Time trends and risk factors for diabetes mellitus in cats presented to veterinary teaching hospitals

Abstract

Veterinary Medical Data Base records of cats with diabetes mellitus (DM) from 1970 through 1999 were reviewed to identify trends in hospital prevalence of DM and potential host risk factors. Hospital prevalence increased from eight cases per 10,000 in 1970 to 124 per 10,000 in 1999 (P < 0.001). Case fatality percent at first visit decreased from 40% to 10% (P < 0.001). Hospital prevalence increased in all age groups (P < 0.002). There was no apparent seasonal pattern in hospital prevalence. Significant risk factors included male gender, increasing age for both genders (P < 0.001), increasing weight for males (P < 0.001), and mixed vs pure breed for females (P = 0.006).

Diabetes mellitus (DM) is a common endocrinopathy of the cat. Initiating factors in the cascade of events leading to feline DM are not completely elucidated, but likely include a combination of genetic and environmental determinants such as obesity, diet, exposure to toxic chemicals or drugs that cause insulin resistance, immune-mediated destruction of islet cells, and destruction of islet cells secondary to pancreatitis.

The incidence of DM in cats in the United States was reported to be one in 400 in a previous epidemiological study; reported risk factors were weight >6.8 kg, increasing age, male gender, and being neutered (Panciera et al 1990). Obesity, considered the most common nutritional disorder in dogs and cats, has also been suggested to be a risk factor for DM in cats (Sloth 1992, Rand 1999). There are anecdotal reports that the incidence of feline DM is increasing, but this has not been substantiated by controlled studies (Rand 1999).

The primary purpose of the present study was to identify temporal trends in the hospital prevalence of DM over a 30-year period using a large hospital data base, and to characterize potential host risk factors. Data were collected from abstracted medical records from the Veterinary Medical Data Base (VMDB). The VMDB, established in 1964 by the National Cancer Institute, contains standardized abstracts of every case seen at the participating veterinary teaching hospitals. Since 1964, 26 veterinary teaching hospitals have submitted over six million records to the Veterinary Medical Data Base – VMDB (2006).

Materials and methods

All VMDB records of cats were reviewed to identify the first hospital visit of cats with DM (VMBD code 870178500) for the period of time from January 1, 1970 through December 31, 1999. The records of all cats seen over the 30-year period were used to calculate the hospital prevalence of DM (number of cats with DM per 10,000 total cats seen per year), and the age specific hospital prevalence of DM (number of cats with DM in each age group per 10,000 total cats seen in the same age group per year). The case fatality percentage was calculated by dividing the total number of cats with DM that died or were euthanased in a given year at the first hospital visit by the total number of cats with DM in the same year then multiplying by 100. Seasonal occurrence was evaluated by grouping the admission date of cats with DM as follows: Spring = March–May; Summer = June–August; Autumn = September–November; Winter = December–February. A χ ² test for trend (Epi Info Version 6, Centers for Disease Control) was used to evaluate time trends.

Potential risk factors (age, gender, weight, neuter status, breed) were evaluated using univariate and multivariable logistic regression (SAS, Cary, NC, Version 9.1) for cat records in the VMDB from January 1, 1990 through December 31, 1999. This 10-year period was chosen in order to evaluate risk factors for the time period during which the hospital prevalence of diabetes mellitus increased the most sharply. Risk factor analysis was previously reported on limited data prior to 1990 (Panciera et al 1990). Records of first hospital visits of cats with any diagnosis other than DM, seen at the same teaching hospital in the same years as the cases, were used as controls for risk factor analyses. In VMDB records, age and weight are recorded only as categorical variables; actual ages and weights are not available. The overlapping age groups coded by the VMBD include ≤1 year, 1–2 years, 2–4 years, 4–7 years, 7–10 years, 10–15 years and >15 years. Overlapping weight groups coded by the VMDB include ≤2.27 kg (≤5 lb), 2.27–6.82 kg (5–15 lb), 6.82–13.64 kg (15–30 lb), and >13.64 kg (30 lb). The VMDB includes records for non-domestic cats. Subsequently, any cat over 13.64 kg (30 lb) was excluded from the analysis, as cats in that weight category were considered most likely to be non-domestic cats. Odds ratios, and 95% confidence limits and P values were calculated for each potential risk factor (SAS, Cary, NC, Version 9.1). Potential risk factors with a P value ≤ 0.1 in the univariate analysis were included in the multivariable analysis. A P value less than 0.05 was considered significant in the multivariable analysis. The breed risk of DM was calculated only for breeds having a minimum of 10 cases or controls. Interaction terms for gender × weight and for gender × breed were considered significant if P ≤ 0.1.

