Journal of Psychiatric Research 52 (2014) 21e27
Contents lists available at ScienceDirect
Journal of Psychiatric Research journal homepage: www.elsevier.com/locate/psychires
Depression, deficits in functional capacity, and impaired glycemic control in urban African Americans with type 2 diabetes Dominique L. Musselman a, *, David C. Ziemer b, Marcia D. McNutt c, Julia S. Seay c, Erica B. Royster d, Bridget Larsen d, Terrika Barham d, Angelo R. Brown d, Octavia L. Vogel d, Lawrence S. Phillips b, Philip D. Harvey a a
Department of Psychiatry and Behavioral Sciences, University of Miami Leonard Miller School of Medicine, Miami, FL, USA Department of Medicine, Division of Endocrinology, Emory University School of Medicine, Atlanta, GA, USA University of Miami, Department of Psychology, Miami, FL, USA d Department of Psychiatry and Behavioral Sciences, Emory University School of Medicine, Atlanta, GA, USA b c
a r t i c l e i n f o
a b s t r a c t
Article history: Received 20 September 2013 Received in revised form 11 December 2013 Accepted 9 January 2014
Background: Effective depression treatment does not reliably reduce glycosylated hemoglobin (HbA1c) in depressed patients with type 2 diabetes, possibly in part due to deficits in functional capacity, i.e. performance of certain everyday living skills, essential for effective diabetes self-management. We sought to determine: a) the magnitude of deficits in functional capacity among urban, African American (AA) patients with type 2 diabetes, and b) whether these deficits were associated with poorer glycemic control. Methods: At their initial visit to an inner-city diabetes clinic, 172 AA patients with type 2 diabetes were assessed with a variety of instruments, including the Mini International Neuropsychiatric Interview (MINI) and the UCSD Performance Skills Assessment-Brief (UPSA-B). They then entered a comprehensive diabetes management intervention, whose success was indexed by HbA1c levels at up to four reassessments over a one-year period. A mixed-effects model repeated-measures method was used to predict HbA1c. Results: The prevalence of depression was 19%; the mean UPSA-B score was 81 17. After multivariate adjustment, increased HbA1c levels over time were predicted by the presence of major depression (B ¼ .911, p ¼ .002) and decreasing (worse) scores on the UPSA-B (B ¼ .016, p ¼ .027), respectively. Further adjustment for increasing the dosage of oral or insulin during the treatment eliminated the association between the UPSA score and HbA1c level (B ¼ .010, p ¼ .115). Conclusions: Depression, as well as deficits in functional capacity, predicted reduced effectiveness of a diabetes self-management intervention. Future studies will determine whether interventions targeted at both improve glycemic control. Ó 2014 Elsevier Ltd. All rights reserved.
Keywords: African Americans Type 2 diabetes Functional capacity HbA1c UPSA-B Major depression Neurocognition Glycemic control MINI structured diagnostic interview Zung Depression Rating Scale
1. Introduction African-Americans (AA) suffer from an increased prevalence of type 2 diabetes (Centers for Disease Control, 1993), and comorbid depression (Gary et al., 2000; Gilles et al., 2006). Irrespective of race, the presence of depression in patients with type 2 diabetes has been associated with acceleration of cognitive decline (Sullivan
* Corresponding author. University of Miami Leonard H. Miller School of Medicine, Department of Psychiatry and Behavioral Sciences, Mental Health Hospital Center, 1695 NW 9th Avenue, Rm 3308L, Miami, FL 33146, USA. Tel.: þ1 404 723 8361; fax: þ1 305 355 9072. E-mail address: [email protected] (D.L. Musselman). 0022-3956/$ e see front matter Ó 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jpsychires.2014.01.006
et al., 2013) as well as poorer compliance with medications which can potentially slow cognitive decline, i.e. antihypertensive drugs (Poon, 2008), aspirin (Kern et al., 2012), and statins (Steenland et al., 2013). Furthermore, in patients with diabetes, neurocognitive function can be impaired by agents with anticholinergic effect (Campbell et al., 2010): prescribed for neuropathic pain (e.g. amitriptyline), hypertension (e.g. atenolol). Less well examined are the consequences of neurocognitive deficits upon adherence to often complex diabetes treatment regimens (Awad et al., 2004; Kodl and Seaquist, 2008). Nevertheless, a rapidly accumulating database indicates that neurocognitive deficits correlate strongly with deficits in functional capacity (Moore et al., 2007), that is, the inability to perform critical daily skills, such as communication, manage medications, and handle finances (r ¼ .60e.65) (Leifker
22
D.L. Musselman et al. / Journal of Psychiatric Research 52 (2014) 21e27
