Reducing the Duration of Untreated First-Episode Psychosis:  Effects on Clinical Presentation FREE

The period from the onset of frank psychotic symptoms to the start ofadequate treatment, the duration of untreated psychosis (DUP), varies considerablyin patients with first-episode psychosis. Several studies have shown meanDUPs of 1 year or more.¹ Staying psychoticfor months or years will obviously influence social, occupational, and interpersonalfunctioning. Most studies on first-episode psychosis show an association betweena long DUP and poorer short-term outcome,^2- 9 althoughsome newer studies have questioned this finding.^10- 13

The mechanism(s) of this association is poorly understood,¹⁴ andthe correlational designs of all existing studies make it impossible to drawany conclusions regarding the direction of the relationship. A long DUP mayonly be an epiphenomenon or a marker of preexisting common factors that contributeto the poor outcome.^1,15 The increasingpsychosocial problems and stigma associated with the untreated psychosis mayalso serve as an indirect link between the DUP and outcome. A long DUP mightalso contribute directly to a poorer prognosis through destructive biologicaland psychological processes.¹ If the DUP iscausally related to outcome, it may be one of the few known prognostic factorsin schizophrenia that theoretically can be modified by interventions. Thishas sparked a number of studies on the early phases in psychosis, includingseveral on early treatment. No study has managed to reduce DUP significantly,¹⁶ but some have been good models for the developmentof early-intervention programs.⁶

A study that aims at disentangling the effects of the DUP from confoundingfactors requires modifying the DUP differentially across patient groups. Themost promising way to change the timing of initial treatment in psychosis(ie, the DUP) is to introduce early detection (ED) programs into an interventionservice. The ideal design to test whether there is a DUP effect would be arandomized controlled study assigning a group of first-episode patients todelayed treatment, but this has major ethical and practical difficulties.An alternative design is quasi-experimental, in which patients from one healthcare sector serving an entire population receive an experimental approach,in this case the ED program, and are compared with patients from another healthcare sector who do not receive the experimental approach. The control populationcan be parallel and consist of patients from a geographically distinct healthcare sector, or it can be historical and consist of patients from the experimentalhealth care sector before the experimental approach was introduced.^17- 18 A combination of the parallel andhistorical control designs in a single study would compensate for severalof the methodological problems inherent in quasi-experimental designs. Noprevious studies have attempted to modify the DUP with a parallel controldesign.

The present report originates from the Early Treatment and Interventionin Psychosis Study, a prospective longitudinal study of first-onset psychosisfrom 4 Scandinavian health care sectors with equivalent first-episode treatmentprograms. Two of the sectors carried out an extensive ED program (ED area)and the other 2 did not (No-ED area) in a parallel control design.¹⁹ A historical control comparison between patientstreated within the first 2 years of the ED program and patients treated withinthe same health care sector 4 years previously showed that the DUP was significantlyshorter for ED program patients.²⁰ This stronglyindicated that the program had an ability to reduce the DUP,²⁰ mostprobably through a reduced threshold for detection. The historical controldesign cannot control for any cohort effects, such as the possibility of ageneral increase in attention to first-episode patients, or the local effectsof introducing an ambitious research program. This effect is minimized bythe parallel control design. The present report addresses the following 2main hypotheses of the parallel control study: (1) Adding an ED program willreduce the DUP for ED-area patients compared with No-ED–area patients.(2) The decrease in the DUP is related to a reduced threshold for enteringtreatment and will be paralleled by less severe symptom levels at baseline,ie, from start of treatment throughout the acute phase, in ED-area patientscompared with No-ED–area patients. The present study does not deal withthe question of differences in treatment outcome during the first year oftreatment; this issue will be addressed in later communications.

The study was carried out on the basis of the specialist psychiatricservices in 4 Scandinavian health care sectors (the 2 health care sectorsin Rogaland County, Norway, with a total of 370 000 inhabitants, andthe Ullevaal health care sector, Oslo County, Norway, and the midsector, RoskildeCounty, Denmark, with a combined total of 295 000 inhabitants). The populationsof the health care sectors lived mainly in urban and suburban areas (Stavanger,Norway, with suburban areas; parts of Oslo, Norway, with suburban areas; andparts of greater Copenhagen, Denmark, suburban areas), with 84% of the EDpopulation living in urban areas vs 96% of the No-ED population (P<.001, Fisher exact test). There were no differences in age andsex distribution between the 2 areas and no differences in main income levelsand unemployment rates in the total population. Patients from the ED areawere less likely to be immigrants from a nonwestern country (4% vs 12%) andto have education after high school (21% vs 31%) (P<.001for both, Fisher exact test).

