S72.1); and (3) subtrochanteric (ICD-10 S72.2). Patients were excluded for any of the following reasons: active cancer, pathological HF, inability to understand or speak Danish without an interpreter, RBC transfusion refusal, fluid overload, irregular erythrocyte antibodies, or previous participation in the trial.
123
Ethics Cognitively unimpaired patients gave written informed consent. Consent was given for impaired patients by a next of kin and the relevant general practitioner. Cognitive capacity was assessed by the project manager at admission or at the latest on the day after surgery. Cognitive impairment was defined by: (1) a Mini Mental State Examination (MMSE) score \20; (2) a Confusion Assessment Method (CAM) indicating delirium; or (3) a clinical cognitive evaluation undertaken by two independent healthcare professionals. The trial was conducted according to the ethical principles of the Declaration of Helsinki. The protocol was approved by the Danish Data Protection Agency and Central Denmark Region Ethical Committee. The study is registered at ClinicalTrials.gov Identifier NCT01102010. Randomization and blinding After giving informed consent, patients with postoperative Hb levels between 9.7 g/dL (6 mmol/L) and 11.3 g/dL (7 mmol/L) were randomly assigned to either the restrictive strategy (Hb \ 9.7 g/dL; 6 mmol/L), which is recommended for elderly hip fracture patients in Denmark [22], or the liberal RBC transfusion strategy (Hb \ 11.3 g/dL; 7 mmol/L). The randomization procedure was provided by an allocation concealment process and conducted electronically in the web-based clinical-trial-support-system ‘TrialPartner’ from Public Health and Quality Improvement in Central Denmark Region. This central computer program using permuted block-sizes stratified the randomization according to gender and type of residence. Results of randomization were available at the electronic patient record for the hospital staff in the Orthopedic and Geriatric wards since the staff should administer the transfusions. The participants, their relatives, and the endpoint assessors were blinded to the result of randomization, and the patients’ Hb measures. Procedures A standard venous Hb analysis was done by cyanmethemoglobin assay (interlaboratory error: 0.2 g/dL) immediately upon admission and postoperatively. Hb, leukocytes, and CRP concentrations were measured daily during the first three postoperative days, at least once during the following four to 6 days, then at least once per week during the subsequent 3 weeks. Transfusions were performed, when necessary, according to group assignments, but not later than 24 h after the relevant Hb determination, and a further Hb measurement preceded each subsequent transfusion. No more than two units per day were administered.
Eur J Orthop Surg Traumatol
In Denmark, one packed RBC unit comprises approximately 300 mL with an erythrocyte concentration C1.65 g/dL per unit collected from one donor and is aged up to 5 weeks. Red cells are separated from whole blood and stored in a citrate-phosphate-dextrose-adenine anticoagulant solution without leukodepletion. The erythrocyte suspension consists of 100 mL SAG-M (sodium chloride, adenine, glucose, and mannitol). Leukocytes in the suspension are reduced to \1200 9 106. Plasma content is \20 mL and contains no functional thrombocytes or coagulation factors. All patients received preoperatively either 2000 mg intravenous dicloxacillin (a narrow-spectrum antibiotic of the penicillin class) or cefuroxime for patients allergic to penicillin. In our hospital, the usual length of hospital stay is 2 days among nursing home residents, due to their needs of returning home to rehabilitation in well-known surroundings to avoid delirium, and 10 days among sheltered housing residents who rehabilitate in a Geriatric ward. The last 10 years the Orthopedic and Geriatric departments have been cooperated (orthogeriatrics). A consultant geriatrician in the Orthopedic ward prescribes the assigned RBC transfusion and after discharge the same geriatrician, hereafter included in a multidisciplinary orthogeriatric team, is still responsible for further prescription of RBC transfusion in the patients’ residence as a standard procedure. The transfusion is accomplished by the nurse from the orthogeriatric team [23]. Before each RBC transfusion, full information is supplied mandatory concerning available blood typing, irregular erythrocyte antibodies, and cross-matching. The designated blood unit is then transferred from the blood bank to the hospital ward, or patient’s residence, and stored in an authorized cool-box until use. The duration of transfusion is approximately 1 h. The collective measurements of infection biomarkers (leukocytes and CRPs) in blood samples were registered together with the Hb measurements on days 3, 10, 17, 24 and 30, or until death. Infection was, according to WHO, defined as: ‘‘An infection occurring in a patient in a hospital or other healthcare facility in whom the infection was not present or incubating at the time of admission and includes infections acquired in the hospital but appearing after discharge’’ [24]. The participants were screened daily by the healthcare professionals for any symptom suggesting infection. UTI was verified by microscopy, pneumonia by stethoscopy and clinical assessment, and other infections including surgical wound infection were verified by clinical assessment. When an infection was assessed by the physician, antibiotic treatment was initiated and the incidence of infection was confirmed in the electronic patient record. The number of all infections during the first 10 postoperative days was counted and divided into three
categories: (1) pneumonia, (2) UTI, and (3) other infections. After completion of recruitment, the Comprehensive Geriatric Assessment (CGA) Frailty Index was used to uncover if the nursing home and the sheltered housing residents were equally frail [21, 25]. Statistical analyses In accordance with the original study design, it was determined that a sample size of 284 patients would provide 85 % power with alpha 0.05, to detect an absolute between-group difference in the primary outcome variable on recovery from physical disability (ClinicalTrials.gov, NCT01102010). Patient baseline characteristics and clinical outcomes were compared as follows: (1) intention-totreat (ITT), (2) randomized study group assignment, (3) residential group, and (4) category of RBC unit number. Continuous variables were compared using Student’s t test, categorical variables by Pearson’s Chi-squared test, and non-normal distributed variables with Wilcoxon ’s ranksum test. Comparisons of leukocytes and CRPs on the five occasions were performed to test for equal development over time, using both parallel curves and equal mean curves. As CRP data were not normally distributed a logarithmic transformation was conducted before the analysis of variance (with repeated measurement). Tests for parallel slopes and equal means across repeated measures were tested by Likelihood Ratios. Outcome measures for all infection incidences were dichotomized and presented with relative risk ratios. The number of pre- and intraoperative RBC units per patient was added to that of postoperative units to analyze a possible association between the total numbers of RBC units and risk of infection. The latter RBC units were then classified at four levels and analyzed by a logistic regression model. Postestimation was made by Hosmer–Lemeshow test. Outcome measures for per-protocol (PP) analyses were compared similarly. Results were expressed as risk ratios (RRs) and odds ratios (ORs) with 95 % confidence intervals (CIs), means with standard deviations (SDs), or medians with interquartile ranges (IQRs). A p value B 0.05 was considered statistically significant. Statistical analyses were performed with Stata software, version 13.0.