et al., 2011), in patients ranging from those with HIV infection (Heaton et al., 2004) to Alzheimer’s disease (Gomar et al., 2011), or schizophrenia (Light et al., 2012; Keefe et al., 2011), as well as older healthy individuals (Leifker et al., 2011). Indeed, deficits in functional skills, when quantified in the laboratory, also predict “real world” outcome, i.e. the capability of patients with schizophrenia and bipolar disorder to live independently and sustain employment (Auslander et al., 2001; Mausbach et al., 2011; Twamley et al., 2002; Bowie et al., 2008, 2010). Of note is that functional capacity has not been previously examined in patients with type 2 diabetes, nor has its impact on their disease self-management, as reflected by another “real world” measurement, HbA1c. Though other psychiatric comorbidities such as anxiety and addictive disorders are common in patients with type 2 diabetes, depression may be most prevalent (Gilles et al., 2006). Thus, we sought to determine among urban, AA patients with type 2 diabetes: a) the prevalence and magnitude of depression, b) as well as deficits in functional capacity, and then c) examine whether depression, or difficulties in performing critical functional skills “in the lab”, would be associated with magnitude of HbA1c at up to one year of follow-up. 2. Method 2.1. Participants The Grady Health System Diabetes Clinic population is predominantly African American, with high rates of retinopathy, nephropathy, poverty and poor functional health literacy (Cook et al., 1999). Fewer than 10% of newly presenting patients have commercial health insurance; approximately 50% have no third party health care coverage (Ziemer et al., 1996). Patients may self-refer or be referred from their primary care or other specialty providers for treatment. Those diagnosed with type 2 diabetes fulfill American Diabetes Association (ADA) guidelines (The Expert Committee on the Diagnosis and Classification of Diabetes Mellitus, 2003): symptoms of diabetes (polyuria, polydipsia, unexplained weight loss) and a plasma glucose concentration 200 mg/dl, or a fasting plasma glucose of 126 mg/dl, or a 2-hour plasma concentration 200 mg/dl during an oral glucose tolerance test using an equivalent of 75 g anhydrous glucose dissolved in water. 2.2. Study design This prospective study was a sub-study of a larger, longitudinal, prospective examination of depression and glycemic control in urban African Americans with type 2 diabetes. Patients new to the Clinic were evaluated during a comprehensive intake appointment, including a medical history, physical exam, retinal photographs, and laboratory testing. Patients fulfilling study inclusion criteria (African American, English speaking, medical confirmation of diabetes, age 18e80 years, and to initiate diabetes care in the Clinic) were offered participation in the parent study beginning January 1, 2004 and ending March 30th, 2008. The convenience sample of the sub-study was recruited from August 2007 to March 2008. AfricanAmerican ancestry was confirmed through self-report. This study was approved by the Emory University Institutional Review Board and the Grady Research Oversight Committee. Written informed consent was obtained. 2.3. Study assessment After providing consent, the 30e45 min diagnostic psychiatric evaluation was administered only once, during the 6e8 h initial clinic visit at the Clinic.
The psychiatric diagnostic evaluation included the Mini International Neuropsychiatric Interview (MINI), a brief, structured, observer-administered questionnaire surveying for the presence of psychiatric disorders classified according to DSM-IV during the prior 12 months (Sheehan et al., 2005); magnitude of depressive symptoms were assessed using the 20-item, dimensional, selfreport Zung Depression Rating Scale (Zung, 1965). Another dimensional screening tool, the Traumatic Events Interview (TEI) (Schwartz et al., 2005), was used to detect past episodes of trauma and early life stress (ELS), defined by physical or sexual abuse at the age of 13 or younger. Participants received $15 compensation for the diagnostic evaluation. Those who wished to participate in the sub-study, received another $15 compensation to undergo the 15min functional capacity assessment consisting of the University of California San Diego Performance Skills Assessment-Brief (UPSA-B) (Sheehan et al., 2005; Mausbach et al., 2007), a performance-based, laboratory assessment of financial (i.e. make change and write a check to pay a bill) and communication (i.e. use a telephone to change a medical appointment) skills, with demonstrated validity regarding concurrent residential status and community engagement in persons with psychotic disorders (Mausbach et al., 2009). The psychometric properties of the UPSA-B have been examined, comparing healthy controls (n ¼ 109; mean age 68 12; mean UPSA-B score: 85 9) vs. patients with schizophrenia (n ¼ 238; mean age 57 9; mean UPSA-B score: 69 20), yielding a testretest reliability of the UPSA-B for that sample of patients of .75 and .73 at 18- and 36-months, respectively (Leifker et al., 2010). Sociodemographic and diabetes-related characteristics were obtained from patient self-report, the Grady Diabetes Clinic Patient Tracking System (DPTS), and the Grady Pharmacy database, including duration of diabetes, type of hypoglycemic treatment, and level of HbA1c. Prescription of medications for comorbid medical disorders was measured using the well-validated Chronic Disease Score (CDS) method by Von Korff and colleagues (Clark et al., 1995) in order to index general burden of illness. Information regarding income was gleaned from the Grady Health System electronic database. 2.4. Referral for depression treatment Patients who fulfilled diagnostic criteria for major depressive disorder were informed of their diagnosis, and received referral in verbal and/or written form for treatment within the Grady Healthcare System, their local mental health center, and other community resources. 