Abstract
Depression is a common condition experienced by people with chronic medical illness. It can be chronic and frequently disabling, affecting the course of medical conditions and hindering progress to recovery [1]. It is well documented that depressive illness [2–14] and symptomatology [15–30] is common in people living with HIV/AIDS (PLWHA). However, most of these reports have been US-based studies and are not necessarily generalisable to the Australian population of PLWHA. To date, there is a limited number of studies examining depression in the Australian population of PLWHA.
Judd and Mijch [23] found that in a sample of 100 PLWHA attending an outpatient clinic in a Melbourne hospital, 44% scored at or above a cut-off score of 14 on the Beck Depression Inventory (BDI). A score of ≥ 14 has commonly been used to identify depression [31]. This finding was later replicated in a similar study in which 49.5% of a sample of 192 scored ≥ 14 [22]. The first study [23] found no association between severity of score (≥ 14 on the BDI) and any medical or sociodemographic variables, perhaps due to low sample size. The second study with a larger sample size (n = 192) [22] found that patients who scored ≥ 14 were more likely to have been diagnosed with AIDS, were on sickness benefits/pension, were not in a current relationship or had a past history of depression.
Studies examining the incidence of depressive illness and associated factors using semistructured interviews have reported that past psychiatric morbidity [2, 3, 6, 12] plays a salient role in predicting future psychiatric morbidity as does intravenous drug use [8]. Interestingly, a relatively consistent finding has been that CD4 cell count (indicating level of immune function) [8, 13, 32] and advanced illness (indicated by a diagnosis of AIDS) [32] are not associated with increased rates of depressive illness.
Depression is commonly first seen and treated in the community by non-mental health professionals who are primarily general practitioners (GPs) [33]. Despite the reported increasing prevalence of depression in the community [34–36], there has been ongoing concern that depression remains under-recognized and under-treated in primary care [33, 37–42]. Given the well-documented detrimental effects of depression on comorbid physical conditions (e.g. non-compliance with treatment [43]) this under-detection and/or under-treatment may have a significant negative impact on the course of HIV.
The aims of the current study were (i) to gain an estimate of the rate of depressive disorders in PLWHA; (ii) investigate factors associated with depressive symptoms; and (iii) determine the ability of non-mental health medical practitioners to detect depressive illness in their patients with HIV/AIDS.
Method
Subjects
Study participants were patients attending four group general medical practice clinics and a large general hospital infectious diseases (ID) clinic. Only persons who had tested seropositive for HIV were eligible to participate in the study.
Setting
The four medical general practices are located in inner-city Melbourne. The general hospital ID clinic also in Melbourne provides a large medical outpatient clinic for treatment of PLWHA. All five sites have clinicians (GPs and physicians) specializing in HIV medicine and were chosen for the study because of their high caseloads of patients with HIV/AIDS.
Measures
1. Two questionnaires were developed for use in the study. The first questionnaire was designed to collect medical and psychiatric data and the second to obtain sociodemographic information from each patient. Medical data included information on current CD4 cell count and HIV RNA (viral load) counts; current antiretroviral therapy and other HIV related medication; stage of illness and duration of known HIV seropositivity. Psychiatric information included information on any current psychiatric diagnosis (made by the treating medical practitioner) and related treatment; treatment providers other than the GP/physician; and any past psychiatric diagnosis and related treatment. Current and past illicit drug and alcohol use was also recorded. Sociodemographic data included current state of employment, living situation and relationship status.
2. The Inventory to Diagnose Depression (IDD) [44] was used to assess the rate of syndromal depression in patients. The IDD is a 22-item, self-report scale with good test–retest reliability and internal consistency [44]. The scale was originally designed to detect Major Depressive Disorder (MDD) based on the DSM-III [45]. The advantage of the IDD is that it can be used in two ways: (i) to map symptoms directly to existing diagnostic nosologies such as DSM; and (ii) to quantify severity of depression. Both approaches were used in this study, that is, participant's total score on the IDD were recorded and the approach by Zimmerman and colleagues [44] was used to map the IDD items to the criteria for a Major Depressive Episode (MDE) based on DSM-IV [46]. A breakdown of the IDD items as they correspond to DSM-IV criteria for MDE is shown in Table 1.
