- Email: [email protected]
The New Zealand minds for minds autism spectrum disorder self-reported cohort
Research in Autism Spectrum Disorders 36 (2017) 1–7
Contents lists available at ScienceDirect
Research in Autism Spectrum Disorders journal homepage: http://ees.elsevier.com/RASD/default.asp
The New Zealand minds for minds autism spectrum disorder self-reported cohort Javier Virues-Ortegaa , Klaus Lehnertb , Brendan Swanb , Michael W. Taylorb , Adrienne Southeec , Dane Douganc , Juliet Taylord, Rosamund Hille, Russell G. Snellb,* , Jessie C. Jacobsenb a
School of Psychology, The University of Auckland, New Zealand Centre for Brain Research and School of Biological Sciences, The University of Auckland, New Zealand Autism New Zealand, Inc, New Zealand d Genetic Health Service New Zealand, Auckland City Hospital, New Zealand e Department of Neurology, Auckland City Hospital, New Zealand b c
A R T I C L E I N F O
A B S T R A C T
Article history: Received 16 May 2016 Received in revised form 9 December 2016 Accepted 19 December 2016 Number of reviews completed is 1 Available online xxx
Background: To improve our understanding of autism spectrum disorder (ASD) in New Zealand, a multi-disciplinary research network, Minds for Minds, was created. This network has established a cohort of self- and proxy-reported individuals and their family members with ASD in New Zealand. The aim of this manuscript is to present the New Zealand Minds for Minds Autism Spectrum Disorder Self-Reported Cohort, M4M cohort for short, and to provide preliminary insights into the trends of ASD in New Zealand through the analysis of diagnostic and sociodemographic information of 972 members (ages 2–83) of this cohort, the majority of which were carer-reported. Method: The participants were recruited via an internet-based questionnaire, and social network analysis was used to visually analyse the mutual interactions of the cohort. Results: We observed the well-reported gender bias and an ethnic structure that reflects New Zealand’s most recent census. Comorbidity patterns were consistent with epidemiological literature: anxiety disorders, depression and epilepsy were highly prevalent amongst individuals with ASD and their families. Conclusions: This is the first national large-scale ASD research cohort, which contains an ethnic composition unique to the country. It is anticipated that the multi-disciplinary research approach of this cohort will help inform health policies in New Zealand and contribute to the international effort to better understand ASD. © 2017 Elsevier Ltd. All rights reserved.
Keywords: Asd M4M cohort New Zealand Comorbidity
1. The New Zealand minds for minds autism spectrum disorder self-reported cohort The prevalence of autism spectrum disorder (ASD) is currently estimated to be 1 in 68 children in the general population ([41_TD$IF] Christensen et al., 2012). The epidemiology of ASD and other neurodevelopmental disorders is not well studied in New Zealand. The centralised health care system in New Zealand means the majority of ASD diagnoses are made by relatively few
* Corresponding author. E-mail address: [email protected] (R.G. Snell). http://dx.doi.org/10.1016/j.rasd.2016.12.003 1750-9467/© 2017 Elsevier Ltd. All rights reserved.
2
J. Virues-Ortega et al. / Research in Autism Spectrum Disorders 36 (2017) 1–7
clinicians, which could improve the consistency and comparability of ASD diagnoses. The New Zealand population is therefore in a unique position to make a significant contribution to the international effort to better define ASD. In 2013, a New Zealand-based research network, Minds for Minds, commenced collecting self- and proxy-reported diagnostic information relating to ASD and other neurological disorders. These data conformed the New Zealand Minds for Minds ASD Self-reported research cohort, M4M cohort for short, with the primary purpose of identifying appropriate individuals and families willing to be involved in specific research projects, with the aim to enable multidisciplinary studies. The aim of this manuscript is to present the characterisation of ASD in New Zealand by way of a series of sociodemographic and family comorbidity descriptive and social network analyses of the M4M cohort. We anticipate the findings resulting from the M4M cohort will inform educational, social and health policies in New Zealand. 2. Methods 2.1. Internet-Based questionnaire Interest to participate in the M4M cohort is collected via an HTML interface (www.mindsforminds.org.nz). Participation is encouraged via public media outlets and through clinical collaboration. Only name, e-mail address, and condition are required for successful registration. Individuals can then choose to enter further details. Individuals either proxy-reported or selfreported their diagnostic status through a four-choice itemwhich included ASD, Asperger syndrome, other neurodevelopmental disability, or not diagnosed. Participants selecting ASD or Asperger syndrome were included in the current analysis. As an additional means to establish the credibility of the diagnostic status, participants were asked about the age at diagnosis and the name of the specialist who diagnosed the condition. The process of formally confirming diagnostic status and consenting the individuals in the cohort for genetic, public health, and behavioural research is on-going. The registration period remains open and individuals from all areas of New Zealand are eligible to register. Access is restricted to Principal Investigators. (Ethics approval Northern B Health and Disability Ethics Committee: 12/NTB/59). Following registration of interest, all individuals are formally consented for relevant research programmes through direct contact. 2.2. Descriptive analysis Sociodemographic and clinical characteristics included: (1) date of birth and age, (2) gender, (3) reporting source (selfreported vs. proxy-reported), (4) target diagnosis (ASD, Asperger syndrome, other neurodevelopmental conditions), (5) age at the time the target diagnosis was made, (6) comorbidity of the proband, and (7) self-reported comorbidity of first-degree relatives (parent, siblings, children). Only individuals reporting the seven variables above were included in the descriptive analysis. 