Fractures of the distal radius are one of the most common fractures in many predominantly white and older populations (Sahlin 1990; Singer 1998). It has been estimated that a 50 year old white woman in the USA or Northern Europe has a 15 per cent lifetime risk of a distal radius fracture; whereas a white man of the same age has a lifetime risk of a little over two per cent (Cummings 1985). A recent prospective survey, conducted in six centres in the UK, of Colles' fracture in patients aged 35 years and above, reported the overall annual incidence of this fracture to be 9/10,000 in men and 37/10,000 in women (O'Neill 2001). Distal radial fractures are usually treated on an outpatient basis, with around 20 per cent of patients (mainly older people) requiring hospital admission (Cummings 1985; O'Neill 2001).
Most fractures of the distal radius in older people result from low-energy trauma, such as a fall from standing height or less. In younger adults, these injuries are usually sustained through high-energy trauma, such as a traffic accident. The pattern of incidence reflects the bone loss from osteoporosis in older people as well as an increased number of falls by older women (Nguyen 2001).
These fractures are generally closed and usually involve displacement of fracture fragments. They may be either extra-articular (leaving the articular surface of the distal radius intact) or intra-articular (the articular surface is disrupted). Numerous classifications have been devised to define and group different fracture patterns (Chitnavis 1999). Simple classifications based on clinical appearance, and often named after those who described them, remain in common use. In particular, "Colles' fracture" is still the terminology used for a fracture in which there is an obvious and typical clinical deformity - of dorsal displacement, dorsal angulation, dorsal comminution (small fragments of bone), and radial shortening.
The majority of distal radial fractures are treated conservatively (non-operatively). This usually involves reduction under anaesthesia of the fracture if displaced, and forearm immobilisation in a plaster cast or brace for around six weeks. Questions over the use and timing of fracture reduction and of conservative methods used to stabilise the reduced fracture are covered in a separate Cochrane review (Handoll 2003a). A separate review of methods of closed reduction, some of which were trialled without anaesthesia, is now available (Handoll 2003c). Also available is a Cochrane review of surgical treatment, which usually involves either closed or open reduction followed by external or internal fixation and a similar period of immobilisation (Handoll 2003b).
A variety of options are available for anaesthesia:
- 'Haematoma block' where a local anaesthetic is injected into the fracture site.
- Intravenous regional anaesthesia (IVRA) involving intravenous infusion of a local anaesthetic into the arm after draining the venous system and applying a tourniquet/cuff at greater than arterial blood pressure. This is commonly referred to as a Bier's or Bier block, after the originator of this general method for limb anaesthesia.
- Regional nerve blocks such as radial, median, and ulnar nerve blocks at the elbow or in the axilla (armpit).
- Brachial plexus nerve blocks using the following approaches (locations for injection): interscalene, supraclavicular, or infraclavicular; and without or with catheter implantation (to enable a 'top-up'). (These are also regional nerve blocks but comprise a distinct category.)
- Sedation. In our protocol we called this 'conscious sedation'. While this expression is widely used in the literature, it has been pointed out that the term 'conscious sedation' is a misnomer as well as being vague and confusing (Green 2002; Shipton 2002). Revised anaesthesia care standards in the USA have replaced 'conscious sedation' with 'moderate sedation/analgesia' (JCAHO 2001). The latter defines moderate sedation as "A drug-induced depression of consciousness during which patients respond purposefully to verbal commands, either alone or accompanied by light tactile stimulation. No interventions are required to maintain a patent airway, and spontaneous ventilation is adequate. Cardiovascular function is usually maintained." The expression 'monitored intravenous anaesthetic care' is also used (Shipton 2002). In this review 'sedation' primarily refers to planned 'moderate sedation'.
- General anaesthesia by total intravenous anaesthesia or use of inhalation anaesthetics, with orotracheal intubation or oropharyngeal intubation (e.g. laryngeal mask airway) as necessary.
Each of these methods has potential complications and may also result in inadequate anaesthesia (loss of sensation), analgesia (pain relief) and/or muscle relaxation, which could compromise the treatment process and results. Adverse drug reactions could also result from any of the agents used. Concerns over the toxicity of local anaesthetics have been raised (Quinton 1988a) and the serious dangers of leakage of anaesthetic from a poorly contained Bier's block, perhaps resulting from a ruptured cuff, are well known (Burke 1988; Heath 1982). There are also potential side effects from tourniquet use (Wakai 2001); specifically, any additional swelling of the forearm on release of the tourniquet may result in a tighter plaster cast and associated finger stiffness. Local damage, including haematoma, blood vessel rupture and damage, and various neuropathies, may ensue from regional nerve blocks. Incorrect administration of a nerve block may also result in the serious consequences of an unintended subarachnoid injection, epidural block, intravascular injection, phrenic nerve block (diaphragm) and, in particular for a supraclavicular block, a pneumothorax (lung collapse). Potential complications of general anaesthesia include difficulties in maintaining or establishing an open airway, intra-operative hypotension or hypertension, aspiration of gastric contents, post-operative nausea and vomiting, respiratory depression, and damage to teeth or upper airways.
Factors affecting the choice of anaesthesia include patient co-morbidity and suitability, patient preferences, local expertise including the availability of anaesthetists, treatment methods including reduction technique (e.g. manual manipulation, 'finger trap' traction) and whether immediate surgery is anticipated for a failed reduction, the throughput of patients, and local facilities.
This review aimed to identify and assess the evidence from randomised and quasi-randomised trials evaluating the main methods of anaesthesia, and associated physical techniques, used during the management of distal radial fractures in adults. Where possible, we made a distinction between conservative and surgical management, and between the treatment of initially displaced fractures versus redisplaced or secondarily displaced fractures.
We aimed to test the following null hypotheses.
1. There is no difference in outcome between the main types of anaesthesia (haematoma block, intravenous regional anaesthesia (IVRA), regional nerve blocks using various approaches to the brachial plexus, sedation, and general anaesthesia).
2. There is no difference in outcome between different physical techniques for the above methods. This includes the position of the injection site for local anaesthesia (haematoma block), IVRA and nerve blocks, and total intravenous versus inhalation general anaesthesia.
3. There is no difference in outcome between the use of drug adjuncts/supplements such as muscle relaxants.
Where possible and appropriate, we planned separate analyses of the outcome of conservative and operative treatment, of initial and secondary treatment, of males and females, of different age groups (younger adults, older adults), of types of fracture, of co-morbidities, and of prior functional and mental status. The type of treatment, including reduction technique, and clinician experience and speciality were also noted.
Criteria for considering studies for this review
Types of studies
We considered any randomised or quasi-randomised (use of a method of allocating participants to a treatment that is not strictly random; e.g. by date of birth, hospital record number, alternation) clinical trials of interventions, as listed above, used for treating distal radial fractures.
Types of participants
Patients of either sex who had completed skeletal growth and who were receiving conservative or surgical treatment for a fracture of the distal radius.
The characteristics of the participants included in the trials were noted with an emphasis on gender, age, type of fracture, functional and mental status, and co-morbidities.
Types of interventions
All randomised and quasi-randomised trials comparing different types and physical techniques of anaesthesia used for treating distal radial fractures. Also included are trials examining the use of drug adjuncts/supplements, such as muscle relaxants.
With one exception, we excluded pharmacological trials comparing different drugs in the same class, or drug dosages.
Types of outcome measures
a. Failed/inadequate anaesthesia. This is based on explicit reports of anaesthesia inadequacy or failure, and/or the need to use another method and/or additional procedures, curtailment or modification of treatment (e.g. of closed reduction), patient complaint of pain and patient dissatisfaction.
b. Anatomical restoration. Quality of initial reduction, final malunion and cosmetic deformity.
c. Adverse effects directly attributed to anaesthetic drugs and techniques; complications resulting directly from the administration of anaesthesia (including cardiac and respiratory arrest).
d. Clinical outcomes. Residual soft tissue swelling, early and late complications associated with distal radial fractures and/or their treatment, including reflex sympathetic dystrophy (RSD) or complex regional pain syndrome type 1.
e. Functional outcomes. Range of movement (wrist, forearm and shoulder mobility), pain, grip strength, and activities of daily living including return to previous employment; use of patient functional assessment instruments such as Short Form-36 (SF-36), the Disability of the Arm, Shoulder, and Hand questionnaire (DASH) and the Patient-Rated Wrist Evaluation (PRWE) (MacDermid 2000).
f. Resource use. Hospital admission, anaesthetist involvement and other costs.
