Trends in long-term prescribing of antidepressant medicines

Abstract

This descriptive report presents analyses of CPRD data, showing the proportion of antidepressant prescribing that is long-term, and visualises trends in this proportion over time.

Full text

1
Trends in long-term
prescribing of
antidepressant medicines

2
Sehmi R, Nguyen A, McManus S, Smith N. (2019) Trends in long-term prescribing of
antidepressant medicines. London: PHRC/ NatCen.
This is independent research commissioned and funded by the National Institute for
Health Research (NIHR) Public Health Research Consortium (PHRC) Policy
Research Programme (PHPEHF50/14). The views expressed are those of the
authors and not necessarily those of the NIHR, the Department of Health and Social
Care or its arm's length bodies, or other Government Departments. Information about
the wider programme of the PHRC is available from http://phrc.lshtm.ac.uk/. Study
summary available at: www.natcen.ac.uk/antidepressants
Cover picture by Rosie Rodgers.
NatCen Social Research
35 Northampton Square
London EC1V 0AX
T 020 7250 1866
www.natcen.ac.uk
A Company Limited by Guarantee
Registered in England No.4392418
A Charity registered in England and Wales (1091768) and Scotland (SC038454)
This project was carried out in compliance with ISO20252.

3
Contents
Executive summary ................................................................. 6
1 Introduction ..................................................................... 17
1.1 Background ................................................................................................... 17
1.2 Previous evidence ......................................................................................... 17
1.3 Aims and objectives ...................................................................................... 18
1.4 Acknowledgements and disclosures ............................................................. 19
2 Methods 20
2.1 Data extraction .............................................................................................. 20
2.2 Profile of the CPRD extract analysed ............................................................ 21
2.3 Estimating length of individual prescriptions ................................................. 22
2.4 Estimating length of continuous prescribing periods ..................................... 22
2.5 Base population ............................................................................................ 23
2.6 Analysis…… ................................................................................................. 24
2.7 Missing data .................................................................................................. 24
3 Trends in ADM prescribing .............................................. 26
3.1 ADM items dispensed in England 2008-2018 ............................................... 26
3.2 Proportion of the population prescribed ADM when interviewed, 1993 to
2014………. .................................................................................................. 28
3.3 Length of prescribing: average number of days ............................................ 30
3.4 Length of prescribing: proportion of prescriptions exceeding thresholds ...... 32
3.5 Length of prescribing: proportion of continuous prescribing periods exceeding
thresholds… .................................................................................................. 33
3.5.1 One-year threshold ........................................................................................ 33
3.5.2 Two-year threshold ........................................................................................ 34
3.5.3 Three-year threshold ..................................................................................... 34
3.5.4 Five-year threshold ........................................................................................ 35
4 Characteristics associated with longer-term ADM
prescribing…. ........................................................................ 36
4.1 Variation in continuous prescribing by gender .............................................. 36

4
4.1.1 SSRIs… ......................................................................................................... 36
4.1.2 Tricyclics ........................................................................................................ 36
4.1.3 Other ADMs ................................................................................................... 37
4.2 Variation in continuous prescribing by age-group ......................................... 38
4.2.1 SSRIs… ......................................................................................................... 38
4.2.2 Tricyclics ........................................................................................................ 38
4.2.3 Other ADMs ................................................................................................... 39
4.3 Characteristics associated with long-term prescribing in 2012 ..................... 40
4.3.1 Gender variations in long-term prescribing .................................................... 40
4.3.2 Age-group variations in long-term prescribing ................................................ 40
4.3.3 Regional variations in long-term prescribing .................................................. 40
5 Conclusion ...................................................................... 42
6 References ...................................................................... 44
7 Appendices ..................................................................... 46
Figures
Figure 1 Number of prescribed ADM items dispensed in England 2008-2018 ........ 27
Figure 2 Number of prescribed ADM items dispensed in England by type 2008-2018
................................................................................................................................ 27
Figure 3 Proportion of people (age 16-64) prescribed ADM at the time of the
interview .................................................................................................................. 28
Figure 4 Proportion of people (age 16-64) prescribed ADM at the time of the
interview, by gender ................................................................................................ 29
Figure 5 Proportion of people (age 16-64) prescribed ADM at the time of the
interview, by age ..................................................................................................... 30
Figure 6 Average number of days of individual prescriptions by ADM type ............. 31
Figure 7 Average number of days of continuous prescribing periods by ADM type 31
Figure 8 Proportion of individual prescriptions exceeding 60 days, by ADM type ... 32
Figure 9 Proportion of continuous prescribing periods exceeding one year, by ADM
type ......................................................................................................................... 33
Figure 10 Proportion of continuous prescribing periods exceeding two years, by
ADM type................................................................................................................. 34
Figure 11 Proportion of continuous prescribing periods exceeding three years, by
ADM type................................................................................................................. 35

5
Figure 12 Proportion of continuous prescribing periods exceeding five years, by
ADM type................................................................................................................. 35
Figure 13 Proportion of patients prescribed SSRIs continuously for periods
exceeding three years, by gender ........................................................................... 36
Figure 14 Proportion of patients prescribed tricyclics continuously for periods
exceeding three years, by gender ........................................................................... 37
Figure 15 Proportion of patients prescribed other ADM continuously for periods
exceeding three years, by gender ........................................................................... 37
Figure 16 Proportion of patients prescribed SSRIs continuously for periods
exceeding three years, by age ................................................................................ 38
Figure 17 Proportion of patients prescribed tricyclics continuously for periods
exceeding three years, by age ................................................................................ 39
Figure 18 Proportion of patients prescribed other ADMs continuously for periods
exceeding three years, by age ................................................................................ 39
Tables
Table 1 British National Formulary (BNF) v68 Section 4.3 medicines ..................... 20
Table 2 Geographical distribution of patients in the CPRD extract .......................... 21
Table 3 Number of people prescribed SSRIs, tricyclics, MAOIs and other ADMs ... 23
Table 4 Availability of information for calculating ADM prescription length .............. 24
Table 5 Availability of information for calculating ADM prescription length, by gender
................................................................................................................................ 25
Table 6 Characteristics of people prescribed SSRIs in 2012, by continuous
prescribing period length ......................................................................................... 40
Table 7 Characteristics of people prescribed Tricyclics in 2012 by continuous
prescribing period length ......................................................................................... 41
Table 8 Characteristics of people prescribed other ADMs in 2012 by continuous
prescribing period length ......................................................................................... 41

6
Executive summary
This descriptive report presents estimates of the proportion of antidepressant
prescribing that is long-term and visualises trends in this over time.
Background
The prescribing of antidepressant medicines (ADM) has increased in England in
recent decades. To understand the nature of this increase it is necessary to draw on
a wide range of indicators that reflect different aspects of ADM prescribing,
including the number of ADM items prescribed, the number of people prescribed to,
and the length of time that ADMs are prescribed for. It is also necessary to use
recent information in order to check whether upward trajectories in prescribing have
continued, and to consider trends separately for different types of ADM. This
descriptive report contributes one perspective by providing temporal trends in the
prevalence of long-term ADM prescribing. These are initial estimates drawing on a
relatively small data extract, the descriptive trend lines presented should not be
viewed as definitive or exhaustive. This report is one of several studies taking place
on this topic. Public Health England have published a fuller review of the evidence,
including a detailed analysis of dispensed prescription data and summary analyses
of other data sources (Taylor et al., 2019).
Aims and objectives
The primary aim of this research was to use patient-level primary care data to
estimate the extent to which ADMs were prescribed to people continuously for long
periods of time (such as in excess of three years). The specific objectives included
to:
Objective 1 Produce temporal trends in the population prevalence of ADM use using
general population survey data.
Objective 2 Produce temporal trends relating to the length of individual prescriptions
and of continuous prescribing periods (defined as an initial prescription plus repeat
prescriptions) by:
• Deriving the average length of individual prescriptions and continuous
prescribing periods for each year, for each type of ADM.
• Estimating the proportion of continuous prescribing periods to exceed one,
two, three, and five years.
Objective 3 Among people prescribed an ADM, identify the characteristics
associated with being prescribed to for longer periods.
The overarching aim of this report was to communicate general trends in
prescribing practice. It was not an aim to draw conclusions on reasons for
prescribing and whether – in general, or in individual cases – prescribing was
appropriate or not.

7
Methods
Clinical Practice Research Datalink: the main data source used was an extract from
the Clinical Practice Research Datalink (CPRD). The extract used for this study
consisted of data on a random sample of about 50,000 patients prescribed any of
the ADMs in Section 4.3 of version 68 of the British National Formulary (BNF),
between 2000 and 2017. While analyses would have been more robust if based on
a larger sample of patients, the charge for a larger sample was beyond the resource
available for this simple initial analysis.
Medicines were grouped for analysis into:
• Tricyclics
• Selective serotonin reuptake inhibitors (SSRIs)
• ‘Other’ ADMs.
The decision to use BNF groupings was reached after consulting with other
analysts, in order to provide comparability with other studies. The length of
individual prescriptions and continuous prescribing periods were derived using
information on consultation dates, the quantity of tablets prescribed, and the
numeric daily dose.
Temporal trends in short-term prescribing are presented for every year of the data
extract (2000 to 2017). However, for longer prescribing lengths (such as in excess
of three or five years) reliable estimates could be produced for fewer years. For
example, if a prescribing period started in 2013, it was not possible to identify with
the available data whether that prescribing period would continue for five years or
more. Estimates for those years would underestimate the extent of longer-term
prescribing as some continuous prescribing periods would not have yet completed,
they would have been truncated in the data extract. Trends were therefore
presented only for the years for which reliable estimates could be generated. The
analyses in this report are descriptive and show the overall prevalence of long-term
prescribing in each year. The CPRD does not link prescriptions directly to
diagnoses or reasons for prescribing; any changes in the indications for which ADM
were prescribed are not captured by these analyses.
In addition to the CPRD extract, we briefly drew on other data sources to provide
context for the main analyses and understanding of trends in other aspects of
prescribing practice:
Adult Psychiatric Morbidity Survey (APMS) data from the 1993, 2000, 2007 and
2014 surveys were analysed as a source on trends in the proportion of people in
England aged 16 to 64 to self-report being in receipt of ADM at the time they were
interviewed. Analyses were replicated using Health Survey for England data as a
check.
Prescription Cost Analysis (PCA): we present results drawing on published NHS
Digital analysis of the volume of ADM items dispensed each year for 2008 to 2018.

