Incidence of Pediatric Urinary Tract Infections Before and During the COVID-19 Pandemic

Key Points Question Did the incidence of pediatric urinary tract infection (UTI) diagnoses and outcomes change during the COVID-19 pandemic? Findings In this cohort study of 13 million children with private insurance, the incidence of UTI was 1.30 cases per 100 patient-years, with notable variation by age, sex, and circumcision status. Compared with prepandemic trends, UTI diagnoses decreased by 33% during the early pandemic without associated changes in disease severity. Meaning This investigation provides updated data on UTI incidence in children; while the mechanism for decreased UTI incidence during the pandemic is unknown, a decrease in misdiagnosis and overdiagnosis may play a role.


eMethods. Statistical Model Details
We used an Interrupted Time Series (ITS) model to estimate the percent change in the rate of each outcome that is attributable to the pandemic.The model is exactly analogous to the one we used in a manuscript by Schroeder et al, 3 and was defined in two steps.
In the first step, we transformed the time series to make it roughly stationary during the pre-pandemic period.This where  early pand. is an indicator variable for months in the early pandemic period, and  mid pand. is the same for the mid pandemic period.We used autocorrelation-robust sandwich estimators for evaluate the 95% confidence intervals. 7Models were run separately for each outcome we reported.
The output of the model is the average percent change in the rate of each outcome in each period compared to the counterfactual in which the pre-pandemic trends had continued unabated (i.e., if  2 =  3 = 0. ) These percent changes are computed by exponentiating  2 and  3 , respectively.
Full details of this statistical modeling approach, including the handling of the March 2020 datapoint, were publicly pre-registered at https://osf.io/7rh3sprior to completion of data cleaning.
was accomplished by taking a logarithm and by applying yearly differencing, in which the seasonal effects are removed by subtracting off the value of the time series shifted by one year; the result is a new times series   = ln   − ln  −12 .The datapoint representing March 2020 was discarded: lockdowns occurred in the middle of March 2020, meaning this datapoint is partially in the pre-Pandemic period and partially in the early Pandemic period.In the second step, we fit a four-parameter linear regression model; the four parameters are: 1) an intercept and 2) slope (to model the pre-pandemic trend), and two step changes for the 3) early and 4) mid pandemic periods, respectively (to model the changes during the pandemic).The final regression model structure was: ln _ () − ln ( − 1 year) ~  +  1  +  2  early pand.+  3  mid pand.,

eTable 2 .
Most Common Antibiotics Prescribed for Children Aged 0-17 With UTI Diagnoses Measures of UTI severity including hospitalization and ICU admission, by age and circumcision status.