Design
This systematic review and subsequent network of meta-analysis was conducted and reported according to the Preferred Reporting Items for Systematic reviews and Meta-analysis (PRISMA) 2015 guideline [15, 16].
Search strategy
A search strategy was developed with health-research librarian, and piloted (see Appendices I, II, III, IV) following the guidelines specified by the Cochrane Handbook for Systematic Reviews [17], and the guidance for undertaking reviews in health care by the Centre for Reviews and Dissemination [18]. The following databases were searched from inception until September 2018, to include studies that reported the effects of aerobic exercise on hs-CRP level and depression scores: Cochrane Central Register of Controlled Trials, PubMed (MEDLINE), PEDRO, EMBASE, Allied and Complementary Medicine Database (AMED), Cumulative Index to Nursing and Allied Health Literature (CINAHL), and Web of Science Core collection. The following search terms were adapted for each database: HIV OR seropositive OR HAART OR anti-retroviral therapy, highly active OR anti-HIV agents AND aerobic exercise OR physical activity OR acute exercise OR exercise training OR physical exertion OR sports AND C-reactive protein OR acute-phase proteins OR reactants OR glycoproteins OR blood proteins OR immunoproteins AND depression OR depressive symptoms OR emotional depression OR depressive syndrome OR neurotic depression OR melancholia OR affective disorders OR adjustment disorders. Additionally, the reference lists of the included articles were hand searched. Attempts were made to collect unpublished data, we searched the National Library of Medicine’s clinical trial registry, directory of open-access repository websites and conference proceedings. However, there was only two ongoing trials [19, 20] still at the recruiting of participants stage that aim to evaluate the effect of aerobic exercise on depression and/or hs-CRP level among PLWH receiving HAART.
Eligibility criteria
The following eligibility criteria were applied in selecting studies.
An article was included in the review if;
a) The study population were adults (≥ 18 years old) with HIV infection receiving HAART.
b) The intervention was aerobic exercise, defined as physical activity of low to high intensity requiring free oxygen and utilization of oxygen to sufficiently meet energy demands during exercise through aerobic metabolism [21]. No particular restriction was considered regarding a particular form of aerobic exercise, dose, frequency, intensity, and duration of intervention or follow up period after intervention. As many such interventions may be complex, including several components, other interventions in addition to aerobic exercise including medications, nutrition, and health education were also included as long as the effect of aerobic exercise intervention could be determined.
c) The primary outcome measure was the hs-CRP level, and/or the secondary outcome was any measures of depression especially generic validated tools (i.e. depression tool not developed for a specific condition, e.g. Cardiac Depression Scale). Examples of generic depression validated tools include, but not limited to; Centre for Epidemiological Studies-Depression Scale, Montgomery-Asberg Depression Scale, Becks Depression Inventory score, and Positive and Negative Affect Schedule.
d) The control or comparator can be any control including non-exercise or resistance exercise.
e) The setting in which the study was conducted were any of the following: hospital (e.g. inpatient, acute or outpatient), rehabilitation homes, community or long-term care.
f) The article was randomized controlled trials (RCTs), and non-RCTs including quasi-RCTs, controlled clinical trials, pre-test post-test clinical trials.
An article was excluded if:
a) The intervention was not aerobic exercise based on the definition provided above.
b) It was an opinion paper, qualitative studies, narrative reviews, systematic reviews, letter to the editors or commentaries.
c) If the control group was not clearly defined.
Study record, selection process and data management
Eligibility questions and forms for the screening of the studies included within the review were developed, piloted and refined as appropriate. Literature search results were exported into RefWorks for deduplication and was later exported into Microsoft Excel (Microsoft Excel. Redmond, Washington: Microsoft, 2010. Computer Software) to facilitate the management and selection of articles for inclusion. The selection of the article was done in two phases: title and abstract screening, followed by full text assessment. While two reviewers (E.M.A and E.M.U) independently screened all titles and abstract, another two reviewers (O.A.E and I.B.N) conducted full text assessment according to the study selection criteria. The kappa for title/abstract and full text screening were 0.83 and 0.82 respectively, suggesting almost perfect agreement [22]. Disagreements at any screening stage were resolved through discussion with other authors (U.O.A, M.E.K).
Data collection process
Risk of Bias Assessment in Individual Studies
Two authors (O.J.A, E.M.A.) independently assessed the risk of bias of the included articles. The Cochrane Collaboration Tool for Risk of Bias Assessment (Table 8.5a of the Cochrane Handbook for Systematic Reviews of Interventions) was used to assess the articles in six key domains: 1) selection bias (random sequence generation, allocation concealment); 2) performance bias (blinding of personnel and participants); 3) detection bias (blinding of outcome assessments); 4) bias due to attrition (incomplete outcome data, including dropouts and withdrawals); 5) reporting bias (selective reporting); and 6) other bias (other sources of bias not elsewhere addressed) [17]. Articles were rated as ‘high risk’ or ‘low risk’ following a well-described procedure (Table 8.5.d of the Cochrane Handbook for Systematic Reviews of Interventions) [17, 23]. Then, summary assessment for each important outcome (across domains) within and across studies was made (Table 8.7.a of the Cochrane Handbook for Systematic Reviews of Interventions) [17, 23]. Again, any disagreement in the decision between reviewers was resolved by discussion and consultation with the third author (O.A.E.).
Data extraction items
Four authors (E.M.A, O.J.A, I.B.N, M.E.K ) used a standardized extraction form to extract the articles’ metadata including authors name, country the study was conducted, participants’ characteristics, study setting, sampling method, sample size, intervention characteristics (e.g. components of the intervention, who delivered the intervention, the duration of the intervention and follow-up) (where available), attrition rate, aspects of outcome assessed, the outcome measurement, study findings (e.g. p value, confidence intervals mean and standard deviation), conclusions and funding sources).
Data synthesis and assessment of heterogeneity
To address the review objectives, the quantitative study outcomes were extracted and presented in a proof table. Mean differences were used for the uniform outcomes with 95% of CI. A meta-analysis was conducted to find pooled effect sizes across the included studies, using a random-effects model depending on the level of heterogeneity of intervention effects. Heterogeneity was assessed using the Cochrane’s χ2 test (10% significance level) and Higgins I2 for which values of 25%, 50%, and 75% shows low, medium and high heterogeneity respectively as specified by the guidance in the Cochrane Handbook for Systemic Reviews of Interventions [17].
Data analysis
Studies were narratively described, and meta-analysis were conducted using random-effects model to evaluate the overall effect of aerobic exercise interventions on depression in patients living with HIV infection [17]. Meta-analysis was possible because of the homogenous nature of study design, intervention, comparator(s) and outcomes of the included studies. A substantial heterogeneity was observed in the two studies that considered hs-CRP as an outcome, therefore, only a narrative synthesis was conducted.
Quality of evidence and strength of recommendation.
Two authors (E.M.A, I.B.N) independently graded the quality of the evidence of the included studies on consistency, design, directness, precision, publication bias and study limitations of the Grading of Recommendations Assessment Development and Evaluation (GRADE) [24]. The included studies were graded as high risk of bias or low risk of bias with evidence statement graded from ‘High Quality’ to ‘Very Low Quality’ according to the criteria (Appendix V)