PREVALENCE AND CLINICAL CHARACTERISTICS OF PSYCHOTIC VERSUS NONPSYCHOTIC MAJOR DEPRESSION IN A GENERAL PSYCHIATRIC OUTPATIENT CLINIC

Abstract

Background

Psychotic major depression (PMD) is a severe mental disorder characterized by high levels of illness severity, chronicity, impairment, and treatment resistance. However, most past research on PMD has been conducted in inpatient hospital samples, and relatively little is known about PMD patients presenting for treatment in the community specifically.

Methods

In this study, we examined the prevalence and clinical characteristics of PMD in a large sample (n = 2,500) of treatment-seeking outpatients who were administered structured clinical interviews by trained diagnosticians.

Results

Of the patients diagnosed with major depression, 5.3% had psychotic features. PMD patients were more likely to be members of a racial/ethnic minority and to have lower educational attainment compared to those with nonpsychotic major depression. In addition, PMD patients were found to have greater current depression severity, suicidal ideation, and social and work impairment. These patients also were more likely to have histories of suicide attempts and psychiatric hospitalizations, to report an earlier age of illness onset, and to meet criteria for chronic depression. In terms of psychiatric comorbidity, PMD patients had higher rates of certain anxiety disorders as well as more somatoform and cluster A personality disorders.

Conclusions

Results indicated that PMD was present in a relatively small percentage of treatment-seeking outpatients but was associated with disproportionately high levels of severity and impairment. Similarities and differences between the current findings and those from past research are discussed, including clinical implications for the identification and treatment of PMD in routine practice settings.

INTRODUCTION

Epidemiological research suggests that 15–19% of patients with major depression exhibit hallucinations and/or delusions.^{[1, 2]} Rates tend to be even higher in hospitalized samples, with Coryell et al.^[3] reporting that 25% of consecutively admitted depressed inpatients met criteria for psychotic major depression (PMD). As major depression is the second most common mental disorder with a lifetime population prevalence of approximately 16%,^[4] PMD clearly affects a substantial number of individuals.

Over the years, a number of studies have contrasted the clinical characteristics of patients with PMD versus nonpsychotic major depression (NMD).^[5–8] PMD patients have been found to exhibit greater overall depression severity and a symptom profile characterized by certain melancholic symptoms.^{[5, 9–11]} In addition, PMD patients tend to be characterized by higher levels of suicidality,^[11] anxiety,^[12] hypochondriasis,^[3] dysfunctional cognitions,^[13] and cluster A personality disorders^[14] compared to individuals with NMD. In general, studies have failed to find many demographic differences among patients with PMD versus NMD,^[5] although there are some exceptions (e.g., Serretti et al.^[14]). In addition, several studies have reported poorer social and occupational functioning in PMD relative to NMD,^{[15, 16]} but others have failed to demonstrate these differences.^{[13, 17]} Studies examining differences in family history of psychiatric illness have also produced equivocal results.^{[18, 19]}

Not surprisingly, PMD patients appear to have a poorer course of illness and higher rates of treatment resistance.^[20] For example, research has shown that these patients spend more time depressed^[17] and have higher relapse rates^{[21, 22]} compared to those without psychotic features. In addition, PMD patients tend to respond less well to antidepressant treatment^[23] and psychotherapy.^[24] Because of the multiple differences typically found between patients with PMD and NMD, some have argued that PMD should be considered a diagnostically distinct form of illness, instead of a subtype of major depression.^[8] Proponents of this viewpoint emphasize the differences often found among depressed patients with versus without psychotic features on various neurobiochemical markers, including greater hypothalamic–pituitary–adrenal axis dysfunction sometimes reported in PMD.^[25] In addition, a number of clinical differences commonly observed among PMD and NMD patients have been shown to be present after controlling for the general severity of illness.^[7] In contrast, others argue for a “quasi-dimensional” model of psychotic symptoms based on research showing that psychotic-like experiences and schizotypal traits are relatively prevalent in the general population and suggest that psychotic mood disorders may fall along a continuum between affective and psychotic illnesses.^[26]

Regardless of the current diagnostic debate, PMD clearly represents a severe form of illness, and its identification and treatment in psychiatric outpatient settings is critical because of the potential risk of future hospitalization^[1] and even mortality.^[27] However, research suggests that patients with PMD are likely to be under-identified in routine practice settings, as psychotic symptoms can be more subtle in mood disorders, and patients may be reluctant to disclose these symptoms because of paranoia or embarrassment without systematic probing from clinicians.^[20] Nevertheless, most of the past literature investigating differences among PMD and NMD patients has been conducted exclusively in inpatient hospital samples (e.g., Coryell et al.^[3]) or less frequently, in samples composed of both inpatients and outpatients taken from specialty clinics (e.g., Thakur et al.^[11]). The clinical characteristics of general psychiatric outpatients with PMD may vary due to potential differences in treatment-seeking behaviors or the treatment setting itself. Unfortunately, relatively little is known about the clinical presentation of PMD patients being treated in outpatient clinics specifically.

