Assessment of routine pre-operative group and save testing in patients undergoing cholecystectomy: a retrospective cohort study

Introduction

Pre-operative evaluation is an opportunity to assess and mitigate a patient’s risk of undergoing anaesthesia and surgery. Blood tests are a common component of this assessment and may include group and save (G&S) testing, which identifies a patient’s red blood cell type and screens for irregular antibodies. It has been suggested that G&S testing is overperformed prior to laparoscopic cholecystectomy and a targeted approach should be adopted^1–14. Yet the United Kingdom (UK) lacks national guidance recommending which patient or procedural factors warrant pre-operative G&S testing.

Guidelines published in 2016 by the UK National Institute for Health and Care Excellence (NICE) on pre-operative testing prior to elective surgery stratified patients by their co-morbidities and the complexity of the proposed surgery¹⁵. The aim was to standardise and avoid unnecessary investigations in low- and intermediate-risk operations, however the committee excluded from the scope of the guidelines specific advice relating to G&S testing. Guidelines provided by the French Society of Anaesthesiology and Intensive Care advise if the risk of bleeding or blood transfusion is perceived to be “nil to low”, G&S testing is unnecessary, whereas if it is “intermediate to high”, these tests should be performed¹⁶. However, they do not give specific advice as to which patient and operative factors should be considered as part of this risk assessment.

Choosing Wisely, an international campaign to reduce unnecessary investigations, published recommendations from the Canadian Society of Transfusion Medicine that included specific advice to avoid routine G&S testing in patients undergoing cholecystectomy, as transfusion rates among these patients are low¹⁷. This advice is infrequently heeded: we recently published a systematic review of the available literature relating to G&S testing in patients who underwent cholecystectomy¹. Over 460,000 patients were included, 37.8% of whom underwent pre-operative G&S testing. Peri-operative transfusions occurred in only 2.1% of patients, and the authors of all the studies included in the review recommended a more targeted approach to pre-operative G&S testing^1–14.

The decision to perform G&S testing relies on the clinical judgement of staff running pre-operative assessment clinics. For fully crossmatched blood to be issued in our unit, patients must have one G&S sample taken within the preceding five days, plus a second sample that can be of any age on the patient’s electronic health records. Without clear guidance, inexperienced staff may err on the side of over-investigation¹⁸. Many patients will have two samples taken pre-operatively due to the erroneous perception that this would decrease the time for crossmatched blood to be issued in the unlikely event of rapid intra-operative haemorrhage. As Ghirardo and his colleagues pointed out, such an event would demand rapid transfusion with uncrossmatched blood products, during which time G&S testing can be performed and crossmatched blood obtained within twenty minutes⁶. Even if G&S samples were taken pre-operatively, the crossmatching of blood products would take the same amount of time in this scenario.

Purported pre-operative risk factors for peri-operative blood transfusion in the available literature include anaemia, regular oral anticoagulant use, cardiovascular co-morbidities, type 2 diabetes mellitus, primary haematological malignancies and increased American Society of Anesthesiologists (ASA) grade^5,6,9,11,14. None of these authors have assessed whether these associations are independent predictors of blood transfusion.

NICE and international guidelines support urgent cholecystectomy over a planned, delayed operation in those presenting with acute biliary pathology (especially cholecystitis and gallstone pancreatitis), as it reduces length of hospital stay, medical costs, number of workdays lost, and prevents recurrent admissions whilst on a waiting list^19–22. A large population-based study in Canada showed patients with acute cholecystitis who were discharged without early cholecystectomy had a 14% chance of acute hospital attendance with gallbladder pathology within 6 weeks or 29% within a year²³. Despite this, there is great variability in inter-hospital practice regarding the proportion of cholecystectomies performed on an emergency basis²⁴. Repeated episodes of acute biliary pathology could increase operative difficulty, and hence increase the risk of bleeding and need for blood transfusion.

This study audited our unit’s peri-operative blood transfusion and G&S testing rates in patients undergoing cholecystectomy. It aimed to identify independent pre-operative risk factors that may predict peri-operative blood transfusion in these patients. We also hypothesised that the odds of transfusion could be predicted by the number of prior emergency attendances to hospital with acute biliary pathology secondary to gallstones. To our knowledge, we are the first to explore this association. Armed with this, and with a greater understanding of local transfusion rates and the financial implications of routine G&S testing, we aim to improve selectivity of pre-operative G&S testing prior to this very common procedure.