Trials reporting only pharmacokinetic and/or physiological outcomes were excluded.
Search methods for identification of studies
We searched the Cochrane Bone, Joint and Muscle Trauma Group specialised register (November 2003), the Cochrane Central Register of Controlled Trials (The Cochrane Library Issue 4, 2003), MEDLINE (1966 to November week 2 2003), EMBASE (1988 to 2003 week 49) and CINAHL (1982 to December week 1 2003). We also searched Current Controlled Trials at http://www.controlled-trials.com (accessed October 2003) and the UK National Research Register at http://www.update-software.com/national/ (up to Issue 4, 2003 ) for ongoing and recently completed trials. We handsearched the British Volume of the Journal of Bone and Joint Surgery supplements (1996 onwards), abstracts of the American Society for Surgery of the Hand annual meetings (2000 to 2003: http:/www.assh.org/ASSH) and abstracts of the American Orthopaedic Trauma Association annual meetings (up to 2003: http://www.ota.org/education/amabstracts.htm). We also scrutinised weekly downloads of "Fracture" articles in new issues of 17 journals (Acta Orthop Scand; Am J Orthop; Arch Orthop Trauma Surg; Clin J Sport Med; Clin Orthop; Emerg Med Clin North Am; Foot Ankle Int; Injury; J Accid Emerg Med; J Am Acad Orthop Surg; J Arthroplasty; J Bone Joint Surg Am; J Bone Joint Surg Br; J Foot Ankle Surg; J Orthop Trauma; J Trauma; Orthopedics) from AMEDEO (http://www.amedeo.com). No language restrictions were applied.
The following search strategies were used:
The search strategy for The Cochrane Library is shown in Appendix 1. This search was used to find all trials of interventions for distal radial fractures, not just those comparing anaesthesia methods.
The search strategy used for EMBASE (OVID WEB) is shown in Appendix 3.
Data collection and analysis
Potentially eligible trials were assessed for inclusion by all three reviewers. Any disagreement was resolved through discussion. Titles of journals, names of authors or supporting institutions were not masked at any stage. The methodological quality of included studies was assessed independently by at least two reviewers. All three reviewers considered any discrepancies and disagreement was resolved by discussion; consistency checks were made by HH. Data were extracted, using a pre-piloted data extraction form, by two reviewers (CD & HH) with arbitration, if necessary, by the third reviewer (RM).
Trialists were contacted for additional details of key items of trial methodology or data.
Quality assessment tool
A modification of the Cochrane Bone, Joint and Muscle Trauma Group quality assessment tool (see Group details) was used in the evaluation of the included studies. The scoring scheme for 12 aspects of trial validity, plus brief notes of coding guidelines for some items, is shown in Table 1. Though the scores of the individual items were summed, this was to gain an overall impression rather than for quantitative purposes.
Where available and appropriate, quantitative data for outcomes listed in the inclusion criteria are presented in the analysis tables. Relative risks and 95 per cent confidence limits were calculated for dichotomous outcomes, and mean differences and 95 per cent confidence limits calculated for continuous outcomes. Results of comparable groups of trials were pooled using the fixed effects model and 95 per cent confidence limits. Heterogeneity between comparable trials was tested using a standard chi-squared test and considered to be statistically significant at p < 0.1. Where there was significant heterogeneity in the results of individual trials, we viewed the results of the random effects model and presented these, when considered appropriate, instead of those from the fixed effects model. No subgroup analyses were undertaken; if they had been, any tests of interaction calculated to determine if the results for subgroups are significantly different would have been based on Peto odds ratio results. Sensitivity analyses examining various aspects of trial and review methodology, including the effects of missing data and study quality were also considered but rarely possible.
Description of studies
The updating of search from June 2003 to November 2003 resulted in the identification of no new trials in the second update of this review.
In all, of 36 eligible studies, 18 are included and 18 are excluded (see Characteristics of Excluded Studies Table).
All of the included studies were fully reported in medical journals; Singh 1992 was published in an online journal. We received additional information from the trialists of seven included trials; including separate results for patients with distal radial fractures from three mixed population trials (Erlacher 2001; Esmaoglu 1995; Pippa 2000). The trials were initially identified in the following ways: bibliography checking (1); Current Contents (1); Cochrane Library (1); EMBASE (1); handsearching or journal browsing (3); MEDLINE (10) and reference from an external referee of the review protocol (1).
The periods over which individual trials were conducted spanned over three decades, from the early 1970s (Bultitude 1972) onwards. With the exception of Kendall 1997, which was conducted in two accident and emergency departments, trials took place at single centres in eight countries (Austria (1 trial); Denmark (1); Finland (1); India (1); Italy (1); Sweden (1); Turkey (1); UK (11)). English translations were obtained for trials reported in Danish (Walther-Larsen 1988) and Turkish (Esmaoglu 1995).
The 18 included studies involved at least 1200, mainly female and older, patients with fractures of the distal radius. There was a minimum of 1186 randomised patients with these injuries. A further 10 non-randomised patients were included in one trial (Bultitude 1972). There may have been another patient included in Funk 1997. The number of randomised patients in Wardrope 1985 was not explicitly stated but was probably the same as deduced from a table in the article. There were an estimated 20 patients with distal radial fracture not included in Hollingworth 1982. This amounts to an estimated maximum of 1217 patients overall.
With the exception of two trials (Pippa 2000; Singh 1992), most trials presenting data on patient gender recruited more female than male patients; the proportion ranged from 65 per cent (Erlacher 2001) to 97 per cent (London 1996). Where provided, median or mean ages of trial populations ranged between 31.5 years (Esmaoglu 1995) and 70.5 years (Eastwood 1986). The youngest patient (9 years) appeared in Bultitude 1972 and the oldest (91 years) in Funk 1997. Lower age limits were set by seven trials (Cobb 1985: 15 years; Funk 1997: 16 years; Hollingworth 1982: 16 years; Jones 1996: 12 years; Kendall 1997: 16 years; Walther-Larsen 1988: 14 years; Wardrope 1985: 45 years). Although children were included in some trials, it was clear that the majority of patients were skeletally mature. An upper limit of 54 years was applied in McGlone 1988.
Fracture type was generally broadly defined as distal radial, Colles', or wrist fracture. A fracture classification scheme was specified in three trials: Older's classification scheme was applied in Abbaszadegan 1990 and Walther-Larsen 1988; and Frykman's in Abbaszadegan 1990 and Kendall 1997. Two trials specified radiological criteria for trial inclusion; Kendall 1997 gave criteria for minimum deformity and Abbaszadegan 1990 stated a maximum for radial shortening. Anaesthesia was for manipulation/reduction in 16 trials; subsequent plaster cast immobilisation was clearly evident in 10 of these, and likely in the other six. Anaesthesia was for surgery in all patients in two trials (Erlacher 2001; Pippa 2000) and perhaps for some of the distal radial fracture patients in Hollingworth 1982. Four trials included patients requiring anaesthesia for other injuries or upper limb surgery: separate data for patients with distal radial fractures have been provided by the trialists of three trials (Erlacher 2001; Esmaoglu 1995; Pippa 2000); and are on request for the fourth trial (Hollingworth 1982). A minority of patients (approximately 10%) in Hollingworth 1982 underwent an operation or manipulation of hand injuries, other forearm fractures or elbow dislocation.
Further details of the individual studies are provided in the Characteristics of Included Studies Table.