8
Survey data and information on dispensed prescription items were available from
NHS Digital without charge.
Results
The number of ADM items dispensed in a year doubled between 2008 and 2018
The volume of SSRI and other ADM items dispensed by community pharmacies in
England increased steeply between 2008 and 2018, while increases in the volume
of tricyclic items dispensed were evident but less pronounced. The number of
monoamine oxidase inhibitors (MAOI) items dispensed fell over time, but the
volume was so much lower than for the other types of ADM examined that while
included in the chart below it is not visible. Prescriptions for this type of ADM were
too few in our CPRD data extract for robust analysis and are not examined
separately in this report.
Number of prescribed ADM items dispensed in England by type 2008-2018
Source: Prescription Cost Analysis data (NHS Digital)
Proportion of adults reporting current ADM use increased in the 1990s, and
again since 2007
General population survey data were used to provide a snapshot of the proportion
of people in England aged 16 to 64 who self-reported current ADM use at specific
points in time. Note that ‘current ADM use’ refers to use around the time of the
interview and not use at any time during that calendar year or in the 12 months prior
to the interview. Such data can provide insight into medications taken (as opposed
to prescribed or dispensed) from any source (including non-NHS). APMS 2014
participants were presented with cards listing ADMs and asked: ‘are you currently
taking any of these medications?’ Interviewers asked to see - and validated coding
against - the medication packaging.
1
1
On the APMS survey this variable included a range of types of psychotropic medication, including
less common ones prescribed for the treatment of psychosis.
0
10,000,000
20,000,000
30,000,000
40,000,000
50,000,000
60,000,000
70,000,000
80,000,000
2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
Tricyclic & Related Antidepressant Drugs Selective Serotonin Re-Uptake Inhibitors Other Antidepressant Drugs

9
We only have APMS data for four time points and so trends presented are crude.
However, they indicate that increases in the prevalence of ADM use occurred in the
1990s, and again in more recent years (since 2007). The proportion of people aged
16 to 64 and living in England prescribed an ADM around the time of the interview
rose from around 1.0% in 1993 (95% CI: 0.8,1.2), to 4.3% in 2000 (95% CI: 3.8,4.9),
4.9% in 2007 (95% CI: 4.3,5.5), and 9.9% in 2014 (95% CI:9.1,10.8). A different
method of assessment was used in 1993, which could have led to an underestimate
in the data for that year.
The trends over time are similar in men and women. However, as the figure below
shows, the increase since 2007 in the proportion of people self-reporting current
ADM prescription has been more pronounced in older people (age 55 to 64) than in
those aged 16-24. Our study did not examine trends in those younger than 16.
Proportion of people currently prescribed ADM, by age (%)
Source: Adult Psychiatric Morbidity Survey 1993, 2000, 2007, 2014
Few individual ADM individual prescriptions exceed 60 days in length
A minority of individual ADM prescriptions exceed 60 days: this was consistently
more likely for tricyclics (2% to 4%) than for SSRIs or other ADMs (both consistently
below 2%). The proportion of tricyclic and other ADM individual prescriptions to
exceed 60 days fluctuated, but with a general downward trend over time. For
SSRIs, the proportion of individual prescriptions to exceed 60 days remained stable
over time.
0
5
10
15
16-24 25-34 35-44 45-54 55-64
%
1993 2000 2007 2014

10
Proportion of individual prescriptions exceeding 60 days, by ADM type (%)
Source: CPRD (Appendix 4)
We estimated the odds of individual prescriptions exceeding 60 days, for each year
compared with 2000 as a baseline reference category. For tricyclics, the odds ratios
(OR) for prescriptions exceeding 60 days reduced over time and were significantly
lower than 2000 from 2007 onwards. ORs for other ADM prescriptions exceeding 60
days were higher up to 2007. For SSRIs, the proportion exceeding 60 days
remained more stable, with few years differing significantly from the rate in 2000.
Estimates to test between-year differences in the proportion of individual prescriptions
exceeding 60 days, by ADM type (OR, C.I.s)
* P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001, § - reference category
0
1
2
3
4
5
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
%
SSRIs Tricyclics Other
Year
SSRIs
Tricyclics
Other ADMs
2000
§
§
§
2001
1.09 [0.94-1.27]
1.11 [0.99-1.24]
1.02 [0.60-1.72]
2002
1.14 [0.98-1.32]
1.15* [1.03-1.28]
1.59* [1.00-2.53]
2003
1.22** [1.06-1.41]
1.09 [0.97-1.22]
1.74* [1.11-2.72]
2004
1.16* [1.01-1.35]
0.95 [0.85-1.06]
1.92** [1.24-2.97]
2005
0.97 [0.83-1.12]
0.95 [0.85-1.06]
2.04** [1.32-3.16]
2006
0.93 [0.80-1.08]
0.90 [0.80-1.01]
2.04** [1.32-3.15]
2007
0.88 [0.76-1.02]
0.87* [0.77-0.98]
1.81** [1.17-2.81]
2008
0.96 [0.83-1.11]
0.78*** [0.69-0.88]
1.25 [0.80-1.96]
2009
0.92 [0.80-1.06]
0.78*** [0.69-0.88]
1.17 [0.75-1.84]
2010
1.03 [0.89-1.18]
0.88* [0.78-0.99]
1.13 [0.72-1.77]
2011
1.02 [0.88-1.18]
0.84** [0.74-0.94]
0.95 [0.60-1.50]
2012
1.10 [0.95-1.26]
0.87* [0.77-0.98]
0.98 [0.62-1.54]
2013
1.11 [0.96-1.28]
0.88* [0.78-1.00]
0.80 [0.50-1.28]
2014
1.20* [1.04-1.39]
0.75*** [0.66-0.87]
0.54* [0.32-0.91]
2015
1.16 [1.00-1.36]
0.72*** [0.61-0.84]
0.85 [0.51-1.41]
2016
1.13 [0.96-1.34]
0.75** [0.63-0.90]
0.63 [0.35-1.14]
2017
1.12 [0.94-1.35]
0.66*** [0.54-0.82]
0.57 [0.31-1.06]

11
Average length of continuous ADM prescribing periods increased
The mean average number of days of SSRI, tricyclic, and other ADM continuous
prescribing periods increased year on year between 2005 and 2011. (As noted
previously, estimates of mean prescribing lengths from 2013 will be underestimates
as continuous prescribing periods that commenced before 2017 and were ongoing
in 2018, were truncated in the data extract. For this reason, estimates of mean
length are not presented for 2013-2017. The estimate for 2012 could also be an
underestimate given that prescribing periods started in that year and continuing
beyond 2017 will not be fully captured in the data extract).
The mean duration of SSRI prescribing periods rose from 281 days in 2005 to a
peak of 322 in 2011. Tricyclics and other ADMs consistently averaged longer
continuous prescribing periods than SSRIs. In 2005, the average number of days
for both types was below 400 days, rising to 452 for tricyclics in 2010 and 466 for
other ADMs in 2011. The figure for 2012 may be an underestimate, as any
prescribing periods started that year and lasting six years or more will be truncated
in the data.
Average number of days of continuous prescribing periods by ADM type
Source: CPRD (Appendix 3)
We tested whether the average length of continuous prescribing periods each year
differed significantly from the baseline rate in 2005. The average number of days of
continuous prescribing periods was significantly higher in all years relative to 2005;
this pattern was consistent across all types of drugs.
0
50
100
150
200
250
300
350
400
450
500
2005 2006 2007 2008 2009 2010 2011 2012
%
SSRIs Tricyclics Other

12
Estimates to test between-year differences in the average number of days of continuous
prescribing periods, by ADM type (betas, SEs)
* P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001, § - reference category
In 2014, one in twelve tricyclics and other ADM prescribing periods exceeded
three years
Tricyclics and other ADMs were more likely to be prescribed continuously for
periods exceeding three years than SSRIs. In 2014, 4% of continuous SSRI
prescribing periods exceeded three years, compared with 8% of prescribing periods
for tricyclics and other ADMs. For these, the proportion of continuous prescribing
periods to exceed three years increased up to 2011/2, while the proportion of SSRI
prescribing periods exceeding this threshold remained more stable.
Proportion of continuous prescribing periods exceeding three years, by ADM type (%)
Source: CPRD (Appendix 7)
There are indications that in recent years the proportion of prescribing periods to
exceed long thresholds may have fallen. This could result from changes such as
greater switching between types of ADM.
We tested whether there were significant differences between years in the
proportion of continuous prescribing exceeding three years, using 2003 as a
baseline reference category. For SSRIs, the proportion of continuous prescribing
periods exceeding three years was significantly higher in almost all years relative to
2003. The exceptions were 2013 and 2014, where the rates were comparable to the
2003. For tricyclics and other ADMs, the proportion of continuous prescribing
0
2
4
6
8
10
12
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
%
SSRIs Tricyclics Other
Year
SSRIs
Tricyclics
Other ADMs
2005
§
§
§
2006
7.57* (3.31)
16.48** (5.60)
43.73*** (8.43)
2007
11.49*** (3.23)
36.92*** (5.54)
47.23*** (8.37)
2008
23.95*** (3.22)
47.04*** (5.53)
74.20*** (8.24)
2009
27.32*** (3.18)
49.26*** (5.57)
85.74*** (8.14)
2010
31.32*** (3.16)
51.69*** (5.58)
93.44*** (7.99)
2011
40.89*** (3.17)
50.82*** (5.59)
107.21*** (7.97)
2012
34.72*** (3.17)
36.96*** (5.65)
98.60*** (7.87)

13
periods to exceed three years was significantly higher in all years relative to the rate
in 2003.
Estimates to test between-year differences in the proportion of continuous prescribing
periods exceeding three years, by ADM type (OR, C.I.s)
* P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001, § - reference category
Among people prescribed to: men were prescribed to for longer
Among people prescribed SSRIs, men were more likely to be prescribed SSRIs for
longer than three years than women. This was the case for each year examined.
There was relatively little variation by gender among those prescribed tricyclics in
the proportion prescribed to for longer. However, a difference has emerged for other
ADM: since 2008 men prescribed the drug were more likely than women to be
prescribed it for longer.
Proportion of patients prescribed SSRIs continuously for periods exceeding three years, by
gender (%)
Source: CPRD (Appendix 11)
0
1
2
3
4
5
6
7
8
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
%Male
Female
Year
SSRIs
Tricyclics
Other ADMs
2003
§
§
§
2004
1.22*** [1.12-1.32]
1.13** [1.05-1.22]
1.19* [1.01-1.39]
2005
1.34*** [1.24-1.45]
1.29*** [1.20-1.39]
1.72*** [1.48-2.00]
2006
1.47*** [1.37-1.59]
1.32*** [1.23-1.42]
2.19*** [1.89-2.54]
2007
1.39*** [1.29-1.49]
1.41*** [1.31-1.51]
2.22*** [1.92-2.57]
2008
1.31*** [1.22-1.41]
1.42*** [1.32-1.53]
2.61*** [2.27-3.01]
2009
1.29*** [1.20-1.39]
1.40*** [1.31-1.51]
2.83*** [2.46-3.26]
2010
1.23*** [1.14-1.32]
1.50*** [1.39-1.61]
3.10*** [2.70-3.56]
2011
1.28*** [1.19-1.37]
1.55*** [1.44-1.66]
3.20*** [2.80-3.67]
2012
1.15*** [1.07-1.24]
1.39*** [1.29-1.49]
3.36*** [2.93-3.84]
2013
1.06 [0.98-1.15]
1.34*** [1.24-1.45]
2.90*** [2.53-3.32]
2014
0.99 [0.91-1.07]
1.42*** [1.32-1.54]
2.48*** [2.15-2.85]

14
Among people prescribed to: older people were prescribed to for longer
Older people prescribed SSRIs were more likely to be prescribed to in excess of
three years than younger people prescribed the drug. This association with age has
remained broadly consistent over time. The pattern for other types of ADM however
is less clear.
Proportion of patients prescribed SSRIs continuously for periods exceeding three years, by
age (%)
Source: CPRD (Appendix 15)
Among people prescribed to: people in the North of England were prescribed
to for longer
Among people prescribed ADM, those living in the North of England were more
likely than those living in other regions to be prescribed SSRIs and tricyclics in
excess of three and five years. For example, among people prescribed SSRIs, 7%
of those living in the North of England were prescribed the drug for longer than
three years compared with less than 5% in other regions. Among people prescribed
other ADMs, those living in the Midlands and East of England were the most likely
to be prescribed to for five years or more (7%).
Limitations
The analyses presented in this report have several limitations, including that they:
• Examine trends by broad class of ADM, and do not examine trends for
specific medicines.
• Provide descriptive visualisations of temporal trends, without extensive
significance testing or controlling for shifts in population demographics.
Further work is needed that takes a more complex approach. This study
focused on trends in prevalence in each year, other studies are needed that
focus on changes in incidence rates over time.
• Are based on a relatively small sample of patients from the CPRD (<50,000).
While it would have been preferable to have used the total CPRD sample or
0
2
4
6
8
10
12
14
16
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
%
18-40 41-60 61-80 81+

15
a larger sample, the costs involved in obtaining a larger data extract were
beyond the resourcing available.
• More exploration of potential biases in the wider CPRD sample coverage is
needed, especially given changes in recent years in the number of practices
participating.
• These trends relate to NHS prescribing practice only, and CPRD data do not
indicate whether medicines were dispensed, whether they were taken, and
whether they were taken by the patient they were prescribed to or by
someone else. Further research is needed on the characteristics of people
prescribed to, examining a broader range of factors at both the patient-level
(e.g. ethnicity) and practice-level (e.g. practice rurality) in larger samples to
improve generalisability.
• The data could not tell us about reasons for prescribing and whether long-
term prescribing was clinically appropriate. Qualitative research with both
patients and prescribers could provide insight on this. Further research
should also examine the potential underlying reasons for why long-term
prescribing is increasing at the population level, including considering the
role of misuse and inappropriate prescribing.
• Narrow and conservatively defined rules on what constitutes a continuous
prescribing period were applied; furthermore, ‘intermittent’ prescribing or a
continuous period of prescribing that comprised transitions between ADM
classes (for example from SSRI to tricyclics) would not have counted as a
continuous prescribing period.
• At least five years of subsequent data need to be available in order to
estimate the proportion of prescribing periods in a given year that will
continue for at least five years. This means that trends in prescribing periods
greater than five years could only be generated up to 2012, and trends for
prescribing periods greater than three years could only be generated up to
2014.
Conclusions
ADM prescribing has increased and is relatively common
The number of ADM items dispensed in England doubled over the past decade.
The prevalence of ADM use among 16-64-year olds increased from one in a
hundred in 1993, to one in twenty in 2000 and 2007, and one in ten in 2014: the
proportion prescribed ADM at any point in the year was higher. These figures relate
to snapshots of prevalence in time, not incidence.
Prescribing periods average about a year
Since 2000, individual prescriptions have been issued for short periods of time -
overwhelmingly for less than 60 days.
Continuous prescribing periods, however, last much longer. In 2011, the average
SSRI prescribing period was just under a year, and the average tricyclic or other
ADM prescribing period lasted well over a year.

