Published on 04.05.17 in Vol 3, No 2 (2017): Apr-Jun
Preprints (earlier versions) of this paper are available at http://preprints.jmir.org/preprint/7150, first published Dec 14, 2016.
GapMap: Enabling Comprehensive Autism Resource Epidemiology
Background: For individuals with autism spectrum disorder (ASD), finding resources can be a lengthy and difficult process. The difficulty in obtaining global, fine-grained autism epidemiological data hinders researchers from quickly and efficiently studying large-scale correlations among ASD, environmental factors, and geographical and cultural factors.
Objective: The objective of this study was to define resource load and resource availability for families affected by autism and subsequently create a platform to enable a more accurate representation of prevalence rates and resource epidemiology.
Methods: We created a mobile application, GapMap, to collect locational, diagnostic, and resource use information from individuals with autism to compute accurate prevalence rates and better understand autism resource epidemiology. GapMap is hosted on AWS S3, running on a React and Redux front-end framework. The backend framework is comprised of an AWS API Gateway and Lambda Function setup, with secure and scalable end points for retrieving prevalence and resource data, and for submitting participant data. Measures of autism resource scarcity, including resource load, resource availability, and resource gaps were defined and preliminarily computed using simulated or scraped data.
Results: The average distance from an individual in the United States to the nearest diagnostic center is approximately 182 km (50 miles), with a standard deviation of 235 km (146 miles). The average distance from an individual with ASD to the nearest diagnostic center, however, is only 32 km (20 miles), suggesting that individuals who live closer to diagnostic services are more likely to be diagnosed.
Conclusions: This study confirmed that individuals closer to diagnostic services are more likely to be diagnosed and proposes GapMap, a means to measure and enable the alleviation of increasingly overburdened diagnostic centers and resource-poor areas where parents are unable to diagnose their children as quickly and easily as needed. GapMap will collect information that will provide more accurate data for computing resource loads and availability, uncovering the impact of resource epidemiology on age and likelihood of diagnosis, and gathering localized autism prevalence rates.
JMIR Public Health Surveill 2017;3(2):e27
Autism spectrum disorder (ASD) has been attracting more interest recently, as a result of skyrocketing prevalence rates. Centers for Disease Control and Prevention (CDC) estimates in 2000 suggested that the condition affected 1 in 150 children; 10 years later, the rate rose to 1 in 68 children . However, many present epidemiological studies suffer from small sample sizes and regional focuses [ - ]. For example, the most recent CDC prevalence estimates were based on a limited number of communities in only 11 states [ ]. However, studies that estimate prevalence on broader geographic scales often rely on aggregate data that omit valuable regional prevalence rates. Although families report concerns as early as 1 year of age and reliable diagnoses can be made during early childhood [ ], the average age of diagnosis for ASD in the United States is more than 4 years and an estimated quarter of children remain undiagnosed at 8 years of age [ - ]. Despite the considerable number of undiagnosed individuals who meet the criteria for autism [ ], studies that calculate prevalence rates often do not include undiagnosed individuals [ ]. This means that individuals without access to diagnostic centers for socioeconomic or geographic reasons are not reported, resulting in underrepresented statistics [ , ]. Instead, an ideal dataset would span a large geographic region (such as the United States) and maintain high specificity (to the city level).
Comprehensive regional autism prevalence rates would be extremely helpful for determining the true prevalence of autism and correlating genetic and environmental factors with higher standards of significance. In particular, comparing geographic trends in prevalence rates with autism resource epidemiology would be invaluable in revealing patient care deficits.
Whereas much effort has been focused on measuring autism incidence and prevalence rates, few have explored resource epidemiology. Yet, studies reveal alarming indicators of significant resource shortages. In metropolitan areas, parents may wait up to 18 months after initial suspicions for a final diagnosis . For families of minority populations or lower socioeconomic status, this process may take even longer [ ]. Children with autism in rural or resource-poor areas are diagnosed much later than their peers in resource-rich areas (if they are diagnosed at all) and face difficulties finding appropriate post diagnosis services [ ].
Finding these resource gaps, regions in which there are limited diagnostic or treatment resources with respect to the demand, can support pushes for congressional change with hard data, allocate resources more efficiently, and provide information to emerging organizations and businesses to let them know where their services are most needed. These efforts can help reduce age at the first diagnosis and ensure that speech and behavioral therapies are started during critical periods when they are maximally impactful [, ].
