Materials and Methods

Selection of cats—Seven cats with diabetes mellitus that ranged in age from 7 to 14 years (median, 12 years) and weighed 4.3 to 7.5 kg (9.5 to 16.5 lb; median, 6.1 kg [13.4 lb]) were included in the study. Five cats were castrated males, and 2 were spayed females; breeds included domestic shorthair (n = 3), Persian (1), Siamese (1), Burmese (1), and Carthusian (1). The diagnosis of diabetes mellitus was made on the basis of characteristic clinical signs, hyperglycemia after withholding of food (fasting), glucosuria, and high serum fructosamine concentration (> 340 μmol/L). Cats were included in the study if owners were willing to learn home monitoring, return cats to our clinic for reevaluations, and generate BGCs at home. Diabetes mellitus had been diagnosed in all cats 85 to 690 days (mean, 252 days) prior to enrollment in the study. Treatment consisted of SC injection of a porcine intermediate-acting insulin^a every 12 hours. The insulin dosage and quality of glycemic control varied among cats during the study.

Home monitoring—The concept of home monitoring was introduced to owners at the time of diagnosis or at the first evaluation at the clinic. One week after the initial evaluation, cats were reevaluated at the clinic and home monitoring was again discussed. At a second reevaluation 3 weeks later, owners were given the opportunity to learn the technique. This usually took a minimum of 30 minutes and consisted of repeated demonstrations of use of the lancing device and PBGM. Owners performed the procedure once or twice on their cats and were also taught how to calibrate the PBGM, check its accuracy with a control strip, and record blood glucose concentrations on prepared forms. In addition, owners received written and illustrated instructions for measuring blood glucose and generating BGCs. After instruction in the clinic, owners were asked to generate a BGC at least once every 3 to 4 weeks from that date. They were also asked to generate additional BGCs for the requirements of this study.

Capillary blood was collected from the ear according to a described method^2,4,6,12 by use of a lancing device,^b and blood glucose concentrations were measured with a commercially available PBGM.^c All owners were able to perform home monitoring without difficulties.

Study design—The study consisted of 2 parts. In part 1, cat owners generated a BGC on each of 2 consecutive days at home. Within 7 days of completion of the home BGCs, cats were admitted to the clinic, where a third BGC was generated. For all BGCs, blood glucose was measured every 2 hours for 12 hours, starting immediately before the morning insulin injection (total, 7 measurements/d). The insulin dose, time of insulin administration, and feeding regimen were constant for each cat in all 3 BGCs.

In part 2, the same protocol (2 BGCs at home and 1 BGC in the clinic) was repeated a minimum of 4 weeks after completion of part 1. The insulin dose, time of insulin administration, and feeding regimen were constant for each cat in all 3 BGCs; however, they were not necessarily the same as in part 1. Collection of capillary blood and measurement of blood glucose concentration at home and in the clinic were performed with the same lancing device and PBGM. Blood glucose curves generated at home were referred to as home curves, and those generated in the clinic were referred to as clinic curves.

Besides generating BGCs, clinical reevaluations included a detailed and updated history; physical examination; and measurement of Hct and serum concentrations of fructosamine, albumin, and total protein. Serum fructosamine analyses were performed by use of an automated analyzer^d and commercial reagents supplied by the manufacturer. Cats were allocated to 2 groups on the basis of glycemic control. Group A consisted of cats that were considered to have good glycemic control; there was resolution or marked improvement in clinical signs, and serum fructosamine concentration was ≤ 500 μmol/L. Group B consisted of cats that were considered to have moderate to poor glycemic control; there was persistence or little improvement in clinical signs, and serum fructosamine concentration was > 500 μmol/L.

