Perioperative fluid volume optimization following proximal femoral fracture
Abstract
Background
Proximal femoral fracture (PFF) is a common orthopaedic emergency, affecting mainly elderly people at high risk of complications. Advanced methods for managing fluid therapy during treatment for PFF are available, but their role in reducing risk is unclear.
Objectives
To compare the safety and effectiveness of different methods of perioperative fluid optimization in adult participants undergoing surgical repair of hip fracture. We considered the following methods: advanced invasive haemodynamic monitoring, such as transoesophageal Doppler and pulse contour analysis; a protocol using standard measures, such as blood pressure, urine output and central venous pressure; and usual care.
Comparisons of fluid types (e.g. crystalloid vs colloid) and other methods of optimizing oxygen delivery, such as blood product therapies and pharmacological treatment with inotropes and vasoactive drugs, are considered elsewhere.
Search methods
We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2012, Issue 9); MEDLINE (1966 to October 2012); and EMBASE (1980 to October 2012) without language restrictions. We ran forward and backward citation searches on identified trials. We contacted authors and searched ClinicalTrials.gov and the WHO International Clinical Trials Registry Platform for unpublished trials. This is an updated version of a review published in 2004. The original search was performed in October 2003.
Selection criteria
We included randomized controlled trials (RCTs) in adult participants undergoing surgical treatment for PFF, which compared any two of advanced haemodynamic monitoring, protocols using standard measures or usual care, irrespective of blinding, language or publication status.
Data collection and analysis
Two review authors assessed the impact of fluid optimization interventions on outcomes of mortality, length of hospital stay, return of participant to pre‐fracture accommodation and mobility at six months and adverse events in hospital. We pooled data using risk ratio or mean difference for dichotomous or continuous data, respectively, based on random‐effects models.
Main results
We included three RCTs with a total of 200 participants. One of these included studies was found to have a high risk of bias; no trial featured all pre‐specified outcomes. We found one trial for which data are awaited for classification and two ongoing trials. One included study with low risk of bias found that compared with usual care, time to medical fitness for discharge was shorter with the use of advanced haemodynamic monitoring (mean reduction 6.20 days, 95% CI 2.3 to 10.1 days; 59 participants, one trial) and with the use of protocols that apply standard measures (mean reduction 3.9 days, 95% CI 0.75 to 7.05; 57 participants, one trial). Our results are consistent with both increased and decreased risk of mortality and adverse events in participants receiving the intervention. No data for other outcomes were available. Our results are limited by the quantity of available data.
Authors' conclusions
Three studies considering a total of 200 participants reveal an absence of evidence that fluid optimization strategies improve outcomes for participants undergoing surgery for PFF. Length of hospital stay may be improved, but lack of good quality data leaves uncertainty. Further research powered to test some of these outcomes is ongoing.
PICOs
Plain language summary
Optimization of fluid levels in people suffering hip fractures
Hip fractures are common in elderly people, who often have medical conditions that put them at risk of developing other problems whilst their fracture is treated. Treatment usually involves an operation to fix the break in the bone, and it is possible that giving too much or too little fluid to a patient around this time may increase the risk of further problems. Healthcare staff can use many approaches in trying to determine how much fluid a patient needs in this situation, but it is not clear if some methods are better than others. For this Cochrane review, researchers from The Cochrane Collaboration looked at research on the effects of different methods of optimizing fluid levels for adult men and women who underwent surgery for any type of hip fracture. We searched the databases to October 2012 and identified three studies (randomized controlled trials) with a total of 200 people, each of which compared two or three methods of guiding fluid therapy. These methods include 'usual care' (where staff use changes in basic measurements, such as heart rate, to decide for themselves how much fluid to give), 'protocols using standard measures' (where staff use changes in basic measurements to give fluid according to a formal set of rules) and 'advanced haemodynamic monitoring' (where staff use equipment, such as specialized blood pressure monitoring devices placed into arteries, to guide how much fluid to give). These trials found no evidence that using one method instead of another reduces harm, including death or number of complications. One study suggests that length of stay in the hospital may be reduced if protocols or advanced haemodynamic methods are used, but because the number of people studied is not large, it is not possible to draw firm conclusions about this. No information was found regarding differences in the time taken for people to return to their previous type of accommodation or level of mobility. Two ongoing studies may provide more information in the future. The quality of evidence in a review may be high, moderate, low or very low. In this review, the evidence was assessed as being of low quality for all outcomes except time to medical fitness for discharge, for which the quality of evidence was moderate. Research findings to this point are insufficient to show how one can best optimize fluid levels in the large number of people around the world suffering from hip fracture. This is an update of a review published in 2004.
Authors' conclusions
Summary of findings
| Advanced haemodynamic monitoring compared with protocol using standard measures for perioperative fluid volume optimisation | ||||||
| Patient or population: patients with proximal femoral fracture | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No. of participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Protocol using standard measures such as CVP | Advanced haemodynamic monitoring | |||||
| All‐cause mortality | Moderatea | RR 0.52 | 61 | ⊕⊕⊝⊝ | ||
| 66 per 1000 | 34 per 1000 | |||||
| Total length of hospital stay | The mean total length of hospital stay in the control groups was | The mean total length of hospital stay in the intervention groups was | 61 | ⊕⊕⊝⊝ | ||
| Time to medical fitness for discharge | The mean time to medical fitness for discharge in the control groups was | The mean time to medical fitness for discharge in the intervention groups was | 61 | ⊕⊕⊝⊝ | ||
| Adverse outcomes Cardiopulmonary－not reported | See comment | See comment | Not estimable | ‐ | See comment | No data suitable for analysis available |
| Adverse outcomes－neurological | Moderated | RR 2.07 | 61 | ⊕⊕⊝⊝ | ||
| 10 per 1000 | 21 per 1000 | |||||
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence: | ||||||
| aBased on mortality rate from Moppett 2012. | ||||||
| Advanced haemodynamic monitoring compared with usual care for perioperative fluid optimization | ||||||
| Patient or population: patients with proximal femoral fracture | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No. of participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Usual care | Advanced haemodynamic monitoring | |||||
| All‐cause mortality | Moderatea | RR 1.03 | 99 | ⊕⊕⊝⊝ | ||
| 66 per 1000 | 68 per 1000 | |||||
| Total length of hospital stay | The mean total length of hospital stay in the control groups was | The mean total length of hospital stay in the intervention groups was | 59 | ⊕⊕⊝⊝ | ||
| Time to medical fitness for discharge | The mean time to medical fitness for discharge in the control groups was | The mean time to medical fitness for discharge in the intervention groups was | 59 | ⊕⊕⊕⊝ | ||
| Adverse outcomes Cardiopulmonary－not reported | See comment | See comment | Not estimable | ‐ | See comment | No data suitable for analysis available |
| Adverse outcomes－neurological | Moderated | RR 1.93 | 59 | ⊕⊕⊝⊝ | ||
| 10 per 1000 | 19 per 1000 | |||||
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence: | ||||||
| aBased on mortality rate from Moppett 2012. | ||||||
| Protocol using standard measures such as CVP compared with usual care for perioperative fluid optimization | ||||||
| Patient or population: patients with proximal femoral fracture | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No. of participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Usual care | Protocol using standard measures such as CVP | |||||
| All‐cause mortality | Moderatea | RR 2.81 | 60 | ⊕⊕⊝⊝ | ||
| 66 per 1000 | 185 per 1000 | |||||
| Total length of hospital stay | The mean total length of hospital stay in the control groups was | The mean total length of hospital stay in the intervention groups was | 57 | ⊕⊕⊝⊝ | ||
| Time to medical fitness for discharge | The mean time to medical fitness for discharge in the control groups was | The mean time to medical fitness for discharge in the intervention groups was | 57 | ⊕⊕⊕⊝ | ||
| Adverse outcomes Cardiopulmonary－not reported | See comment | See comment | Not estimable | ‐ | See comment | Data suitable for analysis not available |
| Adverse outcomes－neurological | Moderated | RR 0.94 | 60 | ⊕⊕⊝⊝ | ||
| 10 per 1000 | 9 per 1000 | |||||
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence: | ||||||
| aBased on mortality rate from Moppett 2012. | ||||||
Background
Description of the condition
Proximal femoral fractures (PFFs), or hip fractures, are fractures of the femur immediately distal to the articular surface of the hip joint, to about 5 cm below the lesser trochanter. They can be subdivided into intracapsular and extracapsular fractures. Intracapsular (also termed transcervical or subcapital) fractures occur proximal to the trochanteric line, and extracapsular (also termed pertrochanteric, subtrochanteric, trochanteric or intertrochanteric) fractures occur distal to the trochanteric line, up to 5 cm below the lower border of the lesser trochanter.