16
Average prescribing periods have gradually increased over time
This rise in the mean number of days of continuous prescribing was evident for all
types of antidepressant examined. The increase was least pronounced for SSRIs,
and most pronounced for other ADMs.
One in twelve tricyclic and other ADM prescribing periods exceed three years
The proportion of SSRI prescribing periods to exceed three years remained around
4% between 2003 and 2014, while the proportion of ADM prescribing periods to
exceed this length doubled between 2003 (4%) and 2014 (8%).
There are indications that in recent years the proportion of prescribing periods to
exceed long thresholds has fallen. This could result from changes such as greater
switching between types of ADM.
There is variation in average prescribing length among those prescribed to
Among people prescribed ADMs, those more likely to be prescribed to continuously
for more than three years tended to be older, male, and living in the North of
England – although this pattern varied by type of drug.

17
1 Introduction
1.1 Background
It has been widely noted that the prescribing of antidepressant medicines (ADMs) in the UK
has long been on an upward trajectory (Middleton et al., 2001), and increased steeply in
recent years (Lacobucci, 2019). Extent of prescribing, however, can be measured in
different ways. Some analyses have found that increases in the volume of ADM items
prescribed were driven more by increases in the duration of time they were prescribed to
people for, than by increases in the number of people starting them (Mars et al., 2017;
Lockhart and Guthrie, 2011). ADMs may be prescribed for a range of indications, and those
indications may change over time. Using ADMs for long periods may be clinically
recommended and beneficial for some people. However, the relative risks and benefits of
long-term use are still not fully understood; and may vary between individuals. Previous
research has focused more on trends in volume of items prescribed than on patterns in
ADM prescribing at the patient-level. This report presents patient-level analyses of primary
care prescribing data from 2000 to 2017; providing temporal trends in the duration of time
that ADMs were continuously prescribed for.
1.2 Previous evidence
Several explanations have been offered for why the number of ADM items prescribed in the
UK has increased in recent decades. This section refers to some of the research on this,
although more detailed and systematic reviews have been conducted elsewhere. One
explanation is that the incidence of diagnosed depression has increased in the population
over time. There is some evidence to suggest that increases in the volume of ADMs
prescribed between 1995/6 and 2000/1 were partially explained by rises in the number of
people diagnosed with depression (Lockhart and Guthrie, 2011). However, incidence rates
for depression have not risen at a comparable rate to the marked increase in the volume of
ADM prescriptions dispensed in the last two decades (Spence, Roberts, Ariti and Bardsley,
2014). The number of ADM prescriptions dispensed increased steeply between 2008 and
2013, but there was only a small increase in consultations for depression (4%) during this
period (Kendrick et al, 2015; Spiers et al., 2016). Trends could have been influenced by
changes in care-seeking behaviour by patients, as well as by improved detection of
depression by practitioners (Munoz-Arroyo, Sutton and Morrison, 2006). It is also the case
that ADM may be increasingly prescribed for indications other than depression. For
instance, tricyclics are used for a range of conditions including chronic pain, fibromyalgia,
irritable bowel syndrome, sedatives and other mental health problems such as
anxiety(Lockhart and Guthrie, 2011).Trends in the volume of ADM ‘items’ prescribed could
also have been affected by changes in the average pack size or dose of items, although
investigation by Spence and colleagues in earlier data did not find this to be a major factor
in trends (2014).
Perhaps the most compelling explanation is that there was an increase in the length of time
that ADMs were being prescribed for. Several studies have used primary care data to
examine whether increases in longer-term prescribing drove increases in the volume of
ADMs prescribed (Lockhart and Guthrie, 2011; Mars et al., 2017). One study found that
most ADM prescriptions issued between 1993 and 2005 were prescribed as part of a long-

18
term period of continuous or intermittent prescribing to patients who had experienced
multiple episodes of depression (Moore et al., 2009). Another study examined changes in
prescribing between 2003 and 2013 following the implementation of NICE guidelines
designed to reduce prescribing to patients with mild depression (NICE, 2004). The rate of
first-onset depressive episodes to be treated fell, while treatment rates for recurrent
depression rose during this period. A recent study also used primary care data but, unlike
previous studies, did not limit prescribing events solely to patients diagnosed with
depression. Again, the authors found that increases in ADM prescribing between 1995 and
2011 were largely explained by greater long-term use (Mars et al., 2017).
Further research was required to provide more up-to-date evidence on the extent of long-
term continuous prescribing periods for ADMs, and the characteristics associated with long-
term ADM use at the patient level. This report provides some initial analyses that aim to
address this gap by providing some evidence on trends in long-term ADM use for 2000 to
2017, extending previous research by examining trends beyond 2011 for some thresholds.
It also examines variations in individual (including age and gender) and contextual (region)
characteristics associated with ADM use in excess of different prescribing length
thresholds, including continuously for more than three years and five years.
1.3 Aims and objectives
This descriptive report presents temporal trends in the prevalence of ADM prescribing that
is long-term. These initial estimates draw on a relatively small data extract, these visual
trend lines should not be viewed as definitive or exhaustive. This report is one of several
studies taking place on this topic. Public Health England have published a fuller review of
the evidence, including a detailed analysis of dispensed prescription data and summary
analyses of other data sources (Taylor et al., 2019).
The primary aim of this research was to produce temporal trends in the length of ADM
prescribing from 2000 to 2017. The specific ADM groups examined are tricyclics, selective
serotonin reuptake inhibitors (SSRIs), and ‘other’ ADMs.
We examine two aspects of prescribing: the length of individual prescriptions and the length
of continuous prescribing periods.
Prescriptions: refers to individual prescriptions, not linked at the patient level.
Continuous prescribing periods: refers to a series of one or more prescriptions (the initial
prescription combined with all repeat prescriptions) making up a continuous episode of
prescribing.
The specific objectives of the research were to:
Objective 1 Describe temporal trends in the population prevalence of ADM use
Objective 2 Produce temporal trends in the length of individual prescriptions and
continuous prescribing periods.
• Derive the average length of individual prescriptions and continuous prescribing
periods for each year, for each type of ADM.
• Estimate the proportion of continuous prescribing periods to exceed one, two, three
or five years.

19
Objective 3 Among people prescribed ADMs, identify the characteristics associated
with being prescribed ADM for longer periods.
• Describe the demographic (age, gender) characteristics and region of people
prescribed ADMs longer-term.
The overarching aim of this report was to communicate general trends in prescribing
practice. It was not an aim to draw conclusions on reasons for prescribing and whether – in
general, or in individual cases – prescribing was appropriate or not.
Methods are described in Chapter 2. Chapter 3 covers trends over time in length of
prescribing, and Chapter 4 covers the characteristics associated with being prescribed
ADMs for longer periods. Detailed tables are provided in the Appendices.
1.4 Acknowledgements and disclosures
This is independent research commissioned and funded by the National Institute for Health
Research (NIHR) Public Health Research Consortium (PHRC) Policy Research Programme
(PHPEHF50/14). The project underwent ethical review with the National Centre for Social
Research internal review committee. The CPRD data extract was provided by Efthalia
Massou and Cassandra Woodfin of the Cambridge Research Methods Hub. APMS 2014
data use was carried out by NatCen under their data use agreement with NHS Digital. NHS
Digital gave permission for us to reproduce figures relating to trends in volume of dispensed
prescribing items. We are also grateful to advice and comments from Martin White, Steve
Taylor, Pete Burkinshaw and Fizz Annand at Public Health England (PHE). Rachel Conner
at the Department of Health and Social Care (DHSC) oversaw the project, and project
administration was handled by Sarah Kennedy of the PHRC at London School of Hygiene
and Tropical Medicine. Many thanks to Katherine Ridout of the National Centre for Social
Research who replicated the population trends in proportion of the population prescribed to
using Health Survey for England data as a validation check on results. The authors have no
conflicts of interest to declare.

20
2 Methods
This report builds on the methodological approach detailed in previous research on long-
term prescribing of dependence forming medicines (Sehmi et al., 2019; Cartagena Farias et
al. 2017).
2.1 Data extraction
The Clinical Practice Research Datalink (CPRD) contains primary care data for
approximately 8% of the UK population (Herrett et al., 2015). A geographically
representative random sample of 49,935 individuals living in England and prescribed any
ADM between 2000 and 2017 was extracted for analysis. While a larger sample would have
been preferable, that was not possible with the resource available for this initial analysis.
ADMs were defined as those medicines listed in Section 4.3 of the British National
Formulary (BNF) version 68 (see Table 1). Analyses focus on medicines combined into the
following groups:
• Any ADM: all medicines listed in Section 4.3
• Tricyclic ADM: all medicines listed in Section 4.3.1
• SSRI ADM: all medicines listed in Section 4.3.3
• Other ADM: all medicines listed in Section 4.3.4
The decision to use BNF groupings was reached after consulting with other analysts to
provide comparability with other studies. The ‘any ADM’ group includes the monoamine
oxidase inhibitors (MAOI) medicines listed in Section 4.3.2, but these were not also
analysed as separate group due to small numbers. Note that analyses of ‘other ADM’ refers
specifically to medicines listed in the ‘other ADM’ BNF Section 4.3.4, and does not also
include MAOIs.
Table 1 British National Formulary (BNF) v68 Section 4.3 medicines
4.3 ADMs
4.3.1 Tricyclics
Amitriptyline (including with perphenazine)
Amoxapine
Clomipramine
Dosulepin
Doxepin
Imipramine
Lofepramine
Maprotiline
Mianserin
Nortriptyline
Protriptyline
Trazodone
Trimipramine
4.3.2 MAOIs
Isocarboxazid
Moclobemide
Phenelzine
Tranylcypromine
4.3.3 SSRIs
Citalopram
Escitalopram
Fluoxetine

21
Fluvoxamine
Paroxetine
Sertraline
4.3.4 Other ADMs2
Agomelatine
Duloxetine
Flupentixol
Mirtazapine
Nefazodone
Oxitriptan
Reboxetine
Tryptophan
Venlafaxine
Vortioxetine
2.2 Profile of the CPRD extract analysed
The CPRD extract contains a random sample of 49,935 patients in England who have been
prescribed at least one of the four types of ADMs at some point between 2000 to 2017. The
extract is geographically representative of patients in England in the CPRD, with the
proportion of patients in each area determined by the proportion of the overall population
living there during the 2011 Census. From the total number of patients requested in the
extract 5% are in the North East and 15% are in the South East (see Table 2). The
sampling was stratified by region to allow for some regional analysis of prescribing trends,
although some pairing of regions was also required.
Table 2 Geographical distribution of patients in the CPRD extract
The patients had an associated 1,118,430 consultations distributed between 2000 and
2017. Of those patients, 64% were women and 69% were 18-60 years old. The median age
in 2000 was 53 years old, which increased to 55 years in 2006 and then remained fairly
2
The decision to use BNF groupings was reached through consultation with others carrying out
parallel analyses, to enable comparability across analyses.
Region %
N
North East
5
2,446
North West
13
6,638
Yorkshire and the Humber
10
4,979
East Midlands
9
4,270
West Midlands
11
5,275
East of England
11
5,510
South West
10
4,980
London
9
4,243
South East
15
7,703
South West
8
3,891
Total
100
49,935

