Are there long-term benefits to prescribing long-acting analogue insulin compared to human insulin?

Diabetes mellitus is the seventh leading cause of death in the United States. Approximately 25.8 million Americans are affected by diabetes, and total cost related to diabetes in 2007 was $174 billion dollars.

The progressive nature of diabetes requires many individuals to need insulin therapy to maintain glycemic control. Providers and patients can choose between long-acting ‘human' isophane insulin (NPH) and long-acting analogue insulins (e.g., glargine or detemir). No generic versions of the analogue insulins are available making them much more expensive compared to NPH. Analogue insulin may be cost-effective if it improves glycemic control or if it prevents diabetes complications compared to NPH. Previous studies have not found significant clinical differences between the two types of insulin in the short-term when focused on glycemic control or hypoglycemia. Ours is the first study to compare the effectiveness of NPH and long-acting insulin on long-term outcomes including mortality and preventable hospitalizations.

This is a retrospective observational study of US veterans who received a prescription for diabetes medication. To overcome the main limitation of observational studies that treatments are not randomly assigned, we use local variation in provider prescribing patterns as an instrumental variable and control for several measures of provider quality. Our quasi-experimental treatment was the proportion of days for which each patient had an analogue insulin prescription between the date they entered the study and a series of “snapshot dates” occurring 90 days, 1 year, 2 years, 3 years, 4 years, 5 years and 6 years after the index date. The instrumental variables approach estimates a pair of simultaneous equations: one for the amount of treatment before the snapshot date and the second for outcomes occurring after the snapshot date. Our treatment equations modeled the proportion of days prescribed analogue insulin as a function of the provider-level prescribing rate and control variables. Mortality and preventable hospitalizations were assessed using Cox proportional survival models.

There are no significant differences in the hazard ratios predicting mortality and preventable hospitalizations when comparing NPH and long-acting analogue insulin. Consequently, long-acting analogue insulins do not offer a significant clinical advantage in long-term health outcomes when compared to NPH. Previous research has not found a significant clinical advantage of long-acting analogue insulin compared to NPH for short-term outcomes including glycemic control. This lack of clinical difference in outcomes along with the high-cost of analogue insulin makes NPH the more cost-effective prescribing choice.