The 37,000 number is correct, but it can be misleading in a couple of ways.
A kg of fat contains 7700 calories of food energy, so to lose that 1kg of fat, you need to burn 7700 more calories than you take in. This is well-understood and agreed upon.
When you start talking about kJ, it gets a bit more complicated and confusing. Kj is a measure of the amount of work done - for example, I have a power meter on my bicycle, and it can measure the amount of work that is done in kj. Let's assume I go out on my bicycle and do 1000 kj of work. Now, a calorie is roughly 4 times bigger than a kj (see note 1), so that would imply that I only used about 250 calories of food energy. However, humans around 20-25% efficient in converting food energy to mechanical energy, which means it takes 250 / 25% = 1000 calories of food energy to do 250 calories of mechanical work. So... what everybody does is just consider them to be equivalent.
Unless you have a way of determining power expenditure, it makes a lot more sense to just deal with calories instead of kj.
note 1: When we talk about calories WRT food, we are really talking about kilocalories in the physics sense, but nobody calls them that.