Demoralization, anhedonia and grief in patients with severe physical illness

Abstract

Within the physically ill population, demoralization has been a valuable concept with which to consider dysphoric states. Moreover, it can be distinguished from other mood states. In this study we replicate previous research and show that demoralization can be distinguished from anhedonic depression and grief. All three correlate more or less equally with measures of depression (e.g., the Beck Depression Inventory), although they are differentiated on styles of coping, and social, family and physical functioning. These results confirm that distinguishing depression with demoralization from depression with anhedonia is both useful and scientifically valid. It sharpens the characterisation of dysphoric states and provides an empirical framework to enhance research into the aetiology and treatment of depression.

Depressive states are common in the medically ill, with prevalence rates between 12 and 30% (1,2). However, there is strong opinion that current or traditional diagnostic systems have failed to adequately capture the range and nature of mood states seen and experienced in the medically ill (3,4) and that there is a lack of "syndromal differentiation" in the classification of depression (5).

In response to these sentiments, we undertook a study of the latent structure of symptoms in the medically ill in order to develop a taxonomy of meaningful syndromes of distress. In a sample of 312 patients hospitalized with a range of physical illnesses, we identified and distinguished dimensions of demoralization, anhedonia, grief, autonomic anxiety and somatic symptoms (6). Demoralization was characterized by feelings of being unable to cope, distress, apprehension, helplessness, hopelessness, personal failing and aloneness; anhedonia by a loss of ability to experience pleasure in things and accompanying loss of interest; grief by feelings of loss with intrusive thoughts about the loss, distress (pangs) and pining (yearning). Demoralized patients were more distressed than patients with anhedonic depression (as measured by the depression and anxiety scales of the General Health Questionnaire), but had less social dysfunction (7). These findings, using a method of numeric taxonomy, suggest that there are different types of depression, distinguishable by prominent anhedonia, demoralization and grief respectively.

In the present study, we aimed to examine whether this taxonomy is replicated in another sample of medically ill patients. Because helplessness and hopelessness are critical to the construct of demoralization, we decided to examine the phenomena in patients for whom no curative treatment was available - cancer patients admitted to a palliative care service and patients with motor neuron disease (MND). A further aim was to examine the convergent and discriminant validity of the factors (provisionally demoralization and anhedonia) by examining the correlations on a range of concurrent measures. We predicted that both anhedonia and demoralization would correlate highly with standard measures of depression which are themselves non-specific for subtype of depression, but would be differentiated on hopelessness and trait optimism. Hopelessness is central to the construct of demoralization. Trait optimism has been shown to be an important predictor of health-related coping (8) and, in the context of this study, it was hypothesized to be protective against demoralization. Other measures included physical functioning, coping style, social support and trait anxiety/ neuroticism.

METHODS

Sample

Consecutive patients with metastatic cancer were recruited at the point of referral to five palliative care services in south-east Melbourne. MND patients were recruited from consecutive attendees at specialist clinics in Melbourne and Sydney, and by invitation through the Motor Neuron Disease Associations in Victoria and Tasmania. Exclusion criteria were being too unwell to complete the interview and self-report questionnaires, and inadequate command of the English language. The demographic and illness particulars of the two groups of patients are summarized in Table 1.

Table 1.

Demographic and illness characteristics of the sample

		Motor neuron disease	Cancer	Total
		(n=134)	(n=137)	(n=271)
Age (years, mean ± SD)		62.7 (±11.4)	67.1 (±12.4)	64.9 (±12.1)
Sex (% female)		38	43	41
Country of birth (% born in non-English speaking country)		19	18	18
Marital status (%)	- Married	71	59	65
	- Widowed	10	24	17
Living with (%)	- Self	13	27	20
	- Spouse/Partner	52	58	55
	- Other adult	0.7	1.5	1.1
	- Family	24	14	19
	- Nursing home	11	0	5.2
Educational level (%)	- Incomplete secondary	33	37	35
	- Completed secondary	45	42	44
	- Completed tertiary	22	21	21
Religion (%)	- Christian	70	77	73
	- Other	1.5	6.6	4.0
	- None	28	17	23
Belief in God (%)		66	69	68
Cancer subtype (%)	- Lung		31
	- Gastrointestinal		18
	- Prostate		8.0
	- Breast		7.3
	- Haematological		4.4
	- Other		31
Motor neuron disease subtype (%)
	- Amyotrophic lateral sclerosis	55
	- Bulbar	15
	- Progressive muscular atrophy	6.7
	- Primary lateral sclerosis	6.0
	- Not recorded	18
Duration of illness (months, mean ± SD)		30.6 (±34.7)	4.8 (±8.6)	17.5 (±28.2)
Current treatment (%)	- Steroids	5.2	48	27
	- Opioids	7.5	56	32
	- Non-opioid analgesics	34	80	57
	- Psychotropics	45	48	47
History of previous psychiatric treatment (%)		24	30	27

