Introduction
Pediatric acute-onset neuropsychiatric syndrome (PANS) is a descriptive entity consisting of acute-onset OCD and/or eating disorder accompanied by a wide range of secondary psychiatric and somatic symptoms [
1]. Initial and recurring symptoms may be severe and lead to significant loss of function [
2,
3]. There are no clearly established evidence-based treatments for PANS [
4] but the long-term prognosis is generally positive, with approximately two thirds of patients presenting with minimal or no symptoms 2–5 years after initial presentation [
5]. However, approximately one third of patients have a chronic clinical course and require additional treatment [
5].
Several measures have been proposed for the characterization of PANS patients [
6‐
8]. The wide range of symptoms represented within the PANS construct constitutes a challenge for both daily clinic work and for the design of clinical trials [
4]. At least two clinician-rated instruments or symptom checklists have been specifically developed for PANS, but their administration is time-consuming, their items cannot be easily collated to calculate total scores, and their psychometric properties have not been established [
9,
10]. The use of gold standard measures for specific psychiatric symptoms, such as the Children’s Yale-Brown Obsessive Compulsive Scale (CY-BOCS) [
11] or the Yale Global Tic Severity Scale (YGTSS) [
12,
13] as single outcome measures is also problematic because they only capture specific symptom clusters and their use may result in both an over- or underestimation of treatment responders [
4,
14]. Conversely, trying to assess every individual symptom cluster may result in an overwhelming number of rating scales for the families to complete, reducing the quality of the collected data and the willingness to participate in research.
Previous experience from our PANS cohort [
3,
5] suggested a potential discrepancy between clinician-rated measures of global functioning, disease severity and improvement scales on the one hand, and subjective reports from parents on the other. It is possible that frequently used instruments may not capture the full extent of the patients’ difficulties [
5]. For some families this may result in frustration and a perceived lack of understanding of the true impact of PANS from the medical community.
In this study, we aimed to investigate the suitability of standard clinical measures for the assessment and follow-up of children with PANS, and formally assess the degree of agreement between multiple informants (child, parent and clinician ratings). Ultimately, we aim to shed some light on the optimal ways to measure the complex presentation of this patient group.
Results
Sample characteristics
Thirty-four out of 46 eligible PANS patients consented to participate in the follow-up study and provided data. Median age at follow-up was 11.5 years (range 6.7–17.1) and 19 (56%) of the participants were male. Further details on the clinical characteristics of the cohort, including duration of illness, comorbidities, current symptoms, family history of psychiatric and autoimmune disease can be found in a previous publication [
5].
Descriptive statistics
At a group level, median and mean scores for most global and specific symptom scales generally indicated low-to-moderate symptom severity and high level of functioning. However, there was substantial variability in the data, suggesting that some individuals experienced impaired symptoms at follow-up. Descriptive statistics for each of the measures are presented in Tables
1 and
2.
Table 1
Descriptive statistics of global severity and adaptive function rating scales at follow-up (n = 34)
CGI-Sa clinician | 34 | 3 (1–6) | 2.8 (1.4) |
