1
Title: Initial therapy, persistence and regimen change in a cohort of newly treated type 2
2
diabetes patients.1
3
Running Title: Antidiabetic therapy use in the treatment naive
4
Author names and affiliations: 1R T Grimes, 2K Bennett, 3L Tilson, 3C Usher, 4 S M Smith,
5
1
Accepted Article
1
M C Henman.
6 7
1. School of Pharmacy and Pharmaceutical Sciences, Trinity College Dublin, Ireland
8
2. Department of Pharmacology and Therapeutics, Trinity Centre for Health Sciences, St
9
James’s Hospital, Dublin 8, Ireland
10
3. National Centre for Pharmacoeconomics, St James’s Hospital, Dublin 8, Ireland
11
4. Department of General Practice, Royal College of Surgeons Ireland, Beaux Lane House,
12
Lower Mercer Street, Dublin 2, Ireland
13 14
Corresponding Author:
15
Ronan Grimes
16
School of Pharmacy and Pharmaceutical Sciences, Trinity College Dublin, Ireland
17
e-mail: [email protected]
18
Telephone: +353863515693
19
Fax: +353 1 896 2810
20
Word Count: 4163
21
Tables and Figures: 4 tables, 0 figures
22
Keywords: Type 2 diabetes mellitus, Real world treatment patterns, Treatment progression This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/bcp.12573
This article is protected by copyright. All rights reserved.
2
Accepted Article
23 24
Structured Summary
25
Aims: To describe the utilisation of antidiabetic agents , in terms of persistence and regimen
26
change in the management of a cohort of newly treated type 2 diabetes patients and to
27
investigate associated socio-demographic and treatment factors.
28
Methods: A population-based retrospective cohort study was conducted using the national
29
pharmacy claims database in Ireland. Subjects were analysed for persistence and regimen
30
change. Cox proportional-hazards regression examined associations of socio-demographic
31
and treatment factors on treatment patterns. Hazard ratios (HR) and 95% CIs are presented.
32
Results: 20947 subjects were identified in the study over a 2 year period. Most were initiated
33
on metformin (76%) or sulphonylureas (22%) and 77% were persistent with therapy 12
34
months after initiation. The likelihood of non-persistence was significantly lower in the
35
youngest (40-49) age groups (ref 60-69) (HR 1.62 [CI 1.42-1.84]) and those on
36
sulphonylureas (HR 1.49 [1.36-1.64]). The likelihood of receiving a regimen change was
37
significantly lower in the older (80+) age groups (HR 0.63 [0.56-0.71]), females (HR 0.91
38
[0.86-0.95], and those with pre-existing CVD (1 vs 0 CVD medicines) (HR 0.82 [0.74-0.90]),
39
and higher in those on sulphonylureas (HR 1.83 [1.73-1.94])
40
Conclusions: Type of treatment, pre-existing CVD and demographic factors are shown to be
41
associated with the observed treatment patterns. Guideline recommended agents were widely
42
used on treatment initiation though a substantial minority were not initiated on the
43
recommended first line agent. Use of guideline recommended agents was not as evident
44
during treatment progression. Further optimisation of initial and subsequent antidiabetic
45
agent prescribing may be possible.
46
This article is protected by copyright. All rights reserved.
3
Accepted Article
47 48
What is already known about this subject:
49
The utilisation patterns of antidiabetic agents for type 2 diabetes have changed over time.
50
Despite published guidelines little is known about antidiabetic agent use in the primary
51 52
care setting.
53
The utilisation patterns of an entire country’s treatment naive type 2 diabetes population
has not been assessed.
54
What this study adds:
55
56
Guideline recommended agents were widely used on treatment initiation though a
substantial minority were not initiated on the recommended first line agent.
57
Choice of initial agent affected the likelihood of and the time to treatment progression.
58
Optimisation of initial and subsequent antidiabetic agent prescribing may be possible.
59 60
This article is protected by copyright. All rights reserved.
