Boundaries of Schizoaffective Disorder Revisiting Kraepelin

Home , Emil Kraepelin, Hypomania, Kraepelinian dichotomy

Research

Original Investigation Boundaries of Schizoaffective Disorder Revisiting Kraepelin

Roman Kotov, PhD; Shirley H. Leong, PhD; Ramin Mojtabai, MD, PhD, MPH; Ann C. Eckardt Erlanger, PsyD; Laura J. Fochtmann, MD; Eduardo Constantino, MD; Gabrielle A. Carlson, MD; Evelyn J. Bromet, PhD

Editorial page 1263

IMPORTANCE Established nosology identifies schizoaffective disorder as a distinct category Author Audio Interview at with boundaries separating it from mood disorders with psychosis and from schizophrenia. jamapsychiatry.com Alternative models argue for a single boundary distinguishing mood disorders with psychosis Supplemental content at from schizophrenia (kraepelinian dichotomy) or a continuous spectrum from affective to jamapsychiatry.com nonaffective psychosis.

OBJECTIVE To identify natural boundaries within psychotic disorders by evaluating associations between symptom course and long-term outcome.

DESIGN, SETTING, AND PARTICIPANTS The Suffolk County Mental Health Project cohort consists of first-admission patients with psychosis recruited from all inpatient units of Suffolk County, New York (72% response rate). In an inception cohort design, participants were monitored closely for 4 years after admission, and their symptom course was charted for 526 individuals; 10-year outcome was obtained for 413.

MAIN OUTCOMES AND MEASURES Global Assessment of Functioning (GAF) and other consensus ratings of study psychiatrists.

RESULTS We used nonlinear modeling (locally weighted scatterplot smoothing and spline regression) to examine links between 4-year symptom variables (ratio of nonaffective psychosis to mood disturbance, duration of mania/hypomania, depression, and psychosis) and 10-year outcomes. Nonaffective psychosis ratio exhibited a sharp discontinuity—10 days or more of psychosis outside mood episodes predicted an 11-point decrement in GAF—consistent with the kraepelinian dichotomy. Duration of mania/hypomania showed 2 discontinuities demarcating 3 groups: mania absent, episodic mania, and chronic mania (manic/hypomanic >1 year). The episodic group had a better outcome compared with the mania absent and chronic mania groups (12-point and 8-point difference on GAF). Duration of depression and psychosis had linear associations with worse outcome.

CONCLUSIONS AND RELEVANCE Our data support the kraepelinian dichotomy, although the study requires replication. A boundary between schizoaffective disorder and schizophrenia was not observed, which casts further doubt on schizoaffective diagnosis. Co-occurring schizophrenia and mood disorder may be better coded as separate diagnoses, an approach Author Affiliations: Department of that could simplify diagnosis, improve its reliability, and align it with the natural taxonomy. Psychiatry and Behavioral Science, Stony Brook University, Stony Brook, New York (Kotov, Fochtmann, Constantino, Carlson, Bromet); Department of Psychiatry, University of Pennsylvania, Philadelphia (Leong); Department of Mental Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland (Mojtabai); Department of Cardiology and Comprehensive Care, New York University, New York (Erlanger). Corresponding Author: Roman Kotov, PhD, Department of Psychiatry and Behavioral Science, Putnam Hall-South Campus, Stony JAMA Psychiatry. 2013;70(12):1276-1286. doi:10.1001/jamapsychiatry.2013.2350 Brook University, Stony Brook, NY Published online October 2, 2013. 11794 ([email protected]).

1276 jamapsychiatry.com

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Boundaries of Schizoaffective Disorder Original Investigation Research

he delineation of schizophrenia (dementia praecox) and was that a natural boundary would manifest as a significant psychotic mood disorders (manic-depressive insanity) drop in the outcome at some point along the spectrum, whereas T as 2 distinct entities was one of Emil Kraepelin’s semi- a continuum would result in a linear decline. Kendell and nal contributions to nosology.1 More than 100 years later, this Brockington found no evidence of a boundary, but their study kraepelinian dichotomy remains highly influential.2 How- was underpowered and analyses were limited to visual inspec- ever, some patients exhibit features of both schizophrenia and tion of graphs.22 The latter shortcoming might explain why this psychotic mood disorders, which led Kasanin3 to propose a new technique has not been widely adopted. More recent devel- category labeled schizoaffective disorder. Conceptualization of opments in statistical methods23 make it possible to test such this condition evolved across editions of the DSM from a sub- data for nonlinearity rigorously. type of schizophrenia to a distinct disorder. DSM-IV4 defines The aim of the present study was to test for the existence it as (A) co-occurrence of schizophrenia symptoms and mood of natural boundaries in psychotic disorders using modern sta- episodes, (B) psychosis present for at least 2 weeks in the ab- tistical methods. We analyzed detailed symptom course data sence of mood symptoms, and (C) mood episodes present for from an epidemiologic cohort of inpatients with psychosis a substantial portion of illness duration. Thus, DSM-IV elabo- monitored prospectively for 10 years after their first hospital- rates on the kraepelinian dichotomy by adding an intermedi- ization. In particular, we examined links between nonaffec- ate condition, with criterion B defining its boundary with psy- tive psychosis ratio during the first 4 years of the study and chotic mood disorder and criterion C with schizophrenia. The outcomes at year 10. The continuum model predicts a linear key to classifying these disorders is the ratio of nonaffective association, the kraepelinian model predicts a single bound- psychosis to mood disturbance: in psychotic mood disorder, ary between psychotic mood disorder and the schizophrenia nonaffective psychosis is absent; in schizoaffective disorder, spectrum, and the DSM-IV model predicts 2 boundaries, one both nonaffective psychosis and mood episodes are promi- between psychotic mood disorder and schizoaffective disor- nent; and in schizophrenia, nonaffective psychosis predomi- der and another between schizoaffective disorder and schizo- nates. However, some have argued that these boundaries are phrenia (Supplement [eFigure 1]). In the latter 2 models, dif- artificial and that psychotic disorders fall along a continuous ferences are expected between groups (eg, low nonaffective spectrum that ranges from psychotic mood disorder to psychosis and high nonaffective psychosis), but no associa- schizophrenia.5,6 tion is predicted between nonaffective psychosis and out- These conflicting accounts inspired a substantial body of come within groups. We constructed statistical models to test literature that evaluated the validity of schizoaffective disor- these hypotheses. We also used this method to explore natu- der using several basic approaches. Investigations of phenom- ral boundaries within depression and mania. enology found support for the continuum model,7 the kraepelinian 2-disorders model,8,9 and the DSM-IV 3-disorders 10 model. Studies of neurobiological and cognitive function- Methods ing, as well as family and genetic research, reported evidence favoring the continuum7,11 and 3-disorders12-14 models. Lon- Participants gitudinal studies of illness course produced the most support Data for this study came from the Suffolk County Mental for the continuum15,16 and 2-disorders8,17-20 models. Thus, to Health Project, an epidemiologic study of first-admission date, the literature is too conflicting to offer firm recommen- psychosis.24-26 Patients were recruited from the 12 psychiat- dations for nosology. Some of these inconsistencies likely re- ric inpatient units of Suffolk County, New York, between Oc- sult from changes in schizoaffective diagnosis, which was de- tober 1989 and December 1995. Inclusion criteria were first ad- fined more broadly by earlier diagnostic manuals. mission, either current or within 6 months; clinical evidence Among diagnostic validators, illness course is of particu- of psychosis; age 15 to 60 years; IQ higher than 70; profi- lar interest. Indeed, it was most central to Kraepelin’s work be- ciency with English; and no apparent general medical etiol- cause he sought to develop diagnoses that would be prognos- ogy. The study was approved annually by the institutional re- tic of future symptoms and functioning (ie, global outcome).2 view boards of Stony Brook University and the participating Unfortunately, existing longitudinal studies were not de- hospitals. Treating physicians determined participants’ ca- signed to answer questions about the natural organization of pacity to provide consent. Written consent was obtained from psychotic disorders. They typically compared outcomes among adults and from parents of patients younger than 18 years. diagnostic groups: schizophrenia, schizoaffective disorder, and We initially interviewed 675 participants (72% of refer- psychotic mood disorder, but such analyses cannot distin- rals); 628 of them met the eligibility criteria. By the 4-year point, guish gradual differences (ie, a continuum) from qualitative 10 participants had died, 29 were untraceable, 41 refused fur- changes (ie, natural boundaries). Indeed, in many studies15,16 ther participation, and 22 provided insufficient information outcome of schizoaffective disorder fell between that of schizo- about symptom course; the remaining 526 participants (83.8%) phrenia and psychotic mood disorder, which is consistent with constituted the course sample. Of them, by the 10-year assess- both the continuum and 3-disorders models. ment, 27 had died, 28 were untraceable, 41 refused further par- Kendell and Brockington21 proposed a solution to this prob- ticipation, and 17 provided insufficient outcome informa- lem. They examined associations between the spectrum rang- tion; the remaining 413 participants (78.5%) compose the ing from typical psychotic mood disorder to typical schizo- outcome sample. These samples were very similar to each other phrenia and continuous outcome measures. Their hypothesis and to the total cohort on the study variables (Table 1). The only

