Catherine K. EttmanpersName, Salma M. AbdallapersName, Gregory H. CohenpersName, Laura SampsonpersName, Patrick M. VivierpersName, Sandro GaleapersName (2020date). “Prevalence of Depression Symptoms in US Adults Before and During the COVID-19 Pandemic.” JAMA Network Opentitle·venue.
Object of study — the prevalence of depression symptomsterm in US adults during the first fortnight of the COVID-19 pandemic, set against a pre-pandemic national baseline measured with THE SAME INSTRUMENT
Research question — how much higher depression symptomterm prevalence was during COVID-19 than before it, and which social and economic resources predicted it
Contribution — the first nationally representative before-and-during comparison of US depression symptomsterm on a single validated screenerterm, and a severity gradient showing the increase is largest at the severe end
phenomenon · claim-type · cross-domain-transfer · mechanism · claim-relation · replication-status · limitation — grounded in the full text| phenomenon | a population-scale rise in depression symptoms concentrated in the lowest-resource groups, rising monotonically with counted stressor exposure — “Experiencing more COVID-19 stressors was also associated with greater odds of depression symptoms compared with people with low stressor exposure (medium: OR, 1.77 [95% CI, 1.16-2.71]); high: OR, 3.05 [95% CI, 1.95-4.77]).” | |
| claim-type | existence — “This survey study found that prevalence of depression symptoms in the US increased more than 3-fold during the COVID-19 pandemic, from 8.5% before COVID-19 to 27.8% during COVID-19.” | |
| method — evidence | — “This nationally representative survey study used 2 population-based surveys of US adults aged 18 or older. During COVID-19, estimates were derived from the COVID-19 and Life Stressors Impact on Mental Health and Well-being study, conducted from March 31, 2020, to April 13, 2020. Before COVID-19 estimates were derived from the National Health and Nutrition Examination Survey, conducted from 2017 to 2018.” | |
| object of study — evidence | — “This nationally representative survey study used 2 population-based surveys of US adults aged 18 or older. During COVID-19, estimates were derived from the COVID-19 and Life Stressors Impact on Mental Health and Well-being study, conducted from March 31, 2020, to April 13, 2020. Before COVID-19 estimates were derived from the National Health and Nutrition Examination Survey, conducted from 2017 to 2018.” | |
| research question — evidence | — “This survey study found that prevalence of depression symptoms in the US increased more than 3-fold during the COVID-19 pandemic, from 8.5% before COVID-19 to 27.8% during COVID-19.” | |
| contribution — evidence | — “Overall, prevalence was 1.5-fold higher for mild depression symptoms, 2.6-fold higher for moderate depression symptoms, 3.7-fold higher for moderately severe depression symptoms, and 7.5-fold higher for severe depression symptoms categories during COVID-19 compared with before COVID-19.” | |
| mechanism | the explanation offered is material rather than clinical - income, savings under $5000, and a count of realised stressors such as job loss - and each is carried by an odds ratio with an interval rather than by a narrative — “Compared with individuals with an annual household income of $75 000 or more, those with a household income of $19 999 or less had 2.4-fold increased odds of depression symptoms (OR, 2.37 [95% CI, 1.26-4.43]). Individuals with household savings less than $5000 had 1.5-fold increased odds of depression symptoms (OR, 1.52 [95% CI, 1.02-2.26]).” | mechanisms[0] |
| mechanism | the dose-response across the stressor score is the strongest internal evidence for the material reading: low is the reference, medium 1.77 and high 3.05, ordered, and BOTH INTERVALS EXCLUDE 1. They do NOT exclude each other - 1.16-2.71 and 1.95-4.77 overlap - so the ordering of the two raised categories is not established by this quote, only their separation from the reference — “Experiencing more COVID-19 stressors was also associated with greater odds of depression symptoms compared with people with low stressor exposure (medium: OR, 1.77 [95% CI, 1.16-2.71]); high: OR, 3.05 [95% CI, 1.95-4.77]).” | mechanisms[1] |
| supports | the 2014 review finding that depression increases during and after traumatic events - the paper says in its own words that it ADDS TO this literature — “Findings from a 2014 review 3 on trauma and mental health suggest that depression increases during and after traumatic events; our study adds to this literature.” | claim_relations[0] |
| extends | Ni et al's before-and-after study of the 2019 Hong Kong protests, WHICH USED THE SAME INSTRUMENT - the paper makes instrument identity the explicit licence for comparing its numbers with theirs — “A 2020 study by Ni et al 8 analyzed depression symptoms before and after political unrest in Hong Kong using the same measure of depression symptoms we deployed in this study. They reported national depression symptoms prevalence before the unrest to be 6.5% (compared with 8.5% in our pre–COVID-19 US sample) and 11.2% in 2019 during unrest (compared with 27.8% in our during–COVID-19 sample).” | claim_relations[1] |
| supports | Lai et al's finding of substantial depression among Chinese health care workers - offered as consistent WHILE naming the population and timing differences that stop the percentages being read against each other — “In a study of health care workers in Hubei province and surrounding regions, Lai et al 15 found similar levels of depression: 49.6% of participants had no depression (vs 47.5% of participants in the CLIMB sample), while 35.6% of participants had mild depression (vs 24.6% in the CLIMB sample), 8.6% of participants had moderate depression (vs 14.8% in the CLIMB sample), and 6.2% of participants had moderately severe depression (vs 7.9% in the CLIMB sample). The Lai et al 15 sample in China included only health care professionals and was concentrated in the Hubei region, while our sample included a representative sample of all US residents and sampled the whole country.” | claim_relations[2] |
| sense | depression symptoms - AS OPERATIONALISED HERE, A PHQ-9 SCORE OF 10 OR GREATER, on an instrument validated BEFORE the study with a published sensitivity of 88% and specificity of 88% AT THAT CUT-OFF. The threshold is inherited, not asserted, and that is the single largest methodological difference between this paper and qiu2020 — “Depression symptoms in both studies were assessed using the Patient Health Questionnaire–9, a clinically validated survey with a sensitivity of 88% and a specificity of 88% at a cutoff score of 10 or higher.” | senses[0] |
| sense | the severity categories - none 0-4, mild 5-9, moderate 10-14, moderately severe 15-19, severe >=20, CITED TO THE VALIDATION PAPER rather than chosen here. The flattened `19` after the list is that citation marker — “Depression symptom categories were defined as none (score, 0-4), mild (score, 5-9), moderate (score, 10-14), moderately severe (score, 15-19), and severe (score, ≥20). Binary classification of depression symptoms was defined by a score of 10 or greater.” | senses[1] |
| sense | COVID-19 stressor exposure - A COUNT, 0 to 13, OF REALISED EVENTS (job loss, a death, financial problems), then CUT INTO THIRDS. AND THE THIRDS ARE NOT THIRDS OF THE SCALE: equal parts of 0-13 would break at about 4 and 9, and the paper breaks at 2 and 4. The cut points are therefore fitted to the RESPONSES, not to the instrument, so `high exposure` means `in the upper part of THIS survey` and does not transfer — “We assessed 13 stressors based on prior studies conducted after traumatic events. 20 , 21 Examples of COVID-19 stressors included losing a job, death of someone close to you owing to COVID-19, and having financial problems. We excluded stressors that were capturing constructs applicable only to specific groups and created a score ranging from 0 to 13 to measure cumulative exposure to COVID-19 stressors. We divided the scores into thirds to measure low, medium, and high exposure to COVID-19-induced stressors. Cutoffs for stressor score categories were low (score, 0-2), medium (score, 3-4), and high (score, 5-13).” | senses[2] |
| sense | `consistent` - USED HERE FOR TWO NUMBERS THAT DIFFER BY NEARLY A FACTOR OF TWO. 4.7% with major depressive disorder with severe impairment against 8.5% with depression symptoms, same country, adjacent years, and the paper calls the pair consistent BECAUSE IT NAMES THE CONSTRUCT DIFFERENCE. This is the correct use of the word and it is rare — “Our estimates of pre–COVID-19 depressive symptoms are consistent with other population-level nationally representative studies that captured different constructs of depression; in 2018, 4.7% of adults aged 18 years or older surveyed in the National Survey on Drug Use and Health had major depressive disorder with severe impairment, compared with 8.5% of adults in the NHANES having depression symptoms.” | senses[3] |
| sense | major depressive disorder - NOT MEASURED HERE and explicitly `a more severe measure of depression` than this paper's outcome. The paper uses the ordering to make a PREDICTION - that MDD estimates should rise too - rather than to claim it found MDD — “While major depressive disorder is a more severe measure of depression, based on the results of our study, we may expect to see increases in the presence of major depressive disorder in other studies conducted during the COVID-19 pandemic.” | senses[4] |
