A study of the safety and harms of antidepressant drugs for older people: a cohort study using a large primary care database

Citation

(2012), "A study of the safety and harms of antidepressant drugs for older people: a cohort study using a large primary care database", Clinical Governance: An International Journal, Vol. 17 No. 1. https://doi.org/10.1108/cgij.2012.24817aaa.004

A study of the safety and harms of antidepressant drugs for older people: a cohort study using a large primary care database

Article Type: Health technology assessment From: Clinical Governance: An International Journal, Volume 17, Issue 1

C.A.C. Coupland, P. Dhiman, G. Barton, R. Morriss, A. Arthur, T. Sach and J. Hippisley-Cox

Background

Depression is a common and debilitating condition in older people. People diagnosed with depression are generally prescribed antidepressant drugs, which they might receive for a long period of time. Adverse drug events may be more common in the treatment of depression in older people than in younger age groups owing to higher levels of comorbidity, physiological changes and polypharmacy, but few studies have been carried out in this group.

The aim of this study was to establish the relative safety and balance of risks for antidepressant treatment in older people.

Objectives

The project had five key objectives:

1. 1.

   to determine the relative and absolute risks of predefined adverse events in older people diagnosed with depression, comparing classes of antidepressant drugs (tricyclic and related antidepressants (TCAs) selective serotonin reuptake inhibitors (SSRIs); monoamine oxidase inhibitors (MAOIs); other antidepressants) as well as commonly prescribed individual drugs with non-use of antidepressant drugs;

2. 2.

   to directly compare the risk of adverse events for SSRIs with TCAs;

3. 3.

   to determine how dose and duration of antidepressant medication are associated with the risk of adverse events;

4. 4.

   to describe patterns of antidepressant use in older people diagnosed with depression, in particular the types and doses prescribed, the durations and the proportions of people switching between different antidepressant classes; and

5. 5.

   to estimate costs of antidepressant medication and primary care visits in older people diagnosed with depression, comparing patients by class of antidepressant drug.

Methods

Design

The study was a retrospective cohort study of patients aged 65 years and over diagnosed with depression and identified using the QResearch primary care database. A self-controlled case-series analysis was nested within the cohort study; this is an analysis “within” patients, which removes the effects of indication bias and unmeasured confounding for variables that do not vary over time.

 Setting

The study was based in 570 general practices in the UK supplying data to the QResearch (version 22) database.

Participants

The cohort study comprised patients diagnosed with a new episode of depression. Patients were eligible if they had a recorded diagnosis of depression between the ages of 65 and 100 years, which was recorded between 1 January 1996 and 31 December 2007 and occurred at least 12 months after registration with a study practice. Patients were excluded if they were temporary residents or had a recorded diagnosis of depression or prescriptions for antidepressants in the previous 12 months or if they had a diagnosis of schizophrenia, bipolar disorder or other types of psychoses. The cohort was followed up until 31 December 2008.

Patients in the cohort who had each of the outcomes of interest constituted the samples for the self-controlled case-series analysis.

Exposures

Details were extracted for all antidepressant medications prescribed during the follow-up period, including information on the prescription date, type of drug, dose and duration. The antidepressant drugs were grouped for analysis according to the major classes (TCAs, SSRIs, MAOIs, other antidepressant drugs) and commonly prescribed individual drugs were identified (TCAs – amitriptyline, dosulepin, lofepramine, trazodone; SSRIs – citalopram, escitalopram, fluoxetine, paroxetine, sertraline; and others – mirtazapine, venlafaxine). Combined treatment with different drugs was also assessed. Duration and dose of the drugs were examined where numbers were sufficient.

