Materials and Methods

Study population—A total of 12,750 horse farms with approximately 400,000 horses in Florida were considered for inclusion in this study. Census data from the 2002 USDA NASS Florida Field Office^12,13 were used for identification of the study farms. Racetracks and farms with > 50 horses were excluded because a low survey response was expected.

Anhidrosis—In this study, horses were classified as affected with anhidrosis by the farm owner or manager in the year 2005. In some instances, the diagnosis of anhidrosis was made by the attending veterinarian. Horses affected with anhidrosis were horses that sweated poorly or not at all under conditions in which a clinically normal horse would sweat freely; horses were treated for anhidrosis and had ≥ 1 of the following signs: rapid shallow breathing (usually ≥ 30 breaths/min) in hot conditions, a rectal temperature > 38.9°C (102°F) when resting, loss of hair over the face or chest, or dull coat, dry flaky skin, or both. Horse farms classified as affected with anhidrosis were farms with ≥ 1 horse classified as affected with this condition in the year 2005. This case definition of anhidrosis is acceptable and justified because in most situations, the diagnosis of anhidrosis in horses is based on clinical findings, which are well recognized by horse owners, farm owners or managers, or veterinarians. Horses that are suspected to have anhidrosis are observed by a farm owner or manager or the attending veterinarian after exercise (observation of either a complete or partial inability to sweat, increased respiratory rate and body temperature for a prolonged period after exercise, and, in some instances, tachypnea).^8,10,11 In this observational study, the case definition did not require results of an intradermal sweat test (with terbutaline sulfate, epinephrine bitartrate, or salbutamol sulfate) because, in clinical practice, most veterinarians do not use intradermal sweat tests for diagnosis of anhidrosis in horses. Intradermal sweat test protocols for diagnosis of anhidrosis in horses are not standardized^7,14,15 or have not been validated in anhidrotic horses.^7,15

Epidemiological questionnaire—A structured questionnaire was developed for collection of epidemiological data. Information requested pertained to farm- and animal-level factors. Farm-level information included the following: geographic location of the premises (northern, central, and southern Florida; Figure 1), type of operation (ranch, boarding, racing, pleasure, instruction, show, or other), whether the horse had access to grass on the premises (yes or no), type of feeding concentrate most commonly used (sweet feed, complete pelleted, alfalfa pellets, corn, oats, or other), type of roughage most commonly used (timothy and alfalfa, alfalfa, timothy, coastal, peanut, orchard grass, or other), source of water (pond, lake, river, well-pumped, well-artesian, city or county supply, or other), access to salt and minerals (none, salt-mineral block, loose salts added to grain, commercial mineral supplements,^a or other), number of horses on the premises, and horses diagnosed with anhidrosis (yes or no) and, if yes, number of horses affected with this condition. For each horse on the premises, the following information was requested: horse affected with anhidrosis (yes or no), age, sex, breed, color, foaling place (southern, northeastern, western, or midwestern United States; foreign country; or unknown), diet (free-salt minerals [yes or no]), use (halter, pleasure, trail riding, or working ranch horse; racing or race training; breeding; nonracing athletic performance; retired; or other), and family history of anhidrosis (sire or dam with history of anhidrosis [yes, no, or unknown]). In addition, for horses affected with anhidrosis in the year 2005, the following information was requested: age (years) when first affected with this condition, month of onset, clinical signs, treatment type considered as the most successful, and duration of treatment. Finally, among farms with horses affected with anhidrosis, the questionnaire was structured to gather information on every anhidrotic horse (maximum of 5 horses) and up to 10 nonaffected horses/farm. Similarly, among farms with horses not affected with anhidrosis, a maximum 10 horses/farm were selected. A list of random numbers for selection of horses not affected with anhidrosis was sent to each farm along with the survey. The farm owner or manager was asked to identify each nonaffected horse with a consecutive number (1 to n), then to select the first 10 horses (number identification) shown on the list of random numbers that matched her or his list. The questionnaire was mailed to each study farm from the USDA NASS during November and December of 2006. All questions pertained to events of anhidrosis and conditions in the horse farm during 2005. Four to 8 weeks after the questionnaires were mailed, farm owners or managers who failed to return the questionnaire were identified by USDA NASS personnel, and an attempt was made to contact them via phone call for an interview to complete the questionnaire. Data collection was recorded electronically by USDA NASS personnel.