These fractures most commonly occur in elderly people with osteoporosis, following a simple mechanical fall. Approximately 1.5 million hip fractures are reported per year worldwide, with a projected increase to 4.5 million by 2050 (Gullberg 1997; Sterling 2011). Incidence varies by country, from about 50 to 500 per 100,000, and is about two times higher in females than males, although this difference varies with race (Kanis 2012; Kannus 1996).
PFF is one of the most common orthopaedic emergencies, and most cases are managed by early surgical fixation to reduce complications from the prolonged immobility associated with conservative treatment. Limited evidence has been obtained from randomized controlled trials (RCTs) to inform this practice, but other types of studies have shown increased risk of death if surgery is delayed; it would be difficult to conduct ethically sound trials comparing operative with conservative treatment (Bottle 2006; Handoll 2008). Generally, medically fit patients should undergo surgery within 24 hours.
Undisplaced intracapsular fractures are usually treated by internal fixation to preserve the femoral head, with the use of screws or pins, with or without plates, to the femur. Displaced intracapsular fractures may be reduced and internally fixated or may undergo replacement arthroplasty. Extracapsular fractures may be fixed with a screw passed up the femoral neck to the head and then attached to a plate on the side of the femur, or an intramedullary nail may be used with a side screw passed up into the femoral head.
Description of the intervention
Age‐related co‐morbidities and dehydration in people presenting with PFF put them at increased risk for peritraumatic and perioperative complications. Providing adequate fluid resuscitation is important in minimizing this risk. The adequacy of fluid therapy may be determined by using simple, readily available clinical measures, such as tissue turgor, heart rate, blood pressure, urine output and central venous pressure (CVP). However, these are non‐specific and poorly sensitive measures of fluid optimization (Marik 2008). Growing evidence suggests that predicting responsiveness to fluid therapy is more important (Funk 2009). The aim here is to use goal‐directed fluid therapy to optimize cardiac output, to avoid overloading the cardiovascular system and precipitating heart failure. Alongside adequate haemoglobin and inspired oxygen levels, this optimizes delivery of oxygen to tissues and organs and may improve outcomes (Green 2010).
One way to assess fluid responsiveness is to use a protocol that combines several simple measures to determine the effect of a standardized fluid bolus and to decide whether additional fluid will provide benefit for the patient. Another method is to use advanced haemodynamic monitoring techniques to detect cardiovascular changes that occur with incremental fluid boluses, to predict responsiveness to increased fluid. Although some of these advanced techniques are in their infancy, a number have become established in clinical practice. These can be split into static measures of cardiac preload (the load placed on the heart by blood returning to it) and dynamic measures of interactions between heart and lung.
Static measures aim to determine cardiac preload but fail to estimate the response to fluids in about one‐half of patients, thus rendering them exposed to the hazards of unnecessary fluid therapy (Eyre 2010). Despite this, many of these measures are in clinical use. Right ventricular end‐diastolic volume can be measured by fast‐response thermistor pulmonary artery catheter or by cardiac scintigraphy. Transoesophageal echocardiography (TOE) can measure left ventricular (LV) end‐diastolic area, which correlates well with left ventricular end‐diastolic volume－a measure of LV preload. Transpulmonary thermodilution using a commercially available device (PiCCO, Pulsion Medical Systems) assesses global end‐diastolic volume (GEDV), the largest volume of blood contained within the four heart chambers, and intrathoracic blood volume; both are validated as indicators of cardiac preload (Bendjelid 2010; Muller 2008). Pulmonary artery (Swan‐Ganz) catheters are inserted into the pulmonary artery to measure pulmonary artery occlusion pressure (PAOP); however, their use has been reduced over recent years, as PAOP has been shown to be a poor marker of left ventricular end‐diastolic volume, and therefore of cardiac preload and cardiac output (Maus 2008).
Dynamic measures are generally superior to static measures in predicting fluid responsiveness, although this has been demonstrated mainly in sedated ventilated patients (Eyre 2010). Various technologies use these measures, which include stroke volume variation (SVV), systolic pressure variation (SPV), pulse pressure variation (PPV), aortic blood velocity (ABV), superior vena cava collapsibility index (SVCCI) and inferior vena cava distensibility index (IVCDI). The commercially available LiDCO device (Vigileo) analyses the waveform of the arterial blood pressure pulse for SVV, SPV and PPV; transthoracic echocardiography (TTE) measures IVCDI; oesophageal Doppler measures SVV and ABV; TOE can measure SVCCI and ABV; and PiCCO can measure SVV. LiDCO and TTE can be used with patients who are awake when undergoing regional anaesthesia and with those who are unconscious when undergoing general anaesthesia. Oesophageal Doppler and TOE can be used only with patients who are undergoing general anaesthesia.
