Confidential: For Review Only - BMJ...Confidential: For Review Only 3 CI, 0.99 to 1.56) and...

Confidential: For Review Only

Sulfonylureas as second-line therapy in type 2 diabetes and

the risk of cardiovascular and hypoglycaemic events: a cohort study

Journal: BMJ

Manuscript ID BMJ.2017.042934.R1

Article Type: Research

BMJ Journal: BMJ

Date Submitted by the Author: 09-Mar-2018

Complete List of Authors: Douros, Antonios; McGill University, Epidemiology Dell'Aniello, Sophie; Jewish General Hospital, Centre for Clinical Epidemiology Yu, Oriana; Jewish General Hospital; McGill University Filion, Kristian; McGill University, Centre for Clinical Epidemiology Azoulay, Laurent; McGill University, Department of Oncology Suissa, Samy; McGill University, Epidemiology and Biostatistics and Medicine; Lady Davis Institute, Centre for Clinical Epidemiology

Keywords: Myocardial infarction, ischaemic stroke, oral antidiabetic drugs, prevalent new-user design, severe hypoglycaemia, population-based

https://mc.manuscriptcentral.com/bmj

BMJ


Figure 1

Abbreviations: HES = Hospital Episode Statistics; ONS = Office for National Statistics

Patients with a first prescription of

metformin between April 1, 1998 and March 31, 2013

and linkable to HES and ONS (n = 137,043)

Exclusions:

Combination use at base cohort entry (n = 9707)

Base cohort entry before age 40 (n = 17,938)

Having <12 months of medical information prior to base cohort

entry or date inconsistencies (n = 31,263)

Women with polycystic ovary syndrome (n = 986)

Metformin monotherapy users

(n = 77,149)

Base cohort with metformin monotherapy users

(n = 77,138)

Study cohort for myocardial infarction (n = 23,551)

Study cohort for ischemic stroke (n = 23,636)

Study cohort for cardiovascular death (n = 23,548)

Study cohort for all-cause mortality (n = 23,592)

Study cohort for severe hypoglycaemia (n = 23,555)

Exclusions:

Trim in myocardial infarction cohort (n = 217)

No eligible matches in myocardial infarction cohort (n = 1931)

Trim in ischemic stroke cohort (n = 226)

No eligible matches in ischemic stroke cohort (n = 1837)

Trim in cardiovascular death cohort (n = 212)

No eligible matches in cardiovascular death cohort (n = 1938)

Trim in all-cause mortality cohort (n = 210)

No eligible matches in all-cause mortality cohort (n = 1897)

Trim in severe hypoglycaemia cohort (n = 213)

No eligible matches in severe hypoglycaemia cohort (n = 1931)

Exclusions:

No follow-up (n = 11)

Patients adding or switching to sulfonylureas

(n = 25,699)

Page 1 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

Confidential: For Review OnlyFigure 2

Page 2 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


1

Sulfonylureas as second-line therapy in type 2 diabetes and the risk of

cardiovascular and hypoglycaemic events: a cohort study

Running title: Safety of sulfonylureas as second-line treatment

Antonios Douros, postdoctoral-fellow1,2,3

, Sophie Dell’Aniello, statistician1, Oriana Hoi Yun Yu,

endocrinologist1,4

, Kristian B. Filion, assistant professor of epidemiology and medicine1,2,5

,

Laurent Azoulay, associate professor of epidemiology and oncology1,2,6

, Samy Suissa, professor

of epidemiology and medicine1,2,5

1. Centre for Clinical Epidemiology, Lady Davis Institute, Jewish General Hospital, Montreal,

Quebec, Canada

2. Department of Epidemiology, Biostatistics, and Occupational Health, McGill University,

Montreal, Quebec, Canada

3. Institute of Clinical Pharmacology and Toxicology, Charité-Universitätsmedizin Berlin,

corporate member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin

Institute of Health, Berlin, Germany

4. Division of Endocrinology, Jewish General Hospital, Montreal, Quebec, Canada

5. Division of Clinical Epidemiology, Department of Medicine, McGill University, Montreal,

Quebec, Canada

6. Gerald Bronfman Department of Oncology, McGill University, Montreal, Quebec, Canada

Correspondence:

Dr. Samy Suissa

Centre for Clinical Epidemiology, Lady Davis Institute – Jewish General Hospital

3755 Cote Ste-Catherine, H-461

Montreal, Québec, Canada H3T 1E2

Tel: 514-340-7593

Fax: 514-340-7564

E-mail: [email protected]

Word count main text: 3614

Word count abstract: 365

April 10, 2018

Page 3 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


2

ABSTRACT

Objective: To assess whether adding or switching to sulfonylureas is associated with an

increased risk of myocardial infarction (MI), ischaemic stroke, cardiovascular death, all-cause

mortality, and severe hypoglycaemia, compared to remaining on metformin monotherapy in

patients with type 2 diabetes.

Design: Population-based cohort study.

Setting: General practices contributing data to the United Kingdom Clinical Practice Research

Datalink.

Participants: Patients with type 2 diabetes initiating metformin monotherapy between 1998 and

2013.

Main outcome measures: Using the prevalent new-user cohort design we matched 1:1 patients

adding or switching to sulfonylureas with those remaining on metformin monotherapy on high-

dimensional propensity score, haemoglobin A1c, and number of previous metformin

prescriptions. The two groups were compared using Cox proportional hazards models to estimate

adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) of the study outcomes. In a

secondary analysis, exposure was sub-classified as adding or switching to sulfonylureas.

Results: Among 77,138 metformin initiators, 25,699 added or switched to sulfonylureas during

study period. During a mean follow-up of 1.1 years, sulfonylureas were associated with an

increased risk of MI (incidence rate, 7.8 versus 6.2 per 1000/year; HR, 1.26; 95% CI, 1.01

to1.56), all-cause mortality (incidence rate, 27.3 versus 21.5 per 1000/year; HR, 1.28; 95% CI,

1.15 to 1.44), and severe hypoglycaemia (incidence rate, 5.5 versus 0.7 per 1000/year; HR, 7.60;

95% CI, 4.64 to 12.44) compared with metformin monotherapy. There was a trend towards

increased risks of ischaemic stroke (incidence rate, 6.7 versus 5.5 per 1000/year; HR, 1.24; 95%

Page 4 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


3

CI, 0.99 to 1.56) and cardiovascular death (incidence rate, 9.4 versus 8.1 per 1000/year; HR,

1.18; 95% CI, 0.98 to 1.43). Switching to but not adding sulfonylureas accounted for the

increased risk of MI (HR, 1.73; 95% CI, 1.32 to 2.26 versus 1.02; 0.79 to 1.31), cardiovascular

death (HR, 1.56; 95% CI, 1.24 to 1.97 versus 0.95; 0.75 to 1.20), and all-cause mortality (HR,

1.60; 95% CI, 1.39 to 1.84 versus 1.09; 0.95 to 1.25). No such differences were observed for

ischaemic stroke or severe hypoglycaemia.

Conclusions: Sulfonylureas as second-line therapy are associated with an increased risk of MI,

all-cause mortality, and severe hypoglycaemia, compared with remaining on metformin

monotherapy. Continuing metformin when introducing sulfonylureas appears to be safer than

switching.

KEYWORDS

Myocardial infarction, ischaemic stroke, oral antidiabetic drugs, prevalent new-user design,

severe hypoglycaemia, population-based

Page 5 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


4

WHAT IS ALREADY KNOWN ON THIS SUBJECT

• Sulfonylureas are widely used second-line oral antidiabetic drugs.

• Previous studies have assessed their cardiovascular and hypoglycaemic safety in the

setting of first-line treatment or in comparison with other second-line antidiabetic drugs.

• The specific risk of treatment intensification with sulfonylureas compared with staying

on metformin monotherapy has rarely been investigated.

WHAT THIS STUDY ADDS

• Our study suggests that sulfonylureas as second-line therapy are associated with a

modestly increased risk of myocardial infarction and all-cause mortality, and a strongly

increased risk of severe hypoglycaemia, compared with remaining on metformin

monotherapy.

• Continuation of metformin when introducing sulfonylureas appears to be safer than

switching.

Page 6 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


5

INTRODUCTION

Sulfonylureas are oral antidiabetic drugs recommended as second-line treatments in

patients with type 2 diabetes.1 Despite the recent approval of several new medications,

sulfonylureas remain the most commonly prescribed antidiabetic drugs after treatment failure

with the first-line agent metformin.2 While the safety of sulfonylureas with respect to adverse

cardiovascular and hypoglycaemic events has been studied extensively,3 4

studies focusing

specifically on their cardiovascular and hypoglycaemic safety as second-line therapy, in

individuals with poorly controlled diabetes in need of pharmacotherapy escalation, have been

sparse and limited.

The randomized controlled trials evaluating sulfonylureas as second-line treatment were

underpowered to assess cardiovascular complications of diabetes or severe hypoglycaemia,

which may contribute to the development of adverse cardiovascular events.5 6

In the setting of

real-world clinical practice, the majority of observational studies assessing these outcomes

compared sulfonylureas with other second-line antidiabetic drugs such as dipeptidyl peptidase-4

(DPP-4) inhibitors or insulin.7-9

Thus, the specific risk of treatment intensification with

sulfonylureas compared with staying on metformin monotherapy has rarely been investigated,

and several of the respective observational studies had methodological limitations such as

selection bias,10 11

exposure misclassification,12

and residual confounding.10 12

Since metformin has been associated with a decreased risk of cardiovascular events and

low rates of hypoglycaemia,13

intensifying treatment with sulfonylureas, a potentially cardiotoxic

class3 with high hypoglycaemic risk,

14 may reverse the positive effects of metformin. Thus, the

objective of our population-based study was to assess whether the use of second-line

sulfonylureas, after metformin, is associated with increased risks of myocardial infarction (MI),

Page 7 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


6

ischaemic stroke, cardiovascular death, all-cause mortality, and severe hypoglycaemia in patients

with type 2 diabetes, compared with continuation of metformin monotherapy.

Page 8 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


7

METHODS

Data sources

We used the United Kingdom (UK) Clinical Practice Research Datalink (CPRD) linked

to the Hospital Episode Statistics (HES) and Office for National Statistics (ONS) databases. The

CPRD is a large primary care database which contains the medical records for over 14 million

people enrolled in over 680 general practices.15

Medical diagnoses and procedures are recorded

using the Read code classification, and drugs prescribed by general practitioners are coded based

on the UK Prescription Pricing Authority dictionary. The CPRD contains information on

anthropometric variables such as body mass index (BMI), and lifestyle variables such as

smoking and alcohol use. CPRD data have been previously validated and shown to be of high

quality.16

The HES contains all inpatient and day case hospital admission information, including

primary and secondary diagnoses (coded using the International Statistical Classification of

Diseases and Health-Related Problems, Tenth Revision [ICD-10]), and hospital-related

procedures (coded using the Office of Population Censuses and Surveys classification of

interventions and procedures, 4th version [OPCS-4]). The ONS contains the electronic death

certificates of all citizens living in the UK and includes the underlying cause of death (coded

using the ICD-9 and ICD-10 classifications). The linkage of the HES and ONS to the CPRD is

possible from April 1, 1997, onward, and is limited to English general practices that have

consented to the linkage scheme (currently representing 75% of all English practices).16

Study population – Base cohort

We first formed a base cohort consisting of patients newly treated for type 2 diabetes

with metformin in monotherapy between April 1, 1998 and March 31, 2013, with follow-up until

Page 9 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


8

March 31, 2014. Base cohort entry was defined by the date of the first metformin prescription.

We excluded all patients less than 40 years of age, those with less than one year of medical

history in the CPRD before cohort entry, and women with a diagnosis of polycystic ovary

syndrome at any time before cohort entry. We also excluded patients prescribed any antidiabetic

drugs at any time before base cohort entry. Since this study aimed to be representative of real-

world practice, patients with a history of cardiovascular disease at cohort entry were not

excluded.

Study population – Study cohort

The study cohort was formed by identifying all subjects from the base cohort of

metformin initiators who subsequently added or switched to a sulfonylurea as second-line

treatment. Patients who added or switched to other antidiabetic drugs were censored. For each

one of the subjects adding or switching to a sulfonylurea, we identified a matched reference

subject who also was a metformin initiator but remained on metformin, using a prevalent new-

user design.17

The potential reference subjects were selected from the corresponding exposure

sets, namely from the metformin initiators in the base cohort who received a metformin

prescription within three months of the date the exposed subjects added or switched to a

sulfonylurea, and who received the same number of metformin prescriptions during time in the

base cohort as the exposed subjects (eFigure 1). Thus, the number of metformin prescriptions

between base cohort entry (monotherapy initiation) and study cohort entry (adding or switching

to a sulfonylurea or matched continuation) was inherently a matching covariate. Moreover,

exposed and reference subjects were matched on haemoglobin A1c (HbA1c) level (≤7%, 7.1-

8%, >8%, unknown) at study cohort entry. Finally, exposed and reference subjects were matched

Page 10 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


9

on high-dimensional propensity score (hdPS).18

The hdPS method empirically selects covariates

based on their prevalence and potential for confounding. For each member of each matched set,

we identified all available information from seven data dimensions (five dimensions from the

CPRD: drug prescriptions, procedures, diagnoses, disease history, administrative information;

two dimensions form the HES: diagnoses, procedures) in the one-year period prior to the date of

the matched set, and applied conditional logistic regression to estimate the propensity of

receiving a sulfonylurea treatment, thereby considering the 500 most likely confounders. Patients

with scores in non-overlapping hdPS regions were trimmed from the cohort. The hdPS procedure

was repeated for each outcome, since this method calculates a bias term that accounts for the

association with a specific outcome. Matched metformin patients were allowed to add or switch

to sulfonylureas later during follow-up. In this case, the follow-up for the metformin

monotherapy group was censored at the point of add-on or switch to sulfonylureas. Then, the

patient was included as a sulfonylurea user from that point onwards while identifying a matched

metformin patient at that point.

