Relationship of criminality to endogeneity and anxiety in patients with unipolar depression
Abstract
Abstract The relationship between criminal behavior on the one hand and endogeneity and anxiety on the other hand was investigated in a sample of patients with unipolar depression to help elucidate factors influencing the criminality rate in this population. A lower criminality rate in patients with higher ratings of endogeneity and anxiety was predicted. Clinical records of 179 male and 99 female psychiatric inpatients were retrospectively evaluated using the Newcastle Scale II and Hamilton Anxiety Scale. A full account of conviction records served as a measure of criminal behavior. Forty per cent of male patients and 7% of female patients were criminally registered. A lower criminality rate was indeed found in male and female patients with endogenous type of depression and in male patients with higher anxiety ratings. In a multivariate evaluation, however, sociodemographic variables in terms of age and social class seem to be more important predictors of criminality and all variables we assessed contributed only marginally to the explanation of the criminality variance. Thus, in patients with unipolar depression, sociodemographic factors seem to be of a greater even though still limited importance regarding criminal behavior compared with the clinical variables of endogeneity and anxiety.
INTRODUCTION
There are few investigations devoted specifically to the study of criminal behavior in patients with affective disorders. One reason for this is the low criminal rate in affective disorders in general; the frequency of affective disorders was estimated 0.0–0.6% in unselected criminal populations and 0.7–13.2% with an unweighted mean of 4.4% in different criminal unprosecuted offender populations.1 The second reason is the lower frequency of criminal behavior in affective rather than in other mental disorders2,3 (e.g. 49% of schizophrenic but only 19% of affective probands were found to have a criminal record).4 Yet there are depression-characteristic offences, in the first place extended suicide5 or attempted extended suicide6,7 and, perhaps, shoplifting.8 Affective and antisocial symptoms are linked in a subgroup of patients,9 and high rates of affective disorder have been found among juvenile offenders.10,11 In these studies, endogenous type of depression6 and adjunctive anxiety7,10 were only exceptionally addressed.
In a previous study,12 we investigated criminal behavior in 261 male patients, diagnosed with affective disorders according to Research Diagnostic Criteria (RDC).13 Compared with matched controls from the general population, a significantly higher criminal rate was found for 82 bipolar patients and for 67 patients with unipolar minor or intermittent depression, but not for 112 patients with unipolar major depression. In the present study, the relationship between endogeneity and anxiety to criminal behavior was investigated in a total of 278 psychiatric inpatients of both sexes and with all types of unipolar depression. We expected the criminal rate to be influenced by both variables; in particular, we predicted the criminal rate to be lower in anxious patients and in patients with endogenous depression, our hypothesis being based on the following reasoning.
The commitment of criminal acts is rare in depression, purportedly as it is in contradiction with the melancholic type of personality.14 This personality type was described in patients with monopolar depression of endogenous type.15 In spite of some clinicians having considered the demarcation between endogenous and non-endogenous as the primary categorical distinction in affective disorders classification,16 the endogenous concept has lost its significance in the last years due to an unreliable identification of the endogenous class and a relative lack of the concordance between the different diagnostic instruments.17 Nevertheless, the persistence of endogenous depression in modern diagnostic systems as an optional diagnosis in the form of depression with somatic syndrome18 or depression with melancholic syndrome19 and the sudden, although short-lived, revival of interest for the concept in connection with the introduction of the dexamethasone suppression test,20,21 point to the reluctance to give it up definitely.
Anxiety may also influence violent and antisocial behavior.22,23 A high proportion of criminal subjects suffer from sociopathy24 and psychopathic individuals were said to be incapable of anxiety and remorse.25 Depressive and anxious symptoms frequently coexist,26–28 even so that the discrimination between both becomes difficult when there is only a mild to moderate symptom intensity.29,30
METHODS
Subjects
Our male sample consisted of 179 psychiatric inpatients described previously.12 Briefly, all patients were hospitalized at least once at the Psychiatric University Hospital in Bern during 1985–1987. They were all between 18 and 78 years of age, received the hospital ICD-931 diagnosis of an affective disorder as their principle diagnosis and the diagnosis was retrospectively confirmed by RDC.13 For the present study only patients who were diagnosed with RDC unipolar major (112 patients) and unipolar minor or intermittent depression (67 patients) were considered. The mean age of the patients was 52 years (SD 16) and their marital status was 44% single, 43% married and 13% widowed. A total of 46% were recruited from the middle social class and 54% from the lower social class.