jamapsychiatry.com JAMA Psychiatry December 2013 Volume 70, Number 12 1277

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Research Original Investigation Boundaries of Schizoaffective Disorder

Table 1. Demographic and Clinical Characteristics of the Sample

No. (%)a Total Course Outcomes Cohort Sample Sample Characteristic (N = 628) (n = 526) (n = 413) Age at baseline, mean (SD), y 29.7 (9.7) 29.2 (9.5) 29.1 (9.5) Male sex 365 (58.1) 299 (56.8) 231 (55.9) White race 470 (74.8) 400 (76.0) 313 (75.8) SES of family of origin: blue collar 284 (45.2) 236 (44.9) 199 (48.2) DSM-IV diagnosis at year 2 Schizophrenia/schizophreniform 199 (33.8) 184 (35.1) 145 (35.1) Schizoaffective 30 (5.1) 26 (5.0) 21 (5.1) Abbreviations: GAF, Global Bipolar with psychosis 148 (25.1) 135 (25.8) 112 (27.1) Assessment of Functioning; MDD with psychosis 104 (17.7) 91 (17.4) 68 (16.5) GAF-F, Global Assessment of Other psychoses 108 (18.3) 88 (16.8) 67 (16.2) Functional Performance; GAS, Global Assessment of Symptom course, mean (SD) Symptoms; MDD, major depressive % Psychosisb NA 36.4 (38.4) 37.4 (38.7) disorder; NA, not applicable; % Maniab NA 7.5 (18.6) 8.3 (19.9) SADS, Schedule for Affective Disorders and Schizophrenia; % Depressionb NA 24.0 (32.6) 24.5 (32.8) SES, socioeconomic status. c % Nonaffective psychosis ratio NA 35.7 (43.4) 34.9 (43.0) a Percentages may vary because of Outcome, mean (SD)d missing data. GAF NA NA 54.8 (16.2) b Percentage of observed interval GAF-F NA NA 57.5 (15.9) from baseline to 4-year point. c Percentage of illness during interval GAS NA NA 57.4 (16.5) from baseline to 4-year point. Psychosocial functioning (SADS) NA NA 2.4 (1.2) d Outcome at 10-year point.

significant differences between the course sample and the rest chotic episode), because this ratio defines the diagnostic of the cohort (n = 102) were slightly younger age (P = .008) and boundaries of schizoaffective disorder in DSM-IV (especially lower prevalence of other psychoses in the sample (P = .044). criterion C). The only significant difference between the outcome sample Overall outcome is particularly relevant to validation of and the rest of the course sample (n = 113) was the slight over- psychotic disorders.1,15,17,18 We examined 3 measures target- representation of patients with low parental socioeconomic sta- ing its different aspects: Global Assessment of Symptoms (GAS) tus in the former (P = 008). indicated overall symptom severity in the best month between the 4-year and 10-year interviews, Global Assessment Measures of Functional Performance (GAF-F) indicated overall social and Face-to-face assessments were conducted by master’s level occupational functioning in the best month between 4-year and mental health professionals at baseline, 6-month, 2-year, 10-year interviews, and Global Assessment of Functioning 4-year, and 10-year follow-up; telephone interviews were per- (GAF) was rated for the best month of the year before the 10- formed every 3 months until the 2-year wave and every 6 year interview considering both symptoms and functioning. months until the 4-year wave. Interviewers were blinded to Each measure was rated on a 0 to 90 scale (with 10 anchors spe- study diagnoses. Medical records and interviews with signifi- cific to that rating) according to the DSM-III-R version of GAF, cant others were also obtained at each major assessment. These which was standard at the start of this study. To ensure that detailed data allowed raters to chart symptom course be- results were not influenced by format, we also evaluated the tween baseline and year 4. At least half of the interval was docu- overall rating of psychosocial functioning from the Schedule mented for everyone in the course sample; 91.7% of them had for Affective Disorders and Schizophrenia (SADS),27 scored as at least 3.5 years of follow-up data. 1, marked chronic condition; 2, moderate chronic condition; Symptom documentation included start and end dates of 3, mild chronic condition; and 4, complete return to highest psychotic, depressive, and manic episodes, each rated sepa- functioning. These ratings were made by consensus of study rately and defined according to DSM-IV criteria except for du- psychiatrists (including L.J.F., E.C., and G.A.C.). Interrater re- ration, which we did not require. Episodes were scored as (1) liability of consensus scores could not be assessed, but reli- percentage of the observed interval psychotic, (2) percentage ability of the individual raters was excellent, ranging intra- of patients depressed, and (3) percentage of patients manic (in- class r = 0.90-0.94 across outcomes. cluding hypomania). Of particular interest was the nonaffec- Primary DSM-IV diagnosis was formulated at the 2-year tive psychosis ratio, scored as percentage of illness psychotic point by consensus of 4 or more psychiatrists (including L.J.F. and not in mood episode (illness was defined as mood or psy- and G.A.C.) using all available information, including Struc-