Patients meeting the study inclusion criteria were followed until the earliest of the

following events: treatment discontinuation (defined either by the absence of a new prescription

by the end of a 60-day period [30 days prescription duration plus 30 days grace period] or by the

addition or the switch to another, non-sulfonylurea antidiabetic drug), occurrence of one of the

study outcomes (defined in detail below), end of registration with the general practice, or end of

the study period (31 March 2014). For the metformin initiators adding or switching to

sulfonylureas, further switches within the sulfonylurea class were permitted.

Study outcomes

Page 11 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


10

We considered five outcomes: hospitalization for MI, hospitalization for ischaemic

stroke, cardiovascular death, all-cause mortality, and severe hypoglycaemia. Hospitalization for

MI and ischaemic stroke were identified using the HES and ONS (MI ICD-9 codes: 410.x; ICD-

10 codes: I21.x; ischaemic stroke ICD-9 codes: 433.x, 434.x, 436.x; ICD-10 codes: I63.x, I64.x;

in primary or secondary position). The diagnostic codes to identify MI in HES have been shown

to be highly valid (>90%),19

while the validity of stroke diagnoses in administrative data is also

high (>80%).20

Cardiovascular death was identified in ONS (underlying cause of death; ICD-9

codes: 390.x-398.x, 401.x-405.x, 410.x-417.x, 420.x-429.x [except 427.5], 430.x-438.x, 440.x-

447.x; ICD-10 codes: I00.x-I77.x [except I46.9]), and all-cause mortality was identified from all

three databases, with the date of death defined by the earliest recording of death in any database.

Severe hypoglycaemia was identified in HES (ICD-10 codes: E16.0, E16.1, E16.2; in primary or

secondary position).

Statistical analysis

We used descriptive statistics to summarize the characteristics of the patients in the

matched groups. Potential post-matching imbalances among covariates were assessed using

standardized mean differences. Incidence rates of each outcome were calculated based on the

Poisson distribution and expressed as number of events per 1000/year. Moreover, we constructed

a Cox proportional hazards regression model for each outcome that estimated the hazard ratio

(HR) and the 95% confidence intervals (CIs) for sulfonylurea therapy versus metformin therapy.

To maximize comparability between the two groups, the models for MI, ischaemic stroke,

cardiovascular death, and severe hypoglycaemia were additionally adjusted for age, sex, hdPS

deciles, and history of the respective outcome in the year prior to cohort entry (or, for the case of

Page 12 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


11

cardiovascular death, history of MI or ischaemic stroke). The model for all-cause mortality was

additionally adjusted for age, sex, and hdPS deciles.

Secondary analyses

We conducted three secondary analyses. First, we assessed the risk of the study outcomes

separately in case of addition of sulfonylureas to metformin and in case of switching to

sulfonylureas from metformin. For this analysis, we used a time-dependent exposure definition

subcategorizing the person-time of sulfonylurea use, which resulted in three mutually exclusive

categories: current use of metformin only (reference), current use of sulfonylureas only, and

concomitant current use of metformin and sulfonylureas. Thus, the same patient could contribute

person-time to different exposure categories. To further corroborate the results of this analysis,

we also conducted a head-to-head comparison between patients adding sulfonylureas and

patients switching to sulfonylureas (described in eMethods 1). Second, to assess a duration-

response relation between switching to or adding on sulfonylureas and the risk of each study

outcome, drug use was further categorized according to three predefined durations (≤3 months,

3.1-12 months, >12 months). Third, given the pharmacologic heterogeneity observed among

different sulfonylureas, the risk of each study outcome was assessed separately for two groups of

sulfonylureas classified by duration of action and pancreas specificity. The first group included

pancreas non-specific, long-acting compounds (i.e., glyburide and glimepiride), while the second

group included pancreas specific, short-acting compounds (i.e., gliclazide, glipizide,

tolbutamide).21-25

This analysis was based on the first sulfonylurea added or switched to.

Switches among sulfonylureas of the same group were allowed during follow-up.

Sensitivity analyses

Page 13 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


12

We performed four sensitivity analyses to assess the robustness of our findings. First, to

assess possible exposure misclassification, we repeated the analyses using a 60-day grace period

between non-overlapping successive prescriptions. Second, the analyses for MI, ischaemic

stroke, cardiovascular death, and severe hypoglycaemia were repeated after excluding patients

with a history of the outcome (or, for the case of cardiovascular death, history of MI or

ischaemic stroke) in the year prior to cohort entry. Third, to assess the potential impact of

residual confounding, we repeated the primary analysis after additionally adjusting for covariates

with a standardized mean difference >5%. Finally, to assess the potential impact of unmeasured

confounding, we conducted a post-hoc sensitivity analysis using the approach proposed by Ding

and VanderWeele (described in eMethods 2).26

Negative control analysis

To further assess the validity of our findings, we conducted an additional analysis using a

negative control outcome.27

Thus, we compared metformin initiators who added or switched to

sulfonylureas to metformin initiators who stayed on metformin monotherapy regarding the risk

of diabetic retinopathy (identified using Read codes from the CPRD and ICD-10 codes from the

HES), since no differential effects between metformin and sulfonylureas have been reported for

this outcome.28

For this analysis, we additionally excluded all patients with previous retinopathy

in order to assess incident disease. All analyses were conducted with SAS, version 9.4 (SAS

Institute, Cary, NC), NC) and R (R Foundation for Statistical Computing, Vienna, Austria;

http://www.r-project.org).

Patient involvement

Page 14 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


13

No patients were involved in setting the research question or the outcome measure, nor

were they involved in developing plans for design or implementation of the study. No patients

were asked to advise on interpretation or writing up of results. There are no plans to disseminate

the results of the research to study participants or the relevant patient community.

Page 15 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


14

RESULTS

The base cohort included 77,138 patients with a first prescription for metformin between

April 1, 1998 and March 31, 2013 (Figure 1). A total of 25,699 patients added or switched to

sulfonylureas during follow-up. For the analysis on all-cause mortality, 2107 (8%) of these

patients were trimmed from the cohort due to non-overlapping hdPS distributions or did not have

any eligible matches. Therefore, the study cohort for this outcome comprised 23,592 patients

who added or switched to sulfonylureas and 23,592 matched patients who remained on

metformin monotherapy, with the size of the study cohorts for the other four outcomes being

similar (Figure 1).

The mean (standard deviation) follow-up was 1.1 (1.4) years, generating a total of

244,150 patient-years. During follow-up, there were 337 MIs (incidence rate: 6.9 per 1000/year,

95% CI: 6.2 to 7.7), 299 ischaemic strokes (incidence rate: 6.1 per 1000/year, 95% CI: 5.5 to

6.9), 429 cardiovascular deaths (incidence rate: 8.7 per 1000/year, 95% CI: 7.9 to 9.6), 1190

deaths from any cause (incidence rate: 24.4 per 1000/year, 95% CI: 23.0 to 25.8), and 150 severe

hypoglycaemic events (incidence rate: 3.1 per 1000/year, 95% CI: 2.6 to 3.6). The most frequent

causes of death were cancer (31%), cardiovascular diseases (31%), and respiratory diseases

(10%). The baseline characteristics of the matched cohorts for the analysis on all-cause mortality

are shown in Table 1. Corresponding tables for the other outcomes (MI, ischaemic stroke,

cardiovascular death, severe hypoglycaemia) were practically identical (eTables 1-4). After

hdPS matching, patients adding or switching to sulfonylureas had a similar baseline profile to

those remaining on metformin monotherapy.

Table 2 and eFigures 2-6 show the results for the five outcomes. Compared with the use

of metformin monotherapy, adding or switching to sulfonylureas was associated with an

Page 16 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


15

increased risk of MI (7.8 versus 6.2 per 1000/year; HR 1.26; 95% CI 1.01 to 1.56), all-cause

mortality (27.3 versus 21.5 per 1000/year; HR 1.28; 95% CI 1.15 to 1.44), and severe

hypoglycaemia (5.5 versus 0.7 per 1000/year; HR 7.60; 95% CI 4.64 to 12.44). There was also a

trend towards increased risks of ischaemic stroke (6.7 versus 5.5 per 1000/year; HR 1.24; 95%

CI 0.99 to 1.56) and cardiovascular death (9.4 versus 8.1 per 1000/year; HR 1.18; 95% CI 0.98

to 1.43).

We also observed that the increase in the risk was driven by the switching to

sulfonylureas and not the addition of sulfonylureas for the outcomes of MI (HR, 1.73; 95% CI,

1.32 to 2.26 versus HR, 1.02; 95% CI, 0.79 to 1.31), cardiovascular death (HR, 1.56; 95% CI,

1.24 to 1.97 versus HR, 0.95; 95% CI, 0.75 to 1.20), and all-cause mortality (HR, 1.60; 95% CI,

1.39 to 1.84 versus HR, 1.09; 95% CI, 0.95 to 1.25) (Table 3). No such differences were

observed for ischaemic stroke (HR, 1.21; 95% CI, 0.89 to 1.65 versus HR, 1.26; 95% CI, 0.97 to

1.63) or severe hypoglycaemia (HR, 8.14; 95% CI, 4.74 to 13.98 versus HR, 7.27; 95% CI, 4.34

to 12.16) (Table 3). A head-to-head comparison between patients adding sulfonylureas to

metformin and patients switching to sulfonylureas from metformin led to similar results (patient

characteristics in eTables 5-9, results in eTable 10).

The analyses based on different durations of use yielded higher point estimates for all

five outcomes for shorter durations of use and especially for the ≤3 months category (eTables

11-15). Classifying sulfonylureas based on important pharmacologic properties provided similar

point estimates for the two sulfonylurea groups (eTable 16). The results of the primary analysis

remained consistent in the sensitivity analyses (summarized in Figure 2 and presented in detail

in eTables 17-19). For MI, the extension of the grace period to 60 days led to a dilution of the

HR, resulting in a non-statistically significant association (HR 1.12; 95% CI 0.94 to 1.32). Based

Page 17 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


16

on a post-hoc analysis, the findings of the primary analysis on MI, all-cause mortality, and severe

hypoglycaemia are unlikely to be the result of an unmeasured confounder under most plausible

exposure-confounder and confounder-outcome associations (eTables 20-22). Finally, we

observed no difference in the risk among sulfonylurea and metformin patients regarding our

negative control outcome diabetic retinopathy (incidence rate 41.4 versus 40.4 per 1000/year;

HR 1.02; 95% CI 0.92 to 1.14).

Page 18 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


17

DISCUSSION

Principal findings

Our study assessed the cardiovascular and hypoglycaemic risk associated with the use of

sulfonylureas as second-line treatment. Among metformin initiators, adding or switching to

sulfonylureas was associated with an increased risk of MI, all-cause mortality and severe

hypoglycaemia, compared with patients remaining on metformin. Moreover, there was a trend

towards increased risks of ischaemic stroke and cardiovascular death. Importantly, the observed

associations with MI, cardiovascular death, and all-cause mortality were driven by the switching

to sulfonylureas and not the addition of sulfonylureas. The results of the primary analysis

remained consistent in sensitivity analyses, as well as after classifying sulfonylureas in two

different groups based on important pharmacologic properties.

Comparison with other studies

Numerous observational studies have evaluated sulfonylurea safety as first-line

treatment.3 However, their results cannot be directly extrapolated to patients having failed initial

monotherapy with metformin, since these patients probably have progressive disease and are

thus at a higher risk of adverse cardiovascular and hypoglycaemic events. Most of the studies

evaluating sulfonylurea safety as second-line treatment used other second-to-third line agents as

active comparators.7-9

Thus, no conclusions can be drawn on the risk of adding or switching to

sulfonylureas, which have been associated with an increased cardiovascular and hypoglycaemic

risk,3 14

as compared to remaining on monotherapy with metformin, a drug with potential

cardioprotective properties and low risk of hypoglycaemia.13

Moreover, among the few studies

assessing the latter risk several featured methodological limitations such as selection bias due to

Page 19 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


18

the inclusion of prevalent users,10 11

exposure misclassification,12

and confounding.10 12

For our

study, we used the recently developed prevalent new-user design.17

To emulate the randomized

controlled trial, this design identifies, at the physician visit that led to the patient on metformin

adding or being switched to sulfonylureas, a comparable patient with the same history of

metformin use and of other characteristics, but who on that visit continued on metformin. Thus,

we were able to specifically address the question of how pharmacotherapy escalation with

sulfonylureas performs in patients having failed metformin monotherapy.