The following is a brief description of the interventions compared in the 18 trials. One trial (Funk 1997) appears in three comparisons (1d, 1f and 3c) as it compared general anaesthesia versus haematoma block versus haematoma plus a sedative.
1. Main types of anaesthesia
1a. Intravenous regional anaesthesia (IVRA) versus haematoma block (local anaesthesia)
Five trials (Abbaszadegan 1990; Cobb 1985; Kendall 1997; Walther-Larsen 1988; Wardrope 1985) made this comparison in 478 patients. Prilocaine was used for both methods in Abbaszadegan 1990, and for IVRA in Cobb 1985, Kendall 1997 and Wardrope 1985; mepivacaine was used for IVRA in Walther-Larsen 1988; lignocaine was used for the haematoma block in the latter four trials. Lignocaine was either alkalinised or neutral in Kendall 1997; this comparison is not included in this review.
1b. Nerve block versus haematoma block
One trial (Haasio 1990) compared nerve block at the elbow region with haematoma block in 35 patients. Prilocaine was used for both techniques.
1c. Sedation versus haematoma block
One trial (Singh 1992) compared an intravenous injection of pentazocine combined with diazepam for sedation versus haematoma block in 67 patients. Lignocaine (Xylocaine) was used for the haematoma block.
1d. General anaesthesia versus haematoma block
Intravenous general anaesthesia was compared with haematoma block in 40 of the 58 or 59 patients in Funk 1997. Propofol was used for general anaesthesia and lignocaine for the haematoma block.
1e. General anaesthesia versus sedation
One vintage trial (Bultitude 1972) compared general inhalation anaesthesia with intravenous diazepam in 71 patients. The barbiturate and inhalation agents for general anaesthesia were not identified. Readers should note that the dosage of diazepam (generally 20mg) would be now considered dangerously high for administration to elderly patients.
1f. General anaesthesia versus haematoma block with sedation
Intravenous general anaesthesia using propofol was compared with haematoma block and sedation in 40 of the 58 or 59 patients of Funk 1997. Lignocaine was used for haematoma block and intravenous midazolam for sedation.
2. Physical techniques of anaesthesia
2a. Site of IVRA (Bier's block) injection
One trial (Blyth 1995) compared the injection of prilocaine into the antecubital fossa (elbow dip) with injection into the more traditional dorsum (back) of the hand site for a Bier's block in 100 patients.
2b. Use of an additional tourniquet for IVRA
One trial (Eastwood 1986), involving 50 patients, tested the application of an extra tourniquet immediately proximal to the fracture site during IVRA.
2c. Technique for brachial plexus block
One trial (Pippa 2000) compared a new subclavian perivascular technique, the proximal cranial needle approach, with the Winnie and Collins technique involving a supraclavicular approach. Seven of the 60 patients in this trial underwent surgery for distal radial fracture.
3. Drug adjuncts or supplements
3a. Muscle relaxant
Two trials (Esmaoglu 1995; McGlone 1988), involving a total of 76 patients, evaluated the addition of a muscle relaxant to IVRA. Vecuronium was tested in Esmaoglu 1995; all patients received intravenous lignocaine. Atracurium was tested in McGlone 1988; prilocaine was used for IVRA in this trial. Seventeen patients in Esmaoglu 1995 had distal radial fractures; the other 23 patients had unspecified forearm fractures.
One trial (Jones 1996) tested the addition of an analgesic, tenoxicam, to prilocaine for IVRA in 45 patients. Tenoxicam was inserted into either the site of the Bier's block or into the equivalent site on the unaffected arm. Though the site for supplementary analgesia is not covered in this review, the results for this comparison are presented in the analyses.
The addition of intravenous midazolam sedation to haematoma block was tested in 38 of the 58 or 59 patients in Funk 1997. Lignocaine was used for the haematoma block.
One trial (London 1996), involving 33 patients, tested the use of an enzyme, hyaluronidase, with haematoma block. Lignocaine was used for the haematoma block.
One trial (Erlacher 2001) tested the addition of clonidine to one of three local anaesthetics (mepivacaine, ropivacaine and bupivacaine) for axillary perivascular brachial plexus block using the Winnie method. Fifty-seven of the 120 patients in this trial underwent surgery for distal radial fracture. The comparison of the three different anaesthetics is not included in this review.
4. Extra comparison
Our intention was not to include comparisons of drugs in the same class. However, we made an exception for one trial (Hollingworth 1982) that compared bupivacaine with prilocaine for IVRA in 200 patients, at least 142 of whom had a distal radial fracture. We included this trial for three main reasons: a) it reported orthopaedic outcomes, b) it appeared to provide some insight into the potential effects on patient outcome of different operators, and c) it tested a formerly standard intervention (bupivacaine for IVRA). It is recognised that the characteristics of drugs in the same drug class can differ importantly; for instance, bupivacaine has a longer duration of action than prilocaine. While the differences between these two drugs are pertinent here, they alone were not sufficient to determine inclusion. Readers should note that bupivacaine is no longer used for IVRA in the UK due to safety concerns (BNF 2000).
Risk of bias in included studies
The methodological quality scores based on trial reports were variable. Lack of confirmation of the concealment of allocation and deficiencies in the assessment of outcome, including only short-term follow-up, were common reasons for lower quality scores. A summary of the individual aspects of trial quality follows the table of the scores for individual trials presented below. Information specific to the first three items of the quality score is given in the methods section of the Characteristics of Included Studies Table.
1 2 3 4 5 6 7 8 9 10 11 12 Total Study ID
1 0 0 0 0 0 1 1 1 1 1 1 7 Abbaszadegan 1990
1 3 0 1 0 0 3 0 3 1 0 0 12 Blyth 1995
0 0 1 0 0 0 0 0 1 1 1 0 4 Bultitude 1972
1 0 0 0 0 0 0 3 1 1 0 0 6 Cobb 1985
1 1 0 0 0 0 3 0 3 1 0 0 9 Eastwood 1986
1 3 1 1 3 3 1 1 1 3 1 0 19 Erlacher 2001
0 1 0 0 0 0 0 3 3 3 1 0 11 Esmaoglu 1995
1 0 1 1 0 0 1 1 1 3 3 0 12 Funk 1997
1 3 0 0 0 0 1 1 3 3 1 0 13 Haasio 1990
3 0 3 0 3 3 0 1 1 3 0 0 17 Hollingworth 1982
1 3 1 3 3 3 3 3 3 3 1 0 27 Jones 1996
0 0 0 3 0 0 3 1 3 3 3 0 16 Kendall 1997
1 1 3 1 3 3 3 0 1 1 0 0 17 London 1996
1 3 3 1 1 3 3 0 3 3 1 0 22 McGlone 1988
0 3 1 0 0 0 1 1 3 3 1 0 13 Pippa 2000
3 3 3 3 3 0 3 3 3 3 1 0 28 Singh 1992
1 3 1 1 0 0 0 3 1 3 1 1 15 Walther-Larsen 1988
0 0 0 1 0 0 0 3 3 3 1 0 11 Wardrope 1985
Allocation was clearly concealed (item 1) in two trials (Hollingworth 1982; Singh 1992). Some potential for disclosure of allocation was considered for two trials (Erlacher 2001; Funk 1997) using computer generated randomisation lists and two trials (Eastwood 1986; McGlone 1988) using envelopes. Seven trials (Abbaszadegan 1990; Blyth 1995; Cobb 1985; Haasio 1990; Jones 1996; London 1996; Walther-Larsen 1988) provided no details of their method of randomisation. Allocation was not concealed in Kendall 1997, where an open list of randomised numbers was used, nor in the three trials using quasi-randomised methods based on either dates of birth (Pippa 2000) or alternation (Bultitude 1972; Wardrope 1985). Concealment of allocation was also considered unlikely in Esmaoglu 1995; further clarification of the method of randomisation has been sought for this trial.
