Changes in Self-Reported Depressive Symptoms Among Adults in the United States from 2005-2016

Abstract

Objective:

To determine national trends in self-reported depressive symptoms.

Methods:

This study examined interview data from the National Health and Nutrition Examination Survey (NHANES) from 2005–2016. Depressive symptoms were assessed using self-reported data on the 9 item Patient Health Questionnaire (PHQ-9), with a total score ≥ 10 and an individual item score of 2 or 3 indicating greater severity.

Results:

A total of 31,191 individuals contributed PHQ-9 data from 2005–2016. The absolute proportion of individuals with total PHQ-9 score ≥ 10 increased from 6.2% to 8.1%. After adjustment for participant demographic characteristics and comorbid conditions, the odds ratio for high PHQ-9 score at the end vs. the beginning of the study interval was 1.27 (95% confidence interval 1.07–1.50). Anhedonia, guilt/worthlessness, appetite, and hypoactivity/hyperactivity had the largest increases in individual item risk after adjusting for demographic and comorbid characteristics.

Conclusions:

There were sizeable increases in the prevalence of self-reported depressive symptoms in the United States over an 11-year period. Further work is needed to understand the reasons for and implications of this increase. However, the results suggest greater efforts should be made by healthcare providers to screen for depressive symptoms that may warrant further assessment, treatment, or referral to mental health services as needed.

Introduction

Major depressive disorder (MDD) is one of the most prevalent psychological disorders experienced, with a lifetime frequency predicted at 20.8% (Kessler et al., 2005). MDD’s societal burden is mediated by the association with a wide variety of chronic physical disorders; increasing costs of depression treatment; and significant depression-related impairments in work performance (American Psychiatric Association [APA], 2013; Kessler, 2012). Depressive symptoms are also commonly experienced with numerous other psychological and medical disorders. Impairment from depressive symptoms can range from very mild and almost undetectable to complete incapacity (APA, 2013). Depressive symptoms are linked with a higher chronic disease burden (Poole & Steptoe, 2018) and are a predictor for worse treatment outcomes in other psychological disorders and drug dependence (Compton, Cottler, Jacobs, Ben-Abdallah, & Spitznagel, 2003). Additionally, depressive symptoms are associated with greater mortality risk (Everson-Rose, House, & Mero, 2004). It is essential to analyze trends in depressive symptoms in order to address this substantial mental health care issue.

Major Depressive Disorder vs. Depressive Symptoms

Studies looking at MDD prevalence trends over time demonstrate mixed results; some indicate significant increases in MDD rates while others indicate no significant change (Mojtabai, Olfson, & Han, 2016; Brody, Pratt, & Hughes, 2018; Center for Behavioral Health Statistics and Quality, 2017). The years and populations of interest vary widely across studies which may explain the variability of results. MDD affects approximately 7% of Americans during a 12-month period (APA, 2013). MDD is associated with an increased risk of morbidity and mortality from other chronic diseases including, but not limited to, cardiovascular disease, diabetes, and chronic pain conditions which contribute to the growing economic burden of depression (Institute for Health Metrics and Evaluation, 2018; Saint Onge, Krueger, & Rogers, 2014; Greenberg, Fournier, Sisitsky, Pike, & Kessler, 2015). The economic burden from MDD, including direct costs, suicide-related costs, and workplace costs, increased by 21.5% from 2005 ($173.2 billion) to 2010 ($210.5 billion) (Greenberg et al., 2015). However, the burden of depressive symptomatology is not limited to individuals with major depression as depressive symptoms can be present in multiple mental and physical health disorders. Furthermore, individuals may experience depressive symptoms without ever meeting criteria for any mental health disorder. Trends in depressive symptoms in the United States have been analyzed through use of large data sets and survey data such as the National Health and Nutrition Examination Survey (NHANES). Shim, Baltrus, Ye, and Rust (2011) examined data from NHANES between 2005 and 2008 which demonstrated that 21.6% of the U.S. population aged 18 years and older (42 million United States adults) demonstrated at least mild (>5 on the PHQ-9) depressive symptoms in the past 2 weeks. Based upon the Shim et al.’s (2011) work, depressive symptoms of at least mild severity appear to be far more pervasive than major depression in the general U.S. population.