To our knowledge, this is the first observational study assessing the risk of MI associated

with sulfonylureas as second-line treatment. Several potential mechanisms could account for the

increase in the risk as compared to patients remaining on metformin monotherapy. First,

sulfonylureas are associated with weight gain, which is an important risk factor for MI.25

Second,

since hypoglycaemia has been implicated in the development of arrhythmias and cardiac

ischemia,29

the hypoglycaemic propensity of sulfonylureas could have contributed to the

increased MI risk. The higher estimates observed for shorter durations of use argue for an

involvement of short-term mechanisms such as arrhythmias and against long-term mechanisms

such as weight gain. Moreover, the similar estimates obtained for the two sulfonylurea groups

classified by pancreas specificity indicate that this pharmacodynamic property does not

necessarily translate into improved clinical outcomes.22

Finally, the absence of an increased risk

of MI associated with the addition of sulfonylureas to metformin, i.e., in case of metformin

continuation, alludes to the established beneficial effects of the biguanide in this regard.13

Interestingly, metformin was recently shown to also positively modify the cardiovascular effects

of a newer class of antidiabetic drugs, the DPP-4 inhibitors.30

Page 20 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


19

Our results on all-cause mortality corroborate a previous study showing an increased risk

associated with use of sulfonylureas only, but not with concomitant use of sulfonylureas and

metformin, when compared to use of metformin only.31

Again, the absence of an increased risk

associated with the addition of sulfonylureas to metformin could reflect the beneficial effects of

the biguanide.13

Moreover, our higher point estimates for shorter durations of use indicate that

short-term mechanisms such as arrhythmias or seizures and falls potentially induced by severe

hypoglycaemia could be involved in the increased mortality risk.

Our results on severe hypoglycaemia are concordant with a recently published

observational study showing an increased risk for second-line sulfonylureas.11

Moreover, the

similar estimates we obtained for the two sulfonylurea groups classified by duration of action

argue against an effect of this pharmacokinetic property on the risk of severe hypoglycaemia in

the setting of second-line treatment. This is in contrast to our recent findings on sulfonylurea

safety in the setting of first-line treatment,22

and underscores the importance of diabetes severity

as a possible effect modifier on the risk of adverse events.

Strengths and weaknesses of our study

Our study has several strengths. First, the population-based design, the inclusion of

patients with previous events, and the few exclusion criteria make its results highly

generalizable. Second, the large sample size allowed the calculation of precise estimates even for

rare outcomes such as severe hypoglycaemia. Third, our separate analysis of two separate

sulfonylurea groups classified based on important pharmacologic properties was able to account

for the high pharmacologic heterogeneity observed this drug class.25

Page 21 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


20

Our study also has some limitations. First, due to its observational nature there is the

potential for residual confounding. However, we went to great lengths to minimize this potential

bias by matching on hdPS, the number of previous metformin prescriptions, and HbA1c level.

Moreover, we observed no difference in the risk regarding our negative control outcome diabetic

retinopathy. Second, due to the relatively short duration of follow-up, we were not able to assess

long-term risk differences between the two groups. However, the length of follow-up reflects

real-world practice of second-line sulfonylureas.7 Finally, since metformin use is contraindicated

in case of severe kidney disease and decompensated heart failure,32

we cannot exclude that such

conditions leading to metformin discontinuation and switching to sulfonylureas may also account

for the observed increased risks.

Conclusions

In summary, our study showed an increased risk of MI, all-cause mortality, and severe

hypoglycaemia associated with the use of second-line sulfonylureas compared to remaining on

metformin monotherapy. The associations with MI and all-cause mortality were driven by the

switching to sulfonylureas and not the addition of sulfonylureas. Thus, in line with current

recommendations on the treatment of type 2 diabetes,1 continuing metformin when introducing

sulfonylureas appears to be safer than switching.

Page 22 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


21

ACKNOWLEDGMENTS

A.D. is the recipient of a Research Fellowship from the German Research Foundation (Deutsche

Forschungsgemeinschaft). K.B.F. holds a a Chercheur Boursier award from the Fonds de

recherche du Québec–Santé (FRQS). L.A. holds a Chercheur Boursier award from the FRQS and

is the recipient of a William Dawson Scholar award. S.S. is the recipient of the James McGill

Professorship award.

COMPETING INTERESTS

“All authors have completed the ICMJE uniform disclosure form at

www.icmje.org/coi_disclosure.pdf and declare: this research was funded in part by grants from

the Canadian Institutes of Health Research, the Canadian Foundation for Innovation, and

Boehringer-Ingelheim. Boehringer-Ingelheim were provided with the opportunity to comment on

the manuscript, but they were not directly involved in the design and conduct of the study, the

collection, management, analysis, and interpretation of the data, or the preparation, review, or

approval of the manuscript; S.S. has received research grants and has participated in advisory

board meetings or as a speaker at conferences for AstraZeneca, Bayer, Boehringer-Ingelheim,

Bristol-Myers Squibb, Merck, and Novartis. No other potential conflicts of interest relevant to

this article were reported.

CONTRIBUTION STATEMENT

A.D., S.D., O.H.Y.Y., K.B.F., L.A., and S.S. contributed to study concept and design, analyzed

and interpreted the data, and critically revised the manuscript. A.D. drafted the manuscript. A.D.,

S.D., and S.S. conducted the statistical analysis. S.S. acquired the data, obtained funding, and

supervised the study.

Page 23 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


22

ETHICAL APPROVAL

The study protocol was approved by the Independent Scientific Advisory Committee of the

CPRD (protocol 14_019AMn) and by the Research Ethics Board of the Jewish General Hospital,

Montreal, Canada.

DATA SHARING

No additional data available.

TRANSPARENCY

The guarantor (S.S.) affirms that the manuscript is an honest, accurate, and transparent account

of the study being reported; that no important aspects of the study have been omitted; and that

any discrepancies from the study as planned (and, if relevant, registered) have been explained.

Page 24 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


23

REFERENCES

1. ADA. 7. Approaches to Glycemic Treatment. Diabetes care 2016;39 Suppl 1:S52-9. doi:

10.2337/dc16-S010 [published Online First: 2015/12/24]

2. Christensen DH, Rungby J, Thomsen RW. Nationwide trends in glucose-lowering drug use,

Denmark, 1999-2014. Clinical epidemiology 2016;8:381-87. doi: 10.2147/clep.s113211

[published Online First: 2016/10/30]

3. Azoulay L, Suissa S. Sulfonylureas and the Risks of Cardiovascular Events and Death: A

Methodological Meta-Regression Analysis of the Observational Studies. Diabetes care

2017;40(5):706-14. doi: 10.2337/dc16-1943 [published Online First: 2017/04/22]

4. Hemmingsen B, Schroll JB, Wetterslev J, et al. Sulfonylurea versus metformin monotherapy

in patients with type 2 diabetes: a Cochrane systematic review and meta-analysis of

randomized clinical trials and trial sequential analysis. CMAJ open 2014;2(3):E162-75.

doi: 10.9778/cmajo.20130073 [published Online First: 2014/10/09]

5. Dluhy RG, McMahon GT. Intensive glycemic control in the ACCORD and ADVANCE trials.

The New England journal of medicine 2008;358(24):2630-3. doi: 10.1056/NEJMe0804182


6. CADTH Optimal Use Reports. Second- and Third-Line Pharmacotherapy for Type 2 Diabetes

- Update of CADTH 2010 Reviews - Project Protocol. Ottawa (ON): Canadian Agency for

Drugs and Technologies in Health 2012.

7. Roumie CL, Greevy RA, Grijalva CG, et al. Association between intensification of metformin

treatment with insulin vs sulfonylureas and cardiovascular events and all-cause mortality

among patients with diabetes. Jama 2014;311(22):2288-96. doi: 10.1001/jama.2014.4312


Page 25 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


24

8. Roumie CL, Min JY, Greevy RA, et al. Risk of hypoglycemia following intensification of

metformin treatment with insulin versus sulfonylurea. CMAJ : Canadian Medical

Association journal = journal de l'Association medicale canadienne 2016;188(6):E104-12.

doi: 10.1503/cmaj.150904 [published Online First: 2016/01/27]

9. Yu OH, Yin H, Azoulay L. The combination of DPP-4 inhibitors versus sulfonylureas with

metformin after failure of first-line treatment in the risk for major cardiovascular events

and death. Canadian journal of diabetes 2015;39(5):383-9. doi: 10.1016/j.jcjd.2015.02.002


10. Sillars B, Davis WA, Hirsch IB, et al. Sulphonylurea-metformin combination therapy,

cardiovascular disease and all-cause mortality: the Fremantle Diabetes Study. Diabetes,

obesity & metabolism 2010;12(9):757-65. doi: 10.1111/j.1463-1326.2010.01230.x


11. Hippisley-Cox J, Coupland C. Diabetes treatments and risk of amputation, blindness, severe

kidney failure, hyperglycaemia, and hypoglycaemia: open cohort study in primary care.

BMJ (Clinical research ed) 2016;352:i1450. doi: 10.1136/bmj.i1450 [published Online

First: 2016/04/01]

12. Evans JM, Ogston SA, Emslie-Smith A, et al. Risk of mortality and adverse cardiovascular

outcomes in type 2 diabetes: a comparison of patients treated with sulfonylureas and

metformin. Diabetologia 2006;49(5):930-6. doi: 10.1007/s00125-006-0176-9 [published

Online First: 2006/03/10]

13. Effect of intensive blood-glucose control with metformin on complications in overweight

patients with type 2 diabetes (UKPDS 34). UK Prospective Diabetes Study (UKPDS)

Page 26 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


25

Group. Lancet (London, England) 1998;352(9131):854-65. [published Online First:

1998/09/22]

14. Risk of hypoglycaemia in types 1 and 2 diabetes: effects of treatment modalities and their

duration. Diabetologia 2007;50(6):1140-7. doi: 10.1007/s00125-007-0599-y [published

Online First: 2007/04/07]

15. Walley T, Mantgani A. The UK General Practice Research Database. Lancet (London,

England) 1997;350(9084):1097-9. doi: 10.1016/s0140-6736(97)04248-7 [published Online

First: 1999/04/23]

16. Herrett E, Thomas SL, Schoonen WM, et al. Validation and validity of diagnoses in the

General Practice Research Database: a systematic review. British journal of clinical

pharmacology 2010;69(1):4-14. doi: 10.1111/j.1365-2125.2009.03537.x [published Online

First: 2010/01/19]

17. Suissa S, Moodie EE, Dell'Aniello S. Prevalent new-user cohort designs for comparative

drug effect studies by time-conditional propensity scores. Pharmacoepidemiology and drug

safety 2017;26(4):459-68. doi: 10.1002/pds.4107 [published Online First: 2016/09/10]

18. Schneeweiss S, Rassen JA, Glynn RJ, et al. High-dimensional propensity score adjustment in

studies of treatment effects using health care claims data. Epidemiology (Cambridge,

Mass) 2009;20(4):512-22. doi: 10.1097/EDE.0b013e3181a663cc [published Online First:

2009/06/03]

19. Herrett E, Shah AD, Boggon R, et al. Completeness and diagnostic validity of recording

acute myocardial infarction events in primary care, hospital care, disease registry, and

national mortality records: cohort study. BMJ : British Medical Journal 2013;346 doi:

10.1136/bmj.f2350

Page 27 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


26

20. Andrade SE, Harrold LR, Tjia J, et al. A systematic review of validated methods for

identifying cerebrovascular accident or transient ischemic attack using administrative data.

