Guideline for the management of knee and hip osteoarthritis

Chapter 2

Method

The (PICO) format

The key questions to be answered in the guideline were determined using the patient/population/problem, intervention, comparison/control, outcome (PICO) format. PICO questions are provided in Appendix 2. PICO.

Literature searches

The initial systematic literature search was designed to build upon the literature in the first edition, and to update the evidence published after the last search date for those guidelines.8 To accomplish this, we searched PubMed, the Cumulative Index to Nursing and Allied Health Literature (CINAHL), and the Cochrane Library (including Cochrane Database of Systematic Reviews and Cochrane Central Register of Controlled Trials [CENTRAL]) for studies published from January 2005 to December 2016. (For detailed information on search strategies, refer to Appendix 1 of the Guideline for the management of knee and hip osteoarthritis: Technical document.)

A medical librarian developed and conducted searches using search terms determined by the working group’s preliminary recommendations, database-specific medical subject headings, free-text terms, and study type filters were applied where appropriate. Studies of adults with hip and/or knee OA that involved one or more therapeutic interventions of interest were sought. Searches were limited to systematic reviews and randomised controlled trials (RCTs), which are classified as Level 1 and Level 2 evidence according to the National Health and Medical Research Council (NHMRC) hierarchy of evidence (Table 1).63


Adapted with permission from National Health and Medical Research Council. Additional levels of evidence and grades for recommendations for developers of guidelines. Canberra: NHMRC, 2009.

No limits were placed on language, and studies published in languages other than English were translated whenever possible. Details of the search strategies are available in Appendix 1 of the Guideline for the management of knee and hip osteoarthritis: Technical document. Electronic searches were supplemented with manual searches of reference lists of recent systematic reviews to ensure all pertinent resources were obtained. Searches were also performed within published supplements of relevant conference proceedings up to and including August 2017. Working group members were consulted regarding the evidence procured for each topic and, based on their expert knowledge of prior and emerging research in the field, reference to any additional resources that were not collected were requested. All electronic searches were updated in August 2017.

The initial searches for most interventions were then expanded to identify studies published prior to 2005. This was done to accommodate the transition of the current guideline to the Grading of Recommendations Assessment, Development and Evaluation (GRADE) methodology, thus to provide a comprehensive assessment of the quality of the entire body of evidence available for a given intervention. Consequently, the working group reviewed a GRADE summary table, which comprised all available RCT evidence regarding a given intervention to date, in order to make the most informed voting decision.

Study selection and PICO question matching

Systematic reviews and RCTs of adults with hip and/or knee OA in which the majority of the enrolled study population (≥80%) was aged 45 years or older were included. Only studies reporting participant health outcomes that were determined to be of interest by the working group’s recommendations were considered eligible for inclusion. Detailed patient health outcomes of interest, and inclusion and exclusion criteria are presented in Appendix 2 of the Guideline for the management of knee and hip osteoarthritis: Technical document.

The results of the literature searches were uploaded onto the Centre for Treatment Comparison and Integrative Analysis (CTCIA) web-based screening platform, which was used for primary and secondary literature screenings. Primary literature screening involved reviewing each record’s title and abstract for eligibility. Primary screening of each record was performed in duplicate by two independent reviewers (any paired combination among the pool of investigators from the working group or named in the Guideline for the management of knee and hip osteoarthritis: Technical document – Xia Wang, Mikala Osani, Elizaveta Vaysbrot and Mia-Cara Musetti), with conflicts resolved by a third reviewer (Raveendhara Bannuru).

Secondary literature screening involved the thorough review of fulltext articles. This was performed on all publications considered potentially eligible during the primary screening. Secondary literature screening followed the same independent duplicate review procedure, with conflict resolution undertaken by the same third reviewer. During secondary screening, all included articles were tagged with PICOrelated terms (eg intervention type(s), reported outcome(s)), to facilitate more efficient matching of the literature with PICO questions. Upon completion of secondary screening, the screening input for the references were exported on a database, and references were sorted in sequence by ‘Study design’, then by ‘Intervention’, ‘Comparator’, and finally ‘Date’.

Preliminary PICO designations were assigned to the references within the sorted document based on their ‘Interventions’ and ‘Comparators’; these designations were verified by a manual review of the included publications. Prior to the initiation of data extraction, the included articles were summarised by order of their matched PICO questions for the members of the working group, who assisted with reconciling possible mismatches or omissions. The study flow diagram in Figure 1, Appendix 2 of the Guideline for the management of knee and hip osteoarthritis: Technical document illustrates in detail the numbers of abstracts identified, full-text manuscripts retrieved, and studies selected for inclusion in the systematic literature review for these guidelines.

Data from eligible studies for each PICO question were extracted into the RevMan software. Risk of bias of the individual studies was assessed using the Cochrane Risk of Bias tool.64 Data extraction and risk of bias ratings were reviewed for consistency, and any discrepancies were resolved by consensus within the working group. Data were extracted on:

  • study and population characteristics
  • intervention dosage and frequency of administration
  • concomitant medications
  • all critical and important efficacy outcomes
  • all critical and important safety outcomes.

Random effects meta-analyses were conducted in anticipation of some heterogeneity among the studies.

