Cargando…

F136. PARSING DUP TO REFINE EARLY DETECTION: QUANTILE REGRESSION OF RESULTS FROM THE SCANDINAVIAN TIPS STUDY

BACKGROUND: Prolonged duration of untreated psychosis (DUP) is associated with poor outcome. The Scandinavian TIPS study deployed an early detection (ED) campaign to halve DUP. However, while reducing DUP will improve outcomes for most patients, there are some for whom prolonged DUP is a byproduct o...

Descripción completa

Detalles Bibliográficos
Autores principales: Ferrara, Maria, Guloksuz, Sinan, Burke, Shadie, Li, Fangyong, Friis, Svein, Hegelstad, Wenche ten Velden, Joa, Inge, Johannessen, Jan Olav, Melle, Ingrid, Simonsen, Erik, Srihari, Vinod
Formato: Online Artículo Texto
Lenguaje:English
Publicado: Oxford University Press 2018
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5888098/
http://dx.doi.org/10.1093/schbul/sby017.667
_version_ 1783312445993910272
author Ferrara, Maria
Guloksuz, Sinan
Burke, Shadie
Li, Fangyong
Friis, Svein
Hegelstad, Wenche ten Velden
Joa, Inge
Johannessen, Jan Olav
Melle, Ingrid
Simonsen, Erik
Srihari, Vinod
author_facet Ferrara, Maria
Guloksuz, Sinan
Burke, Shadie
Li, Fangyong
Friis, Svein
Hegelstad, Wenche ten Velden
Joa, Inge
Johannessen, Jan Olav
Melle, Ingrid
Simonsen, Erik
Srihari, Vinod
author_sort Ferrara, Maria
collection PubMed
description BACKGROUND: Prolonged duration of untreated psychosis (DUP) is associated with poor outcome. The Scandinavian TIPS study deployed an early detection (ED) campaign to halve DUP. However, while reducing DUP will improve outcomes for most patients, there are some for whom prolonged DUP is a byproduct of an insidiously illness rather than a modifiable prognostic factor. It is also unclear whether the success of an ED program relies on targeting those with longer or shorter DUP. Previously, we demonstrated that quantile regression (QR) can both manage skewed distributions and allow analysis for meaningful heterogeneity in DUP. The current study aims to investigate the utility of QR to analyze the impact of an ED campaign across the DUP distribution, using data from the TIPS study. We hypothesized the effectiveness of TIPS’s ED campaign will vary across different quantiles of DUP. METHODS: Between 1997 and 2000, a comprehensive early detection (ED) program with public information campaigns and low-threshold psychosis detection teams was established in one health-care area (ED-area), but not in a comparable area (No-ED area). Users with DSM IV non-organic non-affective first episode psychosis were consecutively recruited. Demographic, social and clinical characteristics of people enrolled in an ED area were compared to those coming from a No-ED area. Quantile regression can model the relationship between conditional quantiles of response and independent variables. Unlike ordinary least-squares regression that focuses on conditional mean response, QR estimates the heterogeneous effects of ED across different quantiles of DUP, rather than presuming a uniform mean effect. It is particularly useful when the differential effect of predictors on lower or upper quantile of outcome are of interest. In this study, we examined the impact of ED across the entire quantiles of DUP, particularly on Q1, Q2, Q3, dividing data into four quartiles. A post hoc analysis of the effect of gender, marital status, premorbid adjustment social level, and social cluster on quartiles of DUP was also conducted. RESULTS: The total sample included 301 subjects, of which 161 belonged to an early detection (ED) area. If compared to users from No-ED area, ED users were younger (mean age 25 vs 31), and mainly unmarried (80% vs 62%). QR highlighted that ED had no effect on the first quartile (Q1) of DUP, with very short DUP, even in the No-ED area. ED was significantly associated with a reduction in the second quartile of DUP (median) (11 weeks reduction, p<0.001), and the third quartile of DUP (Q3) (41 weeks reduction, p=0.01). The effect of ED was significantly stronger on last quartile than Q1(p=0.01) and Q2 (p=0.04) suggesting a stronger effect of campaign on people with longer DUP. After controlling for age and marital status, the ED campaign’s effect on Q3 of DUP significantly differed by gender: only male users in the ED group showed a significant reduction in this quartile of DUP (coefficient [SE] at Q3=−46.6; P = .01), suggesting an interaction between gender and ED campaigns in reducing DUP. DISCUSSION: Quantile regression represents a powerful tool reveal the different effects of an ED campaign across DUP distribution. The upper tail of DUP benefited the most from ED program: users with longer DUP might be hesitant to engage into treatment because of a longstanding active psychosis or failed attempts in the community to receive care. Very short DUP, highly associated with rapidly escalating symptoms, was not affected by the campaign. This could represent a subgroup of patients for which no specific ED efforts are needed or further reductions may need strategies targeting prodromal signs. These findings have been fertile ground for generating hypotheses that could lead to targeted ED efforts.