Intention to treat analysis (item 2) was considered very likely in eight studies (Blyth 1995; Erlacher 2001; Haasio 1990; Jones 1996; McGlone 1988; Pippa 2000; Singh 1992; Walther-Larsen 1988). Reasons for a reduced score for this item were the inclusion of patients from a non-randomised pilot study (Bultitude 1972), discrepancies in the data (Funk 1997; Wardrope 1985), and no or insufficient information on post-randomisation exclusions or loss to follow-up (Abbaszadegan 1990; Cobb 1985; Eastwood 1986; Esmaoglu 1995; Hollingworth 1982; Kendall 1997; London 1996).
Blinding of outcome assessors (item 3) where possible was clearly established in four studies (Hollingworth 1982; London 1996; McGlone 1988; Singh 1992). For six studies (Bultitude 1972; Erlacher 2001; Funk 1997; Jones 1996; Pippa 2000; Walther-Larsen 1988) there was either a lack of information to establish assessor blinding or only some of the assessors were blinded. Aside from Esmaoglu 1995, all the studies scoring zero for this item tested physical methods.
Three trials (Jones 1996; Kendall 1997; Singh 1992) provided sufficient information to indicate comparability between the treatment groups in key baseline characteristics such as gender, age and fracture type (item 4). Eight trials did not score for this item; either through failure to provide baseline data for all randomised patients, or failure to present any or sufficient information to confirm baseline comparability of important confounders.
Blinding of both patients and treatment providers (items 5 and 6) was considered effective in four drug trials (Erlacher 2001; Hollingworth 1982; Jones 1996; London 1996). Blinding was explicitly stated for treatment providers in McGlone 1988; whilst the lower score reflects that there was no report of whether patients were also blinded, we consider that this is likely. Patient blinding took place in Singh 1992 where two methods of anaesthesia were compared; Singh et al. reflected that this approach would be less achievable in most other cultures. Blinding of patients and providers was often not practical for trials that compared different anaesthetic techniques.
Comparability of care programmes (item 7), comprising interventions other than the trial interventions, could not be confirmed in any trial but was considered likely in seven (Blyth 1995; Eastwood 1986; Jones 1996; Kendall 1997; London 1996; McGlone 1988; Singh 1992). The criteria for judging comparability varied according to the length of follow-up; for instance, the duration of plaster cast immobilisation was not considered relevant when patients were only followed up to the end of the manipulation. Information on the technique used for fracture manipulation was provided in only three trials (Eastwood 1986; Kendall 1997; McGlone 1988); for these, and for some of the other trials, we may have assumed incorrectly that the operators used the same methods. There was no or inadequate information to assess care programme comparability in the six trials that scored zero for this item.
Six trials (Cobb 1985; Esmaoglu 1995; Jones 1996; Singh 1992; Walther-Larsen 1988; Wardrope 1985) provided sufficient trial inclusion and exclusion criteria to define the study population (item 8). In assessing this item we put more emphasis on exclusion criteria that included contraindications and less on the radiological criteria used to determine when a fracture required reduction; only Kendall 1997 provided the latter. Studies where the only stated criterion was wrist or distal radial fractures requiring manipulation did not score for this item.
Ten trials provided a good description of the interventions used (item 9); this included sufficient information to identify the discipline of the care providers (see Characteristics of Included Studies Table).
The definition (item 10) of outcome measurement was clear enough to give a good idea of what was recorded in the majority of the studies, but only two trials (Funk 1997; Kendall 1997) were rated as having good quality outcome measurement that included active follow-up (item 11). Five studies (Blyth 1995; Cobb 1985; Eastwood 1986; Hollingworth 1982; London 1996) failed to score for item 11; the loss of 25 out of 200 questionnaires was considered an important failing in Hollingworth 1982. Of note is that the comprehensiveness, aptness or overall validity of outcome assessment was not scored in item 11; hence many trials scored for this item despite their limited measurement of outcome, as long as assessment was clearly active and systematic.
The length of overall follow-up (item 12) was exceedingly short in most trials. Of the four trials (Abbaszadegan 1990; Bultitude 1972; Singh 1992; Walther-Larsen 1988) that followed up patients post-immobilisation, just two (Abbaszadegan 1990; Walther-Larsen 1988), each with six months follow-up, scored for this item.
Effects of interventions
1. Main types of anaesthesia
1a. Intravenous regional anaesthesia (IVRA) versus haematoma block
All five trials (Abbaszadegan 1990; Cobb 1985; Kendall 1997; Walther-Larsen 1988; Wardrope 1985) testing this comparison found that the IVRA patients experienced significantly less pain during fracture manipulation. Data for two trials (Cobb 1985; Wardrope 1985) are presented in the analyses; no pooling was possible due to incompatible outcome measures. Kendall 1997 reported that the pain associated with the administration of anaesthesia was also significantly less in the IVRA group. Two trials (Cobb 1985; Kendall 1997) reported that pain soon after the procedure was similar in both groups and Cobb 1985 found no difference in the duration of post-operative analgesia (see analyses). Similar numbers of patients in the two groups of Wardrope 1985 stated that they would have preferred general anaesthesia.
Four trials (Abbaszadegan 1990; Kendall 1997; Walther-Larsen 1988; Wardrope 1985) found that IVRA enabled better and easier reduction of the fracture; Cobb 1985 reported that an inadequate reduction was rare and not related to the method of anaesthesia. Pooled results from three studies show that significantly fewer fractures needed to be remanipulated during IVRA (11/140 versus 33/131; relative risk (RR) 0.30, 95% confidence interval (CI) 0.16 to 0.57; p = 0.0002). In the IVRA group, statistically better anatomical post-reduction measurements were declared for dorsal angulation in Abbaszadegan 1990 and Kendall 1997, and reported, with confirmatory data, for ulnar angulation and radial length (mean difference 1.8mm, 95%CI 0.6 to 3.0mm) in Wardrope 1985. Based on categories devised by Older et al. (Older 1965), there was a tendency in Walther-Larsen 1988 for fewer people in the IVRA group to have a poor result after reduction, or a poor anatomical result at five weeks (see analyses). Pooled results from three studies show some tendency for fewer later redislocations or re-reductions with IVRA (2/118 versus 8/110; RR 0.32, 95%CI 0.09 to 1.13).
No study reported a failure of anaesthesia; either in itself, or leading to a premature curtailment in the procedure, or use of another method of anaesthesia. Based on an undefined linear analogue scale, Cobb 1985 reported that the casualty officer's impression of a greater effectiveness of IVRA was usually more generous than, but still in agreement with, that of the patients (see analyses). Cobb 1985 found that haematoma block was more popular among the accident service staff who considered that it was simpler and quicker to perform and would avoid some of the risks associated with IVRA.
Both Cobb 1985 and Kendall 1997 found that IVRA took around 22 minutes longer to prepare or to begin compared with haematoma block. There was no significant difference in the timing of the subsequent procedure in either trial, but an additional 41 minutes for tourniquet release and observation was reported in Cobb 1985. Kendall 1997 reported no statistically significant difference between the two groups in the overall time in the accident and emergency department; Wardrope 1985 also claimed to have found no difference. Walther-Larsen 1988 stated that IVRA was more time consuming.
Four trials reported no or no significant adverse effects (infection or systemic complications) from anaesthesia; these were not mentioned in Abbaszadegan 1990. Other complications, such as median nerve compression, were few and occurred in similar numbers in the two anaesthesia groups (see analyses).
Functional outcomes were only reported by the two trials presenting longer-term results (Abbaszadegan 1990; Walther-Larsen 1988). The trialist responsible for the "medium term" follow-up study of Kendall 1997 indicated that the number of patients available at three months follow-up were insufficient to draw meaningful conclusions on the final function and anatomical position. Abbaszadegan 1990 presented contradictory results in the text compared with the tables for pain, but claimed that both groups had similar pain and range of motion at six months. Only grip strength was reported to be significantly stronger for the IVRA group at six months in Abbaszadegan 1990 (percentage of opposite arm: 65% versus 53%; reported p = 0.01). Walther-Larsen 1988 found no statistically significant difference in the functional grades (Older 1965) of the two groups (see analyses).
