Blood transfusion requirements in elderly hip fracture patients

Article Outline

• Abstract
• 1. Introduction
• 2. Patients and methods
• 3. Results
• 4. Discussion
• References
• Copyright

Abstract 

The purpose of this study was to investigate blood transfusion patterns in elderly hip fractured patients and to determine the clinical predictive criteria for blood utilization. This retrospective study involved data analysis of 302 patients, undergoing surgical repair of pertrochanteric and subcapital fractures. Patients of the transfused group had significantly lower admission hemoglobin (P<0.0001) and significantly more pertrochanteric fractures. About 80% of patients of the lowermost quartile of admission hemoglobin were transfused, compared with <20% of the uppermost group (P<0.0001). Admission hemoglobin levels <12 g% and pertrochanteric fractures were independently associated with transfusions (odds ratio (O.R.) 0.475, C.I. 0.377–0.598, P<0.0001 and O.R. 1.885, C.I. 1.05–3.215, P=0.02, respectively). The results support the adoption of a policy considering primarily these two factors. Other factors we studied had no predictive power. Cross-matching may be reserved only for patients presenting with hemoglobin <12 g% and pertrochanteric fractures. This practice seems safe and should help in reducing the extent of unnecessary blood ordering.

Keywords:  Aged, Hip fracture, Blood transfusion

Back to Article Outline

1. Introduction 

Few studies have documented a rapidly increasing incidence of hip fractures, greater than would be expected from the increasing proportion of elderly people in the population alone. Such fractures constitute a leading cause of hospital admissions and length of stay among the elderly. Moreover, they are associated with functional impairments and major disabilities and result in high rates of institutionalization and mortality.

Surgery for hip fractures frequently requires blood transfusion, despite recent advances in techniques of orthopedic surgery and mechanic improvements of implants. Blood may be transfused before, during or following surgery. High rates of blood typing and cross-matching, as well as unnecessary transfusions raise concern of improper habits, rather than proper medical indications.

In the US, surgery for hip fractures ranks second in total number of units of blood administered to patients according to diagnoses groups (Friedman, 1979).

The indications for blood transfusion in elderly with hip fractures had not been standardized before. Blood is usually withdrawn for type and cross-matching, at the emergency room stage, and blood is routinely ordered. This happens without considering the various factors that may affect the need for blood transfusions, such as patient's age, admission hemoglobin, type of fracture, surgical procedure, etc. Previous studies concerning blood requirements in femoral neck fractures (Kurdy and Hokan, 1993, Swain et al., 2000) were confined mainly to type of fracture, rather than other potentially associated factors. Therefore, the objectives of the present study were to determine the demographic characteristics associated with transfusion of blood, and to study factors that may interfere with the decision to transfuse a patient. This would assist in evaluating the needs and assessing the appropriateness of blood transfusions.

Back to Article Outline

2. Patients and methods 

This is a retrospective chart review describing the patterns of blood transfusion in older patients with hip fractures admitted from the emergency ward to an orthogeriatric ward. The ward is a 30-bed unit utilizing an interdisciplinary team approach. The nature and characteristics of this orthogeriatric facility has already been described in details (Adunsky et al., 2002). We have included consecutive elderly hip fractured patients with pertrochanteric (extracapsular) fractures treated by plating and nailing, and subcapital (intracapsular) fractures, treated by hemiarthroplasty. We have excluded from the study patients with undisplaced subcapital fractures treated by pinning that have long been shown to be fractures with low level loss of blood, and transfusion of these patients is infrequent. All patients were 60 years or older. We have included only patients referred directly from the emergency ward of our medical center to the ward with primary inclusion criteria of a fractured hip. Patients admitted for elective hip surgery due to osteoarthritis, etc. were excluded as well as patients admitted for conservative non-surgical treatment. The presence of other acute disabilities (e.g. other fractures) or other acute medical problems (concurrent febrile disease, stroke, etc.) did not exclude the patients, nor did their cognitive level.

Data regarding age, gender, type of fractures, surgical procedures, ASA, duration of surgery, postoperative stability and complications were recorded. In addition, we have considered the time interval from admission to surgery (delay), total in-hospital length of stay. Admission and discharge values of hemoglobin and albumin were recorded, as well as the number of units of transfused blood.