Dichotomous outcomes were analysed using the Mantel–Haenszel method, and were reported as risk ratios (relative risk; RR) with 95% confidence intervals (CIs).65 For all continuous outcomes, the mean change from baseline was extracted; when change values were not available, these were calculated using baseline and followup means. Continuous outcomes were analysed using the DerSimonian and Laird inverse variance method, and reported as standardised mean differences (SMDs) with 95% CI.66 SMDs were calculated to account for variation in the outcome scales. All meta-analyses were also conducted using RevMan.

The results of the analyses were exported from RevMan into GRADEpro, a web-based software, to generate a GRADE Evidence Profile for each PICO question.67 The quality of evidence available for each outcome was assessed in GRADEpro using GRADE quality assessment criteria (Table 3.1, Appendix 3 of the Guideline for the management of knee and hip osteoarthritis: Technical document).68 This assessment was performed in duplicate by two independent reviewers (Investigators Raveendhara Bannuru and Mikala Osani as named in the Guideline for the management of knee and hip osteoarthritis: Technical document), with discrepancies resolved by consensus.

The evidence was rated for each outcome judged by the following criteria:69–72

  • risk of bias assessment of all individual studies
  • inconsistency between trial results
  • indirectness of evidence
  • imprecision of the effect estimate
  • potential reporting bias.

As per GRADE methodology, the eventual quality ratings are sorted into four categories: ‘High’, ‘Moderate’, ‘Low’ and ‘Very low’. These categories reflect the reviewers’ confidence in the effect estimate and its proximity to the true effect of an intervention.73 ‘High’ grade evidence is designated a numerical equivalent of 4, with quality downgrades carrying a weight of –1 for ‘serious’ risk or –2 for ‘very serious’ risk. The ‘Very low’ rating carries a numerical equivalent of 1; once the quality of evidence has been downgraded to this point, it cannot be downgraded further.

Once the quality of evidence has been assessed for all reported outcomes, the overall evidence quality was evaluated based on the lowest quality rating given to any of the outcomes that were designated a priori to be ‘Critical’ by the working group.74 When no data was available for a recommendation, the overall quality of evidence was automatically marked as ‘Very low’ to signify that this recommendation was made based on clinical experience alone. The resulting GRADE Evidence Profile contained:

  • pooled effect estimates calculated for each outcome
  • quality ratings for each outcome
  • footnotes containing brief qualitative summaries of the rationales behind quality downgrades
  • importance of each outcome
  • overall quality of evidence rating.

The final evidence report comprised all the GRADE Evidence Profiles (Appendix 5 of the Guideline for the management of knee and hip osteoarthritis: Technical document).

GRADE methodology specifies that guideline working groups formulate recommendations based on:74

  • consideration of the balance of relative benefits and harms of the treatment options under consideration
  • quality of the evidence (ie confidence in the effect estimates)
  • a person’s values and preferences
  • resource implications.

Key to the recommendation is the trade-off between desirable and undesirable patient outcomes; recommendations require estimating the relative value individuals place in the outcomes.

A recommendation could either be in favour or against the proposed treatment option, and strong or conditional. The recommendation can also be labelled as conditional neutral, where the working group cannot determine the direction of the recommendation.

A GRADE recommendation is categorised as strong if the working group is very confident that the benefits of an intervention clearly outweigh the harms (or vice versa). A conditional recommendation denotes uncertainty over the balance of benefits and harms (eg evidence quality is low or very low), or when personal preferences or costs are expected to affect the decision.75 Thus, conditional recommendations refer to decisions where incorporation of personal preferences is an essential element of decision-making.

Consensus building

The working group received the evidence report for review before meeting to discuss and decide on the final recommendations. For each PICO question, the working group provided initial votes, and discussed the direction of the recommendation during monthly teleconferences and two face-to-face meetings until general consensus was reached. Once the recommendations were drafted, the working group was asked to indicate their extent of support for each recommendation through an online survey voting process.76 An 11-point numerical scale was used to rate the extent of support for each recommendation. A 70% consensus agreement was set as the threshold for accepting a recommendation. If a 70% consensus was not achieved during the initial vote, additional discussions were convened to finalise the recommendation. A second and final voting process (with an agree/ disagree response) was then conducted. Final recommendations were accepted with a 70% consensus agreement by working group members. Details on the survey and voting data are available in Appendix 4 of the Guideline for the management of knee and hip osteoarthritis: Technical document.

In some instances, the working group decided to combine certain treatment options based on the review of the evidence and clinical scenario (eg different types of exercise). In addition, the working group identified a number treatments (and the resultant evidence reviews) not required for the guideline as the clinical scenario was uncommon, irrelevant, or redundant (Table 3.2, Appendix 3 of the Guideline for the management of knee and hip osteoarthritis: Technical document), or because one of the treatment options for that scenario had been eliminated by another recommendation. Consistent with GRADE guidance, there were some treatment options where the working group chose to provide a strong recommendation despite a low-quality rating of evidence.74 In these instances, a written explanation is provided describing the reasons for this decision.

This event attracts CPD points and can be self recorded

Did you know you can now log your CPD with a click of a button?

Create Quick log

Advertising