No cost data were provided. Cobb 1985 reported that additional staff time (mean 41 minutes/patient) was required for IVRA and Kendall 1997 noted that two doctors were required to apply IVRA whereas just one was required for haematoma block.
1b. Nerve block versus haematoma block
Haasio 1990 compared nerve block at the elbow region with haematoma block in 35 patients. Slightly more patients endured moderate (6 versus 4) or severe (3 versus 2) pain during manipulation in the nerve block group but the difference was not statistically different (9/16 versus 6/19; RR 1.78, 95%CI 0.81 to 3.93). The results of pin-prick analgesia at 15 minutes were reported to be better (reported p < 0.05) for the median and ulnar nerves in the nerve block group. Full motor blockade in three nerves (radial, median, ulnar) only occurred in one patient in the nerve block group and in one or two nerves in four others of the same group; no instances of full motor block were found in the haematoma group. Just one patient, belonging to the haematoma block group, was considered by the surgeon to have an insufficiently relaxed wrist to ease manipulation; there was no mention of re-manipulation. Despite five patients experiencing severe pain, all patients were reported as being satisfied with their treatment. There were no cases of systemic toxicity.
1c. Sedation versus haematoma block
Singh 1992 compared intravenous sedation with haematoma block in 67 patients. The exclusion from the reported analyses of the results of one sedation group patient in whom intravenous injection at the fracture site had failed (site chosen for blinding purposes) was stated not to influence the overall results. Patients in the sedation group experienced significantly more pain during reduction of their fractures (mean difference in visual analogue scores (0: no pain to 10: excruciating pain): 5.6, 95%CI 4.5 to 6.6). Only one patient in the sedation group registered a score under three, compared with 25 in the haematoma group (RR 4.00, 95%CI 2.18 to 7.33). The time taken to reduce fractures was greater for patients in the sedation group (patients for whom fracture reduction took at least one minute: 32/33 versus 25/33; RR 1.28, 95%CI 1.05 to 1.57). Whether this reflects operator bias where a gentler reduction technique was intentionally applied in the haematoma group could not be answered, but there were no significant differences in the number of cases of "reduction failure" or cases of deformity sufficient to "bother" patients between the two groups by eight weeks follow-up (see analyses). There were no cases of non-union or significant adverse effects. Similar numbers of patients in the two groups had substantial "rest pain" and stiffness (assumed to be of the wrist) at eight weeks: see analyses.
1d. General anaesthesia versus haematoma block
Intravenous general anaesthesia was compared with haematoma block in 40 patients in Funk 1997. A request for clarification on the number of patients entered and analysed in the trial has been made; where unavailable, the numbers presented in the analyses represent 'best guesses'. No patients receiving general anaesthesia experienced pain during manipulation of their fractures, whereas patients in the haematoma block group experienced significant pain (mean pain visual analogue scores (0: no pain to 10: worst imaginable): 0 versus 3.7). After manipulation, however, the general anaesthesia group experienced significantly greater pain compared with the haematoma block group (mean pain scores: 5.8 versus 1.5; reported p < 0.01). There were no statistically significant differences in the numbers with radial shortening or with residual dorsal deformity (see analyses). Patients allocated general anaesthesia waited on average five hours longer for manipulation (mean difference 300 minutes, 95%CI 67 to 533 minutes; p = 0.01), including over one hour waiting for the arrival of staff (mean difference 73 minutes, 95%CI 46 to 100 minutes). The time for anaesthesia and manipulation was twice as long in the general anaesthesia group (mean difference 10.2 minutes, 95%CI 4.2 to 16.2 minutes; p = 0.0009); Funk 1997 suggested that early manipulations in the haematoma block group may have resulted in the extra pain during manipulation in this group. Recovery time was not recorded. There were no complications. Funk 1997 reported that the cost of administering a general anaesthetic was more than three times that of a haematoma block (£15.56 versus £4.09); mainly attributable to the requirement for anaesthesia specialists.
1e. General anaesthesia versus sedation
Bultitude 1972 compared general inhalation anaesthesia with intravenous diazepam in 71 patients. As well as being a quasi-randomised study, the results from 10 patients given diazepam but who were not in the study could not be separated out. Thus, while we have presented the results of this study, the strong risk of systematic bias needs to be noted. Though most of the patients given sedative reacted in some way (groaning etc), only three of these actually recalled the reduction, two of whom recalled slight pain. No patient attempted to withdraw their arm during reduction of their fracture; but reduction was delayed due to a vigorous reaction in one patient given general anaesthesia. One elderly female patient given diazepam sustained a precipitous fall in systolic blood pressure and respiratory rate from which she recovered. Two other elderly females receiving diazepam were temporarily unrousable - the diazepam dosage (20mg) used in this trial for elderly people would now be viewed as dangerous. There were no statistically significant differences in adverse effects: see analyses. Just one reduction, in the general anaesthesia group, was considered "bad" on inspection of post-reduction films (four films were missing). By six weeks, similar proportions of fractures had redisplaced (RR 0.97, 95%CI 0.62 to 1.50). Two general anaesthesia and four diazepam patients indicated that they would be apprehensive if they were given the same method of anaesthesia in future.
1f. General anaesthesia versus haematoma block with sedation
Intravenous general anaesthesia was compared with haematoma block and midazolam sedation in 40 patients in Funk 1997. As explained earlier (see 1d), where denominators are unavailable for this trial, the numbers presented in the analyses represent 'best guesses'. No patients receiving general anaesthesia experienced pain during manipulation of their fractures, while patients receiving haematoma block with sedation experienced, on average, slight pain only (mean pain visual analogue scores (0: no pain to 10: worst imaginable): 0 versus 0.9). After manipulation, the general anaesthesia group experienced significantly greater pain compared with the haematoma block with sedation group (mean pain scores: 5.8 versus 1.6; reported p < 0.01). There were no statistically significant differences in the numbers with radial shortening or with residual dorsal deformity, although these tended to favour the haematoma block with sedation group (see analyses). Patients allocated general anaesthesia waited on average four hours longer for manipulation (mean difference 240 minutes, 95%CI -11 to 491 minutes; p = 0.06, not significant), including over an hour's wait for the arrival of staff (mean difference 76 minutes, 95%CI 47 to 105 minutes). The time for anaesthesia and manipulation was twice as long in the general anaesthesia group (mean difference 10.7 minutes, 95%CI 4.3 to 17.1 minutes; p = 0.001). Recovery time was not recorded but was described as longer in the sedation group. There were no complications. Funk 1997 reported that the cost of administering a general anaesthetic was three times that of a haematoma block with sedation (£15.56 versus £5.18); mainly attributable to the requirement for anaesthesia specialists.
2. Physical techniques of anaesthesia
2a. Site of IVRA (Bier's block) injection
Blyth 1995 compared IVRA injection into the antecubital fossa with injection into the dorsum of the hand in 100 patients. No statistically significant differences were reported for pain during IVRA at either the fracture site (visual analogue scale (0: no pain to 10: worst imaginable) mean values: 3.45 versus 2.66) or at the cuff (mean values: 3.30 versus 3.16). The two groups were reported to be similar in the number of manipulations, usually one, per patient. The surgeons recorded more procedural problems in the hand group in terms of failed cannulations, and injection site bleeding or haematoma leading to difficulties in the application of a plaster (0/50 versus 10/50; RR 0.05, 95%CI 0.0 to 0.8; p = 0.03). There were no major complications, including signs of systematic toxicity, noted. Longer-term outcome was not recorded.
2b. Use of an additional tourniquet for IVRA
Eastwood 1986 investigated the application of an extra tourniquet immediately proximal to the fracture site during IVRA in 50 patients. Only one outcome was reported in the trial report: the time to achieve clinical anaesthesia (pin prick test) in the additional tourniquet group was about half as long as that for the control group (mean difference -4.2 minutes, 95%CI -5.2 to -3.1 minutes). Contact with one of the trialists revealed that there were no recorded adverse effects.