In this study, we report on the prevalence and clinical characteristics of patients diagnosed with PMD by examining a large sample (n = 2,500) of general psychiatric outpatients who were seeking treatment. The current data were collected as part of the Methods to Improve Diagnostic Assessment and Services (MIDAS) project, which represents an integration of research methods into a community-based outpatient practice affiliated with an academic university.^[28] The use of reliable and valid structured clinical interviews is essential for identifying PMD patients due to the frequent problems with differential diagnosis in this population.^[29] Subjects participating in the MIDAS project completed a comprehensive assessment battery during intake that included structured clinical interviews administered by trained diagnosticians. In the present investigation, outpatients with PMD versus NMD were compared on measures of demographic characteristics, symptom severity, functional impairment, history of illness, psychiatric comorbidity, and family psychiatric history. In addition, the relative importance of various clinical features of PMD was examined in an effort to identify characteristics reliably associated with the diagnosis after taking into account the overall severity of illness.

METHODS

SAMPLE

Participants included 2,500 psychiatric patients presenting for treatment at the outpatient practice of the Rhode Island Hospital Department of Psychiatry. The sample consisted of 1,514 females (60.6%) and 986 (39.4%) males, ranging in age from 18 to 85 (M = 38.3, SD = 12.8). The majority of the sample was Caucasian (n = 2,189; 87.6%), followed by African American (n = 112; 4.5%), Portuguese (n = 80; 3.2%), Hispanic (n = 65; 2.6%), other or mixed ethnicities (n = 35; 1.4%), and Asian (n = 19; 0.8%). Many participants were married (n = 1,040; 41.6%), followed by never married (n = 774; 31.0%), divorced (n = 371; 14.8%), separated (n = 141; 5.6%), living as if married (n = 128; 5.1%), and widowed (n = 46; 1.8%). Over half of the sample (n = 1,573; 62.9%) had a high school degree or equivalency, whereas 355 (14.2%) received a 4-year college degree, 328 (13.1%) had a graduate degree/some graduate education, and 244 (10%) did not graduate from high school. The most frequent, current axis I Diagnostic and Statistical Manual (DSM)-IV diagnoses were NMD (n = 1,054; 42.2%), social phobia (n = 690; 27.6%), generalized anxiety disorder (n = 428; 17.5%), panic disorder with agoraphobia (n = 339; 13.6%), posttraumatic stress disorder (PTSD; n = 315; 12.6%), specific phobia (n = 258; 10.3%), alcohol abuse (n = 202; 8.1%), dysthymic disorder (n = 189; 7.6%), and obsessive–compulsive disorder (OCD; n = 179; 7.2%).

MEASURES

The Structured Clinical Interview for DSM-IV axis I disorders (SCID)^[30] and the Structured Interview for the DSM-IV Personality (SIDP-IV)^[31]; were used to diagnosis psychiatric disorders. The Clinical Global Impression Scale (CGI)^[32]; is an interviewer-rated measure of illness severity based on a 6-point anchored scale, ranging from 0 (none) to 5 (extremely ill). The CGI was rated specifically for depression severity based on the depressive symptoms endorsed in the SCID and not on other symptoms. Social functioning and suicidal ideation were rated by diagnosticians using items from the Schedule for Affective Disorders and Schizophrenia (SADS).^[33] Social functioning (past month) was rated on a 7-point Likert scale, ranging from 1 (superior) to 7 (grossly inadequate). Suicidal ideation (past 2 weeks) was rated on a 7-point Likert scale, ranging from 1 (not at all) to 7 (very extreme). Time out of work (past 5 years) was rated by interviewers on a 10-point Likert scale from the SADS, ranging from 0 (not expected to work) to 9 (worked none or practically none due to psychopathology). Patients not expected to work (e.g., students, those on disability for medical reasons) were excluded from this analysis, thus resulting in a range from 1 (virtually no time out of work) to 9 (worked none). Family history of psychiatric diagnoses was based on patient interviews using the Family History Research Diagnostic Criteria.^[34] Diagnoses assessed included affective, anxiety, psychotic, and substance use disorders for all of the first-degree relatives of patients in the study.

PROCEDURE

New patients to the clinic were offered the opportunity to have a more comprehensive evaluation as part of the clinical-research program, although they were not required to undergo this evaluation. Therefore, not all patients who presented for treatment participated in the study. The varying number of trained diagnostic interviewers available influenced the number of patients who were invited to participate. Also, because one of the goals of the MIDAS project is to develop and study the reliability and validity of self-administered questionnaires, patients with significant cognitive limitations were not included. Nonetheless, as reported elsewhere, patients who did and did not participate in the study were similar in scores on self-administered symptom questionnaires.^[34] Of particular importance, patients who did and did not participate in the MIDAS project did not differ in their scores on the Psychiatric Diagnostic Screening Questionnaire, a self-administered scale that screens for 13 DSM-IV axis I disorders.^{[36, 37]}