Results Between January 18, 2010, and June 6, 2013, the study enrolled a total of 284 patients (Fig. 1). Eight patients dropped out of the study and protocol deviations totalled 16 (eight per transfusion group). Thirty-three patients died
123
Eur J Orthop Surg Traumatol Fig. 1 Flow chart 393 Patients assessed for eligibility
89 Excluded 55 Hb concentration ≥11·4 g/dL 15 Active cancers 3 Fluid overloads 1 Irregular erythrocyte antibodies 2 No surgery 10 Died before or during surgery 1 Jehovah Witness 2 No Danish language
304 Screened for consent 20 Excluded (did not provide consent)
284 Randomised
140 Randomised to liberal transfusion strategy
144 Randomised to restrictive transfusion strategy
Drop-outs: 3 refuse RBC transfusion 1 acute bleeding ulcer
Drop-outs: 4 refuse RBC transfusion
Deviated from protocol: 8 missed 1-2 RBC units outside protocol threshold
Deviated from protocol: 8 received 1-3 RBC units outside protocol threshold
12 died
21 died
116 Included in per-protocol analysis
during the 30 days after surgery; hence, 227 patients remained for the per-protocol (PP) analysis. Baseline characteristics were well balanced between transfusion groups with the exception of mean age, which was lower in the restrictive group (p = 0.02) (Table 1). At randomization mean concentrations of Hb, leukocytes and CRPs were similar for the two groups. The median number
123
111 Included in per-protocol analysis
of RBC units transfused per patient in the restrictive transfusion group was 1 (IQR 1–2) versus 3 (IQR 2–5) for the liberal group. Among the restrictive group, 35 patients (24 %) received no transfusion after randomization, whereas all patients received one or more RBC transfusions in the liberal group. Of all the transfusions, 87 % was administered within the first 10 days after hip surgery.
Eur J Orthop Surg Traumatol 14
Table 1 Baseline characteristics Restrictive strategy (n = 144)
Liberal strategy (n = 140)
36 (25)
34 (24)
108 (75) 85.7 (6.9)
106 (76) 86.9 (9.8)
p value
Gender (%) Male Female Mean age, years (sd)
0.89 0.02
Mean leukocytes (10^9/l)
Characteristics
Liberal Restrictive
12
10
8
Residence (%) Nursing homes
89 (62)
90 (64)
Sheltered housing
55 (38)
50 (36)
Self-sufficient
59 (41)
63 (45)
Dependent
85 (59)
77 (55)
Low
33 (23)
39 (28)
Moderate
78 (54)
75 (54)
Severe
33 (23)
26 (18)
0
5
10
0.67
15
20
25
30
Postoperative days
Activity of daily living (%)
0.51
Dementia (%)
44 (31)
53 (37)
0.16
Underweight (%)
22 (15)
23 (16)
0.79
6 (4–9)
6 (4–9)
0.45
C-reactive protein (mg/L)
Comorbidity (%)
Prescribed medicine, no (IQR)
200
0.64
Liberal Restrictive
150
100
50
0 0
Blood measurements
5
10
15
20
25
30
Postoperative days
Hemoglobin, g/dL (sd)
10.4 (1.31)
10.3 (1.44)
0.99
C-reactive protein, mg/L (sd)
150 (68.7)
152 (64.7)
0.45
Leukocytes, 109/L (sd)
11.4 (6.75)
10.7 (3.01)
0.22
109 (76)
112 (80)
31 (22)
26 (19)
4 (3)
2 (1)
0.38
24 (16)
20 (14)
0.17
Fig. 2 Mean leukocyte concentrations and mean C-reactive protein (CRP) concentrations with 95 % confidence intervals for repeated measurements during 30 days of intervention with restrictive and liberal RBC transfusions strategies
Surgical treatment (%) Internal fixation Arthroplasty Girdlestone Pre- and intra-operative RBC Transfusions (%)
Repeated measurements of Hb concentrations showed that a mean level of 11.31 g/dL (95 % CI, 11.27–11.36) (7.0 mmol/L) was maintained for the restrictive transfusion group versus 12.25 g/dL (95 % CI, 12.20–12.30) (7.6 mmol/L) for the liberal group. A statistically significant higher proportion of nursing home residents (96 %) was assessed frail as compared to sheltered housing residents (78 %) (p \ 0.001). The mean curves of leukocyte counts of both groups were parallel over the 30 days (p = 0.17) and mean counts were equal (p = 0.62). The same situation applied to repeated measurements of CRP concentrations. The test showed parallel curves (p = 0.94) and equal means within the two transfusion groups (p = 0.86) (Fig. 2). The overall risk of a postoperative infection was similar for the two transfusion groups. No incidence difference was found among the types of infection incurred by
residents of either nursing homes or sheltered housing (Table 2). In total, 69 % of the patients in the study experienced at least one infection during the 10 days after HF surgery. The incidence of pneumonia was similar among both transfusion groups: restrictive 21 % versus liberal 20 % (p = 0. 86) and was independent on type of residence. Likewise, the UTI rate was similar among both transfusion groups: restrictive 53 % versus liberal 44 % (p = 0.15). There was no statistically significant difference between rates of other infections: restrictive 1.3 % versus liberal 2.1 % (p = 0.63). Similar results were found with the per-protocol analysis. RBC transfusions given pre- an intra-operatively were administered equally in the groups by 16 % in the restrictive transfusion group versus 14 % in the liberal group (p = 0.58). The numbers of these transfusions combined with the transfusions given during the intervention period showed that the number of RBC units transfused were not associated with higher infection probabilities (Table 3). The incidence of infection was similar for both transfused patients and non-transfused patients (p = 0.29). Pneumonia within 10 days was associated with 90-day mortality (RR 1.73; 95 % CI: 1.09–2.75, p = 0.02) and male gender (RR 1.86; 95 % CI: 1.18–2.95, p = 0.008).
123
Eur J Orthop Surg Traumatol Table 2 (Intention-to-treat) distribution of postoperative infections after restrictive or liberal RBC transfusion thresholds within 10 days after hip fracture surgery and presented as relative risk ratios with 95 % confidence intervals (CI) 10-day outcomes
Nursing home residents
Sheltered housing residents
Restrictive (n = 89)
Liberal (n = 90)
Risk ratio (95 % CI)
Restrictive (n = 55)
Liberal (n = 50)
Risk ratio (95 % CI)
No
25 (28)
34 (38)
1.00 (reference)
15 (27)
13 (26)
1.00 (reference)
Yes
64 (72)
56 (62)
1.15 (0.94–1.42)
40 (73)
37 (74)
0.98 (0.78–1.23)
No
71 (80)
72 (80)
1.00 (reference)
43 (78)
40 (80)
1.00 (reference)
Yes
18 (20)
18 (20)
1.01 (0.56–1.81)
12 (22)
10 (20)
1.09 (0.51–2.30)
No
43 (48)
53 (59)
1.00 (reference)
25 (45)
25 (50)
1.00 (reference)
Yes
46 (52)
37 (41)
1.25 (0.91–1.72)
30 (55)
25 (50)
1.09 (0.76–1.57)
No
87 (98)
87 (97)
1.00 (reference)
55 (100)
50 (100)
1.00 (reference)
Yes
2 (2)
3 (3)
0.67 (0.11–3.93)
0
0
–
Total infections (%)
Pneumonia (%)
UTIa (%)
Other infections
a
Urinary Tract Infection
Table 3 The probability of infections related to the number of intra- and postoperative red blood cell units regardless randomization in 284 frail elderly hip fracture patients analyzed by a logistic regression model and presented as odds ratios with 95 % confidence intervals (CI) 1–10 days after surgery
Red blood cell units None (n = 35)
1–2 (n = 121)
3–4 (n = 71)
[5 (n = 57)
Odds ratioa (95 % CI)
Pneumonia (%)
8 (23)
27 (22)
14 (20)
9 (16)
0.92 (0.80–1.05)
Urinary tract infection (%)
18 (51)
59 (49)
33 (46)
28 (49)
0.97 (0.88–1.08)
Other infections (%)
0
1 (1)
3 (4)
1 (1.5)
1.17 (0.87–1.61)
a
The odds ratio of the outcomes compared to the ordered exposure variable (blood units per patient)
UTI within 10 days was not associated with 90-day mortality (RR 0.85; 95 % CI: 0.63–1.15, p = 0.27), but was associated with female gender (RR 1.47; 95 % CI: 1.05–2.07, p = 0.01). Infections in total were not attributable to the surgical procedures, i.e., arthroplasty versus internal fixation (RR 0.89 (95 % CI: 0.72–1.09, p = 0.25). The causes of death within 30 days were due to sepsis (n = 5, 15 %), cardiovascular disease (n = 8, 24 %), pneumonia (n = 12, 36 %), delirium (n = 5, 15 %), and liver failure (n = 3, 9 %). No deaths were related to bleeding or complications to the RBC transfusions.
