Comorbid mental disorders and substance use disorders: epidemiology, prevention and treatment

What are the most disabling comorbidities?

Page last updated: 2003

It is widely documented in community and in clinical samples that comorbidity is associated with high levels of disability (Bijl et al., 1998; Kessler et al., 1994). In the Australian survey, respondents with more than one disorder reported significantly higher levels of disability, distress and service utilisation, with levels increasing in a linear trend as the number of disorders increased (Andrews et al., 2002). However, what is not addressed in the literature is a way of determining which, if any, combinations of disorders are especially prevalent and disabling and how this information can be used to inform health planning. The following section presents a method for determining the clinical significance of the various groups of mental disorders.

As alluded to at the beginning of this chapter, comorbidity is a different issue to concurrence. Comorbidity refers to the clustering of mental disorders in certain individuals over time. That is, it refers to a history of disorders in the past as well as to the concurrence of disorders in the present (see Wittchen, 1996). The previous discussion has indicated that concurrent associations between disorders are often stronger than successive comorbid associations. Successive comorbidity, the presence of two or more disorders some time during an extended time period, is useful for discussing risk factors, disability or service utilisation but is less useful for determining service delivery priorities. Thus, the current analysis will focus on the prevalence and disability associated with concurrent disorders.

Method
Results
Discussion

Method

Sample

The NSMHWB (Andrews, Henderson et al., 2001) was conducted by the Australian Bureau of Statistics under the terms of their Act that guarantees the privacy of respondents. The survey covered urban and rural areas across Australia. A multistage sample of private dwellings was drawn. Each state and territory was stratified and each dwelling within a stratum had an equal and known probability of selection. In all, 13,624 private dwellings were initially selected in the survey sample, and one adult member randomly selected as the possible respondent: 1,477 people refused, in 558 households contact could not be made with the identified respondent, and in 948 households no interview occurred because the identified respondent could not communicate, there was death or illness in the household, or the interview was prematurely terminated. The sample included people aged 18 years and over who were usual residents of households in the identified private dwellings. The sample did not include persons in hospitals, nursing homes, hotels, jails etc., or residents of households in remote and sparsely settled parts of the country. For this reason Aboriginal and Torres Strait Islander people were under-sampled and are not further identified in this paper. Ten thousand, six hundred and forty one people participated, a response rate of 78.1%. The age and sex characteristics of the sample were weighted to match the age and sex distribution in the national census.

Assessment

The whole interview was administered from a laptop computer. The Composite International Diagnostic Interview (Andrews & Peters, 1998; WHO, 1997) was used to determine, using ICD-10 and DSM-IV criteria, the presence of six anxiety disorders (panic disorder, agoraphobia, social phobia, [simple phobias were not identified], generalised anxiety disorder, obsessive compulsive disorder, post-traumatic stress disorder), two affective disorders (major depression, dysthymia), four substance use disorders (alcohol dependence and drug dependence for three classes of drugs). Screening questions were used to determine personality disorders (Loranger, Janca, & Sartorius, 1997) and an interview for ICD Neurasthenia (Tacchini, Janca, & Isaacs, 1995) was modified to reflect the CDC criteria for Chronic Fatigue Syndrome or DSM-IV undifferentiated somatoform disorder (Hickie et al., 1997).The CIDI module for schizophrenia generates false positive diagnoses (Kendler, Gallagher, Abelson, & Kessler, 1996) and a brief psychosis screener was used instead.

Disability was measured at the beginning of the interview by the Medical Outcomes Study Short Form-12 (SF-12,Ware, Kosinski, & Keller, 1996). The SF-12 is a generic measure of disability that has a mean of 50 and a standard deviation of 10. People who are disabled score less than 50, people who are very well score more than 50. The SF-12 produces two scores, a mental component scale score and a physical component scale score, the present data only refers to the former. It is reliable, valid and sensitive to change and the longer form (SF-36) has been widely used in Australia. We consider that it will become the standard health outcome measure in both mental and physical medicine. The mental health score relies on questions about vitality, social functioning, emotional role and mental health.