22
stable up to 2017. While some patients died during the years covered by the CPRD extract,
others turned eighteen and became eligible for inclusion.
2.3 Estimating length of individual prescriptions
The length of an individual prescription was defined as the number of treatment days a drug
was prescribed for, for a specific therapy event. Each consultation provided information
regarding the total dose and numeric daily dose of a particular drug prescribed to a
particular person. To estimate the length of each individual prescription, the total quantity of
drug prescribed was divided by the numeric daily dose for each given therapy (e.g. 90
tablets / 3 tablets per day = 30 days duration).
2.4 Estimating length of continuous prescribing
periods
The number of days for which a class of drug was continuously prescribed to a person for
was estimated by adding the length of an initial prescription to the length of any repeat
prescriptions. A repeat prescription was identified when the same drug (or another drug in
the same class) was prescribed within a period of time. To identify a series of prescriptions
that could be reliably classified as repeat prescriptions, the following rules were applied:
1. The number of days between each consultation was less than or equal to 35 days, or the
length of the previous prescription is equal to, or less than, the number of the days between
consultations (plus five days
3
). This ensured that a repeat prescription was identified, as
opposed to multiple prescriptions over a longer period.
2. In some instances, consultation identifiers varied despite being held on the same day; it
was assumed that all events that occurred on the same day were part of the same
consultation.
3. For analyses of the classes of drugs (i.e. tricyclics, SSRIs and other ADM medicines) if a
patient switched from one type of tricyclic to another type of tricyclic, this still counted as a
continuous prescribing period; the same applied to other classes of drug.
4. If two different drugs belonging to the same class of drug were prescribed in the same
consultation, the longer of the two derived prescription lengths was used. Similarly, if two
different prescription lengths were prescribed in the same consultation then the longer
length was used.
5. The first consultation date when a drug was prescribed was noted as the start date of a
continuous prescribing period; this ensured that repeat prescriptions following from one
year to the next were counted from the original start date.
When reporting on prescribing periods that span two calendar years, duration length is
based on when the prescribing period ends as opposed to curtailing periods at the end of a
3
A 5-day buffer period was allowed for as it represents a working week and hence was chosen as a
pragmatic time period in which an individual could seek a fresh prescription from their GP once the
previous one had elapsed.

23
given calendar year. As a result, periods spanning two calendar years are counted in both
consecutive years. For example, if someone was prescribed tricyclics from May 2001 to
August 2002, that would be counted as a continuous prescribing period exceeding 12
months in both 2001 and 2002. This phenomenon will affect longer prescribing periods
more. Periods spanning three years are counted three times, in each of the relevant years.
This approach is appropriate for capturing the prevalence (rather than incidence) of long-
term prescribing periods in any given year, and therefore for capturing change over time.
It is important to note that the CPRD data extract captures information up to the end of
2017. A proportion of people being prescribed to in 2017 will not yet have completed their
continuous prescribing period. This is a particular issue when estimating the average
number of days of a prescribing period and the proportion of prescribing periods exceeding
the longer thresholds, such as over one year. Due to this uncertainty, estimates for certain
later years of data have not been discussed in the text of the report (although are
presented, in grey, in the Appendices for reference). Similarly, a significant proportion of
people would not have completed their continuous prescribing period at the start of the
study period. For instance, we would expect low estimates of the proportion of people with
a continuous prescribing period exceeding one year within the first 12 months of the data
extract, as many people would have received their initial prescription before the start date of
the extract.
This report shows the rates for 2001 to 2016 for continuous prescribing periods exceeding
one year due to this uncertainty about the reliability of longer-term estimates for 2000 and
2017. Similarly, rates are presented for 2002 to 2015 for continuous prescribing periods
exceeding two years, 2003 to 2014 for continuous prescribing periods exceeding three
years, and 2005 to 2012 for continuous prescribing periods exceeding five years. Rates for
individual prescriptions, however, include estimates for 2000 and 2017 as truncation should
have had only a minimal impact on this estimate. It should be noted that results for
continuous prescribing periods incorporate prescription lengths within the measure.
2.5 Base population
The unit of analysis for the tables presented varied according to the subject being examined
and is referred to in the title of the tables and the text. Analyses examining trends over time
used either prescriptions or continuous prescribing periods, whereas analyses of the
characteristics of long-term prescribing in 2012 refers to people. For the latter analyses,
prescribing patterns were investigated by region for the patient’s address and by the gender
and age of the person prescribed to.
The table below shows the number of patients in the extract prescribed each type of ADM
at some point during the reporting period. There were 34,484 people prescribed an SSRI,
24,847 prescribed a tricyclic, 64 people prescribed an MAOI and 9,129 prescribed any
other ADM.
Table 3 Number of people prescribed SSRIs, tricyclics, MAOIs and other ADMs
SSRIs
Tricyclics
MAOIs
Other ADMs
N
34,384
24,847
64
9,129

24
2.6 Analysis
The average length of individual prescriptions and continuous prescribing periods for each
class of ADM (or any combination of these to allow for co-prescribing) were estimated for
each year of the data available. The results examining trends in length of prescriptions and
continuous prescribing periods are descriptive, unadjusted and not tested for statistical
significance.
Further analysis estimated the proportion of prescribing events for each class of ADM which
were either prescriptions or continuous prescribing periods in excess of one, two, three or
five years respectively. These thresholds are not mutually exclusive, meaning that a
continuous prescribing period in excess of three years, for example, was also counted as
being a prescribing period in excess of one year and two years. A further analysis
examined the demographic (age, gender) and area-level (region) characteristics associated
with trends in long-term prescribing in excess of three years. Due to sample size limitation,
the ten available regions in the CPRD were banded into pairs to create five larger regions,
North of England, Midlands and East of England, London, South East and South West. Age
was banded into four periods, 18-40, 41-60, 61-80 and 81 and over, with each period
spanning approximately 20 years of the life course.
A different analytic approach was taken to describe the characteristics of people prescribed
ADMs long-term in a specific year, 2012. We made comparisons by characteristic
separately for each threshold. For example, we tested whether the proportion of women
prescribed SSRIs in excess of one year was significantly different to the proportion of men
prescribed SSRIs in excess of one year. We did not adjust the model for any covariates. A
reference category was assigned where appropriate in order to make comparisons, in the
case of gender, ‘male’ was assigned as the reference category. A continuous measure of
age was used to examine whether there were age-related variations in prescribing, so no
reference category was assigned. North of England was chosen as the reference category
to examine regional variations as it was the most populated region in this sample.
2.7 Missing data
In order to estimate the duration of a prescription information was needed on both the
quantity of pills prescribed and the daily dose. If information was missing for either or both
of these, length could not be estimated.
Table 2 shows that MAOIs had the least completed length information, with 24.0% of the
total number of consultations having missing data. Tricyclics and other ADM had about 17%
with missing data, while SSRIs had the least data missing (9.4%). Overall, there may be a
very slight systematic bias towards underestimating continuous prescribing periods, and
that this would have had greater impact on the estimated length of MAOI, tricyclic and other
ADM prescribing periods, and less impact on length of SSRI prescribing periods.
Table 4 Availability of information for calculating ADM prescription length
SSRIs
Tricyclics
MAOIs
Other ADMs
Consultations with length
information
544,929
292,535
1071
161,159
Total consultations
601,764
354,098
1409
134,142
% missing data
9.4
17.4
24.0
16.8

25
Table 3 shows the amount of missing data by gender for each type of ADM. Women tend to
have slightly less data missing than men, although gender differences were not
pronounced. There was also very little variation in extent of missing data by patient age:
this ranged from 12% among those aged 18 to 40, to 14% of patients aged 61 to 80.
Table 5 Availability of information for calculating ADM prescription length, by gender
SSRIs
Tricyclics
MAOIs
Other ADMs
Male
Female
Male
Female
Male
Female
Male
Female
% missing data
9.7
9.3
17.1
17.5
25.8
23.3
17.7
16.3
Practices based in the North of England (12%) and in Midlands and the East of England
(11%) had the least missing data, followed by London (14%). In this data extract, practices
based in the South East (15%) and South West (18%) were the most likely to have data
missing (data not shown). While these variations were not pronounced, bias resulting from
missing data cannot be ruled out for other characteristics for which we did not have data,
such as practice size.

26
3 Trends in ADM prescribing
This chapter presents several different indicators relating to temporal trends in the
prescribing of ADM in England. The first section (3.1) presents trends in the overall volume
of ADM items dispensed each year between 2008 and 2018, drawing on Prescription Cost
Analysis data published by NHS Digital. Section 3.2 presents estimates of the proportion of
the population (aged 16 to 64) taking prescribed ADM when interviewed on surveys
conducted in 1993, 2000, 2007 and 2014. The remaining sections in this chapter examine
temporal trends in prescribing duration; and draw on analyses of the CPRD extract. These
sections cover the mean average number of days of ADM individual prescriptions and
continuous prescribing periods (3.3); the proportion of individual prescriptions to exceed 60
days (3.4); and the proportion of continuous prescribing periods to exceed one, two, three
and five years (3.5).
3.1 ADM items dispensed in England 2008-2018
NHS Digital derived the overall volume of ADM items dispensed between 2008 and 2018
using Prescription Cost Analysis (PCA) data, which covers the number of items dispensed
in the community in England (NHSD 2019). In this section we present these figures for
ADMs to provide context for the rest of the report. While this report focuses on duration of
prescribing periods, it is important to consider these patterns in the context of trends in the
volume of items prescribed and the proportion of the population prescribed to (considered
in Section 3.2).
The total volume of ADM items prescribed, including tricyclics, SSRIs, MAOIs, and other
ADM, has almost doubled in the last ten years in England; over 35 million items were
prescribed in 2008, rising to over 70 million items prescribed in 2018 (

27
Figure 1). It should be noted that the definition of an ‘item’ in the PCA data is any quantity of
a particular medicine issued at one point in time; no adjustment was made for any changes
in the average number of tablets issued at a time.

28
Figure 1 Number of prescribed ADM items dispensed in England 2008-2018
Source: Prescription Cost Analysis data (NHS Digital)
Figure 2 shows that most ADM prescription items are SSRIs, followed by tricyclics and
other ADMs. MAOI prescribing is relatively rare and the volume of items prescribed has
declined from around 65,000 in 2008 to about 40,000 in 2018 (Appendix 1). We have not
examined MAOI prescribing in this report, as the numbers of people prescribed to were too
few in our CPRD extract for robust analyses. There has been some variation between drug
types in the extent to which the volume of items prescribed has increased. The greatest
increase in prescribed items from 2008 to 2018 has occurred for SSRIs and other ADMs.
The volume of SSRI items issued doubled from over 19 million items in 2008 to over 38
million in 2018, and other ADMs tripled from 5 million in 2008 to 16 million in 2018. Tricyclic
prescribed items increased from 11 million in 2008 to 16 million in 2018, representing an
almost 50% increase in the volume of prescribing items issued over time.
Figure 2 Number of prescribed ADM items dispensed in England by type 2008-2018
Source: Prescription Cost Analysis data (NHS Digital)
0
10,000,000
20,000,000
30,000,000
40,000,000
50,000,000
60,000,000
70,000,000
80,000,000
2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
0
10,000,000
20,000,000
30,000,000
40,000,000
50,000,000
60,000,000
70,000,000
80,000,000
2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
Tricyclic & Related Antidepressant Drugs Selective Serotonin Re-Uptake Inhibitors
Other Antidepressant Drugs