CGI-S parent | 34 | 2 (0–5) | 2.4 (1.5) |
CGI-S child | 33 | 1 (0–6) | 2.2 (1.6) |
SDQ-Pb total | 34 | 9.5 (3–23) | 11.1 (6.4) |
SDQ-P emotional symptoms | 34 | 3.5 (0–10) | 4 (2.3) |
SDQ-P hyperactivity/inattention | 34 | 4.5 (0–10) | 4.1 (3.1) |
SDQ-P peer problems | 34 | 1 (0–6) | 1.6 (1.7) |
SDQ-P conduct problems | 34 | 1 (0–6) | 1.4 (1.6) |
SDQ-P prosocial behavior | 34 | 9 (4–10) | 8.2 (1.8) |
SDQ-Sc total | 22 | 10 (4–18) | 10.3 (4.5) |
SDQ-S emotional symptoms | 22 | 4 (0–10) | 3.8 (2.3) |
SDQ-S hyperactivity/inattention | 22 | 3.5 (0–8) | 3.9 (2.2) |
SDQ-S peer problems | 22 | 1.5 (0–6) | 1.5 (1.4) |
SDQ-S conduct problems | 22 | 1 (0–5) | 1.3 (1.2) |
SDQ-S prosocial behavior | 22 | 9 (6–10) | 8.7 (1.2) |
CGASd | 34 | 61 (28–80) | 60.9 (13.5) |
KIDSCREEN-10 parent total | 34 | 38.5 (25–47) | 37.5 (5.8) |
KIDSCREEN-10 parent general well-being | 34 | 3 (1–5) | 3.1 (1.1) |
KIDSCREEN-10 child total | 33 | 39 (24–46) | 38.3 (6.1) |
KIDSCREEN-10 child general well-being | 33 | 3 (1–5) | 3.3 (1) |
WSAS-Pe | 34 | 12.5 (0–30) | 12.9 (9.2) |
WSAS-Yf | 33 | 9 (0–31) | 10.8 (9.7) |
Table 2
Descriptive statistics of symptom-specific rating scales at follow-up (n = 34)
CY-BOCSa | 34 | | 8 (0–30) | 8.2 (8) |
YGTSSb | 34 | | 4.5 (0–65) | 10.4 (14) |
OCI-CVc | 33 | | 11 (0–18) | 9.5 (5.1) |
SMFQ-Pd | 34 | | 4.5 (0–14) | 5.9 (4.4) |
SMFQ-Ce | 31 | | 4 (1–19) | 5.5 (4.7) |
SAAI-Pf | 33 | | 10 (0–35) | 12.5 (10.9) |
SAAI-Cg | 33 | | 8 (0–36) | 10.6 (9.4) |
ISI-Ch | 33 | | 5 (0–18) | 5.9 (4.6) |
AQ-10i | 31 | | 3 (0–8) | 3 (2) |
SNAP-IVj | 34 | | 18.5 (0–57) | 21.9 (14.3) |
SNAP-IV inattention | 34 | | 1 (0–2.3) | 1.1 (0.7) |
SNAP-IV hypertactivity/impulsivity | 34 | | 0.4 (0–2.4) | 0.7 (0.7) |
SNAP-IV ODD | 34 | | 0.5 (0–2.1) | 0.7 (0.6) |
ECBIk intensity scale | 33 | | 90 (41–169) | 95.9 (34.5) |
ECBI problem scale | 34 | | 7 (0–25) | 7.9 (7.4) |
The overall ICC (1,3) of CGI-S ratings made by clinician, parent and child was 0.57 (95% CI 0.37–0.74). The ICC (1,2) of clinician and parent CGI-S ratings was 0.69 (0.46–0.83), the ICC (1,2) of clinician and child CGI-S ratings was 0.47 (95% CI 0.17–0.70) and the ICC (1,2) of parent and child CGI-S ratings was 0.55 (95% CI 0.26–0.75). Thus, the overall agreement between CGI-S ratings made by clinician, parent and child was only fair.
Regarding the SDQ-P/S for participants > 11 years (n = 22), the ICC (1,2) of parent and child ratings was 0.64 (95% CI 0.31–0.83), representing a moderately good agreement.
The ICC (1,2) of parent and child ratings of the KIDSCREEN-10 was 0.81 (95% CI 0.65–0.90) and of parent and child ratings of the WSAS-P/Y 0.82 (95% CI 0.66–0.91), both representing excellent agreements.
The ICC (1,2) of SMFQ-P/C ratings made by parent and child was 0.46 (95% CI 0.14–0.70), only representing a fair agreement. In contrast, the ICC (1,2) of SAAI-P/C ratings was 0.88 (95% CI 0.77–0.94), representing an excellent agreement.
Correlations between measures
Inter-correlations between clinician-rated CGAS and both parent-rated SDQ-P and child-rated SDQ-S were poor, including subscales. There was a fair association between clinician-rated CGI-S and parent-rated total SDQ (
ρ = 0.448,
p < 0.008), but not at a subscale level. The parent-rated functional scales KIDSCREEN-10 and WSAS-P had stronger inter-correlations than did the general symptom-severity scale SDQ-P with clinician-rated global symptom and functional scales. There was a moderate-to-strong association between parent-rated KIDSCREEN-10 and clinician-rated CGI-S (
ρ = − 0.663,
p < 0.001) and a very strong association between WSAS-P and clinician-rated CGI-S (
ρ = 0.811, p < 0.001). The same was true for the child-rated versions of the functional scales. See Tables
3 and
4 (subscales available in Tables
S1 and
S2).