Importance of Depressive Symptom Burden Assessment

Disease burden of depressive symptoms is predicted to grow over time (Mathers & Loncar, 2006). As life expectancy increases, as a result of advanced medical care, it is anticipated that trends in both medical comorbidities and depressive symptoms will increase in the general population as well. In 2030, it is estimated that unipolar depressive disorders will be the second leading cause of disease burden (as measured in disability-adjusted life years) (Mathers & Loncar, 2006). However, treatment rates remain low—in one study, only 35% of persons with severe depressive symptoms reported seeing a mental health professional in the last year—and the treatment received is often inadequate (Shim et al., 2011; Pratt & Brody, 2014). This study examined national trends in depressive symptoms in the United States from 2005–2016, using a cross-sectional collection of surveys (NHANES). Analysis of trends in depressive symptoms has important implications in setting future clinical and policy priorities.

Methods

Overview

This study evaluated 11-year trends in self-reported depressive symptoms using a standardized rating scale (9-item Patient Health Questionnaire) in a nationally representative U.S. dataset.

NHANES

The National Health and Nutrition Examination Survey (NHANES) is a program of the National Center for Health Statistics that began in 1960 to assess the health and nutritional status of individuals in the United States. The NHANES is a cross-sectional collection of surveys and other health examination data for a nationally representative sample of the resident, civilian, non-institutionalized U.S. population with approximately 5,000 individuals sampled each year (Brody et al., 2018; Centers for Disease Control and Prevention [CDC], 2005–2016). The NHANES is conducted in consecutive two-year cycles (i.e. 2005–2006, 2007–2008).

PHQ-9

NHANES interviewers assessed self-reported depressive symptoms using the 9 item Patient Health Questionnaire (PHQ-9), a widely used self-report instrument to screen for and to evaluate depressive symptom severity. The questionnaire is reliable (Cronbach’s alpha of 0.89) and valid (score ≥10 has sensitivity of 88% and specificity of 88% for major depression) (Kroenke, Spitzer & Williams, 2001). Although previous studies have considered a total PHQ-9 score of ≥ 10 as a depression diagnosis (Brody et al., 2018; Kroenke et al., 2001), no confirmatory diagnostic test or clinical evaluation is performed during the NHANES assessment. Accordingly, in our study, we consider this score as an indicator of depressive symptom severity only.

Individual PHQ-9 items correspond with Diagnostic and Statistical Manual of Mental Disorders (DSM) diagnostic criteria for major depression. A score of 2 (more than half the days) or 3 (nearly every day) on 5 or more individual PHQ-9 items is suggestive of major depressive disorder when at least one of those items is anhedonia or depressed mood (Kroenke et al., 2001). Therefore, we assume a score of 2 or 3 on an individual PHQ-9 item represents greater individual symptom severity.

Analysis

We assessed interview data from the NHANES reflecting self-reported depressive symptoms over an 11-year time frame (2005–2016) (CDC, 2005–2016). In addition to this data, we obtained NHANES demographic and comorbidity data for the same years of interest. Demographic data obtained included gender, age, and race/ethnicity. The total number of comorbidities was determined using data for arthritis, angina, emphysema, chronic bronchitis, liver conditions, chronic obstructive pulmonary disease, asthma, congestive heart failure, coronary artery disease, stroke, cancer, kidney disease, and diabetes. We analyzed data from two concatenated NHANES cycles (2005–2006 and 2015–2016) to determine national trends in total and individual depressive symptom scores over the entire study interval. To examine trends in depressive symptom scores over time, we performed logistic regression analyses for absolute proportion of respondents scoring ≥ 10 (no vs. yes) and for percentage of respondents scoring 2 or 3 on an individual PHQ-9 item (no vs. yes). Odds Ratios were adjusted for gender, age, race/ethnicity, and comorbid conditions using multivariate regression models. All statistical analyses were completed using JMP, a statistical software program {version 14, SAS, Cary, NC}. The NHANES has received IRB/ERB approval from the National Center of Health Statistics Research Ethics Review Board (ERB) (Centers for Disease Control and Prevention [CDC], 2017).

Results

Sample Description

A total of 60,936 individuals were interviewed for the NHANES from 2005–2016. 24,649 individuals were excluded due to age being younger than 18 years. Out of 36,287 interviewees aged 18 years or older, 34,963 individuals contributed NHANES PHQ-9 data from 2005–2016. 3,772 additional individuals were excluded from analysis due to missing answers to one or more PHQ-9 items for a total of 31,191 individuals included in the analysis. Table 1 provides participant demographic and comorbidity data for all cycle years from 2005–2006 to 2015–2016. The majority of participants were female, between 18 and 60 years old, non-Hispanic White, and had no comorbidities.

Table 1.

