Depressive symptoms and the incidence of adult-onset asthma in African American women

Introduction

Asthma and depression are significant burdens on the health of African American women, with estimated prevalences of 11.6%¹ (current asthma) and 13.1%² (lifetime prevalence of major depressive disorder). Depression commonly coexists with asthma and is associated with more severe asthma and poorer asthma management.^3–5 Some data suggest that depressed or sad moods elicited under laboratory conditions can produce pulmonary effects consistent with decreased airway function, but overall the evidence is weak.^6–8 Stress is a known trigger of asthma exacerbation in patients with asthma⁹; it has been hypothesized that stress effects on the immune and autonomic nervous systems are also relevant to the development of asthma.^10,11 Major depressive disorder leads to overproduction of corticotropin-releasing hormone and hyperactivity of the hypothalamic–pituitary–adrenal axis,^12,13 hallmarks of stress.

Although the influence of maternal and perinatal stress on asthma onset in the young child has received most study,¹¹ some epidemiologic evidence suggests that increased risk of adult-onset asthma is associated with stress, including perceived stress,¹⁴ war-related stress,¹⁵ abuse during childhood,¹⁶ and experiences of racism.¹⁷ Three prospective cohort studies—the Canadian National Population Health Survey,¹⁸ the National Health and Nutrition Examination Survey (NHANES),¹⁹ and a German cohort study²⁰— reported that the risk of new-onset asthma was increased in adults with a history of depression by 24%²⁰ to approximately 200%.^18,19 No such data are available in an African American population.

The purpose of the present analysis was to assess the association between depressive symptoms and adult-onset asthma in a prospective cohort study of African American women, the Black Women’s Health Study (BWHS). In previous work, the authors found that high body mass index (BMI),²¹ experiences of child abuse,¹⁶ and experiences of racism¹⁷ were associated with an increased incidence of adult-onset asthma.

Methods

Results

Compared with women in the lowest category of depressive symptoms (Table 1), those with higher levels were younger, heavier, had fewer years of education and lower income, and engaged in fewer hours of vigorous exercise. Women with higher levels also were more likely to smoke, to have been exposed to secondhand smoke at all ages, to report sleep apnea, to have used female hormone supplements, and were less likely to have a regular physician and health insurance. All covariates were significantly associated with CES-D score (P < .001), with the exception of parental history of asthma (P = .26).

There was a positive relation between CES-D score and asthma incidence (Table 2). The IRR adjusted only for age and calendar time was 2.59 (95% CI 1.98–3.40) for the highest compared with the lowest category of CES-D score, with a significant trend (P <.0001). The association was slightly attenuated after controlling for multiple confounding factors: the IRR in the highest category was 2.08 (95% CI 1.58–2.74, P for trend < .0001). Results from an analysis confined to women who had CES-D scores for 1999 and 2005 (n = 20,477) were almost identical to the main results (data not shown).

The authors assessed the joint effect of CES-D score (<16 vs ≥16) and use of antidepressant medication (yes vs no; Table 3). Twelve percent of participants had used an antidepressant during follow-up: 9% of those with a 1999 CES-D score lower than 16 and 20% of those with a CES-D score of at least 16. Compared with women with a CES-D score lower than 16 who did not take antidepressants (reference), the multivariable IRR for women with a CES-D score of at least 16 who used antidepressants was 2.77 (95% CI 2.05–3.75). The IRR also was significantly increased, although less so, for women with a CES-D score lower than 16 who used antidepressants and for women with a CES-D score of at least 16 who did not take antidepressants. There was no evidence of statistical interaction between antidepressant use and CES-D score (P for interaction = .61).

