original article
A meta-analysis of the hypoglycaemic risk in randomized controlled trials with sulphonylureas in patients with type 2 diabetes M. Monami1 , I. Dicembrini2 , L. Kundisova1 , S. Zannoni1 , B. Nreu1 & E. Mannucci2 1 Section of Geriatric Medicine, Careggi Teaching Hospital, Florence, Italy 2 Diabetes Agency, Careggi Teaching Hospital, Florence, Italy
Aim: To assess hypoglycaemic risk with sulphonylureas in comparison with other drugs in randomized controlled trials. Methods: Randomized trials with a duration ≥24 weeks, enrolling patients with type 2 diabetes, comparing sulphonylureas with placebo or active drugs different from other sulphonylureas. The principal outcome was the effect of sulphonylureas on the incidence of any or severe hypoglycaemia. Cumulative incidence of hypoglycaemia was estimated combining sulphonylurea groups of different trials with a random effect model and used for meta-regression analyses. Results: The incidence of severe hypoglycaemia in patients treated with sulphonylureas was 1.2 [1.0–1.6]%. The overall risk of severe hypoglycaemia was increased more than threefold with sulphonylureas than with comparators. The proportion of patients with at least one hypoglycaemia while on sulphonylureas was 17.4 [14.5–20.8]%. The overall risk (Mantel–Haenszel Odds Ratio) of any hypoglycaemia with sulphonylureas versus comparators was 3.69 [3.47–3.93] (p < 0.001). Meta-regression analysis suggested that the incidence of any hypoglycaemia was higher in trials enrolling patients with higher body mass index (BMI) and lower haemoglobin A1c (HbA1c). Conclusions: In conclusion, hypoglycaemia, including severe hypoglycaemia, is frequent in patients treated with sulphonylureas, particularly when baseline HbA1c levels are lower and BMI levels higher. Further studies are needed to characterize predictors for the identification of patients at higher risk. Keywords: meta-analysis, sulphonylureas Date submitted 14 January 2014; date of first decision 27 January 2014; date of final acceptance 8 March 2014
Introduction Hypoglycaemia which is one of the most frequent adverse events of pharmacological treatments for diabetes, is also a limiting factor for glycaemic targets. In fact, many guidelines and consensus documents recommend the personalization of therapeutic goals, allowing for higher blood glucose levels in patients at higher risk of hypoglycaemia, or with higher risk of severe consequences from hypoglycaemia, such as elderly, frail subjects with comorbidities [1]. It is well known that drugs used for type 2 diabetes have a different effect on hypoglycaemic risk, mainly because of differences in the mechanism of action. Insulin, and drugs which stimulate insulin secretion in a glucose-independent fashion, such as sulphonylureas and glinides, are associated with increased hypoglycaemic risk [1,2]. Conversely, insulin sensitizers and drugs which enhance glucose-induced insulin secretion, as well as agents which modulate carbohydrate intestinal absorption and glucose renal excretion, do not appear Correspondence to: Edoardo Mannucci, MD, Diabetes Agency, Careggi Teaching Hospital, Via delle Oblate 4, 50141 Florence, Italy. E-mail: edoardo.mannucci@unifi.it
to induce hypoglycaemia, unless they are combined with insulin or insulin secretagogues [1]. Despite this adverse event, sulphonylureas are still widely used in the treatment of type 2 diabetes. There is a widespread belief that, although clinically relevant, the actual incidence of hypoglycaemia in type 2 diabetes with sulphonylureas is substantially lower than with insulin [1]. Considering the low cost and the good short-term efficacy, sulphonylureas are still considered a viable option in metformin monotherapy failure [1], and they are even recommended as a preferred choice by some institutions [3]. Therapeutic decisions should be based, whenever possible, on a comprehensive assessment of available evidence from randomized trials. In the case of sulphonylureas, a collection of data from the many available intervention studies for the determination of the hypoglycaemic risk in comparison with other drugs can provide useful information in this regard. In addition, such analysis allows the estimation of the actual incidence of hypoglycaemia, that is, the absolute risk, during clinical trial. On the basis of those data, an attempt can be made at identifying the characteristics of patients at higher risk for hypoglycaemia.
original article
Diabetes, Obesity and Metabolism 16: 833–840, 2014. © 2014 John Wiley & Sons Ltd
original article Aim of this meta-analysis is to collect and summarize available data from randomized trials on the risk of hypoglycaemia with sulphonylurea, exploring possible predictors.