| control | the instrument is held identical across the two arms, and identity of measurement is named as the thing that makes the comparison possible — held — “Depression symptoms in both studies were assessed using the Patient Health Questionnaire–9, a clinically validated survey with a sensitivity of 88% and a specificity of 88% at a cutoff score of 10 or higher.” — rules out instrument change as the explanation of the before-during difference - the one confound qiu2020 and gao2020 cannot exclude against each other — “We used identical measures of outcome assessment to enable cross-study comparability.” | controls[0] |
| control | whether the two samples are drawn from comparable frames — partial — “First, we compare 2 cross-sectional data sources; by comparing NHANES and CLIMB, we are assessing different individuals. We do not survey the same individuals over time, so the NHANES data can provide only a proxy for the true baseline level of depression for CLIMB participants. Differences in prevalence of depression between groups may in part be driven by differences in demographic characteristics, random selection, recruitment, and sampling frame.” — rules out nothing empirically - both frames are probability-based and address-based, which is offered as 'mitigating this concern', but the arms are DIFFERENT INDIVIDUALS and the baseline is called a PROXY by the authors — “However, these are both population-representative surveys, drawn randomly with probability-based sampling using address-based sampling of US households, mitigating this concern.” | controls[1] |
| control | whether the panel arm was already more depressed than the population before COVID-19 — failed — “It is impossible to anticipate how the difference in previous depression could, if at all, affect the magnitude of depression found in our study. It is possible that individuals with a known history of depression were more likely to have ongoing depression or have had better access to treatment that would mitigate the current trauma.” — rules out nothing - measured at 18.8% prior diagnosis against a national 16.5%, and the authors state it is IMPOSSIBLE TO ANTICIPATE how that difference bears on the result — “Second, some differences in the NHANES and CLIMB depression estimates may be owing to the differences in baseline mental health across participants; for example, estimates of responses to the assessment of previous diagnosis of depression from 2016 suggest that participants in the AmeriSpeak panel had a 2.3–percentage point higher prevalence of previous diagnosis of depression than the national baseline as measured through the Behavioral Risk Factor Surveillance System (18.8% in AmeriSpeak vs 16.5% in the national baseline).” | controls[2] |
| control | the multivariable model controls sex, age, race/ethnicity, household size, education, marital status, income, savings and stressor score simultaneously — held — “Experiencing more COVID-19 stressors was also associated with greater odds of depression symptoms compared with people with low stressor exposure (medium: OR, 1.77 [95% CI, 1.16-2.71]); high: OR, 3.05 [95% CI, 1.95-4.77]).” — rules out demography as the explanation of the income, savings and stressor gradients - all three survive adjustment with intervals excluding 1 — “Among the CLIMB sample, after controlling for sex, age, race/ethnicity, household size, education, marital status, household income, household savings, and COVID-19 stressor score, participants with lower social and economic resources and with higher COVID-19 stressor scores had higher odds of depression symptoms compared with participants with higher social and economic resources or lower COVID-19 stressor scores” | controls[3] |
| weight | the 2014 review finding that depression increases during and after traumatic events - the paper says in its own words that it ADDS TO this literature — 0.8 — “Findings from a 2014 review 3 on trauma and mental health suggest that depression increases during and after traumatic events; our study adds to this literature.” | claim_relations[0] |
| weight | Ni et al's before-and-after study of the 2019 Hong Kong protests, WHICH USED THE SAME INSTRUMENT - the paper makes instrument identity the explicit licence for comparing its numbers with theirs — 0.85 — “A 2020 study by Ni et al 8 analyzed depression symptoms before and after political unrest in Hong Kong using the same measure of depression symptoms we deployed in this study. They reported national depression symptoms prevalence before the unrest to be 6.5% (compared with 8.5% in our pre–COVID-19 US sample) and 11.2% in 2019 during unrest (compared with 27.8% in our during–COVID-19 sample).” | claim_relations[1] |
| weight | Lai et al's finding of substantial depression among Chinese health care workers - offered as consistent WHILE naming the population and timing differences that stop the percentages being read against each other — 0.7 — “In a study of health care workers in Hubei province and surrounding regions, Lai et al 15 found similar levels of depression: 49.6% of participants had no depression (vs 47.5% of participants in the CLIMB sample), while 35.6% of participants had mild depression (vs 24.6% in the CLIMB sample), 8.6% of participants had moderate depression (vs 14.8% in the CLIMB sample), and 6.2% of participants had moderately severe depression (vs 7.9% in the CLIMB sample). The Lai et al 15 sample in China included only health care professionals and was concentrated in the Hubei region, while our sample included a representative sample of all US residents and sampled the whole country.” | claim_relations[2] |
| edge | depression symptoms —operationalised-as→ Patient Health Questionnaire–9 — “Depression symptoms in both studies were assessed using the Patient Health Questionnaire–9, a clinically validated survey with a sensitivity of 88% and a specificity of 88% at a cutoff score of 10 or higher.” | concept_edges[0] |
| edge | COVID-19 stressors —operationalised-as→ stressor score — “We assessed 13 stressors based on prior studies conducted after traumatic events. 20 , 21 Examples of COVID-19 stressors included losing a job, death of someone close to you owing to COVID-19, and having financial problems. We excluded stressors that were capturing constructs applicable only to specific groups and created a score ranging from 0 to 13 to measure cumulative exposure to COVID-19 stressors. We divided the scores into thirds to measure low, medium, and high exposure to COVID-19-induced stressors. Cutoffs for stressor score categories were low (score, 0-2), medium (score, 3-4), and high (score, 5-13).” | concept_edges[1] |
| edge | major depressive disorder —distinct-from→ depression symptoms — “Our estimates of pre–COVID-19 depressive symptoms are consistent with other population-level nationally representative studies that captured different constructs of depression; in 2018, 4.7% of adults aged 18 years or older surveyed in the National Survey on Drug Use and Health had major depressive disorder with severe impairment, compared with 8.5% of adults in the NHANES having depression symptoms.” | concept_edges[2] |
| outcome | control — held — “We used identical measures of outcome assessment to enable cross-study comparability.” — “Depression symptoms in both studies were assessed using the Patient Health Questionnaire–9, a clinically validated survey with a sensitivity of 88% and a specificity of 88% at a cutoff score of 10 or higher.” | controls[0] |
| outcome | control — partial — “However, these are both population-representative surveys, drawn randomly with probability-based sampling using address-based sampling of US households, mitigating this concern.” — “First, we compare 2 cross-sectional data sources; by comparing NHANES and CLIMB, we are assessing different individuals. We do not survey the same individuals over time, so the NHANES data can provide only a proxy for the true baseline level of depression for CLIMB participants. Differences in prevalence of depression between groups may in part be driven by differences in demographic characteristics, random selection, recruitment, and sampling frame.” | controls[1] |
| outcome | control — failed — “Second, some differences in the NHANES and CLIMB depression estimates may be owing to the differences in baseline mental health across participants; for example, estimates of responses to the assessment of previous diagnosis of depression from 2016 suggest that participants in the AmeriSpeak panel had a 2.3–percentage point higher prevalence of previous diagnosis of depression than the national baseline as measured through the Behavioral Risk Factor Surveillance System (18.8% in AmeriSpeak vs 16.5% in the national baseline).” — “It is impossible to anticipate how the difference in previous depression could, if at all, affect the magnitude of depression found in our study. It is possible that individuals with a known history of depression were more likely to have ongoing depression or have had better access to treatment that would mitigate the current trauma.” | controls[2] |