Discussion Among frail elderly patients with HF, we found that leukocytes and CRP biomarkers of infection remained similar within the 30 days following either the restrictive strategy (Hb \ 9.7 g/dL;6 mmol/L) or the liberal RBC transfusion strategy (Hb \ 11.3 g/dL;7 mmol/L). Also, postoperative infections which required antibiotic
123
treatment were not statistically significantly associated with either RBC transfusion per se or the number of RBC units transfused. A recent large randomized trial of RBC transfusion in high-risk patients after HF surgery (FOCUS trial) who walked without human assistance prior to HF and had either a clinical evidence of or risk factors for cardiovascular disease, found no association to serious adverse events, e.g., wound infection, myocardial infarction, and pneumonia [26]. Only 10 % of this study population was the nursing home residents. All analyses were made on the total population, and unfortunately, no subgroup analysis of the nursing home residents was performed and results of infection in only the frail population are missing. Our study is the first RCT that investigated the effect of RBC transfusions on specifically the frail elderly HF patients with high age, high rate of comorbidity, frequent state of dementia, and impaired walking ability. In the FOCUS trial, their liberal Hb target was similar to our restrictive target. Rohde and colleagues reported in their meta-analysis a reduced risk of serious infection with a restrictive RBC
Eur J Orthop Surg Traumatol
transfusion strategy (Hb threshold range: 6.4–9.7 g/dL) (4–6 mmol/L) compared to a liberal strategy (Hb threshold range: 9.0–11.3 g/dL) (5.6–7 mmol/L) [20]. They had included preliminary result from our TRIFE study, unfortunately using our hazard ratio with inverse outcome [27]. Another limit of this meta-analysis was that the included trials reported varied infectious outcomes, and some trials listed all infections, whereas others only reported specific types of infections. Several studies only include the outcome of infection during hospitalization, e.g., in the FOCUS trial, there was 4 days from randomization to discharge in US, and 12 days in Canada. Subsequently, the information on infections that occurred after randomization and within 30 days was obtained by telephone follow-up from patients or relatives. We agree with Carson that outcomes such as mortality, myocardial infarction, and function should be considered in the overall risk–benefit analysis of RBC transfusion [28]. In the nursing home residents, we have found that survival was improved by a liberal strategy [21], and patients with acute coronary syndrome too might improve survival from higher Hb levels [29]. Anemia is known to be associated with an increased risk of nosocomial infections [4]. Postoperative Hb concentrations \ 10 g/dL are associated with complications such as UTI and pneumonia [30], and pneumonia is linked to death [18]. Frailty, too, can increase morbidity, disability, and mortality. Frail elderly patients have impaired immune responses [31, 32], and frequently succumb to infections, as shown in our study. More patients from nursing home were frail than the sheltered housing residents; however, it did not relate to the infection risk. The leukocyte counts in our serial blood samples showed a minor trend toward non-parallel curves over time, i.e., liberal transfusion patients showed higher mean leukocyte concentrations toward the end of the intervention period. An explanation could be that more patients survived in the liberal group, hence with the ITT analysis more survivors became exposed to infections during the intervention period thus increasing the leukocyte counts. Alternatively, leukocytes may have accumulated from the additional RBC units transfused.
Limitations Our sample size calculation was based on the primary objective of recovery from physical disability. A post hoc sample size calculation with 69 % power based on risk of all serious infections, combined, showed that 3075 patients were required in each group to prove no type II errors [20]. Unfortunately, it was impossible for us to recruit as many participants, and furthermore, it was unethically continuing
the recruitment of patients since the nursing home residents survived by the liberal transfusion strategy. Nevertheless, our analyses showed no tendency at all suggesting an increase of bacterial infections. The RCT design allowed us to evaluate the effects of each transfusion strategy since patients with potentially confounding health conditions, with the exception of age, were distributed equally between both treatment groups. Age did not modify or confound the effect of blood transfusion strategies on infection rate. No selection bias intruded since all assessments were performed daily in the Orthopedic ward throughout the entire study period.
Conclusion We conclude that more RBC transfusions according to the liberal strategy are not associated with infections following hip fracture surgery neither in nursing home residents nor in sheltered housing residents. Since the infection risk is not increased it may not avoid that we provide the most frail elderly life-saving blood transfusions. Acknowledgments We gratefully acknowledge the kind participation of patients, their relatives, and their general practitioner. We would also like to thank the contributions to this study of our colleagues at the Orthopedic ward and the Geriatric Ward, especially the Orthogeriatric Team. We have received grants from the Helga and Peter Korning Foundation for medical equipment (HemoCue portable photometer). Costs of data collection, analyses, and article writing were borne by the Department of Geriatrics at Aarhus University Hospital. Conflict of interest interest.