2.3. Social network analysis Social network graphing can help to visually analyse the mutual interactions of an array of co-occurring conditions in a group of individuals (Cramer, Waldorp, van der Maas, & Borsboom, 2010). We used force-directed social network graphs (Fruchterman & Reingold, 1991) with the following attributes: (1) Bidimensional space. Diagnostic entities (nodes) and individual instances of comorbidity (connecting lines or connections) were represented over a bi-dimensional space with a tendency to become circular as the complexity of the network (number of nodes and connecting lines) increases. (2) Minimal distance. A node’s location within the network results from minimizing the distances between all connected nodes. The length of a particular connecting line is the function of the number of times that the two nodes (diagnostic conditions) connected by the line co-occur. For example, in the event that dyspraxia co-occurs with depression twice as often as it does with hyperactivity, the distance between dyspraxia and depression will be half as long as that between dyspraxia and hyperactivity. (3) Distribution of nodes. Conditions that tend to co-occur with multiple other conditions (which may themselves be related or unrelated) will tend to acquire relatively central locations within the wider network. By contrast, conditions that rarely co-occur with others will acquire peripheral positions in the network. (4) Degree. Number of times that a given diagnostic entity co-occurs with any other diagnosis. For example, if dyspraxia co-occurs with epilepsy in five individuals, with depression in eight individuals, and with gastrointestinal disorders in 10 individuals, the degree of dyspraxia would be 23. Force-directed graphs minimize the distances between those diagnoses that co-occur frequently. As a result, it is possible to determine if a particular diagnosis operates as a comorbidity “hub” within the network or unrelated to other diagnoses. If a target diagnosis frequently occurs in the shortest paths that connect a given pair of comorbidities, the target condition would be a greater contributor to the comorbidity network structure. Such nodes are referred to as having a relatively high level of centrality. Eigenvector centrality estimates the influence of a node based on the number of connections received and on the degree of the connecting nodes (Hansen, Shneiderman, & Smith, 2011). We used the Girvan and Newman (2002) cluster analysis method for small networks to identify clusters or conditions with a tendency to correlate with one another. In order to provide an indication of the comprehensiveness of clusters in the
J. Virues-Ortega et al. / Research in Autism Spectrum Disorders 36 (2017) 1–7
3
comorbidity network, we computed the percentage of connections in the network that are unique to the cluster. Specifically, if cluster A is composed of three nodes and 10 connections, while the complete network is composed of 10 nodes and 100 connections, cluster A would contain 10% of the connections in the network. All social network analyses were conducted with NodeXL v. 1.0.1.167 (NodeXL team, 2011). 3. Results The data reported here describe 972 individuals who submitted online questionnaires between May 2013 and February 2015. Among these, a total of 754 cases recorded information for gender, age, ethnicity, source (self-reported, proxyreported), and target diagnosis (ASD, Asperger syndrome, or other neurodevelopmental disorder), hereinafter referred to as the cohort. The cohort included 490 individuals with ASD, 183 diagnosed with Asperger syndrome, and 81 with other neurodevelopmental disorders. The cohort was composed of individuals of all ages (mean age SD: 16.1 12.2, range: 2– 83): 116 individuals were less than 8 years of age, 422 were preadolescents and adolescents between 8 and 17 years of age, and 216 were adults (18 years and older). The cohort was composed of 595 males (78.9%) and 159 females (21.1%). The majority of individuals were of European descent (93.6%, n = 706). There were also 127 individuals with Maori and Pacific Island ancestry (15.3%, n = 115) and a fraction of probands were of other ancestry (n = 30, 4.0%). Of all individuals 19.4% reported mixed ancestry (n = 146). The questionnaire was completed by a proxy in 87.4% of cases (n = 659) and by a proband in 12.6% of cases (n = 95). Not surprisingly, 90.5% (n = 86) of self-reported cases were adults. Only 24 cases with ASD were selfreported (4.9%). The fraction of self-reporting was higher in Asperger syndrome (28.4%, n = 52), and individuals with Asperger syndrome comprised the majority of self-reported cases in the cohort (68.4%). All individuals contributed more information than the minimum required. Most individuals that omitted key information including age, gender, and ethnicity were self-reported cases (74.4%, n = 163). Table 1 presents detailed descriptive information. 3.1. Comorbidity across age groups A total of 471 individuals (48.5%) reported at least one comorbid condition. Individuals of up to 7 years of age (n = 116) showed comorbidity with nine distinct conditions. The most common conditions for this age group were attention deficit and hyperactivity disorder (ADHD) (27 individuals), dyspraxia (23), and depression and anxiety (20). By contrast, individuals within 8–17 years of age showed comorbidity in 14 distinct conditions. Emerging comorbidities observed in more than one participant included migraine, bipolar disorder, and Tourette syndrome (Fig. 1). The most common conditions for this age group were depression and anxiety (90), ADHD (83), and dyspraxia (50). Finally, adults aged 18 years or more showed a level of comorbidity consistent with the preceding age group: 13 distinct conditions. The most common conditions for adults were depression and anxiety (116), gastrointestinal symptoms (42), and ADHD (37). The social network revealed a growing level of complexity in the comorbidity patterns with increasing age among all individuals in the M4M cohort (Fig. 1). The number of co-occurrences of two comorbid conditions within the same individual (i.e., edges of the social network) increased across age groups (Table 2). The number of conditions per individual also increased with age: children showed 43 edges (0.17 per individual), adolescents 222 (0.60 per individual), and adults 266 (1.03 per individual). The increased network complexity among adults is apparent in Fig. 1. The cluster analysis revealed that a single cluster accounted for 93%, 92%, and 73% of all instances of comorbidity in the analysis among the first, second, and third age groups respectively. ADHD, depression and anxiety, and dyspraxia were part of the core of this cluster across all three age groups (Fig. 1). The conditions with greater eigenvector centrality across age groups were: depression and anxiety (0.17), dyspraxia (0.16), and ADHD (0.15) for children; epilepsy (0.13), depression and anxiety (0.12), and gastrointestinal
Table 1 Study sample characteristics by target diagnosis and age group. ASD
Gender Male Female Ethnicity European Maori & Pacific Other Mixed Registrands Proxy Proband
Asperger Syndrome
Other
All
0–7
8–17
18
0–7
8–17
18
0–7
8–17
18
0–7
8–17
18
87 16
256 38
73 20
5 0
70 16
54 38
6 2
28 14
16 15
98 18
354 68
143 73
95 26 4 95
276 45 12 276
80 13 8 80
4 3 0 4
86 8 0 86
85 9 5 85
8 4 0 8
41 6 1 41
31 1 0 31
107 33 4 107
403 59 13 403
196 23 13 196
101 2
291 3
74 19
5 0
85 1
41 51
7 1
40 2
15 16
113 3
416 6
130 86
Notes. Cases with missing values excluded. The category European included individuals that were either European or part European. The category Maori & Pacific included individuals that were either Maori & Pacific or part Maori & Pacific.