4
Accepted Article
61 62
Introduction
63
Clinical trials aim to show the clinical efficacy and safety of a medicine while real world drug
64
utilisation studies show how these drugs are used outside the controlled setting of a clinical
65
trial (1). Numerous classes of antidiabetic agents are licensed to manage type 2 diabetes, all
66
with varying clinical efficacy, profiles and costs (2). Studies have shown a change in specific
67
drug utilisation trends and an increase in prescribing for type 2 diabetes over time (3–7).
68
Guidelines recommending preferred treatment patterns for initial treatment and disease
69
progression are available (2,8,9), however, it has been argued that the benefits and risks of
70
new therapies are not being fully assessed prior to marketing, leaving prescribers with limited
71
information on which to base their treatment decisions (10). In addition, results from several
72
clinical trials have shown limited improvement in cardiovascular disease (CVD) with
73
intensive glucose control (11–13). Pre-existing CVD and management of newly diagnosed
74
type 2 diabetes can therefore be a difficult issue for physicians (14). Little information has
75
been published regarding antidiabetic agent use in the primary care setting amongst newly
76
treated type 2 diabetes patients (15–18), and the majority of information published has come
77
from the United States with no studies having looked at how pre-existing CVD medicine use
78
affects treatment patterns. This study aims to provide prescribers and other health care
79
professionals with an understanding of drug utilisation in newly treated type 2 diabetes, at the
80
population level, in a European country.
81
The primary objectives of this study were: 1) to describe the
82
antidiabetic agents used at initiation and to examine persistence, addition and switching of
83
therapy in the management of a cohort of newly treated type 2 diabetes patients and 2) to
84
investigate socio-demographic factors and utilisation of cardiovascular drugs on the observed
85
treatment patterns. This article is protected by copyright. All rights reserved.
utilisation patterns of
5
Accepted Article
86 87
Subjects, Materials and Methods
88
Study design, subjects and setting
89
A retrospective, observational cohort study was conducted using subjects from the two
90
relevant schemes in the Irish Health Services Executive (HSE) Primary Care Reimbursement
91
Services (PCRS) pharmacy claims database: the General Medical Services (GMS) scheme
92
and the Long-Term Illness (LTI) scheme (19). The HSE-PCRS reimburses pharmacists for
93
dispensed medicines in the primary care setting under various schemes. The GMS scheme is
94
means-tested and provides free primary care, including free prescription medication, for
95
eligible participants. Persons suffering from certain chronic illnesses, including type 2
96
diabetes, are entitled to receive medicines to treat their condition free of charge under the LTI
97
scheme. The HSE-PCRS database records demographic details such as age and sex, in
98
addition to details and dates of drugs dispensed within the scheme. A maximum of one
99
month’s supply of medication can be dispensed at a time on the GMS and LTI scheme. No
100
information on clinical diagnosis or outcomes is recorded. Ethical approval was not required
101
as all analysis was carried out on anonymised data.
102
The study participants were identified from the GMS and LTI schemes and included all
103
newly treated type 2 diabetes patients aged 40 years and over for a two year period from
104
January 1st 2008 to December 31st 2009. Some subjects receiving metformin for the first time
105
may have been receiving it for conditions other than type 2 diabetes (20), excluding subjects
106
aged less than 40 years removes the ambiguity around classification. We defined newly
107
treated subjects as those receiving monotherapy with antidiabetic agents with WHO ATC
108
code A10B who had no antidiabetic medication dispensed in the preceding 12 months.
109
Subjects initially prescribed insulin, more than one antidiabetic agent or a combination agent
This article is protected by copyright. All rights reserved.
6
were excluded from the study as it could not be guaranteed that these subjects were treatment
111
naive. Subjects were followed until treatment discontinuation or November 2012 whichever
112
came first. Date of treatment discontinuation was defined as the day of last medication supply
113
plus the number of day’s medication supplied on that day. Subjects were categorised and
114
analysed according to coverage scheme, sex, age and the initial oral antidiabetic medicine.