Pharmacoepidemiology and drug safety 2012;21 Suppl 1:100-28. doi: 10.1002/pds.2312


21. Abdelmoneim AS, Hasenbank SE, Seubert JM, et al. Variations in tissue selectivity amongst

insulin secretagogues: a systematic review. Diabetes, obesity & metabolism

2012;14(2):130-8. doi: 10.1111/j.1463-1326.2011.01496.x [published Online First:

2011/09/20]

22. Douros A, Yin H, Yu OHY, et al. Pharmacologic Differences of Sulfonylureas and the Risk

of Adverse Cardiovascular and Hypoglycemic Events. Diabetes care 2017 doi:

10.2337/dc17-0595

23. Gribble FM, Reimann F. Pharmacological modulation of K(ATP) channels. Biochemical

Society transactions 2002;30(2):333-9. doi: 10.1042/ [published Online First: 2002/05/25]

24. Gribble FM, Reimann F. Sulphonylurea action revisited: the post-cloning era. Diabetologia

2003;46(7):875-91. doi: 10.1007/s00125-003-1143-3 [published Online First: 2003/06/24]

25. Krentz AJ, Bailey CJ. Oral antidiabetic agents: current role in type 2 diabetes mellitus. Drugs

2005;65(3):385-411. [published Online First: 2005/01/27]

26. Ding P, VanderWeele TJ. Sensitivity Analysis Without Assumptions. Epidemiology

(Cambridge, Mass) 2016;27(3):368-77. doi: 10.1097/EDE.0000000000000457

27. Lipsitch M, Tchetgen Tchetgen E, Cohen T. Negative controls: a tool for detecting

confounding and bias in observational studies. Epidemiology (Cambridge, Mass)

2010;21(3):383-8. doi: 10.1097/EDE.0b013e3181d61eeb [published Online First:

2010/03/26]

Page 28 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


27

28. Solomon SD, Chew E, Duh EJ, et al. Diabetic Retinopathy: A Position Statement by the

American Diabetes Association. Diabetes care 2017;40(3):412-18. doi: 10.2337/dc16-2641


29. Frier BM. Hypoglycaemia in diabetes mellitus: epidemiology and clinical implications.

Nature reviews Endocrinology 2014;10(12):711-22. doi: 10.1038/nrendo.2014.170


30. Crowley MJ, Williams JW, Kosinski AS, et al. Metformin Use May Moderate the Effect of

DPP-4 Inhibitors on Cardiovascular Outcomes. Diabetes care 2017;40(12):1787-89. doi:

10.2337/dc17-1528

31. Azoulay L, Schneider-Lindner V, Dell'aniello S, et al. Combination therapy with

sulfonylureas and metformin and the prevention of death in type 2 diabetes: a nested case-

control study. Pharmacoepidemiology and drug safety 2010;19(4):335-42. doi:

10.1002/pds.1834 [published Online First: 2010/01/07]

32. Summary of product characteristics - Glucophage 2017 [updated 23.05.2017. Available

from: https://www.medicines.org.uk/emc/medicine/20952 accessed 06.12.2017.

Page 29 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


28

Table 1 Characteristics of users of sulfonylureas as second-line treatment and metformin as

first-line treatment in patients with type 2 diabetes (all-cause mortality study cohort)*

Sulfonylureas

(n = 23,592)

Metformin

(n = 23,592) SMD

Age (years), mean (SD) 64.0 (11.5) 63.6 (11.6) 0.0358

Female sex, n (%) 10,269 (43.5) 10,018 (42.5) 0.0215

Diabetes duration (years), mean (SD) 3.5 (3.6) 3.4 (3.5) 0.0281

Alcohol-related disorders, n (%) 1284 (5.4) 1274 (5.4) 0.0019

Smoking status, n (%)

Ever 12,944 (54.9) 12,810 (54.3) 0.0114

Never 10,435 (44.2) 10,559 (44.8) -0.0106

Unknown 213 (0.9) 223 (0.9) -0.0044

Body mass index (kg/m2), n (%)

<25 2694 (11.4) 2182 (9.2) 0.0713

25-30 7914 (33.5) 7559 (32.0) 0.0321

≥30 12,554 (53.2) 13,425 (56.9) -0.0743

Unknown 430 (1.8) 426 (1.8) 0.0013

Haemoglobin A1c (%), mean (SD) 8.7 (1.7) 8.4 (1.5)

≤7, n (%) 2028 (8.6) 2028 (8.6) **

7.1-8.0, n (%) 7106 (30.1) 7106 (30.1) **

>8, n (%) 12,439 (52.7) 12,439 (52.7) **

Unknown, n (%) 2019 (8.6) 2019 (8.6) **

Medical history, n (%)

Heart failure 1623 (6.9) 1413 (6.0) 0.0363

Arterial hypertension 15,089 (64.0) 14,804 (62.8) 0.0251

Coronary artery disease 5174 (21.9) 4994 (21.2) 0.0186

Atrial fibrillation or flutter 2217 (9.4) 2052 (8.7) 0.0244

Hyperlipidaemia 7424 (31.5) 7212 (30.6) 0.0194

Chronic obstructive pulmonary disease 3396 (14.4) 3275 (13.9) 0.0147

Cancer 2328 (9.9) 2211 (9.4) 0.0168

Anaemia 399 (1.7) 360 (1.5) 0.0131

Thyroid disease 2526 (10.7) 2396 (10.2) 0.0180

Severe hypoglycaemia 35 (0.1) 18 (0.0) 0.0215

Myocardial infarction 1190 (5.0) 1020 (4.3) 0.0341

Ischaemic stroke 434 (1.8) 385 (1.6) 0.0159

Medications, n (%)

Angiotensin-converting enzyme inhibitors 10,523 (44.6) 10,037 (42.5) 0.0416

Angiotensin II receptor blockers 3300 (14.0) 3108 (13.2) 0.0238

Page 30 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


29

Sulfonylureas

(n = 23,592)

Metformin

(n = 23,592) SMD

Beta-blockers 6163 (26.1) 5751 (24.4) 0.0402

Calcium-channel blockers 6751 (28.6) 6366 (27.0) 0.0364

Diuretics 8410 (35.6) 7924 (33.6) 0.0433

Cardiac glycosides 1069 (4.5) 962 (4.1) 0.0223

Nitrates 1453 (6.2) 1348 (5.7) 0.0188

Statins 16,402 (69.5) 15,800 (67.0) 0.0548

Acetylsalicylic acid 9187 (38.9) 8886 (37.7) 0.0262

Clopidogrel 875 (3.7) 693 (2.9) 0.0430

Warfarin 1368 (5.8) 1228 (5.2) 0.0260

Acetaminophen 8350 (35.4) 7845 (33.3) 0.0451

Non-steroidal anti-inflammatory drugs 4987 (21.1) 4981 (21.1) 0.0006

Opioids 7544 (32.0) 7019 (29.8) 0.0482

Diabetic complications, n (%)

Neuropathy 3040 (12.9) 2888 (12.2) 0.0194

Peripheral vascular disease 1439 (6.1) 1297 (5.5) 0.0258

Nephropathy 3082 (13.1) 2561(10.9) 0.0681

Retinopathy 5566 (23.6) 5433 (23.0) 0.0133

* The two groups were matched 1:1 on high-dimensional propensity score, level of haemoglobin

A1c, and number of metformin prescriptions before the first sulfonylurea prescription. Slight

differences exist in the populations for each outcome due to trimming.

** Matching variable.

Abbreviations: SD = standard deviation; SMD = standardized mean difference

Page 31 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


30

Table 2 Crude and adjusted hazard ratios for the association between the use of sulfonylureas as second-line treatment and

the risk of the study outcomes

Exposure Patients Events Person-years Incidence rate (95% CI)

[per 1000 person-years]

Crude HR

(95% CI)

Adjusted HR

(95% CI)*

Myocardial infarction

Metformin 23,551 152 24,673 6.2 (5.3 to 7.2) 1.00 (reference) 1.00 (reference)

Sulfonylureas 23,551 185 23,858 7.8 (6.7 to 9.0) 1.25 (1.01 to 1.55) 1.26 (1.01 to 1.56)

Ischaemic stroke



Cardiovascular death



All-cause mortality



Severe hypoglycaemia



* The models for myocardial infarction, ischaemic stroke, cardiovascular death, and severe hypoglycaemia were adjusted for age, sex,

hdPS deciles, and history of the respective outcome in the year prior to cohort entry (or, for the case of cardiovascular death, history of

myocardial infarction or ischaemic stroke). The model for all-cause mortality was adjusted for age, sex, and hdPS deciles.

Abbreviations: CI = confidence interval; HR = hazard ratio; hdPS = high-dimensional propensity score.

Page 32 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


31

Table 3. Crude and adjusted hazard ratios for the association between the use of sulfonylureas and the risk of the study

outcomes separately for patients adding sulfonylureas and patients switching to sulfonylureas

Exposure Events Person-years Incidence rate (95% CI)


Crude HR

(95% CI)

Adjusted HR

(95% CI)*


Metformin 152 24,673 6.2 (5.3 to 7.2) 1.00 (reference) 1.00 (reference)

Adding sulfonylureas 100 17,725 5.6 (4.6 to 6.9) 0.90 (0.70 to 1.16) 1.02 (0.79 to 1.31)

Switching to sulfonylureas 85 6133 13.9 (11.2 to 17.1) 2.30 (1.76 to 3.00) 1.73 (1.32 to 2.26)

Ischaemic stroke


Adding sulfonylureas 104 17,829 5.8 (4.8 to 7.1) 1.05 0.82 to 1.36) 1.26 (0.97 to 1.63)






All-cause mortality








Page 33 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


32



myocardial infarction or ischaemic stroke). The model for all-cause mortality was adjusted for age, sex, and hdPS deciles.


Page 34 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


33

Figure 1

Figure 2

Patients with a first prescription of

metformin between April 1, 1998 and March 31, 2013

and linkable to HES and ONS (n = 137,043)

Exclusions:

Combination use at base cohort entry (n = 9707)

Base cohort entry before age 40 (n = 17,938)

Having <12 months of medical information prior to base cohort

entry or date inconsistencies (n = 31,263)

Women with polycystic ovary syndrome (n = 986)

Metformin monotherapy users

(n = 77,149)

Base cohort with metformin monotherapy users

(n = 77,138)

Study cohort for myocardial infarction (n = 23,551)

Study cohort for ischaemic stroke (n = 23,636)

Study cohort for cardiovascular death (n = 23,548)

Study cohort for all-cause mortality (n = 23,592)

Study cohort for severe hypoglycaemia (n = 23,555)

Exclusions:

Trim in myocardial infarction cohort (n = 217)

No eligible matches in myocardial infarction cohort (n = 1931)

Trim in ischaemic stroke cohort (n = 226)

No eligible matches in ischaemic stroke cohort (n = 1837)

Trim in cardiovascular death cohort (n = 212)

No eligible matches in cardiovascular death cohort (n = 1938)

Trim in all-cause mortality cohort (n = 210)

No eligible matches in all-cause mortality cohort (n = 1897)

Trim in severe hypoglycaemia cohort (n = 213)

No eligible matches in severe hypoglycaemia cohort (n = 1931)

Exclusions:

No follow-up (n = 11)

Patients adding or switching to sulfonylureas

(n = 25,699)

Page 35 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


34

Page 36 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


35

FIGURE LEGENDS

Figure 1

Flowchart describing the construction of base and study cohorts.

Abbreviations: HES = Hospital Episode Statistics; ONS = Office for National Statistics

Figure 2

Forest plot summarizing the primary analysis and all sensitivity analyses.

Abbreviations: HR = hazard ratio; CI = confidence interval; MI = myocardial infarction

Page 37 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


1

Table of contents

eMethods 1. Head-to-head comparison between patients adding sulfonylureas and patients

switching to sulfonylureas .............................................................................................................. 3

eMethods 2. Sensitivity analysis without assumptions ................................................................... 3

eFigure 1. Illustration of the prevalent new-user study design ....................................................... 5

eFigure 2. Cumulative incidence of myocardial infarction among users of sulfonylureas and

metformin. ....................................................................................................................................... 6

eFigure 3. Cumulative incidence of ischemic stroke among users of sulfonylureas and

metformin. ....................................................................................................................................... 7

eFigure 4. Cumulative incidence of cardiovascular death among users of sulfonylureas and

metformin. ....................................................................................................................................... 8

eFigure 5. Cumulative incidence of all-cause mortality among users of sulfonylureas and

metformin. ....................................................................................................................................... 9

eFigure 6. Cumulative incidence of severe hypoglycaemia among users of sulfonylureas and

metformin. ..................................................................................................................................... 10

eTable 1. Characteristics of users of sulfonylureas as second-line treatment and metformin as

first-line treatment in patients with type 2 diabetes (myocardial infarction study cohort)* ......... 11


first-line treatment in patients with type 2 diabetes (ischemic stroke study cohort)* .................. 13


first-line treatment in patients with type 2 diabetes (cardiovascular death study cohort)* .......... 15


first-line treatment in patients with type 2 diabetes (severe hypoglycaemia study cohort)* ........ 17

eTable 5. Characteristics of patients adding sulfonylureas to metformin and patients switching to

sulfonylureas from metformin (myocardial infarction study cohort) ........................................... 19


sulfonylureas from metformin (ischaemic stroke study cohort) ................................................... 21


sulfonylureas from metformin (cardiovascular death study cohort) ............................................. 23


sulfonylureas from metformin (all-cause mortality study cohort) ................................................ 25


sulfonylureas from metformin (severe hypoglycaemia study cohort) .......................................... 27

eTable 10. Crude and adjusted hazard ratios for the association between switching versus adding

sulfonylureas and the risk of the study outcomes ......................................................................... 29

Page 38 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


2

eTable 11. Crude and adjusted hazard ratios for the association between the use of sulfonylureas

and the risk of myocardial infarction according to duration of use .............................................. 30


and the risk of ischaemic stroke according to duration of use ...................................................... 31


and the risk of cardiovascular death according to duration of use ................................................ 32


and the risk of all-cause mortality according to duration of use ................................................... 33


and the risk of severe hypoglycaemia according to duration of use ............................................. 34

eTable 16. Crude and adjusted hazard ratios for the association between the use of different

sulfonylureas classified by duration of action and pancreas specificity and the risk of the study

outcomes ....................................................................................................................................... 35


and the risk of the study outcomes (60-day grace period) ............................................................ 37


and the risk of the study outcomes (exclusion of previous outcomes) ......................................... 38


and the risk of the study outcomes (adjusting for additional covariates) ..................................... 39

eTable 20. Bounds on corrected estimates and 95% confidence intervals for unmeasured

confounding for the analysis on the risk of myocardial infarction (sensitivity analysis without

assumptions)* ............................................................................................................................... 40


confounding for the analysis on the risk of all-cause mortality (sensitivity analysis without

assumptions)* ............................................................................................................................... 41


confounding for the analysis on the risk of severe hypoglycaemia (sensitivity analysis without

assumptions)* ............................................................................................................................... 42

Page 39 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


3

eMethods 1. Head-to-head comparison between patients adding sulfonylureas and patients

switching to sulfonylureas

For this analysis, we further classified patients initiating sulfonylureas as second-line treatment

either as ‘adders’ or switchers. We defined addition of sulfonylureas as a metformin prescription

in the first month after the first sulfonylurea prescription. Switching to sulfonylureas was defined

by the absence of a metformin prescription in the same period. Next, we performed a

multivariate logistic regression to estimate the probability (propensity score) of switching to

sulfonylureas versus adding sulfonylureas conditional on all variables listed in the manuscript.