| outcome | control — held — “Among the CLIMB sample, after controlling for sex, age, race/ethnicity, household size, education, marital status, household income, household savings, and COVID-19 stressor score, participants with lower social and economic resources and with higher COVID-19 stressor scores had higher odds of depression symptoms compared with participants with higher social and economic resources or lower COVID-19 stressor scores” — “Experiencing more COVID-19 stressors was also associated with greater odds of depression symptoms compared with people with low stressor exposure (medium: OR, 1.77 [95% CI, 1.16-2.71]); high: OR, 3.05 [95% CI, 1.95-4.77]).” | controls[3] |
| limitation | THE TWO ARMS ARE TWO DIFFERENT SURVEYS OF DIFFERENT PEOPLE, and the paper says so first: `we are assessing different individuals ... the NHANES data can provide only a PROXY for the true baseline`. The instrument is held constant and the sampling frame is not — “First, we compare 2 cross-sectional data sources; by comparing NHANES and CLIMB, we are assessing different individuals. We do not survey the same individuals over time, so the NHANES data can provide only a proxy for the true baseline level of depression for CLIMB participants. Differences in prevalence of depression between groups may in part be driven by differences in demographic characteristics, random selection, recruitment, and sampling frame.” | limitations[0] |
| limitation | the during-COVID arm is conditioned on things the baseline arm is not: respondents had to SPEAK ENGLISH and to have COMPLETED AN AMERISPEAK SURVEY IN THE PREVIOUS SIX MONTHS, and 94.2% answered over the internet. NHANES imposes none of these — “Adults aged 18 years or older who could speak English and who had completed an AmeriSpeak survey in the past 6 months were eligible to take the survey. In total, 1470 participants completed the survey, representing a survey completion rate of 64.3% of sampled panelists. The survey was conducted mainly over the internet (1385 participants [94.2%]) with a small number conducted over the telephone (85 participants [5.8%]).” | limitations[1] |
| limitation | and the panel was PAID - a cash equivalent of $3 - which NHANES participants are not. The paper reports the payment in the methods and does not carry it into the limitations — “As members of the AmeriSpeak panel, participants are invited to participate in several surveys per month and were paid a cash equivalent of $3 for completing this survey.” | limitations[2] |
| limitation | the panel also came in with MORE PRIOR DEPRESSION than the national baseline, 18.8% against 16.5%. The paper reports the gap, calls it small, and then states plainly that `It is impossible to anticipate how the difference ... could, if at all, affect the magnitude` - an unresolved confound, declared as unresolved — “Second, some differences in the NHANES and CLIMB depression estimates may be owing to the differences in baseline mental health across participants; for example, estimates of responses to the assessment of previous diagnosis of depression from 2016 suggest that participants in the AmeriSpeak panel had a 2.3–percentage point higher prevalence of previous diagnosis of depression than the national baseline as measured through the Behavioral Risk Factor Surveillance System (18.8% in AmeriSpeak vs 16.5% in the national baseline).” | limitations[3] |
| limitation | the comparability argument is an ARGUMENT, not a demonstration: both frames are probability-based and address-based, therefore `mitigating this concern`. No empirical check that the two frames yield the same PHQ-9 distribution on a common population is offered, and none could be from these data — “However, these are both population-representative surveys, drawn randomly with probability-based sampling using address-based sampling of US households, mitigating this concern.” | limitations[4] |
| limitation | the instrument is a SCREENER and the paper is the one in this group that says so: `a diagnosis of depression is ultimately a diagnosis made by a clinician`. 27.8% is 27.8% scoring at or above 10 on the PHQ-9 — “Fourth, we are using a well-established and validated depression screener. While this provides probably the best possible assessment of population-level burden of depression, a diagnosis of depression is ultimately a diagnosis made by a clinician.” | limitations[5] |
finding · quantity · variable · definition · key-term — grounded in the full text| finding | — “This survey study found that prevalence of depression symptoms in the US increased more than 3-fold during the COVID-19 pandemic, from 8.5% before COVID-19 to 27.8% during COVID-19.” | content_tags.findings[0] |
| finding | — “Overall, prevalence was 1.5-fold higher for mild depression symptoms, 2.6-fold higher for moderate depression symptoms, 3.7-fold higher for moderately severe depression symptoms, and 7.5-fold higher for severe depression symptoms categories during COVID-19 compared with before COVID-19.” | content_tags.findings[1] |
| finding | — “Our estimates of pre–COVID-19 depressive symptoms are consistent with other population-level nationally representative studies that captured different constructs of depression; in 2018, 4.7% of adults aged 18 years or older surveyed in the National Survey on Drug Use and Health had major depressive disorder with severe impairment, compared with 8.5% of adults in the NHANES having depression symptoms.” | content_tags.findings[2] |
| finding | — “A 2020 study by Ni et al 8 analyzed depression symptoms before and after political unrest in Hong Kong using the same measure of depression symptoms we deployed in this study. They reported national depression symptoms prevalence before the unrest to be 6.5% (compared with 8.5% in our pre–COVID-19 US sample) and 11.2% in 2019 during unrest (compared with 27.8% in our during–COVID-19 sample).” | content_tags.findings[3] |
| finding | — “Experiencing more COVID-19 stressors was also associated with greater odds of depression symptoms compared with people with low stressor exposure (medium: OR, 1.77 [95% CI, 1.16-2.71]); high: OR, 3.05 [95% CI, 1.95-4.77]).” | content_tags.findings[4] |
| finding | — “Compared with individuals with an annual household income of $75 000 or more, those with a household income of $19 999 or less had 2.4-fold increased odds of depression symptoms (OR, 2.37 [95% CI, 1.26-4.43]). Individuals with household savings less than $5000 had 1.5-fold increased odds of depression symptoms (OR, 1.52 [95% CI, 1.02-2.26]).” | content_tags.findings[5] |
| finding | — “Among the CLIMB sample, after controlling for sex, age, race/ethnicity, household size, education, marital status, household income, household savings, and COVID-19 stressor score, participants with lower social and economic resources and with higher COVID-19 stressor scores had higher odds of depression symptoms compared with participants with higher social and economic resources or lower COVID-19 stressor scores” | content_tags.findings[6] |
| finding | — “Adults aged 18 years or older who could speak English and who had completed an AmeriSpeak survey in the past 6 months were eligible to take the survey. In total, 1470 participants completed the survey, representing a survey completion rate of 64.3% of sampled panelists. The survey was conducted mainly over the internet (1385 participants [94.2%]) with a small number conducted over the telephone (85 participants [5.8%]).” | content_tags.findings[7] |
| finding | — “Second, some differences in the NHANES and CLIMB depression estimates may be owing to the differences in baseline mental health across participants; for example, estimates of responses to the assessment of previous diagnosis of depression from 2016 suggest that participants in the AmeriSpeak panel had a 2.3–percentage point higher prevalence of previous diagnosis of depression than the national baseline as measured through the Behavioral Risk Factor Surveillance System (18.8% in AmeriSpeak vs 16.5% in the national baseline).” | content_tags.findings[8] |
| quantity | the PHQ-9 score at or above which a respondent is counted as having depression symptoms - VALIDATED, with 88% sensitivity and 88% specificity reported at it — “Depression symptoms in both studies were assessed using the Patient Health Questionnaire–9, a clinically validated survey with a sensitivity of 88% and a specificity of 88% at a cutoff score of 10 or higher.” | content_tags.quantities[0] |
| quantity | percent with depression symptoms during COVID-19, against 8.5% before — “This survey study found that prevalence of depression symptoms in the US increased more than 3-fold during the COVID-19 pandemic, from 8.5% before COVID-19 to 27.8% during COVID-19.” | content_tags.quantities[1] |
| quantity | fold increase in SEVERE depression symptoms - the largest of the four severity ratios and five times the mild one — “Overall, prevalence was 1.5-fold higher for mild depression symptoms, 2.6-fold higher for moderate depression symptoms, 3.7-fold higher for moderately severe depression symptoms, and 7.5-fold higher for severe depression symptoms categories during COVID-19 compared with before COVID-19.” | content_tags.quantities[2] |
| quantity | odds ratio for depression symptoms at high stressor exposure against low, 95% CI 1.95-4.77 — “Experiencing more COVID-19 stressors was also associated with greater odds of depression symptoms compared with people with low stressor exposure (medium: OR, 1.77 [95% CI, 1.16-2.71]); high: OR, 3.05 [95% CI, 1.95-4.77]).” | content_tags.quantities[3] |