Methods

Results

Our data have been made publicly available on Open Science Framework at https://osf.io/5uva4/²⁵.

Five of the 453 patients (1.1%) received a blood transfusion within 30 days of their operation. Table 1 summarises demographic characteristics of the patients, stratified by whether they received a blood transfusion. Those who received a transfusion tended to be more co-morbid, as measured by ASA grade (χ² (3) = 90.3, p < 0.001). One patient in each group had an underlying primary haematological malignancy predisposing to bleeding, which was found to be a statistically significant risk factor for receiving a blood transfusion (χ² (1) = 44.0, p < 0.001): the patient who received a transfusion had myelodysplasia characterized by chronic anaemia and thrombocytopaenia. Similar rates of cardiovascular co-morbidities (χ² (1) = 0.2, p = 0.630) and type 2 diabetes mellitus (χ² (1) = 0.4, p = 0.548) were seen between the two groups.

As would be expected, those patients requiring a blood transfusion had greater drops in post-operative haemoglobin (-36 g/l vs. -8 g/l, U = 77, p = 0.001), though the two groups started with similar pre-operative haemoglobin levels (132 g/l vs. 134 g/l, U = 957.5, p = 0.651). Blood transfusions were associated with longer inpatient stays (median 8 days vs. 1 day, U = 268, p = 0.002).

The patient characteristics and indications for transfusion for each of the five cases who received a blood transfusion are detailed in the Supplementary Table. None required emergency O negative blood intra-operatively or post-operatively. At the time of identification of the need for a blood transfusion, each patient was haemodynamically stable; as such, sufficient time would have been present for two G&S samples to be taken and for blood to be fully crossmatched prior to transfusion.

In 116 (25.6%) cases, no G&S samples were taken in the preceding five days of the operation. One hundred and seventy-nine (39.5%) patients had one G&S sample taken and 158 (34.9%) had two. 301 of the 404 patients with ASA grade I or II had at least one G&S sample taken. The cost of analysing a G&S sample in our trust is £11, excluding laboratory staffing costs. Within our cohort, analysis of pre-operative G&S samples therefore cost in the region of £3,800 per year in materials.

Patients who required a peri-operative blood transfusion had more prior emergency hospital attendances in the preceding three years with gallstone complications (median 4 attendances vs. 1 attendance, U = 186, p < 0.001). A logistic regression was performed to determine whether this association held independently of the effects of age, gender, urgency of operation, indication, ASA grade, presence of cardiovascular co-morbidities, type 2 diabetes mellitus, primary haematological malignancy, regular oral anticoagulant or antiplatelet use, or pre-operative haemoglobin level. The regression model was statistically significant (χ² (17) = 28.3, p = 0.042), explained 53.3% (Nagelkerke R²) of the variance in blood transfusion requirement and correctly classified 98.9% of cases. Following adjustment for the other co-variants, we found that with each additional emergency hospital attendance, the odds of requiring a blood transfusion increased by a factor of 4.6 (95% confidence interval 1.3 to 16.3, Wald χ² (1) = 5.5, p = 0.019). No other co-variants added significantly to the model.

ROC curve analysis showed an AUC of 0.92 (95% confidence interval 0.79 – 1.00) for number of prior emergency hospital attendances as a predictor of the need for post-operative blood transfusion (Figure 1). Three or more prior attendances predicted those patients who would require a post-operative blood transfusion with 60.0% sensitivity and 98.0% specificity.

Figure 1. Receiver Operating Characteristic (ROC) curve analysis of number of prior emergency hospital attendances and its impact on the requirement for post-operative blood transfusion.

Area under the curve 0.92 (95% confidence interval 0.79 – 1.00).

Discussion

To our knowledge, we are the first to demonstrate that the number of pre-operative emergency attendances with acute gallstone complications can be a strong independent predictor of the need for a post-operative blood transfusion following cholecystectomy. Within our cohort, three or more emergency hospital attendances with biliary pathology was highly predictive of this outcome. Indeed, this was the only pre-operative factor that we explored demonstrating a statistically significant independent association following logistic regression. It is likely that frequent flare-ups of biliary inflammation creates more local fibrosis and adhesions, setting up a technically more challenging procedure, as was found in Lo and colleagues’ randomised trial of early versus delayed cholecystectomies²⁶. Fibrosis and oedema in chronic cholecystitis destroy clear operative planes between the gallbladder and the liver bed, which may increase the risk of bleeding from the raw liver surface²⁷. Increased vascular fragility and bleeding may result from the vascular congestion and angiogenesis associated with chronic inflammatory processes^27,28. Four of the 5 patients who required a blood transfusion in our cohort developed post-operative subhepatic haematomas. It may be that these cases were technically challenging, and therefore achieving adequate haemostasis was more difficult; or, even if haemostasis were achieved initially, a high risk of rebleeding followed abdominal closure.

