the prognosis of schizophrenia - Oxford Academic

56 SCHIZOPHRENIA BULLETIN

the prognosis of schizophrenia: rationale for a multidimensionalconcept*

John S. Strauss and William T. Carpenter, Jr.

What is the prognosis of schizophrenia? Thisquestion might best be answered by reviewingthe way in which the concept of schizophreniahas evolved. Originally, as indicated by its ear-lier name "dementia praecox," schizophrenia wasconsidered to have an extremely poor prognosis.In fact, this characteristic was used as a majorcriterion for considering it to be a disease.

Kraepelin's combining of prognosis and present-ing picture to define dementia praecox marked amajor shift from the earlier 19th century trendin psychiatry to conceptualize many types ofpsychosis as separate diseases based on theircross-sectional symptom pictures alone. It ap-peared to Kraepelin and several others that suchcross-sectional diagnostic criteria were not ade-quate to differentiate basic disease processes. AsSydenham had proposed two centuries before,diseases as usually conceived have a naturalhistory. Particular symptoms, individually or assyndromes, can reflect many different diseaseprocesses, or conversely a single disease maypresent with various symptom pictures.

But even from the first description of the de-mentia praecox concept, two questions arose.First, it rapidly became clear that the prognosticpieture was in fact more complex than the nameimplied. Kraepelin noted that some patientsconsidered as schizophrenic by his complex setof criteria (including such characteristics ascatatonic symptoms, indifference, and lack ofinsight) did not deteriorate. Second, he recog-nized that many clinicians believed the diverse

• Reprint requests should be addressed to the seniorauthor at Yale Psychiatric Institute, P.O. Box 12A, YaleStation, New Haven, Conn. 06520.

syndromes he had grouped together under thecategory "dementia praecox" did not in factrepresent a single disorder.

Later, in changing the name of dementia prae-cox to "schizophrenia" or "the group of schizo-phrenias," Bleuler (1950) expanded the conceptand described what he believed to be the under-lying pathological processes. The expectation ofpoor outcome, however, was maintained. Thus,the two basic questions already posed about thisdisorder were still not resolved: how many ill-nesses are found under the rubric "schizophrenia,"and how can the concept of schizophrenia remainviable when some patients so designated do nothave poor outcomes?

This review of the more recent studies on out-come and its prediction in schizophrenia will notprovide definitive answers to these questions, sinceimportant data are still not available. Such areview can, however, clarify the key issues asthey have been developed and modified by thefindings of many investigations. It can also sug-gest a way in which these findings might best besynthesized.

Three Major Views About the Prognosis ofSchizophrenia

Since the time of Kraepelin, opinions aboutprognosis in schizophrenia have taken three dif-ferent directions: (1) those related to the prog-nostic implications of the composite Kraepelinianconcept of schizophrenia involving specific symp-toms viewed cross-sectionally plus course of thedisorder; (2) those related to concepts of schizo-phrenia based on cross-sectional symptom pic-

Dow

nloaded from https://academ

ic.oup.com/schizophreniabulletin/article/4/1/56/1843799 by guest on 01 O

ctober 2022

VOL. 4, NO. 1, 1978 57

ture alone; and (3) those related primarily tothe longitudinal aspects of symptoms and socialfunctioning. It is these three orientations to prog-nosis, not the often—but mistakenly—supposedprognostic differences between broad and narrowdiagnostic concepts of schizophrenia that webelieve are crucial for clarifying prognostic fac-tors in this disorder.

The Composite Concept of Schizophrenia,Characteristic Symptoms Plus Course

Both Kraepelin and Bleuler observed deteri-orating course in the vast majority of schizo-phrenic patients, and both used this observationto validate the nosologic concept. The same ori-entation has since been followed by many others,including Langfeldt (1937), Kleist (1960), andLeonhard (1966), in their refinements of Kraepe-lin's views.

Once diagnosis and poor prognosis are seen asinextricably entwined, a question arises wheneverdiagnosed schizophrenic patients recover. Kleist(1960) and Leonhard (1966) accounted for thisphenomenon by asserting that patients who re-covered had been misdiagnosed. Extending thisview and attempting to demonstrate it em-pirically, Langfeldt (1937 and 1969) carried outa series of studies to show that particular diag-nostic criteria could be used prospectively toseparate the poor-outcome from good-outcomepatients. The diagnostic criteria specified by Lang-feldt (1937) included certain "characteristic" hal-lucinations and delusions; for example, the delu-sion of being controlled by an outside force. Insome of his writings, Langfeldt also noted thatchanges in personality may be diagnostically im-portant, although he pointed out that in mostpatients a careful consideration of the "character-istic" symptoms was decisive for the prognosis(Langfeldt 1969). The good-outcome patients whoat first glance might be considered schizophrenicwould not have these symptoms, he believed, andshould be more correctly labeled as havingschizophreniform psychoses.

Although undeniably valuable, Langfeldt's im-portant studies had some significant methodolog-ical shortcomings that raised questions about

their conclusions. Diagnoses were based on caserecord reviews and hence could easily have beencontaminated by prognostically relevant informa-tion other than the stated diagnostic criteria. Forexample, even the size of a case record providesclues to how long the patient has been hospital-ized or rehospitalized, and hence presumably tothe duration of severe incapacitation. Anotherproblem was that although the diagnostic criteriawere denned with considerable clarity for theirtime, a definitive demonstration of their relia-bility and prognostic value would have requiredthem to be operationalized still further.