View Full Size

Map of Florida showing the northern, central, and southern regions in the state.

Citation: Journal of the American Veterinary Medical Association 236, 10; 10.2460/javma.236.10.1091

• Download Figure
• Download figure as PowerPoint slide

Study design—This study was designed as a cross-sectional study. To accomplish the first objective, the epidemiological questionnaire was mailed to 1,207 horse farm owners or managers in Florida. A sample size of 649 horse farms was calculated by use of the following assumptions: population size of 12,750 farms, 20% prevalence of anhidrosis at the farm level, 95% confidence, and accepted error of 3%. On the basis of a projected 33% response proportion, an initial sample of 1,947 farms was selected. However, because of funding limitations to conduct follow-up phone interviews, the number of farms was reduced to 1,207 farms, which were randomly selected from a total of 12,750 horse farms. A computer program^b was used to generate random numbers (ie, select farms) needed in this portion of the study. The projected response proportion of 33% was based on results from a previous prevalence study¹⁶ that used a similar survey method. Prevalence of anhidrosis was estimated by dividing the number of affected premises and horses by the number of premises and horses included in this study, respectively.

The approach to the second objective was to compare the frequency of investigated factors between 56 and 444 farms classified as affected and not affected with anhidrosis, respectively, as well as between 79 and 2,365 horses classified as affected and not affected with this condition, respectively.

In addition, among 56 farms with horses affected with anhidrosis, the frequency of investigated animal-level factors was compared between 79 horses affected and 327 horses not affected with anhidrosis, matched by farm. The rationale was to control for possible confounding effects that farm conditions might have had on anhidrosis.

Statistical analysis—Under objective 1, a study population of 500 farms and 4,620 horses was used to estimate the prevalence of anhidrosis. Prevalence of anhidrosis at the farm level was estimated by dividing the number of farms with ≥ 1 affected horses by the total number of study farms. Prevalence of anhidrosis in horses was estimated by dividing the number of affected horses by the total number of study horses. For each prevalence estimate, a 95% CI was calculated. Prevalence of anhidrosis was compared among regions (northern, central, and southern Florida) by use of a χ² test. Differences in response proportions were compared among regions by use of a χ² test. Values of P ≤ 0.05 were considered significant.

Under objective 2, a study population of 500 farms and 2,444 horses (79 affected horses and 2,365 nonaffected horses) with complete data was used in the epidemiological analysis. Unconditional logistic regression was used to model the odds of being a farm or a horse affected with anhidrosis. Initial screening of factors for anhidrosis was performed by use of univariable logistic regression. Initially, variables associated (P ≤ 0.20) with the outcome of interest (anhidrosis) were entered into a multivariable model. A forward stepwise approach was used to identify which variables were associated with anhidrosis by use of 2-sided P values (0.05 to enter and 0.10 to remove variables). Two-way interaction terms were included in the final model and tested for significance (P ≤ 0.05). The goodness of fit of the multivariable models was explored by use of the Hosmer-Lemeshow goodness-of-fit χ² statistic. In the final model, adjusted OR and 95% CI were reported. In this study, an OR was used as an epidemiological measure of association between the investigated factor and anhidrosis. Thus, if a factor was not associated with diagnosis of anhidrosis, the OR was 1. The greater the departure from 1 (either larger or smaller), the stronger the association was between the factor and the odds of diagnosis of anhidrosis. Upper and lower limits of the 95% CI were used to indicate that there was 95% confidence in the assertion that the true OR fell within this interval.