2.5. Diabetes Management Education Intervention Patients were then scheduled return appointments to the Grady Diabetes Clinic every three months for their diabetes treatment involving the Diabetes Management Education Intervention: three educational classes, two designed to give information on diabetes, and another focused on diet and nutrition (Ziemer et al., 1996). Combined, the classes provided 5e6 h of diabetes education and were taught by either a certified diabetes educator or a dietician. Topics included the signs and symptoms of hypo- and hyperglycemia, the “ABCs” of diabetes, including what a HbA1c is and its importance, support group resources, and the importance of ADA goals and goal setting, the importance of self-management activities including medication compliance, nutrition and diet, physical exercise, stress management, how to perform the blood glucose test, maintenance of personal logs tracking glucose levels and other important medical indices, and recognition of patterns of their glucose levels. Clinic patients were given the opportunity to ask questions during sessions and were administered pre-, and
D.L. Musselman et al. / Journal of Psychiatric Research 52 (2014) 21e27
post-test, assessments of their diabetes knowledge. At their followup visits and during any telephone communications that might occur between visits, the health care providers reinforced diabetes information to the patient. Blood pressure and cholesterol levels, as well as HbA1c results, and goals were also discussed with the patient during follow-up visits. In addition, during these visits, antidiabetic and antihypertensive medications were adjusted as necessary, and patients receive foot and dilated eye exams. 2.6. Measurement of HbA1c Following the National Glycohemoglobin Standardization Program, HbA1c of the study participants was measured using a turbidometric, immuno-inhibition assay that recognizes glycated A, S and C hemoglobins equally (Beckman Synchron, Fullerton, CA) and shows no cross-reactivity with HbA0, HbA1a, HbA1b, acetylated Hb, carbamylated Hb, glycated albumin, HbA1d, or an acetylaldehyde Hb adduct. Total hemoglobin was measured using cell lysate and absorption at 560 nm. These assays were performed in the Grady Health System Clinical Laboratory, a College of American Pathology (CAP) accredited and licensed by the Georgia Department of Human Resources (DHR).
23
Table 1 Sociodemographic and medical characteristics of study population at intake. Characteristics
Total (n ¼ 172)
Age (years) (mean þ SD) Female Gender n (%) UPSA Score (mean þ SD) UPSA Communication Skills domain (mean þ SD) UPSA Financial Skills domain (mean þ SD) Chronic Disease Score (mean þ SD) Major Depressive Disorder-current n (%) Zung Depression Rating Score (mean þ SD)a Early Life Stressb e n (%) Anxiety Disordersc e n (%) Addictive Disordersd e n (%) Diabetes Duration (years) (median) Baseline HbA1c (%) (mean þ SD) Anti-diabetic Regimen n (%) None Oral agents alone Insulin alone or with oral agents
50 (10) 107 (62) 81 þ 17 40 þ 11 42 þ 8 3016 þ 1666 33 (19) 46 þ 13 25 (14) 24 (14) 23 (13) 2.9 8.9 þ 2.7 17 (10) 84 (49) 71 (41)
a Zung score: 20e49 (normal). 50e59 (mildly depressed). 60e69 (moderately depressed), 70þ (severely depressed). b Early life stress ¼ physical or sexual abuse at the age of 13 or below. c Anxiety Disorders ¼ panic disorder, agoraphobia, social phobia, post-traumatic stress disorder, obsessive compulsive disorder, generalized anxiety disorder. d Addictive disorders alcohol and/or drug abuse, alcohol and/or drug dependence.
2.7. Planned analyses A mixed-effects model repeated-measures method (MMRM) examined the course of HbA1c during the 12-month intervention, given that repeat assessments were scheduled at regular intervals, but actual occurrences of these assessments were somewhat variable. Certain variables were measured at baseline only: age, sex, HbA1c, body mass index (BMI), duration of diabetes, presence of major depression, general illness burden, and performance on the UPSA-B. These variables, well as number of subsequent clinic visits, and initiation of oral or insulin treatment, were used as covariates to predict the course of HbA1c over the follow-up. As HbA1c was not normally distributed, analyses were performed using generalized mixed models. As is the case with MMRM, baseline HbA1c was adjusted for, all available observations were used, and different numbers of participants were examined at each examination (Fig. 1). Specific analyses examined the level of HbA1c during 1 year of follow-up, and associated covariables, with the second part of the two-step analysis examining whether intensification of dose (of an
existing oral hypoglycemic or insulin) was associated with magnitude of HbA1c. Lastly, a “cut-off” score of the UPSA-B was explored for level of HbA1c during the year of follow-up (Mausbach et al., 2009). 3. Results Of 4341 consecutive, newly presenting African American patients to the clinic, 2112 (49% participation rate) provided consent for the parent study. Sixty-eight patients did not fulfill inclusion criteria. Other eligible patients refused (n ¼ 220), and others could not be scheduled for a research appointment in a timely manner (n ¼ 112). Two patients did not complete the USPA-B and were excluded; another 5 patients did not have a contemporaneous HbA1c drawn at clinic intake, resulting in 172 patients with both an UPSA-B score and a HbA1c at baseline. Of the patients who refused to participate (n ¼ 1276), certain of their socio-ecologic and medical characteristics were statistically
Fig. 1. Correlations between UPSA and Zung (panel A, left), and UPSA and HbA1c (panel B, right).
