Depression and Physical Illness available in Paperback
- ISBN-10:
- 0521603609
- ISBN-13:
- 9780521603607
- Pub. Date:
- 10/26/2006
- Publisher:
- Cambridge University Press
- ISBN-10:
- 0521603609
- ISBN-13:
- 9780521603607
- Pub. Date:
- 10/26/2006
- Publisher:
- Cambridge University Press
Paperback
Buy New
$90.99Overview
Product Details
| ISBN-13: | 9780521603607 |
|---|---|
| Publisher: | Cambridge University Press |
| Publication date: | 10/26/2006 |
| Pages: | 434 |
| Product dimensions: | 6.85(w) x 9.72(h) x 0.83(d) |
About the Author
Read an Excerpt
Cambridge University Press
978-0-521-60360-7 - Depression and Physical Illness Edited - by Andrew Steptoe
Index
part 1
Introduction to depression and
its determinants
1
Depression in the medically ill
Francis Creed and Chris Dickens
Introduction
Psychiatric disorders of all types are more common in people with physical illness compared with the general population. Depression is the most common disorder, accounting for approximately 50% of psychopathology in the medically ill, with the remainder made up of various anxiety disorders and mixed subsyndromal symptoms of anxiety and depression. The importance of depression in the medically ill lies in its adverse effect on outcome, most notably health-related quality of life, combined with the fact that it is rarely detected and treated adequately in people who have physical illness.
The prevalence of depression in medically ill populations varies greatly according to the definition of depression and the type of measure used [1–4]. Variation in the definition and measuring instrument are the main reasons for the large variation in the prevalence figures quoted in the literature[5]. A higher prevalence of depression has been reported by studies that have used a self-administered questionnaire compared with those that used standardised research interviews administered by a trained interviewer [6]. The prevalence of depression also varies according to sociodemographic characteristics of the sample and the location of the survey (out-patient, in-patient, community) [3]. Only after all of these factors have been taken into account is it possible to assess whether the prevalence of depression varies according to type of medical disorder, its chronicity or severity [2,3].
In this chapter we indicate the ways in which different definitions and different modes of measurement used in previous research can affect the prevalence of depression. We examine the prevalence of depression in different groups (in-patients, out-patients, population-based samples) and review briefly the few studies that have examined the incidence of depression in the medically ill. Finally, we mention the effect of depression on outcome – assessed in terms of health-related quality of life and healthcare costs. This topic is discussed more fully in later chapters.
‘Depressive disorder’ or ‘depression’ in the medically ill
In published studies, the prevalence of depression in the medically ill ranges between 15% and 61% [7]. The first figure comes from a study that used a standardised research interview, the latter figure from a study that used self-administered questionnaires with cut-off points indicating probable depression. This variation cannot be attributed to a variation in disorders, as the same phenomenon can be observed in a single disorder – rheumatoid arthritis – where the prevalence of depression was reported as being between 19% and 53% in studies using self-administered questionnaires compared with a narrower variation of 17% and 27% in studies that used a standardised research interview [6].
These types of measure reflect, in part, a conceptual difference regarding depression. Questionnaires usually measure depression as a continuous variable – more symptoms indicate more marked depression, and this is regarded as a continuum, like blood pressure or heart rate. The concept of a ‘case’ of depression, however, implies that a certain number of symptoms, present for a certain duration, amount to a threshold, above which depressive disorder is diagnosed. The idea of the threshold is that it identifies the point beyond which depression carries complications, mostly an increased risk of impaired daily function and increased suicide risk. When a questionnaire is used, the results are expressed as a mean score for the medically ill population; e.g. the mean depression score on the Hamilton rating scale for depression was 11.8 for stroke patients two months after the stroke [8]. When the concept of a case is used, the results are expressed as the proportion of the population that has been classified as depressed, e.g. 27% of stroke patients classified as having a depressive disorder after a research interview, or 29% scored greater than 17 on the Hamilton rating scale for depression [8].
These two methods of measuring depression merge into one if a cut-off score on a self-administered questionnaire is used to determine the proportion of the population who are probably cases of depressive disorder. Such data are frequently reported in the literature, but it is important to be aware that such a translation from one method to another may lead to misleading results. The sensitivity and specificity of a self-administered questionnaire refer to the ability of the questionnaire to detect cases (determined by research interview) and its accuracy in doing so, when a particular cut-off score is used. It is necessary to consider these parameters because many patients are misclassified as having depressive disorder that is not confirmed by interview (false positives), and vice versa (false negatives) [1,9]. There is no clear consensus as to which method is preferable in physically ill people. This is not so different from the measurement of blood pressure – although there is an upper limit of normal blood pressure, above which treatment is recommended, not everyone with raised blood pressure will develop complications.