Outcome measures

Outcomes were extracted from the primary-care computer records and linked death certificates of patients in the cohort if they occurred after the index date and up until 31 December 2008. There were 13 predefined outcome measures:

1. 1.

   all-cause mortality;

2. 2.

   sudden cardiac death;

3. 3.

   suicide;

4. 4.

   attempted suicide/self-harm;

5. 5.

   myocardial infarction;

6. 6.

   stroke/transient ischaemic attack (TIA);

7. 7.

   falls;

8. 8.

   fractures;

9. 9.

   upper gastrointestinal bleeding;

10. 10.

    epilepsy/seizures;

11. 11.

    road traffic accidents;

12. 12.

    adverse drug reactions (ADRs); and

13. 13.

    hyponatraemia.

Analysis

In the cohort analysis, Cox proportional hazard models were used to estimate the associations of the outcomes with antidepressant use, treating antidepressant use as time varying and adjusting for potential confounding variables. Conditional Poisson regression was used to estimate relative incidence rates for defined time periods of risk after antidepressant prescriptions in the self-controlled case-series analyses.

Analysis of costs

In the base-case analysis, the cohort study data were used to estimate prescription costs for all antidepressant drugs over one- and five-year post-diagnosis periods compared with those prescribed no antidepressants, after controlling for differences between patients prescribed different antidepressants. For each outcome measure the subsequently calculated incremental costs were combined with estimates of the incremental number of averted events, to estimate the incremental cost per adverse event averted.

Results

A total of 60,746 patients were included in the study cohort. Of these patients, 54,038 (89.0 per cent) received at least one prescription for an antidepressant drug during follow-up, and 6,708 (11.0 per cent) received no antidepressant prescriptions. A total of 1,398,359 antidepressant prescriptions were received during the follow-up period: 54.7 per cent for SSRIs, 31.6 per cent for TCAs, 0.2 per cent for MAOIs and 13.5 per cent for the group of other antidepressant drugs. The median duration of treatment with antidepressants during follow-up was 364 days. Patients prescribed SSRIs were less likely than patients prescribed TCAs or other antidepressants to either stop after a single prescription or switch to another drug class in the year following their first prescription (37 per cent for SSRIs, 48 per cent for TCAs, 50 per cent for the group of other antidepressants).

The associations with the adverse outcomes were significantly different between the classes of antidepressant drugs for seven outcomes. For these outcomes use of SSRIs had the highest adjusted hazard ratios (HRs) for falls (1.66, 95 per cent confidence interval (CI) 1.58 to 1.73) and hyponatraemia (1.52, 95 per cent CI 1.33 to 1.75), and the group of other antidepressants had the highest adjusted HRs for overall mortality (1.66, 95 per cent CI 1.56 to 1.77), attempted suicide/self-harm (5.16, 95 per cent CI 3.90 to 6.83), stroke/TIA (1.37, 95 per cent CI 1.22 to 1.55), fracture (1.63, 95 per cent CI 1.45 to 1.83) and epilepsy/seizures (2.24, 95 per cent CI 1.60 to 3.15), all compared with when antidepressants were not being used. TCAs did not have the highest HR for any of the outcomes. Use of a combination of antidepressant drugs had higher HRs than any of the three main classes for eight outcomes.

The associations with the adverse outcomes were also significantly different between the individual drugs for seven outcomes. Trazodone (a tricyclic-related antidepressant) was associated with the highest adjusted HR for all-cause mortality and one of the highest HRs for attempted suicide/self-harm. Mirtazapine (in the group of other antidepressants) was associated with the highest rate of attempted suicide/self-harm and one of the highest rates for all-cause mortality and stroke/TIA. Venlafaxine (also in the group of other antidepressants) was associated with higher rates of stroke/TIA, fracture and epilepsy/seizures than the other drugs and one of the highest rates for all-cause mortality, attempted suicide/self-harm and falls. Citalopram (an SSRI) was associated with the highest rate of falls, but rates were similar for all of the SSRIs. Three SSRIs (citalopram, escitalopram and fluoxetine) were associated with significantly increased risks of hyponatraemia but paroxetine and sertraline were not. There was some evidence of increased rates of ADRs associated with lofepramine and sertraline. The TCAs, amitriptyline and dosulepin, had the lowest rates for many of these outcomes.