4
J. Virues-Ortega et al. / Research in Autism Spectrum Disorders 36 (2017) 1–7
[(Fig._1)TD$IG]
Fig. 1. Comorbidity network of all individuals in the database aged 0–7 (n = 248), 8–17 (n = 372), and above 18 years of age (n = 258). Conditions within the same cluster are highlighted in blue. ADD = Attention deficit and hyperactivity disorder; BIP = Bipolar disorder; DEA = Depression and anxiety disorders; DLX = Dyslexia; DPR = Dyspraxia; DWN = Down syndrome; EPI = Epilepsy; FXS = Fragile X syndrome; GAS = Gastrointestinal disorder; MGR = Migraine; OCD = Obsessive-compulsive disorder; OTH = Other neurological disorders; PWS = Prader-Willi syndrome; RET = Rett syndrome; TOU = Tourette syndrome; TUR = Turner syndrome. Probands were not excluded due to missing data.
Table 2 Comorbidity: social network structure across age groups. Age (years)
0–7 8–17 18
n
248 372 258
Overall network
Main Cluster
Conditions
Edges
Conditions
Edges
9 14 13
43 222 266
7 9 9
40 205 195
symptoms (0.12) for adolescents; and Tourette syndrome (0.1), gastrointestinal symptoms (0.1), and depression and anxiety (0.1) for adults. 3.2. Comorbidities of first-Degree relatives A total of 894 (91.2%) respondents reported one or more conditions among the probands’ first-degree relatives. The conditions most frequently reported among the relatives of respondents with ASD were depression and anxiety (108 relatives with the condition), ASD (71), migraine (47), and Asperger syndrome (45) (Fig. 2). In total, 13 distinct conditions were reportedly present in one or more relatives. The cluster analysis revealed that all conditions with the exceptions of migraine, gastrointestinal disorder, and epilepsy could be grouped in a single cluster accounting for 80.0% of the connections in the network. The conditions that co-occurred more frequently across all relatives of respondents with ASD were ADHD with depression and anxiety (18), ASD with depression and anxiety (15), Asperger syndrome with depression and anxiety (15), and dyspraxia with depression and anxiety (15). The conditions with the greatest eigenvector centrality within the network of respondent’s relatives with ASD were obsessive-compulsive disorder (OCD; 0.100), dyslexia (0.097), and dyspraxia (0.097). A point of departure of the first-degree relatives comorbidity network is the greater centrality and prevalence of OCD and dyslexia. OCD was reported less often among probands. Interestingly, on occasions when OCD was
J. Virues-Ortega et al. / Research in Autism Spectrum Disorders 36 (2017) 1–7
[(Fig._2)TD$IG]
5
Fig. 2. Comorbidity network of first-degree relatives of probands with autism spectrum disorder (n = 518) and Asperger syndrome (n = 201), and across all individuals in the database (972). Probands were not excluded due to missing data. Graph conventions are the same as for Fig. 1.
reported among probands, other conditions were unlikely to be reported, which explains the minimal centrality of OCD within the network (Fig. 1). The conditions more frequently reported among the respondent’s relatives with Asperger syndrome were depression and anxiety (48), Asperger syndrome (46), ASD (18), and dyslexia (23). While the presence of clinical conditions among firstdegree relatives of probands with Asperger syndrome is lower than that for relatives of probands with ASD, the most prominent conditions largely coincided (Fig. 2). A total of 13 distinct conditions were present among one or more relatives. A single cluster accounting for 95.6% of all connections in the network amalgamated all conditions with the exception of gastrointestinal disorder, migraine and Prader-Willi syndrome. The comorbidity network of relatives of probands with Asperger syndrome was consistent with that of the relatives of probands with ASD. The structure of the network was slightly less peripheral than the one found for relatives of probands with ASD, as suggested by the relatively higher eigenvector centrality values. The conditions with the greatest eigenvector centrality were ASD (0.104), Asperger syndrome (0.103), depression and anxiety (0.103) and OCD (0.102) (Table 3). 4. Discussion The general comorbidity pattern reported here is consistent with evidence from the epidemiological literature. Depression and anxiety are highly prevalent comorbid conditions in individuals with ASD and their families (Matson & Williams, 2014). Overall, our findings are consistent with the psychiatric comorbidity found in register-based studies (Meier et al., 2015).