In general, the two methods of assessment can reach broad agreement as long as the cut-off score on self-administered questionnaires is adjusted for the physically ill population being screened (see below) [10,11]. In practice, the choice of method may reflect the purpose for the study. A self-administered questionnaire is required to screen a large population of physically ill patients [12] and may be used as the first stage of a two-phase survey, which includes research interviews to determine the actual cases of depressive disorder [13]. The latter method is also required to identify cases for inclusion in a treatment trial.
Standardised research interviews
The standardised interview method of ascertaining depressive disorder tends to be the approach used most often by psychiatrists. In epidemiological research, the identification of cases of depressive disorder within a given population of physically ill patients allows (a) the comparison of the prevalence of depressive disorder between physically ill and healthy controls and (b) the comparison between cases and non-cases within the physically ill group, which is an essential step to identifying risk factors. This approach forms the bulk of the research presented in this chapter; the tables below provide data indicating the prevalence of depressive disorder in different groups of medically ill populations.
Although used widely, there are problems with this approach. The main problem is that the threshold for depressive disorder, established in research performed in the general population or psychiatric patients, may not be that associated most closely with impairment in the physically ill; subthreshold disorders may also lead to impairment [3,9,14–16]. Further studies are required to identify the threshold of depression that is associated with impaired daily functioning, increased suicide risk and increased healthcare use [9,12]. There are also several different but related psychiatric diagnoses that may be relevant. These include major and minor depressive disorder, dysthymia and depressive adjustment disorder; it is not clear whether all or only some of these diagnoses correlate closely with impairment in the physically ill.
Self-administered questionnaires
The alternative approach views depression as a continuum without any clear cut-off separating depressed and non-depressed populations. This approach is probably a more accurate representation of the true picture [12].
Most studies using this approach have employed self-rated questionnaires that were originally designed for use in the general population, or in psychiatric populations, but not for use in the medically ill. These assessments (e.g. Beck Depression Inventory (BDI), Centre for Epidemiological Studies Depression Scale) contain a number of questions relating to physical (bodily) symptoms of depression, such as fatigue, weight loss, pain and anorexia. Since these physical symptoms may be a direct result of physical illnesses, such as renal failure or rheumatoid arthritis, it is inappropriate to include them as contributors to a diagnosis of depression. Doing so inflates the depression score simply because a physical illness is present. This phenomenon has been termed ‘criterion contamination’ and is responsible for some studies finding extremely high rates of depression in medically ill populations. For example, early studies that used the Minnesota Multiphase Personality Inventory in rheumatoid arthritis reported a high prevalence of depression until the following items were excluded from the definition of depression: inability to work as usual, failure to be in as good health as previously, easy fatigability, and presence of pain [17]. All of these symptoms could be attributable directly to the rheumatoid arthritis, and it is misleading to use them as symptoms of depression.
Table 1.1 Endicott criteria: symptoms of depression in the medically ill
A number of strategies have been used to overcome this problem of criterion contamination. First, Endicott [18] suggested alternative criteria to diagnose depression in cancer patients, whereby biological symptoms of depression were replaced by cognitive and emotional symptoms and signs of depression (see Table ). This may be done with other questionnaires (e.g. BDI), but doing so alters the psychometric properties of the questionnaires, and the validity of the measures used in this way cannot be assumed without formal testing against a gold-standard measure. Studies suggest removing some of the BDI items (weight loss, sleep disturbance, work inhibition) to increase its validity in chronic patients with back pain [19] and with headache [20], but the full BDI has been found to be satisfactory in patients with diabetes and multiple sclerosis [21,22].
Second, some questionnaires have been designed for, and validated in, medical patients, such as the Hospital Anxiety and Depression Scale (HADS) [23], or for use in populations where physical health problems are common, such as the Geriatric Depression Scale (GDS) [24,25]. The HADS is probably the most commonly used, as it excludes many of the physical symptoms of depression (loss of appetite and weight, sleep disturbance, fatigue, decreased sexual drive, poor concentration, psychomotor changes) included in more conventional depression questionnaires [26–28]. Cut-off scores were determined that best identified possible cases (scores of 8 or above) or probable cases (scores of 11 or above) on the depression and anxiety subscales, although these may need to be revalidated for the population in question [27]. Finally, it is possible to use a well-known self-administered questionnaire but with a modified cut-off score. When the General Health Questionnaire (GHQ) (28-item version) score was used in neurological patients, the cut-off score had to be raised from the usual 4/5 to 11/12 [10].