Accepted Article
110
115 116
Utilisation of antidiabetic agents
117
We analysed utilisation of antidiabetic agents (WHO ATC code A10) over the study period.
118
In line with a previous study examining newly treated type 2 diabetes non-persistence was
119
defined as a treatment gap of greater than 12 weeks occurring within 365 days of treatment
120
initiation (21). Also, subjects on the GMS scheme can receive 3 monthly prescriptions from
121
their physician, with a maximum of one month’s medication being dispensed at a time, thus
122
persistent patients should have returned to their physician within this 12 week period.
123
Subjects receiving only one dispensing were excluded from persistence analysis. Only
124
subjects initiated on metformin or a sulphonylurea were included for analysis of persistence,
125
addition and switching, as only 2% of subjects were initiated on alternative antidiabetic
126
therapy. Subjects receiving treatment addition or switching were excluded from persistence
127
analysis.
128
Subjects were considered to have received add-on therapy if they were dispensed medication
129
from a different therapeutic class while continuing to receive their initial treatment. Subjects
130
moving from single to triple therapy or from double to triple therapy were excluded from
131
analysis due to the diverse range of regimens being prescribed.
This article is protected by copyright. All rights reserved.
7
Treatment switching was defined as discontinuation of initial antidiabetic agent and initiation
133
on an alternative agent from a different drug class. Subjects switching back to their initial
134
antidiabetic agent within 90 days of switching were not classified as having switched therapy.
135
For the purpose of this study, treatment addition and switching are collectively referred to as
136
regimen change.
137
Within treatment class changes were analysed for the treatment addition and switching
138
groups. Therapy was considered to have changed within its treatments class if a subject
139
received an increased or decreased dose of their antidiabetic agent, or received a different
140
antidiabetic agent within the same treatment class.
141
Statistical methods
142
Descriptive analysis of subject characteristics and initial treatment patterns are presented.
143
Subjects were analysed in four groups: those who remained on initial therapy and were
144
persistent for the first twelve months; those who remained on initial therapy and were non-
145
persistent in the first twelve months; those switching therapy; those receiving add-on therapy.
146
Only subjects remaining on initial therapy were examined for persistence. Time to treatment
147
addition or switching is reported in median number of days and interquartile range (IQR).
148
Cox proportional-hazards regression was used to examine differences in time to treatment
149
addition and switching for those initiated on metformin compared to a sulphonylurea
150
adjusting for age, sex and coverage scheme. Adjusted logistic regression was used to examine
151
the association between initial antidiabetic agent (and age, sex and coverage scheme, odds
152
ratio (OR) with 95% confidence interval (CI) is presented.
153
Cox proportional-hazards regression was used to examine time to (i) treatment non-
154
persistence, (ii) receiving a regimen change, and (iii) dose change prior to treatment
155
switching or treatment addition. . These analyses were adjusted for the following variables of
Accepted Article
132
This article is protected by copyright. All rights reserved.
8
interest and available within the database: duration of follow up, sex, age, coverage scheme
157
and initial antidiabetic agent. Similar variables have found to be of significance in previous
158
studies (15–18) Hazard ratio (HR) and CIs are presented. Metformin (vs sulphonylurea),
159
GMS scheme (vs LTI), males, and subjects aged 60-69 were the reference groups.
160
Using Cox proportional-hazards regression we examined the association of cardiovascular
161
drug use in the GMS population on regimen changes, controlling for age, antidiabetic agent
162
and gender. We used number of different classes of CVD agents prescribed prior to
163
antidiabetic agent initiation as an indication of pre-existing CVD. The following classes of
164
CVD agents were included in our study: Aspirin (B01AC06, N02BA01), Nitrates (C01DA),
165
lipid lowering drugs (Statins (C10AA), Non-Statin Lipid Lowering Drugs (C10AB, A10AC,
166
A10AD, C10AX)), Beta-Blockers (C07), Calcium Channel Blockers (C08), agents acting on
167
the renin-angiotensin system (C09), Diuretics (C03). Analyses were performed using SAS,
168
version 9.2, significance at p
Title: Initial therapy, persistence and regimen change in a cohort of newly treated type 2
2
diabetes patients.1
3
Running Title: Antidiabetic therapy use in the treatment naive
4
Author names and affiliations: 1R T Grimes, 2K Bennett, 3L Tilson, 3C Usher, 4 S M Smith,
5
1
Accepted Article
1
M C Henman.
