The Impact of Early Detection (ED) Campaigns on Care Presentations: Beyond DUP Reduction

Abstract

We examined the effects of an early detection (ED) campaign (Mindmap), that successfully shortened the duration of untreated psychosis (DUP), on patient presentation profiles at two receiving coordinated specialty care (CSC) services. Data were collected between 2015-2019 during a test of ED delivered at one CSC (STEP, n = 147) compared to usual detection at another CSC (PREP, n = 63). Regression models were used to test the effects of ED and DUP on presentation. Before the launch of ED, there were no differences in presentation between STEP and PREP. However, the ED changed the profile of presentations to STEP such that patients were admitted with better negative and total symptoms scores, but worse GAF current and GAF social and with a greater decline in function over the prior year (GAF-Δ). Site-by-time interaction effects were not significant. During the campaign years, STEP vs. PREP recruited patients with better negative and total symptoms, GAF role, and pre-morbid adjustment scores but with worse positive symptoms, GAF current, and GAF-Δ. Nonetheless, mediation analysis revealed that DUP reduction accounted for very little (< 8%) of these differences in presentation. Early detection campaigns while successfully reducing access delays, can have salutary effects on presentation independent of DUP reduction.

1. Introduction

First Episode Services (FES) utilize dedicated teams to deliver a package of interventions for early-course schizophrenia and have emerged as an international standard of care (Correll et al., 2018). In the U.S., such service models, also called Coordinated Specialty Care (CSC), have undergone extensive and rapid implementation (Read and Kohrt, 2022). Yet, prolonged delays between psychosis onset and entry into care can diminish the effectiveness of CSC (Kane et al., 2016). This delay known as the duration of untreated psychosis (DUP-total) encompasses delays to the prescription of antipsychotic medication and delays in enrollment to FES (DUP-supply). These delays in either the supply or the demand side contribute to delays in the overall DUP (DUP-demand).

The pioneering Scandinavian Treatment and Intervention in Psychosis (TIPS) early detection (ED) campaign (Melle et al., 2005, 2004a) which halved DUP across a healthcare catchment (Ferrara et al., 2019), found that reduction in DUP resulted in recruiting patients with better clinical and functional presentations (Melle et al., 2004a), but with similar levels of social functioning (Melle et al., 2005). The benefits of reducing the DUP were evident not only during presentation, but also in improved outcomes observed up to a decade later (Hegelstad et al., 2012; Ten Velden Hegelstad et al., 2013).

We previously reported the first controlled demonstration of DUP reduction across a U.S. community (Srihari et al., 2022a, 2014a). In this prospective controlled design, we measured DUP for consecutive admissions at two equivalent CSC for 1 year before (pre-ED) and during 4-years of early detection (ED) (Mindmap) targeting one clinic’s catchment area (Srihari et al., 2022b, 2014b). Mindmap combined three elements of mass & social media messaging, professional outreach & detailing of referral sources, and performance improvement to reduce wait times (Ferrara et al., 2021; Mathis et al., 2022). The duration from onset of psychosis to admission at STEP’s CSC (DUP-Total) fell from a median of 312 to 149 days (44.5 to 21 weeks or 10 to 5 months) by the end of the campaign. The quasi-experimental design and careful measurement of DUP enabled us to test, in the current paper, the effect of ED and DUP reduction on admission profiles. We hypothesized that the reduction in DUP at the ED site would result in lower levels of distress and dysfunction at clinical presentation.

2. Methods

2.1. Setting and Study Design

Data for this analysis were drawn from a quasi-experimental study with a contemporaneous control site. DUP was measured for consecutive admissions to two equivalent CSCs located in Boston, Massachusetts (Prevention and Recovery in Early Psychosis, PREP) and New Haven, Connecticut (clinic for Specialized Treatment Early in Psychosis, STEP) for 1 year before (Feb 1, 2014–Jan 31, 2015), and 4-years (Feb 1, 2015–Jan 31, 2019) after the launch of an ED campaign (Mindmap) targeting only STEP’S catchment. PREP continued to offer usual detection. A study protocol detailing the design and analytic approach was published before implementation (Srihari et al., 2014b). All subjects provided written informed consent. All study procedures were approved and monitored by the Institutional Review Boards of Yale University, Beth Israel Deaconess Medical Center, and the Massachusetts Department of Mental Health.

2.2. Participants

STEP and PREP recruited comparable, diverse samples reflective of local demography (Srihari et al., 2022c). These were young (M = age 22.2 years, SD = 3.5) and predominantly male (71%) patients with a notable proportion of first-generation immigrants (17%).