The authors declare that there is no conflict of
References 1. Marik PE, Corwin HL (2008) Efficacy of red blood cell transfusion in the critically ill: a systematic review of the literature. Crit Care Med 36:2667–2674 2. Carson JL, Altman DG, Duff A, Noveck H, Weinstein MP, Sonnenberg FA et al (1999) Risk of bacterial infection associated with allogeneic blood transfusion among patients undergoing hip fracture repair. Transfusion 39:694–700 3. Taylor RW, O’Brien J, Trottier SJ, Manganaro L, Cytron M, Lesko MF et al (2006) Red blood cell transfusions and nosocomial infections in critically ill patients. Crit Care Med 34:2302–2308 4. Izuel Rami M, Garcia Erce JA, Gomez-Barrera M, Cuenca Espierrez J, Abad Sazatornil R, Rabanaque Hernandez MJ (2008) Relationship between allogeneic blood transfusion, iron deficiency and nosocomial infection in patients with hip fracture. Med Clin (Barc) 131:647–652 5. Weber WP, Zwahlen M, Reck S, Misteli H, Rosenthal R, Buser AS et al (2009) The association of preoperative anemia and perioperative allogeneic blood transfusion with the risk of surgical site infection. Transfusion 49:1964–1970
123
Eur J Orthop Surg Traumatol 6. Villanueva C, Colomo A, Bosch A, Concepcion M, HernandezGea V, Aracil C et al (2013) Transfusion strategies for acute upper gastrointestinal bleeding. N Engl J Med 368:11–21 7. Johnston P, Wynn-Jones H, Chakravarty D, Boyle A, Parker MJ (2006) Is perioperative blood transfusion a risk factor for mortality or infection after hip fracture? J Orthop Trauma 20:675–679 8. Dillon MF, Collins D, Rice J, Murphy PG, Nicholson P, Mac Elwaine J (2005) Preoperative characteristics identify patients with hip fractures at risk of transfusion. Clin Orthop Relat Res 439:201–206 9. Ali ZA, Lim E, Motalleb-Zadeh R, Ali AA, Callaghan CJ, Gerrard C et al (2004) Allogenic blood transfusion does not predispose to infection after cardiac surgery. Ann Thorac Surg 78:1542–1546 10. WHO (2011) Haemoglobin concentrations for the diagnosis of anaemia and assessment of severity, www.who.int/vmnis/indica tors/haemoglobin. Accessed 2 April 2014 11. Barr PJ, Donnelly M, Cardwell C, Alam SS, Morris K, Parker M et al (2011) Drivers of transfusion decision making and quality of the evidence in orthopedic surgery: a systematic review of the literature. Transfus Med Rev 25:304–316 12. Halm EA, Wang JJ, Boockvar K, Penrod J, Silberzweig SB, Magaziner J et al (2004) The effect of perioperative anemia on clinical and functional outcomes in patients with hip fracture. J Orthop Trauma 18:369–374 13. Kumar D, Mbako AN, Riddick A, Patil S, Williams P (2011) On admission haemoglobin in patients with hip fracture. Injury 42:167–170 14. Adunsky A, Lichtenstein A, Mizrahi E, Arad M, Heim M (2003) Blood transfusion requirements in elderly hip fracture patients. Arch Gerontol Geriatr 36:75–81 15. Foss NB, Kristensen MT, Kehlet H (2008) Anaemia impedes functional mobility after hip fracture surgery. Age Ageing 37:173–178 16. Halm EA, Wang JJ, Boockvar K, Penrod J, Silberzweig SB, Magaziner J, Koval KJ, Siu AL (2003) Effects of blood transfusion on clinical and functional outcomes in patients with hip fracture. Transfusion 43:1358–1365 17. Petersen MB, Jorgensen HL, Hansen K, Duus BR (2006) Factors affecting postoperative mortality of patients with displaced femoral neck fracture. Injury 37:705–711 18. Berry SD, Samelson EJ, Bordes M, Broe K, Kiel DP (2009) Survival of aged nursing home residents with hip fracture. J Gerontol A Biol Sci Med Sci 64:771–777 19. Carson JL, Carless PA, Hebert PC (2012) Transfusion thresholds and other strategies for guiding allogeneic red blood cell transfusion. Cochrane Database Syst Rev 4:CD002042
123
20. Rohde JM, Dimcheff DE, Blumberg N, Saint S, Langa KM, Kuhn L et al (2014) Health care-associated infection after red blood cell transfusion: a systematic review and meta-analysis. JAMA 311:1317–1326 21. Gregersen M, Borris L, Damsgaard EM (2015) Postoperative blood transfusion strategy in frail anemic elderly with hip fracture: the TRIFE randomized controlled trial. Acta Orthop 14:1–10 22. Sundhedsstyrelsen. Vejledning nr. 10333 om blodtransfusion. 2008, http://www.sst.dk/Tilsyn/Transfusionsmedicin. Accessed 20 December 2009 23. Gregersen M, Zintchouk D, Borris LC, Damsgaard EM (2011) A geriatric multidisciplinary and tailor-made hospital-at-home method in nursing home residents with hip fracture. Geriatr Orthop Surg Rehabil 2:148–154 24. World Health Organisation (2010) International Statistical Classification of Diseases and Related Health Problems 10th Revision, http://apps.who.int/classifications/icd10/browse/2010/en#/ S70-S79. Accessed 20 December 2010 25. Kristjansson SR, Ronning B, Hurria A, Skovlund E, Jordhoy MS, Nesbakken A et al (2012) A comparison of two pre-operative frailty measures in older surgical cancer patients. J Geriatr Oncol 3:1–7 26. Carson JL, Terrin ML, Noveck H, Sanders DW, Chaitman BR, Rhoads GG et al (2011) Liberal or restrictive transfusion in highrisk patients after hip surgery. N Engl J Med 365:2453–2462 27. Gregersen M, Borris LC, Damsgaard EM (2014) Red blood cell transfusion strategies and health care-associated infection. JAMA—J Am Med Assoc 312:2041 28. Carson JL (2014) Blood transfusion and risk of infection: new convincing evidence. JAMA 311:1293–1294 29. Carson JL, Brooks MM, Abbott JD, Chaitman B, Kelsey SF, Triulzi DJ et al (2013) Liberal versus restrictive transfusion thresholds for patients with symptomatic coronary artery disease. Am Heart J 165:964–971 30. Bjorkelund KB, Hommel A, Thorngren KG, Lundberg D, Larsson S (2009) Factors at admission associated with 4 months outcome in elderly patients with hip fracture. AANA J 77:49–58 31. Ensrud KE, Ewing SK, Taylor BC, Fink HA, Stone KL, Cauley JA et al (2007) Frailty and risk of falls, fracture, and mortality in older women: the study of osteoporotic fractures. J Gerontol A Biol Sci Med Sci 62:744–751 32. Makary MA, Segev DL, Pronovost PJ, Syin D, Bandeen-Roche K, Patel P et al (2010) Frailty as a predictor of surgical outcomes in older patients. J Am Coll Surg 210:901–908
ORIGINAL ARTICLE • HIP - FRACTURES
Blood transfusion and risk of infection in frail elderly after hip fracture surgery: the TRIFE randomized controlled trial Merete Gregersen • Else Marie Damsgaard Lars Carl Borris
•
Received: 17 December 2014 / Accepted: 29 January 2015 Springer-Verlag France 2015
Abstract Objectives It is still under debate that red blood cell (RBC) transfusions might increase the risk of healthcareassociated infections after hip fracture surgery. Previously, we found that a liberal RBC transfusion strategy improved survival in nursing home residents. Our aim, therefore, was to investigate whether a more liberal RBC transfusion strategy was associated with a higher infection risk in frail elderly hip fracture patients. Design Prospective, assessor-blinded, randomized and controlled trial. Settings Orthopedic ward, Geriatric ward, and Hospitalat-home. Patients 284 consecutively hospital-admitted elderly with hip fracture from nursing homes or sheltered housing facilities were included. Intervention A restrictive RBC transfusion strategy (hemoglobin \9.7 g/dL; 6 mmol/L) compared with a liberal strategy (hemoglobin\11.3 g/dL; 7 mmol/L) administered within 30 days after surgery. Main outcome measurements Leukocytes and C-reactive protein (CRP) in repeated blood samples within 30 days, and number of all infections (pneumonia, urinary tract infection, and other infections) within 10 days. Results 88 % of the patients received a RBC transfusion. A median of 1 RBC unit (interquartile range (IQR): 1–2) was transfused for the restrictive strategy group versus 3 M. Gregersen (&) E. M. Damsgaard Department of Geriatrics, Aarhus University Hospital, ˚ rhus C, Denmark P.P. Oerumsgade 11, 8000 A e-mail: [email protected] L. C. Borris Department of Orthopaedics, Aarhus University Hospital, ˚ rhus C, Denmark Nørrebrogade 44, 8000 A
RBC units (IQR: 2–5) for the liberal group. Leukocytes and CRP measurements were similar for both groups. Rates of infection were 72 % for the restrictive group compared to 66 % for the liberal group (risk ratio 1.08; 95 % confidence interval 0.93–1.27, p value 0.29). Conclusions A more liberal RBC transfusion strategy was not associated with higher risk of infection among residents from nursing homes or sheltered housing undergoing hip fracture surgery. Keywords Infection Hip fracture Frail elderly Blood transfusion
Introduction Red blood cell (RBC) transfusion might reduce a recipient’s immune response and thereby increase the risk of infection [1]. Some studies have found that allogeneic RBC transfusions do increase the risk of postoperative bacterial infections, such as pneumonia, deep wound infection, urinary tract infection (UTI), and sepsis, compared to rates for non-transfused patients [1–5], whereas other studies have found no such association [6–9]. RBC transfusions are given to immediately correct anemia and operative hemorrhage. The World Health Organization (WHO) defines anemia as hemoglobin (Hb) concentrations \12 g/dL (7.5 mmol/L) for women, and \13 g/dL (8.1 mmol/L) for men [10]. In general, independent predictors of RBC transfusions are as follows: advanced age, dependency on activities of daily living (ADL), trauma, surgery, renal insufficiency, low body weight, presence of comorbidities, anemia at hospital admission, and cardiovascular disease (CVD) [7, 9, 11–14]. For hip fracture (HF) patients, the incidence of
123
Eur J Orthop Surg Traumatol
postoperative RBC transfusions varies between 48 and 69 % [14–16]. Postoperative infections remain a serious problem for surgical patients. The most frequent infection after HF surgery is UTI involving 26 % of elderly patients [4], most commonly women [17]. Pneumonia after HF is associated with death (18) and is the most common infection for men [17]. Previous randomized controlled trials (RCTs) and metaanalyses, with the aim to examine the association between RBC transfusion strategies and healthcare-associated infections, compare a restrictive Hb thresholds range from 7 to 9 g/dL (4.3 to 5.6 mmol/L) with a liberal thresholds range from 9 to 10 g/dL (5.6 to 6.2 mmol/L [19]. The most recent meta-analysis which includes 18 trials concludes that among hospitalized patients, a restrictive transfusion strategy is associated with a reduced risk of serious infections compared to a liberal strategy [20]. The populations span patients with low birth weight, cardiac disease, gastrointestinal bleeding, sepsis, sickle cell anemia, patients in critical care, and postpartum and orthopedic patients. However, in the frailest elderly HF patients, it seems that a restrictive transfusion strategy is fatal compared to a liberal strategy [21], and risk of infection is not examined solely in this population. Therefore, we hypothesized that for the frailest elderly HF patients, a restrictive RBC transfusion strategy compared to an even more liberal postoperative RBC transfusion strategy, than used in previous studies, is associated with infection.