Methods The meta-analysis was reported following the PRISMA checklist [4]. No review protocol was published elsewhere. A detailed protocol is available from the authors, upon request.
Data Sources and Searches An extensive Medline, Embase and Cochrane database search for ‘sulfonylureas OR sulphonylureas OR glimepiride OR glyburide OR glibenclamide OR carbutamide OR tolazamide OR tolbutamide OR gliclazide OR glipizide OR chlorpropamide OR gliquidone OR acetohexamide’ was performed, collecting all randomized clinical trials on humans up to 30 November 2013. The identification of relevant abstracts, the selection of studies based on the criteria described above, and the subsequent data extraction were performed independently by two of the authors (L. K. and M. M.), and conflicts resolved by the third investigator (E. M.).
Study Selection A meta-analysis was performed including all randomized clinical trials with a duration of at least 24 weeks, enrolling patients with type 2 diabetes, comparing sulphonylureas with placebo or active drugs (oral hypoglycaemic agents, GLP1 receptor agonists, and/or insulin) different from other sulphonylureas. Trials enrolling nondiabetic, or type 1 diabetic, subjects were also excluded.
Data Extraction and Quality Assessment Results of unpublished trials were retrieved, if available, on www.clinicaltrials.gov. This source was also used to complete information on results of published trials, when not reported in publications (including the primary trial publications, and subsequent reviews and/or pooled analyses reporting data on individual trials). For all published trials, results reported in papers were used as the primary source of information, when available. Extracted data included the number of patients experiencing at least one episode of overall or severe hypoglycaemia, together with the definitions used for those events. The quality of trials was assessed using some of the parameters proposed by Jadad et al. [5]. The score was not used as a criterion for the selection of trials, whereas some items were used only for descriptive purposes.
Data Synthesis and Analysis The principal outcome of this analysis was the effect of sulphonylureas, compared either with placebo or active drugs, on the incidence of any or severe hypoglycaemia. For all those endpoints, further analyses were performed excluding trials in which sulphonylureas were used as add-on to insulin. Subgroup analyses were performed for different classes of comparators
834 Monami et al.
DIABETES, OBESITY AND METABOLISM
and individual sulphonylureas, whenever more than one trial was available. Heterogeneity (on any and severe hypoglycaemia) was assessed by using I2 statistics. Even when a low heterogeneity was detected, a random-effects model was applied, because the validity of tests of heterogeneity can be limited with a small number of component studies. To estimate possible publication/disclosure bias we used funnel plots and the Begg adjusted rank correlation test [6,7]. However, because these tests have low statistical power when the number of trials is small, undetected bias may still be present. Mantel–Haenszel odds ratio with 95% Confidence Interval (MH-OR) was calculated for the events defined above, on an intentionto-treat basis, excluding trials with zero events. In addition, cumulative incidence (i.e. proportion of subjects with at least one event) was estimated combining sulphonylurea groups of different trials with a random effect model. Estimates of cumulative incidence were calculated separately in trials using different definitions and methods for assessment of any hypoglycaemia. Estimated incidences were also used for metaregression analyses, confronting hypoglycaemia with some baseline characteristics of patients enrolled [age, duration of diabetes, body mass index (BMI) and haemoglobin A1c (HbA1c)]. Multivariate analyses were performed using a stepwise linear regression model. All analyses were performed using Comprehensive Metaanalysis Version 2, Biostat, (Englewood, NJ, USA), except multiple linear regression, performed with spss 20.0.
Results Out of 1623 identified abstracts, 91 trials fulfilling inclusion criteria were selected (Figure 1). Of those, 69 and 66 reported information on overall and severe hypoglycaemia, respectively. The characteristics of the retrieved trials (including parameters of trials quality) and the number of events recorded are reported in supporting information (Tables S1 and S2.) Retrieved trials included 19 801 and 23 603 patients in sulphonylurea and comparator groups, respectively; mean trial duration was 69 weeks. The mean age, duration of diabetes, baseline HbA1c and BMI of enrolled patients were 56.9 years, 5.8 years, 8.4% and 29.9 kg/m2 , respectively.