Finally, among farms with horses affected with anhidrosis, the frequency of investigated animal-level factors was compared between horses affected and horses not affected with anhidrosis (matched by farm) by use of conditional logistic regression analysis. In the final model, adjusted OR and 95% CI were reported. Values of P ≤ 0.05 were considered significant.

Results

A total of 500 of 1,207 horse farms participated in the study; the response proportion was 41%. The frequency with which questionnaires were returned via regular mail or completed via phone call was not recorded. Response proportions in northern (45%), central (39%), and southern Florida (41%) were not significantly (P = 0.60) different.

Prevalence of anhidrosis—At the farm level, the prevalence of anhidrosis was 11.2% (56/500 farms) with a 95% CI of 8.4% to 13.9%. The prevalence of the condition was significantly different among regions (northern Florida, 5.2%; central Florida, 13.4%; and southern Florida, 26.5%).

At the animal level, the prevalence of anhidrosis was 1.8% (83/4,620 horses) with a 95% CI of 1.4% to 2.2%. The prevalence of this condition was significantly different among regions (northern Florida, 0.08%; central Florida, 1.7%; and southern Florida, 4.3%). Among 56 farms with ≥ 1 horses affected with anhidrosis, 83 of 1,213 (7%) horses had clinical signs of anhidrosis. The median age when horses were first affected with anhidrosis was at 10 years old (first and third quartiles of 5 and 18 years old, respectively). Most horses were first diagnosed as affected with anhidrosis during the warmest months of the year (May through September; Figure 2).

View Full Size

Month of the year in 2005 when horses (n = 75) were first determined to be affected by anhidrosis during the study period: January to December 2005. Month of diagnosis known for 75 of the 83 affected horses. Letters on the x-axis indicate the months of January through December in sequential order.

Citation: Journal of the American Veterinary Medical Association 236, 10; 10.2460/javma.236.10.1091

• Download Figure
• Download figure as PowerPoint slide

Epidemiological analysis at the farm level—In the univariable analysis, the following variables were associated with anhidrosis: region, number of horses per operation, type of operation, type of feeding concentrate, type of roughage fed, and source of water. In the multivariable analysis, after controlling for number of horses on the farm, the odds of anhidrosis were 2.13 times as high in farms located in central Florida (95% CI, 0.99 to 4.59) and 4.40 times as high in farms located in southern Florida (95% CI, 1.58 to 12.23), compared with farms in northern Florida (Table 1). In addition, the odds of anhidrosis were 5.26 times as high in show operations and 15.40 times as high in riding instruction operations, compared with ranch operations. The odds of anhidrosis were 3.69 times as high in racing operations, compared with ranch operations (95% CI, 0.94 to 14.53), but this association was not significant (P = 0.06). The P value of the Hosmer-Lemeshow statistic was 0.43, which supported the overall goodness of fit of the model.

Epidemiological analysis at the animal level—In the univariable analysis, the following variables were associated with anhidrosis: sex, breed, foaling place, use, and family history of anhidrosis. In the multivariable analysis, breed, foaling place, use, and family history of anhidrosis were associated with diagnosis of anhidrosis (Table 2). The P value of the Hosmer-Lemeshow statistic was 0.89, which supported the overall goodness of fit of the model.

Among horses from farms affected with anhidrosis, the following variables were associated with this condition in the univariable analysis: breed, use, and family history of anhidrosis. In the multivariable analysis, breed and family history of anhidrosis were confirmed as factors associated with this condition. The odds of anhidrosis were 4.37 times as high in Thoroughbreds (95% CI, 1.23 to 15.45) and 13.88 times as high in warmblood horses (95% CI, 2.48 to 77.47), compared with Quarter Horses. The odds of anhidrosis were 6.87 times as high in horses with a family history of anhidrosis (95% CI, 1.82 to 25.92), compared with the odds in horses without a history.

