Prospective double-blind randomized study of the effects of four intravenous fluids on platelet function and hemostasis in elective hip surgery


Professor Stan Heptinstall, Cardiovascular Medicine, Queen's Medical Center, Nottingham NG7 2UH, UK.
Tel.: +44 115 970- 9340; fax: + 44 115 970 9384; e-mail:


Summary.  A prospective randomized double-blind study was performed to determine the effects of three colloids, Haemaccel, Gelofusine and albumin, and also saline on platelet activation, platelet aggregation (induced by adenosine diphosphate (ADP), epinephrine, collagen) platelet agglutination by ristocetin and other hemostatic variables in 55 patients undergoing primary unilateral total hip replacement. The fluids were administered according to normal clinical practice and assessments were made immediately before, at the end, and 2 h after the end of surgery. Surgery was accompanied by thrombin generation (increases in thrombin/antithrombin III complex, prothrombin F1 +2 fragment) platelet activation (βTG) and compromised coagulation. Generally, the platelet activation appeared to result in platelet desensitization and brought about a persistent reduction in platelet aggregation to ADP and epinephrine, irrespective of the fluid used. Additionally, Haemaccel and Gelofusine inhibited ristocetin-induced platelet agglutination and albumin inhibited collagen-induced platelet aggregation. Gross inhibitory effects of Haemaccel that had been predicted from an earlier in vitro study did not occur. Particular fluids had selective additional effects on the hemostatic system. Albumin infusion served to maintain plasma albumin at normal concentrations postsurgery. The two gelatin preparations, Haemaccel and Gelofusine, maintained plasma viscosity. All three colloids led to a transient increase in activated partial thromboplastin time postsurgery and also a transient fall in the concentration of factor VIII, which were accompanied by a transient increase in bleeding time, but there was no measurable increase in blood loss. Inhibition of platelet aggregation by certain colloids may provide additional protection against the increased thrombotic risk in patients following major surgery.


Major surgery is always associated with administration of intravenous fluids to maintain blood volume, blood pressure and tissue perfusion. Several different types of fluid are used including crystalloids such as saline, or colloids such as Haemaccel, Gelofusine and albumin. There is continuing debate as to which fluids should be used in terms of their potential for enhancing blood loss and/or protection against postsurgical thrombosis [1–9]. The results of a previous study [8] performed wholly in vitro suggested that Haemaccel, in particular, might affect hemostasis and thrombosis through pronounced inhibition of platelet aggregation as a consequence of the high levels of ionized calcium present in the preparation (6.25 mmol L−1) [10]. We conducted a prospective randomized double-blind study to determine the effects of four different fluids administered during hip surgery on platelet aggregation and other hemostatic parameters. The effects of the fluids on bleeding time (BT) and blood loss were also monitored. Assessments were made immediately before surgery (Pre), at the end of surgery (Post) and 2 hours after the end of surgery (Last).

Patients and methods

Patients and procedures

After obtaining local ethics committee approval and informed consent, we randomized consecutive eligible patients undergoing primary unilateral cemented total hip replacement in a prospective randomized double-blind study to receive either albumin (4.5% w/v human serum albumin; BPL, Elstree, UK; n = 13), Gelofusine (4% w/v succinylated gelatin; B Braun Medical, Sheffield, UK; n = 14), Haemaccel (3.5% w/v polygeline with 6.25 mmol L−1 calcium; Beacon Pharmaceuticals, Tunbridge Wells, UK; n = 14), or saline (0.9% w/v sodium chloride; Baxter Health Care, Thetford, UK; n = 14). Randomization was achieved by the use of numbered envelopes. Data for one patient (randomized to albumin) were incomplete and excluded from the analysis.

Following randomization, the patient groups were as follows: albumin group (3M, 10F; median age 69.2 years; median bodyweight 66 kg); Gelofusine group (7M, 7F; median age 73.9 years; median bodyweight 75.8 kg); Haemaccel group (4M, 10F; median age 73.8 years; median bodyweight 77.8 kg); and saline group (11M, 3F; median age 69.1 years; median bodyweight 82.3 kg). There were no significant differences in ASA grade, initial ionized calcium and magnesium concentrations, serum albumin concentrations or hematocrit when classified by fluid group. Crosstabulation of sex, ABO blood group and Rhesus blood group showed no significant bias for either of the blood groups but a significant bias for sex.