24
D.L. Musselman et al. / Journal of Psychiatric Research 52 (2014) 21e27
different, but clinically similar, to those who did participate. The mean HbA1c of nonparticipants was 8.4% vs. 8.7% (p ¼ .04) of study participants; a clinically insignificant difference given that a difference of .5% is predictive of microvascular complications (The Diabetes Control and Complications Trial Research Group, 1993). The mean body mass index (BMI) of participants vs. nonparticipants was 33 vs. 32 (p ¼ .001), respectively. The mean albumin/creatinine ratio of nonparticipants was worse (statistically) compared to participants, i.e. .11 vs. .07 mg/mg (p ¼ .02). The mean diastolic blood pressure of participants was 73 mm Hg vs. 75 mm Hg of nonparticipants (p ¼ .03). The relatively good control of both reflected, at least in part, initiation of antihypertensive interventions by Grady primary care providers prior to patient referral to the Diabetes Clinic. Table 1 shows that the African American type 2 diabetic patients at intake who participated in the sub-study were on average 50 years old, with 62% being female. Most had a high school degree or less (60%), and most were unmarried (83%). Fifty-nine percent had an UPSA-B score < 80. With an average BMI of 36, the participants of this study were generally “obese”; their average LDL cholesterol was greater than 100 mg/dl. In contrast, their systolic blood pressure averaged 127 mm Hg, a magnitude of relatively good control. Of the 77% of study participants prescribed antihypertensive medication, they were prescribed an average of 2.4 antihypertensive medications each. Thirty-four percent were receiving antidepressant treatment; 11 of these 61 patients were receiving either the tricyclic antidepressant amitriptyline or nortriptyline in low dosage for diabetic neuropathy. At baseline, the correlation between the UPSA-B score and HbA1c was significant (r ¼ .106, p ¼ .021), but not between the UPSA-B score and Zung depression scale score (r ¼ .008, p ¼ .916) (Fig. 1). The UPSA-B score was not correlated with age (r ¼ .12; p ¼ .10), Chronic Disease Score (CDS) (r ¼ .06, p ¼ .42) or Zung score (r ¼ .04; p ¼ . 57), but did correlate with lifetime educational attainment (r ¼ .33; p < .001). Over the course of their one-year follow-up, participants attended an average of 5.5 4 clinic visits. Factors at baseline potentially associated with the likelihood of follow-up during the subsequent year were examined. For the most part, baseline medical characteristics, i.e. age, gender, duration of diabetes, LDL, BMI and CDS, were not different between the patients who did, and did not, have a follow-up clinic visit, except that patients with a return visit exhibited a lower baseline systolic blood pressure vs. those who did not return to the clinic (p ¼ .002). Persons also more likely to return the clinic were those treated with insulin (p ¼ .035). Thirty-four (or 20%) changed from diet to oral hypoglycemic; 22 (or 13%) began insulin in addition to their pre-existing oral hypoglycemic medications. The number of return visits to the Diabetes Clinic was associated with greater CDS scores (r ¼ .267, p ¼ .000), as well as magnitude of the UPSA-B score (r ¼ .171, p ¼ .000). At baseline, 36% of the patients were found to have a HbA1c of less than 7.0% (American Diabetes Association, 1995), while 53% of the patients reached this ADA goal at their endpoint. Thus, 17% of the patients demonstrated a clinically significant improvement in HbA1c. Overall, the mean HbA1c decreased to 7.5 2.1% (Fig. 2). Mixed model analyses revealed a statistically significant time effect, reflecting significant overall improvement in HbA1c over the intervention (Table 2). Multiple significant covariate effects affected the course of HbA1c. Those with an increasing number of visits to the Clinic, those with more co-existing medical disorders, having a longer duration of diabetes, younger age, and those undergoing an intensification of treatment, either starting an oral hypoglycemic agent or beginning insulin therapy, exhibited an increased HbA1c
Fig. 2. Mean haemoglobin A1c (%) at baseline and reassessment visits during 1-year follow-up.
