The scope of this review, originally published in Issue 4, 1999, was expanded in the second update, published in Issue 4, 2001, to cover other methods of anaesthesia. The main focus remains the comparison of regional versus general anaesthesia.
The term proximal femoral fracture, or 'hip fracture', refers to a fracture of the femur in the area of bone immediately distal to the articular cartilage of the hip, to a level of about five centimetres below the lower border of the lesser trochanter. The majority of these fractures occur in an elderly population with an average age of around 80 years. Females predominate over males by about four to one (Parker 1993) and the injury is usually the result of a simple fall. Whilst the hip fracture is usually the only injury, the patients frequently have many other medical problems associated with aging.
An estimated 1.7 million hip fractures occurred worldwide in the year 1990 (WHO study group 1994). The number of hip fracture patients continues to rise, due to a combination of an increasingly elderly population and an increase in the age specific incidence. A prediction for global numbers of 6.26 million hip fractures by the year 2050 has been made (Melton 1993). The majority of these fractures are treated surgically; thus hip fracture surgery represents one of the most common emergency orthopaedic procedures. Surgical treatment may be either fixation of the fracture or replacement of the femoral head with an arthroplasty. Internal fixation involves using screws or pins, either alone or in combination with a side plate applied to the femur, or by the use of an intramedullary nail with a cross screw inserted into the femoral head. Arthroplasty involves excision of the fractured area of bone and replacement with a partial or total hip replacement, which may be cemented in place.
General anaesthesia refers to the use of a variety of intravenous and or inhalation drugs to render the patient unconscious. The patient may breathe spontaneously or require mechanical ventilation following the administration of neuromuscular blocking agents. Potential complications of general anaesthesia include adverse reactions to the drugs used, difficulty in maintaining or establishing an airway, intraoperative hypotension, aspiration of gastric contents, postoperative nausea, respiratory depression and damage to the teeth or upper airways.
Regional anaesthesia for hip fracture surgery refers to the injection of a local anaesthetic into the epidural or subarachnoid space at the lumbar spine. Injection into the subarachnoid space, often termed spinal anaesthesia, is the most commonly used method for hip fracture surgery. In some cases the patient also receives sedatives whilst the block is inserted and possibly during the surgery itself. The main complication of a regional technique is intraoperative hypotension, which may lead to cerebrovascular or myocardial ischaemia or infarction. Other problems may be an inadequate regional block, the rare complications of damage to local structures and headache secondary to leakage of cerebrospinal fluid from the dural puncture site. Specific advantages of regional anaesthesia may be a reduction in the incidence of thrombotic episodes and a reduced operative blood loss (Modig 1988). These may be a consequence of an increased peripheral limb blood flow in combination with reduced venous tone. Alternatively they may arise from an alteration of blood viscosity and coagulability, as a result of changes in the metabolic and neurohumoral responses to surgery (Modig 1983).
Other forms of anaesthesia used for hip fracture surgery are the insertion of local nerve blocks around the hip. These may be supplemented with sedatives, analgesics or other parenteral drugs. A lumbar plexus block refers to injection of a local anaesthetic agent into the area of the lumbar plexus close to the transverse process of the fourth lumbar vertebra (Winnie 1974). Only the plexus on the side of the fracture needs to be blocked, which may reduce the incidence of complications such as intraoperative hypotension. A sacral plexus block refers to the injection of a local anaesthetic agent in the area around the sacral nerves (Mansour 1993). The use of nerve blocks preoperatively or in conjunction with general anaesthesia is considered in another Cochrane review (Parker 2001).
An alternative type of anaesthetic involves the use of intravenous ketamine on its own. Ketamine renders the patient unconscious, thereby acting as a general anaesthetic, and has analgesic effects.
No consensus exists as to which is the best method of anaesthesia. Thus the choice of anaesthesia used for hip fracture surgery is often determined by the personal preference of the anaesthetist concerned, following assessment of the patient's medical state and preferably, if possible, consultation with the patient. Thus the choice of anaesthesia used for hip fracture surgery is often determined by the personal preference of the anaesthetist concerned, following assessment of the patient's medical state and, if possible, after consultation with the patient. A general review of anaesthesia for hip fracture surgery (Covert 1989) summarised the possible advantages of different anaesthetic methods using information from eight of the randomised trials on this subject. In a meta-analysis, using Bayesian methods, of 11 randomised trials of regional versus general anaesthesia for surgical repair of hip fractures, Sorensen 1992 concluded that the superiority of one method over the other was unproven. Not all currently available randomised trials were included and, moreover, some trial data from two studies were duplicated in the analysis. A more recent meta-analysis of randomised trials for all types of surgery has demonstrated a reduction of early postoperative mortality and morbidity with epidural or spinal anaesthesia (Rodgers 2000).
To determine the optimum anaesthetic technique for hip fracture surgery. Different types of anaesthesia, namely regional (either spinal or epidural), inhalation general anaesthesia, local nerve blocks and intravenous ketamine anaesthesia were compared. Variations in anaesthetic drug dosage and delivery or supplementary regional blocks were not considered within this review.
The following null hypotheses were tested within the trials included so far in this review:
(1) There is no difference in outcome between regional anaesthesia (spinal or epidural) and general anaesthesia.
(2) There is no difference in outcome between regional anaesthesia (spinal or epidural) supplemented with a 'light' general anaesthetic and general anaesthesia alone.
(3) There is no difference in outcome between regional anaesthesia (spinal or epidural) and regional nerve blocks alone.
(4) There is no difference in outcome between anaesthesia using ketamine (with or without a benzodiazepine) and inhalation general anaesthesia.
Criteria for considering studies for this review
Types of studies
All randomised controlled trials comparing different methods of anaesthesia were included. Quasi-randomised trials (for example, alternation), and trials in which the treatment allocation was inadequately concealed, were considered for inclusion.
Types of participants
Skeletally mature patients undergoing hip fracture surgery.
Types of interventions
(1) Regional anaesthesia (if necessary supplemented by sedatives) achieved by injection of local anaesthetic into the epidural or subarachnoid spaces. This type of anaesthesia is also referred to as 'spinal' or 'epidural'.
(2) General anaesthesia using intravenous or inhalation agents to render the patient unconscious. Unless otherwise stated, general anaesthesia refers to general anaesthesia using inhalation agents in this review.
(3) Intravenous ketamine.
(4) Local nerve blocks (if necessary supplemented by sedatives) when used as the primary method of anaesthesia.
Trials testing other methods of anaesthesia as the primary method of anaesthesia were considered for inclusion. Trials comparing the use of local nerve blocks in conjunction with general anaesthesia and the use of nerve blocks preoperatively, are evaluated in another Cochrane review (Parker 2001). Also not considered in this review were trials comparing different types of drugs or techniques of individual methods of anaesthesia.
Types of outcome measures
The primary outcome measure was mortality (at 1 month, 3 months, 6 months and 1 year).
In addition, data were extracted from each study for outcomes in the following four categories. The majority of outcomes in the first category (peri-operative outcomes) are surrogate or intermediate outcomes: these are marked with an asterisk (*). As such they have an inexact relationship with important clinical outcomes that would be directly experienced by the patient. Some of these surrogate outcomes, such as fall in haemoglobin levels and hypotension, mainly serve to prompt remedial intervention to reduce the risk of a serious clinical event. We have included these outcomes in order to provide a full picture of the results of the included trials.