In all 4 health care sectors, the treatment systems were catchment areabased and publicly funded. The core basis of the psychiatric specialist treatmentsystem was subsector catchment area–based outpatient units, with theaddition of short-term, acute-care inpatient units and a restricted numberof specialized long-term wards (4 different outpatient units and 2 hospitalsin the ED area and 5 outpatient units and 2 hospitals in the No-ED area).The treatment system was based on referral from general practitioners, butthe assessment and treatment of first-episode psychosis were considered atask for the specialized psychiatric treatment system, with immediate transferof recognized cases as standard practice. There were no differences in theuse of inpatient psychiatric services across the areas.

From January 1, 1997, the specialized psychiatric services of the 4health care sectors established equivalent treatment programs for patientswith first-episode psychosis. All first-episode patients from all 4 sectorsunderwent assessment by trained personnel as soon as possible after firstcontact with the specialized treatment system and were assigned to the treatmentprogram without delay. The program consisted of defined treatment algorithmsfor antipsychotic medication (low-dose second-generation antipsychotic medication),individual psychosocial treatment (a trained psychiatric case worker offeringweekly sessions), and psychoeducational family work (multifamily groups).The program was based on experiences from other first-episode treatment programsand the recommendations of the Schizophrenia Patients Outcomes Research Team.²¹ In 2 of the health care sectors (Rogaland County),an extensive ED program was added that was not performed in the 2 other participatinghealth care sectors (Oslo and Roskilde counties).

The ED program consisted of educational campaigns about psychotic symptomsand their treatment directed at the general population through newspaper,radio, and cinema advertisements and targeted information campaigns directedat general practitioners, social workers, and high school health care personnel.In addition, specialized low-threshold ED teams were established that couldbe reached by a single telephone call from potential patients, family, andfriends, and the telephone number was made part of the information campaign.The organization of the ED program was based on the idea that treatment couldbe delayed owing to lack of awareness or to prejudices about psychotic disordersand their treatments in the psychotic person, in the person's network, and/oramong first-line health care and social welfare personnel. Treatment couldsecondarily be delayed owing to difficulties in accessing psychiatric services.A tertiary delaying factor could be delays in identifying cases and/or resourcelimitations in assigning adequate treatment for first-episode patients atthe level of the specialized psychiatric services. The ED program addressedthe first 2 factors, with the intention of bringing first-episode patientsearlier into the specialized psychiatric services assessment and treatmentprograms.¹⁹ Establishing equivalent treatmentprograms in the ED and No-ED areas would necessarily reduce delays on thelevel of the specialized services equally and thus carried the possibilityof shortening the DUP in the No-ED area.

The study included all eligible patients meeting study criteria forthe period of January 1, 1997, through December 31, 2000. Participants werefollowed up with personal interviews after 3 months and 1 and 2 years. Five-yearfollow-up has started, and 10-year follow-up is planned. Only data from baselineand the 3-month follow-up are parts of this report. The Regional Committeefor Research Ethics approved the study. All patients entering the study gavewritten informed consent. Information regarding nonparticipants is based ondata gathered anonymously for the purpose of bias testing.

Inclusion criteria consisted of living in the catchment area of 1 ofthe 4 health care areas (south and north sectors, Rogaland County [ED area],and Ullevaal sector, Oslo County, and midsector, Roskilde County [No-ED area]);age 18 to 65 years; meeting the DSM-IV criteria forschizophrenia, schizophreniform disorder, schizoaffective disorder (narrowschizophrenia spectrum disorders) or brief psychotic episode, delusional disorder,affective psychosis with mood-incongruent delusions, or psychotic disordernot otherwise specified; being actively psychotic, as measured by a Positiveand Negative Syndrome Scale (PANSS)²² scoreof 4 or more on at least 1 of positive subscale items 1 (delusions), 3 (hallucinatorybehavior), 5 (grandiosity), or 6 (suspiciousness/persecution) or general subscaleitem 9 (unusual thought content); not receiving previous adequate treatmentfor psychosis (defined as antipsychotic medication of >3.5 haloperidol equivalentsfor >12 weeks or until remission of the psychotic symptoms); having no neurologicalor endocrine disorders with relationship to the psychosis; having no contraindicationsto antipsychotic medication; understanding and/or speaking a Scandinavianlanguage; having an IQ score of above 70; and being willing and able to giveinformed consent.