Patients were not included in the study if there was a pre-existing defect in platelet function or coagulation, or if they were on warfarin or heparin therapy or on aspirin that could not be stopped for 2 weeks prior to the operation. No nonsteroidal agents were given on the morning of surgery, and low molecular weight heparin prophylaxis was given on the evening after surgery as was the normal practice in the Orthopaedic Unit. This was after all blood samples had been taken for the study reported here.

All patients received a standard general anesthetic comprising temazepam (10–20 mg) premedication and propofol (2.0–2.5 mg kg−1), and were maintained with isoflurane and nitrous oxide (1.0–2.5% isoflurane in nitrous oxide with 20–30% oxygen), to avoid effects of the anesthetic on BT and platelet function [11].

Immediately prior to surgery a dedicated 16G venous cannula was inserted and 56 mL blood taken into a polypropylene syringe and distributed as follows: 15 mL into a tube (Bibby Sterilin Ltd, Stone, UK) containing 150 µL recombinant hirudin (Revasc™, a gift from Novartis, Farnborough, UK, final concentration 50 µg mL−1) for platelet aggregometry; 2.7 mL into a tube containing EDTA (Sarstedt, Leicester, UK) for a full blood count (FBC); 4 mL into a tube (Sarstedt) containing 0.106 mol L−1 citrate (achieving a final dilution ratio 1 part citrate: 4 parts blood) for plasma viscosity (PV); 11 mL into tubes (Sarstedt) containing lithium–heparin; 9.4 mL into serum gel tubes (Sarstedt) containing kaolin-coated beads for urea and electrolytes, calcium, and albumin; 9 mL into tubes (Sarstedt) containing 0.106 mol L−1 citrate for coagulation tests; 4.5 mL into a CTAD tube containing a mixture of sodium citrate/citric acid supplemented with the platelet aggregation inhibitors theophylline, adenosine and dipyridamole (Becton Dickinson, Cowley, UK) for measurement of platelet factor 4 (PF4) and beta-thromboglobulin (βTG).

The citrate tubes and CTAD tubes were centrifuged at 2500 g for 30 min at 2 °C. Plasma was then aspirated and frozen. These were then stored at −80 °C. Ionized calcium and magnesium were measured from the lithium–heparin sample on an AVL 988–4 analyzer. The hirudinized sample was placed in a flask at 37 °C for transport and then in a water bath at the same temperature. Those for FBC and PV were kept at room temperature whilst the others were placed on ice.

The patient was transferred to the operating room and one arm placed on an arm board, with continuous monitoring of skin-surface temperature. Under the influence of an arm cuff inflated to 40 mmHg, a BT determination was performed using Simplate IIR device (Organon-Teknika, Cambridge, UK) [12], with the cuts arranged longitudinally on the forearm. This determination and blood sample constituted the Pre timepoint. The determination of BT was always performed by the same investigator.

Surgery then commenced and 2 L fluid infused during the operative period, the fluid choice of either saline, Haemaccel, Gelofusine or 4.5% human albumin being determined by the contents of the sealed envelope, and set up by an independent operator and covered with an opaque black bag. The amount of fluid used (2 L) corresponds to ATLS guidelines for fluid administration [13].

Repeat BT was performed and a further 56-mL blood removed from the cannula following skin closure (approximately 1 h after initiation of surgery). These determinations were the Post timepoint.

Two hours later, BT and blood sampling were repeated this being the Last timepoint. Following this final sampling, nonsteroidal analgesics and/or blood were given as required. Transfusion requirements were noted, as were the effects of these upon hemoglobin and hematocrit as assessed by the usual standard postoperative checks on full blood count.

Blood loss in theatre was determined by volume of blood and fluids in suction containers, blood on swabs, and volume of washing solutions used. The nursing staff on the wards, using a dedicated protocol sheet with set timepoints, noted postoperative blood loss into the drains. This was part of the usual recording of wound drainage, which had been ongoing for the previous year.