For all outcomes, rates tended to be highest in the first 28 days of starting antidepressant drugs, and also within 28 days of stopping medication. The absolute and excess risks were highest for all-cause mortality, falls, fracture, stroke/TIA and attempted suicide, and were low for the other outcomes. The self-controlled case-series results were generally consistent with the results from the cohort study analyses, but differed for attempted suicide/self harm and stroke/TIA.

The mean incremental cost (for all antidepressant prescriptions) ranged between £51.58 (amitriptyline) and £641.18 (venlafaxine) over the five-year post-diagnosis period. None of the 11 most commonly prescribed antidepressant drugs were estimated to consistently be the most cost-effective across the different adverse outcomes studied.

Conclusions

This study has found significant associations between use of antidepressant drugs and a number of adverse events in people aged 65 years and older with depression. There was no evidence that the use of SSRIs or drugs in the group of other antidepressants was associated with a reduced risk of any of the adverse outcomes compared with TCAs; however, these drugs may be associated with an increased risk for certain outcomes. Examination of individual drugs has found that trazodone, mirtazapine and venlafaxine were associated with the highest rates for several outcomes.

Limitations of this study include possible indication and channelling bias, and residual confounding. The presence and severity of depression change over time and this is likely to affect comparisons between treated and untreated periods of time. Differences in characteristics between patients prescribed different antidepressant drugs may account for some of the differences in associations between the drugs and the adverse outcomes, although the analyses adjusted for many potential confounding variables.

The risks of prescribing an antidepressant drug need to be weighed against the potential benefits of these drugs.

Implications for health care

In this study, SSRIs and drugs in the group of other antidepressants were not associated with a reduced risk of any of the adverse outcomes compared with TCAs, and they may even be associated with an increased risk for certain outcomes. This implies a careful evaluation of benefits and adverse outcomes is needed when prescribing antidepressants to older people, which should include consideration of TCAs and tailoring of drugs to individual patients.

In this study, mirtazapine, venlafaxine and trazodone were associated with higher rates than the other antidepressants for a number of outcomes including all-cause mortality and attempted suicide/self-harm. These potential risks should be considered when prescribing these drugs.

Use of a combination of antidepressants was associated with an increased risk for many of the adverse events studied; although this may reflect increased severity of depression and lack of response to monotherapy, it is a matter of concern and use of a higher dose of a single antidepressant should be considered as an alternative to combined treatment where appropriate.

This study found that rates of most outcomes were highest in the first 28 days after starting an antidepressant, which would support careful monitoring during the first weeks after prescribing antidepressants in older people.

Recommendations for research (in priority order)

1. 1.

   A long-term community-based randomised clinical trial is needed to compare benefits and common adverse effects between a low-dose TCA and an SSRI for older people with depression.

2. 2.

   Meta-analyses of randomised controlled trials of antidepressants in relation to adverse events in older people should be carried out to confirm these findings.

3. 3.

   These findings should be confirmed using other data sources of older people in a community setting.

4. 4.

   Further studies are needed to develop algorithms to individualise the risks associated with antidepressant use, so that patients at highest risk of these adverse events can be monitored closely.

5. 5.

   Further research could be conducted to estimate the loss in utility associated with different types of adverse events. This would enable the health economic analysis to be based on quality-adjusted life-years and so allow estimation of the relative cost-effectiveness of different antidepressants.

Funding

The National Institute for Health Research Technology programme.

© Crown Copyright

C.A.C. Coupland, P. Dhiman and J. Hippisley-Cox are all based in the Division of Primary Care, University of Nottingham, Nottingham, UKG. Barton and T. Sach are both based at the Health Economics Group (HEG), School of Medicine, Health Policy and Practice, University of East Anglia, Norwich, UKR. Morriss, is based in the Division of Psychiatry, University of Nottingham, Nottingham, UKA. Arthur is based in the Division of Nursing, University of Nottingham, Nottingham, UK