As part of the work in the Washington Centerof the International Pilot Study of Schizophrenia(IPSS) (World Health Organization 1973), wewere partially able to test Langfeldt's (1969)hypothesis using operationalized and reliablesymptom criteria (Strauss and Carpenter 1974aand Hawk, Carpenter, and Strauss 1975). Aprospective design minimized any unwitting con-tamination from nondiagnostic factors associatedwith chronicity. The prognostic significance of"true" schizophrenia compared to schizophreni-form psychosis was tested by dividing schizo-phrenic patients1 according to whether or notLangfeldt's "characteristic" symptoms were ratedas present in the Present State Examination(Wing, Cooper, and Sartorius 1974) interviewgiven on admission to the study. The charac-teristic symptom criteria used in the investigationare described in the original articles, but it isworth noting that the most important were delu-sions of thoughts or actions being controlled byan outside force and severe depersonalization orderealization.

Items considered to reflect Langfeldt's criteriawere selected by us from the interview. Theselection procedure was modified following helpfulcommunication with Professor Langfeldt andothers who had used his system. Two years andagain 5 years after the initial assessment, thepatients were evaluated with standardized inter-views for assessing outcome. The results were

1 Patients were diagnosed using relatively broad, cross-sectional DSM II (American Psychiatric Association 1968)definitions.

Dow



ctober 2022


surprising in that outcome in the group of "true"schizophrenic patients with "characteristic" symp-toms was similar to the outcome of schizophrenicpatients without such symptoms. Since it wasdifficult to judge whether the degree of deper-sonalization/derealization met Langfeldt's cri-teria, we reanalyzed the data without using thesesymptoms and obtained the same results.

Other criteria besides characteristic delusionsand hallucinations, such as personality change,had also been mentioned by Langfeldt (1937) ashelpful in diagnosing schizophrenia, but thesecharacteristics have been difficult to operationa-lize and could not be empirically evaluated inour study. The results of our investigation did,however, challenge the assumption of an inex-tricable bond between "characteristic" symptomsand poor outcome in schizophrenia and suggestedthat a major part of the prognostic value of suchcomposite diagnostic systems may have comefrom patient characteristics other than the spe-cific symptoms considered as diagnostic.

Concepts of Schizophrenia Based onCross-sectional Symptomatology

An approach for diagnosing schizophrenia thatrelies on symptom criteria described cross-sec-tionally was developed by Kurt Schneider (1959).In his writings, Schneider focused on the diag-nostic importance of first-rank symptoms—delu-sions and hallucinations he considered pathogno-monic of schizophrenia in the absence of anorganic syndrome. In focusing on these specificsymptoms, Schneider deemphasized the diag-nostic role of affect, personality, interpersonalrelations, and other behavioral characteristics.The result was a clearly defined, reliable, highlyoperational system for diagnosing schizophrenia(Carpenter, Strauss, and Muleh 1973).

In a sense, Schneider and his followers can beviewed as having pursued to an extreme Lang-feldt's emphasis on characteristic symptoms ofschizophrenia by deemphasizing the chronicityand personality characteristics in the concept ofschizophrenia described by Kraepelin and Lang-feldt. This approach to diagnosing schizophrenia

improved the reliability of the diagnostic process,but the question remained: did deterioratingcourse and poor outcome still validate the con-cept of schizophrenia so defined? While Schneiderdid not deal with this issue at any length in hiswritings, it was often supposed that the greaterdiagnostic clarity made possible by his systemwould enhance the association between diagnosisand prognosis.

Using a procedure similar to that employed forevaluating Langfeldt's criteria, we failed to sup-port the hypothesis that schizophrenia defined bySchneider's symptom criteria could be validatedby course and outcome. In an analysis of symptomdata from the sample described earlier, patientsconsidered schizophrenic by Schneider's criteriawere compared with other diagnosed schizophrenicpatients not having first-rank symptoms. Therewas no significant difference between Schneider-positive and Schneider-negative schizophrenicpatients at either 2 years (Strauss and Carpenter1974a) or 5 years (Hawk, Carpenter, and Strauss1975).

Mellor's (1970) study of Schneider's diagnosticconcepts is compatible with these conclusionsin that he found schizophrenic patients withfirst-rank symptoms slightly more acute thanschizophrenic patients without first-rank symp-toms. Taylor's (1972) apparently opposite findingthat first-rank symptoms were associated withpoor outcome is open to question since he didnot classify patients based on cross-sectional,first-rank symptoms alone. Rather, he comparedpatients who had both poor premorbid functioning(longitudinally characterized) and first-ranksymptoms (cross-sectionally evaluated) with pa-tients who had both good premorbid function andabsence of first-rank symptoms.

If the narrower symptom diagnostic criteria ofSchneider and Langfeldt define groups as schizo-phrenic that have outcomes similar to patientsconsidered schizophrenic by the broad DSM II(American Psychiatric Association 1968) guide-lines, the breadth or stringency of symptom diag-nostic criteria may not be so crucial for prognosisas some have maintained. Many clinicians andinvestigators, especially in the United States,have used broad diagnostic criteria of schizo-

Dow



ctober 2022

VOL. 4, NO. 1, 1978 59

phrenia including characteristics such as mode ofrelating, and have even diagnosed as schizophrenicthose patients without delusions, hallucinations,severe thought disorder, or bizarre behavior. It isby no means certain that such broad concepts ofschizophrenia are invalid; indeed, they appearto be no less valid prognostically than the diag-nostic criteria of Langfeldt or Schneider. Nonethe-less, it has been so difficult to ensure reliabilityof the broadly based diagnostic criteria that morespecific study of their prognostic implications hasnot been possible.

