4 Evidence and interpretation

The Appraisal Committee (Appendix A) considered evidence from a number of sources (see Appendix B).

4.1 Clinical effectiveness

Pre-hospital or hospital IV fluid replacement

4.1.1 The Assessment Report identified randomised controlled trials (RCTs) and systematic reviews of RCTs that compared pre-hospital IV fluid replacement with withheld (no pre-hospital) fluid. Observational studies cited in the evidence base of the Consensus Statement and JRCALC guidelines were also critically appraised. Seven studies were identified: two RCTs comparing immediate pre-hospital IV fluid replacement with delayed replacement; two RCTs comparing the use of different volumes of fluid for IV fluid replacement; two systematic reviews of RCTs of IV fluid replacement in humans and animals; and one observational study.

4.1.2 One US trial randomised (according to day of the week) 598 trauma patients with penetrating injuries either to receive IV fluid before surgery (en route to hospital or in a trauma centre), or to have fluid withheld until surgical intervention at hospital. This was the most methodologically sound of all of the studies, with appropriate randomisation and protocol compliance, although the study population was not representative of the majority of trauma patients in the UK, who have blunt injuries. Delayed IV fluid replacement was associated with a significant improvement in mortality until discharge (70% survival compared with 62%, p = 0.04).

4.1.3 In a UK crossover RCT of 1309 trauma patients with mainly blunt injuries, paramedics were randomised either to withhold IV fluid until arrival at hospital (delayed fluid group) or to give pre-hospital fluids to those who would normally receive them (immediate fluid group), and the paramedics were 'crossed over' to the other protocol half way through the trial. The trial reported no statistically significant differences in mortality between groups (adjusted odds ratio, 0.93; 95% confidence interval, 0.58 to 1.49). However, poor adherence to the protocol meant that only about 10% more patients in the immediate fluid group received pre-hospital fluid than in the delayed fluid group.

4.1.4 Two US studies compared the effect of IV fluid volume administered after hospital arrival on patient mortality. The studies did not appear to take into account fluid administered before arrival. One RCT randomised 36 hypovolaemic patients to the rapid infusion system or to a conventional infusion system. The rapid infusion system administered IV fluid via one catheter, which resulted in a higher rate of fluid infusion in the first hour than conventional infusion but a lower total volume administered over 24 hours. There were no significant differences in mortality between groups (5/16 deaths with rapid infusion and 4/20 deaths with conventional infusion), but there was a trend towards fewer complications among survivors of the rapid infusion group. In the other RCT of 110 patients with uncontrolled haemorrhage, IV fluid was administered to achieve a target systolic blood pressure of 70 mmHg in the intervention group and more than 100 mmHg in the control group. Again, there were no differences in mortality between groups. However, interpretation of both studies was hindered by methodological limitations, including the absence of details of randomisation, concealment, compliance and differences in surgical interventions between groups.

4.1.5 One systematic review of RCTs comparing immediate and delayed IV fluid replacement included the four RCTs considered above and an additional two RCTs that did not meet the inclusion criteria for this appraisal because their focus was on blood transfusion.

4.1.6 Another systematic review of animal models of IV fluid replacement in uncontrolled haemorrhage reported an improvement in mortality associated with early replacement, but this was not statistically significant (risk ratio, 0.88; 95% confidence interval, 0.73 to 1.07). Early IV fluid replacement appeared to improve survival in severe haemorrhage but to increase the risk of death with less severe haemorrhage. It is not clear, however, whether the findings are relevant to humans.

4.1.7 A Canadian retrospective cohort study compared the effect of administering or withholding pre-hospital IV fluid in 360 patients with matched pre-hospital injury (PHI) scores. Pre-hospital administration of IV fluid was associated with a significant increase in mortality (adjusted odds ratio, 2.33; 95% confidence interval, 1.02 to 5.28). Despite the matching of the PHI scores of the two groups, there remained important differences in terms of age, injury severity score, mechanism and anatomical location of the injury, all of which are predictors of trauma-related mortality.

4.1.8 In summary, there was insufficient evidence to draw definitive conclusions about the effectiveness of pre-hospital or delayed IV fluid administration in trauma. Although the most methodologically sound RCT provided some evidence that in certain circumstances pre-hospital IV fluid resuscitation may be harmful, it is not clear how different subgroups would be affected.

