Table 1.
Summary of included studies.
| Author and Country | Study Type | Study Population | Autistic Cohort Characteristics | Data Source | Diagnostic Methods | Statistical Methods | Salient Findings | STROBE Score |
|---|---|---|---|---|---|---|---|---|
| T1DM-specific Findings | ||||||||
| Chen et al (2013) [19], Taiwan | Retrospective case-control study. | 1,598 autistic patients and 6,392 age and gender matched controls. | Mean age = 18 years; Age range = Not reported (NR); Male % = 80; ID % = NR; Ethnicity distribution (ED) = NR. | Electronic health records from 1996-2010. | Diagnoses were recorded in electronic records according to ICD-9-CM criteria [36]. | For between-group comparisons, the Pearson’s Chi square test or Fisher’s exact test were used for nominal variables, such as presence of T1DM. A logistic regression model was used to investigate the OR and 95% CI of T2DM in autistic patients. | Autistic patients were borderline significantly (p=0.056) more likely to have T1DM than non-autistic controls (Autistic group: 0.3%, 4/1,598; control group: 0.1%, 4/6,392). Modified OR (after adjusting for age and gender) = 4.00 (95% CI 1.00-16.00). | 17.5/27; 65% |
| Kohane et al (2012) [20], United States | Retrospective case-control study. | 14,381 autistic patients and 2,379,397 non-matched non-autistic controls. | Mean age = NR; Age range = 0-34 years; Male % = 79; ID % = NR; ED = NR. | Electronic healthcare records. | ICD-9 [37]. | Prevalence of DM was compared between autistic cases and controls using chi-square testing. | Autistic patients were significantly (p<0.00001) more likely to have T1DM (Autistic group: n=114, 0.79%; Control group: n=8058, 0.34%; 95% CI 0.3-0.6%). | 16/26; 62% |
| Supekar et al (2017) [21], United States | Cross-sectional study. | 4,790 autistic individuals and 1,842,575 non-matched non-autistic controls. | Mean age = NR; Age range = NR; Male % = NR; ID % = NR; ED = NR. | Electronic hospital database. | ICD-9 [37]. | Descriptive analysis, including percentage values (though not raw data). | The authors reported a T1DM prevalence of 0.42% among the autistic group, and 0.59% among the non-autistic controls. | 18.5/27; 69% |
| Zerbo et al (2015) [13], United States | Retrospective case-control. | 5,565 autistic individuals, and 27,825 controls, matched on age, sex and enrolment time. | Mean age = 12 years; Age range = NR; Male % = 82; ID % = NR; ED = NR. | Electronic health records. | ICD-9-CM [36]. | Chi-square testing was used to evaluate differences between the autistic and non-autistic groups. | Of the autistic group, 0.22% (12/5,565) had T1DM, compared with 0.19% (52/27,825) among the control group. This difference was not statistically significant (OR 1.15; 95% CI 0.62-2.17). | 21/27; 78% |
| T2DM-specific Findings | ||||||||
| Brondino et al (2019) [22], Italy | Cross sectional observational study | 191 autistic patients. No non-autistic controls were included. | Mean age = 24 years; Age range = NR; Male % = 75; ID % = 421; ED = NR. | Clinical records. | Autism diagnosis was according to DSM-5 [1],whereas T2DM was according to ICD-10 [38]. | Descriptive analysis, including frequencies. | Autistic patients had a T2DM prevalence of 0.5% (1/191). | 11/24; 46% |
| Shedlock et al (2016) [23], United States | Retrospective-case control study. | 48,762 autistic children and 243,810 controls matched by age, sex and enrolment time. | Mean age = NR; Age range = NR; Male % = 80; ID % = NR; ED = NR. | Electronic database. | ICD-9-CM [36]. | Conditional logistic regression was used to calculate OR and 95% CI for T2DM in comparing autistic patients and non-autistic controls. | Autistic children were significantly (p≤0.05) more likely to have T2DM (n=515; 1.06%) than non-autistic controls (n=970; 0.40%). Both results represent an underestimate of T2DM prevalence, as insulin treatment was a study exclusion criterion. | 25.5/32; 80% |
| T1DM-and T2DM-specific Findings | ||||||||