We also used the data described above toevaluate the prognostic value of one other cross-sectional diagnostic system. In so doing, we hopedto maximize the opportunity of determiningwhether a narrow cross-sectional symptom diag-nosis of schizophrenia is associated with worseoutcome than a broadly defined cross-sectionalsymptom diagnosis. Using the 12-point systemderived from the IPSS by Carpenter, Strauss,and Bartko (1973), we compared "most certain"schizophrenics (8 or more points) with "leastcertain" schizophrenics (4 or fewer points), omit-ting from the study patients with intermediatescores. Once again, we failed to find a significantdifference in outcome (Hawk, Carpenter, andStrauss 1975).

None of the cross-sectional symptom diag-noses described above defined a group of schizo-phrenics with homogeneously poor outcome, andno system was prognostically superior to theothers. Nevertheless, all the diagnostic systemsused identified a group of patients with somewhatpoorer outcomes than nonschizophrenics in thesample, although there was much overlap inoutcome scores between the nonschizophrenicand the schizophrenic groups.

Although the power of any cross-sectionalsymptom-based diagnosis to identify a group ofpatients with poor outcome is limited, symptomsdo have another, perhaps even more interesting,prognostic value. The symptom picture tends topredict itself. Followup data from all centers inthe International Pilot Study of Schizophrenia(Sartorius, in press) show that if patients re-lapse, they are highly likely to have symptoms

similar to those occurring in earlier episodes ofthe disorder.

How then are we to understand the prognosticrole of symptoms considered cross-sectionally?Generally speaking, it appears that patients diag-nosed schizophrenic by any of a wide variety ofcross-sectional symptom criteria will have asomewhat less favorable course and outcome thanpatients diagnosed as having other functionalpsychiatric disorders. But the difference is lessmarked than once thought, and it does not appearto relate to which cross-sectional criteria—broador narrow—are used. Symptoms do seem to beespecially useful prognostically, not so much forpredicting likelihood of future psychopathology,but for predicting what symptoms a person willhave if any do recur.

Prognostic Importance of LongitudinalCharacteristics: History of Social Function

and Symptom Course

Our work has suggested that cross-sectionalsymptom criteria alone have far more limitedprognostic importance than many had believed.What about the possible longitudinal componentsof a diagnostic system, variables such as priorsocial function and symptom course? Thesecharacteristics have often been confounded withtype of symptoms so that any special prognosticrole they might have may have been overlooked.

One diagnostic system, for example (Feighneret al. 1972), includes, in addition to cross-sec-tional symptom criteria, the requirement thatsymptoms be continuously present for 6 monthsor more before reaching a diagnosis of schizo-phrenia. This system, which has made the im-portant contribution of operationalizing complexdiagnostic criteria, has been used effectively bya large group of investigators, including McCabeet al. (1971), Winokur and Tsuang (1975),Tsuang, Dempsey, and Rauscher (1976), andRobins et al. (1977). If Langfeldt's and Schnei-der's work is seen as having shifted emphasis toan increasing degree toward the cross-sectionalsymptom aspect of Kraepelin's dementia praecox,

Dow



ctober 2022


Robins et al. (1977) can be seen as exploringanother orientation in their introduction of opera-tionally defined longitudinal criteria related tothe patient's functioning before the time ofdiagnosis.

The Feighner et al. (1972) criteria have resultedin important prognostic and other distinctionsbetween schizophrenics and nonschizophrenics.At first, this system's ability to distinguish apoor-prognosis type of schizophrenia was at-tributed to its use of narrower criteria than havecommonly been employed for diagnosing schizo-phrenia in the United States. It appeared thatthe narrowly denned schizophrenic group wouldnot be diluted with patients who might be con-sidered as having better prognosis, such as thosediagnosed as affective disorders or reactivepsychosis.

However, the criterion of 6 months' durationof symptoms may have largely accounted for theprognostic validity of the system of Feighner etal. (1972). There is overwhelming evidence, forexample, that established chronicity is by itselfan important prognostic indicator (Brown et al.1966, Strauss and Carpenter 19746 and Vaillant1964). It is possible, therefore, that criteriacombining symptom type and prior chronicitydo not carry a unique degree of prognostic validitybeyond the predictive value of each characteristicwhen it occurs alone. That is, the prognosesassociated with symptom type and prior chronic-ity, respectively, may merely be additive factorsnot necessarily reflecting a specific disease withpoor outcome. The independent prognostic con-tributions of the various diagnostic criteria ofFeighner et al. have not been studied, and thusno one knows how much of this system's prognosticvalidity is due to symptom criteria, how much tochronicity criteria (which could apply to anysymptoms), or whether there is a unique prog-nostic value to combining these characteristics.

The system of Feighner et al. represents a steptoward incorporating longitudinal criteria asdiagnostic requirements for schizophrenia. Otherinvestigators have more extensively explored theprognostic implications of pre-illness functioning.Phillips (1966) developed a scale to test the im-portance of premorbid social adjustment in dis-

tinguishing prognostic groups in schizophreniaand other disorders. The validity of the good/poorpremorbid dichotomy as measured by this scalehas been repeatedly demonstrated in relation tooutcome and several treatment and biologicalcharacteristics (Garmezy 1968, Higgins 1969,Klorman, Strauss, and Kokes 1977, Kokes,Strauss, and Klorman 1977, and Strauss et al.1977). Although the predictive value of thePhillips Scale may be somewhat less than wasfirst believed (Serban and Gidynski 1975 andStrauss and Carpenter 19746), the prognostic im-portance of premorbid functioning appears to beestablished. Moreover, newer scales are nowavailable that may have a broader range of ap-plicability than the original instrument devisedby Phillips (Gittelman-Klein and Klein 1969,Goldstein,1 and Kokes and Strauss, in prepa-ration).

