Depression treatment patterns among women veterans with cardiovascular conditions or diabetes

Abstract

The study aimed to examine treatment patterns for depression among women veterans diagnosed with cardiovascular conditions or diabetes. We used longitudinal data from the 2002-2003 merged Veteran Health Administration (VHA) and Medicare files. Chi-square tests and multinomial logistic regression were performed to analyse depression treatment among veteran women with incident depressive episode and one of the following chronic conditions: diabetes or coronary artery disease or hypertension. Overall, 77% received treatment for depression, 54% with only antidepressants, 4% with only psychotherapy, and 19% with both. Multinomial logistic regression revealed that African American women were more likely to be in the no treatment group and were more likely than white women to receive psychotherapy rather than antidepressants. Older women and women with coronary artery disease only were less likely to receive treatment.

The worldwide higher prevalence of depression in women compared to men has been well established 1. However, research on treatment for depression among women, specifically when depression co-exists with medical illnesses, has not received much attention. One can piece together some indirect evidence on treatment patterns for depression among women through studies that analyse gender differences in depression or studies that assess quality of depression care. One study from the 1990s reported that women in primary care were more likely to receive a prescription for antidepressant medication than men 2. In adults over 60 years of age, women were more likely than men to receive depression care 3. In randomized controlled trials that assess quality improvement for depression care, women were more likely to receive depression care than men over time 4. Based on psychiatrists’ responses to video vignettes depicting an elderly patient with late-life depression, it was concluded that patient’s gender did not influence depression management 5.

However, none of those studies examined treatment for depression when chronic illnesses co-exist with the depressive syndrome. An analysis of depression treatment rates and patterns in cardiovascular diseases and diabetes is needed. One in five women have some form of cardiovascular disease 6 and there are more women with diabetes than men 7. In a sample of women with diabetes, major depression was an independent risk factor that accelerated the development of coronary heart disease 8. In addition, the first clinical trial testing the relationship between depression treatment and mortality in coronary heart disease patients reported a significant treatment-sex interaction, suggesting that women may have had worse outcomes compared to men 9.

The primary objective of this study was to report rates of treatment for depression (antidepressants and/or psychotherapy) among women with diabetes or coronary artery disease or hypertension and analyse variations in these rates by demographic, socioeconomic and health status characteristics. The population of women veterans is of special interest because research on women veterans with chronic physical conditions is scant 10 and, specifically, treatment issues in women veterans with co-occurring physical and mental illnesses have remained an understudied area of research within the Veteran Health Administration (VHA) 11.

We studied women veterans who used the VHA clinics and were diagnosed with either diabetes or coronary artery disease or hypertension in fiscal year 2002 and evaluated depression treatment for incident depression episodes in fiscal year 2003.

METHODS

Data for our study come from the merged VHA and Medicare data in fiscal years 2002-2003. Fiscal year 2002 represented the year from October 1, 2001 to September 30, 2002; fiscal year 2003 represented the year from October 1, 2002 to September 30, 2003. We excluded women veterans who died by the end of fiscal year 2003. Among dual VHA/Medicare enrollees, we included only those women veterans who had fee-for-service enrollment for all of the 12 months in 2002 and 2003.

We identified women veterans with diabetes or coronary artery disease or hypertension during fiscal year 2002 using an algorithm based on disease-specific ICD-9-CM codes, the number of clinical care encounters, site of care (inpatient versus outpatient), and level of diagnosis (primary versus other). For example, to identify women with coronary artery disease, we required that there was at least one inpatient visit or one outpatient visit with primary diagnosis of coronary artery disease or two outpatient visits with diagnosis of primary or secondary coronary artery disease.