We then trimmed patients with non-overlapping propensity score distributions. The remaining

patients were followed from one month after the initial sulfonylurea prescription until they added

or switched to a non-metformin, non-sulfonylurea antidiabetic drug or experienced one of the

study outcomes, whichever occurred earlier. ‘Adders’ were additionally censored at metformin

or sulfonylurea discontinuation, and switchers were additionally censored in case of metformin

re-initiation or sulfonylurea discontinuation.

Finally, the hazard ratio of the study outcomes was estimated using a Cox proportional hazards

model adjusted for age, sex, history of the outcome in the year prior to cohort entry (in the case

of cardiovascular death it was defined as history of myocardial infarction or ischaemic stroke; in

the case of all-cause mortality this variable was not included in the model), propensity score

deciles, and for variables with a standardized mean differences >5% in Table 1 (body mass

index, diuretics, statins, acetaminophen, opioids, and nephropathy).

Patient characteristics and standardized mean differences before trimming are presented in

eTables 5-9 and the results of the risk analysis are presented in eTable 10.

eMethods 2. Sensitivity analysis without assumptions

We performed a post-hoc sensitivity analysis to assess the impact of unmeasured confounding on

our point estimates using the approach proposed by Ding and VanderWeele.1 This approach does

not impose any assumptions on the unmeasured confounder or confounders; such as having an

unmeasured confounder that is binary or having no interaction between the effects of the

exposure and the confounder on the outcome or having only one unmeasured confounder. The

analysis derives a ‘joint bounding factor’ and a sharp inequality such that the sensitivity analysis

parameters must satisfy the inequality if an unmeasured confounder is to explain away the

observed effect estimate or reduce it to a particular level. We found that in order to completely

explain away the observed increased risks of myocardial infarction, all-cause mortality, and

severe hypoglycaemia associated with sulfonylureas as second-line treatment (i.e., to reduce the

observed risks to hazard ratio = 1), an unmeasured confounder would need to be strongly

associated with both the exposure and the outcome (eTables 20-22). For example, with respect

to myocardial infarction, given an association between a potential unmeasured confounder and

the exposure of 1.5, the association between this confounder and the outcome would need to be

at least 2.5 (eTable 20).

Reference

Page 40 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


4

1. Ding P, VanderWeele TJ. Sensitivity Analysis Without Assumptions. Epidemiology

(Cambridge, Mass). 2016;27(3):368-377.

Page 41 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


5

Time in base cohort

Time in study cohort

Metformin prescription

Sulfonylurea prescription

Patient 1

Patient 3

Patient 2

Patient 5

Patient 4

Patient 6

Patient 7

Patient 8

Metformin prescription

from the exposure set

1 2 7 5 3 6 4

Number of prescriptions

eFigure 1. Illustration of the prevalent new-user study design

Initiators of metformin who added or switched to sulfonylureas are matched to initiators of

metformin who remained on metformin on the number of metformin prescriptions before the

first sulfonylurea prescription*.

* In the present study additional matching was performed on high-dimensional propensity score

and levels of haemoglobin A1c.

Page 42 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


6

eFigure 2. Cumulative incidence of myocardial infarction among users of sulfonylureas and

metformin.

Page 43 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


7

eFigure 3. Cumulative incidence of ischemic stroke among users of sulfonylureas and

metformin.

Page 44 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


8

eFigure 4. Cumulative incidence of cardiovascular death among users of sulfonylureas and

metformin.

Page 45 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


9

eFigure 5. Cumulative incidence of all-cause mortality among users of sulfonylureas and

metformin.

Page 46 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


10

eFigure 6. Cumulative incidence of severe hypoglycaemia among users of sulfonylureas and

metformin.

Page 47 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


11

eTable 1. Characteristics of users of sulfonylureas as second-line treatment and metformin

as first-line treatment in patients with type 2 diabetes (myocardial infarction study

cohort)*

Sulfonylureas

(n = 23,551)

Metformin

(n = 23,551) SMD

Age (years), mean (SD) 64.0 (11.5) 63.6 (11.7) 0.0328

Female sex, n (%) 10,229 (43.4) 10,111 (42.9) 0.0101


Alcohol-related disorders, n (%) 1293 (5.5) 1362 (5.8) -0.0127


Ever 12,923 (54.9) 12,840 (54.5) 0.0071

Never 10,415 (44.2) 10,482 (44.5) -0.0057

Unknown 213 (0.9) 229 (1.0) -0.0070


<25 2692 (11.4) 2222 (9.4) 0.0653

25-30 7903 (33.6) 7570 (32.1) 0.0301

≥30 12,522 (53.2) 13,315 (56.5) -0.0677

Missing 434 (1.8) 444 (1.9) -0.0031


≤7%, n (%) 2036 (8.6) 2036 (8.6) **

7.1-8.0%, n (%) 7107 (30.2) 7107 (30.2) **

>8%, n (%) 12,378 (52.6) 12,378 (52.6) **

Unknown, n (%) 2030 (8.6) 2030 (8.6) **


Heart failure 1641 (7.0) 1440 (6.1) 0.0345






Cancer 2349 (10.0) 2210 (9.4) 0.0200

Anaemia 392 (1.7) 386 (1.6) 0.0020

Thyroid disease 2509 (10.7) 2405 (10.2) 0.0144




Medications, n (%)


Page 48 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


12

Sulfonylureas

(n = 23,551)

Metformin

(n = 23,551) SMD


Beta-blockers 6111 (25.9) 5768 (24.5) 0.0335


Diuretics 8388 (35.6) 7950 (33.8) 0.0391


Nitrates 1446 (6.1) 1351 (5.7) 0.0171

Statins 16,363 (69.5) 15,761 (66.9) 0.0549


Clopidogrel 873 (3.7) 718 (3.0) 0.0364

Warfarin 1374 (5.8) 1211 (5.1) 0.0304

Acetaminophen 8370 (35.5) 7739 (32.9) 0.0565


Opioids 7563 (32.1) 6959 (29.5) 0.0556


Neuropathy 3033 (12.9) 2955 (12.5) 0.0099


Nephropathy 3109 (13.2) 2615 (11.1) 0.0642

Retinopathy 5542 (23.5) 5450 (23.1) 0.0092




** Matching variable

Abbreviations: SD = standard deviation; SMD = standardized mean difference.

Page 49 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


13


as first-line treatment in patients with type 2 diabetes (ischemic stroke study cohort)*

Sulfonylureas

(n = 23,636)

Metformin

(n = 23,636) SMD

Age (years), mean (SD) 64.0 (11.6) 63.7 (11.7) 0.0300

Female sex, n (%) 10,281 (43.5) 10,096 (42.7) 0.0158




Ever 12,957 (54.8) 12,875 (54.5) 0.0070

Never 10,466 (44.3) 10,552 (44.6) -0.0073

Unknown 213 (0.9) 209 (0.9) 0.0018


<25 2698 (11.4) 2226 (9.4) 0.0654

25-30 7919 (33.5) 7608 (32.2) 0.0280

≥30 12,590 (53.3) 13367 (56.6) -0.0661

Missing 429 (1.8) 435 (1.8) -0.0019


≤7%, n (%) 2029 (8.6) 2029 (8.6) **

7.1-8.0%, n (%) 7107 (30.1) 7107 (30.1) **

>8%, n (%) 12,466 (52.7) 12,466 (52.7) **

Unknown, n (%) 2034 (8.6) 2034 (8.6) **


Heart failure 1638 (6.9) 1419 (6.0) 0.0377






Cancer 2351 (9.9) 2204 (9.3) 0.0211

Anaemia 397 (1.7) 391 (1.7) 0.0020

Thyroid disease 2530 (10.7) 2410 (10.2) 0.0166




Medications, n (%)



Page 50 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


14

Sulfonylureas

(n = 23,636)

Metformin

(n = 23,636) SMD

Beta-blockers 6156 (26.0) 5761 (24.4) 0.0385


Diuretics 8427 (35.7) 7842 (33.2) 0.0521


Nitrates 1456 (6.2) 1314 (5.6) 0.0256

Statins 16,432 (69.5) 15,888 (67.2) 0.0495


Clopidogrel 890 (3.8) 707 (3.0) 0.0429

Warfarin 1368 (5.8) 1255 (5.3) 0.0209

Acetaminophen 8386 (35.5) 7859 (33.3) 0.0470


Opioids 7596 (32.1) 7033 (29.8) 0.0515


Neuropathy 3044 (12.9) 2957 (12.5) 0.0111


Nephropathy 3123 (13.2) 2605 (11.0) 0.0672

Retinopathy 5575 (23.6) 5431 (23.0) 0.0144






Page 51 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


15


as first-line treatment in patients with type 2 diabetes (cardiovascular death study cohort)*

Sulfonylureas

(n = 23,548)

Metformin

(n = 23,548) SMD

Age (years), mean (SD) 64.0 (11.5) 63.6 (11.7) 0.0384

Female sex, n (%) 10,263 (43.6) 10,099 (42.9) 0.0141




Ever 12,917 (54.9) 12,820 (54.4) 0.0083

Never 10,417 (44.2) 10,506 (44.6) -0.0076

Unknown 214 (0.9) 222 (0.9) -0.0035


<25 2686 (11.4) 2156 (9.2) 0.0742

25-30 7914 (33.6) 7565 (32.1) 0.0316

≥30 12,518 (53.2) 13,410 (56.9) -0.0762

Missing 430 (1.8) 417 (1.8) 0.0042


≤7%, n (%) 2027 (8.6) 2027 (8.6) **

7.1-8.0%, n (%) 7112 (30.2) 7112 (30.2) **

>8%, n (%) 12,379 (52.6) 12,379 (52.6) **

Unknown, n (%) 2030 (8.6) 2030 (8.6) **


Heart failure 1641 (7.0) 1428 (6.1) 0.0367






Cancer 2323 (9.9) 2167 (9.2) 0.0226

Anaemia 387 (1.6) 386 (1.6) 0.0003

Thyroid disease 2523 (10.7) 2441 (10.4) 0.0113




Medications, n (%)



Page 52 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


16

Sulfonylureas

(n = 23,548)

Metformin

(n = 23,548) SMD

Beta-blockers 6119 (26.0) 5836 (24.8) 0.0276


Diuretics 8364 (35.5) 7948 (33.8) 0.0371


Nitrates 1446 (6.1) 1367 (5.8) 0.0142

Statins 16,358 (69.5) 15,730 (66.8) 0.0573


Clopidogrel 879 (3.7) 716 (3.0) 0.0383

Warfarin 1368 (5.8) 1265 (5.4) 0.0190

Acetaminophen 8341 (35.4) 7784 (33.1) 0.0499

Non-steroidal anti-inflammatory drugs 4973 (21.1) 5075 (21.6) -0.0106

Opioids 7546 (32.0) 6930 (29.4) 0.0567


Neuropathy 3025 (12.8) 2858 (12.1) 0.0215


Nephropathy 3098 (13.2) 2575 (10.9) 0.0683

Retinopathy 5540 (23.5) 5408 (23.0) 0.0133






Page 53 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


17


as first-line treatment in patients with type 2 diabetes (severe hypoglycaemia study

cohort)*

Sulfonylureas

(n = 23,555)

Metformin

(n = 23,555) SMD

Age (years), mean (SD) 64.0 (11.5) 63.7 (11.7) 0.0291

Female sex, n (%) 10,250 (43.5) 10,104 (42.9) 0.0125


Alcohol-related disorders, n (%) 1301 (5.5) 1316 (5.6) -0.0028


Ever 12,928 (54.9) 12,874 (54.7) 0.0046

Never 10,413 (44.2) 10,472 (44.5) -0.0050

Unknown 214 (0.9) 209 (0.9) 0.0023


<25 2704 (11.5) 2152 (9.1) 0.0771

25-30 7912 (33.6) 7659 (32.5) 0.0228

≥30 12,510 (53.1) 13,305 (56.5) -0.0679

Missing 429 (1.8) 439 (1.9) -0.0032


≤7%, n (%) 2040 (8.7) 2040 (8.7) **

7.1-8.0%, n (%) 7097 (30.1) 7097 (30.1) **

>8%, n (%) 12,396 (52.6) 12,396 (52.6) **

Unknown, n (%) 2022 (8.6) 2022 (8.6) **


Heart failure 1637 (6.9) 1377 (5.8) 0.0451






Cancer 2331 (9.9) 2211 (9.4) 0.0173

Anaemia 394 (1.7) 392 (1.7) 0.0007

Thyroid disease 2537 (10.8) 2395 (10.2) 0.0197




Medications, n (%)