Several investigators have attempted to deter-mine whether characteristics other than thosedescribed by Phillips (1966) or some optimal groupof several characteristics might be still morevaluable in differentiating schizophrenics withpoor outcome from those who improve. The workof Stephens and Astrup (1963) and Vaillant(1964) has been among the most successful inpursuing this goal. In a series of carefully per-formed studies, these investigators have describeda variety of characteristics, such as onset ofsymptoms more than 6 months before evaluationand poor early social relationships, that help todistinguish good- and poor-outcome schizo-phrenics. Their studies are described elsewhere inthis issue and will not be presented in detail here.Two facts are important to note, however: (1)These investigators have emphasized the over-riding prognostic importance of characteristicsother than specific symptoms; in fact, the in-dependence of the prognostic characteristics fromspecific symptomatic variations within schizo-phrenia has been noted by Vaillant (1964).(2) No studies have been carried out by theseinvestigators to evaluate whether similar charac-teristics are important prognostically for patients

1 Personal communication.

Dow



ctober 2022

VOL. 4, NO. 1, 1978 61

with other psychiatric disorders than schizo-phrenia.

These facts raise the possibility that there arenot two diseases—good- and poor-prognosisschizophrenia—but that there is one disorderwith heterogeneous prognosis. Moreover, certaincharacteristics may exist, such as poor socialfunction, that are prognostically important notonly for schizophrenia but for other disorders.Patients who have poor-prognostic traits plusschizophrenic symptoms do not necessarily havea different disease than good-premorbid functionpatients with the same symptoms—any more thana debilitated, elderly person with pneumococcalpneumonia has a different form of the diseasethan a younger, more vigorous individual. Avail-able evidence about this question has been re-viewed in more detail elsewhere (Strauss et al.,in press), but it can be noted here that evidencedoes suggest the general—not specific—impor-tance of the nonsymptom prognostic indicatorsfor many disorders. A study currently beingcarried out, the First Admission Study (Strausset al. 1976), is collecting more definitive data onthis issue.

Another recurring question relevant to theissue of good and poor premorbid schizophreniaas two distinct diseases is whether measuring thevarious prognostic variables results in a bimodaldistribution. Although one study (Rosen et al.1969) reported such a bimodal distribution, thisfinding may have been an artifact of using caserecord data not collected in a structured manner(Strauss and Carpenter 19746). The prognosticdata reported by Vaillant (1964) were originallyanalyzed as good and poor prognostic dichotomies,suggesting perhaps two discrete processes. Re-viewing these data suggests, however, that thereis not a bimodal distribution of prognostic scoresor prognostic outcome correlations, but a con-tinuous relationship between prognostic ratingsand outcome. Tabulating the data presented inthat study demonstrates that three patientgroups can be defined, ranging from worst throughintermediate to best outcome, and that the prog-nostic scores for each group demonstrate a corre-sponding stepwise increment. Our findings(Strauss et al., in press) also suggest that the

relations between predictors and outcomes areroughly linear and not bimodal or skewed. Theconcept of schizophrenia as a specific poor-out-come disease is thus further challenged.

Another important part of Vaillant's (1964)and Stephens and Astrup's (1963) work was toidentify more operationally than had been donepreviously a number of prognostically importantvariables. Vaillant described seven such varia-bles: acute onset, confusion, depressive heredity,nonschizoid personality, depression, precipitat-ing factors, and concern with death. Stephens,Astrup, and Mangrum (1967) noted nine variablesof prognostic value, five overlapping with theprognostic variables identified by Vaillant, andfour additional variables: ever married, low IQ,good premorbid work record, and no emotionalblunting.

We also evaluated a number of potential prog-nostic variables other than presenting signs andsymptoms in our studies including variables sug-gested by the work of Vaillant and Stephens,Astrup, and Mangrum. Our results showed thatthe following three characteristics were con-sistently important prognostically: previous dura-tion of psychiatric hospitalization, poor pre-admission social relations functioning, and poorpre-admission work functioning.

These and other studies are impressively con-sistent in demonstrating the prognostic im-portance of certain variables. Previous durationof disorder and social relations functioning are thetwo variables most consistently found to be ofprognostic value, with occupational functioningalso frequently noted as prognostically important.The prognostic significance of these characteris-tics has been noted in several cultures, includingthe United States, Europe (Brown et al. 1966,and Stephens, Astrup, and Mangrum 1967), andAfrica (Murphy and Raman 1971). Their valuehas also been shown recently in an oriental cul-ture as well (Strauss and Chen 1977).

It seems to us that several of the other variablesconsidered of prognostic importance by some in-vestigators may have only limited value in thisregard or may be largely overlapping with thethree listed above. For example, the variable ofprecipitating events, often considered prognos-

Dow



ctober 2022


tically valuable, was in our experience extremelydifficult to rate reliably or validly (Strauss,Carpenter, and Nasrallah, in press). That ourexperience is not unique is suggested bythe difficulties demonstrated in life eventsresearch where elaborate interview and ratingprocedures have been required to obtainreliable data. It seems possible that at least someof the prognostic value associated with precipitat-ing events is an artifact of data collection inwhich such events are sought from, and reportedmore readily by, patients without a history ofchronicity or poor social relationships.

Several other characteristics, such as bluntedaffect, absence of confusion, and absence of de-pression, may have derived much of their prog-nostic value from a relationship to establishedchronicity. For example, in a recent study offirst admissions (Strauss et al. 1976) there werevery few patients (8 percent) first entering in-patient status who did not show major affectivesymptoms and signs. Those few without suchsigns tended to have psychiatric symptoms oflong duration even though hospitalization hadnot previously been required.