Ethics committee approval was gained from all institutions involved and written informed consent obtained from all participants.

Interview and assessment procedure

The interview was an adaptation of the previously used Monash Interview for Liaison Psychiatry (MILP) (9). The original interview was simplified with the removal of duration and cause questions. Symptoms were considered present if they were present "over the past month". The attribution of "cause" of physical symptoms (whether the symptom is due to a medical illness, medication, drugs or alcohol, psychogenic or unexplained) is a difficult judgement in this group of patients and consequently we used an "inclusive" approach (10).

The original latent class analysis produced a dimension we called "anhedonia", which consisted of four items. To potentially strengthen this construct we added five additional items relating to interest and enjoyment developed after reviewing other questionnaires, of most relevance being the depression subscale of the Hospital Anxiety and Depression Scale (11).

The interview consisted of a demographic and illness section, and 86 questions in the domains of somatic symptoms, mood, self-concept, suicidal ideation and functioning. All symptom items were coded on a scale of zero to four if they were present "never", "almost never", "sometimes", "fairly often", or "very often". Most interviews were conducted in the patients' homes. For those patients in hospital, a private setting was sought. All patients were allowed to have a friend or relative present during the interview if they wished. The Beck Scale for Suicidal Ideation (SSI), described below, was administered as part of the interview. Self-report instruments were completed thereafter and returned by mail. In a number of cases, because of death or worsening health, self-report questionnaires were not completed. At conclusion of the study we had 271 completed interview schedules and 251 questionnaires with mostly complete data.

Instruments

Aspects of physical health and functioning were measured using the Physical Functioning Scale of the European Organization for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire Core 30 Items (QLQ-C30) and a visual analogue pain scale.

The QLQ-30 is a validated self-report measure of a range of core quality-of-life domains (12). We used the five physical functioning items each scored on a scale of one to four (QLQ-C30 version 3.0). The items are summed and then converted to a score with a range of 0-100 with a higher score indicating better physical functioning. This scale had a Cronbach's alpha of .84.

The pain analogue scale was a 10 cm line on which patients rated their current level of pain. This was measured in millimetres and given a score of 0-100.

Depression was assessed with two measures. The Beck Depression Inventory (BDI) provides a dimensional measure of depression and has been well validated over many years' experience (13). We used the BDI-II (14), a revised 21-item version using a 4-point response, with items covering domains of mood (2 items), anhedonia (2 items), hopelessness (2 items), cognitive functioning (2 items), self-concept (6 items), physical symptoms (4 items), psychomotor retardation (1 item), crying (1 item) and loss of interest in sex (1 item). The BDI-II had a Cronbach's alpha of .86 in our sample.

The Patient Health Questionnaire (PHQ-9, 15) is a selfreport questionnaire derived from the Primary Care Evaluation of Mental Disorders (PRIME-MD, 16). Its nine questions assess the presence of the DSM-IV criteria for major depressive disorder across a two-week period using a 4- point Likert scale response. When items are summed, a score of 10 or more corresponds to a diagnosis of major depressive disorder with a sensitivity of 88% and a specificity of 88% (15). The PHQ had a Cronbach's alpha of .78.

The degree of hopelessness was measured using the Beck Hopelessness Scale (BHS, 17), a 20-item true-false questionnaire assessing negative expectations about the future. Several studies have demonstrated a high degree of internal consistency and construct validity (e.g. correlations between .62 and .74 with clinical ratings of hopelessness) (18). The Cronbach's alpha of the BHS in our sample was .89.

Suicidal thought and desire for death were measured using the SSI (19). The original scale comprised 19 items, each scored zero, one or two. We removed nine items which seemed inappropriate or too confronting for people with a terminal illness and used items 1, 2, 3, 4, 6, 7, 8, 9, 10 and 14. Internal reliability of these ten items in our sample, using the Cronbach's alpha, was .86.

