- Email: [email protected]
Medication management of depression
Journal of Psychosomatic Research 57 (2004) 565 – 571
Medication management of depression The impact of comorbid chronic medical conditions Paul A. Kurdyak*, William H. Gnam Department of Psychiatry, University of Toronto, Toronto ON, Canada Received 3 February 2003; accepted 26 April 2004
Abstract Objective: This paper addresses the following question: Does quality of care for depression differ between depressed persons with and without chronic medical conditions (CMCs)? Methods: We used a population-based mental health survey to identify respondents aged 18 to 64 with the diagnosis of major depression in the past year (N = 278). In our model, the dependent variable was guideline-level medication management of depression. Determinants for guideline-level care were modeled using multivariate logistic regression. Results: Depressed persons with CMCs were significantly more likely to receive guideline-level care for depression than were the depressed persons without CMCs (OR = 1.46; 95% C.I. = 1.12–1.90). This increased likelihood did
not persist when the sample excluded persons seeing physicians at more than eight visits per year (OR = 0.81; 95% CI = 0.35 – 1.90). Previous psychiatric hospitalization was the only other significant determinant of guideline-level care. Conclusion: Depressed persons with comorbid CMCs are more likely to receive guidelinelevel care for depression than are depressed persons without comorbid CMCs. However, the association did not persist once we excluded respondents who were high utilizers. This finding implies that further understanding of the interaction between depression care and comorbid CMCs will require a longitudinal focus on repeated physician–patient interactions. D 2004 Elsevier Inc. All rights reserved.
Keywords: Depression; Chronic medical illness; Community sample; Quality care guidelines
Introduction Major depression is increasingly viewed as a chronic, recurrent illness resulting in considerable morbidity [1]. The World Health Organization Collaborative Project predicts that by 2020, the disability associated with depression will be second only to ischemic heart disease [2]. Despite a growing awareness of the burden of depression, treatment of depression remains inadequate [3,4]. The development and dissemination of treatment guidelines [5–7] and controlled experiments to improve depression care [8–10] have only resulted in modest improvements in care quality, suggesting that further research into the determinants of depression care is warranted.
* Corresponding author. Health Systems Research and Consulting Unit, ARF Division, Centre for Addiction and Mental Health, 33 Russell Street, Toronto ON, Canada M5S 2S1. Tel.: +1 416 535 8501x6134; fax: +1 416 979 4703. E-mail address: [email protected] (P.A. Kurdyak). 0022-3999/04/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.jpsychores.2004.04.367
Based on community surveys, persons with chronic medical conditions (CMCs) have been demonstrated to have higher rates of depression than persons without CMCs do [11–13]. Persons with CMCs and depression have greater morbidity, reduced quality of life, and greater functional impairment than do persons with CMCs alone [14,15]. The implications for health service delivery are less well understood. Klinkman and others have proposed a model of bcompeting demandsQ to predict the likelihood of receiving treatment for depression in primary care settings [16,17]. These authors suggest that the inadequate detection and treatment of depression in primary care settings are not due to diagnostic difficulty, but rather to a complex set of factors such as patient resistance, somatic manifestations of psychic distress, the short length of physician visits, and multiple competing demands during a single visit for the physician’s attention, among others [17]. This model predicts that primary care physicians are less likely to detect depression as the number of competing demands (such as comorbid medical illnesses) increases. Indeed, there is evidence that any preexisting medical or psychiatric
566
P.A. Kurdyak, W.H. Gnam / Journal of Psychosomatic Research 57 (2004) 565–571
condition reduces the likelihood of any additional conditions being treated [18]. There is further evidence, based upon studies conducted in clinical settings, that competing demands, including CMCs, reduce the likelihood of receiving treatment for depression [19,20]. While these studies provide some support for the competing demands model, they are prone to the selection bias that arises from studying clinical samples. Recent evidence from a community survey indirectly suggests that depressed persons with CMCs may have a higher probability of receiving guideline-level depression care compared with depressed persons without CMCs [21]. However, we are aware of no existing population-based studies that examine this question systematically. The adequacy of depression treatment in a population with comorbid medical conditions is particularly important, given the substantial evidence that adequate treatment of major depression results in lower risks of relapse [22], improvement in quality of life, and restoration of functioning [23,24]. This paper addresses the following questions: (1) In what way do depressed persons with and without comorbid CMCs differ in their utilization of mental health services? (2) Are depressed persons with CMCs more or less likely to receive guideline-level medication management of depression than do depressed people without CMCs? Motivated by the model of competing demands of Klinkman [16], we hypothesized that, despite having a higher probability of any health care utilization, depressed populations with CMCs would have a lower probability of receiving depression care meeting minimum guideline standards compared with a depressed population without CMCs.
Methods Data Data for this analysis were taken from the 1990/1991 Mental Health Supplement (the Supplement) to the Ontario Health Survey (OHS) [25]. The Supplement is an epidemiologic mental health survey of 9953 randomly selected household residents within the Province of Ontario, Canada. The methodology of the Supplement is described in detail elsewhere [26]. Sample selection for the OHS Supplement was based on a stratified, clustered design in which the 42 public health units of the Province were divided into urban and rural strata [27]. Enumeration areas were randomly selected within each stratum, and households were randomly selected within enumeration areas. For the Supplement, one resident aged 15 or older was randomly selected from each participating household. The overall response rate was 67.4%. The prevalence and time of onset of major depression was evaluated using the Composite International Diagnostic Interview, as modified by the University of
Michigan (UM-CIDI), a structured interview designed to be administered by laypersons. The UM-CIDI has been shown to generate valid postinterview mental disorder diagnoses based on DSM-III-R criteria [28]. The results reported in this paper are restricted to respondents aged 18 to 64 years (N = 8116) because the elderly subpopulation did not receive the full diagnostic battery for psychiatric disorders. Our study population included all respondents who suffered from major depression in the past 12 months, resulting in a sample of 314 respondents. We subsequently excluded respondents who met lifetime criteria for bipolar affective disorder, which reduced the study sample to 278 respondents. Variables The outcome of interest for this analysis was quality of care for major depressive disorder. Because the quality of care could not be directly observed, as a proxy, we developed a measure of the adherence to treatment guidelines using survey data. Similar measures have been used in several depression care studies using epidemiological or administrative data sets [4,22]. We classified a survey respondent as receiving guideline-level care if she reported that an antidepressant drug was prescribed and if she had at least four follow-up visits with a medical doctor for monitoring of medication. Restricting the visits criterion to physicians makes sense in the Canadian context because physicians provide the vast majority of services related to antidepressant management and depression care, and because other practitioners (such as specialist outpatient nurse-practitioners) are rare. The Supplement differentiated between visits to a physician for any reason and visits to a physician specifically for mental health-related reasons. For our proxy definition of guideline-level care, only visits related to mental health issues were used. During the administration of the Supplement, respondents were shown a typewritten list of antidepressants. This was designed to increase the likelihood of proper identification of medication. Thus, efforts were made in the design of the survey to distinguish mental health-related and general medical visits to physicians. The primary independent variable in our model is a count variable of self-reported, co-occurring CMCs. A count variable was used to reflect one of the competing demands [16] to depression treatment; hypothetically, the greater number of comorbid CMCs will reduce the likelihood of receiving treatment. The following CMCs were included, based on previous studies of chronic medical illnesses [29,30]: arthritis/rheumatism, stroke, asthma, chronic obstructive pulmonary disease, circulatory problems, heart problems, diabetes, urinary problems, stomach ulcer/digestive problems, eye problems, hypertension, and cancer. As well as the above conditions, epilepsy, low-back pain, and thyroid problems or goitre were also included because of their known association with depression [31–33]. The above
P.A. Kurdyak, W.H. Gnam / Journal of Psychosomatic Research 57 (2004) 565–571
conditions were classified as chronic if the respondent reported that the condition had been present for at least 12 months. Skin allergies or other skin diseases and hay fever or other allergies were excluded. The empirical literature indicates that numerous respondent characteristics will influence the utilization of mental health care and, possibly, the quality of depression care [34–36]. Based upon a literature review and the data available to us, we included demographic factors (age, sex, education, cohabitation status, and income), severity of depressive disorder (previous hospitalizations for depressive disorders, suicide attempts made over a lifetime), and perception of mental health care. The perception of mental health care was derived from a question that asked the respondents what percentage of people with emotional problems benefited from professional help. Studies in Canada have demonstrated that access to specialty care is strongly influenced by income and socioeconomic status [37]. There were 21 missing income variable observations among respondents with major depression. For the missing personal income observations, we imputed a variable for the household income of each respondent in the study population [4].