2c. Technique for brachial plexus block
Pippa 2000 compared the proximal cranial needle approach (PCN) with the Winnie and Collins technique (W+C). Seven of the 60 patients in this trial underwent surgery for distal radial fracture. All seven patients were rated as having either excellent or good anaesthesia after 20 minutes. The one serious complication in these seven patients was a subclavian artery puncture in one patient of the W+C group. In the whole trial population, serious complications occurred only in the W+C group: phrenic nerve block with respiratory failure (6/30 versus 0/30); subclavian artery puncture (9/30 versus 0/30); there were no cases of pneumothorax (see analyses).
3. Drug adjuncts or supplements
3a. Muscle relaxant
Two trials (Esmaoglu 1995; McGlone 1988) evaluated the addition of a muscle relaxant to IVRA in a total of 76 patients. There were no anaesthesia failures, in that there was no recourse to general anaesthesia, in Esmaoglu 1995; but the fractures of two control patients could not be initially reduced in McGlone 1988. One control group patient with a distal radial fracture received an additional analgesic in Esmaoglu 1995 and overall a non-statistically significant tendency for less pain and discomfort in the muscle relaxant group was reported for this trial. Better analgesia during reduction was also found for the muscle relaxant group in McGlone 1988 (mean pain scores (0: painless to 10: worst imaginable): 0.6 versus 1.8; reported p <0.05). Easier reduction of fractures, as measured on a visual analogue scale (0: impossible to 10: very easy), in the muscle relaxant group was also noted in McGlone 1988 (mean 8.4 versus 6; reported p < 0.025). Two patients, both in the control group, had an unacceptable anatomical result in McGlone 1988; no anatomical outcomes were reported in Esmaoglu 1995.
No statistically significant difference was found in the overall tourniquet time for the whole trial population in Esmaoglu 1995 (see analyses). Esmaoglu 1995 found that sensorial and motor blockade were faster in the muscle relaxant group by around one and 10 minutes respectively (see analyses). The recovery time from the motor block was 10 minutes longer in the muscle relaxant group in Esmaoglu 1995 (see analyses) and, defined in terms of return to fine movement, around 20 minutes longer (mean time 25.8 versus 4.8 minutes; reported p< 0.001) in McGlone 1988.
Adverse effects of anaesthesia were few and transient (see analyses). No long-term or functional outcomes were recorded in these two trials.
Jones 1996 examined supplementary analgesia where tenoxicam was injected either into the IVRA injection site or into the opposite arm or not at all in 45 patients. The results for the three comparisons in this trial are presented in the analysis tables. Aside from tourniquet time, which was similar in all three groups, there was no mention of outcomes during anaesthesia or orthopaedic results; the focus of the trial was on post-reduction analgesia. Tenoxicam has a relatively long half-life that allows a once-daily administration; the choice of a 24 hour follow-up may reflect this. Injecting tenoxicam into the affected arm at the same time as the local anaesthetic (prilocaine) provided better analgesia than either no tenoxicam or injecting tenoxicam into the opposite arm (see analyses): patients waited longer before taking extra analgesics and took statistically significantly fewer analgesics over 24 hours; they also had a significantly lower mean pain score, with fewer patients assessing their pain as moderate (none of the trial patients had severe pain). The injection of tenoxicam into the unaffected arm was relatively ineffective in providing analgesia. No adverse effects were reported by any of the trial patients.
The addition of intravenous midazolam sedation to haematoma block was tested in 38 patients in Funk 1997. As explained above (see 1d), where denominators are unavailable for this trial, the numbers presented in the analyses represent 'best guesses'. Patients receiving midazolam experienced significantly less pain during manipulation of their fractures (mean pain visual analogue scores (0: no pain to 10: worst imaginable): 0.9 versus 3.7; reported p <0.01). The pain scores in the two groups were similar after fracture manipulation (mean pain scores: 1.6 versus 1.5). There were no statistically significant differences in the numbers with radial shortening or with residual dorsal deformity (see analyses). There were no statistically significant differences between the two groups in either of the two measures of waiting time (see analyses), and the mean times for anaesthesia and manipulation were similar in the two groups. Recovery time was not recorded but was observed to be longer in the sedation group. There were no complications. The cost per person of supplementary sedation was about one pound sterling more (£5.18 versus £4.09).
London 1996 tested the use of hyaluronidase for enhancing haematoma block in 33 patients. One patient in the enzyme group was excluded due to an unsuccessful manipulation (cause not stated) and another because they did not understand visual analogue scales. London 1996 reported that there were no significant differences (reported p > 0.05) between the two groups in any of the three measures of subjectively assessed pain. There were no side effects noted.
Erlacher 2001 tested the addition of clonidine to either mepivacaine, ropivacaine or bupivacaine for axillary brachial plexus block in 57 patients undergoing surgery for distal radial fracture. The results in the analysis tables are presented for each of the three local anaesthetics, as well as overall. The onset of sensory blockade was not statistically significantly different between the clonidine and control groups when the results from the three comparisons were pooled: mean visual analogue scale (0: complete blockade to 100: no blockade) scores for sensory blockade at 10 minutes (weighted mean difference (WMD) in score 6.5, 95%CI -2.4 to 15.5) and at 30 minutes (WMD 1.2, 95%CI -1.0 to 3.36). But, the onset of sensory blockade was statistically significantly slower when clonidine was added to bupivacaine (see analyses). When the results of the three comparisons were pooled, motor blockade was significantly prolonged in patients given clonidine (WMD 199 minutes, 95%CI 167 to 231 minutes). These results, however, are significantly heterogeneous for the three anaesthetics; the addition of clonidine to ropivacaine appeared to have little effect on the duration of motor blockade (see analyses). There were no side effects (specifically: hypotension, nausea, vomiting) recorded.
4. Extra comparison
Hollingworth 1982 compared bupivacaine with prilocaine for Bier's block (IVRA) in 200 patients with forearm injuries. Separate results for the 142, or more, patients with fractures of the distal radius are not currently available but have been requested. Outcome assessment was by questionnaire, and only 175 of these were "retrieved". There were no statistically significant differences between the two groups in the incidence of 'great' pain as assessed by the doctor or patient (see analyses). No bupivacaine patient versus nine prilocaine patients had a poor (remanipulation/abandoned procedure) result (RR 0.05, 95%CI 0.0 to 0.92; P = 0.04). However, similar numbers in the two groups had a poor or unacceptable result (49/85 versus 47/88; RR 1.08, 95%CI 0.83 to 1.41). A worst case analysis, where it is assumed that the patients missing from the analysis in the bupivacaine group had a poor result whereas those in the prilocaine group did not, shows the non robust nature of the results for 'poor result' (15/100 versus 9/100; RR 1.67, 95%CI 0.77 to 3.63). No major side effects were recorded in Hollingworth 1982. There was a non significant tendency for slightly more side effects (notably tinnitus and feeling 'funny') in patients given bupivacaine (20/85 versus 12/90; RR 1.76, 95%CI 0.92 to 3.39). Hollingworth 1982 examined the variability in performance of the 20 doctors who administered IVRA in the trial. The poor performances of two doctors, explicitly in relation to pain assessed by the patients, were highlighted. Hollingworth 1982 reported that the findings of the trial were not significantly affected by the variation in the performance of the doctors but provided no confirmatory evidence.
Most of the comparisons addressed by the 18 included trials are relevant to current practice, at least in some parts of the world. Safety concerns mean that bupivacaine for Bier's block (Hollingworth 1982) and the high dosage of diazepam used in Bultitude 1972 for sedation are no longer sanctioned; at any rate not in the UK (BNF 2000). In agreement with comments received at editorial review, we also question the acceptability of intravenous sedation using drugs with little or no analgesic effect such as diazepam and midazolam without supplementary analgesia (Shipton 2002).