The length of the study recruitment period (4 years) was based on apredefined estimate of the expected number of first-episode patients (15/100 000per year, with 25% refusal of informed consent) and a power analysis of patientsnecessary to demonstrate clinically significant differences in outcome (aminimum of 100 patients in each of the 2 study groups).

A total of 874 persons with psychosislike symptoms seeking help fromthe specialized psychiatric services underwent screening by the study's assessmentteams (131/100 000). As expected, because of campaigns that encouragedpatients with possible psychotic symptoms to seek help, the number of personsundergoing screening was higher in the ED area. However, slightly fewer patientsmet the study criteria in the ED area (186 [50/100 000] vs 194 [66/100 000]from the No-ED area). The yearly incidence varied considerably from studyyear to study year across and within the 2 study areas (ED area highest, 17/100 000[year 1]; ED area lowest, 10/100 000 [year 3]; No-ED area highest, 18/100 000[year 4]; No-ED area lowest, 13/100 000 [year 1]). These incidence ratesare within the expected range for schizophrenia and related disorders.²³ These observed differences in incidence may be dueto natural variations in the base rate of relatively rare disorders. Alternatively,the trend toward a higher incidence in the No-ED areas may be accounted forby the higher degree of urbanization.²⁴

Most patients (n = 249 [66%]) meeting study criteria were hospitalizedat first contact; there were no significant differences for the ED area (n= 113 [61%]) vs the No-ED area (n = 136 [70%]) (P =.07). Of the 380 eligible patients, 284 gave informed consent to enter, but3 later withdrew this, leaving 281 patients aged 18 to 65 years who gave informedconsent (141 in the ED area and 140 in the No-ED area [74% of all patientsmeeting study criteria]). As reported elsewhere,²⁵ patientswho did not enter the study had significantly longer DUPs than patients whoentered (median, 26 weeks [range, 0-936 weeks] and 10 weeks [range, 0-1196weeks], respectively; P<.001, Mann-Whitney test).This finding was replicated in both areas when they were examined separately.We found no other significant differences between patients who did and didnot enter the study.

Assessment teams in Rogaland County; the Ulleval sector, Oslo County;and the midsector, Roskilde County, consisted of clinically experienced andtrained research personnel who performed all evaluations.²⁶ Difficultevaluations were discussed in regular team meetings to arrive at consensusratings. The Structured Clinical Interview for the DSM-IV Axis I Disorders was used for diagnostic purposes.²⁷ Symptomlevels were measured by means of the PANSS.²² Globalfunctioning was measured by the Global Assessment of Functioning Scale (GAF),²⁸ and the scores were split into symptom (GAFs) andfunction (GAFf) scores to improve psychometric properties. Misuse of alcoholand other drugs was measured by the Drake Scale.²⁹

The DUP was measured as the time from onset of psychosis until the startof adequate treatment. Onset of psychosis was equated with the first appearanceof positive psychotic symptoms, defined as the first week with symptoms correspondingto a PANSS score of 4 or more on positive subscale items 1, 3, 5, or 6 oron general subscale item 9. Adequate treatment was defined as the start ofstructured treatment with antipsychotic medications or the start of hospitalizationin highly staffed psychiatric wards organized to manage disturbing psychoticsymptoms. All available sources, including semistructured personal interviewswith patients and relatives, were used to ascertain the length of this period.In the rare case of previous short, self-remitting psychotic episodes, thelengths of these episodes were added to the DUP.