DSM-IV inclusion criteria for a major depressive episode and corresponding items in the IDD†
Procedure
Enrolment for the study occurred over a 1-month period at each of the sites. All patients attending the GP/outpatient clinics who were HIV positive, aged 18 and over, capable of reading and writing English and able to give informed consent were eligible to participate in the study. Consecutive eligible patients attending regular practice appointments were informed by their GPs/physicians of the study. When a patient consented to participate, the GP/physician completed the questionnaire relating to the patient's medical and psychiatric history. The project Research Assistant recorded participants’ sociodemographic details. Each participant also completed the IDD.
Data analysis
Statistical analyses were conducted using non-parametric methods to compare those who met criteria for MDE (IDD positive) and those who did not (IDD negative). Variables considered for inclusion were participants’ HIV-related medical variables (CD4 cell counts and HIV RNA viral loads, duration of known HIV seropositivity, a diagnosis of AIDS, whether on a combination of antiretroviral medication), and psychiatric data (IDD total score, any current psychiatric disorder as diagnosed by GP/physician, a past psychiatric history, current and past use of illicit drugs). Sociodemographic characteristics (age, gender, employment status, living situation, relationship status) were also assessed. Almost all variables were categorical except age, duration of HIV diagnosis and IDD total scores. Although CD4 cell counts and HIV RNA viral load counts are also continuous variables, in this study they were treated as categorical (see below). Continuous variables were compared using Kruskal–Wallis Test. Categorical data was compared using χ 2 tests.
Potential predictors of depression (IDD positive) were initially screened using conventional univariate logistic regression methods. In the case of continuous or quasi-continuous variables, a further term Xln(X) (where X is the predictor) was entered into the analysis to test for the presence of non-linear scaling in accordance with the recommendations of Hosmer and Lemeshow [47]. The variables that had statistically significant (p ≤ 0. 05) odds ratios (OR) were then included in stepwise logistic regression analyses to explore models of best fit for medical, psychiatric and substance use and sociodemographic factors before fitting a final model for all variables. Examination of the laboratory indicators of HIV disease, viral load and CD4 counts indicated that the viral load variable was best used dichotomously by classifying viral loads simply as detected or not detected. The majority of the patients’ viral loads were detected using the Quantiplex TM HIV RNA assay, version 3.0 (bDNA; Bayer Diagnostics, Emeryville, CA, US) which has a detection threshold of 50 copies/ml. Therefore, viral loads below the threshold are deemed ‘undetectable’ and those above ‘detectable’. Viral load analyses on 20 cases were performed using COBAS Amplicor HIV Monitor assay, version 1.5 (RT-PCR; Roche Diagnostics, Branchburg, NJ, US); the method used had a detection limit of 400 copies/ml. Although theoretically, viral loads below the detection threshold of this assay may be deemed ‘undetectable’, for the purpose of consistency in this study, we have coded those 20 cases as having ‘detectable’ viral loads. The CD4 variable was partitioned into clinically indicative levels (i.e. 0–50, 51–200, 201–400, < 400) where the strata of CD4 chosen for analysis were based on levels of risk for opportunistic infection; 0-50 predicts very severe immunosuppression and high risk of opportunistic infection (e.g. CMV retinitis); 51-200 some level of risk of opportunistic infection; 201-400 minimal to no risk of opportunistic infection; and < 400 normal common function, and the effects at each level were explored.
Results
In total, from the five sites, 322 HIV seropositive persons consented to participate in the study. In order to preserve confidentiality, it was not possible to obtain the number or details of those who refused to participate in the study.
Sociodemographic data
The study sample was predominantly male with an average age of 41.4. No data was collected on risk group or ethnic background. Approximately half (54.3%) were in paid employment either on a full-time or part-time basis but a substantial number (42%) were recipients of sickness/disability pension and had been so for an average of five years. Forty-one per cent of the group were involved in an intimate relationship and also lived with their partners. A smaller number (2.5%) lived with partners and children or lived with their parents (6.5%).
Medical data
The majority of the participants were medically well. Although the CD4 counts and viral loads were widely varied, the median CD4 (476 cells/ μ L) and viral load (235 copies/mL) counts indicated the immune function of most participants was at a normal level. One-third of the group had undetectable viral loads and less than 20% had advanced HIV illness (i.e. AIDS). Approximately 66% were on a combination of three or more antiretroviral medications, while one-fifth were on no antiretroviral therapy at the time of study. Approximately 10% were co-infected with Hepatitis C.