(a) Peri-operative outcomes:
- length of operation (in minutes) *
- hypotension (intraoperative or immediately postoperative) *
- operative blood loss (in millilitres) *
- transfusion requirements
- fall in haemoglobin level*
- need for supplementary drugs to complete anaesthetic *
- changes in body temperature *
- pre- and postoperative arterial blood gases *
- changes in catecholamines and other stress response chemicals during and after surgery *
- intraoperative cardiac arrhythmias *
- time to mobilisation
- length of hospital stay (in days)
(b) Complications specific to the method of treatment:
- aspiration pneumonia
- post-dural puncture headache
- damage to the upper airways or mouth from devices used for general anaesthesia
- secondary intervention required for anaesthetic complications or failure
- any other adverse effects as detailed in each study (new in second update)
(c) General postoperative complications:
(unless otherwise specified, the definition for these complications will be as detailed in each study, or by post-mortem)
- myocardial infarction
- cerebrovascular accident
- congestive cardiac failure
- renal failure
- cardiac arrhythmias
- acute confusional state
- urine retention (requiring catheterization)
- postoperative nausea and /or vomiting
- deep vein thrombosis (diagnosis confirmed by post-mortem, venography, isotope scanning, ultrasound or plethysmography, whether this was performed routinely or only as clinically indicated)
- pulmonary embolism (diagnosed by isotope scanning, angiography or post-mortem)
(d) Final outcome measures:
- mortality (primary outcome)
- change in mental function
- functional status
- return of patient to their pre-fracture place of residence
Search methods for identification of studies
We searched the Cochrane Bone, Joint and Muscle Trauma Group specialised register (November 2003). The specialised register is compiled from multiple databases, including regular searches of the Cochrane Central Register of Controlled Trials in The Cochrane Library, MEDLINE (which combines subject specific terms with the optimal trial search strategy (Alderson 2004a)), EMBASE and CINAHL, and handsearch results. For further details see the search strategy in the group's module in The Cochrane Library.
In addition we searched MEDLINE (1996 to February week 2 2004), EMBASE (1988 to 2004 week 10) and reference lists of relevant articles. In MEDLINE (OVID-WEB) the following search strategy was combined with the first two levels of the optimal trial search strategy (Alderson 2004a).
1. exp Hip Fractures/
2.((hip$ or femur$ or femoral$ or trochant$ or pertrochant$ or intertrochant$ or subtrochant$ or intracapsular$ or extracapsular$) adj4 fracture$).tw.
4. exp Anesthesia/
5. ((an?esthet$ or an?esthesia) adj4 (regional$ or local$ or general or spinal or epidural)).tw.
The generic hip fracture search strategy for EMBASE is shown in Appendix 1.
Articles of all languages were considered and translated if necessary.
Data collection and analysis
Data for the outcome measures listed above were independently extracted by two reviewers, and checked by at least one of the other two reviewers. In addition each trial was assessed without masking for its quality of methodology. Any differences were resolved by discussion between the reviewers.
The main assessment was by the quality of concealment of allocation which was scored either A, B or C according to the criteria in the Cochrane Reviewers' Handbook (Alderson 2004b), or 3, 2, 1 or 0 as described below (item 1). A further eight aspects of methodology were also rated. Though the scores of the individual items were summed, this was to gain an overall impression rather than for quantitative purposes.
(1) Trials with clear concealment of allocation (e.g. numbered sealed opaque envelopes drawn consecutively) were coded as A and scored 3. Those in which there was a possible chance of disclosure of assignment were coded as B and scored 2. Those in which allocation concealment was not stated, or unclear, were coded as B and scored 1. Those where allocation was clearly not concealed, such as trials using quasi-randomisation (e.g. even or odd date of birth), were coded as C and scored 0.
(2) Were the inclusion and exclusion criteria clearly defined? Score 1 if text states type of patients included and those excluded; otherwise score 0.
(3) Were the outcomes of trial participants who withdrew or who were excluded after allocation described and included in an intention to treat analysis? This particularly applies to people allocated to regional anaesthesia where it was not achieved due to technical difficulties. Score 1 if these people were either detailed separately or included in the analysis group to which they were allocated, or if text states that no withdrawals occurred; otherwise score 0.
(4) Were the treatment and control groups adequately described at entry? Score 1 if a minimum of four admission details were given (e.g. age, sex, mobility, fracture type, function score, ASA grade, mental test score); otherwise score 0.
(5) Were the care programmes other than trial options identical? Score 1 if text states they were; otherwise score 0.
(6) Were the outcome measures clearly defined in the text? Score 1 if yes; otherwise score 0.
(7) Were the outcome assessors blind to treatment group? Score 1 if yes; otherwise score 0.
(8) Was the timing of outcome measures appropriate? This was considered to be a minimum of three-months follow up for all surviving trial participants. Score 1 if yes; otherwise score 0.
(9) Was loss to follow up reported and if so were less than five per cent of trial participants lost to follow up? Score 1 if yes; otherwise score 0.
Heterogeneity between comparable trials was tested using a standard chi squared test and, latterly, the I-squared test (Higgins 2003). Relative risks and 95% confidence intervals were calculated for dichotomous outcomes. Mean differences and 95% confidence intervals were calculated for continuous outcomes. Results of comparable groups of trials were pooled using fixed effect and random effects models and 95% confidence intervals. Both Peto odds ratio and relative risk plots were viewed and a note was taken of where there was statistically significant heterogeneity (P < 0.1) using either method. The results for the random effects model are presented when there is significant heterogeneity in the results of individual trials. Any tests of interaction, calculated to determine if the results for subgroups were significantly different, are based on odds ratio results.
Description of studies
Of 13 newly identified studies for this update, four studies (Biffoli 1998; Casati 2003; Kamitani 2003; Svarting 1986) were included, seven excluded and two placed in 'References to studies awaiting assessment'. Further details have been requested for one study (Dougall 1988) in the latter category; we have already received confirmation that this was a different trial to McKenzie 1984. The other potential trial (Yao 1997) in this category is reported in Chinese. One study (Wajima 1995) previously in 'Studies awaiting assessment' is now included upon being translated from Japanese. Another article, reporting mortality data for an additional 61 trial participants, was identified for McLaren 1978.
In all, 50 studies were identified of which 26 trials were included in this review, 22 were excluded and two are pending. Of the 22 excluded studies: three were not randomised trials; 14 involved comparisons outside the scope of this review; two (Tonczar 1981; Wickstrom 1982) involved neuroleptic general anaesthesia which was considered to be no longer appropriate for hip fracture surgery; one (Darling 1994) only reported one outcome, the rate of clearance of injected indocyanine green, which was considered not to have direct clinical relevance; one (El-Zahaar 1995) involving a mixed population of orthopaedic patients did not provide separate data for hip fracture patients; and one (Dyson 1988) with a factorial design which focused on a comparison outside the review scope, did not provide any results for the spinal versus general anaesthesia comparison. Further details of these are given in the 'Characteristics of excluded studies' table.
The 26 included trials involved a total of 2746 predominantly female and elderly hip fracture patients. Translations were obtained for three trial reports in French, one in German, one in Italian and two in Japanese. Twenty two trials were published as full reports in peer-reviewed journals; the four exceptions (Brichant 1995; Eyrolle 1998; Tasker 1983; Ungemach 1993) being only available as conference abstracts. Two trial reports were available for Davis 1981, one of which focused on a subgroup of trial participants monitored for deep vein thrombosis. Four references, one again which focused on a subgroup of trial participants monitored for deep vein thrombosis, were available for McKenzie 1984. Though these at first appeared to be reports of separate trials, further details supplied by another trialist indicated that all the references applied to one study.