2.3. Measures

2.3.1. Duration of Untreated Psychosis (DUP)

Three measures of delay were utilized: DUP-Total (time in days from psychosis onset to enrollment in CSC), DUP-Demand (time in days from psychosis onset to first antipsychotic medication), and DUP-Supply (time in days from first antipsychotic medication to enrollment in CSC). DUP-Demand and DUP-Supply thus quantify successive phases of delay in pathways to care. The Structured Interview for Psychosis-risk Syndromes (SIPS) (McGlashan et al., 2014) was used to establish the date of transition to psychosis. Psychosis onset was defined as meeting the Presence of Psychotic Syndrome (POPS) criteria i.e., at least 6 in severity on scales P1-P5 of SIPS, with at least one of these symptoms occurring over a period of one month for at least one hour per day at a minimum average frequency of 4 days per week or leading to serious disorganization or dangerousness. The POPS date was determined with input from all available stakeholders in the pathway to care, identified within a structured questionnaire. A medication questionnaire was used to date the first use of antipsychotic medication for psychosis. For those participants who were antipsychotic naïve at presentation (DUP-Supply = 0), DUP-Demand was equal to DUP-Total.

2.3.2. Presentation measures

Symptom severity was measured using the Positive and Negative Symptom Scale (PANSS)(Kay et al., 1987). The Global Functioning (GAF: Hall, 1995) (integrated within the SIPS) measured current and recent social, occupational, and psychological functioning (range 0 to 100). The Global Functioning: Social Scale (GF: Social) and the Global Functioning: Role Scale (GF: Role) measured the current, lowest, and highest social and occupational functioning respectively, in the past year (range 0 to 10) (Cornblatt et al., 2007). GAF-Δ was calculated as the 12-month GAF – current GAF scores and quantified functional deterioration in the year before. Premorbid Adjustment Scale (PAS)(Cannon-Spoor et al., 1982) score was used to define the level of premorbid functioning only for the childhood epoch (age 5–12) preceding the usual onset of prodromal symptoms. Suicidality was assessed with the Columbia Suicide Severity Scale (C-SSRS)(Posner and Brent, 2003).

2.4. Statistical analysis

Clinical presentation was analyzed using two-way ANOVA for continuous outcomes. Site (STEP or PREP) or campaign (pre or during) effects and their interaction were included as independent variables. The interaction term was used to assess the effect of the campaign in STEP relative to changes taking place during the same period in PREP. Similarly, the Cochran-Mantel-Haenszel test was used to compare the rate of suicide attempts within and between sites. Relative risk and 95% confidence intervals were computed. To examine the contribution of DUP to the overall effect of the campaign on presentation, we performed an exploratory mediation analysis to compare changes in the regression coefficient of the campaign effect after including DUP as a covariate. We pre-specified a minimum of 10% change as a clinically meaningful mediation effect. All analyses were performed using SAS 9.4. Statistical significance was set as p < 0.05, two-sided. Given the exploratory intent, we did not correct for multiple comparisons.

3. Results

During the baseline study period (pre-ED), there were no statistically significant differences (Table 1) in presentation at STEP (n = 24) vs. PREP (n = 12). Patients enrolled at each CSC had similar levels of PANSS (positive, negative, general, total), GAF (role, social, current), suicidality (frequency, severity, and number of attempts), and childhood pre-morbid adjustment scores.

Table 1.

Main and interaction effects of the site (STEP vs. PREP) and ED (pre-ED vs during ED)