Severe Hypoglycaemia Of the 69 trials reporting information on severe hypoglycaemia, 24 trials reported at least one event. I2 was 20.0 (p = 0.19). Funnel plot (Figure 2, panel A) and Begg adjusted rank correlation test (Kendall’s tau: −0.20; p = 0.16) suggested no publication bias. The overall risk of severe hypoglycaemia was increased more than threefold with sulphonylureas than with comparators. Significant differences were observed in direct comparisons with DPP4 inhibitors and thiazolidinediones, whereas the number of subjects experiencing at least one episode of major hypoglycaemia was not significantly different between sulphonylureas and insulin in head-to-head trials (Figure 3, panel A).
Volume 16 No. 9 September 2014
original article A
0.0
Standard error
DIABETES, OBESITY AND METABOLISM
0.5
MEDLINE/EMBASE
1.0
1.5
Retrieved Excluded for:
n=1,623
2.0 -0.4
-0.3
-0.2
Not Human
-0.1
0.0
1.0
2.0
3.0
4.0
2.0
3.0
4.0
n=6
B
0.0
Standard error
MH Log odds ratio
0.5
Not RCT n=546 Not T2 diabetes n=93 Duration < 24weeks n= 264 Duplicate
1.0
1.5
2.0
n=95
-0.4
-0.3
-0.2
-0.1
0.0
1.0
MH Log odds ratio No external comparison n=119
Figure 2. Funnel plots of standard errors for any (Panel A) and severe (Panel B) hypoglycaemia.
Noton sulfonylureas n=337 Not data on hypoglycaemia n=51 Differentadd-on n=21
INCLUDED n= 91
Figure 1. Trial flow summary.
The cumulative incidence (i.e. proportion of patients with at least one event) of severe hypoglycaemia in patients treated with sulphonylureas was 1.2 [1.0–1.6]%; the corresponding figure after the exclusion of trials in which sulphonylureas were added to insulin was 1.2[0.9–1.6]%. The cumulative incidence of severe hypoglycaemia did not increase with trial duration (Figure 5, panel A).
Any hypoglycaemia Information on the proportion of patients with at least one hypoglycaemic episode was available in 70 trials, 5 of which reported 0 events. I2 was 0.0 (p = 0.98). Funnel plot (Figure 2, panel B) and Begg adjusted rank correlation test (Kendall’s tau: 88.9; p < 0.001) suggested possible major publication bias. The definition of hypoglycaemia differed across trials: 25 studies defined hypoglycaemia on the basis of symptoms only; in 12 trials, hypoglycaemia was defined on the basis of results of glucose self-monitoring, irrespective of symptoms. In 10 trials,
Volume 16 No. 9 September 2014
only confirmed symptomatic hypoglycaemia (i.e. with both symptoms and low glucose readings) were considered, whereas in 8 trials both symptomatic unconfirmed (i.e. without glucose measurement) and asymptomatic hypoglycaemic episodes were included. Finally, 15 trials did not specify any definition of hypoglycaemia. In those trials which considered glucose selfmonitoring data (either alone or combined with symptoms) for the definition of hypoglycaemia, the diagnostic thresholds ranged from 2.7 to 4.0 mmol/l. The overall risk (MH-OR) of hypoglycaemia with sulphonylureas versus comparators was 3.69 [3.47–3.93] (p < 0.001). The increase in risk was statistically significant in comparisons with placebo/no therapy and any active drug, including glinides, with the only exception of insulin (Figure 3, panel B). Similar results were obtained when excluding trials in which sulphonylureas and comparator were used in combination with insulin (data not shown). Separate analyses for each sulphonylureas are reported in Figure 4. All drugs were associated with a significant increase in the risk of hypoglycaemia, with the only exception of chlorpropamide (panel A); however, when trials comparing sulphonylureas with insulin and glinides were excluded, the difference between chlorpropamide and comparators reached statistical significance (panel B). The proportion of patients with at least one hypoglycaemic episode while on sulphonylureas in randomized trials was 17.4 [14.5–20.8]%. The estimated cumulative incidence varied depending on diagnostic criteria for hypoglycaemia: trials in which both symptomatic unconfirmed and asymptomatic episodes were included in the definition of hypoglycaemia provided a higher estimate of incidence than the rest of the studies (Table 1). Similar results were obtained when excluding trials on combined sulphonylurea-insulin treatment (Table 1).
