Of the 1,055 veterans in the initial sampling frame deemed appropriate for the study, 658 (62.4%) responded to letters of invitation to participate and 381 (57%) agreed to be contacted. Of these, 211 presented to VAPHS for signed informed consent, additional screening, and baseline measurements. The 150 consenting veterans who had a capillary A1C ≥7.5% at the baseline were randomly assigned to the ACM+HT (n
= 73) or CC (n
= 77) groups. Of these, 3 ACM+HT and 2 CC participants were excluded because they were subsequently found to meet baseline exclusion criteria; 2 CC participants withdrew before the initial education session and 6 ACM+HT participants withdrew afterward. This analysis includes the remaining 64 ACM+HT and 73 CC participants (supplementary Table 4A, available in an online appendix at http://care.diabetesjournals.org/cgi/content/full/dc09-1012/DC1
All participants completed the baseline assessment; 6 ACM+HT and 4 CC participants missed the 3-month assessment and 8 ACM+HT and 7 CC participants missed the 6-month assessment. A total of 8 A1C values in the ACM+HT group and 9 A1C values in the CC group were missing, and 10 A1C values were truncated.
Baseline patient characteristics
There were no significant differences by treatment group for age, sex, race, or any of the other baseline characteristics as shown in supplementary Table A1. Approximately one-third of the participants in both groups were aged ≥65 years; the vast majority were male and non-Hispanic white. The predominant comorbidities were coronary artery disease and congestive heart failure.
Most participants in each group were taking oral hypoglycemic agents (predominantly glyburide and metformin) and antihypertensive and lipid-lowering medications at baseline, 3 months, and 6 months; >50% were using insulin (supplementary Table A2). There were no significant differences by medication class at any time point (P > 0.14 for each). By 6 months, ACM+HT participants had significantly more medication or dose changes on average involving antihypertensive agents (3.1 for ACM+HT vs. 1.9 for CC participants, P = 0.02), but not lipid-lowering agents (1.4 for ACM+HT vs. 1.1 for CC participants, P = 0.29) or oral hypoglycemic agents (1.8 for ACM+HT vs. 1.8 for CC participants, P = 0.91).
At baseline, 39 ACM+HT and 40 CC participants were using insulin. By 6 months, 1 ACM+HT and 1 CC participant had discontinued insulin, whereas 5 ACM+HT and 3 CC participants had begun insulin. Although the average daily insulin dose was similar in both groups at baseline, the average daily dose for ACM+HT participants was ~18 IU higher than that for CC participants at 3 and 6 months (P = 0.02 and P = 0.048, respectively). The average number of adjustments in insulin dose was also higher in ACM+HT (6.6) than in CC (2.8) participants (P < 0.001). However, no significant correlation was found between the frequency of insulin adjustment and A1C at 6 months in either ACM+HT (r = 0.12; P = 0.43) or CC (r = 0.14; P = 038) participants.
Dotplots of individual values for A1C, weight, blood pressure, and lipids are shown by treatment group for each time point in . Baseline values were similar for both groups (P > 0.45 for each) (). A1C was significantly lower for ACM+HT than for CC participants at both 3 and 6 months (0.7% lower at each time point, P < 0.001 for each). Significantly greater decreases in A1C were observed in the ACM+HT group relative to the CC group at 3 months (1.7 vs. 0.7%) and 6 months (1.7 vs. 0.8%), corresponding to differential decreases of ~0.9% (P < 0.001 for each) (supplementary Table A3). There was no significant interaction between baseline insulin usage and treatment response at any time point (P > 0.39 for each) (supplementary Fig. 1).
Figure 1 Dot plots of the primary outcome measures (A1C, weight, systolic blood pressure, diastolic blood pressure, cholesterol, HDL, LDL, and triglycerides) at baseline, 3 months, and 6 months by treatment group. ●, ACM + HT group; ○, CC group. (more ...)
Time-specific primary outcomes by treatment group
None of the other primary outcomes differed significantly by treatment group at either 3 or 6 months (). However, except for weight and HDL cholesterol levels, the direction of the differences favored the ACM+HT group. Within both treatment groups, A1C, blood pressure, cholesterol, and LDL improved significantly at 3 and 6 months relative to baseline, whereas HDL decreased (supplementary Table A3). Triglycerides declined significantly from baseline only in the ACM+HT group. A 4-lb mean weight increase in the ACM+HT group was the only significant within-group change between 3 and 6 months.
Similar proportions of ACM+HT and CC participants had A1C levels <8 or <9% at baseline (). However, at 6 months, 20.3% of ACM+HT and 5.5% of CC participants achieved A1C <7% (P = 0.01). Significantly more ACM+HT than CC participants also reached A1C levels of <8 and <9% at both 3 and 6 months (P ≤ 0.03 for each). Less than half of the participants had systolic blood pressure ≤130 mmHg at any time point, whereas a majority met the targets for diastolic blood pressure, LDL, and triglycerides. A higher percentage of ACM+HT than CC participants met the LDL treatment target of <100 mg/dl at 6 months (79.7 vs. 59.4%, respectively; P = 0.02).
Number of participants achieving each identified clinical target at baseline, 3 months, and 6 months by treatment group
SMBG among ACM+HT participants
Seven ACM+HT participants (10.9%) never transmitted any SMBG data after initial training. Another 9 participants (14.1%) performed SMBG on average <1 time per day, whereas 75.0% performed SMBG between 1 and 4 times per day (average 2.3 times daily) during the period in which they transmitted measurements. Among the 57 participants who transmitted measurements, 35 (61.4%) transmitted SMBG <50 mg/dl on at least 1 day (median 1 day) and 16 (28.1%) transmitted SMBG between 50 and 70 mg/dl (median 10 days). Within the ACM+HT group, the frequency of SMBG did not correlate significantly with reduction in A1C (r = −0.11; P = 0.39).
Nurse-to-participant telephone contact time was substantially greater in ACM+HT than CC participants (~1.3 vs. 0.3 h/participant/month, respectively). In the ACM+HT group, telephone contact was triggered by transmitted suboptimal SMBG or blood pressure levels. Thus, contact time was disproportionately high in this subgroup of ACM+HT participants.
One participant in the ACM+HT group died at home suddenly 7 months after entry in the study. No postmortem examination was obtained. The participant had diabetic neuropathy, stage IV renal insufficiency, and congestive heart failure for which he had been hospitalized recently. He was treated with insulin alone and thus was taking no oral hypoglycemic agent that may have complicated his heart failure. His A1C levels fell from 11.8% at baseline to 6.5% at 3 months. Of his 468 transmitted SMBG values, 4 (0.85%) were <50 mg/dl, a frequency similar to that of ACM+HT participants overall (0.66%) and that (0.59%) of 7 other ACM+HT participants using insulin who had rapid declines in A1C (≥3%) over 3 months. No SMBG <50 mg/dl occurred during the month before his death. The relationships, if any, between the rapid decline in A1C, hypoglycemia, and sudden death in this patient are uncertain.