Table 3 First degree relatives comorbidity: social network structure. n
ASD ASP All
502 195 916
Overall network
Main Cluster
Conditions
Edges
Conditions
Edges
13 13 15
256 204 676
9 10 10
204 194 608
6
J. Virues-Ortega et al. / Research in Autism Spectrum Disorders 36 (2017) 1–7
The co-occurrence of epilepsy among individuals with ASD is well established in the literature, and the patterns of comorbidity reported here are consistent with these findings (e.g., Besag, 2015). The proposed cluster analysis had some construct validity in that there was a degree of differentiation between psychopathological conditions (e.g. depression and anxiety, ADHD) and other conditions (e.g. epilepsy, gastrointestinal). A higher-than-normal prevalence of OCD among close relatives of individuals with ASD is also well documented in the literature (Meier et al., 2015). Initial recruitment relied on self- and proxy-reported data. Recent studies have indicated the reliability of self- and proxyreporting, demonstrating recruitment of families with reliable and valid diagnoses that correlate with clinical reports (Daniels et al., 2011; Lee et al., 2010; [29_TD$IF]Warnell et al., 2015). Moreover, the overall ethnic structure of the cohort reflects New Zealand’s most recent census data ([30_TD$IF]Statistics New Zealand, 2013) suggesting that the sample was not geographically biased. Also, the sex ratio reflects the widely reported male-to-female bias of ASD and Asperger syndrome. Interestingly, we have fewer than expected registrations from individuals for whom ASD is not their primary diagnosis, such as Tuberous Sclerosis and Rett syndrome, which suggests that a more targeted recruitment may be necessary to reach these clinical populations. Other potential limitations are the restricted number of categories that participants could select in the process of completing an expression of interest, which may mean that unknown or low-prevalence comorbid conditions were underreported. However, this risk is partially mitigated by a free-text field that allows the registrand to record any condition. The capability of large research cohorts to advance understanding of ASD has been demonstrated through large international ASD initiatives around the world (reviewed by [29_TD$IF]Warnell et al., 2015). Despite the high prevalence, little is known about the epidemiology of ASD in New Zealand. The self-reported data mirror sociodemographic and comorbidity patterns of international cohorts, suggesting that the epidemiology of ASD in New Zealand is similar to that described for European populations. We anticipate that this will contribute to health decision-makers in New Zealand, advising key clinical needs for this population as a whole (e.g., high incidence of depression and anxiety), and also within specific ethnic groups. We also foresee further academic research on this cohort, for instance the genetic analysis, will advise diagnostic and clinical management protocols, thereby improving the speed and accuracy for which a result is returned to a family. Also, the cohort could provide the basis for outcome research studies evaluating the effects of various services and interventions, including early intensive behavioural intervention. Given the unique ethnic composition of New Zealand (i.e., Maori and Pacific) and the centralised health-care system, we anticipate that the subsequent research projects resulting from this cohort will encourage the development of a formalised national database and epidemiological study of ASD in New Zealand and contribute to the international understanding of the condition. Conflict of interest The authors declare that they have no conflict of interest.[31_TD$IF] Acknowledgements We would like to thank all the families involved in the research, our four founding community organizations; Autism New Zealand, Children’s Autism Foundation, Autism Intervention Trust, Altogether Autism and the Minds for Minds charitable trust. JCJ is supported by a Rutherford Discovery Fellowship from Government funding, administered by the Royal Society of New Zealand. The research was funded by the Minds for Minds charitable Trust, the Neurological Foundation of New Zealand, the Oakley Mental Health Research Foundation,The University of Auckland, New Zealand Federation of Women’s Institutes and the Julius Brendel Trust. References Besag, F. M. (2015). Current controversies in the relationships between autism and epilepsy. Epilepsy & Behavior, 47, 143–146. http://dx.doi.org/10.1016/j. yebeh.2015.05.032. Christensen, D. L., Baio, J., & Braun, K. V., et al. (2012). Prevalence and Characteristics of Autism Spectrum Disorder [34_TD$IF]Among Children Aged 8 Years — Autism and Developmental Disabilities Monitoring Network, [35_TD$IF]11 Sites, United States, MMWR Surveill Summ 2016, 65 (No. SS-3) (No. SS-3) 1–23. DOI: http://dx. doi.org/10.15585/mmwr.ss6503a1. Cramer, A. O., Waldorp, L. J., & van der Maas, H. L. (2010). Comorbidity: A network perspective. Behavior and Brain Sciences, 33, 137–150. http://dx.doi.org/ 10.1017/S0140525x09991567. Daniels, A. M., Rosenberg, R. E., Law, J. K., Lord, C., Kaufmann, W. E., & Law, P. A. (2011). Stability of initial autism spectrum disorder diagnoses in community settings. Journal of Autism and Developmental Disorders, 41, 110–121. http://dx.doi.org/10.1007/s10803-010-1031-x. Fruchterman, T. M. J., & Reingold, E. M. (1991). Graph drawing by force-directed placement. Software: Practice and Experience, 21, 1129–1164. http://dx.doi. org/10.1002/spe.4380211102. Girvan, M., & Newman, M. E. J. (2002). Community structure in social and biological networks. Proceedings of the National Academy of Sciences of the United States of America, 99, 7821–7826. http://dx.doi.org/10.1073/pnas.122653799. Hansen, D., Shneiderman, B., & Smith, M. A. (2011). Analysing social media networks with NodeXL. Burlington, MA: Elsevier. Lee, H., Marvin, A. R., Watson, T., Piggot, J., Law, J. K., Law, P. A., et al. (2010). Accuracy of phenotyping of autistic children based on Internet implemented parent report. American Journal of Medical Genetics Part B: Neuropsychiatric Genetics, 153b, 1119–1126. http://dx.doi.org/10.1002/ajmg.b.31103. Matson, J. L., & Williams, L. W. (2014). Depression and mood disorders among persons with autism spectrum disorders. Research in Developmental Disabilities, 35, 2003–2007. http://dx.doi.org/10.1016/j.ridd.2014.04.020.Epub 2014 May 24. Meier, S. M., Petersen, L., Schendel, D. E., Mattheisen, M., Mortensen, P. B., & Mors, O. (2015). Obsessive-Compulsive disorder and autism spectrum disorders: Longitudinal and offspring risk. PLoS One, 10, e0141703. http://dx.doi.org/10.1371/journal.pone.0141703.
J. Virues-Ortega et al. / Research in Autism Spectrum Disorders 36 (2017) 1–7
7
NodeXL team, Mircrosoft Corporation (2011). NodeXL: Network overview, discovery and exploration for Excel, version 1.0.1.167. Retrieved from http://www. codeplex.com/nodexl. Statistics New Zealand [39_TD$IF](2013). 2013 Census. Retrieved from http://www.stats.govt.nz/Census/2013-census.aspx. Warnell, F., George, B., McConachie, H., Johnson, M., Hardy, R., & Parr, J. R. (2015). Designing and recruiting to UK autism spectrum disorder [40_TD$IF]research databases: do they include representative children with valid ASD diagnoses? British Medical Journal, 5(9) . http://dx.doi.org/10.1136/bmjopen-2015008625.