How the intervention might work
Major surgery and critical illness are associated with increased oxygen demand due to a systemic inflammatory response, the stress response, and increased metabolic activity. Inadequate fluid resuscitation and cardiopulmonary disease may reduce the supply of adequate tissue blood flow and delivery of oxygen. This may result in cellular dysfunction, organ damage, organ failure and ultimately death. Fluid overload is also harmful, potentially causing cardiac performance to fall as the result of extreme right shift on the Starling myocardial performance curve, respiratory failure due to fluid accumulation in the lungs, gastric dysmotility and poor wound healing. Growing evidence indicates that standardized methods to optimize fluid and oxygen delivery to tissues may decrease morbidity and mortality in a variety of clinical settings, particularly among high‐risk surgical patients and those with critical illness or sepsis (Lees 2009).
Why it is important to do this review
Protocols, or advanced methods for managing fluid therapy, may improve various outcomes in the large number of people who suffer from PFF each year. However, these methods also have the potential for harm and incur financial cost. A systematic evaluation of the current evidence is needed to assist clinicians in attempting to optimize fluid volume status in people undergoing surgery for PFF. The outcomes that we included were selected according to their frequency of use in studies of PFF and their usefulness in clinical decision making (Liem 2012).
This is an update to a Cochrane review first published in 2004 (Price 2004). Because new monitoring techniques and revised methods have been introduced within The Cochrane Collaboration, we have re‐run the searches, including extra search terms, and have used different methods to assess study quality.
Objectives
To compare the safety and effectiveness of different methods of perioperative fluid optimization in adult participants undergoing surgical repair of hip fracture. We considered the following methods: advanced invasive haemodynamic monitoring, such as transoesophageal Doppler and pulse contour analysis; a protocol using standard measures, such as blood pressure, urine output and central venous pressure; and usual care.
Comparisons of fluid types (e.g. crystalloid vs colloid) and other methods of optimizing oxygen delivery, such as blood product therapies and pharmacological treatment with inotropes and vasoactive drugs, are considered elsewhere.
Methods
Criteria for considering studies for this review
Types of studies
We included only randomized controlled trials (RCTs), including cluster‐randomized trials. Quasi‐randomized trials (e.g. alternation) and trials in which treatment allocation was inadequately concealed were also considered for inclusion. We included unpublished studies and studies published only in abstract form if adequate method and results data could be obtained. We did not expect to identify any cross‐over trials for this condition.
Types of participants
We included studies on adults who underwent acute surgical treatment of any type for PFF while under regional or general anaesthesia.
Types of interventions
We included studies that compared the use of any two of the following.
-
Advanced invasive haemodynamic monitoring, such as transoesophageal Doppler and pulse contour analysis.
-
A protocol using standard measures, such as blood pressure, urine output and central venous pressure.
-
Usual care.
We undertook reviews of three different comparisons.
-
Advanced haemodynamic monitoring versus a protocol using standard measures.
-
Advanced haemodynamic monitoring versus usual care.
-
A protocol using standard measures versus usual care.
Types of outcome measures
Primary outcomes
-
All‐cause mortality (within 30 days if reported, otherwise as reported in the trial).
-
Length of hospital stay.
-
-
Total length of hospital stay.
-
Time to medical fitness for discharge.
-
-
Return of participant to pre‐fracture category of accommodation at six months.
-
Return to pre‐fracture mobility at six months.
Secondary outcomes
-
Major adverse events in hospital.
-
-
Iatrogenic (related to intervention, e.g. pneumothorax, haemothorax, upper limb thrombosis, line sepsis, local haematoma).
-
Cardiopulmonary (e.g. myocardial infarction, cardiac or respiratory failure, thromboembolic event).
-
Neurological (e.g. delirium, postoperative cognitive dysfunction, cerebrovascular accident).
-
We also recorded any complications reported in the study, including minor events.
Outcomes did not form part of the study eligibility assessment. Studies that met design, participant and intervention criteria were included in the review even if they did not report any relevant outcomes.
Search methods for identification of studies
Electronic searches
We searched for relevant randomized trials published in any language. We searched the Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library (2012, Issue 10, see Appendix 1); MEDLINE via Ovid SP (1966 to October 2012, see Appendix 2); and EMBASE via Ovid SP (1982 to October 2012, see Appendix 3). For searching in MEDLINE, we combined our topic‐specific key words with the Cochrane highly sensitive search strategy for identifying RCTs (Higgins 2011). We modified this filter for use in EMBASE and used specific keywords to identify potential studies (see Appendix 1; Appendix 2; Appendix 3).
We searched for ongoing clinical trials and unpublished studies on the following Internet sites (on 19 October 2012).
-
ClinicalTrials.gov trials registry (see Appendix 4).
-
The World Health Organization's International Clinical Trials Registry Platform (see Appendix 5).
Searching other resources
We undertook backward and forward citation searching for key review articles identified through the initial searches (see Appendix 6). We used Web of Science for forward citation searching. We read the reference lists of articles selected for backward citation, paying particular attention to the articles included in systematic reviews.
We contacted investigators to ask for details of ongoing studies and any unpublished data needed for our analyses.
Data collection and analysis
Selection of studies
Two review authors (AB, AN) independently screened all titles and abstracts identified by the searches (to October 2012) for potentially eligible trials. A pilot screening of 100 articles was performed initially to clarify criteria for discarding articles at this stage. We removed studies that were very unlikely to be eligible. If no abstract was available but the title was possibly relevant, the full text of the article was obtained. Then the full texts of all remaining articles were independently examined by the same review authors. A joint decision was made at that time regarding inclusion, with disagreements resolved by a third review author (AFS).
Data extraction and management
AB (content area specialist) and AN (methodologist) independently extracted and collected data on a standardized paper form (see Appendix 7). No blinding of the author, the institution or the publication source of the studies was performed. If relevant information or data were not available in the paper, we contacted the lead author to request additional details. We resolved disagreements by discussion and consensus, and finally with the involvement of a third review author (AFS).
Multiple reports of the same study were extracted directly onto a single data collection form, thereby constructing a composite data set for that study.
Assessment of risk of bias in included studies
Two review authors (AB, AN) independently assessed risk of bias using the Cochrane 'Risk of bias' tool (Higgins 2011a). The following six domains were assessed: sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data and selective outcome reporting.
Blinding and incomplete outcome data were considered separately for each outcome. Blinding of participants was of particular importance for patient‐reported outcomes such as mobility. Blinding of assessors was particularly important for outcomes such as cognitive function that may be prone to detection bias.
Measures of treatment effect
For dichotomous outcomes (e.g. mortality, adverse outcomes), we entered numbers of events and total number within each randomization group into RevMan 5.1 (RevMan 5.1) and calculated risk ratios (RRs) with 95% confidence intervals (CIs) to express the effect size. If data were presented in other forms, such as hazard or odds ratios, and if we were unable to obtain the required tabular data from the study authors, we planned to enter these and to use the generic inverse variance option in RevMan 5.1. For continuous measures, such as length of stay, weighted mean differences were calculated if means and standard deviations were available. Standard deviations were calculated from 95% CIs using the methods described in Section 7.7.3.2 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).