doi:10.1111/dom.12287 835
original article
DIABETES, OBESITY AND METABOLISM
A
MH-OR
p
# trials
patient/yr
# events
Yearly incidence
DPP-4i
10.97[4.37;27.51]
A meta-analysis of the hypoglycaemic risk in randomized controlled trials with sulphonylureas in patients with type 2 diabetes M. Monami1 , I. Dicembrini2 , L. Kundisova1 , S. Zannoni1 , B. Nreu1 & E. Mannucci2 1 Section of Geriatric Medicine, Careggi Teaching Hospital, Florence, Italy 2 Diabetes Agency, Careggi Teaching Hospital, Florence, Italy
Aim: To assess hypoglycaemic risk with sulphonylureas in comparison with other drugs in randomized controlled trials. Methods: Randomized trials with a duration ≥24 weeks, enrolling patients with type 2 diabetes, comparing sulphonylureas with placebo or active drugs different from other sulphonylureas. The principal outcome was the effect of sulphonylureas on the incidence of any or severe hypoglycaemia. Cumulative incidence of hypoglycaemia was estimated combining sulphonylurea groups of different trials with a random effect model and used for meta-regression analyses. Results: The incidence of severe hypoglycaemia in patients treated with sulphonylureas was 1.2 [1.0–1.6]%. The overall risk of severe hypoglycaemia was increased more than threefold with sulphonylureas than with comparators. The proportion of patients with at least one hypoglycaemia while on sulphonylureas was 17.4 [14.5–20.8]%. The overall risk (Mantel–Haenszel Odds Ratio) of any hypoglycaemia with sulphonylureas versus comparators was 3.69 [3.47–3.93] (p < 0.001). Meta-regression analysis suggested that the incidence of any hypoglycaemia was higher in trials enrolling patients with higher body mass index (BMI) and lower haemoglobin A1c (HbA1c). Conclusions: In conclusion, hypoglycaemia, including severe hypoglycaemia, is frequent in patients treated with sulphonylureas, particularly when baseline HbA1c levels are lower and BMI levels higher. Further studies are needed to characterize predictors for the identification of patients at higher risk. Keywords: meta-analysis, sulphonylureas Date submitted 14 January 2014; date of first decision 27 January 2014; date of final acceptance 8 March 2014
Introduction Hypoglycaemia which is one of the most frequent adverse events of pharmacological treatments for diabetes, is also a limiting factor for glycaemic targets. In fact, many guidelines and consensus documents recommend the personalization of therapeutic goals, allowing for higher blood glucose levels in patients at higher risk of hypoglycaemia, or with higher risk of severe consequences from hypoglycaemia, such as elderly, frail subjects with comorbidities [1]. It is well known that drugs used for type 2 diabetes have a different effect on hypoglycaemic risk, mainly because of differences in the mechanism of action. Insulin, and drugs which stimulate insulin secretion in a glucose-independent fashion, such as sulphonylureas and glinides, are associated with increased hypoglycaemic risk [1,2]. Conversely, insulin sensitizers and drugs which enhance glucose-induced insulin secretion, as well as agents which modulate carbohydrate intestinal absorption and glucose renal excretion, do not appear Correspondence to: Edoardo Mannucci, MD, Diabetes Agency, Careggi Teaching Hospital, Via delle Oblate 4, 50141 Florence, Italy. E-mail: edoardo.mannucci@unifi.it
to induce hypoglycaemia, unless they are combined with insulin or insulin secretagogues [1]. Despite this adverse event, sulphonylureas are still widely used in the treatment of type 2 diabetes. There is a widespread belief that, although clinically relevant, the actual incidence of hypoglycaemia in type 2 diabetes with sulphonylureas is substantially lower than with insulin [1]. Considering the low cost and the good short-term efficacy, sulphonylureas are still considered a viable option in metformin monotherapy failure [1], and they are even recommended as a preferred choice by some institutions [3]. Therapeutic decisions should be based, whenever possible, on a comprehensive assessment of available evidence from randomized trials. In the case of sulphonylureas, a collection of data from the many available intervention studies for the determination of the hypoglycaemic risk in comparison with other drugs can provide useful information in this regard. In addition, such analysis allows the estimation of the actual incidence of hypoglycaemia, that is, the absolute risk, during clinical trial. On the basis of those data, an attempt can be made at identifying the characteristics of patients at higher risk for hypoglycaemia.
original article
Diabetes, Obesity and Metabolism 16: 833–840, 2014. © 2014 John Wiley & Sons Ltd
original article Aim of this meta-analysis is to collect and summarize available data from randomized trials on the risk of hypoglycaemia with sulphonylurea, exploring possible predictors.
