Other PHQ-9 item pairings are better than the PHQ-2: A Machine Learning analysis

Glavin, Darragh; Maekawa, Eduardo; Grua, Eoin Martino; Nakamura, Carina Akemi; Scazufca, Márcia; Araya, Ricardo; Peters, Tim; Hollingworth, William; Ven, Pepijn van de

doi:10.1016/j.procs.2022.09.089

Cited by 2 publications

(6 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The ML methodology applied to the primary data set in this study was based on the work done in our previous analysis [12]. As in the previous analysis, all 36 unique PHQ-9 item pairings were iterated.…”

Section: Discussionmentioning

confidence: 99%

“…In our previous analysis, the phq2&4 and phq2&8 were found to best classify individuals into screen positive and screen negative cases of depressive symptomatology [12]. The phq2&4 contains the second and fourth items of the PHQ-9, and the phq2&8 contains the second and eighth items.…”

Section: Ml-based Pairingsmentioning

confidence: 97%

“…To our knowledge, our previous analysis is the only one to date that has compared the performance of the PHQ-2 with alternative PHQ-9 item pairings and explored their combination with ML algorithms to predict depressive symptomatology [12]. ML algorithms have previously been combined with PHQ-9 items, but it was used to accurately predict suicidal ideation [13].…”

Section: Prior Workmentioning

confidence: 99%

“…ML algorithms have previously been combined with PHQ-9 items, but it was used to accurately predict suicidal ideation [13]. Our earlier work provided a data-driven ML analysis of all 36 possible PHQ-9 item pairings to predict depressive symptomatology [12]. Depressive symptomatology was defined as PHQ-9 scores ≥10 and was the reference standard used.…”

Section: Prior Workmentioning

confidence: 99%

“…The first objective was to re-evaluate the ML-based phq2&4 and phq2&8 on the primary data set with oversampling removed. Our previous analysis [12] used random oversampling to balance the output classes (PHQ-9<10 and PHQ-9≥10) and prevent the more frequent class from inflating performance scores, such as accuracy. However, as ROC curve analysis and area under the curve (AUC) are irrespective of class balance, there was no need to oversample; therefore, the analysis of the primary data set was rerun in this study.…”

Section: Objectivesmentioning

confidence: 99%

See 4 more Smart Citations

Patient Health Questionnaire-9 Item Pairing Predictiveness for Prescreening Depressive Symptomatology: Machine Learning Analysis

Glavin¹,

Grua²,

Nakamura³

et al. 2023

JMIR Ment Health

Self Cite

View full text Add to dashboard Cite

Background Anhedonia and depressed mood are considered the cardinal symptoms of major depressive disorder. These are the first 2 items of the Patient Health Questionnaire (PHQ)–9 and comprise the ultrabrief PHQ-2 used for prescreening depressive symptomatology. The prescreening performance of alternative PHQ-9 item pairings is rarely compared with that of the PHQ-2. Objective This study aims to use machine learning (ML) with the PHQ-9 items to identify and validate the most predictive 2-item depressive symptomatology ultrabrief questionnaire and to test the generalizability of the best pairings found on the primary data set, with 6 external data sets from different populations to validate their use as prescreening instruments. Methods All 36 possible PHQ-9 item pairings (each yielding scores of 0-6) were investigated using ML-based methods with logistic regression models. Their performances were evaluated based on the classification of depressive symptomatology, defined as PHQ-9 scores ≥10. This gave each pairing an equal opportunity and avoided any bias in item pairing selection. Results The ML-based PHQ-9 items 2 and 4 (phq2&4), the depressed mood and low-energy item pairing, and PHQ-9 items 2 and 8 (phq2&8), the depressed mood and psychomotor retardation or agitation item pairing, were found to be the best on the primary data set training split. They generalized well on the primary data set test split with area under the curves (AUCs) of 0.954 and 0.946, respectively, compared with an AUC of 0.942 for the PHQ-2. The phq2&4 had a higher AUC than the PHQ-2 on all 6 external data sets, and the phq2&8 had a higher AUC than the PHQ-2 on 3 data sets. The phq2&4 had the highest Youden index (an unweighted average of sensitivity and specificity) on 2 external data sets, and the phq2&8 had the highest Youden index on another 2. The PHQ-2≥2 cutoff also had the highest Youden index on 2 external data sets, joint highest with the phq2&4 on 1, but its performance fluctuated the most. The PHQ-2≥3 cutoff had the highest Youden index on 1 external data set. The sensitivity and specificity achieved by the phq2&4 and phq2&8 were more evenly balanced than the PHQ-2≥2 and ≥3 cutoffs. Conclusions The PHQ-2 did not prove to be a more effective prescreening instrument when compared with other PHQ-9 item pairings. Evaluating all item pairings showed that, compared with alternative partner items, the anhedonia item underperformed alongside the depressed mood item. This suggests that the inclusion of anhedonia as a core symptom of depression and its presence in ultrabrief questionnaires may be incompatible with the empirical evidence. The use of the PHQ-2 to prescreen for depressive symptomatology could result in a greater number of misclassifications than alternative item pairings.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Ml-based Pairingsmentioning