Table 3
Spearman correlations between clinician-rated global symptom and functional scales and parent-rated KIDSCREEN-10, WSAS-P and SDQ-P and child-rated functional scales KIDSCREEN-10, WSAS-Y and SDQ-S, n = 34
CGASa | 1 | | | | | |
CGI-Sb clinician | −0.911 | 1 | | | | |
CGI-S parent | −0.555 | 0.664 | 1 | | | |
KIDSCREEN-10 parent | 0.567 | −0.663 | − 0.308 | 1 | | |
WSAS-Pc | −0.669 | 0.811 | 0.660 | −0.710 | 1 | |
SDQ-Pd | −0.374 | 0.448 | 0.423 | −0.564 | 0.528 | 1 |
Table 4
Spearman correlations between clinician-rated global symptom and functional scales and child-rated KIDSCREEN-10, WSAS-Y and SDQ-S, n = 22 (excludes participants < 11 years)
CGASa | 1 | | | | | |
CGI-Sb clinician | −0.929 | 1 | | | | |
CGI-S child | −0.511 | 0.473 | 1 | | | |
KIDSCREEN-10 child | 0.627 | −0.626 | − 0.257 | 1 | | |
WSAS-Yc | −0.570 | 0.640 | 0.315 | −0.569 | 1 | |
SDQ-Sd | −0.304 | 0.215 | 0.344 | −0.195 | 0.196 | 1 |
Clinician-rated global symptom and functional scales had moderate associations with clinician-scored CY-BOCS but not to child-rated OCI-CV. There was only a fair association between CY-BOCS and OCI-CV (ρ = 0.468,
p < 0.006). See Table
S3.
There was a fair-to-moderate association between clinician-rated CGAS and CGI-S and both parent- and child-rated SMFQ-P/C. Symptoms of separation anxiety and sleep disorder were more uncommon in the cohort, and inter-correlations between the symptom-specific measures SAAI-C and ISI-C and the global symptom and functional scales were lower. See Table
5.
Table 5
Spearman correlations between clinician-rated global and functional scales and SMFQ-P/C, SAAI-P/C and ISI-C, n = 30
CGASa | 1 | | | | | | |
CGI-Sb clinician | −0.904 | 1 | | | | | |
SMFQ-Pc | −0.447 | 0.516 | 1 | | | | |
SMFQ-C | −0.582 | 0.457 | 0.527 | 1 | | | |
SAAI-Pd | −0.362 | 0.485 | 0.461 | 0.146 | 1 | | |
SAAI-C | −0.241 | 0.364 | 0.258 | 0.871 | 0.125 | 1 | |
ISI-Ce | −0.198 | 0.297 | 0.362 | 0.297 | 0.225 | 0.253 | 1 |
There was a fair-to-moderate association between CGAS and CGI-S and AQ-10, ECBI and SNAP-IV, specifically on the SNAP-IV inattention subscale. ECBI had a very strong association to SNAP-IV, highest on the SNAP-IV total but also on hyperactivity and conduct subscales, as expected. See Table
6.
Table 6
Spearman correlations between clinician-rated global and functional scales and AQ-10, SNAP-IV and ECBI, n = 30
CGASa | 1 | | | | | | | | |
CGI-Sb clinician | −0.904 | 1 | | | | | | | |
AQ-10c | −0.421 | 0.558 | 1 | | | | | | |
SNAP-IVd | −0.398 | 0.511 | 0.330 | 1 | | | | | |
SNAP-IV inattention | −0.463 | 0.562 | 0.289 | 0.813 | 1 | | | | |
SNAP-IV hyper | −0.238 | 0.300 | 0.220 | 0.916 | 0.652 | 1 | | | |
SNAP-IV conduct | −0.261 | 0.300 | 0.240 | 0.760 | 0.391 | 0.668 | 1 | | |
ECBIe | −0.331 | 0.432 | 0.378 | 0.874 | 0.660 | 0.809 | 0.745 | 1 | |
ECBI problem | −0.373 | 0.503 | 0.584 | 0.813 | 0.614 | 0.720 | 0.710 | 0.937 | 1 |
In summary, the global clinician-rated symptom and functional measures tended to have stronger inter-correlations with parent- and child-rated functional measures than with symptom-specific measures.