Any depression diagnosis was identified using ICD-9-CM codes 296.2 (major depressive disorder, single episode), 296.3 (major depressive disorder, recurrent episode), 311 (depression not elsewhere classified), 309.1 (prolonged depressive reaction), 300.4 (neurotic depression), 309.0 (adjustment disorder with depression), and 298.0 (depressive type psychosis). To define an incident episode of depression in fiscal year 2003, we followed an algorithm used in our prior published work: the criteria consisted of 120-day negative depression and/or antidepressant medication history on or before the first depression diagnosis date in fiscal year 2003 12. To allow for sufficient follow-up for depression treatment, we included only those who had at least 6 months of follow-up period in fiscal year 2003.

Thus, the final study population for analysis consisted of women veterans who used VHA clinics and were diagnosed with diabetes or coronary artery disease or hypertension, of whom 8,147 had an incident depression diagnosis in fiscal year 2003.

Antidepressant medications were identified using drug names from the VHA Pharmacy Benefit Management (PBM) files. During the period of observation, Medicare did not provide prescription drug benefits. Therefore, antidepressants were derived only from VHA files. Among women veterans with incident depression in fiscal year 2003 and diagnosed with diabetes or coronary artery disease or hypertension in fiscal year 2002, any woman with at least one prescription for antidepressant medications on or after the depression diagnosis date in 2003 was considered to be receiving antidepressant treatment for depression.

We compiled a comprehensive list of psychotherapy codes using current procedural terminology (CPT) codes from both the VHA and Medicare files to assess psychotherapy treatment for depression. The list of codes was reviewed by psychologists and psychiatrists of the research team. Again, among women veterans with incident episodes of depression, any visit with psychotherapy codes within 180 days on or after the start date of the incident episode of depression in fiscal year 2003 was defined as psychotherapy use.

Based on antidepressant prescriptions and psychotherapy visits, we classified women veterans into four mutually exclusive categories: no depression treatment; antidepressant use only; psychotherapy use only; and both antidepressants and psychotherapy. Due to very small number of women veterans (n=297; 3.6%) using psychotherapy only, subgroup differences in psychotherapy only yielded many cells with less than two individuals. Therefore, we combined the groups receiving psychotherapy with and without antidepressant, and analysed depression treatment with three categories.

Independent variables consisted of race/ethnicity (white, African American, Latino, other, and missing), age (less than 50 years, 50-64 years, 65-74 years, and 75 or older), Medicare enrollment (12 months fee-for-service Medicare enrollment versus only VHA coverage), region (Northeast/Midwest/South/West), index diagnosis in fiscal year 2002 (diabetes only, coronary artery disease only, hypertension only, diabetes and coronary artery disease, diabetes and hypertension, coronary artery disease and hypertension, or all three conditions), other physical conditions (none, one, two or more), and psychiatric comorbidities (anxiety disorders, bipolar disorder, psychoses other than schizophrenia, post-traumatic stress disorder (PTSD), schizophrenia, substance abuse). All medical and psychiatric comorbidities were based on ICD-9-CM codes in fiscal year 2002.

Group differences in depression treatment were tested with the chi-square statistic. To examine depression treatment patterns among women veterans, we used multinomial logistic regression. Bivariate and multivariate analyses were conducted with the 3-level depression treatment variable (“no treatment”, “antidepressants only”, and “psychotherapy with or without antidepressants”). The parameter estimates from the multinomial logistic regression were transformed to odds ratios and their corresponding 95% confidence intervals. In the multinomial regression, for the dependent variable, the reference group was “antidepressant use only”.

RESULTS

A description of our study population of 8,147 women veterans with incident depression episode in fiscal year 2003 and a diagnosis of diabetes or coronary artery disease or hypertension in fiscal year 2002 is provided in Table 1. A majority of women veterans were white (69% white versus 19% African American and 1.9% Latino); 25% of women veterans were 65 years or older. About half (55%) of the women veterans were enrolled in the Medicare fee-for-service system. An overwhelming majority of women veterans (84%) had hypertension, either alone (55%) or in combination with diabetes or coronary artery disease (29%). Thirty-nine percent of the women veterans qualified for VHA enrollment because of low-income status. The most highly prevalent psychiatric comorbidity was anxiety disorders (20%). Another common condition was substance use disorders: 20% had clinical encounters with substance use diagnosis (alcohol, drugs and/or tobacco) codes.