Twenty two included trials involving 2567 patients compared spinal or epidural anaesthesia with general anaesthesia. One study (White 1980) of 40 participants, which compared a 'light' general anaesthetic in conjunction with spinal anaesthesia versus general anaesthesia, is considered separately. A further group of 20 trial participants were allocated to receive a psoas nerve block in conjunction with general anaesthesia, which is outside the scope of this review but included in another Cochrane review on localised nerve blocks (Parker 2001). Two studies compared spinal anaesthesia with nerve blocks (de Visme 2000; Eyrolle 1998). The remaining trial (Spreadbury 1980) compared ketamine anaesthesia with inhalation general anaesthesia in 60 patients.
Further details of the individual trials are given in the 'Characteristics of included studies' table.
Additional information on trial methodology and results would be welcomed from the authors of any of the studies, or from authors of trials that have not been identified.
Risk of bias in included studies
Treatment allocation was considered to be definitely concealed (Cochrane code A) in only one study (McKenzie 1984), which used sealed envelopes and random numbers. Allocation concealment was possible (Cochrane code B) in a further six studies (Brown 1994; Casati 2003; Couderc 1977; de Visme 2000; Maurette 1988; Racle 1986) which gave incomplete details of their methods of randomisation, as well as the 14 studies which did not provide any details. Allocation was not concealed in the only overtly quasi-randomised trial (Adams 1990) which allocated treatment by the date of operation.
The methodology scores using the scoring system described earlier were:
Regional versus general anaesthesia
1 2 3 4 5 6 7 8 9 Total (maximum 11)
0 0 0 1 0 0 0 0 1 2 Adams 1990
1 1 0 1 1 1 1 1 0 7 Berggren 1987
1 1 0 0 0 1 0 0 1 4 Biffoli 1998
1 1 0 1 1 1 1 1 0 7 Bigler 1985
1 1 0 1 1 1 0 0 1 6 Bredahl 1991
1 1 0 0 0 1 1 0 0 4 Brichant 1995
2 1 0 1 0 1 0 0 1 6 Brown 1994
2 1 0 1 1 1 0 0 1 7 Casati 2003
1 1 0 1 1 1 0 0 1 6 Davis 1981
2 1 0 1 0 1 0 1 0 6 Davis 1987
1 1 0 1 1 1 1 0 0 6 Juelsgaard 1998
1 0 0 1 0 1 0 0 1 4 Kamitani 2003
2 1 0 1 0 1 0 0 1 6 Maurette 1988
1 0 0 1 0 1 0 0 1 4 McLaren 1978
3 0 0 0 0 1 0 1 1 6 McKenzie 1984
2 1 0 0 1 1 0 1 1 7 Racle 1986
1 1 1 1 1 1 0 0 1 7 Svarting 1986
1 0 0 0 0 1 0 0 0 2 Tasker 1983
1 0 0 0 0 0 0 0 0 1 Ungemach 1993
1 1 0 1 0 1 1 1 1 7 Valentin 1986
1 0 0 0 0 1 0 0 1 3 Wajima 1995
'Light' general anaesthesia combined with spinal anaesthesia versus general anaesthesia
1 2 3 4 5 6 7 8 9 Total (maximum 11)
1 1 0 1 0 1 0 0 0 4 White 1980
Ketamine versus general anaesthesia
1 2 3 4 5 6 7 8 9 Total (maximum 11)
1 1 0 1 1 0 0 0 1 5 Spreadbury 1980
Two items meriting specific comment are items 3 (intention to treat) and 7 (assessor blinding). Only one trial satisfied the criteria for the first item. The other trials scored zero. Some because no information was available for trial participants or on whether any participants were withdrawn from the study. Others because trial participants who had been withdrawn or excluded were not included in the baseline or outcome analyses, or because an intention to treat analysis was not done. The extent of assessor blinding was usually limited to select outcomes in most of the trials scoring on this item.
Effects of interventions
The outcome measures listed earlier were extracted for each study and, where appropriate data were available, summarised in the graphs. The results are presented using the fixed effect model except where there is statistically significant heterogeneity between study results (P < 0.1) where the random effects model is applied. Since the primary outcome for this review, as stated in the protocol, is mortality, this is considered first. Other outcomes are presented in the categories listed under 'Types of outcome measures'. As explained, these include surrogate or intermediate outcomes, such as peri-operative hypotension, body temperature and arterial blood gases. Although such outcomes may be predictive of important clinical outcomes, the relationship is usually not an exact one and some conditions, such as operative hypotension, may be remedied to reduce the risk of a serious clinical event occurring. Thus the results of such outcomes are not accurate guides of 'hard' clinical outcomes and may be misleading.
Regional versus general anaesthesia
Mortality was not reported in the nine short-term studies, involving 392 patients. (The primary foci of these trials were body temperature (Bredahl 1991), deep vein thrombosis (Brichant 1995), oxygen saturation (Brown 1994), blood pressure and plasma levels of cortisol associated with cementation of a hip prosthesis (Svarting 1986) and cognitive/psychological function (Biffoli 1998; Casati 2003; Kamitani 2003; Maurette 1988; Wajima 1995) respectively.)
Where possible, data for mortality up to one, three, six and 12 months were deduced or extracted from study reports, and pooled, for these four pre-specified time periods. Data for three months and beyond were extracted from graphs for two studies (Davis 1987; Valentin 1986). Additional mortality data were obtained for McKenzie 1984 from another trialist. Mortality data for undefined follow-up periods, or for under one month were provided by four studies (Adams 1990; Bigler 1985; Tasker 1983; Ungemach 1987). The data for the first two studies, which were for early deaths during hospital stay, and those for Ungemach 1987, which were at two weeks, were pooled with those for one month in an extra analysis. Tasker 1983 reported, without providing denominators, that the difference in mortality was not statistically different between the two groups (4 versus 6).
Results for all these studies are shown in the graphs (01.01 to 01.06). The reduced mortality for regional anaesthesia at one month (56/811 (6.9%) versus 86/857 (10.0%)) was of borderline statistical significance when evaluated using the fixed effect method (relative risk (RR) 0.69, 95% confidence interval (CI) 0.50 to 0.95), but not statistically significant when using the random effects model (RR 0.68, 95% CI 0.44 to 1.05). Based on the I-squared statistic (31%), there was some but not substantial heterogeneity/inconsistency between the studies. There was a similar pattern when the results from the three studies (Adams 1990; Bigler 1985; Ungemach 1987), which provided data on deaths during hospital stay or under one month, were pooled with the data for one month mortality (see graph 01.06). The difference in mortality between the two groups was smaller and not statistically significant at subsequent follow-up times: three-months mortality (graph 01.03: 86/726 (11.8%) versus 98/765 (12.8%), RR 0.92, 95% CI 0.71 to 1.21); six-months mortality (graph 01.04: 103/613 (16.8%) versus 115/651 (16.1%), RR 1.04, 95% CI 0.81 to 1.33); and 12-months mortality (graph 01.05: 80/354 (22.6%) versus 78/372 (21.0%), RR 1.07, 95% CI 0.82 to 1.41). Notably, the number of trials and associated data for pooling shrank at each time interval, with only the two largest trials (McKenzie 1984; Valentin 1986) contributing to the 12-months analysis.