All participants were evaluated with the full SCID and a subsample was evaluated with the SIDP-IV (as this assessment was instituted at a later point in the study). Diagnosticians had bachelor’s degrees in the social or biological sciences or were doctoral-level clinical psychologists. Diagnosticians were trained for a period of 3 months, which included reviewing written cases, discussing item-by-item administration with the principal investigator (M. Z.), observing at least five interviews, and administering 15–20 interviews while being observed and supervised. Diagnosticians were then required to demonstrate exact or near-exact inter-rater reliability with a senior diagnostician for five consecutive interviews. Diagnosticians received ongoing supervision of interviews via a weekly case conference. PMD was diagnosed according to DSM-IV criteria, which were assessed based on the Mood and Psychotic Modules of the SCID. Furthermore, diagnosticians carefully considered the differential diagnosis of PMD versus co-occurring conditions that can be confused with the disorder (e.g., bipolar disorder, schizoaffective disorder, substance-induced mood disorder, OCD, PTSD).

Inter-rater reliability information was collected over the course of the entire project. From 47 joint-interview reliability evaluations of the SCID, the reliability coefficients of the major axis I disorders were major depression κ = 0.91; panic disorder κ = 1.0; social phobia κ = 0.84; OCD κ = 1.0; specific phobia κ = 0.93; generalized anxiety disorder κ = 0.93; PTSD κ = 0.91; alcohol abuse/dependence κ = 0.64; drug abuse/dependence κ = 0.73; impulse control disorders κ = 1.0; and somatoform disorder κ = 1.0. Reliability estimates for the personality disorders from 29 joint interviews of the SIDP-IV were any personality disorder κ = 0.77; cluster A κ = 1.0; cluster B κ = 0.61; and cluster C κ = 0.97. In addition, reliabilities for SIDPIV dimensional scores were very high (ICC = 0.89–0.97).

STATISTICAL ANALYSES

The two diagnostic subgroups (PMD versus NMD) were compared using independent samples t-tests or χ² tests on demographic variables, comorbid axes I and II disorders, current Global Assessment of Functioning (GAF), CGI for depression, age of onset of major depression, chronic depression (>2 years in episode), current suicidal ideation, s, past psychiatric hospitalizations, time out of work, current social functioning, and family history of psychiatric disorders. Cohen’s d statistic or odds ratio (OR) and their 95% confidence intervals were also computed for group differences to describe the magnitude of effects. All tests were two-tailed, and α was set a priori at .01 to reduce the familywise error due to multiple comparisons. A post hoc power analysis indicated that statistical power was .88 for detecting a medium effect size (d = 0.5) between the groups (α = .01), which was judged to be a clinically meaningful difference. Adjusted t-test values were reported if Levene’s test for inequality of variances was found to be significant. Logistic regression analysis based on a forward selection method was used to examine variables that differentiated the diagnostic groups.

Morbid risks (MR) for family history of psychiatric disorders were computed using Weinberg’s shorter method.^[38] An age-corrected denominator (or Bezugsziffer) is calculated, in which relatives over the risk period for a disorder (U₃) make twice the contribution of those within the risk period (U₂), and those below the risk period are not counted (U₁). MR is calculated by dividing the number of affected relatives (A) by the age-corrected number of relatives (Bezugsziffer): MR = A/(A + U₂/2 + U₃). The period of risk for major depression was 25–44 years, for bipolar disorder was 15–50, for anxiety disorders was 8–44, for psychotic disorders was 15–50 years, and for substance use disorders was 20–34 based on the age distribution of illness onset in the probands^[39] and past research.^{[38, 40]}

RESULTS

DIAGNOSTIC PREVALENCE AND FREQUENCY OF PSYCHOTIC SYMPTOMS

Of the total sample (n = 2,500), 42.1% (n = 1052)¹ were diagnosed with NMD and 2.4% (n = 60) with PMD based on lifetime history. Thus, 5.3% of patients diagnosed with major depression in the current sample reported psychotic features. Of the patients with PMD who had detailed psychotic symptom data available (n = 57), 68% (n = 41) reported hallucinations in the past month and 80% (n = 48) reported a lifetime history of hallucinations. In addition, 20% (n = 12) reported delusions in the past month and 32% (n = 19) reported a lifetime history of delusions. Seventeen percent (n = 10) reported both hallucinations and delusions during their lifetime. See Table 1 for a breakdown of hallucination and delusion subtypes. The most frequently reported type of hallucination was auditory, with 65% (n = 39) reporting a lifetime history of this symptom. Persecutory (25%; n = 15) was the most frequent type of delusion reported in the sample. Seventy-seven percent (n = 46) of PMD patients reported psychotic symptoms within the past month, whereas the remaining patients reported a past history of psychotic symptoms during a depressive episode according to the SCID.

TABLE 1.