Results

Diabetes mellitus was diagnosed in 2576 cats of the 618,814 cats whose VMDB records were available for the 30-year period between January 1, 1970 and December 31, 1999. Nine hundred and fifty-five (37%) diabetic cats were female and 1612 (62.5%) were male. Eighty-seven percent of male and female cats were neutered. There were nine diabetic cats for which gender was not reported. Most (2186; 85%) of the diabetic cats were mixed breed cats, 389 (15%) were purebred, and breed was not reported for one diabetic cat. There were 575 diabetic cats (22.3%) for which weight was not reported. Most diabetic cats (94.7% of diabetic cats, for which weight was reported) weighed between 2.27 and 13.64 kg (Table 1).

Table 1.

Summary of characteristics of all 2576 cats with diabetes mellitus from the Veterinary Medical Data Base from January 1, 1970 through December 31, 1999

Characteristic	Number of cats	Percent of cats for which characteristic was reported
Weight (kg)
≤2.27	86	4.3
2.27–6.82	1640	82
6.82–13.64	254	12.7
≥13.64	21	1
Not reported	575	NA *
Neuter status
Male
Intact	201	12.5
Neutered	1411	87.5
Female
Intact	123	12.9
Neutered	832	87.1
Not reported	9	NA
Age (years)
≤1	32	1.3
1–2	34	1.3
2–4	83	3.3
4–7	288	11.3
7–10	570	22.5
10–15	1220	48.1
15+	311	12.2
Not reported	38	NA
Breed
Mixed	2186	84.9
Purebred	389	15.1
Not reported	1	NA

^*NA = not applicable.

Time trends

The hospital prevalence of DM in cats during this time period increased from eight cases per 10,000 in 1970 to 124 per 10,000 in 1999 (P < 0.001) (Fig 1). The age specific hospital prevalence increased significantly from 1970 to 1999 in all age groups (P < 0.001) (Fig 2). Most (82%) diabetic cats in this study were ≥7 years old. Five hundred and seventy cats (22%) were in the 7–10 year age group, 1220 (47%) cats were in the 10- to 15-year age group, and 311 (12%) cats were in the 15+ year age group. There were 38 cats for which age was not recorded. The case fatality percent decreased from 40% in 1970 to 10% in 1999 (P < 0.001) (Fig 3). There were no apparent seasonal trends in the prevalence of DM (P > 0.3).

Fig 1.

Hospital prevalence of feline diabetes mellitus (DM), VMDB 1970–1999. Prevalence of DM increased during the 30-year study period. χ ² test for linear trend = 1378.051; P < 0.00001; N = 2576.

Fig 2.

Age specific hospital prevalence of cats with diabetes mellitus during 1970–1999. Hospital prevalence increased in all age groups. Age information is missing for 38 cats. χ ² analysis P < 0.0001 for all groups except group 1–2 years (P = 0.0017); N = 2576.

Fig 3.

Case fatality percentage for feline diabetes mellitus (DM), VMDB, 1970–1999. The case fatality percent for cats with DM decreased during the 30-year study period. χ ² test for linear trend = 46.06; P < 0.001; N = 2576.