Of the 83 horses with anhidrosis, complete data on clinical signs were available for 79 horses. Of these 79 horses, 70 had tachypnea (as the most common clinical sign), 33 had a history of profuse sweating followed by no sweating, 31 had a dull coat or skin scaling, 23 appeared fatigued, 22 had alopecia, 9 had hyperthermia, 9 had a decreased appetite, and 6 had a decrease in water intake.

Discussion

This study provides new information on the prevalence of anhidrosis and factors associated with this condition in horses in the state of Florida. Compared with previous studies,^10,11 the main strength of this study is the random sampling approach used to estimate the prevalence of anhidrosis. At the farm level, study results revealed that the odds of anhidrosis were higher in farms located in central and southern Florida, compared with farms in northern Florida. Furthermore, the odds of anhidrosis were higher in show and riding instruction operations, compared with ranch operations. At the animal level, diagnosis of anhidrosis was associated with breed (Thoroughbreds and warmblood horses), use (retired horses and nonracing performance horses), and foaling place (western or midwestern regions of the United States) as well as family history of anhidrosis.

This study has several limitations. First, in the epidemiological analysis, the cross-sectional nature of the study limited our ability to examine a temporal relationship between time of diagnosis of anhidrosis and exposure factors such as diet or current use of the horse in the year 2005 (eg, retired). Second, the examination of age as a factor associated with anhidrosis was not possible because the age of anhidrotic horses was that when horses were first affected with this condition in their lifetime, whereas the age of nonaffected horses was the same as in the year 2005. In the final version of the questionnaire, the variable for age of affected horses in the year 2005 was not (in-error) recorded. Because the estimated mean age of horses when they were first diagnosed with anhidrosis was 10 years, it is possible that the median age of anhidrotic horses at the time of the survey was > 10 years. Another limitation is external validity, as this study was limited to horse premises in Florida. Thus, prevalence estimates of anhidrosis reported here cannot be extrapolated to other states in the United States. Furthermore, racetracks were not included in this study; prevalence of anhidrosis in racetrack horses has been reported to be as high as 20%.¹⁰ Finally, in this study, diagnosis of anhidrosis was made by farm owners or managers, and in some instances, the diagnosis was confirmed by the attending veterinarian. Thus, it is possible that some horses may have been misclassified as not affected or affected with anhidrosis by farm owners and managers. For example, it is possible that horses with partial signs of anhidrosis may not have been detected as anhidrotic (false-negative results). If this scenario were real, then the prevalence of anhidrosis was underestimated. Furthermore, in the epidemiological analysis, if this scenario were an important bias, then the factors for anhidrosis identified in this study would apply only to horses with severe signs of anhidrosis. A scenario where horses with complete loss of sweat function were not detected by farm managers is less likely. A different scenario of misclassification (false-positive results) could be horses with respiratory tract disease classified (in-error) as anhidrotic; the magnitude of this potential type of misclassification in this study is not known.

Our study results revealed that the prevalence of anhidrosis was 11% at the farm level and 2% at the animal level. To our knowledge, there are no other reports on the prevalence of this condition at the farm level in the United States for those states bordering the Gulf of Mexico, where the condition is commonly diagnosed. In a previous study¹¹ conducted on 4 farms with a history of anhidrosis in Florida, 52 of 834 (6%) horses had clinical signs of anhidrosis. In that study,¹¹ the main objective was not to estimate the prevalence of anhidrosis, but rather to compare clinical findings and results from a clinicopathologic analysis between affected and nonaffected horses in a selected number of farms with affected horses. Furthermore, in that study,¹¹ Thoroughbreds were the only breed represented in the study population. In our study, among 56 farms with ≥ 1 horses affected with anhidrosis, 83 of 1,213 (7%) horses had clinical signs of anhidrosis. It is difficult to compare these 2 prevalence estimates, as the study objectives, sampling methods, and diagnostic procedures all differed.