Small trial sizes of 50 or fewer patients in each intervention group, flawed methodology and deficiencies in the assessment of outcome hamper the ability of individual trials to address their own questions, let alone the objectives of this review. Although some pooling was possible for the two comparisons (IVRA versus haematoma block; muscle relaxant adjunct for IVRA), the number of patients remains low. Trial methodology was inadequately reported in most trials but, even with feedback from some of the trialists involved, the possibility of systematic bias resulting in flawed evidence cannot be ruled out. Notable is the failure to ensure or confirm the concealment of allocation in all but two trials. Another issue is that of confounding, where factors other than the interventions under investigation influence the trial results. Examples of confounders are imbalances in baseline characteristics, such as fracture type, of patients, and differences in care programmes including reduction methods, their timing (as considered in Singh 1992) and the expertise of, and/or delivery of health care by, the health professionals involved (as reported in Hollingworth 1982). Our methodological quality scheme did not score the comprehensiveness of outcome assessment in individual trials; instead we looked at the potential for ascertainment bias due to a lack of blinding of outcome assessors, and the quality of measurement of the actual outcomes recorded. Nonetheless, the included trials did not provide a full, or sufficient, 'picture' of the relative effects of the interventions under investigation. This deficiency was usually associated with an inadequate length of follow-up. Some trials such as Eastwood 1986, which primarily reported on the time to achieve clinical anaesthesia, were particularly limited. While it is tempting to focus on very short term outcomes, such as pain during manipulation, for evaluating anaesthetic interventions, it is important to collect long-term outcomes given that inadequate anaesthesia could have long-term consequences; for instance, resulting from inadequate reduction. Some side effects may also have longer-term implications. (The confounding effects of differences in subsequent treatment hamper the interpretation of long-term follow-up results; thus measures are needed to ensure comparability of care programmes other than the trial interventions.) Trials that fail to record the success of the procedure, usually reduction, for which anaesthesia is required are certainly incomplete and potentially misleading.
Our prime focus is on examining clinical outcomes. However, information on resource use and costs that would enable a cost-effectiveness analysis is also desirable; and particularly useful when there is no strong evidence for important differences in the effectiveness of interventions. Though some information on care providers and timing was available from other trials, only one trial (Funk 1997) provided some cost data: namely, the average total procedural costs for the three interventions under test. The entry for this trial in the NHS Economic Evaluation Database (NHS EED 2003) points out several deficiencies relating to the cost evaluation in this trial. These include the absence of information on the trial dates and price year, insufficient information on the methods used for calculating costs, and no consideration of the generalisability to other settings or countries. It is noteworthy that the NHS EED reviewer(s) commented that it was "impossible to state whether all relevant costs were included in the analysis". Trials examining costs and resource use should aim for a complete assessment of costs, covering the whole treatment period.
Finally, the interpretation and application of trial findings is limited where, as is often the case with the trials included in this review, there are inadequate details of the intended and actual trial population and interventions.
These general points need to be considered when interpreting the results of individual comparisons. To aid the reader, a table summarising the types of outcome data available for the interventions tested within randomised controlled trials plus some comments is available ( Table 2). These comments are expanded on in this discussion.
1. Main types of anaesthesia
A search for surveys of anaesthesia practice for these fractures identified two studies conducted in the UK. Responses were obtained from 86 accident and emergency (A&E) departments to a questionnaire on the methods of anaesthesia used in the manipulation of Colles' fractures in 1994 (Kendall 1995). These revealed that Bier's block (IVRA) was given to one third of patients, haematoma block to another third, general anaesthetic to 24 per cent and intravenous sedation to seven per cent. Five years previously, a study of 54 UK centres showed that general anaesthesia was given to 44 per cent of patients, IVRA to 33 per cent, intravenous sedation to 13 per cent and haematoma block to just seven per cent (Hunter 1989). The swing to haematoma block, primarily at the expense of general anaesthesia, was associated with a reduction in the involvement of anaesthetists and reduction in numbers of patients requiring fasting and/or hospital admission. More A&E doctors also performed IVRA. Kendall et al. also found that monitoring (ECG, blood pressure and oxygen saturation) was carried out in a third of patients given haematoma blocks compared with 85 per cent of Bier's blocks and all those given general anaesthesia or intravenous sedation. Though these surveys reveal trends in current practice in just one country, this review aimed to examine the evidence supporting the various choices for anaesthesia worldwide.
There was evidence from five trials comparing IVRA with haematoma block that IVRA provided better analgesia during fracture manipulation and enabled better and easier reduction of the fracture. There was some indication of superior post-reduction anatomical measurements and a reduced risk of later redislocation or re-reduction. Neither method was associated with anaesthesia failure or significant adverse effects. There was some evidence that haematoma block was quicker and easier to perform and less resource intensive. One study found enhanced long-term grip strength in the IVRA group. Overall, there was not enough evidence here to confirm a clinical superiority of IVRA, or to establish the relative safety of the two methods. But, there is some indication that haematoma block provides poorer analgesia and can compromise reduction.
One trial that compared nerve block (this method is relatively rare for these fractures and did not appear in Kendall's UK survey (Kendall 1995)) with haematoma block found no statistically significant differences in pain during manipulation; there was one case of insufficient muscle relaxation. These limited results provide no basis to judge the relative effectiveness of the two methods.
One well conducted trial comparing intravenous sedation with haematoma block found that patients in the sedation group experienced significantly more pain during reduction of their fractures. No significant differences in the "reduction failure", deformity, "rest pain" or stiffness were evident at eight weeks. The time taken for fracture manipulation was greater in the sedation group and there is a possibility of some confounding due to doctors doing a different, perhaps gentler, manipulation in the haematoma block group. The numbers were too few and outcomes measures too vague to examine this or to detect long term differences. Also relevant is that this trial was carried out in a 'developing' country with scarce resources and involved a relatively young trial population. The choices available will differ in other circumstances and for other populations.
Inadequate information on methodology, interventions and patient numbers, mean that the results of a trial comparing intravenous general anaesthesia with haematoma block must be viewed very cautiously. However, it is plausible that general anaesthesia gives better pain relief during manipulation but involves longer waits and takes longer, with higher procedural costs compared with haematoma block. There was some indication that there was more pain post manipulation after general anaesthesia. Overall, there were too few patients to determine if the lack of statistically significant differences in radiological outcomes was a true result, as well as an absence of evidence on eventual clinical outcome.
As noted previously, the evidence from a dated, quasi-randomised study comparing general inhalation anaesthesia versus sedation with diazepam is likely to be seriously biased, and the dosage of diazepam is excessive for elderly persons. Nonetheless some of the observations from this study still appear to have merit. For instance, the serious event after diazepam in one elderly woman given sedation, the contrast between the reactions observed during fracture manipulation and recall in sedated patients, and the enumeration of redisplaced fractures at six weeks.
The same reservations listed for the above comparison of general anaesthesia versus haematoma block apply to the comparison of general anaesthesia versus haematoma block with sedation. The absence of quantitative data on recovery times means that it is not possible to comment on the clinical and administrative consequences of the longer recovery time for patients given supplementary sedation.
2. Physical techniques of anaesthesia
A trial comparing two different injection sites for IVRA found more procedural problems during plaster cast application when the more traditional injection site, into the dorsum of the hand, was used. However, the small sample size, the lack of blinding and the incomplete assessment of outcome mean that the lack of difference between the two injection sites in other reported outcomes cannot be considered conclusive evidence that injection into the antecubital fossa is a safe and effective alternative.
The inadequate assessment of outcome in a study (Eastwood 1986) of the use of an extra tourniquet for IVRA has already been mentioned.
Although the one trial examining two techniques for brachial plexus block was small and involved only seven patients undergoing surgery for distal radial fractures, the excess in the number of serious complications associated with the Winnie and Collins technique in this trial is notable; both on its own and in the context of an absence of serious complications for the proximal cranial needle approach. Since the successful application of regional nerve blocks is particularly dependent on the skill and experience of the operator, the promising results for the new approach need to be tested by other anaesthetists.
3. Drug adjuncts or supplements
The addition of two different muscle relaxants and one analgesic was tested for IVRA, one sedative and hyaluronidase for haematoma block and clonidine for brachial plexus block. All trials evaluating adjuncts were too small and failed to provide evidence on eventual clinical outcome.