Dispositional or trait optimism was measured using the revised Life Orientation Test (LOT-R, 20). This is a 10-item self-report questionnaire, each item scored zero to four, giving a range for the scale of 0-40. Cronbach's alpha for this scale in our sample was .71. Dispositional optimism has been shown to correlate positively with active coping and negatively with avoidance in cancer patients (21).

Trait anxiety was measured with the State-Trait Anxiety Inventory (STAI, 22). The "trait" scale asks about "how you generally feel about yourself", with 20 items scored one to four giving a scale range of 20-80. Its validity based on correlation with other trait anxiety measures is good, with correlations ranging between .73 and .85 (22). Trait anxiety and neuroticism are strongly associated with vulnerability to depression, both in the presence (23) and absence (24,25) of physical illness. The Cronbach's alpha for the trait anxiety scale in our sample was .90.

Coping styles were measured using the Medical Coping Modes Questionnaire (26), a 32-item self-report measure using a 4-point Likert scale producing three subscales: confrontation (with eight items, score range 8-32, Cronbach's alpha= .66); avoidance (seven items, range 7-28, Cronbach's alpha= .61); acceptance-resignation (four items, range 4-16, Cronbach's alpha= .61). In general, a confrontational style is used more frequently by patients with acute life-threatening illness compared to patients with chronic illness (27). On the other hand, acceptanceresignation is typically seen in patients with little expectation of recovery. Effective coping in acute life-threatening situations is negatively linked with the use of avoidance or acceptance-resignation, while patients with chronic illness cope better using acceptance-resignation (28).

Social network and support was measured using the Indices of Social Functioning and Resources from the Health and Daily Living Form (29). From this, three scales were derived: number of close relationships (close friends and "people that can be counted on for real help in time of trouble"), number of social activities with friends (in the last month), and the quality of significant relationship. The latter was derived from six questions rated zero to four, giving a range of 0-24 with a Cronbach's alpha of .73 in our sample. A fourth scale, the number of social network contacts, had too much missing data to be used.

Quality of family relationships was measured using the three scales from the Family Environment Scale (30,31). This has been usefully employed to characterise family functioning in cancer families (32). From the 12-item true-false scale were derived scales for Family Cohesiveness (Cronbach's alpha= .44), Family Expressiveness (Cronbach's alpha= .53) and Family Conflict (Cronbach's alpha= .57).

Statistical analysis

All data were double-entered into a database and checked. Because the items used in this study were different in content and coding from the original study (6), we undertook exploratory rather than confirmatory factor analysis (principal component analysis with varimax rotation). Cattell's scree test was used to determine the number of components to be retained (33). Scores on retained components were computed for each patient by summing the raw scores of those items that loaded ≥ .45. The constructs defined by these dimensions were then examined using a series of bivariate and regression analyses. Comparisons of components from these analyses and those of the original study (6) were effected by means of the coefficient of congruence (CC, 34) between respective loadings.

Although there are no strict rules for calculation of sample size for factor analyses, samples in excess of 200 are generally required to ensure stability of estimates (35). Following Green (36), the sample size required in a multiple regression equation with up to 10 predictor variables should exceed 130. This assumes moderate effect sizes (e.g. regression beta weights of 0.20, moderate intercorrelations between predictors and alpha=.05). We therefore sought a sample size of 250: 125 cancer patients and 125 motor neuron disease patients.

RESULTS

Factor analysis

Two hundred and seventy one cases had sufficiently complete data to be included in the principal component analysis (PCA). Two items relating to sexual function were infrequently answered and removed. Questions pertaining to grief were only asked if a "gatekeeper" question (Does it feel as if you have lost something?) was answered affirmatively. These items were removed from the main PCA and were considered separately. This left 76 items for the main PCA.

The application of Cattell's scree test to the PCA results clearly indicated that three components should be retained. These three components accounted for 13.2%, 8.3% and 6.8% of the variance, and collectively 28.3% (see Table 2). The first component corresponds well to the concept of demoralization, with its highest loadings being on items of hopelessness, despairing, brooding, depressed, angry, pessimistic, discouraged, tearful, anxious, unable to cope and death wish. The second component is clearly anhedonia, with the highest loadings on lessened interest and inability to enjoy activities either with self or others. The third component consisted of physical symptoms (gastrointestinal symptoms, tiredness). Thirty-five items, nearly half, did not achieve loadings of .45 or more on any of the three components.