format Online
Article
Text
id pubmed-5888098
institution National Center for Biotechnology Information
language English
publishDate 2018
publisher Oxford University Press
record_format MEDLINE/PubMed
spelling pubmed-58880982018-04-11 F136. PARSING DUP TO REFINE EARLY DETECTION: QUANTILE REGRESSION OF RESULTS FROM THE SCANDINAVIAN TIPS STUDY Ferrara, Maria Guloksuz, Sinan Burke, Shadie Li, Fangyong Friis, Svein Hegelstad, Wenche ten Velden Joa, Inge Johannessen, Jan Olav Melle, Ingrid Simonsen, Erik Srihari, Vinod Schizophr Bull Abstracts BACKGROUND: Prolonged duration of untreated psychosis (DUP) is associated with poor outcome. The Scandinavian TIPS study deployed an early detection (ED) campaign to halve DUP. However, while reducing DUP will improve outcomes for most patients, there are some for whom prolonged DUP is a byproduct of an insidiously illness rather than a modifiable prognostic factor. It is also unclear whether the success of an ED program relies on targeting those with longer or shorter DUP. Previously, we demonstrated that quantile regression (QR) can both manage skewed distributions and allow analysis for meaningful heterogeneity in DUP. The current study aims to investigate the utility of QR to analyze the impact of an ED campaign across the DUP distribution, using data from the TIPS study. We hypothesized the effectiveness of TIPS’s ED campaign will vary across different quantiles of DUP. METHODS: Between 1997 and 2000, a comprehensive early detection (ED) program with public information campaigns and low-threshold psychosis detection teams was established in one health-care area (ED-area), but not in a comparable area (No-ED area). Users with DSM IV non-organic non-affective first episode psychosis were consecutively recruited. Demographic, social and clinical characteristics of people enrolled in an ED area were compared to those coming from a No-ED area. Quantile regression can model the relationship between conditional quantiles of response and independent variables. Unlike ordinary least-squares regression that focuses on conditional mean response, QR estimates the heterogeneous effects of ED across different quantiles of DUP, rather than presuming a uniform mean effect. It is particularly useful when the differential effect of predictors on lower or upper quantile of outcome are of interest. In this study, we examined the impact of ED across the entire quantiles of DUP, particularly on Q1, Q2, Q3, dividing data into four quartiles. A post hoc analysis of the effect of gender, marital status, premorbid adjustment social level, and social cluster on quartiles of DUP was also conducted. RESULTS: The total sample included 301 subjects, of which 161 belonged to an early detection (ED) area. If compared to users from No-ED area, ED users were younger (mean age 25 vs 31), and mainly unmarried (80% vs 62%). QR highlighted that ED had no effect on the first quartile (Q1) of DUP, with very short DUP, even in the No-ED area. ED was significantly associated with a reduction in the second quartile of DUP (median) (11 weeks reduction, p<0.001), and the third quartile of DUP (Q3) (41 weeks reduction, p=0.01). The effect of ED was significantly stronger on last quartile than Q1(p=0.01) and Q2 (p=0.04) suggesting a stronger effect of campaign on people with longer DUP. After controlling for age and marital status, the ED campaign’s effect on Q3 of DUP significantly differed by gender: only male users in the ED group showed a