| quantity | percent of US adults with major depressive disorder with severe impairment in 2018 - set beside this paper's own 8.5%, and called CONSISTENT because the construct differs — “Our estimates of pre–COVID-19 depressive symptoms are consistent with other population-level nationally representative studies that captured different constructs of depression; in 2018, 4.7% of adults aged 18 years or older surveyed in the National Survey on Drug Use and Health had major depressive disorder with severe impairment, compared with 8.5% of adults in the NHANES having depression symptoms.” | content_tags.quantities[4] |
| variable | household savings — “Compared with individuals with an annual household income of $75 000 or more, those with a household income of $19 999 or less had 2.4-fold increased odds of depression symptoms (OR, 2.37 [95% CI, 1.26-4.43]). Individuals with household savings less than $5000 had 1.5-fold increased odds of depression symptoms (OR, 1.52 [95% CI, 1.02-2.26]).” | content_tags.variables[0] |
| variable | income — “Compared with individuals with an annual household income of $75 000 or more, those with a household income of $19 999 or less had 2.4-fold increased odds of depression symptoms (OR, 2.37 [95% CI, 1.26-4.43]). Individuals with household savings less than $5000 had 1.5-fold increased odds of depression symptoms (OR, 1.52 [95% CI, 1.02-2.26]).” | content_tags.variables[1] |
| definition | depression symptoms — “Depression symptoms in both studies were assessed using the Patient Health Questionnaire–9, a clinically validated survey with a sensitivity of 88% and a specificity of 88% at a cutoff score of 10 or higher.” | content_tags.definitions[0] |
| definition | COVID-19 stressor score — “We assessed 13 stressors based on prior studies conducted after traumatic events. 20 , 21 Examples of COVID-19 stressors included losing a job, death of someone close to you owing to COVID-19, and having financial problems. We excluded stressors that were capturing constructs applicable only to specific groups and created a score ranging from 0 to 13 to measure cumulative exposure to COVID-19 stressors. We divided the scores into thirds to measure low, medium, and high exposure to COVID-19-induced stressors. Cutoffs for stressor score categories were low (score, 0-2), medium (score, 3-4), and high (score, 5-13).” | content_tags.definitions[1] |
| concept | Patient Health Questionnaire–9 — not defined in this paper's own words; the depression screener used in BOTH arms, binarised at a score of 10 or higher; validated before this study, with sensitivity and specificity both reported at 88% at exactly that cut-off — “Depression symptoms in both studies were assessed using the Patient Health Questionnaire–9, a clinically validated survey with a sensitivity of 88% and a specificity of 88% at a cutoff score of 10 or higher.” | concepts[0] |
| concept | sensitivity — not defined in this paper's own words; the validation property the cut-off carries, reported at 88% - the thing qiu2020's Cronbach alpha is not — “Depression symptoms in both studies were assessed using the Patient Health Questionnaire–9, a clinically validated survey with a sensitivity of 88% and a specificity of 88% at a cutoff score of 10 or higher.” | concepts[1] |
| concept | specificity — not defined in this paper's own words; reported at 88% at the same cut-off, jointly fixing the operating point — “Depression symptoms in both studies were assessed using the Patient Health Questionnaire–9, a clinically validated survey with a sensitivity of 88% and a specificity of 88% at a cutoff score of 10 or higher.” | concepts[2] |
| concept | depression symptoms — not defined in this paper's own words; the outcome as named throughout - deliberately not 'depression', and the paper maintains the distinction to its last limitation — “This survey study found that prevalence of depression symptoms in the US increased more than 3-fold during the COVID-19 pandemic, from 8.5% before COVID-19 to 27.8% during COVID-19.” | concepts[3] |
| concept | major depressive disorder — not defined in this paper's own words; the clinical category the paper explicitly does NOT measure, invoked to explain why a national estimate of 4.7% and its own 8.5% are both right — “Our estimates of pre–COVID-19 depressive symptoms are consistent with other population-level nationally representative studies that captured different constructs of depression; in 2018, 4.7% of adults aged 18 years or older surveyed in the National Survey on Drug Use and Health had major depressive disorder with severe impairment, compared with 8.5% of adults in the NHANES having depression symptoms.” | concepts[4] |
| concept | COVID-19 stressors — not defined in this paper's own words; the exposure the paper counts - realised events such as losing a job, a death from COVID-19, financial problems - taken from prior post-disaster instruments — “We assessed 13 stressors based on prior studies conducted after traumatic events. 20 , 21 Examples of COVID-19 stressors included losing a job, death of someone close to you owing to COVID-19, and having financial problems. We excluded stressors that were capturing constructs applicable only to specific groups and created a score ranging from 0 to 13 to measure cumulative exposure to COVID-19 stressors. We divided the scores into thirds to measure low, medium, and high exposure to COVID-19-induced stressors. Cutoffs for stressor score categories were low (score, 0-2), medium (score, 3-4), and high (score, 5-13).” | concepts[5] |
| concept | stressor score — not defined in this paper's own words; the 0-13 summation of those 13 stressors, cut at 2 and 4 - WHICH ARE NOT THE THIRDS OF 0-13. The boundaries are fitted to the responses, so 'high' is a rank in this survey rather than a level of exposure — “We assessed 13 stressors based on prior studies conducted after traumatic events. 20 , 21 Examples of COVID-19 stressors included losing a job, death of someone close to you owing to COVID-19, and having financial problems. We excluded stressors that were capturing constructs applicable only to specific groups and created a score ranging from 0 to 13 to measure cumulative exposure to COVID-19 stressors. We divided the scores into thirds to measure low, medium, and high exposure to COVID-19-induced stressors. Cutoffs for stressor score categories were low (score, 0-2), medium (score, 3-4), and high (score, 5-13).” | concepts[6] |
| concept | screener — not defined in this paper's own words; what the paper calls its own instrument in the limitation that separates it from a clinician's diagnosis — “Fourth, we are using a well-established and validated depression screener. While this provides probably the best possible assessment of population-level burden of depression, a diagnosis of depression is ultimately a diagnosis made by a clinician.” | concepts[7] |
| OR | — “Experiencing more COVID-19 stressors was also associated with greater odds of depression symptoms compared with people with low stressor exposure (medium: OR, 1.77 [95% CI, 1.16-2.71]); high: OR, 3.05 [95% CI, 1.95-4.77]).” | content_tags.statistics[0] |
| 95% CI | — “Experiencing more COVID-19 stressors was also associated with greater odds of depression symptoms compared with people with low stressor exposure (medium: OR, 1.77 [95% CI, 1.16-2.71]); high: OR, 3.05 [95% CI, 1.95-4.77]).” | content_tags.statistics[1] |
| sensitivity | — “Depression symptoms in both studies were assessed using the Patient Health Questionnaire–9, a clinically validated survey with a sensitivity of 88% and a specificity of 88% at a cutoff score of 10 or higher.” | content_tags.statistics[2] |
| specificity | — “Depression symptoms in both studies were assessed using the Patient Health Questionnaire–9, a clinically validated survey with a sensitivity of 88% and a specificity of 88% at a cutoff score of 10 or higher.” | content_tags.statistics[3] |
| claim type | existence — a second reading reaching the same value; recorded: existence — “In this survey study that included 1441 respondents from during the COVID-19 pandemic and 5065 respondents from before the pandemic, depression symptom prevalence was more than 3-fold higher during the COVID-19 pandemic than before.” | alternatives.claim_type |
| method | survey-study — a second reading DIFFERING from the recorded value; recorded: survey — “This survey study examines the prevalence of depression symptoms among US adults during the coronavirus disease 2019 (COVID-19) pandemic compared with before the pandemic.” | alternatives.method |
| phenomenon | the burden of depression symptoms among US adults during the COVID-19 pandemic compared with before it — a second reading DIFFERING from the recorded value; recorded: a population-scale rise in depression symptoms concentrated in the lowest-resource groups, rising monotonically with cou — “What is the burden of depression symptoms among US adults during the coronavirus disease 2019 (COVID-19) pandemic compared with before COVID-19, and what are the risk factors associated with depression symptoms? Findings In this survey study that included 1441 respondents from during the COVID-19 pandemic and 5065 respondents from before the pandemic, depression symptom prevalence was more than 3-fold higher during the COVID-19 pandemic than before.” | alternatives.phenomenon |
Full text — tagged
The reference list is not part of the text we hold for this paper. Our copy was extracted from the publisher's page and the bibliography did not survive that extraction, so the article below ends before its references. Nothing has been removed by us.