567
varying the requisite number of follow-up visits from zero to five. Finally, to explore the possibility that the results were affected by depressed respondents receiving psychotherapy (that was not measured directly in the survey), we reestimated the logistic regression model excluding the respondents who reported seeing their family doctor or psychiatrist 12 or more times in the past year for emotional reasons. Twelve visits were chosen because a review of the literature on manualized forms of psychotherapy (such as cognitive therapy or interpersonal therapy) suggests that this represents the minimum number of treatments required to sustain treatment benefit [38,39]. All analyses were conducted using STATA version 6.0 software [40].
Results Table 1 compares the characteristics of depressed respondents with and without CMCs. For most of the sociodemographic variables, the two populations are similar. Respondents with depression and comorbid CMCs were significantly older than were depressed persons without CMCs (39.5 vs. 33.1 years; P b.005) and were more likely
Analysis We separated the population of respondents who reported major depression in the past 12 months into those with CMCs (n = 107) and those without CMCs (n = 171). We compared the means of the dependent and independent variables between these two groups using unpaired t tests. The multivariate model specification used a limited number of independent variables because of the relatively small size of the study population. We used logistic regression to model guideline-level care (prescription of an antidepressant, plus at least four visits with a medical doctor) with the following independent variables: age, gender, cohabitation status, education, income, perception of mental health care, a severity variable, and a count measure of CMCs. Because the respondents were asked to report income in ranges, we created categorical income variables, with the lowest income bracket (US$0–11,999) used as the reference. Education was also recorded as a categorical variable and was coded in a similar manner. To determine whether group differences in adherence to guideline level care were explained simply by a higher probability of receiving any health care, we reestimated the logistic regression model using the conditional sample of depressed persons who had eight or fewer visits with a medical doctor for any reason in the previous year. This limit on visits is arbitrary, but, given the distribution of visits observed, was a natural break point at which to distinguish high versus normal utilization. However, we also conducted sensitivity analyses by making the break point six or seven visits. We also conducted a sensitivity analysis of the dependent variable, guideline-level care for depression, by
Table 1 Demographic and severity measures for the two study samples—depressed persons with and without comorbid chronic medical conditions Major depression and chronic medical condition, n = 107
Major depression and no chronic medical condition, n = 171
39.5 (1.1)*** 72.1 (4.5) 73.8% (4.3) 78.7 (15.3)
33.1 (0.8) 70.2% (3.4) 69.0% (3.5) 76.9 (11.1)
Education High school Some college Degree
7.5% (2.6) 32.7% (4.6) 20.6% (3.9)
5.8% (1.8) 28.1% (3.4) 25.1% (3.3)
Income US$0 – 11,999 US$12,000 – 29,999 US$30,000 – 59,999 US$60,000+
38.3% 34.6% 21.5% 5.6%
36.8% 29.2% 26.3% 7.6%
Age (years) Sex (female) Married/Common-Law Attitude towards mental healtha
Severity of depression Suicide attempt in lifetime Psychiatric hospitalization in life time
(4.7) (4.6) (4.0) (2.2)
29.0% (4.4)*** 23.4% (4.1)
(3.7) (3.5) (3.4) (2.0)
14.7% (2.7) 15.8% (2.8)
Standard error values are presented in brackets. a Attitude toward mental health was determined by using a question that asked respondents what percentage of people who see a professional are helped with their serious emotional problems. *** P b.005.
568
P.A. Kurdyak, W.H. Gnam / Journal of Psychosomatic Research 57 (2004) 565–571
Table 2 Measures of health and mental health service utilization for depressed persons with and without comorbid chronic medical conditions
Saw family doctor in past year for any health problem Mean number of visits for any health problem Saw family doctor in past year for emotional problems Mean number of visits with family doctor in past year for emotional problems Saw psychiatrist in past year for emotional problems Saw other MD in past year for emotional problems Saw any MD in past year for emotional problems Mean number of visits with any MD in past year for emotional problems On an antidepressant in past year Antidepressant from family doctor Antidepressant from psychiatrist On antidepressant and had four or more follow-up visits
Major depression and chronic medical condition
Major depression and no chronic medical condition
98.1% (1.3)***
86.5% (2.6)
10.2 (1.1)*
46.7% (4.8)**
7.1 (0.8)
30.4% (3.5)
6.7 (0.9)
5.5 (1.1)
15.0% (3.5)
15.2% (2.8)
6.5% (2.4)
2.3% (1.2)
52.3% (4.9)**
36.3% (3.7)
10.5 (1.6)
11.5 (2.5)
25.5% (4.3)*
14.4% (2.7)
19.2% (3.9)*** 8.1% (2.8) 20.6% (3.9)**
6.7% (2.0)
with antidepressants was quite low. This suggests that the rate of guideline-level treatment for depression was low for both specialists and primary care physicians. This finding is consistent with previous studies looking at treatment rates in both Canada and the United States [4,41]. In bivariate comparisons (Table 2), significantly more depressed persons with CMCs received an antidepressant in the past year (25.5% vs. 14.4%; P b.05). These respondents were also more likely to receive an antidepressant from their primary care physician (19.2% vs. 6.7%; P b.005). In terms of our outcome variable, more depressed persons with CMCs than depressed persons without CMCs received guideline-level care from any physician (20.6% vs. 9.4%; P b.01). Both groups, however, reported relatively low rates of guideline-level care. The results of the multivariate logistic regression model (Table 3) were consistent with the bivariate comparisons. Depressed persons with CMCs were more likely than depressed persons without CMCs to receive guidelinelevel care (OR = 1.46; 95% CI = 1.12–1.90). To control for the higher likelihood of service utilization among depressed respondents with CMCs, the logistic regression model was reestimated with a sample limited to respondents who had seen a primary care physician in the past year, from one to eight times, for any health reason.
