Logo Logo
Hilfe
Kontakt
Switch language to English
Opportunities and challenges of ecological momentary assessment for research on psychological interventions in depression
Opportunities and challenges of ecological momentary assessment for research on psychological interventions in depression
Ecological Momentary Assessment (EMA) has gained increasing popularity in depression research due to its ability to capture symptoms in real time and its potential to mitigate recall bias present in retrospective clinical assessments. In our EMA substudy, conducted within a large randomized controlled trial comparing three psychotherapeutic interventions - cognitive behavioral therapy (CBT), schema therapy (ST), and individual supportive therapy (IST) - we examined the opportunities and challenges of EMA in supporting psychotherapy research and practice in depression. Over the course of seven weeks of psychotherapy, 106 moderately to severely depressed patients provided momentary self-reports of depressive symptoms and repetitive negative thinking (RNT) three times daily. In addition, a comprehensive test battery, including weekly questionnaire assessments (WQA) of depressive symptoms and RNT, and clinical interviews of global functioning, was assessed before and at the end of the intervention. RNT is a transdiagnostic cognitive process that plays an important role in the development and maintanence of depression. Defined as repetitive, intrusive, relatively uncontrollable, and of negative content, RNT is an umbrella term for rumination and worry. The collected data was analyzed in three different studies: Study I compared the results of EMA and WQA in terms of measuring changes in depressive symptoms and RNT. We found that EMA was more sensitive to detecting between-group differences in intervention effects. Specifically, it revealed a superior reduction of RNT in the ST group compared to CBT and IST, which was not detected by WQA. The higher sensitivity of EMA for intervention effects may stem from a higher measurement reliability due to its real-time assessments, which avoid recall bias inherent in retrospective questionnaires. In contrast, WQA proved more effective in predicting changes in clinician-rated global functioning, potentially due to the common retrospective nature of the two measures. These findings highlight the complementary strengths of EMA, WQA, and clinical interviews and suggest that integrating these methods into clinical assessments could accelerate the comparison of intervention effects in clinical trials by improving measurement reliability. Study II focused on predicting treatment response (versus non-response) based on early improvements in depressive symptoms as assessed by EMA versus WQA. Our analyses showed that early improvements assessed by either method significantly predicted treatment response within three weeks of treatment initiation. However, WQA provided a clearer pattern of the optimal predictive change rate, indicating that a 10% symptom improvement at four weeks resulted in a true negative rate of 22% compared with a false negative rate of 0%. EMA provided comparable predictive power but lacked clarity in its pattern of an optimal predictive change rate. These findings demonstrate the potential of WQA and EMA for early treatment prediction, while suggesting that the clarity of the prediction pattern may depend on the measure of treatment response, which in our study was equivalent to the WQA predictor (both were operationalized with the BDI-II). Study III used EMA to explore the temporal relationships between momentary levels of depressive symptoms and concreteness levels of RNT, as well as their changes over the intervention course. Depressed patients tend to ruminate and worry in a less concrete manner than healthy individuals, i.e., their RNT is more unclear, aggregated, cross-situational and less solution-oriented. Our study showed that RNT concreteness explains additional variance in momentary levels of depressive symptoms that is not explained by the process of RNT itself. Notably, changes in RNT concreteness over the course of therapy interacted with patients' improvement in depression severity: patients who improved more than average showed a slight increase in concreteness, while those who improved less showed a decrease. In addition, higher levels of momentary depressive symptoms predicted subsequent decreases in momentary levels of concreteness, but not vice versa. These findings suggest that future studies should examine the long-term dynamics between RNT concreteness and depression. Based on these investigations, several strategies for refining EMA approaches in future studies to improve data quality and patient adherence are discussed. In all three studies, EMA’s ability to capture real-time fluctuations in depressive symptoms and RNT provided new insights into the assessment, treatment and understanding of depression. In addition, the combination of EMA with emerging technologies, such as passive data tracking and AI-based text analyses to automate complex rating procedures such as RNT concreteness ratings, offers significant potential for advancing EMA approaches. Besides, providing continuous personalized feedback on symptom progression, delivering just-in-time recommendations, and optimizing treatment module allocation based on EMA data are promising strategies for developing personalized, potentially more effective treatments. The temporal dynamics between depressive symptoms measurable with EMA support a growing shift from traditional latent-disease models to a network perspective on mental disorders, in which transdiagnostic factors like RNT and global functioning gain an increased role. Nevertheless, the rapid adoption of disruptive technologies like EMA and AI underscores the need for careful investigation of their opportunities and challenges in psychotherapy research and practice for depression.
Depression, Ecological Momentary Assessment, Psychotherapy
Tamm, Jeanette
2025
Englisch
Universitätsbibliothek der Ludwig-Maximilians-Universität München
Tamm, Jeanette (2025): Opportunities and challenges of ecological momentary assessment for research on psychological interventions in depression. Dissertation, LMU München: Fakultät für Psychologie und Pädagogik
[thumbnail of Tamm_Jeanette.pdf]
Vorschau
PDF
Tamm_Jeanette.pdf