Statistical analysis—For each BGC, 7 variables were determined: blood glucose concentration determined after fasting, nadir glucose concentration, time to nadir concentration, maximum blood glucose concentration, mean blood glucose concentration during 12 hours, difference between fasting blood glucose concentration and nadir concentration (fasting-nadir), and area under the blood glucose curve. For parts 1 and 2, differences between the 7 variables were calculated for the following 3 pairs of curves: the 2 home curves, first home curve and clinic curve, and second home curve and clinic curve. For each of the 6 pairs of curves, the mean, SD, and coefficient of variation were calculated for the differences between the corresponding variables. A normality test performed by use of statistical software^e revealed no significant difference between the values and a normal distribution; therefore, parametric tests were used. A paired t test and 1-way ANOVA were used to analyze differences between values obtained in parts 1 and 2, between values of pairs of home curves, and between values of pairs of a home curve and clinic curve.^f To identify significant sources of variation in the paired curves, a factorial ANOVA was performed, followed by a Bonferroni-Dunn post hoc test.^e Differences were considered significant at values of P ≤ 0.05. Scatterplots^g were used for graphical presentation of the data, with a horizontal line indicating the mean. To examine possible clinical implications of day-to-day variations, a theoretical recommendation was made for adjustment of the insulin dose, which was based on the results of each BGC. The recommendation was to increase or decrease the insulin dosage or to leave it unchanged when the nadir was ≥ 9.0 μmol/L, < 5.0 μmol/L, or 5.0 to 8.9 mmol/ L, respectively.^7,13

To determine possible causes of variability in the BGCs, the mean and nadir blood glucose concentration of each home curve were compared with the corresponding clinic curve in each cat. Thus, on the basis of the mean and nadir concentrations, values on the clinic curve could be higher or lower than one or both of the home curves. The same criteria were used to compare the home curves.

Results

Comparison of differences between home and clinic curves—Differences among the 7 variables from BGCs obtained on 3 days (2 home curves and 1 clinic curve) would be small if there was minimal day-to-day variability of the BGCs. However, differences between the variables were large for all pairs of BGCs. This was reflected by SDs of the differences that were nearly as large as the means and by large coefficients of variation that ranged from 69% to 101% (Table 1). To test our hypothesis that there would be better agreement between paired BGCs generated at home than between curves generated at home and those generated in the clinic, differences between variables of the home curves were compared with differences between variables of the home and clinic curves. There were no significant differences, indicating that agreement between the home curves was not better than between the home and clinic curves and that the hypothesis was to be rejected. Comparison of the differences between the paired home curves generated in parts 1 and 2 of the study revealed that the difference between the maximum blood glucose concentration in part 2 was significantly (P = 0.045) higher than that in part 1; no other significant differences were detected (Figure 1). There were also no significant differences between parts 1 and 2 in comparison to the home and clinic curves or fructosamine concentrations.

View Full Size

Diagram depicting the differences between values for fasting (food withheld) blood glucose concentrations (Diff fasting), nadir concentrations (Diff nadir), maximum blood glucose concentration (Diff max), fasting-nadir concentrations (Diff fast-nadir), and mean blood glucose concentrations (Diff mean) obtained from paired home BGCs in 7 cats with diabetes mellitus. Seven paired BGCs were generated for each cat on 2 occasions (parts 1 and 2). Notice that the difference between maximum blood glucose concentrations in the 2 BGCs were significantly higher in the part 2 curves. Horizontal bars indicate the mean. 1, 2 = Parts 1 and 2, respectively.

Citation: Journal of the American Veterinary Medical Association 230, 7; 10.2460/javma.230.7.1011

• Download Figure
• Download figure as PowerPoint slide

Comparison of home and clinic curves—For these comparisons, absolute values of the BGC variables were used. The results of parts 1 and 2 were considered together because corresponding BGCs of the 2 parts (eg, first home curve of part 1 and first home curve of part 2) did not differ with respect to any of the 7 variables. Values for fasting blood glucose concentration, maximum blood glucose concentration, mean blood glucose concentration, and area under the blood glucose curve were significantly lower in the first home curve than in the second home curve (Figure 2). The fasting blood glucose and maximum blood glucose concentrations of the first home curve were significantly lower than those of the clinic curve. Values for nadir concentration, fasting-nadir concentration, and time from insulin injection to nadir concentration did not differ significantly between the first and second home curve. The nadir concentration, fasting-nadir concentration, time from insulin injection to nadir concentration, mean blood glucose concentration, and area under the curve did not differ significantly between the first home curve and clinic curve. There were no significant differences between the second home curve and clinic curve with respect to any of the 7 variables.