Feedback obtained from an external referee (Ballantyne 2004) prompted some consideration of these results in terms of the vintage of the trials contributing data and the high mortality figures in McLaren 1978. Two actions were taken. Firstly, the trials were ordered according to their year of publication (Reader, please sort by year for graph 01.01). This reveals a potential, but statistically untested, trend in the results towards a reduced early mortality for regional anaesthesia in earlier studies. Secondly, removal of the data for McLaren 1978, which has an unusually high mortality rate in the general anaesthesia group (28%), resulted in a statistically non-significant difference in mortality at one month (RR 0.79, 95% CI 0.56 to 1.12: graph not shown).
(a) Peri-operative outcomes
Length of operation
Most studies that recorded this outcome reported a statistically non-significant increase in the time taken to complete the operation for regional anaesthesia (Adams 1990; Berggren 1987; Bigler 1985; Maurette 1988; McKenzie 1984; Racle 1986). One study (Svarting 1986) had a significant increase and two studies (Bredahl 1991; Kamitani 2003) a non-significant increase for general anaesthesia. Five studies found no difference between the two groups (Biffoli 1998; Casati 2003; Davis 1981; Juelsgaard 1998; White 1980). Prior to the inclusion of data from two new trials (Kamitani 2003; Svarting 1986), the pooling of data from six studies showed a statistically significant increase of around five minutes for regional anaesthesia (weighted mean difference 4.8 minutes, 95% CI 1.1 to 8.6 minutes). This has now changed in that there is now no statistically significant difference between the two groups. Moreover, the result from Svarting 1986 is significantly different from the other trials; the addition of this trial changed the chi squared value from 7.45 (P = 0.28) to 15.95 (P = 0.03) and the I-squared value from 20% (low heterogeneity) to 56% (substantial heterogeneity). The adoption of the random effects model (shown in graph 01.07) shows minimal difference between the two groups (weighted mean difference 0.8 minutes, 95% CI -5.4 to 6.9 minutes).
The definition of hypotension, when stated, was a greater than a 30% reduction in systolic blood pressure (Berggren 1987; Svarting 1986); a 33% fall (Juelsgaard 1998); a 40 mmHg fall (Couderc 1977); and a 20% fall from the baseline in four studies (Casati 2003; Davis 1987; Maurette 1988; Racle 1986).
Two studies (Adams 1990; Davis 1981) stated, without data for pooling, that the drop in systolic blood pressure was significantly greater in the regional anaesthesia group. Bigler 1985 reported no significant difference in the maximum drop of systolic blood pressure (48 versus 51 mmHg). Pooling of data from 11 studies (Berggren 1987; Biffoli 1998; Brown 1994; Casati 2003; Couderc 1977; Davis 1987; Juelsgaard 1998; Maurette 1988; McLaren 1978; Racle 1986; Svarting 1986) showed hypotension to be more common after regional anaesthesia. This difference was statistically significant when viewed using the fixed effect model (graph 01.08: 172/501 (34.3%) versus 137/521 (26.3%), RR 1.30, 95% CI 1.08 to 1.55) but not when adopting the random effects model (graph 01.09: RR 1.10, 95% CI 0.79 to 1.55), which is probably more appropriate given the significant heterogeneity of trial results (chi squared = 21.23, P = 0.01; I-squared = 57.6%). This significant heterogeneity persisted when we explored the effect of removing each of the trials in turn.
Operative blood loss
Pooled data for five studies (Bredahl 1991; Davis 1981; Kamitani 2003; McKenzie 1984; Svarting 1986) showed a statistically significant decrease in operative blood loss for regional anaesthesia (graph 01.10: weighted mean difference -85 ml, 95% CI -162 to -9 ml). There was substantial heterogeneity in these trial results (I-squared = 53%). Five other studies contained insufficient data to enable pooling. Adams 1990 and Juelsgaard 1998 reported a non-significant increase in blood loss for regional anaesthesia; McLaren 1978 reported no significant difference; Ungemach 1987 reported no difference; and Casati 2003 and Valentin 1986 reported a significantly increased blood loss in the general anaesthesia group.
Nine studies gave data for blood transfusion, which are presented as either the numbers of patients who were transfused in four studies (Adams 1990; Bigler 1985; Davis 1981; Svarting 1986), or the mean volume of blood transfused (transfusion requirement) (Couderc 1977; Juelsgaard 1998; Kamitani 2003; Maurette 1988; Racle 1986). Similar proportions of patients received transfusion in each group in the first four studies (graph 01.11: 64/123 (52.0%) versus 73/135 (54.1%)). In contrast the transfusion requirements were greater in the regional anaesthesia group but there was significant heterogeneity (chi squared = 30.27, P < 0.00001; I-squared = 90.1%) in the trial results and the pooled result applying the random effects model was not statistically significant (graph 01.12: weighted mean difference 100 ml, 95% CI -53 to 252 ml). Juelsgaard 1998 reported statistically non-significantly lower mean values of blood volume transfused over the operative and peri-operative period for the regional anaesthesia group (237 ml versus 257 ml). Bigler 1985 reported the mean falls in haemoglobin to be greater in the regional anaesthesia group (22% versus 19%, not significant). Kamitani 2003 found no difference between the two groups in postoperative haemoglobin levels.
Pre- and postoperative arterial blood gases
The reports of eight studies (Berggren 1987; Brown 1994; Couderc 1977; Davis 1981; Kamitani 2003; McLaren 1978; McKenzie 1984; Svarting 1986) contained data for blood gases taken either preoperatively, operatively or postoperatively. In addition, Biffoli 1998 reported on postoperative hypoxia. Pooled data from Biffoli 1998 and Berggren 1987, which reported numbers of trial participants with postoperative arterial oxygen tension of less than 60 mmHg, are presented in graph 01.13 (11/58 (19%) versus 17/59 (29%), RR 0.67, 95% CI 0.36 to 1.22). Brown 1994, in a study of postoperative oxygen saturation in 20 people, found significantly lower oxygen saturation for the group who received general anaesthesia. Davis 1981 reported that the general anaesthesia group showed a postoperative fall in oxygen saturation in the early postoperative period, which was not seen after regional anaesthesia. By the first postoperative day there was no significant difference between the two groups. Kamitani 2003 reported no statistically significant difference between the two groups in oxygen saturation levels postoperatively and at 12 and 18 hours; however, the oxygen saturation was 1.6% higher in the general anaesthesia group at six hours. McKenzie 1984 reported a significant decrease in the oxygen saturation at one hour postoperatively in those who received general anaesthesia compared with those who received regional anaesthesia. In contrast, two studies (Couderc 1977; McLaren 1978) reported no difference in the mean arterial oxygen or carbon dioxide tensions for the two types of anaesthesia. Svarting 1986 observed a distinct deterioration in arterial oxygen tension in both groups on cementation of a Thomson prosthesis; additional oxygen was considered necessary for eight spinal anaesthesia patients during their operations.