1278 JAMA Psychiatry December 2013 Volume 70, Number 12 jamapsychiatry.com

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Boundaries of Schizoaffective Disorder Original Investigation Research

Table 2. Multiple Linear Regression Analyses of 4-Year Course Predicting 10-Year Outcomes

Global Assessment Functioning (Overall) Functional Performance Symptoms Psychosocial Functioning Predictor/Outcome BβP Value B β P Value B β P Value B β P Value Intercept 65.54 69.05 69.48 3.33 % Psychosis −16.20 −0.39 <.001a −14.08 −0.35 <.001a −17.17 −0.40 <.001a −1.08 −0.34 <.001a % Mania −3.95 −0.05 .28a −2.07 −0.03 .56 −1.83 −0.02 .61a −0.21 −0.03 .44 % Depression −6.43 −0.13 .02a −9.88 −0.21 <.001a −8.30 −0.17 .002a −0.77 −0.20 <.001a % Nonaffective psychosis ratio −7.90 −0.21 .001a −10.96 −0.30 <.001a −10.45 −0.27 <.001a −0.89 −0.31 <.001a

Abbreviations: B, unstandardized regression coefficient; β, standardized regression coefficient. a P < .01 considered significant.

tured Diagnostic Interview for DSM-IV28 with participants, the entire spectrum. The nonaffective psychosis ratio was U- medical records, and significant others.26 Diagnoses were shaped: 51.3% of the patients were psychotic only while in mood grouped into 5 categories: schizophrenia/schizophreniform, episodes, 20.7% had only nonaffective psychosis, and 28.0% had schizoaffective, bipolar with psychosis, depression with psy- psychosis both in and outside of mood episodes. chosis, and other psychoses (eg, psychosis not otherwise speci- Ten-year outcomes ranged widely: GAF scores from 21 to 90, fied and substance-induced psychosis). In assigning schizoaf- GAF-F from 30 to 90, and GAS from 25 to 90. Distributions were fective disorder diagnosis, psychiatrists interpreted criterion positively skewed with modes in the low 40s. On SADS, 35.2% C (substantial portion) as requiring mood disturbance to be pre- of participants were rated marked; 20.0%, moderate; 15.2%, mild; sent for more than 30% of illness duration. and 29.5%, remitted (returned to highest functioning). Demographic characteristics were also considered in analyses. They included age at baseline, sex, race, and socioeco- Linear and Nonlinear Associations With Outcome nomic status of the head of household. First, we examined linear associations between the 4 symptom course variables and the 4 outcomes by conducting mul- Statistical Analysis tiple regression analyses, with the 4 predictors entering si- First, we examined relationships between the 4 symptom multaneously and each outcome serving as the dependent course predictors and 4 outcomes using locally weighted scat- variable in turn. The strongest predictor was psychosis dura- terplot smoothing (LOESS),29,30 which uses weighted least tion (β = −0.34 to −0.40), followed by nonaffective psychosis squares to fit linear functions within a fixed neighborhood of (−0.21 to −0.31) and finally depression (−0.13 to −0.21); coef- each data point. If LOESS indicated nonlinearity of the asso- ficients for mania were not significant (Table 2). Zero-order cor- ciation, we evaluated its exact form using spline regression.31-33 relations are given in the Supplement (eMethods). Spline regression is a piecewise regression that fits polyno- To test for nonlinearity of these associations, we esti- mial functions onto segments of the predictor variable. In com- mated LOESS models and compared them with linear mod- paring fit of different spline models, we used 4 fit indices: the els. The LOESS smoothed scatterplots for each predictor out- generalized cross-validation criterion, the Akaike informa- come pair and took whatever shape summarized the data best. tion criterion, the Akaike information criterion corrected 1, and For psychosis and depression, LOESS showed no improve- the Bayesian information criterion.34-39 Analyses were per- ment over the linear model: change in fit was small and non- formed using commercial software (SAS, version 9.2, with significant (Supplement [eTable]). For mania and nonaffec- PROC LOESS and PROC NLIN; SAS Institute Inc). tive psychosis, LOESS was significantly superior across all outcomes, and the improvement in fit ranged from Akaike In- formation Criterion Corrected 1 of 6.69 (substantial) to 51.57 Results (very substantial). Consistent with the fit indices, LOESS curves for psychosis and depression were essentially linear (Figure 1). Descriptive Characteristics Mania curves had an initial rise that plateaued and then gradu- The total duration of illness (psychotic or mood episodes) ranged ally returned to the starting level. Nonaffective psychosis from 2 days to 4 years. On average, participants were in an epi- curves showed an initial drop that soon leveled. An apparent sode for a mean (SD) of 48.4% (39.0%) of the follow-up period. discontinuity in nonaffective psychosis contradicted the con- The distribution of psychosis duration was U-shaped (Supple- tinuum model and was most consistent with the kraepelin- ment [eFigure 2]); 28.7% of participants were psychotic briefly ian model. However, more rigorous modeling was needed to (<5% of the follow-up period), 18.4% were psychotic con- understand the exact form of the nonlinearity. stantly (>95%), and 52.9% were between these subgroups. The distribution of mania/hypomania was L-shaped: 58.6% had none Spline Models in the interval, and others were spread across the entire spec- We used spline regression to more precisely evaluate nonlin- trum of duration. Depression had a similar distribution, with earity detected by LOESS for nonaffective psychosis and ma- 31.4% of the participants not depressed and others spread across nia. Psychosis and depression were not considered further be-

jamapsychiatry.com JAMA Psychiatry December 2013 Volume 70, Number 12 1279

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Research Original Investigation Boundaries of Schizoaffective Disorder

Figure 1. Locally Weighted Scatterplot–Smoothed Curves for Global Assessment of Functioning (GAF) and 4 Predictors

75 75

65 65

55 55 GAF Score GAF Score

45 45

35 35 0 20 40 60 80 100 0 20 40 60 80 100 Psychosis, % Mania, %

75 75

65 65

55 55 GAF Score GAF Score

45 45

35 35 0 20 40 60 80 100 0 20 40 60 80 100 Depression, % Nonaffective Psychosis Ratio, %

Dashed lines are 95% confidence band around the curve.

