Intervention Review

You have free access to this content

Pre-operative traction for hip fractures in adults

  1. Helen HG Handoll1,*,
  2. Joseph M Queally2,
  3. Martyn J Parker3

Editorial Group: Cochrane Bone, Joint and Muscle Trauma Group

Published Online: 7 DEC 2011

Assessed as up-to-date: 1 JUL 2011

DOI: 10.1002/14651858.CD000168.pub3


How to Cite

Handoll HHG, Queally JM, Parker MJ. Pre-operative traction for hip fractures in adults. Cochrane Database of Systematic Reviews 2011, Issue 12. Art. No.: CD000168. DOI: 10.1002/14651858.CD000168.pub3.

Author Information

  1. 1

    Teesside University, Health and Social Care Institute, Middlesborough, Tees Valley, UK

  2. 2

    Royal College of Surgeons in Ireland, Higher Surgical Training Group, Dublin, Ireland

  3. 3

    Peterborough and Stamford Hospitals NHS Foundation Trust, Department of Orthopaedics, Peterborough, Cambridgeshire, UK

*Helen HG Handoll, Health and Social Care Institute, Teesside University, Middlesborough, Tees Valley, TS1 3BA, UK. h.handoll@tees.ac.uk. H.Handoll@ed.ac.uk.

Publication History

  1. Publication Status: New search for studies and content updated (conclusions changed)
  2. Published Online: 7 DEC 2011

SEARCH

 
Characteristics of included studies [ordered by study ID]
Anderson 1993

MethodsRandomised by last digit of patient's registration number.


Participants252 people with a proximal femoral fracture
Hospital in Leicester, UK
Period of study: November 1991 to July 1993
Excluded: patients refusing consent, senile patients, and patients with conditions which contraindicated the use of skin traction as specified in the published article
Mean age: 81 years
Females: 77%
Intracapsular fractures: 46%

Assigned: 101/151 [traction / control]


InterventionsPre-operative skin traction using 2.3 kg weight of traction applied via Hamilton-Russell traction versus those nursed free in bed (exact method of nursing the injured limb not specified)


OutcomesLength of follow-up: unknown (until discharge)
Daily pain scores (visual analogue score 0 [no pain] to 10 [worst pain])
Analgesic use on day 1, 2 and 3
Pressure sores (grades 1 and 2; reference to Morison 1989. Grade 1 = "simple erythema")
Difficulty in fracture reduction
Length of hospital stay


NotesPain scale direction inferred.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)High risk“The patients were randomised using the last digit of their registration number.”

Allocation concealment (selection bias)High riskQuasi-randomised - based on hospital registration number.

Blinding (performance bias and detection bias)
Subjective outcomes
High riskNo blinding except surgeon unaware of the preoperative regimen.

Blinding (performance bias and detection bias)
Objective outcomes
Low riskUnlikely to be affected: mortality, duration of operation, bleeding.

Incomplete outcome data (attrition bias)
Short term outcomes
Unclear riskActive surveillance but poor reporting with reliance on graphs.

Incomplete outcome data (attrition bias)
Post-operative outcomes
Unclear riskNo mention of mortality with respect to hospital length of stay.

Selective reporting (reporting bias)High riskNo protocol. Some possible post-hoc decisions, such as presenting fracture reduction data for one surgeon; and reporting of pressure sore results.

Balanced baseline characteristics?Unclear riskThe baseline characteristics appeared balanced (“well matched”). With reference to the imbalance in patient numbers in the two groups the authors stated: “The reason for the difference is unclear and may relate to the allocation of registration numbers by the hospital administration.”

Free of performance bias?Low risk“In all cases definitive surgical treatment was planned for the next available operating list.”
Results for ease of reduction only given for one surgeon.

Draper 1997

MethodsRandomised using a random numbers table to order colour coded booklets (pink = traction; blue = no traction) into a pile. The top envelope of the pile was withdrawn when a patient was admitted into the study.
Possibility of post-randomisation exclusions or transfer of patients from one group to another (see Notes).