Psychiatric data
Seventy people (21.7%) met criteria for a MDE (DSM-IV defined) on the IDD and a slightly higher number (23.9%) were also diagnosed as depressed by their GPs/physicians. The mean IDD total score for the whole group was 18.
More than one-third of the group (34.5%) was diagnosed by their GPs as having some form of psychiatric illness, the most common being depression (23.9%) and ‘stress’ (10.6%). General practitioners (18%) were the main providers of treatment for psychological and/or psychiatric problems experienced by the study participants, followed by counsellors (6.5%), private psychiatrists (6.2%) and the public mental health service (5.6%). Only 1.9% were seeing a psychologist. Twenty-two per cent were taking some form of psychotropic medication; the most common being antidepressants (15.8%). Almost half the group (43.8%) had experienced a psychiatric illness in the past.
Substance use
Less than 10% of participants were currently drinking alcohol at levels considered harmful by the National Health and Medical Research Council (NH & MRC) guidelines, but more than one-third (37%) were currently using illicit drugs. This was mostly cannabis (23.3%) used daily/weekly, while a small but significant number used stimulants (1.9%) and opioids (1.2%) on a daily/weekly basis. Forty-three per cent also used illicit drugs in the past.
Depressed versus non-depressed
Of the 322 participants, 70 (21.7%) met criteria for syndromal depression (IDD positive) and were classified as depressed. There were no significant differences between the depressed and non-depressed groups with regards to medical variables (i.e. CD4 and viral load counts, whether on combination antiretroviral therapy, duration of known HIV seropositivity or stage of illness). Although the two groups were similar in age and gender composition, there were significant differences with regards to other sociodemographic factors, psychiatric factors and total IDD scores (see Table 2).
Comparison between depressed (IDD positive) and non-depressed (IDD negative) study participants
Compared with the non-depressed group, depressed participants were more likely to have a history of psychiatric illness and illicit drug use. They were less likely to be in paid full-time employment and more likely to be recipients of disability/sickness benefits. The nondepressed participants were significantly more likely to be involved in intimate relationships and living with their partners.
Predictors of depression
Univariate logistic models indicated no significant association between detectable (< 50 copies/mL) HIV viral loads, CD 4 cell count, advanced HIV illness, or duration of known HIV seropositivity and odds of depression (IDD defined). There was no significant relationship between depression and the majority of the antiretroviral drugs. Having a past history of psychiatric illness (OR = 6.40, CI = 3.46–11.84) and a current GPs' diagnosis of ‘stress’ (OR = 5.11, CI = 2.44–10.67) were strongly related to increased odds of depression. Current (OR = 2.37, CI = 1.38–4.06) and past (OR = 3.33, CI = 1.91–5.81) illicit drug use seemed to double and triple, respectively, the odds of depression.
Factors that were associated with lowered odds of depression were full-time (but not part-time) employment (OR = 0.54, CI = 0.30–0.98), being in an intimate relationship (OR = 0.52, CI = 0.31–0.90), and living with one's partner (OR = 0.51, CI = 0.29–0.90). Increased odds of depression were associated with being on sickness/disability pension (OR = 2.01, CI = 1.18–3.43), and living alone (OR = 0.51, CI = 0.29–0.90).
For those with partners, there appeared to be no effect of the partner's physical or mental health on depression.
All the factors explored in the univariate models above were considered in a final multivariate model. Only three variables, being in an intimate relationship (OR = 0.43, CI = 0.23–0.81), a history of illicit drug use (OR = 2.98, CI = 1.60–5.54), and a GP diagnosis of ‘stress’ (OR = 5.65, CI = 2.50–12.77) retained any significant predicative effect on the increased or decreased odds of depression.
The above results were confirmed by exploring the relationship of the possible risk factors with IDD scores on a continuous scale. Higher scores on the IDD scale (indicating greater severity of depressive symptoms) were strongly related to a diagnosis of ‘stress’, and having a history of psychiatric illness. Past illicit drug use was also significantly related to severity of IDD score and although current and past excessive alcohol use did not reach significance, the trends were in the expected direction (i.e. alcohol use associated with more depressive symptoms). Full-time work was associated with lower IDD scores and hence less symptoms of depression.
