Something as seemingly safe as water can be as dangerous as molten Sodium. The following is an explanation of the Chernobyl accident, which in part was due to using water as a coolant:
Efforts to increase the power to the level originally planned for the test were frustrated by a combination of xenon poisoning, reduced coolant void and graphite cooldown. Many of the control rods were withdrawn to compensate for these effects, resulting in a violation of the minimum operating reactivity margin (ORM, see Positive void coefficient section in the information page on RBMK Reactors) by 01:00 – although the operators may not have known this. At 01:03, the reactor was stabilised at about 200 MWt and it was decided that the test would be carried out at this power level. Calculations performed after the accident showed that the ORM at 01:22:30 was equal to eight manual control rods. The minimum permissible ORM stipulated in the operating procedures was 15 rods. The test commenced at 01:23:04; the turbine stop valves were closed and the four pumps powered by the slowing turbine started to run down. The slower flowrate, together with the entry to the core of slightly warmer feedwater, may have caused boiling (void formation) at the bottom of the core. This, along with xenon burnout, could have resulted in a runaway increase in power. An alternative view is that the power excursion was triggered by the insertion of the control rods after the scram button was pressed (at 01:23:40). At 01:23:43, the power excursion rate emergency protection system signals came on and power exceeded 530 MWt and continued to rise. Fuel elements ruptured, leading to increased steam generation, which in turn further increased power owing to the large positive void coefficient. Damage to even three or four fuel assemblies would have been enough to lead to the destruction of the reactor. The rupture of several fuel channels increased the pressure in the reactor to the extent that the 1000t reactor support plate became detached, consequently jamming the control rods, which were only halfway down by that time. As the channel pipes began to rupture, mass steam generation occurred as a result of depressurisation of the reactor cooling circuit. Two explosions were reported, the first being the initial steam explosion, followed two or three seconds later by a second explosion, possibly from the build-up of hydrogen due to zirconium-steam reactions.
As soon as you involve radiation, you have some severe compromises to make, and the choices of elements you can use are limited.
Sodium is not chosen because of its dangers. Sodium was chosen despite the dangers, because the engineers very very carefully looked at other options and the other options were worse for the needs of their design. https://xkcd.com/793/ and http://johnsalvatier.org/blog/2017/reality-has-a-surprising-... seem relevant.
Something as seemingly safe as water can be as dangerous as molten Sodium. The following is an explanation of the Chernobyl accident, which in part was due to using water as a coolant:
https://world-nuclear.org/information-library/safety-and-sec...