Platelet function

Platelet function tests began 30 min after sampling from the patient. Platelet aggregation/agglutination in whole blood was measured using the platelet-counting technique described by Fox et al. [14] and the fixation technique described by Bevan and Heptinstall [15]. Blood (480 µL) was added to LP3 polystyrene tubes (LIP Equipment and Services Ltd, Shipley, UK) containing 20 µL of varying concentrations of agonist, and the sample placed in a multisample agitator operating at 37 °C. At appropriate timepoints, 1 mL fixative solution was added. Platelet counting was performed using an Ultra-Flo100 Whole Blood Platelet Counter within 24 h. Percentage aggregation/agglutination was expressed as the percentage fall in the number of single platelets compared with the starting count. Results were expressed as response curves to increasing agonist dose. Mean values were plotted for each agonist concentration. The following agonist concentrations and timepoints were used: adenosine 5′ diphosphate (ADP) 0.3, 1.0, 3.0, 10 µmol L−1, reaction stopped at 2 min; collagen 0.25, 0.5, 1.0, 2.0, 4.0 µg mL−1, reaction stopped at 2 min; epinephrine 3 µmol L−1+0.1 µmol L−1 ADP, epinephrine 10 µmol L−1+0.1 µmol L−1 ADP, and epinephrine 30 µmol L−1+0.1 µmol L−1 ADP, reaction stopped at 4 min; ristocetin 0.6, 0.8, 1.0, 1.2, 1.4 mg mL−1, reaction stopped at 4 min.

Collagen was obtained from Nycomed (Munich, Germany). Ristocetin was obtained from Forum Products Ltd (Redhill, Surrey, UK), and ADP (as sodium salt) and epinephrine were obtained from the Sigma Chemical Co (Poole, Dorset, UK). The platelet fixative solution contained 150 mmol L−1 NaCl, 4.6 mmol L−1 Na2EDTA, 4.5 mmol L−1 Na2HPO4, 1.6 mmol L−1 KH2PO4 and 0.16% w/v formaldehyde, pH 7.4.

Platelet activation

Platelet factor 4 (PF4) and beta-thromboglobulin (βTG) were measured using, respectively, Asserachrom PF-4 and Asserachrom βTG enzyme immunoassay kits (Diagnostica Stago, Asnières, France). Absorbance for the enzyme immunoassays was measured using an Argus 300 microtitre plate reader.

Other hemostatic parameters

The activated partial thromboplastin time (APTT), prothrombin time (PT), factor (F)VIII and fibrinogen were measured on an MDA 180 analyzer (bioMèrieux, Basingstoke, UK). Total von Willebrand antigen was measured using a monoclonal antibody from Dako as part of an ELISA assay. FXIII was measured using an activity assay (Berichrom Dade Behring, Milton Keynes, UK) on a Cobas-Mira analyzer (ABX Diagnostics, Bedfordshire, UK). Thrombin/antithrombin III complex (TAT) and prothrombin F1 +2 fragment were measured using Enzygnost TAT micro and Enzygnost F1 +2 micro enzyme ELISA immunoassay kits (Dade Behring, Milton Keynes, UK).

Other analyses (albumin, full blood count and plasma viscosity) were measured routinely at the Leicester Royal Infirmary.

Statistical analysis

Platelet aggregation dose–response curves were compared by analysis of variance (anova) using SPSS. Other data were sometimes non-normally distributed and for consistency were always expressed as medians and analyzed by the non-parametric Kruskal–Wallis test using UniStat 4 and S-plus 2000.


For all three colloid fluids there was a significant prolongation of BT at the Post timepoint compared with the Pre timepoint, but no difference between Pre and Last (Table 1). No significant change in BT was seen in those patients receiving saline, and no significant differences were identified between the prolongation for albumin, Haemaccel or Gelofusine. A sex-based difference in BT has been reported [16] (females tending to have a longer BT than males), but although there was significant sex bias between the fluid groups in this study there was no significant difference in the baseline BTs.

Table 1.  Platelet factors: cross-sectional and longitudinal median effects of surgical trauma and fluid therapy, median differences tested by Kruskal–Wallis (a non-parametric comparison)
TimeFluid group
  1. Longitudinal comparison is of the time-points Pre (immediately prior to surgery), Post (immediately after surgery) and Last (2 h after Post sample): significant differences P < 0.05; ††P < 0.005. Cross-sectional comparison is of the four resuscitation fluids used: significant differences *P < 0.05; **P < 0.005. Bracketed figures are first and third quartiles.