29
3.2 Proportion of the population prescribed ADM
when interviewed, 1993 to 2014
One way to estimate the proportion of people in the general population who are taking
prescribed ADM is to use general population data like that from the Adult Psychiatric
Morbidity Survey (APMS; McManus et al. 2016). The survey was funded by the Department
for Health and Social Care (DHSC) and commissioned by NHSD. It has been carried out
using comparable methods in 1993, 2000, 2007, and 2014, with response rates of 79%,
69%, 57%, and 57% respectively. Those aged 16 to 64 and living in England were included
on every wave and have been selected for the current analysis: there were 8,903
participants in 1993, 6,175 in 2000, 5,419 in 2007, and 5,294 in 2014. The restriction of
analysis to this age range (16-64) should be noted in any comparisons with other data
sources, as well as the focus on current prescribing at the time of the interview. Such data
can provide insight into the medications taken (as opposed to prescribed or dispensed)
from any source (including non-NHS). APMS 2014 participants were presented with cards
listing ADMs and asked: ‘are you currently taking any of these medications?’ Interviewers
asked to see - and validated coding against - the medication packaging.
4
A strength of the series is that it covers the population living in private households,
irrespective of whether they are registered with a primary care provider. Limitations include
that ADM use is self-reported, and therefore may be unreliable (although interviewers did
check medication packaging, where possible, to check correct coding). The rates derived
represent the proportion of people aged 16 to 64 who reported being in receipt of ADM at
the time of the interview, which would be a lower rate than the proportion being prescribed
an ADM at any time during that year. Our analysis of the survey data was weighted and
controlled for complex survey design.
Figure 3 Proportion of people (age 16-64) prescribed ADM at the time of the interview
Source: Adult Psychiatric Morbidity Survey 1993, 2000, 2007, 2014
4
On the APMS survey this variable included a range of types of psychotropic medication, including
less common ones prescribed for the treatment of psychosis.
0
5
10
1993 2000 2007 2014
%

30
The proportion of people aged 16 to 64 and living in England prescribed an ADM around
the time of the interview increased over time (Figure 3). Rates rose from around 1.0% in
1993 (95% CI: 0.8,1.2), to 4.3% in 2000 (95% CI: 3.8,4.9) and 4.9% in 2007 (95% CI:
4.3,5.5) and 9.9% in 2014 (95% CI: 9.1,10.8). A different method of assessment was used
in 1993 and this could have led to an underestimate in the data for that year.
Analyses of the APMS data reported elsewhere, indicate that while some population
subgroups have experienced a significant increase in the prevalence of common mental
disorder over time, overall rates of depression and anxiety have only increased slightly. The
proportion of people with severe symptoms of anxiety and depress did change significantly
between 2007 and 2014. However, the longer-term trend has been one of gradual increase
(6.9% of 16 to 64-year olds in 1993, 7.9% in 2000; 8.5% in 2007; 9.3% in 2014) (McManus
et al., 2016). It is the treatment gap between having a disorder and receiving treatment that
has reduced over time (Lubian et al., 2016). The results are also consistent with ADM being
prescribed for a wider range of indications.
Figure 4 Proportion of people (age 16-64) prescribed ADM at the time of the interview, by
gender
Source: Adult Psychiatric Morbidity Survey 1993, 2000, 2007, 2014
Women were about twice as likely to be prescribed ADM at the time of the interview as men
across all years (Figure 4) (1993: p=0.001; 2000, 2007 and 2014: p<0.001). For example,
7% of men and 13% of women reported that they were currently being prescribed ADMs
when interviewed in 2014. The proportion of people prescribed ADMs increased at a similar
rate for both men and women over time.
0
5
10
15
1993 2000 2007 2014
%
Male Female

31
Figure 5 Proportion of people (age 16-64) prescribed ADM at the time of the interview, by age
Source: Adult Psychiatric Morbidity Survey 1993, 2000, 2007, 2014
The proportion of people prescribed ADM at the time of the interview increased over time
across all age groups (Figure 5). However, the extent to which this proportion increased
varied. People aged 55 to 64 years had the largest increase in the proportion currently
prescribed to, rising from 1% in 1993 to 14% in 2014. People aged 16 to 24 years were the
least likely to be prescribed ADMs when interviewed at each wave, this proportion
increased four-fold from less than 1% in 1993 to 4% in 2014.
The prevalence and pattern in ADM prescribing to emerge in our analysis of APMS data is
similar to that found using data from the Health Survey for England 2015 and 2016, with
10% of adults (age 16 or more) taking a prescribed antidepressant in the week before
interview. They also found rates to be higher in women (13%) than men (6%), and to be
highest in those aged 45 and over (Moody et al., 2017).
3.3 Length of prescribing: average number of
days
The average number of days of individual prescriptions and continuous prescribing periods
by type of ADM are presented in Appendices 2 and 3.
Average length of individual prescriptions
The average length of individual prescriptions for SSRIs and tricyclics tended to be longer
than for other ADMs (Figure 6). The average length of individual SSRI prescriptions
increased marginally over time, from 32 days in 2000 to 35 days in 2017. For tricyclics and
other ADMs, the average length of prescriptions between 2000 and 2017 remained
relatively stable over time; fluctuating between 33 and 36 days for tricyclics, and between
29 and 32 days for other ADMs.
0
5
10
15
16-24 25-34 35-44 45-54 55-64
%
1993 2000 2007 2014

32
Figure 6 Average number of days of individual prescriptions by ADM type
Source: CPRD (Appendix 2)
Average length of continuous prescribing periods
For trends in the average length of continuous prescribing periods, analyses were restricted
to 2005 to 2012 because longer prescribing periods were more likely to be truncated, and
therefore underestimated, in the data extract.
Overall, tricyclics and other ADMs tended to be continuously prescribed for longer periods
than SSRIs (Figure 7). Increases in the average length of continuous prescribing periods
were evident for all types of ADM. This increase was most marked for other ADMs, for
which the average continuous prescribing period rose from 359 days in 2005 to 457 days in
2012. Increases for SSRIs (from 281 in 2005 to 316 in 2012) and tricyclics (from 393 in
2005 to 430 in 2012) were also evident, but slightly less pronounced than for other ADMs.
Figure 7 Average number of days of continuous prescribing periods by ADM type
Source: CPRD (Appendix 3)
0
5
10
15
20
25
30
35
40
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
SSRIs Tricyclics Other
0
50
100
150
200
250
300
350
400
450
500
2005 2006 2007 2008 2009 2010 2011 2012
SSRIs Tricyclics Other

33
3.4 Length of prescribing: proportion of
prescriptions exceeding thresholds
Trends were produced for the proportion of individual prescriptions each year exceeding 60
days. A small minority of individual ADM prescriptions exceeded this threshold: this was
more likely for tricyclics (2% to 4%) than for SSRIs or other ADMs (both consistently below
2%).
The proportion of tricyclic and other ADM prescriptions to exceed 60 days fluctuated over
time but with a general downward trend over time (Figure 8). For tricyclics, the proportion
fell from 3.4% in 2000 to 2.3% in 2017, with a sharp decline between 2002 and 2008. For
other ADMs, the proportion rose from 0.8% in 2000 to 1.7% in 2005, and then steadily fell
to 0.5% in 2017. For SSRIs, the proportion of individual prescriptions to exceed 60 days
was more stable, ranging between 1.3% and 1.7% from 2000 to 2017.
Figure 8 Proportion of individual prescriptions exceeding 60 days, by ADM type
Source: CPRD (Appendix 4)
0
1
2
3
4
5
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
%
SSRIs Tricyclics Other

34
3.5 Length of prescribing: proportion of continuous
prescribing periods exceeding thresholds
The trends for the proportion of continuous prescribing periods to exceed thresholds of one,
two, three and five years are presented for the three types of ADM examined in Figures 9 to
12.
3.5.1 One-year threshold
For each type of ADM, the proportion of continuous prescribing periods to exceed one year
increased between 2001 and 2011/2. Since 2012 there is evidence of a possible decline in
the proportion of continuous prescribing periods to exceed this threshold (Figure 9). For
SSRIs, the proportion to exceed one year rose from 14% in 2001 to 23% in 2011, then fell
to 19% in 2016. For tricyclics, the proportion to exceed one year rose from 24% in 2001 to
32% in 2008 and remained stable at this rate until 2012, before declining to 25% in 2016.
Other ADMs showed a more marked increase from 15% in 2001 to 35% in 2011, followed
by a fall to 24% in 2016.
The indications that in recent years the proportion of prescribing periods to exceed longer
thresholds may have fallen is somewhat surprising. However, these changes could be
explained by a range of factors, such as a greater willingness to switch between types of
ADM or to prescribe ADM more intermittently.
Figure 9 Proportion of continuous prescribing periods exceeding one year, by ADM type
Source: CPRD (Appendix 5)
0
5
10
15
20
25
30
35
40
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
%
SSRIs Tricyclics Other

35
3.5.2 Two-year threshold
The proportion of continuous prescribing periods to exceed two years was about half of the
proportion in excess of one year. For example, the highest proportion of SSRI continuous
prescribing periods in excess of two years was 10% in 2011, and in the same year 23%
exceeded one year.
Trends over time for continuous prescribing periods in excess of two years showed a
similar trend to the proportion exceeding one year. For all ADM types, prescribing in excess
of two years tended to increase from 2002 to 2011/12 and subsequently decreased in the
following years up to 2015 (Figure 10). The proportion of SSRI continuous prescribing
periods to exceed two years rose from 6% in 2002 to 10% in 2011, then declined to 8% in
2015. Similarly, the proportion of tricyclic prescribing periods to exceed two years rose from
11% in 2002 to 16% in 2008 and remained at this rate until 2012, before falling to 13% in
2015. The proportion of other ADMs prescribed in excess of two years more than doubled
from 7% in 2002 to 18% in 2012, and then steadily decreased to 12% in 2015.
Figure 10 Proportion of continuous prescribing periods exceeding two years, by ADM type
Source: CPRD (Appendix 6)
3.5.3 Three-year threshold
Prescribing trends in excess of three years varied between drugs types (Figure 11).
Between 4% and 6% of SSRI continuous prescribing periods exceeded three years. While
this proportion remained quite stable over time, the proportion of continuous prescribing
periods for tricyclics and other types of ADM increased steeply. For tricyclics this peaked in
2011, when 10% of continuous prescribing periods lasted more than three years. And for
other types of ADM, this peaked in 2012 when 11% of continuous prescribing periods
exceeded three years. While the proportion of continuous prescribing periods to exceed
three years appeared to be falling for SSRIs and other types of ADM, the trend for tricyclics
was less clear.
0
2
4
6
8
10
12
14
16
18
20
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
%
SSRIs Tricyclics Other

36
Figure 11 Proportion of continuous prescribing periods exceeding three years, by ADM type
Source: CPRD (Appendix 7)
3.5.4 Five-year threshold
The proportion of prescribing periods to exceed five years again varied by drug type (Figure
12). Overall, tricyclics were consistently about twice as likely as SSRIs to be prescribed
continuously for five years or more. The proportion to be prescribed for this duration
remained quite stable over time. Between 2005 and 2012, the proportion of continuous
SSRI prescribing periods to exceed five years ranged between 1-2% and tricyclics ranged
between 3-4%. The proportion of other drugs prescribed in excess of five years rose from
2% in 2005 to 4% in 2009, and then remained around 3-4% until 2012.
Figure 12 Proportion of continuous prescribing periods exceeding five years, by ADM type
Source: CPRD (Appendix 8)
0
2
4
6
8
10
12
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
%
SSRIs Tricyclics Other
0
1
2
3
4
5
2005 2006 2007 2008 2009 2010 2011 2012
%
SSRIs Tricyclics Other

37
4 Characteristics associated with longer-
term ADM prescribing
This chapter focuses on patients prescribed ADMs and identifies what characteristics are
associated with being prescribed to for longer periods. The focus here is on profiling those
patients prescribed to continuously for more than three years, as this threshold is the
longest to also yield a sample sufficient for robust subgroup analysis. There were too few
people in the sample prescribed to in excess of five years to ensure estimates were robust
once stratified by gender, age, or region. Results for thresholds exceeding one, two- and
five- year prescribing periods are available in the appendices for reference (Appendices 9
to 24).
4.1 Variation in continuous prescribing by gender
4.1.1 SSRIs
Among people prescribed SSRIs, men were more likely to be prescribed them for longer
than three years than women. This was the case for nearly every year examined (Figure
13).
Figure 13 Proportion of patients prescribed SSRIs continuously for periods exceeding three
years, by gender
Source: CPRD (Appendix 11)
4.1.2 Tricyclics
Among people prescribed tricyclics, the proportion of women and men to be prescribed
them for periods in excess of three years tended to be similar (Figure 14).
0
1
2
3
4
5
6
7
8
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
%Male
Female

