Materials competition results as well. It was not unusual that
276 L
. Wallin et al. Livestock Production Science 63 2000 275 –289
compensation had been granted. These estimates were never found 30. Among those traced, the
should not be confused with the average expected response frequency to the questionnaires was 78.
life span of all horses in the population, because of Ninety-nine horses had been exported and were not
lack of information on horses still alive and on included in the study due to lack of information
horses not compensated for at death. Such estimates about their health status. The number of horses
would therefore most likely underestimate length of finally included in the study was 1847, 503 of whom
life Ohlsson, 1990. were dead. The distribution between mares and
Survival analysis, a statistical technique that can males was fairly even 864 983. Of the males 74
use information on horses still alive, so-called cen- were stallions at the time of the test. Information
sored data, is a more appropriate way to estimate the about the horses was available until 1990. Horse
length of life of horses. More accurate estimates of owners were asked if particular horses were alive or
length of life would bring better information to the dead and, if dead, to state the cause and year of
debate about equine longevity. death. When analysing the questionnaires, no distinc-
Reports summarising causes of death in sports tion was made between horses that died and those
horses are rare. Many studies mention a specific that were put down, because death was considered to
disease affecting horses and are often based on a be due to the causes mentioned in the questionnaires.
relatively selected group of horses Baker and Ellis, The date of birth ranged from 1968 to 1982 Fig.
1981. Available statistics are based mostly on 1. The horses were usually 4 years old when they
insurance data or veterinary clinical studies. Accord- participated in the test, the oldest horses in the
ing to insurance statistics, diseases of the musculo- material could have reached an age of 22, whereas
skeletal system are the predominant cause of death of horses born in 1982 could not have been more than 8
sports horses Clausen et al., 1990; AGRIA, 1995; years old. When the test started in 1973, participation
Heisele, 1995. Among ‘companion’ horses kept as a was limited because all horses were judged at only
hobby, diseases of the digestive system are the most one location in Sweden. Later, more places became
frequent cause of death Baker and Ellis, 1981. available and interest in participating increased.
Other than insurance statistics, recent reports on The horses included in the study had been used for
length of life and the most important causes of death different purposes — many of them for more than
in horse populations are scarce. This emphasizes the one purpose Table 1. Most horses 85 had been
need to find unselected materials in order to establish competing in different disciplines. It was quite
appropriate benchmarks. The aims of this inves- common that horses competing at a novice level had
tigation, therefore, were to estimate median length of results in several disciplines, whereas horses compet-
life and to analyse and compare causes of death in ing at an advanced level had results only in one of
different Swedish horse populations representative of the disciplines. An earlier study Ohlsson and Philip-
their breeds and normal use. sson, 1992 showed that horses with high scores in
the test had competed more successfully than the others at advanced levels. Many brood-mares had