In this study, most horses were first diagnosed as affected with anhidrosis during the warmer months of May through September. This finding was expected, and it is in line with those of a previous study,¹⁷ where a group of 8 horses with a history of anhidrosis was exposed to riding school training during the summer, autumn, and winter seasons in Hong Kong. Results of that study,¹⁷ and a previous study¹⁸ with similar results, suggested that the atrophy of sweat glands and the blockage of sweat gland ducts is the consequence, rather than the cause, of anhidrosis.

At the farm level, the odds of anhidrosis were 2.13 and 4.40 times as high in horse farms located in central and southern Florida, respectively, compared with the odds in farms in northern Florida. One explanation for this observed association is exposure to increasingly hot and humid conditions in central and southern Florida, compared with the northern region in Florida. For example, in the year 2005, the number of weeks when the weekly average maximum temperature was ≥ 25°C (≥ 77°F) was 6, 7, and 9 weeks in northern, central, and southern Florida, respectively.^c Anhidrosis results from prolonged stimulation of the sweat glands by epinephrine secreted as a response to conditions of hot, humid climate.⁶ It is hypothesized that the sweat glands accommodate to the high epinephrine content of the blood, thus becoming insensitive to it.⁶ The clinical signs can be reversed by rest in a cooler and less humid climate.⁶

The odds of anhidrosis were 5.26 and 15.40 times as high in show and riding instruction operations, respectively, compared with the odds in ranch operations. Diet and exercise have been identified as aggravating factors, rather than risk factors for anhidrosis.⁴ The condition of anhidrosis in horses can be aggravated by factors that tend to increase the basal metabolism of affected horses (high-protein feed, disease, and exercise), exacerbate water loss (polyuria), or influence heat loss and help to dislocate body heat control.^4,19,20 In this study, the number of riding instruction operations was small (3 farms with affected horses and 1 farm with nonaffected horses), and this result affected the precision of the point estimate (OR, 15.40), as reflected by the wide 95% CI (1.20 to 197.20). Finally, the association between racing operations and diagnosis of anhidrosis (OR, 3.69) was not significant (P = 0.06); however, it is possible that the odds of diagnosis were higher in racing operations, compared with the odds in ranch operations, but the relatively small number of racing operations (n = 7) with affected horses in this study prevented the detection of a significant association.

At the animal level, the odds of being diagnosed with anhidrosis were higher in Thoroughbreds and warmblood horses than in Quarter Horses. Thoroughbreds, in particular, have consistently been the focus of several studies^1,10,21 on anhidrosis. Warmblood horses typically have a larger body size than other breeds of horses included in this study. A larger body size results in a decreased skin surface area-to-body mass ratio, which can effectively limit the thermoregulatory capacity of the sweat mechanism.²² The decreased thermoregulatory efficiency of the sweat mechanism can lead to increased body heat storage, placing horses at a higher risk of anhidrosis.²² In the present study, a definition for warmblood horses was not provided in the survey; the final designation of a horse as a warmblood was made by the owner. In Florida, warmblood horses (eg, Hanoverian, Trakenhner, Westphalian, Holsteneiner, Dutch Warmblood) are often used as sport horses in dressage, jumping, or eventing disciplines. By comparison, Quarter Horses are often used for trail riding or the less strenuous disciplines of western performance (eg, western pleasure, halter). A high-energy diet and strenuous exercise are 2 aggravating factors⁴ that can increase the frequency of anhidrosis observed in Thoroughbreds and warmblood horses, compared with Quarter Horses.

None of the 190 Arabian horses included in this study were classified as anhidrotic. One possible explanation for this observation could be the differences in breeding regimes between Arabian horses and other breeds. The Arabian horse was first domesticated in what was known as the Fertile Crescent, a crescent-shaped area of land extending from present-day Iraq to Egypt.²³ The breed has been evolving in this area for several centuries.²³ The Arabian horse has adapted to survive in much higher temperatures than have other breeds, especially Thoroughbreds and warmblood horses.^23,24 Over many generations of selection in moderate climates, breeds partially derived from Arabian bloodlines (eg, Thoroughbreds and Quarter Horses) may have lost, to varying degrees, resistance to anhidrosis. Possibly, these differences in evolution may have affected the sweat gland epithelia of these breeds and allowed for polymorphisms among the genes controlling sweating (eg, β₂-adrenergic receptors or aquaporin-5 channel).⁵