Table 3 Logistic regression model of guideline-level care, with odds ratios for independent variables in the model Variable
Odds ratio
95% confidence interval
Chronic medical illness Age Sex Attitude towards mental healtha Marital status Suicide attempt in lifetime Psychiatric hospitalization in lifetime
1.46** 1.00 1.29 1.00 0.52 0.66 4.48***
1.12 – 1.90 0.96 – 1.04 0.54 – 3.10 0.99 – 1.00 0.23 – 1.16 0.25 – 1.75 1.88 – 10.72
Education Completed primary school Some secondary education Completed secondary school Some postsecondary education Completed postsecondary degree
Reference 0.35 0.78 0.44 0.68
0.08 – 1.54 0.16 – 3.75 0.08 – 2.36 0.15 – 3.12
Income US$0 – 11,999 US$12,000 – 29,999 US$30,000 – 59,999 US$60,000+
Reference 1.52 2.57 3.00
0.57 – 4.03 0.95 – 6.94 0.75 – 12.02
7.9% (2.1) 9.4% (2.2)
* P b.05. ** P b.01. *** P b.005.
to have made a suicide attempt over their lifetime (29.0% vs. 14.7%; P b.005). Table 2 summarizes utilization rates and patterns. Depressed respondents with CMCs were more likely to visit their primary care physicians in the past year for any health problem (98.1% vs. 86.5%; P b.005) and had a higher mean number of visits. A higher proportion of depressed persons with CMCs saw their primary care physicians for emotional problems than depressed persons without CMCs did (46.7% vs. 30.4%; P b.01). In general, psychiatrists treat a minority of patients with depression (Table 2), suggesting that most cases of depression do not get referred to specialists. There was no difference in the percentage of depressed persons who were treated by psychiatrists. In both samples, the percentage of depressed persons visiting a psychiatrist who were treated
The sample being modeled included persons between the ages of 18 and 64 years of age with a diagnosis of major depression within the past year. N = 277; Prob N v 2 = 0.02; Pseudo R 2 = .12. a Attitude toward mental health was determined by using a question that asked respondents what percentage of people who see a professional are helped with their serious emotional problems. ** P b.01. *** P b.005.
P.A. Kurdyak, W.H. Gnam / Journal of Psychosomatic Research 57 (2004) 565–571
Indeed, when the sample was restricted in this way, depressed persons with CMCs were no longer at an increased likelihood of receiving adequate treatment (OR = 0.81; 95% CI = 0.35–1.90). The odds ratios for the other independent variables were substantially unchanged (data not shown). Furthermore, the OR for receiving treatment was similarly reduced when the conditional sample was limited to six visits. A sensitivity analysis of the definition of guideline-level care was performed. The results of the model were robust to variations in the definition of guideline-level care (e.g., receiving an antidepressant alone, or an antidepressant plus one, two, three, or five physician visits within the last year). In addition, we excluded respondents from the baseline analysis who might have been receiving psychotherapy by reestimating the model on those respondents who had seen a primary care physician or psychiatrist 11 or fewer times in the past year. Again, the results of the analysis did not differ substantially from the baseline results (data not shown).
Discussion Our major finding is that, based upon a representative population sample, depressed persons with CMCs are more likely to receive guideline-level antidepressant treatment than do depressed persons without CMCs. The difference is robust to various proxy definitions of guideline-level care and does not appear to be biased by access to psychotherapy as an alternate form of treatment. These findings are poignant reminders that conclusions drawn from clinical samples may not hold for entire populations. However, the increase in likelihood was not robust when we excluded high utilizers (defined as persons who visited physicians more than eight times in the past year). When the sample was limited in this way, comorbid CMCs had no effect on the likelihood of receiving treatment. In contradiction to existing published research, our major finding does not support the model of competing demands. The competing demands hypothesis predicts that as the number of CMCs increases, the likelihood of receiving treatment for depression will decrease. In a comprehensive study of the competing demands hypothesis based upon administrative data [18], not all existing medical conditions were universally associated with a reduced probability of treatment of another condition. For example, persons with diabetes mellitus were more likely to be treated with lipid-lowering agents, as expected from the increased likelihood of hyperlipidemia associated with diabetes mellitus [18]. Psychotic illness was associated with a universal reduction in the likelihood of treatment, no matter what the second condition was, but this could be related to patient factors, such as disorganization, cognitive deficits, or communication barriers, rather than competing demands.
569
As stated above, the primary association was not maintained when high utilizers were excluded. This finding is novel and intriguing. One possibility is that high differ from normal utilizers by higher depression severity or other related characteristics. Another plausible explanation is that a high number of interactions increases the likelihood that numerous (competing) health demands are adequately addressed. The competing demands model does not explicitly account for repeated physician/patient interactions longitudinally, although Klinkman [16] himself suggests that depression might be both recognized and treated over repeated visits, and his model does not account explicitly for such a prediction. Our study has several strengths and limitations. The most significant advantage is that the data are drawn from a representative community survey. As such, the results are not subject to bias commonly found in clinical samples, in which persons utilizing health services are typically more ill. Our study also featured instruments known to reproduce valid DSM-III-R diagnoses linked to extensive demographic data. However, in common with most epidemiological surveys, the data from the Supplement and the OHS are cross-sectional and based on self-report. As with all selfreport data, there is the potential for recall bias. We were also limited in the degree to which we could determine the temporal relationship between depressive disorder, comorbid medical conditions, and utilization of health services because the survey data are cross-sectional. For example, we could not determine the temporal sequence of follow-up visits with respect to the antidepressant prescription. It is possible, with this cross-sectional data, that the bfollow-upQ visits occurred prior to receiving an antidepressant. Finally, the response rate to the Supplement and OHS was 67.4%. This response rate is reasonable for an epidemiologic survey but may nonetheless have an impact on the findings. Low response rates are problematic if the nonresponders were significantly different from the responders. However, an analysis of Supplement responders and nonresponders found little or no difference between these groups on key measures of health status [26]. The UM-CIDI of the Supplement generated a prevalence of depression that is consistent with other reported prevalence rates of depression [42]. This study demonstrates that having a comorbid CMC is associated with a higher likelihood of receiving guidelinelevel care for depression. Based on the universally low treatment rates for depression, our study also demonstrates the need for further research into the factors associated with treatment initiation and access. The reasons why so many people do not receive adequate treatment for depression are multifactorial [21]. However, our findings in respondents with CMCs suggest that further insight into these issues will require attention to the repeated interactions between health care provider and depressed patient over time. Future research should also include the measurement of patient attributes beyond sociodemographic information and diagnosis, e.g., disability as a predictor of depression
570
P.A. Kurdyak, W.H. Gnam / Journal of Psychosomatic Research 57 (2004) 565–571
treatment. To address these complex issues empirically, innovative research designs that employ person-specific longitudinal data linking clinical and administrative sources will be required.
Acknowledgments The authors wish to acknowledge the helpful comments of Drs. Paula Goering and David Streiner, and the members of the Health Systems Research and Consulting Unit at the Centre for Addiction and Mental Health. This study was supported by a SSHRC Institutional Grant. Dr. Paul Kurdyak receives salary support from a Canadian Institutes of Health Research (CIHR) Rx&D/ AstraZenecka/Canadian Psychiatric Research Foundation Fellowship Award. He is also a trainee, and receives support from the Research in Addictions and Mental Health Policy and Services (RAMPS) training program, a CIHR Strategic Training Program.