Discussion
In this study we analyzed data from a PANS cohort that had been followed-up for 2–5 years after initial presentation [
5]. We examined the correspondence between clinician, parent and child measures of global symptom severity, adaptive functioning and specific psychiatric symptoms. This is critical because it is still unclear how to best measure the complex symptom presentation of the syndrome. Using the appropriate outcome measures has important implications for both clinical practice and the design of clinical trials.
Overall, median ratings for measures assessing global symptom severity and adaptive functioning indicated low symptom burden and a rather high level of everyday functioning in our sample. However, there was a large variability in the data, particularly in the symptom-specific measures, reflecting the heterogeneity of symptom presentations and clinical courses that are characteristic of the syndrome. These findings confirm and extend the findings of our previous study on the same cohort [
5]. Specifically, we had previously reported that approximately one third of participants in the follow-up study had clinically significant symptoms and required additional treatment.
CGI-S is a gold standard measure of psychiatric illness severity, most frequently assessed by the clinician. The modest agreement between CGI-S ratings across informants in our sample suggests that it may be helpful complementing the clinician rating with ratings made by the parent and child. Overall, agreement between ratings made by parent and child were excellent for functional scales, but only fair-to-moderate for global symptom severity and symptom-specific scales, with the exception of the separation anxiety measure, which had excellent agreement between informants.
As expected, because of the previously mentioned heterogeneity of symptom presentations, global clinician-rated symptom and functional measures tended to have stronger inter-correlations with parent- and child-rated functional measures than with symptom-specific ones.
Parent- and child-rated functional scales KIDSCREEN-10 and WSAS-P/Y correlated well with clinician-rated global symptom and functional scales, and the agreements between ratings made by parent and child were excellent. Despite both scales being useful in the study, KIDSCREEN-10 may be more easily accessible to a younger patient group. Multiple parents commented on the suitability of the WSAS-P/Y, and indicated that its items may be less suitable for the younger patients. By contrast, the KIDSCREEN-10 was perceived as simple and straightforward for both younger children and teenagers.
Somewhat surprisingly, SDQ-P/S seemed to have weak inter-correlations with global symptom and functional scales and the agreement between informants was only moderate. This may be due to SDQ-P/S being more symptom-oriented than the other global measures used in the study. Results suggest that it may be less clinically useful in this particular patient group, but it should also be noted that the SDQ impact supplement was not used and therefore is not included in our analyses.
Obsessive-compulsive symptoms are part of main PANS criteria but were surprisingly rare in our sample at follow-up. Assessing obsessive-compulsive symptoms with a self-rated scale as a complement to CY-BOCS may not add a lot of information within the PANS patient group at follow-up. At onset, OCD is generally a more pervasive part of the symptom presentation, and more time should be devoted to a comprehensive OCD assessment.
We did not include a measure specifically measuring eating disorder symptoms because, in our clinical experience, these tend to be OCD-related and without the defining features of a typical eating disorder such as fears of gaining weight (resembling avoidant/restrictive food avoidance disorder). For selected patients with eating difficulties, it may be useful to measure these symptoms in order to track their improvement.
The agreement between SMFQ-P/C ratings made by parent and child was only fair, indicating the importance of having both parent- and child ratings of depression and thus avoid underestimating these symptoms in PANS patients. Conversely, the agreement between SAAI-P/C ratings made by parent and patient was excellent, suggesting that either parent or child ratings may suffice for clinical purposes.
Previous longitudinal data have shown a high comorbidity with neuropsychiatric disorders, combined with intensification of related symptoms during PANS flares [
5,
45]. We therefore recommend measures that can screen for, and assess the severity of, autistic behaviors, inattention, hyperactivity and conduct problems when following up PANS. In our sample ECBI had a very strong association to SNAP-IV, suggesting that the simpler SNAP-IV may sufficiently cover the patients’ oppositional behaviors for clinical purposes. When detecting potentially severe oppositional defiant behaviors, ECBI can be used as a complement.
Our study has some limitations. First, we analyzed data from a small sample of patients from a single clinic. Second, the age range was such that the results of the child-rated measures should be interpreted with caution; despite our efforts, it is possible that the younger children received help from their parents to fill in their questionnaires. Third, we were not able to calculate internal consistency of the scales included in this study (only total scores were available). We could thus not examine the psychometric properties of the scales in this particular sample. Future studies would benefit from conducting such psychometric analyses.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.