Unit of analysis issues
For any cluster trials included, we extracted data directly only if the analysis properly accounted for the cluster design, using methods such as multi‐level modelling or generalized estimating equations. If these adjustments were not made within the report, we performed approximate analyses by recalculating standard errors or sample sizes based on the design effect (Section 16.3.6, Higgins 2011). The resulting effect estimates and their standard errors were analysed using the generic inverse variance method in RevMan 5.1 (RevMan 5.1).
In studies in which participants were randomly assigned to multiple intervention groups, each pair‐wise comparison was made separately but with shared intervention groups divided out approximately among the comparisons. For example, if multiple intervention groups shared a common control group, the number of participants and the number of events in the control group were divided equally; thus the number of subgroups in the control group matched the number of intervention groups (Higgins 2011).
Dealing with missing data
We attempted to contact the first author or the contact person for all trials with missing data before making a decision about trial eligibility. A modified intention‐to‐treat (ITT) analysis was undertaken, and all participants who did not withdraw consent for trial inclusion before the time of surgery were included. For missing outcome data, where possible, we compared the effects of complete case analysis, worst case scenario and last observation carried forward options on the results of any individual study and on any meta‐analysis undertaken.
Assessment of heterogeneity
The trials that were found may not have been carried out according to common protocols, thus introducing differences in participant groups, clinical settings, concomitant care, etc. Important potential sources of heterogeneity include participant characteristics, differences in control or intervention protocols and duration of perioperative fluid optimization.
Heterogeneity between studies was described on the basis of participant group, setting and type of intervention. This was then assessed statistically when data allowed, using the Chi2 test and the I2 statistic. Important heterogeneity (Chi2 P < 0.1 and I2 > 50%) was investigated, when possible, by subgroup analyses and by meta‐regression.
Assessment of reporting biases
Reporting bias may occur within studies, with certain outcomes not reported. When a report or the original protocol suggested that data on an outcome were collected but were not reported in the paper, we contacted the authors to request the data.
When an adequate number of trials had been identified for inclusion, funnel plots were constructed and were examined visually to assess the presence of publication bias; Egger’s test was used to test for asymmetry.
Data synthesis
We attempted meta‐analysis for outcomes for which we had comparable effect measures from more than one study and when measures of heterogeneity indicated that pooling of results was appropriate. A value of I2 > 80% would argue against presentation of an overall estimate. When we had identified sufficient studies to allow combination of results, differences between studies related to duration and methods of fluid optimization and participant characteristics were likely to suggest that random‐effects models would be the most suitable choice. Mantel‐Haenszel models were used when possible for dichotomous outcomes.
Subgroup analysis and investigation of heterogeneity
If data were sufficient, we investigated the following subgroups, which may account for heterogeneity between studies.
-
Duration of monitoring and protocol use.
-
Timing of outcome measurement. If the timing of outcome measures varied between studies, then the outcome was analysed only as a subgroup analysis.
Any differences in effect size between subgroups were assessed in RevMan, using I2 estimates (Higgins 2011).
Sensitivity analysis
When possible, we performed the following sensitivity analyses.
-
Reanalysis excluding studies with a high risk of bias.
-
Reanalysis excluding unpublished studies.
Summary of findings table
We used the principles of the GRADE system (Guyatt 2008) to assess the quality of the body of evidence associated with the specific outcomes in our review.
-
All‐cause mortality.
-
Length of hospital stay.
-
Return to pre‐fracture category of accommodation at six months.
-
Return to pre‐fracture mobility at six months.
-
Major adverse events in hospital.
We constructed a 'Summary of findings' (SoF) table by using the GRADE software (gradepro.org). The GRADE approach appraises the quality of a body of evidence based on the extent to which one can be confident that an estimate of effect or association reflects the item being assessed. The quality of a body of evidence is based on within‐study risk of bias (methodological quality), the directness of the evidence, the heterogeneity of the data, the precision of effect estimates and the risk of publication bias.
Results
Description of studies
Results of the search
See Characteristics of included studies; Characteristics of ongoing studies; Characteristics of studies awaiting classification.
Searches of electronic databases revealed 667 records. An additional seven records were identified from the references of potentially relevant articles (backwards citation), and 481 had cited important articles (forward citation). Searches of clinical trials databases identified two ongoing studies, one of which had two publications. A total of 1151 unique titles and/or abstracts were reviewed, and 41 publications met the criteria for further assessment. From these, we included three trials that randomly assigned a total of 200 participants (Schultz 1985; Sinclair 1997; Venn 2002). One study is awaiting classification (see Figure 1).
I Study flow diagram.
Included studies
None of the three included trials were published in abstract form only (see Characteristics of included studies). All trials included solely adult participants who were undergoing surgery for PFF. Two trials were conducted in the UK (Sinclair 1997; Venn 2002) and one in the USA (Schultz 1985). All were published in the English language. The interval between the first and the last trial was approximately 17 years. Ages of participants ranged from 40 to 102 years. Each trial made different comparisons: Swan‐Ganz monitoring versus CVP monitoring (Schultz 1985); oesophageal Doppler monitoring versus conventional fluid management (Sinclair 1997); and oesophageal Doppler monitoring versus CVP monitoring versus conventional fluid management (Venn 2002). These trial comparisons correspond to the following comparisons in our review: advanced haemodynamic monitoring (Swan‐Ganz, oesophageal Doppler); a protocol using standard measures (CVP monitoring); and usual care (conventional fluid management). Two trials studied only intraoperative fluid optimization (Sinclair 1997; Venn 2002); one trial studied preoperative, intraoperative and postoperative fluid optimization (Schultz 1985). All participants underwent general anaesthesia, although this was not explicitly stated in one trial (Schultz 1985). The surgical techniques used to treat PFF included dynamic hip screw, arthroplasty and AO cannulated screw. All trials investigated mortality, although at different time points: undefined "postoperative" (Schultz 1985) and in‐hospital (Sinclair 1997; Venn 2002). We excluded in‐hospital deaths that occurred more than 30 days postoperatively (Sinclair 1997). On the basis of total hospital stays and ranges reported in Venn 2002, we assumed that all deaths and adverse events in this trial occurred within 30 days of operation. Two trials compared both total length of hospital stay and time until medically fit for discharge (Sinclair 1997; Venn 2002). One reported as medians and interquartile ranges (Sinclair 1997), and the other as means with 95% confidence intervals (Venn 2002). We calculated standard deviations using the formula for the T distribution in Section 7.7.3.2 in Higgins 2011. Two trials investigated morbidity, although the time frame for Schultz was described only as postoperative (Schultz 1985; Venn 2002). In Venn 2002, postoperative adverse events were reported, but data were given as episodes of cardiopulmonary events rather than as numbers of participants, so we were not able to calculate risk ratios. Two trials compared changes in intraoperative physiological parameters (Sinclair 1997; Venn 2002). One trial compared both time from admission to surgery and length of operation (Schultz 1985).