Length of hospital stay
Most studies reporting this found no difference in the length of hospital stay. Juelsgaard 1998 observed that the results for hospital stay were affected by a lack of rehabilitation facilities. Adams 1990 reported 21 days for regional versus 20 days for general anaesthesia. Berggren 1987 stated there was no difference in length of hospital stay between the two groups, as did Casati 2003 (median stay: 12 versus 14 days). Davis 1987 reported an average of 16 days for both groups, and Racle 1986, 20 days for both groups. Valentin 1986 reported a median stay of 10 days for regional anaesthesia and 11 days for general anaesthesia. Finally, McKenzie 1984 recorded a mean of 38 days for regional anaesthesia against 43 days for general anaesthesia. Summation of the two studies which quoted standard deviations (McKenzie 1984; Racle 1986), shown in the graphs, demonstrated no difference in the length of hospital stay between groups (graph 01.14: weighted mean difference -0.2 days, 95% CI -5.2 to 4.8 days).
Other peri-operative outcomes
Other peri-operative outcomes recorded were changes in body temperature (Bredahl 1991), serum catecholamine and endocrine levels (Adams 1990; Svarting 1986; Tasker 1983), bradycardia (Casati 2003), ECG changes (Juelsgaard 1998), pain and analgesic use (Casati 2003; Kamitani 2003), and time to ambulation (Bigler 1985; Valentin 1986). Ungemach 1993 used a scoring system "calculated from data on consciousness, respiration, circulation, laboratory tests and blood loss".
Bredahl 1991, who recorded body temperatures of 30 patients, concluded that temperature changes during the peri-operative period were unrelated to the type of anaesthesia.
Adams 1990 reported raised serum adrenaline and noradrenaline levels at the end of the operation for a subgroup of 32 trial participants, the rise in levels being greater in those who received a general anaesthetic. Svarting 1986 found significantly increased plasma cortisol levels after cementation in the general anaesthesia group. Tasker 1983, in a study of 100 patients, reported a significantly greater increase in plasma noradrenaline and cortisol levels after general anaesthesia in comparison with regional anaesthesia. There was no report of intraoperative cardiac arrhythmias.
Bradycardia (heart rate < 50 beats/min) was observed in three patients during general anaesthesia in Casati 2003. Juelsgaard 1998 reported a significant increase in the overall number of ST segment depressions for those in the spinal anaesthesia group (125 versus 16 events).
Pain at one hour post surgery was worse in the general anaesthesia group in Casati 2003 but not at three, six or 12 hours. Kamitani 2003 reported there was no difference between the two groups in diclofenac use.
Bigler 1985 reported a significant reduction in the mean time from surgery to ambulation of 3.3 days after regional anaesthesia versus 5.1 days after general anaesthesia. Valentin 1986 however reported no difference in the time to ambulation for patients in the two groups.
Postoperative scores (calculated from "data on consciousness, respiration, circulation, laboratory tests and blood loss") in Ungemach 1993 were reported as "better" in the spinal group, but it was not clear by how much and how this was manifested.
(b) Complications specific to the method of treatment
Davis 1981 was the only study to report on aspiration pneumonia, with two cases in the general anaesthesia group. These have been included under the complication of pneumonia. A persistent headache, lasting three days, in one person in the spinal anaesthesia group was noted in Bigler 1985. McLaren 1978 reported that there were no post-anaesthetic headaches. There was no mention within the included studies of other complications such as damage to the upper airways or mouth from equipment used during general anaesthesia.
Failure of spinal anaesthesia, usually resulting in the secondary use of general anaesthesia, was reported in both studies conducted by Davis et al (Davis 1981; Davis 1987). Spinal anaesthesia, often performed by junior staff, was unsuccessful in eight out of 72 patients (11.1%) in Davis 1981 and in 30 out of 259 patients (11.6%) in Davis 1987. Davis 1987 also referred to a 10% failure rate in the study of Valentin 1986. The treatment of these spinal anaesthesia failures in the analyses presented by these three trials has further implications regarding intention to treat analysis. For instance, it may be that the excluded trial participants had different characteristics and outcomes than those participants in which spinal anaesthesia was successful. The eight patients in Davis 1981 were incorrectly analysed in the general anaesthesia group, whereas the 30 patients in Davis 1987 were analysed in the spinal anaesthesia group, and lastly, Valentin 1986 chose to exclude them from the analysis.
(c) General postoperative complications
Data for most of the life threatening complications such as pneumonia, myocardial infarction, cerebral vascular accident, congestive cardiac failure and pulmonary embolism were only available as causes for deaths in many of the trial reports. To reflect this, the data from fatal events have been subgrouped separately from those listed as complications, or not wholly associated with deaths, in trial reports.
Pneumonia or 'chest infection' was reported in nine studies (Adams 1990; Berggren 1987; Bigler 1985; Davis 1981; Davis 1987; Juelsgaard 1998; McKenzie 1984; McLaren 1978; Racle 1986). Pooling of the results indicates no statistically significant difference between the two anaesthetic methods (graph 0.15: 21/574 (3.7%) versus 29/612 (4.7%), RR 0.76, 95% CI 0.44 to 1.30).
This complication was reported in seven studies (Couderc 1977; Davis 1981; Davis 1987; Juelsgaard 1998; McKenzie 1984; McLaren 1978; Racle 1986). Summation of the results from six trials showed no statistically significant difference in myocardial infarction between the two groups (graph 01.16: 5/502 (1.0%) versus 11/531 (2.1%), RR 0.55, 95% CI 0.22 to 1.37).
This complication was reported in seven studies (Berggren 1987; Bigler 1985; Couderc 1977; Davis 1981; Davis 1987; McKenzie 1984; Racle 1986). Pooling of results showed no statistically significant difference in cerebrovascular accidents between the two groups (graph 01.17: 10/529 (1.9%) versus 6/556 (1.1%), RR 1.51, 95% CI 0.64 to 3.57).
Congestive cardiac failure
This complication was reported in seven studies (Adams 1990; Berggren 1987; Bigler 1985; Davis 1981; Davis 1987; Juelsgaard 1998; Racle 1986). Pooling of data showed no statistically significant difference between the two groups (graph 01.18: 12/454 (2.6%) versus 12/477 (2.5%), RR 1.05, 95% CI 0.49 to 2.23).
Renal failure was reported in five studies (Adams 1990; Davis 1981, Davis 1987; McLaren 1978; Racle 1986). Summation of results in the graph demonstrated no statistically significant difference between anaesthetic techniques (graph 01.19: 3/438 (0.7%) versus 5/474 (1.1%), RR 0.76, 95% CI 0.23 to 2.49).
Post operative cardiac arrhythmia
More abnormal cardiac rhythms were detected in the general anaesthesia group in Couderc 1977. However, Couderc 1977 reported that there was no difference in the overall electrocardiographic results; these included results for other peri-operative changes in the cardiogram.
Acute confusional state
This complication and/or underlying cognitive dysfunction was reported in seven small studies (Berggren 1987; Biffoli 1998; Bigler 1985; Casati 2003; Kamitani 2003; Racle 1986; Wajima 1995). Summation of the limited results showed a significant reduction in the regional anaesthesia group (graph 01.20: 11/117 (9.4%) versus 23/120 (19.2%), RR 0.50, 95% CI 0.26 to 0.95). In the subgroup of 38 patients who were not confused at trial entry, Biffoli 1998 found slightly better result in the spinal anaesthesia group (mean difference -0.97, 95% CI -1.91 to -0.03) in the 'Organic Brain Syndrome' score (0: no confusion to 36: total confusion) at 48 hours. Wajima 1995 reported no statistically significant differences between the two anaesthetic groups in the Hasegawa Dementia Scale scores at one week post surgery.