Methods The meta-analysis was reported following the PRISMA checklist [4]. No review protocol was published elsewhere. A detailed protocol is available from the authors, upon request.
Data Sources and Searches An extensive Medline, Embase and Cochrane database search for ‘sulfonylureas OR sulphonylureas OR glimepiride OR glyburide OR glibenclamide OR carbutamide OR tolazamide OR tolbutamide OR gliclazide OR glipizide OR chlorpropamide OR gliquidone OR acetohexamide’ was performed, collecting all randomized clinical trials on humans up to 30 November 2013. The identification of relevant abstracts, the selection of studies based on the criteria described above, and the subsequent data extraction were performed independently by two of the authors (L. K. and M. M.), and conflicts resolved by the third investigator (E. M.).
Study Selection A meta-analysis was performed including all randomized clinical trials with a duration of at least 24 weeks, enrolling patients with type 2 diabetes, comparing sulphonylureas with placebo or active drugs (oral hypoglycaemic agents, GLP1 receptor agonists, and/or insulin) different from other sulphonylureas. Trials enrolling nondiabetic, or type 1 diabetic, subjects were also excluded.
Data Extraction and Quality Assessment Results of unpublished trials were retrieved, if available, on www.clinicaltrials.gov. This source was also used to complete information on results of published trials, when not reported in publications (including the primary trial publications, and subsequent reviews and/or pooled analyses reporting data on individual trials). For all published trials, results reported in papers were used as the primary source of information, when available. Extracted data included the number of patients experiencing at least one episode of overall or severe hypoglycaemia, together with the definitions used for those events. The quality of trials was assessed using some of the parameters proposed by Jadad et al. [5]. The score was not used as a criterion for the selection of trials, whereas some items were used only for descriptive purposes.
Data Synthesis and Analysis The principal outcome of this analysis was the effect of sulphonylureas, compared either with placebo or active drugs, on the incidence of any or severe hypoglycaemia. For all those endpoints, further analyses were performed excluding trials in which sulphonylureas were used as add-on to insulin. Subgroup analyses were performed for different classes of comparators
834 Monami et al.
DIABETES, OBESITY AND METABOLISM
and individual sulphonylureas, whenever more than one trial was available. Heterogeneity (on any and severe hypoglycaemia) was assessed by using I2 statistics. Even when a low heterogeneity was detected, a random-effects model was applied, because the validity of tests of heterogeneity can be limited with a small number of component studies. To estimate possible publication/disclosure bias we used funnel plots and the Begg adjusted rank correlation test [6,7]. However, because these tests have low statistical power when the number of trials is small, undetected bias may still be present. Mantel–Haenszel odds ratio with 95% Confidence Interval (MH-OR) was calculated for the events defined above, on an intentionto-treat basis, excluding trials with zero events. In addition, cumulative incidence (i.e. proportion of subjects with at least one event) was estimated combining sulphonylurea groups of different trials with a random effect model. Estimates of cumulative incidence were calculated separately in trials using different definitions and methods for assessment of any hypoglycaemia. Estimated incidences were also used for metaregression analyses, confronting hypoglycaemia with some baseline characteristics of patients enrolled [age, duration of diabetes, body mass index (BMI) and haemoglobin A1c (HbA1c)]. Multivariate analyses were performed using a stepwise linear regression model. All analyses were performed using Comprehensive Metaanalysis Version 2, Biostat, (Englewood, NJ, USA), except multiple linear regression, performed with spss 20.0.
Results Out of 1623 identified abstracts, 91 trials fulfilling inclusion criteria were selected (Figure 1). Of those, 69 and 66 reported information on overall and severe hypoglycaemia, respectively. The characteristics of the retrieved trials (including parameters of trials quality) and the number of events recorded are reported in supporting information (Tables S1 and S2.) Retrieved trials included 19 801 and 23 603 patients in sulphonylurea and comparator groups, respectively; mean trial duration was 69 weeks. The mean age, duration of diabetes, baseline HbA1c and BMI of enrolled patients were 56.9 years, 5.8 years, 8.4% and 29.9 kg/m2 , respectively.