View Full Size

Diagram depicting fasting blood glucose concentration, nadir concentration, maximum blood glucose concentration, fasting nadir concentration, and mean blood glucose concentration in home curves generated on 2 consecutive days in the same 7 cats as in Figure 1. Results of parts 1 and 2 of the study were combined, yielding 14 values per variable and day of measurement. Notice that fasting blood glucose concentration, maximum blood glucose concentration, and mean blood glucose concentration on the second day (h2) were significantly higher than on the first day (h1). See Figure 1 for key.

Citation: Journal of the American Veterinary Medical Association 230, 7; 10.2460/javma.230.7.1011

• Download Figure
• Download figure as PowerPoint slide

Theoretical recommendations for adjustment of insulin dose—Fourteen sets of paired home curves (7 from part 1 and 7 from part 2) were compared to determine a theoretical recommendation for adjustment of the insulin dose. For 6 of the 14 paired curves, the same recommendations for adjustment of insulin dose resulted, whereas theoretical recommendations were different for 8 of the 14 paired curves. In 6 of the latter 8 paired curves, there would have been no dose adjustment on the basis of results of one of the home curves, and there would have been an increase or decrease in the insulin dose on the basis of results of the other home curve. In the remaining 2 paired curves, an opposite theoretical recommendation for insulin dose adjustment resulted; assessment of the results of one of the home curves led to a theoretical recommendation for a reduction in the insulin dose, whereas results of the other home curve led to a recommendation for an increase in the dose. Twenty-eight curve comparisons were made between home and clinic curves. Evaluation of results of the home and clinic curves led to the same recommendation for adjustment of insulin dose in 14 instances. In the other 14 instances, recommendations differed: in 7 instances, results of the home curves indicated no adjustment and an increase or decrease in the insulin dose would have been recommended on the basis of results of the clinic curves. In 1 instance, no adjustment of the insulin dose was made on the basis of the clinic curve, whereas a reduction in the insulin dose was recommended on the basis of the home curve. Evaluation of 6 paired curves led to opposite recommendations; in 3 instances, an increase in insulin dose was recommended on the basis of the clinic curve and a decrease in insulin dose was recommended on the basis of the home curve. In the other 3 instances, a decrease in the insulin dose was recommended on the basis of the clinic curve and an increase was recommended on the basis of the home curve.

Quality of glycemic control—Two cats in parts 1 and 2, 1 cat in part 1, and 1 cat in part 2 (different cats) were considered to have good glycemic control. Thus, when the home curves of parts 1 and 2 were considered together, there were 6 home curve pairs from cats with good glycemic control (group A) and 8 from cats with moderate to poor glycemic control (group B). In group A, 4 of the 6 home curve pairs led to the same theoretical recommendations for insulin dose adjustment and the remaining 2 home curve pairs led to different recommendations. In group B, the same recommendation for insulin dose adjustment was made in only 2 of 8 home curve pairs; a different recommendation was made in the other 6 instances. Of the 28 home curve and clinic curve pairs, 12 were from cats of group A and 16 were from cats of group B. In group A, 8 of the 12 home and clinic curve pairs and in group B, 6 of the 16 home and clinic curve pairs led to identical recommendations for insulin dose adjustment. There were no significant differences between the 7 variables obtained from the home curves in groups A and B; however, differences between all of the variables in group A tended to be smaller than those in group B (Figure 3).

View Full Size

Diagram depicting the differences between fasting blood glucose concentration, nadir concentration, maximum blood glucose concentration, fasting-nadir concentration, and mean blood glucose concentration in BGCs generated at home on 2 consecutive days in the same cats as in the preceding figures. Six pairs of curves were obtained from cats with good glycemic control (group A), and 8 pairs of curves were obtained from cats with moderate to poor glycemic control (group B). Notice that for all variables, although the differences were typically lower in cats with good glycemic control, the differences were not significant. See Figure 1 for key.