Ongoing studies
Two studies are ongoing. The first is comparing routine perioperative fluid therapy and goal‐directed haemodynamic therapy in terms of morbidity, mortality, length of hospital stay, activity of daily living, health‐related quality of life, cognitive function and the need for social services until 12 months postoperatively after fixation of PFF in elderly participants (GDHT study). The second is comparing stroke volume－guided intraoperative fluid management using a calibrated cardiac output monitor (LiDCOplus) against routine fluid administration in terms of length of acute hospital stay, numbers of complications and total costs of care (NOTTS study; Characteristics of ongoing studies).
Studies awaiting classification
One study that included a mixed high‐risk surgical population is awaiting classification; we have been unable to contact the authors to request adequate data about participants within the orthopaedic group who were treated for PFF (Sandham 2003; Characteristics of studies awaiting classification).
Excluded studies
We excluded 34 full text articles identified for further assessment (Figure 1). These articles provided the wrong intervention, included the wrong study population or were not RCTs. Specific groups of excluded trials were those investigating blood product transfusion strategies, vasopressor therapies and bundles of perioperative care, including nursing care, rehabilitation and nutritional strategies. Eight RCTs that were excluded because of incorrect intervention or participant group are described in the Characteristics of excluded studies table.
Risk of bias in included studies
The various bias domains are presented in the ’Risk of bias’ graph and a ’Risk of bias’ summary figure. The risk of bias was evaluated on the basis of major sources of bias (domains), as described above. For a more detailed description of individual trial qualities, see Characteristics of included studies (see Figure 2, Figure 3).
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Allocation
The methods used for random sequence generation were adequately reported in only one study (Venn 2002) and were inadequately reported or questionable in two studies (Schultz 1985; Sinclair 1997). Allocation concealment was not adequately reported in any study.
Blinding
Defining adequate blinding in trials of fluid optimization was challenging. Two trials were pragmatic in their attempts to blind the primary anaesthetist in theatre to fluid administered (performance bias), as it would not be practical in our view to safely blind the attending clinician to this (Sinclair 1997; Venn 2002); no information was provided about blinding of the operating surgeon(s) in any trial. Two trials provided enough information to allow assessment of blinding of outcome assessment (detection bias) as adequate (Sinclair 1997; Venn 2002). One trial provided no information about blinding of participants, attending clinicians or investigators (Schultz 1985) (see Characteristics of included studies).
Incomplete outcome data
Complete follow‐up was reported for mortality, morbidity, adverse events and length of stay for two trials (Sinclair 1997; Venn 2002). One trial provided no information about exclusions due to deviations from protocol (Schultz 1985). No participant was lost to follow‐up. Only one trial explicitly performed analyses in accordance with the ITT method (Venn 2002) (see Characteristics of included studies).
Selective reporting
Two trials reported all expected outcomes (Sinclair 1997; Venn 2002). One trial provided inadequate information about expected outcomes; therefore we assessed the risk of selective reporting as unclear (Schultz 1985) (see Characteristics of included studies).
Some of our analyses were subject to limitations because length of stay data were published in graphical form without adequate corresponding numerical data (Sinclair 1997). One trial did not provide details on length of follow‐up in terms of mortality, but data were sufficient for analysis; data regarding morbidity and adverse events were adequate (Schultz 1985).
Other potential sources of bias
One trial received support from the Special Trustees of the Middlesex Hospital, defined as not‐for‐profit (Sinclair 1997). Funding sources for the remaining trials were defined as unknown. Sample size calculation was reported in two trials (Sinclair 1997; Venn 2002). No trial was stopped early as the result of benefits or difficulty in recruiting participants. Trials were too few to permit construction of funnel plots to facilitate assessment of publication bias, or to perform Egger's test for asymmetry.
Analyses of the benefits of fluid optimization in this group of participants were limited by differences in study design. Two trials involved intraoperative optimization (Sinclair 1997; Venn 2002). One involved both intraoperative and postoperative optimization and was seriously limited by the fact that no detail was given about the protocol used (Schultz 1985). Between trials, differences were noted in outcome definitions, in time points for mortality and length of stay reporting and in types of adverse events reported. In addition, all trials involved relatively low numbers of participants (see Characteristics of included studies).
Effects of interventions
See: Summary of findings for the main comparison Advanced haemodynamic monitoring compared with protocol using standard measures for perioperative fluid volume optimisation; Summary of findings 2 Advanced haemodynamic monitoring compared with usual care for perioperative fluid optimization; Summary of findings 3 Protocol using standard measures such as CVP compared with usual care for perioperative fluid optimization
See also summary of findings Table for the main comparison; summary of findings Table 2; summary of findings Table 3.
Comparison 1. Advanced haemodynamic monitoring vs protocol using standard measures
All‐cause mortality
Two trials reported mortality (Schultz 1985; Venn 2002). For one study, the follow‐up period was unclear but was reported as "postoperative" (Schultz 1985), so we were unable to pool results. This trial showed a significant reduction in mortality (RR 0.1, 95% CI 0.01 to 0.74; 70 participants, one trial); however, we had serious concerns about its quality (Schultz 1985). In the other study (Venn 2002), the time frame for death was described as postoperative, and the results were consistent with both increased and decreased risk of mortality in the intervention group (RR 0.52, 95% CI 0.14 to 1.88; 61 participants, one trial). (See Analysis 1.1.)
Length of hospital stay
Only one trial reported this outcome, and it found that the mean difference for hospital stay was 0.2 days longer in the advanced haemodynamic group (95% CI 5.1 days shorter to 5.5 days longer; 61 participants, one trial) and for time to medical fitness was 2.30 days shorter (95% CI 5.90 days shorter to 1.30 days longer; 61 participants, one trial) (Venn 2002; Table 1).
| Outcomes reported in Venn 2002: comparison 1 | Advanced haemodynamic－Doppler N = 30 | Protocol－ CVP N = 31 | Effect estimate (95% CI) | ||
|
| Mean | SD | Mean | SD | Mean difference |
| Length of hospital stay (days) | 13.5 | 8.8 | 13.3 | 12.1 | 0.20 (‐5.10 to 5.50) |
| Time to fitness to discharge | 7.7 | 8.6 | 10 | 5.3 | ‐2.30 (‐5.90 to 1.30) |
|
|
|
|
|
|
|
|
| Events |
| Events |
| MH relative risk |
| Mortality | 3 |
| 6 |
| 0.52 (0.14 to 1.88) |
| Adverse events |
|
|
|
|
|
| · Cardiopulmonary－episodes | 6 |
| 6 |
| N/A |
| · Neurological－participants | 2 |
| 1 |
| 2.07 (0.20 to 21.61) |
| · Any, including minor－participants | 7 |
| 8 |
| 0.90 (0.37 to 2.18) |
Return of participant to pre‐fracture category of accommodation at six months; return to pre‐fracture mobility at six months
No trial reported data for these outcomes.