Patients and methods Study design The ‘Transfusion Requirements In Frail Elderly (TRIFE) study’ of HF patients is a prospective, assessor-blinded, randomized and controlled trial. Patients were enrolled consecutively from an Orthopedic ward. Eligible patients (aged 65 ? years) were admitted for surgical repair of unilateral HFs with subsequent Hb levels \11.3 g/dL (7 mmol/L) during the first six postoperative days, with a home-address of nursing homes or sheltered housing facilities. Fractures were categorized as three subtypes (International Code of Disease (ICD 10) according to fracture line location: (1) femoral neck (ICD-10 S72.0); (2) intertrochanteric (ICD-10 S72.1); and (3) subtrochanteric (ICD-10 S72.2). Patients were excluded for any of the following reasons: active cancer, pathological HF, inability to understand or speak Danish without an interpreter, RBC transfusion refusal, fluid overload, irregular erythrocyte antibodies, or previous participation in the trial.
123
Ethics Cognitively unimpaired patients gave written informed consent. Consent was given for impaired patients by a next of kin and the relevant general practitioner. Cognitive capacity was assessed by the project manager at admission or at the latest on the day after surgery. Cognitive impairment was defined by: (1) a Mini Mental State Examination (MMSE) score \20; (2) a Confusion Assessment Method (CAM) indicating delirium; or (3) a clinical cognitive evaluation undertaken by two independent healthcare professionals. The trial was conducted according to the ethical principles of the Declaration of Helsinki. The protocol was approved by the Danish Data Protection Agency and Central Denmark Region Ethical Committee. The study is registered at ClinicalTrials.gov Identifier NCT01102010. Randomization and blinding After giving informed consent, patients with postoperative Hb levels between 9.7 g/dL (6 mmol/L) and 11.3 g/dL (7 mmol/L) were randomly assigned to either the restrictive strategy (Hb \ 9.7 g/dL; 6 mmol/L), which is recommended for elderly hip fracture patients in Denmark [22], or the liberal RBC transfusion strategy (Hb \ 11.3 g/dL; 7 mmol/L). The randomization procedure was provided by an allocation concealment process and conducted electronically in the web-based clinical-trial-support-system ‘TrialPartner’ from Public Health and Quality Improvement in Central Denmark Region. This central computer program using permuted block-sizes stratified the randomization according to gender and type of residence. Results of randomization were available at the electronic patient record for the hospital staff in the Orthopedic and Geriatric wards since the staff should administer the transfusions. The participants, their relatives, and the endpoint assessors were blinded to the result of randomization, and the patients’ Hb measures. Procedures A standard venous Hb analysis was done by cyanmethemoglobin assay (interlaboratory error: 0.2 g/dL) immediately upon admission and postoperatively. Hb, leukocytes, and CRP concentrations were measured daily during the first three postoperative days, at least once during the following four to 6 days, then at least once per week during the subsequent 3 weeks. Transfusions were performed, when necessary, according to group assignments, but not later than 24 h after the relevant Hb determination, and a further Hb measurement preceded each subsequent transfusion. No more than two units per day were administered.
Eur J Orthop Surg Traumatol
In Denmark, one packed RBC unit comprises approximately 300 mL with an erythrocyte concentration C1.65 g/dL per unit collected from one donor and is aged up to 5 weeks. Red cells are separated from whole blood and stored in a citrate-phosphate-dextrose-adenine anticoagulant solution without leukodepletion. The erythrocyte suspension consists of 100 mL SAG-M (sodium chloride, adenine, glucose, and mannitol). Leukocytes in the suspension are reduced to \1200 9 106. Plasma content is \20 mL and contains no functional thrombocytes or coagulation factors. All patients received preoperatively either 2000 mg intravenous dicloxacillin (a narrow-spectrum antibiotic of the penicillin class) or cefuroxime for patients allergic to penicillin. In our hospital, the usual length of hospital stay is 2 days among nursing home residents, due to their needs of returning home to rehabilitation in well-known surroundings to avoid delirium, and 10 days among sheltered housing residents who rehabilitate in a Geriatric ward. The last 10 years the Orthopedic and Geriatric departments have been cooperated (orthogeriatrics). A consultant geriatrician in the Orthopedic ward prescribes the assigned RBC transfusion and after discharge the same geriatrician, hereafter included in a multidisciplinary orthogeriatric team, is still responsible for further prescription of RBC transfusion in the patients’ residence as a standard procedure. The transfusion is accomplished by the nurse from the orthogeriatric team [23]. Before each RBC transfusion, full information is supplied mandatory concerning available blood typing, irregular erythrocyte antibodies, and cross-matching. The designated blood unit is then transferred from the blood bank to the hospital ward, or patient’s residence, and stored in an authorized cool-box until use. The duration of transfusion is approximately 1 h. The collective measurements of infection biomarkers (leukocytes and CRPs) in blood samples were registered together with the Hb measurements on days 3, 10, 17, 24 and 30, or until death. Infection was, according to WHO, defined as: ‘‘An infection occurring in a patient in a hospital or other healthcare facility in whom the infection was not present or incubating at the time of admission and includes infections acquired in the hospital but appearing after discharge’’ [24]. The participants were screened daily by the healthcare professionals for any symptom suggesting infection. UTI was verified by microscopy, pneumonia by stethoscopy and clinical assessment, and other infections including surgical wound infection were verified by clinical assessment. When an infection was assessed by the physician, antibiotic treatment was initiated and the incidence of infection was confirmed in the electronic patient record. The number of all infections during the first 10 postoperative days was counted and divided into three
categories: (1) pneumonia, (2) UTI, and (3) other infections. After completion of recruitment, the Comprehensive Geriatric Assessment (CGA) Frailty Index was used to uncover if the nursing home and the sheltered housing residents were equally frail [21, 25]. Statistical analyses In accordance with the original study design, it was determined that a sample size of 284 patients would provide 85 % power with alpha 0.05, to detect an absolute between-group difference in the primary outcome variable on recovery from physical disability (ClinicalTrials.gov, NCT01102010). Patient baseline characteristics and clinical outcomes were compared as follows: (1) intention-totreat (ITT), (2) randomized study group assignment, (3) residential group, and (4) category of RBC unit number. Continuous variables were compared using Student’s t test, categorical variables by Pearson’s Chi-squared test, and non-normal distributed variables with Wilcoxon ’s ranksum test. Comparisons of leukocytes and CRPs on the five occasions were performed to test for equal development over time, using both parallel curves and equal mean curves. As CRP data were not normally distributed a logarithmic transformation was conducted before the analysis of variance (with repeated measurement). Tests for parallel slopes and equal means across repeated measures were tested by Likelihood Ratios. Outcome measures for all infection incidences were dichotomized and presented with relative risk ratios. The number of pre- and intraoperative RBC units per patient was added to that of postoperative units to analyze a possible association between the total numbers of RBC units and risk of infection. The latter RBC units were then classified at four levels and analyzed by a logistic regression model. Postestimation was made by Hosmer–Lemeshow test. Outcome measures for per-protocol (PP) analyses were compared similarly. Results were expressed as risk ratios (RRs) and odds ratios (ORs) with 95 % confidence intervals (CIs), means with standard deviations (SDs), or medians with interquartile ranges (IQRs). A p value B 0.05 was considered statistically significant. Statistical analyses were performed with Stata software, version 13.0.