38
Figure 14 Proportion of patients prescribed tricyclics continuously for periods exceeding
three years, by gender
Source: CPRD (Appendix 11)
4.1.3 Other ADMs
Up to 2007, there was little variation by gender in the extent to which other ADM were
prescribed long-term. However, from 2008 men prescribed ADM were more likely than
women prescribed ADM to be prescribed it for longer (Figure 15).
Figure 15 Proportion of patients prescribed other ADM continuously for periods exceeding
three years, by gender
Source: CPRD (Appendix 11)
0
2
4
6
8
10
12
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
%Male
Female
0
2
4
6
8
10
12
14
16
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
%
Male Female

39
4.2 Variation in continuous prescribing by age-
group
4.2.1 SSRIs
Older people prescribed SSRIs were more likely to be prescribed to in excess of three
years than younger people prescribed the drug. This association with age has remained
broadly consistent over time (Figure 14).
Figure 16 Proportion of patients prescribed SSRIs continuously for periods exceeding three
years, by age
Source: CPRD (Appendix 15)
4.2.2 Tricyclics
Overall, older people prescribed tricyclics were more likely to be prescribed to for periods in
excess of three years than younger people prescribed the drug. This association was not
consistent over time. For instance, people age over 80 and who were prescribed tricyclics
were the most likely to be prescribed to continuously in excess of three years up to 2010,
after which rates almost halved to below that of people aged 41-60 (Figure 15).
0
2
4
6
8
10
12
14
16
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
%
18-40 41-60 61-80 81+

40
Figure 17 Proportion of patients prescribed tricyclics continuously for periods exceeding
three years, by age
Source: CPRD (Appendix 19)
4.2.3 Other ADMs
Figure 18 Proportion of patients prescribed other ADMs continuously for periods exceeding
three years, by age
Source: CPRD (Appendix 23)
0
2
4
6
8
10
12
14
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
%
18-40 41-60 61-80 81+
0
2
4
6
8
10
12
14
16
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
%
18-40 41-60 61-80 81+

41
4.3 Characteristics associated with long-term
prescribing in 2012
Further analyses focused on data from 2012 was carried out. This also considered regional
variation and involved significance testing. It sought to identify the characteristics – among
those people prescribed each type of ADM – of being prescribed to for more than one, two,
three or five years. These results are presented in Tables 4, 5, and 6.
4.3.1 Gender variations in long-term prescribing
Among people prescribed ADMs in 2012, men were generally more likely than women to be
prescribed to for periods exceeding three and five years. This was the case for all drug
types apart from the proportion of continuous SSRI prescribing periods in excess of five
years, where women (1.5%) prescribed the drug were more likely than men to be
prescribed to for more than five years (1.1%).
4.3.2 Age-group variations in long-term prescribing
Among people prescribed ADMs in 2012, older people were generally more likely than
younger people to be prescribed to for periods exceeding five years; this was the case for
all types of ADM.
4.3.3 Regional variations in long-term prescribing
Among people prescribed ADM, those living in the North of England were more likely than
those living in other regions to be prescribed for SSRIs and tricyclics in excess of three and
five years. For example, among people prescribed SSRIs, 7% of those living in the North of
England were prescribed the drug for longer than three years compared with less than 5%
in other regions. Among people prescribed other ADMs, those living in the Midlands and
East of England were the most likely to be prescribed to for five years or more (7%).
Table 6 Characteristics of people prescribed SSRIs in 2012, by continuous prescribing period
length
SSRIs
Over 1
year
Over 2
Years
Over 3
Years
Over 5
Years
Mean age (years)
60.4***
63.0***
62.2***
65.7***
Gender (%)
Male §
24.4
10.8
5.7
1.1
Female
21.2***
8.8***
3.8***
1.5**
Region (%)
North of England §
28.8
12.7
6.6
2.4
Midlands and East of
England
19.8***
7.6***
3.3***
1.0***
London
22.4***
10.6***
4.5***
0.6***
South East
17.3***
7.9***
4.0***
1.4***
South West
21.2***
7.0***
1.7***
0.5***
* P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001, § - reference category Source: CPRD

42
Table 7 Characteristics of people prescribed Tricyclics in 2012 by continuous prescribing
period length
Tricyclics
Over 1
year
Over 2
Years
Over 3
Years
Over 5
Years
Mean age (years)
63.1***
62.7***
61.2
58.1***
Gender (%)
Male §
29.7
15.6
9.6
3.6
Female
30.2
16.2
8.5**
3.0*
Region (%)
North of England §
34.8
20.4
11.4
4.4
Midlands and East of
England
30.6***
16.5***
8.2***
2.7***
London
29.0***
16.9***
8.4***
3.7
South East
23.9***
10.9***
7.0***
2.8***
South West
27.0***
10.5***
6.7***
1.2***
* P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001, § - reference category
Source: CPRD
Table 8 Characteristics of people prescribed other ADMs in 2012 by continuous prescribing
period length
Other ADMs
Over 1
year
Over 2
Years
Over 3
Years
Over 5
Years
Mean age (years)
57.2***
57.3***
56.0
50.3***
Gender (%)
Male §
37.0
20.9
13.7
5.3
Female
31.3***
15.7***
8.8***
2.4***
Region (%)
North of England §
40.4
19.6
11.4
2.8
Midlands and East of
England
33.4***
20.6
12.3
6.6***
London
26.4***
15.8**
10.8
0.1***
South East
23.0***
11.8***
7.5***
0.9***
South West
39.6
18.7
11.0
5.0***
* P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001, § - reference category
Source: CPRD

43
5 Conclusion
Limitations
The analyses presented in this report have several limitations, including that they:
• Examine trends by broad class of ADM, and do not examine trends for
specific medicines.
• Provide descriptive visualisations of temporal trends, without extensive
significance testing or controlling for shifts in population demographics.
Further work is needed that takes a more complex approach. This study
focused on trends in prevalence in each year, other studies are needed that
focus on changes in incidence rates over time.
• Are based on a relatively small sample of patients from the CPRD (<50,000).
While it would have been preferable to have used the total CPRD sample or
a larger sample, the costs involved in obtaining a larger data extract were
beyond the resourcing available.
• More exploration of potential biases in the wider CPRD sample coverage is
needed, especially given changes in recent years in the number of practices
participating.
• These trends relate to NHS prescribing practice only, and CPRD data do not
indicate whether medicines were dispensed, whether they were taken, and
whether they were taken by the patient they were prescribed to or by
someone else. Further research is needed on the characteristics of people
prescribed to, examining a broader range of factors at both the patient-level
(e.g. ethnicity) and practice-level (e.g. practice rurality) in larger samples to
improve generalisability.
• The data could not tell us about reasons for prescribing and whether long-
term prescribing was clinically appropriate. Qualitative research with both
patients and prescribers could provide insight on this. Further research
should also examine the potential underlying reasons for why long-term
prescribing is increasing at the population level, including considering the
role of misuse and inappropriate prescribing.
• Narrow and conservatively defined rules on what constitutes a continuous
prescribing period were applied; furthermore, ‘intermittent’ prescribing or a
continuous period of prescribing that comprised transitions between ADM
classes (for example from SSRI to tricyclics) would not have counted as a
continuous prescribing period.
• At least five years of subsequent data need to be available in order to
estimate the proportion of prescribing periods in a given year that will
continue for at least five years. This means that trends in prescribing periods
greater than five years could only be generated up to 2012, and trends for
prescribing periods greater than three years could only be generated up to
2014.

44
Conclusions
ADM prescribing has increased and is relatively common
The number of ADM items dispensed in England doubled over the past decade.
The prevalence of ADM use among 16-64-year olds increased from one in a
hundred in 1993, to one in twenty in 2000 and 2007, and one in ten in 2014: the
proportion prescribed ADM at any point in the year was higher. These figures relate
to snapshots of prevalence in time, not incidence.
Prescribing periods average about a year
Since 2000, individual prescriptions have been issued for short periods of time -
overwhelmingly for less than 60 days.
Continuous prescribing periods, however, last much longer. In 2011, the average
SSRI prescribing period was just under a year, and the average tricyclic or other
ADM prescribing period lasted well over a year.
Average prescribing periods have gradually increased over time
This rise in the mean number of days of continuous prescribing was evident for all
types of antidepressant examined. The increase was least pronounced for SSRIs,
and most pronounced for other ADM.
One in twelve tricyclic and other ADM prescribing periods exceed three years
The proportion of SSRI prescribing periods to exceed three years remained around
4% between 2003 and 2014, while the proportion of ADM prescribing periods to
exceed this length doubled between 2003 (4%) and 2014 (8%).
There are indications that in recent years the proportion of prescribing periods to
exceed long thresholds has fallen. This could result from changes such as greater
switching between types of ADM.
Variation among those prescribed to:
Those more likely to be prescribed ADM beyond a three-year threshold tended to
be older, male, and living in the North of England – although this pattern varied by
type of drug.

45
6 References
BMA (2015) Prescribed drugs associated with dependence and withdrawal –
building a consensus for action.
http://bmaopac.hosted.exlibrisgroup.com/exlibris/aleph/a23_1/apache_media
/H6IB5G1BL8SX1KJ7VY4MCRCXVG6EV7.pdf
Cartagena Farias J, Porter L, McManus S, Strang J, Hickman M, Reed K, Smith N.
(2017) Prescribing patterns in dependence forming medicines. London:
NatCen.
Herrett E, Gallagher AM, Bhaskaran K, Forbes H, Mathur R, van Staa T, Smeeth L.
(2015). Data resource profile: Clinical practice research datalink (CPRD).
International Journal of Epidemiology, 44(3), 827-836.
https://doi.org/10.1093/ije/dyv098
Kendrick T. (2015). Long-term antidepressant treatment: time for a review.
Prescriber 26(19):7–10.
Lacobucci G. (2019). NHS prescribed record number of antidepressants last year.
BMJ, 364, 1508. https://doi.org/10.1136/bmj.l1508
Lockhart P, Guthrie B. (2011). Trends in primary care antidepressant prescribing
1995–2007: A longitudinal population database analysis. Br J Gen Pract, 61,
e565–72.
Lubian K, Weich S, Stansfeld S, Bebbington P, Brugha T, Spiers N, McManus S,
Cooper C. (2016) ‘Chapter 3: Mental health treatment and services’ in
McManus S, Bebbington P, Jenkins R, Brugha T. (eds.) Mental health and
wellbeing in England: Adult Psychiatric Morbidity Survey 2014. Leeds: NHS
Digital.
Mars B, Heron J, Kessler D, Davies NM, Martin RM, et al. (2017). Influences on
antidepressant prescribing trends in the UK: 1995–2011. Soc. Psychiatry
Psychiatr Epidemiol, 52, 193–200.
McManus S, Bebbington P, Jenkins R, Brugha T. (2016) Mental health and
wellbeing in England: Adult Psychiatric Morbidity Survey 2014. NHS Digital:
Leeds.
Middleton N, Gunnell D, Whitley E, Dorling D, Frankel S (2001) Secular trends in
antidepressant prescribing in the UK, 1975–1998, Journal of Public Health, 23,
4:1,262–267.
Moore M, Yuen HM, Dunn N, Mullee MA, Maskell J, Kendrick T. (2009). Explaining
the rise in antidepressant prescribing: a descriptive study using the general
practice research database. Brit. Med. J., 339, b3999.

