The International Tokamak Physics Activity topical group on integrated operational scenarios has compiled a database of stationary H-mode discharges at q(95) similar to 3 from AUG, C-Mod, DIII-D, JET and JT-60U, for both carbon wall and high-Z metal wall experiments with similar to 3300 entries. The analyses focus on discharges that are stationary for. 5 thermal energy confinement times to evaluate the baseline scenario proposed for ITER at 15 MA for achieving its goals of Q = 10, fusion power of 500 MW at normalised pressure, beta(N) = 1.8 and normalised confinement as predicted by the standard H-mode scaling, H-98y2 = 1. With the data restricted to stationary H-modes at q(95) similar to 3, the database shows significant variation of thermal energy confinement compared to the standard H-mode scaling (IPB98(y, 2)) in dimensionless form. The data show similar scaling with normalised gyro-radius, but more favourable scaling towards lower collision frequency and more favourable scaling with plasma beta. Using all the engineering variables employed in IPB98(y, 2), results in an overfit due to correlations among the data. Moreover, there are significant residual trends in the confinement for plasma current, device size, loss power, and in particular for the plasma density. Significant differences between results obtained for devices with a carbon wall and high-Z metal wall are observed in the data, with data from carbon wall devices providing a larger operating space, encompassing ITER parameters or even exceeding them. H-modes in high-Z metal wall devices have, so-far, not accessed conditions at low collision frequencies, have lower normalised confinement (H-98y2 similar to 0.8-0.9) at low input power or beta, achieving H-98y2 similar to 1.0 only at input powers two times the L-to H-mode transition scaling predictions and at beta(N) similar to 2.0. Hence, only the best H-modes with high-Z metal walls reach ITER baseline performance requirements. The data show that operating at high plasma density, with line-averaged density at 85% of Greenwald density is achievable for H-98y2 > 0.95 for a range of plasma configurations, and that operation at low plasma inductance with l(i)(3) similar to 0.7-0.75 is feasible. Scenario simulations employed for projecting the plasma performance in ITER should incorporate a lower thermal confinement at low plasma beta for the entry to burn and provide projections using higher levels of plasma core radiation by plasma impurities. Moreover, ITER projections should not subtract the core radiation in the evaluation of the thermal confinement time and H-98y2, to allow a fair comparison with experimental data currently available. From the data presented here, it is likely that in ITER the energy confinement time will not increase with plasma density and will have no degradation with plasma beta. The analyses indicate that the data at q(95) similar to 3 are consistent with achievement of the ITER mission goals at 15 MA.