Participants303 people with a proximal femoral fracture
Hospital in Hull, UK
Period of study: not stated
Excluded: patients refusing consent; multiple fractures or injuries; presence of pressure sores graded 3 or above at hospital admission; transfer from another hospital; fracture not considered suitable for surgical treatment; absence, paralysis or severe contraction of lower limb; presence of skin condition severe enough to prevent application of skin extension tapes.
Mean age: 80.5 years
Female: 84.5%
Intracapsular fractures : 48.5%

Assigned: 121/182 [traction / control]


InterventionsPre-operative skin traction using 2.5 kg weight of traction applied via Hamilton-Russell traction versus those nursed free in bed (injured limb placed on pillow; patient encouraged to adopt position of greatest comfort)


OutcomesLength of follow-up: 7th post-operative day
Pain scores (visual analogue scale: 0 [no pain] to 10 [excruciating pain]): at rest and during movement in bed, pre-operative only.
Analgesic consumption
Pressure sores (22 pressure sites inspected and graded. Graded using Torrance 1983. Score 1 = blanching (skin goes white on pressure) erythema, score 2 = non blanching erythema, score 3 = ulceration)


NotesAdditional report of trial provided by main author.
The number of participants in the two groups were dissimilar (121 versus 182), as was the number of participants with low (0) mental test scores (2 versus 37). The authors suggested the reason for this might be failure by nursing staff to apply traction to mentally confused participants allocated to the traction group; these were then included in the non-traction group. Because of this potential bias in the different characteristics of the two groups, analysis of co-variance was undertaken by the authors to try to correct for the imbalance.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk“We used a random numbers table. We had colour coded booklets for data collection – pink for traction and blue for control group. We used a random numbers table to place the book on order. Odd numbers = pink = traction, even numbers = blue = no traction – and then as patients were admitted to the study we allocated the next book that was on top of the pile.” (personal communication)

Allocation concealment (selection bias)High riskThere was no concealment of allocation.

Blinding (performance bias and detection bias)
Subjective outcomes
High riskNo blinding.

Blinding (performance bias and detection bias)
Objective outcomes
Unclear riskNo blinding and no safeguards for pressure sore measurement.

Incomplete outcome data (attrition bias)
Short term outcomes
High riskLarge difference in the numbers in each intervention group (121 versus 182) suggests a failure of randomisation methodology. Difference in mental cognitive scores also: number of participants with low (0) mental test scores (2 versus 37). It is likely that an unknown number of participants allocated traction were transferred to the non-traction group.

Incomplete outcome data (attrition bias)
Post-operative outcomes
High riskAs above, but post-operative pressures sore results not given.

Selective reporting (reporting bias)Unclear riskProtocol was clearly in place but outcome analysis modified to address imbalance in mental cognitive score status.

Balanced baseline characteristics?High riskMajor imbalance in mental test scores. Higher [better] scores in traction group (p < 0.001). Authors speculated that “possible explanation is that nurses decided not to apply traction to patients with very low mental test scores who had been randomized to the traction group.” “Number of patients in the traction group with a mental test score of 0 was two, while the number in the no-traction group was 37.”

Free of performance bias?Unclear riskMention in Nursing Times article reporting on experience at trial hospital site that “The traction study, however, was seen as a hindrance because, in the absence of the researchers, ward staff were expected to apply skin traction. Staff had [initial] difficulty with this as they felt skin traction prejudiced skin integrity.”

Finsen 1992

MethodsRandomised by the use of random numbers.
38 post-randomisation exclusions: 21 impacted fractures, 17 who had surgery within 6 hours of admission.