Types of psychotic symptoms in patients with psychotic major depression

	Past month		> Past month
	%	n	%	n
Hallucinations
Any	68.3	41	25.0	15
Auditory	53.3	32	11.7	7
Visual	21.7	13	3.3	2
Tactile	21.7	13	6.7	4
Olfactory	13.3	8	6.7	4
Gustatory	3.3	2	5.0	3
Delusions
Any	20.0	12	15.0	9
Persecutory	15.0	9	10.0	6
Reference	11.7	7	5.0	3
Bizarre	3.3	2	1.7	1
Being controlled	3.3	2	0	0
Thought insertion	3.3	2	0	0
Thought broadcasting	0	0	3.3	2
Jealous	0	0	1.7	1
Mind reading	0	0	1.7	1

Note: n = 57;

psychotic symptoms based on Structured Clinical Interview for DSM-IV axis I disorders interview.

DEMOGRAPHIC CHARACTERISTICS

See Table 2 for the results for demographic and clinical variables between groups. Analyses failed to identify significant differences on most demographic variables, including age, gender, or marital status. However, a significant difference was found on race/ethnicity, with significantly more non-Caucasian patients being diagnosed with PMD (P<.001; OR = 3.4). In addition, a significant difference was found for education, with PMD patients being less likely to have obtained a post-high school degree (P = .01; OR = 2.2). These results suggest that PMD patients were similar to NMD patients overall on demographic variables, although they were more likely to be a member of a racial/ethnic minority and to not have obtained a college degree.

TABLE 2.

Comparisons of demographic and clinical variables in patients with psychotic versus nonpsychotic major depression

	PMD (n = 60)		NMD (n = 1,052)		Analysis
Variablesa	M/%	SD/n	M/%	SD/n	t/χ2	dfs	P	Effect	95% CI
Demographic
Age	37.0	11.7	39.5	12.2	1.52	1,110	.129	d = 0.21	0.05–0.47
Gender (%)
Male	26.7	16	34.5	364	1.56	1	.211	OR = 1.45	0.81–2.61
Female	73.3	44	65.5	690
Race/ethnicity (%)
Caucasian	65.0	39	86.1	907	45.44	5	.000**	OR = 3.37b	1.92–5.89
African American	13.3	8	5.0	53				OR = 2.90c	1.31–6.41
Hispanic	18.3	11	3.0	32				OR = 7.16d	3.41–15.0
Portuguese	0	0	0.6	6
Asian	1.7	1	3.4	37
Other	1.7	1	1.8	19
Education (%)
Less than HS	15.0	9	11.1	117	13.38	4	.010*	OR = 2.20e	1.30–3.71
HS graduate	40.0	24	24.6	259
Some college	31.7	19	30.6	322
College graduate	10.0	6	26.3	277
Graduate degree	3.3	2	7.5	79
Marital status (%)
Single	21.7	13	26.9	284	3.93	3	.269	OR = 1.20f	0.71–2.03
Married/living together	43.3	26	47.8	504
Divorced/separated	30.0	18	23.1	243
Widowed	5.0	3	2.2	23
Clinical
Age of onset	20.7	12.1	26.1	13.8	3.00	1,110	.003*	d = 0.42	0.16–0.68
CGI (depression)	3.8	.5	3.1	.6	8.42	1,112	.000**	d = 1.27	1.00–1.53
GAF (current)	37.5	9.6	50.4	7.6	10.22j	1,111	.000**	d = 1.50	1.22–1.76
Social impairment (current)	4.0	1.7	3.2	1.2	3.29j	1,109	.002*	d = 0.54	0.28–0.80
Chronic depression (%)	63.3	38	31.6	332	25.75	1	.000**	OR = 3.74	2.18–6.42
Current SI (%)	16.7	10	4.8	51	15.29	1	.000**	OR = 3.93	1.88–8.19
Past suicide attempts (%)
0	63.3	38	88.0	925	49.45	2	.000**	OR = 4.25g	2.43–7.42
1	15.0	9	8.6	90
>1	26.5	13	3.4	36
Past hospitalizations (%)
0	45.8	27	73.6	791	29.96	2	.000**	OR = 2.48h	1.46–4.20
1	25.4	15	15.8	161
>1	28.8	17	10.5	100
Time out of work (%)
Virtually none	14.3	7	31.6	304	11.11	2	.004*	OR = 2.34i	1.29–4.36
<1 year	51.0	25	40.1	481
>1 year	34.7	17	18.3	176

Note α = .01;

^*P<.01;

^**P<.001.

PMD, psychotic major depression; NMD, nonpsychotic major depression; HS, high school; CGI, Clinical Global Impression Scale; GAF, Global Assessment of Functioning Scale; chronic depression = episode > 2 years; SI, suicidal ideation; OR, odds ratio.

^aSample sizes for some analyses vary slightly due to missing data.

^bCaucasian versus non-Caucasian.

^cAfrican American versus non-African American.

^dHispanic versus non-Hispanic.

^eCollege versus no college degree.

^fMarried/living together versus not married.

^gHospitalized versus not hospitalized.

^hSuicide attempt versus no suicide attempt.

