Effect modifiers and prognostic factors
Indirect comparisons are fair only when the trials behind them are alike in the characteristics that change treatment effects. These characteristics were fixed in the protocol before any result was extracted. Each analysis page shows their baseline values trial by trial and flags imbalance between comparisons.
Baseline HbA1c
HbA1c falls more with higher baseline HbA1c (PIONEER subgroups: about 2 points above 9 percent against about 1 point below 8 percent); baseline HbA1c also sets the chance of reaching HbA1c targets.
Subgroup evidence in the included trials
| Outcome | Trials with subgroups | With interaction P below 0.05 | Median spread |
|---|---|---|---|
| HbA1c change | 17 | 9 | 0.41 |
| Body weight change in kilograms (diabetes) | 6 | 0 | 1.03 |
| All cause mortality | 1 | 0 | 0.06 |
| Three point MACE | 1 | 0 | 0.09 |
Baseline BMI or body weight
Absolute weight loss grows with baseline weight; trial level meta-regression links higher baseline BMI with greater weight loss; BMI is prognostic for blood pressure and cardiometabolic outcomes.
Subgroup evidence in the included trials
| Outcome | Trials with subgroups | With interaction P below 0.05 | Median spread |
|---|---|---|---|
| HbA1c change | 11 | 0 | 0.34 |
| Body weight change in kilograms (diabetes) | 8 | 0 | 0.97 |
| Percent body weight change | 4 | 1 | 2.00 |
| Three point MACE | 3 | 1 | 0.28 |
| Apnea hypopnea index change | 1 | 0 | 7.40 |
Type 2 diabetes share in weight trials
Meta-regression across obesity trials found about 4 to 5 kg less weight loss with type 2 diabetes for GLP-1 and dual agonists; mixed obesity trials differ in their diabetes share.
Sex
Weight loss is greater in women (STEP-HFpEF pooled: 9.6 against 7.2 percent; interaction p 0.006); female dominant cohorts lose more weight in meta-regression; gastrointestinal adverse events are more frequent in women.
Subgroup evidence in the included trials
| Outcome | Trials with subgroups | With interaction P below 0.05 | Median spread |
|---|---|---|---|
| HbA1c change | 6 | 0 | 0.23 |
| Percent body weight change | 4 | 4 | 5.62 |
| Body weight change in kilograms (diabetes) | 3 | 2 | 1.20 |
| Motor function (Parkinson disease) | 1 | 1 | 1.50 |
| Three point MACE | 1 | 0 | 0.06 |
| Apnea hypopnea index change | 1 | 0 | 5.10 |
Background glucose lowering therapy
Insulin and sulfonylurea background raises hypoglycemia risk and limits achievable HbA1c reduction; handled by the background stratum key of P01 glycemic and weight networks and checked within networks.
Subgroup evidence in the included trials
| Outcome | Trials with subgroups | With interaction P below 0.05 | Median spread |
|---|---|---|---|
| HbA1c change | 16 | 1 | 0.18 |
| Body weight change in kilograms (diabetes) | 12 | 3 | 1.07 |
| Three point MACE | 4 | 0 | 0.14 |
| Kidney composite outcome | 2 | 0 | 0.40 |
| All cause mortality | 1 | 0 | 0.00 |
| Kidney composite outcome (kidney disease populations) | 1 | 0 | 0.45 |
Diabetes duration
Longer duration reflects lower beta cell reserve and is prognostic for glycemic response; subgroup evidence for modification of relative HbA1c effects is weak and inconsistent.
Subgroup evidence in the included trials
| Outcome | Trials with subgroups | With interaction P below 0.05 | Median spread |
|---|---|---|---|
| HbA1c change | 5 | 0 | 0.17 |
| Body weight change in kilograms (diabetes) | 3 | 0 | 0.58 |
| Three point MACE | 3 | 0 | 0.26 |
Asian ancestry or region
Some trials show larger HbA1c reductions in Asian participants; Asian participants discontinued more often for adverse events in SUSTAIN; lower BMI thresholds change baseline weight in East Asian programs.
Subgroup evidence in the included trials
| Outcome | Trials with subgroups | With interaction P below 0.05 | Median spread |
|---|---|---|---|
| HbA1c change | 8 | 1 | 0.17 |
| Body weight change in kilograms (diabetes) | 3 | 1 | 0.48 |
| Percent body weight change | 2 | 0 | 2.02 |
| Three point MACE | 2 | 0 | 0.63 |
Age
Age is a strong prognostic factor for death and cardiovascular events; relative effects on MACE are consistent across age; HbA1c reduction was greater in people aged 60 or younger in SUSTAIN 6 and PIONEER 6.
Subgroup evidence in the included trials
| Outcome | Trials with subgroups | With interaction P below 0.05 | Median spread |
|---|---|---|---|
| HbA1c change | 12 | 3 | 0.27 |
| Body weight change in kilograms (diabetes) | 8 | 1 | 0.82 |
| Percent body weight change | 6 | 1 | 2.65 |
| Motor function (Parkinson disease) | 1 | 1 | 2.81 |
| Percent body weight change after randomization in maintenance trials | 1 | 0 | 1.50 |
| All cause mortality | 1 | 0 | 0.12 |
| Three point MACE | 1 | 0 | 0.07 |
| Apnea hypopnea index change | 1 | 0 | 6.70 |
Established cardiovascular disease
Established cardiovascular disease raises baseline cardiovascular risk; GLP-1 receptor agonist relative effects on MACE show no heterogeneity by cardiovascular history.
Subgroup evidence in the included trials
| Outcome | Trials with subgroups | With interaction P below 0.05 | Median spread |
|---|---|---|---|
| Three point MACE | 1 | 0 | 0.16 |
Kidney function (eGFR)
Lower eGFR raises the risk of MACE and kidney events while relative MACE effects stay consistent; weight reduction with semaglutide differed by baseline eGFR (interaction p below 0.001).
Subgroup evidence in the included trials
| Outcome | Trials with subgroups | With interaction P below 0.05 | Median spread |
|---|---|---|---|
| All cause mortality | 2 | 0 | 0.35 |
| Body weight change in kilograms (diabetes) | 1 | 1 | 0.85 |
| Three point MACE | 1 | 1 | 0.41 |
| eGFR total slope | 1 | 0 | 0.40 |
Albuminuria
Albuminuria is a strong prognostic factor for MACE and kidney progression; semaglutide effects on MACE and eGFR slope were consistent across albuminuria categories.
Subgroup evidence in the included trials
| Outcome | Trials with subgroups | With interaction P below 0.05 | Median spread |
|---|---|---|---|
| All cause mortality | 2 | 1 | 0.70 |
| Three point MACE | 1 | 0 | 0.37 |
Treatment duration
Weight effects accrue for more than a year; handled by the protocol windows and checked as the spread of time points within each network.
Lifestyle co-intervention intensity
An intensive behavioral program enlarges the placebo response and can shrink the relative weight effect.
Left ventricular ejection fraction
Ejection fraction defines the heart failure phenotype and prognosis; semaglutide effects in obesity related HFpEF were consistent across LVEF subgroups.
Fibrosis stage
Fibrosis stage determines the chance of histological resolution and of fibrosis improvement.
Baseline AHI and airway pressure use
Baseline AHI sets the room for reduction; positive airway pressure changes measured AHI and the reference response.