1.5.3 Evidence synthesis

Systematic reviews of the evidence

The WHO Guideline Steering Group formulated PICO questions to guide the systematic reviews. The following technical advisory meetings held in 2014 and 2015 contributed to this process:

  • a consultation on the treatment of HIV among adolescents (5);
  • a consultation on new strategies to optimize care in the postnatal period: infant prophylaxis, feeding and diagnosis;
  • Paediatric ARV Drug Optimization 2 (6);
  • The future with PrEP in combination HIV prevention, WHO and UNAIDS; Scoping for the development of further recommendations on the use of PrEP for the prevention of sexual transmission of HIV and the Technical Advisory Group for Pre-Exposure Prophylaxis;
  • a scoping consultation on care packages for people living with HIV (7);
  • a scoping consultation on chronic comorbidities in people living with HIV (8);
  • a scoping consultation on priority areas and required work on ARV toxicity (9);
  • a scoping meeting on the use of viral load and CD4 testing in the management of HIV; and
  • a consultation on strengthening the quality of HIV clinical services in resource-limited settings.

Systematic review teams (Web Supplement B) developed protocols and conducted reviews. PRISMA guidelines for systematic reviews and meta-analyses were used for reporting of reviews (10). Web Supplement B includes search strategies, quality assessment and synthesis of findings for all systematic reviews conducted in 2015. Data from the systematic reviews were summarized and presented as evidence profiles using the GRADE (Grading of Recommendation, Assessment, Development and Evaluation) approach (11).

Several reviews were conducted using a network meta-analysis to evaluate the inference of the comparative effectiveness of interventions that may or may not have been evaluated against each other. The WHO GRADE working group was consulted on the interpretation of findings and use of GRADE in evaluating the overall quality of evidence from these reviews.

Diagnostic test accuracy reviews were conducted following Cochrane methods. Summary-of-evidence profiles were developed using an adapted GRADE approach in consultation with the WHO Guidelines Review Committee and the methodologist in the Clinical Guideline Development Group.

Values of outcomes

A list of potential outcomes of interest was circulated to members of both Guideline Development Groups. The members were asked to score the importance on a scale of 1 (not important) to 9 (critical) from the perspective of individuals living with HIV, to consider the importance of the values to service users. The average of the scores and variability for each outcome were used to determine the outcomes critical to decision-making.

Overall quality of the evidence

The GRADE method was used to rate the overall quality of the evidence (Table 1.1).

The quality of evidence is defined as the confidence that the reported estimates of effect are adequate to inform a specific recommendation. The GRADE system classifies the quality of evidence as high, moderate, low and very low. Randomized controlled trials are initially rated as high-quality evidence but may be downgraded for several reasons, including risk of bias, inconsistency of results, indirectness of evidence, imprecision and publication bias. Observational studies are initially rated as low-quality evidence but may be upgraded if the magnitude of the treatment effect is very large, if evidence indicates a dose–response relationship or if all plausible residual confounding would reduce the demonstrated effected or increase the effect if no effect were observed.

The strength of a recommendation reflects the degree of confidence of the Guideline Development Group that the desirable effects of the recommendation outweigh the undesirable effects (12). Desirable effects (potential benefits) may include beneficial health outcomes (such as reduced incidence of HIV and reduced morbidity and mortality); reduction of the burden on the individual and/or health services; and potential cost savings for the individual, communities, programme and/or health system. Undesirable effects (potential harm) include those affecting individuals, families, communities or health services. Harm may include the resource use and cost implications

of implementing the recommendations; adverse clinical outcomes (such as drug resistance or drug toxicity); and legal ramifications, in which certain practices are criminalized.

Strength of the recommendations

The strength of a recommendation can be either strong or conditional.

A strong recommendation is one for which there is confidence that the desirable effects of adherence to the recommendation clearly outweigh the undesirable effects.

A conditional recommendation is one for which the Guideline Development Group concludes that the desirable effects of adherence to the recommendation probably outweigh the undesirable effects or are closely balanced, but the Groups are not confident about these trade-offs in all situations. At implementation, monitoring and rigorous evaluation is needed to address these uncertainties, which are likely to provide new evidence that may change the calculation of the balance of trade-offs and to suggest how to overcome any implementation challenges.

Key information sources

Modelling of potential impact

Modelling data on the impact of interventions were used to support decision-making. A causal modelling analysis comparing the effect of various treatment initiation criteria on death and growth response for children and adolescents aged 5–16 years old was used to support the recommendation on when to start ART for children and adolescents. Models developed to simulate how immediate ART initiation regardless of CD4 cell count could affect HIV mortality and transmission in sub-Saharan Africa contributed to decision-making on the recommendation on when adults should start ART. The benefits of alternative approaches to monitoring people receiving ART were also modelled, including prediction of one-year mortality using the current CD4 percentage and count among children receiving ART. The results of this work informed the CD4 criteria for switching therapy in children without access to viral load monitoring. The costs and consequences of viral load monitoring with dried blood spot versus plasma specimens in low- and middle-income countries were also used to support decision-making.