References [1] Parikh SV, Lam RW, CANMAT Depression Work Group. Clinical guidelines for the treatment of depressive disorders: 1. Definitions, prevalence, and health burden. Can J Psychiatry 2001;46(Supplement 1):13S – 20S. [2] Murray CJ, Lopez AD. Alternative projections of mortality and disability by cause 1990 – 2020: global burden of disease study. Lancet 1997;349:1498 – 504. [3] Lin E, Parikh SV. Sociodemographic, clinical, and attitudinal characteristics of the untreated depressed in Ontario. J Affect Disord 1999;53: 153 – 62. [4] Katz SJ, Kessler RC, Lin E, Wells KB. Medication management of depression in the United States and Ontario. J Gen Intern Med 1998; 13:77 – 85. [5] Depression Guidelines Panel. Depression in primary care: I Detection and diagnosis. Rockville (MD)7 US Department of Health and Human Services, Agency for Health Care Policy and Research, 1993. [6] Reesal RT, Lam RW, CANMAT Depression Work Group. Clinical guidelines for the treatment of depressive disorders: principles of management. Can J Psychiatry 2001;46(Supplement 1):21S – 8S. [7] American Psychiatric Association. Practice guidelines for major depressive disorder in adults. Am J Psychiatry 1993;150(Suppl): S1 – 26. [8] Lin EHB, Simon GE, Katzelnick DJ, Pearson SD. Does physician education on depression management improve treatment in primary care? J Gen Intern Med 2001;16:614 – 9. [9] Schulberg HC, Katon W, Simon GE, Rush J. Treating major depression in primary care practice: an update of the Agency for Health Care Policy and Research Practice Guidelines. Arch Gen Psychiatry 1998;55:1121 – 7. [10] Meredith EL, Jackson-Triche M, Duan N, Rubenstein LV, Camp P, Wells KB. Quality improvement for depression enhances long-term treatment knowledge for primary care clinicians. J Gen Intern Med 2000;15:868 – 77. [11] Patten SB. Long-term medical conditions and major depression in the Canadian population. Can J Psychiatry 1999;44:151 – 7. [12] Patten SB. Long-term medical conditions and major depression in a Canadian population study at waves 1 and 2. J Affect Disord 2000;63: 35 – 41.
[13] Wells KB, Golding JM, Burnam MA. Psychiatric disorder and limitations in physical functioning in a sample of the Los Angeles population. Am J Psychiatry 1988;145:712 – 7. [14] Hays RD, Wells KB, Sherbourne CD, Rogers W, Spritzer K. Functioning and well-being outcomes of patients with depression compared with chronic general medical illnesses. Arch Gen Psychiatry 1995;52:11 – 9. [15] Judd LL, Paulus MP, Wells KB, Rapaport MH. Socioeconomic burden of subsyndromal depressive symptoms and major depression in a sample of the general population. Am J Psychiatry 1996;153(11): 1411 – 7. [16] Klinkman MS. Competing demands in psychosocial care. A model for the identification and treatment of depressive disorders in primary care. Gen Hosp Psychiatry 1997;19(2):98 – 111. [17] Klinkman MS, Schwenk TL, Coyne JC. Depression in primary care— more like asthma than appendicitis: the Michigan depression project. Can J Psychiatry 1997;42(9):966 – 73. [18] Redelmeier DA, Tan SH, Booth GL. The treatment of unrelated disorders in patients with chronic medical diseases. N Engl J Med 1998;338(21):1516 – 20. [19] Rost K, Nutting P, Smith J, Coyne JC, Cooper-Patrick L, Rubenstein L. The role of competing demands in the treatment provided primary care patients with major depression. Arch Fam Med 2000;9(2): 150 – 4. [20] Nutting PA, Rost K, Smith J, Werner JJ, Elliot C. Competing demands from physical problems: effect on initiating and completing depression care over 6 months. Arch Fam Med 2000;9(10): 1059 – 64. [21] Young AS, Klap R, Sherbourne C, Wells KB. The quality of care for depressive and anxiety disorders in the United States. Arch Gen Psychiatry 2001;58:55 – 61. [22] Melfi CA, Chawla AJ, Croghan TW, Hanna MP, Kennedy S, Stredl K. The effects of adherence to antidepressant treatment guidelines on relapse and recurrence of depression. Arch Gen Psychiatry 1998;55: 1128 – 32. [23] Schulberg HC, Block MR, Madonia MJ, Scott CP, Rodriguez E, Imber SD, Perel J, Lave J, Houck PR, Coulehan JL. Treating major depression in primary care practice: eight-month clinical outcomes. Arch Gen Psychiatry 1996;53:913 – 9. [24] Coulehan J, Schulberg HC, Block MR, Madonia MJ, Rodriguez E. Treating depressed primary care patients improves their physical, mental, and social functioning. Arch Intern Med 1997;157:1113 – 20. [25] Ontario Ministry of Health E. Microdata manual: user’s guide, The Ontario health survey. Toronto7 Queen’s Printer for Ontario, 1992. [26] Boyle MH, Offord DR, Campbell D, Catlin G, Goering P, Lin E, Racine YA. Mental health supplement to the Ontario health survey: methodology. Can J Psychiatry 1996;41(9):549 – 58. [27] Statistics Canada. Profile of urban and rural areas: A. Canadian provinces and territories: 1991 census. Ottawa7 Ministry of Industry, Science, and Technology, 1993. [28] Wittchen HU. Reliability and validity studies of the WHO—Composite International Diagnostic Interview (CIDI): a critical review. J Psychiatr Res 1994;28(1):57 – 84. [29] Verbrugge LM, Patrick DL. Seven chronic conditions: their impact on US adults’ activity levels and use of medical services. Am J Public Health 1995;85(2):173 – 82. [30] Tomiak M, Berthelot J-M, Mustard C. A profile of health care utilization of the disabled population in Manitoba. Med Care 1998; 36(9):1383 – 97. [31] Mendez MF, Doss RC, Taylor JL, Salguero P. Depression in epilepsy Relationship to seizures and anticonvulsant therapy. J Nerv Ment Dis 1993;181:444 – 7. [32] Herr KA, Mobily PR, Smith C. Depression and the experience of chronic back pain: a study of related variables and age differences. Clin J Pain 1993;9:104 – 14. [33] Prange AJ, Lipton MA, Nemeroff CB, Wilson IC. The role of hormones in depression. Life Sci 1973;20(8):1305 – 18.
P.A. Kurdyak, W.H. Gnam / Journal of Psychosomatic Research 57 (2004) 565–571 [34] Katz SJ, Kessler RC, Frank RG, Leaf P, Lin E. Mental health care use, morbidity, and socioeconomic status in the United States and Ontario. Inquiry 1997;34(1):38 – 49. [35] Rhodes A, Goering P. Gender differences in the use of outpatient mental health services. J Mental Health Adm 1994;21(4):338 – 46. [36] Leaf PJ, Bruce ML, Tischler GL. The differential effect of attitudes on the use of mental health services. Soc Psychiatry 1986;21(4): 187 – 92. [37] Dunlop S, Coyte PC, McIsaac W. Socioeconomic status and the utilization of physicians’ services: results from the Canadian national population health survey. Soc Sci Med 2000;51:123 – 33.
571
[38] Beck JS. Cognitive therapy: basics and beyond. New York7 The Guilford Press, 1995. [39] Weissman MM, Markowitz JC, Klerman GL. Comprehensive guide to interpersonal psychotherapy. New York7 Basic Books, 2000. [40] Stata statistical software: release 6.0. College Station, TX7 Stata, 1999. [41] Parikh SV, Lin E, Lesage AD. Mental health treatment in Ontario: selected comparisons between the primary care and specialty sectors. Can J Psychiatry 1997;42:929 – 34. [42] Blazer DG, Kessler RC, McGonagle KA, Swartz MS. The prevalence and distribution of major depression in a national community sample: the national comorbidity survey. Am J Psychiatry 1994;151(7):979 – 86.