| Taggart et al (2013) [24], Northern Ireland | Quantitative questionnaire. | 186 individuals with ID, including autistic persons (though the number with autism is not reported). | No autistic cohort characteristics reported other than entire population having ID (overall study population characteristics reported only). | Postal questionnaire. | No details reported; questionnaire design suggests carers were just asked whether their patient had autism and/or DM. | The precise statistical technique used to examine the association between autism and DM is not reported2. | The authors report that significantly (p<0.05) more people with autism had T1DM (13%) compared with T2DM (5%). Raw data regarding the overall number of autistic patients, as well as those with T1DM and T2DM was not reported. | 17.5/27; 65% |
| Non-specific DM Findings | ||||||||
| Akobirshoev et al (2020) [25], United States. | Retrospective case-control study. | 34,237 adults with autism and 102,711 age and sex-matched controls. | Mean age = 33 years; Age range = NR; Male % = 75; ID % = NR; ED = 65% White non-Hispanic, 11% Black non-Hispanic, 5% Hispanic, 4% Other non-Hispanic, 15% Unknown ethnicity. | Electronic hospital database records from 2004-2014. | Autism diagnosis was according to ICD-9-CM criteria [36]. | Differences across categorical variables between the two groups were assessed using the Chi square test; Differences across continuous variables were assessed using the t-test. Logistic regression was used to assess mortality risk for different medical comorbidities in autistic adults relative to matched controls. | Among adults whom experienced in-hospital mortality, 11.5-13.6% (53-63/462) of the autistic group had a diagnosis of DM, compared to 15.0% (145/967) of the control group. Of these, 53 of the autistic group and 114 of the control group had DM without chronic complications (a statistically significantly [p≤0.05] increased risk in the autistic group). 0-103 of the autistic group and 31 of the control group had DM with chronic complications (a non-statistically significant difference). Overall data for all autistic adults (i.e. including those whom did not experience in-hospital mortality) was not reported. | 13.5/30; 45% |
| Alabaf et al (2019) [26], Sweden | Case-control study. | 23,049 children (301 with autism), from the Child and Adolescent Twin Study [39]. | Mean age = NR; Age range = 9-12; Male % = NR; ID % = 34; ED = NR. | Autism, Tics and other Comorbidities (A-TAC) [40] interview with parents. | Autism diagnosis was via the A-TAC, based on DSM-IV [41] and ICD-10 [38] criteria. DM diagnosis was based on parental self-report. | The Pearson Chi square or the Fisher’s exact tests were used to test for statistical significance between the prevalence of DM across subgroups. | 0.66% (2/301) of the autistic subgroup had a diagnosis of DM, compared to 0.4% (91/22028) of the control group. This difference was not statistically significant (p≥0.05). | 21.5/28; 77% |
| Croen et al (2015) [6],United States | Retrospective case-control study. | 1,507 autistic adults and 15,070 age and sex matched controls. | Mean age = 29 years; Age range = ≥18 years; Male % = 73; ID % = 19; ED = 66% White non-Hispanic, 11% Asian, 8% Black, 4% White Hispanic, 12% Other. | Electronic health records from January 2008 – December 2012. | Diagnoses were recorded in electronic records according to ICD-9-CM criteria [36] for autism, and ICD-9 criteria [37] for DM. | Prevalence of DM was compared between autistic cases and controls using chi-square tests. Also, a multivariate logistic regression model was run to compare the odds for DM between cases and controls. | Autistic adults were significantly (p<0.001) more likely to have DM than their non-autistic peers (Autistic group: n=114, 7.6%; control group: n=653, 4.3%). Adjusted OR (after adjusting for sex, age and race/ethnicity) = 2.18 (99% CI 1.62-2.93). | 20.5/26; 79% |