Outcomes	Within-site comparison	STEP	PREP	Site Difference	P value
Symptom severity
Positive symptoms	Pre-ED	20.1 (17.7 to 22.5)	18.5 (15.1 to 21.9)	−1.6 (−5.7 to 2.6)	0.45
	ED	19.6 (18.6 to 20.5)	17.3 (15.8 to 18.8)	−2.3 (−4.0 to −0.5)	0.01
	Within-site difference	−0.5 (−3.1 to −2.1)	−1.2 (−4.9 to −2.5)		0.76
	P value	0.69	0.52
Negative symptoms	Pre-ED	19.3 (16.7 to 21.9)	20.4 (16.8 to 24.0)	1.1 (−3.3 to −5.6)	0.61
	ED	16.2 (15.2 to 17.2)	20.3 (18.7 to 21.9)	4.1 (2.2 to 6.0)	<.0001
	Within-site difference	−3.1 (−5.9 to −0.3)	−0.1 (−4.1 to 3.8)		0.22
	P value	0.03	0.95
General PANSS	Pre-ED	37.7 (34.2 to 41.1)	39.0 (34.2 to 43.8)	1.3 (−4.6 to 7.3)	0.65
	ED	34.0 (32.7 to 35.4)	36.8 (34.7 to 39.0)	2.8 (0.3 to 5.3)	0.03
	Within-site difference	−3.6 (−7.3 to 0.0)	−2.2 (−7.4 to 3.1)		0.65
	P value	0.05	0.42
Total PANSS	Pre-ED	77.0 (70.5 to 83.6)	77.9 (68.6 to 87.2)	0.9 (−10.5 to 12.2)	0.88
	ED	69.8 (67.1 to 72.5)	74.4 (70.4 to 78.5)	4.6 (−0.2 to 9.5)	0.06
	Within-site difference	−7.3 (−14.3 to −0.2)	−3.5 (−13.6 to 6.6)		0.55
	P value	0.045	0.49
		Function
GAF-Δ	Pre-ED	−10.5 (−17.8 to −3.2)	−3.7 (−13.9 to 6.6)	6.8 (−1.4 to 1.9)	0.28
	ED	−23.9 (−26.9 to −21.0)	−6.4 (−11.0 to −1.8)	17.6 (−1.1 to 0.3)	<.0001
	Within-site difference	−13.4 (−0.4 to 1.6)	−2.7 (−1.5 to 1.4)
	P value	0.0009	0.64		0.12
GF:Role	Pre-ED	4.3 (3.5 to 5.2)	4.2 (3.0 to 5.4)	−0.2 (−1.7 to 1.3)	0.82
	ED	4.3 (4.0 to 4.7)	2.9 (2.4 to 3.4)	−1.4 (−2.1 to −0.8)	<.0001
	Within-site difference	−0.01 (−0.9 to 0.9)	−1.3 (−2.6 to 0.04)
	P value	0.97	0.06		0.12
GF : social	Pre-ED	5.7 (54 to 6.3)	5.2 (4.3 to 6.0)	−0.5 (−1.6 to 0.5)	0.30
	ED	5.0 (4.7 to 5.2)	5.1 (4.7 to 5.5)	0.1 (−0.3 to 0.6)	0.55
	Within-site difference	−0.7 (−1.4 to −0.1)	−0.1 (−1.0 to 0.9)		0.23
	P value	0.02	0.90
Current GAF	Pre-ED	36.7 (32.4 to 40.9)	37.6 (31.6 to 43.6)	0.9 (−6.4 to 8.2)	0.80
	ED	30.4 (28.7 to 32.1)	39.0 (36.4 to 41.7)	8.6 (5.4 to 11.8)	<.0001
	Within-site difference	−6.2 (−10.8 to - 1.7)	1.5 (−5.1 to 8.0)		0.06
	P value	0.007	0.66
Pre-morbid adjustment scale
Childhood	Pre-ED	0.2 (0.2 to 0.3)	0.3 (0.2 to 0.3)	0.0 (−0.1 to 0.2)	0.47
	ED	0.2 (0.2 to 0.2)	0.3 (0.3 to 0.3)	0.1 (0.1 to 0.2)	<.0001
	Within-site difference	−0.02 (−0.09 to 0.04)	0.04 (−0.05 to 0.14)		0.25
	P value	0.47	0.37
Suicidality
The severity of suicide ideation	Pre-ED	3.6 (2.4 to 4.8)	2.4 (1.2 to 3.6)	−1.1 (−2.9 to 0.6)	0.18
	ED	2.8 (2.5 to 3.2)	3.3 (2.8 to 3.9)	0.5 (−0.1 to 1.2)	0.11
	Within-site difference	−0.7 (−2.0 to 0.5)	0.9 (−0.4 to 2.2)		0.07
	P value	0.24	0.17
Frequency of suicide ideation	Pre-ED	2.0 (0.9 to 3.1)	2.6 (1.5 to 3.7)	0.6 (−1.0 to 2.2)	0.47
	ED	2.2 (1.9 to 2.6)	2.7 (2.2 to 3.2)	0.5 (−0.1 to 1.1)	0.11
	Within-site difference	0.2 (−0.9 to 1.4)	0.2 (−1.1 to 1.4)		0.92
	P value	0.68	0.80
				Relative Risk (RR)
Suicide Attempt	Pre-ED, N (%)	6 (25)	2 (16.7)	1.5 (0.36 to 6.35)	0.57
	ED, N (%)	26 (17.7)	14 (22.2)	0.80 (0.45 to 1.42)	0.44
	RR (95% CI)	1.41 (0.65 to 3.07)	0.75 (0.2 to 2.88)		0.70
	P value	0.3944	0.66

^**Data are presented as least squared mean and 95% Confidence interval unless otherwise specified in the table.