Medication management of depression The impact of comorbid chronic medical conditions Paul A. Kurdyak*, William H. Gnam Department of Psychiatry, University of Toronto, Toronto ON, Canada Received 3 February 2003; accepted 26 April 2004
Abstract Objective: This paper addresses the following question: Does quality of care for depression differ between depressed persons with and without chronic medical conditions (CMCs)? Methods: We used a population-based mental health survey to identify respondents aged 18 to 64 with the diagnosis of major depression in the past year (N = 278). In our model, the dependent variable was guideline-level medication management of depression. Determinants for guideline-level care were modeled using multivariate logistic regression. Results: Depressed persons with CMCs were significantly more likely to receive guideline-level care for depression than were the depressed persons without CMCs (OR = 1.46; 95% C.I. = 1.12–1.90). This increased likelihood did
not persist when the sample excluded persons seeing physicians at more than eight visits per year (OR = 0.81; 95% CI = 0.35 – 1.90). Previous psychiatric hospitalization was the only other significant determinant of guideline-level care. Conclusion: Depressed persons with comorbid CMCs are more likely to receive guidelinelevel care for depression than are depressed persons without comorbid CMCs. However, the association did not persist once we excluded respondents who were high utilizers. This finding implies that further understanding of the interaction between depression care and comorbid CMCs will require a longitudinal focus on repeated physician–patient interactions. D 2004 Elsevier Inc. All rights reserved.
Keywords: Depression; Chronic medical illness; Community sample; Quality care guidelines
Introduction Major depression is increasingly viewed as a chronic, recurrent illness resulting in considerable morbidity [1]. The World Health Organization Collaborative Project predicts that by 2020, the disability associated with depression will be second only to ischemic heart disease [2]. Despite a growing awareness of the burden of depression, treatment of depression remains inadequate [3,4]. The development and dissemination of treatment guidelines [5–7] and controlled experiments to improve depression care [8–10] have only resulted in modest improvements in care quality, suggesting that further research into the determinants of depression care is warranted.
* Corresponding author. Health Systems Research and Consulting Unit, ARF Division, Centre for Addiction and Mental Health, 33 Russell Street, Toronto ON, Canada M5S 2S1. Tel.: +1 416 535 8501x6134; fax: +1 416 979 4703. E-mail address: [email protected] (P.A. Kurdyak). 0022-3999/04/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.jpsychores.2004.04.367
Based on community surveys, persons with chronic medical conditions (CMCs) have been demonstrated to have higher rates of depression than persons without CMCs do [11–13]. Persons with CMCs and depression have greater morbidity, reduced quality of life, and greater functional impairment than do persons with CMCs alone [14,15]. The implications for health service delivery are less well understood. Klinkman and others have proposed a model of bcompeting demandsQ to predict the likelihood of receiving treatment for depression in primary care settings [16,17]. These authors suggest that the inadequate detection and treatment of depression in primary care settings are not due to diagnostic difficulty, but rather to a complex set of factors such as patient resistance, somatic manifestations of psychic distress, the short length of physician visits, and multiple competing demands during a single visit for the physician’s attention, among others [17]. This model predicts that primary care physicians are less likely to detect depression as the number of competing demands (such as comorbid medical illnesses) increases. Indeed, there is evidence that any preexisting medical or psychiatric
566
P.A. Kurdyak, W.H. Gnam / Journal of Psychosomatic Research 57 (2004) 565–571
condition reduces the likelihood of any additional conditions being treated [18]. There is further evidence, based upon studies conducted in clinical settings, that competing demands, including CMCs, reduce the likelihood of receiving treatment for depression [19,20]. While these studies provide some support for the competing demands model, they are prone to the selection bias that arises from studying clinical samples. Recent evidence from a community survey indirectly suggests that depressed persons with CMCs may have a higher probability of receiving guideline-level depression care compared with depressed persons without CMCs [21]. However, we are aware of no existing population-based studies that examine this question systematically. The adequacy of depression treatment in a population with comorbid medical conditions is particularly important, given the substantial evidence that adequate treatment of major depression results in lower risks of relapse [22], improvement in quality of life, and restoration of functioning [23,24]. This paper addresses the following questions: (1) In what way do depressed persons with and without comorbid CMCs differ in their utilization of mental health services? (2) Are depressed persons with CMCs more or less likely to receive guideline-level medication management of depression than do depressed people without CMCs? Motivated by the model of competing demands of Klinkman [16], we hypothesized that, despite having a higher probability of any health care utilization, depressed populations with CMCs would have a lower probability of receiving depression care meeting minimum guideline standards compared with a depressed population without CMCs.
Methods Data Data for this analysis were taken from the 1990/1991 Mental Health Supplement (the Supplement) to the Ontario Health Survey (OHS) [25]. The Supplement is an epidemiologic mental health survey of 9953 randomly selected household residents within the Province of Ontario, Canada. The methodology of the Supplement is described in detail elsewhere [26]. Sample selection for the OHS Supplement was based on a stratified, clustered design in which the 42 public health units of the Province were divided into urban and rural strata [27]. Enumeration areas were randomly selected within each stratum, and households were randomly selected within enumeration areas. For the Supplement, one resident aged 15 or older was randomly selected from each participating household. The overall response rate was 67.4%. The prevalence and time of onset of major depression was evaluated using the Composite International Diagnostic Interview, as modified by the University of
Michigan (UM-CIDI), a structured interview designed to be administered by laypersons. The UM-CIDI has been shown to generate valid postinterview mental disorder diagnoses based on DSM-III-R criteria [28]. The results reported in this paper are restricted to respondents aged 18 to 64 years (N = 8116) because the elderly subpopulation did not receive the full diagnostic battery for psychiatric disorders. Our study population included all respondents who suffered from major depression in the past 12 months, resulting in a sample of 314 respondents. We subsequently excluded respondents who met lifetime criteria for bipolar affective disorder, which reduced the study sample to 278 respondents. Variables The outcome of interest for this analysis was quality of care for major depressive disorder. Because the quality of care could not be directly observed, as a proxy, we developed a measure of the adherence to treatment guidelines using survey data. Similar measures have been used in several depression care studies using epidemiological or administrative data sets [4,22]. We classified a survey respondent as receiving guideline-level care if she reported that an antidepressant drug was prescribed and if she had at least four follow-up visits with a medical doctor for monitoring of medication. Restricting the visits criterion to physicians makes sense in the Canadian context because physicians provide the vast majority of services related to antidepressant management and depression care, and because other practitioners (such as specialist outpatient nurse-practitioners) are rare. The Supplement differentiated between visits to a physician for any reason and visits to a physician specifically for mental health-related reasons. For our proxy definition of guideline-level care, only visits related to mental health issues were used. During the administration of the Supplement, respondents were shown a typewritten list of antidepressants. This was designed to increase the likelihood of proper identification of medication. Thus, efforts were made in the design of the survey to distinguish mental health-related and general medical visits to physicians. The primary independent variable in our model is a count variable of self-reported, co-occurring CMCs. A count variable was used to reflect one of the competing demands [16] to depression treatment; hypothetically, the greater number of comorbid CMCs will reduce the likelihood of receiving treatment. The following CMCs were included, based on previous studies of chronic medical illnesses [29,30]: arthritis/rheumatism, stroke, asthma, chronic obstructive pulmonary disease, circulatory problems, heart problems, diabetes, urinary problems, stomach ulcer/digestive problems, eye problems, hypertension, and cancer. As well as the above conditions, epilepsy, low-back pain, and thyroid problems or goitre were also included because of their known association with depression [31–33]. The above
P.A. Kurdyak, W.H. Gnam / Journal of Psychosomatic Research 57 (2004) 565–571
conditions were classified as chronic if the respondent reported that the condition had been present for at least 12 months. Skin allergies or other skin diseases and hay fever or other allergies were excluded. The empirical literature indicates that numerous respondent characteristics will influence the utilization of mental health care and, possibly, the quality of depression care [34–36]. Based upon a literature review and the data available to us, we included demographic factors (age, sex, education, cohabitation status, and income), severity of depressive disorder (previous hospitalizations for depressive disorders, suicide attempts made over a lifetime), and perception of mental health care. The perception of mental health care was derived from a question that asked the respondents what percentage of people with emotional problems benefited from professional help. Studies in Canada have demonstrated that access to specialty care is strongly influenced by income and socioeconomic status [37]. There were 21 missing income variable observations among respondents with major depression. For the missing personal income observations, we imputed a variable for the household income of each respondent in the study population [4].
