Experience with Automated Insulin Dose Adjustment based upon Computerized Formula in a Tertiary Care Diabetes Center

Background: With increasing burden of diabetes, existing system with limited trained medical personals especially in a developing country with poor health system will not be able to cope patient load. So an automated insulin dose adjustment based upon computerized algorithms might be a need of the hour. Objective: To determine the efficacy and safety of automated insulin dose adjustment model based upon computerized formula in diabetic out-patient clinic at a tertiary care center Material and Methods: This was case control study, conducted from (16 October 2020 to 15 May 2021) on 120 diabetic patients. Patients were attended thoroughly after addressing all ethical issues (described below) and prescription was given to them either generated by software or compiled by diabetes experienced fellow endocrinology. Then cases were followed after seven days for diabetes control. Results: Our study was conducted on 120 cases, 60 in each group. There was no significant difference in our outcome parameters i.e. Mean fasting blood sugar, mean 2-hours post lunch blood sugar and mean 2-hours post dinner blood sugar and episodes of hypoglycemia between the two groups. In Group-S, 1.6% (n=1) cases developed episodes of hypoglycemia and in Group-F, 1.6% (n=1) cases developed episodes of hypoglycemia (p=1.00). Mean fasting blood sugar was 121.95 ± 16.22 mg/dl in Group-S and was 121.60 ± 16.46 mg/dl in Group-F (p=0.91). Mean 2-hours post lunch blood sugar was 182.45 ± 36.43 mg/dl in Group-S and was 181.45 ± 36.44 mg/dl in Group-F (p=0.88). Mean 2-hours post dinner blood sugar was 182.32 ± 29.66 mg/dl in Group-S and was 180.31 ± 28.66 mg/dl in Group-F (p=0.71). Conclusion: So we concluded that use of automated insulin dose initiation and adjustment models based upon computerized algorithms are comparable to insulin dose initiation and adjustment by experienced physician. So our automated insulin dose initiation and adjustment model might help clinician at heavily burdened diabetic clinics. But we recommend it supervised use in such clinical settings