In 1995, Nassar and colleagues published a grading system classifying laparoscopic cholecystectomies according to intra-operative technical difficulty, taking into consideration the density of adhesions and features of the gallbladder and cystic pedicle that make for a more complex dissection²⁹. Their team have shown that patients who undergo emergency laparoscopic cholecystectomy are more likely to have a higher grade of difficulty if they have had multiple hospital admissions with acute biliary disease compared to those for operated on in their first admission, and that a higher Nassar scale grade is associated with a higher risk of bleeding^30,31.

In one case (Case 3, Supplementary Table), the need for post-operative blood transfusion was attributed to underlying primary haematological malignancy. Granted, the number of patients with this co-morbidity in our cohort is small, but its association with receiving a blood transfusion reached statistical significance, though our regression model did not show that this was an independent predictor of transfusion. Others have also found primary haematological cancers to be a risk factor for peri-operative transfusion: in an observational study of 4,462 patients in Scotland, this co-morbidity was present in 6 of the 48 (12.5%) cases who required a blood transfusion¹¹. A single-centre retrospective study of 1,167 patients undergoing cholecystectomy in the USA also found that one of their five patients who required a peri-operative blood transfusion had a diagnosis of chronic leukaemia with thrombocytopaenia⁶. We also found an association between higher ASA grade and need for post-operative transfusion in cholecystectomy, confirming findings by other groups, though this association also did not hold through our regression model^9,11.

Other cohort studies have found associations between pre-operative anaemia, obstructive jaundice as an indication for surgery, regular oral anticoagulant use, and underlying type 2 diabetes mellitus^5,6,9,11,14. Our study did not show these to be predictors of post-operative blood transfusion. This may be owing to the rarity of patients requiring blood transfusion in our cohort, and a more highly powered study may be required to confirm or refute these associations.

Our study also did not correlate presence of cardiovascular co-morbidities with the need for blood transfusion. Wan and colleagues noted that many post-cholecystectomy patients developed a drop in haemoglobin and haematocrit, even when intra-operative overt blood loss was negligible³². Seeking factors that could be attributable to this “hidden blood loss”, they found that patients with a history of controlled hypertension had over 130 millilitres’ greater estimated blood loss than those with no history of hypertension. The authors suggested that people with chronic hypertension may have more fragile vessels that predispose to post-operative bleeding. Relating these findings to our cohort, perhaps the increased blood loss, overt or hidden, associated with cardiovascular disease is insufficient to warrant a transfusion.

At 1.1%, peri-operative blood transfusion following cholecystectomy in our cohort is rare, and comparable to that in the available literature^1–14. In all of our cases, following identification of the need for peri-operative blood transfusion, there is usually enough time for two G&S samples to be taken, typed and fully crossmatched blood procured, which corroborates the findings of other cohort studies^5,6,12. In the rare cases of intra-operative major vascular injury, it would be detrimental to the patient to await crossmatched blood, and therefore even with two valid G&S samples available, rapid transfusion of O negative blood would be required^1,6,12,14. G&S testing pre-operatively is therefore not required in most cases^1–14.

There are important limitations that must be considered when interpreting the study findings. This study was retrospective in nature with an adequate sample size over a 19-month period. The main number of patients included is comparable to many other cohort studies investigating rates of peri-operative blood transfusion, fortunately the need for blood products following cholecystectomy is rare^2,8,9,13. Therefore, where associations between risk factors and blood transfusion are statistically insignificant, it can be difficult to refute a correlation that may not be seen due to underpowering of our study. Future prospective national studies, with a larger sample size, is required to make a further assessment of whether G&S routine testing is required and its associated safety.

Conclusions

Our study suggests that routine G&S testing prior to cholecystectomy, despite this being common practice in many units, may not be essential. A more selective approach to pre-operative G&S testing would be safe and avoid wasting resources. It can also be inferred that performing cholecystectomy on an urgent basis following emergency admission with acute biliary pathology should be considered where possible to avoid recurrent episodes, as the frequency of prior emergency hospital attendances with gallstone complications increases the risk of receiving post-operative blood transfusion. Further research is required to determine a specific patient selection criteria for G&S testing in other surgical specialties.