46
Munoz-Arroyo R, Sutton M, Morrison J. (2005). Exploring potential explanations for
the increase in antidepressant prescribing in Scotland using secondary
analysis of routine data. Br J Gen Pract, 56(527): 423–428.
NICE (2004). Depression: Management of depression in primary and secondary
care National Clinical Practice Guideline Number 23.
http://guidance.nice.org.uk/ CG23
NHS Digital (2017). Prescriptions dispensed in the community – Statistics for
England, 2006–2016. https://digital.nhs.uk/data-and-information/publications/
statistical/prescriptions-dispensed-in-the-community/prescriptions-dispensed-
in-thecommunity-statistics-for-england-2006-2016-pas
Moody A, Mindell J, Faulding S. (2017) Health Survey for England 2016: Prescribed
medicines. NHS Digital https://files.digital.nhs.uk/pdf/3/c/hse2016-pres-
med.pdf
NHS Digital (2019). Prescription cost analysis-England 2018.
https://digital.nhs.uk/data-and-information/publications/statistical/prescription-
cost-analysis/2018.
Sarginson J, Webb R, Stocks SJ, Esmail A, Garg S, Ashcroft D. Temporal trends in
antidepressant prescribing to children in UK primary care, 2000-2015. Journal
of Affective Disorders. 2017. https://doi.org/10.1016/j.jad.2016.12.047
Sehmi R, Nyugen A, McManus S, Smith N. (2019). Trends in long-term prescribing
of dependence forming medicines. National Institute of Health Research.
Spence R, Roberts A, Ariti C, Bardsley M. (2014) Focus on: Antidepressant
Prescribing. Trends in the prescribing of antidepressants in primary care. The
Health Foundation, Nuffield Trust.
https://www.health.org.uk/sites/default/files/QualityWatch_FocusOnAndidepres
santPrescribing.pdf
Spiers N, Qassem T, Bebbington PE, McManus S, King M, Jenkins R, Brugha TS.
Increasing treatment, continuing burden: the prevalence and treatment of
common mental disorders in the English national population, 1993 to 2007.
British Journal of Psychiatry. 2016: 209, 150-6.
Taylor S, Annand F, Burkinshaw P, Greaves F, Kelleher M, Knight J, Perkins C,
Tran A, White M, Marsden J. An evidence review of dependence and
withdrawal associated with some prescribed medicines. 2019. Public Health
England, London.

47
7 Appendices
Appendix 1 Monoamine-oxidase inhibitors (MAOI)
prescribing items in England 2008-2018
Appendix 2 Average number of days of prescriptions
by ADM type
Year
SSRIs
Tricyclics
Other
2000
32.4
35.2
30.8
2001
33.0
34.9
30.8
2002
33.4
35.7
31.6
2003
33.7
35.1
31.5
2004
33.8
34.8
31.2
2005
33.4
34.5
31.2
2006
33.5
34.3
30.2
2007
33.5
33.8
30.7
2008
33.4
33.5
30.2
2009
33.6
33.8
29.7
2010
33.9
34.1
29.4
2011
34.0
33.9
29.1
2012
34.1
34.2
28.9
2013
34.0
33.8
29.6
2014
34.2
33.7
29.5
2015
34.3
33.4
30.3
2016
34.3
34.0
30.6
2017
34.5
34.4
29.9
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

48
Appendix 3 Average number of days of continuous
prescribing periods by ADM type
Year
SSRIs
Tricyclics
Other
2000
178.5
221.3
172.4
2001
209.2
290.6
229.4
2002
227.9
322.8
265.9
2003
258.4
348.6
282.5
2004
269.7
366.9
297.5
2005
281.4
392.9
358.6
2006
289.0
409.4
402.4
2007
292.9
429.8
405.9
2008
305.4
439.9
432.8
2009
308.7
442.1
444.4
2010
312.7
444.6
452.1
2011
322.3
443.7
465.8
2012
316.1
429.8
457.2
2013
313.9
420.4
447.5
2014
310.5
433.3
464.6
2015
292.1
414.6
423.1
2016
282.6
394.0
380.5
2017
238.2
320.5
321.1
Appendix 4 Proportion of individual prescriptions
exceeding 60 days, by ADM type
Year
SSRIs
Tricyclics
Other
2000
1.4
3.4
0.8
2001
1.5
3.7
0.8
2002
1.6
3.9
1.3
2003
1.7
3.7
1.4
2004
1.6
3.2
1.6
2005
1.4
3.2
1.7
2006
1.3
3.0
1.6
2007
1.3
3.0
1.5
2008
1.4
2.7
1.0
2009
1.3
2.7
1.0
2010
1.5
3.0
0.9
2011
1.4
2.9
0.8
2012
1.6
3.0
0.8
2013
1.6
3.0
0.7
2014
1.7
2.6
0.4
2015
1.6
2.5
0.7
2016
1.6
2.6
0.5
2017
1.6
2.3
0.5

49
Appendix 5 Proportion of continuous prescribing
periods exceeding 1 year, by ADM type
Year
SSRIs
Tricyclics
Other
2000
10.5
15.0
10.9
2001
14.4
24.3
15.1
2002
15.5
25.7
18.7
2003
17.7
26.8
21.3
2004
17.7
26.8
22.1
2005
19.1
27.4
27.4
2006
20.0
28.2
31.0
2007
20.6
29.9
28.3
2008
22.2
29.5
31.9
2009
21.6
30.0
33.3
2010
21.8
30.2
35.0
2011
22.5
29.9
35.3
2012
22.2
30.0
33.5
2013
22.3
28.3
29.2
2014
22.3
27.1
25.7
2015
20.6
25.5
24.5
2016
19.2
24.8
23.8
2017
13.7
19.5
18.8
Appendix 6 Proportion of continuous prescribing
periods exceeding 2 years, by ADM type
Year
SSRIs
Tricyclics
Other
2000
2.7
4.5
1.4
2001
4.4
8.6
5.3
2002
5.9
10.8
6.8
2003
7.0
12.0
6.7
2004
7.5
12.8
8.4
2005
8.2
14.0
11.5
2006
8.9
14.8
13.3
2007
8.4
15.4
12.9
2008
8.8
15.7
15.5
2009
9.1
16.1
16.7
2010
9.1
16.1
16.6
2011
9.6
16.1
16.9
2012
9.4
16.0
17.7
2013
8.9
14.9
15.4
2014
8.5
13.7
13.7
2015
7.5
12.6
12.4
2016
6.4
11.7
11.1
2017
4.2
8.5
8.0

50
Appendix 7 Proportion of continuous prescribing
periods exceeding 3 years, by ADM type
Year
SSRIs
Tricyclics
Other
2000
1.2
1.9
0.5
2001
1.9
3.8
2.3
2002
2.8
5.6
2.6
2003
3.8
6.5
3.4
2004
4.6
7.3
4.0
2005
5.1
8.3
5.8
2006
5.5
8.5
7.2
2007
5.2
8.9
7.3
2008
5.0
9.0
8.5
2009
4.9
8.9
9.1
2010
4.6
9.5
9.9
2011
4.8
9.8
10.2
2012
4.4
8.8
10.7
2013
4.1
8.6
9.3
2014
3.8
9.1
8.1
2015
3.1
8.3
5.3
2016
2.6
7.0
3.9
2017
1.9
4.3
3.0
Appendix 8 Proportion of continuous prescribing
periods exceeding 5 years, by ADM type
Year
SSRIs
Tricyclics
Other
2000
0.3
0.6
0.0
2001
0.4
1.0
0.4
2002
0.9
1.4
0.9
2003
1.6
1.6
1.4
2004
1.8
2.3
1.4
2005
2.0
3.3
1.8
2006
2.0
3.7
2.6
2007
1.9
3.7
3.4
2008
1.8
4.0
3.9
2009
1.7
3.9
4.0
2010
1.5
4.0
3.4
2011
1.6
3.8
3.4
2012
1.4
3.2
3.5
2013
1.1
2.9
3.2
2014
1.2
2.9
3.1
2015
0.9
2.4
2.0
2016
0.9
1.4
2.0
2017
0.8
1.0
1.8

51
Appendix 9 Proportion of patients prescribed
continuously for periods exceeding 1 year, by ADM
type and gender
SSRIs
Tricyclics
Other
Years
Male
Female
Male
Female
Male
Female
2000
11.3
10.2
12.9
15.8
17.2
7.3
2001
16.6
13.4
23.8
24.5
14.9
15.1
2002
18.5
14.2
26.3
25.5
17.2
19.6
2003
21.1
16.1
25.1
27.4
20.1
21.9
2004
21.8
15.7
28.2
26.2
20.6
22.9
2005
23.3
17.2
27.2
27.4
24.3
28.9
2006
22.9
18.6
27.7
28.4
28.1
32.4
2007
22.9
19.5
29.6
30.0
26.0
29.5
2008
23.9
21.5
28.9
29.8
34.4
30.7
2009
22.4
21.3
30.8
29.6
33.2
33.4
2010
23.7
20.9
29.6
30.5
38.6
33.1
2011
22.7
22.4
27.9
30.7
39.2
33.2
2012
24.4
21.2
29.7
30.2
37.0
31.3
2013
24.9
21.1
30.3
27.4
31.6
27.7
2014
25.3
21.0
26.6
27.4
27.9
24.3
2015
24.1
18.9
25.2
25.7
26.3
23.4
2016
21.7
18.0
24.4
25.0
26.0
22.4
2017
14.1
13.5
19.8
19.4
22.8
16.1
Appendix 10 Proportion of patients prescribed
continuously for periods exceeding 2 years, by ADM
type and gender
SSRIs
Tricyclics
Other
Years
Male
Female
Male
Female
Male
Female
2000
3.6
2.4
4.6
4.4
1.3
1.5
2001
5.5
4.0
8.2
8.7
4.1
5.9
2002
8.6
4.7
10.4
10.9
5.9
7.3
2003
9.6
5.7
11.8
12.0
5.8
7.2
2004
10.0
6.4
13.5
12.5
6.8
9.3
2005
10.8
7.0
13.6
14.1
10.2
12.1
2006
11.2
7.8
13.4
15.4
9.8
15.0
2007
9.7
7.8
13.8
16.2
11.4
13.7
2008
9.7
8.3
13.8
16.5
17.0
14.7
2009
9.5
8.9
15.9
16.2
18.5
15.7
2010
10.3
8.6
15.7
16.2
18.6
15.5
2011
10.6
9.2
15.4
16.4
19.5
15.3
2012
10.8
8.8
15.6
16.2
20.9
15.7
2013
10.2
8.3
15.7
14.6
17.5
14.2
2014
10.2
7.7
13.1
14.0
15.2
12.8
2015
8.9
6.8
11.0
13.4
12.6
12.4
2016
7.9
5.7
10.5
12.2
13.4
9.6
2017
5.3
3.7
6.7
9.4
11.9
5.6

52
Appendix 11 Proportion of patients prescribed
continuously for periods exceeding 3 years, by ADM
type and gender
SSRIs
Tricyclics
Other
Years
Male
Female
Male
Female
Male
Female
2000
1.5
1.0
3.1
1.4
0.0
0.8
2001
2.5
1.6
5.1
3.3
1.0
3.0
2002
4.6
2.0
6.6
5.2
1.4
3.3
2003
6.0
2.8
6.8
6.4
2.6
3.9
2004
6.5
3.7
7.7
7.2
3.2
4.5
2005
7.3
4.0
8.4
8.2
5.1
6.1
2006
7.3
4.7
7.8
8.7
5.9
7.9
2007
6.3
4.7
7.6
9.5
7.2
7.4
2008
5.3
4.8
7.4
9.7
9.7
7.9
2009
4.9
4.9
7.5
9.5
11.6
7.9
2010
5.1
4.4
8.3
10.0
12.8
8.4
2011
5.9
4.4
9.4
9.9
13.2
8.5
2012
5.7
3.8
9.6
8.5
13.7
8.8
2013
4.9
3.7
9.8
8.0
11.6
7.9
2014
4.7
3.3
9.5
8.9
10.1
6.8
2015
3.8
2.7
8.8
8.1
6.3
4.7
2016
3.5
2.2
8.0
6.5
6.1
2.5
2017
2.6
1.6
4.2
4.4
5.4
1.5
Appendix 12 Proportion of patients prescribed
continuously for periods exceeding 5 years, by ADM
type and gender
SSRIs
Tricyclics
Other
Years
Male
Female
Male
Female
Male
Female
2000
0.4
0.3
1.5
0.2
0.0
0
2001
0.6
0.4
2.6
0.3
1.0
0.1
2002
1.7
0.5
3.4
0.6
1.3
0.7
2003
2.5
1.1
3.4
0.9
1.6
1.3
2004
2.6
1.4
3.6
1.8
1.8
1.2
2005
3.1
1.5
4.6
2.8
2.9
1.3
2006
3.0
1.5
4.6
3.3
3.2
2.3
2007
2.6
1.6
3.9
3.6
4.5
2.8
2008
2.0
1.7
3.9
4.1
5.6
3
2009
1.4
1.8
4.1
3.8
5.9
3
2010
1.3
1.7
4.3
3.9
4.9
2.6
2011
1.4
1.7
4.1
3.7
5.2
2.3
2012
1.1
1.5
3.6
3.0
5.3
2.4
2013
1.0
1.2
2.9
2.9
4.8
2.2
2014
1.3
1.1
2.3
3.1
5.2
1.8
2015
0.7
0.9
1.7
2.7
3.2
1.3
2016
0.9
0.9
1.0
1.6
4.1
0.6
2017
0.9
0.7
0.7
1.1
4.1
0.3