significant reduction in this quartile of DUP (coefficient [SE] at Q3=−46.6; P = .01), suggesting an interaction between gender and ED campaigns in reducing DUP. DISCUSSION: Quantile regression represents a powerful tool reveal the different effects of an ED campaign across DUP distribution. The upper tail of DUP benefited the most from ED program: users with longer DUP might be hesitant to engage into treatment because of a longstanding active psychosis or failed attempts in the community to receive care. Very short DUP, highly associated with rapidly escalating symptoms, was not affected by the campaign. This could represent a subgroup of patients for which no specific ED efforts are needed or further reductions may need strategies targeting prodromal signs. These findings have been fertile ground for generating hypotheses that could lead to targeted ED efforts. Oxford University Press 2018-04 2018-04-01 /pmc/articles/PMC5888098/ http://dx.doi.org/10.1093/schbul/sby017.667 Text en © Maryland Psychiatric Research Center 2018. http://creativecommons.org/licenses/by/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
spellingShingle Abstracts
Ferrara, Maria
Guloksuz, Sinan
Burke, Shadie
Li, Fangyong
Friis, Svein
Hegelstad, Wenche ten Velden
Joa, Inge
Johannessen, Jan Olav
Melle, Ingrid
Simonsen, Erik
Srihari, Vinod
F136. PARSING DUP TO REFINE EARLY DETECTION: QUANTILE REGRESSION OF RESULTS FROM THE SCANDINAVIAN TIPS STUDY
title F136. PARSING DUP TO REFINE EARLY DETECTION: QUANTILE REGRESSION OF RESULTS FROM THE SCANDINAVIAN TIPS STUDY
title_full F136. PARSING DUP TO REFINE EARLY DETECTION: QUANTILE REGRESSION OF RESULTS FROM THE SCANDINAVIAN TIPS STUDY
title_fullStr F136. PARSING DUP TO REFINE EARLY DETECTION: QUANTILE REGRESSION OF RESULTS FROM THE SCANDINAVIAN TIPS STUDY
title_full_unstemmed F136. PARSING DUP TO REFINE EARLY DETECTION: QUANTILE REGRESSION OF RESULTS FROM THE SCANDINAVIAN TIPS STUDY
title_short F136. PARSING DUP TO REFINE EARLY DETECTION: QUANTILE REGRESSION OF RESULTS FROM THE SCANDINAVIAN TIPS STUDY
title_sort f136. parsing dup to refine early detection: quantile regression of results from the scandinavian tips study
topic Abstracts
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5888098/
http://dx.doi.org/10.1093/schbul/sby017.667
work_keys_str_mv AT ferraramaria f136parsingduptorefineearlydetectionquantileregressionofresultsfromthescandinaviantipsstudy
AT guloksuzsinan f136parsingduptorefineearlydetectionquantileregressionofresultsfromthescandinaviantipsstudy
AT burkeshadie f136parsingduptorefineearlydetectionquantileregressionofresultsfromthescandinaviantipsstudy
AT lifangyong f136parsingduptorefineearlydetectionquantileregressionofresultsfromthescandinaviantipsstudy
AT friissvein f136parsingduptorefineearlydetectionquantileregressionofresultsfromthescandinaviantipsstudy
AT hegelstadwenchetenvelden f136parsingduptorefineearlydetectionquantileregressionofresultsfromthescandinaviantipsstudy
AT joainge f136parsingduptorefineearlydetectionquantileregressionofresultsfromthescandinaviantipsstudy
AT johannessenjanolav f136parsingduptorefineearlydetectionquantileregressionofresultsfromthescandinaviantipsstudy
AT melleingrid f136parsingduptorefineearlydetectionquantileregressionofresultsfromthescandinaviantipsstudy
AT simonsenerik f136parsingduptorefineearlydetectionquantileregressionofresultsfromthescandinaviantipsstudy
AT sriharivinod f136parsingduptorefineearlydetectionquantileregressionofresultsfromthescandinaviantipsstudy