Our female sample consisted of 99 psychiatric inpatients, all of whom were hospitalized at the same psychiatric hospital in 1987. They were all between 18 and 78 years of age and received the hospital ICD-931 diagnosis of a unipolar affective disorder as their principle diagnosis (ICD-9 nos 296.1, 298.0, 300.4, 301.1, 309.0/1 and 311). The mean age of the patients was 59 years (SD 15) and their marital status was 32% single, 38% married, 11% widowed and 18% divorced. A total of 44% were recruited from the middle social class and 56% from the lower social class.
Instruments
The original Newcastle Scale (NCS),32,33 and its later version NCS-II34 were used to separate patients with endogenous and non-endogenous depression. Compared with other instruments, the validity of the NCS as the measure of endogeneity was most frequently supported.17,35 In comparison to the original NCS, the NCS-II was found to be more reliable and more homogenous; inter-rater reliability assessments yielded kappa = 0.44 for the NCS and kappa = 0.66 for NCS-II. Compared with each other NCS-II had a higher specificity but a lower sensitivity.34 Both scales are composed of 10 items that are given different diagnostic weights.
The level of anxiety was assessed with the help of the Hamilton Anxiety Scale (HAMA),36,37 containing 14 items; psychic and somatic anxiety symptoms are registered with seven items each. Inter-rater reliability measures yielded a weighted mean correlation coefficient of 0.89. Deviating from the original scale all items were rated as ‘absent’ or ‘present’ in this retrospective evaluation. Thus, the total score ranged from 0 (all symptoms absent) to 14 (all symptoms present). As the level of anxiety can fluctuate, its highest degree noted during the patient’s hospital stay was considered.
A full account of conviction records served as a global measure of criminal behavior. The data indicating the type and number of law infringements and sentences were provided by the Swiss Central Criminal Records Department and the Division of Penal Control of the Canton of Bern. The details are given in our previous publication.12 As the criminal records cover a wide variety of crimes and violations (except for trivial offences leading to small fines, e.g., in the case of traffic law violation), the criminal behavior has been categorized and will also be presented by type of offence.
Procedure
Hospital clinical records of all 278 male and female patients were scrutinized by the same investigator (CT), who was blind with regard to the criminal status of the patients. As demonstrated, clinical charts give information of sufficient validity and reliability provided they are properly recorded.38 Our clinical charts were of acceptable quality to allow the completion of both rating scales. The data were evaluated separately for men and women with regard to the relationship of endogeneity and anxiety to criminal behavior. The study design corresponded to the valid regulations of the Swiss Government for the treatment of personal data for scientific purposes and was approved by the representative of the Canton of Berne responsible for data protection. Full confidentiality was strictly maintained throughout the study.
Statistical evaluation
Statistical analysis used the χ2 test (with continuity correction when appropriate) and Fisher’s exact test (two-tailed) for categorical variables. The t-test and the non-parametric Kruskal–Wallis χ2 approximation test were used for continuous variables. Non- parametric multiple stepwise regression analysis was carried out to identify the set of variables best explaining ‘criminal record’ as a dependent variable. The significance level of P < 0.15 was chosen for a variable to enter and to stay in the model.
RESULTS
A total of 71 (40%) out of 179 men and seven (7%) out of 99 women were criminally registered. In men, five (3%) patients were registered because of violent crimes, five (3%) because of sexual offences, 30 (17%) had committed crimes against property, 43 (24%) had violated traffic laws, 10 (6%) had violated drug laws and 31 (17%) had violated other laws. In women, the corresponding figures were one (1%) for violent crimes and one (1%) for crimes against property. Three (3%) patients had violated traffic laws, two (2%) drug laws and three (3%) other laws.
More patients were diagnosed with depression of endogenous type with NCS than with NCS-II scoring 6 or more (NCS) or – 20 or less (NCS-II); in men 51 (28%) versus 39 (22%), in women 35 (35%) versus 33 (33%). The HAMA score range was 0–11 (median 4.5) in the male sample; the HAMA score range was 2–11 (median 5.5) in the female sample.