ⁱ<1 year out of work versus >1 year out of work.

^jEqual variances not assumed.

SYMPTOM SEVERITY, FUNCTIONAL IMPAIRMENT, AND HISTORY OF ILLNESS

In terms of illness severity, the PMD group had significantly higher current GAF scores (P<.001; d = 1.5) and greater depression severity based on the CGI compared to the NMD group (P<.001; d = 1.3). The age of onset of major depression was significantly lower in the PMD group compared to the NMD group (P<.01; d = 0.4). In addition, patients with PMD were significantly more likely to be experiencing chronic depression, defined by the DSM-IV as an episode lasting for more than 2 years (P<.001; OR = 3.7). Furthermore, results showed that PMD patients had higher levels of current suicidal ideation (i.e., SADS ≥ 3; clear suicidal intent or plan) (P<.001; OR = 3.9) and were more likely to have a history of suicide attempts (P<.001; OR = 4.3) compared to the NMD group. PMD patients were also significantly more likely to have been hospitalized in the past for psychiatric reasons (P<.001; OR = 2.5). In terms of functional impairment, PMD patients were significantly more impaired socially (P<.001; d = 0.5). PMD patients were also significantly more likely to exhibit chronic work impairment (>1 year out of work due to psychiatric illness) compared to NMD patients (P<.01; OR = 2.3). These results clearly demonstrate that patients with PMD were more severely ill at the time of intake, had a poorer past course of illness, and had higher levels of functional impairment due to their illness (see Table 2).

PSYCHIATRIC COMORBIDITY

See Table 3 for group differences on the most frequent axis I and II comorbid diagnoses in the sample. Results indicated that PMD patients had significantly higher rates of PTSD (OR = 2.3) and OCD (OR = 2.6) (P<.01). In addition, the PMD group was significantly more likely to be diagnosed with a somatoform disorder (P<.01; OR = 2.7). No significant differences were found between groups for rates of other anxiety disorders, substance use disorders, eating disorders, or impulse control disorders. However, there were trends observed for higher rates of panic disorder and social anxiety disorder in PMD patients, in contrast to higher rates of generalized anxiety disorder in NMD patients. Personality disorder rates were analyzed based on their clusters using the subsample with available data (n = 793). Results demonstrated that the PMD had a significantly higher rate of cluster A personality disorders (P<.01; OR = 3.3); but there were no significant differences for clusters B or C. Follow-up analyses demonstrated that the only individual cluster A personality disorder that was significantly higher in the PMD group was paranoid personality disorder (13.5 versus 4.4%, χ² = 6.32, df = 1, P = .01; OR = 3.4, 95% CI = 1.2–9.2).

TABLE 3.

Lifetime psychiatric comorbidity in patients with psychotic versus nonpsychotic major depression

	PMD (n = 60)		NMD (n = 1,052)		Analysis
Variables	%	n	%	n	χ2 (df = 1)	P	OR	95% CI
Axis I
Posttraumatic stress disorder	56.7	34	25.0	263	29.08	.000**	3.92	2.31–6.66
Social anxiety disorder	48.3	29	36.4	383	3.46	.063	1.63	0.97–2.75
Panic disorder	43.3	26	28.5	300	6.01	.014	1.92	1.13–3.25
Obsessive–compulsive disorder	21.7	13	9.5	100	9.20	.002*	2.63	1.38–5.03
Specific phobia	20.0	12	13.3	140	2.15	.142	1.63	0.84–3.14
Generalized anxiety disorder	8.3	5	21.4	225	5.90	.015	2.99	1.18–7.56
Substance use disorders	55.0	33	44.2	465	2.68	.102	1.54	0.91–2.60
Eating disorders	21.7	13	14.3	150	2.49	.115	1.66	0.88–3.15
Somatoform disorders	20.0	12	8.5	89	9.15	.002*	2.71	1.39–5.28
Impulse control disorders	16.7	10	23.7	249	1.56	.212	1.55	0.77–3.10
Axis IIa
Cluster A	17.1	6	5.9	45	6.98	.008*	3.28	1.29–8.30
Cluster B	20.9	7	15.2	115	0.61	.437	1.40	0.60–3.28
Cluster C	32.4	11	24.5	187	1.06	.303	1.47	0.70–3.01

Note α = .01;

^*P<.01;

^**P<.001.

PMD, psychotic major depression; NMD, nonpsychotic major depression; OR, odds ratio.

^aAnalyses vary based on the subsample with available data.

FAMILY HISTORY OF ILLNESS

We compared the MR for major depression, bipolar disorder, anxiety disorders (panic disorder, agoraphobia, generalized anxiety disorder, PTSD, specific phobia, social phobia, or OCD), schizophrenia-spectrum disorders (schizophrenia, schizoaffective disorder, or psychosis not otherwise specified), and substance use disorders (alcohol abuse/dependence or drug abuse/dependence) in the first-degree relatives of those with PMD versus NMD. Family members were considered positive for the disorder if they received treatment for it. Table 4 reports that there were no significant differences between the groups on any of the disorders examined. Thus, individual psychotic, anxiety, or substance use disorders were not tested separately to reduce the overall number of tests conducted.