Median bleeding time (BT) (min)
Pre6.5 (5.2–7.8)5.78 (5.0–6.9)7.2 (5.4–8.8)5.9 (4.8–6.5)
Post10 (8.5–11.0)11.8 (6.8–15.9)10.1 (8.0–12.7)7.0 (5.6–8.9)*
Last7 (5.2–7.8)††8 (5.0–10.0)6.5 (5.0–8.5)6 (4.9–6.6)
Median β-thromboglobulin (βTG) (IU mL−1)
Pre47.5 (27.7–54.5)58.6 (49.9–66.0)39.3 (24.6–63.9)43.8 (32.7–57.6)
Post92.0 (70.5–117.5)75.5 (67.6–109.2)154.4 (83.0–202.3)76.1 (60.8–97.7)
Last64.4 (51.5–183.2)81.7 (68.7–155.6)123.4 (84.4–193.7)51.8 (51.2–100.0)
Median platelet factor 4 (PF4) (IU mL−1)
Pre13.0 (9.0–36.7)21.2 (10.5–29.9)10.1 (6.7–25.8)11.5 (8.9–18.3)
Post17.5 (12.9–31.7)20.0 (10.2–35.5)37.2 (26.2–110.9)14.8 (12.7–18.1)
Last18.4 (9.5–32.6)44.0 (10.7–77.7)25.9 (19.6–43.4)17.0 (10.2–22.0)
Median VWF (U dL−1)
Pre75.1 (52.6–150.1)100.8 (74.7–118.7)90.8 (56.9–119.9)70.4 (52.5–86.8)
Post62.7 (49.2–129.9)79.6 (72.9–104.2)86.4 (70.8–107.7)54.8 (45.4–78.6)
Last65.9 (57.0–193.4)105.4 (84.0–138.6)94.4 (76.2–101.8)71.0 (62.8–93.4)

There were no significant differences in median blood loss between the fluid groups (median total blood losses: albumin group 840 mL, Gelofusine group 945 mL, Haemaccel group 932.5 mL, saline group 885 mL; P = 0.5587).

The effects of the different fluid therapies on platelet aggregation varied with both type of fluid and stimulating agonist. With aggregation in response to ADP, the effect seen in all fluid groups was a shift of the dose–response curve to the right, i.e. a reduced response of platelets to agonist at the low to mid range of agonist concentration. However, maximal aggregation was unaffected (Fig. 1). This reduced response to ADP was evident at both the Post and Last time-points.

Figure 1.

Effect of surgical trauma on platelet aggregation induced by ADP (at 0, 0.3, 1.0, 3.0 and 10 µmol L−1) in blood from patients receiving albumin, Gelofusine, Haemaccel and saline.

Aggregation in response to the combination of epinephrine and ADP was qualitatively different to that obtained with ADP alone. There was a significant reduction in aggregation in all fluid groups at all agonist concentrations, with a tendency to a greater reduction at higher agonist concentrations, particularly so for the albumin-treated group (Fig. 2).

Figure 2.

Effect of surgical trauma on platelet aggregation induced by epinephrine (at 0, 3.0, 10 and 30 µmol L−1, each with 0.1 µmol L−1 ADP) in blood from patients receiving albumin, Gelofusine, Haemaccel and saline.

Only one of the fluid therapies, albumin, significantly and markedly affected aggregation in response to collagen. In the Gelofusine group there appeared to be a small and transient inhibition of aggregation. In the group treated with albumin there was a much greater inhibition of aggregation at all but the highest concentration of agonist, and this effect persisted (Fig. 3).

Figure 3.

Effect of surgical trauma on platelet aggregation induced by collagen (at 0.25, 0.5, 1.0, 2.0 and 4.0 µg mL−1) in blood from patients receiving albumin, Gelofusine, Haemaccel and saline.

Platelet agglutination in response to ristocetin was completely and persistently suppressed in the Gelofusine and Haemaccel groups (Fig. 4). In contrast, there was a small but significant increase in this parameter in the groups treated with albumin and saline at both Post and Last timepoints.

Figure 4.

Effect of surgical trauma on platelet agglutination induced by ristocetin (at 0.6, 0.8, 1.0, 1.2 and 1.4 mg ml−1) in blood from patients receiving albumin, Gelofusine, Haemaccel and saline.

In the colloid groups, albumin, Gelofusine and Haemaccel, there were significant increases in βTG at the Post timepoint, which tended to remain high at Last. In addition, the increase in βTG with saline was almost significant. There were no significant changes in PF4, nor in von Willebrand factor (VWF) (Table 1).

