RAPID Frequently Asked Questions

Can I install this on my desktop/laptop computer? Why is it necessary to install RAPID as a server?

RAPID stores protected health information (PHI). RAPID is designed to be HIPAA compliant and meet meaningful use criteria, if you decide to certify your installation of RAPID. Because of the very significant security requirements for PHI, RAPID was designed from the beginning to meet these standards. Furthermore, there are extensive access logs should a security audit be necessary.

It is possible to install this virtual machine (VM) on a desktop/laptop machine but is strongly inadvisable as it likely violates your institution’s security policies for PHI.

What additional data does my center have to provide to RAPID? Do I need to manually enter data?

RAPID currently only loads data you have already submitted to UNOS, and SRTR PSRs. No additional data are necessary.

How often should I run the RAPID dashboards?

RAPID is updated as often as the underlying data is updated. The UNOS data is released once a month and the SRTR PSR should be released once every 6 months.

Does RAPID share data or does it remain private? Where is the data stored?

RAPID is a completely closed system that does not require any internet access to operate, except to link to ASTS documentation web pages. No data is uploaded, or is accessible, to any other system. RAPID is not cloud-based and stores all data locally within the virtual machine. There are no “backdoors” for the developers to access a RAPID installation once you change the administrator password.

Can I change the dbuser password?

Starting with RAPID 2014.5, we have substantially changed user accounts. There is now support for multiple users. There are several “back end” and administrative user accounts that allow reports, data loading, and system administration to be accomplished. Accounts and default actions are outlined below.

New user accounts in RAPID 2014.5

SQLEDW – User account used to run SSIS jobs, ability to login is disabled

ReportViewer – Windows User (not an admin account) with ability to view Dashboard Reports either via remote desktop connection or Web Browser. User will be required to change the password at first login.

DataLoader – Windows User (not an admin account) with ability to view Dashboard Reports and Load Data either via remote desktop connection or via Web Browser. User will be required to change the password at first login.

Existing users accounts in RAPID

dbuser (admin account) - User will be required to change the password at first login.

administrator (admin account) - User will be required to change the password at first login.

Is this software truly free?

All the software that Northwestern has developed is free, and ASTS will not charge you a fee to download the software. If you intend to use this software outside of evaluation testing (that is, for actual production or enterprise use), you will need to purchase a Microsoft Windows Server and a Microsoft SQL Server license. In many instances, your institution already has some existing license for both software and it is straightforward to piggy back on the institutional license at a pre-negotiated fee. Beyond software licensing, any IT support costs are your own.

Can I modify this software?

You are free to modify the software in whatever fashion that you would like, for example, to implement your local security policies and procedures or to create new dashboards and reports. You cannot sell the software. If you modify the software, you are bound by the GNU Affero General Public License version 3 <https://www.gnu.org/licenses/agpl-3.0.html>, which indicates that any modifications must be made freely and publicly available. Dr. Bing Ho at Northwestern can step you through the process.

What other conditions are there to use the software?

We ask that you register before downloading the software and that you cite RAPID in any manuscripts. The citation is:

Bing Ho (1), Nathan Sisterson, Ania Nassiri, Prasanth Nannapaneni, Christopher Mitchell, Anna Pawlowski. RAPID (Real-time Analysis and Process Improvement Dashboard) – an analytics platform for solid organ transplantation. 2014. Northwestern University, Chicago, Illinois, USA. URL http://www.asts.org/rapid. (1) Corresponding author.

Why doesn’t the green dot match the blue dot?!

There are four potential sources of error between SRTR (green) and UNOS (blue) data.

  1. Methodology. We (the developers) may have misinterpreted the published SRTR methodology or there is unpublished methodology that affects your SRTR report. If you believe that there is a methodology, please contact the developers and we will investigate with you and the SRTR.
  2. Cohort size. If cohorts do not match the SRTR cohorts, there may be variable effects on observed and expected depending on the missing patients. The two most likely scenarios is that a patient was a previous transplant recipient from another center (they contribute to graft survival but not patient survival) as this information is not found within the UNOS data set (which includes only transplants done at your center), a pediatric-listed patient was transplanted at your adult center, or there was a methodology error.
  3. Observed. The SRTR uses additional data sets beyond UNOS data for the PSR – two major sources are SSDMF and USRDS. SSDMF may provide patient deaths that you have not reported to UNOS, and USRDS will provide data on patients that have returned to dialysis that you may not have reported.  RAPID supports SSDMF if your institution has access to it (please contact the developers for instructions on loading it). Updating your UNOS data with these events will increase the accuracy of the projected PSR.
  4. Expected. The expected number of events is the sum of over 40 variables that combine data from multiple data sets. Small differences in these variables will translate into potentially significant differences in projected expected and actual SRTR PSR expected. For example, for a medium sized program transplanting 60 patients a year (150 patients in the 2.5 year cohort) with an average expected of 0.1 for each patient, if the expected calculation is off by a mere 5%, there will be a total error of 0.75 expected. Depending on the direction of the error, this could lead to the O-E and O/E being off by a considerable margin. Larger programs will experience greater error. The most common differences between UNOS and SRTR data that we are seeing involve recipient year of ESRD and peak PRA. Updating your UNOS data will enhance the accuracy of the projected PSR.

Lastly, if you correct a data point in UNOS after a PSR is made publicly available, the SRTR will continue to use the originally published value. Your UNOS data could be either more or less accurate than the SRTR PSR!

What is the difference between using UNOS and SRTR data in projected PSR reports?

Using just UNOS data will allow you to understand circumstances where your UNOS submitted data differs from the SRTR PSR. This will be helpful during draft PSR periods where you can still correct any discrepancies. If you choose to use published SRTR data (when available) in lieu of the UNOS data, your near-term projected PSR statistics will likely be more accurate at the expense of understanding why your UNOS data differs from the SRTR data.

How do I get the reports more accurate?

RAPID is limited by the accuracy and completeness of the data you have submitted to UNOS. If you notice any discrepancies between UNOS and SRTR data, you can update your data one of two ways. During the draft period before any PSR public release, you can correct errors with UNOS and SRTR. Outside of the draft period, any corrections to the UNOS data will improve the following SRTR PSR.

We need more support than just downloading and starting the VM! It would be great if we could just implement this one tiny feature. Can we contract with Northwestern for custom development and support?

Yes, if you are an academic or nonprofit organization, the developers, administrators, architects, analysts, and statisticians can be contracted at a standard, internal hourly research rate for custom development. Please contact Bing Ho at Northwestern University if you want further details.

Why are the confidence intervals so large?

To paraphrase the US Census Bureau, a confidence interval is a range of values that describes the uncertainty that surround an estimate. If we were to repeatedly make new estimates using exactly the same procedure (by drawing a new sample and calculating new estimates and new confidence intervals), the confidence intervals would contain the average of all the estimates 95% of the time. As the sample size increases, the possible range of estimate (O-E, O/E) is larger and larger. The small confidence interval for the small incomplete cohorts indicates the range of estimates for a cohort of that size, not the anticipated cohort size when the cohort is complete.

How do you calculate the risk adjustment?

The beta coefficients used to calculate the risk adjustment are derived from the complete national data set. In order to calculate the risk adjustment for future cohorts, we use the last published beta coefficients in your PSR. Although the beta coefficients do change over time, they typically do so very slowly and serve as reasonable estimates.

How was the development of RAPID funded?

RAPID’s development is supported internally by the Northwestern Medicine Comprehensive Transplant Center.