The amount a given load will need to be backed off depends on a number of factors that include the temperature sensitivity of the specific powder used, the actual charge weight and internal volume of cartridge, and how close to MAX pressure the original load that was developed in colder temperatures was. It's probably better to think of such an amount in terms of a percentage of the total charge weight, rather than some specific number of grains. For example, if you have worked up a good load at safe operating pressure in cooler temps, a good place to start is usually to drop the charge weight by 1%, 2%, 3%, etc., at the higher temperature until the velocity is equal to what it was originally at the cooler temp.
The most critical problems in using a load worked up at cooler temps when it's really hot outside are safety issues including increased pressure and velocity, although the load may also show poor precision. Going the other direction in terms of temperature (i.e. high>>>low) is usually less an issue of safety, and more an issue of poor precision. In my hands, loads that were tuned at close to, if not quite, the very highest summer temps, let's say from around 80-85 degrees F, will usually still shoot reasonably well when the mercury hits 95 degrees, or even slightly higher. Strangely enough, those loads still seem to shoot reasonably well down into the 50s or upper 40s, whereas the loads I've worked up in the upper 40s to low 50s usually shoot quite poorly at 80+ degrees.
With a given cartridge and load, it's not too difficult to measure actual velocity for a few different charge weights during both cold and hot temps, to make a good estimate of how much the load will need to be changed. In fact, it is not uncommon to see a wide range of daily temperature extremes within a short amount of time either early in the spring, or late in the fall. That is when I typically try to get such testing done.