Page 54 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


18

Sulfonylureas

(n = 23,555)

Metformin

(n = 23,555) SMD


Beta-blockers 6123 (26.0) 5640 (23.9) 0.0474


Diuretics 8393 (35.6) 7942 (33.7) 0.0402


Nitrates 1462 (6.2) 1355 (5.8) 0.0192

Statins 16,382 (69.5) 15,833 (67.2) 0.0501


Clopidogrel 875 (3.7) 715 (3.0) 0.0376

Warfarin 1367 (5.8) 1252 (5.3) 0.0213

Acetaminophen 8329 (35.4) 7810 (33.2) 0.0464


Opioids 7542 (32.0) 6964 (29.6) 0.0532


Neuropathy 3042 (12.9) 2956 (12.5) 0.0110


Nephropathy 3100 (13.2) 2579 (10.9) 0.0680

Retinopathy 5516 (23.4) 5390 (22.9) 0.0127






Page 55 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


19

eTable 5. Characteristics of patients adding sulfonylureas to metformin and patients

switching to sulfonylureas from metformin (myocardial infarction study cohort)

Adding

sulfonylureas

(n = 13,210)

Switching to

sulfonylureas

(n = 9765)

SMD

Age (years), mean (SD) 62.3 (11.1) 66.4 (11.7) -0.3557

Female sex, n (%) 5256 (39.8) 4720 (48.3) -0.1728

Diabetes duration (years), mean (SD) 3.4 (3.4) 3.6 (3.8) -0.0502



Ever 7351 (55.6) 5239 (53.7) 0.0401

Never 5748 (43.5) 4427 (45.3) -0.0367

Unknown 111 (0.8) 99 (1.0) -0.0181


<25 1303 (9.9) 1301 (13.3) -0.1082

25-30 4297 (32.5) 3429 (35.1) -0.0547

≥30 7392 (56.0) 4830 (49.5) 0.1304

Missing 218 (1.7) 205 (2.1) -0.0331

Haemoglobin A1c (%), mean (SD) 9.0 (1.7) 8.4 (1.6) 0.3120

≤7%, n (%) 644 (4.9) 1325 (13.6) -0.3039

7.1-8.0%, n (%) 3902 (29.5) 3099 (31.7) -0.0477

>8%, n (%) 7647 (57.9) 4412 (45.2) 0.2563

Unknown, n (%) 1017 (7.7) 929 (9.5) -0.0647


Heart failure 734 (5.6) 854 (8.7) -0.1240

Arterial hypertension 8142 (61.6) 6558 (67.2) -0.1155

Coronary artery disease 2675 (20.2) 2362 (24.2) -0.0948

Atrial fibrillation or flutter 1055 (8.0) 1094 (11.2) -0.1094

Hyperlipidaemia 4066 (30.8) 3172 (32.5) -0.0366

Chronic obstructive pulmonary disease 1770 (13.4) 1522 (15.6) -0.0622

Cancer 1122 (8.5) 1131 (11.6) -0.1029

Anaemia 186 (1.4) 192 (2.0) -0.0434

Thyroid disease 1291 (9.8) 1153 (11.8) -0.0656

Severe hypoglycaemia 16 (0.1) 15 (0.2) -0.0088

Myocardial infarction 619 (4.7) 534 (5.5) -0.0357

Ischaemic stroke 231 (1.7) 190 (1.9) -0.0146

Medications, n (%)

Angiotensin-converting enzyme inhibitors 5946 (45.0) 4311 (44.1) 0.0174

Page 56 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


20

Adding

sulfonylureas

(n = 13,210)

Switching to

sulfonylureas

(n = 9765)

SMD

Angiotensin II receptor blockers 1693 (12.8) 1538 (15.8) -0.0839

Beta-blockers 3199 (24.2) 2777 (28.4) -0.0960

Calcium-channel blockers 3630 (27.5) 2910 (29.8) -0.0514

Diuretics 4286 (32.4) 3880 (39.7) -0.1522

Cardiac glycosides 505 (3.8) 546 (5.6) -0.0836

Nitrates 708 (5.4) 707 (7.2) -0.0775

Statins 9520 (72.1) 6494 (66.5) 0.1208


Clopidogrel 405 (3.1) 440 (4.5) -0.0755

Warfarin 647 (4.9) 689 (7.1) -0.0911

Acetaminophen 4363 (33.0) 3769 (38.6) -0.1164


Opioids 3979 (30.1) 3348 (34.3) -0.0892


Neuropathy 1608 (12.2) 1361 (13.9) -0.0524

Peripheral vascular disease 739 (5.6) 665 (6.8) -0.0504

Nephropathy 1401 (10.6) 1627 (16.7) -0.1772

Retinopathy 3071 (23.2) 2353 (24.1) -0.0200


Page 57 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


21


switching to sulfonylureas from metformin (ischaemic stroke study cohort)

Adding

sulfonylureas

(n = 13,307)

Switching to

sulfonylureas

(n = 9776)

SMD

Age (years), mean (SD) 62.3 (11.1) 66.4 (11.7) -0.3552

Female sex, n (%) 5287 (39.7) 4752 (48.6) -0.1795




Ever 7402 (55.6) 5228 (53.5) 0.0431

Never 5793 (43.5) 4450 (45.5) -0.0400

Unknown 112 (0.8) 98 (1.0) -0.0168


<25 1307 (9.8) 1308 (13.4) -0.1113

25-30 4326 (32.5) 3425 (35.0) -0.0534

≥30 7460 (56.1) 4840 (49.5) 0.1315

Missing 214 (1.6) 203 (2.1) -0.0348


≤7%, n (%) 642 (4.8) 1327 (13.6) -0.3063

7.1-8.0%, n (%) 3899 (29.3) 3103 (31.7) -0.0530

>8%, n (%) 7747 (58.2) 4413 (45.1) 0.2639

Unknown, n (%) 1019 (7.7) 933 (9.5) -0.0673


Heart failure 741 (5.6) 849 (8.7) -0.1213






Cancer 1129 (8.5) 1135 (11.6) -0.1041

Anaemia 185 (1.4) 200 (2.0) -0.0505

Thyroid disease 1300 (9.8) 1167 (11.9) -0.0697




Medications, n (%)


Page 58 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


22

Adding

sulfonylureas

(n = 13,307)

Switching to

sulfonylureas

(n = 9776)

SMD


Beta-blockers 3240 (24.3) 2787 (28.5) -0.0945


Diuretics 4319 (32.5) 3897 (39.9) -0.1546


Nitrates 720 (5.4) 711 (7.3) -0.0765

Statins 9599 (72.1) 6501 (66.5) 0.1224


Clopidogrel 415 (3.1) 449 (4.6) -0.0766

Warfarin 636 (4.8) 691 (7.1) -0.0971

Acetaminophen 4382 (32.9) 3778 (38.6) -0.1194


Opioids 4008 (30.1) 365 (34.4) -0.0921


Neuropathy 1616 (12.1) 1366 (14.0) -0.0543


Nephropathy 1411 (10.6) 1634 (16.7) -0.1787

Retinopathy 3098 (23.3) 2366 (24.2) -0.0216


Page 59 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


23


switching to sulfonylureas from metformin (cardiovascular death study cohort)

Adding

sulfonylureas

(n = 13,224)

Switching to

sulfonylureas

(n = 9784)

SMD

Age (years), mean (SD) 62.3 (11.1) 66.3 (11.7) -0.3516

Female sex, n (%) 5268 (39.8) 4754 (48.6) -0.1769




Ever 7349 (55.6) 5254 (53.7) 0.0376

Never 5764 (43.6) 4430 (45.3) -0.0340

Unknown 111 (0.8) 100 (1.0) -0.0190


<25 1298 (9.8) 1305 (13.3) -0.1103

25-30 4310 (32.6) 3439 (35.1) -0.0540

≥30 7403 (56.0) 4834 (49.4) 0.1320

Missing 213 (1.6) 206 (2.1) -0.0366


≤7%, n (%) 642 (4.9) 1325 (13.5) -0.3041

7.1-8.0%, n (%) 3895 (29.5) 3117 (31.9) -0.0522

>8%, n (%) 7671 (58.0) 4408 (45.1) 0.2614

Unknown, n (%) 1016 (7.7) 934 (9.5) -0.0664


Heart failure 738 (5.6) 858 (8.8) -0.1238






Cancer 1110 (8.4) 1131 (11.6) -0.1058

Anaemia 180 (1.4) 196 (2.0) -0.0499

Thyroid disease 1298 (9.8) 1164 (11.9) -0.0669




Medications, n (%)


Page 60 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


24

Adding

sulfonylureas

(n = 13,224)

Switching to

sulfonylureas

(n = 9784)

SMD


Beta-blockers 3196 (24.2) 2795 (28.6) -0.1000


Diuretics 4267 (32.3) 3898 (39.8) -0.1582


Nitrates 712 (5.4) 711 (7.3) -0.0774

Statins 9517 (72.0) 6515 (66.6) 0.1168


Clopidogrel 405 (3.1) 448 (4.6) -0.0792

Warfarin 641 (4.8) 691 (7.1) -0.0937

Acetaminophen 4347 (32.9) 3774 (38.6) -0.1192


Opioids 3960 (29.9) 3366 (34.4) -0.0955


Neuropathy 1602 (12.1) 1366 (14.0) -0.0549


Nephropathy 1400 (10.6) 1626 (16.6) -0.1766

Retinopathy 3067 (23.2) 2368 (24.2) -0.0238


Page 61 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


25


switching to sulfonylureas from metformin (all-cause mortality study cohort)

Adding

sulfonylureas

(n = 13,247)

Switching to

sulfonylureas

(n = 9805)

SMD

Age (years), mean (SD) 62.3 (11.1) 66.3 (11.7) -0.3497

Female sex, n (%) 5277 (39.8) 4756 (48.5) -0.1753




Ever 7352 (55.5) 5274 (53.8) 0.0344

Never 5784 (43.7) 4432 (45.2) -0.0310

Unknown 111 (0.8) 99 (1.0) -0.0180


<25 1307 (9.9) 1304 (13.3) -0.1074

25-30 4310 (32.5) 3438 (35.1) -0.0535

≥30 7416 (56.0) 4858 (49.5) 0.1292

Missing 214 (1.6) 205 (2.1) -0.0352


≤7%, n (%) 641 (4.8) 1327 (13.5) -0.3045

7.1-8.0%, n (%) 3885 (29.3) 3122 (31.8) -0.0546

>8%, n (%) 7707 (58.2) 4430 (45.2) 0.2623

Unknown, n (%) 1014 (7.7) 926 (9.4) -0.0640


Heart failure 733 (5.5) 844 (8.6) -0.1202






Cancer 1121 (8.5) 1126 (11.5) -0.1010

Anaemia 189 (1.4) 198 (2.0) -0.0456

Thyroid disease 1300 (9.8) 1167 (11.9) -0.0672




Medications, n (%)


Page 62 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


26

Adding

sulfonylureas

(n = 13,247)

Switching to

sulfonylureas

(n = 9805)

SMD


Beta-blockers 3231 (24.4) 2806 (28.6) -0.0959


Diuretics 4307 (32.5) 3903 (39.8) -0.1522


Nitrates 721 (5.4) 709 (7.2) -0.0735

Statins 9550 (72.1) 6529 (66.6) 0.1196


Clopidogrel 406 (3.1) 443 (4.5) -0.0761

Warfarin 640 (4.8) 691 (7.0) -0.0939

Acetaminophen 4364 (32.9) 3770 (38.4) -0.1151


Opioids 3960 (29.9) 3371 (34.4) -0.0962


Neuropathy 1610 (12.2) 1371 (14.0) -0.0543


Nephropathy 1383 (10.4) 1627 (16.6) -0.1807

Retinopathy 3087 (23.3) 2374 (24.2) -0.0214


Page 63 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


27


switching to sulfonylureas from metformin (severe hypoglycaemia study cohort)

Adding

sulfonylureas

(n = 13,222)

Switching to

sulfonylureas

(n = 9775)

