Nearly a quarter of patients in the sample were prescribed an AD on at least one occasion during the study period. The prevalence of AD prescribing doubled between 1995 and 2011, although levels remained relatively stable between 2002 and 2005, when there was a notable reduction in prescribing to those under 30 years. The overall rise in prescribing was largely driven by an increase in SSRIs and other ADs. Our findings suggest the observed rise in prescribing is not due to an increased number of people starting meditation, but rather appears to be explained by an increase in the duration of treatment.
Strengths and limitations
The study included data from a large anonymised database of Primary Care Patients, which enabled examination of AD trends according to drug class, and also by age and gender. Trends were also examined over a long period of time (17 years). We examined AD prescribing regardless of indication, which is important, given that ADs are prescribed for a range of indications other than depression; only 67% of the patients in our study had a depression-related Read code, and only 39% of patients who started on AD had one recorded in the year prior to their first prescription. Sensitivity analysis conducted in the subgroup of patients with depression and excluding those prescribed low doses of amitriptyline found similar results.
Findings must also be interpreted in light of several limitations. First, our analysis is based on prescriptions issued in Primary Care only, and we do not have information about the dispensing of medications or patient compliance. Second, although trends were examined over a long period of time, data were only available until 2011. Third, the results may not generalise to practices that do not contribute to the CPRD, or to other countries which have different healthcare systems. Fourth, when calculating duration of treatment, we chose a minimum period of 4 months between prescriptions to indicate the end of a treatment episode. However, we are unable to say whether patients actually achieved remission during the treatment period. Findings from sensitivity analysis using a minimum duration of 30 days and 6 months were similar (results available on request). Finally, incident AD users were defined as those with no previous AD prescription during the study period. This may have led to a bias of selecting proportionally more ‘real’ new starters and fewer re-starters in the later part of the follow-up period.
Comparison with existing literature
The rise in prevalence of AD prescriptions found in this study is consistent with existing literature [1–10]. Studies regarding trends in incidence have been less consistent, with some studies reporting an increase in incident prescriptions over time [4, 6, 8], and others finding stable rates or a decrease [2, 10–12]. Our findings are also in line with previous reports that the rise in AD prescribing is due to an increase in the proportion of patients receiving long-term treatment [2, 4, 6–8, 12]. For example, a previous study using this database  found that the increase in AD prescriptions between 1993 and 2005 was explained by an increase in the proportion of patients receiving long-term prescriptions. Another UK study examining prescribing rates between 2003 and 2013  found a reduction in AD prescribing for incident depression, and an increase in prescribing for recurrent depression. We extend these studies by examining trends over a longer time period, which overlaps with several important external events, and by not limiting our analysis to those with depression. Examining prescribing trends in the whole population is important, as a large proportion of patients prescribed ADs do not have a depression diagnosis. Moreover, GPs increasingly use symptom rather than diagnostic codes [16–18], which could result in cases being missed. There is also evidence to suggest that the introduction of QOF performance indicators for depression may have influenced prescribing . This was found to be the case for both genders, and for both younger and older adults .
The prevalence and incidence of AD prescriptions was consistently higher amongst females than males. Although, as found in some previous studies [25, 26] there was a decrease in the ratio of female to male prescribing over time, indicating that AD prescribing has increased more in males. This could suggest there has been an increase in help-seeking behaviour in males, an increase in depression, or an increase in the number of males prescribed an AD for other indications.
The limited evidence on AD trends according to age suggests that AD prescriptions increase with increasing patient age [26–28]. We found an increase in prevalence of AD for all age groups, with the exception of those under 18 years [4, 5, 26]. Antidepressant prescribing in adolescents has received considerable attention following MHRA advice in 2003 against the initiation of SSRIs except fluoxetine in this age group. Consistent with our findings, studies from Europe, USA and Australia show that the regulatory warnings were associated with a reduction in the prescribing of ADs to children and adolescents [17, 29–33]. Our data also suggest that the warnings had a spill-over effect into other age groups [34, 35], with a join point indicating a change in trend in 2002. Several studies have found the reduction in AD prescribing following the regulatory warnings was not associated with a rise in adolescent suicides or non-fatal self-harm [31, 36, 37]. Rates began to rise following a second join point in 2005, which could suggest that concerns about a possible increased risk of suicidality have reduced.
We also explored whether trends in AD use were affected by a number of other events overlapping with the study period. These included the introduction of QOF performance indicators for depression in 2006, the 2008 recession, and improved access to cognitive behavioural therapy (CBT) through the IAPT initiative in 2006. Some studies have found an increase in AD prescribing or depression following the recession [18, 38]. For example, Kendrick et al.  found a rise in the prevalence of depression after 2008 in younger men, associated with increased unemployment. In this study, we found the best fitting model for trends in AD prevalence included a join point at 2008, following which the rate of prescribing accelerated. However, unlike Kendrick et al., our data suggest there was an increase in prevalence for both males and females, and for all age groups. Similar to previous research, we found no relationship between trends in AD prescribing and the introduction of IAPT services [38, 39], although it is possible that IAPT availability may have attenuated the recent rise in incident prescribing.
Implications for research and practice
Guidelines for depression recommend that patients continue medication for at least 6 months after remission to reduce the risk of relapse . Our findings of longer treatment periods for patients who began taking ADs in later years of the study suggest there is improved adherence to practice guidelines. While encouraging, it is important to note that the majority (65%) of patients who began AD treatment in 2009 discontinued treatment before the recommended time, with 32% of patients being prescribed ADs for 30 days or less. The increase in duration could also be attributed to the introduction of newer ADs, which may be better tolerated, or to changes in patient/GP attitudes regarding the treatment of mental illness. Alternatively, it could reflect failure by GPs to adequately follow-up patients and monitor treatment, with several studies finding that many patients on long-term AD treatment have not had a recent medication review [28, 41].
While long-term prescribing may be appropriate for some patients, currently little is known about the risks and benefits of taking AD medication long term. In the future, research, guidelines, and performance indicators should focus more on the appropriateness of long-term prescribing, and ensure regular review of patients who become established on long-term treatments.