Once the key is submitted, the data will be made available.
This section provides the user with tools necessary to build stratified tables. First, select the variable on which to stratify the tables, then select the row and column variables.
Results will appear in the tab labelled Stratified Tables.
Marginal tables are given in the tab labelled Marginals.
Records whose first/second placement combination was considered a disruption were classified as an event but were otherwise considered censored. Records that were missing second placement dates were censored in the following manner. If a record’s custody end date fell before the date of the latest CANS report, that record was censored at the custody end date. Otherwise, the record was censored at the date of the latest CANS report. To evaluate the probability of disruption, we fit a Cox proportional hazards model containing the following covariates:• Age • Race • Gender • Adjudication Status • CANS Recommendation Level • Grand Region CANS Adherence analysis
This a secondary disruption analysis which explores the probability of disruption conditional on whether an individual was placed in a level of care matching the CANS Recommendation level.
The Hazard Ratio (HR) indicates the change in risk of disruption, relative to a baseline category, associated with a one unit increase in a variable, while holding the values of all other variables constant.
The Kaplan-Meier Plot provides an estimate of the proportion of children that have not yet had events (disruptions) at each time. Since all children are not observed for the same amount of time, the estimate at each time point is based on the proportion of the population that is 'at risk.' All individuals begin in the 'at risk' population and remain there up until (and including) the time of disruption or their last record. The Kaplan-Meier curve can be interpreted as an estimate of the percentage of the original cohort of children that is disruption-free at each time period, using all data up to that time period, rather than the crude percentage without disruption at each point.
It should be noted that as the number of individuals at risk decreases, the precision of the estimate similarly decreases, though it remains unbiased. Additionally, while the Kaplan-Meier curve is stratified by CANS Recommendation Level and Strata in the 'CANS.Disruption' and 'CANS.Adherence' analyses, respectively, it does not control for covariates such as 'Age' or 'Race' specified in the 'Survival Analysis' tab.
Heatmap of the percent of records receiving CANS assessment within 30 days of custody date for fiscal year 2011 to fiscal year 2018.
Forest plot summarizing regression estimates of the association between covariates and no CANS assessment within 30 days. For each covariate, the first level is considered the baseline to which subsequent levels are compared in the estimated odds ratios (OR). Black dots in the graph represent the odds ratio point estimates and the grey bars represent 95 percent confidence limits.
CANS placement level recommendations are categorized into one of four levels, roughly equivalent to kinship/foster care (level 1), theraputic foster care (level 2), congregate care/residential treatment (level 3), and subacute hospital/inpatient care (level 4). The heatmap and table below show the percent of CANS recommendations falling into levels 3 or 4, stratified by region. The black points on the heatmap indicate the physical location of the level 3 and level 4 treatment facilities.