Training of interviewers and data analysis

All interviewers were experienced interviewers employed by the Australian Bureau of Statistics. Supervisors for each State and Territory were trained to criterion at the WHO Training and Reference Centre for CIDI in Sydney and then had a subsidiary course on how to train field staff. Routine data analysis procedures were used but as a result of the complex sample design and weighting, special software was required to estimate standard errors (SE).The SE of prevalence estimates and proportions were estimated using delete-1 jackknife repeated replication in 30 design-based sub-samples (Kish & Frankel, 1974).These calculations used the SUDAAN software package (Shah, Barnwell, & Bieler, 1997).
Top of page

Results

The data is presented in Table 1 and will be considered column by column. The first column is about the frequency of the diagnostic groups in the population. Anxiety disorders are the most common mental disorders and in any month, 739,000 Australian adults report symptoms that meet criteria for a DSM-IV anxiety disorder. People could have more than one anxiety disorder, say Post-Traumatic Stress Disorder (PTSD) and Obsessive Compulsive Disorder, but in this table they would still be counted as having an anxiety disorder. The frequency of personality disorders has never before been estimated in any population survey and while the prevalence (5.3%) is close to what was expected, more work is required to know exactly who was being identified. The diagnoses of affective, substance use and somatoform disorders are standard and as expected, and the rates are likely to be correct. Psychosis was the rarest disorder, 14 times less common than anxiety disorders. Although the rate of psychosis in this study (0.4%) was identified via a brief screening instrument, it is similar to that reported in the recent low prevalence survey (Jablensky et al., 2000) and is likely to be correct.

The second column is about disability. The results in column 2 are the mean SF-12 deviations from the population mean in standard deviation units. For illustrative purposes, 0–1 is considered to indicate mild disability, 1–2 moderate disability and a score of more than 2 severe disability. Remember that these are group means, and individual scores will be distributed above and below the mean value. In this column the affective disorders generate the highest scores and the substance use disorders the lowest. We have evidence that such self-report measures do not accurately represent the true disability associated with psychosis and have arbitrarily assigned a score of 3 (severe disability) to cases of psychosis. The significance of this decision will become apparent in later tables.

The third column is about the total disability in the Australian population attributed to people with the various disorders. When the number of cases is multiplied by the average level of disability of those cases, the affective and anxiety disorders are principal causes of disability in the community and, psychosis aside, substance use and somatoform disorders the least. But even if one substitutes an SF-12 value of 3 for all persons with psychosis, psychosis still generates less total disability than any other group of disorders, simply because it is a rare disorder. The lowest disability score of any individual in the national mental health survey was 4.2 standard deviations below the mean of 50, thus an average score of 3 for the whole psychosis group is very low indeed.

The bottom row shows that 1.7 million (1,660,000) people in Australia met criteria for any current mental disorder, their average SF-12 score is 0.8 and the product of these scores is 1.5 million disability units. In the sub-total line above, we show the total number of diagnoses as 2.5 million (50% greater) as though half the people had symptoms that meet criteria for two diagnosis groups. Actually some meet criteria for three or four diagnoses and rather fewer have two diagnoses. In the right hand column the population disability units are 2,739,000, twice as high as in the bottom or 'any disorder' row, demonstrating that people with comorbid disorders are more likely to have higher disability scores, higher even than the concurrence of two diagnoses would suggest.

Table 2a shows exactly the same people as in Table 1, now cross tabulated according to their concurrent diagnoses. In the top row 183,000 Australians met criteria for an affective disorder only and their mean disability score was 1.4; 245,000 Australians met criteria for concurrent anxiety and affective disorders and their mean disability score was 1.8 and so on. On the diagonal, in bold, are the disability scores for people who met criteria for only one current diagnosis. The top two disability values were psychosis (remember we re-scored all of them as severely disabled) and affective disorders. The least disabling single disorders were substance dependence and personality disorders, respondents with substance dependence and no other disorder returning an average score of 0.1 standard deviation drop on the SF-12. Thus while some might have regarded themselves as very well and others as disabled, it was the average of the group that was close to zero, not that all individuals with substance use disorders as their only mental disorder scored close to zero.