TABLE 4.

Morbid risks for psychiatric disorders in first-degree relatives of patients with psychotic versus nonpsychotic major depression

	PMD (n = 59)			NMD (n = 1,032)			Analysis
Diagnoses	A	BZ	MR (%)	A	BZ	MR (%)	χ2 (df = 1)	P	OR	95% CI
Major depression	27	212.0	12.7	702	3,955.0	17.8	2.56	.110	1.39	0.93–2.10
Bipolar disorder	6	206.0	2.9	69	3,784.0	1.8	1.20	.285a	1.60	0.69–3.72
Anxiety disorder	8	238.0	3.4	290	4,297.5	6.7	3.80	.051	2.01	0.98–4.10
Psychotic disorder	2	202.0	1.0	23	3,769.0	0.6	0.44	.376a	1.63	0.38–6.93
Substance use disorder	22	238.0	9.2	323	4,385.0	7.4	0.83	.363	1.25	0.80–1.97

Note α = .01;

A = number of affected relatives; BZ = age-corrected number of relatives; MR = morbid risk (A/BZ); OR, odds ratio.

^aFisher’s exact test used as expected values in some cells were less than 5.

PREDICTORS OF PSYCHOTIC SUBTYPE

A logistic regression analysis was conducted to examine the relative utility of various clinical and demographic variables in differentiating between patients with PMD versus NMD. Demographic and clinical variables found in previous analyses to be significantly different between the groups were tested: education (college degree versus no college degree), race/ethnicity (Caucasian versus non-Caucasian), PTSD (yes versus no), OCD (yes versus no), any somatoform disorder (yes versus no), chronic work impairment (yes versus no), chronic depression (yes versus no), current suicidal ideation (yes versus no), CGI for depression, current GAF, current social impairment, number of past suicide attempts, number of past psychiatric hospitalizations, and age of major depression onset. Cluster A personality disorders were not included in the regression because data were available only for a subsample of patients. The aim was to identify which variables were most useful for differentiating between the diagnostic subtypes while controlling for the influence of the other variables in the model. The correlation matrix for variables included in the regression was examined and no problems with multicolinearity were identified. The Hosmer–Lemeshow test (where P>.05 is expected) showed acceptable model fit (χ² = 3.96, df = 8, P = .86). The final model explained 39% (Nagelkerke R²) of the variability in major depression subtypes. Results from the logistic regression analysis are presented in Table 5. Race/ethnicity, education, depression CGI, current GAF, and chronic depression variables significantly differentiated between the diagnostic groups while controlling for the other variables in the model (P<.01). PMD patients were more likely to be a member of a racial/ethnic group, have less than a college degree, be diagnosed with chronic depression, and have greater current depression severity (CGI) and poorer overall functioning (GAF).

TABLE 5.

Logistic regression of demographic and clinical variables associated with major depression subtype

Variables	β	SE	Wald χ2 (df = 1)	P	Hazard ratio	95% CI
In the final model
Race/ethnicity (non-Caucasian)	−1.08	0.35	9.85	.002	0.34	0.17–0.67
Education (college degree)	1.22	0.46	6.97	.008	3.38	1.37–8.33
Clinical Global Impression (depression)	1.02	0.29	12.34	.000	2.77	1.57–4.88
Global assessment of functioning (current)	−0.12	0.02	46.07	.000	0.88	0.85–0.92
Chronic depression (>2 years)	−0.88	0.32	7.67	.006	0.41	0.22–0.77
Not in the final model
Posttraumatic stress disorder	3.80			.051
Obsessive–compulsive disorder	2.08			.150
Somatoform disorder	0.28			.598
Current social impairment	0.95			.330
Current suicidal ideation	1.61			.204
Past suicide attempts	0.12			.732
Past psychiatric hospitalizations	2.17			.141
Time out of work	0.21			.650
Age of illness onset	0.23			.633
Constant	0.56	1.61		.728	0.57

Note. α = .01, n = 1,112.

Overall model: χ² = 157.68, df = 5, P<.001, −2 log likelihood = 303.19, Nagelkerke R² =.39.

DISCUSSION

Comprehensive diagnostic assessments indicated that PMD was present in 2.4% of the current treatment-seeking outpatient sample or 5.3% of those diagnosed with major depression specifically. PMD patients were found to differ on certain demographic variables, including race/ethnicity and educational attainment. Overall, PMD patients tended to be more severely ill currently, to have a worse past course of illness, and to show higher functional impairment and disability related to their illness. PMD patients also had higher rates of certain comorbid conditions, although no differences were found for family history of disorders. After controlling for a variety of demographic and clinical variables, race/ethnicity, education, current depression severity, global functioning, and depression chronicity remained significantly associated with diagnostic status.

