This website was developed as a place to reach international consensus amongst clinicians; academics/researchers; and members of the interdisciplinary team involved in ICU patient management (nurses; intensivists) on a best practice physiotherapy clinical algorithm.

NOTE: The author had access to the clinical guidelines developed by the European respiratory Society of Intensive Care Medicine task force which was recently published (Gosselink et al 2008). These recommendations were incorporated into the algorithm where applicable.

	Delphi Study Results
The way forward These results will be collated into management algorithms using the following key words based on the consensus median rating DO Rating 1 SHOULD DO Rating 2 CONSIDER Rating 3 UNIMPORTANT Rating 4 DETRIMENTAL Rating 5 The evidence based protocol will be implemented over 4 three week periods in a surgical ICU starting 1 November 2008. Patient outcomes from protocol care will be compared to usual care. The abdominal surgery; rehabilitation and pulmonary dysfunction groups are currently working on publications for submission this year.

Site content:

• Why is it necessary to have an internationally validated best practice physiotherapy clinical algorithm?
• Process of development
• Structure of the site
• The electronic Delphi research process
• How will this internationally validated best practice physiotherapy clinical algorithm (BPPCA) be evaluated?
• Which outcomes will be measured?
• References

Process of development

Seven clinical algorithms for the management of patients admitted to a surgical ICU were developed based on the following process:

Step 1
A conceptual frame work was developed by the researcher based on own clinical experience; interdisciplinary discussions and extensive reading.

Step 2
Based on this conceptual framework suitable systematic review questions (PICO format) were formulated. A protocol for each of the questions were developed which included:

• A specific search strategy including databases; search words; inclusion and exclusion criteria.
• Identified studies were screened independently by two researchers.
• Consensus was reached in case of dispute
• The relevant data was extracted to spreadsheets
• The quality of the studies were initially assessed with the following three scales:
• PEDro for all experimental studies http:/ptwww.cchs.usyd.edu.au/pedro/scaleitems.htm
• AMSTAR for systematic reviews (Shea et al 2007)
• AGREE for clinical guidelines (Agree instrument 2001)

 
Although, these scales are valuable in assessing the internal validity of an experimental study (PEDro); the methodological quality of systematic reviews (Shea et al 2007) or guidelines (Agree collaboration 2001), a general problem in interpreting the total scores is that  all criteria do not provide equally important information about the validity of the study findings. Therefore, specific criteria were used in the formulation of the algorithms (refer to step 5 for the criteria used).

Step 4
Recommendations were formulated based on the Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) system as adopted by the American Thoracic Society (ATS) (Schu¨nemann et al 2005). This two tier system incorporates both the quality of the evidence and an assessment of the balance between the benefits versus the downside of the intervention. The judgment in terms of the strength of the recommendation is placed first followed by the quality of the evidence.

The strength of the recommendation was either classified as strong or weak. When the researcher was confident that the benefits (outcome of interest; cost; burden of implementation) were greater than the undesirable effects (harms; costs and burden) a recommendation was qualified as strong. When benefits were measured but the researcher was not convinced that adherence to the recommendation would in all instances result in desirable effects outweighing the undesirable effects (Schu¨nemann et al 2005), the recommendation was qualified as weak.

The quality of the evidence was described based on the basic study structure and the precision of execution documented in the study. In the GRADE system all relevant clinical studies or observations provides evidence, the quality of which varies. The quality of evidence reflects the confidence that can be placed on the benefits and potential harm resulting from an intervention. The quality was described as high; moderate; low or very low (Schu¨nemann et al 2005). Table 1 reflects the initial quality rating assigned to a specific study structure and the factors considered in the downgrading or upgrading of the quality rating (Schu¨nemann et al 2006)

Table 1: GRADE quality assessment criteria

* 1 = move up or down one grade (for example from high to moderate) 2 = move up or down two grades (for example from high to low)

The quality of the evidence was downgraded in the presence of methodological issues; indirectness of the evidence; heterogeneity of studies that could not be explained, and imprecision of the study (Schu¨nemann et al 2006). These indicators were used as follows to determine the quality of the evidence.

The methodological issues that were considered in the evaluation of the quality of the evidence in RCT’s were:

• Randomisation (sequence generation) and concealed allocation
• Intention to treat analysis of the data.
• Patients lost to follow up
• It is not always possible to blind physiotherapists administering the care or patients in physiotherapy research. Therefore if blinding was not applied because it was impossible, the study was not downgraded due to this limitation.

In cross-over RCT designs, the methodological issues  that were considered in the evaluation of the quality of the evidence were (Lathyris et al 2007):

• Suitability of design for the question asked based on the stability of the patient’s condition and the outcome investigated
• Possible carry over effect/fatigue/learning. The evaluation of the sufficiency of the washout period will be assessed based on baseline values.
• Random sampling method with sufficient regard for possible confounders (time on ventilater; APACHE etc)

The directness of the evidence refers to how similar the study sample is to the sample in question and the precision of the documented intervention. If the study population differed from a surgical ICU population the quality of the evidence was downgraded. If there were more than one study and the interventions were not the same for example dosage and application it also affected the quality of the evidence.

Unexplained heterogeneity of results. When there were two or more studies presenting with different results and no plausible reasons for the observed variation in the results could be identified, the quality rating was downgraded.

The precision refers to sample size and effect size.  If a study was insufficiently powered to detect an effect or if the confidence intervals were sufficiently wide that the confidence in the effect is affected the quality rating was downgraded.

In observational studies quality ratings were upgraded when (Schu¨nemann et al 2006):

• a statistically significant relative risk of >2 (< 0.5) was reported, based on consistent evidence from two or more observational studies, with no plausible confounders
• A statistically significant relative risk of > 5 (< 0.2) based on direct evidence with no major threats to validity was reported
• There was evidence of a dose response gradient.

Step 5
The ultimate aim of the study is to evaluate the effect of an internationally validated physiotherapy management protocol based on the best available evidence on patient outcome compared to usual care. For the purpose of the formulation of the management algorithms used in this project the following criteria were used:

• If a recommendation could be formulated based on the primary or secondary research, even if the quality of the available evidence was low, these recommendations were collated into the clinical algorithms. (Do or might do depending on the strength of the recommendation)
• The recommendations formulated in clinical guidelines scoring at least 70% in the rigor of development domain on the AGREE instrument, were collated into the algorithms. (Might do)
• When no objective data could be identified expert opinion was considered (Oxman et al 2007) and collated into the algorithms. Opinion papers, reviews, surveys, and clinical guidelines that did not reach the minimum score on the AGREE instrument were considered as expert opinion. (Use as basis for interdisciplinary discussion before implementation)

In total seven clinical algorithms were formulated, which included one overall algorithm and six independent management algorithms.