This finding is consistent with results from theWashington Center of the IPSS. There we usedan analysis of variance technique to determinewhich signs and symptoms were associated withgood or poor outcome in patients who, whenfirst evaluated, had disorders of relatively recentonset. Of all the symptoms assessed, only re-stricted affect (flattened or partly flattened af-fect) was significantly associated with outcome.The more restricted the patient's affective ex-pression was on admission to the study, the worsewas his 5-year outcome. Other symptoms andsigns, such as anxiety and depression, were notpredictive of outcome (Carpenter et al. 1976).

Considered together, the findings describedabove lead to the following conclusions: (a) theKraepelinian definition of schizophrenia, whichcombines course and symptoms, requires thatkey variables be further differentiated to deter-mine the sources of prognostic validity; (b)cross-sectional, symptom-based diagnosis has onlymodest prognostic power; and (c) longitudinalmeasures of function and symptom course have

significant prognostic value when consideredalone or when incorporated in diagnostic criteria.But more information is now available that helpsin further understanding the prognosis of schizo-phrenia. This information has been derived fromstudies of the outcome state itself.

The Nature of Outcome

To take the next step in understanding theprognosis of schizophrenia, one must review in-formation that has recently become availableabout the nature of the outcome to be predicted.Although outcome has often been considered asa single characteristic, recent studies have shownthat view to be inaccurate. An investigation(Strauss and Carpenter 1972) that measured fouroutcome variables (symptom severity, duration ofhospitalization, social relations functioning, andoccupational functioning) found a surprisingly lowcorrelation between these characteristics, andsimilar results have been noted in other studies(Lamb and Goertzel 1971 and Schwartz, Myers,and Astrachan 1976). Clinicians have long beenaware of the relative independence of variousoutcome measures in their patients, as, forexample, in patients who work well despite severesymptoms and in virtually asymptomatic patientswho deteriorate socially nevertheless. Based onthese findings, we suggested that different aspectsof outcome may reflect separate psychologicalprocesses; each of these aspects may bear somerelationship to the others, but may still be largelyindependent. The question arises: If outcomefunctions are to a significant extent independent ofeach other, what does that imply about prognosis?

Relationships of Predictors toOutcome Variables

Evaluating the relationships of the variouspredictors with each aspect of outcome showedthat each of three outcome variables—hospitaliza-tion, level of social relations, and level of occu-pational function—was best predicted by its

Dow



ctober 2022

VOL. 4, NO. 1, 1978 63

corresponding predictor variable. Thus, social re-lations before admission were the best predictorof social relations at outcome; work functionbefore admission was the best predictor of workfunction at outcome; and duration of hospitaliza-tion before the initial evaluation was the bestpredictor of duration of hospitalization the yearbefore followup. Other results from the literaturelend support to these findings (e.g., Monck 1963).Our results suggest that predictor-outcome rela-tionships actually represent a group of longitudinalprocesses that might best be conceptualized asopen-linked systems.

The specific predictor-outcome relationshipsdescribed above were originally found in analyz-ing predictors of data collected at the 2-yearfollowup. The same predictor-outcome relation-ships were found again in analyzing the predictorsof 5-year outcome and also in comparing the 2-and 5-year outcome findings. Although no measureof symptom severity had been made at initialevaluation, the longitudinal continuity of thisvariable, as with the others, was demonstratedby the correlations between the 2- and 5-yearfollowup results.

These findings indicate the longitudinal, con-sistent, long-term nature of the several predictor-outcome processes. The specific patterns describedabove predominated, but there was also somecrossover of prognostic relationships. It waspossible to construct a composite prognosticvariable (previous hospitalization, past socialrelations, and prior work function) of value inhelping to predict all outcome characteristics.

Are these predictor-outcome relationships spe-cific to schizophrenia? If a particular prognosisis important to the definition of schizophrenia,then one might expect the predictor-outcomeassociations in schizophrenia to be different fromthose in other diagnostic groups. To evaluate howspecific the processes described above were toschizophrenia, we investigated whether prog-nostic characteristics important for schizophreniawere also found in other functional psychiatricdisorders. A review of the relevant literature andpreliminary data from a study currently in itsconcluding phases suggested that, in fact, prog-nostic variables important for schizophrenia are

also significant in other psychiatric disorders(Strauss et al., in press).

Schizophrenia: Do Specific Symptoms TogetherWith Longitudinal Variables Reflect a Unique

Prognostic Phenomenon?

To evaluate whether there was a unique effectin combining longitudinal prognostic factors withcross-sectional symptom type in defining poor-prognosis schizophrenia as a disease, we matcheda group of schizophrenic patients and a group ofpatients with other functional psychiatric dis-orders to equate their longitudinal prognosticcharacteristics (Strauss et al., in press). Patientsseen in the early part of the First AdmissionStudy on whom 2-year followup data were alreadyavailable and patients from the WashingtonCenter of the IPSS with 2- and 5-year followupinformation were used for this analysis. Compar-ing the outcome scores of the schizophrenic andnonschizophrenic groups showed that neithertheir 2-year outcomes nor their 5-year outcomeswere significantly different in terms of socialrelations, symptoms, or a total outcome score.Nonschizophrenics were slightly better in havingless hospitalization in the second followup year(but not in the fifth) and were slightly better at5 years in employment (but not at 2 years).