| Davignon et al (2018) [27], United States | Retrospective case-control study. | 47,509 children and young adults, including 4,123 with autism, 20,615 with ADHD, 2,156 with DM and 20,615 controls with none of these conditions, matched on age. | Mean age = 18 years; Age range = 14-25 years; Male % = 81; ID % = 13; ED = 54% White non-Hispanic, 16% Asian, 8% African American, 7% White Hispanic, 16% Other. | Electronic health records from 2013-2015. | Diagnoses were recorded in electronic records according to ICD-9 criteria [37]. | Prevalence of DM was compared between autistic cases and controls using chi-square testing. Additionally, a multivariate logistic regression model was run to compare the adjusted odds for DM between cases and controls. | 0.6% (25/4123) of the autistic group had DM; compared to 0.5% (107/20,615) of the ADHD group (Autism vs. ADHD adjusted OR 1.18; 95% CI 0.76-1.83). Due to the control group not having a diagnosis of DM by definition, a comparison between the autistic and control groups is not meaningful. | 24.5/29; 84% |
| Flygare Wallen et al (2018) [28], Sweden | Retrospective case-control study. | 13,921 autistic patients with no ID and 1,996,140 non-autistic non-matched controls4 | Mean age = NR; Age range = NR; Male % = 66; ID % = 05; ED = NR. | Electronic database. | Diagnoses were recorded according to ICD-10 criteria [38]. | Age-adjusted odds ratios with 95% CI’s were calculated using logistic regression analyses, to compare the prevalence of DM different groups. | Autistic patients had a DM prevalence of 2.87% (399/13,921), compared with a control group prevalence of 5.87% (117,148/1,996,140). However, the age-adjusted OR for DM in autistic patients was significantly (p<0.05) greater than for the control group (Autistic males – OR 1.705, 95% CI 1.50-1.93; Autistic females – OR 1.596, CI 1.33-1.92). Both DM prevalence estimates likely represent overestimations, as a diagnosis of DM, hypertension or obesity was an inclusion criterion for the study. | 16/24; 67% |
| Guinchat et al (2015) [29], France | Mixed retrospective and prospective cohort study. | 58 autistic patients (44 male and 14 female), 56 of which had co-occurring ID. No non-autistic controls were included. | Mean age = 16 years; Age range = 11-37 years; Male % = 76; ID % = 97; ED = NR. | Medical records. | ICD-10 criteria [38] confirmed by the Childhood Autism Rating Scale [42]. | Descriptive analysis, including frequencies. | Autistic patients had a DM prevalence of 1.7% (1/58). | 25/28; 89% |
| Gurney et al (2006) [30],United States | Cross-sectional analysis of survey. | 324,000 autistic children and 61,100,0006 non-autistic non-matched controls. | Mean age = NR; Age range = 3-17 years; Male % = 79; ID % = NR; ED = 74% White, 15% Black, 2% Multiracial, 3% Other, 7% Missing. | The 2003-2004 National Survey of Children’s Health (NSCH) [43],a population-based, cross-sectional telephone survey. | No specific diagnostic criteria were used. Autistic cases were identified from asking parents if a health professional has ever told them that their child has autism. | Prevalence values were calculated using the stratified weighted sampling fractions detailed in the NSCH public use data set. A multivariate logistic regression model was used to estimate OR. | Of the autistic group, 0.4% had DM, compared with 0.3% of the control group (raw values are not reported in the article). The difference between the groups was not statistically significant (OR 1.1; 95% CI 0.4-3.1). | 21.5/25; 86% |
| Jones et al (2016) [31], United States | Cross-sectional analysis of cohort. | 92 autistic adults. | Mean age = NR; Age range = 24-51 years; Male % = 75; ID % = 62; ED = NR. | Patient or caregiver questionnaire. | DSM-III [44] or DSM-IV-TR [45] for autism. No specific diagnostic criteria for DM were reported as being used. | Goodness-of-fit testing was used to test for any differences in the frequency of categorical variables, such as DM. | At follow-up, 9.8% (9/92) of patients had DM. Though the researchers did not have a control group for their study, they compared this finding with that of Ford et al.(2013), whom found a DM prevalence of 12% in the general US adult population. | 24.5/30; 82% |