Results Between January 18, 2010, and June 6, 2013, the study enrolled a total of 284 patients (Fig. 1). Eight patients dropped out of the study and protocol deviations totalled 16 (eight per transfusion group). Thirty-three patients died
123
Eur J Orthop Surg Traumatol Fig. 1 Flow chart 393 Patients assessed for eligibility
89 Excluded 55 Hb concentration ≥11·4 g/dL 15 Active cancers 3 Fluid overloads 1 Irregular erythrocyte antibodies 2 No surgery 10 Died before or during surgery 1 Jehovah Witness 2 No Danish language
304 Screened for consent 20 Excluded (did not provide consent)
284 Randomised
140 Randomised to liberal transfusion strategy
144 Randomised to restrictive transfusion strategy
Drop-outs: 3 refuse RBC transfusion 1 acute bleeding ulcer
Drop-outs: 4 refuse RBC transfusion
Deviated from protocol: 8 missed 1-2 RBC units outside protocol threshold
Deviated from protocol: 8 received 1-3 RBC units outside protocol threshold
12 died
21 died
116 Included in per-protocol analysis
during the 30 days after surgery; hence, 227 patients remained for the per-protocol (PP) analysis. Baseline characteristics were well balanced between transfusion groups with the exception of mean age, which was lower in the restrictive group (p = 0.02) (Table 1). At randomization mean concentrations of Hb, leukocytes and CRPs were similar for the two groups. The median number
123
111 Included in per-protocol analysis
of RBC units transfused per patient in the restrictive transfusion group was 1 (IQR 1–2) versus 3 (IQR 2–5) for the liberal group. Among the restrictive group, 35 patients (24 %) received no transfusion after randomization, whereas all patients received one or more RBC transfusions in the liberal group. Of all the transfusions, 87 % was administered within the first 10 days after hip surgery.
Eur J Orthop Surg Traumatol 14
Table 1 Baseline characteristics Restrictive strategy (n = 144)
Liberal strategy (n = 140)
36 (25)
34 (24)
108 (75) 85.7 (6.9)
106 (76) 86.9 (9.8)
p value
Gender (%) Male Female Mean age, years (sd)
0.89 0.02
Mean leukocytes (10^9/l)
Characteristics
Liberal Restrictive
12
10
8
Residence (%) Nursing homes
89 (62)
90 (64)
Sheltered housing
55 (38)
50 (36)
Self-sufficient
59 (41)
63 (45)
Dependent
85 (59)
77 (55)
Low
33 (23)
39 (28)
Moderate
78 (54)
75 (54)
Severe
33 (23)
26 (18)
0
5
10
0.67
15
20
25
30
Postoperative days
Activity of daily living (%)
0.51
Dementia (%)
44 (31)
53 (37)
0.16
Underweight (%)
22 (15)
23 (16)
0.79
6 (4–9)
6 (4–9)
0.45
C-reactive protein (mg/L)
Comorbidity (%)
Prescribed medicine, no (IQR)
200
0.64
Liberal Restrictive
150
100
50
0 0
Blood measurements
5
10
15
20
25
30
Postoperative days
Hemoglobin, g/dL (sd)
10.4 (1.31)
10.3 (1.44)
0.99
C-reactive protein, mg/L (sd)
150 (68.7)
152 (64.7)
0.45
Leukocytes, 109/L (sd)
11.4 (6.75)
10.7 (3.01)
0.22
109 (76)
112 (80)
31 (22)
26 (19)
4 (3)
2 (1)
0.38
24 (16)
20 (14)
0.17
Fig. 2 Mean leukocyte concentrations and mean C-reactive protein (CRP) concentrations with 95 % confidence intervals for repeated measurements during 30 days of intervention with restrictive and liberal RBC transfusions strategies
Surgical treatment (%) Internal fixation Arthroplasty Girdlestone Pre- and intra-operative RBC Transfusions (%)
Repeated measurements of Hb concentrations showed that a mean level of 11.31 g/dL (95 % CI, 11.27–11.36) (7.0 mmol/L) was maintained for the restrictive transfusion group versus 12.25 g/dL (95 % CI, 12.20–12.30) (7.6 mmol/L) for the liberal group. A statistically significant higher proportion of nursing home residents (96 %) was assessed frail as compared to sheltered housing residents (78 %) (p \ 0.001). The mean curves of leukocyte counts of both groups were parallel over the 30 days (p = 0.17) and mean counts were equal (p = 0.62). The same situation applied to repeated measurements of CRP concentrations. The test showed parallel curves (p = 0.94) and equal means within the two transfusion groups (p = 0.86) (Fig. 2). The overall risk of a postoperative infection was similar for the two transfusion groups. No incidence difference was found among the types of infection incurred by
residents of either nursing homes or sheltered housing (Table 2). In total, 69 % of the patients in the study experienced at least one infection during the 10 days after HF surgery. The incidence of pneumonia was similar among both transfusion groups: restrictive 21 % versus liberal 20 % (p = 0. 86) and was independent on type of residence. Likewise, the UTI rate was similar among both transfusion groups: restrictive 53 % versus liberal 44 % (p = 0.15). There was no statistically significant difference between rates of other infections: restrictive 1.3 % versus liberal 2.1 % (p = 0.63). Similar results were found with the per-protocol analysis. RBC transfusions given pre- an intra-operatively were administered equally in the groups by 16 % in the restrictive transfusion group versus 14 % in the liberal group (p = 0.58). The numbers of these transfusions combined with the transfusions given during the intervention period showed that the number of RBC units transfused were not associated with higher infection probabilities (Table 3). The incidence of infection was similar for both transfused patients and non-transfused patients (p = 0.29). Pneumonia within 10 days was associated with 90-day mortality (RR 1.73; 95 % CI: 1.09–2.75, p = 0.02) and male gender (RR 1.86; 95 % CI: 1.18–2.95, p = 0.008).