These findings suggest that the longitudinalvariables of prognostic value are important inboth schizophrenic and nonschizophrenic dis-orders and do not combine in some specific waywith schizophrenic symptoms to suggest a uniquedisease, poor-outcome schizophrenia. This lastconclusion is still only speculative, however, sincethe nonschizophrenics in the groups comparedmay not have been representative of nonschizo-phrenic inpatients with functional psychiatricdisorder. This issue will be tested when followupdata, now in the final stages of collection, areavailable from the entire sample in the FirstAdmission Study.

Considering the prognostic and outcome re-search described above, we believe that concep-tualizing poor prognosis as part of a diseaseschizophrenia—or even of one type of schizo-

Dow



ctober 2022


phrenia—although possible, is not the mostparsimonious interpretation of current informa-tion. Rather, it appears to us that prognosticfactors reflect psychobiological systems that areprobably important in many functional psychi-atric disorders. When poor functioning is pre-dicted by prognostic factors, then a chroniccourse is more likely—regardless of the type offunctional psychiatric disorder. These processesdo not seem to be specific to schizophrenia but,rather, reflect a more general group of "prognosticsystems" in human functioning, at least in termsof their importance for psychiatric disorder.There is some evidence (Cassel 1964) that in factsuggests these factors may also operate in certainphysical illnesses, such as tuberculosis.

But it is important to go one step further. If,as has been claimed, "diagnosis is prognosis"—and this seems to be an important considerationin defining a diagnostic or disease entity—thenit is valuable to take these prognostic findingsand return to the concept of schizophrenia.

In an earlier paper (Strauss and Carpenter19746), we speculated that schizophrenia as de-fined by Kraepelin was a diagnostic concept bring-ing together several processes that are actuallyrelatively independent. Three such processes weresuggested: (1) "positive symptoms" (Jackson1887), such as specific hallucinations and delu-sions; (2) "negative symptoms" (Jackson 1887),such as flat affect, which we have suggested maybe indicators of established chronicity; and (3)level of social relations functioning. Perhaps afourth process, level of work functioning, shouldalso have been included. Whatever one choosesto call schizophrenia—a group of positive symp-toms, a group of negative symptoms, poor capacityfor social relations, or some combination of theseand perhaps other characteristics—it is importantnot to lose sight of the relatively independentbasic processes that appear to be involved. Theseprocesses, each possibly with its own etiological,diagnostic, and treatment implications, may needto be dealt with separately in considering patientsdiagnosed as schizophrenic.

Although such an approach to schizophreniamay appear unduly complex, more and morefrequently investigators are turning to multi-

dimensional models of psychopathology. Thesemodels are attractive because they surpass anycomposite set of diagnostic criteria in providingimpressive relationships with the growing bodyof descriptive, etiologic, and treatment informa-tion that is becoming available. For example,Bowers (1974) has suggested three dimensionsuseful for understanding the complex relation-ships found in phenomenological data describingthe schizophrenic experience. Cooper and Sar-torius (1977) have used a multidimensional orien-tation in considering cultural factors and schizo-phrenia, and Rutter, Shaffer, and Shepherd (1973)have recommended a multidimensional frame-work for classifying childhood psychiatric dis-orders.

It is interesting that of all the more traditionaldiagnostic approaches, Schneider's (1959) is mostreadily adaptable to a multisystem model sinceit limits itself primarily to certain "positive"symptoms of schizophrenia, leaving the otherimportant processes to be covered separately. Amultiaxial diagnostic model (Strauss 1975) canbe used to provide a framework for integratingsuch a number of processes into a multidimen-sional diagnostic system.

What is the prognosis of schizophrenia? Al-though over half of outcome variance cannot beaccounted for by even the best prognostic meth-ods, the best answer to the question of prognosisclearly depends on what one calls schizophrenia.However, whatever choice is made about diag-nosis, it seems most accurate to conceptualizeprognosis not as a unitary entity but as severalsemi-independent processes.

Summary

Three main ways of viewing schizophrenia existthat have different prognostic, practical, andconceptual implications. The mixed cross-sec-tional/longitudinal view of Kraepelin and othersstates essentially that there is a disease, schizo-phrenia, that can be identified by consideringboth characteristic symptoms and course. Prog-nosis in the disorder so defined is guarded, butbecause of the composite diagnostic criteria used,

Dow



ctober 2022

VOL. 4, NO. 1, 1978 65

the characteristics most important for relating tooutcome remain unknown. The second viewdefines schizophrenia primarily in terms of thecross-sectional symptom picture alone. It hasbeen shown empirically that defining a disorder inthis way has relatively limited prognostic implica-tions. The third approach identifies schizophreniain terms of several "axes," such as type ofschizophrenic symptoms, their duration, the per-son's social relations, and his work function. Theseaxes are viewed as relatively independent pro-cesses with somewhat different determinants,implying the need for different specific treatments.The authors believe this last view is the one mostsupported by currently available data.

References

American Psychiatric Association. DSM II: Diag-nostic and Statistical Manual of Mental Disorders. 2ded. Washington, D.C.: The American PsychiatricAssociation, 1968.

Bleuler, E. Dementia Praecox or the Group ofSchizophrenias. (1911) Translated by J. Zinkin. NewYork: International Universities Press, Inc., 1950.

Bowers, M. Retreat From Sanity. New York:Human Sciences Press, 1974.

Brown, G.; Bone, M.; Dalison, B.; and Wing, J.Schizophrenia and Social Care. London: OxfordUniversity Press, 1966.

Carpenter, W.T., Jr.; Bartko, J.J.; Strauss, J.S.;and Hawk, A.B. "Do Symptoms Really PredictOutcome?" Presented at the Annual Meeting of theAmerican Psychiatric Association, Miami Beach, Fla.,May 1976.