There was a significant increase in APTT for the three colloid-treated groups at the Post timepoint, but this was only transient, returning to normal at the Last timepoint. FVIII activity decreased significantly, and also transiently. No significant changes in APTT or FVIII activity were seen in the group receiving saline (Table 2).

Table 2.  Clotting parameters: cross-sectional and longitudinal median effects of surgical trauma and fluid therapy, median differences tested by Kruskal–Wallis (a non-parametric comparison)
TimeFluid groupw
  1. Longitudinal comparison is of the time-points Pre (immediately prior to surgery), Post (immediately after surgery) and Last (2 h after Post sample): significant differences P < 0.05; ††P < 0.005. Cross-sectional comparison is of the four resuscitation fluids used: significant differences *P < 0.05; **P < 0.005. Bracketed figures are first and third quartiles.

Median APTT (s)
Pre28.2 (26.6–30.6)29.7 (25.6–31.6)28.4 (27.4–30.9)28.9 (27.5–32.2)
Post34.7 (33.0–38.6)37.2 (31.5–39.9)35.6 (31.0–41.3)32 (29.2–34.7)
Last31.2 (29.4–35.0)31.1 (27.9–32.8)30.7 (28.7–32.2)28.3 (26.0–37.8)
Median FVIII (IU dL−1)
Pre154.1 (118.2–164.3)151 (104.7–166.2)141.6 (124.1–187.4)139.6 (114.9–146.7)
Post94 (77.8–102.9)87.8 (82.4–103.2)91.4 (70.4–102.5)116 (87.6–132.4)
Last108.6 (93.5–117.6)††118.4 (100.2–142.3)123.8 (111.0–148.1)††143.3 (99.7–159.6)
Median prothrombin F1 ± 2 fragment (mmol L−1)
Pre2.44 (2.26–2.57)2.57 (2.40–2.76)2.24 (2.09–2.83)2.45 (2.22–2.67)
Post3.13 (2.73–3.52)2.93 (2.69–3.46)3.65 (3.15–4.78)3.09 (2.63–3.30)
Last3.43 (2.98–5.07)††3.4 (3.01–3.76)††4.02 (3.34–5.78)††3.02 (2.85–3.51)††
Median TAT complex (TAT) (g L−1)
Pre10.6 (2.9–15.7)12 (4.0–17.3)9.7 (5.2–23.0)4.84 (3.2–24.1)
Post56.5 (37.4–64.1)44.1 (33.6–55.6)58.3 (48.3–78.6)41.1 (35.2–56.6)
Last61.8 (40.5–90.0)††47.8 (25.9–62.0)††54.7 (39.3–67.4)††27.3 (21.9–60.1)††
Median fibrinogen (g L−1)
Pre2.9 (2.6–3.1)3.1 (2.6–3.5)2.95 (2.1–3.1)3.05 (2.5–3.5)
Post1.65 (1.5–1.8)1.55 (1.4–1.9)1.3 (1.3–1.9)2.3 (2.0–2.6)**
Last1.65 (1.4–1.7)††1.5 (1.1–1.7)††1.3 (1.2–1.4)††2.3 (1.9–2.5)**††
Median prothrombin time (PT) (s)
Pre13.6 (13.5–13.9)14 (13.7–14.6)14.6 (14.0–14.8)14.2 (13.6–14.6)
Post15.8 (15.1–16.5)16.3 (15.8–16.8)17 (15.9–17.3)15.2 (14.8–15.6)*
Last15.4 (15.3–16.8)††16.3 (15.8–16.9)††16.3 (15.8–17.2)††15 (14.0–15.3)*
Median factor XIII (%)
Pre130 (120–176)159 (137–174)157 (152–174)135.5 (117–150)
Post92.5 (78–110)101 (85–115)101.5 (92–114)105 (99–132)
Last87.5 (69–107)††105.5 (84–118)††113 (91–114)118.5 (98–143)

There were significant increases in prothrombin F1 +2 fragment and in TAT in all the fluid groups at the Post timepoint, which were sustained (Table 2).

Effects on fibrinogen and PT showed that for all fluid groups there was a significant change at the Post timepoint, a decrease for fibrinogen and an increase for PT, and this was maintained through to the Last timepoint. However the magnitude of this change was lower for the saline group with respect to fibrinogen, and for the saline and albumin groups with respect to PT (Table 2).