during the subsequent 12 months. Having current major depression (B ¼ .911, p ¼ .002), and having a lower UPSA-B score (B ¼ .016, p ¼ .027), was also associated with higher HbA1c, while sex, education, baseline BMI, and blood pressure were not. In the second step, when further intensification of dosage of an existing oral hypoglycemic or insulin was added to the model, having a lower UPSA-B score was not associated with severity of HbA1c during the year of follow-up (B ¼ .010, p ¼ .115). Furthermore, neither UPSA communication (p ¼ .15) or financial (p ¼ .14) domain scores were significantly associated with magnitude of HbA1c during the year of follow-up. Based on a prior report by Mausbach, we explored whether the cut-off score of 60 would be associated with HbA1c levels. At baseline, the HbA1c scores of the persons with UPSA scores (0e59 vs. 60þ) did not significantly differ [t (170) ¼ .840, p ¼ .402]. However, during the year of follow-up, those with UPSA cut-off score of 60þ exhibited significantly lower HbA1C vs. those with an UPSA score
Contents lists available at ScienceDirect
Journal of Psychiatric Research journal homepage: www.elsevier.com/locate/psychires
Depression, deficits in functional capacity, and impaired glycemic control in urban African Americans with type 2 diabetes Dominique L. Musselman a, *, David C. Ziemer b, Marcia D. McNutt c, Julia S. Seay c, Erica B. Royster d, Bridget Larsen d, Terrika Barham d, Angelo R. Brown d, Octavia L. Vogel d, Lawrence S. Phillips b, Philip D. Harvey a a
Department of Psychiatry and Behavioral Sciences, University of Miami Leonard Miller School of Medicine, Miami, FL, USA Department of Medicine, Division of Endocrinology, Emory University School of Medicine, Atlanta, GA, USA University of Miami, Department of Psychology, Miami, FL, USA d Department of Psychiatry and Behavioral Sciences, Emory University School of Medicine, Atlanta, GA, USA b c
a r t i c l e i n f o
a b s t r a c t
Article history: Received 20 September 2013 Received in revised form 11 December 2013 Accepted 9 January 2014
Background: Effective depression treatment does not reliably reduce glycosylated hemoglobin (HbA1c) in depressed patients with type 2 diabetes, possibly in part due to deficits in functional capacity, i.e. performance of certain everyday living skills, essential for effective diabetes self-management. We sought to determine: a) the magnitude of deficits in functional capacity among urban, African American (AA) patients with type 2 diabetes, and b) whether these deficits were associated with poorer glycemic control. Methods: At their initial visit to an inner-city diabetes clinic, 172 AA patients with type 2 diabetes were assessed with a variety of instruments, including the Mini International Neuropsychiatric Interview (MINI) and the UCSD Performance Skills Assessment-Brief (UPSA-B). They then entered a comprehensive diabetes management intervention, whose success was indexed by HbA1c levels at up to four reassessments over a one-year period. A mixed-effects model repeated-measures method was used to predict HbA1c. Results: The prevalence of depression was 19%; the mean UPSA-B score was 81 17. After multivariate adjustment, increased HbA1c levels over time were predicted by the presence of major depression (B ¼ .911, p ¼ .002) and decreasing (worse) scores on the UPSA-B (B ¼ .016, p ¼ .027), respectively. Further adjustment for increasing the dosage of oral or insulin during the treatment eliminated the association between the UPSA score and HbA1c level (B ¼ .010, p ¼ .115). Conclusions: Depression, as well as deficits in functional capacity, predicted reduced effectiveness of a diabetes self-management intervention. Future studies will determine whether interventions targeted at both improve glycemic control. Ó 2014 Elsevier Ltd. All rights reserved.
Keywords: African Americans Type 2 diabetes Functional capacity HbA1c UPSA-B Major depression Neurocognition Glycemic control MINI structured diagnostic interview Zung Depression Rating Scale
1. Introduction African-Americans (AA) suffer from an increased prevalence of type 2 diabetes (Centers for Disease Control, 1993), and comorbid depression (Gary et al., 2000; Gilles et al., 2006). Irrespective of race, the presence of depression in patients with type 2 diabetes has been associated with acceleration of cognitive decline (Sullivan
* Corresponding author. University of Miami Leonard H. Miller School of Medicine, Department of Psychiatry and Behavioral Sciences, Mental Health Hospital Center, 1695 NW 9th Avenue, Rm 3308L, Miami, FL 33146, USA. Tel.: þ1 404 723 8361; fax: þ1 305 355 9072. E-mail address: [email protected] (D.L. Musselman). 0022-3956/$ e see front matter Ó 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jpsychires.2014.01.006
et al., 2013) as well as poorer compliance with medications which can potentially slow cognitive decline, i.e. antihypertensive drugs (Poon, 2008), aspirin (Kern et al., 2012), and statins (Steenland et al., 2013). Furthermore, in patients with diabetes, neurocognitive function can be impaired by agents with anticholinergic effect (Campbell et al., 2010): prescribed for neuropathic pain (e.g. amitriptyline), hypertension (e.g. atenolol). Less well examined are the consequences of neurocognitive deficits upon adherence to often complex diabetes treatment regimens (Awad et al., 2004; Kodl and Seaquist, 2008). Nevertheless, a rapidly accumulating database indicates that neurocognitive deficits correlate strongly with deficits in functional capacity (Moore et al., 2007), that is, the inability to perform critical daily skills, such as communication, manage medications, and handle finances (r ¼ .60e.65) (Leifker
22
D.L. Musselman et al. / Journal of Psychiatric Research 52 (2014) 21e27
