"Disease" is a big word. I'd like to address this question by focusing on the difference, or lack of difference, between a poison, a disease, and a yummy thing to eat. It turns out that they may all be the same. Yet different.
Phenylketonuria (fee-null-keet-o-noo-ria), mercifully also known as "PKU" (pee - kay - you) is a disorder in which the amino acid phenylalanine is not broken down by an enzyme (phenylalanine hydroxylase) and thus accumulates in the body as phenylpyruvic acid.
This is bad because phenylpyruvic acid interferes with normal development of neural tissues.
In western settings, newborns are tested for PKU. If a newborn has PKU, steps are taken to avoid negative effects. This involves restricting the intake of phenylalanine, which is found in animal products but not in plant products.
What kind of disease is PKU? Since it is caused by a genetically "broken" gene, PKU is a genetic disease. However, it is also the case that PKU is caused by having a certain diet. What if this allele arose in a primate that ate mostly leaves and fruit, and never consumed animal tissue, and became the common form of that gene? PKU would not be a disease in this primate because they never consume phenylalanine. If an individual in this species had nothing else to eat but animal products and was thus forced through starvation to eat meat, then it would suffer the effects of PKU. In this case, phenylalanine is a poison that has negative effects when consumed in quantity by a juvenile primate of this hypothetical species.
This may seem semantic, but consider this: People enjoy a wide range of plant tissues, especially leaves, which have wonderful aromatic flavors or a kind of edgy bitterness (eg. spinach, basil, thyme, and sage). However, the molecules that provide these qualities are often evolved products to limit herbivory by insects. One species' culinary enjoyment is another species' poison. An insect that eats a bunch of basil and dies from it is an insect that died from consuming poison, not an insect with a genetic disease.
There are many parallel examples of evolutionary change in the relationship between a molecule and its consumer. Oxygen went from being a poisonous waste product in early bacteria to a useful molecule in many forms of life, for instance. A more complicated set of examples can be seen with pathogens such as various bacteria. One kind of bacteria may "infect" a certain kind of host, but live in symbiosis in a different host. Indeed, endosymbiosis, the permanent cooperative relationship between bacteria and proto-eucaryotic cells that gives us todays' mitochondria is an example: Is it a disease, or is it an organelle?
In short, infectious and genetic diseases are not properly thought of as "things" or "conditions" but rather, as aspects of dynamic evolutionary processes. (I quickly add: this entire discussion, applied only to viruses, may lead to an entirely different conclusion.)
Sometimes semantics are important. We could start by distinguishing pathogens from the pathological conditions which they can cause. Next, we could distinguish conditions that predispose one to dysfunction from the dysfunctional states to which one is so predisposed. As soon as we draw such distinctions, it becomes obvious that ordinary linguistic practices won't underwrite rigorous definitions.
I've only just discovered the 'Ask a scienceblogger' section, and thus this post, which is a shame because it's really very good.
You just made me think about 'disease' in a whole different way, and it makes so much sense.
Have you written a follow up about the viruses? I'd like to see that too.