Table 2.

Loadings of items on principal component analysis

	1	2	3
Hopelessness	.718	.276	.014
Despairing	.711	.226	.090
Brooding	.684	.261	.045
Depressed, down, sad, miserable	.635	.316	.141
Angry	.612	.234	.045
Pessimistic	.605	.310	.114
Discouraged, despondent	.593	.266	.098
Tearful	.583	.013	.071
Anxious, apprehensive	.551	.233	.180
Unable to cope	.547	.286	.141
Wish rather not be alive	.547	.231	-.002
Restless, unable to relax	.535	.091	.045
Panic anxiety	.532	.035	.225
Feelings easily hurt	.525	.217	.006
Self-pity	.513	.294	.112
Keyed up, on edge	.507	.224	.141
Irritable	.504	.118	.117
Detached, depersonalized	.500	.245	.071
Suicidal thought	.481	.043	-.066
Thoughts of death	.477	.110	.149
Helplessness	.475	.261	.058
Feelings of choking	.453	-.073	.230
Less interested in doing things with others	.114	.762	.007
Unable to enjoy doing things with others	-.021	.728	.051
Unable to enjoy doing things by oneself	.073	.723	.164
Less interested in doing things by oneself	.218	.652	.171
Difficult to enjoy things	.211	.583	.192
Socially withdrawn	.137	.552	.099
No longer look forward to things	.415	.543	.042
Lost interest in appearance	.213	.510	.041
Difficulty doing things	.139	.496	.172
Unable to laugh	.212	.457	-.007
Abdominal pain or discomfort	.051	-.019	.614
Nausea	-.124	.044	.596
Vomiting or regurgitation	-.052	-.028	.556
Flushes, chills	.182	-.044	.494
Pain or discomfort in chest	.045	.079	.484
Tired, fatigued	-.021	.240	.480
Dry mouth	.094	.044	.474
Tire easily	-.024	.176	.471
Bloated, distended	.072	.024	.460
Not fully restored after rest	.157	.373	.344
Tense, sore, achy	.392	-.082	.270
Tremble, twitch, shaky	.388	-.145	.206
Back pain	.252	.019	.391
Pains in joints	.396	-.054	.310
Stiffness in joints	.405	-.106	.216
Butterflies, churning in stomach	.333	-.070	.438
Lump in throat	.186	-.040	.305
Trouble swallowing	.371	-.104	.122
Coated tongue, bad taste in mouth	.104	.024	.425
Loose bowels	.074	.011	.271
Frequent bowel movements	-.023	.017	.267
Palpitations, tachycardia	.165	.122	.340
Headaches	.169	.121	.346
Dizziness, light-headed	.090	.135	.291
Numbness, tingling	.136	.056	.304
Difficulty breathing, shortness of breath	-.006	.214	.404
Blurred vision	.000	.181	.332
Loss of mood reactivity	.288	.332	.097
Difficulty sharing others’ happiness	.266	.414	-.040
Unable to enjoy a good book	.076	.333	.136
Less talkative than usual	.302	.208	.039
Psychomotor retardation	.163	.242	.256
Psychomotor agitation	.439	.071	.116
Early or middle insomnia	.190	.005	.331
Early morning waking	.097	.182	.348
Easily startled	.334	.087	.128
Losing mind, losing control	.449	.137	.013
Feeling of impending death	.384	.151	.198
Trouble with thinking	-.042	.341	.324
Agoraphobia	.348	.235	-.063
Worthlessness	.337	.279	.001
Lost confidence	.393	.370	.000
Guilt	.306	.282	.068
Suicidal behaviour	.130	.126	.024

For the 212 patients who did acknowledge a "loss", the five grief items were submitted to principal component analysis in the same way as above. All items loaded ≥ .45 on a single factor, accounting for 53% of the variance (see Table 3).

Table 3.

Loadings of grief items on principal component analysis

Thinks about object of loss	.76
Pangs (distress caused by thinking about object of loss)	.82
Memories, mental pictures or dreams about object of loss	.55
Pining and yearning, longing for return of object of loss	.80
Cries about loss	.68

A comparison of component 1 from the 3-factor solution with the demoralization dimension from our original study (6) showed a CC of .87. A comparison of component 2 with the anhedonia dimension from the original study showed a CC of .60. A comparison of the grief component with the grief dimension of the original study showed a CC of .98.