et al., 2011), in patients ranging from those with HIV infection (Heaton et al., 2004) to Alzheimer’s disease (Gomar et al., 2011), or schizophrenia (Light et al., 2012; Keefe et al., 2011), as well as older healthy individuals (Leifker et al., 2011). Indeed, deficits in functional skills, when quantified in the laboratory, also predict “real world” outcome, i.e. the capability of patients with schizophrenia and bipolar disorder to live independently and sustain employment (Auslander et al., 2001; Mausbach et al., 2011; Twamley et al., 2002; Bowie et al., 2008, 2010). Of note is that functional capacity has not been previously examined in patients with type 2 diabetes, nor has its impact on their disease self-management, as reflected by another “real world” measurement, HbA1c. Though other psychiatric comorbidities such as anxiety and addictive disorders are common in patients with type 2 diabetes, depression may be most prevalent (Gilles et al., 2006). Thus, we sought to determine among urban, AA patients with type 2 diabetes: a) the prevalence and magnitude of depression, b) as well as deficits in functional capacity, and then c) examine whether depression, or difficulties in performing critical functional skills “in the lab”, would be associated with magnitude of HbA1c at up to one year of follow-up. 2. Method 2.1. Participants The Grady Health System Diabetes Clinic population is predominantly African American, with high rates of retinopathy, nephropathy, poverty and poor functional health literacy (Cook et al., 1999). Fewer than 10% of newly presenting patients have commercial health insurance; approximately 50% have no third party health care coverage (Ziemer et al., 1996). Patients may self-refer or be referred from their primary care or other specialty providers for treatment. Those diagnosed with type 2 diabetes fulfill American Diabetes Association (ADA) guidelines (The Expert Committee on the Diagnosis and Classification of Diabetes Mellitus, 2003): symptoms of diabetes (polyuria, polydipsia, unexplained weight loss) and a plasma glucose concentration 200 mg/dl, or a fasting plasma glucose of 126 mg/dl, or a 2-hour plasma concentration 200 mg/dl during an oral glucose tolerance test using an equivalent of 75 g anhydrous glucose dissolved in water. 2.2. Study design This prospective study was a sub-study of a larger, longitudinal, prospective examination of depression and glycemic control in urban African Americans with type 2 diabetes. Patients new to the Clinic were evaluated during a comprehensive intake appointment, including a medical history, physical exam, retinal photographs, and laboratory testing. Patients fulfilling study inclusion criteria (African American, English speaking, medical confirmation of diabetes, age 18e80 years, and to initiate diabetes care in the Clinic) were offered participation in the parent study beginning January 1, 2004 and ending March 30th, 2008. The convenience sample of the sub-study was recruited from August 2007 to March 2008. AfricanAmerican ancestry was confirmed through self-report. This study was approved by the Emory University Institutional Review Board and the Grady Research Oversight Committee. Written informed consent was obtained. 2.3. Study assessment After providing consent, the 30e45 min diagnostic psychiatric evaluation was administered only once, during the 6e8 h initial clinic visit at the Clinic.
The psychiatric diagnostic evaluation included the Mini International Neuropsychiatric Interview (MINI), a brief, structured, observer-administered questionnaire surveying for the presence of psychiatric disorders classified according to DSM-IV during the prior 12 months (Sheehan et al., 2005); magnitude of depressive symptoms were assessed using the 20-item, dimensional, selfreport Zung Depression Rating Scale (Zung, 1965). Another dimensional screening tool, the Traumatic Events Interview (TEI) (Schwartz et al., 2005), was used to detect past episodes of trauma and early life stress (ELS), defined by physical or sexual abuse at the age of 13 or younger. Participants received $15 compensation for the diagnostic evaluation. Those who wished to participate in the sub-study, received another $15 compensation to undergo the 15min functional capacity assessment consisting of the University of California San Diego Performance Skills Assessment-Brief (UPSA-B) (Sheehan et al., 2005; Mausbach et al., 2007), a performance-based, laboratory assessment of financial (i.e. make change and write a check to pay a bill) and communication (i.e. use a telephone to change a medical appointment) skills, with demonstrated validity regarding concurrent residential status and community engagement in persons with psychotic disorders (Mausbach et al., 2009). The psychometric properties of the UPSA-B have been examined, comparing healthy controls (n ¼ 109; mean age 68 12; mean UPSA-B score: 85 9) vs. patients with schizophrenia (n ¼ 238; mean age 57 9; mean UPSA-B score: 69 20), yielding a testretest reliability of the UPSA-B for that sample of patients of .75 and .73 at 18- and 36-months, respectively (Leifker et al., 2010). Sociodemographic and diabetes-related characteristics were obtained from patient self-report, the Grady Diabetes Clinic Patient Tracking System (DPTS), and the Grady Pharmacy database, including duration of diabetes, type of hypoglycemic treatment, and level of HbA1c. Prescription of medications for comorbid medical disorders was measured using the well-validated Chronic Disease Score (CDS) method by Von Korff and colleagues (Clark et al., 1995) in order to index general burden of illness. Information regarding income was gleaned from the Grady Health System electronic database. 2.4. Referral for depression treatment Patients who fulfilled diagnostic criteria for major depressive disorder were informed of their diagnosis, and received referral in verbal and/or written form for treatment within the Grady Healthcare System, their local mental health center, and other community resources. 