The cumulative disability associated with each single and double disorder is shown in Table 2b in the same population disability units as used in Table 1 column 3. In fact the total disability scores by diagnosis are exactly the same as in the right hand column in Table 1. The largest contributor to disability at the population level is the combination of anxiety and affective disorders. The least significant is substance dependence alone (we have ignored cells with less than 10,000 people because the numbers in the survey on which they were based are too small to be reliable). While it is easy to identify the highest and the lowest single diagnoses, and the highest and lowest combinations of diagnoses that contribute to psychiatric disability, it is very difficult to form a judgement about the totality of the data in Table 2b, important as it is.

The prevalence and mean disability scores for people with only one current diagnosis are displayed in the first column of Table 3a and the population disability units in the first column of Table 3b. They are exactly the same numbers that were on the diagonals in Table 2a and Table 2b.Top of page

In clinical practice, patients prioritise their symptoms and emphasise the symptoms that trouble them the most. It is that group of symptoms that is the focus of treatment. In the survey, once all diagnoses had been established, each person who was likely to meet criteria for more than one of the listed diagnoses was asked, "you mentioned having problems like (listing their groups of symptoms). Which troubled you the most?" Their response to this question was recorded as the main problem for those with concurrent disorders, and the numbers of people, and their mean disability and total disability units are displayed in column two of Table 3a and Table 3b.

When people have two or more disorders, what proportion chose a particular group as their main disorder? At some level this gives an indication of what they might seek treatment for or the disorder they would most like to be without, not necessarily what might disable them the most. Seventy-seven per cent of people with a concurrent anxiety said that was their main complaint, 61% of people with psychosis and 54% of people with affective disorders said likewise. These three disorders were of greatest importance to the sufferer. Forty per cent of people with a concurrent substance use dependence chose it as their main complaint, 28% of people with a personality disorder and 27% of people with a concurrent somatoform disorder did likewise. That is, in these three groups of disorders, other comorbid disorders were judged to be more troubling. Mostly these were the comorbid disorders listed in Table 2a, but sometimes people identified a concurrent physical disorder as their main complaint.

In Table 3b we list the population disability units for single or only disorders, for the identified main disorder when there were two or more present, and the total for the two classes. Nobody is counted twice. The total gives the sum of population disability units attributable to each group of disorders. The total, 1,643,000 population disability units, is greater than the sum of the averages in Table 1, because now we include only the disorders that the respondents see as primary, presumably most severe. However, it is less than the subtotal in Table 1 because there is no double counting of disability. The affective and anxiety disorders are the largest, accounting for 38% and 35% of the population total of disability respectively, or 73% in all. The remaining 27% is divided among the other three classes: personality disorders 12%, psychosis (even with the higher loading) 8%, somatoform disorders 4%, and substance use disorders 2%.

Table 1: Population prevalence and relative disability for 1 month diagnoses, NSMHWB

Population Prevalence - '000 (%)Mean SF-12 DeviationPopulation Disability Units - '000
Affective
518 (3.8)
1.7
881
Anxiety
739 (5.5)
1.1
813
Substance dependence
297 (2.2)
0.6
178
Personality
709 (5.3)
0.8
566
Psychosis*
56 (0.4)
1.0
56
Somatoform
164 (1.2)
1.5
245
Total of above diagnoses
2483 (18.4)
2739
Any mental disorder
1660 (12.3)
0.8
1494

† People who met criteria for abuse without dependence are not included in this analysis
* If a weighting of severe disability for psychosis is used the population disability units are 3.0 x 56 = 168,000