Major adverse events in hospital
Two trials investigated complications, reporting overall morbidity and cardiovascular or neurological outcomes; however, iatrogenic events were not reported by intervention/control groups (Schultz 1985; Venn 2002). Once again, it was not possible to pool data because of the unclear period of follow‐up in Schultz 1985. Both studies reported results consistent with increased and decreased risks of adverse events in the intervention groups. In Venn 2002, the relative risk for neurological events was 2.07 (95% CI 0.20 to 21.61) and for all complications 0.90 (95% CI 0.37 to 2.18) (Table 2). (See Analysis 1.2.)
| Outcomes reported in Venn 2002: comparison 2 | Advanced haemodynamic－Doppler N = 30 | Standard care
N = 29 | Effect estimate (95% CI) | ||
|
| Mean | SD | Mean | SD | Mean difference |
| Length of hospital stay (days) | 13.5 | 8.8 | 17.5 | 13.8 | ‐4.00 (‐9.93 to 1.93) |
| Time to fitness to discharge | 7.7 | 8.6 | 13.9 | 6.6 | ‐6.20 (‐10.10 to ‐2.30) |
|
|
|
|
|
|
|
|
| Events |
| Events |
| MH relative risk |
| Mortality | 3 |
| 2 |
| 1.45 (0.26 to 8.06) |
| Adverse events |
|
|
|
|
|
| · Cardiopulmonary－episodes | 6 |
| 7 |
| N/A |
| · Neurological－participants | 2 |
| 1 |
| 1.93 (0.19 to 20.18) |
| · Any, including minor－participants | 7 |
| 14 |
| 0.48 (0.23 to 1.02) |
Comparison 2. Advanced haemodynamic monitoring vs usual care
All‐cause mortality
Two trials reported in‐hospital mortality (Sinclair 1997; Venn 2002). We excluded two deaths from Sinclair 1997, one each from the intervention and control groups, as they occurred more than 30 days postoperatively. Only three deaths were reported in each group, and the pooled results are consistent with both increased and decreased risks of mortality in participants who received advanced haemodynamic monitoring (RR 1.03, 95% CI 0.23 to 4.66; 99 participants, two trials). (See Analysis 2.1.)
Length of hospital stay
Two trials investigated both total hospital stay and time to medical fitness for discharge (Sinclair 1997; Venn 2002). One trial found a reduction in time to medical fitness for discharge (6.20 days shorter, 95% CI 10.1 to 2.30 days shorter; 59 participants, one trial), but not for total inpatient stay (4.00 days shorter, 95% CI 9.93 days shorter to 1.93 days longer; 59 participants, one trial), in the advanced haemodynamic group (Venn 2002; Table 2). The other trial provided data in the form of median and interquartile ranges, which were not suitable for inclusion in a meta‐analysis (Sinclair 1997; Table 3), but reported a reduction of five days in median time to fitness for discharge (from 15 to 10 days) and a reduction of eight days in total hospital stay (from 20 to 12); the authors reported significant differences at P < 0.05 (Mann Whitney U test). Hence no meta‐analysis was carried out, and only a narrative summary is offered for this outcome.
| Time to medical fitness for discharge (days) | ||
| Control group (18 participants) | Advanced haemodynamic monitoring group (19 participants) | |
| Extremes | 6 to 125 | 4 to 26 |
| Quartiles | 10 to 32 | 8 to 10 |
| Median | 15 | 10 |
| Total hospital stay | ||
| Control group (18 participants) | Advanced haemodynamic monitoring group (19 participants) | |
| Extremes | 5 to 220 | 4 to 24 |
| Quartiles | 10 to 33 | 8 to 15 |
| Median | 20 | 12 |
Values visually estimated by box‐and‐whisker plots in published trial.
Return of participant to pre‐fracture category of accommodation at six months; return to pre‐fracture mobility at six months
No trial reported data for these outcomes.
Major adverse events in hospital
These were reported by only one trial, and results were consistent with increased and decreased risk in participants who had received advanced haemodynamics monitoring for neurological events (RR 1.93, 95% CI 0.19 to 20.18; 59 participants, one trial) and for all complications (RR 0.48, 95% CI 0.23 to 1.02; 59 participants, one trial) (Venn 2002; Table 2).
Comparison 3. A protocol using standard measures vs usual care
All‐cause mortality
Only one trial reported on this outcome (Venn 2002) and found no difference in mortality between participants who received care according to the protocol and standard care (RR 2.81, 95% CI 0.61 to 12.81; 60 participants, one trial) (Table 4).
| Outcomes reported in Venn 2002: comparison 3 | Protocol－ CVP N = 31 | Standard care
N = 29 | Effect estimate (95% CI) | ||
|
| Mean | SD | Mean | SD | Mean difference |
| Length of hospital stay (days) | 13.3 | 12.1 | 17.5 | 13.8 | ‐4.20 (‐11.0 to 2.60) |
| Time to fitness to discharge | 10 | 5.3 | 13.9 | 6.6 | ‐3.90 (‐7.05 to ‐0.75) |
|
|
|
|
|
|
|
|
| Events |
| Events |
| MH relative risk |
| Mortality | 6 |
| 2 |
| 2.81 (0.61 to 12.81) |
| Adverse events |
|
|
|
|
|
| · Cardiopulmonary－episodes | 6 |
| 7 |
| N/A |
| · Neurological－participants | 1 |
| 1 |
| 0.94 (0.06 to 14.27) |
| · Any, including minor－participants | 8 |
| 14 |
| 0.53 (0.26 to 1.08) |
Length of hospital stay
One trial reported a reduction in time to medical fitness (3.9 days shorter, 95% CI 7.05 to 0.75 days shorter; 60 participants, one trial) but not in total hospital stay (4.2 days shorter, 95% CI 11.0 days shorter to 2.60 days longer; 60 participants, one trial) (Venn 2002; Table 4).
Return of participant to pre‐fracture category of accommodation at six months; return to pre‐fracture mobility at six months
No trial reported data for these outcomes.
Major adverse events in hospital
These were reported by only one trial, and results were consistent with increased and decreased risk in participants who had received care according to a protocol for neurological events (RR 0.94, 95% CI 0.06 to 14.27; 60 participants, one trial) and for all complications (RR 0.53, 95% CI 0.26 to 1.08; 60 participants, one trial) (Venn 2002; Table 4).
