Short-acting insulin analogues versus regular human insulin for type 2 diabetes mellitus

Review question

Are short-acting insulin analogues better than regular human insulin for adult, non-pregnant people with type 2 diabetes?


Short-acting insulin analogues act more quickly than regular human insulin. They can be injected immediately before meals and lead to lower blood sugar levels after food intake. Whether people with diabetes really profit from these newer insulins is debated.

Study characteristics

We found 10 randomised controlled trials (clinical studies where people are randomly put into one of two or more treatment groups) comparing the short-acting insulin analogues insulin lispro, insulin aspart, or insulin glulisine to regular human insulin in 2751 participants. The people in the included trials were monitored (followed) for 24 to 104 weeks.

This evidence is up to date as of 31 October 2018.

Key results

We are uncertain whether short-acting insulin analogues are better than regular human insulin for long-term blood glucose control or for reducing the number of times blood sugar levels drop below normal (hypoglycaemic episodes). The studies were too short to reliably investigate death from any cause. We found no clear effect of insulin analogues on health-related quality of life. We found no information on late diabetes complications, such as problems with the eyes, kidneys, or feet. No study reported on socioeconomic effects, such as costs of the intervention and absence from work.

Certainty of the evidence

The overall certainty of the included studies was low or very low for most outcomes, mainly because all studies were carried out in an open-labelled fashion (study participants and study personnel knew who was getting which treatment). Several studies also showed inconsistencies in the reporting of methods, and results were imprecise.

Authors' conclusions: 

Our analysis found no clear benefits of short-acting insulin analogues over regular human insulin in people with type 2 diabetes. Overall, the certainty of the evidence was poor and results on patient-relevant outcomes, like all-cause mortality, microvascular or macrovascular complications and severe hypoglycaemic episodes were sparse. Long-term efficacy and safety data are needed to draw conclusions about the effects of short-acting insulin analogues on patient-relevant outcomes.

Read the full abstract...

The use of short-acting insulin analogues (insulin lispro, insulin aspart, insulin glulisine) for adult, non-pregnant people with type 2 diabetes is still controversial, as reflected in many scientific debates.


To assess the effects of short-acting insulin analogues compared to regular human insulin in adult, non-pregnant people with type 2 diabetes mellitus.

Search strategy: 

For this update we searched CENTRAL, MEDLINE, Embase, the WHO ICTRP Search Portal, and to 31 October 2018. We placed no restrictions on the language of publication.

Selection criteria: 

We included all randomised controlled trials with an intervention duration of at least 24 weeks that compared short-acting insulin analogues to regular human insulin in the treatment of people with type 2 diabetes, who were not pregnant.

Data collection and analysis: 

Two review authors independently extracted data and assessed the risk of bias. We assessed dichotomous outcomes by risk ratios (RR), and Peto odds ratios (POR), with 95% confidence intervals (CI). We assessed continuous outcomes by mean differences (MD) with 95% CI. We assessed trials for certainty of the evidence using the GRADE approach.

Main results: 

We identified 10 trials that fulfilled the inclusion criteria, randomising 2751 participants; 1388 participants were randomised to receive insulin analogues and 1363 participants to receive regular human insulin. The duration of the intervention ranged from 24 to 104 weeks, with a mean of about 41 weeks. The trial populations showed diversity in disease duration, and inclusion and exclusion criteria. None of the trials were blinded, so the risk of performance bias and detection bias, especially for subjective outcomes, such as hypoglycaemia, was high in nine of 10 trials from which we extracted data. Several trials showed inconsistencies in the reporting of methods and results.

None of the included trials defined all-cause mortality as a primary outcome. Six trials provided Information on the number of participants who died during the trial, with five deaths out of 1272 participants (0.4%) in the insulin analogue groups and three deaths out of 1247 participants (0.2%) in the regular human insulin groups (Peto OR 1.66, 95% CI 0.41 to 6.64; P = 0.48; moderate-certainty evidence). Six trials, with 2509 participants, assessed severe hypoglycaemia differently, therefore, we could not summarise the results with a meta-analysis. Overall, the incidence of severe hypoglycaemic events was low, and none of the trials showed a clear difference between the two intervention arms (low-certainty evidence).

The MD in glycosylated haemoglobin A1c (HbA1c) change was -0.03% (95% CI -0.16 to 0.09; P = 0.60; 9 trials, 2608 participants; low-certainty evidence). The 95% prediction ranged between -0.31% and 0.25%. The MD in the overall number of non-severe hypoglycaemic episodes per participant per month was 0.08 events (95% CI 0.00 to 0.16; P = 0.05; 7 trials, 2667 participants; very low-certainty evidence). The 95% prediction interval ranged between -0.03 and 0.19 events per participant per month. The results provided for nocturnal hypoglycaemic episodes were of questionable validity. Overall, there was no clear difference between the two short-acting insulin analogues and regular human insulin. Two trials assessed health-related quality of life and treatment satisfaction, but we considered the results for both outcomes to be unreliable (very low-certainty evidence).

No trial was designed to investigate possible long term effects (all-cause mortality, microvascular or macrovascular complications of diabetes), especially in participants with diabetes-related complications. No trial reported on socioeconomic effects.