Table 4. Evidence profile for adherence and clinical outcomes in HIV, hypertension and asthma.

Certainty assessment							№ of patients		Effect		Certainty	Importance
№ of studies	Study design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	EMM interventions	usual care	Relative (95% CI)	Absolute (95% CI)
Adherence
27	randomised trials	seriousa	seriousb	not serious	not seriousc	strong association	1267	1317	-	SMD 0.93 higher (0.69 higher to 1.17 higher)	⨁⨁⨁◯ Moderate	IMPORTANT
Proportion of patients with undetectable HIV viral load
5	randomised trials	not seriousd	not seriouse	not serious	very seriousf	none	117/164 (71.3%)	107/167 (64.1%)	not estimable		⨁⨁◯◯ Low	CRITICAL
Change in blood pressure from baseline—Systolic BP
4	randomised trials	not seriousg	serioush	not serious	seriousi	none	340	326	-	MD 2.2 lower (6.99 lower to 2.59 higher)	⨁⨁◯◯ Low	CRITICAL
Change in blood pressure from baseline—Diastolic BP
4	randomised trials	not seriousg	seriousj	not serious	seriousi	none	340	326	-	MD 2.44 lower (7.8 lower to 2.92 higher)	⨁⨁◯◯ Low	CRITICAL
Change in asthma control from baseline
5	randomised trials	seriousk	not serious	serious	seriousl	all plausible residual confounding would reduce the demonstrated effect	309	335	-	SMD 0.09 higher (0.07 lower to 0.24 higher)	⨁⨁◯◯ Low	CRITICAL

CI: confidence interval; SMD: standardised mean difference

Explanations

a. Allocation concealment was unclear in most studies, with most studies at high risk of performance bias due to the inability to blind participants to the EMM intervention. This may have affected adherence behaviour in participants over and above the EMM intervention effect

b. Imprecision—heterogeneity is high with I-squared = 86%

c. The boundaries of the CI are on the same side of their decision-making threshold and generally consistent across studies

d. Most studies included that assess this clinical outcome have unclear selection bias and are at high risk of performance bias due to the inability to blind participants to the EMM intervention. However, the clinical outcome (viral load) in this case is objective and therefore at low risk of detection bias.

e. Studies generally consistent with an I-square statistic of 40% suggesting studies generally homogenous

f. Studies generally small and events are few; risk difference is low with the confidence interval crossing zero

g. Included studies assessing this clinical outcome have unclear risk of selection bias, but generally low risk of bias in the other domains. Of note, the outcome blood pressure is unlikely to be affected by performance or detection bias

h. Although I-square is 41%, there is some inconsistency between studies

i. Relatively wide confidence interval crossing the no effect line, sample size less than 400

j. High heterogeneity in studies for diastolic BP outcomes, with I-square of 86%

k. Studies are at unclear risk of selection bias and high risk of performance and detection bias due to the nature of the EMM intervention. The outcome of interest is self reported asthma control; as such has potential to be affected by knowledge of intervention group

l. Wide confidence interval that crosses no effect line with relatively few number of studies