SMD

Age (years), mean (SD) 62.3 (11.1) 66.4 (11.7) -0.3560

Female sex, n (%) 5265 (39.8) 4740 (48.5) -0.1753




Ever 7357 (55.6) 5247 (53.7) 0.0395

Never 5754 (43.5) 4428 (45.3) -0.0358

Unknown 111 (0.8) 100 (1.0) -0.0191


<25 1311 (9.9) 1304 (13.3) -0.1070

25-30 4310 (32.6) 3431 (35.1) -0.0529

≥30 7386 (55.9) 4837 (49.5) 0.1280

Missing 215 (1.6) 203 (2.1) -0.0334


≤7%, n (%) 638 (4.8) 1336 (13.7) -0.3089

7.1-8.0%, n (%) 3883 (29.4) 3108 (31.8) -0.0527

>8%, n (%) 7684 (58.1) 4405 (45.1) 0.2634

Unknown, n (%) 1017 (7.7) 926 (9.5) -0.0636


Heart failure 722 (5.5) 861 (8.8) -0.1303






Cancer 1118 (8.5) 1124 (11.5) -0.1017

Anaemia 189 (1.4) 195 (2.0) -0.0436

Thyroid disease 1311 (9.9) 1165 (11.9) -0.0643




Medications, n (%)


Page 64 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


28

Adding

sulfonylureas

(n = 13,222)

Switching to

sulfonylureas

(n = 9775)

SMD


Beta-blockers 3194 (24.2) 2795 (28.6) -0.1008


Diuretics 4288 (32.4) 3894 (39.8) -0.1546


Nitrates 716 (5.4) 720 (7.4) -0.0798

Statins 9537 (72.1) 6507 (66.6) 0.1209


Clopidogrel 400 (3.0) 449 (4.6) -0.0820

Warfarin 638 (4.8) 686 (7.0) -0.0930

Acetaminophen 4333 (32.8) 3769 (38.6) -0.1210


Opioids 3955 (29.9) 3364 (34.4) -0.0965


Neuropathy 1609 (12.2) 1374 (14.1) -0.0559


Nephropathy 1387 (10.5) 1630 (16.7) -0.1813

Retinopathy 3054 (23.1) 2356 (24.1) -0.0237


Page 65 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


29

eTable 10. Crude and adjusted hazard ratios for the association between switching versus adding sulfonylureas and the risk of

the study outcomes



Crude HR

(95% CI)

Adjusted HR

(95% CI)*


Adding sulfonylureas 13,203 57 11,442 5.0 (3.8 to 6.5) 1.00 (reference) 1.00 (reference)

Switching to sulfonylureas 9759 68 5138 13.2 (10.4 to 16.8) 2.65 (1.86 to 3.78) 1.51 (1.03 to 2.24)

Ischaemic stroke






Alltocause mortality




Adding sulfonylureas 13,215 39 11440 3.4 (2.5 to 4.7) 1.00 (reference) 1.00 (reference)



propensity score deciles, history of the respective outcome in the year prior to cohort entry (or, for the case of cardiovascular death,

history of myocardial infarction or ischaemic stroke), body mass index category, diuretics, statins, acetaminophen, opioids, and

nephropathy. The model for all-cause mortality was adjusted for age, sex, propensity score deciles, body mass index category,

diuretics, statins, acetaminophen, opioids, and nephropathy.

Abbreviations: CI = confidence interval; HR = hazard ratio.

Page 66 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


30

eTable 11. Crude and adjusted hazard ratios for the association between the use of sulfonylureas and the risk of myocardial

infarction according to duration of use



Crude HR

(95% CI)

Adjusted HR

(95% CI)*



≤3 months of use

Metformin 23,551 33 5122 6.4 (4.6 to 9.1) 1.00 (reference) 1.00 (reference)

Sulfonylureas 23,551 50 5267 9.5 (7.2 to 12.5) 1.47 (0.95 to 2.29) 1.45 (0.93 to 2.25)

3.1-12 months of use



>12 months of use

Metformin 7133 78 11,558 6.7 (5.4 to 8.4) 1.00 (reference) 1.00 (reference)

Sulfonylureas 6548 77 10,900 7.1 (5.7 to 8.8) 1.04 (0.76 to 1.43) 1.05 (0.77 to 1.45)

* The model was adjusted for age, sex, hdPS deciles, and history of myocardial infarction in the year prior to cohort entry.


Page 67 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


31

eTable 12. Crude and adjusted hazard ratios for the association between the use of sulfonylureas and the risk of ischaemic

stroke according to duration of use



Crude HR

(95% CI)

Adjusted HR

(95% CI)*



≤3 months of use






>12 months of use



* The model was adjusted for age, sex, hdPS deciles, and history of ischemic stroke in the year prior to cohort entry.


Page 68 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


32

eTable 13. Crude and adjusted hazard ratios for the association between the use of sulfonylureas and the risk of

cardiovascular death according to duration of use



Crude HR

(95% CI)

Adjusted HR

(95% CI)*



≤3 months of use






>12 months of use



* The model was adjusted for age, sex, hdPS deciles, and history of myocardial infarction or ischemic stroke in the year prior to cohort

entry.


Page 69 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


33

eTable 14. Crude and adjusted hazard ratios for the association between the use of sulfonylureas and the risk of all-cause

mortality according to duration of use



Crude HR

(95% CI)

Adjusted HR

(95% CI)*



≤3 months of use






>12 months of use



* The model was adjusted for age, sex, and hdPS deciles.


Page 70 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


34

eTable 15. Crude and adjusted hazard ratios for the association between the use of sulfonylureas and the risk of severe

hypoglycaemia according to duration of use



Crude HR

(95% CI)

Adjusted HR

(95% CI)*



≤3 months of use






>12 months of use



* The model was adjusted for age, sex, hdPS deciles, and history of severe hypoglycaemia in the year prior to cohort entry.


Page 71 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


35

eTable 16. Crude and adjusted hazard ratios for the association between the use of different sulfonylureas classified by

duration of action and pancreas specificity and the risk of the study outcomes



Crude HR

(95% CI)

Adjusted HR

(95% CI)*



Specific, short-acting sulfonylureas** 21,365 164 21,102 7.8 (6.7 to 9.1) 1.26 (1.01 to 1.57) 1.23 (0.99 to 1.54)

Non-specific, long-acting sulfonylureas*** 2185 18 2493 7.2 (4.5 to 11.5) 1.17 (0.72 to 1.91) 1.37 (0.84 to 2.23)

Ischaemic stroke








All-cause mortality










myocardial infarction or ischemic stroke). The model for all-cause mortality was adjusted for age, sex, and hdPS deciles.

Page 72 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


36

** Pancreas specific, short-acting sulfonylureas included gliclazide, glipizide, and tolbutamide.

*** Pancreas non-specific, long-acting sulfonylureas included glyburide and glimepiride.


Page 73 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


37

eTable 17. Crude and adjusted hazard ratios for the association between the use of sulfonylureas and the risk of the study

outcomes (60-day grace period)



Crude HR

(95% CI)

Adjusted HR

(95% CI)*




Ischaemic stroke






All-cause mortality








myocardial infarction or ischemic stroke). The model for all-cause mortality was adjusted for age, sex, and hdPS deciles.


Page 74 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


38


outcomes (exclusion of previous outcomes)



Crude HR

(95% CI)

Adjusted HR

(95% CI)*


Metformin 23316 143 24418 5.9 (5.0 to 6.9) 1.00 (reference) 1.00 (reference)

Sulfonylureas 23261 174 23542 7.4 (6.4 to 8.6) 1.26 (1.01 to 1.57) 1.28 (1.02 to 1.60)

Ischaemic stroke









* The models were adjusted for age, sex, and hdPS deciles.


Page 75 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


39


outcomes (adjusting for additional covariates)



Crude HR

(95% CI)

Adjusted HR

(95% CI)*




Ischaemic stroke






All-cause mortality







hdPS deciles, history of the respective outcome in the year prior to cohort entry (or, for the case of cardiovascular death, history of

myocardial infarction or ischemic stroke), body mass index category, diuretics, statins, acetaminophen, opioids, and nephropathy. The

model for all-cause mortality was adjusted for age, sex, hdPS deciles, body mass index category, diuretics, statins, acetaminophen,

opioids, and nephropathy.


Page 76 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


40

eTable 20. Bounds on corrected estimates and 95% confidence intervals for unmeasured confounding for the analysis on the

risk of myocardial infarction (sensitivity analysis without assumptions)*

1.2 1.3 1.5 1.8 2.0 2.5 3.0 4.0 5.0 6.0 8.0 10.0

1.2 1.23 (0.98 to

1.52)

1.21 (0.97 to

1.50)

1.19 (0.95 to

1.47)

1.17 (0.94 to

1.44)

1.16 (0.93 to

1.43)

1.13 (0.91 to

1.40)

1.12 (0.90 to

1.39)

1.10 (0.88 to

1.37)

1.09 (0.88 to

1.35)

1.09 (0.87 to

1.34)

1.08 (0.86 to

1.33)

1.07 (0.86 to

1.33)

1.3 1.21 (0.97 to

1.50)

1.19 (0.96 to

1.48)

1.16 (0.93 to

1.44)

1.13 (0.91 to

1.40)

1.11 (0.89 to

1.38)

1.09 (0.87 to

1.34)

1.07 (0.85 to

1.32)

1.04 (0.84 to

1.29)

1.03 (0.82 to

1.27)

1.02 (0.82 to

1.26)

1.01 (0.81 to

1.25)

1.00 (0.80 to

1.24)

1.5 1.19 (0.95 to

1.47)

1.16 (0.93 to

1.44)

1.12 (0.90 to

1.39)

1.07 (0.86 to

1.33)

1.05 (0.84 to

1.30)

1.01 (0.81 to

1.25)

0.98 (0.79 to

1.21)

0.95 (0.76 to

1.17)

0.92 (0.74 to

1.14)

0.91 (0.73 to

1.13)

0.89 (0.72 to

1.11)

0.88 (0.71 to

1.09)

1.8 1.17 (0.94 to

1.44)

1.13 (0.91 to

1.40)

1.07 (0.86 to

1.33)

1.01 (0.81 to

1.25)

0.98 (0.79 to

1.21)

0.92 (0.74 to

1.14)

0.89 (0.71 to

1.10)

0.84 (0.67 to

1.04)

0.81 (0.65 to

1.01)

0.79 (0.64 to

0.98)

0.77 (0.62 to

0.95)

0.76 (0.61 to

0.94)

2.0 1.16 (0.93 to

1.43)

1.11 (0.89 to

1.38)

1.05 (0.84 to

1.30)

0.98 (0.79 to

1.21)

0.95 (0.76 to

1.17)

0.88 (0.71 to

1.09)

0.84 (0.67 to

1.04)

0.79 (0.63 to

0.98)

0.76 (0.61 to

0.94)

0.74 (0.59 to

0.91)

0.71 (0.57 to

0.88)

0.69 (0.56 to

0.86)

2.5 1.13 (0.91 to

1.40)

1.09 (0.87 to

1.34)

1.01 (0.81 to

1.25)

0.92 (0.74 to

1.14)

0.88 (0.71 to

1.09)

0.81 (0.65 to

1.00)

0.76 (0.61 to

0.94)

0.69 (0.56 to

0.86)

0.66 (0.53 to

0.81)

0.63 (0.51 to

0.78)

0.60 (0.48 to

0.74)

0.58 (0.46 to

0.72)

3.0 1.12 (0.90 to

1.39)

1.07 (0.85 to

1.32)

0.98 (0.79 to

1.21)

0.89 (0.71 to

1.10)

0.84 (0.67 to

1.04)

0.76 (0.61 to

0.94)

0.70 (0.56 to

0.87)

0.63 (0.51 to

0.78)

0.59 (0.47 to

0.73)

0.56 (0.45 to

0.69)

0.53 (0.42 to

0.65)

0.50 (0.40 to

0.62)

4.0 1.10 (0.88 to

1.37)

1.04 (0.84 to

1.29)

0.95 (0.76 to

1.17)

0.84 (0.67 to

1.04)

0.79 (0.63 to

0.98)

0.69 (0.56 to

0.86)

0.63 (0.51 to

0.78)

0.55 (0.44 to

0.68)

0.50 (0.40 to

0.62)

0.47 (0.38 to

0.59)

0.43 (0.35 to

0.54)

0.41 (0.33 to

0.51)

5.0 1.09 (0.88 to

1.35)

1.03 (0.82 to

1.27)

0.92 (0.74 to

1.14)

0.81 (0.65 to

1.01)

0.76 (0.61 to

0.94)

0.66 (0.53 to

0.81)

0.59 (0.47 to

0.73)

0.50 (0.40 to

0.62)

0.45 (0.36 to

0.56)

0.42 (0.34 to

0.52)

0.38 (0.30 to

0.47)

0.35 (0.28 to

0.44)

6.0 1.09 (0.87 to

1.34)

1.02 (0.82 to

1.26)

0.91 (0.73 to

1.13)

0.79 (0.64 to

0.98)

0.74 (0.59 to

0.91)

0.63 (0.51 to

0.78)

0.56 (0.45 to

0.69)

0.47 (0.38 to

0.59)

0.42 (0.34 to

0.52)

0.39 (0.31 to

0.48)

0.34 (0.27 to

0.42)

0.32 (0.25 to

0.39)

8.0 1.08 (0.86 to

1.33)

1.01 (0.81 to

1.25)

0.89 (0.72 to

1.11)

0.77 (0.62 to

0.95)

0.71 (0.57 to

0.88)

0.60 (0.48 to

0.74)

0.53 (0.42 to

0.65)

0.43 (0.35 to

0.54)

0.38 (0.30 to

0.47)

0.34 (0.27 to

0.42)

0.30 (0.24 to

0.37)

0.27 (0.21 to

0.33)

10.0 1.07 (0.86 to

1.33)

1.00 (0.80 to

1.24)

0.88 (0.71 to

1.09)

0.76 (0.61 to

0.94)

0.69 (0.56 to

0.86)

0.58 (0.46 to

0.72)

0.50 (0.40 to

0.62)

0.41 (0.33 to

0.51)

0.35 (0.28 to

0.44)

0.32 (0.25 to

0.39)

0.27 (0.21 to

0.33)

0.24 (0.19 to

0.30)

* Rows correspond to increasing strength of the risk ratio of unmeasured confounding on the outcome and columns correspond to

increasing strength of risk ratio of unmeasured confounding on the exposure.