6 7
1. School of Pharmacy and Pharmaceutical Sciences, Trinity College Dublin, Ireland
8
2. Department of Pharmacology and Therapeutics, Trinity Centre for Health Sciences, St
9
James’s Hospital, Dublin 8, Ireland
10
3. National Centre for Pharmacoeconomics, St James’s Hospital, Dublin 8, Ireland
11
4. Department of General Practice, Royal College of Surgeons Ireland, Beaux Lane House,
12
Lower Mercer Street, Dublin 2, Ireland
13 14
Corresponding Author:
15
Ronan Grimes
16
School of Pharmacy and Pharmaceutical Sciences, Trinity College Dublin, Ireland
17
e-mail: [email protected]
18
Telephone: +353863515693
19
Fax: +353 1 896 2810
20
Word Count: 4163
21
Tables and Figures: 4 tables, 0 figures
22
Keywords: Type 2 diabetes mellitus, Real world treatment patterns, Treatment progression This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/bcp.12573
This article is protected by copyright. All rights reserved.
2
Accepted Article
23 24
Structured Summary
25
Aims: To describe the utilisation of antidiabetic agents , in terms of persistence and regimen
26
change in the management of a cohort of newly treated type 2 diabetes patients and to
27
investigate associated socio-demographic and treatment factors.
28
Methods: A population-based retrospective cohort study was conducted using the national
29
pharmacy claims database in Ireland. Subjects were analysed for persistence and regimen
30
change. Cox proportional-hazards regression examined associations of socio-demographic
31
and treatment factors on treatment patterns. Hazard ratios (HR) and 95% CIs are presented.
32
Results: 20947 subjects were identified in the study over a 2 year period. Most were initiated
33
on metformin (76%) or sulphonylureas (22%) and 77% were persistent with therapy 12
34
months after initiation. The likelihood of non-persistence was significantly lower in the
35
youngest (40-49) age groups (ref 60-69) (HR 1.62 [CI 1.42-1.84]) and those on
36
sulphonylureas (HR 1.49 [1.36-1.64]). The likelihood of receiving a regimen change was
37
significantly lower in the older (80+) age groups (HR 0.63 [0.56-0.71]), females (HR 0.91
38
[0.86-0.95], and those with pre-existing CVD (1 vs 0 CVD medicines) (HR 0.82 [0.74-0.90]),
39
and higher in those on sulphonylureas (HR 1.83 [1.73-1.94])
40
Conclusions: Type of treatment, pre-existing CVD and demographic factors are shown to be
41
associated with the observed treatment patterns. Guideline recommended agents were widely
42
used on treatment initiation though a substantial minority were not initiated on the
43
recommended first line agent. Use of guideline recommended agents was not as evident
44
during treatment progression. Further optimisation of initial and subsequent antidiabetic
45
agent prescribing may be possible.
46
This article is protected by copyright. All rights reserved.
3
Accepted Article
47 48
What is already known about this subject:
49
The utilisation patterns of antidiabetic agents for type 2 diabetes have changed over time.
50
Despite published guidelines little is known about antidiabetic agent use in the primary
51 52
care setting.
53
The utilisation patterns of an entire country’s treatment naive type 2 diabetes population
has not been assessed.
54
What this study adds:
55
56
Guideline recommended agents were widely used on treatment initiation though a
substantial minority were not initiated on the recommended first line agent.
57
Choice of initial agent affected the likelihood of and the time to treatment progression.
58
Optimisation of initial and subsequent antidiabetic agent prescribing may be possible.
59 60
This article is protected by copyright. All rights reserved.
