Table 5.
Marginal Effects from Logit Model Predicting ED Disposition Among Pediatric Psychiatric ED Visits from 2018-2020
| Explanatory Variables | Admission/Transfer versus Discharge | ||
|---|---|---|---|
| Point Estimate | Delta-Method SE | 95% CI | |
| Average Marginal Effects For Community Sites | |||
| Non-Psych Ped Daily Demand | −0.003* | 0.001 | [−0.006, −0.001] |
| Adult Psych Daily Demand | −0.002 | 0.001 | [−0.004, 0.000] |
| Ped Psych Daily Demand | −0.000 | 0.005 | [−0.011, 0.010] |
| ED Occupancy Rate | 0.001 | 0.000 | [−0.000, 0.001] |
| Average Marginal Effects For Academic Sites | |||
| Non-Psych Ped Daily Demand | −0.004* | 0.002 | [−0.007, −0.000] |
| Adult Psych Daily Demand | 0.000 | 0.002 | [−0.003, 0.003] |
| Ped Psych Daily Demand | −0.009* | 0.004 | [−0.018, −0.001] |
| ED Occupancy Rate | 0.001 | 0.001 | [−0.000, 0.002] |
| Average Marginal Effects Across All ED Sites | |||
| Academic Indicator | 0.035 | 0.094 | [−0.149, 0.220] |
| Age | |||
| 5-8 | −0.025 | 0.0 14 | [−0.053, 0.003] |
| 9-12 | −0.018 | 0.010 | [−0.038, 0.001] |
| 13-15 | 0.016 | 0.009 | [−0.001, 0.033] |
| 16-17 | Reference | Reference | Reference |
| Sex (Male) | −0.028*** | 0.008 | [−0.043, −0.013] |
| Race | |||
| White or Caucasian | Reference | Reference | Reference |
| African American | −0.019* | 0.009 | [−0.037, −0.002] |
| Other | −0.016 | 0.013 | [−0.042, 0.009] |
| Hispanic/Latino | 0.009 | 0.014 | [−0.019, 0.037] |
| Insurance | |||
| Commercial | 0.008 | 0.009 | [−0.009, 0.025] |
| Medicaid | Reference | Reference | Reference |
| Tricare/Medicare/Other | 0.051** | 0.019 | [0.013, 0.089] |
| None Listed | −0.105*** | 0.016 | [−0.137, −0.073] |
| Mode of Arrival | |||
| Ambulance/Flight | 0.079*** | 0.011 | [0.058, 0.100] |
| Walk-In | Reference | Reference | Reference |
| Police | 0.050*** | 0.012 | [0.027, 0.073] |
| Arrival Time | |||
| Overnight (12am-5:59am) | 0.067*** | 0.014 | [0.040, 0.094] |
| Morning (6am-11:59am) | −0.001 | 0.010 | [−0.022, 0.019] |
| Afternoon (12pm-5:59pm) | 0.018* | 0.008 | [0.002, 0.034] |
| Evening (6pm-11:59pm) | Reference | Reference | Reference |
| Weekend Visits | −0.008 | 0.009 | [−0.026, 0.010] |
| Acuity | |||
| Emergent | 0.445*** | 0.010 | [0.425, 0.465] |
| Urgent | 0.200*** | 0.010 | [0.181, 0.219] |
| Less Urgent | 0.020* | 0.008 | [0.004, 0.036] |
| Non-Urgent | Reference | Reference | Reference |
| Neurodev. Comorbidity | −0.027** | 0.008 | [−0.042, −0.011] |
| Hospital Site | |||
| Site A | Reference | Reference | Reference |
| Site B | −0.168 | 0.095 | [−0.354, 0.017] |
| Site C | −0.174 | 0.092 | [−0.354, 0.006] |
| Site D | −0.087 | 0.100 | [−0.283, 0.109] |
| Site E | −0.064 | 0.101 | [−0.262, 0.135] |
| Site F | −0.151 | 0.096 | [−0.339, 0.036] |
| Site G | −0.152 | 0.096 | [−0.341, 0.037] |
| Site H | −0.088 | 0.105 | [−0.293, 0.118] |
| Mar-Dec 2020 (PHE) | −0.029 | 0.017 | [−0.062, 0.004] |
| Visit Year | |||
| 2018 | Reference | Reference | Reference |
| 2019 | −0.016 | 0.009 | [−0.034, 0.002] |
| 2020 | 0.004 | 0.015 | [−0.025, 0.034] |
| N | 12,487 | ||
| Pseudo R^2 | 0.239 | ||
p<0.05
p<0.01
p<0.001
Notes: Point estimates indicate the change in the likelihood of admission/transfer compared to discharge. The ED Occupancy Rate variable was calculated by dividing the daily ED volume by the historical daily average ED volume at the same ED site on the same day of the week over the 30 days prior to the observed visit date. The logit model included interaction terms between the ED demand variables and an academic site indicator. Academic and Non-Academic site AMEs were calculated for the interacted demand variables by holding the academic indicator either equal to 1 or 0. The majority of patients in the Tricare/Medicare/Other insurance group had Tricare. CI = Confidence Interval, Neurodev = Neurodevelopmental, PHE=Public Health Emergency (COVID-19 Pandemic), SE=Standard Error.