Participants118 people with a proximal femoral fracture
Hospital in Orkanger, Norway

Period of study: not stated
Mean age: 79 years
Female: 74%
Intracapsular fractures: 58%

Assigned: ?/?/?
Assessed: 26/29/25 [skin traction / skeletal traction / control]


InterventionsPre-operative skin traction using 3 kg weight of traction applied via a pulley at the end of the bed versus pre-operative skeletal traction (10% of body weight) applied via a Steinman pin versus those nursed free in bed (injured leg placed on pillow)


OutcomesLength of follow-up: included surgery; analgesic data for 24 hours
Analgesic consumption
Time taken in fracture reduction
Operation time
Operative blood loss
Mortality


NotesReply received: no new information


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk"On admission they were allocated with the aid of random numbers to one of three preoperative treatment groups."

Allocation concealment (selection bias)Unclear riskNo details given

Blinding (performance bias and detection bias)
Subjective outcomes
High riskNo indication of blinding. Note though that “Analgesic medication was standardized and recorded.”

Blinding (performance bias and detection bias)
Objective outcomes
Unclear riskUnlikely to be affected: mortality, duration of operation, bleeding.

Incomplete outcome data (attrition bias)
Short term outcomes
High riskWhile “five people were not treated according to protocol”, “all protocol deviants were retained in their intended group for evaluation of the results.”
However, given randomisation was on admission, it is possible that the exclusion of  “17 patients who underwent operation within 6 h of admission” was after randomisation. The same may apply to 21 patients excluded because they had impacted fractures.

Incomplete outcome data (attrition bias)
Post-operative outcomes
Unclear riskNo report of these outcomes

Selective reporting (reporting bias)High riskNo protocol. Unclear information on exclusion criteria.

Balanced baseline characteristics?Unclear riskBaseline data may not have been provided for all randomised patients. Otherwise, similar baseline characteristics in the three groups aside from “the proportion of patients who had received analgesic medication before admission was somewhat higher in the skeletal traction group than in the other two groups”.

Free of performance bias?Unclear riskMean time to surgery was comparable in the three groups but no mention of operator or surgeon's experience.

Ghnaimat 2005

MethodsRandomised by last digit of hospital admission number.


Participants74 people with a proximal femoral fracture
Hospital in Zarqa, Jordan
Period of study: February 2002 to October 2004
Excluded: patients refusing consent, patients with conditions which contraindicated the use of skin traction: e.g. skin ulceration, severe oedema or peripheral arterial disease, lower limb deformities, allergy to adhesive bandages
Mean age: 73 years
Females: 57% (see Notes)
Intracapsular fractures: 19% (see Notes)

Assigned: 36/38 [traction / control]


InterventionsPre-operative skin traction using 6 lb weight of traction applied via longitudinal traction versus those nursed free in bed (exact method of nursing the injured limb not specified)


OutcomesLength of follow-up: unknown (until discharge or up to 7 days?)
Daily pain scores (visual analogue score 0 [no pain] to 10 [worst pain])
Analgesic use on days 1 to 7
Pressure sores (grades 1 and 2; no mention of classification system)
Difficulty in fracture reduction (see Notes)
Complications related to traction
Length of hospital stay


NotesText and tables contradictory for sex ratio, present values are calculated from the text (ratio female to male: 2/1.5). This also applied to the number of intracapsular fractures, again the number in the text was used in the review.
Percentages given for the fracture reduction results do not tally with the numbers randomised.
Pain scale direction inferred.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)High risk"patients with odd admission numbers received skin traction and those with even numbers were nursed in bed without skin traction."

Allocation concealment (selection bias)High riskQuasi-randomised - based on hospital admission number.

Blinding (performance bias and detection bias)
Subjective outcomes
High risk"A subjective evaluation about the ease of reduction or performing hemiarthroplasty was done by a senior surgeon who was not informed about which patient received preoperative skin traction.” But no indication of safeguards and only for ease of reduction.

Blinding (performance bias and detection bias)
Objective outcomes
Low riskLength of stay reported - less likely to be at risk of bias.

Incomplete outcome data (attrition bias)
Short term outcomes
High riskIncomplete data and inconsistent percentages.