Aims of This Study
The specific aims of this study were to (1) obtain an early approximation of the disconnect between autism resources and diagnosed individuals by determining the average distance between an individual with autism and the nearest diagnostic center based on public census data in the United States and the United Kingdom, (2) define useful metrics that can be used to determine whether a center is overloaded or whether a region is underserved, and (3) propose an online prototype to collect information pertaining to geographic variations of autism prevalence and the geographic resource utilization of autism resources.
Our proposed app, GapMap , will contribute to the growing field of participatory surveillance and epidemiology [ - ] by prospectively collecting locational, diagnostic, and resource use information from individuals with autism in order to compute accurate prevalence rates and better understand autism prevalence rates and resource epidemiology.
Computing the Average Distance From an Individual to the Nearest Diagnostic Center
A simulation of the average distance from an individual in the United States to the nearest diagnostic center was constructed using county-level 2015 Census Data  and a list of 840 developmental medical centers that were collected through Autism Speaks [ ] and Autism Source [ ]. County populations and land area were used to model randomly uniform distributions of individuals within county borders; individuals were then assigned to the nearest diagnostic center. The average and standard deviation were then computed for the distance from an individual to the nearest diagnostic center.
Computing the Average Distance From an Individual With ASD to the Nearest Diagnostic Center
A simulation of the average distance from an individual with an ASD to the nearest diagnostic center was constructed for the United States and the United Kingdom. Data from the United Kingdom are comprehensively collected and publicly available, which allowed us to conduct a proof-of-concept study with a more complete dataset to highlight disparities between families with a clinical diagnosis of autism and autism-related resources. City-level geographic approximations of individuals with autism were obtained by screen scraping information from the Internet and social networks. Lists of 840 and 135 diagnostic centers in the United States and United Kingdom, respectively, were collected through Autism Speaks , Autism Source [ ], and the National Autistic Society (in the United Kingdom) [ ]. Individuals were similarly assigned to the nearest diagnostic center, and statistics were computed. Maps of the individuals and centers were generated for both countries.
Computing Resource Load and Availability
Resource load was computed for 840 diagnostic centers in the United States using county-level 2015 Census Data [- ]. Individuals were randomly assigned to a center near them, where nearby centers were defined as all the centers within a radius of 25, 100, 500, or 3000 km (the smallest radius was chosen so that at least one center existed within that distance). The maximum and average distance an individual traveled to each center was calculated in addition to the resource load.
Resource load was calculated based on an approximation of the number of caseloads per clinician in a year and an estimation of the number of clinicians at a single diagnostic center. These numbers vary greatly by clinician, services, and the size of the diagnostic centers, but families are most often referred to general practitioners, pediatricians, speech and language therapists, psychologists, and psychiatrists . We calculated our statistics based on an estimate that the average diagnostic center is composed of 5 specialists who can attend to about 200 patients each per year. The actual number is likely lower. Two hospitals in Ohio, both of which have 4-5 clinicians on their teams for diagnostics, experience months-long waitlists, with an estimated 40-100 referrals for ASD each month and about 698-1100 patients with ASD who need ongoing care per year [ ]. A school district in Texas, staffs 52 diagnosticians and 45 speech language pathologists across about 60 schools, and estimates that each diagnostician has a caseload of 87.5 students per year and each speech language pathologist has a caseload of 53.3 students per year [ ], yet the American Speech-Language-Hearing Association estimates that the median monthly caseload of a speech language pathologist working full-time in the United States was 47 cases per month [ ]. For analysis, we set the percent of individuals in need of an autism screening set to 0.195% (given that 6.5% of individuals are at an age appropriate for diagnosis from US Census Data, with 3% of those children needing an autism screening).
Resource availability statistics were also computed (Equation 2 in “Results” section) for each county in the United States and used to generate a heatmap. The results from the previous analyses were used for the parameters (z-average distance from an individual with autism) and (RL-resource load).