Back to Article Outline

3. Results 

The data regarding a total number of 296 patients was included in the final analysis. Table 1 shows some of the general characteristics of the study population. Means of transfusion index (Ti, mean units of blood transfused for total number of patients) and actual transfusion index (ATi, mean units of blood transfused per number of actually transfused patients) are presented in Fig. 1. Values of these indexes were persistently above 0.5, indicating that a routine cross-match cannot be avoided.

Table 1. General characteristics of the study population

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 1. 

  Transfusion indexes.

Table 2 shows the data of 155 non-transfused compared with 141 transfused patients. Patients of the transfused group had a significantly lower admission hemoglobin (P<0.0001), yet, their mean hemoglobin value upon discharge was similar to that of the non-transfused group. Transfused patients had significantly more pertrochanteric fractures, nailing procedures, postoperative nonweight-bearing fractures and a significantly higher complication rate.

Table 2. Comparison of transfused and non-transfused groups

When the patients were analyzed according to fracture type (not shown), we found a significant difference with regards to admission hemoglobin (11.8 g% for pertrochanteric fractures, 12.6 g% for subcapital fractures, P<0.0001), length of stay (P=0.001) and admission albumin values (P=0.03). There were no differences as for other parameters such as age or gender.

To further characterize the patients, they were divided into quartiles, stratified by admission hemoglobin (Table 3). No difference was observed between the quartiles when age, gender, complication rates and other variables were tested. A significant trend was observed upon moving from lower to higher quartiles with respect to two parameters: (1) Diagnosis. There were statistically significant more patients with pertrochanteric fractures in the lowermost compared with the uppermost quartiles (P=0.001). (2) About 80% of patients of the lowermost quartile of admission hemoglobin quartile were transfused, compared with <20% of the uppermost group (P<0.0001, Fischer's exact test).

Table 3. Comparison of quartiles of patients, stratified by admission hemoglobin

Table 4 shows the multiple regression analysis for factors predicting the necessity for blood transfusions. The results show that none of the variables we tested proved predictive for transfusion, with the exception of admission hemoglobin levels when cutoff was set to 12 g% (O.R. 0.475, C.I. 0.377–0.598, P<0.0001) and pertrochanteric fractures (O.R. 1.885, C.I. 1.05–3.215, P=0.02).

Table 4. O.R. for blood transfusion derived from a multiple logistic regression analysis (in order of importance)

We have also calculated the ratio between number of cross-matches and the transfusions. This ratio serves as an indicator of blood ordering efficiency. Values above 2.0 (Roberts et al., 2000) or 2.5 (Rouault and Gruenhagen, 1978) are usually considered as indicator of over order of blood. The ratios in our study population were 1.27 for pertrochanteric fractures and 1.38 for subcapital ones.

Back to Article Outline

4. Discussion 

Fractures of the femoral necks frequently require the utilization of blood. However, patients with normal preoperative hemoglobin values may undergo the surgical procedure without blood transfusion. The early identification of such patients may save unnecessary blood cross-matching, which is done at the time of the patient's admission.

The results of the present study are instructive regarding some aspects of blood transfusion requirements in this population:

(a) Admission hemoglobin. Values <12 g% proved as best predictor for blood transfusion and approached a 5-fold risk for transfusion. It therefore seems that patients with normal (>12 g%) hemoglobin may undergo surgery after typing without cross-matching and/or transfusion, particularly when a subcapital fracture is involved. This is in accordance with previous observation (Robbins and Steingold, 1986) that the likelihood of using blood in patients with >11 g% hemoglobin is only 1 in 9.