Abstract. Key Points Question What is the burden of depression symptomsterm among US adults during the coronavirus disease 2019 (COVID-19) pandemic compared with before COVID-19, and what are the risk factors associated with depression symptomsterm? Findings In this surveyterm·mth study that included 1441 respondents from during the COVID-19 pandemic and 5065 respondents from before the pandemic, depression symptomterm prevalence was more than 3-fold higher during the COVID-19 pandemic than before. Lower incometerm·var, having less than $5000 in savings, and having exposure to more stressors were associated with greater risk of depression symptomsterm during COVID-19. Meaning These findings suggest that there is a high burden of depression symptomsterm in the US associated with the COVID-19 pandemic and that this burden falls disproportionately on individuals who are already at increased risk. This surveyterm·mth study examines the prevalence of depression symptomsterm among US adults during the coronavirus disease 2019 (COVID-19) pandemic compared with before the pandemic. Importance The coronavirus disease 2019 (COVID-19) pandemic and the policies to contain it have been a near ubiquitous exposure in the US with unknown effects on depression symptomsterm. Objective To estimate the prevalence of and risk factors associated with depression symptomsterm among US adults during vs before the COVID-19 pandemic. Design, Setting, and Participants This nationally representative surveyterm·mth study used 2 population-based surveysterm·mth of US adults aged 18 or older. During COVID-19, estimates were derived from the COVID-19 and Life Stressors Impact on Mental Health and Well-being study, conducted from March 31, 2020, to April 13, 2020. Before COVID-19 estimates were derived from the National Health and Nutrition Examination Surveyterm·mth, conducted from 2017 to 2018. Data were analyzed from April 15 to 20, 2020. Exposures The COVID-19 pandemic and outcomes associated with the measures to mitigate it. Main Outcomes and Measures Depression symptomsterm, defined using the Patient Health Questionnaire-9term cutoff of 10 or higher. Categories of depression symptomsterm were defined as none (score, 0-4), mild (score, 5-9), moderate (score, 10-14), moderately severe (score, 15-19), and severe (score, ≥20). Results A total of 1470 participants completed the COVID-19 and Life Stressors Impact on Mental Health and Well-being surveyterm·mth (completion rate, 64.3%), and after removing those with missing data, the final during–COVID-19 sample included 1441 participants (619 participants [43.0%] aged 18-39 years; 723 [50.2%] men; 933 [64.7%] non-Hispanic White). The pre–COVID-19 sample included 5065 participants (1704 participants [37.8%] aged 18-39 years; 2588 [51.4%] women; 1790 [62.9%] non-Hispanic White). Depression symptomterm prevalence was higher in every category during COVID-19 compared with before (mild: 24.6% [95% CI, 21.8%-27.7%] vs 16.2% [95% CI, 15.1%-17.4%]; moderate: 14.8% [95% CI, 12.6%-17.4%] vs 5.7% [95% CI, 4.8%-6.9%]; moderately severe: 7.9% [95% CI, 6.3%-9.8%] vs 2.1% [95% CI, 1.6%-2.8%]; severe: 5.1% [95% CI, 3.8%-6.9%] vs 0.7% [95% CI, 0.5%-0.9%]). Higher risk of depression symptomsterm during COVID-19 was associated with having lower incometerm·var (odds ratio, 2.37 [95% CI, 1.26-4.43]), having less than $5000 in savings (odds ratio, 1.52 [95% CI, 1.02-2.26]), and exposure to more stressors (odds ratio, 3.05 [95% CI, 1.95-4.77]). Conclusions and Relevance These findings suggest that prevalence of depression symptomsterm in the US was more than 3-fold higher during COVID-19 compared with before the COVID-19 pandemic. Individuals with lower social resources, lower economic resources, and greater exposure to stressors (eg, job loss) reported a greater burden of depression symptomsterm. Post–COVID-19 plans should account for the probable increase in mental illness to come, particularly among at-risk populations. Introduction Coronavirus disease 2019 (COVID-19) and the policies to contain it have been a near ubiquitous exposure for people in the US in 2020. As an event that can cause physical, emotional, and psychological harm, the COVID-19 pandemic can itself be considered a traumatic event. In addition, the policies created to prevent its spread introduced new life stressors and disrupted daily living for most people in the US. As of April 13, 2020, 42 states were under stay-at-home advisories or shelter-in-place policies, affecting at least 316 million people in the US, or approximately 96% of the population. The unemployment rate was reaching record highs in the US, with more than 20 million people filing for unemployment between the start of COVID-19 and mid-April 2020. Mental health is sensitive to traumatic events and their social and economic consequences. Previous studies on disruptions to life owing to disasters, epidemics, or civil unrest suggest that exposure to large-scale traumatic events are associated with increased burden of mental illness in the populations affected. For example, after September 11, 2001, 9.6% of Manhattan residents reported symptoms consistent with depression and 7.5% reported symptoms consistent with posttraumatic stress disorder. Residents living closer to the World Trade Centers had higher prevalence of mental illness. 4 , 5 Similarly, after natural disasters, populations affected by hurricanes report an increase in symptoms consistent with mental illness. Increases in mental illness have also been documented after other epidemics, such as the Ebola virus and SARS outbreak. 6 , 7 In addition, social disruptions in day-to-day living after civil unrest, for example, have been found to be associated with mental illness. Data from Hong Kong show greater levels of depression, anxiety, and psychological distress during the 2019 Hong Kong civil protests. 8 , 9 Importantly, the mental health consequences of mass traumatic events are not evenly distributed across populations. Having lower incometerm·var and less wealth are associated with greater burden of mental illness. Mental illness has been well documented in the wake of previous financial recessions, particularly among individuals who are unemployed and are otherwise affected by social and economic adversity. 11 , 12 , 13 , 14 Early evidence from published studies suggests that COVID-19 is associated with mental illness. Among health care workers in China who were exposed to patients with COVID-19, 50.4% reported symptoms of depression. 15 A study of medical students in China identified elevated prevalence of anxiety. Another study found that reduced sleep was associated with greater levels of anxiety and stress among health care workers in China. To date, most studies regarding mental health and COVID-19 have been conducted in Asia and have focused on specific subpopulations, such as college students 16 and medical workers. 15 , 17 Published studies from the US on mental health have been in purposive samples. Most relevant to this study, a study by Nelson et al 18 analyzed concerns about COVID-19, symptoms, and responses to the pandemic across 9009 completed surveysterm·mth distributed over social media. They found that 67.3% of participants were very or extremely concerned about COVID-19 and that 48.8% of participants were self-isolating most of the time to avoid COVID-19. To our knowledge, the mental health of the broader US population during COVID-19 has not been documented. Aiming to address this gap in understanding, we assessed the burden of depression symptomsterm in the US during COVID-19 using the same measures deployed in representative national surveysterm·mth before COVID-19 began. We also aimed to understand the factors associated with depression symptomsterm during and before COVID-19. Methods This study was approved by the institutional review boards of NORC at the University of Chicago and Boston University. All AmeriSpeak participants provided written informed consent during the enrollment process to join the AmeriSpeak standing panel. All NHANES participants provided written informed consent first for the household interview and then for the health examination. This study followed the American Association for Public Opinion Research ( AAPOR ) reporting guideline. Population Sample During the COVID-19 Pandemic The primary sample for this study was a nationally representative group of US adults aged 18 years or older using the AmeriSpeak standing panel. The AmeriSpeak panel is a probability-based panel that is representative of the US population by design. Households are randomly selected with a known, nonzero probability from the NORC National Frame, covering approximately 97% of US households. People excluded from the sampling frame were people with PO box–only addresses, some addresses not listed in the US Postal Service Delivery Sequence File, and some newly constructed homes. Adults aged 18 years or older who could speak English and who had completed an AmeriSpeak surveyterm·mth in the past 6 months were eligible to take the surveyterm·mth. In total, 1470 participants completed the surveyterm·mth, representing a surveyterm·mth