By having individual families submit their diagnostic and demographic information through GapMap , we can eventually calculate a more accurate prevalence rate with more granularity. A simple Web portal was created to allow individuals with ASD to enter relevant data and to crowdsource from the social and global databases. Caregivers (aged ≥18 years) of persons with a clinical diagnosis of ASD are allowed to submit data, including gender, date of birth, location including city, state, and zip code, specific (clinically defined) diagnosis of an ASD, diagnostic tools, date of diagnosis, any comorbid conditions, email address, and a list of ratable local services used to care for the subject (where local services include medical specialists, therapists, support resources, and “autism-friendly” generic services). We will then validate the disclosed ASD diagnosis by having participants complete a peer-reviewed autism screening measure such as the Social Communication Questionnaire (SCQ) [ ] or the Social Responsiveness Scale (SRS) [ ]. IP addresses, date and time of submission, and similarity of data submitted will also be used to detect duplicate or flag anomalies as potentially falsified data. These two measures will ensure higher quality data, as crowdsourced-data have been shown to match the quality of expert-curated data with proper instructions for data submission and reasonable validation on input data [ - ].
On validation, we will include the family’s data in GapMap; however, if they do not meet ASD criteria, we will not limit the participant from reviewing local resources. Providing families with an accurate list of local resources gives undiagnosed or untreated individuals more chances for earlier diagnosis and an earlier onset of therapy. In addition, we account for the undiagnosed populations within a community by collecting data from participants who select “No diagnosis, but suspicious” under the clinical ASD diagnosis option.
Mainstream social networking platforms, including Facebook, Twitter, colleague networks, consortia, related events and conferences, websites, and fan-based networks, will be used to reach families who have a family member with a clinical diagnosis of autism.
GapMap’s home page also allows users to search and edit its database of autism resources, as shown in. , which features a Google map with markers and marker clusters for resources and a heatmap overlay, shows resource availability. The metrics resource load, resource gap, and resource availability were defined to aid in this purpose, and estimates were calculated with limited datasets to provide more evidence for the importance of our tool.
Average Distance to the Nearest Diagnostic Center
The distance an individual must travel to the nearest autism diagnostic center is important for several reasons. Studies have shown that individuals are less likely to take advantage of medical services when they are located far away, particularly for low-income families [, ]. Local care is convenient, and families that must travel long distances to diagnostic facilities may wait to observe more risk factors in their child to schedule an appointment [ ]. It may also be the case that areas without any autism resources may lack general knowledge of autism or its characteristic behaviors. These delays may dramatically increase the time to diagnosis, as we know to be the case for other disorders in rural regions [ ]. However, given the scarcity of autism diagnostic resources, the typical distance individuals travel for an autism diagnosis may be greater than realized.
Our simulation constructed from US Census data revealed the average distance from an individual in the United States to the nearest diagnostic center to be approximately 182 km (50 miles), with a standard deviation of 235 km (146 miles). The list of diagnostic centers included hospitals and clinics with dedicated autism diagnosis personnel, but may have missed many solo and small-town diagnosticians. As complete and precise population maps (at the city level) were not available, the simulation also assumed that populations within counties are uniformly distributed, whereas populations are likely to be clustered even within counties. Both of these limitations may slightly overinflate the resulting distances. Even with a fair amount of overestimation, these distances are alarming, and suggest that autism resource allocation is a central issue that needs to be addressed.
The average distance from an individual with ASD to the nearest diagnostic center was also calculated for the United States and the United Kingdom using locational information scraped from the Internet. The dataset for the United States was comprised of 47,622 individuals with autism and 840 developmental medical centers; the dataset for the United Kingdom was comprised of 737 individuals with autism and 135 diagnostic centers. The pool of individuals sampled was restricted to English speakers with Internet access, which can skew toward cities (where autism resources are likely to be more abundant) and younger adults. The results are shown in.
|Location||United States (miles)||United Kingdom (miles)|
|Average distance||32 km (20)||31 km (19)|
|Median distance||13 km (8)||11 km (7)|
|Maximum distance||1,819 km (11,130)||519 km (322)|
|Percent of individuals living within 30 km (19 miles) of a diagnostic center (z)||70%||74%|
Notably, the average distances from an individual with autism to the nearest diagnostic center are lesser than the average distance from any individual to the nearest diagnostic center. Although some researchers have hypothesized that this higher rate might be a result of environmental factors in urban areas contributing to this higher incidence [, ], and distances may be underestimated as a result of biases in sampling methods, these results suggest that individuals who live closer to the diagnostic services are more likely to be diagnosed. This interpretation supports previous research [ ], but can also be explained by the incentive for families to move closer to health services if they have concerns for ASD, or other correlational factors such as more advanced parental age (older parents may be more likely to live in urban areas). In order to rule out alternative explanations such as sampling biases or environmental factors and determine whether such an association exists, future work should look at the relationship between age of diagnosis and distance from the nearest diagnostic center. If such a connection does exist, we would expect to see that the average age of diagnosis is lower for individuals closer to a diagnostic provider than for those who must travel extensively. Although we were not able to collect the date of diagnosis for the individuals in this study, the release of GapMap will soon enable such analyses. GapMap will be an effective tool to support or challenge correlations to certain environmental factors or clusters of high autism incidence. As GapMap collects data increasingly, we will also be able to determine the average distance for individual states in the United States.