Incomplete outcome data (attrition bias)
Post-operative outcomes
High riskIncomplete data

Selective reporting (reporting bias)High riskNo protocol and some concerns regarding reporting of outcomes.

Balanced baseline characteristics?Low riskProbably OK - balance in age, gender, fracture type and pressure sores.

Free of performance bias?Unclear riskNo statements on this. Time to surgery appeared similar in the two groups. This may apply also to surgical experience.

Jerre 2000

MethodsRandomised by sealed opaque envelopes and involving the use of a telephone line.
No post-randomisation exclusions.


Participants120 people with a proximal femoral fracture
Hospital in Goteborg, Sweden
Period of study: not stated
Mean age: 80 years (range 50 to 96)
Female: 76%
Intracapsular fractures: 50%

Assigned: 60/60 [traction / control]


InterventionsPre-operative skin traction using 3 kg weight of traction applied to the leg via a foam rubber boot and straps and the leg placed in a traction sled versus those nursed free in bed (exact method of nursing the injured limb not specified)


OutcomesLength of follow-up: 4 months
Pain scores (visual analogue scale: 0 [pain-free] to 10 [worst pain imaginable]) at 1, 4 and 12 hours after 'treatment' (application of traction or allocation to control group)
Supplementary analgesic consumption
Pre-operative complications (pressure sores: all grade 1; classification system not stated)
Post-operative complications (e.g. urinary tract infections, red spots, pressure sores)
Fracture healing complications
Quality of fracture reduction
Wound infection
Thrombosis
Fracture compression
Mortality


NotesReply from authors of trial with supplementary information of trial methodology and results. 30 cervical fractures and 30 trochanteric fractures were allocated to traction and 30 cervical fractures and 30 trochanteric fractures to 'no traction'.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear riskRandomised. “sealed opaque envelope” (personal communication)

Allocation concealment (selection bias)Low risk“sealed opaque envelope and telephone line” Inexact details of process but seems likely that allocation was concealed.

Blinding (performance bias and detection bias)
Subjective outcomes
High riskNo mention of blinding.

Blinding (performance bias and detection bias)
Objective outcomes
Unclear riskNo mention of blinding, but for the outcomes measured (operative time, fracture healing and ease of reduction), lack of blinding is unlikely to be important.

Incomplete outcome data (attrition bias)
Short term outcomes
Low riskNo losses or exclusions

Incomplete outcome data (attrition bias)
Post-operative outcomes
Low riskFew lost to follow-up. Reasonable balance.

Selective reporting (reporting bias)Unclear riskNo protocol available

Balanced baseline characteristics?Unclear riskWell matched groups in terms of age, sex  and fracture types but no information on co-morbidities

Free of performance bias?Low riskComparable time to operation and care programmes

Needoff 1993

MethodsRandomisation based on case note number.
3 post-randomisation exclusions: not operated on.


Participants67 people with a proximal femoral fracture
Hospital in Nottingham, UK
Period of study: not stated
Excluded: 33 patients with mini-mental state examination score of 23 or less out of 30 points
Mean age: 78 years
Female: 77%
Intracapsular fractures: 50%

Assigned: 32/35 [traction / control]


InterventionsPre-operative skin traction using 2.5 kg weight of traction applied via skin traction over a pulley at the end of the bed versus those nursed free in bed (injured leg in comfortable position - flexion, abduction and external rotation - with pillow under thigh)


OutcomesLength of follow-up: unknown (2 days?)
Pain scores (visual analogue scale: 0 (pain-free) to 10 (worst pain possible))
Analgesic consumption
Pressure sores (no description of classification system or monitoring of these)
Operating time


NotesReply received: method of randomisation, no difference in care programmes, or pressure sores.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)High risk"Case home number" (personal communication)

Allocation concealment (selection bias)High risk“Case home number” (personal communication)

Blinding (performance bias and detection bias)
Subjective outcomes
High riskNo mention.

