Outliers

Note: JMP Clinical uses a special protocol for data including non-unique Findings test names. Refer to How does JMP Clinical handle non-unique Findings test names? for more information.

Running this report with the Nicardipine sample setting generates the report shown below.

The Report contains the following elements:

Shows a volcano plot used to determine the deviation of the proportion of outliers at each study site with the overall proportions for the whole study. The X axis is calculated using p(site) - p(reference), where p(site) is the proportion (or percent) of outliers in the site in question and p(reference) is the proportion of outliers in all other sites.

The volcano plot is used to determine the deviation in the proportion of outliers at each site from the proportion of outliers at all of the other sites. The X axis represents the difference in proportions for measurements that are considered outliers between the site and the reference set (all other sites).in the proportion of outliers at each site from the average proportion at all of the other sites.

You can mouse over points of interest to see their labels or select points by dragging a mouse rectangle over them (as shown above). Selected rows are highlighted in other plots and you can also then click on various Down Buttons for further analyses on those specific sites. For example, if you click the Outlier Bar Charts (

) drill down (circled above), the following opens with comparisons of sites 3 and 40 with all sites not site 3 or 40, respectively.

In these plots, 0 represents findings that are not outliers and 1 represents findings that are outliers. Both sites have an excessive number of outliers for QRS Duration.

Displays histograms showing distribution of ECG findings across study sites, tests, and visit numbers, as well as the frequencies and types of outliers.

This section contains the following elements.

•

Distributions: Displays histograms showing distribution of ECG findings across study sites, tests, and visit numbers, study sites, categories and tests. The last two histograms show the relative proportion of outliers and whether they are outliers because the value is too high (Outlier: Above) or small (Outlier: Below). Clicking Show Subjects will open a data table that will show the range of non-outlier values (Lower and Upper Outlier Limits).

Displays histograms showing distribution of laboratory test findings across study sites, tests, and visit numbers, as well as the frequencies and types of outliers.

This section contains the following elements.

•

Distributions: Displays histograms showing distribution of LB findings across study sites, tests, and visit numbers, study sites, categories, and tests. The last two histograms show the relative proportion of outliers and whether are outliers because the value is too high (Outlier: Above) or small (Outlier: Below). Clicking Show Subjects will open a data table that will show the range of non-outlier values (Lower and Upper Outlier Limits).

Displays histograms showing distribution of vital signs findings across study sites, tests, and visit numbers, as well as the frequencies and types of outliers.

This section contains the following elements.

•

Distributions: Displays histograms showing distribution of VS findings across study sites, tests, and visit numbers, study sites, categories, and tests. The last two histograms show the relative proportion of outliers and whether are outliers because the value is too high (Outlier: Above) or small (Outlier: Below). Clicking Show Subjects will open a data table that will show the range of non-outlier values (Lower and Upper Outlier Limits).

Action buttons, provide you with an easy way to drill down into your data. The following action buttons are generated by this report:

•

Show Sites: Shows the rows of the data table for the selected points from the volcano plot. Clicking

opens the following table:

•

Frequencies Bar Charts: Select points on the volcano plot and click

to display a bar chart for each point showing the proportions of outliers for the site as compared with the average of all of the other sites.

•

Profile Subjects: Select subjects and click

to generate the patient profiles. See Profile Subjects for additional information.

•

Click

to view the associated data tables. Refer to View Data for more information. Note: Numeric data for individual subjects come directly from the xxSTRESN¹ variable from the domain data sets.

•

Click

to generate a standardized pdf- or rtf-formatted report containing the plots and charts of selected sections.

•

Click the gray border to the left of the Options tab to open a dynamic report navigator that lists all of the reports in the review. Refer to Report Navigator for more information.

This report compares the outlier status of tests with each site compared to all other sites taken together as a reference. A point is an outlier if it exceeds the upper outlier limit (Q3+1.5×IQR) or is less than the lower outlier limit (Q1-1.5×IQR), where Q1 and Q3 are first and third quartiles and IQR =Q3-Q1. It then compares the rate of outliers using Fisher's exact test. FDR p-values are calculated and the reference line is determined as described in How does JMP Clinical calculate the False Discovery Rate (FDR)?.

By default the report is set to analyze all tests from all findings domains. You can opt however, to restrict the search to specific Findings Tests.

You can specify whether to search all results in original units when you Analyze: the data or restrict the search to either character results (in standard format) or numeric results (in standard units).

Unscheduled visits can occur for a variety of reasons. By default, these are excluded from this analysis. However, by unchecking the Remove unscheduled visits box, you have the option of including them.

You can opt to Consider BY variables in the analysis. This option, which assumes that BY variables (left vs. right arm for collecting blood pressure data, for example) are included in the experimental design, is selected by default. You can uncheck this option to ignore BY variables.

Use the Only include BY variables if they are domain keys option to subset the available variables to only include those variables that are domain keys. If the option is unchecked, the report uses the cross-classification of xxCAT, xxSCAT, xxLOC, xxMETHOD, xxPOS, xxSPEC, and xxTPT for creating by groups for all variables that are available (as it had in the past).

The Summarize sites with at least this many subjects: option enables you to set a minimal threshold for the sites to be analyzed. Only those sites which exceed the specified number of subjects are included. This feature is useful because it enables you to exclude smaller sites, where small differences due to random events are more likely to appear more significant than they truly are. In larger sites, observed differences from expected attendance due to random events are more likely to be significant because any deviations due to random events are less likely to be observed.

The Alpha option is used to specify the significance level by which to judge the validity of the outliers revealed by this report. The meaning of alpha depends on the adjustment method that you select. Alpha can be set to any number between 0 and 1, but is most typically set at 0.001, 0.01, 0.05, or 0.10. The higher the alpha, the higher the error rate but also higher the power for detecting significant differences. You will need to decide on the best trade-off for your experiment.