Carpenter, W.T., Jr.; Strauss, J.S.; and Bartko, J.J.A flexible system for the identification of schizo-phrenia: A report from the International Pilot Studyof Schizophrenia. Science, 182:1275-1278, 1973.

Carpenter, W.T., Jr.; Strauss, J.S.; and Muleh, S.Are there pathognomonic symptoms of schizophrenia?Archives of General Psychiatry, 29:443-449, 1973.

Cassel, J. Social science theory as a source ofhypotheses in epidemiological research. AmericanJournal of Public Health, 9(54):1482-1488, 1964.

Cooper, J., and Sartorius, N. Cultural and temporalvariations in schizophrenia: A speculation on theimportance of industrialization. British Journal ofPsychiatry, 130:50-55, 1977.

Feighner, J.; Robins, E.; Guze, S.; Woodruff, R.,

Jr.; Winokur, G.; and Munoz, R. Diagnostic criteriafor use in psychiatric research. Archives of GeneralPsychiatry, 26:57-63, 1972.

Garmezy, N. Process and reactive schizophrenia:Some conceptions and issues. In: Katz, M.M.; Cole,J.O.;and Barton, W.E., eds. The Role and Methodologyof Classification in Psychiatry and Psychopathology.Washington, D.C.: U.S. Public Health Service Publi-cation No. 1584, Superintendent of Documents, U.S.Government Printing Office, 1968. pp. 419-466.

Gittelman-Klein, R., and Klein, D.F. Premorbidasocial adjustment and prognosis in schizophrenia.Journal of Psychiatric Research, 7:35-53, 1969.

Hawk, A.B.; Carpenter, W.T., Jr.; and Strauss, J.S.Diagnostic criteria and 5-year outcome in schizo-phrenia: A report from the International Pilot Studyof Schizophrenia. Archives of General Psychiatry,32:343-347, 1975.

Higgins, J. Process-reactive schizophrenia: Recentdevelopments. Journal of Nervous and Mental Disease,149:450-472, 1969.

Jackson, H. Remarks on evolution and dissolutionof the nervous system. Journal of Mental Science,33:25-48, 1887.

Kleist, K. Schizophrenic symptoms and cerebralpathology. Journal of Mental Science, 106:246-255,1960.

Klorman, R.; Strauss, J.S.; and Kokes, R.F.Premorbid adjustment in schizophrenia:' Concepts,measures, and implications: Part IV. Some biologicalapproaches to research on premorbid functioning inschizophrenia. Schizophrenia Bulletin, 3(2):226-239,1977.

Kokes, R.F., and Strauss, J.S. Developmental Pre-morbid Adjustment Scale. University of Rochester,Rochester, N.Y. In preparation.

Kokes, R.F.; Strauss, J.S.; and Klorman, R. Pre-morbid adjustment in schizophrenia: Concepts, mea-sures, and implications: Part II. Measuring premorbidadjustment: The instruments and their development.Schizophrenia Bulletin, 3(2):186-213, 1977.

Kraepelin, E. Dementia Praecox. Translated byR.M. Barclay. Edinburgh: E.S. Livingstone, Ltd.,1919.

Lamb, H.R., and Goertzel, V. Discharged mentalpatients—are they really in the community? Archivesof General Psychiatry, 24:29-34, 1971.

Langfeldt, G. The Prognosis in Schizophrenia andthe Factors Influencing the Course of the Disease.Copenhagen: E. Munksgaard, 1937.

Langfeldt, G. Schizophrenia, diagnosis and prog-nosis. Behavioral Science, 14:173-182, 1969.

Dow



ctober 2022


Leonhard, K. The question of prognosis in schizo-phrenia. International Journal of Psychiatry, 2:633-635, 1966.

McCabe, M.S.; Fowler, R.C.; Cadoret, R.J.; andWinokur, G. Familial differences in schizophreniawith good and poor prognosis. Psychological Medicine,1:326-332, 1971.

Mellor, C.S. First-rank symptoms of schizophrenia..British Journal of Psychiatry, 17:15-23, 1970.

Monck, E.M. Employment experiences of 127discharged schizophrenic men in London. BritishJournal of Preventive Social Medicine, 17:101-110,1963.

Murphy, H.B.M., and Raman, A.C. Course ofschizophrenia in indigenous tropical peoples. BritishJournal of Psychiatry, 118:489-497, 1971.

Phillips, L. Social competence, the process-reactivedistinction, and the nature of mental disorder. In:Hoch, P.H., and Zubin, J., eds. Psychopathology ofSchizophrenia. New York: Grune & Stratton, Inc.,1966. pp. 471-481.

Robins, E.; Gentry, K.A.; Munoz, R.A.; andMarten, S. A contrast of the three more commonillnesses with the ten less common in a study and18-month follow-up of 314 psychiatric emergency roompatients. Archives of General Psychiatry, 34:259-265,1977.

Rosen, B.; Klein, D.F.; Levenstein, S.; andShahinian, S.P. Social competence and posthospitaloutcome among schizophrenic and nonschizophrenicpsychiatric patients. Journal of Abnormal Psychology,74:401-404, 1969.

Rutter, M.; Shaffer, D.; and Shepherd, M. An eval-uation of the proposal for a multiaxial classification ofchild psychiatric disorders. Psychological Medicine,3:244-250, 1973.

Sartorius, N. Validity of diagnostic concepts acrosscultures: A preliminary report from the InternationalPilot Study of Schizophrenia. In: Wynne, L.C.;Cromwell, R.L.; and Matthysse, S., eds. Nature ofSchizophrenia: New Findings and Future Strategies.New York: John Wiley & Sons, Inc., in press.