There was a significant decrease in FXIII at the Post timepoint for the colloid fluids, albumin, Gelofusine and Haemaccel, but no significant difference between these fluid groups at any of the timepoints. This reduction in FXIII persisted through to the Last timepoint. The trend with saline was similar but not significant (Table 2).

There was a significant decrease in ionized calcium concentration in the albumin, Gelofusine and saline groups at both Post and Last timepoints. However there was a significant increase in ionized calcium at these timepoints in the Haemaccel group. The magnitude of the changes was about 10% in either direction. There were no significant changes in ionized magnesium concentrations (Table 3).

Table 3.  Blood biochemistry: cross-sectional and longitudinal median effects of surgical trauma and fluid therapy, median differences tested by Kruskal–Wallis (a non-parametric comparison)
TimeFluid group
  1. Longitudinal comparison is of the time-points Pre (immediately prior to surgery), Post (immediately after surgery) and Last (2 h after Post sample): significant differences P < 0.05; ††P < 0.005. Cross-sectional comparison is of the four resuscitation fluids used: significant differences *P < 0.05; **P < 0.005. Bracketed figures are first and third quartiles.

Median ionized calcium (mmol L−1)
Pre1.21 (1.08–1.22)1.17 (1.11–1.21)1.17 (1.12–1.24)1.21 (1.09–1.24)
Post1.04 (0.98–1.08)1.05 (0.99–1.09)1.28 (1.23–1.39)1.09 (1.05–1.14)**
Last1.035 (0.98–1.11)1.07 (1.03–1.08)1.215 (1.11–1.28)1.09 (1.01–1.15)*
Median ionized magnesium (mmol L−1)
Pre0.64 (0.59–0.66)0.66 (0.61–0.71)0.59 (0.56–0.76)0.60 (0.59–0.66)
Post0.59 (0.55–0.63)0.62 (0.54–0.63)0.62 (0.58–0.74)0.55 (0.54–0.60)
Last0.61 (0.60–0.65)0.62 (0.56–0.66)0.62 (0.60–0.68)0.57 (0.56–0.65)
Median albumin (g L−1)
Pre39 (37–40)40 (38–40)39 (38–41)39 (36–41)
Post40 (39–43)22 (21–24)25 (25–27)30 (26–33)**
Last42 (39–44)24 (22–26)††28 (26–29)††34 (31–34)**††

Whilst there was a significant decrease in serum albumin between Pre and Post, and Pre and Last timepoints in the groups treated with Gelofusine, Haemaccel and saline, there was no such decrease in the albumin-treated group (Table 3).

Both hematocrit and hemoglobin decreased significantly between Pre and Post, and Pre and Last timepoints in all the fluid groups, however, this fall was less in the saline group. Similar changes occurred with respect to the effects on platelet count. Plasma viscosity decreased significantly at the Post timepoint in groups receiving albumin and saline. This was maintained through to the Last timepoint. However there was no significant change in groups receiving the gelatin based colloids, Gelofusine and Haemaccel (Table 4).

Table 4.  Hematological parameters: cross-sectional and longitudinal median effects of surgical trauma and fluid therapy, median differences tested by Kruskal–Wallis (a non-parametric comparison)
TimeFluid group
  1. Longitudinal comparison is of the time-points Pre (immediately prior to surgery), Post (immediately after surgery) and Last (2 h after Post sample): significant differences P < 0.05; ††P < 0.005. Cross-sectional comparison is of the four resuscitation fluids used: significant differences *P < 0.05; **P < 0.005. Bracketed figures are first and third quartiles.