Top of page

Table 2a: Prevalence and disability of concurrent 1 month diagnoses, NSMHWB

Affective
('000)
Anxiety
('000)
Substance dependence
('000)
Psychosis
('000)
Personality
('000)
Somatoform
('000)
Total
'000)
Affective
183
245
64
185
20
74
518
Affective - SF-12 deviation
1.4
1.8
1.9
1.8
1.9
2.2
Anxiety
324
77
241
15
89
739
Anxiety - SF-12 deviation
0.6
1.5
1.5
1.9
1.8
Substance dependence
165
72
18
297
Substance dependence - SF-12 deviation
0.1
1.2
2.1
Personality
378
16
57
708
Personality - SF-12 deviation
0.3
1.8
2.0
Psychosis
23
56
Psychosis - SF-12 deviation
3.0
Somatoform
47
164
Somatoform - SF-12 deviation
0.8

† < 10,000

Table 2b: Population disability units for concurrent 1 month diagnoses from the NSMHWB

Affective
('000)
Anxiety
('000)
Substance dependence
('000)
Psychosis
'000)
Personality
('000)
Somatoform
('000)
Total
('000)
Affective
256
441
122
333
38
163
881
Anxiety
194
116
362
29
160
813
Substance dependence
17
86
38
178
Personality
113
29
114
566
Psychosis
69
168
Somatoform
38
245
Top of page

† < 10,000

Table 3a: Prevalence and disability of only or main problem diagnosis, NSMHWB

Diagnosis as only problem ('000)Diagnosis as main problem ('000)Affective with other diagnosis as main problem ('000)Anxiety with other diagnosis as main problem ('000)Substance dependence with other diagnosis as main problem ('000)Personality with other diagnosis as main problem ('000)Psychosis with other diagnosis as main problem ('000)Somatoform with other diagnosis as main problem ('000)
Affective
183
192
107
13
Affective - SF-12 deviation
1.4
1.9
1.9
2.2
Anxiety
324
324
80
20
Anxiety - SF-12 deviation
0.6
1.2
2.1
1.6
Substance dependence
165
53
23
35
Substance dependence - SF-12 deviation
0.1
0.4
2.0
1.5
Personality
378
99
62
111
13
Personality - SF-12 deviation
0.3
0.9
1.8
1.5
0.7
Psychosis
23
20
Psychosis - SF-12 deviation
3.0
3.0
Somatoform
47
32
29
34
Somatoform - SF-12 deviation
0.8
1.0
2.4
1.7

† < 10,000

Table 3b. Population disability units for only or main problem diagnosis, NSMHWB

Population disability units: Diagnosis as only problem ('000)Population disability units: Diagnosis as main problem ('000)Total population disability units ('000)% total population disability units (%)
Affective
256
365
621
38
Anxiety
194
389
583
35
Substance dependence
17
21
38
2
Personality
113
89
202
12
Psychosis
69
60
129
8
Somatoform
38
32
70
4
Total
687
956
1643
100
Top of page

Discussion

The current analysis takes a population approach to determining the prevalence, disability and clinical significance associated with comorbid disorders. As shown in Table 2a, the disability associated with comorbidity among any two disorder groups is generally higher than that associated with any disorder group alone. These findings are not necessarily new. As mentioned in the introduction, results from the NCS and the NEMESIS epidemiological surveys of mental disorders showed that, regardless of disorder combinations, as the number of disorders increases so too does the level of functional disability (Bijl et al., 1998; Kessler et al., 1994). When the disability is combined with the prevalence of each comorbid disorder group combination the resultant population disability units show that the combination of affective with anxiety disorders or personality disorders with affective disorders or personality with anxiety disorders produce the greatest population burden (Table 2b). When the analysis was restricted to a single nominated main problem for each respondent it can be seen that affective and anxiety disorders are again the most prevalent and most disabling and therefore account for the highest amount of population disability (Table 3a and Table 3b).