Contents lists available at ScienceDirect
Research in Autism Spectrum Disorders journal homepage: http://ees.elsevier.com/RASD/default.asp
The New Zealand minds for minds autism spectrum disorder self-reported cohort Javier Virues-Ortegaa , Klaus Lehnertb , Brendan Swanb , Michael W. Taylorb , Adrienne Southeec , Dane Douganc , Juliet Taylord, Rosamund Hille, Russell G. Snellb,* , Jessie C. Jacobsenb a
School of Psychology, The University of Auckland, New Zealand Centre for Brain Research and School of Biological Sciences, The University of Auckland, New Zealand Autism New Zealand, Inc, New Zealand d Genetic Health Service New Zealand, Auckland City Hospital, New Zealand e Department of Neurology, Auckland City Hospital, New Zealand b c
A R T I C L E I N F O
A B S T R A C T
Article history: Received 16 May 2016 Received in revised form 9 December 2016 Accepted 19 December 2016 Number of reviews completed is 1 Available online xxx
Background: To improve our understanding of autism spectrum disorder (ASD) in New Zealand, a multi-disciplinary research network, Minds for Minds, was created. This network has established a cohort of self- and proxy-reported individuals and their family members with ASD in New Zealand. The aim of this manuscript is to present the New Zealand Minds for Minds Autism Spectrum Disorder Self-Reported Cohort, M4M cohort for short, and to provide preliminary insights into the trends of ASD in New Zealand through the analysis of diagnostic and sociodemographic information of 972 members (ages 2–83) of this cohort, the majority of which were carer-reported. Method: The participants were recruited via an internet-based questionnaire, and social network analysis was used to visually analyse the mutual interactions of the cohort. Results: We observed the well-reported gender bias and an ethnic structure that reflects New Zealand’s most recent census. Comorbidity patterns were consistent with epidemiological literature: anxiety disorders, depression and epilepsy were highly prevalent amongst individuals with ASD and their families. Conclusions: This is the first national large-scale ASD research cohort, which contains an ethnic composition unique to the country. It is anticipated that the multi-disciplinary research approach of this cohort will help inform health policies in New Zealand and contribute to the international effort to better understand ASD. © 2017 Elsevier Ltd. All rights reserved.
Keywords: Asd M4M cohort New Zealand Comorbidity
1. The New Zealand minds for minds autism spectrum disorder self-reported cohort The prevalence of autism spectrum disorder (ASD) is currently estimated to be 1 in 68 children in the general population ([41_TD$IF] Christensen et al., 2012). The epidemiology of ASD and other neurodevelopmental disorders is not well studied in New Zealand. The centralised health care system in New Zealand means the majority of ASD diagnoses are made by relatively few
* Corresponding author. E-mail address: [email protected] (R.G. Snell). http://dx.doi.org/10.1016/j.rasd.2016.12.003 1750-9467/© 2017 Elsevier Ltd. All rights reserved.
2
J. Virues-Ortega et al. / Research in Autism Spectrum Disorders 36 (2017) 1–7
clinicians, which could improve the consistency and comparability of ASD diagnoses. The New Zealand population is therefore in a unique position to make a significant contribution to the international effort to better define ASD. In 2013, a New Zealand-based research network, Minds for Minds, commenced collecting self- and proxy-reported diagnostic information relating to ASD and other neurological disorders. These data conformed the New Zealand Minds for Minds ASD Self-reported research cohort, M4M cohort for short, with the primary purpose of identifying appropriate individuals and families willing to be involved in specific research projects, with the aim to enable multidisciplinary studies. The aim of this manuscript is to present the characterisation of ASD in New Zealand by way of a series of sociodemographic and family comorbidity descriptive and social network analyses of the M4M cohort. We anticipate the findings resulting from the M4M cohort will inform educational, social and health policies in New Zealand. 2. Methods 2.1. Internet-Based questionnaire Interest to participate in the M4M cohort is collected via an HTML interface (www.mindsforminds.org.nz). Participation is encouraged via public media outlets and through clinical collaboration. Only name, e-mail address, and condition are required for successful registration. Individuals can then choose to enter further details. Individuals either proxy-reported or selfreported their diagnostic status through a four-choice itemwhich included ASD, Asperger syndrome, other neurodevelopmental disability, or not diagnosed. Participants selecting ASD or Asperger syndrome were included in the current analysis. As an additional means to establish the credibility of the diagnostic status, participants were asked about the age at diagnosis and the name of the specialist who diagnosed the condition. The process of formally confirming diagnostic status and consenting the individuals in the cohort for genetic, public health, and behavioural research is on-going. The registration period remains open and individuals from all areas of New Zealand are eligible to register. Access is restricted to Principal Investigators. (Ethics approval Northern B Health and Disability Ethics Committee: 12/NTB/59). Following registration of interest, all individuals are formally consented for relevant research programmes through direct contact. 2.2. Descriptive analysis Sociodemographic and clinical characteristics included: (1) date of birth and age, (2) gender, (3) reporting source (selfreported vs. proxy-reported), (4) target diagnosis (ASD, Asperger syndrome, other neurodevelopmental conditions), (5) age at the time the target diagnosis was made, (6) comorbidity of the proband, and (7) self-reported comorbidity of first-degree relatives (parent, siblings, children). Only individuals reporting the seven variables above were included in the descriptive analysis. 