cause their associations with outcomes were purely linear. much better than LOESS, indicating further support for Spline regression allowed us to specify basic shapes of the these specific types of nonlinearity. curves to test target models and likely alternatives (Supple- ment [eMethods]). Diagnostic Comparisons For nonaffective psychosis, the fit indices consistently sup- Next, we examined concordance between empirical groups ported the kraepelinian model across the outcomes (Table 3). identified by spline regression and DSM-IV diagnoses. Be- The only exception was GAS, for which 3 indices favored the cause diagnoses were assigned at the 2-year point, we scored DSM-IV model, but the fit of the kraepelinian model was nearly empirical groups from the first 2 years of course data using the identical and superior on the Bayesian Information Criterion— aforementioned cutoffs (1.5% on nonaffective psychosis, and the most parsimonious index. We named the identified groups 0.8% and 27.0% on mania). nonaffective psychosis absent and nonaffective psychosis present. Overall, concordance between the empirical groups and The boundary between them was at 1.5% of nonaffective psy- DSM-IV diagnoses was high. Nearly all (88.6) participants with chosis ratio, that is, 10 days of psychosis outside of mood epi- schizophrenia or schizoaffective disorder diagnosis were in the sodes (Figure 2). nonaffective psychosis present group (Table 4); those who were For mania, the fit indices consistently supported the assigned to nonaffective psychosis absent either had nonaf- 3-group model over all alternatives (Table 3). The only fective psychosis before the first hospitalization—including the exception was SADS, for which 3 indices favored the 4-group 5 schizoaffective cases—or had prominent negative symp- model, but fit of the 3-group model was nearly identical and toms outside mood episodes. Almost all (97.3%) cases of psy- the Bayesian Information Criterion favored 3 groups. We chotic mood disorders were in the nonaffective psychosis ab- named the 3 groups mania absent, episodic mania, and sent group; the remaining 2.7% had only brief periods of chronic mania. The boundaries between them were 0.8% (11 nonaffective psychosis and their mood symptoms were much days) and 27.0% (394 days) manic (Figure 2). In the episodic more severe than psychotic symptoms, resulting in psychotic group, elevated mood consisted primarily of mania (mean, mood disorder diagnosis. 65.4% of time in episodes), with the rest being mixed state Of participants with bipolar disorder, 20.7% were in the (23.5%) or hypomania (11.1%). In the chronic group, mania chronic mania group. Others were in the episodic mania group, (35.0%), mixed state (37.3%), and hypomania (27.7%) were except for 4 patients who had mania only before the first hos- evenly represented. The selected spline models fit the data pitalization and thus were classified in the absent group. Ap-

1280 JAMA Psychiatry December 2013 Volume 70, Number 12 jamapsychiatry.com

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Boundaries of Schizoaffective Disorder Original Investigation Research

Table 3. Comparison of Spline Regression Modelsa

Global Assessment Functioning (Overall) Functional Performance Symptoms Psychosocial Functioning Predictor/Outcome GCV AICC1 AIC BIC GCV AICC1 AIC BIC GCV AICC1 AIC BIC GCV AICC1 AIC BIC Nonaffective psychosis LOESSb 0.588 2684 2268 2282 0.548 2624 2212 2226 0.601 2625 2218 2232 0.00325 553 130 143 2 Flat (kraepelinian)c 0.581 2678 2261 2265 0.539 2618 2205 2209 0.595 2621 2213 2217 0.00319 544 120 124 3Flat(DSM-IV) 0.584 2680 2263 2271 0.541 2619 2206 2214 0.593 2619 2211 2219 0.00320 546 122 130 2 Linear 0.585 2681 2266 2278 0.542 2620 2209 2221 0.598 2623 2216 2229 0.00321 548 126 138 3 Linear 0.590 2685 2268 2284 0.544 2622 2213 2233 0.597 2622 2216 2228 0.00328 557 128 140 1 Quadratic + 1 linear 0.588 2684 2266 2278 0.541 2620 2211 2227 0.601 2625 2217 2229 0.00323 550 127 144 2 Quadratic + 1 linear 0.593 2687 2270 2290 0.549 2625 2212 2232 0.606 2628 2220 2240 0.00326 555 131 151 1 Cubic + 1 linear 0.585 2681 2270 2290 0.544 2622 2211 2227 0.604 2627 2219 2235 0.00323 550 129 150 Mania LOESS 0.577 2676 2258 2282 0.548 2625 2212 2237 0.615 2635 2227 2254 0.00342 575 151 178 2 Flat 0.596 2689 2272 2276 0.560 2634 2221 2225 0.628 2642 2234 2238 0.00345 578 154 158 3 Flat 0.561 2664 2247 2255 0.538 2617 2204 2212 0.603 2626 2218 2226 0.00336 567 143 151 4 Flat 0.566 2668 2251 2263 0.539 2618 2205 2217 0.608 2630 2221 2233 0.00335 566 142 154 2 Linear 0.576 2675 2262 2270 0.546 2623 2214 2222 0.613 2633 2229 2237 0.00343 576 152 160 3 Linear 0.569 2670 2259 2279 0.541 2619 2206 2218 0.607 2629 2223 2239 0.00336 567 143 155 4 Linear 0.580 2678 2261 2277 0.539 2618 2211 2231 0.614 2634 2226 2242 0.00338 569 145 161 1 Quadratic + 1 linear 0.589 2684 2269 2285 0.566 2638 2227 2243 0.625 2641 2235 2251 0.00340 572 150 166 2 Quadratic + 1 linear 0.570 2671 2260 2288 0.539 2618 2211 2235 0.611 2632 2224 2244 0.00341 573 151 167 1 Cubic + 1 linear 0.591 2686 2271 2287 0.576 2645 2234 2250 0.635 2647 2241 2257 0.00353 587 163 179

Abbreviations: AIC, Akaike information criterion; AICC1, Akaike information substantial, and greater than 10 is very substantial.39 criterion corrected 1; BIC, Bayesian information criterion; GCV, generalized cross b LOESS uses weighted least squares to fit linear functions within a fixed validation criterion; LOESS, locally weighted scatterplot smoothing. neighborhood of each data point, as determined by the smoothing parameter a Bold indicates best fit of the series. These indices are derived from different (percentage of the sample included). We examined a range of smoothing statistical theories and are scaled differently.34,36,38,39 However, all 4 can be parameters and selected 60%, as greater inclusion did not increase fit of the decomposed into 2 components: a measure of fit between the model and data curve. and a penalty for model complexity. Based on the latter, the indices can be c Polynomials of degree 0, 1, 2, or 3 (flat, linear, quadratic, or cubic function) ordered from least to most parsimonious: AIC, GCV, AICC1, and BIC. There are were fit onto each segment. Other than the flat function regressions, all no absolute cutoffs on these indices, but they can be used to compare models, regressions were restricted to be continuous. The locations of break points 38,39 with lower values representing better fit. Conventional guidelines suggest and the slope of each segment were freely estimated. that on AIC, AICC1, and BIC, a difference less than 6 is small, 6 to 10 is