In analyses stratified by smoking (Table 4), the multivariable IRRs for the CES-D category of at least 33 vs lower than 16 were 2.73 (95% CI 1.85–4.04) in current and former smokers and 1.68 (95% CI 1.12–2.50) in never smokers. The IRRs for the same comparison were 2.74 (95% CI 1.82–4.12) in women with a BMI lower than 30 kg/m² and 1.68 (95% CI 1.15–2.45) in obese women. IRRs were similar in normal-weight (BMI <25 kg/m²) and overweight (BMI 25–<30 kg/m²) women (data not shown). IRRs for the highest vs lowest CES-D categories were 2.35 (95% CI 1.71–3.22) in women at least 40 years old and 1.48 (95% CI 0.84–2.62) in younger women. IRRs were 1.98 (95% CI 1.28–3.05) in those who scored below the median on the coping scale (poorer coping skills) and 1.42 (95% CI 0.92–2.21) in those who scored at or above the median. There was no evidence of statistical interaction between CES-D score and any of the stratifying variables (Table 4).

In sensitivity analyses, a more stringent asthma case definition was used that included only cases who reported use of a preventive asthma medication and specified the medication name (eg, fluticasone; n = 521): the results were similar to the main results, with a multivariable IRR of 1.96 (95% CI 1.38–2.78) for a CES-D category of at least 33 vs lower than 16. In a second sensitivity analysis, the case definition was expanded to include all physician-diagnosed asthma regardless of medication use (n = 1,456): the IRR for the same comparison was 1.69 (95% CI 1.36–2.10). The authors also conducted a lagged analysis in which the 1999 CES-D score was the exposure and cases diagnosed during the first 4 years of follow-up were excluded. The IRRs for CES-D categories 16 to 22, 23 to 32, and at least 33 were 1.13 (95% CI 0.86–1.49), 1.06 (95% CI 0.75–1.51), and 1.73 (95% CI 1.16–2.60), respectively (P for trend < .0001).

Discussion

In this large follow-up study of African American women, there was a positive association between CES-D score and incidence of adult-onset asthma. The association was present in all strata of age, BMI, and smoking. The IRRs were highest in women who reported antidepressant use, which may indicate the presence of depression or of more severe depression.

The overall findings are consistent with positive results from 3 previous prospective studies of depressive symptoms and asthma.^18–20 In the NHANES, depressive symptoms were assessed at baseline (NHANES I, 1982–1984) using a 4-item “cheerful versus depressed” scale and classified as low, medium, or high.¹⁹ Over an average of 9.4 years of follow-up, 181 of 5,231 adults reported new physician-diagnosed asthma. The multivariable relative risk for the high vs low depressive category was 2.08 (95% CI 1.14–3.79) in nonsmokers, but there was no association in smokers. The Canadian National Population Health Survey included 15,254 participants older than 12 years interviewed in 1994 and followed every 2 years until 2002.¹⁸ The Composite International Diagnostic Interview Short Form (2 screening questions followed by 7 questions for those who screened positive) was used to assess past-year major depressive episodes at every interview, and presence of major depression was modeled as varying by time. Over 8 years of follow-up, the hazard ratio for incident asthma in those reporting major depression was 1.7 (95% CI 1.2–2.4) after adjusting for sex, age, and health care use. During the first 2 years of follow-up, 283 cases occurred, but the total number of cases during the 8 years of follow-up was not given. In the most recent study,²⁰ 5,114 German adults 40 to 65 years old completed a baseline questionnaire in the early 1990s, including a 16-question depression scale, and newly diagnosed asthma was ascertained on a second questionnaire administered 8.5 years later. There was a nonsignificant 65% increase in the multivariable relative risk for the highest of 3 categories of depressive symptoms, based on 63 total cases. When analyzed as a continuous z score, the relative risk per 1-SD increment in the depressive symptom score was 1.24 (95% CI 1.02–1.50).