Computing Resource Shortages
Although we know autism resources are scarce in many regions, there have not been any attempts to quantify the magnitude or scope of these resource shortages. It will be useful, then, to define several key metrics to determine the existence and magnitude of resource shortages.
We will define resource load (RL) as a measure of how well a center can meet the resource demand placed on it. This is a simple ratio of demand to supply, where demand is the number of individuals who require autism services (Np) and supply is equal to the product of the number of specialists who work at the center (s) and the number of individuals a specialist can attend yearly (o). A resource load of 1 would indicate, then, that a center is perfectly able to meet the demand. A resource load of 6, however, means that the center is overburdened with 6 times the number of patients it can comfortably handle. We can expect that centers with a resource load greater than one have longer wait times for appointments. Seefor the equation.
Resource load may be underestimated for some centers and overestimated for others because individuals are often willing to travel greater distances to receive the “best” medical care for their children. In future, we can use ratings and local services usage information collected from GapMap to help explore the extent of this possible phenomenon by checking for correlations between ratings of and distance traveled to each resource.
Although resource load is useful from the perspective of a resource center, it is also important to consider the availability of resources from the perspective of a family looking for ASD resources. We will therefore define resource availability (RA) with respect to a given location. The equation is shown in.
In short, resource availability is the sum of the resources available in a given location, where each resource’s contribution is adjusted by its resource load and its distance to the given location. Resources that are very far away and resources that are overburdened are given little weight. Resource availability will be high when resources with reasonable resource load are nearby, and will be at least equal to one if one resource exists within a reasonable distance with a resource load of one. A resource gap, or region in which there exist limited diagnostic or treatment resources with respect to the demand, is indicated by a resource availability of less than 1. These regions are where we should focus most on improving the accessibility of resources.
The parameter z accounts for how far the average individual will travel to seek out a resource. Note that this distance may vary with the type of resource—for example, individuals may be willing to travel farther for behavioral therapy (once a child has been diagnosed) than for diagnostic resources, when they are acting only on suspicions of developmental delays. For this study, we estimate z for diagnostic centers at 30 km by using the results of our previous analysis, where we estimate that 70% of individuals with autism in the United States live within 30 km of a diagnostic center.
We computed resource load for 840 developmental medical centers in the United States. About 0.5% of the centers had a resource load of less than one, and even after giving some leeway to account for the limited information on diagnosticians and diagnostic centers, only 5% of the centers had a resource load of less than 5. Alarmingly, 18% of centers had a resource load greater than 25, or experienced 25 times the demand they could handle. The average center had a resource load of 18 and a standard deviation of 14. The median center had a resource load of 14, and the maximum resource load experienced by a center was 113. These numbers are distressing, but not particularly surprising, given the difficulty and time involved in obtaining an autism diagnosis. Regardless, they make a convincing case for further research in and reassessment of our investment in diagnostic resources.
We are missing the information about diagnostic capabilities in smaller facilities and rural areas. Specialists ranging from general practitioners to developmental pediatricians can (but are not required to) be certified to diagnose ASD, and most use brief, unstandardized assessment instruments to diagnose ASD [, ]. Not all certified specialists feel comfortable giving a diagnosis and will refer parents to another professional for diagnosis [ ]. Because of this, it is difficult to know which clinicians are also autism diagnosticians. This phenomenon is particularly true in rural areas, where facilities dedicated specifically to ASD are few or nonexistent and generalist clinicians are more prevalent [ , ]. This is problematic because parents who are referred from their pediatrician to a specialist are more likely to receive a diagnosis sooner [ ]. In these very rural areas, parents may need to rely on a family doctor or pediatrician to recognize risk factors or early signs of ASD. Soon, we will be able to use GapMap to learn more about these complex dynamics and come to a better understanding of autism diagnoses, small town diagnosticians, and rural specialists who provide referrals.
