The overall mean (8.25 years) and median (8.40 years) life expectancy of BMDs in the present study was higher than that reported in most studies. However, life expectancy of BMDs appears to have increased over time from 5.7 years [18] and 6.8 years [5] in early studies to between 7.0 and 8.0 years in recent years [1, 4, 7, 19]. This apparent increase in life expectancy was also noted in previous studies and has been attributed to more advanced veterinary care and an increasing social value of dogs [5, 20]. Despite this increase, the BMD remains in the group of dogs with a relatively low life expectancy compared to breed-independent values, estimated to be a mean of 10.0 years [5] or a median of 11.0 years [6, 21]. The age at death used in data analysis in the present study was based on pedigree records from the Swiss Bernese Mountain Dog Club for the date of birth and the date reported by veterinarians or owners for the date of death. The accuracy of the date of death reported could not be verified.

The low life expectancy in the BMD has been attributed to a high prevalence of neoplasia, reported to be between 26.1 % and 55.1 % [1, 3–5, 18, 19, 22], compared to other purebred and general dog populations, in which tumour prevalence is estimated between 14.5 % and 27 % [1, 6, 21]. Data in the present study corroborates these findings with death due to neoplasia in 58.3 % of the target population. This was somewhat higher than previously reported. However, the reliability of diagnoses in the present study was highly variable, and most cases with a diagnosis of neoplasia were not confirmed on cytology or histology. The true incidence of neoplastic disorders is therefore likely to be somewhat lower or higher than that found in the present study.

Female BMDs in the present study had a higher life expectancy than males, corroborating findings of a previous study in which female BMDs were reported to live about 1 year longer than males [5]. Although a higher hazard ratio was found for intact males compared to intact and spayed females, and castrated males compared to intact females, this was not detected for castrated males compared to spayed females. However, a large proportion of dogs, for which the gonadectomy status was unknown, was excluded from this analysis. Moreover, the time at which gonadectomy took place was not taken into consideration. The effect, if any, of long-term presence or absence of sex hormones on survival is therefore unclear. Possible explanations for longer survival in females include a generally higher life expectancy in female dogs, as well as the higher risk of neoplasia in males and the associated lower survival in males in the present study. This finding contradicts findings of a previous study in which no difference in risk of neoplasia was found between males and females [4]. Moreover, it is in contrast to findings of different cancer registries, which reveal a predisposition for all neoplasms in female dogs [2, 5, 23]. However, this female predisposition is largely due to mammary cancer, which was not found in any of the dogs in the present study and was found to have a low prevalence in BMDs in a previous study [5]. This finding suggests that, although BMDs may be predisposed to some neoplasms, they may have a diminished risk for certain other types of neoplastic disorders or simply not live long enough for neoplasms associated with high age.

The most frequently diagnosed neoplasia in the present study was HS (10.4 % of all neoplasms). This was similar to previous investigations in Dutch and Norwegian BMD populations, in which HS accounted for 15.3 % [4] and 10.7 % [7], respectively. In a study of BMDs presented to a German university hospital, HS accounted for 8.2 % of histologically-confirmed tumours but up to 15.7 % when including tumours not histologically confirmed but highly suspected based on typical lesion distribution [22]. As the majority of tumours diagnosed in the present study lacked histologic confirmation and many dogs died without any diagnosis, the true prevalence in this population is unclear.

Of dogs with confirmed HS, the median survival of 7.2 years was somewhat higher than the previously reported 6.3 years [4]. Previous studies reported a wide range of ages of BMDs diagnosed with HS with one report of dogs aged between 10 months and 14.7 years with the majority between 5 and 8 years old [22]. Findings in the present study were similar with nearly half of the dogs dying between 7 and 8 years old with a range between 3.5 and 10.8 years old. The higher relative risk of male dogs with HS in the present study does not corroborate findings of previous recent studies in which females were reported with a higher prevalence or no difference in sex was found [11, 12, 22]. The reason for this apparent discrepancy is unclear but the large numbers of dogs in the present study for which no cause of death was available introduces a large uncertainty with regards to disease prevalence and sex distributions. Lymphoma was the second most commonly identified cancer, corroborating findings in previous studies [22].