Blinding (performance bias and detection bias)
Objective outcomes
Unclear riskNo report of these measures, except operation time.

Incomplete outcome data (attrition bias)
Short term outcomes
High riskActive surveillance but incomplete data.

Incomplete outcome data (attrition bias)
Post-operative outcomes
High riskAs above. No data provided.

Selective reporting (reporting bias)High riskWell set up but no protocol and changes in outcome collection (no long term data).

Balanced baseline characteristics?Unclear riskBalanced age, gender, mental text scores, fracture type, pain at admission.

Free of performance bias?Low riskAttempt at standardisation in terms of pain medication and assessment.

Resch 1998

MethodsRandomised by closed envelopes. After X-ray diagnosis, the nurse in the emergency department taking care of the patient drew an envelope from a pre-prepared set of closed envelopes.


Participants78 people with a displaced proximal femoral fracture
Hospital in Lund, Sweden
Period of study: not stated
Excluded: 75 patients were excluded mainly due to senile confusion. refusal to participate and skin problems such as leg ulcers.
Mean age: 81 years
Female: 73%
Intracapsular fractures: 55%

Assigned: 40/38 [skin traction / skeletal traction]


InterventionsPre-operative skin traction using 3 kg weight of traction applied to the leg via a foam rubber boot and straps and the leg placed in a traction sled versus pre-operative skeletal traction (5% to 10% of body weight) applied via a K-wire though the proximal tibia


OutcomesLength of follow-up: 3 to 4 days
Pain scores (visual analogue scale: 0 [pain-free] to 10 [worst pain imaginable])
Use of analgesics
Complications related to traction
Length of operation
Time spent in hospital departments


NotesAbstract only available for the first version of review.
Small discrepancies in denominators for numbers of participants who found the application of traction painful. Text gives 7/35 versus 16/43. Possible intention-to-treat problem but also could be results for trochanteric (35) versus cervical (43) fractures. Results from abstract (and summary of full report) retained for this outcome.
Information on method of randomisation received indirectly from authors. Also confirmation that no patient was lost to follow-up.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear risk“closed envelopes” (personal communication)

Randomised by closed envelopes. After X-ray diagnosis, the nurse in the emergency department taking care of the patient drew an envelope from a pre-prepared set of closed envelopes. (personal communication)

Allocation concealment (selection bias)Unclear risk“closed envelopes” (personal communication)

Randomised by closed envelopes. After X-ray diagnosis, the nurse in the emergency department taking care of the patient drew an envelope from a pre-prepared set of closed envelopes. (personal communication)

Blinding (performance bias and detection bias)
Subjective outcomes
High riskNo blinding mentioned

Blinding (performance bias and detection bias)
Objective outcomes
Unclear riskNo blinding but uncertain risk of bias

Incomplete outcome data (attrition bias)
Short term outcomes
High riskUnresolved discrepancies regarding percentages and numbers available at follow-up.

Incomplete outcome data (attrition bias)
Post-operative outcomes
High risk[Complications from chart review and on ward.] Uncertain about risk of bias.

Selective reporting (reporting bias)Unclear riskNo protocol available – although some consistency with approach in the two trials

Balanced baseline characteristics?Unclear riskBalanced regarding gender, fracture type, initial pain but no information otherwise

Free of performance bias?Unclear riskComparable in time to operation and follow-up, but insufficient to tell.

Resch 2005

MethodsRandomised by closed envelopes. After X-ray diagnosis, the nurse in the emergency department taking care of the patient drew an envelope from a pre-prepared set of closed envelopes. Queried imbalance in the numbers in the 3 intervention groups with the trialists but this remained unexplained (49; 21; 53).