Although these analyses and metrics highlight the lack of resources in much of the United States and the overburdening of many centers, they are not enough. We have built GapMap as a tool to collect important information and visually display the results. The collected location information, diagnosis, diagnostic tools, and comorbid conditions will be used to obtain both widespread and highly localizable autism prevalence rates. Date of diagnosis and age will be aggregated and used to obtain localizable average age of diagnosis, a measure that correlates with difficulty obtaining a diagnosis and can be used to help approximate geographic differences in resource accessibility. We will also collect the specific diagnoses of individuals who submit their data, including diagnostic and statistical manual of mental disorders (DSM-IV and DSM-5) diagnoses and an option for “no diagnosis, but suspicious.” This information will allow us to track individuals who remain undiagnosed, providing us with valuable data that is often overlooked in studies of prevalence rates. Ratings and local services will be used to estimate resource usage trends with respect to geography and resource density. Prevalence rates and local service usage will also be used to calculate resource load and availability for different resource types such as behavioral therapy.
There is a dearth of ASD resources, as well as a lack of understanding of the extent and epidemiology of these resource gaps. Statistics computed from simulations and web scraping suggest that individuals located close to diagnostic centers are more likely to be diagnosed. By quantitatively defining resource load and resource availability, we provide a means to measure and enable the alleviation of increasingly overburdened diagnostic centers and resource-poor areas where parents are unable to diagnose their children as quickly and easily as needed. The release of GapMap will collect crucial information that will provide more accurate data for computing resource loads and availability, uncovering the impact of resource epidemiology on age and likelihood of diagnosis, and gathering localized autism prevalence rates, both from families that have already received a diagnosis and families that haven’t received an official diagnosis yet.
The authors acknowledge support from Stanford University School of Medicine. They thank Anish Nag, Anika Kumar, and Sylvia Illouz for their work in mining autism-related-resource databases. They also extend thanks to Matthew Hoying, for his advice and contributions with database design and construction. All authors read and approved the final manuscript. The work was supported in part by Stanford’s Child Health Research Institute.
Conflicts of Interest
- Center for Disease Control and Prevention (CDC). 2016. Prevalence of autism spectrum disorder among children aged 8 years? autism and developmental disabilities monitoring network, 11 sites, United States, 2010 URL: https://www.cdc.gov/mmwr/preview/mmwrhtml/ss6302a1.htm [accessed 2016-12-13] [WebCite Cache]
- White S. Anxiety in children and adolescents with autism spectrum disorders. Clin Psychol Rev 2009;29(3):216-229. [CrossRef]
- Elsabbagh M, Divan G, Koh Y, Kim YS, Kauchali S, Marcín C, et al. Global prevalence of autism and other pervasive developmental disorders. Autism Res 2012 Jun;5(3):160-179 [FREE Full text] [CrossRef] [Medline]
- Maenner MJ, Yeargin-Allsopp M, Van Naarden BK, Christensen DL, Schieve LA. Development of a machine learning algorithm for the surveillance of autism spectrum disorder. PLoS One 2016 Dec;11(12):e0168224 [FREE Full text] [CrossRef] [Medline]
- Autism and Developmental Disabilities Monitoring Network Surveillance Year 2008 Principal Investigators, Centers for Disease Control and Prevention. Prevalence of autism spectrum disorders--Autism and Developmental Disabilities Monitoring Network, 14 sites, United States, 2008. MMWR Surveill Summ 2012 Mar 30;61(3):1-19 [FREE Full text] [Medline]