Resource use

The prices for drugs in ART regimens were collected from the WHO Global Price Reporting Mechanism: http://apps.who.int/hiv/amds/price/hdd/.

Equity and acceptability

Evidence on the acceptability and views of service users was collected using a mixedmethods approach. The WHO Guideline Steering Group reviewed all PICO questions to identify those for which collecting service user data would best inform decision-making. The service users considered were people living with HIV, caregivers of children living with HIV and health workers.

Four qualitative evidence syntheses were conducted on the following topics: 1) the role of ARV drug toxicity in influencing adherence among people living with HIV; 2) barriers to and facilitators of interventions to improve linkage to care; 3) barriers to and facilitators of interventions to improve ARV adherence; and 4) barriers to and facilitators of interventions to improve retention in care for people receiving ART. The Confidence in the Evidence from Reviews of Qualitative Research (CERQual) (13) approach was used to assess the confidence in the findings from the qualitative evidence syntheses, using four factors (Table 1.2).

Qualitative literature reviews (published and grey literature) were conducted on the following topics: 1) the timing of ART initiation for all populations (including pregnant and breastfeeding women); 2) the duration of infant prophylaxis; 3) PrEP; 4) the frequency of clinic visits and medication pickups; 5) initiating ART on the same day as HIV testing; 6) task shifting for sample collection and diagnostic tests, and 7) delivering sexual transmitted infection and family planning services in HIV clinic settings.

A community consultation was conducted on when to start ART and the use of viral load testing for monitoring of people living with HIV. Seven networks of people living with HIV, supported by a global research organization, conducted 24 workshops to assess the acceptability of initiating ART earlier for people living with HIV, caregivers and service providers in eight countries (India, Indonesia, Kenya, Peru, Portugal, Ukraine, Zambia and Zimbabwe). An additional global consultation on adolescent treatment and care and a facility-based situational analysis also contributed to the evidence base on the acceptability of initiating ART earlier among adolescents.

Community networks conducted focus group discussions to explore the views on early infant diagnosis of women living with HIV and the mothers of infants exposed to HIV on in Kenya, Namibia and Nigeria. An online survey on the quality of HIV care experienced by people living with HIV collated responses from 534 individuals and was used to support the guidance on quality of care in Chapter 6.


Programmatic data from country implementation experience were used to support decision-making. Programme data from Brazil, Rwanda, Thailand, Uganda and VietNam were presented to the Guideline Development Groups (14). Qualitative interviews and an online survey were used to ascertain the views of HIV programme managers in three WHO regions and the conclusions from these assisted discussions on the feasibility of interventions.

WHO also conducted a survey in low- and middle-income countries on the availability and use of HIV diagnostics (15) and ARV drugs. The outcomes from a meeting with diagnostic manufacturers informed the decision-making (16).

Meetings of the Guideline Development Groups

The Clinical and Operational Guideline Development Groups met in Geneva, Switzerland in June 2015 to review evidence and formulate recommendations. The Operational Guideline Development Group met following the Clinical Guideline Development Group so that participants could consider the implementation considerations of decisions made in the clinical meeting.

The Clinical Guideline Development Group considered evidence for PICO questions A1.1 to E1.1. The Operational Guideline Development Group considered evidence for PICO questions F1.1 to F6.2. The Clinical Guideline Development Group held an additional virtual meeting in September 2015 to review two PICO questions that were not fully reviewed in June. A quorum of the Clinical Guideline Development Group attended this meeting and a ‘round-robin’ approach was used to gain consensus on the recommendations. Email communication followed to ensure that all members of the Clinical Guideline Development Group agreed with the recommendations made by virtual discussion.

The systematic reviews and evidence-to-decision-making tables prepared in accordance with the GRADE process (Table 1.3) were presented at the meetings, and the respective methodologist facilitated discussions. Web Supplement A provides all evidence-todecision-making tables, including GRADE evidence profile tables for all PICO questions that led to a recommendation.

The Guideline Development Groups made decisions by consensus. The Clinical Guideline Development Group voted as a decision-making aid for the question of when adults should initiate ART in relation to the strength of the recommendation (with a majority vote of 70%). No other votes were held to decide strength or directionality of recommendations.

Good practice statements

Good practice statements were made when the Guideline Development Groups considered the benefits to substantially outweigh any undesirable consequences, in many cases considering a large body of indirect evidence with indirect comparisons to strongly support the decision. Principles for developing good practice statements were applied (17). Formal GRADE methods have not been applied to these statements. Where statements have been referenced from other guidelines (often referred to as good practice principles or practices), the source is noted.

Content Type
Hide share