Participants123 people with a displaced proximal femoral fracture
Hospital in Lund, Sweden
Period of study: not stated
Excluded: patients unable to give informed consent; local problems that would prohibit the use of skin traction such as ulcers, eczema or perivascular disease
Mean age: 81 years
Female: 73%
Intracapsular fractures: 46%

Assigned: 49/21/53 [skin traction / Lasse pillow / control]


InterventionsPre-operative skin traction using 3 kg weight of traction applied to the leg via a foam rubber boot and straps and the leg placed in a traction sled versus those with their lower leg placed in a special foam pillow (Lasse pillow) that allowed some movement while preventing inadvertent movement of the injured leg versus those nursed free in bed (injured leg in resting position with pillow placed under thigh).
Leg position - 30 degrees flexion and slight outward rotation of the hip and supported by pillows - same in the 3 groups


OutcomesLength of follow-up: 3 to 4 days
Pain scores (visual analogue scale: 0 [pain-free] to 10 [worst pain imaginable])
Use of analgesics
Complications (erythema (2), oedema (1) and paraesthesia (1))
Length of operation
Time spent in hospital departments


NotesPossible intention-to-treat problem. Reply received: method of randomisation, some baseline characteristics and complications split by treatment group.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear risk“closed envelopes” (personal communication)

Randomised by closed envelopes. After X-ray diagnosis, the nurse in the emergency department taking care of the patient drew an envelope from a pre-prepared set of closed envelopes. (personal communication)

Allocation concealment (selection bias)Unclear riskclosed envelopes” Randomised by closed envelopes. After X-ray diagnosis, the nurse in the emergency department taking care of the patient drew an envelope from a pre-prepared set of closed envelopes. (personal communication

Blinding (performance bias and detection bias)
Subjective outcomes
High riskNo blinding mentioned

Blinding (performance bias and detection bias)
Objective outcomes
Unclear riskNo blinding but uncertain risk of bias

Incomplete outcome data (attrition bias)
Short term outcomes
High riskUnresolved discrepancies including numbers randomised

Incomplete outcome data (attrition bias)
Post-operative outcomes
Unclear riskComplications data obtained

Selective reporting (reporting bias)Unclear riskNo protocol available – although some consistency with approach in the two trials

Balanced baseline characteristics?Unclear riskBalanced regarding pain but no information on population

Free of performance bias?Unclear riskComparable in time to operation and follow-up, but insufficient to tell.

Rosen 2001

MethodsRandomised using a computer programme to randomly assign 100 sequential 'slots' to #1 or #2; these were then assigned as traction or pillow.


Participants100 people with a proximal femoral fracture
Hospital in New York, USA

Period of study: June 1995 to February 1997
Excluded: patients younger than 50 years of age, underlying dementia, other concomitant injury, presentation more than 24 hours after the initial injury. Patients had to have adequate cognitive function to be considered for inclusion.
Mean age: 78 years (range 50-97 years)
Female: 78%
Intracapsular fractures: 55% (in text), 43% (in table)

Assigned: 50/50 [traction / control]


InterventionsPre-operative skin traction using 5 pounds weight of traction applied via a foam traction boot versus those nursed free in bed (injured leg in resting position with pillow placed under thigh)


OutcomesLength of follow-up: till surgery
Pain scores (visual analogue scale: 0 [no pain] to 10 [extreme pain]): 15 minutes after application traction or leg rested on pillows and the following morning.
Pain on application traction/rest on pillows
Analgesic consumption
Complications of traction


NotesReply received: method of randomisation, all patients received allocated treatment, identical pre-operative care.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk“randomization was performed using a computer-generated program”.

Allocation concealment (selection bias)Unclear riskNo other details given.

Blinding (performance bias and detection bias)
Subjective outcomes
High riskNo blinding.

Blinding (performance bias and detection bias)
Objective outcomes
Unclear riskThese outcomes were not assessed.

Incomplete outcome data (attrition bias)
Short term outcomes
Unclear riskSome discrepancies in the data (pain at baseline & types of fracture).

Incomplete outcome data (attrition bias)
Post-operative outcomes
Unclear riskOutcomes not assessed.

Selective reporting (reporting bias)Unclear riskNo protocol available.