Advanced life support (ALS) versus basic life support (BLS)

4.1.9 Studies that compared the effectiveness of ALS (where additional interventions including the administration of pre-hospital IV fluid may be performed) with BLS (where no pre-hospital IV fluid is administered) were considered as indirect information on the effectiveness of pre-hospital and withheld IV fluid respectively. Six studies were identified: two systematic reviews, and the four observational studies that formed the evidence base of the Consensus Statement and JRCALC guidelines.

4.1.10 One of the systematic reviews contained just one RCT, in which 2045 trauma patients in three areas of England (covering urban, suburban and rural areas) were randomised to treatment by paramedics or EMTs. Although the study was designed as an RCT, the results were analysed as a cohort study because poor protocol compliance meant only 16 patients were successfully randomised. When data from all areas were aggregated, the study showed that attendance by paramedics was associated with a non-significant increase in mortality (adjusted odds ratio, 1.74; 95% confidence interval, 0.89 to 3.41) but there was substantial regional variation (odds ratio 3.1 in area 1 and 0.78 in area 3).

4.1.11 The other systematic review included 13 observational studies and two reports of one RCT. Of these studies, in terms of mortality, 3/15 favoured ALS and 12/15 supported BLS (overall unadjusted odds ratio 2.92, favouring BLS). Limiting the analysis to well-designed studies produced a crude odds ratio of 1.89 (favouring BLS to a lesser extent) but as confidence intervals were not stated it is not clear whether differences in mortality were statistically significant.

4.1.12 Four observational studies were cited in the Consensus Statement and JRCALC guidelines: two reported higher mortality in patients attended by paramedics, one favoured LS and the other found that pre-hospital time did not affect survival. The results of these studies are included in the systematic reviews above. All the studies were critically appraised and were considered to have serious methodological flaws that increased the likelihood of bias, or to have controlled inadequately for confounding factors.

4.1.13 In summary, studies comparing ALS with BLS care of trauma patients provide insufficient evidence to demonstrate the benefit or harm of paramedic interventions. There was a trend towards poorer outcomes with ALS but it is not possible to determine whether this is due to the delay associated with ALS, or to the additional procedures themselves, or because patients who have additional procedures may be more severely injured and have a poorer prognosis.

Intravenous infusion with different fluid types

4.1.14 As a subsidiary issue, systematic reviews of the effectiveness of IV infusion with different fluid types in a variety of settings were assessed.

4.1.15 Ten systematic reviews of RCTs were identified that compared different IV fluid types: four reviews were general comparisons of crystalloid and colloid solutions; the other reviews compared more specific IV fluid types (for example, isotonic crystalloid versus colloid; albumin-based colloid versus non-albumin solutions; comparisons of different classes of colloid; and hypertonic crystalloid [with or without dextran] versus isotonic crystalloid). The four systematic reviews that were general comparisons of IV fluid types showed a potential trend towards crystalloids being more effective than colloids, although making general comparisons may have obscured the effect of individual crystalloid or colloid solutions. The Assessment Report concluded that there was insufficient evidence of benefit of a particular IV fluid because of clinical heterogeneity between studies (such as case-mix, additional interventions received, resuscitation protocols, amounts of IV fluid administered, and different types of colloids and crystalloids administered) and the fact that different types of patients were combined in the meta-analyses.

4.2 Cost effectiveness

4.2.1 The Assessment Report identified two Health Technology Assessment reports of the cost effectiveness of pre-hospital IV fluid replacement from an NHS perspective.

4.2.2 The first study, published in 1998, assessed the cost and effectiveness of treatment of trauma patients by paramedics, compared with treatment by EMTs.

4.2.3 The additional cost of paramedic treatment (costs associated with trauma-related ALS training, salary and additional pre-hospital interventions) was presented per paramedic crew and per call out. The average unit cost of the ALS crew at £2.44 per minute was similar to the cost of a BLS crew at £2.43 per minute. There was an increase of £3 in the call out cost of an ALS crew (average of £81.08 per ALS call out) compared with a BLS crew call out (average of £78.02), because ALS crews spent more time at the scene. The total costs (pre-hospital and hospital costs combined) were also estimated. There was a non-significant increase of £22 in the average total costs for patients attended by a paramedic-crewed ambulance (£2231 per patient, compared with £2209 per patient attended by EMTs alone). Between 20% and 30% of the cost of paramedic training and salary was attributed to trauma. However, reductions in the level of trauma-related training had little effect on the overall cost of training.