Pooling of the data from the two studies (Berggren 1987; Bigler 1985) reporting this complication showed similar results for the two anaesthetic techniques (graph 01.21: 10/48 (20.8%) versus 10/49 (20.4%), RR 1.02, 95% CI 0.47 to 2.23). Kamitani 2003 found no difference in the urine output of the two groups.
Pooling of the data from the two studies (Bigler 1985; McLaren 1978) reporting this complication again showed similar results for the two anaesthetic techniques (graph 01.22: 2/46 (4.3%) versus 3/49 (6.1%), RR 0.70, 95% CI 0.12 to 3.94).
Deep vein thrombosis
Deep vein thrombosis was the primary outcome for one study (Brichant 1995), and for two subgroups of patients from a further two studies (Davis 1981; McKenzie 1984). Awareness of the risk of deep vein thrombosis was evident in several other studies who did not report this outcome, with various prophylactic interventions being deployed: Dextran 70 (Berggren 1987); early mobilisation (Bigler 1985); anti-vitamin K and early mobilisation (Couderc 1977); heparin and active movement (Racle 1986) and anti-embolic stockings (Valentin 1986). Patients in Brichant 1995 also received thromboembolic prophylaxis with low molecular weight heparin and anti-embolism stockings. Venography screening was used to detect deep vein thrombosis in two studies (Brichant 1995; McKenzie 1984) and fibrinogen scanning in Davis 1987. Pooled data, grouped by method of diagnosis, include two deaths whose underlying cause was deep vein thrombosis from McLaren 1978. Significantly fewer thromboses were detected in patients in the regional anaesthesia group (graph 01.23: 39/129 (30%) versus 61/130 (47%); RR 0.64, 95% CI 0.48 to 0.86). Though the difference in incidence rates was consistent between trials, whether measured by venography, fibrinogen update or at post-mortem, these results have to be viewed with caution since these were the results of subgroups of patients for whom data from venography or fibrinogen were available. In turn, the patients specially monitored for deep vein thrombosis were also subgroups of the trial populations in two studies (Davis 1981; McKenzie 1984).
Pulmonary embolism was reported in 10 studies (Adams 1990; Berggren 1987; Bigler 1985; Brichant 1995; Couderc 1977; Davis 1981; Davis 1987; McKenzie 1984; McLaren 1978; Racle 1986) but mostly as a reason for death rather than through active monitoring for non-fatal pulmonary embolism. Pooling the results from nine studies using Peto odds ratios (graph 01.24) showed statistically significant heterogeneity (chi squared = 15.11, P = 0.06; I-squared = 47.1%). Summation of results from nine studies using the random effects model to allow for this heterogeneity showed no statistically significant difference in overall incidence of pulmonary embolism in the two groups (graph 01.25: 9/605 (1.5%) versus 13/640 (2.0%), RR 0.88, 95% CI 0.32 to 2.39). The source of heterogeneity resides mainly in the significantly different results in trials presenting results for fatal pulmonary embolism, and those presenting results for non-fatal pulmonary embolism. These are analysed in separate subgroups in graph 01.26 (test for interaction, based on Peto odds ratio results: P = 0.003).
Ungemach 1993 used a scoring system that was "based on laboratory tests, cardiopulmonary evaluation and complications such as heart failure, thrombosis and apoplexy". No difference between the two groups was found in the scores at two weeks.
(d) Final outcome measures
Mortality has already been considered above.
Changes in mental function
Two studies (Bigler 1985; Maurette 1988) reported on long-term changes in mental function. Bigler 1985 reported that there was no persistent impairment in mental function, and no significant differences between the two groups in the mental scores achieved at three months. Maurette 1988 performed psychological evaluations on 33 patients and found no significant difference relating to the type of anaesthesia.
No study reported on the difference in functional outcomes between groups. Only McKenzie 1984 provided limited data on the location of trial participants at 12 months, but not for the return of participants to their previous residence.
'Light' general anaesthesia combined with spinal anaesthesia versus general anaesthesia
The only study (White 1980) in this category involved only 20 patients in each group. No trial participants died within the one-month follow up period of the study. The mean length of operation was 58 minutes in both groups. There was no significant difference in the mean postoperative blood oxygen or carbon dioxide levels between the two groups. Complications reported were pneumonia (4 versus 5 cases), confusional states (3 in each group), deep vein thrombosis (1 in the general anaesthesia group) and postoperative vomiting (1 in each group). Results for most of these outcomes are presented in the graphs (02.01 to 02.05).
Regional (spinal) anaesthesia versus local nerve blocks
Two studies, involving 79 patients, were included. One study (Eyrolle 1998) compared spinal anaesthesia with a lumbar plexus block in 50 people; supplementary intravenous propofol sedation was performed when necessary. The other study (de Visme 2000) compared spinal anaesthesia with a lumbar plexus block in conjunction with a sacral plexus block and iliac crest block (for lateral cutaneous nerve of the thigh). Intravenous alfentanil or sedatives were also used if necessary. Both studies only reported on outcome during the peri-operative period and did not report on postoperative complications or mortality. Results where available and appropriate are given in the graphs.
In Eyrolle 1998, the need for propofol supplementation, of dosage greater than 1 mg/kg/hr, was significantly less common in the spinal group (5/25 versus 19/25). No cases of incomplete or unsatisfactory anaesthesia in the spinal group were reported in de Visme 2000 as opposed to four cases of incomplete anaesthesia and one case, requiring repeated sedation that was judged as unsatisfactory, in the nerve block group (0/14 versus 5/15). Overall, the need for supplementary sedation was significantly less in the spinal group (graph 03.01: 5/39 versus 24/40; RR 0.23, 95% CI 0.10 to 0.50).
A fall in mean arterial blood pressure of more than 20% occurred in significantly more patients in the spinal group (graph 03.02: 18/25 versus 3/25; RR 6.0, 95% CI 2.02 to 17.83) in Eyrolle 1998. The mean fall in arterial blood pressure was also significantly greater in the spinal group in de Visme 2000 (graph 03.03: mean difference 16 mmHg, 95% CI 1.3 to 30.7 mmHg). In both trials, significantly higher doses of ephedrine were used to stabilise blood pressure in the spinal group (graph 03.04: weighted mean difference 5.96 mg, 95% CI 4.46 to 7.45 mg).
Pain as measured by the visual analogue scale (VAS) was stated as showing no difference between groups in Eyrolle 1998. Eleven trial participants failed to complete the VAS in de Visme 2000, who considered that VAS rating for pain was unsatisfactory when there were cases of "sensorial" deficiency.
"Insertion difficulty" was significantly more common in the spinal group in Eyrolle 1998 (10/25 cases versus 3/25). In contrast, the mean time to administer the spinal was reported as being statistically significantly lower in the spinal group in de Visme 2000 (12 versus 18 minutes; reported P = 0.013).
Adverse effects, including five cases of urinary retention, were more common in the spinal group in Eyrolle 1998 (graph 03.05: 6/25 versus 1/25; RR 6.00, 95% CI 0.78 to 46.29). No adverse effects of the techniques were reported by de Visme 2000.
Postoperatively, similar numbers of trial participants had impaired cognitive function in de Visme 2000 (graph 03.06: 5/14 versus 6/15); this was reflected in the comparable mini-mental test scores (mean 15.5 versus 14.5).