123
Eur J Orthop Surg Traumatol Table 2 (Intention-to-treat) distribution of postoperative infections after restrictive or liberal RBC transfusion thresholds within 10 days after hip fracture surgery and presented as relative risk ratios with 95 % confidence intervals (CI) 10-day outcomes
Nursing home residents
Sheltered housing residents
Restrictive (n = 89)
Liberal (n = 90)
Risk ratio (95 % CI)
Restrictive (n = 55)
Liberal (n = 50)
Risk ratio (95 % CI)
No
25 (28)
34 (38)
1.00 (reference)
15 (27)
13 (26)
1.00 (reference)
Yes
64 (72)
56 (62)
1.15 (0.94–1.42)
40 (73)
37 (74)
0.98 (0.78–1.23)
No
71 (80)
72 (80)
1.00 (reference)
43 (78)
40 (80)
1.00 (reference)
Yes
18 (20)
18 (20)
1.01 (0.56–1.81)
12 (22)
10 (20)
1.09 (0.51–2.30)
No
43 (48)
53 (59)
1.00 (reference)
25 (45)
25 (50)
1.00 (reference)
Yes
46 (52)
37 (41)
1.25 (0.91–1.72)
30 (55)
25 (50)
1.09 (0.76–1.57)
No
87 (98)
87 (97)
1.00 (reference)
55 (100)
50 (100)
1.00 (reference)
Yes
2 (2)
3 (3)
0.67 (0.11–3.93)
0
0
–
Total infections (%)
Pneumonia (%)
UTIa (%)
Other infections
a
Urinary Tract Infection
Table 3 The probability of infections related to the number of intra- and postoperative red blood cell units regardless randomization in 284 frail elderly hip fracture patients analyzed by a logistic regression model and presented as odds ratios with 95 % confidence intervals (CI) 1–10 days after surgery
Red blood cell units None (n = 35)
1–2 (n = 121)
3–4 (n = 71)
[5 (n = 57)
Odds ratioa (95 % CI)
Pneumonia (%)
8 (23)
27 (22)
14 (20)
9 (16)
0.92 (0.80–1.05)
Urinary tract infection (%)
18 (51)
59 (49)
33 (46)
28 (49)
0.97 (0.88–1.08)
Other infections (%)
0
1 (1)
3 (4)
1 (1.5)
1.17 (0.87–1.61)
a
The odds ratio of the outcomes compared to the ordered exposure variable (blood units per patient)
UTI within 10 days was not associated with 90-day mortality (RR 0.85; 95 % CI: 0.63–1.15, p = 0.27), but was associated with female gender (RR 1.47; 95 % CI: 1.05–2.07, p = 0.01). Infections in total were not attributable to the surgical procedures, i.e., arthroplasty versus internal fixation (RR 0.89 (95 % CI: 0.72–1.09, p = 0.25). The causes of death within 30 days were due to sepsis (n = 5, 15 %), cardiovascular disease (n = 8, 24 %), pneumonia (n = 12, 36 %), delirium (n = 5, 15 %), and liver failure (n = 3, 9 %). No deaths were related to bleeding or complications to the RBC transfusions.
Discussion Among frail elderly patients with HF, we found that leukocytes and CRP biomarkers of infection remained similar within the 30 days following either the restrictive strategy (Hb \ 9.7 g/dL;6 mmol/L) or the liberal RBC transfusion strategy (Hb \ 11.3 g/dL;7 mmol/L). Also, postoperative infections which required antibiotic
123
treatment were not statistically significantly associated with either RBC transfusion per se or the number of RBC units transfused. A recent large randomized trial of RBC transfusion in high-risk patients after HF surgery (FOCUS trial) who walked without human assistance prior to HF and had either a clinical evidence of or risk factors for cardiovascular disease, found no association to serious adverse events, e.g., wound infection, myocardial infarction, and pneumonia [26]. Only 10 % of this study population was the nursing home residents. All analyses were made on the total population, and unfortunately, no subgroup analysis of the nursing home residents was performed and results of infection in only the frail population are missing. Our study is the first RCT that investigated the effect of RBC transfusions on specifically the frail elderly HF patients with high age, high rate of comorbidity, frequent state of dementia, and impaired walking ability. In the FOCUS trial, their liberal Hb target was similar to our restrictive target. Rohde and colleagues reported in their meta-analysis a reduced risk of serious infection with a restrictive RBC
Eur J Orthop Surg Traumatol
transfusion strategy (Hb threshold range: 6.4–9.7 g/dL) (4–6 mmol/L) compared to a liberal strategy (Hb threshold range: 9.0–11.3 g/dL) (5.6–7 mmol/L) [20]. They had included preliminary result from our TRIFE study, unfortunately using our hazard ratio with inverse outcome [27]. Another limit of this meta-analysis was that the included trials reported varied infectious outcomes, and some trials listed all infections, whereas others only reported specific types of infections. Several studies only include the outcome of infection during hospitalization, e.g., in the FOCUS trial, there was 4 days from randomization to discharge in US, and 12 days in Canada. Subsequently, the information on infections that occurred after randomization and within 30 days was obtained by telephone follow-up from patients or relatives. We agree with Carson that outcomes such as mortality, myocardial infarction, and function should be considered in the overall risk–benefit analysis of RBC transfusion [28]. In the nursing home residents, we have found that survival was improved by a liberal strategy [21], and patients with acute coronary syndrome too might improve survival from higher Hb levels [29]. Anemia is known to be associated with an increased risk of nosocomial infections [4]. Postoperative Hb concentrations \ 10 g/dL are associated with complications such as UTI and pneumonia [30], and pneumonia is linked to death [18]. Frailty, too, can increase morbidity, disability, and mortality. Frail elderly patients have impaired immune responses [31, 32], and frequently succumb to infections, as shown in our study. More patients from nursing home were frail than the sheltered housing residents; however, it did not relate to the infection risk. The leukocyte counts in our serial blood samples showed a minor trend toward non-parallel curves over time, i.e., liberal transfusion patients showed higher mean leukocyte concentrations toward the end of the intervention period. An explanation could be that more patients survived in the liberal group, hence with the ITT analysis more survivors became exposed to infections during the intervention period thus increasing the leukocyte counts. Alternatively, leukocytes may have accumulated from the additional RBC units transfused.
Limitations Our sample size calculation was based on the primary objective of recovery from physical disability. A post hoc sample size calculation with 69 % power based on risk of all serious infections, combined, showed that 3075 patients were required in each group to prove no type II errors [20]. Unfortunately, it was impossible for us to recruit as many participants, and furthermore, it was unethically continuing
the recruitment of patients since the nursing home residents survived by the liberal transfusion strategy. Nevertheless, our analyses showed no tendency at all suggesting an increase of bacterial infections. The RCT design allowed us to evaluate the effects of each transfusion strategy since patients with potentially confounding health conditions, with the exception of age, were distributed equally between both treatment groups. Age did not modify or confound the effect of blood transfusion strategies on infection rate. No selection bias intruded since all assessments were performed daily in the Orthopedic ward throughout the entire study period.