53
Appendix 13 SSRIs: Proportion of patients prescribed
continuously for periods exceeding 1 year, by age
Year
18-40
41-60
61-80
81+
2000
4.6
9.6
15.9
25.9
2001
8.3
12.2
18.8
37.4
2002
10.1
13.4
20.3
35.2
2003
10.5
15.6
27.2
33.6
2004
9.9
15.7
27.7
30.8
2005
11.1
17.3
26.5
36.6
2006
11.9
18.1
27.0
38.1
2007
12.6
18.6
26.6
40.6
2008
12.8
18.9
30.8
41.3
2009
11.5
18.5
30.8
40.2
2010
11.8
18.4
29.4
45.1
2011
13.3
18.4
31.7
44.3
2012
12.6
19.3
31.2
41.6
2013
14.1
20.0
30.0
37.4
2014
12.1
20.5
28.4
42.5
2015
9.7
19.8
27.2
37.8
2016
8.6
18.3
26.3
39.9
2017
6.9
14.1
16.6
28.1
Appendix 14 SSRIs: Proportion of patients prescribed
continuously for periods exceeding 2 years, by age
Year
18-40
41-60
61-80
81+
2000
1.5
2.8
5.1
2.1
2001
3.1
4.1
7.2
5.1
2002
4.5
5.4
7.0
11.2
2003
4.3
6.7
9.4
13.3
2004
4.1
6.3
11.6
16.2
2005
4.6
6.4
11.8
20.0
2006
5.2
7.8
12.0
18.3
2007
5.1
7.4
11.1
17.0
2008
4.5
7.7
12.7
16.0
2009
4.4
7.0
14.2
18.4
2010
4.5
7.0
13.2
21.5
2011
5.1
6.8
14.0
24.2
2012
4.6
7.4
13.1
23.0
2013
5.2
7.2
12.4
19.1
2014
4.5
7.4
11.5
16.4
2015
3.9
6.2
10.6
15.8
2016
3.3
4.4
10.4
15.5
2017
2.6
3.3
5.9
9.9

54
Appendix 15 SSRIs: Proportion of patients prescribed
continuously for periods exceeding 3 years, by age
Year
18-40
41-60
61-80
81+
2000
0.8
1.6
1.4
0.7
2001
1.9
1.7
1.9
2.7
2002
3.2
2.2
3.6
2.6
2003
3.0
3.3
5.8
4.7
2004
3.2
3.6
7.4
7.6
2005
3.2
4.1
7.4
9.8
2006
3.2
4.6
7.9
11.8
2007
2.6
4.2
7.3
13.5
2008
2.3
4.2
6.8
11.4
2009
2.5
3.9
6.3
12.1
2010
2.1
3.8
6.4
11.0
2011
2.8
3.6
6.9
10.8
2012
2.4
3.5
5.9
10.3
2013
2.8
2.9
5.9
8.5
2014
2.0
3.2
5.5
6.9
2015
1.4
2.5
5.2
5.3
2016
1.1
1.6
5.4
4.7
2017
1.1
1.7
2.6
4.5
Appendix 16 SSRIs: Proportion of patients prescribed
continuously for periods exceeding 5 years, by age
Year
18-40
41-60
61-80
81+
2000
0.0
0.7
0.2
0.0
2001
0.5
0.4
0.5
0.0
2002
1.4
0.7
0.7
0.3
2003
1.4
1.1
2.9
1.0
2004
1.4
1.1
3.8
1.1
2005
1.7
1.3
3.7
2.2
2006
1.7
1.2
3.1
3.3
2007
1.3
1.2
2.8
4.9
2008
0.8
1.3
2.7
4.6
2009
0.6
1.4
2.5
3.9
2010
0.3
1.5
2.4
3.2
2011
0.4
1.2
2.8
3.8
2012
0.5
0.9
2.4
3.7
2013
0.5
0.7
2.2
2.1
2014
0.5
0.8
2.4
1.6
2015
0.3
0.6
1.8
1.3
2016
0.5
0.4
1.7
1.6
2017
0.6
0.5
1.2
1.8

55
Appendix 17 Tricyclics: Proportion of patients
prescribed continuously for periods exceeding 1 year,
by age
Year
18-40
41-60
61-80
81+
2000
7.1
15.1
17.9
15.2
2001
13.4
25.1
27.2
25.8
2002
13.5
25.9
28.4
31.4
2003
12.4
26.5
29.8
35.1
2004
14.9
25.2
29.9
33.0
2005
18.0
24.0
31.3
34.2
2006
21.5
25.2
29.8
37.8
2007
21.5
25.8
32.1
39.8
2008
20.7
26.5
31.6
37.8
2009
22.3
25.8
33.5
36.8
2010
18.9
26.2
34.3
38.4
2011
18.0
27.2
32.8
38.4
2012
21.5
28.7
33.3
31.0
2013
21.7
27.6
30.0
30.5
2014
18.2
28.0
29.1
26.4
2015
17.3
25.0
27.3
28.6
2016
18.4
22.3
29.2
22.6
2017
14.4
15.5
22.5
25.7
Appendix 18 Tricyclics: Proportion of patients
prescribed continuously for periods exceeding 2
years, by age
Year
18-40
41-60
61-80
81+
2000
1.0
5.1
5.9
2.6
2001
3.3
9.5
10.8
5.5
2002
4.9
11.9
11.8
11.2
2003
5.1
12.6
12.8
15.6
2004
7.9
12.7
13.5
16.1
2005
9.8
12.5
15.9
16.6
2006
7.6
14.2
16.0
19.7
2007
11.4
14.1
15.3
22.5
2008
12.3
13.9
16.1
21.6
2009
11.3
13.5
18.0
21.2
2010
8.1
14.5
18.4
20.7
2011
9.8
15.3
17.9
18.4
2012
9.7
16.4
17.8
15.0
2013
10.6
14.1
17.3
13.5
2014
6.5
14.1
16.2
11.5
2015
4.6
13.8
13.7
12.7
2016
2.1
11.3
14.1
12.5
2017
1.2
7.1
10.7
11.2

56
Appendix 19 Tricyclics: Proportion of patients
prescribed continuously for periods exceeding 3
years, by age
Year
18-40
41-60
61-80
81+
2000
0.7
2.0
2.7
0.7
2001
2.4
3.8
4.7
2.7
2002
3.9
6.0
5.7
5.9
2003
3.5
7.6
6.3
7.7
2004
5.1
7.8
7.1
9.0
2005
6.4
7.7
9.1
9.4
2006
6.3
8.7
8.6
9.5
2007
8.1
8.6
8.5
11.7
2008
9.1
7.8
9.5
10.7
2009
8.0
7.5
9.4
12.0
2010
5.7
9.1
10.0
12.4
2011
7.4
9.5
10.5
10.2
2012
6.1
10.2
8.9
7.5
2013
6.9
9.8
8.5
6.9
2014
4.5
10.7
9.6
6.9
2015
3.4
10.1
8.9
5.5
2016
2.1
7.0
9.4
2.7
2017
1.2
3.6
6.2
2.6
Appendix 20 Tricyclics: Proportion of patients
prescribed continuously for periods exceeding 5
years, by age
Year
18-40
41-60
61-80
81+
2000
0.0
0.6
1.0
0.0
2001
0.7
1.0
1.3
0.6
2002
1.9
1.3
1.4
1.5
2003
1.8
1.5
1.4
2.4
2004
2.9
1.7
2.2
4.1
2005
2.9
2.3
4.0
5.1
2006
2.6
2.9
4.3
5.1
2007
3.4
2.9
4.2
4.4
2008
4.2
3.5
4.5
4.0
2009
3.8
4.1
3.8
3.8
2010
3.9
4.5
3.9
3.3
2011
4.1
4.2
4.1
1.7
2012
3.9
3.5
3.3
1.0
2013
5.0
3.3
2.6
0.8
2014
2.7
4.5
2.2
1.0
2015
1.7
4.1
1.8
0.0
2016
0.0
1.8
1.9
0.0
2017
0.0
0.7
1.7
0.0

57
Appendix 21 Other ADMs: Proportion of patients
prescribed continuously for periods exceeding 1 year,
by age
Year
18-40
41-60
61-80
81+
2000
4.2
10.5
9.3
36.5
2001
9.3
13.8
18.7
29.5
2002
10.0
20.3
23.2
33.2
2003
15.3
19.9
24.2
42.3
2004
16.2
19.1
27.0
42.5
2005
19.6
23.2
33.9
50.1
2006
28.2
27.2
33.8
45.2
2007
22.7
25.1
33.8
38.5
2008
20.8
32.2
36.7
39.6
2009
25.2
33.2
35.4
43.5
2010
29.4
32.6
37.6
48.5
2011
30.8
32.2
41.4
41.6
2012
32.4
31.5
37.4
33.3
2013
24.8
28.5
31.3
34.0
2014
16.8
25.0
30.6
30.4
2015
14.7
25.7
26.4
28.8
2016
17.3
24.7
26.5
25.9
2017
14.5
18.7
22.6
16.1
Appendix 22 Other ADMs: Proportion of patients
prescribed continuously for periods exceeding 2
years, by age
Year
18-40
41-60
61-80
81+
2000
0.0
1.3
1.6
5.6
2001
1.4
5.2
5.5
16.5
2002
1.7
8.0
9.0
14.6
2003
3.9
6.6
8.9
12.2
2004
6.1
5.7
13.5
17.0
2005
6.1
10.2
17.6
16.9
2006
7.9
13.6
15.4
18.8
2007
11.5
10.5
16.8
16.0
2008
13.4
13.3
19.5
18.0
2009
15.2
17.0
15.6
20.8
2010
18.0
17.4
14.9
15.0
2011
16.7
18.3
16.7
12.4
2012
16.2
17.7
18.3
19.2
2013
10.9
14.8
16.7
22.0
2014
6.7
13.0
17.1
19.4
2015
5.9
12.8
14.7
13.8
2016
8.0
11.5
14.6
5.4
2017
7.9
6.6
12.1
1.3

58
Appendix 23 Other ADMs: Proportion of patients
prescribed continuously for periods exceeding 3
years, by age
Year
18-40
41-60
61-80
81+
2000
0.0
0.1
0.0
5.3
2001
0.0
0.8
3.1
13.4
2002
0.0
1.8
5.4
9.9
2003
2.4
2.7
5.0
7.4
2004
3.8
1.9
7.1
8.5
2005
3.3
4.0
10.7
8.1
2006
4.7
6.6
7.7
14.3
2007
7.4
5.6
8.7
10.3
2008
10.4
7.2
9.4
7.8
2009
12.0
7.6
8.6
10.1
2010
14.4
7.9
9.9
7.8
2011
12.3
8.4
12.4
7.7
2012
12.9
7.7
14.0
8.4
2013
8.8
8.1
11.3
9.6
2014
5.1
7.7
9.8
10.2
2015
4.2
5.4
5.4
5.1
2016
6.6
4.1
2.4
2.2
2017
6.2
3.2
1.3
1.0
Appendix 24 Other ADMs: Proportion of patients
prescribed continuously for periods exceeding 5
years, by age
Year
18-40
41-60
61-80
81+
2000
0.0
0.1
0.0
0.0
2001
0.0
0.2
1.6
0.0
2002
0.0
0.7
3.1
0.0
2003
1.0
1.3
2.5
0.0
2004
1.6
0.8
3.2
0.0
2005
2.0
1.2
3.6
0.0
2006
3.4
2.2
2.7
2.5
2007
5.3
2.6
3.4
2.3
2008
6.5
3.4
3.5
1.8
2009
6.7
3.7
3.3
2.0
2010
6.8
2.6
2.6
1.1
2011
5.6
3.1
2.9
1.3
2012
6.2
2.4
3.9
0.9
2013
4.5
3.1
3.4
1.0
2014
3.6
2.8
4.3
0.4
2015
3.8
2.1
1.7
0.0
2016
5.7
1.6
0.6
0.0
2017
5.1
2.0
0.0
0.0