Our study results support current knowledge from the literature¹⁰ that horses originating from more temperate climates can be at risk of anhidrosis when transported to hot, humid climates for racing or exercise. In this study, the odds of anhidrosis were higher in horses that originated from the western or midwestern regions of the United States, compared with the odds in horses that originated from the southern region. In a previous study¹⁰ in Florida, 15 of 24 horses affected with anhidrosis were native horses and the remaining 9 horses were imported horses. Because that study¹⁰ did not include healthy horses, it was not possible to estimate and compare the prevalence of anhidrosis between these 2 subpopulations of horses. Thus, although other studies^1,2 show that native horses are susceptible to develop anhidrosis, our study results provide epidemiological evidence that horses from temperate climates are more likely to be diagnosed with this condition when moved into Florida, compared with native horses.

In this study, the odds of anhidrosis were higher in retired horses and horses used for nonracing performance than in horses used for halter, pleasure, or trail riding or working ranch horses. Among 240 horses classified as retired, most were Quarter Horses (30%) and Thoroughbreds (20%). Ageassociated conditions such as hirsutism likely affect thermoregulatory efficiency in geriatric horses and thereby increase the risk of anhidrosis. It is likely that horses diagnosed with anhidrosis were retired after the diagnosis, more so if they remained in a hot and humid environment. Anhidrosis is a severe condition that can affect performance, possibly ending a horse's performance career. In addition, it is a welfare issue of concern because of the considerable distress experienced by affected horses.²⁵ The fact that nonracing performance horses were more likely to be diagnosed with anhidrosis can be attributed to diet and exercise, 2 factors that have been identified as aggravating factors, rather than risk factors for anhidrosis.⁴ Finally, in the present study, the reference group of horses identified at the lowest risk of anhidrosis (OR, 1.00) included 303 working ranch horses, all of which were classified as not affected with anhidrosis; ranch horses are seldom worked hard enough to sweat.

Our study results support a previous hypothesis that there is a hereditary predisposition associated with anhidrosis in horses. A hereditary predisposition to anhidrosis was first suggested in a review paper⁴ published in 1976. In our study, results of the epidemiological analysis revealed a strong association between family history of anhidrosis and risk of this condition. In 2 separate analyses, the odds of anhidrosis were 21.67 and 6.87 times as high in horses with a family history of anhidrosis than in horses without a history. We examined the foaling place data of the 30 horses with a family history of this condition, and almost all horses (n = 29) were foaled in the southern region of the United States. Thus, we could expect that horses in other states (ie, southeastern United States) with a family history of anhidrosis may be at high risk of anhidrosis when exposed to exercise in a hot and humid climate. Future studies comparing the degree of aquaporin-5 expression and lack of sweat production from sweat glands between horses with and without a family history of anhidrosis may further advance our limited, current knowledge on the genetic susceptibility as a causative factor for anhidrosis in horses.

Tachypnea was the most common clinical sign observed in horses affected with anhidrosis. In anhidrotic horses, tachypnea is recognized as a sign of persisting thermal distress even at rest.⁵ In horses, cutaneous evaporation of sweat is the primary heat loss mechanism; a minor portion of heat load is lost via respiratory evaporation and convection, mechanisms that are enhanced by increased respiration and air movement in hot and humid conditions.²⁵

In summary, our study results revealed that anhidrosis is a condition that is commonly diagnosed in horses during the warmer months in Florida. This condition is more frequently diagnosed in farms in central and southern Florida as well as in show and riding instruction operations. Thoroughbreds and warmblood horses, horses that originated in the western or midwestern regions of the United States and later moved to Florida, and horses with a family history of anhidrosis were more likely to be diagnosed with anhidrosis.