proximately half (53.8%) of participants with schizoaffective tween ratio of nonaffective psychosis to mood disturbance and disorder diagnosis were in the episodic or chronic group. Mania later outcome in our first-admission cohort with psychotic dis- was rare in other disorders. orders. Specifically, we observed a qualitative difference in out- With regard to outcomes, nonaffective psychosis present come between cases in which psychosis is limited to mood epi- had notably worse scores than nonaffective psychosis absent sodes and cases in which at least some psychosis is nonaffective. (Table 4). The differences were more than 10 points on GAF, No other discontinuities emerged in analyses of nonaffective GAF-F, and GAS, and one level on SADS (ie, between moder- psychosis. These findings clearly support the kraepelinian di- ate and mild condition). Similar differences were observed be- chotomy over the DSM-IV and continuum accounts. We found tween mania absent and episodic mania. In contrast, the no evidence of a distinct schizoaffective disorder. Judged by out- chronic mania group was similar to mania absent on all outcome, this diagnosis appears to be a part of the schizophrenia comes. Outcomes for DSM-IV schizoaffective disorder were spectrum. Other definitions of schizoaffective disorder that do similar to those of nonaffective psychosis present, whereas out- not rely on nonaffective psychosis are possible and were not comes for schizophrenia were slightly worse (4-5 points on GAF evaluated here. The analyses also revealed 2 distinct types of metric). Participants with bipolar disorder did about as well mania: episodic and chronic. In contrast, duration of psycho- as the episodic mania group. sis and depression both had linear associations with outcomes and did not demarcate natural boundaries within psychotic disorders. Discussion If replicated in other samples and with other validators, our results would have several implications for future edi- Using modern statistical techniques—LOESS and spline regres- tions of the DSM.Giventhelackofvalidityofschizoaffective sion—we detected strong nonlinearity in the relationship be- disorder diagnosis observed in this study and questionable

jamapsychiatry.com JAMA Psychiatry December 2013 Volume 70, Number 12 1281

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Research Original Investigation Boundaries of Schizoaffective Disorder

Figure 2. Curves for the Best-Fitting Spline Models for Each Outcome

75 75

65 65

55 55 GAF Score GAF Score

45 45

35 35 0 20 40 60 80 100 0 20 40 60 80 100 Nonaffective Psychosis Ratio, % Mania, %

75 75

65 65

55 55 GAF-F Score GAF-F Score GAF-F

45 45

35 35 0 20 40 60 80 100 0 20 40 60 80 100 Nonaffective Psychosis Ratio, % Mania, %

75 75

65 65

55 55 GAS Score GAS Score

45 45

35 35 0 20 40 60 80 100 0 20 40 60 80 100 Nonaffective Psychosis Ratio, % Mania, %

4 4

3 3 SADS Score SADS Score 2 2

1 1 0 20 40 60 80 100 0 20 40 60 80 100 Nonaffective Psychosis Ratio, % Mania, %

Outcome expected at each level of symptom is shown. GAF indicates Global Assessment of Functioning; GAF-F, Global Assessment of Functional Performance; GAS, Global Assessment of Symptoms; and SADS, Schedule for Affective Disorders and Schizophrenia.

1282 JAMA Psychiatry December 2013 Volume 70, Number 12 jamapsychiatry.com

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Boundaries of Schizoaffective Disorder Original Investigation Research

Table 4. DSM-IV Diagnoses, Empirical Groups, and Outcomesa

No. (%) Outcomes, Mean (SD) Nonaffective Psychosis Groups Mania Groups Global Assessment Functioning Functional Psychosocial Group Absent Present Absent Episodic Chronic (Overall) Performance Symptoms Functioning Primary DSM-IV diagnosis Schizophrenia 19 (10.3) 165 (89.7) 159 (86.4) 21 (11.4) 4 (2.2) 44.7 (11.0) 46.8 (11.2) 45.9 (12.4) 1.6 (0.9) Schizoaffective 5 (19.2) 21 (80.8) 12 (46.2) 10 (38.5) 4 (15.4) 49.4 (14.8) 52.6 (12.9) 51.4 (15.0) 2.0 (1.2) Bipolar with psychosis 131 (97.0) 4 (3.0) 4 (3.0) 103 (76.3) 28 (20.7) 65.2 (15.2) 67.8 (13.7) 68.7 (13.6) 3.1 (1.1) MDD with psychosis 89 (97.8) 2 (2.2) 91 (100) 0 0 58.0 (14.8) 61.9 (14.5) 62.4 (13.8) 2.7 (1.2) Other psychoses 36 (40.9) 52 (59.1) 68 (77.3) 11 (12.5) 9 (10.2) 57.8 (16.1) 60.3 (15.6) 59.9 (15.6) 2.7 (1.2) Nonaffective psychosis group Absent NA NA 134 (47.9) 111 (39.6) 35 (12.5) 59.6 (15.8) 62.8 (15.0) 62.9 (15.1) 2.9 (1.2) Present NA NA 201 (81.7) 35 (14.2) 10 (4.1) 49.1 (14.8) 50.9 (14.4) 50.5 (15.5) 1.9 (1.1) Mania group Absent NA NA NA NA NA 51.3 (15.1) 53.6 (15.4) 53.6 (16.0) 2.2 (1.2) Episodic NA NA NA NA NA 62.8 (15.9) 65.8 (14.4) 65.6 (15.2) 2.9 (1.2) Chronic NA NA NA NA NA 54.1 (15.5) 57.5 (13.4) 57.7 (14.6) 2.5 (1.1)

Abbreviations: MDD, major depressive disorder; NA, not applicable. mania); dichotomous outcomes are presented as row percentages; diagnosis a Number of patients: 280 (nonaffective psychosis absent), 246 (nonaffective was made at 2-year point; nonaffective psychosis and mania groups were psychosis present), 335 (mania absent), 146 (episodic mania), and 45 (chronic scored based on illness course between baseline and 2-year point.