Page 77 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


41


risk of all-cause mortality (sensitivity analysis without assumptions)*

1.2 1.3 1.5 1.8 2.0 2.5 3.0 4.0 5.0 6.0 8.0 10.0

1.2 1.24 (1.12 to

1.40)

1.23 (1.11 to

1.38)

1.21 (1.09 to

1.36)

1.19 (1.06 to

1.33)

1.17 (1.05 to

1.32)

1.15 (1.04 to

1.30)

1.14 (1.02 to

1.28)

1.12 (1.01 to

1.26)

1.11 (1.00 to

1.25)

1.10 (0.99 to

1.24)

1.09 (0.98 to

1.23)

1.09 (0.98 to

1.22)

1.3 1.23 (1.11 to

1.38)

1.21 (1.09 to

1.36)

1.18 (1.06 to

1.33)

1.15 (1.03 to

1.29)

1.13 (1.02 to

1.27)

1.10 (0.99 to

1.24)

1.08 (0.97 to

1.22)

1.06 (0.95 to

1.19)

1.04 (0.94 to

1.17)

1.03 (0.93 to

1.16)

1.02 (0.92 to

1.15)

1.01 (0.91 to

1.14)

1.5 1.21 (1.09 to

1.36)

1.18 (1.06 to

1.33)

1.14 (1.02 to

1.28)

1.09 (0.98 to

1.23)

1.07 (0.96 to

1.2)

1.02 (0.92 to

1.15)

1.00 (0.89 to

1.12)

0.96 (0.86 to

1.08)

0.94 (0.84 to

1.06)

0.92 (0.83 to

1.04)

0.91 (0.81 to

1.02)

0.90 (0.81 to

1.01)

1.8 1.19 (1.06 to

1.33)

1.15 (1.03 to

1.29)

1.09 (0.98 to

1.23)

1.03 (0.92 to

1.16)

1.00 (0.89 to

1.12)

0.94 (0.84 to

1.06)

0.90 (0.81 to

1.01)

0.85 (0.77 to

0.96)

0.82 (0.74 to

0.93)

0.81 (0.72 to

0.91)

0.78 (0.70 to

0.88)

0.77 (0.69 to

0.86)

2.0 1.17 (1.05 to

1.32)

1.13 (1.02 to

1.27)

1.07 (0.96 to

1.20)

1.00 (0.89 to

1.12)

0.96 (0.86 to

1.08)

0.90 (0.81 to

1.01)

0.85 (0.77 to

0.96)

0.80 (0.72 to

0.90)

0.77 (0.69 to

0.86)

0.75 (0.67 to

0.84)

0.72 (0.65 to

0.81)

0.70 (0.63 to

0.79)

2.5 1.15 (1.04 to

1.30)

1.10 (0.99 to

1.24)

1.02 (0.92 to

1.15)

0.94 (0.84 to

1.06)

0.90 (0.81 to

1.01)

0.82 (0.74 to

0.92)

0.77 (0.69 to

0.86)

0.70 (0.63 to

0.79)

0.67 (0.60 to

0.75)

0.64 (0.58 to

0.72)

0.61 (0.55 to

0.68)

0.59 (0.53 to

0.66)

3.0 1.14 (1.02 to

1.28)

1.08 (0.97 to

1.22)

1.00 (0.89 to

1.12)

0.90 (0.81 to

1.01)

0.85 (0.77 to

0.96)

0.77 (0.69 to

0.86)

0.71 (0.64 to

0.80)

0.64 (0.58 to

0.72)

0.60 (0.54 to

0.67)

0.57 (0.51 to

0.64)

0.53 (0.48 to

0.60)

0.51 (0.46 to

0.58)

4.0 1.12 (1.01 to

1.26)

1.06 (0.95 to

1.19)

0.96 (0.86 to

1.08)

0.85 (0.77 to

0.96)

0.80 (0.72 to

0.90)

0.70 (0.63 to

0.79)

0.64 (0.58 to

0.72)

0.56 (0.50 to

0.63)

0.51 (0.46 to

0.58)

0.48 (0.43 to

0.54)

0.44 (0.40 to

0.50)

0.42 (0.37 to

0.47)

5.0 1.11 (1.00 to

1.25)

1.04 (0.94 to

1.17)

0.94 (0.84 to

1.06)

0.82 (0.74 to

0.93)

0.77 (0.69 to

0.86)

0.67 (0.60 to

0.75)

0.60 (0.54 to

0.67)

0.51 (0.46 to

0.58)

0.46 (0.41 to

0.52)

0.43 (0.38 to

0.48)

0.38 (0.35 to

0.43)

0.36 (0.32 to

0.40)

6.0 1.10 (0.99 to

1.24)

1.03 (0.93 to

1.16)

0.92 (0.83 to

1.04)

0.81 (0.72 to

0.91)

0.75 (0.67 to

0.84)

0.64 (0.58 to

0.72)

0.57 (0.51 to

0.64)

0.48 (0.43 to

0.54)

0.43 (0.38 to

0.48)

0.39 (0.35 to

0.44)

0.35 (0.31 to

0.39)

0.32 (0.29 to

0.36)

8.0 1.09 (0.98 to

1.23)

1.02 (0.92 to

1.15)

0.91 (0.81 to

1.02)

0.78 (0.70 to

0.88)

0.72 (0.65 to

0.81)

0.61 (0.55 to

0.68)

0.53 (0.48 to

0.60)

0.44 (0.40 to

0.50)

0.38 (0.35 to

0.43)

0.35 (0.31 to

0.39)

0.30 (0.27 to

0.34)

0.27 (0.24 to

0.31)

10.0 1.09 (0.98 to

1.22)

1.01 (0.91 to

1.14)

0.90 (0.81 to

1.01)

0.77 (0.69 to

0.86)

0.70 (0.63 to

0.79)

0.59 (0.53 to

0.66)

0.51 (0.46 to

0.58)

0.42 (0.37 to

0.47)

0.36 (0.32 to

0.40)

0.32 (0.29 to

0.36)

0.27 (0.24 to

0.31)

0.24 (0.22 to

0.27)



Page 78 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


42


risk of severe hypoglycaemia (sensitivity analysis without assumptions)*

1.2 1.3 1.5 1.8 2.0 2.5 3.0 4.0 5.0 6.0 8.0 10.0

1.2 7.38 (4.51 to

12.08)

7.30 (4.46 to

11.95)

7.17 (4.38 to

11.74)

7.03 (4.30 to

11.51)

6.96 (4.25 to

11.39)

6.83 (4.18 to

11.19)

6.75 (4.12 to

11.05)

6.64 (4.06 to

10.88)

6.58 (4.02 to

10.77)

6.54 (4.00 to

10.70)

6.48 (3.96 to

10.62)

6.45 (3.94 to

10.57)

1.3 7.30 (4.46 to

11.95)

7.19 (4.39 to

11.77)

7.01 (4.28 to

11.47)

6.81 (4.16 to

11.16)

6.71 (4.10 to

11.00)

6.54 (4.00 to

10.71)

6.42 (3.93 to

10.52)

6.28 (3.84 to

10.28)

6.19 (3.78 to

10.14)

6.13 (3.75 to

10.04)

6.06 (3.70 to

9.92)

6.01 (3.68 to

9.85)

1.5 7.17 (4.38 to

11.74)

7.01 (4.28 to

11.47)

6.75 (4.12 to

11.05)

6.47 (3.95 to

10.59)

6.33 (3.87 to

10.36)

6.07 (3.71 to

9.94)

5.90 (3.61 to

9.67)

5.69 (3.48 to

9.32)

5.57 (3.40 to

9.12)

5.48 (3.35 to

8.98)

5.38 (3.29 to

8.80)

5.31 (3.25 to

8.70)

1.8 7.03 (4.30 to

11.51)

6.81 (4.16 to

11.16)

6.47 (3.95 to

10.59)

6.09 (3.72 to

9.97)

5.90 (3.61 to

9.67)

5.57 (3.40 to

9.12)

5.34 (3.27 to

8.75)

5.06 (3.09 to

8.29)

4.89 (2.99 to

8.01)

4.78 (2.92 to

7.83)

4.64 (2.84 to

7.60)

4.55 (2.78 to

7.46)

2.0 6.96 (4.25 to

11.39)

6.71 (4.10 to

11.00)

6.33 (3.87 to

10.36)

5.90 (3.61 to

9.67)

5.69 (3.48 to

9.32)

5.31 (3.25 to

8.70)

5.06 (3.09 to

8.29)

4.74 (2.90 to

7.77)

4.55 (2.78 to

7.46)

4.43 (2.71 to

7.25)

4.27 (2.61 to

6.99)

4.17 (2.55 to

6.84)

2.5 6.83 (4.18 to

11.19)

6.54 (4.00 to

10.71)

6.07 (3.71 to

9.94)

5.57 (3.40 to

9.12)

5.31 (3.25 to

8.70)

4.86 (2.97 to

7.96)

4.55 (2.78 to

7.46)

4.17 (2.55 to

6.84)

3.95 (2.41 to

6.46)

3.80 (2.32 to

6.22)

3.61 (2.20 to

5.90)

3.49 (2.13 to

5.72)

3.0 6.75 (4.12 to

11.05)

6.42 (3.93 to

10.52)

5.90 (3.61 to

9.67)

5.34 (3.27 to

8.75)

5.06 (3.09 to

8.29)

4.55 (2.78 to

7.46)

4.22 (2.58 to

6.91)

3.80 (2.32 to

6.22)

3.54 (2.17 to

5.80)

3.37 (2.06 to

5.52)

3.16 (1.93 to

5.18)

3.04 (1.86 to

4.97)

4.0 6.64 (4.06 to

10.88)

6.28 (3.84 to

10.28)

5.69 (3.48 to

9.32)

5.06 (3.09 to

8.29)

4.74 (2.90 to

7.77)

4.17 (2.55 to

6.84)

3.80 (2.32 to

6.22)

3.32 (2.03 to

5.44)

3.04 (1.86 to

4.97)

2.85 (1.74 to

4.66)

2.61 (1.60 to

4.27)

2.47 (1.51 to

4.04)

5.0 6.58 (4.02 to

10.77)

6.19 (3.78 to

10.14)

5.57 (3.40 to

9.12)

4.89 (2.99 to

8.01)

4.55 (2.78 to

7.46)

3.95 (2.41 to

6.46)

3.54 (2.17 to

5.80)

3.04 (1.86 to

4.97)

2.73 (1.67 to

4.47)

2.53 (1.55 to

4.14)

2.28 (1.39 to

3.73)

2.13 (1.30 to

3.48)

6.0 6.54 (4.00 to

10.70)

6.13 (3.75 to

10.04)

5.48 (3.35 to

8.98)

4.78 (2.92 to

7.83)

4.43 (2.71 to

7.25)

3.80 (2.32 to

6.22)

3.37 (2.06 to

5.52)

2.85 (1.74 to

4.66)

2.53 (1.55 to

4.14)

2.32 (1.42 to

3.80)

2.06 (1.26 to

3.37)

1.90 (1.16 to

3.11)

8.0 6.48 (3.96 to

10.62)

6.06 (3.70 to

9.92)

5.38 (3.29 to

8.80)

4.64 (2.84 to

7.60)

4.27 (2.61 to

6.99)

3.61 (2.20 to

5.90)

3.16 (1.93 to

5.18)

2.61 (1.60 to

4.27)

2.28 (1.39 to

3.73)

2.06 (1.26 to

3.37)

1.78 (1.09 to

2.91)

1.61 (0.99 to

2.64)

10.0 6.45 (3.94 to

10.57)

6.01 (3.68 to

9.85)

5.31 (3.25 to

8.70)

4.55 (2.78 to

7.46)

4.17 (2.55 to

6.84)

3.49 (2.13 to

5.72)

3.04 (1.86 to

4.97)

2.47 (1.51 to

4.04)

2.13 (1.30 to

3.48)

1.90 (1.16 to

3.11)

1.61 (0.99 to

2.64)

1.44 (0.88 to

2.36)



Page 79 of 79


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

Confidential: For Review Only - BMJ...Confidential: For Review Only 3 CI, 0.99 to 1.56) and...

Documents

Transcript of Confidential: For Review Only - BMJ...Confidential: For Review Only 3 CI, 0.99 to 1.56) and...