Median hematocrit
Pre0.379 (0.370–0.392)0.395 (0.356–0.418)0.368 (0.346–0.387)0.392 (0.365–0.420)
Post0.280 (0.258–0.291)0.278 (0.244–0.316)0.291 (0.285–0.323)0.330 (0.312–0.364)**
Last0.276 (0.255–0.283)††0.292 (0.267–0.319)††0.299 (0.283–0.312)††0.347 (0.319–0.372)**††
Median hemaglobin (g dL−1)
Pre13.0 (12.4–13.5)13.8 (11.7–14.3)12.5 (12.0–13.3)13.2 (12.5–14.5)
Post9.4 (8.9–10.0)9.6 (8.2–10.4)9.9 (9.5–11.2)11.4 (10.7–12.3)**
Last9.2 (8.7–9.6)††9.9 (8.9–11.0)††10.4 (9.8–10.6)††11.4 (11.0–12.6)**††
Median platelets (109 L−1)
Pre236 (206–252)215 (182–266)196 (176–246)221.5 (184–262)
Post170.5 (153–202)150 (131–217)188 (162–201)204 (180–216)
Last172 (137–197)160 (151–213)187 (160–222)208 (178–238)
Median plasma viscosity (centipoise)
Pre1.63 (1.58–1.72)1.68 (1.61–1.73)1.59 (1.56–1.64)1.64 (1.59–1.71)
Post1.46 (1.40–1.49)1.69 (1.65–1.70)1.60 (1.55–1.62)1.42 (1.40–1.47)**
Last1.46 (1.42–1.49)††1.67 (1.60–1.72)1.59 (1.51–1.67)1.47 (1.42–1.53)**††


The first aim of this present investigation was to establish the extent to which effects of fluids on platelet aggregation observed in previous studies in vitro were reproduced ex vivo after administration to patients undergoing hip replacement. At the same time we sought to obtain additional information on the effects of various fluids on other hemostatic variables. These included measures of platelet activation, thrombin generation and several other hematological indices. The effects of the fluids on BT and on blood loss were also determined. Studies were performed immediately before the start of surgery, at the end of surgery and 2 hours after the end of surgery. The fluid groups used were saline, Haemaccel, Gelofusine and 4.5% albumin.

Based on our previous in vitro findings [8] of the effects of Haemaccel and Gelofusine compared with saline on platelet aggregation, we had expected to see pronounced inhibition by Haemaccel of the platelet aggregation induced by agents such as ADP, epinephrine and collagen, and also pronounced inhibition by Haemaccel and Gelofusine of the platelet agglutination induced by ristocetin. In fact we did see profound effects of both Haemaccel and Gelofusine on ristocetin-induced responses at both Post and Last timepoints. Inhibition by gelatin colloids of platelet agglutination by ristocetin is a well-established observation [2,4], and confirms our previous in vitro work [8]. It is thought this may be due to an interaction of the gelatins with VWF [2,7].

However, the expected effects of Haemaccel on the aggregation induced by other agents were not evident. There was significant inhibition of the aggregation induced by ADP and by epinephrine, but no more than was seen with saline, Gelofusine or albumin. The explanation for the different results obtained in our previous experiments conducted wholly in vitro and these experiments ex vivo is that the increases in plasma calcium ions obtained in vitro that resulted from the high concentrations of calcium in Haemaccel were not evident ex vivo. In our in vitro system Haemaccel caused a rise in ionized calcium to 2.8 mmol L−1, whereas in the ex vivo system, despite a similar dilutional load, median ionized calcium rose to only 1.28 mmol L−1 (maximum concentration 1.63 mmol L−1). In a previous in vivo study of trauma patients resuscitated with the calcium-containing fluid Haemaccel ionized calcium concentrations in excess of 2 mmol L−1 were obtained [17]; however, these patients were critically injured and received greater volumes of Haemaccel than those in our current study. Although our patients had significant surgical trauma, this was not severe enough to affect their homeostatic compensatory mechanisms. Clearly, homeostatic mechanisms had resulted in avoidance of large increases in plasma calcium, despite large amounts (2 L) of Haemaccel being infused. This resulted in less inhibition of aggregation by Haemaccel in the blood from the patients receiving this particular fluid.

There was persistent inhibition of platelet aggregation induced by ADP and by epinephrine in all fluid groups. We believe the most likely explanation of this reduced response is that desensitization to the aggregatory effects of these agents occurred consequent to prior platelet activation. There was clear evidence of platelet activation during surgery as judged by release of βTG, even though PF4 did not increase. This was probably because of the short half-life of PF4 in plasma due to its binding to glycosaminoglycans on the vascular endothelium. It is certainly possible that ADP had been generated both through release from damaged tissue during surgery and via secretory mechanisms following platelet aggregation, resulting in desensitization of ADP receptors. Additionally, epinephrine receptors may have been desensitized consequent to the surgical trauma.

Collagen-induced aggregation was markedly inhibited following albumin infusion, which suggests that maintenance of plasma albumin concentrations results in a further decrease in platelet function. This may be related to the ability of albumin to bind and neutralize platelet agonists such as thromboxane A2[18,19].