support in the literature,7,40,41 continued use of this category type defined by being manic for at least a year. This subtype is difficult to justify. Indeed, prior research7-20 considered vari- has prognostic significance because it was associated with a ous validators: phenotypic, outcome, cognitive, neural, and distinctly worse outcome. Of note, all of these findings were genetic, and only 9 of 256 studies of this question concluded consistent across several outcome measures, strengthening that schizoaffective disorder is a distinct condition.40 Our find- conclusions of the study. These measures reflect a single vali- ings suggest that patients who currently are assigned a diag- dator—global outcome—and are not independent replica- nosis of schizoaffective disorder would be better described as tions, but they helped to ensure that the present results are having schizophrenia (or schizophreniform disorder) with co- not due to characteristics of a particular rating scale. morbid mood disorder. This nosologic change would reflect a We did not hypothesize the chronic mania group a priori, growing recognition of the important role that mood comor- and it requires confirmation, but this finding aligns well with bidities play in schizophrenia42-44 and permit a flexible clas- the extant literature. Chronic mania was recognized as a dis- sification of psychotic illnesses without invoking an appar- tinct category in the 19th century.46 More recently, it has been ently arbitrary diagnostic category. Continuous ratings of operationalized as a manic episode lasting at least 2 years, and severity for mood disorders and schizophrenia could further 6% to 13% of patients with bipolar I disorder fit this subtype.46 increase informational value of such a classification. Indeed, Of note, the 2-year definition of chronic mania was proposed such ratings have been proposed for the DSM-5. With regard based on a zone of rarity in distribution of episode length, but to schizoaffective diagnosis, the only significant revision con- the zone ranged from 1 to 2 years.47 By the 1-year definition, sidered for the DSM-5 is to make it explicitly a lifetime prevalence of chronic mania is approximately 15%,48 which is diagnosis,45 and this is how the disorder was approached in comparable to the estimate in our cohort (20.7% of bipolar I the present study. Our findings argue for reconsideration of disorder). schizoaffective disorder, but more research is needed. The observed empirical groups were defined by symp- In contrast, we found a clear discontinuity between schizo- toms only. Nevertheless, both nonaffective psychosis and ma- phrenia spectrum disorders and psychotic mood disorder. In nia categories showed the anticipated convergence with our data, even 10 days of nonaffective psychosis resulted in a DSM-IV diagnoses. There was only a handful of inconsisten- qualitatively worse outcome. This is consistent with DSM-IV cies resulting from symptoms present before the first hospi- criteria for demarcating schizoaffective disorder and psy- talization or to highly prominent symptoms that received spe- chotic mood disorder (ie, 2 weeks of nonaffective psychosis). cial weight in diagnostic decision making. In addition, mania Bipolar disorder with psychosis also was clearly distin- groups included some patients with schizophrenia and other guished from other psychotic disorders, even with several days psychoses, which reflects the presence of comorbid mood dis- of manic symptoms forecasting qualitatively better out- orders in these cases. comes. This finding is consistent with research10,16,17 indicat- These empirical groupings had substantial predictive va- ing favorable outcomes for this disorder relative to other psy- lidity, forecasting more than 10-point differences in GAF among choses. In addition, we observed a discontinuity within the both nonaffective psychosis and mania groups years later. The bipolar spectrum, suggesting existence of a chronic mania sub- DSM-IV diagnosis was somewhat more predictive, with schizo-

jamapsychiatry.com JAMA Psychiatry December 2013 Volume 70, Number 12 1283

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Research Original Investigation Boundaries of Schizoaffective Disorder

phrenia outcome being 5 GAF points lower than the nonaffec- hensive evaluation has to include other characteristics, such tive psychosis present group. Schizophrenia diagnosis explic- as genetic risk factors, neural substrates, and treatment itly requires marked deterioration of functioning (criterion B), response.53 This effort requires integration of findings from dif- which likely explains why this group fared worse than the non- ferent research paradigms, and the present study is a step to- affective psychosis present group. Altogether, it is remark- ward this goal. Third, the present report focused on global out- able that simple classification rules based solely on symptom- comes, as these have been the primary benchmarks for other atology were almost as predictive as full DSM-IV diagnosis. longitudinal studies of schizoaffective disorder.15,17,18 We also The observed sharp distinction between no nonaffective collected fine-grained information and will investigate spe- psychosis and any nonaffective psychosis and the large effect cific outcomes in subsequent studies. Fourth, each outcome it had on outcome suggests differences in etiologies of these was a single rating, and such variables tend to have low reli- groups. For instance, psychotic symptoms in schizophrenia ability. To ensure strong psychometric properties, the pre- spectrum disorders may result from neurodevelopmental sent ratings were made by consensus of research psychia- pathologic factors, whereas psychosis in psychotic mood dis- trists based on all available information. Fifth, consensus order may be induced by stress.49,50 These findings contra- diagnosis was available at the 2-year rather than 4-year point, dict the continuum view of psychotic disorders, but psy- so we had to limit analyses comparing diagnoses and empiri- chotic mood disorder and nonaffective psychosis may share cal groups to 2 years of symptom course. Sixth, we could not some risk factors and pathophysiologic processes. Many such investigate treatment effects in this naturalistic study, and it commonalities have been documented51,52 and may explain is important to confirm present findings in randomized trials, their substantial comorbidity.42-44 In our sample, 57% of the controlling for treatment experiences. Finally, generalizabil- nonaffective psychosis group experienced at least one mood ity of the present results was limited by attrition. Fortunately, episode. The degree of overlap versus distinction among these attrition during the 10-year study was modest and had little conditions can be further explicated by applying nonlinear effect on study variables. modeling to other validators. In conclusion, if replicated, our findings would provide Our rejection of the 3-disorder model in favor of the krae- clear support for the kraepelinian dichotomy, and this sharp pelinian dichotomy seems to be at odds with studies15,16 report- boundary presents a significant challenge for the continuum ing better outcomes in schizoaffective disorder compared with view of psychotic disorders. Also, absence of the boundary be- schizophrenia. Importantly, schizoaffective disorder is de- tween schizophrenia and schizoaffective disorder calls valid- fined only by symptom pattern and, unlike schizophrenia, does ity of the latter into question. Schizoaffective disorder was an not require marked functional impairment or 6-month dura- early advance that recognized the co-occurrence of schizo- tion, which likely explains differences in outcome. Indeed, in phrenia and mood disorders. It was an imperfect solution, how- our cohort, outcome of schizoaffective disorder was no differ- ever, and the present findings suggest that coding of comor- ent from the outcome of the rest of the nonaffective psychosis bid schizophrenia (or schizophreniform disorder) and mood group. Quantitative distinctions among patients with psy- disorder as 2 separate diagnoses may serve the field better than chotic disorders also must be recognized. We found that of all the schizoaffective category. In fact, the DSM-IV already per- variables considered, duration of psychosis was the most im- mits such coding, and this proposal would extend it to cases portant predictor of outcome. Clinicians need to remain vigi- currently diagnosed as schizoaffective disorder. This change lant to long-term disability associated with chronic psychosis. also would streamline differential diagnosis for psychotic dis- Strengths of this investigation include a first-admission epi- orders. Indeed, the reliability of schizoaffective disorder di- demiologic cohort that was followed long-term and a pains- agnosis is remarkably poor.26,54 Much of this difficulty stems taking tracking of symptoms and functioning using inter- from criterion C,54 which separates schizoaffective disorder views, informant reports, and medical records. Nevertheless, from schizophrenia with comorbid mood disorder. Our re- the present findings need to be considered against the study’s sults suggest that this distinction is superfluous, which may limitations. First, detailed documentation of symptoms was explain the associated unreliability. Thus, by abolishing the limited to 4 years and sometimes did not include illness on- schizoaffective disorder category while maintaining the quali- set. This investigation targeted a crucially important period of tative distinction between psychotic mood disorder and schizo- illness course, but close tracking of symptoms over a long term phrenia spectrum disorders, it may be possible to align the no- would provide a more definitive test of diagnostic boundar- sology with the natural taxonomy of psychoses, simplify ies. Second, validation of diagnostic distinctions was limited diagnosis, and improve its reliability. This contention re- to long-term outcome. Kraepelin1,2 considered illness course quires verification in other samples and with a variety of the key consideration for diagnostic validity, but a compre- validators.