Characteristics of NHANES Participants, 2005–2016

	2005–2006		2007–2008		2009–2010		2011–2012		2013–2014		2015–2016
	n	%	n	%	n	%	n	%	n	%	n	%
Gender
Male	2312	48.2%	2687	49.6%	2759	49.7%	2483	50.4%	2585	48.1%	2508	48.9%
Female	2487	51.8%	2728	50.4%	2787	50.3%	2442	49.6%	2787	51.9%	2626	51.1%
Age, years
18–39	2164	45.1%	1924	35.5%	2007	36.2%	1893	38.4%	2025	37.7%	1867	36.4%
40–59	1324	27.6%	1652	30.5%	1787	32.2%	1487	30.2%	1705	31.7%	1602	31.2%
60–79	1038	21.6%	1521	28.1%	1431	25.8%	1272	25.8%	1360	25.3%	1369	26.7%
80+	273	5.7%	318	5.9%	321	5.8%	273	5.5%	282	5.2%	296	5.8%
Race/Ethnicity
Non-Hispanic White	2311	48.2%	2532	46.8%	2683	48.4%	1836	37.3%	2315	43.1%	1710	33.3%
Non-Hispanic Black	1143	23.8%	1123	20.7%	986	17.8%	1319	26.8%	1087	20.2%	1096	21.3%
Hispanic	1161	24.2%	1569	29.0%	1606	29.0%	987	20.0%	1234	23.0%	1590	31.0%
Other Race	184	3.8%	191	3.5%	271	4.9%	783	15.9%	736	13.7%	738	14.4%
Number of Comorbidities
0	2908	60.6%	2936	54.2%	3131	56.5%	2846	57.8%	3030	56.4%	2846	55.4%
1	1125	23.4%	1371	25.3%	1382	24.9%	1149	23.3%	1308	24.3%	1184	23.1%
2	454	9.5%	613	11.3%	591	10.7%	514	10.4%	594	11.1%	633	12.3%
3	176	3.7%	291	5.4%	256	4.6%	232	4.7%	251	4.7%	248	4.8%
4+	136	2.8%	204	3.7%	186	3.4%	184	3.7%	189	3.6%	223	4.4%

Change in Depressive Symptom Severity

Over the entire study period of interest, the absolute proportion of individuals with total PHQ-9 score ≥10 increased from 6.2% to 8.1%. After adjusting for participant demographic characteristics and comorbid conditions, the odds ratio for high PHQ-9 score (≥10) at the end vs. the beginning of the study interval was 1.27 (95% confidence interval (CI), [1.07–1.50]). Table 2 presents the absolute proportion of survey respondents with a total PHQ-9 score ≥10 for each survey year broken down by demographic characteristic and comorbidity load.

Table 2.

Change in proportion of adults with total PHQ9 score≥10 over time, NHANES 2005–2016

	2005–2006	2007–2008	2009–2010	2011–2012	2013–2014	2015–2016
All participants	6.2%	9.7%	9.4%	8.9%	9.5%	8.1%
Gender
Male	5.2%	6.8%	6.7%	6.4%	6.5%	6.5%
Female	7.2%	9.7%	12.1%	11.5%	12.3%	9.6%
Age, years
18–39	5.5%	9.9%	9.2%	7.7%	7.5%	7.6%
40–59	9.0%	12.5%	12.3%	11.6%	10.7%	8.6%
60–79	4.9%	7.7%	7.2%	8.4%	11.2%	8.6%
80+	2.9%	2.8%	5.0%	5.5%	9.2%	6.1%
Race/Ethnicity
Non-Hispanic White	5.1%	9.0%	8.3%	9.6%	9.7%	8.9%
Non-Hispanic Black	8.3%	9.7%	10.0%	8.8%	9.9%	7.7%
Hispanic	6.4%	10.8%	11.0%	10.9%	10.9%	8.7%
Other Race	5.4%	8.4%	8.9%	5.1%	6.0%	5.4%
Number of Comorbidities
0	4.2%	5.9%	7.1%	6.1%	5.3%	5.2%
1	7.4%	11.2%	8.9%	8.8%	11.1%	8.9%
2	9.5%	15.0%	13.0%	14.4%	16.0%	11.9%
3	15.9%	18.6%	21.1%	18.5%	20.3%	12.9%
4+	16.2%	24.0%	25.8%	26.1%	31.2%	25.1%

When odds ratios were analyzed for risk of respondents scoring 2 or 3 on individual PHQ-9 items (Table 3), the largest increases over the 11-year study interval were for anhedonia (1.48, 95% CI, [1.27–1.73]), guilt/worthlessness (1.37, 95% CI, [1.13–1.67]), appetite (1.37, 95% CI, [1.17–1.59]), and hypoactivity/hyperactivity (1.32, 95% CI, [1.04–1.67]).