Subgroup and sensitivity analyses
We were particularly concerned about the quality of one study in terms of the risk of bias (Schultz 1985). No details were given about methods of randomization, and important baseline differences between intervention and control groups were noted. In addition, the nature of intervention was not fully reported, and staff were not blinded to the intervention group (see Characteristics of included studies). Because only two studies were included in the comparison of advanced haemodynamic monitoring, we were not able to perform sensitivity analyses; however, we have included results from Venn 2002 in the 'Summary of findings' tables.
We obtained no unpublished data, and so it was not possible to carry out this subgroup analysis.
Discussion
Summary of main results
The conclusions of this updated review remain the same as those of the original review (Price 2004). We did not find a benefit for the use of fluid optimization strategies in participants undergoing surgery for PFF in terms of mortality or adverse events. We did find a possible benefit in terms of length of hospital stay; however, only limited data are available. Furthermore, we were unable to conduct relevant subgroup and sensitivity analyses because of lack of data. Currently, no convincing evidence of safety or effectiveness is available to support the routine use of advanced monitoring or protocols to guide fluid therapy in adult patients undergoing surgery for PFF. Length of hospital stay may be reduced, but the evidence is not strong enough to allow evidence‐based recommendations to be made regarding fluid optimization in this patient group (see also summary of findings Table for the main comparison; summary of findings Table 2; summary of findings Table 3).
Overall completeness and applicability of evidence
Since the time of the original review, the method of reporting Cochrane systematic reviews has changed; therefore we re‐ran our search strategy from the inception of the databases, rather than using the date of the previous search. Despite this, we found no new published studies to include, and it is clear that good quality clinically relevant evidence on this subject is insufficient. We did identify at least two ongoing studies, and it is hoped that they will provide more data in the near future (see Characteristics of ongoing studies).
Among the three studies and 200 participants analysed, outcome data were variable in terms of quality and definition. Each trial reported mortality data, but they were defined by different time points or were not defined at all. Only one study reported an a priori power calculation for mortality (Sinclair 1997). It is arguable that any mortality reduction due to the interventions in our review would be small because of the many other factors that put PFF patients at relatively high risk of death. If an in‐hospital mortality of 6.6% is assumed (Moppett 2012), a study with 80% power to detect a 50% decrease in in‐hospital mortality (from 6.6% to 3.3%) would require randomization of 678 participants into each group (α = 0.05). Therefore much larger studies than the ones presented in this review are likely to be needed to show benefit derived from these interventions. Similarly, to detect a 50% reduction in adverse event incidence (from 15% to 7.5%) (Lawrence 2002; Roche 2005), 278 participants would be required for each group. On the other hand, the two studies investigating length of stay were adequately powered (Sinclair 1997; Venn 2002).
Other outcomes were investigated only by one or two trials, or not at all, making it difficult for investigators to draw conclusions. It could be contested that we should have looked at outcomes of greater "orthopaedic relevance", but these tend to be reported over the longer term, and it seems logical to assume that intraoperative fluid optimization is more likely to affect shorter‐term in‐hospital outcomes. Whilst these may influence longer‐term outcomes, it would be more difficult for studies to gain evidence of such effects over longer time periods in the presence of other confounding factors. This may be why studies have not investigated time to return to pre‐fracture mobility/accommodation.
Caution should be exercised in the applicability of our results in countries that are less well developed. Furthermore, it should be appreciated that usual care in some countries, or even between clinicians in the same hospital, may differ.
Quality of the evidence
Our review has several limitations, and our findings are limited by the quality and quantity of available evidence. All trials recruited participants from similar populations in well‐developed countries, but detail was not uniformly provided about the interventions that we investigated, particularly the protocol used in the trial by Schultz et al (Schultz 1985). As has been mentioned, the quality of outcome reporting was variable. Mortality data were reported but to different time points, making combination of data from different studies impossible for a single comparison (Analysis 1.1). The two trials reporting significant differences in time to medical fitness for discharge include relatively low numbers (130 participants) (Sinclair 1997; Venn 2002), and in one trial, length of stay data were estimated from graphical data, which we were unable to incorporate into a meta‐analysis (Sinclair 1997). No trials reported on two of our primary outcome measures: time to return to pre‐fracture mobility and accomodation. The secondary outcome of adverse events was reported only well enough to allow analysis for the neurological event subgroup and total adverse events, and again was limited to only one or two trials, depending on the comparison. The data were not reported well enough to allow analysis of subgroups of iatrogenic and cardiopulmonary events. We were unable to contact the authors of included trials to ask for unpublished outcome data. Trials were too few to permit subgroup, heterogeneity or sensitivity analyses to be performed.
Lack of available information therefore significantly limits this systematic review. Broadening the scope of the review to include a greater number of clinical groups would increase the data set but would be clinically less useful to the reader interested in the specific management of PFF. It would also further increase heterogeneity. It is hoped that this can be avoided in the future when ongoing and future studies are published (see Characteristics of ongoing studies), or when additional data from existing studies become available (see Characteristics of studies awaiting classification).
Potential biases in the review process
To the best of our knowledge, no potential biases arose from the review process.
Agreements and disagreements with other studies or reviews
The authors of the original version of this review stated that invasive methods of fluid optimization may shorten hospital stay, but their effects on other important, patient‐centred, longer‐term outcomes are uncertain. We would agree in general with this but urge caution in the interpretation of hospital stay data that are limited in scale and in some cases are not adequate for detailed analysis. We are not aware of any other good quality studies or systematic reviews investigating perioperative fluid optimization after PFF.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Comparison 1 Advanced haemodynamic monitoring vs protocol using standard measures, Outcome 1 All‐cause mortality.
Comparison 1 Advanced haemodynamic monitoring vs protocol using standard measures, Outcome 2 Adverse outcomes.
Comparison 2 Advanced haemodynamic monitoring vs usual care, Outcome 1 All‐cause mortality.