Rates of PMD found in the current sample were somewhat lower than those reported in previous treatment-seeking samples 16–54%.^[41] The lower rate of PMD was not completely unexpected as the highest rates of the disorder generally have been found in inpatients. However, the prevalence of PMD in the current sample was lower than that demonstrated in previous epidemiological studies, which have reported that between 15 and 19% of people with major depression have psychotic features.^{[1, 2]} There are several potential explanations for this discrepancy. First, the use of structured clinical interviews administered by trained diagnosticians in this study may have improved differential diagnosis of PMD patients, especially in comparison with the diagnostic methods employed in past large-scale epidemiological surveys.^[41] Alternatively, it is possible that depressed patients with psychotic features are less likely to seek treatment in outpatient settings,^[27] which may partially explain their higher prevalence in hospital samples. In addition, the lower PMD prevalence in our sample may have been the result of the particular characteristics of patients treated at the recruitment site. The psychiatry outpatient practice predominantly treats patients with medical insurance (including Medicare) on a fee-for-service basis. It is possible that PMD patients are disproportionately uninsured or underinsured. Moreover, the highest rates of PMD have been reported in elderly clinical samples. However,^{[41, 43]} the average age of patients with this diagnosis in this study was 37, and the clinic does not specialize in the treatment of geriatric patients.

The finding that significantly more PMD patients were members of racial/ethnic minorities is consistent with past epidemiological research showing a higher proportion of African American patients with PMD versus NMD.^[1] In a previous report from this data set based on a smaller subsample of PMD patients (n = 15), a significantly higher number of Hispanic compared to Caucasian patients were found to have a PMD diagnosis.^[44] These results were extended in the current larger sample of PMD patients (n = 60). Given the broader literature showing racial/ethnic differences in primary psychotic disorders, the potential role of culture in the presentation and interpretation of symptoms requires further investigation.^[45] The lower proportion of minority patients in our sample could also account for the lower prevalence of PMD found. Furthermore, results demonstrated that PMD patients were less likely to have obtained a college degree. Serretti et al.^[14] also found lower educational attainment in their sample of PMD patients. It is likely that these ethnic and educational differences sometimes found between PMD and NMD patients may be at least partly explained by differences in socioeconomic status, as shown by Johnson et al.^[1] However, it should be noted that race/ethnicity remained significantly associated with diagnosis even after education level, which represents a proxy for socioeconomic status, was controlled for in analyses.

Consistent with past research^{[9, 11]} and as expected, PMD patients had greater overall depression severity. Although some studies have failed to find differences in the onset of illness,^{[1, 14]} PMD patients in this study had a significantly earlier age of depression onset compared with NMD patients. The finding that PMD patients had nearly four times greater odds than NMD patients of chronic depression was consistent with past research demonstrating that PMD patients tend to experience much longer depressive episodes.^{[21, 22]} Furthermore, past research has indicated that patients with PMD have higher levels of suicidal ideation^[11] and are two to five times more likely to have a suicide attempt ^{[1, 46]} compared to those with NMD. This is consistent with current results showing that PMD patients had significantly higher current suicidal ideation severity as well as over four times greater odds of having a past suicide attempt.

Goldberg and Harrow^[47] reported that functional impairment in PMD patients contributed to poorer perceived quality of life. However, past comparisons of PMD and NMD patients on measures of psychosocial functioning have produced conflicting results. Some studies have reported that PMD patients possess greater functional impairment over long-term follow-up,^{[16, 46]} whereas others have failed to find functional differences over time in comparisons with NMD patients.^[18] In this study, PMD patients had significantly poorer social functioning and were more likely to evidence chronic occupational impairment (i.e., greater than 1 year out of work due to psychiatric illness). These discrepant findings in the literature may be explained by the clinical characteristics of different PMD samples. For example,^[13] recently failed to find greater psychosocial impairment in PMD compared with NMD patients currently hospitalized, most likely because of the overall high severity of illness observed in the sample and possible “ceiling effects” on these measures. The current outpatient sample had a greater heterogeneity of illness, which may have increased the ability to differentiate PMD and NMD patients on measures of functional impairment.

This study represents one of the most comprehensive and detailed examinations of psychiatric comorbidity in PMD patients that has been reported in the literature to date. Patterns of comorbidity found in the current PMD sample, including the higher rates of anxiety and somatoform disorders, were largely consistent with past research. For example, other studies have reported elevated anxiety and hypochondriasis symptoms in PMD patients.^{[3, 9, 12]} Furthermore Johnson et al.^[1] found that PMD patients had higher rates of simple phobia, OCD, and somatization disorder. Consistent with the findings of Serretti et al.^[14] PMD patients in this study also had higher rates of cluster A disorders, particularly paranoid personality disorder. In contrast, we failed to find any consistent evidence for differences between PMD and NMD patients in family history of psychiatric illness. Although some studies have found higher family prevalence of affective disorders (e.g., Leckman et al.^[19]) the majority have not.^[5] It should be noted that self-reports of family history in this study were not independently verified.