4
Accepted Article
61 62
Introduction
63
Clinical trials aim to show the clinical efficacy and safety of a medicine while real world drug
64
utilisation studies show how these drugs are used outside the controlled setting of a clinical
65
trial (1). Numerous classes of antidiabetic agents are licensed to manage type 2 diabetes, all
66
with varying clinical efficacy, profiles and costs (2). Studies have shown a change in specific
67
drug utilisation trends and an increase in prescribing for type 2 diabetes over time (3–7).
68
Guidelines recommending preferred treatment patterns for initial treatment and disease
69
progression are available (2,8,9), however, it has been argued that the benefits and risks of
70
new therapies are not being fully assessed prior to marketing, leaving prescribers with limited
71
information on which to base their treatment decisions (10). In addition, results from several
72
clinical trials have shown limited improvement in cardiovascular disease (CVD) with
73
intensive glucose control (11–13). Pre-existing CVD and management of newly diagnosed
74
type 2 diabetes can therefore be a difficult issue for physicians (14). Little information has
75
been published regarding antidiabetic agent use in the primary care setting amongst newly
76
treated type 2 diabetes patients (15–18), and the majority of information published has come
77
from the United States with no studies having looked at how pre-existing CVD medicine use
78
affects treatment patterns. This study aims to provide prescribers and other health care
79
professionals with an understanding of drug utilisation in newly treated type 2 diabetes, at the
80
population level, in a European country.
81
The primary objectives of this study were: 1) to describe the
82
antidiabetic agents used at initiation and to examine persistence, addition and switching of
83
therapy in the management of a cohort of newly treated type 2 diabetes patients and 2) to
84
investigate socio-demographic factors and utilisation of cardiovascular drugs on the observed
85
treatment patterns. This article is protected by copyright. All rights reserved.
utilisation patterns of
5
Accepted Article
86 87
Subjects, Materials and Methods
88
Study design, subjects and setting
89
A retrospective, observational cohort study was conducted using subjects from the two
90
relevant schemes in the Irish Health Services Executive (HSE) Primary Care Reimbursement
91
Services (PCRS) pharmacy claims database: the General Medical Services (GMS) scheme
92
and the Long-Term Illness (LTI) scheme (19). The HSE-PCRS reimburses pharmacists for
93
dispensed medicines in the primary care setting under various schemes. The GMS scheme is
94
means-tested and provides free primary care, including free prescription medication, for
95
eligible participants. Persons suffering from certain chronic illnesses, including type 2
96
diabetes, are entitled to receive medicines to treat their condition free of charge under the LTI
97
scheme. The HSE-PCRS database records demographic details such as age and sex, in
98
addition to details and dates of drugs dispensed within the scheme. A maximum of one
99
month’s supply of medication can be dispensed at a time on the GMS and LTI scheme. No
100
information on clinical diagnosis or outcomes is recorded. Ethical approval was not required
101
as all analysis was carried out on anonymised data.
102
The study participants were identified from the GMS and LTI schemes and included all
103
newly treated type 2 diabetes patients aged 40 years and over for a two year period from
104
January 1st 2008 to December 31st 2009. Some subjects receiving metformin for the first time
105
may have been receiving it for conditions other than type 2 diabetes (20), excluding subjects
106
aged less than 40 years removes the ambiguity around classification. We defined newly
107
treated subjects as those receiving monotherapy with antidiabetic agents with WHO ATC
108
code A10B who had no antidiabetic medication dispensed in the preceding 12 months.
109
Subjects initially prescribed insulin, more than one antidiabetic agent or a combination agent
This article is protected by copyright. All rights reserved.
6
were excluded from the study as it could not be guaranteed that these subjects were treatment
111
naive. Subjects were followed until treatment discontinuation or November 2012 whichever
112
came first. Date of treatment discontinuation was defined as the day of last medication supply
113
plus the number of day’s medication supplied on that day. Subjects were categorised and
114
analysed according to coverage scheme, sex, age and the initial oral antidiabetic medicine.