Schneider, K. Clinical Psychopathology. Translatedby M.W. Hamilton. New York: Grune & Stratton,Inc., 1959.

Schwartz, C.C.; Myers, J.K.; and Astrachan, B.M.Concordance of multiple assessments of outcome ofschizophrenia. Archives of General Psychiatry, 32:1221-1227, 1976.

Serban, G., and Gidynski, C.B. Differentiatingcriteria for acute-chronic distinction in schizophrenia.Archives of General Psychiatry, 32:705-712, 1975.

Stephens, J., and Astrup, C. Prognosis in processand non-process schizophrenia. American Journal ofPsychiatry, 119:945-953, 1963.

Stephens, J.; Astrup, C ; and Mangrum, J.C.Prognosis in schizophrenia: Prognostic scales cross-validated in American and Norwegian patients.Archives of General Psychiatry, 16:693-698, 1967.

Strauss, J.S. A comprehensive approach to psy-chiatric diagnosis. American Journal of Psychiatry,132 (11):1193-1197, 1975.

Strauss, J.S., and Carpenter, W.T., Jr. Predictionof outcome in schizophrenia: I. The nature of outcome.Archives of General Psychiatry, 27: 739-746, 1972.

Strauss, J.S., and Carpenter, W.T., Jr. Character-istic symptoms and outcome in schizophrenia. Archivesof General Psychiatry, 30:429-434, 1974a.

Strauss, J.S., and Carpenter, W.T., Jr. The predic-tion of outcome in schizophrenia. II. Relationshipsbetween predictor and outcome variables: A reportfrom the WHO International Pilot Study of Schizo-phrenia. Archives of General Psychiatry, 31:37-42,1974b.

Strauss, J.S.; Carpenter, W.T., Jr.; and Bartko, J.J.An approach to the diagnosis and understanding ofschizophrenia: Part III. Speculations on the processesthat underlie schizophrenic symptoms. SchizophreniaBulletin, 1 (Experimental Issue No. ll):61-70, 1974.

Strauss, J.S.; Carpenter, W.T., Jr.; and Nasrallah,A.T. How reliable is the psychiatric history? Com-prehensive Psychiatry, in press.

Strauss, J.S., and Chen, C.C. "Cross-CulturalPredictors of Outcome in Schizophrenia." Presentedat the VI World Congress of Psychiatry, Honolulu,Hawaii, September 1977.

Strauss, J.S.; Kokes, R.F.; Carpenter, W.T., Jr.;and Ritzier, B. The course of schizophrenia as adevelopmental process. In: Wynne, L.C.; Cromwell,R.L.; and Matthysse, S., eds. Nature of Schizophrenia:New Findings and Future Strategies. New York: JohnWiley & Sons, Inc., in press.

Strauss, J.S.; Kokes, R.F.; Klorman, R.; and Sack-steder, J. Premorbid adjustment in schizophrenia:Concepts, measures, and implications: Part I. Theconcept of premorbid adjustment. SchizophreniaBulletin, 3{2):182-185, 1977.

Strauss, J.S.; Kokes, R.F.; Ritzier, B.A.; Harder,D.W.; and VanOrd, A. "Crucial Patterns of Psy-chiatric Disorder." Presented at the Annual Meetingof the American Psychiatric Association, MiamiBeach, Fla., May 1976.

Taylor, M. Schneiderian first-rank symptoms andclinical prognostic features in schizophrenia. Archives

Dow



ctober 2022

VOL. 4, NO. 1, 1978 67

of General Psychiatry, 26:64-67, 1972.Tsuang, M.T.; Dempsey, G.M.; and Rauscher, F.

A study of "atypical schizophrenia." Archives ofGeneral Psychiatry, 33:1157-1160, 1976.

Vaillant, G.E. Prospective prediction of schizo-phrenic remission. Archives of General Psychiatry,11:509-518, 1964.

Wing, J.K.; Cooper, J.E.; and Sartorius, N. TheMeasurement and Classification of Psychiatric Symp-toms. London: Cambridge University Press, 1974.

Winokur, G., and Tsuang, M. The Iowa 500:Suicide in mania, depression, and schizophrenia.American Journal of Psychiatry, 132(6) :650-651, 1975.

World Health Organization. International PilotStudy of Schizophrenia. Vol.1. Geneva: World HealthOrganization Press, 1973.

Acknowledgment

This study was supported in part by NationalInstitute of Mental Health Grants MH 25466-02and MH 00006-01.

The Authors

John S. Strauss, M.D., is Professor ofPsychiatry and Director, Yale PsychiatricInstitute, New Haven, Conn. William T.Carpenter, Jr., M.D., is Director, Mary-land Psychiatric Research Center, Bal-timore, Md.

community psychology program

Community Psychology is a developing field within psychology which incorporates knowledge fromthe various social sciences to effect environmental and social system change through action-orientedanalysis, research, and implementation. It is also a field of graduate study leading to a Ph.D. inpsychology.

It offers the opportunity to develop a program of academic study and practical experience that is con-gruent with personal and professional goals, as well as a conceptual framework and practical analytictool for working in and with the social systems that include, impinge upon, and affect the individualfamily and group.

The field is of special interest to persons who are interested in how systems, institutions, and environ-ments affect the individual; who have a commitment to social change through active research and/orintervention into human service delivery systems; and who have practical exposure to or research interestin issues of social control, social institutions, and social systems.

For further information call or write: Community Psychology Program (GF), University of Michigan,529 Thompson, Ann Arbor, MI 48109; or call (313) 763-0264.

Dow



ctober 2022

the prognosis of schizophrenia - Oxford Academic

Documents

Transcript of the prognosis of schizophrenia - Oxford Academic