The finding of higher rates of PTSD in patients with PMD extended the findings from a previous report from this data set that examined a smaller subsample (n = 19) of these patients.^[48] It also is consistent with other research showing similar trends of high PTSD comorbidity in PMD.^[49] In the current study, diagnosticians were trained to carefully distinguish between psychotic symptoms and common PTSD symptoms, such as flashbacks and dissociative experiences. Only when the perceptual disturbances were outside the realm of the trauma-related material was PMD diagnosed. Still, high rates of PTSD comorbidity remained in the PMD sample, suggesting that trauma exposure may represent a diathesis for psychosis, which is suggested by emerging evidence.^[50]

Especially in patients with mood disorders, clinicians often focus on the differentiation of “real” or “true” psychotic symptoms from the so-called “pseudo” symptoms (e.g., transient psychotic experiences sometimes reported in individuals with borderline personality disorder). However, research suggests that patients classified as having “false” psychotic symptoms are still at risk for the later development of threshold psychotic symptoms and the need for treatment.^[51] Furthermore, research examining internally versus externally perceived hallucinations fails to support the validity of this often used clinical heuristic.^[52] Based on research showing higher than expected rates of “psychotic-like” experiences and schizotypal personality traits in the general population, there is a growing recognition that psychotic symptoms may best be viewed as falling along a continuum of severity.^[26] The study of patients with PMD may be particularly helpful for investigating the quasi-dimensional qualities of psychotic symptoms.^[13]

One interesting finding was that the majority of PMD patients in this sample exhibited hallucinations (80%) in contrast to delusions (32%). Current DSM-IV criteria require only that delusions or hallucinations to be present for a diagnosis of PMD. However, early studies on PMD tended to focus on samples of patients with delusional depression specifically.^[53] Furthermore, many past studies of PMD samples have included patients with higher rates of delusions relative to hallucinations.^{[2, 3, 14]} Again, these differences in psychotic feature frequencies may be attributable to our examination of an entirely outpatient sample; whereas most previous studies were conducted in acutely ill inpatient samples. As discussed above, the differences observed between PMD and NMD outpatients in this study were largely consistent with those reported in past research in inpatient samples. However, some of the discrepancies in our findings may be partly due to the types and frequencies of psychotic symptoms in the current sample. Some have argued that there may exist important differences between PMD patients exhibiting primarily delusions versus hallucinations.^[54] It will be important for future research to compare and contrast the clinical characteristics of PMD patients based on their psychotic symptom presentation.

There are several important clinical implications that can be derived from the current findings. First, results indicated that DSM-IV diagnoses of PMD were present in a relatively small group of treatment-seeking outpatients, but that these individuals were experiencing very high levels of severity and impairment. PMD outpatients can be viewed as similar to other severely mentally ill populations, such as schizophrenia and bipolar disorder, with lower absolute prevalence rates but disproportionately high levels of illness chronicity and associated disability. The high rates of comorbidity found in PMD patients indicate that the use of comprehensive semi-structured interviews administered by trained diagnosticians may be necessary to properly identify and differentially diagnose patients with PMD. As such assessments are not commonly employed in routine practice settings, the identification of outpatients with this disorder may continue to prove difficult.

Second, the high levels of severity and impairment observed in PMD outpatients are likely to be associated with treatment resistance.^[20] PMD patients in our sample had approximately four times greater odds of chronic depression and suicide attempts and over two times greater odds of psychiatric hospitalization and chronic work impairment. Given the decreased efficacy of standard pharmacological treatments with PMD patients,^{[7, 55]} possible adjunctive interventions require careful consideration. Systematic reviews have concluded that the current evidence for the efficacy of adjunctive antipsychotic medication remains unclear.^[56] Electroconvulsive therapy has been shown to be efficacious for PMD patients,^[57] but the problems posed by treatment acceptability, potential side effects, and limited availability in many areas keep it from being a first-line treatment choice. Other approaches, such as adjunctive psychotherapy for depression appear particularly promising, although research on the efficacy of psychosocial interventions in PMD patients specifically remains in the preliminary stages.^{[24, 58]} Given the high levels of anxiety disorder comorbidity and recent advances made in the adaptation of behavioral interventions for psychotic patients,^[59] psychotherapy that can be provided in addition to pharmacotherapy may be an attractive and feasible option for many outpatients.

Finally, it is important to note that our findings indicate that PMD patients are more likely to be members of disadvantaged groups who may be experiencing unique challenges that may also be related to their socioeconomic status. This finding combined with the lower rates of PMD found in the current outpatient sample suggests that specific community outreach efforts may be needed for individuals suffering from this illness. It is clear that aggressive treatment of PMD remains essential. Vythilingam et al.^[27] found that PMD patients followed prospectively for up to 15 years after an index hospitalization had over twice the mortality rate of NMD patients (41 versus 20%) even after controlling for age and medical illness. Thus, clear challenges are faced by clinicians working with outpatients with PMD, and improvements in screening and treatment are critically needed for this high-risk clinical population.