Ketamine versus general anaesthesia
The only study included in this category (Spreadbury 1980) involved 60 female patients. The limited results available are summarised in the graphs (04.01 to 04.05). Data were presented for early deaths (within 14 days) and late deaths (time unspecified, in hospital). These showed no difference in the overall mortality during hospital stay (9/30 (30%) versus 9/30 (30%)). Data presented for the complications of myocardial infarction (1 case), congestive cardiac failure (2 cases) and pulmonary embolism (3 cases) were all derived from causes of death for the seven early deaths.
The mean length of hospital stay for the 39 trial participants who returned home was 36 days for the ketamine group against 24 days for the general anaesthesia group. This difference is statistically significant and is related to the higher incidence of unsatisfactory surgical results in the ketamine group (see below). Although the general anaesthesia group mobilised more quickly than the ketamine group, Spreadbury 1980 reported that the differences were not statistically significant. The proportions of people who returned home were similar (19/30 versus 20/30).
Spreadbury 1980 also reported that the numbers of patients who experienced dreams and hallucinations were similar for the two groups (4 versus 5 patients). They stated however that the dreams were more likely to be unpleasant after general anaesthesia. Spreadbury 1980 also reported the incidence of unsatisfactory surgical results, either due to later dislocation of the prosthesis or an unstable fixation, which subsequently required bed rest or traction. There were 7/30 (23%) such cases for the ketamine group against 3/30 (10%) for general anaesthesia.
Regional versus general anaesthesia
Many of the studies within this review involved small numbers of participants and reported only a few outcome measures. The trial reports of all studies indicated a poor level of methodological rigour, in particular regarding concealment of allocation, assessor blinding and intention to treat analysis. Despite these limitations, there is a reasonable agreement between trials for many of the outcome measures reported, particularly for mortality. It remains possible that some of the differences in outcome within the studies could be related to the differences in the experience, and competence, of the anaesthetists. Inexperience with the anaesthetic techniques could be inferred in some studies. For example, there was a high failure rate of spinal anaesthesia, often performed by junior staff, of over 11% in both Davis 1981 and Davis 1987. However, there was no evidence that the seniority of the anaesthetists applying the different methods of anaesthesia differed in any given trial. A further consideration, raised at editorial review (Ballantyne 2004), is that many of the included trials are relatively old and do not represent contemporary practice nor account for the advances in safety in the field of anaesthesia.
Hip fractures occur predominantly in the frail elderly who have multiple other medical conditions. The high mortality within this group of patients often results from these other medical conditions rather than being a direct consequence of the hip fracture and its treatment. Regional anaesthesia may reduce short-term mortality, yet this finding is borderline in that it is statistically significant when using the fixed effect model but not with the random effects model. Notably, both models give very similar point estimates of effect (relative risk = 0.69 (fixed) and 0.68 (random)). The effect of the removal of the oldest trial (McLaren 1978), which has an excessive mortality in the general anaesthesia group, also shows the weakness of the evidence. Nonetheless, the three-months mortality result retains a potential for a reduction in mortality in the regional anaesthesia group. There is no evidence of substantial differences between regional and general anaesthesia in terms of long-term mortality, although the small numbers of people with long-term follow up, available from two trials of poor methodology, means that we cannot exclude clinically relevant differences. It is plausible that changing one aspect of hip fracture treatment (the type of anaesthesia) could affect long-term mortality: potentially, regional anaesthesia could enable the group of very frail elderly to survive the initial surgery, only for death to ensue later from other medical complications.
In their comprehensive review of regional anaesthesia, Rodgers et al (Rodgers 2000) found that postoperative mortality up to 30 days was significantly reduced for all types of surgery (general, orthopaedic, urological and vascular) by about a third (odds ratio = 0.70, 95% CI 0.54 to 0.90). This is consistent with the point estimate of effect in this review. Rodgers et al concluded that their findings supported a "more widespread use of neuraxial blockade [spinal/epidural anaesthesia]". It was notable that over half the trials with at least 10 deaths per trial involved patients with hip fracture; thus enhancing the contribution of the findings of these trials to the overall result. Rodgers 2000 considered that a lack of statistical power in individual trials and meta-analyses could be the principal reason for a conclusion that "neuraxial blockade had no important effect on mortality". In fact, our conclusions are still phrased in a more tentative way than Rodgers et al imply and, although there is a lack of statistical power in our review, we also consider that there is an important lack of longer-term outcome data. Given that the evidence from the trials in our review of hip fracture is insufficient to show a statistically significant reduction in mortality at one month, an important consideration is whether we should draw on the results from Rodgers 2000 in making our conclusions. The comparison of regional versus general anaesthesia has been identified as a clinical question where the focus on a subgroup of studies (i.e. on hip fracture patients) could miss an important effect; namely the mortality associated with general anaesthesia compared with regional anaesthesia across different surgical procedures (Oxman 2001). There are others, however, who are not convinced that the findings of Rodgers 2000 apply or should be applied to all surgical groups (McCullock 2001; Hughes 2000). Also questioned, and pertinent to both reviews, is the relevance of older trials to current anaesthetic practice (Higham 2001). Anaesthetic techniques, equipment and drugs have changed in recent years, with important advances in safety, and there is often greater use of antithrombotic prophylaxis. The inclusion of older studies in Rodgers 2000 and in previous versions of this review may therefore have biased results in favour of regional anaesthesia. There is general agreement that further research is warranted.
Returning to this review on hip fractures. Because of the low incidence of many of the complications following surgery, no individual study had numbers large enough to determine if any difference exists. As much of the data for many of these complications was for fatal complications, these results are far from complete. Some possible, although unconfirmed, trends for regional anaesthesia were for less myocardial infarction, more cerebrovascular accidents and less fatal pulmonary embolism but more non-fatal pulmonary embolism. Pooled data from five small trials, involving a total of 237 patients, showed significantly fewer cases of acute confusion when regional anaesthesia was used. This was a consistent finding between five trials, of which four specifically focused on mental functioning, and was supported by the results of two other trials with the same focus. These attributes strengthen the finding of lower incidence of acute confusion with regional anaesthesia but it would still be prudent to have confirmation of this and, importantly too, to gather evidence on whether there is persistent impairment in mental function.
Pooled results for deep vein thrombosis showed a statistically significant reduction in the incidence of deep vein thrombosis in the regional anaesthesia group. This should not be considered conclusive as the data were from subgroups of trial participants who had been 'selected' by their compliance with a method of diagnosis, and thus the effect, and certainly the effect size, may have been distorted. The effects of thromboembolic prophylaxis may also affect the incidence of thromboembolic complications. The routine use of thromboembolic prophylaxis was mentioned in six studies (Berggren 1987; Bigler 1985; Brichant 1995; Couderc 1977; Racle 1986; Valentin 1986). It is also possible that thromboembolic prophylaxis may have been withheld in those receiving regional anaesthesia in some studies. The results do suggest a trend towards a reduced risk of thromboembolic complications with regional anaesthesia but, because of the small number of trials that reported this outcome and the heterogeneity of results, firm conclusions cannot be made for this outcome.