ARTICLE INFORMATION Author Contributions: Dr Kotov takes Drafting of the manuscript: Kotov, Leong, Erlanger, Submitted for Publication: November 15, 2012; responsibility for the integrity of the data and the Bromet. final revision received February 6, 2103; accepted accuracy of the data analysis. Critical revision of the manuscript for important April 2, 2013. Study concept and design: Kotov, Mojtabai, Carlson. intellectual content: Leong. Mojtabai, Erlanger, Acquisition of data: Kotov, Fochtmann, Fochtmann, Constantino, Carlson, Bromet. Published Online: October 2, 2013. Constantino, Bromet. Statistical analysis: Kotov, Leong. doi:10.1001/jamapsychiatry.2013.2350. Analysis and interpretation of data: Kotov, Leong, Obtained funding: Kotov, Bromet. Mojtabai, Erlanger.

1284 JAMA Psychiatry December 2013 Volume 70, Number 12 jamapsychiatry.com

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Boundaries of Schizoaffective Disorder Original Investigation Research

Administrative, technical, and material support: 12. Bora E, Yucel M, Fornito A, Berk M, Pantelis C. Schizophrenia. Arch Gen Psychiatry. Kotov, Erlanger. Major psychoses with mixed psychotic and mood 1978;35(7):837-844. Study supervision: Kotov, Erlanger, Bromet. symptoms: are mixed psychoses associated with 28. First MB, Spitzer RL, Gibbon M, Williams JBW. Conflict of Interest Disclosures: None reported. different neurobiological markers? Acta Psychiatr Structured Clinical Interview for DSM-IV Axis I Scand. 2008;118(3):172-187. Funding/Support: National Institutes of Health Disorders—Patient Edition (SCID-I/P, Version 2.0). grant MH094398 to Dr Kotov and MH44801 to Dr 13. Hamshere ML, Green EK, Jones IR, et al; New York, NY: Biometrics Research Dept, New York Bromet. Wellcome Trust Case Control Consortium. Genetic State Psychiatric Institute; 1995. utility of broadly defined bipolar schizoaffective Role of the Sponsor: The National Institutes of 29. Cleveland WS. Robust locally weighted disorder as a diagnostic concept. Br J Psychiatry. regression and smoothing scatterplots. J Am Stat Health had no role in the design and conduct of the 2009;195(1):23-29. study; collection, management, analysis, and Assoc. 1979;74:829-836. interpretation of the data; preparation, review, or 14. Kendler KS, McGuire M, Gruenberg AM, Walsh 30. Cleveland WS, Devlin SJ. Locally weighted approval of the manuscript; and decision to submit D. Examining the validity of DSM-III-R regression: an approach to regression analysis by the manuscript for publication. schizoaffective disorder and its putative subtypes local fitting. JAmStatAssoc. 1988;83:596-610. in the Roscommon Family Study. Am J Psychiatry. Additional Contributions: Greg Perlman, PhD, 1995;152(5):755-764. 31. Marsh LC, Cormier DR. Spline regression Adam Gonzalez, PhD, and Camilo Ruggero, PhD, models. In: Lewis-Beck MS, ed. Sage University provided feedback on the manuscript. We thank 15. Harrow M, Grossman LS, Herbener ES, Davies Papers Series on Quantitative Application in the the mental health professionals in Suffolk County, EW. Ten-year outcome: patients with Social Sciences. Thousand Oaks, CA: Sage the project psychiatrists and staff, and most of all, schizoaffective disorders, schizophrenia, affective Publications; 2001:137. the study participants and their families and disorders and mood-incongruent psychotic symptoms. Br J Psychiatry. 2000;177:421-426. 32. Muggeo VMR. Estimating regression models friends. The project psychiatrists and staff received with unknown break-points. Stat Med. their usual salary support; no others were paid for 16. Tsuang MT, Dempsey GM. Long-term outcome 2003;22(19):3055-3071. their services. of major psychoses: II: schizoaffective disorder compared with schizophrenia, affective disorders, 33. Toms JD, Lesperance ML. Piecewise regression: REFERENCES and a surgical control group. Arch Gen Psychiatry. a tool for identifying ecological thresholds. Ecology. 2003;84:2034-2041. 1. Kraepelin E. Psychiatrie. 6th ed. Leipzig, 1979;36(12):1302-1304. Germany: Barth; 1899. 17. Ledda MG, Fratta AL, Pintor M, Zuddas A, 34. Craven P, Wahba G. Smoothing noisy data with spline functions: estimating the correct degree of 2. Greene T. The kraepelinian dichotomy: the twin Cianchetti C. Early-onset psychoses: comparison of clinical features and adult outcome in 3 diagnostic smoothing by the method of generalized pillars crumbling? Hist Psychiatry.2007;18(71,pt cross-validation. Numerische Mathematik. 3):361-379. groups. Child Psychiatry Hum Dev. 2009;40(3):421-437. 1979;31:377-403. 3. Kasanin J. The acute schizoaffective psychoses. 35. Hu L, Bentler PM. Cutoff criteria for fit indexes Am J Psychiatry. 1994;151(6)(suppl):144-154. 18. Sim K, Chan YH, Chong SA, Siris SG. A 24-month prospective outcome study of in covariance structure analysis: conventional 4. American Psychiatric Association. Diagnostic first-episode schizophrenia and schizoaffective criteria versus new alternatives. Struct Equ and Statistical Manual of Mental Disorders. ed 4. disorder within an early psychosis intervention Modeling. 1999;6:1-55. Washington, DC; American Psychiatric Association; program. J Clin Psychiatry. 2007;68(9):1368-1376. 36. Hurvich CM, Simonoff JS, Tsai CL. Smoothing 1994. 19. Williams PV, McGlashan TH. Schizoaffective parameter selection in nonparametric regression 5. Crow TJ. The continuum of psychosis and its psychosis: I: comparative long-term outcome. Arch using an improved Akaike information criterion. JR implication for the structure of the gene. Br J Gen Psychiatry. 1987;44(2):130-137. Stat Soc Series B Stat Methodol. 1998;60:271-293. Psychiatry. 1986;149:419-429. 20. Tsuang D, Coryell W. An 8-year follow-up of 37. Marsh HW, Hau KT, Wen Z. In search of golden 6. Peralta V, Cuesta MJ. Exploring the borders of patients with DSM-III-R psychotic depression, rules: comment on hypothesis-testing approaches the schizoaffective spectrum: a categorical and schizoaffective disorder, and schizophrenia. Am J to setting cutoff values for fit indexes and dangers dimensional approach. J Affect Disord. Psychiatry. 1993;150(8):1182-1188. in overgeneralizing Hu and Bentler’s (1999) 2008;108(1-2):71-86. findings. Struct Equ Modeling. 2004;11:320-341. 21. Kendell RE, Brockington IF. The identification of 7. Cheniaux E, Landeira-Fernandez J, Lessa Telles disease entities and the relationship between 38. Akaike H. A new look at the statistical model L, et al. Does schizoaffective disorder really exist? a schizophrenic and affective psychoses. Br J identification. IEEE Trans Automat Contr. systematic review of the studies that compared Psychiatry. 1980;137:324-331. 1974;19:716-723. schizoaffective disorder with schizophrenia or 22. Grove WM. A critique of Kendell and 39. Burnham KP, Anderson DR. Model Selection mood disorders. J Affect Disord. 2008;106(3): and Multimodel Inference: A Practical 209-217. Brockington’s procedure for identification of disease entities. http://www.psych.umn.edu/faculty Information-Theoretic Approach. 2nd ed. New York, 8. Conus P, Abdel-Baki A, Harrigan S, Lambert M, /grove/pubs.htm. Published March 19, 1993. NY: Springer-Verlag; 2002. McGorry PD, Berk M. Pre-morbid and outcome Accessed October 26, 2012. 40. Lake CR, Hurwitz N. Schizoaffective disorders correlates of first episode mania with psychosis: is a are psychotic mood disorders; there are no distinction between schizoaffective and bipolar I 23. Anderson R. Nonparametric methods for modeling nonlinearity in regression analysis. Annu schizoaffective disorders. Psychiatry Res. disorder valid in the early phase of psychotic 2006;143(2-3):255-287. disorders? J Affect Disord. 2010;126(1-2):88-95. Rev Sociol. 2009;35:67-85. 24. Bromet EJ, Kotov R, Fochtmann LJ, et al. 41. Malhi GS, Green M, Fagiolini A, Peselow ED, 9. Jablensky A, Woodbury MA. Dementia praecox Kumari V. Schizoaffective disorder: diagnostic and manic-depressive insanity in 1908: a Grade of Diagnostic shifts during the decade following first admission for psychosis. Am J Psychiatry. issues and future recommendations. Bipolar Disord. Membership analysis of the kraepelinian 2008;10(1, pt 2):215-230. dichotomy. Eur Arch Psychiatry Clin Neurosci. 2011;168(11):1186-1194. 1995;245(4-5):202-209. 25. Bromet EJ, Schwartz JE, Fennig S, et al. The 42. Buckley PF, Miller BJ, Lehrer DS, Castle DJ. epidemiology of psychosis: the Suffolk County Psychiatric comorbidities and schizophrenia. 10. Benabarre A, Vieta E, Colom F, Martínez-Arán Schizophr Bull. 2009;35(2):383-402. A, Reinares M, Gastó C. Bipolar disorder, Mental Health Project. Schizophr Bull. schizoaffective disorder and schizophrenia: 1992;18(2):243-255. 43. Kessler RC, Birnbaum H, Demler O, et al. The epidemiologic, clinical and prognostic differences. 26. Schwartz JE, Fennig S, Tanenberg-Karant M, prevalence and correlates of nonaffective psychosis Eur Psychiatry. 2001;16(3):167-172. et al. Congruence of diagnoses 2 years after a in the National Comorbidity Survey Replication (NCS-R). Biol Psychiatry. 2005;58(8):668-676. 11. Van Snellenberg JX, de Candia T. Meta-analytic first-admission diagnosis of psychosis. Arch Gen evidence for familial coaggregation of Psychiatry. 2000;57(6):593-600. 44. Sands JR, Harrow M. Depression during the schizophrenia and bipolar disorder. Arch Gen 27. Endicott J, Spitzer RL. A diagnostic interview: longitudinal course of schizophrenia. Schizophr Bull. Psychiatry. 2009;66(7):748-755. the Schedule for Affective Disorders and 1999;25(1):157-171.