4.2.4 The second study was conducted alongside an RCT (discussed in Section 4.1.3) in which paramedics were randomised to different resuscitation protocols (pre-hospital IV fluid versus no pre-hospital fluid) to evaluate the cost effectiveness of pre-hospital IV fluid therapy.

4.2.5 Although there was no difference in the median ambulance call-out time of 55 minutes, there was a 2-minute increase in the mean call-out time associated with pre-hospital IV fluid replacement. Costs were presented as initial costs (ambulance costs, consumables, and accident and emergency costs) and total costs (which also included inpatient costs). The cost of pre-hospital IV fluid replacement was higher (but not significantly higher) than that of delayed fluid replacement, by £3 in the initial phase of treatment (£419 compared with £416) and by £28 overall (£2706 compared with £2678).

4.2.6 The Assessment Report did not include an economic model because it was considered that there was insufficient evidence on the effectiveness of IV fluid replacement therapy to inform such a model, and because the additional costs associated with pre-hospital IV fluid therapy, such as consumables and paramedic training, were thought to be minor – particularly because paramedics would be required to stock IV fluid and to undergo training in cannulation and IV fluid administration for the treatment of non-trauma patient groups. Adherence to a conservative pre-hospital IV fluid policy could, however, increase ambulance efficiency to a small extent by improving response times.

4.3 Consideration of the evidence

4.3.1 The Committee reviewed the evidence, including the views of experts, on the clinical and cost effectiveness of IV fluid administered in a pre-hospital setting. In its considerations, the Committee was mindful of the need to take account of the effective use of NHS resources.

4.3.2 The Committee considered that there was some evidence of benefit associated with delaying the initiation of IV fluid until hospital arrival.

4.3.3 The Committee considered the extent to which evidence from the most methodologically sound RCT, which demonstrated a benefit in delaying IV fluid resuscitation in penetrating injuries, could be generalised to blunt injuries. The Committee heard, however, that the trial was based on the administration of larger quantities of IV fluid than would now be considered appropriate, and that it was, therefore, difficult to generalise the results to current clinical practice. In the absence of high-quality evidence on effectiveness, the Committee considered that guidance should take into account professional consensus, although the Institute did not formally evaluate the Consensus Statement and JRCALC guidelines.

4.3.4 The Committee heard from experts that it would not be clinically appropriate to withdraw the use of IV fluid in a pre-hospital setting. The experts emphasised that there was a small proportion of severely hypovolaemic trauma patients at high risk of immediate death who might benefit from pre-hospital initiation of IV fluid therapy; they explained that the aim of IV fluid in these circumstances is to prevent further circulatory collapse without attempting to restore circulating volume fully back to normal or to achieve normal physiology.

4.3.5 The Committee considered how these patients with severe hypovolaemic shock should best be identified and treated. The Committee heard that there are a number of physiological indicators of haemorrhagic shock such as pallor, tachycardia and capillary refill time, although the most readily available physiological measure was considered to be the absence of a palpable radial pulse. The Committee understood that presence or absence of a radial pulse has been used as an approximate guide to whether the systolic blood pressure is above or below 80–90 mmHg but that this is not fully validated. The Committee concluded, therefore, that IV fluid should be administered in the pre-hospital setting only if a radial pulse (or a central pulse, in penetrating injuries of the torso) is not palpable. The Committee was persuaded that in the presence of severe hypovolaemia, which is considered to be immediately life threatening, clinical judgement as to the best course of action would be required.

4.3.6 The experts further advised the Committee that consideration should be given to how well the haemorrhage is controlled. Haemorrhages can be described as controlled (for example, by external pressure applied to a wound), self-limiting (for example, bleeding from a closed femoral fracture), potentially uncontrolled (when the bleeding has stopped but might start again if the blood pressure increased, and the injury is at a site where applying pressure would not stop the bleeding), or uncontrolled. The Committee heard that the risks associated with administration of IV fluid are different for controlled and uncontrolled haemorrhage, and it may be difficult to distinguish between these types of haemorrhage in the pre-hospital setting. Patients with severe uncontrolled bleeding with circulatory collapse are at risk of immediate cardiac arrest and death from extreme hypovolaemia/exsanguination. But the experts also highlighted that in patients with uncontrolled or potentially uncontrolled bleeding, vigorous fluid therapy may exacerbate bleeding by diluting blood clotting factors, reducing the concentration of circulating blood platelets, and by dislodging early clots forming at the site of haemorrhage. In patients with controlled or self-limiting bleeding, the infusion of IV fluid may help to restore tissue perfusion without exacerbating bleeding.