567
varying the requisite number of follow-up visits from zero to five. Finally, to explore the possibility that the results were affected by depressed respondents receiving psychotherapy (that was not measured directly in the survey), we reestimated the logistic regression model excluding the respondents who reported seeing their family doctor or psychiatrist 12 or more times in the past year for emotional reasons. Twelve visits were chosen because a review of the literature on manualized forms of psychotherapy (such as cognitive therapy or interpersonal therapy) suggests that this represents the minimum number of treatments required to sustain treatment benefit [38,39]. All analyses were conducted using STATA version 6.0 software [40].
Results Table 1 compares the characteristics of depressed respondents with and without CMCs. For most of the sociodemographic variables, the two populations are similar. Respondents with depression and comorbid CMCs were significantly older than were depressed persons without CMCs (39.5 vs. 33.1 years; P b.005) and were more likely
Analysis We separated the population of respondents who reported major depression in the past 12 months into those with CMCs (n = 107) and those without CMCs (n = 171). We compared the means of the dependent and independent variables between these two groups using unpaired t tests. The multivariate model specification used a limited number of independent variables because of the relatively small size of the study population. We used logistic regression to model guideline-level care (prescription of an antidepressant, plus at least four visits with a medical doctor) with the following independent variables: age, gender, cohabitation status, education, income, perception of mental health care, a severity variable, and a count measure of CMCs. Because the respondents were asked to report income in ranges, we created categorical income variables, with the lowest income bracket (US$0–11,999) used as the reference. Education was also recorded as a categorical variable and was coded in a similar manner. To determine whether group differences in adherence to guideline level care were explained simply by a higher probability of receiving any health care, we reestimated the logistic regression model using the conditional sample of depressed persons who had eight or fewer visits with a medical doctor for any reason in the previous year. This limit on visits is arbitrary, but, given the distribution of visits observed, was a natural break point at which to distinguish high versus normal utilization. However, we also conducted sensitivity analyses by making the break point six or seven visits. We also conducted a sensitivity analysis of the dependent variable, guideline-level care for depression, by
Table 1 Demographic and severity measures for the two study samples—depressed persons with and without comorbid chronic medical conditions Major depression and chronic medical condition, n = 107
Major depression and no chronic medical condition, n = 171
39.5 (1.1)*** 72.1 (4.5) 73.8% (4.3) 78.7 (15.3)
33.1 (0.8) 70.2% (3.4) 69.0% (3.5) 76.9 (11.1)
Education High school Some college Degree
7.5% (2.6) 32.7% (4.6) 20.6% (3.9)
5.8% (1.8) 28.1% (3.4) 25.1% (3.3)
Income US$0 – 11,999 US$12,000 – 29,999 US$30,000 – 59,999 US$60,000+
38.3% 34.6% 21.5% 5.6%
36.8% 29.2% 26.3% 7.6%
Age (years) Sex (female) Married/Common-Law Attitude towards mental healtha
Severity of depression Suicide attempt in lifetime Psychiatric hospitalization in life time
(4.7) (4.6) (4.0) (2.2)
29.0% (4.4)*** 23.4% (4.1)
(3.7) (3.5) (3.4) (2.0)
14.7% (2.7) 15.8% (2.8)
Standard error values are presented in brackets. a Attitude toward mental health was determined by using a question that asked respondents what percentage of people who see a professional are helped with their serious emotional problems. *** P b.005.
568
P.A. Kurdyak, W.H. Gnam / Journal of Psychosomatic Research 57 (2004) 565–571
Table 2 Measures of health and mental health service utilization for depressed persons with and without comorbid chronic medical conditions
Saw family doctor in past year for any health problem Mean number of visits for any health problem Saw family doctor in past year for emotional problems Mean number of visits with family doctor in past year for emotional problems Saw psychiatrist in past year for emotional problems Saw other MD in past year for emotional problems Saw any MD in past year for emotional problems Mean number of visits with any MD in past year for emotional problems On an antidepressant in past year Antidepressant from family doctor Antidepressant from psychiatrist On antidepressant and had four or more follow-up visits
Major depression and chronic medical condition
Major depression and no chronic medical condition
98.1% (1.3)***
86.5% (2.6)
10.2 (1.1)*
46.7% (4.8)**
7.1 (0.8)
30.4% (3.5)
6.7 (0.9)
5.5 (1.1)
15.0% (3.5)
15.2% (2.8)
6.5% (2.4)
2.3% (1.2)
52.3% (4.9)**
36.3% (3.7)
10.5 (1.6)
11.5 (2.5)
25.5% (4.3)*
14.4% (2.7)
19.2% (3.9)*** 8.1% (2.8) 20.6% (3.9)**
6.7% (2.0)
with antidepressants was quite low. This suggests that the rate of guideline-level treatment for depression was low for both specialists and primary care physicians. This finding is consistent with previous studies looking at treatment rates in both Canada and the United States [4,41]. In bivariate comparisons (Table 2), significantly more depressed persons with CMCs received an antidepressant in the past year (25.5% vs. 14.4%; P b.05). These respondents were also more likely to receive an antidepressant from their primary care physician (19.2% vs. 6.7%; P b.005). In terms of our outcome variable, more depressed persons with CMCs than depressed persons without CMCs received guideline-level care from any physician (20.6% vs. 9.4%; P b.01). Both groups, however, reported relatively low rates of guideline-level care. The results of the multivariate logistic regression model (Table 3) were consistent with the bivariate comparisons. Depressed persons with CMCs were more likely than depressed persons without CMCs to receive guidelinelevel care (OR = 1.46; 95% CI = 1.12–1.90). To control for the higher likelihood of service utilization among depressed respondents with CMCs, the logistic regression model was reestimated with a sample limited to respondents who had seen a primary care physician in the past year, from one to eight times, for any health reason.