Two trials provided some limited and preliminary evidence in favour of a muscle relaxant adjunct to IVRA in terms of analgesia and ease of fracture reduction. Muscle relaxant prolonged recovery time from motor blockade and was associated with transient double-vision in some patients.
There was limited evidence from one small trial that injection of tenoxicam directly into the IVRA injection site improved post-reduction analgesia in the first 24 hours. Other relevant outcomes, such as effectiveness of anaesthesia, were not reported.
The reservations listed for the above comparison of general versus haematoma block also apply to the examination within the same trial of the addition of sedation to haematoma block. Sedation with midazolam may give better pain relief during manipulation but there were too few patients to determine if the lack of statistically significant differences in radiological outcomes was a true result, and a lack of information to judge the importance of the longer recovery times for patients given sedation.
Given the localised nature of the fracture site and manipulation, the rationale for the use of hyaluronidase for enhancing haematoma block through the breakdown of local connective tissue is fairly unconvincing especially in view of the potential for allergic reactions. However, the trial evaluating this option did not provide the robust evidence required to settle the question.
One trial provided some limited evidence of the effect on neural blockade of supplementary clonidine for axillary brachial plexus block in patients undergoing surgery or distal radial fracture. While the effects on neural blockade varied according to the anaesthetic used (bupivacaine, mepivacaine, ropivacaine), clonidine does not appear to enhance and may delay the speed of sensory blockade, and may unnecessarily prolong motor blockade for the relatively short operations for these injuries. Whether clonidine affected the onset of motor blockade was not reported.
4. Extra comparison
At study selection, we considered that the inclusion of a trial comparing bupivacaine with prilocaine for IVRA was merited due to a) the more comprehensive assessment of clinical outcomes, especially orthopaedic outcomes; b) the recording of patient outcomes for the different care providers in this trial; and c) the special relevance of the comparison for IVRA. Though prilocaine was proposed as "first-line medication" due to low toxicity, both drugs are listed as being suitable for IVRA for upper limb trauma in a recent publication (Mollmann 2000). However, bupivacaine is not approved for IVRA in the UK (BNF 2000). There was some concern expressed at editorial review that making this exception, to our general ban on comparisons of drugs within the same class, could lay us open to charges of "selection bias". We counter this by stressing that this is an exception and that we consider our reasons for including it remain valid. Two alternatives would have been to a) exclude the trial but summarise its findings in the Discussion and b) formulate an additional hypothesis based on important differences in the characteristics of these two drugs. We decided against these as neither was satisfactory; either resulting in the loss of the valuable insights gained through the systematic processing and full presentation of the trial, or engendering further hypotheses based on other differences in characteristics of other drugs in the same class.
In the event, we found that Hollingworth 1982 had serious methodological flaws and that the evidence was not robust. Nonetheless, there was some indication that the pain experienced by patients could depend to some extent on the doctor reducing the fracture.
Implications for practice
There was insufficient robust evidence from comparisons tested within randomised trials to establish the relative effectiveness of different methods of anaesthesia or of different techniques of individual methods, or to judge the use of different drug adjuncts in the treatment of distal radial fractures.
Though there was not enough evidence to confirm a clinical superiority of IVRA, nor to establish the relative safety of IVRA to haematoma block, there is some indication that haematoma block provides poorer analgesia and can compromise reduction. Clinicians need to be mindful of this in view of the potential attractiveness of haematoma block in terms of technical ease and lower cost of administration.
Implications for research
Research on anaesthetic interventions needs to be set in the context of the overall management of these fractures. There are many unresolved issues such as:
Patient characteristics and preferences, fracture type, and local expertise and resources influence treatment choices and must be considered. We suggest an integrated programme of research, which includes consideration of anaesthesia options, for the management of these fractures is the way forward.
For researchers contemplating specific research on anaesthesia options for these fractures, we suggest that priority should be given to large scale, preferably, multi-centre randomised comparisons of the main methods (haematoma block, IVRA, general anaesthesia, and intravenous sedation (with analgesia)) in common use. In addition, any new promising developments in anaesthesia must be rigorously evaluated before general implementation. Such developments include combinations of the main methods (e.g. intravenous sedation plus analgesia with a haematoma block or elbow block), an expansion in the use of regional nerve blocks especially in the light of improved techniques for delivery, and multimodal analgesia (Shipton 2002). Other developments at the pharmacological level include the use of single enantiomers (Burke 2002). For example, there is some preliminary evidence that levobupivacaine, the levo or S-enantiomer of bupivacaine, has less cardiac toxicity than bupivacaine (Burke 2002; Whiteside 2001). Any trials testing anaesthesia options should abide by the methodological criteria for a well-conducted and well-reported randomised trial (Altman 2001). As well as anaesthesia outcomes, trials should record the effectiveness of the orthopaedic procedure being performed under anaesthesia, and subsequent clinical outcomes for at least six months, as well as patient satisfaction and all relevant costs. Stratification by key patient characteristics and by care provider should be considered.
We thank Lesley Gillespie for her help with the search strategy and trial retrieval. We thank the following for feedback at the editorial and external review of the protocol: Ted Shipton, Lesley Gillespie, Bill Gillespie, Peter Herbison, Finn Molke Bjorgbjerg, Leeann Morton and Janet Wale. We thank the following for feedback and help at the editorial and external review of the review and review updates: Mike Bennett, Bill Gillespie, Peter Herbison, Finn Molke Bjorgbjerg, Kate Rowntree, Ted Shipton, Marc Swiontkowski and Janet Wale. We thank Hilda Bastian for her help with the Synopsis.
We are grateful to the following for providing further information on their trials: Patricia Allen, C Bhattacharjee, Owen Brady, Marcus Cope, Aliye Esmaoglu, Lennard Funk, Nick James, Jason Kendall, Christopher Khoo, Raymond McGlone, P Pippa and Keith Porter.
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 The Cochrane Library
#1. RADIUS FRACTURES explode all trees (MeSH)
#2. WRIST INJURIES explode all trees (MeSH)
#3. (#1 or #2)
#4. ((distal near radius) or (distal near radial))
#5. (colles:ti or smith:ti or smiths:ti)
#6. (colles:ab or smith:ab or smiths:ab)
#8. (#4 or #5 or #6 or #7)
#10. (#8 and #9)
#11. (#3 or #10)
Appendix 2. Search strategy for MEDLINE (OVID WEB)
1. Exp Radius Fractures/
2. Wrist Injuries/
3. (((distal adj3 (radius or radial)) or wrist or colles$ or smith$2) adj3 fracture$).tw
5. exp Anesthetics/
6. exp Anesthesia/
Appendix 3. Search strategy for EMBASE (OVID WEB)
Last assessed as up-to-date: 5 January 2004.
Protocol first published: Issue 4, 2001
Review first published: Issue 3, 2002
Contributions of authors
Helen Handoll (HH) initiated and coordinated the production of the review, starting with the compilation of the first draft of the protocol and subsequent revisions in RevMan. Rajan Madhok (RM) arranged funding for the review. RM and Chris Dodds (CD) critically reviewed and rewrote the protocol.
HH, with some assistance from Lesley Gillespie, located the review studies. All three reviewers partook in study selection, critically reviewed the included studies and piloted the quality assessment and data extraction forms. HH and CD extracted trial details and results. HH contacted trialists for further information. HH compiled the first draft and all subsequent revisions in RevMan. CD and RM critically reviewed and checked all review drafts.
The first update of the review, including the extension of the literature search, contact of trialists and preparation of the first draft was performed by HH. CD and RM critically reviewed the update. All three reviewers are guarantors of the review.
Declarations of interest
Sources of support
- University of Teesside, Middlesbrough, UK.
- East Riding and Hull Health Authority, UK.
Medical Subject Headings (MeSH)
MeSH check words
Aged; Female; Humans; Male
* Indicates the major publication for the study