Table 2. Association Between Screening Intervention and Autism Spectrum Disorder (ASD) Incidence in Early Intervention (EI) Settingsa.
| Independent variable | Incidence of ASD per 1000 children per quarter | Difference-in-difference | Interaction | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Intervention sites | Comparison sites | IRR | P value | Marginal effect | IRR | P value | |||
| Preintervention | Postintervention | Preintervention | Postintervention | ||||||
| Overall screening intervention | 12.7 (10.8-15.1) | 21.8 (19.4-24.5) | 12.9 (10.9-15.2) | 17.0 (15.6-18.5) | 1.6 (1.3-2.1) | <.001 | 8.1 (5.4-10.8) | NA | NA |
| Descriptive subgroup analyses | |||||||||
| EI intervention | |||||||||
| Site 1 | 13.3 (10.8-16.5) | 27.5 (22-34.3) | NA | NA | 1.5 (1.1-2.1) | .02 | 8.0 (1.3-14.6) | 0.8 (0.5-1.3) | .44 |
| Site 2 | 10.9 (8.2-14.5) | 21.9 (17.7-27) | NA | NA | 2.0 (1.3-2.9) | <.001 | 12.8 (4.9-20.6) | 1.3 (0.8-2.1) | .27 |
| Site 3 | 14.4 (9-23.2) | 18 (15.3-21.4) | NA | NA | 1.4 (0.8-.5) | .24 | 5.6 (−2.5 to 13.7) | 0.9 (0.5-1.6) | .61 |
| Preferred language | |||||||||
| Spanish | 7.1 (4.7-10.8) | 19.5 (15.4-24.7) | 9.9 (7.5-13) | 13.2 (11.1-15.7) | 2.6 (1.6-4.3) | <.001 | 15.4 (5.3-25.4) | 1.8 (1.0-3.1) | .04 |
| English | 14.5 (12.1-17.5) | 22.8 (20-26.1) | 14.0 (11.9-16.4) | 18.4 (16.8-20.1) | 1.5 (1.1-1.9) | .005 | 6.9 (1.9-12) | [Reference] | NA |
| Child sex | |||||||||
| Female | 8.8 (6.2-12.6) | 10.0 (7.4-13.5) | 7.9 (6.4-9.7) | 9.8 (8.5-11.3) | 1.1 (0.6-1.7) | .84 | 0.5 (−3.9 to 4.8) | [Reference] | NA |
| Male | 16.5 (13.7-19.9) | 31.9 (28-36.2) | 17.8 (15.1-21.1) | 23.9 (21.9-26.1) | 1.8 (1.4-2.3) | <.001 | 14.8 (7.4-22.1) | 1.7 (1-2.9) | .04 |
Abbreviations: IRR, incidence rate ratio; NA, not applicable.
a
Preestimates and postestimates represent the number of new ASD diagnoses per 1000 children per quarter. The difference-in-difference IRR reflects the estimate representing the interaction between screening status and postperiod status. Difference-in-difference (marginal effect) represents an estimate of the additional number of ASD diagnoses per quarter associated with the independent variable. The interaction (IRR) represents the interaction of the difference-in-difference term with select covariates, which tests whether associations vary by subgroup. An IRR >1 indicates that screening was associated with higher incidence of ASD.