- Developmental, Disabilities Monitoring Network Surveillance Year, and 2010 Principal Investigators. Prevalence of autism spectrum disorder among children aged 8 years-autism and developmental disabilities monitoring network, 11 sites, United States, 2012. MMWR Surveill Summ 2016;65(3):1-23. [CrossRef]
- Mazurek MO, Handen BL, Wodka EL, Nowinski L, Butter E, Engelhardt CR. Age at first autism spectrum disorder diagnosis: the role of birth cohort, demographic factors, and clinical features. J Dev Behav Pediatr 2014;35(9):561-569. [CrossRef] [Medline]
- Shattuck PT, Durkin M, Maenner M, Newschaffer C, Mandell DS, Wiggins L, et al. Timing of identification among children with an autism spectrum disorder: findings from a population-based surveillance study. J Am Acad Child Adolesc Psychiatry 2009 May;48(5):474-483 [FREE Full text] [CrossRef] [Medline]
- Tonge BJ, Bull K, Brereton A, Wilson R. A review of evidence-based early intervention for behavioural problems in children with autism spectrum disorder: the core components of effective programs, child-focused interventions and comprehensive treatment models. Curr Opin Psychiatry 2014 Mar;27(2):158-165. [CrossRef] [Medline]
- Russell G, Ford T, Steer C, Golding J. Identification of children with the same level of impairment as children on the autistic spectrum, and analysis of their service use. J Child Psychol Psychiatry 2010 Jun;51(6):643-651. [CrossRef] [Medline]
- Miller L, Reynolds J. Autism and vaccination-the current evidence. J Spec Pediatr Nurs 2009 Jul;14(3):166-172. [CrossRef] [Medline]
- Hutton AM, Caron SL. Experiences of families with children with autism in rural New England. Focus Autism Other Dev Disabl 2005 Jan 01;20(3):180-189. [CrossRef]
- Mandell DS, Novak MM, Zubritsky CD. Factors associated with age of diagnosis among children with autism spectrum disorders. Pediatrics 2005 Dec;116(6):1480-1486 [FREE Full text] [CrossRef] [Medline]
- Wiggins LD, Baio J, Rice C. Examination of the time between first evaluation and first autism spectrum diagnosis in a population-based sample. J Dev Behav Pediatr 2006 Apr;27(2 Suppl):S79-S87. [Medline]
- Bernier R, Mao A, Yen J. Psychopathology, families, and culture: autism. Child Adolesc Psychiatr Clin N Am 2010 Oct;19(4):855-867. [CrossRef] [Medline]
- Howlin P. Children with Autism and Asperger's Syndrome: A Guide for Practitioners and Carers. Chichester, UK: Wiley; 1998.
- Pinto-Martin JA, Young LM, Mandell DS, Poghosyan L, Giarelli E, Levy SE. Screening strategies for autism spectrum disorders in pediatric primary care. J Dev Behav Pediatr 2008 Oct;29(5):345-350. [CrossRef] [Medline]
- GapMap. 2017. The Wall Lab URL: https://gapmap.stanford.edu/ [accessed 2017-03-28] [WebCite Cache]
- Jost CC, Mariner JC, Roeder PL, Sawitri E, Macgregor-Skinner GJ. Participatory epidemiology in disease surveillance and research. Rev Sci Tech 2007 Dec;26(3):537-549. [Medline]
- Freifeld CC, Chunara R, Mekaru SR, Chan EH, Kass-Hout T, Ayala IA, et al. Participatory epidemiology: use of mobile phones for community-based health reporting. PLoS Med 2010 Dec;7(12):e1000376 [FREE Full text] [CrossRef] [Medline]
- Paolotti D, Carnahan A, Colizza V, Eames K, Edmunds J, Gomes G, et al. Web-based participatory surveillance of infectious diseases: the Influenzanet participatory surveillance experience. Clin Microbiol Infect 2014 Jan;20(1):17-21 [FREE Full text] [CrossRef] [Medline]
- U.S. Census Bureau, Population Division. Factfinder. 2016. Annual estimates of the resident population: April 1, 2010 to July 1, 2015 URL: https://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?src=bkmk [accessed 2016-11-28] [WebCite Cache]
- Autism Speaks. 2016. URL: https://www.autismspeaks.org/ [accessed 2016-11-29] [WebCite Cache]
- Autism Source. 2004. URL: http://www.autismsource.org [accessed 2016-11-13] [WebCite Cache]
- Autism. 2016. The National Autistic Society URL: http://www.autism.org.uk/ [accessed 2016-12-13] [WebCite Cache]
- Crane L, Chester J, Goddard L, Henry L, Hill E. Experiences of autism diagnosis: a survey of over 1000 parents in the United Kingdom. Autism 2016;2(20):153-162. [CrossRef]