Citation: Journal of the American Veterinary Medical Association 230, 7; 10.2460/javma.230.7.1011

• Download Figure
• Download figure as PowerPoint slide

Comparison of individual blood glucose curves—Regarding comparison of home and clinic curves, the clinic curves in 2 cats were higher than the home curves in both parts of the study. In 1 other cat, the clinic curve was considerably higher than the first home curve but corresponded to the second home curve. The mean and nadir concentrations of the clinic curves were considerably lower than those of the home curves in both parts of the study in 1 cat and in 1 part of the study in another cat. In the remaining 2 cats, the clinic curves corresponded to those home curves with the lower mean and nadir concentration in both parts of the study.

Regarding comparison of the home curves, the second home curve in 3 cats was much higher than the first; the difference was even more pronounced in part 2 of the study. In 1 cat, the second home curve was slightly higher than the first. In 2 cats, the home curves were the same in both study parts. In the remaining cat, the first home curve was slightly higher than the second curve in the first part of the study and the second home curve was slightly higher than the first curve in the second part.

Discussion

Results indicate that there is considerable day-today variability in BGCs in diabetic cats, even when factors such as insulin dose and meal size remain constant and the cat is at home in a stress-free environment. There was a large difference between the values of home curves obtained on 2 consecutive days. In particular, the maximum blood glucose concentration, time from insulin injection to nadir concentration, and fasting blood glucose concentration differed considerably between the 2 home curves. Contrary to our hypothesis, there was no greater agreement between the home curves than between the home and clinic curves. In humans with insulin-dependent diabetes mellitus, variations in blood glucose concentrations occur within a 24-hour period as well as from day to day and are associated primarily with activity level, meal composition and size, stress, and intake of certain medications.^9-11,14 However, studies^15–17 have revealed that even when these factors remain constant, day-to-day variability in glucose concentration persists. Causes include variable rate of insulin absorption when different injection sites are used, variation in the length of insulin activity, variation in insulin sensitivity among individuals, and variation in residual beta-cell function.

In humans with diabetes, consistent SC insulin injection in the abdominal region results in faster absorption and smaller dayto-day fluctuations in blood glucose concentrations than when the injection site is rotated.¹⁵ In the present study, owners varied the SC injection site regularly, which may have resulted in variable rates of insulin absorption. This aspect of administration has not been investigated in diabetic cats, to the authors' knowledge. The duration of action of the insulin^a used is approximately 12 hours, but may be shorter in some cats.¹⁸ The duration of insulin action was < 12 hours in all 3 BGCs generated during part 1 of the present study in 1 cat. Moreover, the blood glucose concentration returned to baseline after 8 to 10 hours, resulting in a marked difference in the morning fasting blood glucose concentration.

Because most cats have type 2 diabetes mellitus, the insulin dose depends on the severity of insulin resistance and the amount of residual beta-cell function.^19,20 The degree of insulin sensitivity varies diurnally and also varies from day to day in humans with diabetes.^16,17 Healthy cats also have substantial day-today variability in insulin sensitivity, and the same may also be true in diabetic cats, although, to the authors' knowledge, this has not been investigated.²¹

In addition to internal factors, reproducibility of the BGCs may have also been affected by external factors. The activity level of the cats was difficult to control, and it may have differed on the consecutive days of blood glucose measurement at home. Although the insulin dose remained constant, there may have been slight errors in the amount of insulin as drawn up by the owners.