The present finding of a positive association in smokers and nonsmokers conflicts with the findings from the NHANES, in which a positive association was observed only in nonsmokers.¹⁹ The other 2 studies did not stratify by smoking status and none of the 3 stratified by BMI. The present observation of a stronger association between CES-D score and asthma in nonobese compared with obese women might reflect the lower background risk of asthma in nonobese women.²¹ A weak association in women with more coping skills may reflect the amelioration of stress effects by better coping strategies. Given that lung function decreases with age,³¹ the stronger association observed in older women may reflect a greater sensitivity of the aging lung to stress. However, none of the observed variations across strata were statistically significant, and apparent differences may be chance variation.

The present study was larger than the previous follow-up studies, providing excellent statistical power. The authors also controlled for some important potential confounding factors. The authors were not able to control for occupational or other environmental exposures that might be relevant, including indoor and outdoor air contaminants, and they did not have information on the presence of allergy or rhinitis. The ascertainment of depressive symptoms before asthma diagnosis precluded recall bias. Diagnostic bias might have contributed to the present findings if people with high levels of depressive symptoms are more likely to see a physician and then to be diagnosed with asthma than people with fewer depressive symptoms. The authors attempted to address this possibility by conducting a lagged analysis in which cases diagnosed in the first 4 years of follow-up were excluded. The IRRs were decreased by approximately 20%, compatible with some degree of detection bias. However, having symptoms is not the same as physician-diagnosed depression that would bring a person into the medical system; indeed, women in the highest CES-D category at baseline were less likely to report that they had a regular physician and health insurance (Table 1). Also, results were unchanged in an analysis using a stricter case definition that might have indicated the presence of more severe disease that would be diagnosed regardless of depressive symptoms.

The validity of the CES-D as a screening tool for detecting depressive symptoms has been demonstrated,^23,24 as has the validity of this tool within the BWHS.²⁷ There is a high correlation between the CES-D and the Major Depression Inventory.²⁵

A limitation of the study was reliance on self-reported physician-diagnosed asthma. However, self-report is the standard method of identifying asthma cases in large national cohort studies^32–36 because it is infeasible to examine all cases to verify asthma or to examine all noncases to identify undiagnosed asthma. In a small validation study of 43 women in the BWHS who met the present case definition and who gave the authors permission to contact their physicians, 39 (91%) were confirmed by their physicians as having asthma. In addition, overall results were consistent when a more restrictive case definition was used.

The authors hypothesized that stress might be a pathway linking depression and asthma. Psychological stress leads to neuroendocrine alterations of the hypothalamic–pituitary–adrenal axis and of the sympathetic and adrenomedullary systems, alterations that in turn appear to have immune system effects that could increase allergy and airway inflammation.¹⁰ Chronic stress is associated with increased basal cortisol concentration, which has been reported to affect antigen presentation³⁷ and bias the immune system toward T-helper type 2 cytokine responses.³⁸ In addition, central nervous system effects of stress may adversely affect autonomic nervous system control of airway smooth muscle tone, and stress also may lead directly to neurogenic airway inflammation through the effects of the nonadrenergic noncholinergic nervous system and tachykinins such as substance P.^10,38 Although a stress or depression link is plausible, it is speculative for at least 2 reasons. First, most of the data indicating a biological role for stress in asthma incidence pertain to prenatal and early-life stressors and their effect on asthma development in young children;¹ whether these mechanisms are relevant to stress experienced as adults and to adult-onset asthma is unknown. Second, whether depression causes significant chronic stress is not clear. Abnormalities in stress hormones (eg, elevated plasma cortisol levels) have been observed in some groups of patients with depression but are apparently not typical for all patients with depression.³⁹

In conclusion, the present results support the hypothesis that depression increases the risk of adult-onset asthma. Depression and asthma are substantial public health burdens on the US population, and the incidence and prevalence of the 2 conditions are higher in women than in men.^40,41 If the hypothesis is correct, depression could contribute substantially to the burden of asthma in adults. Additional population-based research and laboratory and clinical studies explicating the mechanisms by which stress increases asthma risk might indicate subgroups at high risk for asthma and generate new therapies for asthma prevention.

Supplementary Material