TCA antidepressants are superior to placebo in treating depression in physical illness
Why is the treatment of depression in patients with physical illness important?
There is an increased risk of depression in people with a physical illness. Depression is a risk factor for poor prognosis of physical diseases. It is associated with reduced treatment adherence, poor prognosis, increased disability and higher mortality in many physical illnesses. Therefore, treatment of depression in this context should be an important goal for general medical services. Antidepressants are effective to treat depression in a physically healthy population, but there is less clarity regarding their effectiveness in physically ill patients.
Do TCA antidepressants work in the treatment of depression in patients with physical illness?
TCA antidepressants are more effective than placebo in treating depression. However, fewer patients receiving placebo dropped out compared to patients treated with antidepressants. Dry mouth was more common in patients treated with antidepressants.
Equity: do TCA antidepressants for depression work in the disadvantaged?
None of the included studies were conducted in low-income countries and the review does not provide a description of the participants according to the PROGRESS-Plus criteria.
- All relevant randomized controlled trials with depression as the primary outcome were included.
- All types of TCA antidepressants were eligible for inclusion.
- The duration of the trial had to be at least 4 weeks.
Population and setting
- Participants had to be adults (18 or older) with diagnoses of Major Depressive Disorder, Adjustment Disorder and Dysthymic Disorder based on standardized criteria (e.g. DSM-IV or ICD-10) and/or according to participant scores indicative of a diagnosis on validated tools (e.g. HRSD or BDI).
- Participants had one or more physical illness (a physical health problem known to have a biological underpinning).
- Trials in which the main comorbidity was a symptom-based diagnosis were excluded.
- Trials in which antidepressants were prescribed primarily to treat symptoms other than depression were excluded.
- Physical healthcare setting (in-patient and out-patient) in high and middle income countries (Denmark, UK, USA, Romania, Canada, France, Austria, the Netherlands, Italy, Sweden, Turkey, Belgium, China, Venezuela)
Summary of Findings [SOF] Table: TCA antidepressants versus placebo for patients with physical illness
Patient or population: Adults (18 or older)
Settings: High and middle income countries
Intervention: All types of TCA antidepressants
Anticipated absolute effects
No of Participants
Quality of the evidence
Risk with placebo
Risk difference with TCA antidepressants
Individuals who attained a 50% improvement of depressive symptoms at 6 to 8 weeks (TCA antidepressant efficacy 6-8 weeks) assessed with: validated measures (such as HDRS, the MADRS and the HADS)
|233 per 1000|
306 more per 1000 (130 more to 472 more)
Number of drop-outs
213 per 1000
101 more per 1000 (3 fewer to 229 more)
OR 1.69 (0.98 to 2.92)
299 (6 RCTs)
Very Low d
Number of adverse events (dry mouth)
212 per 1000
310 more per 1000 (90 more to 521 more)
OR 4.06 (1.61 to 10.21)
325 (9 RCTs)
The risk in the intervention group (and its 95% cofidence interval is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI = Confidence interval
OR = Odds ratio
a. Study protocols were available for very few trials, therefore risk of bias from selective outcome reporting was difficult to assess. The review authors' judgements about each methodological quality item presented risk of bias judgements for all studies included in the review.
b. The total number of events is less than 300 (a threshold rule of thumb value).
c. Funnel plots of the response to treatment efficacy analysis showed a scarcity of smaller negative trials. This indicates risk of bias due to small studies without statistically significant results remaining unpublished.
d. The total number of events is less than 300 and the 95% confidence interval around the pooled effect includes no effect.
Comments on this summary? Please contact Jennifer Petkovic.