2.3. Social network analysis Social network graphing can help to visually analyse the mutual interactions of an array of co-occurring conditions in a group of individuals (Cramer, Waldorp, van der Maas, & Borsboom, 2010). We used force-directed social network graphs (Fruchterman & Reingold, 1991) with the following attributes: (1) Bidimensional space. Diagnostic entities (nodes) and individual instances of comorbidity (connecting lines or connections) were represented over a bi-dimensional space with a tendency to become circular as the complexity of the network (number of nodes and connecting lines) increases. (2) Minimal distance. A node’s location within the network results from minimizing the distances between all connected nodes. The length of a particular connecting line is the function of the number of times that the two nodes (diagnostic conditions) connected by the line co-occur. For example, in the event that dyspraxia co-occurs with depression twice as often as it does with hyperactivity, the distance between dyspraxia and depression will be half as long as that between dyspraxia and hyperactivity. (3) Distribution of nodes. Conditions that tend to co-occur with multiple other conditions (which may themselves be related or unrelated) will tend to acquire relatively central locations within the wider network. By contrast, conditions that rarely co-occur with others will acquire peripheral positions in the network. (4) Degree. Number of times that a given diagnostic entity co-occurs with any other diagnosis. For example, if dyspraxia co-occurs with epilepsy in five individuals, with depression in eight individuals, and with gastrointestinal disorders in 10 individuals, the degree of dyspraxia would be 23. Force-directed graphs minimize the distances between those diagnoses that co-occur frequently. As a result, it is possible to determine if a particular diagnosis operates as a comorbidity “hub” within the network or unrelated to other diagnoses. If a target diagnosis frequently occurs in the shortest paths that connect a given pair of comorbidities, the target condition would be a greater contributor to the comorbidity network structure. Such nodes are referred to as having a relatively high level of centrality. Eigenvector centrality estimates the influence of a node based on the number of connections received and on the degree of the connecting nodes (Hansen, Shneiderman, & Smith, 2011). We used the Girvan and Newman (2002) cluster analysis method for small networks to identify clusters or conditions with a tendency to correlate with one another. In order to provide an indication of the comprehensiveness of clusters in the
J. Virues-Ortega et al. / Research in Autism Spectrum Disorders 36 (2017) 1–7
3
comorbidity network, we computed the percentage of connections in the network that are unique to the cluster. Specifically, if cluster A is composed of three nodes and 10 connections, while the complete network is composed of 10 nodes and 100 connections, cluster A would contain 10% of the connections in the network. All social network analyses were conducted with NodeXL v. 1.0.1.167 (NodeXL team, 2011). 3. Results The data reported here describe 972 individuals who submitted online questionnaires between May 2013 and February 2015. Among these, a total of 754 cases recorded information for gender, age, ethnicity, source (self-reported, proxyreported), and target diagnosis (ASD, Asperger syndrome, or other neurodevelopmental disorder), hereinafter referred to as the cohort. The cohort included 490 individuals with ASD, 183 diagnosed with Asperger syndrome, and 81 with other neurodevelopmental disorders. The cohort was composed of individuals of all ages (mean age SD: 16.1 12.2, range: 2– 83): 116 individuals were less than 8 years of age, 422 were preadolescents and adolescents between 8 and 17 years of age, and 216 were adults (18 years and older). The cohort was composed of 595 males (78.9%) and 159 females (21.1%). The majority of individuals were of European descent (93.6%, n = 706). There were also 127 individuals with Maori and Pacific Island ancestry (15.3%, n = 115) and a fraction of probands were of other ancestry (n = 30, 4.0%). Of all individuals 19.4% reported mixed ancestry (n = 146). The questionnaire was completed by a proxy in 87.4% of cases (n = 659) and by a proband in 12.6% of cases (n = 95). Not surprisingly, 90.5% (n = 86) of self-reported cases were adults. Only 24 cases with ASD were selfreported (4.9%). The fraction of self-reporting was higher in Asperger syndrome (28.4%, n = 52), and individuals with Asperger syndrome comprised the majority of self-reported cases in the cohort (68.4%). All individuals contributed more information than the minimum required. Most individuals that omitted key information including age, gender, and ethnicity were self-reported cases (74.4%, n = 163). Table 1 presents detailed descriptive information. 3.1. Comorbidity across age groups A total of 471 individuals (48.5%) reported at least one comorbid condition. Individuals of up to 7 years of age (n = 116) showed comorbidity with nine distinct conditions. The most common conditions for this age group were attention deficit and hyperactivity disorder (ADHD) (27 individuals), dyspraxia (23), and depression and anxiety (20). By contrast, individuals within 8–17 years of age showed comorbidity in 14 distinct conditions. Emerging comorbidities observed in more than one participant included migraine, bipolar disorder, and Tourette syndrome (Fig. 1). The most common conditions for this age group were depression and anxiety (90), ADHD (83), and dyspraxia (50). Finally, adults aged 18 years or more showed a level of comorbidity consistent with the preceding age group: 13 distinct conditions. The most common conditions for adults were depression and anxiety (116), gastrointestinal symptoms (42), and ADHD (37). The social network revealed a growing level of complexity in the comorbidity patterns with increasing age among all individuals in the M4M cohort (Fig. 1). The number of co-occurrences of two comorbid conditions within the same individual (i.e., edges of the social network) increased across age groups (Table 2). The number of conditions per individual also increased with age: children showed 43 edges (0.17 per individual), adolescents 222 (0.60 per individual), and adults 266 (1.03 per individual). The increased network complexity among adults is apparent in Fig. 1. The cluster analysis revealed that a single cluster accounted for 93%, 92%, and 73% of all instances of comorbidity in the analysis among the first, second, and third age groups respectively. ADHD, depression and anxiety, and dyspraxia were part of the core of this cluster across all three age groups (Fig. 1). The conditions with greater eigenvector centrality across age groups were: depression and anxiety (0.17), dyspraxia (0.16), and ADHD (0.15) for children; epilepsy (0.13), depression and anxiety (0.12), and gastrointestinal
Table 1 Study sample characteristics by target diagnosis and age group. ASD
Gender Male Female Ethnicity European Maori & Pacific Other Mixed Registrands Proxy Proband
Asperger Syndrome
Other
All
0–7
8–17
18
0–7
8–17
18
0–7
8–17
18
0–7
8–17
18
87 16
256 38
73 20
5 0
70 16
54 38
6 2
28 14
16 15
98 18
354 68
143 73
95 26 4 95
276 45 12 276
80 13 8 80
4 3 0 4
86 8 0 86
85 9 5 85
8 4 0 8
41 6 1 41
31 1 0 31
107 33 4 107
403 59 13 403
196 23 13 196
101 2
291 3
74 19
5 0
85 1
41 51
7 1
40 2
15 16
113 3
416 6
130 86
Notes. Cases with missing values excluded. The category European included individuals that were either European or part European. The category Maori & Pacific included individuals that were either Maori & Pacific or part Maori & Pacific.