Regarding the NCS-II superiority over the NCS in some important aspects,30 the results obtained with NCS-II will be presented concerning the above mentioned 39 men and 33 women with endogenous depression and their 140 male and 66 female non-endogenous depressed counterparts. However, very similar results were obtained with NCS. In Table 1, NCS-II endogenous and non-endogenous depression groups are compared with each other separately for men and women with regard to a few basic demographic variables, criminal behavior, and HAMA scores. Patients of both sexes with endogenous depression were significantly older and they less frequently had a criminal record. Table 1 indicates that men with endogenous depression violated significantly less frequently several individual laws, while no significant differences were found between endogenous and non-endogenous groups with regard to the violation of individual laws in women. Patients of both sexes with endogenous depression also scored higher on HAMA; higher scores indicating higher levels of anxiety.
| NCS-II | NCS-II | ||||
|---|---|---|---|---|---|
| endogenous depression | non-endogenous depression | Significance | |||
| Men | n1 = 39 (100) | n2 = 140 (100) | χ2t* | d.f. | P |
| Age (years): Mean ± SD | 63 ± 16 | 49 ± 15 | 5.26* | 177 | < 0.0001 |
| Marital status: married | 29 (74) | 48 (34) | 18.38 | 1 | < 0.0001 |
| Low social class | 17 (44) | 80 (57) | NS | ||
| Criminal record | 6 (15) | 65 (46) | 11.02 | 1 | 0.0009 |
| Violent crimes | 0 (0) | 5 (4) | NS | ||
| Sexual offences | 0 (0) | 5 (4) | NS | ||
| Crimes against property | 1 (3) | 29 (21) | 5.96 | 1 | 0.015 |
| Violations of traffic law | 3 (8) | 40 (29) | 6.19 | 1 | 0.013 |
| Violations of drug law | 0 (0) | 10 (7) | Fisher | 0.080 | |
| Violations of other laws | 2 (5) | 29 (21) | 4.14 | 1 | 0.042 |
| Age at first offencea (years): Mean ± SD | 44 ± 16 | 29 ± 10 | 3.42* | 69 | 0.0005 |
| HAMA score: Mean ± SD | 7.4 ± 2.0 | 5.0 ± 2.2 | 32.08 | 1 | 0.0001 |
| NCS-II | NCS-II | ||||
|---|---|---|---|---|---|
| endogenous depression | non-endogenous depression | Significance | |||
| Women | n1 = 33 (100) | n2 = 66 (100) | χ2t* | d.f. | P |
| Age [years]: Mean ± SD | 65 ± 15 | 56 ± 15 | 2.80* | 97 | 0.006 |
| Marital status: married | 16 (48) | 22 (33) | NS | ||
| Lower social class | 15 (46) | 40 (61) | NS | ||
| Criminal record | 0 (0) | 7 (11) | Fisher | 0.052 | |
| Age at first offenceb (years): Mean ± SD | – | 35 ± 11 | |||
| HAMA score: Mean ± SD | 7.0 ± 1.9 | 5.9 ± 2.0 | 6.53 | 1 | 0.011 |
- Percentages are given in parentheses.
- a n 1 = 6, n2 = 65; b n1 = 0, n2 = 7.
Both the male and female sample were divided by HAMA median into ‘anxious’ and ‘non-anxious’ groups. The proportion of criminal individuals was higher among non-anxious than among anxious men with regard to both total criminal record and several particular laws (Table 2). In contrast, the proportion of criminal individuals was practically the same among non-anxious and anxious women (8%vs 6%; n.s.), where no significant differences were also found with regard to particular laws.
| Anxious group | Non-anxious group | Significance | |||
|---|---|---|---|---|---|
| n1 = 83 (100) | n2 = 96 (100) | χ2 | d.f. | P | |
| Criminal record | 22 (27) | 49 (51) | 10.20 | 1 | 0.0014 |
| Violent crimes | 2 (2) | 3 (3) | NS | ||
| Sexual offences | 0 (0) | 5 (5) | Fisher | 0.062 | |
| Crimes against property | 11 (13) | 19 (20) | NS | ||
| Traffic law | 9 (11) | 34 (35) | 13.41 | 1 | 0.0003 |
| Drug law | 1 (1) | 9 (9) | 4.19 | 1 | 0.041 |
| Other laws | 8 (10) | 23 (24) | 5.41 | 1 | 0.020 |
- Percentages are given in parentheses.
Thus, there seems to be a negative relationship between endogeneity and criminal behavior. However, patients with endogenous depression were older and presented a higher level of anxiety. Age is a well-known variable negatively correlated with criminality and the same seems to be the case for anxiety as evidenced by the relationship between HAMA and criminal records just mentioned. Therefore, the multiple stepwise regression analysis was carried out separately for both sexes, the fact of being criminally registered served as a dependent variable. The following variables were included into the stepwise regression procedure: age, marital status, social class, NCS-II scores and HAMA scores. The results are presented in Table 3. Age and social class as well as the endogeneity scores are included in the final model in both sexes. In the male sample, age makes the highest contribution, followed by the HAMA scores. The proportion of the total variance explained by the model is very modest, however, 13.5% in men and 9% in women.
