Here is a research document on mammary tumors in dogs.
Thanks for that, it's a very interesting read. Not too often you see people actually come up with studies! Now I don't want to threadjack, but I have to say I have a very different take-away then you. I think you're basing your conclusion on the numbers coming out in table 2 (the number of dogs with vs without tumors in the different risk factor classes), which are rather different then the results of their statistical analysis in table 3 (whether or not those risk factors are actually statistically significant).
The study looked at 8 variables in a population of dogs in specific geographic areas (Brazil) to see if they could determine if they were significant risk factors to the occurrence of mammary tumors. The 3 factors they found to be statistically significant in this specific case were:
-age: the older the dog, the more likely to have a tumor.
-being "unspayed": they specify a significant difference in spaying before 3rd estrous cycle vs after, but also state that a similar relationship wasn't observed in some other studies.
-being overweight: fatter dogs were more likely to have tumors.
The factors they were unable to demonstrate to be significant in their analysis (I'd point out this doesn't mean these factors are not significant, just that their study wasn't "good enough" to show it) were:
- mixed breed vs pure breed (they didn't look at specific breeds and note tumors are known to be more frequent in certain breeds)
- diet: though they didn't show diet to be a significant cause directly, They note that 80% of the obese dogs in the study were fed a home diet, suggesting nutrition could contribute to other factors that are significant like obesity.
- contraceptive use, history of false pregnancy and history of having given birth were actually NOT found to be statistically significant factors in their analysis (despite the numbers in table 2 really looking like they were, there's like 2-2.5x more tumors in those groups). They unfortunately offer very little information on the matter in their discussion and point out that other studies have had different results.
I'm not knowledgeable enough in statistics to understand why table 2 and table 3 are so different, but my guess is that 1. the overall population in the study is fairly small, 2. the distribution in some of the categories is really skewed (like there's 7% of the dogs on contraceptives vs 93% not on them). Those 2 things would also explain why the confidence intervals are so large.
If you look at "being unspayed" in table 3, the odds ratio suggests you're 9.3x more likely then spayed individuals of having a tumor. The 95% confidence interval suggests that if you did the same study 100 times (picked same number of dogs with the same criteria etc), then 95 times out of 100 that odds ratio would be between 3.4-25. It's a huge gap, but it's basically always more likely, therefore statistically significant according to their model.
Compare that to having a history of false pregnancies. This specific study has an odds ratio close to 1 (so just as likely to get tumors one way or the other), but if you did the same study a bunch of times, sometimes you'd be half as likely to get tumors, sometimes you'd be twice as likely. If you were to do a stronger study, maybe using a larger number of dogs, you might narrow that interval down to something more conclusive.
I still believe that OSS will NOT reduce the risk of mammary tumors or false pregnancies in dogs. I'll grant you that if gestation contributes to the likeliness of getting these tumors, OSS would prevent that from happening, but OSS vs "intact and not having puppies" would be the same.