- Austin J, Manning-Courtney P, Johnson M, Weber R, Johnson H, Murray D, et al. Improving access to care at autism treatment centers: a system analysis approach. Pediatrics 2016;137(2):S149-S157 [FREE Full text] [CrossRef]
- Fort Bend ISD. 2010. Staffing review URL: http://www.fortbendisd.com/cms/lib09/TX01917858/Centricity/Domain/96/tasb-salay-study.pdf [accessed 2017-02-24] [WebCite Cache]
- American Speech-Language-Hearing Association. 2016. ASHA 2014 schools survey: speech language pathologist caseload characteristics report URL: http://www.asha.org/uploadedFiles/2014-Schools-Survey-SLP-Caseload-Characteristics.pdf [accessed 2017-02-23] [WebCite Cache]
- Bölte S, Holtmann M, Poustka F. The social communication questionnaire (SCQ) as a screener for autism spectrum disorders: additional evidence and cross-cultural validity. J Am Acad Child Adolesc Psychiatry 2008 Jun;47(6):719-720. [CrossRef] [Medline]
- Aldridge FJ, Gibbs VM, Schmidhofer K, Williams M. Investigating the clinical usefulness of the social responsiveness scale (SRS) in a tertiary level, autism spectrum disorder specific assessment clinic. J Autism Dev Disord 2012 Feb;42(2):294-300. [CrossRef] [Medline]
- Comber A, Brunsdon C, See L, Fritz S, McCallum I. Comparing expert and non-expert conceptualisations of the land: an analysis of crowdsourced land cover data. In: Tenbrink T, Stell J, Galton A, Wood Z, editors. Spatial Information Theory. COSIT 2013. Lecture Notes in Computer Science, vol 8116. Switzerland: Springer International Publishing; 2013:243-260. [CrossRef]
- Behrend TS, Sharek DJ, Meade AW, Wiebe EN. The viability of crowdsourcing for survey research. Behav Res Methods 2011 Sep;43(3):800-813. [CrossRef] [Medline]
- Swan M. Crowdsourced health research studies: an important emerging complement to clinical trials in the public health research ecosystem. J Med Internet Res 2012;14(2):e46. [CrossRef]
- Currie J, Reagan P. Distance to hospital and children's use of preventive care: is being closer better, and for whom? Economic Inquiry 2003;41(3):378-391. [CrossRef]
- Roberts A, Blunt I, Bardsley M. Focus On: Distance from home to emergency care. London: The Health Foundation and the Nuffield Trust; 2014.
- Pfeiffer P. Impact of distance facility of initial diagnosis on depression treatment. Health Serv Res 2011;46(3):768-786 [FREE Full text] [CrossRef]
- Rice CE, Baio J, Van Naarden BK, Doernberg N, Meaney FJ, Kirby RS. A public health collaboration for the surveillance of autism spectrum disorders. Paediatr Perinat Epidemiol 2007 Mar;21(2):179-190. [CrossRef] [Medline]
- Nevison CD. A comparison of temporal trends in United States autism prevalence to trends in suspected environmental factors. Environ Health 2014 Sep;13(1):73 [FREE Full text] [CrossRef] [Medline]
- Mazumdar S, Winter A, Liu K, Bearman P. Spatial clusters of autism births and diagnoses point to contextual drivers of increased prevalence. Soc Sci Med 2013 Oct;95:87-96 [FREE Full text] [CrossRef] [Medline]
|API: application program interface|
|ASD: Autism spectrum disorders|
|AWS S3: Amazon Web Services Simple Storage Serve|
|CDC: The Centers for Disease Control and Prevention|
|DSM: diagnostic and statistical manual of mental disorders|
|RDS: Amazon Relational Database Service|
|SCQ: Social Communication Questionnaire|
|SRS: Social Responsiveness Scale|
Edited by G Eysenbach; submitted 14.12.16; peer-reviewed by K Bowers; comments to author 22.02.17; revised version received 15.03.17; accepted 16.03.17; published 04.05.17
©Nikhila Albert, Jena Daniels, Jessey Schwartz, Michael Du, Dennis P Wall. Originally published in JMIR Public Health and Surveillance (http://publichealth.jmir.org), 04.05.2017.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Public Health and Surveillance, is properly cited. The complete bibliographic information, a link to the original publication on http://publichealth.jmir.org, as well as this copyright and license information must be included.