(b) Type of fracture. In our series, patients with pertrochanteric fractures were transfused significantly more blood than the patients with subcapital fractures. Transfusion index were 1.09 and 0.72 for pertrocanteric and intracapsular fractures, respectively. Transfusion index of <0.5 usually indicates that a routine cross-matching may be avoided, as less than half a unit of blood is used per procedure. This means that there is a stronger case for cross-matching in pertrocanteric fractures. Really, the hazard risk for transfusion was 1.89 greater for pertrochanteric fractures (P=0.02). The combined odds for a patient with pertrochanteric fracture and admission hemoglobin <12 g% is 12-folds greater to be transfused. This probably results from the fact that the first group presented with lower hemoglobin, or alternatively from the fact that pertrochanteric fractures and associated procedures may involve a greater blood loss (Kurdy and Hokan, 1993) due to surgical procedures involving extensive cutting of the vastus lateralis.

(c) Age. Despite earlier reports regarding blood transfusion in relation to age >80, we could not ascertain this data, as well as a different need for blood with regards to sex. Swain reported a significant greater need for patients aged 80 years and above. Their observation could not be supported by our results as well as by previous studies. Patients >80 years old had a mean hemoglobin level of 12.00±1.75 g% compared with 12.39±1.51 g% in younger patients.

(d) Actual transfusion index (units per transfused patient). Each transfused patient received a mean of 1.99 units of blood. This value is slightly lower, as compared with previous publications reporting transfusion indexes of 2.09, 2.22 and 2.57. Possible explanations may not necessarily indicate better surgical skills, but rather a less aggressive correction of hemoglobin values. A study by Levi (1996) reported 19% decrease postoperative hemoglobin, as compared with a 9% decrease in our group. However, Levi's study measured the immediate postoperative hemoglobin while ours accounted for hemoglobin at discharge, so that no conclusions may be made.

To conclude, the decision-making process regarding blood transfusion in this population should consider primarily admission hemoglobin and type of fracture. Other factors such as age, gender, ASA, length of stay, etc. have no predictive power. Blood typing is mandatory, yet cross-matching and blood ordering may be reserved to patients presenting with hemoglobin <12 g% and pertrochanteric fractures.

A great concern has been paid to the cross-matching issue, since it bears important implications for the daily operation of blood banks. Unnecessary cross-matching is time wasting and expensive, causing blood waste by possibly exceeding the expiry date of blood units. It may also be hazardous by delaying other surgical procedures, thereby increasing morbidity and mortality. A British study (Robbins and Steingold, 1986) concluded that patients with normal preoperative hemoglobin levels can undergo operative treatment for fractures of the neck of the femur (other than pinning) after typing, but without the necessity of cross-matching blood. The results of their study, coupled with ours, mean that in about one third of cases the ‘force of habit’ still dictates an automatic request for blood typing and cross-matching that is unjustified, and should be substituted by the adoption of a more rationale approach.

Back to Article Outline

References 

1. Adunsky A, Levi R, Cecic A, Arad M, Noy S, Barell V. The Sheba model of comprehensive orthogeriatric care for elderly hip fracture patients: a preliminary report. Isr. Med. Assoc. J. 2002;4:259–261
   • View In Article
   • MEDLINE
2. Friedman BA. An analysis of surgical blood use in United States hospitals with application to the maximum surgical blood order schedule. Transfusion. 1979;19:268
   • View In Article
3. Kurdy NMG, Hokan R. A cross matching policy for fractures of the proximal third of the femur. Injury. 1993;24:521
   • View In Article
   • MEDLINE
   • CrossRef
4. Levi N. Blood transfusion requirements in intracapsular femoral neck fractures. Injury. 1996;27:709–711
   • View In Article
   • Abstract
   • Full-Text PDF (277 KB)
   • CrossRef
5. Roberts M, Ahya R, Greaves M, Maffulli N. A one-centre prospective audit of peri- and postoperative blood loss and transfusion practice in patients undergoing hip or nee replacement surgery. Ann. R. Surg. Engl. 2000;82:44–48
   • View In Article
6. Robbins J, Steingold RF. Blood use in urgent operation for patients with fractures of the femoral neck. Injury. 1986;17:265–266
   • View In Article
   • MEDLINE
   • CrossRef
7. Rouault C, Gruenhagen J. Reorganization of blood ordering practices. Transfusion. 1978;18:448–453
   • View In Article
8. Swain DG, Nightingale PG, Patel JV. Blood transfusion requirements in femoral neck fracture. Injury. 2000;31:7–10
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (102 KB)
   • CrossRef