confidence: 97%

Section: Prior Workmentioning

confidence: 99%

Section: Prior Workmentioning

confidence: 99%

Section: Objectivesmentioning

confidence: 99%

See 3 more Smart Citations

Patient Health Questionnaire-9 Item Pairing Predictiveness for Prescreening Depressive Symptomatology: Machine Learning Analysis

Glavin¹,

Grua²,

Nakamura³

et al. 2023

JMIR Ment Health

Self Cite

View full text Add to dashboard Cite

show abstract

Patient Health Questionnaire-9 Item Pairing Predictiveness for Prescreening Depressive Symptomatology: Machine Learning Analysis (Preprint)

Glavin¹,

Grua²,

Nakamura³

et al. 2023

Preprint

View full text Add to dashboard Cite

BACKGROUND Anhedonia and depressed mood are considered the cardinal symptoms of major depressive disorder. These are the first two items of the Patient Health Questionnaire-9 (PHQ-9) and make up the ultra-brief PHQ-2 questionnaire used for pre-screening depressive symptomatology. The pre-screening performance of alternative PHQ-9 item pairings is rarely compared with that of the PHQ-2. OBJECTIVE Use a data-driven Machine Learning (ML) approach on the PHQ-9 items to find and validate the most predictive two-item depressive symptomatology ultra-brief questionnaire. Test the generalisability of the best pairings found on the primary dataset, with six external datasets from different populations to validate their use as pre-screening instruments. METHODS All thirty-six possible PHQ-9 item pairings (each yielding scores of 0-6) were investigated using Machine Learning-based methods with logistic regression models. Their performances were evaluated on the classification of depressive symptomatology, defined as PHQ-9 scores ≥10. This gave each pairing equal opportunity and avoided any bias in item pairing selection. RESULTS The ML-based phq2&4, the depressed mood and low energy item pairing, and phq2&8, the depressed mood and psychomotor retardation or agitation item pairing, were found to be the best on the primary dataset training split. They generalised well on the primary dataset test split with Area Under the Receiver Operating Characteristic Curves (AUC) of 0.954 and 0.946, respectively, compared with an AUC of 0.942 for the PHQ-2. The phq2&4 had a higher AUC than the PHQ-2 on all six external datasets and the phq2&8 was higher than the PHQ-2 on three. The logistic regression probability thresholds that maximised Youden’s index (an unweighted average of sensitivity and specificity) during cross-validation on the primary dataset were applied to the phq2&4 and phq2&8 models. The phq2&4 had the highest Youden’s index on two external datasets and the phq2&8 was highest on another two. The PHQ-2 ≥2 also had the highest Youden’s index on two external datasets, joint highest with the phq2&4 on one, but its performance fluctuated the most. The PHQ-2 ≥3 had the highest Youden’s index on one external dataset. The sensitivity and specificity achieved by the phq2&4 and phq2&8 were more evenly balanced than the PHQ-2 ≥2 and ≥3 cutpoints. For the PHQ-2, the former cutpoint excessively favoured sensitivity over specificity and vice versa for the latter. CONCLUSIONS The PHQ-2 was not a superior pre-screening instrument to other PHQ-9 item pairings. Evaluating all item pairings showed that, compared with alternative partner items, the anhedonia item underperformed alongside the depressed mood item. This suggests the inclusion of the anhedonia item in ultra-brief questionnaires may be a relatively arbitrary choice. The use of the PHQ-2 to pre-screen for depressive symptomatology could result in a greater number of misclassifications than alternatives such as the phq2&4 and phq2&8.

show abstract

Other PHQ-9 item pairings are better than the PHQ-2: A Machine Learning analysis

Cited by 2 publications

References 19 publications

Patient Health Questionnaire-9 Item Pairing Predictiveness for Prescreening Depressive Symptomatology: Machine Learning Analysis

Patient Health Questionnaire-9 Item Pairing Predictiveness for Prescreening Depressive Symptomatology: Machine Learning Analysis

Patient Health Questionnaire-9 Item Pairing Predictiveness for Prescreening Depressive Symptomatology: Machine Learning Analysis (Preprint)

Contact Info

Product

Resources

About