Premorbid functioning was measured by the Premorbid Adjustment Scale(PAS).³⁰ The PAS subdivides the premorbid periodinto the following: childhood (age, <11 years), early adolescence (age,12-15 years), late adolescence (age, 16-18 years), and adulthood (age, ≥19years). The last year before onset of psychosis was not taken into accountin the PAS evaluation to avoid interference by functional loss in the prodromalphase. Scores for adulthood are not reported here, since a relatively largegroup of patients had onset of psychosis before 18 years of age. The PAS resultsare most often reported as an index for each period. Recent reports indicatethat the PAS covers the following 2 discrete areas of functioning: academic(school performance and school adaptation) and social (social accessibility/isolation,peer relationships, and capacity to establish sociosexual relationships [thelast item was not rated for childhood]).³¹ Adivision of the PAS into academic and social adaptation (mean scores acrosschildhood and early and late adolescence) were used in the statistical analyses.

All raters were trained to reliability in the use of study instrumentsby rating previously prepared case notes and audiotapes/videotapes beforeentering the study assessment teams.²⁶ Reliabilityfor the PANSS scores was measured by the rating of actual videotaped interviewsof first-episode patients by all raters. Reliability for the diagnosis, GAFscores, and DUP was measured by the rating of actual case notes by maskedraters (S.F. and S.O.) with long clinical research experience who had notparticipated directly in the patient assessments. Written vignettes were usedbecause the multiple sources for the DUP ascertainment could not be coveredfully by a taped interview. Reliability of measurements ranged from fair tovery good (DUP, 0.99; GAFs score, 0.63; GAFf score, 0.75; PANSS positive sumscore, 0.88; PANSS negative sum score, 0.76; and PANSS general sum score,0.56 [all intraclass correlations, 1.1]; for diagnostic categories, κ= 0.76). Measurements of the duration of the prodromal phase did not achievesufficient reliability, and duration of untreated illness (including the prodrome)was taken out of the analyses.

Analyses were performed with the statistical package SPSS (version 11.0;SPSS Inc, Chicago, Ill). The applied methods are reported for all group comparisons.All tests were 2-tailed. We used nonparametric tests for data without normaldistribution. As noted in several other studies, the DUP does not have a normaldistribution, whereas its natural logarithm does.¹⁴ Allanalyses that included the DUP were nonparametric, with the exception of multiplelinear regression analyses. In these, the DUP was transformed to its naturallogarithm. The multiple linear regression analyses were performed for 2 explicitreasons. The first was to find a statistical model with a good predictionof the DUP. The second and most important reason was to investigate whetherdifferences between ED and No-ED areas were due to confounding factors. Themultiple linear regression analyses were thus performed hierarchically inseveral steps, with coming from the ED area entered last. First, all variableswere entered that showed group differences between the ED and No-ED samples,that had a significant bivariate relationship to one of the dependent variablesin the present study, or that demonstrated potential predictive power fromprevious studies. The variable of coming from the ED area was entered last.Because this procedure included a large number of variables, they were organizedinto different domains or blocks (eg, demographics, premorbid adjustment,diagnostics, personality factors and social integration, and ED status).

Most independent variables were known predictors of outcome in schizophreniaspectrum psychosis. The personality factors and social integration domainincluded variables that might be indicators of differences in help seekingor detection. The PANSS baseline items of lack of insight, poor volition,hostility, withdrawal, passive/apathetic withdrawal, and active social avoidancewere chosen as signs of personality factors that could be related to reluctanceto seek treatment. The variables of being married, frequency of family contact,frequency of family telephone contact, frequency of contact with others, frequencyof telephone contact with others, working the last year before first treatment,and studying the last year before first treatment were chosen as signs ofcontacts with other persons who might intervene and aid help seeking.

In most cases, we found multiple intercorrelations among variables withina domain, and we chose a subset of variables to represent the domain on thebasis of the strength of their relationship to the dependent variable andthe goodness of fit of the model. This implies that slightly different subsetswere possible, but changing the number and content of subsets did not haveany influence on the main findings regarding differences between areas. Continuousindependent variables were examined for nonlinear relationships with the dependentvariable and transformed (to its natural logarithm or to a binary variableat identified thresholds) if necessary. The final model was examined for interactioneffects and the effects of outliers and influential observations, includingleverages. The strength of prediction of an individual domain or variablewas based on a judgment of changes in R² andF for blocks and by the adjusted β coefficient for individual variables.Only the final models are presented herein.

