The methodology is quite interesting--it's based on dogs that have health insurance, and the rates of pyometra and mammary tumors are based on the payment of health claims for those two diseases. The article is peer-reviewed. I'll quote some of the more relevant sections below for others to read:
Materials and Methods
The insurance process, benefits and limitations of using insurance data in veterinary research has earlier been described in detail (Egenvall et al.
2000,
2009). The Agria Insurance database contains data from over 260 000 female dogs (≤10 years). In the present study, female dogs enrolled for both veterinary care and life insurance during 1995–2006 were included. Each year a dog was insured contributed to one dog-year at risk (DYAR). All insured bitches were considered at potential risk for developing pyometra. Variables used were: gender, breed, date of birth, dates when dogs entered or left the insurance program, information on the type of insurance for which dogs were enrolled, and whether claims were reimbursed. Breeds were classified according to the Swedish Kennel Club breed classification system. If dogs had reimbursed claims for pyometra, they were considered as pyometra-cases and likewise for MTs. The crude- and breed-specific proportions (percentages) of dogs that had insurance claims for pyometra or MTs up to certain ages (6, 8, and 10 years) were calculated using the base-line survival statement from Cox regression (without independent variables). Incidence rates, crude and breed-specific mean ages were calculated for development of pyometra and MTs and either or both of the two diseases.
...
Results
The database contained data from over 260 000 female dogs, resulting in just over 1 000 000 DYAR calculated for dogs up to 10 years of age. There were 110 breeds with over 1000 DYAR in the database, constituting the basis for the analyses.
...
Discussion
It is common practise to insure dogs in Sweden and approximately 40% of all dogs are covered by Agria Insurance. Approximately 90% (all ages) of the Swedish dog population is intact, hence most dogs are susceptible to reproductive organ disease (Egenvall et al.
1999). The Agria database thus offers unique possibilities for studies concerning such diseases. The incidences of pyometra and MTs have been shown to vary between different breeds, suggesting a genetic background (Egenvall et al.
2001; Rivera et al.
2009). The overall proportion of dogs that had developed pyometra in this study was slightly lower (19%) than previously reported (23–24%) which possibly reflects that more low-risk breeds were included in the present study (Egenvall et al.
2001). Although based on different study periods, the overall proportion of dogs developing MTs in the present study was 13%, which is the same as in a previous report (Egenvall et al.
2005).
...
Elective spaying is commonly performed in many countries, and will prevent development of pyometra, and depending on at which age the surgery is performed, also have a protective effect for MTs (Schneider et al.
1969). However, recently the protective effects by spaying on MT development have been questioned (Beauvais et al.
2012). In the present study, high incidences (up to 73%) of pyometra and MTs were demonstrated in some giant and large breeds of which are commonly affected by post spaying urinary incontinence (Thrusfield et al.
1998). Since the pros and cons of elective spaying will vary by breed, knowledge of breed-variations will be clinically useful in the decision process for each dog.
Due to insurance age limitations, only data from dogs up to 10 years of age are included in the database, which is a study limitation. The true mean ages of diagnosis and incidences are likely to be higher and larger, respectively, and especially in breeds with long life expectancy.