Risk factors

Risk factor analysis was performed using VMDB records of cats diagnosed with DM from January 1, 1990 through December 31, 1999. The records of 1526 cats diagnosed with DM (cases) and 1539 cats with diagnoses other than DM from the same institutions seen in the same years as DM cases (controls) were used for risk factor analysis. Numbers of cases and controls are unequal because cats weighing >16.34 kg (30 lb) were excluded from the analysis after cases and controls were selected. The excluded cases were considered most likely to be non-domestic cats based on their weight of >16.34 kg.

Univariate analysis indicated that increasing age, increasing weight, being male and being neutered were significantly associated (P < 0.001) with a diagnosis of DM (Table 2). Purebred cats were at decreased risk for DM compared with mixed breed cats (P = 0.002). When multivariable analysis was performed with male and female cats analyzed in the same model, male cats were found to be at significantly increased risk for DM, compared to female cats (P < 0.001). Significant interactions between gender and breed (P = 0.06) and gender and weight P ≤ 0.01) were observed. Therefore, multivariable risk factor analysis was also performed separately for female and male cats (Table 3). Increasing age was identified as a significant risk factor for DM for both genders. Neutered cats were not at significantly increased risk of DM compared with intact cats (male cats P = 0.21, female cats P = 0.56). The risk for DM was significantly decreased for female purebred cats (P = 0.006) compared with female mixed breed cats, but not for male purebred cats compared with male mixed breed cats (P = 0.88). For male cats, the risk of DM increased significantly with increasing weight (P < 0.001). In contrast, no significant increase in risk of DM was observed with increasing weight for female cats (P = 0.27).

Table 2.

Summary of univariate analyses of risk factors

Risk factor	Cases * (n = 1526)	Controls (n = 1539)	Odds ratio	95% CI †	P value
Weight (kg)
≤2.27	55	168	1.00	RG ‡	–
2.27–6.82	911	804	3.46	2.52, 4.76	<0.001
6.82–13.64	181	113	4.89	3.33, 7.19	<0.001
Gender §
Female	557	722	1.00	RG	–
Male	968	802	1.573	1.36, 1.82	<0.001
Neuter status
Male
Intact	82	172	1.00	RG	–
Neutered	886	630	2.95	2.22, 3.91	<0.001
Female
Intact	47	164	1.00	RG	–
Neutered	510	558	3.17	2.24, 4.48	<0.001
Age (years)
≤1	18	312	1.00	RG	–
1–2	18	125	2.49	1.26, 4.95	0.009
2–4	40	205	3.38	1.89, 6.06	<0.001
4–7	149	240	10.75	6.41, 18.03	<0.001
7–10	330	203	28.15	16.97, 46.70	<0.001
10–15	752	292	44.60	27.22, 39.46	<0.001
15+	219	162	23.41	13.97, 39.46	<0.001
Breed
Mixed	1344	1296	1.00	RG	–
Pure	182	243	0.72	0.59, 0.89	0.002
Abyssinian	7	21	0.32	0.14, 0.76	0.01
Himalayan	19	39	0.47	0.27, 0.82	0.008
Maine Coon	11	9	1.18	0.48, 2.85	0.72
Persian	18	61	0.28	0.17, 0.48	<0.001
Siamese	89	69	1.24	0.90, 1.72	0.19
Season
Fall	380	378	0.00	RG	–
Winter	352	373	0.94	0.77, 1.15	0.54
Spring	368	406	0.90	0.74, 1.10	0.31
Summer	426	382	1.11	0.91, 1.35	0.31

Data from the VMBD 1/1/90–12/31/99.

^*The numbers of cases and controls are not equal due to the elimination of cats >13.64 kg.

^†CI = confidence interval.

^‡RG = reference group.

^§Gender not reported for one diabetic cat.

Table 3.