4
J. Virues-Ortega et al. / Research in Autism Spectrum Disorders 36 (2017) 1–7
[(Fig._1)TD$IG]
Fig. 1. Comorbidity network of all individuals in the database aged 0–7 (n = 248), 8–17 (n = 372), and above 18 years of age (n = 258). Conditions within the same cluster are highlighted in blue. ADD = Attention deficit and hyperactivity disorder; BIP = Bipolar disorder; DEA = Depression and anxiety disorders; DLX = Dyslexia; DPR = Dyspraxia; DWN = Down syndrome; EPI = Epilepsy; FXS = Fragile X syndrome; GAS = Gastrointestinal disorder; MGR = Migraine; OCD = Obsessive-compulsive disorder; OTH = Other neurological disorders; PWS = Prader-Willi syndrome; RET = Rett syndrome; TOU = Tourette syndrome; TUR = Turner syndrome. Probands were not excluded due to missing data.
Table 2 Comorbidity: social network structure across age groups. Age (years)
0–7 8–17 18
n
248 372 258
Overall network
Main Cluster
Conditions
Edges
Conditions
Edges
9 14 13
43 222 266
7 9 9
40 205 195
symptoms (0.12) for adolescents; and Tourette syndrome (0.1), gastrointestinal symptoms (0.1), and depression and anxiety (0.1) for adults. 3.2. Comorbidities of first-Degree relatives A total of 894 (91.2%) respondents reported one or more conditions among the probands’ first-degree relatives. The conditions most frequently reported among the relatives of respondents with ASD were depression and anxiety (108 relatives with the condition), ASD (71), migraine (47), and Asperger syndrome (45) (Fig. 2). In total, 13 distinct conditions were reportedly present in one or more relatives. The cluster analysis revealed that all conditions with the exceptions of migraine, gastrointestinal disorder, and epilepsy could be grouped in a single cluster accounting for 80.0% of the connections in the network. The conditions that co-occurred more frequently across all relatives of respondents with ASD were ADHD with depression and anxiety (18), ASD with depression and anxiety (15), Asperger syndrome with depression and anxiety (15), and dyspraxia with depression and anxiety (15). The conditions with the greatest eigenvector centrality within the network of respondent’s relatives with ASD were obsessive-compulsive disorder (OCD; 0.100), dyslexia (0.097), and dyspraxia (0.097). A point of departure of the first-degree relatives comorbidity network is the greater centrality and prevalence of OCD and dyslexia. OCD was reported less often among probands. Interestingly, on occasions when OCD was
J. Virues-Ortega et al. / Research in Autism Spectrum Disorders 36 (2017) 1–7
[(Fig._2)TD$IG]
5
Fig. 2. Comorbidity network of first-degree relatives of probands with autism spectrum disorder (n = 518) and Asperger syndrome (n = 201), and across all individuals in the database (972). Probands were not excluded due to missing data. Graph conventions are the same as for Fig. 1.
reported among probands, other conditions were unlikely to be reported, which explains the minimal centrality of OCD within the network (Fig. 1). The conditions more frequently reported among the respondent’s relatives with Asperger syndrome were depression and anxiety (48), Asperger syndrome (46), ASD (18), and dyslexia (23). While the presence of clinical conditions among firstdegree relatives of probands with Asperger syndrome is lower than that for relatives of probands with ASD, the most prominent conditions largely coincided (Fig. 2). A total of 13 distinct conditions were present among one or more relatives. A single cluster accounting for 95.6% of all connections in the network amalgamated all conditions with the exception of gastrointestinal disorder, migraine and Prader-Willi syndrome. The comorbidity network of relatives of probands with Asperger syndrome was consistent with that of the relatives of probands with ASD. The structure of the network was slightly less peripheral than the one found for relatives of probands with ASD, as suggested by the relatively higher eigenvector centrality values. The conditions with the greatest eigenvector centrality were ASD (0.104), Asperger syndrome (0.103), depression and anxiety (0.103) and OCD (0.102) (Table 3). 4. Discussion The general comorbidity pattern reported here is consistent with evidence from the epidemiological literature. Depression and anxiety are highly prevalent comorbid conditions in individuals with ASD and their families (Matson & Williams, 2014). Overall, our findings are consistent with the psychiatric comorbidity found in register-based studies (Meier et al., 2015).