Table 3 Logistic regression model of guideline-level care, with odds ratios for independent variables in the model Variable
Odds ratio
95% confidence interval
Chronic medical illness Age Sex Attitude towards mental healtha Marital status Suicide attempt in lifetime Psychiatric hospitalization in lifetime
1.46** 1.00 1.29 1.00 0.52 0.66 4.48***
1.12 – 1.90 0.96 – 1.04 0.54 – 3.10 0.99 – 1.00 0.23 – 1.16 0.25 – 1.75 1.88 – 10.72
Education Completed primary school Some secondary education Completed secondary school Some postsecondary education Completed postsecondary degree
Reference 0.35 0.78 0.44 0.68
0.08 – 1.54 0.16 – 3.75 0.08 – 2.36 0.15 – 3.12
Income US$0 – 11,999 US$12,000 – 29,999 US$30,000 – 59,999 US$60,000+
Reference 1.52 2.57 3.00
0.57 – 4.03 0.95 – 6.94 0.75 – 12.02
7.9% (2.1) 9.4% (2.2)
* P b.05. ** P b.01. *** P b.005.
to have made a suicide attempt over their lifetime (29.0% vs. 14.7%; P b.005). Table 2 summarizes utilization rates and patterns. Depressed respondents with CMCs were more likely to visit their primary care physicians in the past year for any health problem (98.1% vs. 86.5%; P b.005) and had a higher mean number of visits. A higher proportion of depressed persons with CMCs saw their primary care physicians for emotional problems than depressed persons without CMCs did (46.7% vs. 30.4%; P b.01). In general, psychiatrists treat a minority of patients with depression (Table 2), suggesting that most cases of depression do not get referred to specialists. There was no difference in the percentage of depressed persons who were treated by psychiatrists. In both samples, the percentage of depressed persons visiting a psychiatrist who were treated
The sample being modeled included persons between the ages of 18 and 64 years of age with a diagnosis of major depression within the past year. N = 277; Prob N v 2 = 0.02; Pseudo R 2 = .12. a Attitude toward mental health was determined by using a question that asked respondents what percentage of people who see a professional are helped with their serious emotional problems. ** P b.01. *** P b.005.
P.A. Kurdyak, W.H. Gnam / Journal of Psychosomatic Research 57 (2004) 565–571
Indeed, when the sample was restricted in this way, depressed persons with CMCs were no longer at an increased likelihood of receiving adequate treatment (OR = 0.81; 95% CI = 0.35–1.90). The odds ratios for the other independent variables were substantially unchanged (data not shown). Furthermore, the OR for receiving treatment was similarly reduced when the conditional sample was limited to six visits. A sensitivity analysis of the definition of guideline-level care was performed. The results of the model were robust to variations in the definition of guideline-level care (e.g., receiving an antidepressant alone, or an antidepressant plus one, two, three, or five physician visits within the last year). In addition, we excluded respondents from the baseline analysis who might have been receiving psychotherapy by reestimating the model on those respondents who had seen a primary care physician or psychiatrist 11 or fewer times in the past year. Again, the results of the analysis did not differ substantially from the baseline results (data not shown).
Discussion Our major finding is that, based upon a representative population sample, depressed persons with CMCs are more likely to receive guideline-level antidepressant treatment than do depressed persons without CMCs. The difference is robust to various proxy definitions of guideline-level care and does not appear to be biased by access to psychotherapy as an alternate form of treatment. These findings are poignant reminders that conclusions drawn from clinical samples may not hold for entire populations. However, the increase in likelihood was not robust when we excluded high utilizers (defined as persons who visited physicians more than eight times in the past year). When the sample was limited in this way, comorbid CMCs had no effect on the likelihood of receiving treatment. In contradiction to existing published research, our major finding does not support the model of competing demands. The competing demands hypothesis predicts that as the number of CMCs increases, the likelihood of receiving treatment for depression will decrease. In a comprehensive study of the competing demands hypothesis based upon administrative data [18], not all existing medical conditions were universally associated with a reduced probability of treatment of another condition. For example, persons with diabetes mellitus were more likely to be treated with lipid-lowering agents, as expected from the increased likelihood of hyperlipidemia associated with diabetes mellitus [18]. Psychotic illness was associated with a universal reduction in the likelihood of treatment, no matter what the second condition was, but this could be related to patient factors, such as disorganization, cognitive deficits, or communication barriers, rather than competing demands.
569
As stated above, the primary association was not maintained when high utilizers were excluded. This finding is novel and intriguing. One possibility is that high differ from normal utilizers by higher depression severity or other related characteristics. Another plausible explanation is that a high number of interactions increases the likelihood that numerous (competing) health demands are adequately addressed. The competing demands model does not explicitly account for repeated physician/patient interactions longitudinally, although Klinkman [16] himself suggests that depression might be both recognized and treated over repeated visits, and his model does not account explicitly for such a prediction. Our study has several strengths and limitations. The most significant advantage is that the data are drawn from a representative community survey. As such, the results are not subject to bias commonly found in clinical samples, in which persons utilizing health services are typically more ill. Our study also featured instruments known to reproduce valid DSM-III-R diagnoses linked to extensive demographic data. However, in common with most epidemiological surveys, the data from the Supplement and the OHS are cross-sectional and based on self-report. As with all selfreport data, there is the potential for recall bias. We were also limited in the degree to which we could determine the temporal relationship between depressive disorder, comorbid medical conditions, and utilization of health services because the survey data are cross-sectional. For example, we could not determine the temporal sequence of follow-up visits with respect to the antidepressant prescription. It is possible, with this cross-sectional data, that the bfollow-upQ visits occurred prior to receiving an antidepressant. Finally, the response rate to the Supplement and OHS was 67.4%. This response rate is reasonable for an epidemiologic survey but may nonetheless have an impact on the findings. Low response rates are problematic if the nonresponders were significantly different from the responders. However, an analysis of Supplement responders and nonresponders found little or no difference between these groups on key measures of health status [26]. The UM-CIDI of the Supplement generated a prevalence of depression that is consistent with other reported prevalence rates of depression [42]. This study demonstrates that having a comorbid CMC is associated with a higher likelihood of receiving guidelinelevel care for depression. Based on the universally low treatment rates for depression, our study also demonstrates the need for further research into the factors associated with treatment initiation and access. The reasons why so many people do not receive adequate treatment for depression are multifactorial [21]. However, our findings in respondents with CMCs suggest that further insight into these issues will require attention to the repeated interactions between health care provider and depressed patient over time. Future research should also include the measurement of patient attributes beyond sociodemographic information and diagnosis, e.g., disability as a predictor of depression
570
P.A. Kurdyak, W.H. Gnam / Journal of Psychosomatic Research 57 (2004) 565–571
treatment. To address these complex issues empirically, innovative research designs that employ person-specific longitudinal data linking clinical and administrative sources will be required.
Acknowledgments The authors wish to acknowledge the helpful comments of Drs. Paula Goering and David Streiner, and the members of the Health Systems Research and Consulting Unit at the Centre for Addiction and Mental Health. This study was supported by a SSHRC Institutional Grant. Dr. Paul Kurdyak receives salary support from a Canadian Institutes of Health Research (CIHR) Rx&D/ AstraZenecka/Canadian Psychiatric Research Foundation Fellowship Award. He is also a trainee, and receives support from the Research in Addictions and Mental Health Policy and Services (RAMPS) training program, a CIHR Strategic Training Program.