2.5. Diabetes Management Education Intervention Patients were then scheduled return appointments to the Grady Diabetes Clinic every three months for their diabetes treatment involving the Diabetes Management Education Intervention: three educational classes, two designed to give information on diabetes, and another focused on diet and nutrition (Ziemer et al., 1996). Combined, the classes provided 5e6 h of diabetes education and were taught by either a certified diabetes educator or a dietician. Topics included the signs and symptoms of hypo- and hyperglycemia, the “ABCs” of diabetes, including what a HbA1c is and its importance, support group resources, and the importance of ADA goals and goal setting, the importance of self-management activities including medication compliance, nutrition and diet, physical exercise, stress management, how to perform the blood glucose test, maintenance of personal logs tracking glucose levels and other important medical indices, and recognition of patterns of their glucose levels. Clinic patients were given the opportunity to ask questions during sessions and were administered pre-, and
D.L. Musselman et al. / Journal of Psychiatric Research 52 (2014) 21e27
post-test, assessments of their diabetes knowledge. At their followup visits and during any telephone communications that might occur between visits, the health care providers reinforced diabetes information to the patient. Blood pressure and cholesterol levels, as well as HbA1c results, and goals were also discussed with the patient during follow-up visits. In addition, during these visits, antidiabetic and antihypertensive medications were adjusted as necessary, and patients receive foot and dilated eye exams. 2.6. Measurement of HbA1c Following the National Glycohemoglobin Standardization Program, HbA1c of the study participants was measured using a turbidometric, immuno-inhibition assay that recognizes glycated A, S and C hemoglobins equally (Beckman Synchron, Fullerton, CA) and shows no cross-reactivity with HbA0, HbA1a, HbA1b, acetylated Hb, carbamylated Hb, glycated albumin, HbA1d, or an acetylaldehyde Hb adduct. Total hemoglobin was measured using cell lysate and absorption at 560 nm. These assays were performed in the Grady Health System Clinical Laboratory, a College of American Pathology (CAP) accredited and licensed by the Georgia Department of Human Resources (DHR).
23
Table 1 Sociodemographic and medical characteristics of study population at intake. Characteristics
Total (n ¼ 172)
Age (years) (mean þ SD) Female Gender n (%) UPSA Score (mean þ SD) UPSA Communication Skills domain (mean þ SD) UPSA Financial Skills domain (mean þ SD) Chronic Disease Score (mean þ SD) Major Depressive Disorder-current n (%) Zung Depression Rating Score (mean þ SD)a Early Life Stressb e n (%) Anxiety Disordersc e n (%) Addictive Disordersd e n (%) Diabetes Duration (years) (median) Baseline HbA1c (%) (mean þ SD) Anti-diabetic Regimen n (%) None Oral agents alone Insulin alone or with oral agents
50 (10) 107 (62) 81 þ 17 40 þ 11 42 þ 8 3016 þ 1666 33 (19) 46 þ 13 25 (14) 24 (14) 23 (13) 2.9 8.9 þ 2.7 17 (10) 84 (49) 71 (41)
a Zung score: 20e49 (normal). 50e59 (mildly depressed). 60e69 (moderately depressed), 70þ (severely depressed). b Early life stress ¼ physical or sexual abuse at the age of 13 or below. c Anxiety Disorders ¼ panic disorder, agoraphobia, social phobia, post-traumatic stress disorder, obsessive compulsive disorder, generalized anxiety disorder. d Addictive disorders alcohol and/or drug abuse, alcohol and/or drug dependence.
2.7. Planned analyses A mixed-effects model repeated-measures method (MMRM) examined the course of HbA1c during the 12-month intervention, given that repeat assessments were scheduled at regular intervals, but actual occurrences of these assessments were somewhat variable. Certain variables were measured at baseline only: age, sex, HbA1c, body mass index (BMI), duration of diabetes, presence of major depression, general illness burden, and performance on the UPSA-B. These variables, well as number of subsequent clinic visits, and initiation of oral or insulin treatment, were used as covariates to predict the course of HbA1c over the follow-up. As HbA1c was not normally distributed, analyses were performed using generalized mixed models. As is the case with MMRM, baseline HbA1c was adjusted for, all available observations were used, and different numbers of participants were examined at each examination (Fig. 1). Specific analyses examined the level of HbA1c during 1 year of follow-up, and associated covariables, with the second part of the two-step analysis examining whether intensification of dose (of an
existing oral hypoglycemic or insulin) was associated with magnitude of HbA1c. Lastly, a “cut-off” score of the UPSA-B was explored for level of HbA1c during the year of follow-up (Mausbach et al., 2009). 3. Results Of 4341 consecutive, newly presenting African American patients to the clinic, 2112 (49% participation rate) provided consent for the parent study. Sixty-eight patients did not fulfill inclusion criteria. Other eligible patients refused (n ¼ 220), and others could not be scheduled for a research appointment in a timely manner (n ¼ 112). Two patients did not complete the USPA-B and were excluded; another 5 patients did not have a contemporaneous HbA1c drawn at clinic intake, resulting in 172 patients with both an UPSA-B score and a HbA1c at baseline. Of the patients who refused to participate (n ¼ 1276), certain of their socio-ecologic and medical characteristics were statistically
Fig. 1. Correlations between UPSA and Zung (panel A, left), and UPSA and HbA1c (panel B, right).