| Variable | Partial R2 | Model R2 | F | P |
|---|---|---|---|---|
| Men | ||||
| Age | 0.082 | 0.082 | 15.78 | 0.0001 |
| HAMA score | 0.030 | 0.111 | 5.86 | 0.016 |
| NCS-II score | 0.013 | 0.124 | 2.58 | 0.11 |
| Social class | 0.011 | 0.135 | 2.17 | 0.14 |
| Women | ||||
| NCS-II score | 0.038 | 0.038 | 3.84 | 0.053 |
| Social class | 0.028 | 0.067 | 2.94 | 0.09 |
| Age | 0.023 | 0.090 | 2.46 | 0.12 |
- HAMA, Hamilton Anxiety Scale; NCS-II, Newcastle Scale-II
DISCUSSION
A total of 40% of depressed male patients were identified in the present study to have a criminal record. This high proportion will be due to several factors: (i) complete criminal records were evaluated and they encompassed, except for trivial offences, all kinds of criminal behavior including traffic law violations; (ii) 37% of the male sample suffered from minor or intermittent depression (in these patients there could be a considerable comorbidity with personality disorders) and (iii) the majority of patients were recruited from the lower social class.
Our results indicate that even in patients with unipolar affective disorder the criminal rate is higher in the younger age and lower social class. It also depends on the type of depressive illness that the patient suffers from; as had been predicted the criminal rate was found to be lower in patients with endogenous depression. The predicted relationship between lower criminal rate in anxious patients was confirmed for male patients but not for female patients. The inverse relationship between criminal behavior and endogeneity and anxiety was found for a variety of criminal acts. In men with a higher criminal rate, in general the proportion of offending patients was lower in endogenous and anxious groups for all types of offences, even though the level of statistical significance was not always reached. The influence of the demographic and clinical variables on the criminal behavior seems, however, to be only marginal, and demographic variables in terms of age and social class seem to play a more significant role than endogeneity or anxiety.
Some of these findings are in correspondence with the results by other authors. Age composition of the population has been claimed to be the most significant factor affecting crime rate24 and it has been demonstrated that criminality rate declines with age.39 A higher criminality rate in the younger age group was demonstrated for both sexes,40 for criminal offenders with and without a psychiatric diagnosis24 and for offending psychiatric patients.41 In the general population, male sex, younger age and lower social economic status were all found to be significant predictors of violent behavior42 and the correlates of crime among the mentally ill appear to be the same demographic and life-history factors including age and social class that are the correlates of crime among any other group.43 Lower social class increases the risk of criminality in men but not in women.44 In the present study, social class was defined by professional level; correspondingly, lower rearing socio-economic status was found to be related to lower intellectual performance and this relationship mediated social class-criminality relationship.45
The inverse relationship between endogeneity and criminal behavior, which has been confirmed in the present study, corresponds with our previous results.12 Compared with controls a higher criminality rate was found in unipolar patients with minor or intermittent depression but not in unipolar patients with major depression. The proportion of endogenous type depression will be higher in major rather than in minor or in intermittent depression. We are not aware of any other study that directly addresses the question of the relationship between endogeneity and criminality. Regarding anxiety, we found its low level to be correlated with criminal behavior. However, this could be demonstrated for the male sample only and the contribution of anxiety scores to the ‘criminal record’ variance was only marginal. The inverse relationship between psychopathy and anxiety has not always been confirmed,46 but also, using HAMA (i.e., a clinical rating scale), state anxiety was assessed. The decision to do so was based on the type of data available for the analysis in the clinical records and assessed retrospectively. Even though the state and the trait anxiety commonly merge into one another or occur together,47 it might be preferable to assess trait and not state anxiety to attain a more pronounced result. Similarly, the endogeneity will also only play the role of an interposed variable mediating the influence of certain personality features. Thus, and understandably so, the low impact of both the endogenous type of depression and the state of anxiety on criminality will be explainable by their indirect, only mediating effect they exert. We would expect the explanatory power of the variables they probably stand for (i.e. the personality of melancholic type and the trait anxiety) to be higher regarding the occurrence of criminal behavior. While the latter factors operate in a constant way, depression and anxiety mostly represent only temporary states which again, except for rare instances of illness-bound criminal acts, reduces their potential effect on criminal behavior.
CONCLUSIONS
Even though there seems to be a negative relationship of criminal behavior to endogenous type of the illness and the presence of anxiety syndrome in patients with unipolar depression including intermittent depression, this relationship appears to be very weak. In fact, sociodemographic factors in terms of age and social class seem to make a greater contribution and to explain a higher proportion of variance with regard to the criminal behavior as a dependent variable.