Table 3.

Change in proportion of adults scoring 2 or 3 on specific PHQ9 symptoms over time, NHANES 2005–2016

Symptom	2005–2006 n=4799	2007–2008 n= 5415	2009–2010 n= 5546	2011–2012 n= 4925	2013–2014 n= 5372	2015–2016 n= 5134	Odds Ratio (2005/06 vs. 2015/16)
							Odds Ratio [95% CI]	p value
1. Anhedonia	6.3%	7.4%	8.6%	8.4%	9.9%	9.6%	1.48 [1.27–1.73]	<.001
2. Feel depressed	6.1%	8.0%	8.1%	7.4%	7.7%	7.2%	1.12 [0.95–1.33]	.163
3. Sleep disturbance	12.5%	17.2%	17.0%	15.1%	15.8%	15.0%	1.21 [1.07–1.36]	0.002
4. Fatigue	14.4%	17.5%	16.5%	15.6%	17.0%	17.5%	1.26 [1.12–1.41]	<.001
5. Appetite	6.8%	9.4%	9.5%	9.2%	9.7%	9.3%	1.37 [1.17–1.59]	<.001
6. Guilt/worthlessness	3.8%	6.2%	5.8%	6.1%	6.0%	5.5%	1.37 [1.13–1.67]	.002
7. Concentration	4.5%	5.8%	6.1%	6.2%	7.1%	6.1%	1.28 [1.07–1.54]	.008
8. Hypoactivity/Hyperactivity	2.6%	4.0%	4.0%	4.4%	4.4%	3.8%	1.32 [1.04–1.67]	.022
9. Self-harm	0.9%	1.3%	1.1%	1.4%	1.2%	1.3%	1.29 [0.87–1.91]	.201

The largest year-to-year increase in proportion of respondents scoring ≥ 10 on the total PHQ-9 survey and 2 or 3 on each individual PHQ-9 item occurs from 2005–2006 to 2007–2008 (Table 2 and 3). There are variations in the magnitude and duration of this initial upward trend depending on the variable observed (i.e. the large initial increase is continued to 2009–2010 in females and those with zero or 3 comorbidities). However, following the initial large increase in severity of self-reported scores between 2005/06 and 2007/08 most variables have smaller variations with increasing and decreasing proportion of respondents meeting our threshold for greater severity. All demographic and comorbidity groups appear to have overall increasing trends in severity of self-reported depressive symptoms during the study interval except for persons aged 40–59, Non-Hispanic Blacks, persons with 3 comorbidities (end proportion responding ≥ 10 less than beginning of study interval for all three groups), and persons identifying as Other race (beginning and end proportion responding ≥ 10 equal).

Discussion

Over the 11-year period of interest, using the results of a large nationally representative survey series, there were sizeable increases in self-reported depressive symptoms in the United States. Overall, the odds of having a total depressive symptom score indicating greater severity increased by 27% over this period of time. Individual symptoms with increased risk for greater severity were anhedonia, guilty/worthlessness, appetite, and hypoactivity/hyperactivity.

The use of the NHANES provides generalizable data that is nationally representative of the United States. This allows for trends in depressive symptoms to be estimated, analyzed, and generalized to the entire United States non-institutionalized population. Our study contributes to the literature regarding U.S. national trends in depressive symptoms. Using an 11-year study interval, we were able to examine relative change and total score trends from beginning to end, as well as, year to year. By analyzing each depressive symptom individually, we were not only able to examine changes in total depressive symptom score but also trends in risk for individual depressive symptom severity. This allowed us to identify symptoms that may have particular impact on total PHQ-9 score changes.

Trends in depressive symptom severity vary depending on the study reviewed (Mojtabai et al., 2016; Brody et al., 2018; Center for Behavioral Health Statistics and Quality, 2017). However, we believe this can be explained by the variability in populations of interest and in reports of overall symptom severity from year to year. Meertens, Scheepers, and Tax (2003) found similar fluctuations in the overall longitudinal depressive symptom trends in the Netherlands from 1975 to 1996. We hypothesize that the large increase in percentage of respondents scoring 2 or 3 on every individual PHQ-9 item and those having a total score of ≥ 10 between NHANES years 2005–2006 and 2007–2008 may correspond to the financial recession beginning in 2008. Due to the cross-sectional nature of the data set, we cannot determine causality or verify associations. We note that risk of reporting more severe depressive symptoms on individual PHQ-9 items and having total score ≥ 10 does not return to baseline following the initial increase and remains plateaued with little variation until the end of the study interval except in four individual demographic and comorbidity groups. The potential link between mental health, particularly depressive symptoms, and the economic recession of 2008 has been examined elsewhere in the literature. Katikireddi, Niedzwiedz, and Popham (2012) examined trends in population mental health before and after the 2008 recession and found that men’s mental health deteriorated within the following two years. Similarly, two recent studies found that foreclosures secondary to the 2008 recession were associated with a higher risk of developing depressive symptoms (Osypuk, Caldwell, Platt, & Misra, 2012; Cagney, Browning, Iveniuk, & English, 2014).