Among those with incident episodes of depression in fiscal year 2003, 54% were prescribed antidepressants only, 23% had psychotherapy with or without antidepressants and 23% had no depression treatment (Table 2). All bivariate subgroup differences in depression treatment were significant; therefore we highlight only those that were significant in multinomial logistic regression (Table 3). A significantly higher percentage of African Americans women compared to white women (26.5% vs. 17.1%) had no prescriptions for antidepressants or psychotherapy visits within 180 days of follow-up. The adjusted odds ratio for no treatment for African Americans was 1.25 (95% CI=1.05, 1.49) compared to only antidepressants (Table 3). Psychotherapy rather than antidepressant use was more likely among African American women compared to white women (32.0% versus 18.9%), with an adjusted odds ratio of 1.38 (95% CI=1.19, 1.59).

Older women (65 years and older compared to those in the age group 50 years and younger) were more likely to have “no depression treatment”. The adjusted odds ratio for no treatment was 1.40 (95% CI=1.03, 1.90) for veteran women with coronary artery disease only compared to veteran women with hypertension only. Variations in the relationships between type of mental illness and depression treatment were observed. For example, those with anxiety disorders or PTSD were less likely to receive “no depression treatment”, whereas those with psychoses were more likely to receive “no depression treatment”. Those with any substance use disorders were less likely to receive “no depression treatment”.

Table 1.

Table 1 Description of study population (N=8,147)

	N	%
Race/ethnicity
White	5,598	68.7
African American	1,530	18.8
Latino	153	1.9
Other	73	0.9
Age
< 50 years	3,018	37.0
50-64	2,253	27.7
65-74 years	874	10.7
75, +	2,002	24.6
Priority status
Disabled	3,706	45.5
Poor	3,157	38.8
Co-pay	1,188	14.
Index diagnosis
Diabetes only	683	8.4
Coronary artery disease only	295	3.6
Hypertension only	4,435	54.4
Diabetes plus coronary artery disease	53	0.7
Diabetes plus hypertension	1,199	14.7
Coronary artery disease plus hypertension	927	11.4
All the three conditions	555	6.8
Other physical conditions
None	252	3.1
One	1,688	20.7
Two or more	6,207	76.2
Other mental disorders
Anxiety disorders	1,662	20.4
Bipolar disorder	733	9.0
Psychoses other than schizophrenia	527	6.5
Post-traumatic stress disorder	897	11.0
Schizophrenia	524	6.4
Any substance abuse	1,647	20.2

Table 2.