Balanced baseline characteristics?Low riskGroups were well matched with respect to baseline data, including initial pain.

Free of performance bias?Low riskActive, standardised medication and follow-up; similar time to operation.

Saygi 2010

MethodsRandomised by order of admission to the hospital


Participants108 people with a proximal femoral fracture

Two hospitals in Istanbul, Turkey

Excluded: patients refusing consent, patients with cognitive inadequacy

Mean age: 76 (range 19 to 100)

Females: 67%

Intracapsular fractures: 44%

Assigned: 36/36/36 [traction / placebo traction / control]


InterventionsPre-operative skin traction using 2 kg weight of traction along with a pillow beneath the leg versus application of skin traction without the use of weights along with a pillow beneath the injured leg versus pillow placed beneath the injured leg


OutcomesLength of follow-up: unclear (up to operation?)

Pain scores (visual analogue scale: 0 [no pain] to 10 [worst pain]): 1, 4 and 12 hours after application
Number of analgesic injections / day

Adverse effects: pressure sores and neuropraxia


NotesTiming of follow-up and measurement of adverse outcomes were not described in the trial methods


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)High risk“The 108 patients who participated in the study were randomly allocated into three groups, according to the order of admission to the hospital.”

Allocation concealment (selection bias)High riskAs above. Quasi-randomised method.

Blinding (performance bias and detection bias)
Subjective outcomes
High riskNo mention of blinding

Blinding (performance bias and detection bias)
Objective outcomes
Unclear riskUnclear risk of bias from lack of blinding of adverse effects

Incomplete outcome data (attrition bias)
Short term outcomes
Low riskPre-randomisation exclusions given – seven patients excluded due to refusal to give consent or cognitive inadequacy. Loss of follow-up unlikely

Incomplete outcome data (attrition bias)
Post-operative outcomes
Unclear riskNo post-op outcomes

Selective reporting (reporting bias)Unclear riskNo protocol available.

Balanced baseline characteristics?Unclear riskGroups were well matched with respect to age, sex and type of fracture. No mention of medical co-morbidities.

Free of performance bias?Low risk“Anti-thrombotic prophylaxis was routinely administered to all patients. The patients also received paracetamol tablets orally 3-times a day routinely.” Time to operation was similar in the three groups (mean 53.2 hours for whole population).

Yip 2002

MethodsRandomised by last digit of patient's registration number.


Participants311 people with a proximal femoral fracture
Hospital in Hong Kong, China
Period of study: August 1995 to December 1997
Excluded: patients with senile dementia or taking regular analgesia prior to admission
Mean age: 79 years
Females: 66%
Intracapsular fractures: % not stated

Assigned: 166/145 [traction / control]


InterventionsPre-operative skin traction using 2 kg weight of traction applied via a foam boot versus those nursed free in bed (injured leg placed on pillow)


OutcomesLength of follow up: one year (however, only data up to one week from admission were presented).
Pain scores (visual analogue score: 0 [no pain] to 2 [worst pain]) assessed 4 times each day pre-operatively
Analgesic use pre-operatively
Operative blood loss
Operative time
Complications (no description of recording of these, including the monitoring of pressure sores)


NotesReply received: no difference in care programmes, confirmation of no loss to follow-up, no pressure sores, full results for operative time.
Top end (2) of pain scale inferred from graph in article.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)High risk“Patients were randomised into two study arms depending on whether their hospital admission number was an odd or even number”.

Allocation concealment (selection bias)High riskQuasi-randomised - based on hospital registration number.

Blinding (performance bias and detection bias)
Subjective outcomes
High riskNo mention in paper. Assessor blinding claimed in an email from the trial investigator but with no supporting evidence. A research assistant collected the data,

Blinding (performance bias and detection bias)
Objective outcomes
Unclear riskOutcomes were not assessed.

Incomplete outcome data (attrition bias)
Short term outcomes
Unclear riskLack of unable data. No pressure sores?