Studies examining trends in the treatment and control of depressive symptoms demonstrated similar yearly variability. One study found that antidepressant use increased from 1999 to 2014 (7.7% and 12.7%, respectively) (Pratt, Brody, & Gu, 2017). Another study found that mental-health spending, primarily driven by pharmaceutical spending on antipsychotics, grew by 29% between 1996 and 2005 (Fullerton, Busch, Normand, McGuire, & Epstein, 2011). However, Marcus and Olfson (2010) found that, despite similar increases in Medicare expenditures for depression treatment, the percentage of patients in treatment for depression who used antidepressants changed little from 1998 (73.8%) to 2007 (75.3%).

In contrast, trends in the percentage of people receiving psychotherapy consistently declined. Using two nationally representative surveys, Marcus and Olfson (2010) found that the percentage of persons being treated for depression who received psychotherapy declined between the years 1998 (53.6%) and 2007 (43.1%). Similarly, Fullerton et al. (2011) examined data from the Florida Medicaid program and found that the percentage of enrollees using psychotherapy decreased from 9% (1996) to 5% (2005). Usage of psychotherapy declined despite increases in depression prevalence, mental health expenditures, and economic burden of depression over similar time periods (Greenberg et al., 2015; Fullerton et al., 2011; Marcus & Olfson, 2010). These trends may be, in part, due to the high cost and relative scarcity of mental health and psychotherapy providers.

Our findings, in combination with previous work looking at prevalence, treatment, and control of depressive symptoms, suggest that continued efforts should be made to identify and improve management of depressive symptoms in the United States. It is important to evaluate and understand trends in depressive symptoms over time in order to assess need for intervention and where and how an intervention may be most effective. The increases we found suggest that depressive symptoms in the United States are increasing in prevalence and new and more effective interventions and/or combination of interventions are needed to address this problem.

Nurses, already on the frontlines of care management in all practice areas, play a vital role in developing and implementing creative solutions to the problem of depressive symptom identification and management. Nurses act as trusted advocates and “gatekeepers to vital health information that patients may not share with anyone else” (Hill, 2013). Improving rates of case identification through routine depression screening may lead to earlier identification of depressive symptoms and decrease overall depressive symptom burden to the patient, the medical system, and society (Haber & Billings, 1995). Both general and psychiatric practice nurses can leverage their role to improve rates of depression screening in all patients then advocate for those patients, using screening data, to provide symptomatic management, psychosocial support, and appropriate referral to mental health services as needed (Hill, 2013).

Limitations

Several limitations must be considered in interpreting our study. The PHQ-9 is a screening instrument, and our study is unable to determine depression diagnoses because no clinical assessments or diagnostic interviews were performed. Respondents may over- or under-report symptom frequency on the PHQ-9 leading to over- or under-estimation of depressive symptoms in our study. Furthermore, the cross-sectional design of the NHANES survey prevents us from examining changes over time in the same participants or attributing any causality to changes in depressive symptom trends. The NHANES also lacks PHQ-9 data for the years immediately preceding and following the years of interest. Expanding the interval of study may have allowed for improved understanding and graphical visualization of trends in self-reported depressive symptom severity over time. Based on previous study results and the relative change results from our own NHANES data sets, we suspect low response rates to the PHQ-9 item assessing suicidality (item 9) may have limited our ability to determine increased risk for scoring high (2 or 3) on the suicidality screening question at the 95% confidence interval.

Conclusion

In conclusion, self-reported depressive symptom severity fluctuates between NHANES survey years, but the overall trend appears to be increasing. Further work is needed to understand the reasons for and full implications of the increases in total depressive symptom scores and in individual depressive symptoms.

Our results suggest greater efforts should be made by healthcare providers to screen for depressive symptoms that may warrant further assessment, treatment or referral to mental health services as needed. However, to avoid overtreatment of non-pathological grief or sadness, as well as depressive symptoms that may respond without medication, actions should be taken to evaluate for any biopsychosocial or environmental stressors resulting in elevated depressive symptom scores.