Table 3 First degree relatives comorbidity: social network structure. n
ASD ASP All
502 195 916
Overall network
Main Cluster
Conditions
Edges
Conditions
Edges
13 13 15
256 204 676
9 10 10
204 194 608
6
J. Virues-Ortega et al. / Research in Autism Spectrum Disorders 36 (2017) 1–7
The co-occurrence of epilepsy among individuals with ASD is well established in the literature, and the patterns of comorbidity reported here are consistent with these findings (e.g., Besag, 2015). The proposed cluster analysis had some construct validity in that there was a degree of differentiation between psychopathological conditions (e.g. depression and anxiety, ADHD) and other conditions (e.g. epilepsy, gastrointestinal). A higher-than-normal prevalence of OCD among close relatives of individuals with ASD is also well documented in the literature (Meier et al., 2015). Initial recruitment relied on self- and proxy-reported data. Recent studies have indicated the reliability of self- and proxyreporting, demonstrating recruitment of families with reliable and valid diagnoses that correlate with clinical reports (Daniels et al., 2011; Lee et al., 2010; [29_TD$IF]Warnell et al., 2015). Moreover, the overall ethnic structure of the cohort reflects New Zealand’s most recent census data ([30_TD$IF]Statistics New Zealand, 2013) suggesting that the sample was not geographically biased. Also, the sex ratio reflects the widely reported male-to-female bias of ASD and Asperger syndrome. Interestingly, we have fewer than expected registrations from individuals for whom ASD is not their primary diagnosis, such as Tuberous Sclerosis and Rett syndrome, which suggests that a more targeted recruitment may be necessary to reach these clinical populations. Other potential limitations are the restricted number of categories that participants could select in the process of completing an expression of interest, which may mean that unknown or low-prevalence comorbid conditions were underreported. However, this risk is partially mitigated by a free-text field that allows the registrand to record any condition. The capability of large research cohorts to advance understanding of ASD has been demonstrated through large international ASD initiatives around the world (reviewed by [29_TD$IF]Warnell et al., 2015). Despite the high prevalence, little is known about the epidemiology of ASD in New Zealand. The self-reported data mirror sociodemographic and comorbidity patterns of international cohorts, suggesting that the epidemiology of ASD in New Zealand is similar to that described for European populations. We anticipate that this will contribute to health decision-makers in New Zealand, advising key clinical needs for this population as a whole (e.g., high incidence of depression and anxiety), and also within specific ethnic groups. We also foresee further academic research on this cohort, for instance the genetic analysis, will advise diagnostic and clinical management protocols, thereby improving the speed and accuracy for which a result is returned to a family. Also, the cohort could provide the basis for outcome research studies evaluating the effects of various services and interventions, including early intensive behavioural intervention. Given the unique ethnic composition of New Zealand (i.e., Maori and Pacific) and the centralised health-care system, we anticipate that the subsequent research projects resulting from this cohort will encourage the development of a formalised national database and epidemiological study of ASD in New Zealand and contribute to the international understanding of the condition. Conflict of interest The authors declare that they have no conflict of interest.[31_TD$IF] Acknowledgements We would like to thank all the families involved in the research, our four founding community organizations; Autism New Zealand, Children’s Autism Foundation, Autism Intervention Trust, Altogether Autism and the Minds for Minds charitable trust. JCJ is supported by a Rutherford Discovery Fellowship from Government funding, administered by the Royal Society of New Zealand. The research was funded by the Minds for Minds charitable Trust, the Neurological Foundation of New Zealand, the Oakley Mental Health Research Foundation,The University of Auckland, New Zealand Federation of Women’s Institutes and the Julius Brendel Trust. References Besag, F. M. (2015). Current controversies in the relationships between autism and epilepsy. Epilepsy & Behavior, 47, 143–146. http://dx.doi.org/10.1016/j. yebeh.2015.05.032. Christensen, D. L., Baio, J., & Braun, K. V., et al. (2012). Prevalence and Characteristics of Autism Spectrum Disorder [34_TD$IF]Among Children Aged 8 Years — Autism and Developmental Disabilities Monitoring Network, [35_TD$IF]11 Sites, United States, MMWR Surveill Summ 2016, 65 (No. SS-3) (No. SS-3) 1–23. DOI: http://dx. doi.org/10.15585/mmwr.ss6503a1. Cramer, A. O., Waldorp, L. J., & van der Maas, H. L. (2010). Comorbidity: A network perspective. Behavior and Brain Sciences, 33, 137–150. http://dx.doi.org/ 10.1017/S0140525x09991567. Daniels, A. M., Rosenberg, R. E., Law, J. K., Lord, C., Kaufmann, W. E., & Law, P. A. (2011). Stability of initial autism spectrum disorder diagnoses in community settings. Journal of Autism and Developmental Disorders, 41, 110–121. http://dx.doi.org/10.1007/s10803-010-1031-x. Fruchterman, T. M. J., & Reingold, E. M. (1991). Graph drawing by force-directed placement. Software: Practice and Experience, 21, 1129–1164. http://dx.doi. org/10.1002/spe.4380211102. Girvan, M., & Newman, M. E. J. (2002). Community structure in social and biological networks. Proceedings of the National Academy of Sciences of the United States of America, 99, 7821–7826. http://dx.doi.org/10.1073/pnas.122653799. Hansen, D., Shneiderman, B., & Smith, M. A. (2011). Analysing social media networks with NodeXL. Burlington, MA: Elsevier. Lee, H., Marvin, A. R., Watson, T., Piggot, J., Law, J. K., Law, P. A., et al. (2010). Accuracy of phenotyping of autistic children based on Internet implemented parent report. American Journal of Medical Genetics Part B: Neuropsychiatric Genetics, 153b, 1119–1126. http://dx.doi.org/10.1002/ajmg.b.31103. Matson, J. L., & Williams, L. W. (2014). Depression and mood disorders among persons with autism spectrum disorders. Research in Developmental Disabilities, 35, 2003–2007. http://dx.doi.org/10.1016/j.ridd.2014.04.020.Epub 2014 May 24. Meier, S. M., Petersen, L., Schendel, D. E., Mattheisen, M., Mortensen, P. B., & Mors, O. (2015). Obsessive-Compulsive disorder and autism spectrum disorders: Longitudinal and offspring risk. PLoS One, 10, e0141703. http://dx.doi.org/10.1371/journal.pone.0141703.
J. Virues-Ortega et al. / Research in Autism Spectrum Disorders 36 (2017) 1–7
7
NodeXL team, Mircrosoft Corporation (2011). NodeXL: Network overview, discovery and exploration for Excel, version 1.0.1.167. Retrieved from http://www. codeplex.com/nodexl. Statistics New Zealand [39_TD$IF](2013). 2013 Census. Retrieved from http://www.stats.govt.nz/Census/2013-census.aspx. Warnell, F., George, B., McConachie, H., Johnson, M., Hardy, R., & Parr, J. R. (2015). Designing and recruiting to UK autism spectrum disorder [40_TD$IF]research databases: do they include representative children with valid ASD diagnoses? British Medical Journal, 5(9) . http://dx.doi.org/10.1136/bmjopen-2015008625.