Although there was a pronounced inhibitory effect on ristocetin-induced agglutination of platelets by the gelatin colloids, in contrast there was some potentiation of the effects of ristocetin following infusion of saline and albumin. This is likely to be a consequence of hemodilution.

Overall in this study, platelet activation as a consequence of surgery was seen to be accompanied by a general downregulation of platelet aggregation, together with selective additional inhibition of collagen-induced aggregation by albumin and of the ristocetin-induced platelet agglutination by the gelatin colloids. Such inhibition of platelet function might be interpreted as providing less risk of untoward thrombotic events, but with the potential to increase bleeding (but see below).

In addition to changes in platelet aggregation, several other changes were recorded that might also be expected to influence risk of thrombosis and/or bleeding. Overall there was significant and prolonged hemodilution, particularly with the three colloids, which was reflected in decreases in fibrinogen, FXIII, albumin (except during albumin infusion) and platelet count, and an increase in prothrombin time. Decreases in fibrinogen and albumin appeared to be somewhat larger than could be accounted for by simple hemodilution. Changes in FVIII and APTT also occurred but here there was some recovery between Post and Last, which could not be associated with hemodilution. Consumption of coagulation factors also occurred, as reflected by increases in TAT complex and prothrombin F1 +2 fragment. Generally there was a small reduction in ionized calcium (except with Haemaccel), and infusion of saline or albumin produced a reduction in plasma viscosity. There were no significant changes in ionized magnesium concentrations. It was thus interesting to determine the effects of these various changes on BT and on blood loss.

We found that BT was significantly prolonged in the groups that received colloid fluids, but not in the group receiving saline. However, this was only a transient phenomenon, with BT returning to baseline by 2 h after the end of surgery. Between colloid fluids there was very little difference in degree of prolongation. The fact that BT was not increased in the saline group confirms that the effect is due to colloid infusion and is in accord with earlier data that indicated that BT can increase in patients receiving colloids in both traumatic and non-traumatic situations [5,7].

It is tempting to speculate that the increases in BT following colloid infusion relates to the various inhibitory effects of the colloids on platelet aggregation and the effects on the various coagulation parameters referred to above. However, while the effects on most of the parameters that were measured were seen at both the Post and Last timepoints, the increases in BT had returned to Pre values at the Last timepoint. This was mirrored by the changes in FVIII and the APTT measurements, which also showed some recovery at Last. Also, although not statistically significant, the changes in VWF followed the same trend.

Although we saw a transient increased BT in the colloid groups, this was not mirrored by increased blood loss in these groups compared with the group receiving saline. This concurs with results found by others. Boldt et al. [20] found that there was no significant increase in blood loss when either gelatin or albumin were infused immediately prior to cardiopulmonary bypass in patients undergoing elective aortocoronary bypass grafting. Furthermore, Scott et al. [21] found no significant difference in blood transfusion requirement following coronary bypass grafting between patients in whom the coronary artery bypass pump was primed with crystalloid, Haemaccel or albumin.

In summary, despite considerable changes in platelet, coagulation and other parameters brought about by colloid infusion, the effects on BT were only transient and there was no measurable effect on blood loss in this controlled study of hip arthroplasty. Surgery alone resulted in some reduction in platelet aggregation to agents such as ADP and epinephrine, probably as a consequence of platelet desensitization. Albumin had an additional inhibitory effect on collagen-induced platelet aggregation, and Haemaccel and Gelofusine inhibited ristocetin-induced platelet agglutination. This further inhibition of platelet function by the colloids may have the potential to provide additional protection against the increased thrombotic risk in patients following major surgery, and should be taken into account when assessing the need for and level of heparin prophylaxis following surgery.


This research was supported by a grant from BPL, Elstree UK.

Contributions by authors. This study was devised and supervised by P.A. Evans and S. Heptinstall who also co-ordinated the writing of this manuscript. E.C. Crowhurst entered the patients into the study, allocated them to groups, took all blood samples, performed BTs, recorded blood loss and collated the experimental data. The surgical operations were performed under the supervision of J. Hoskinson and C.M. Stray. Specialist assays that required particular expertise were conducted and/or supervised by J.R. Glenn, W. Madira, S.J. Davidson and J. Burman. T. Davies performed the statistical analyses.