During the ED period, STEP (n =147) vs. PREP (n = 63) recruited patients with better negative symptoms (difference = 4.1, 95% CI: 2.2 to 6.0 p < .0001), general symptoms (difference = 2.8, 95% CI: 0.3 to 5.3 p = .03), GAF role (difference = −1.4 95% CI: −2.1 to −0.8, p < 0.001), and childhood premorbid adjustment scores (difference = 0.1, 95% CI: 0.1 to 0.2, p < 0.001) but with worse positive symptoms (difference = −2.3, 95% CI: −4.0 to −0.5, p = 0.01), GAF current (difference =8.6, 95% CI: 5.4 to 11.8, p < 0.0001), and GAF-Δ (difference = 17.6, 95% CI: −1.1 to −0.3, p < 0.001). No differences were found in levels of suicidality (frequency, severity, and number of attempts). Across all variables tested, there were no significant interaction effects between site and ED on presentation (P value of DIFF > .05). That is, exposure to ED did not explain differences in presentation between STEP and PREP across the study period.

No significant changes were observed within PREP over these 5 years. However, in STEP, the presentation profile shifted during the campaign. During ED vs pre-ED periods, STEP recruited patients with less severe negative (difference = −3.1, 95% CI: −5.9 to −0.3 p = 0.03) and total PANSS symptom scores (difference = −7.3, 95% CI: −14.3 to −0.2, p = 0.045) but with worse GAF social (−0.7 CI: −1.4 to −0.1, p = 0.02), GAF current (difference= −6.2, 95% CI: −10.8 to 1.7, p = 0.008) and GAF-Δ scores (difference = −13.4, 95% CI: −0.4 to −1.6, p = 0.0009). No difference was found in the level of suicidality (frequency, severity, and number of attempts).

Mediation analysis showed that changes in presentation profile to STEP during ED vs pre-ED periods were not well explained by reductions in DUP (Table 2). A shorter DUP-Total was associated with reductions in negative and total PANSS scores, GAF social, current, and increased GAF-Δ at the conventional level of statistical significance (p < .05). However, the magnitude of the contribution of DUP-Total to these effects on presentation was small (< 8%). Similar results were found for DUP-Demand and DUP-Supply (see supplementary material).

Table 2.

Mediation effect of DUP-Total on presentation profile at STEP

	ED total effect	ED+Dup Total (p-value)	Difference (%)
Negative Symptoms	−3.10	−3.012 (.03)	3.0
PANSS total score	−7.24	−6.82 (.03)	5.9
GF: Social	−0.73	−0.73 (.02)	0.7
Current GAF	−6.23	−6.10 (.009)	2.0
GAF-Δ	−13.49	−12.41 (.001)	7.6

4. Discussion

The main findings of this analysis are that an early detection campaign was associated with notable changes in the profile of presentation to care at a first-episode service, but that much of these changes were not explained by reductions in DUP. The relationship between early detection campaigns and within this, the specific effects of DUP reduction, on the health status of those presenting to first-episode services, is complex.

Comparisons of presentations to STEP during ED vs. pre-ED epochs demonstrated that patients had better negative and total PANSS scores, but worse GAF current and social scores and notably, a greater decline in function over the prior year (GAF-Δ). This profile is consistent with a more acutely ill sample, whose rate of functional decline is more rapid and evident, and whose positive symptoms are more labile (e.g., related to variable medication adherence or substance use rather than DUP). Notably, the negative symptom burden at presentation was reduced over the course of the campaign. This profile is known to predict better response to treatment and is consistent with the findings of the only other successful reduction of community-level DUP (Johannessen et al., 2005; Melle et al., 2004). In contrast to the pre-ED group, patients who presented to STEP during the ED epoch had better negative and total PANSS scores, indicating a lower severity of psychotic symptoms. However, they also had worse GAF current and social scores, suggesting a poorer level of functioning in various domains. Notably, they also presented a greater decline in function over the prior year (GAF-Δ), which may have been more easily evident to caregivers. Interestingly, the negative symptom burden at presentation was reduced over the course of the campaign, which may imply that the ED campaign was effective in reaching out to patients who were less impaired by negative symptoms or who developed them later in their illness course. A plausible interpretation of our findings is that patients in the initial stages of their illness may undergo a significant decline in functioning. This occurs even when their overall symptom severity is comparable to, or slightly better than, that observed at non-ED sites. Another potential explanation could be the combined effect of the ED campaign and the noticeable loss of function. This combination might have encouraged healthcare providers to refer patients for treatment earlier, without any further deterioration in symptoms. In other words, patients were able to access treatment without being in a worse symptomatic state. This profile is consistent with a more acutely ill sample, whose positive symptoms are more labile (e.g., related to variable medication adherence or substance use rather than DUP). Surprisingly, this profile is also known to predict better response to treatment and is consistent with the findings of the only other successful reduction of community-level DUP (Johannessen et al., 2005; Melle et al., 2004).