Summary of potential risk factors for diabetes mellitus in multivariable analysis using logistic regression

Risk factor	Odds ratio	95% CI *	P value	Odds ratio	95% CI	P value
Male	Male	Male	Female	Female	Female
Weight (kg)
≤2.27	1.00	RG †	–	1.00	RG	–
2.27–6.82	2.74	1.55, 4.82	<0.001	0.72	0.40, 1.29	0.27
6.82–13.64	3.10	1.64, 5.86	<0.001	0.88	0.42, 1.83	0.73
Neuter status
Intact	1.00	RG	–	1.00	RG	–
Neutered	1.36	0.86, 2.15	0.19	1.18	0.70, 2.01	0.54
Age (years)
≤1	1.00	RG	–	1.00	RG	–
1–2	1.57	0.55, 4.50	0.40	3.96	0.99, 15.86	0.052
2–4	2.16	0.88, 5.34	0.09	5.36	1.55, 18.65	0.008
4–7	7.13	3.20, 15.88	<0.001	19.16	6.27, 58.56	<0.001
7–10	22.14	9.97, 49.17	<0.001	40.40	13.27, 122.94	<0.001
10–15	33.93	15.56, 73.97	<0.001	74.77	25.01, 223.50	<0.001
15+	17.81	7.74, 40.98	<0.001	43.82	14.44, 132.95	<0.001
Breed
Mixed	1.00	RG	–	1.00	RG	–
Pure	0.97	0.65, 1.45	0.88	0.56	0.37, 0.84	0.006

Data from the VMBD 1/1/90–12/31/99.

^*CI = confidence interval.

^†RG = reference group.

Discussion

The hospital prevalence of feline DM at veterinary teaching hospitals increased significantly from 1970 through 1999. This may reflect an increased willingness of pet owners over the 30 years studied to seek advanced veterinary care. Because cats in the VMBD are mostly referred from private veterinary hospitals, these data may also reflect changes in referral patterns. In addition, the prevalence of some potential risk factors for DM, such as obesity, may have increased over this same time period. Results of a 1973 survey of 429 cats in the United Kingdom showed that 9–12% of cats were overweight (Anderson 1973). Twenty-one years later, results of a multi-institution survey conducted in the United States showed that 25% of 2024 cats were overweight and 5% were obese (Scarlett et al 1994). This may reflect an increasingly sedentary lifestyle of pet cats; decreasing activity was significantly associated with obesity in cats (Scarlett et al 1994). It is also possible that cats in the United Kingdom are more active or are offered less food than cats in the United States. The observed increase in hospital prevalence of DM may also be related to changes in nutrient intake. Production and sales of cat foods in the United States began in the 1930s and greatly expanded in the 1960s and 1970s (Case et al 2000). With this expansion, nutrient intake from commercial diets increased among pet cats. The nutrient proportions in commercial cat foods ingested by cats can differ compared with the nutrient proportions ingested by cats subsisting on prey. The composition of expected prey during the evolution of domestic cats includes small rodents which are composed of 64–76% water, 14–18% protein, 6–18% fat, 1–5% minerals and minimal amounts of carbohydrate (Vondruska 1987, Kane 1989). There is marked variability in nutrient proportions among commercial cat foods, for example dry extruded foods typically have 25–60% of calories from carbohydrates, while canned foods can have from <1% to approximately 35% of calories from carbohydrate (Debraekeleer 2000). As obligate carnivores, cats reportedly have a lesser ability to incorporate glucose into glycogen, and longer glucose elimination times in glucose tolerance tests, than do omnivores (Thiess et al 2004). As cats are fed more commercially prepared foods, containing substantial proportions of fats or carbohydrates compared with the proportion of protein, their risk of DM may increase (Rand et al 2004). Farrow et al (2002) found that cats fed a high carbohydrate diet had significantly higher blood glucose concentrations than cats fed high protein or high fat diets. It has also been noted that diabetic cats fed a low carbohydrate diet had improved glycemic control based on clinical signs and fructosamine levels (Bennett et al 2006). However, results of another recent study showed normal glucose tolerance test results in cats fed high carbohydrate or high fat diets (Thiess et al 2004). It is difficult to directly compare results of nutritional studies because conditions, including diet compositions and signalment of cats, differ among studies. Further long-term studies are necessary to clarify the effects of diet on carbohydrate and lipid metabolism of cats and on the development of DM in cats.

