Abstract
Background: Pharmacoepidemiological studies often require detailed information on drug use patterns over time. However, many health care databases lack information on the prescribed dosage regimen, making it difficult to estimate duration of drug use adequately.
Objectives: To assess the influence of methods used to define duration of a single antidepressant (AD) prescription on exposure (mis)classification.
Methods: From the Dutch PHARMO-RLS database, all patients initiating ADs in 2001 were selected. Two methods were used to estimate the theoretical duration of an AD prescription; M1) using the actual dosage regimen and number of units dispensed and M2) assuming use of 1 defined daily dose per day. Treatment episodes were constructed allowing for a 30 day gap between prescriptions. Median duration of use (days) for the first AD treatment episode and persistence rates at 182 days were compared for both methods using Mann-Whitney U and Chi-square tests. Subsequently, we studied the impact of type of AD on these measures.
Results: 28,948 AD initiators were included (67.0% female, mean age 47.9 [Standard Deviation 16.4]) years. The median duration of the first AD treatment episodes was 102.0 (Interquartile range 303.0) days using M1 and 80.0 (290.0) days using M2 (p < 0.001). Patients starting TCAs had significantly longer (p < 0.001) median treatment episodes duration when using M1 compared to M2 (44.0 vs. 19.0 days, respectively). However, no significant difference (p = 0.64) was found for the median duration for patients initiating SSRIs when using M1 (139.0 days) compared to M2 (137.0 days). AD persistence at 182 days was higher using M1 (38.1% vs. 35.7%, p < 0.001) However, among patients initiating on TCAs the 182 day persistence was higher for M1 (27.1% vs. 10.9%, p < 0.001), but no difference was found for SSRI initiators (44.2% vs 44.3%).
Conclusions: Dose assumptions that are made due to lack of information on prescribed dose can lead to exposure misclassification when estimating antdepressant use, specifically for TCAs. How this misclassification influences risk estimates in pharmacoepidemiological studies needs further investigation.
Objectives: To assess the influence of methods used to define duration of a single antidepressant (AD) prescription on exposure (mis)classification.
Methods: From the Dutch PHARMO-RLS database, all patients initiating ADs in 2001 were selected. Two methods were used to estimate the theoretical duration of an AD prescription; M1) using the actual dosage regimen and number of units dispensed and M2) assuming use of 1 defined daily dose per day. Treatment episodes were constructed allowing for a 30 day gap between prescriptions. Median duration of use (days) for the first AD treatment episode and persistence rates at 182 days were compared for both methods using Mann-Whitney U and Chi-square tests. Subsequently, we studied the impact of type of AD on these measures.
Results: 28,948 AD initiators were included (67.0% female, mean age 47.9 [Standard Deviation 16.4]) years. The median duration of the first AD treatment episodes was 102.0 (Interquartile range 303.0) days using M1 and 80.0 (290.0) days using M2 (p < 0.001). Patients starting TCAs had significantly longer (p < 0.001) median treatment episodes duration when using M1 compared to M2 (44.0 vs. 19.0 days, respectively). However, no significant difference (p = 0.64) was found for the median duration for patients initiating SSRIs when using M1 (139.0 days) compared to M2 (137.0 days). AD persistence at 182 days was higher using M1 (38.1% vs. 35.7%, p < 0.001) However, among patients initiating on TCAs the 182 day persistence was higher for M1 (27.1% vs. 10.9%, p < 0.001), but no difference was found for SSRI initiators (44.2% vs 44.3%).
Conclusions: Dose assumptions that are made due to lack of information on prescribed dose can lead to exposure misclassification when estimating antdepressant use, specifically for TCAs. How this misclassification influences risk estimates in pharmacoepidemiological studies needs further investigation.
| Original language | English |
|---|---|
| Article number | 66 |
| Pages (from-to) | 36-37 |
| Number of pages | 2 |
| Journal | Pharmacoepidemiology and Drug Safety |
| Volume | 23 |
| Issue number | S1 |
| DOIs | |
| Publication status | Published - Oct 2014 |
| Externally published | Yes |