With regards to non-neoplastic disease processes, dogs with symptomatic conditions included 13 dogs that were euthanised or died with paraparesis, paraplegia or tetraparesis. The median life expectancy of these dogs was 8.4 years, suggesting that at least some of these dogs may have suffered from degenerative myelopathy, previously described in older dogs of this breed [24]. However, this cannot be confirmed as post-mortem histopathology was not performed in any of these dogs. The BMDs with urogenital disorders had the shortest life span and most were attributed symptomatically to chronic or acute kidney injury. The extent to which any of these dogs was affected by hereditary membranoproliferative glomerulopathy in this breed is unclear [25, 26]. All in all, the 3.6 % of deaths due to kidney injury is somewhat lower (6.9 %) than in a previous report [7]. Five dogs suffered from immune-mediated diseases, of which 4 were diagnosed with immune-mediated haemolytic anaemia or Evans syndrome, resulting in a particularly low life expectancy in this disease process group. However, the number of dogs was very low and the degree to which an underlying neoplastic process was ruled out is not known.

In contrast to 1 previous study [7], cardiovascular diseases, behavioural problems, and infectious diseases occurred extremely rarely or not at all in our limited sample population. However, behavioural problems as cause of death (euthanasia) might not have been reported, as euthanasia for this reason is considered somewhat taboo in our society. In addition, a high vaccination coverage for common infectious diseases in Switzerland and good access to veterinary care might diminish the rate of lethal infectious diseases. Moreover, treatment of disorders such as degenerative joint diseases may significantly diminish their prevalence as a cause of death even though they may be widespread chronic disorders within the population. As the present study only evaluated diseases associated with death, the data cannot therefore be interpreted as general prevalence data.

The reason for a short lifespan in the BMD is likely multifactorial but significant factors include the high prevalence of tumours, as only few other breeds, including flat-coated retrievers and boxers, exhibit a comparable high prevalence [2–4]. Furthermore, BMDs with neoplasia died with a median age of 1.2 years younger than those that died of causes other than neoplasia in the present study. The shorter life span for dogs of larger body size is an additional factor [1, 8–10, 27–29].

In the present study, information regarding the cause of death that was purely based on information acquired from dog owners was considered of low diagnostic reliability. Interestingly, dogs with higher longevity were more frequently classified in the lowest quality of diagnoses. In contrast, dogs that died at an early age were classified in the highest categories. This observation may suggest that dog owners and veterinarians are more willing to invest time and money in the work-up and diagnosis of diseases in dogs presenting at a younger age. However, it is also possible that far more detailed data was transmitted by veterinarians when dogs were euthanized at a younger age.

To the authors’ knowledge, this is the first study to investigate only purebred BMDs with pedigree born in a single country and representing a large proportion of all dogs born within the same period. As the vast majority of dogs (all but 8) in the present study had died by the end of the study, bias due to right-censored data was minimal in the present study. Major limitations of the study include a relatively large proportion of owners who chose not to participate or could not be located, creating possible responder bias. In addition, slightly over 20 % of dogs died or were euthanised without any specific diagnosis, and the diagnosis of neoplasia was mostly made without histologic confirmation. Moreover, a large proportion of diagnoses were of poor or moderate reliability. Possible reasons include unwillingness of owners to confirm diagnoses in cases of presumed poor prognosis, and unwillingness to perform post-mortem examinations in cases in which no specific antemortem diagnosis was evident. For owners unwilling to perform necropsy for emotional and/or spiritual reasons, minimally invasive digital imaging, such as X-Ray, Ultrasound, CT and MRI in combination with fine needle aspiration and needle biopsy, may constitute a promising tool to improve post-mortem diagnoses [30]. In addition, overdiagnosis of neoplasia by veterinarians because of a perceived high risk of tumours in BMDs may have artificially inflated prevalence of neoplasia in this study. At the same time, some animals with unknown cases of death or those dying with only symptomatic diagnoses may indeed have had undiscovered neoplasia, leading to underestimation of its true prevalence.

Besides analyses on the causes of death, the collection of health histories from birth to death, as well as clinical and genetic material for analyses of specific diseases may be valuable to more precisely assess breed health. Further studies, prospectively collecting health data from defined health populations or cohorts may enable breeding clubs to more accurately assess breed health.