Standardised observer-rated research interviews
Use of observer-rated interview assessments carries the advantage that such assessments allow a trained interviewer to ask clarifying questions to determine whether a symptom has arisen directly as the result of a comorbid medical illness. Semi-standardised research interviews, such as the Schedule for Clinical Assessment in Neuropsychiatry (SCAN) interview [29], enable researchers to ask follow-up questions to ascertain whether a bodily symptom is likely to arise as the result of a medical illness as opposed to a psychological disorder. Such interview-based measures are considered to be the most reliable method of diagnosing depression; in general, studies using such rigorous assessments provide the most conservative estimates of the prevalence of depressive disorder in medically ill populations. The disadvantages of such interview assessments are that interpretation of any follow-up questions relating to the cause of a physical symptom relies on the judgement of the interviewer. Since it is not always possible to determine whether a bodily symptom is attributable to depression or to a comorbid medical illness, the diagnosis of depression may rest on the interviewer’s judgement. This subjectivity could undermine the value of delivering a standardised interview, however. The other main disadvantage of semi-standardised interview assessments is the long time taken to perform the assessment, which makes them acceptable for research purposes but too unwieldy for clinical use.
Depressive disorders in medical in-patients
The prevalence of depressive disorders in medical in-patients is shown in Table . There are differences between studies in the prevalence rates, which can be explained partly by the different diagnostic schema employed in different studies. Arolt and colleagues [30] state that their findings (16.4% had depressive disorder) are essentially similar to those of Feldman and colleagues [31] (14% with depressive disorder), assuming, apparently, that dysthymia and depressive adjustment disorder in the Composite International Diagnostic Interview (CIDI) are similar to the depressive disorder diagnosis of the Present State Examination (PSE). This may be reasonable as the difference between these categories might be a single symptom and depression in those with physical illness tends to be chronic, making a diagnosis of dysthymia (fluctuating depression present for two years or more) a common form of depression in this population. The more recent SCAN interview [29] provides diagnoses of, according to the 10th revision of the International Classification of Diseases (ICD-10), mild, moderate and severe depression. Mild depression is the most common form of the disorder in physically ill patients and is likely to overlap with depressive adjustment disorder.
Prevalence of depressive disorders in medical in-patients
| Study | Instrument | No. of patients | Prevalence of depressive disorder | Associated features of depressive disorder | Comment |
| Feldman et al. [31] | PSE | 453 | All depressive disorders = 14% | Younger women Unrelated to severity of medical illness | |
| Silverstone et al. [34] | SCAN/DSM-IV | 186 | Depressive disorder = 9.7% | Young age, female Unrelated severity of illness or medical diagnosis | Nurses (33%) than doctors (22%) better at recognising depression |
| Silverstone [32] | SCAN/DSM-IV | 313 | Major depressive disorder = 5.1%; adjustment disorder = 13.7% | Young age Unrelated to sex, severity of illness or medical diagnosis | Endicott criteria used Low prevalence, as depression present before admission often resolved after hospital admission |
| Arolt et al. [30] | CIDI/ICD-10 | 250 | Depressive episode = 4.8%; dysthymia = 4%; depressive adjustment disorder = 7.6%; Total depression = 16.4% | Divorced/widowed status | Worst social functioning associated with dysthymia, young age, depression preceding medical illness and chronic disease |
| Martucci et al. [7] | CIDI/ICD-10 diagnoses | 1039 (298 interviewed) | Depressive disorder = 12.8% | Female sex Recent life events, low family support | Disability days increased in cases and subthreshold cases |
| Creed et al. [9] | SCAN | 263 | Depressive disorder = 20% | Female sex, severity and number of medical illnesses | Impaired functioning associated in dose–response relationship with cases/subthreshold cases and controls. |
| Hansen et al. [42] | SCAN | 294 | Depressive disorders (including dysthymia) = 8.3% | Young age, female sex and life- threatening illness but not chronic illness | Of depressed patients only 18% recognised by physicians |
| Nair and Pillay [33] | SCID | 230 | Depressive disorders = 7% | Females | |
| Unrelated to severity of medical illness | |||||
CIDI, Composite International Diagnostic Interview; DSM-IV, Diagnostic and Statistical Manual of Mental Disorders, 4th edn; ICD-10, International Classification of Diseases, 10th revision; PSE, Present State Examination; SCAN, Schedule for Clinical Assessment in Neuropsychiatry; SCID, Structured Clinical Interview for the revised version of the Diagnostic and Statistical Manual of Mental Disorders.
The studies that used a diagnosis of major depressive disorder reported prevalence figures of 5.1%, 4.8%, and 7.0% [32–34], whereas those that used the ICD-10 diagnosis of depressive disorder (all ranges of severity) reported figures of 4.8%, 12.8%, and 20% [7,9,30]. The latter of these may be particularly high because of the selection of patients from a deprived inner-city area [9], but the figures have been summarised as mostly being in the range 5–14%, which is clearly higher than that in the general population (2–5%) [35]. The prevalence rate is generally higher among younger patients, and there is the expected excess in females. Typically in people with chronic physical diseases there is an absence of the other usual risk factors for depression, such as age and marital status [9,31] (see below).