4.3.7 Taking these factors into account, the Committee considered that a pragmatic approach would be to withhold IV fluid if the signs of shock are not marked. If the signs of shock are more severe (as illustrated by an absent radial pulse) administering IV fluid may prevent extreme hypovolaemia with its risk of producing organ damage and cardiac arrest. The experts advised that giving small volumes in this situation (and repeating, if necessary) may keep the patient alive without unduly exacerbating the bleeding, even if the bleeding is uncontrolled or potentially uncontrolled. The Committee concluded that IV fluid should be administered according to the above criteria in boluses of no more than 250 ml and should normally be started en route to hospital to avoid delays at the scene. This agrees with the recommendations in the Consensus Statement; the revised JRCALC guidelines (due in 2004) are also expected to include this recommendation. The Committee also concluded that administration of the first bolus should be followed by reassessment and administration of further boluses until a radial pulse (or a central pulse in penetrating injuries of the torso) becomes palpable. The Committee was advised that good patient-handling techniques in pre-hospital care were essential to minimise the risk of continued haemorrhage before the administration of IV fluid.

4.3.8 The treatment of trauma in patients with isolated closed head injury was not considered in detail by the Committee because it fell outside the remit of this appraisal. However, the Committee heard from experts that consideration should be given to trauma patients with multiple injuries in whom both haemorrhagic shock and head injury might coexist. Altered consciousness in trauma may be indicative of severe haemorrhagic shock or of head injury, and it may be difficult to distinguish between the two in the pre-hospital setting. There was concern that if IV fluid was withheld from trauma patients with multiple injuries including head injury, this might have a deleterious effect on the outcome of the head injury because of low perfusion of the brain. The Committee considered cautious administration of IV fluid (with 250 ml boluses titrated against the presence of a radial pulse, with reassessment) was appropriate for trauma patients with uncontrolled bleeding and concomitant head injury in the same way as other hypovolaemic trauma patients.

4.3.9 Experts advised that the guidance should take into account factors – such as the patient being trapped – that might delay arrival at hospital. The Committee concluded that there was inadequate evidence on which to base recommendations on the use of IV fluid in these circumstances. It was appreciated, however, that in these circumstances, other emergency procedures would be initiated that fall outside the remit of this guidance.

4.3.10 On the balance of the evidence on the relative effectiveness of crystalloid and colloid solutions administered in the pre-hospital setting, the Committee was persuaded that the merits of different IV fluid types should be based on cost and risk of adverse events. Crystalloid solutions are not associated with the hypersensitivity reactions seen in some patients when colloids are infused, and they are less expensive than colloid solutions. The Committee therefore considered intravenous infusion with crystalloid solutions to be the preferred option. On balance of the evidence on the relative effectiveness of different crystalloids, the Committee considered normal saline, which has the lowest cost, to be the favoured option.

4.3.11 Consideration was given to the use of pre-hospital IV fluid in young children and infants. The Committee heard from the experts that children, particularly those younger than 8 years, should be considered as a separate group because their physiology is different and different methods are needed to assess hypovolaemic shock. The experts advised that it would not be appropriate to use the absence of a radial pulse or an increased heart rate, in isolation, as criteria to determine whether pre-hospital IV fluid should be administered. The Committee discussed in detail both the available evidence and the experts' views on the appropriate use of IV fluid in the pre-hospital setting for young children and infants. In the absence of adequate evidence and any professional consensus, the Committee considered that they were unable to make specific recommendations for this group. However, they concluded that transfer to hospital should not be delayed in order that IV fluid can be administered.

4.3.12 The Committee considered the issue of training for those administering pre-hospital IV fluid therapy, and concluded that only healthcare professionals with appropriate training in ALS techniques and pre-hospital care should administer IV fluid therapy to trauma patients in the pre-hospital setting.

4.3.13 The Committee noted that the information available on cost effectiveness was poor. The Committee considered the costs associated with the IV fluid used for pre-hospital administration to be minimal. It also considered that costs associated with paramedic training would be unchanged regardless of whether IV fluid is administered for trauma in a pre-hospital setting, because all paramedics should have undergone the necessary training as part of routine preparation for pre-hospital care. The Committee concluded that, although the costs associated with pre-hospital IV fluid administration were minimal, there was an opportunity cost to be considered in terms of potential improvements in response times, throughput, and overall efficiency as a result of longer call-out times when IV fluid therapy was administered.