Severe Hypoglycaemia Of the 69 trials reporting information on severe hypoglycaemia, 24 trials reported at least one event. I2 was 20.0 (p = 0.19). Funnel plot (Figure 2, panel A) and Begg adjusted rank correlation test (Kendall’s tau: −0.20; p = 0.16) suggested no publication bias. The overall risk of severe hypoglycaemia was increased more than threefold with sulphonylureas than with comparators. Significant differences were observed in direct comparisons with DPP4 inhibitors and thiazolidinediones, whereas the number of subjects experiencing at least one episode of major hypoglycaemia was not significantly different between sulphonylureas and insulin in head-to-head trials (Figure 3, panel A).
Volume 16 No. 9 September 2014
original article A
0.0
Standard error
DIABETES, OBESITY AND METABOLISM
0.5
MEDLINE/EMBASE
1.0
1.5
Retrieved Excluded for:
n=1,623
2.0 -0.4
-0.3
-0.2
Not Human
-0.1
0.0
1.0
2.0
3.0
4.0
2.0
3.0
4.0
n=6
B
0.0
Standard error
MH Log odds ratio
0.5
Not RCT n=546 Not T2 diabetes n=93 Duration < 24weeks n= 264 Duplicate
1.0
1.5
2.0
n=95
-0.4
-0.3
-0.2
-0.1
0.0
1.0
MH Log odds ratio No external comparison n=119
Figure 2. Funnel plots of standard errors for any (Panel A) and severe (Panel B) hypoglycaemia.
Noton sulfonylureas n=337 Not data on hypoglycaemia n=51 Differentadd-on n=21
INCLUDED n= 91
Figure 1. Trial flow summary.
The cumulative incidence (i.e. proportion of patients with at least one event) of severe hypoglycaemia in patients treated with sulphonylureas was 1.2 [1.0–1.6]%; the corresponding figure after the exclusion of trials in which sulphonylureas were added to insulin was 1.2[0.9–1.6]%. The cumulative incidence of severe hypoglycaemia did not increase with trial duration (Figure 5, panel A).
Any hypoglycaemia Information on the proportion of patients with at least one hypoglycaemic episode was available in 70 trials, 5 of which reported 0 events. I2 was 0.0 (p = 0.98). Funnel plot (Figure 2, panel B) and Begg adjusted rank correlation test (Kendall’s tau: 88.9; p < 0.001) suggested possible major publication bias. The definition of hypoglycaemia differed across trials: 25 studies defined hypoglycaemia on the basis of symptoms only; in 12 trials, hypoglycaemia was defined on the basis of results of glucose self-monitoring, irrespective of symptoms. In 10 trials,
Volume 16 No. 9 September 2014
only confirmed symptomatic hypoglycaemia (i.e. with both symptoms and low glucose readings) were considered, whereas in 8 trials both symptomatic unconfirmed (i.e. without glucose measurement) and asymptomatic hypoglycaemic episodes were included. Finally, 15 trials did not specify any definition of hypoglycaemia. In those trials which considered glucose selfmonitoring data (either alone or combined with symptoms) for the definition of hypoglycaemia, the diagnostic thresholds ranged from 2.7 to 4.0 mmol/l. The overall risk (MH-OR) of hypoglycaemia with sulphonylureas versus comparators was 3.69 [3.47–3.93] (p < 0.001). The increase in risk was statistically significant in comparisons with placebo/no therapy and any active drug, including glinides, with the only exception of insulin (Figure 3, panel B). Similar results were obtained when excluding trials in which sulphonylureas and comparator were used in combination with insulin (data not shown). Separate analyses for each sulphonylureas are reported in Figure 4. All drugs were associated with a significant increase in the risk of hypoglycaemia, with the only exception of chlorpropamide (panel A); however, when trials comparing sulphonylureas with insulin and glinides were excluded, the difference between chlorpropamide and comparators reached statistical significance (panel B). The proportion of patients with at least one hypoglycaemic episode while on sulphonylureas in randomized trials was 17.4 [14.5–20.8]%. The estimated cumulative incidence varied depending on diagnostic criteria for hypoglycaemia: trials in which both symptomatic unconfirmed and asymptomatic episodes were included in the definition of hypoglycaemia provided a higher estimate of incidence than the rest of the studies (Table 1). Similar results were obtained when excluding trials on combined sulphonylurea-insulin treatment (Table 1).
doi:10.1111/dom.12287 835
original article
DIABETES, OBESITY AND METABOLISM
A
MH-OR
p
# trials
patient/yr
# events
Yearly incidence
DPP-4i
10.97[4.37;27.51]