Table 2 Description of population by depression treatment categories

	No treatment		Antidepressants only		Psychotherapy
	N	%	N	%	N	%
	1,887	23.2	4,388	53.9	1,872	23.0	p
Race/ethnicity							<0.001
White	1,486	26.5	3,055	54.6	1.057	18.9
African American	261	17.1	780	51.0	489	32.0
Latina	27	17.6	79	51.6	47	30.7
Other	17	23.3	39	53.4	17	23.3
Age							<0.001
< 50 years	421	13.9	1,648	54.6	949	31.4
50-64	354	15.7	1,298	57.6	601	26.7
65-74 years	290	33.2	449	51.4	135	15.4
75, +	822	41.4	993	49,6	187	9.3
Married							<0.001
Married	605	23.4	1,427	55.3	550	21.3
Widowed	478	35.3	694	51.3	181	13.4
Divorced/Separated	484	18.4	1,439	54.8	702	26.7
Never married	309	20.1	803	52.3	422	27.5
Priority status							<0.001
Disabled	731	19.7	2,023	54.6	952	25.7
Poor	792	25.1	1,646	52.1	719	22.8
Co-pay	347	29.2	662	55.7	179	15.1
Missing	17	17.7	57	59.4	22	22.9
Index diagnosis							<0.001
Diabetes only	126	18.4	368	53.9	189	27.7
Coronary artery disease only	74	25.1	147	49.8	74	25.1
Hypertension only	920	20.7	2,418	54.5	1,097	24.7
Diabetes plus coronary artery disease	20	37.7	29	54.7	4	7.5
Diabetes plus hypertension	280	23.4	659	55.0	260	21.7
Coronary artery disease plus hypertension	294	31.7	480	51.8	153	16.5
All the three conditions	173	31.2	287	51.7	95	17.1
Other physical conditions							<0.001
None	33	13.1	134	53.2	85	33.7
One	289	17.1	914	54.1	485	28.7
Two or more	1,565	25.2	3,340	53.8	1,302	21.0
Anxiety disorders							<0.05
Yes	344	20.7	938	56.4	380	22.9
No	1,543	23.8	3,450	53.2	1,492	23.0
Bipolar disorder							<0.01
Yes	148	20.2	440	60.0	145	19.8
No	1,739	23.5	3,948	53.3	1,727	23.3
Psychoses other than schizophrenia							<0.001
Yes	163	30.9	273	51.8	91	17.3
No	1,724	22.6	4,115	54.0	1,781	23.4
Post-traumatic stress disorder							<0.001
Yes	94	10.5	583	65.0	220	24.5
No	1,793	24.7	3,805	52.5	1,652	22.8
Schizophrenia							<0.05
Yes	118	22.5	307	58.6	99	18.9
No	1,769	23.2	4,081	53.5	1,773	23.3
Any substance abuse							<0.001
Yes	273	16.6	923	56.0	451	27.4
No	1,614	24.8	3.465	53.3	1,421	21.9

Table 3.

Table 3 Adjusted odds ratios (AOR) from multinomial logistic regression on depression treatment

	No treatment			Psychotherapy use
	AOR	95% CI	P	AOR	95% CI	p
Race/ethnicity
White
African American	125	1.05,1.49	<0.05	1.38	1.19,1.59	<0.001
Latina	1.05	0.66,1.67		1.35	0.92,1.98
Other	1.33	0.73,2.40		1.03	0.58,1.85
Age
< 50 years
50 – 64	0.99	0.84,1.17		0.86	0.75,0.98	<0.05
65 - 74 years	1.62	1.29,2.03	<0.001	0.77	0.60,0.99	<0.05
75, +	2.01	1.63,2.47	<0.001	0.49	0.39,0.61	<0.001
Index diagnosis
Diabetes only	1.22	0.96,1.56		1.03	0.82,1.28
Coronary artery disease only	1.40	1.03,1.90	<0.05	1.15	0.84,1.15
Hypertension only
Diabetes plus coronary artery disease	1.62	0.88,2.97		0.38	0.13,1.09
Diabetes plus hypertension	1.03	0.87,1.21		0.98	0.83,1.15
Coronary artery disease plus hypertension	1.08	0.91,1.29		1.04	0.85,1.28
All the three conditions	1.06	0.86,1.31		1.12	0.87,1.45
Other physical conditions
None
One	1.41	0.91,2.19		0.89	0.63,1.24
Two or more	1.42	0.92,2.20		0.88	0.63,1.23
Anxiety disorders
Yes	0.81	0.70,0.94	<0.01	1.07	0.93,1.23
No
Bipolar disorder
Yes	0.92	0.74,1.14		0.79	0.65,0.98	<0.05
No
Psychoses other than schizophrenia
Yes	1.30	1.04,1.62	<0.05	1.05	0.81,1.37
No
Post-traumatic stress disorder
Yes	0.45	0.36,0.57	<0.001	0.77	0.65,0.9]	<0.01
No
Schizophrenia
Yes	1.00	0.78,1.28		0.81	0.63,1.05
No
Any substance abuse
Yes	0.84	0.72,0.99	<0.05	1.12	0.98,1.29
No

DISCUSSION

In this study, we set out to examine depression treatment patterns in women veterans with cardiovascular conditions or diabetes and incident depression. Because very few observational studies have analysed depression treatment in women with chronic illnesses, we are unable to compare the rates of depression treatment in our study to other studies. Although not directly comparable, in one study based on 18 women with well-controlled diabetes and depression living in Northeastern Connecticut, only 18% were treated with psychotherapy and antidepressants 13, which is similar to our overall rates of psychotherapy estimated at 23%.