Depressive disorders in medical out-patients and primary care
Table shows that the prevalence of depressive disorder in medical out-patients varies with the reason for the clinical presentation. For patients whose symptoms are explained by organic disease, the prevalence is 4–12%, not dissimilar from that in medical in-patients. The prevalence of depressive disorders is higher (10–24%) in patients whose symptoms are not explained by organic disease.
One study indicates the particularly high prevalence (26%) of major depressive disorder in patients attending a neurology clinic [36] compared with the more usual figure (13%) for general medical clinics [37]. The study also found that minor depression (8%) and dysthymia (15%) were common among new out-patients at a neurology clinic. These prevalences can be explained in part by the high proportion of patients with medically unexplained symptoms attending a neurology clinic but also by the high prevalence of depressive disorders associated with chronic and disabling neurological diseases. The high prevalence of depression in patients with medically unexplained symptoms is often associated also with a large number of bodily symptoms (somatisation), which may lead to medical help-seeking and increased healthcare use [38].
Prevalence of depressive disorders in medical out-patients
| Prevalence of depressive disorder | |||||
| Study | Instrument | Sample | Organic disease explains symptoms | MUS | Comments |
| Van Hemert et al. [65] | PSE | Randomly selected new patients at general medicine clinic | 4% (n = 91) | 24% (n = 100) | High prevalence of anxiety, somatisation and hypochondriacal disorders in MUS patients |
| Nimnuan et al. [66] | HADS (depression score >10) | Consecutive new patients at 7 out-patient clinics | 12.6% (n = 254) | 10% (n = 278) | |
| Feder et al. [67] | CIDI | Inner-city primary care | 9% (n = 130) | 22% (n = 42) | |
| Kooiman et al. [68] | HADS (depression score >10) | New general medical out-patients clinic | 7% (n = 152) | 15% (n = 169) | |
| Strik et al. [69] | ICDL/DSM-IV | Consecutive first myocardial infarction patients | 11.1% MDD; 7.8% minor depression (n = 206) | Total HADS gave better results than the separate subscales | |
| Lowe et al. [37] | ICDL/ICD-10 | Out-patients at university hospital clinic and few GPs | 15.8% depressive episode (mild = 5.4%, moderate = 5.6%, severe = 4.8%); MDD = 13.2%; any depressive disorder = 25.1% (n = 501) | PHQ, HADS and WBI-5 all distinguished satisfactorily the 3 types of depressive disorder | |
| Escobar et al. [70;71] | CIDI | Out-patients at a university-affiliated primary care clinic | 19% major depressive disorder (n = 1455) | 13.6% of non-somatisers and 37.5% of somatisers had depressive disorder | |
| Carson et al. [36] | PRIME-MD, DSM-IV | Consecutive new patients at 5 neurology clinics | Major depression 26%; minor depression 8%; dysthymia 5% (n = 300) | 38% of organic group and 60% of patients with MUS had some form of psychiatric disorder | |
CIDI, Composite International Diagnostic Interview; DSM-IV, Diagnostic and Statistical Manual of Mental Disorders, 4th edn; HADS, Hospital Anxiety and Depression Scale; ICD-10, International Classification of Diseases, 10th revisions; ICDL, International Diagnostic Checklist for ICD-10; MDD, Major depressive disorder; MUS, medically unexplained symptoms; PHQ, Patient Health Questionnaire; PSE, Present State Examination; WBI-S, Well-Being Index, version 5.
Depressive disorder is the most common psychiatric disorder among patients attending primary care worldwide [39]. Of the 25 916 patients attending primary care in a study by the World Health Organization (WHO), the estimated prevalence of depressive disorder was 10.4%, with generalised anxiety disorder (7.9%) and neurasthenia (chronic fatigue; 5.4%) being the next most common psychiatric disorders [39]. In the UK centre of the WHO study, the estimated prevalence of depressive disorder was 18.3% of attenders; approximately half of the patients with a psychiatric disorder also had a concurrent physical illness compared with one-third of patients without a psychiatric disorder.
Population-based studies
Cross-sectional studies demonstrate a close association between depressive disorders and physical illness in population-based studies. In the first large study, people with cancer, heart disease, neurological disorder or physical handicap were found to have a significantly higher prevalence of anxiety and depressive disorders (30.3%–37.5%) than people without a chronic physical illness (17.5%) [16]. In another population-based study, chronic disabling physical illness emerged together with lack of a close confidant as the two predictors of depression and anxiety [40]. The nature of the association between physical and psychiatric disorders cannot be discerned from cross-sectional studies, but evidence from a large (n;=7076) population-based study suggests that some common generic factors (low educational attainment, high neuroticism) may be associated independently with psychiatric and physical disorders [41].
© Cambridge University Press