The decreased case fatality percent from 40% to 10% suggests that more cat owners and veterinarians are now willing to undertake long-term management of diabetic cats rather than electing euthanasia. This may reflect a better understanding by cat owners and veterinarians of treatment options for DM, including recent advances in dietary management for better glycemic control, resulting in fewer complications (Feldman and Nelson 2004). Increased awareness by veterinarians of the importance of managing concurrent problems such as infections, dental disease, and hyperthyroidism also may contribute to the increased success of long-term management of DM, thereby, increasing willingness to treat (Feldman and Nelson 2004). Increased screening of middle aged and geriatric cats through yearly blood chemistry profiles and urinalysis may result in earlier detection of DM in some cats. Earlier detection and intervention may also result in a decrease in case fatality percent.

No seasonal pattern of prevalence of feline DM was detected in this study. The VMDB data do not include the date of onset of clinical signs, and lack of such information might mask a true seasonal pattern of DM in cats. However, if time between initial diagnosis of DM and presentation of a cat with DM to a referral institution is a random event, then a strong seasonal pattern might have been detected in this large sample of cats with DM. Previous epidemiological studies of DM in dogs also failed to show a seasonal pattern of DM (Marmor et al 1982, Guptill et al 2003), in contrast to reports for human beings with DM. More people are diagnosed with DM in the colder months (Unger and Foster 1998). The reported seasonal pattern observed in human beings prompted evaluation of the VMDB data for evidence of a seasonal pattern of DM in cats.

Two previous retrospective case series reported that the mean age at diagnosis of DM in cats was >10 years of age (Crenshaw and Peterson 1996, Goossens et al 1998). An epidemiological study of 333 diabetic cats found that cats >10 years old were at significantly increased risk for DM compared with cats <7 years old (Panciera et al 1990). In the study reported here, cats ≥7 years of age were at the highest risk of having DM compared with cats in the reference category (≤1 year of age), and the magnitude of risk increased with increasing age group. As age increases, however, so does the incidence of other conditions that cause insulin resistance such as obesity, infection, hyperadrenocorticism, and acromegaly (Crenshaw and Peterson 1996, Duesberg and Peterson 1997, Feldman and Nelson 2000, 2004). Older cats with these conditions may be more likely to develop DM because they cannot adequately respond to increased insulin requirements and maintain glucose homeostasis.

Male cats were at increased risk for having DM compared with female cats. Gender differences in weight gain and insulin sensitivity may explain why male cats appear to have a greater risk of developing DM than do female cats. Although increased weight was previously reported to be risk factors for DM in cats of both genders, increased weight in the current study was a significant risk factor for DM only in male cats (Panciera et al 1990). A recent study showed that male cats were more likely to gain weight than female cats (Appleton et al 2001). In addition, male cats of any weight had higher basal insulin concentrations and lower insulin sensitivity compared with female cats, suggesting that males may be naturally more insulin resistant than female cats. Insulin sensitivity decreased and insulin concentrations increased in male cats as they gained weight, and compared with females, male cats gained more weight and had a greater fat mass (Appleton et al 2001). Male cats accumulate greater amounts of fat than female cats due to increased glucose oxidation, glycogenesis and lipogenesis in response to insulin compared with female cats (Hoenig et al 2006). These differences in metabolism may, in part, explain the greater risk of male cats for developing DM compared with female cats.