24
D.L. Musselman et al. / Journal of Psychiatric Research 52 (2014) 21e27
different, but clinically similar, to those who did participate. The mean HbA1c of nonparticipants was 8.4% vs. 8.7% (p ¼ .04) of study participants; a clinically insignificant difference given that a difference of .5% is predictive of microvascular complications (The Diabetes Control and Complications Trial Research Group, 1993). The mean body mass index (BMI) of participants vs. nonparticipants was 33 vs. 32 (p ¼ .001), respectively. The mean albumin/creatinine ratio of nonparticipants was worse (statistically) compared to participants, i.e. .11 vs. .07 mg/mg (p ¼ .02). The mean diastolic blood pressure of participants was 73 mm Hg vs. 75 mm Hg of nonparticipants (p ¼ .03). The relatively good control of both reflected, at least in part, initiation of antihypertensive interventions by Grady primary care providers prior to patient referral to the Diabetes Clinic. Table 1 shows that the African American type 2 diabetic patients at intake who participated in the sub-study were on average 50 years old, with 62% being female. Most had a high school degree or less (60%), and most were unmarried (83%). Fifty-nine percent had an UPSA-B score < 80. With an average BMI of 36, the participants of this study were generally “obese”; their average LDL cholesterol was greater than 100 mg/dl. In contrast, their systolic blood pressure averaged 127 mm Hg, a magnitude of relatively good control. Of the 77% of study participants prescribed antihypertensive medication, they were prescribed an average of 2.4 antihypertensive medications each. Thirty-four percent were receiving antidepressant treatment; 11 of these 61 patients were receiving either the tricyclic antidepressant amitriptyline or nortriptyline in low dosage for diabetic neuropathy. At baseline, the correlation between the UPSA-B score and HbA1c was significant (r ¼ .106, p ¼ .021), but not between the UPSA-B score and Zung depression scale score (r ¼ .008, p ¼ .916) (Fig. 1). The UPSA-B score was not correlated with age (r ¼ .12; p ¼ .10), Chronic Disease Score (CDS) (r ¼ .06, p ¼ .42) or Zung score (r ¼ .04; p ¼ . 57), but did correlate with lifetime educational attainment (r ¼ .33; p < .001). Over the course of their one-year follow-up, participants attended an average of 5.5 4 clinic visits. Factors at baseline potentially associated with the likelihood of follow-up during the subsequent year were examined. For the most part, baseline medical characteristics, i.e. age, gender, duration of diabetes, LDL, BMI and CDS, were not different between the patients who did, and did not, have a follow-up clinic visit, except that patients with a return visit exhibited a lower baseline systolic blood pressure vs. those who did not return to the clinic (p ¼ .002). Persons also more likely to return the clinic were those treated with insulin (p ¼ .035). Thirty-four (or 20%) changed from diet to oral hypoglycemic; 22 (or 13%) began insulin in addition to their pre-existing oral hypoglycemic medications. The number of return visits to the Diabetes Clinic was associated with greater CDS scores (r ¼ .267, p ¼ .000), as well as magnitude of the UPSA-B score (r ¼ .171, p ¼ .000). At baseline, 36% of the patients were found to have a HbA1c of less than 7.0% (American Diabetes Association, 1995), while 53% of the patients reached this ADA goal at their endpoint. Thus, 17% of the patients demonstrated a clinically significant improvement in HbA1c. Overall, the mean HbA1c decreased to 7.5 2.1% (Fig. 2). Mixed model analyses revealed a statistically significant time effect, reflecting significant overall improvement in HbA1c over the intervention (Table 2). Multiple significant covariate effects affected the course of HbA1c. Those with an increasing number of visits to the Clinic, those with more co-existing medical disorders, having a longer duration of diabetes, younger age, and those undergoing an intensification of treatment, either starting an oral hypoglycemic agent or beginning insulin therapy, exhibited an increased HbA1c
Fig. 2. Mean haemoglobin A1c (%) at baseline and reassessment visits during 1-year follow-up.
during the subsequent 12 months. Having current major depression (B ¼ .911, p ¼ .002), and having a lower UPSA-B score (B ¼ .016, p ¼ .027), was also associated with higher HbA1c, while sex, education, baseline BMI, and blood pressure were not. In the second step, when further intensification of dosage of an existing oral hypoglycemic or insulin was added to the model, having a lower UPSA-B score was not associated with severity of HbA1c during the year of follow-up (B ¼ .010, p ¼ .115). Furthermore, neither UPSA communication (p ¼ .15) or financial (p ¼ .14) domain scores were significantly associated with magnitude of HbA1c during the year of follow-up. Based on a prior report by Mausbach, we explored whether the cut-off score of 60 would be associated with HbA1c levels. At baseline, the HbA1c scores of the persons with UPSA scores (0e59 vs. 60þ) did not significantly differ [t (170) ¼ .840, p ¼ .402]. However, during the year of follow-up, those with UPSA cut-off score of 60þ exhibited significantly lower HbA1C vs. those with an UPSA score