In our previous update (Issue 4, 2001), where we reported that operations with regional anaesthesia took approximately 5 to 10 minutes longer than general anaesthesia, we stated that such a finding would be expected from clinical practice. We suggested that may be due to the time taken to administer the regional anaesthesia and then the time taken for the analgesic effect to occur. The inclusion of evidence from two new trials, which resulted in a finding of no significant difference, undermines our previous statements and emphasises the lack of robustness in the underlying evidence. Another statistically significant finding is the increased blood loss for general anaesthesia from data pooled from five trials. However, where reported, similar numbers received blood transfusion and transfusion requirements were greater in the regional anaesthesia group in some trials. Though these results seem contradictory, "one explanation might be that, because regional anaesthesia decreased blood pressure more than general anaesthesia, blood transfusion was triggered due to the haemodynamic state rather than as a consequence of blood loss, particularly if during regional anaesthesia the patient's blood had become diluted with clear intravenous fluids" (Carlisle 2004).
Regional anaesthesia results in vasodilatation of the lower limbs and this results in an increased tendency to operative hypotension, as demonstrated by the results. In addition, the increased blood flow to the lower limbs with alterations in coagulability and viscosity of the blood, may be the reason for the reduced incidence of venous thrombosis. It is possible that the benefits of the reduced thromboembolic complications may be negated if thromboembolic prophylaxis is used .
There was a tendency for more hypotension with regional anaesthesia. This may result in a predisposition to an increased incidence of cerebrovascular complications as hypotension is one of the aetiological factors for this complication. However, there are insufficient data to confirm this in this review and the care that needs to be taken in the interpretation of surrogate outcomes, such as hypotension, has already been mentioned (see start of 'Results').
Juelsgaard 1998 specifically targeted patients with known coronary artery disease. Whilst appropriate, the numbers of participants in the trial were too small to determine which type of anaesthesia is best for this specific patient population.
None of the trials evaluated cost. Though regional anaesthesia is cheaper with respect to drug costs incurred during the administration of the anaesthetic, cost evaluation should cover the whole process.
'Light' general anaesthesia combined with spinal anaesthesia versus general anaesthesia
The sole study to address this question (White 1980) involved only 20 participants in each group. There was no statistically significant difference between techniques for any of the outcome measures reported. Because of the small numbers of participants involved, no conclusions about the lack of difference between the two techniques can be made.
Regional (spinal) anaesthesia versus local nerve blocks
The two included trials (de Visme 2000; Eyrolle 1998) involved only 79 participants in total. In addition there was incomplete reporting of outcomes and no follow up of trial participants. The limited results available suggest that the local nerve blocks are associated with a reduced risk of operative hypotension but have a greater risk of incomplete or unsatisfactory analgesia. Because of the limited information, no conclusions can be made on the use of nerve blocks compared with spinal anaesthesia.
Ketamine versus general anaesthesia
The sole trial (Spreadbury 1980) comparing ketamine with general anaesthesia involved only 60 participants. The only key difference was a reduction in the 14-day mortality for ketamine, which related to a reduction in the risk of early fatal thromboembolic complications. However, this difference in mortality did not persist, and the mortality during hospital stay was equal in both groups. The numbers of participants were too small to show if the increase in 'unsatisfactory surgical results' in the ketamine group was a significant factor of ketamine use.
Implications for practice
Overall, there was insufficient evidence available from randomised trials comparing regional versus general anaesthesia for hip fracture surgery to confirm or rule out clinically important differences. In addition, the relevance of evidence from older trials in the context of current anaesthetic and peri-operative practice is unclear. Based on the available evidence, regional anaesthesia may reduce acute postoperative confusion but no definite conclusions can be drawn for mortality or other outcomes.
Due to the limited data available, it is not possible to determine the roles of nerve blocks, ketamine or spinal anaesthesia with 'light' general anaesthesia for hip fracture anaesthesia.
Implications for research
Well designed randomised trials, with active follow up of at least six months, of regional versus general anaesthesia involving large numbers of patients and which record, at minimum, the primary clinical outcomes of death, postoperative complications, and long-term outcomes, would help clarify the relative merits of regional and general anaesthesia in contemporary health care practice. Large trials with subgroup analysis may be able to determine if patients with specific medical conditions (such as cardiac disease, previous stroke) are better managed with one of these two forms of anaesthesia.
Given the importance of assessing outcome based on what matters to patients, qualitative studies of patient preferences and values would inform this topic.
We thank Susan Urwin for her contribution as a co-reviewer to the original version of the review.
We would like to thank the following for useful comments from editorial review of the original review: Gordon Drummond (Department of Anaesthetics, University of Edinburgh), William Gillespie, Rajan Madhok, Gordon Murray, Tom Pedersen (Department of Anaesthesiology, Copenhagen University Hospital) and Marc Swiontkowski. We thank William Gillespie, Leeann Morton and Lesley Gillespie for their help with the first update. For the second update, we are grateful to Lesley Gillespie, William Gillespie, Peter Herbison, Leeann Morton, Tom Pedersen, Janet Wale and Tony Wildsmith for their assistance and helpful feedback at editorial review.
For this update, we are very grateful to Jayne Elms, William Gillespie, Lesley Gillespie, Peter Herbison and Janet Wale of the Musculoskeletal Injuries Group, and Jane Ballantyne, John Carlisle, Jane Cracknell and Ann Moller of the Anaesthesia Group for their assistance and/or helpful and insightful feedback at editorial review.
We are indebted to Lesley Gardner and Norifumi Kuratani for translations.
Helen Handoll's work on the first two versions of the review was supported by the Chief Scientist Office, Department of Health, The Scottish Office, UK.
Data and analyses
- Top of page
- Authors' conclusions
- Data and analyses
- What's new
- Contributions of authors
- Declarations of interest
- Sources of support
- Index terms
Appendix 1. Search strategy for EMBASE (OVID-WEB)
Last assessed as up-to-date: 11 June 2004.
Protocol first published: Issue 4, 1997
Review first published: Issue 4, 1999
Contributions of authors
Martyn Parker (MP) initiated the review and wrote the first draft of the protocol. Helen Handoll (HH) identified the trial studies. Susan Urwin and Richard Griffiths independently assessed trial quality and extracted data. The other two reviewers (HH and MP) independently checked these results and entered the review into RevMan. All reviewers critically reviewed successive drafts of the review. The updates were compiled by MP and HH with RG independently extracting data. Martyn Parker is the guarantor of the review.
Declarations of interest
Sources of support
- University of Teesside, Middlesbrough, UK.
- Peterborough and Stamford Hospitals NHS Foundation Trust, Peterborough, UK.
- No sources of support supplied
This review and first update was published under the title: "General versus spinal/epidural anaesthesia for surgery for hip fractures in adults". The title was changed in the second update to reflect an expansion in the scope of the review to include comparisons of all forms of anaesthesia.
This review was first updated in Issue 4, 2000. The trial search was updated to August 1999 and one small trial (Juelsgaard 1999) was included. A consumer synopsis was added and relative risks instead of Peto odds ratios were presented for dichotomous outcomes. There were no significant changes to the conclusions of the review.
The second update, first appearing in Issue 4, 2001, involved an expansion of the scope of the review to include comparisons of all forms of anaesthesia; as reflected in the changed review title. The trial search was updated to December 2000. Three new trials were included; one comparing general versus spinal anaesthesia (Ungemach 1993) and two (Eyrolle 1998; de Visme 2000) comparing spinal anaesthesia with lumbar plexus blocks. Considerations of surrogate outcomes led to a slight amendment to the conclusions of the review.
Medical Subject Headings (MeSH)
MeSH check words
Aged; Female; Humans; Male
* Indicates the major publication for the study