Accepted Article
110
115 116
Utilisation of antidiabetic agents
117
We analysed utilisation of antidiabetic agents (WHO ATC code A10) over the study period.
118
In line with a previous study examining newly treated type 2 diabetes non-persistence was
119
defined as a treatment gap of greater than 12 weeks occurring within 365 days of treatment
120
initiation (21). Also, subjects on the GMS scheme can receive 3 monthly prescriptions from
121
their physician, with a maximum of one month’s medication being dispensed at a time, thus
122
persistent patients should have returned to their physician within this 12 week period.
123
Subjects receiving only one dispensing were excluded from persistence analysis. Only
124
subjects initiated on metformin or a sulphonylurea were included for analysis of persistence,
125
addition and switching, as only 2% of subjects were initiated on alternative antidiabetic
126
therapy. Subjects receiving treatment addition or switching were excluded from persistence
127
analysis.
128
Subjects were considered to have received add-on therapy if they were dispensed medication
129
from a different therapeutic class while continuing to receive their initial treatment. Subjects
130
moving from single to triple therapy or from double to triple therapy were excluded from
131
analysis due to the diverse range of regimens being prescribed.
This article is protected by copyright. All rights reserved.
7
Treatment switching was defined as discontinuation of initial antidiabetic agent and initiation
133
on an alternative agent from a different drug class. Subjects switching back to their initial
134
antidiabetic agent within 90 days of switching were not classified as having switched therapy.
135
For the purpose of this study, treatment addition and switching are collectively referred to as
136
regimen change.
137
Within treatment class changes were analysed for the treatment addition and switching
138
groups. Therapy was considered to have changed within its treatments class if a subject
139
received an increased or decreased dose of their antidiabetic agent, or received a different
140
antidiabetic agent within the same treatment class.
141
Statistical methods
142
Descriptive analysis of subject characteristics and initial treatment patterns are presented.
143
Subjects were analysed in four groups: those who remained on initial therapy and were
144
persistent for the first twelve months; those who remained on initial therapy and were non-
145
persistent in the first twelve months; those switching therapy; those receiving add-on therapy.
146
Only subjects remaining on initial therapy were examined for persistence. Time to treatment
147
addition or switching is reported in median number of days and interquartile range (IQR).
148
Cox proportional-hazards regression was used to examine differences in time to treatment
149
addition and switching for those initiated on metformin compared to a sulphonylurea
150
adjusting for age, sex and coverage scheme. Adjusted logistic regression was used to examine
151
the association between initial antidiabetic agent (and age, sex and coverage scheme, odds
152
ratio (OR) with 95% confidence interval (CI) is presented.
153
Cox proportional-hazards regression was used to examine time to (i) treatment non-
154
persistence, (ii) receiving a regimen change, and (iii) dose change prior to treatment
155
switching or treatment addition. . These analyses were adjusted for the following variables of
Accepted Article
132
This article is protected by copyright. All rights reserved.
8
interest and available within the database: duration of follow up, sex, age, coverage scheme
157
and initial antidiabetic agent. Similar variables have found to be of significance in previous
158
studies (15–18) Hazard ratio (HR) and CIs are presented. Metformin (vs sulphonylurea),
159
GMS scheme (vs LTI), males, and subjects aged 60-69 were the reference groups.
160
Using Cox proportional-hazards regression we examined the association of cardiovascular
161
drug use in the GMS population on regimen changes, controlling for age, antidiabetic agent
162
and gender. We used number of different classes of CVD agents prescribed prior to
163
antidiabetic agent initiation as an indication of pre-existing CVD. The following classes of
164
CVD agents were included in our study: Aspirin (B01AC06, N02BA01), Nitrates (C01DA),
165
lipid lowering drugs (Statins (C10AA), Non-Statin Lipid Lowering Drugs (C10AB, A10AC,
166
A10AD, C10AX)), Beta-Blockers (C07), Calcium Channel Blockers (C08), agents acting on
167
the renin-angiotensin system (C09), Diuretics (C03). Analyses were performed using SAS,
168
version 9.2, significance at p