Development of dimension scores

Scale scores for demoralization, anhedonia and grief were derived by summing the scores for each item loading greater than .45. In the case of grief, only the patients who answered the gatekeeper question affirmatively (n=212) were able to have a grief score computed. In the case of demoralization, the item "feelings of choking" was excluded because it seemed conceptually unrelated, its loading on the PCA was marginal (.453), and it had the weakest correlations with the other items. Cronbach's alpha was .92, .86 and .77 for demoralization, anhedonia and grief respectively. The demoralization and anhedonia scales were correlated .51, demoralization and grief .66 and anhedonia and grief .35.

Bivariate analyses

Correlations of scale scores with concurrent measures are shown in Table 4. Each scale correlated strongly with measures of depression (PHQ-9 and BDI), hopelessness and suicidality. They also correlated with trait anxiety and, to a lesser but significant extent, with trait optimism and acceptance-resignation. Physical functioning and pain correlated with demoralization and anhedonia, but not with grief.

Table 4.

Correlations of demoralization and anhedonia with dimensional measures

	Demoralization	Anhedonia	Grief
	(n=229-270)	(n=229-270)	(n=181-212)
Demographic and illness variables
Age	-0.23***	-0.01	-0.28***
Duration of illness	0.17**	0.01	0.14*
Physical functioning	-0.29***	-0.25***	-0.09
Pain	0.20**	0.22***	0.05
Mood and hopelessness
BDI-II	0.62***	0.59***	0.37***
PHQ-9	0.55***	0.50***	0.35***
Hopelessness (BHS)	0.45***	0.37***	0.33***
Suicidal ideation (SSI)	0.42***	0.42***	0.42***
Trait or personality measures
Trait optimism (LOT-R)	-0.36***	-0.29***	-0.22**
Trait anxiety (STAI)	0.64***	0.46***	0.34***
Confrontation coping	0.09	0.01	0.10
Avoidance coping	0.25***	0.09	0.32***
Acceptance-resignation coping	0.45***	0.32***	0.32***
Social and family attributes
No. social activities/month	0.08	-0.18**	0.18*
No. close relationships	-0.15*	-0.21***	-0.07
Quality of significant relationships	-0.29***	-0.17**	-0.22**
Family cohesiveness	-0.20**	-0.10	-0.06
Family expressiveness	-0.26***	-0.15*	-0.18*
Family conflict	0.11	0.06	0.09

BDI-II – Beck Depression Inventory – II; BHS – Beck Hopelessness scale; LOT-R – Life Orientation Test - Revised; PHQ-9 – Patient Health Questionnaire – 9; SSI – Scale for Suicidal Ideation; STAI – State-Trait Anxiety Inventory

* p<0.05

** p<0.01

*** p<0.001

Demoralization and anhedonia were differentiated on a number of variables, with age, duration of diagnosis and avoidance coping all correlating significantly with demoralization (and grief) but not anhedonia. With respect to social and family measures, numbers of social activities and close relationships were more strongly (negatively) correlated with anhedonia, while perceived quality of significant relationships was significantly, but not highly, correlated with all dimensions. On the other hand, measures of family functioning, specifically family cohesiveness and expressiveness, were more highly (negatively) correlated with demoralization.

For the categorical variables (Table 5), there were significant differences for diagnosis (cancer patients being more anhedonic, MND patients being more demoralized), living arrangements (people living with families having highest demoralization and grief), belief in God (those without experiencing more demoralization and anhedonia), past psychiatric history (associated with higher scores on demoralization and anhedonia), major depressive disorder (associated with higher scores on all measures) and current treatments (antidepressants being associated with demoralization and grief).

Table 5.