jamapsychiatry.com JAMA Psychiatry December 2013 Volume 70, Number 12 1285

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Research Original Investigation Boundaries of Schizoaffective Disorder

45. American Psychiatric Association. Diagnostic follow-up of a large European cohort (EMBLEM). 52. Craddock N, Owen MJ. The kraepelinian and Statistical Manual of Mental Disorders. 5th ed. WorldJBiolPsychiatry. 2008;9(4):313-320. dichotomy—going, going...but still not gone. Br J Arlington, VA: American Psychiatric Association; 49. Lewis DA, Levitt P. Schizophrenia as a disorder Psychiatry. 2010;196(2):92-95. 2013. of neurodevelopment. Annu Rev Neurosci. 53. Andrews G, Goldberg DP, Krueger RF, et al. 46. Goodwin FK, Jamison KR. Manic-Depressive 2002;25:409-432. Exploring the feasibility of a meta-structure for Illness. New York: Oxford University Press; 1990. 50. Murray RM, Sham P, Van Os J, Zanelli J, Cannon DSM-V and ICD-11: could it improve utility and 47. Perugi G, Akiskal HS, Rossi L, et al. Chronic M, McDonald C. A developmental model for validity? Psychol Med. 2009;39(12):1993-2000. mania: family history, prior course, clinical picture similarities and dissimilarities between 54. Maj M, Pirozzi R, Formicola AM, Bartoli L, Bucci and social consequences. Br J Psychiatry. schizophrenia and bipolar disorder. Schizophr Res. P. Reliability and validity of the DSM-IV diagnostic 1998;173:514-518. 2004;71(2-3):405-416. category of schizoaffective disorder: preliminary 48. Van Riel WG, Vieta E, Martinez-Aran A, et al. 51. Carpenter WT Jr, Bustillo JR, Thaker GK, van Os data. J Affect Disord. 2000;57(1-3):95-98. Chronic mania revisited: factors associated with J, Krueger RF, Green MJ. The psychoses: cluster 3 of treatment non-response during prospective the proposed meta-structure for DSM-V and ICD-11. Psychol Med. 2009;39(12):2025-2042.

1286 JAMA Psychiatry December 2013 Volume 70, Number 12 jamapsychiatry.com

Downloaded From: https://jamanetwork.com/ on 09/27/2021