|
|
|
|
Over the years, I've recommended optimizing laboratory QC for each individual test as the most cost-effective way to manage the analytical quality in your laboratory. That approach allows you to account for the test quality required by your patient population, the imprecision and inaccuracy observed for the methods in your laboratory, and the known rejection characteristics for the control rules being used.
I've also advocated the use of multi-stage QC designs to deal with difficult applications where it's impossible to achieve high error detection and low false rejection with a single QC procedure. The multi-stage strategy is to have one design for high error detection and another for low false rejection, then to switch back and forth depending on which design is more appropriate at the time. A practical multi-stage approach is to initially use a high error detection design when starting up an analyzer and then switch to a low false rejection design to monitor ongoing performance. See the Sage Advice discussion on this website for more information.
Many analysts have been stymied in their efforts to implement these approaches because of limitations in the QC software available on their instruments, data management work stations, or laboratory information systems. To give you hope for the future, here's a preliminary description of our new EZ Runs program.
EZ Runs is a continuation of the development of our Validator technology, which first demonstrated the feasibility of an automatic design process in the QC Validator 2.0 program and then demonstrated a new user interface in the EZ Rules program. EZ Runs makes use of the new user interface and the question/answer format for the setup functions that are performed infrequently. For easy navigation, EZ Runs also makes use of a "folder" format. For efficient data analysis, EZ Runs provides a single data form to record control results, display them graphically, and provide error flags when results are out-of-control. Simplicity and power - those are the benefits of EZ Runs.
EZ Runs is available in a standalone PC version that allows manual entry of data and also as a Dynamic Linked Library (DLL) that can be embedded in other software. The DLL version makes it possible to provide automatic acquisition of the control data, as well as automatic processing to determine control status on an ongoing basis.
Here's a walk-through of the program showing how to setup a new test, setup control materials, select QC rules, and enter control results. This walk-through includes most of the screens that would be used in the standalone PC version. The DLL version would provide access to appropriate screens, depending on the information that needs to be entered into EZ Runs.
All analysts are identified through their own user password, which is initially entered as part of the signon form. Supervisors have special access to the setup functions that are used initially when beginning with the program and then only periodically for maintaining ongoing operation.
Notice the organization of the screen. The folders at the top let you navigate easily between the major functions of the program. Typically the Setup Test, Materials, and Rules functions would be restricted to supervisors or lead analysts. The Enter data and Actions are routinely used by analysts. Both analysts and supervisor will make use of the View Charts and Reports functions.
This particular screen is for entering descriptive information about the test, units, method name, manufacturer, and test/method code, if applicable. When EZ Runs is integrated into other software, this descriptive information will most likely already exist and can be transferred to EZ Runs.
Notice the format of the screen. Within the Materials folder, there are a series of screens that request information. This requested information is entered by the user in the appropriate places. The bottom of the screen provides instructions that pertain to the information on this screen. In this case, it's information about the identity of a control material, including manufacturer, lot number, and expiration date.
For each control material, you can enter a manufacturer's "Labeled" values, your own "Assigned values," or you can indicate values to be calculated from a "Fixed interval", "Moving interval," or "Cumulative" to date. Typically, you might begin by assigning values while you're collecting the initial results for calculating your own values. Then you would change the setting to a fixed interval for the first collection period, then later change to a moving interval or to cumulative values.
Here's where you would actually indicate the mean and SD that are to be used in setting up the control chart and calculating the control limits. You select from a drop-down list that contains all the possible settings. You can have different settings for the mean and SD, i.e., you could initially use a labeled mean and a user assigned SD.
The first screen asks whether you want to use automatic selection or manual selection. Automatic selection is an option that can be added to enhance the basic EZ Runs program. The automatic selection makes use of the EZ Rules program with some minor changes to reduce the number of parameter entry screens. See the description of EZ Rules. In the demonstration here, the manual selection option is shown.
EZ Runs contains a long list of single-rule and multi-rule procedures, approximately one hundred combinations of rules and Ns. You select a particular QC procedure by scrolling down the list, then clicking on the one you want to implement. In this example, a multirule procedure with 3 control measurements per run is selected for the STARTUP design.
Multi-stage QC is another option that's available in EZ Runs. Multi-stage is a QC strategy that's employed when it becomes impossible to get both high error detection and low false rejection from a single QC procedure. Multi-stage QC refers to the use of two or more QC designs, e.g., a STARTUP design that is optimized for high error detection and a MONITOR design that is optimized for low false rejections. The idea is to switch back and forth between the designs, depending on whether it's more important to have high error detection or low false rejection. It's a simple matter to utilize multi-stage QC designs in EZ Runs. Just repeat the selection cycle, using either manual or automatic selection.
Here's a second design that is selected as a MONITOR design. In this example, the rules selected happen to be ones that are regulatory requirements in a particular country. The 7t refers to a "trend" rule where control measurements increase systematically or decrease systematically over a series of 7 measurements.
The selected QC designs are automatically available and ready for use in assessing your control results. The data entry form includes automatic entry of date, time, and analyst, selection of the QC design from a drop-down list, selection of the control material from a drop-down list, and entry of control result, in this case manually. The z-value is calculated automatically and the control result plotted on the chart at the right, which is control chart displayed vertically, rather than the traditional horizontal display. Two QC designs are illustrated in this screen capture, as shown by the different control rules on the chart. Example error flags are shown for a 31s rule violation at entry 44 and 7t at entry 56.
Any violation of a control rule triggers a message box with choices to correct/review the entry or go to the action log to document correction action. The action log contains the entry number, date, time, and analyst identification, along with a free text message that is entered by the analyst. Out-of-control data are automatically flagged in the data entry form and are excluded from any further calculations of the data.
The analyst and/or supervisor can also display and review control data in the format of the traditional Levey-Jennings control chart. The time sequence of control data is maintained in this display, i.e., the points are plotted sequentially versus the entry number (or versus the date if that option is selected). Note that there is no additional flagging on these control charts - all flagging takes place on the data entry form for immediate notification, review, and action by the analyst.
A flexible report generator provides many options for the data and time period to be included in a report. An essential part of the documentation are the control rules and numbers of control measurements that are employed in the different designs. This first page of a report shows the STARTUP and MONITOR designs that were used as examples here. The entire data entry form, action log, and individual control charts can be printed, along with a comments page that provides free text entry.