These results have implications for the Global Burden of Disease (GBD) project. Murray and Lopez (1996) showed that mental disorders were the principal cause of Years lived with disability and that, because of this, mental disorders ranked high in any table of the global burden of disease. It actually may have overestimated the burden of mental disorders because it did not control for concurrent disorders, and hence, while they took care to attribute years of life lost to only one disease, years lived with a disability were multiply attributed to all diseases a person currently had. There have been a number of attempts to rectify this (Andrews, Sanderson, & Beard, 1998). The recent Australian Burden of Disease study (Mathers & Vos, 1999) took a straightforward approach, apportioning the average disease weight between all disorders present. For this reason, as well as for other methodological changes, it calculated the burden of mental disorders in Australia at 15% of the total, third in importance after heart disease and cancer, a proportion that indicates the public health importance of mental disorders. Burden of disease calculations and health service planning require concurrent comorbidity to be addressed, and concurrent comorbidity is what the clinician must deal with. The current chapter presents one method for addressing the problem of concurrent comorbidity.

The results shown in Table 3b are not dissimilar to the years lived with disability proportions in the Australian Burden of Disease study. Mathers et al, (1999, Annex Table G) estimated that the affective disorders accounted for 38% of the years lived with a disability due to a mental disorder, the anxiety disorders 26%, the substance use disorders 21%, personality disorders 6% and psychosis 6%. They did not estimate the disability attributable to somatoform disorder. The disability weights used in that study came, not from self-report, but from judgements made by experts as to the impact of each disorder on the functioning of the average sufferer. The main difference between those results and the present results based on self-report is that people with substance use disorders underestimate the impact of them on their functioning. Both studies agree about the importance of the affective and anxiety disorders, and both note that psychosis is not the pre-eminent cause of disability attributed to mental disorders, because even though disabling, it is a rare disorder.

The size of the disability attributed to a particular disease group may not be a perfect indicator of relative importance. Merikangas et al, (1998) analysed data from seven community surveys in six countries and concluded that while there are strong comorbidities between mental disorders and substance use disorders, the mental disorders typically have an onset at an earlier age and are significant predictors of subsequent substance use disorders — probably by hastening the progression from use to problem use and from problem use to dependence. Simulations on the basis of this data suggest that about half of all drug dependence is associated with prior mental disorder: conduct disorder/adult antisocial behaviour in men, and conduct/antisocial behaviour and anxiety and mood disorders in women. These findings raise the possibility of prevention of substance use by early intervention with the mental disorder before the onset of the substance use disorder.

Australia spends 5% to 7% of its total health budget (public and private practice, specialist and general practitioner, in-patient and outpatient, veterans' affairs and the pharmaceutical benefits scheme) on mental health. This is half of the amount of money per capita that Canada and the United Kingdom spend. About half this money is spent on psychosis and substance dependence treatment: disorders that produce a substantial amount of individual suffering but do not account for a great deal of the total human suffering or disablement. If we were to respond to suffering or to the public health approach of relieving the burden of disease, we would prioritise both the anxiety and the affective disorders. The preferential funding of psychosis and substance use exists because, in a democracy, funds are allocated partly in response to voter demand. Families of young people who develop psychosis or substance dependence are rightly affronted by the visible suffering in their loved one. Other families are afraid their children might develop these disorders. Together, they form a potent advocacy group. But the wider society is also sensitive to these concerns. Fear of the 'crazed psychotic' or 'drug addict' is rightly, or wrongly, an important societal concern and protection from this perceived fear is seen as legitimate expenditure of taxes.Top of page

Rosenheck (1999) edited a series of articles on the 'care of the least well off '. He agreed that the relatively small number of people with the most serious illnesses (psychosis) consume a disproportionately large volume of health care services. He argued that there should be a balance between improving efficiency and maintaining intensive services for those with the greatest needs and put forward seven principles that could be used to guide resource allocation decisions. He rightly argued for the autonomy of individual patient welfare "...that one should never withhold treatment from a patient to achieve some other goal", even if it is the potential to receive more benefit from the same resources. The argument between equity and efficiency has just been joined and there is no obvious solution. What is obvious is that data now exists to inform the argument and that advocacy alone is no longer sufficient (Andrews & Henderson, 2000).