An unexpected finding from our study was that the reduction in DUP accounted for a relatively small proportion (less than 8%) of the observed differences in presentation. This has several implications for the effects of early detection campaigns on presentations to care. The relative reduction of DUP-Total across our US catchment (median of 44.5 to 21 weeks) was comparable to the Scandinavian effort (median of 16 vs 5 weeks)(Srihari et al., 2022). However, our absolute levels of DUP (four times the TIPS post-ED DUP) may have been too high to demonstrate a mediational effect of DUP reduction on presentations to care. Indirect evidence in support of this is a recent analysis suggesting that differences in long-term outcomes were detectable across an absolute DUP cutoff of as low as 3 weeks (Golay et al., 2022). The Mindmap campaign demonstrated progressive reductions in DUP over 4 years and it is possible that extending the campaign may have reduced DUP to low enough levels to reveal a stronger mediation effect. Conversely, it is easy to conceive of the salutary effects that a community-wide ED campaign like Mindmap might have had on pathways to care and hence presentation, independent of its effects on DUP. For example, the intensive efforts to educate lay and clinical stakeholders in STEP’s catchment may have improved identification and help-seeking on behalf of those individuals with the greatest decline in functioning (GAF-Δ), or before more severe negative symptoms were apparent.

Additionally, the ED campaign at STEP included a specific and successful focus on reducing wait times after referral to less than seven days (Ferrara et al., 2022). This performance improvement effort often necessitated individuals' and families' engagement before discharge from acute care settings, such as inpatient units and emergency rooms. This contrasted with usual detection at PREP, which may have favored enrollment for those patients and families who were able to navigate and/or tolerate routine and slower outpatient admission procedures. Thus, the control site had a lower patient volume. This possible source of sampling bias illustrates the challenge of maintaining representative recruitment at a CSC without some proactive efforts to engage young individuals who often lack illness awareness or are ambivalent about psychiatric treatment. Notably, PREP's median DUP-Total (~64 weeks) was comparable to the largest reported US sample.

Finally, we have previously reported on the complex and interactive roles of clinical and non-clinical actors within our regional network (Mathis et al., 2022) that influenced referral and engagement into care; and the differentially better impact of ED on the DUP of those first-episode patients who began help-seeking during the prodromal vs post-psychosis phases of their illness (Ferrara et al., 2021). While beyond the scope of this analysis, these reports demonstrate that the intentionally multi-pronged Mindmap campaign had complex effects on the behavior of actors across both the ‘demand’ (patients, families and community referral sources) and ‘supply’ (healthcare providers and agencies) sides of regional pathways to care. This may have contributed to presentation profiles at STEP independent of effects on quantitative delay (DUP).

A few limitations should be noted. During the ED period, STEP compared to PREP, enrolled patients with better negative and general symptoms, GAF role, and childhood pre-morbid adjustment scores but with worse positive symptoms, GAF current, and GAF-Δ. However, low recruitment at the control site resulted in a lack of statistical power to detect the effect of exposure to ED on these differences between the sites. Also, while ED appeared to recruit individuals with greater recent declines in functioning (and so perhaps in greatest need of care), and fewer negative symptoms, additional analysis will be necessary to determine if these profiles are associated with better responsiveness to CSC. Lastly, given the difference in populations, especially in PAS childhood score, it is possible to argue that the ED campaign did not truly impact the FEP in our catchment area, but rather, it merely redirected a sample that appeared to be affected to the STEP program. The way to counter this argument would be to gather similar measures for FEP from other clinics within our catchment, a task that was unfeasible in our context. Nonetheless, we have compelling indirect evidence suggesting that such a sampling bias was not at play. Since 2014, STEP has been operating as a population health service with a defined geographic catchment. The FEP service has been active in this region since 2006. Throughout the campaign, we observed an increase in referral volume and a broader range of DUP than what was historically observed. These factors collectively suggest that our sample is more, not less, representative of our target catchment (comprising 10 towns).

In summary, while successfully reducing access delays, the Mindmap early detection campaign had specific effects on presentation to the receiving first-episode service that were largely independent of DUP reduction. While DUP reduction remains an important public health goal, ED campaigns should embrace and measure other salutary effects on local pathways to care that can be leveraged to engage all individuals and families who can benefit from specialty first-episode services.