completion rate of 64.3% of sampled panelists. The surveyterm·mth was conducted mainly over the internet (1385 participants [94.2%]) with a small number conducted over the telephone (85 participants [5.8%]). Twenty-nine participants missing data for depression were excluded from the sample. The final COVID-19 study sample included 1441 participants. The surveyterm·mth was distributed by NORC at the University of Chicago from March 31, 2020, through April 13, 2020, assessing COVID-19 exposure, stressors, and mental health using the COVID-19 and Life Stressors Impact on Mental Health and Well-being (CLIMB) study questionnaire. Participants were contacted via web surveyterm·mth and follow-up was conducted via telephone interview. Several demographic questions (eg, sex, age, self-reported race/ethnicity, educational status, and marital status) were previously assessed for all AmeriSpeak panel members. As members of the AmeriSpeak panel, participants are invited to participate in several surveysterm·mth per month and were paid a cash equivalent of $3 for completing this surveyterm·mth. Comparison Sample Before the COVID-19 Pandemic The comparison sample for this study, measuring mental health before COVID-19, was the National Health and Nutrition Examination Surveyterm·mth (NHANES), a nationally representative group of noninstitutionalized civilian US adults aged 18 years or older. The NHANES is an annual cross-sectional surveyterm·mth conducted by the US government. Participants are selected through a 4-stage probability sampling design, selecting primary sampling units by the county-level, then by census block–level, and then by households in the 50 states and the District of Columbia. The AmeriSpeak panel’s sampling frame included 97% of US households, in all 50 states and the District of Columbia, and used a 2-stage probability sample design, first at the county-level and then at the census block–level. Thus, in this way, the NHANES sample is a suitable comparison group for the nationally representative AmeriSpeak panel. Collection of NHANES is administered through household interviews and interviews in a mobile examination center, measuring physical and mental health with a Computer Assisted Personal Interview for the depression screenerterm, similar to an online surveyterm·mth questionnaire. There were 9254 participants in the NHANES 2017 to 2018 cycle. The NHANES sample used in this study excluded 3398 individuals (36.7%) younger than 18 years and 791 individuals (8.5%) missing depression data. The final NHANES sample included 5065 participants. Key Definitions Depression Symptomsterm Depression symptomsterm in both studies were assessed using the Patient Health Questionnaire–9term, a clinically validated surveyterm·mth with a sensitivityterm of 88% and a specificityterm of 88% at a cutoff score of 10 or higher. Depression symptomterm categories were defined as none (score, 0-4), mild (score, 5-9), moderate (score, 10-14), moderately severe (score, 15-19), and severe (score, ≥20). Binary classification of depression symptomsterm was defined by a score of 10 or greater. COVID-19 Stressorsterm Score We assessed 13 stressors based on prior studies conducted after traumatic events. 20 , 21 Examples of COVID-19 stressorsterm included losing a job, death of someone close to you owing to COVID-19, and having financial problems. We excluded stressors that were capturing constructs applicable only to specific groups and created a score ranging from 0 to 13 to measure cumulative exposure to COVID-19 stressorsterm. We divided the scores into thirds to measure low, medium, and high exposure to COVID-19-induced stressors. Cutoffs for stressor scoreterm categories were low (score, 0-2), medium (score, 3-4), and high (score, 5-13). Demographic Characteristics Sex was defined as a binary variable for men or women. Age was defined as a categorical variable with 3 groups: 18 to 39 years, 40 to 59 years, or 60 years or older. Race/ethnicity was defined as a categorical variable across 5 mutually exclusive categories: non-Hispanic White, non-Hispanic Black, Hispanic, non-Hispanic Asian, and other race/ethnicity (including multiple races/ethnicities). Education was defined as a categorical variable with 4 groups: less than high school graduate, high school graduate or general education diploma equivalent, some college, and college graduate or higher. Marital status was defined as a categorical variable with 4 groups: married; widowed, divorced, or separated; never married; and living with partner. Household incometerm·var was defined as a categorical variable with 4 groups, divided approximately at the interquartile range: $0 to $19 999, $20 000 to $44 999, $45 000 to $74 999, and $75 000 or more. Household savingsterm·var was defined as a binary variable with savings of at least $5000. Savings was defined as “money in all types of accounts, including cash, savings, or checking accounts, stocks, bonds, mutual funds, retirement funds (such as pensions, IRAs, 401Ks, etc), and certificates of deposit.” Household size was defined as the number of people living in a home with categories from 1 to 7 or more to protect participant identity. Statistical Analysis First, we calculated the demographic characteristics of the NHANES and CLIMB samples. Mobile examination center surveyterm·mth weights were used for NHANES data, and probability surveyterm·mth weights were used for CLIMB data. Second, we conducted bivariable χ 2 analysis to assess the association between demographic characteristics and depression symptomsterm in CLIMB and NHANES samples. Third, we estimated the prevalence and 95% CIs of depression symptomsterm in the US across categories before and during COVID-19 using CLIMB and NHANES samples. Fourth, we calculated the difference and ratio between estimates of depression symptomsterm during and before COVID-19. Fifth, we estimated the distribution of depression symptomterm categories before and during COVID-19 in the US using CLIMB and NHANES samples. Sixth, we used multivariable logistic regression to estimate odds ratios (ORs) and 95% CIs for the association between COVID-19–induced life stressors and depression symptomsterm, controlling for demographic characteristics and resources; the model controlled for sex, age, race/ethnicity, household size, education, marital status, household incometerm·var, household savingsterm·var, and COVID-19 stressor scoreterm·def. We used complete case analysis for the multivariable analysis. Stata statistical software version 16.1 (StataCorp) was used for statistical analyses. P values were 2-sided, and statistical significance was set at P = .05. Data were analyzed from April 15 to 20, 2020. Results A total of 1470 participants completed the CLIMB surveyterm·mth for a completion rate of 64.3%. Of these, 1441 participants were included in the final sample. A total of 619 participants (43.0% unweighted, 38.0% weighted) in the total sample were aged 18 to 39 years, 723 (50.2% unweighted, 48.1% weighted) were men, and 933 (64.7% unweighted, 62.9% weighted) were non-Hispanic White. The pre–COVID-19 NHANES sample included 5065 participants (1704 participants [37.8%] aged 18-39 years; 2588 [51.4%] women; 1790 [62.9%] non-Hispanic White). Table 1 presents demographic characteristics of the NHANES and CLIMB study participants, prevalence of depression symptomsterm for each sample weighted to the US population, and distribution of depression symptomsterm by demographic groups. Our sample was representative of the US population and similar in distribution of demographic characteristics to that of the NHANES sample. The CLIMB sample had a lower incometerm·var than the NHANES sample ( Table 1 ). Table 1. Demographic Characteristics of Nationally Representative Samples of US Adults Before and After the COVID-19 Pandemic and Association With Depression Symptomsterm Characteristic Before COVID-19 a During COVID-19 b No. (%) c P value e No. (%) c P value e Total (n = 5065) Depression symptomsterm (n = 458) d Total (n = 1441) Depression symptomsterm (n = 382) d Sex Men 2477 (48.6) 181 (6.9) .02 723 (48.1) 149 (21.9) <.001 Women 2588 (51.4) 277 (10.1) 718 (51.9) 233 (33.3) Age, y 18-39 1704 (37.8) 149 (9.0) .62 619 (38.0) 219 (38.8) <.001 40-59 1542 (34.2) 144 (8.5) 462 (32.4) 113 (26.8) ≥60 1819 (28.0) 165 (7.9) 360 (29.7) 50 (14.9) Race/ethnicity Non-Hispanic White 1790 (62.9) 186 (8.4) .01 933 (62.9) 225 (26.5) .32 Non-Hispanic Black 1176 (11.1) 97 (8.4) 143 (11.9) 36 (24.2) Hispanic 1155 (15.8) 109 (8.4) 255 (16.6) 84 (34.0) Non-Hispanic Asian 674 (5.3) 26 (4.4) 36 (3.1) 8 (23.1) Other or multiple 270 (4.8) 40 (16.0) 74 (5.6) 29 (34.4) Education <High school 958 (11.0) 117 (13.3) .004 65 (9.8) 22 (29.2) <.001 High school or GED 1292 (28.4) 129 (9.3) 274 (27.9) 85 (35.0) Some college 1636 (30.5) 158 (9.8) 637 (27.8) 186 (32.0) ≥College 1173 (30.1) 52 (4.7) 465 (34.5) 89 (18.3) Marital status Married 2404 (53.4) 146 (5.5) <.001 712 (47.8) 124 (18.3) <.001 Widowed, divorced, or separated 1093 (18.5) 156 (13.9) 247 (18.4) 75 (31.5) Never married 870 (18.7) 91 (10.5) 344 (24.2) 130 (39.8) Living with partner 450 (9.4) 42 (10.7) 138 (9.7) 53 (37.7) Household incometerm·var, $ ≤19 999 868 (12.9) 130 (16.8) <.001 246 (19.8) 116 (46.9) <.001 20 000-44 999 1319 (24.0) 133 (10.1) 357 (25.8) 109 (31.1) 45 000-74 999 887 (19.8) 61 (8.3) 357 (25.1) 83 (23.3) ≥75 000 1354 (43.4) 68 (4.8) 447 (29.3) 67 (16.9) Household savingsterm·var, $ ≤4999 NA NA NA 577 (43.2) 227 (40.4) <.001 ≥5000 NA NA 819 (56.8) 146 (19.3) COVID-19 stressor scoreterm·def f Low NA NA NA 450 (30.7) 64 (15.5) <.001 Medium NA NA 545 (37.5) 132 (25.1) High NA NA 443 (31.9) 185 (42.9) Household size, mean (SD) 3.1 (0.6) NA NA 3.2 (0.6) NA NA Abbreviations: COVID-19, coronavirus disease 2019; GED, general education diploma; NA, not available. a Before COVID-19 estimates derived from the National Health and Nutrition Examination Surveyterm·mth (NHANES) from 2017 to 2018. Missing data in the NHANES sample included 637 participants with missing household incometerm·var, 248 participants with missing marital status, and 6 participants with missing education level. b During COVID-19 estimates derived from the COVID-19 and Life Stressors Impact on Mental Health and Well-being (CLIMB) study collected from March 31 to April 13, 2020. Missing data in the CLIMB study sample included 34 participants with missing household incometerm·var, 45 participants with missing household savingsterm·var, and 3 pariticiapnts with missing COVID-19 stressor scoreterm·def. c Frequencies are unweighted. Percentages are weighted. Categories may not add up to total number owing to missing data. Percentages may not add up to 100 owing to rounding. d Defined as Patient Health Questionnaire–9term score of 10 or greater. e Two-tailed χ 2 analysis conducted for significance testing. f Calculated from stressor summation (range, 0-13); stratified as low (score, 0-2), medium (score, 3-4), and high (score, 5-13). A total of 458 participants (8.5%) had depression symptomsterm before COVID-19, and 382 participants (27.8%) had depression symptomsterm during COVID-19. Higher levels of depression symptomsterm were observed in all demographic groups during COVID-19 compared with before, with more than 3-fold higher prevalence of depression symptomsterm in general. The distribution of depression symptomsterm within demographic categories was consistent with patterns observed before COVID-19. For example, before and during COVID-19, women were more likely to have depression symptomsterm than men (before: 277 women [10.1%] vs 181 men [6.9%]; during: 233 women [33.3%] vs 149 men [21.9%]). For race/ethnicity, non-Hispanic Asian individuals saw an 18.7–percentage point higher prevalence of depression symptomsterm during COVID-19 compared with before COVID-19 (8 participants [23.1%] vs 26 participants [4.4%]). In general, having more resources was associated with a lower prevalence of depression symptomsterm, both before and during COVID-19. During COVID-19, married individuals had a lower rate of depression symptomsterm (124 participants with depression [18.3%]), compared with those who were widowed, divorced, or separated (75 participants with depression [31.5%]), never married (130 participants with depression [39.8%]), or living with a partner (53 participants with depression [37.7%]). Similarly, 116 individuals in the lowest incometerm·var category (46.9%) had depression symptomsterm, while 67 individuals in the highest incometerm·var category (16.9%) had depression symptomsterm. Individuals with household savingsterm·var of $5000 or more were less likely to have depression symptomsterm (146 participants [19.3%]) than individuals with savings less than $5000 (227 participants [40.4%]). Having experienced more COVID-19 stressorsterm was associated with greater burden of depression symptomsterm (low: 64 participants [15.5%]; medium: 132 participants [25.1%]; high: 185 participants [42.9%]). Prevalence of depression symptomsterm in the US was higher in every category during COVID-19 than before COVID-19 ( Table 2 ). During COVID-19, most US residents (52.5% [95% CI, 49.1%-55.8%]) had symptoms of mild depression or greater; before COVID-19, 24.7% (95% CI, 22.9%-26.6%) of US residents had symptoms of mild depression or greater. Mild depression symptomsterm prevalence was 16.2% (95% CI, 15.1%-17.4%) before COVID-19 compared with 24.6% (95% CI, 21.8%-27.7%) during COVID-19; moderate depression symptomsterm prevalence was 5.7% (95% CI, 4.8%-6.9%) before COVID-19 compared with 14.8% (95% CI, 12.6%-7.4%) during COVID-19; moderately severe depression symptomterm prevalence was 2.1% (95% CI, 1.6%-2.8%) before COVID-19 compared with 7.9% (95% CI, 6.3%-9.8%) during COVID-19; and severe depression prevalence was 0.7% (95% CI, 0.5%-0.9%) before COVID-19 compared with 5.1% (95% CI, 3.8%-6.9%) during COVID-19. Overall, prevalence was 1.5-fold higher for mild depression symptomsterm, 2.6-fold higher for moderate depression symptomsterm, 3.7-fold higher for moderately severe depression symptomsterm, and 7.5-fold higher for severe depression symptomsterm categories during COVID-19 compared with before COVID-19. Table 2. Prevalence of Depression Symptomsterm in US Adults Before and During the COVID-19 Pandemic Depression symptomsterm a % (95% CI) b Difference Before COVID-19 c During COVID-19 d Absolute, % Relative None 75.3 (73.3-77.1) 47.5 (44.2-50.9) −27.7 0.6 Mild 16.2 (15.1-17.4) 24.6 (21.8-27.7) 8.4 1.5 Moderate 5.7 (4.8-6.9) 14.8 (12.6-17.4) 9.1 2.6 Moderately severe 2.1 (1.6-2.8) 7.9 (6.3-9.8) 5.7 3.7 Severe 0.7 (0.5-0.9) 5.1 (3.8-6.9) 4.4 7.5 Abbrevation: COVID-19, coronavirus disease 2019. a Depression symptomsterm assessed using the Patient Health Questionnaire–9term and categorized as none (score, 0-4), mild (score, 5-9), moderate (score, 10-14), moderately severe (score, 15-19), and severe (score, ≥20). b Percentages weighted to the noninstitutionalized US adult population. c Before COVID-19 estimates from the National Health and Nutrition Examination Surveyterm·mth from 2017 to 2018. d During COVID-19 estimates from the COVID-19 and Life Stressors Impact on Mental Health and Well-being study collected from March 31, 2020, to April 13, 2020. The Figure presents depression symptomterm scores from the Patient Health Questionnaire–9term grouped by category before and during COVID-19. There was a greater distribution of scores between 0 and 4 before COVID-19; the CLIMB sample had a greater distribution of depression symptomsterm than the NHANES sample for all scores greater than 5. Thus, we saw a rightward shift in symptom burden in the sample during COVID-19 compared with the sample before COVID-19. Figure. Depression Symptomsterm in US Adults Before and During the Coronavirus Disease 2019 (COVID-19) Pandemic Before COVID-19 estimates from the National Health and Nutrition Examination Surveyterm·mth (NHANES) from 2017-2018. During COVID-19 estimates from the COVID-19 and Life Stressors Impact on Mental Health and Well-being (CLIMB) study collected from March 31 to April 13, 2020. Depression symptomsterm categories calculated using the Patient Health Questionnaire–9term: none (0-4), mild (5-9), moderate (10-14), moderately severe (15-19), and severe (≥20). Percentages weighted to the population of noninstitutionalized US adults aged 18 years or older. Among the CLIMB sample, after controlling for sex, age, race/ethnicity, household size, education, marital status, household incometerm·var, household savingsterm·var, and COVID-19 stressor scoreterm·def, participants with lower social and economic resources and with higher COVID-19 stressor scoresterm·def had higher odds of depression symptomsterm compared with participants with higher social and economic resources or lower COVID-19 stressor scoresterm·def ( Table 3 ). Compared with married individuals, individuals who were widowed, divorced, or separated had 2.1-fold increased odds of depression symptomsterm (OR, 2.08 [95% CI, 1.29-3.36]) and individuals who had never married had 1.9-fold increased odds of depression symptomsterm (OR, 1.85 [95% CI, 1.17-2.94]). Compared with individuals with an annual household incometerm·var of $75 000 or more, those with a household incometerm·var of $19 999 or less had 2.4-fold increased odds of depression symptomsterm (OR, 2.37 [95% CI, 1.26-4.43]). Individuals with household savingsterm·var less than $5000 had 1.5-fold increased odds of depression symptomsterm (OR, 1.52 [95% CI, 1.02-2.26]). Experiencing more COVID-19 stressorsterm was also associated with greater odds of depression symptomsterm compared with people with low stressor exposure (medium: OR, 1.77 [95% CI, 1.16-2.71]); high: OR, 3.05 [95% CI, 1.95-4.77]). Table 3. Odds of Depression Symptomsterm by Resources and Exposure to COVID-19 Stressorsterm Resource OR (95% CI) a P value Education <High school 0.86 (0.41-1.77) .68 High school graduate or GED 1.58 (0.96-2.60) .08 Some college 1.45 (0.97-2.16) .07 ≥College 1 [Reference] NA Marital status Married 1 [Reference] NA Widowed, divorced, or separated 2.08 (1.29-3.36) .003 Never married 1.85 (1.17-2.94) .008 Living with partner 1.27 (0.74-2.18) .39 Household incometerm·var, $ ≤19 999 2.37 (1.26-4.43) .007 20 000-44 999 1.32 (0.78-2.24) .31 45 000-74 999 1.04 (0.63-1.72) .88 ≥75 000 1 [Reference] NA Household savingsterm·var, $ ≤4999 1.52 (1.02-2.26) .04 ≥5000 1 [Reference] NA COVID-19 stressorsterm Low 1 [Reference] NA Medium 1.77 (1.16-2.71) .008 High 3.05 (1.95-4.77) <.001 Abbreviations: COVID-19, coronavirus disease 2019; GED, general education diploma; NA, not applicable; OR, odds ratio. a Complete case analysis used for multiple logistic regression resulting included 1386 participants for this model. Model controls for demographic characteristics (ie, sex, age, race/ethnicity, and household size). Depression symptomsterm defined as PHQ-9 score cutoff of 10 or greater. COVID-19 stressor scoreterm·def calculated from stressor summation ranging from 0-13; categories represent low (score, 0-2), medium (score, 3-4), and high (score, 5-13) exposure to stressors due to COVID-19. Data on household incometerm·var were missing for 34; on household savingsterm·var, 45 participants; and COVID-19 stressor scoreterm·def, 3 participants. Discussion This surveyterm·mth study found that prevalence of depression symptomsterm in the US increased more than 3-fold during the COVID-19 pandemic, from 8.5% before COVID-19 to 27.8% during COVID-19. To our knowledge, this is the first nationally representative study that assessed depression symptomsterm using the Patient Health Questionnaire–9term in US adults before and during the COVID-19-pandemic. We found a shift in depression symptomsterm, with fewer people with no symptoms and more people with more symptoms during COVID-19 than before COVID-19. We also found that lower incometerm·var groups were at greater risk of depression symptomsterm than higher incometerm·var groups, and that having less than $5000 in household savingsterm·var was associated with 1.5-fold increased odds of depression symptomsterm, or 50% greater risk. Additionally, we found that people with exposure to more stressors had greater odds of depression symptomsterm. Findings from a 2014 review 3 on trauma and mental health suggest that depression increases during and after traumatic events; our study adds to this literature. A 2020 study by Ni et al 8 analyzed depression symptomsterm before and after political unrest in Hong Kong using the same measure of depression symptomsterm we deployed in this study. They reported national depression symptomsterm prevalence before the unrest to be 6.5% (compared with 8.5% in our pre–COVID-19 US sample) and 11.2% in 2019 during unrest (compared with 27.8% in our during–COVID-19 sample). This suggests that the impact of COVID-19 on the US population may be substantially larger than that after other large-scale events. This may reflect the greater ubiquity of COVID-19 and its effects on the US population than prior recorded large-scale traumatic events. Our findings are consistent with studies in Asia showing a substantial burden of psychological distress following COVID-19. 16 , 17 , 22 In a study of health care workers in Hubei province and surrounding regions, Lai et al 15 found similar levels of depression: 49.6% of participants had no depression (vs 47.5% of participants in the CLIMB sample), while 35.6% of participants had mild depression (vs 24.6% in the CLIMB sample), 8.6% of participants had moderate depression (vs 14.8% in the CLIMB sample), and 6.2% of participants had moderately severe depression (vs 7.9% in the CLIMB sample). The Lai et al 15 sample in China included only health care professionals and was concentrated in the Hubei region, while our sample included a representative sample of all US residents and sampled the whole country. Lai et al 15 sampled earlier in the pandemic, when the global socioeconomic effects of COVID-19 had yet to materialize fully. Our findings are also consistent with a body of literature showing that having fewer assets and more exposure to life stressors are associated with more depression during times of social duress and during low-intensity periods. 10 , 23 , 24 , 25 Our estimates of pre–COVID-19 depressive symptoms are consistent with other population-level nationally representative studies that captured different constructs of depression; in 2018, 4.7% of adults aged 18 years or older surveyed in the National Surveyterm·mth on Drug Use and Health had major depressive disorderterm with severe impairment, compared with 8.5% of adults in the NHANES having depression symptomsterm. While major depressive disorderterm is a more severe measure of depression, based on the results of our study, we may expect to see increases in the presence of major depressive disorderterm in other studies conducted during the COVID-19 pandemic. Limitations This study has some limitations. First, we compare 2 cross-sectional data sources; by comparing NHANES and CLIMB, we are assessing different individuals. We do not surveyterm·mth the same individuals over time, so the NHANES data can provide only a proxy for the true baseline level of depression for CLIMB participants. Differences in prevalence of depression between groups may in part be driven by differences in demographic characteristics, random selection, recruitment, and sampling frame. However, these are both population-representative surveysterm·mth, drawn randomly with probability-based sampling using address-based sampling of US households, mitigating this concern. We used identical measures of outcome assessment to enable cross-study comparability. Second, some differences in the NHANES and CLIMB depression estimates may be owing to the differences in baseline mental health across participants; for example, estimates of responses to the assessment of previous diagnosis of depression from 2016 suggest that participants in the AmeriSpeak panel had a 2.3–percentage point higher prevalence of previous diagnosis of depression than the national baseline as measured through the Behavioral Risk Factor Surveillance System (18.8% in AmeriSpeak vs 16.5% in the national baseline). This suggests that the AmeriSpeak population may have had a slightly higher history of previous depression than the national comparison group; however, the difference is small, and our study examined past 2-week depression symptomsterm. It is impossible to anticipate how the difference in previous depression could, if at all, affect the magnitude of depression found in our study. It is possible that individuals with a known history of depression were more likely to have ongoing depression or have had better access to treatment that would mitigate the current trauma. Third, the COVID-19 pandemic is peaking in different cities at different times; responses to stressors and COVID-19 may differ across regions. This study was not designed to assess regional differences. Fourth, we are using a well-established and validated depression screenerterm. While this provides probably the best possible assessment of population-level burden of depression, a diagnosis of depression is ultimately a diagnosis made by a clinician. Conclusions In this population-representative surveyterm·mth study of US adults, we found that prevalence of depression symptomsterm was more than 3-fold higher during COVID-19 compared with the most recent population-based estimates of mental health in the US. This increase in depression symptomterm prevalence is higher than that recorded after previous mass traumatic events, likely reflecting the far more pervasive influence of COVID-19 and its social and economic consequences than other, previously studied mass traumatic events. This study was conducted at the beginning of the COVID-19 pandemic in the US. On the final day of the surveyterm·mth, April 13, 2020, the death toll was more than 23 000 individuals in the US, with more than 600 000 cases confirmed. As of June 8, 2020, the US has experienced more than 113 000 deaths, with more than 2 000 000 cases confirmed. While too few people in the study had been diagnosed with COVID-19 themselves to meaningfully comment on differences in mental health between those who had and had not been diagnosed with COVID-19, we imagine that as the virus spreads and more cases of COVID-19 are confirmed, so too may mental illness increase among those with COVID-19 and those around them. These results suggest that context matters, and that the combined context of the COVID-19 pandemic and its economic consequences have resulted in an increase in mental illness in US adults. Addressing treatment for these individuals will be an interesting and important discussion for health professionals, particularly if a large number of depression cases are due to situational factors. While further data will be needed to assess the trajectory of depression in the US population and potential treatment for affected populations, it seems important to recognize the potential for the mental health consequences of COVID-19 to be large in scale, to recognize that these effects can be long-lasting, and to consider preventative action to help mitigate its effects. In particular, this burden is being borne by economically and socially marginalized groups, suggesting that individuals with low incometerm·var and with fewer resources may benefit from particular policy attention in coming months.