Concordance between GPs'/physicians’ diagnosis and patients’ self-report
There was moderate [41] concordance (Kappa = 0.480; p < 0.001) between GP's or treating clinicians’ diagnosis of depression and results on the IDD for those who met criteria for depression. Forty-four (62.9%) of the participants who met criteria on the IDD for depression were also diagnosed as depressed by their doctors. However, another 33 patients who were diagnosed with depression by their doctors did not meet criteria for depression on the IDD. Eleven of the 33 were diagnosed with more than one psychiatric disorder by their GPs. Other diagnosis included ‘stress’ (n = 6), anxiety disorder (n = 4), mania (n = 1), psychotic illness (n = 2), dementia (n = 1) and personality disorder (n = 1). The mean IDD score of the group with comorbid diagnosis was 17. Conversely, 26 participants not diagnosed as depressed by their doctors met criteria for depression (IDD positive). More than half (54%) were not considered to have any psychiatric disorder by their doctors although the mean IDD score of this group was quite high at 35.
Discussion
The rate of depression in this cohort of PLWHA was high, with 22% meeting criteria for a MDE (DSMIV). This finding is consistent with previous studies’ reports of high rates of depressive disorder in PLWHA [22, 23, 48, 49].
It is possible that several factors may have biased the detected rate. The GPs/physicians may have encouraged patients whom they considered depressed to participate in the study thus inflating the number of those reporting depressive symptoms. On the other hand, there is anecdotal evidence that GPs/physicians tended to be more protective of those patients whom they considered too depressed or ‘unwell’ and were less likely to approach them about the study. This problem was noted in an earlier study [50] although in that study, patients had to be referred by their doctors whereas in this study, patients did have a chance to volunteer for the study.
Factors that appeared protective against depression were being in an intimate relationship and/or living with one's partner, and employed in a full-time, but not parttime, capacity. These results are consistent with existing literature which suggest that in HIV seropositive men, lack of supportive or confiding relationships [21, 29, 51] render individuals more vulnerable to depression while full-time employment has a protective effect [52].
In this study, indicators of immune function (i.e. CD4 cell and HIV RNA counts), disease stage (i.e. whether diagnosed with AIDS) and duration of HIV illness did not have any significant association with depressive symptoms. Some previous studies [22, 23, 29] have also reported no association between duration of HIV diagnosis and depression but not all [53]. No association between CD4 cell count on depression has been reported by some researchers [20, 23, 48] but others have found that experiencing an AIDS defining illness was associated with greater depressive symptomatology [22, 54]. It is possible there may have been a ceiling effect with regards to immune function in this study as the majority of participants’ CD4 and HIV RNA were at normal levels. Perhaps if patients with more compromised immune function had participated in the study, an effect of medical variables may have been noted.
A history of illicit drug use, primarily cannabis use in this study sample, remained a significant predictor of depression even when taking into account other factors. The literature is divided on whether cannabis use/abuse is a risk factor for depression or whether depressed individuals are more likely to use cannabis. Crosssectional studies appear to support the latter hypothesis while the former theory is supported by longitudinal data [55]. The association between cannabis use/abuse and psychiatric disorders, particularly affective disorders, has been noted in several recent studies [55–61]. Some researchers suggest that the association between cannabis use/abuse and depression may be mediated by other factors such as use of other drugs [57, 62] and sociodemographic factors [62]. Green and Ritter [62] reported that early use but not adult use of marijuana was associated (albeit weakly) with increased depressive symptoms in adulthood. It is interesting that in this study, a history of any drug use but not current use, was predictive of depression.
A GP/practitioner diagnosis of ‘stress’ also remained a significant independent predictor of MDE. The term ‘stress’ has been used within inverted commas as it is unclear what this term really means. It is possible that a diagnosis of ‘stress’ may have been used for patients who were depressed.
Several factors pertaining to both patient and doctor may affect how a psychiatric diagnosis is formed, including the way in which patients present their distress, interviewing style/skills of GPs/physicians, knowledge of the sociocultural characteristics of their patients and a reluctance to use psychiatric ‘labels’ [63]. Added to these factors is the lack of a useful system in primary care for diagnosing psychiatric disorders. The current diagnostic nosologies of DSM-IV and ICD-10 used primarily in research are not easily or necessarily translatable into use in primary care settings [64].