2MB

Abstract

Ecological Momentary Assessment (EMA) has gained increasing popularity in depression research due to its ability to capture symptoms in real time and its potential to mitigate recall bias present in retrospective clinical assessments. In our EMA substudy, conducted within a large randomized controlled trial comparing three psychotherapeutic interventions - cognitive behavioral therapy (CBT), schema therapy (ST), and individual supportive therapy (IST) - we examined the opportunities and challenges of EMA in supporting psychotherapy research and practice in depression. Over the course of seven weeks of psychotherapy, 106 moderately to severely depressed patients provided momentary self-reports of depressive symptoms and repetitive negative thinking (RNT) three times daily. In addition, a comprehensive test battery, including weekly questionnaire assessments (WQA) of depressive symptoms and RNT, and clinical interviews of global functioning, was assessed before and at the end of the intervention. RNT is a transdiagnostic cognitive process that plays an important role in the development and maintanence of depression. Defined as repetitive, intrusive, relatively uncontrollable, and of negative content, RNT is an umbrella term for rumination and worry. The collected data was analyzed in three different studies: Study I compared the results of EMA and WQA in terms of measuring changes in depressive symptoms and RNT. We found that EMA was more sensitive to detecting between-group differences in intervention effects. Specifically, it revealed a superior reduction of RNT in the ST group compared to CBT and IST, which was not detected by WQA. The higher sensitivity of EMA for intervention effects may stem from a higher measurement reliability due to its real-time assessments, which avoid recall bias inherent in retrospective questionnaires. In contrast, WQA proved more effective in predicting changes in clinician-rated global functioning, potentially due to the common retrospective nature of the two measures. These findings highlight the complementary strengths of EMA, WQA, and clinical interviews and suggest that integrating these methods into clinical assessments could accelerate the comparison of intervention effects in clinical trials by improving measurement reliability. Study II focused on predicting treatment response (versus non-response) based on early improvements in depressive symptoms as assessed by EMA versus WQA. Our analyses showed that early improvements assessed by either method significantly predicted treatment response within three weeks of treatment initiation. However, WQA provided a clearer pattern of the optimal predictive change rate, indicating that a 10% symptom improvement at four weeks resulted in a true negative rate of 22% compared with a false negative rate of 0%. EMA provided comparable predictive power but lacked clarity in its pattern of an optimal predictive change rate. These findings demonstrate the potential of WQA and EMA for early treatment prediction, while suggesting that the clarity of the prediction pattern may depend on the measure of treatment response, which in our study was equivalent to the WQA predictor (both were operationalized with the BDI-II). Study III used EMA to explore the temporal relationships between momentary levels of depressive symptoms and concreteness levels of RNT, as well as their changes over the intervention course. Depressed patients tend to ruminate and worry in a less concrete manner than healthy individuals, i.e., their RNT is more unclear, aggregated, cross-situational and less solution-oriented. Our study showed that RNT concreteness explains additional variance in momentary levels of depressive symptoms that is not explained by the process of RNT itself. Notably, changes in RNT concreteness over the course of therapy interacted with patients' improvement in depression severity: patients who improved more than average showed a slight increase in concreteness, while those who improved less showed a decrease. In addition, higher levels of momentary depressive symptoms predicted subsequent decreases in momentary levels of concreteness, but not vice versa. These findings suggest that future studies should examine the long-term dynamics between RNT concreteness and depression. Based on these investigations, several strategies for refining EMA approaches in future studies to improve data quality and patient adherence are discussed. In all three studies, EMA’s ability to capture real-time fluctuations in depressive symptoms and RNT provided new insights into the assessment, treatment and understanding of depression. In addition, the combination of EMA with emerging technologies, such as passive data tracking and AI-based text analyses to automate complex rating procedures such as RNT concreteness ratings, offers significant potential for advancing EMA approaches. Besides, providing continuous personalized feedback on symptom progression, delivering just-in-time recommendations, and optimizing treatment module allocation based on EMA data are promising strategies for developing personalized, potentially more effective treatments. The temporal dynamics between depressive symptoms measurable with EMA support a growing shift from traditional latent-disease models to a network perspective on mental disorders, in which transdiagnostic factors like RNT and global functioning gain an increased role. Nevertheless, the rapid adoption of disruptive technologies like EMA and AI underscores the need for careful investigation of their opportunities and challenges in psychotherapy research and practice for depression.