References [1] Parikh SV, Lam RW, CANMAT Depression Work Group. Clinical guidelines for the treatment of depressive disorders: 1. Definitions, prevalence, and health burden. Can J Psychiatry 2001;46(Supplement 1):13S – 20S. [2] Murray CJ, Lopez AD. Alternative projections of mortality and disability by cause 1990 – 2020: global burden of disease study. Lancet 1997;349:1498 – 504. [3] Lin E, Parikh SV. Sociodemographic, clinical, and attitudinal characteristics of the untreated depressed in Ontario. J Affect Disord 1999;53: 153 – 62. [4] Katz SJ, Kessler RC, Lin E, Wells KB. Medication management of depression in the United States and Ontario. J Gen Intern Med 1998; 13:77 – 85. [5] Depression Guidelines Panel. Depression in primary care: I Detection and diagnosis. Rockville (MD)7 US Department of Health and Human Services, Agency for Health Care Policy and Research, 1993. [6] Reesal RT, Lam RW, CANMAT Depression Work Group. Clinical guidelines for the treatment of depressive disorders: principles of management. Can J Psychiatry 2001;46(Supplement 1):21S – 8S. [7] American Psychiatric Association. Practice guidelines for major depressive disorder in adults. Am J Psychiatry 1993;150(Suppl): S1 – 26. [8] Lin EHB, Simon GE, Katzelnick DJ, Pearson SD. Does physician education on depression management improve treatment in primary care? J Gen Intern Med 2001;16:614 – 9. [9] Schulberg HC, Katon W, Simon GE, Rush J. Treating major depression in primary care practice: an update of the Agency for Health Care Policy and Research Practice Guidelines. Arch Gen Psychiatry 1998;55:1121 – 7. [10] Meredith EL, Jackson-Triche M, Duan N, Rubenstein LV, Camp P, Wells KB. Quality improvement for depression enhances long-term treatment knowledge for primary care clinicians. J Gen Intern Med 2000;15:868 – 77. [11] Patten SB. Long-term medical conditions and major depression in the Canadian population. Can J Psychiatry 1999;44:151 – 7. [12] Patten SB. Long-term medical conditions and major depression in a Canadian population study at waves 1 and 2. J Affect Disord 2000;63: 35 – 41.
[13] Wells KB, Golding JM, Burnam MA. Psychiatric disorder and limitations in physical functioning in a sample of the Los Angeles population. Am J Psychiatry 1988;145:712 – 7. [14] Hays RD, Wells KB, Sherbourne CD, Rogers W, Spritzer K. Functioning and well-being outcomes of patients with depression compared with chronic general medical illnesses. Arch Gen Psychiatry 1995;52:11 – 9. [15] Judd LL, Paulus MP, Wells KB, Rapaport MH. Socioeconomic burden of subsyndromal depressive symptoms and major depression in a sample of the general population. Am J Psychiatry 1996;153(11): 1411 – 7. [16] Klinkman MS. Competing demands in psychosocial care. A model for the identification and treatment of depressive disorders in primary care. Gen Hosp Psychiatry 1997;19(2):98 – 111. [17] Klinkman MS, Schwenk TL, Coyne JC. Depression in primary care— more like asthma than appendicitis: the Michigan depression project. Can J Psychiatry 1997;42(9):966 – 73. [18] Redelmeier DA, Tan SH, Booth GL. The treatment of unrelated disorders in patients with chronic medical diseases. N Engl J Med 1998;338(21):1516 – 20. [19] Rost K, Nutting P, Smith J, Coyne JC, Cooper-Patrick L, Rubenstein L. The role of competing demands in the treatment provided primary care patients with major depression. Arch Fam Med 2000;9(2): 150 – 4. [20] Nutting PA, Rost K, Smith J, Werner JJ, Elliot C. Competing demands from physical problems: effect on initiating and completing depression care over 6 months. Arch Fam Med 2000;9(10): 1059 – 64. [21] Young AS, Klap R, Sherbourne C, Wells KB. The quality of care for depressive and anxiety disorders in the United States. Arch Gen Psychiatry 2001;58:55 – 61. [22] Melfi CA, Chawla AJ, Croghan TW, Hanna MP, Kennedy S, Stredl K. The effects of adherence to antidepressant treatment guidelines on relapse and recurrence of depression. Arch Gen Psychiatry 1998;55: 1128 – 32. [23] Schulberg HC, Block MR, Madonia MJ, Scott CP, Rodriguez E, Imber SD, Perel J, Lave J, Houck PR, Coulehan JL. Treating major depression in primary care practice: eight-month clinical outcomes. Arch Gen Psychiatry 1996;53:913 – 9. [24] Coulehan J, Schulberg HC, Block MR, Madonia MJ, Rodriguez E. Treating depressed primary care patients improves their physical, mental, and social functioning. Arch Intern Med 1997;157:1113 – 20. [25] Ontario Ministry of Health E. Microdata manual: user’s guide, The Ontario health survey. Toronto7 Queen’s Printer for Ontario, 1992. [26] Boyle MH, Offord DR, Campbell D, Catlin G, Goering P, Lin E, Racine YA. Mental health supplement to the Ontario health survey: methodology. Can J Psychiatry 1996;41(9):549 – 58. [27] Statistics Canada. Profile of urban and rural areas: A. Canadian provinces and territories: 1991 census. Ottawa7 Ministry of Industry, Science, and Technology, 1993. [28] Wittchen HU. Reliability and validity studies of the WHO—Composite International Diagnostic Interview (CIDI): a critical review. J Psychiatr Res 1994;28(1):57 – 84. [29] Verbrugge LM, Patrick DL. Seven chronic conditions: their impact on US adults’ activity levels and use of medical services. Am J Public Health 1995;85(2):173 – 82. [30] Tomiak M, Berthelot J-M, Mustard C. A profile of health care utilization of the disabled population in Manitoba. Med Care 1998; 36(9):1383 – 97. [31] Mendez MF, Doss RC, Taylor JL, Salguero P. Depression in epilepsy Relationship to seizures and anticonvulsant therapy. J Nerv Ment Dis 1993;181:444 – 7. [32] Herr KA, Mobily PR, Smith C. Depression and the experience of chronic back pain: a study of related variables and age differences. Clin J Pain 1993;9:104 – 14. [33] Prange AJ, Lipton MA, Nemeroff CB, Wilson IC. The role of hormones in depression. Life Sci 1973;20(8):1305 – 18.
P.A. Kurdyak, W.H. Gnam / Journal of Psychosomatic Research 57 (2004) 565–571 [34] Katz SJ, Kessler RC, Frank RG, Leaf P, Lin E. Mental health care use, morbidity, and socioeconomic status in the United States and Ontario. Inquiry 1997;34(1):38 – 49. [35] Rhodes A, Goering P. Gender differences in the use of outpatient mental health services. J Mental Health Adm 1994;21(4):338 – 46. [36] Leaf PJ, Bruce ML, Tischler GL. The differential effect of attitudes on the use of mental health services. Soc Psychiatry 1986;21(4): 187 – 92. [37] Dunlop S, Coyte PC, McIsaac W. Socioeconomic status and the utilization of physicians’ services: results from the Canadian national population health survey. Soc Sci Med 2000;51:123 – 33.
571
[38] Beck JS. Cognitive therapy: basics and beyond. New York7 The Guilford Press, 1995. [39] Weissman MM, Markowitz JC, Klerman GL. Comprehensive guide to interpersonal psychotherapy. New York7 Basic Books, 2000. [40] Stata statistical software: release 6.0. College Station, TX7 Stata, 1999. [41] Parikh SV, Lin E, Lesage AD. Mental health treatment in Ontario: selected comparisons between the primary care and specialty sectors. Can J Psychiatry 1997;42:929 – 34. [42] Blazer DG, Kessler RC, McGonagle KA, Swartz MS. The prevalence and distribution of major depression in a national community sample: the national comorbidity survey. Am J Psychiatry 1994;151(7):979 – 86.