Obese cats develop insulin resistance regardless of gender. However, male cats have a greater propensity for weight gain compared with female cats (Nelson et al 1990, Appleton et al 2001, Rand and Appleton 2001). Obesity could not be evaluated as a potential risk factor in the present study because body condition is not reported in VMDB records. Obesity is dependent not only on weight, but also body size and fat deposition. High body condition scores (BCS) are used as an indicator of obesity, with a score of >4 (using a scale of 1–5) consistent with obesity (Burkholder and Toll 2000). A recent review of cat medical records at Purdue University showed that a good correlation exists between cat body weight and BCS (using a 1–5 scale) across all ages, genders, and weight classes (Guptill L, Glickman L, personal communication, January 2006). For example, for male cats, weight of <5 lb (2.27 kg) had a mean BCS of 2.6 ± 0.5, weight of 5–15 lb (2.27–6.82 kg) had a mean BCS of 2.9 ± 0.6; weight of ≥15 lb (6.82 kg) had a mean BCS of 4.2 ± 0.8. For female cats, weight of <5 lb (2.27 kg) had a mean BCS of 2.67 ± 0.6; weight of 5–15 lb (2.27–6.82 kg) had a mean BCS of 3 ± 0.6, and weight of ≥15 lb (6.82 kg) had a BCS of 4.1 ± 0.7. These data suggest that the cats in the present study that weighed ≥15 lb (6.82 kg) were most likely obese and lend support to the hypothesis that obesity is a risk factor for DM in cats. It is possible that male cats are more likely than females to develop a form of DM similar to type II DM of human beings which has been linked to obesity.

There was no significantly increased risk for DM for neutered cats of either gender in the multivariable analysis. This finding is at odds with a previous risk factor analysis (Panciera et al 1990). However, the previous study differed with respect to sample size and the age groups used. The present study evaluated risk factors for 1526 cats with DM while the previous study evaluated only 333 cats (Panciera et al 1990). Being neutered is highly associated with age. In the previous study, cats were grouped into only three age groups (<7 years, 7–10 years and >10 years), whereas seven age groups were used in the current study. As a result of stratifying age into additional categories, the analysis in the present study provided better control for confounding by age, of the relationship between neutering and the risk of developing DM. The report by Fettman et al (1997), that glucose tolerance did not change after cats were neutered, is consistent with the finding that neutered cats are not at increased risk for developing DM. Similarly, Hoenig and Ferguson (2002) reported that glucose tolerance test results in cats kept at the same weight before and after neutering were unchanged and Thiess et al (2004) reported similar glucose tolerance test results for neutered and sexually intact male cats.

Univariate analysis suggested that certain purebred cats were at decreased risk for DM compared with mixed breed cats. However, after correcting for interaction between gender and breed, being a purebred was only protective for female cats. An increased risk of DM was reported for Burmese cats in Australia (Rand et al 1997, Baral et al 2003) and in New Zealand (Wade et al 1999). Rand et al (1997) found 20% of 45 diabetic cats were Burmese cats, which was significantly higher (P < 0.001) than the proportion of Burmese cats in a population of normoglycemic cats (Rand et al 1997). There were not enough Burmese cats in the present study to make any conclusions regarding their risk for DM.

As is the case for any epidemiological study, the results of this study are only strictly applicable to the population that was sampled, ie, pet cats presented to veterinary teaching hospitals. The results may not be applicable to all pet cats observed in primary care practices.

In conclusion, the hospital prevalence of DM in cats increased significantly over the 30 years evaluated, while the case fatality percent decreased significantly. Significant risk factors for DM in cats included increased age, male gender, increased weight for male cats, and mixed breed for female cats. Veterinarians should counsel owners of cats with these risk factors to make them aware of the importance of keeping their older male cat from becoming overweight, and to consider screening for DM in cats over the age of 7 years. Owners should also be made aware of the clinical signs associated with DM. Finally, the results of this study suggest for the first time that increased weight is a risk factor for DM for male but not female cats, and that being neutered is not by itself a risk factor for DM among male or female cats.