Association of demoralization and anhedonia with categorical measures (t test and ANOVA)

		Demoralization	Anhedonia		Grief
	N	Mean	Mean	N	Mean
Sex
-Female	109	20.82	12.45	183	6.55
-Male	160	20.16	13.10	127	7.46
Living with
- Alone	54	16.28	14.24	40	5.43
- Spouse/partner	149	19.36	12.38	111	6.93
- Other adult	3	20.33	16.67	3	8.33
- Family	51	27.59	13.71	45	9.36
- Nursing home	14	24.36***	8.79	13	6.08***
Diagnosis
- Cancer	137	16.91	14.08	96	5.72
- Motor neuron disease	134	24.30****	11.62*	116	8.29****
Belief in God
- Yes	177	19.12	11.79	137	6.61
- No	85	23.32*	14.51**	67	7.99
Past psychiatric history
- Yes	71	25.32	15.45	60	7.92
- No	193	18.55****	11.68****	145	6.74
Major depressive disorder
- Yes	37	35.51	21.65	32	9.81
- No	212	17.71****	11.49****	164	6.49****
Current treatment with antidepressants
- Yes	63	26.16	13.22	54	8.35
- No	208	18.87****	12.75	158	6.71*
Current treatment with sedatives or anxiolytics
- Yes	91	24.34	15.21	68	8.47
- No	180	18.65***	11.68**	144	6.49***

* p<0.05

** p<0.02

***p<0.01

****p<0.001

Multivariate analysis

We undertook a series of multiple regression analyses, with demoralization, anhedonia and grief as the dependent variables and various sets of independent variables. Conventional (p < 0.05) criteria for statistical significance were adopted. The models with highest explanatory power are shown in Table 6.

Table 6.

Results of regression analyses

Demoralization	Coefficient	t	p
Trait anxiety	0.73	8.69	<0.001
Age	-0.26	-4.21	<0.001
Acceptance-resignation coping	1.00	3.36	0.001
Avoidance coping	0.62	2.99	0.003
Diagnosis	4.08	2.79	0.006
Quality of significant relationships	-0.46	-2.44	0.015
Family cohesiveness	-1.78	-2.17	0.031
Confrontation coping	0.39	2.13	0.034
Past psychiatric history	3.13	1.97	0.05
Anhedonia
Trait anxiety	0.28	4.92	<0.001
Diagnosis	-3.67	-3.72	<0.001
Physical functioning	0.05	-2.93	0.004
Acceptance-resignation coping	0.51	2.38	0.018
Past psychiatric history	2.57	2.33	0.021
Belief in God	2.01	-1.95	0.052
Grief
Diagnosis	2.52	4.25	<0.001
Trait anxiety	0.12	3.99	<0.001
Avoidance coping	0.34	3.97	<0.001
Quality of significant relationships	-0.23	-3.03	0.003
Age	-0.06	-2.27	0.025
Confrontation coping	0.17	2.15	0.033

For demoralization, trait anxiety and age remained strong contributors, as did, to lesser extents, past psychiatric history, diagnosis (high demoralization in MND patients), coping style (confrontation, acceptance-resignation, avoidance), family cohesiveness and quality of significant relationships. This model accounted for 57% of the variance. For anhedonia, trait anxiety, diagnosis (high anhedonia in cancer patients) and past psychiatric history remained strong contributors, as did physical functioning. Belief in God also contributed, but age did not. This model accounted for 30% of the variance. For grief, trait anxiety, age and diagnosis remained significant, as did confrontation and avoidance coping, and quality of significant relationships, accounting for 33% of the variance.

DISCUSSION

In this study we have sought to re-examine the latent structure of what are principally dysphoric symptoms in a medically ill population. The results mirror the previous finding (6) of three distinct dimensions of demoralization, anhedonia and grief. Although this is a replication study, the form of the data here is not identical to that used previously. As described, we chose two groups of patients in whom helplessness and hopelessness might be expected to be common. Consequently, we did not screen for psychiatric morbidity, but included all available patients. This makes the results more generalizable. Second, symptoms were coded by severity rather than being rated present or absent, more truly representing their dimensional nature. Third, we did not attempt to make attribution judgements about the physical symptoms. Despite these differences, a very similar symptom structure emerged.