| McDermott et al (2007) [32], United States | Retrospective cohort study. | 1,303 patients with disabilities (45 with autism) and 1,828 non-disabled patients. | Mean age = 28 years; Age range = NR; Male % = 76; ID % = NR; ED = NR for autistic cohort. | Medical records. | ICD-9-CM [36]. | Data was analysed using chi-square testing and logistic regression modelling. | Of the 45 autistic patients within the study population, 0% (0/45) had DM. This compares to a prevalence of 15.5% (202/1303) for the overall disability group and 14.5% (265/1828) for the non-disabled group. | 15/28; 54% |
| McManus et al. (2009) [33], England | Cross-sectional survey. | 19 autistic adults, and 599 non-autistic adults7. | Mean age = 48 years; Age range = 17-78 years; Male % = 79; ID % = NR; ED = 100% White British. | National survey data. | An ADOS [46]score of ≥10 was used as a proxy measure for autism. | No specific statistical methods were applied in comparing the rates of DM between autistic and non-autistic patients; the findings were reported as raw data. | Of the 19 autistic patients, 0% (0/19) had DM. This compares to a prevalence of 6.3% (38/599) for the non-autistic group. | Not applicable8 |
| Tyler et al (2011) [34], United States | Retrospective case-control study. | 108 autistic adults and 206 paired non-autistic controls matched for age, sex and insurance status. | Mean age = 29 years; Age range = NR; Male % = 71; ID % = 25; ED = 77% Caucasian, 14% African American, 2% Hispanic, 7% Other. | Electronic healthcare records. | ICD-9 [37]. | Precise details of statistical technique implemented are not reported9. | No significant difference in DM prevalence (p=0.68) between autistic group (n=7; 6.5%) and control group (n=16; 7.8%). | 23/31; 74% |
| Vohra et al (2017) [35], United States | Retrospective matched cohort study. | 1,772 adults with autism and 5,320 non-autistic adult controls matched by age, gender and race. | Mean age = NR; Age range = 22-64 years; Male % = 71; ID % = NR10; ED = 37% White, 21% African American, 42% Other. | Electronic healthcare records from 2000-2008. | ICD-9-CM [36]. | Chi-square tests of association were used for categorical variables, and t-tests for continuous variables, were used to assess differences between the autistic and non-autistic groups. | Of the 1,772 autistic adults, 3.6% (63/1772) had DM, compared with 4.7% (250/5320) for non-autistic controls. This difference was statistically significant (p<0.05). | 20.5/27; 76% |
1 The percentage of the study population with recorded ‘cognitive impairment’, which is a broader term than ID, also encompassing dementia-related illness and brain injury acquired following the developmental period. 2 The authors write ‘A series of chi-squared tests, independent t-tests and one-way ANOVAs were used to examine for significant differences between T1DM and T2DM, gender, age, level of ID, accommodation, BMI and hypertension across a number of quality diabetes care indicators.’ 3 To maintain confidentiality, the authors could not report precise numbers for cells with <11 cases. 4 There were also separate patient groups with ID and no Down syndrome (n= 11,785) and Down syndrome (n= 1282), though these have not been considered for the purposes of this review. 5 The authors report data for a separate group with ID, rather than as part of the autistic group or the non-autistic controls. 6 The sample sizes reported by the authors are US national estimates, based on sampling fractions and weighted extrapolation of parental reports of 483 autistic children withand 84,789 children without autism. 7 Autistic adults were defined as those scoring ≥10 on the Autism Diagnostic Observation Schedule, rather than having a clinical diagnosis of autism. Controls were defined as persons with ADOS scores of ≤9. 8 The autism and DM findings from the Adult Psychiatric Morbidity Survey were not reported in a specific journal article but rather gleaned from the overall dataset, available online, and as such a STROBE score could not be calculated. 9 Authors write ‘Using standard descriptive statistics, the two cohorts were compared in their documentation of selected biophysical data, chronic disease diagnoses, and pharmacotherapeutic data.’ 10 Though 1,231 (70%) of the autistic cohort were classified as having ‘developmental disorders,’