The point about botulinum toxin is that it grows best in anaerobic conditions. (Detection, Identification and Differentiation of Clostridia Species) This is typically associated with canning and other forms of preservation, where there is sufficient time for toxic amounts of the toxin to be produced.
The other point about anaerobic conditions is that they inhibit other forms of spoilage, such as moulds (Can molds grow in the absence of air?). This means that vacuum-sealing, as with canning, is a form of food preservation, where food might appear not to have spoiled after a long period of time, because the visual signs of spoilage are absent. Botulism-spoiled food may or may not have a bad smell (Do foods containing botulism have a bad odor or taste?).
It is common to vacuum-seal meat stored in commercial chillers, giving a shelf life of up to 27 days (for lamb, UK practice). A study in the UK found that while food is recommended to be stored below 5C, domestic refrigerators are rarely this cold. However even after 50 days, the vacuum-sealed meat did not show toxic levels of botulism.
The study compiled a long list of previous botulism outbreaks (inter alia):
- chopped garlic in oil stored for 8 months unrefrigerated in a restaurant (Botulism from Chopped Garlic: Delayed Recognition of a Major Outbreak), in Canada in 1985, and again USA 1990
- fresh yogurt with canned hazelnut - the source was the bad canning
- a canned cheese product which was not refrigerated after opening
- a sealed commercial clam chowder, which was supposed to be refrigerated, was stored at room temperature and then eaten
- a bean dip, and a pâté, likewise
- fresh fish botulism
The fish botulism case is obviously very relevant to the OP's question, and likely to inform CDC advice.
Three people in Hawaii were hospitalised with botulism poisoning, treated with antitoxin, and survived.
The details were:
- the fish was grilled and eaten on July 17, and it still contained intestines
- two of the patients ate the intestines, one patient did not.
- the fish had been sold to the market between July 2 - 13
- the temperature of the fish at the market was found to be 11C
- it is not stated when the fish was caught, but it was found to have been sold by fishermen sometime between July 2-13
- the fish, it is reasonable to believe, was not vacuum-packed
Another outbreak was caused by foil-wrapped baked potatoes. Here it was noted that during baking the environment improves for botulism, since the inside of a foil-wrapped potato never gets hot enough to kill the spores, and subsequently the cooled, but warm spores, are stimulated to grow in an environment where competing microorganisms have been eliminated. In the study it was found that the 'tightly-wrapped' potatoes would produce sufficient toxin after at least three days at room temperature.
There were several more cases involving vacuum-packed fish. In 1997 (https://www.sciencedirect.com/science/article/abs/pii/S0168160598000804) two Germans were poisoned when they consumed hot-smoked Canadian whitefish, which had been prepared in Finland. The botulism was found to be a North American strain.
The details of the processing were as follows:
- Fish imported, frozen, on 20 December
- Fish defrosted in cold water for 3 hours, on 9th January, then brined
- Fish was then subject to 40C for 30 minutes, and 60-75C for 140 minutes, and vacuum-packed early the next morning.
- The fish was exported to Germany in well-controlled conditions under 5C, and mostly around 3C.
- It was purchased on the day of arrival in Germany, 13 January, and eaten on 15 January. The purchaser's fridge was found to be at 4C, a safe temperature.
The study found good practice with no evidence that the fish had ever been improperly handled. The study notes that more salt would have eliminated the risk, but this is unacceptable to many consumers.
In a similar case in France, the people got sick again after eating Finnish-smoked Canadian fish. The fish was purchased in Finland on 22 August, stored with ice packs in the family car during the 14 hour drive, and eaten on 6 September, two days before the expiry date. This case (https://www.researchgate.net/publication/40034952_Botulism_and_hot-smoked_whitefish_A_family_cluster_of_type_E_botulism_in_France_September_2009/link/5811c5d108ae009606bee685/download) was less thorough than the previous one, and the family fridge temperature was not checked. However, they concluded that the likely >5C temperature during transport, as well as the likely >5C temperature in the family fridge for 2 weeks, could have given rise to the poisonous levels of botulism.
A third case, (https://www.eurosurveillance.org/content/10.2807/esw.11.29.03004-en?crawler=true) also involving Finnish-produced vacuum-packed smoked Canadian white fish, found nothing wrong with processing, and once again postulated that consumer storage was to blame.
- there are several cases associated with specifically vacuum-packed frozen fish, however in the reported cases the fish was first smoked
- there is at least one case associated with fresh fish.
- It seems that there is a greater risk from fish at temperatures over 3C compared to meat, which is safely stored at temperatures up to 8C. It does not seem likely that defrosting fish and cooking and eating it the same day would cause botulism poisoning, however in favourable conditions (e.g., three days at room temperature for anaerobic baked potatoes), botulism can occur.
This study (https://www.researchgate.net/publication/279955141_Clostridium_botulinum_in_packaged_seafood_risk_profile) seems to be the best source on botulism in vacuum-packaged seafood, and shows that botulism toxin is produced in just a few days at 30C, and in 10 days at 8C temperatures. Growth as low as 3.0C is possible, given weeks of storage. For seafood stored specifically in vacuum packaging that is not frozen or below 3C, the recommendation is to use a 3.5% brine.
Recommended storage conditions in vacuum are then 5 days at up to 10C, 10 days at 5C, or below 3C or frozen.
This graph shows botulism toxin production in vacuum packed seafood:
- there are numerous reputable sources (e.g. https://datcp.wi.gov/Documents/StoringThawingVacuumPkgdFish.pdf) recommending that vacuum-packed fish is removed from packaging before defrosting
- the simplified advice that defrosting vacuum-packed fish in its packaging appears to be accurate insofar as 'defrosting' does not imply anything about the conditions or length of time during which this is done. While under optimal conditions defrosting in a sealed package, then cooking and eating, poisoning seems to be unlikely, in practice people might defrost things improperly, leave them in the fridge for several days, and then cook them, and it appears totally possible that failure to release the vacuum could be the proximate cause of a botulism poisoning incident.
- by eliminating the anaerobic conditions, the scenario where improperly stored, vacuum-sealed fish, provides the best possible environment for the botulism present in certain fish to grow to the exclusion of other less deadly bacteria, moulds, etc., which might otherwise cause the food to be thrown away because it smells bad.