In addition, we wanted to examine concurrent - both convergent and discriminant - validity. The first thing to notice is that all three dimensions correlated significantly with common measures of depression, the BDI and the PHQ-9. The BDI is a general depression scale with a number of items in areas covered by demoralization (hopelessness, negative self concept) and anhedonia (loss of interest, loss of pleasure). The PHQ-9 reflects the nine symptom criteria of DSM-IV major depressive episode and therefore includes diminished interest and pleasure, worthlessness and thoughts of death. Furthermore, all three dimensions had a strong association with trait anxiety. The instrument we used to measure this, the STAI, is, in fact, not a pure measure of anxiety but also captures depression (37). These results therefore support the notion that the three constructs - demoralization, anhedonia and grief - are each components of current conceptualizations of depression. It is of note that it was demoralization that was associated with the greater use of antidepressants rather than anhedonia. This seems to go against a tradition of anhedonia as a marker of a so-called biogenic depression (38) and suggests that, at least intuitively, clinicians are identifying demoralized states as depressions worthy of treatment.

On the other hand, there were differences we can note, particularly in the associations of concurrent measures with demoralization and anhedonia. Physical dysfunction contributed uniquely to anhedonia, whilst lack of family cohesiveness, quality of social supports and avoidance coping were uniquely associated with demoralization. Demoralization itself was not significantly associated with severity of illness or physical functioning in the multivariate analysis. This is not to say that physical disease has nothing to do with demoralization. On the contrary, it is the physical disease and the experience of it that provides the context from which the feelings of helplessness arise. What this result suggests, however, is that, given a situation of serious physical illness, a range of psychological and social factors contribute to the development of demoralization more strongly than physical illness factors. Following the model described by Lazarus and Folkman (39), coping depends on a person's appraisal of the situation - whether it is a threat or a challenge - and on his coping behaviours. Greer and Watson (40) have shown that if patients see their cancer as a threat over which they have little control, they will likely feel helpless and hopeless. How a person copes with an illness therefore depends on the beliefs and assumptions he brings to the situation - beliefs and assumptions about himself and the disease, about his life and what it should look like, his sense of competence and control (41,42). Serious illness challenges many assumptions, especially of a younger person, leading readily to feelings of helplessness, aloneness and shame (43). The effects of these may be buffered by strong and supportive social relationships. In work not yet published, we have found that demoralization is strongly associated with a weak view of the manageability, comprehensibility and meaningfulness of the world - what Antonovsky has called a "sense of coherence" (44). The idea of control and "controllability of illness" is also important in the development of depression in people with schizophrenia (45).

The concept of demoralization was introduced into psychiatry by Jerome Frank (46), who said "the state of demoralization, in short, is one of hopelessness, helplessness, and isolation in which the person is preoccupied with merely trying to survive". The epidemiological work of Dohrenwend et al confirmed the validity of the concept (47). McHugh and Slavney incorporated demoralization into their Perspectives of Psychiatry framework (48), in which some psychiatric problems are considered diseases of the brain (they include major depression in this group) and some are considered dimensional and psychological in nature. On the other hand, Fava et al (49) and Kissane et al (50) have each developed operationalized criteria for a category of demoralization. The extensive work of Fava et al shows that demoralization is common in the medically ill, with prevalences of 29% in patients with breast cancer (51) and 34% in patients with endocrine disorders (52). The criteria used in these latter studies require that the feeling of demoralization (being unable to cope, hopeless, helpless, giving up, failing to meet expectations) antedates the manifestation of the medical disorder or exacerbates its symptoms. This is consistent with traditional psychosomatics and the idea of Engel (53), but seems a bit restrictive. Our experience is that the syndrome of demoralization occurs commonly in the setting of medical illness and that it is a useful concept even when considered a secondary phenomenon and a response to the predicament of illness.

One of the difficulties has been whether to consider demoralization as normal or as a "disorder". It is common to hear people say that someone is "just demoralized" as if it is not a serious matter. Slavney (54) believes that demoralization is a normal reaction and places it in the hierarchy of psychiatric disorders below adjustment disorders. This is despite acknowledging that it may involve neurovegetative symptoms, pessimism and suicidality. Other authors describe demoralization as a grief reaction or a situational or reactive depression (e.g., 55). This conflict highlights a slightly different word usage, as well as some confusion surrounding the concept of "normal". Nevertheless, the evidence for the construct validity of the concept (i.e., that the phenomenon exists) is strong. Further research is required to ascertain the level of dysfunction associated with demoralization and its value in predicting course and treatment response. Preliminary data suggests that the level of distress associated with demoralization is comparable with or greater than other forms of depression, but the level of social dysfunction is less (7).