The GAFs score was chosen as the dependent variable representing clinicalstatus, owing to the high correlations with GAFf and PANSS scores measuredsimultaneously. There were no differences regarding main results if the GAFsscore was exchanged with the PANSS positive or the PANSS total symptom scoresin complementary analyses. These analyses are not presented herein.

The main differences between study participants from the ED and No-EDareas were that ED-area patients more often were Scandinavian, younger, andsingle and misused alcohol or other drugs (Table 1).

The DUP was short in both areas, with a combined median of 10 weeks(range, 0-1196 weeks; mean, 49.1 weeks; SD, 120.6 weeks). The DUP was significantlyshorter in the ED area than in the No-ED area, with a median of 5 weeks (range,0-1196 weeks) compared with 16 (range, 0-966 weeks) (P =.003, Mann-Whitney test). Symptom levels were high and global functioningwas low at the start of first treatment, but patients from the ED area hadsignificantly lower symptom levels and higher functioning across all measuredclinical domains. These differences in clinical status were still presentbut less prominent after 3 months (Table2).

There was a high level of intercorrelation between the DUP, ED area,and important demographic and clinical variables (Table 3). All potentially significant variables were thus enteredin a hierarchical stepwise multiple linear regression analysis with the DUPas the dependent variable (Table 4).Sex, premorbid adjustment (academic), a narrow schizophrenia spectrum diagnosis,PANSS active social avoidance, weekly telephone contacts with family, andworking during the last year before contact each had a significant relationshipto the DUP. The direction of relationships were that female sex, good premorbidfunctioning, the absence of a diagnosis of narrow schizophrenia spectrum disorder,less social avoidance, more family contacts, and working during the last yearbefore treatment start were associated with a shorter DUP. Coming from theED area retained its statistically significant influence (coming from ED areaassociated with shorter DUP) when entered on the last step of the analysis.

Concerning clinical status at the start of first treatment, only a narrowschizophrenia spectrum diagnosis, the DUP, and coming from the ED area hada significant influence so that the absence of a narrow schizophrenia spectrumdiagnosis, a longer DUP, and coming from the ED area were associated witha better clinical status at start of treatment (Table 5). Coming from the ED area was the strongest predictor, evenwhen entered at the last step.

At 3 months, premorbid functioning (social), working the previous year,and DUP emerged as statistically significant predictors (Table 6). The direction of the association was that better premorbidadjustment, employment, and a shorter DUP were associated with higher 3-monthGAFs scores. Coming from the ED area retained its significant relationshipwith a better clinical status, even with correction for baseline differencesin the GAFs score and when entered at the last step of the analysis.

The results of the present study demonstrate that it is possible toinfluence the DUP by means of a comprehensive ED program. The study designensures that the DUP differences are not caused by general or specific cohorteffects such as changes in awareness of psychotic disorder or effects on recruitmentthrough better assessment and treatment programs in the psychiatric services.These factors are probably important, as the median DUP observed in this study'sNo-ED group is among the shorter DUPs registered.¹⁴

There are some group differences between ED and No-ED patients. Thelarger number of patients with a non-Scandinavian origin in the No-ED groupis probably based on real population differences. The difference in age atfirst contact and the associated differences in marital status and misuseof alcohol and other drugs (single status and misuse are significantly morecommon in younger patients) are more difficult to explain, since there areno observed differences in the age distribution of catchment area populations.The median age at first contact is within the expected range insofar as severallarge studies report these medians in the first half of the fourth decadeof life,^32- 33 even if mediansin the middle of the third decade are more common epidemiologically. As first-episodeschizophrenia spectrum psychoses are relatively rare, these differences couldreflect random variations in small annual samples. This could pose a problemif there were interactions between age at onset and the effect of the ED program,but to our knowledge there are no studies that indicate differences in help-seekingbehavior between people in their second and third decades of life. The resultsof this study indicate that demographic, premorbid, diagnostic, and personalityfactors have separate influences on the DUP, in addition to the ED program,but that the effect of the ED program is strongest. We believe that it isreasonable to conclude that the observed shortening of the DUP does not appearto be based on confounding factors such as group differences in demographicor clinical characteristics between recruited patients.

