4. Discussion
Our results showed that the risk of falling in older users of TCAs is possibly associated with drug exposure, whereas no association was found in users of SSRIs. These results need to be interpreted with caution considering the small sample size and subsequently confinement to unadjusted analysis for the TCA users.
To our knowledge, this is the first study to assess whether antidepressant blood concentrations are associated with fall risk within antidepressant users. Falls have been established to be an important ADE in community-dwelling older persons. Previous studies have consistently established the association between antidepressant use and falls4. However, literature linking antidepressant plasma concentrations and falls is scarce. A first indication that measuring antidepressant plasma concentrations might be of value in falls prevention was demonstrated in a case series of patients with drug-induced falls. This study suggested that TDM may help identify patients with drug-induced falls and confirm clinical suspicion of an ADE 21. The observed possible association between TCA concentrations and fall risk is in line with earlier studies that addressed the role of antidepressant plasma concentrations in fall-related outcomes, e.g. orthostatic hypotension (OH)22. Significant correlations between serum concentrations of amitriptyline and fluvoxamine and orthostatic blood pressure drop have been demonstrated 22. OH is considered an important fall risk factor and pathway in drug-related falls 23. OH results in transient cerebral hypoperfusion upon standing, which may result in a syncope and/or fall. OH is a prevalent ADE in antidepressants, best known for TCAs, but also seen with SSRIs 7. Furthermore, anticholinergic activity, which can increase fall risk, by among others risk of delirium and visual disturbances, increases with increasing plasma concentrations of nortriptyline (even at therapeutic levels) 24.
We observed a possible increased fall risk for users with higher plasma concentrations of TCA , which is in line with our hypothesis. However, the sample size was small. Furthermore, especially for amitriptyline the measured concentrations were below reference range for therapeutic effect or in therapeutic range, thus we cannot generalize our findings to potentially toxic concentrations. However, older adults are more at risk for adverse events due to altered pharmacokinetics and – dynamics25. TDM for TCAs is well-established in clinical practice, both for therapeutic effect and toxicity26-29. For SSRIs, the clinical use of TDM is less frequently employed, especially in older adults 30. For most SSRIs therapeutic reference ranges are wide, evidence for a relationship between drug concentration and therapeutic outcome is weak and risk for toxicity is relatively low compared to TCAs29, 31, 32. So in general, available data do not suggest benefit for (routine) monitoring of SSRI plasma concentrations31. However, TDM for SSRIs is recommended in specific populations like advanced age 13, 32. Since, studies have shown higher exposure to antidepressant concentrations compared to younger patients 33, 34. And also adequate use of TDM has been shown to be cost effective in older adults 35, 36. Thus, the lack of applying TDM and lack of lower blood concentrations as target, can possibly contribute to the risk of falls, an important ADE.
An important strength of our study is the prospective nature, using fall calendars. In falls research this method is considered the golden standard to avoid recall bias 37. Also, data on medication use was collected thoroughly, using both pharmacy prescription data and self-reported medication lists. Our study also has some important limitations. First, limited power was a problem. Despite the large B-PROOF database, the remaining group of participants that met the inclusion criteria was small as prevalence of antidepressant usage was low. Due to small groups, we needed to pool the data and needed to standardize and categorize the plasma concentrations. Second, information about plasma concentrations during the fall incident was not available, since such an unplanned event cannot be anticipated for. Thus we only had blood samples from baseline and follow-up study visit. Baseline or follow-up plasma concentrations might not be representative for the plasma concentration at the time of the fall incidents. Also, for some participants follow-up samples were missing. Thus it was unsure whether they were an actual antidepressant user at the time of the fall incident. However, this issue probably did not affect our results, since when repeating the analyses in the participants with baseline and follow-up samples, the results did not change significantly (data not shown). Third, for the association between delta concentrations and fall risk cannot be ruled out that concentrations changes did not reflect clinical relevant differences. By coding them binary (either increase or decrease) it could be that not relevant changes are mixed with clinical relevant changes. Fourth, our cohort consisted of relatively healthy community-dwelling older people. Frail older adults, can exhibit different patient characteristics, different pharmacokinetics which could lead to different results 25.