In the individual comparison of home and clinic curves, different patterns were observed: in the 3 cats in which the nadir and mean blood glucose concentrations in the clinic curves were higher than those in 1 or both home curves, stress-induced hyperglycemia caused by hospitalization was suspected. In 2 cats, the mean and nadir concentrations in the clinic curves were lower than both home curves. In the remaining 2 cats, the clinic curves corresponded to the home curves with the lower mean and nadir. Three of those 4 cats did not eat or ate later in the day while they were hospitalized for generation of the clinic curves. It is possible that the lower blood glucose concentrations were attributable to reduced food intake while cats were in the clinic. In an earlier study,⁷ some variables of BGCs generated in the clinic in diabetic cats were significantly lower than those variables in curves generated at home, a finding that was thought to be a result of decreased food intake in the clinic.

Two of the 3 cats in which the second home curve was much higher than the first in both study parts also had noticeably high blood glucose concentrations in the clinic in 1 or both parts of the study, which was thought to be stress induced. It is plausible that in these cats, frequent pricking of the ear was stressful, even at home. These cats had been subjected to home monitoring for several months, but had not undergone 2 consecutive days of blood collection prior to the study. Information on whether the cats were cooperative during blood collection was not available. However, fractiousness is not the only indicator of stress; some cats can be anxious and stressed while remaining cooperative and quiet.²²

The fact that 3 of 7 cats had higher values in the second home curve than in the first led to significant differences in the statistical analysis: fasting blood glucose concentration, maximum blood glucose concentration, mean blood glucose concentration, and area under the curve were significantly higher on the second day of home testing than on the first, further supporting the likelihood of stress-induced hyperglycemia on the second day of home testing.

The nadir concentration, time to nadir concentration, and difference between fasting blood glucose concentration and nadir concentration are critical for evaluation of insulin doses. The theoretical recommendation for insulin dose adjustment in the present study was similar to that used in a recent review¹³ on cats with diabetes mellitus. Comparison of the home curves led to the same recommendation on only 6 of 14 occasions, and comparison of home and clinic curves resulted in the same recommendation on only 14 of 28 occasions. In an earlier study,²³ the reproducibility of BGCs in 10 diabetic dogs was investigated by comparing 3 pairs of BGCs generated on 2 consecutive days in the clinic. In that study, the theoretical recommendation for insulin dose adjustment was the same on 17 of 30 occasions. In addition, 20 sets of paired curves from dogs with good glycemic control (nadir concentration < 10 μmol/L) were compared, and the theoretical recommendation was the same in only 7 of those curves.²³ This led to the conclusion that BGCs vary greatly from day to day, particularly in dogs with good glycemic control. In contrast to the results of that study,²³ there was better agreement between the BGCs of cats with good glycemic control in the present study; 6 of 14 pairs of home curves led to the same theoretical adjustment of insulin dose, and 4 of those were curves from cats with better glycemic control. In cats with moderate to poor glycemic control, comparison of the home curves led to the same recommendation in only 2 of 8 paired curves. This suggests that there is less day-to-day variability in blood glucose concentrations in cats with good glycemic control. There was a nonsignificant finding that the differences between variables of the home curve pairs were smaller in cats with good glycemic control. A larger number of diabetic cats or paired BGCs may have yielded more explicit results. These results, however, are comparable to those found in humans; in a 1972 study,⁹ day-to-day variability in blood glucose concentration curves differed among healthy subjects and those who had good and poor glycemic control. Healthy humans had considerably less day-to-day variability in blood glucose concentrations than humans with diabetes mellitus, and human diabetics with poor glycemic control had significantly more variability than those with good glycemic control.⁹

Overall day-to-day variability in blood glucose concentration curves was substantial in the cats with diabetes mellitus in the present study. This finding is in agreement with those in dogs and humans with the disease. Stress associated with hospitalization was thought to be a cause of the high blood glucose concentrations in some cats, although blood glucose concentrations in the clinic may also have been lower than those at home because of low food intake. The day-to-day variability between the home and clinic curves was not larger than that between paired home curves. The reproducibility of home curves in diabetic cats with good glycemic control was somewhat better than that of cats with moderate to poor control. We therefore assume that BGCs generated at home in cats with good glycemic control are more reliable than those in cats with poorer glycemic control. Results also indicate that serial blood collection on 2 consecutive days may result in stress-related hyperglycemia in diabetic cats.