| Advanced haemodynamic monitoring compared with protocol using standard measures for perioperative fluid volume optimisation | ||||||
| Patient or population: patients with proximal femoral fracture | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No. of participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Protocol using standard measures such as CVP | Advanced haemodynamic monitoring | |||||
| All‐cause mortality | Moderatea | RR 0.52 | 61 | ⊕⊕⊝⊝ | ||
| 66 per 1000 | 34 per 1000 | |||||
| Total length of hospital stay | The mean total length of hospital stay in the control groups was | The mean total length of hospital stay in the intervention groups was | 61 | ⊕⊕⊝⊝ | ||
| Time to medical fitness for discharge | The mean time to medical fitness for discharge in the control groups was | The mean time to medical fitness for discharge in the intervention groups was | 61 | ⊕⊕⊝⊝ | ||
| Adverse outcomes Cardiopulmonary－not reported | See comment | See comment | Not estimable | ‐ | See comment | No data suitable for analysis available |
| Adverse outcomes－neurological | Moderated | RR 2.07 | 61 | ⊕⊕⊝⊝ | ||
| 10 per 1000 | 21 per 1000 | |||||
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence: | ||||||
| aBased on mortality rate from Moppett 2012. | ||||||
| Advanced haemodynamic monitoring compared with usual care for perioperative fluid optimization | ||||||
| Patient or population: patients with proximal femoral fracture | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No. of participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Usual care | Advanced haemodynamic monitoring | |||||
| All‐cause mortality | Moderatea | RR 1.03 | 99 | ⊕⊕⊝⊝ | ||
| 66 per 1000 | 68 per 1000 | |||||
| Total length of hospital stay | The mean total length of hospital stay in the control groups was | The mean total length of hospital stay in the intervention groups was | 59 | ⊕⊕⊝⊝ | ||
| Time to medical fitness for discharge | The mean time to medical fitness for discharge in the control groups was | The mean time to medical fitness for discharge in the intervention groups was | 59 | ⊕⊕⊕⊝ | ||
| Adverse outcomes Cardiopulmonary－not reported | See comment | See comment | Not estimable | ‐ | See comment | No data suitable for analysis available |
| Adverse outcomes－neurological | Moderated | RR 1.93 | 59 | ⊕⊕⊝⊝ | ||
| 10 per 1000 | 19 per 1000 | |||||
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence: | ||||||
| aBased on mortality rate from Moppett 2012. | ||||||
| Protocol using standard measures such as CVP compared with usual care for perioperative fluid optimization | ||||||
| Patient or population: patients with proximal femoral fracture | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No. of participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Usual care | Protocol using standard measures such as CVP | |||||
| All‐cause mortality | Moderatea | RR 2.81 | 60 | ⊕⊕⊝⊝ | ||
| 66 per 1000 | 185 per 1000 | |||||
| Total length of hospital stay | The mean total length of hospital stay in the control groups was | The mean total length of hospital stay in the intervention groups was | 57 | ⊕⊕⊝⊝ | ||
| Time to medical fitness for discharge | The mean time to medical fitness for discharge in the control groups was | The mean time to medical fitness for discharge in the intervention groups was | 57 | ⊕⊕⊕⊝ | ||
| Adverse outcomes Cardiopulmonary－not reported | See comment | See comment | Not estimable | ‐ | See comment | Data suitable for analysis not available |
| Adverse outcomes－neurological | Moderated | RR 0.94 | 60 | ⊕⊕⊝⊝ | ||
| 10 per 1000 | 9 per 1000 | |||||
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence: | ||||||
| aBased on mortality rate from Moppett 2012. | ||||||
| Outcomes reported in Venn 2002: comparison 1 | Advanced haemodynamic－Doppler N = 30 | Protocol－ CVP N = 31 | Effect estimate (95% CI) | ||
|
| Mean | SD | Mean | SD | Mean difference |
| Length of hospital stay (days) | 13.5 | 8.8 | 13.3 | 12.1 | 0.20 (‐5.10 to 5.50) |
| Time to fitness to discharge | 7.7 | 8.6 | 10 | 5.3 | ‐2.30 (‐5.90 to 1.30) |
|
|
|
|
|
|
|
|
| Events |
| Events |
| MH relative risk |
| Mortality | 3 |
| 6 |
| 0.52 (0.14 to 1.88) |
| Adverse events |
|
|
|
|
|
| · Cardiopulmonary－episodes | 6 |
| 6 |
| N/A |
| · Neurological－participants | 2 |
| 1 |
| 2.07 (0.20 to 21.61) |
| · Any, including minor－participants | 7 |
| 8 |
| 0.90 (0.37 to 2.18) |
| Outcomes reported in Venn 2002: comparison 2 | Advanced haemodynamic－Doppler N = 30 | Standard care
N = 29 | Effect estimate (95% CI) | ||
|
| Mean | SD | Mean | SD | Mean difference |
| Length of hospital stay (days) | 13.5 | 8.8 | 17.5 | 13.8 | ‐4.00 (‐9.93 to 1.93) |
| Time to fitness to discharge | 7.7 | 8.6 | 13.9 | 6.6 | ‐6.20 (‐10.10 to ‐2.30) |
|
|
|
|
|
|
|
|
| Events |
| Events |
| MH relative risk |
| Mortality | 3 |
| 2 |
| 1.45 (0.26 to 8.06) |
| Adverse events |
|
|
|
|
|
| · Cardiopulmonary－episodes | 6 |
| 7 |
| N/A |
| · Neurological－participants | 2 |
| 1 |
| 1.93 (0.19 to 20.18) |
| · Any, including minor－participants | 7 |
| 14 |
| 0.48 (0.23 to 1.02) |
| Time to medical fitness for discharge (days) | ||
| Control group (18 participants) | Advanced haemodynamic monitoring group (19 participants) | |
| Extremes | 6 to 125 | 4 to 26 |
| Quartiles | 10 to 32 | 8 to 10 |
| Median | 15 | 10 |
| Total hospital stay | ||
| Control group (18 participants) | Advanced haemodynamic monitoring group (19 participants) | |
| Extremes | 5 to 220 | 4 to 24 |
| Quartiles | 10 to 33 | 8 to 15 |
| Median | 20 | 12 |
| Values visually estimated by box‐and‐whisker plots in published trial. | ||
| Outcomes reported in Venn 2002: comparison 3 | Protocol－ CVP N = 31 | Standard care
N = 29 | Effect estimate (95% CI) | ||
|
| Mean | SD | Mean | SD | Mean difference |
| Length of hospital stay (days) | 13.3 | 12.1 | 17.5 | 13.8 | ‐4.20 (‐11.0 to 2.60) |
| Time to fitness to discharge | 10 | 5.3 | 13.9 | 6.6 | ‐3.90 (‐7.05 to ‐0.75) |
|
|
|
|
|
|
|
|
| Events |
| Events |
| MH relative risk |
| Mortality | 6 |
| 2 |
| 2.81 (0.61 to 12.81) |
| Adverse events |
|
|
|
|
|
| · Cardiopulmonary－episodes | 6 |
| 7 |
| N/A |
| · Neurological－participants | 1 |
| 1 |
| 0.94 (0.06 to 14.27) |
| · Any, including minor－participants | 8 |
| 14 |
| 0.53 (0.26 to 1.08) |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 All‐cause mortality Show forest plot | 2 | Risk Ratio (M‐H, Random, 95% CI) | Totals not selected | |
| 2 Adverse outcomes Show forest plot | 2 | Risk Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 2.1 Cardiopulmonary | 1 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 2.2 Neurological | 2 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 2.3 Any complications, including minor | 2 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 All‐cause mortality Show forest plot | 2 | 99 | Risk Ratio (M‐H, Random, 95% CI) | 1.03 [0.23, 4.66] |