Conclusion We conclude that more RBC transfusions according to the liberal strategy are not associated with infections following hip fracture surgery neither in nursing home residents nor in sheltered housing residents. Since the infection risk is not increased it may not avoid that we provide the most frail elderly life-saving blood transfusions. Acknowledgments We gratefully acknowledge the kind participation of patients, their relatives, and their general practitioner. We would also like to thank the contributions to this study of our colleagues at the Orthopedic ward and the Geriatric Ward, especially the Orthogeriatric Team. We have received grants from the Helga and Peter Korning Foundation for medical equipment (HemoCue portable photometer). Costs of data collection, analyses, and article writing were borne by the Department of Geriatrics at Aarhus University Hospital. Conflict of interest interest.
The authors declare that there is no conflict of
References 1. Marik PE, Corwin HL (2008) Efficacy of red blood cell transfusion in the critically ill: a systematic review of the literature. Crit Care Med 36:2667–2674 2. Carson JL, Altman DG, Duff A, Noveck H, Weinstein MP, Sonnenberg FA et al (1999) Risk of bacterial infection associated with allogeneic blood transfusion among patients undergoing hip fracture repair. Transfusion 39:694–700 3. Taylor RW, O’Brien J, Trottier SJ, Manganaro L, Cytron M, Lesko MF et al (2006) Red blood cell transfusions and nosocomial infections in critically ill patients. Crit Care Med 34:2302–2308 4. Izuel Rami M, Garcia Erce JA, Gomez-Barrera M, Cuenca Espierrez J, Abad Sazatornil R, Rabanaque Hernandez MJ (2008) Relationship between allogeneic blood transfusion, iron deficiency and nosocomial infection in patients with hip fracture. Med Clin (Barc) 131:647–652 5. Weber WP, Zwahlen M, Reck S, Misteli H, Rosenthal R, Buser AS et al (2009) The association of preoperative anemia and perioperative allogeneic blood transfusion with the risk of surgical site infection. Transfusion 49:1964–1970
123
Eur J Orthop Surg Traumatol 6. Villanueva C, Colomo A, Bosch A, Concepcion M, HernandezGea V, Aracil C et al (2013) Transfusion strategies for acute upper gastrointestinal bleeding. N Engl J Med 368:11–21 7. Johnston P, Wynn-Jones H, Chakravarty D, Boyle A, Parker MJ (2006) Is perioperative blood transfusion a risk factor for mortality or infection after hip fracture? J Orthop Trauma 20:675–679 8. Dillon MF, Collins D, Rice J, Murphy PG, Nicholson P, Mac Elwaine J (2005) Preoperative characteristics identify patients with hip fractures at risk of transfusion. Clin Orthop Relat Res 439:201–206 9. Ali ZA, Lim E, Motalleb-Zadeh R, Ali AA, Callaghan CJ, Gerrard C et al (2004) Allogenic blood transfusion does not predispose to infection after cardiac surgery. Ann Thorac Surg 78:1542–1546 10. WHO (2011) Haemoglobin concentrations for the diagnosis of anaemia and assessment of severity, www.who.int/vmnis/indica tors/haemoglobin. Accessed 2 April 2014 11. Barr PJ, Donnelly M, Cardwell C, Alam SS, Morris K, Parker M et al (2011) Drivers of transfusion decision making and quality of the evidence in orthopedic surgery: a systematic review of the literature. Transfus Med Rev 25:304–316 12. Halm EA, Wang JJ, Boockvar K, Penrod J, Silberzweig SB, Magaziner J et al (2004) The effect of perioperative anemia on clinical and functional outcomes in patients with hip fracture. J Orthop Trauma 18:369–374 13. Kumar D, Mbako AN, Riddick A, Patil S, Williams P (2011) On admission haemoglobin in patients with hip fracture. Injury 42:167–170 14. Adunsky A, Lichtenstein A, Mizrahi E, Arad M, Heim M (2003) Blood transfusion requirements in elderly hip fracture patients. Arch Gerontol Geriatr 36:75–81 15. Foss NB, Kristensen MT, Kehlet H (2008) Anaemia impedes functional mobility after hip fracture surgery. Age Ageing 37:173–178 16. Halm EA, Wang JJ, Boockvar K, Penrod J, Silberzweig SB, Magaziner J, Koval KJ, Siu AL (2003) Effects of blood transfusion on clinical and functional outcomes in patients with hip fracture. Transfusion 43:1358–1365 17. Petersen MB, Jorgensen HL, Hansen K, Duus BR (2006) Factors affecting postoperative mortality of patients with displaced femoral neck fracture. Injury 37:705–711 18. Berry SD, Samelson EJ, Bordes M, Broe K, Kiel DP (2009) Survival of aged nursing home residents with hip fracture. J Gerontol A Biol Sci Med Sci 64:771–777 19. Carson JL, Carless PA, Hebert PC (2012) Transfusion thresholds and other strategies for guiding allogeneic red blood cell transfusion. Cochrane Database Syst Rev 4:CD002042
123
20. Rohde JM, Dimcheff DE, Blumberg N, Saint S, Langa KM, Kuhn L et al (2014) Health care-associated infection after red blood cell transfusion: a systematic review and meta-analysis. JAMA 311:1317–1326 21. Gregersen M, Borris L, Damsgaard EM (2015) Postoperative blood transfusion strategy in frail anemic elderly with hip fracture: the TRIFE randomized controlled trial. Acta Orthop 14:1–10 22. Sundhedsstyrelsen. Vejledning nr. 10333 om blodtransfusion. 2008, http://www.sst.dk/Tilsyn/Transfusionsmedicin. Accessed 20 December 2009 23. Gregersen M, Zintchouk D, Borris LC, Damsgaard EM (2011) A geriatric multidisciplinary and tailor-made hospital-at-home method in nursing home residents with hip fracture. Geriatr Orthop Surg Rehabil 2:148–154 24. World Health Organisation (2010) International Statistical Classification of Diseases and Related Health Problems 10th Revision, http://apps.who.int/classifications/icd10/browse/2010/en#/ S70-S79. Accessed 20 December 2010 25. Kristjansson SR, Ronning B, Hurria A, Skovlund E, Jordhoy MS, Nesbakken A et al (2012) A comparison of two pre-operative frailty measures in older surgical cancer patients. J Geriatr Oncol 3:1–7 26. Carson JL, Terrin ML, Noveck H, Sanders DW, Chaitman BR, Rhoads GG et al (2011) Liberal or restrictive transfusion in highrisk patients after hip surgery. N Engl J Med 365:2453–2462 27. Gregersen M, Borris LC, Damsgaard EM (2014) Red blood cell transfusion strategies and health care-associated infection. JAMA—J Am Med Assoc 312:2041 28. Carson JL (2014) Blood transfusion and risk of infection: new convincing evidence. JAMA 311:1293–1294 29. Carson JL, Brooks MM, Abbott JD, Chaitman B, Kelsey SF, Triulzi DJ et al (2013) Liberal versus restrictive transfusion thresholds for patients with symptomatic coronary artery disease. Am Heart J 165:964–971 30. Bjorkelund KB, Hommel A, Thorngren KG, Lundberg D, Larsson S (2009) Factors at admission associated with 4 months outcome in elderly patients with hip fracture. AANA J 77:49–58 31. Ensrud KE, Ewing SK, Taylor BC, Fink HA, Stone KL, Cauley JA et al (2007) Frailty and risk of falls, fracture, and mortality in older women: the study of osteoporotic fractures. J Gerontol A Biol Sci Med Sci 62:744–751 32. Makary MA, Segev DL, Pronovost PJ, Syin D, Bandeen-Roche K, Patel P et al (2010) Frailty as a predictor of surgical outcomes in older patients. J Am Coll Surg 210:901–908