The lower likelihood of antidepressant only treatment among African American women compared to white women is similar to findings of racial disparities in the general population 14,15, in individuals with heart disease 16, and in veterans with diabetes 12. All women in our study had prescription drugs coverage. In the presence of equalized access in terms of insurance coverage, one of the factors that could explain the racial differences may be the greater mental health stigma among African Americans in general 17,18 and specifically in African American women compared to white women 19.

However, African American women were as likely as white women to receive psychotherapy. Among both men and women, African Americans are less likely than whites to accept antidepressant medication for depression treatment 17. A majority of women who received psychotherapy reported that the treatment is acceptable 20 and racial minorities in general prefer counseling for depression over antidepressants 21. Taken together, our findings and those from the literature suggest that promoting psychotherapy in this population could be one way to reduce racial disparities in depression treatment.

Compared to women with hypertension only, women with coronary artery disease were more likely to be in the no depression treatment category. Further research needs to explore whether such lower likelihood of treatment in these women is related to the adverse outcomes observed in the ENRICHD trial. Women with coronary artery disease in our study were as likely as women with hypertension only to use psychotherapy. Given the beneficial effects of psychotherapy found in the ENRICHD trial 9,22, psychotherapy could be encouraged for depression treatment in this population of women veterans 23. In addition, as recommended by the advisory panel for heart disease and depression care, these women need to be closely monitored for both mental and cardiac health outcomes 24.

Our study findings did not support the general perception that multimorbidity may lead to less treatment for depression 25. In our study, women veterans with combinations of diabetes, coronary artery disease, and hypertension as well as increased number of other physical conditions did not have a lower likelihood of depression treatment. This is consistent with a recently published study that concluded that depression treatment in primary care settings is not influenced by competing demands for care for other comorbid medical conditions 26. It is possible that lack of significant difference in treatment by comorbidity could be due to close monitoring of individuals with serious mental illness such as depression and chronic diseases 27.

Although improvement in depressive symptoms due to depression treatment within integrated settings (i.e. mental health and primary care) like the VHA has been observed in older adults 28 and in those with chronic illnesses 29, we found differences in depression treatment by age. These differences were similar to those observed in the general population 30 and in the veteran population with diabetes 12. Multiple barriers to depression care among older individuals are documented 30. However, older individuals generally prefer integrated treatment in the context of heart disease 29. Interventions to improve depression treatment in older adults could include education materials highlighting depression as a cardiovascular risk factor.

The study had many advantages, such as the use of a large nationwide database of women veterans, the availability of information from Medicare and VHA to help capture complete utilization data, and the ability to identify diagnosed diabetes, coronary artery disease, hypertension and depression. One of the major limitations of the study is that we only observed prescription for antidepressants and we do not have information on the actual use of the medications. Similarly, we only observed actual psychotherapy visits and we do not know whether women veterans were referred to psychotherapy and did not follow through on the referral. Another limitation of our study is the generalizability of the findings outside of the VHA system. Furthermore, the use of diagnostic codes from specific years to identify chronic diseases has its drawbacks: these data cannot be used to derive variables such as date of onset, severity and duration of physical and mental illnesses. Severity of depression could have attenuated the relationship between low depression treatment rates and coronary artery disease only. Similarly, severity of physical illnesses could be a barrier to depression care in the presence of multimorbidity 26.

Despite these limitations, our study contributes to the nascent literature that has begun to explore treatment patterns among women with chronic diseases. Subgroup differences in depression treatment of women veterans with chronic diseases were generally similar to the patterns observed in the general population of men and women. Our study findings suggest the need for further research in improving depression care among women with heart diseases, integrating depression care in the context of cardiovascular risk reduction, and promoting psychotherapy use to reduce racial disparities in depression treatment.