With McHugh and Slavney (48) and in contrast to Fava et al (49), we consider demoralization to be best understood as a dimensional phenomenon. However, we do not consider it always to be a "normal" response, even though it may be understandable. Demoralization can be mild or it can be severe, and when it is severe it can lead to profound existential despair and a giving up of life. Consistent with this dimensional view and with the construct described in this paper, we have developed a 24-item self report measure of demoralization measuring aspects of dysphoria, disheartenment, helplessness, sense of personal failure, and loss of meaning and purpose. The scale has high internal reliability, with a Cronbach's alpha of .94 in a sample of 100 patients with advanced cancer (56).

We turn now to consider anhedonia, which was also clearly identified in the factor analysis. The concept of anhedonia was introduced by Ribot in 1896 to refer to an insensibility to pleasure, in an analogy with analgesia (38). Since then the term has been used more broadly, sometimes referring to a trait of hedonic capacity (57) and sometimes to a state phenomenon (58). In the latter context, Klein has described it as a "sharp, unreactive, pervasive impairment of the capacity to experience pleasure or to respond affectively to the anticipation of pleasure" (58). The importance of anhedonia within the classification of depression has varied over the years (59). It was introduced into DSM-III as a necessary criterion for melancholia (60); in DSM-III-R it was reduced to being one of nine symptoms (61); and in DSM-IV it is one of two symptoms, with mood reactivity, of which at least one is required (62). It seems therefore that the argument by Klein (58) and others (38) that it is the hallmark of a so-called biogenic depression is not fully accepted at this stage.

In this study, anhedonia was associated with worse physical functioning after partialling out the effects of other variables in the regression analysis. Anhedonia has been considered, together with apathy, in the medically ill, as being a disorder with a neurological basis (63,64). Latestage cancer patients often feel generally ill and have multi-organ disease, in contrast to patients with MND (65), and it is possible therefore that anhedonia in this setting has a very "physical" cause. It is interesting that the two clinical samples differed, with cancer patients being significantly more anhedonic and MND patients more demoralized. This difference may, in part, be related to physical illness factors and in part to the psychological experience of illness, with MND patients tending to be physically healthier, younger and having a longer duration of illness. Issues of differing illness experience require further investigation.

The notion that depression with anhedonia can be distinguished from depression with demoralization also seems pertinent in the elderly. It has been reported that the most frequently experienced negative emotions in older adults with chronic medical conditions are sadness and loneliness (66), both characteristics of demoralization. Conversely, the Baltimore site of the Epidemiologic Catchment Area Program found, in a community survey of patients over 50 years of age with major depression, two distinct groups: those with sadness and those without sadness (termed "depression without sadness") (67). The latter group had anhedonia to meet criteria for major depression. Both groups had similar and significant functional impairment. Further follow-up studies are required to examine whether these two syndromes have different illness courses.

Our results suggest influence from different styles of coping and religiosity. Acceptance-resignation was related to both demoralization and anhedonia, and this is perhaps unremarkable in the sense that some degree of passive "giving-up" may be an adaptive form of behaviour during acute stages of serious illness (68). Demoralization was, in addition, associated with avoidance and confrontation, suggesting that in a demoralized state a wide variety of coping behaviours are used. Belief in God was significantly associated with less anhedonia and less demoralization, although only remained significant for anhedonia in the regression analyses. The role of religious and spiritual beliefs in coping with physical illness deserves more attention, as evidence suggests that spirituality is related to more active rather than passive coping (69) and to better health and well-being (70). Trait optimism is generally associated with maintenance of subjective well-being in times of stress (71,72) and in this study was negatively associated with both anhedonia and demoralization - although not significantly in the multivariate analyses.

This study does suffer from the limitations of cross-sectional research design whereby it is presumptive to make causal judgements. Nevertheless, in a large sample we have confirmed a previously found distinction between depression with anhedonia and depression with demoralization, and again identified the experience of grief in patients with two serious medical illnesses. Concurrent measures show that demoralization and anhedonia are both components of depression as commonly conceived, and yet have different distributions across two different illnesses and different associations with coping and social and family functioning. These findings support the discriminant validity of the concepts. We would argue that research into both the aetiology and treatment of depression would be better advanced by considering these subtypes of depression. The two concepts have different historical roots, with anhedonia associated with beliefs about biogenicity, while demoralization is more cognitively constructed and linked with the role of psychotherapy (46). Nevertheless, empirical evidence is required to confirm whether this or any other important distinction is valid.