The relationships among patient characteristics, the DUP, an ED program,and clinical status in the baseline period (at start of first treatment andat 3-month follow-up) are complex and warrant some comments. The DUP stillvaries considerably within both areas, as several ED patients have long DUPsand most No-ED patients have short DUPs. As differences in DUP are not fullyexplained by the ED program, we may still expect to find significant relationshipsbetween the DUP and clinical status. At 3 months, we find a strong negativecorrelation between the DUP and level of improvement, in line with most studiesin this area. Few studies, however, have examined the relationship betweenthe DUP and symptoms at start of treatment, and most of these have not foundany significant associations.^34- 35 Theonly exception found a weak positive correlation between the DUP and symptomlevels,³⁶ whereas the present study found aweak negative correlation. A negative correlation makes some sense from aclinical point of view, as it is reasonable to expect that very symptomaticpatients enter treatment more rapidly, but it is too early to draw any conclusionsregarding the relationship between the DUP and symptoms at start of firsttreatment.

The study's main finding is that with all other factors being equal,early detection efforts will bring people into treatment at lower symptomlevels. It is thus reasonable to conclude that the threshold for seeking and/orentering treatment appears to have been lowered by the ED program, and thatthe differences in clinical states are not based on confounding factors andappear to be stable through the initial phase of treatment. Whether the mainpivotal point for the lowered threshold resides with the patient, the family,the patient's network, or the general practitioner warrants further study.

From a methodological point of view, a randomized controlled trial wouldbe a stronger design than the present quasi-experiment. A randomized controlledtrial was discussed in the planning process of this study, but the idea wasdiscarded for several reasons. It is not possible to randomize the effectof a mass media educational campaign or the detection of first-episode patientsby professionals. Randomization would have to take place after detection,by experimentally delaying the start of adequate treatment. It was judgedto be unethical to address the public with information on probable harmfuleffects of a long-standing psychosis, followed by a request to submit to therisk of delayed treatment. Most patients would in this situation probablyrefuse to give informed consent to enter the randomized controlled trial.

The strength of the study is its basis in a catchment area treatmentsystem that constitutes all possible treatment services for first-episodepatients from that particular geographical area. The study sample is thusan epidemiological sample. The publicly paid treatment system ensures thatthere are few major differences in quality of care offered in the participatingsectors. This is strengthened by the introduction of a common treatment programfor first-episode patients. If follow-up data confirm that the 2 patient groupshave received equivalent treatments, the basis is provided for a study ofthe relationship between the DUP and outcome where it is possible to disentanglethe effects of DUP from potential confounding factors for the first time.

Corresponding author and reprints: Ingrid Melle, MD, Division ofPsychiatry, Ullevaal University Hospital, Kirkeveien 166, N-0407 Oslo, Norway(e-mail: Ingrid.melle@psykiatri.uio.no).

Submitted for publication December 9, 2000; final revision receivedMay 4, 2003; accepted August 26, 2003.

This study was supported by grant 133897/320 from the Norwegian NationalResearch Council (Oslo); the Norwegian Department of Health and Social Affairs(Oslo); grant 1997/41 from the National Council for Mental Health/Health andRehabilitation (Oslo); Rogaland County (Stavanger, Norway) and Oslo County(Oslo) (Drs Vaglum, Johannessen, Friis, Larsen, Melle, and Opjordsmoen); theTheodore and Vada Stanley Foundation (Bethesda, Md); the Regional Health ResearchFoundation for Eastern Region (Hilleroed, Denmark); Roskilde County (Roskilde,Denmark); Helsefonden Lundbeck Pharma (Hellerup, Denmark); Eli Lilly Denmark(Lyngby); Janssen-Cilag Pharmaceuticals Denmark (Birkeroed) (Drs Simonsenand Haahr); a National Alliance for Research on Schizophrenia and Depression(NARSAD) Distinguished Investigator Award (Great Neck, NY); grant MH-01654from the National Institute of Mental Health (Rockville, Md) (Dr McGlashan);and a Young Investigator Award from NARSAD (Dr Larsen).

This study is part of the Early Treatment and Intervention in PsychosisStudy project with the following research group: Thomas McGlashan, MD (principalinvestigator); Per Vaglum, MD (principal investigator); Svein Friis, MD; UlrikHaahr, MD; Jan Olav Johannessen, MD; Tor K. Larsen, MD; Ingrid Melle, MD;Stein Opjordsmoen, MD; Bjørn Rishovd Rund, PhD; and Erik Simonsen,MD.