Incomplete outcome data (attrition bias)
Post-operative outcomes
Unclear riskOutcomes not assessed.

Selective reporting (reporting bias)Unclear riskNo protocol although study appeared well set up.

Balanced baseline characteristics?Unclear riskGroups were well matched in age and sex. There was a discrepancy in the numbers for each group (166 v 145) which probably reflects the inadequate randomisation methodology. No mention of co-morbidities.

Free of performance bias?Low riskActive, standardised medication and follow-up; similar type of operations ratios and time to operation.

 
Comparison 1. Pre-operative traction versus no traction

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Pain (Visual Analogue Scale: 0: none to 10: worst imaginable)3Mean Difference (IV, Fixed, 95% CI)Subtotals only

    1.1 Soon after immobilisation
3274Mean Difference (IV, Fixed, 95% CI)0.11 [-0.27, 0.50]

    1.2 At 12 hours
172Mean Difference (IV, Fixed, 95% CI)0.24 [-0.13, 0.61]

 2 Analgesic use on ward (participants)2Risk Ratio (M-H, Fixed, 95% CI)Totals not selected

    2.1 Day 1
1Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]

    2.2 Day 2
1Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]

    2.3 Until surgery
1Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]

    2.4 High analgesic use
1Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]

 3 Analgesic use on ward (doses / injections)2Mean Difference (IV, Fixed, 95% CI)Totals not selected

    3.1 Doses on ward
1Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]

    3.2 Injections / day
1Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]

 4 Pressure sores2Risk Ratio (M-H, Fixed, 95% CI)Totals not selected

    4.1 Pre-operative
1Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]

    4.2 Post-operative
1Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]

 5 Difficulty in fracture reduction2183Risk Ratio (M-H, Fixed, 95% CI)0.92 [0.51, 1.67]

 6 Poor quality fracture reduction1Risk Ratio (M-H, Fixed, 95% CI)Totals not selected

 7 General complications3Risk Ratio (M-H, Fixed, 95% CI)Totals not selected

    7.1 Neurapraxia
1Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]

    7.2 Post-operative (including pressure sores)
2Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]

    7.3 Complications at 3-4 days
1Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]

 8 Fracture fixation failure1Risk Ratio (M-H, Fixed, 95% CI)Totals not selected

    8.1 All fractures
1Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]

    8.2 Intracapsular fracture
1Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]

    8.3 Extracapsular fracture
1Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]

 9 Length of operation (minutes)2413Mean Difference (IV, Fixed, 95% CI)1.28 [-4.82, 7.39]

 10 Intra-operative blood loss (ml)1Mean Difference (IV, Fixed, 95% CI)Totals not selected

 11 Length of hospital stay (days)1Mean Difference (IV, Fixed, 95% CI)Totals not selected

 
Comparison 2. Traction versus no or 'placebo' traction

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Pain (VAS) - soon after immobilisation1Mean Difference (IV, Fixed, 95% CI)Totals not selected

    1.1 'Placebo' traction - no weights added
1Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]

    1.2 No traction - pillow only
1Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]

 2 Pain (VAS) - at 12 hours1Mean Difference (IV, Fixed, 95% CI)Totals not selected

    2.1 'Placebo' traction - no weights added
1Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]

    2.2 No traction - pillow only
1Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]

 3 Analgesic use - injections / day1Mean Difference (IV, Fixed, 95% CI)Totals not selected

    3.1 'Placebo' traction - no weights added
1Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]

    3.2 No traction - pillow only
1Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]

 
Comparison 3. Skin traction versus skeletal traction

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Pain soon after traction (Visual Analogue Scale: 0: none to 10: worst imaginable)1Mean Difference (IV, Fixed, 95% CI)Totals not selected

 2 Analgesic use on ward (number of doses)1Mean Difference (IV, Fixed, 95% CI)Totals not selected

 3 Length of surgery (minutes)1Mean Difference (IV, Fixed, 95% CI)Totals not selected