Review: Long-acting insulin analogues do not improve glycemic control but do reduce nocturnal hypoglycemia in diabetesPDF
ACP J Club. 2008 Mar-Apr;148:37. doi:10.7326/ACPJC-2008-148-2-037
Clinical Impact Ratings
Tran K, Banerjee S, Li H, et al. Long-acting insulin analogues for diabetes mellitus: meta-analysis of clinical outcomes and assessment of cost-effectiveness. Technology report #92. Ottawa: Canadian Agency for Drugs and Technologies in Health (CADTH); 2007.
In patients with diabetes mellitus, what is the efficacy of long-acting insulin analogues (LAIAs) compared with human insulin or oral antidiabetic agents?
Data sources: MEDLINE, EMBASE/Excerpta Medica, BIOSIS Previews, PASCAL, PubMed, and Cochrane Database of Systematic Reviews (1990 to February 2006); Health Economics Evaluation Database; Web sites of regulatory and health technology assessment agencies and professional associations; and drug manufacturers.
Study selection and assessment: Randomized controlled trials (RCTs) comparing LAIAs (insulin glargine or insulin detemir) with conventional human insulin (including NPH) or oral antidiabetic agents in patients with type 1, type 2, or gestational diabetes. 34 RCTs met efficacy selection criteria: 14 of insulin glargine and 9 of insulin detemir in type 1 diabetes (n = 7142); 9 of insulin glargine and 2 of insulin detemir in type 2 diabetes (n = 4729). 31 trials involved adults only. Average quality of the 28 fully published RCTs was low (mean 2.3 on Jadad scale).
Outcomes: Included glycemic control (hemoglobin A1c [HbA1c] level) and hypoglycemic episodes.
Meta-analyses showed that LAIAs and NPH did not differ for change in HbA1c levels (Table); results of individual trials varied, but none reported important, large differences (HbA1c reduction ≥ 1%). The range of endpoint HbA1c levels was similar for insulin glargine vs NPH (16 trials, 6.4% to 9.0% vs 7.0% to 9.1%) and insulin detemir vs NPH (10 trials, 6.6% to 8.3% vs 6.5% to 8.4%). Meta-analyses also showed that risk for nocturnal hypoglycemia was reduced by insulin detemir in type 1 diabetes and by insulin glargine in type 2 diabetes compared with NPH (Table); in type 1 diabetes, severe hypoglycemia was reduced with insulin detemir (Table) and with insulin glargine using human insulin as bolus (5 trials, relative risk reduction 27%, 95% CI 5 to 45). Insulin glargine and NPH did not differ for nocturnal hypoglycemia in type 1 diabetes or severe hypoglycemia overall (Table).
Long-acting insulin analogues do not reduce HbA1c levels more than NPH insulin but do reduce nocturnal hypoglycemia.
Source of funding: Canadian federal, provincial, and territorial governments.
For correspondence: Dr. K. Tran, Canadian Agency for Drugs and Technologies in Health, Ottawa, Ontario, Canada. E-mail email@example.com.
Table. Long-acting insulin analogues (LAIAs) vs conventional insulin (NPH) in diabetes*
|Outcomes||Number of trials (n)||Diabetes type: LAIA||Weighted mean difference (95% CI)†|
|Change in HbA1c level‡||8 (2937)||Type 1: detemir||−0.05% (−0.12 to 0.03)|
|2 (980)||Type 2: detemir||0.11% (−0.03 to 0.26)|
|7 (2967)||Type 2: glargine||0.05% (−0.07 to 0.16)|
|RRR/RRI (CI)||NNT/NNH (CI)|
|Nocturnal hypoglycemia||7 (2590)||Type 1: detemir||RRR 11% (3 to 18)||15 (9 to 34)|
|1 (505)||Type 2: detemir||RRR 34% (4 to 54)||13 (7 to 133)|
|7 (2826)||Type 1: glargine||RRR 8% (−4 to 19)||Not significant|
|5 (2099)||Type 2: glargine||RRR 43% (26 to 56)||9 (7 to 12)|
|Severe hypoglycemia§||8 (2708)||Type 1: detemir||RRR 25% (5 to 41)||50 (25 to 100)|
|6 (2701)||Type 1: glargine||RRR 22% (−5% to 42)||Not significant|
|4 (1885)||Type 2: glargine||RRI 9% (−44 to 112)||Not significant|
*HbA1c = hemoglobin A1c; NPH = neutral protamine Hagedorn; other abbreviations defined in Glossary. RRR, RRI, NNT, NNH, and CI calculated from data in article; all data combined using
a random-effects model.
†Negative values indicate a benefit for insulin glargine or insulin detemir over NPH.
‡Trials of insulin glargine vs NPH in type 1 diabetes were heterogeneous and not pooled.
§Trials of insulin detemir in type 2 diabetes did not report severe hypoglycemia.
Tran and colleagues used an appropriate search strategy, bias protection measures, and pooling methods, but did not contact authors of primary studies for unreported data, increasing the risk for reporting bias. They found that LAIAs have minimal impact on glycemic control compared with older agents but reduce the incidence of hypoglycemia, mainly nocturnal. The evidence regarding quality of life, long-term morbidity, death, and cost-effectiveness is limited and inconclusive.
The conclusions of this review should be interpreted with caution considering the potential heterogeneity of included trials in terms of dosing schedules in the control group, targeted level of glycemic control, level of patient training and education for managing insulin therapy, and baseline risk for hypoglycemia. Moreover, in patients at high risk for hypoglycemia, clinicians should consider continuous insulin delivery (insulin pump) as a viable treatment alternative to LAIAs to minimize hypoglycemia.
The review highlights important issues. There is a lack of well-designed trials that demonstrate clinical and economic benefits for some diabetes interventions that have been widely adopted into practice. The extension of trial data, mostly from outpatient settings, to suggest that LAIAs are optimal agents for in-hospital glycemic control (1) is, at best, premature. In addition, diabetes trials need to improve in quality and shift their almost-exclusive focus on HbA1c to include important patient outcomes, such as hypoglycemia, death, vascular events, quality of life, loss of vision, and patient satisfaction. Of note, only 1 in 5 trials published or in progress considered such outcomes as primary endpoints (2).
Steven A. Smith, MD
M. Hassan Murad, MD, MPH
Rochester, Minnesota, USA
1. Clement S, Braithwaite SS, Magee MF, et al. Management of diabetes and hyperglycemia in hospitals. Diabetes Care. 2004;27:553-591. [PubMed ID: 14747243]
2. Montori VM, Gandhi GY, Guyatt GH. Patient-important outcomes in diabetes—time for consensus. Lancet. 2007;370:1104-6. [PubMed ID: 17905147]