In this study, there was agreement between GPs'/practitioners' diagnosis and self-reports of depression in 63% of patients. In only 8% of cases, symptoms were not recognized by GPs/practitioners as constituting any psychiatric disorder even though the patients’ self-report met criteria for MDE. These results stand in contrast to those previously reported in Australian general practice [37, 40] in which the recognition rate of psychiatric disorders by GPs was low. Consistent with existing literature that suggests most common mental disorders such as depression and anxiety continue to be managed within primary care [33], GPs/physicians were also more likely to provide treatment to their patients for psychiatric/psychological problems than to refer them to other mental health service providers. The higher recognition rate of psychiatric disorders by GPs/practitioners in this study might partially be due to the fact that the Alfred Hospital clinic is colocated with a psychiatric clinic, providing ample opportunity for secondary consultation with psychiatrists. Many of the GPs also spend time in this clinic. Better detection rates of mental illness have been noted in centres where there has been close collaboration between psychiatrists and primary care physicians [33]. In addition, the GPs/physicians were practitioners in HIV medicine and thus more likely to be aware of the psychosocial problems linked to HIV/AIDS. Much of the literature on HIV/AIDS in medical journals/texts has emphasized the psychosocial issues as have major conferences on HIV/AIDS. Because of the specialist nature of their practice (i.e. HIV medicine), it is highly likely that GPs/physicians maintained more constant links with their patients, akin to a personal style of service found to foster better detection of disorders in primary care [33].
Several limitations of the study must be acknowledged. First, small sample size and restriction of the study to metropolitan Melbourne dictates these results are not necessarily generalizable to the larger population of PLWHA in the community. Furthermore, the study cohort consisted predominantly of middle-aged males and thus does not address issues specific to females infected with HIV, younger and older males with HIV/AIDS, and individuals from non-English speaking backgrounds were precluded from the study by the exclusion criteria of non-proficiency in English. Second, there was no independent clinical assessment of the IDD results or standardization of GPs'/practitioners' diagnoses. Third, not all patients, or even consecutive patients, attending each site participated in the study.
Conclusion
Depression was common in this group of PLWHA with more than one-fifth meeting criteria for a major depressive episode. These results are consistent with the existing literature on the prevalence of depression in males with HIV/AIDS. Treating GPs/physicians were also in agreement with the majority of cases.
Investigation into factors associated with depression indicated that in this study cohort, psychosocial, but not medical variables, were linked to depression. This is perhaps not surprising given that significant advances in the last decade in medical treatment and management has led to better medical prognosis particularly for those in developed countries. However, this in turn has presented survivors of HIV/AIDS with problems of a different nature compared to the early years of the pandemic two decades ago [65].
The majority of participants in this study were medically well with immune function indicators at normal levels. Illicit drug use, and ‘stress’ appeared to be predictive of depression whereas being involved in a relationship was protective. Because of the cross-sectional nature of this study, no causal relationships can be assumed between these variables and depression. It could be argued that depressed individuals are more likely to ‘self-medicate’ by taking drugs, less likely to sustain or develop intimate relationships and be more vulnerable to life stressors. On the other hand, it is equally plausible that chronic drug use, stressors and lack of relationships may predispose individuals to depression. The persistence of high rates of depression despite the markedly improved medical treatment of management of HIV/AIDS is of concern. The results of this study and others suggest that psychosocial issues play a central role in the development and maintenance of depression. The exact nature of the relationship between these factors and depressive illness remains unclear and needs elucidation through more longitudinal research. Nonetheless, the importance of giving greater attention to improving support and services by the appropriate agencies to people living with HIV/AIDS is highlighted.
In this study, there was a high recognition rate of depressive symptoms in patients by their GPs/physicians. Furthermore, GPs/physicians were also more likely to provide treatment for psychological/psychiatric problems. This is an encouraging finding given that undetected and untreated depression can have detrimental effects on the course of a physical illness such as HIV/AIDS.
Footnotes
Acknowledgements
We thank the general practitioners and physicians from The Carlton Clinic, Prahran Market Clinic, Middle Park Clinic, The Centre Clinic, St Kilda, Infectious Diseases Clinic, The Alfred Hospital. Thanks also to all the administrative staff at the clinics and particularly Ms Katie Costello, for their assistance.