Quality management of bedside glucose testing.
FIVE YEARS AGO, authors Belsey and Baer discussed the technologist's role in the quality management of bedside testing.(1-3) After several years' experience, we would like to share a few observations about the development and operation of the bedside testing programs at our respective facilities. We will be discussing some elements that have worked for us and a few that have been less successful.
* Improved control. Why is capillary blood glucose testing so important? In authors Baer's and Eyberg's hospital, the Veterans Affairs Medical Center (VAMC), a 500-bed facility in Portland, Ore., about 17% of patients have diabetes. At times of stress, change in activity or diet, or concomitant illness, diabetic control can become problematic.
Blood glucose testing performed several times during the day and night, with subsequent adjustment of insulin, has been shown to improve the control of diabetes.(4) Testing reduces the duration of the patient's illness and hospitalization during acidosis.(5)
Test results, especially in patients receiving insulin and those who have ketoacidosis, need to be timely as well as accurate. Thus, the lower turnaround time associated with bedside results needs to be balanced with the precision and accuracy that can be achieved by the hospital's central laboratory.
* Capillary glucose. To facilitate this balance, bedside capillary blood glucose testing has become a standard of care in hospitals. Laboratories are in a position to assist nurses in identifying the limitations of bedside testing methods while achieving reliable test results.
Our previous bedside testing articles in MLO encouraged a multistep process that involves establishing an ancillary testing committee responsible for considering issues surrounding the development of a bedside testing program.(1-3,6) We proposed that the group identify current problems concerning result reliability, patient care needs, risk management, and potential cost and reimbursement. The group should establish evaluation criteria, including clinical reliability of test results and clinical reliability standards for new analytic systems.
We felt that the most visible portion of the group's work is the development of a quality management program to promote the consistent production of results that meet clinical reliability standards. We recommended that the quality management program be primarily the responsibility of the nursing department, with technical assistance from the laboratory. The lab should advise on the use of quality control (QC) materials, establish a QC reporting mechanism, analyze quality control results, and report the conclusions of the analysis to the appropriate nursing department leaders.
At VAMC, we have used one type of capillary blood glucose instrument for several years. This reagent test strip device electronically collects QC data for control tests at several different glucose levels. It stores these data for later recall, download, and analysis. The instrument also allows the entry of QC limits. When these limits are exceeded, the instrument gives a visible and audible warning that clearly identifies out-of-range tests.
* Nurse mistakes and noncompliance. The biggest problem that we encountered with such a system was the nurses' lack of compliance in performing the quality control tests. The policies in our two hospitals require that each day a quality control test be performed successfully before testing a patient. VAMC requires the testing of a single elevated level of quality control material, while authors Belsey's and Petersen's facility, the 400-bed Oregon Health Sciences University (OHSU), requires a low and elevated level.
A second problem was the need for nurses to enter an identification number into the instrument. In the same way, the nurses were required to indicate whether a specimen belonged to a patient or was a QC or proficiency testing (PT) analyte. We found that even nurses who were highly motivated to comply made frequent errors entering these data. Consequently, it has been difficult for us to analyze the results and accept them as valid statistics.
Finally, the nurses in both our hospitals have maintained a skeptical attitude about the test system. They do not understand the need for QC; they view it as an unnecessary, non-patient care activity, and are passionately outspoken about their belief that patient test results are reliable without quality control.
* Technical advances. Since our first experiences with bedside testing quality management, technology has advanced and we have made some equipment changes. Several reagent strip systems have been introduced that do not require wiping off the specimen prior to reading the results. This eliminates numerous sources of error, including using poor technique in wiping off the blood specimen and inaccurate timing.
One new system is able to distinguish between a red blood specimen and a blue control specimen. It correctly flags each in the electronic memory of the instrument. This system was put into use at the VAMC in the summer of 1991.
* Quality management. At VAMC an ancillary testing committee was developed in 1987 to sanction, organize, and oversee all testing performed outside the clinical lab.
Members of the committee included author Baer, who is the chief of laboratory service, an endocrinologist, the nurse diabetes educator, and several representatives of the nursing service. According to VA rules, the chief of laboratory service is responsible for the quality of all testing in the medical center, whether it is performed in the clinical laboratory or in an ancillary lab unassociated with the laboratory service.
The ancillary testing committee established medical center policies concerning bedside testing and secondary laboratories. The committee also developed procedures for the quality control of bedside capillary glucose testing that include the following requirements:
* One nurse per nursing unit is responsible for the maintenance of the bedside testing equipment.
* Each morning that nurse must clean the optical surfaces and check to see that reagents are present and that the reagent strip code is the same as that designated by the instrument.
* Every nurse performing capillary testing must run one QC test using a high control material on any day that he or she does patient testing.
We emphasize that quality control testing must be performed by each operator since, in our opinion, the operator is the most variable factor in testing.
* Recording data. Each time an assay is performed, the meter automatically records the test number, test result, time, date, and whether the result was from a patient or a control specimen. The instrument senses the latter data by noting whether the sample was red (patient) or not red (control).
Every meter can hold 250 results in memory. Once a month (more often when a heavy testing load requires it) author Eyberg, the ancillary testing QA coordinator, downloads the meters' memories to a lap-top computer. He identifies in the computer the location of each meter and prints a consolidated report for all hospital meters upon returning to the laboratory.
After downloading and printing the report, Eyberg visually scans through the reports for the following information:
* Noncompliance with QC testing procedures, indicated by the lack of a control test result prior to a patient result on any nursing shift.
* Out-of-range QC test results, indicated by a control value outside the expected range.
* No follow-up on outliers, indicated by an out-of-range control report without an immediate, subsequent, within-range report made prior to patient testing.
After reviewing the report, the coordinator prepares a memo for the nursing unit supervisor. We use a printed form on which Eyberg checks one of the following options:
* Monthly QC data look good. Appropriate number of QC tests were performed. QC test results outside expected range were followed up with satisfactory results.
* Greater than expected number of QC tests were outside range for normal results.
* Discrepant QC values not followed up with satisfactory results prior to patient testing.
* No QC tests performed on shift prior to patient testing.
* Abnormally low QC values observed. TABULAR DATA OMITTED Suspect inadequate patient specimens read by meter as control solution.
* No date of opening on high control or test strip container.
The form includes space for the nursing service to document corrective action, if required. The report is then forwarded to the nursing unit supervisor, who follows up with appropriate action.
* Manual log. The individual nurses performing QC testing are also required to enter information on a manual quality control log. The log serves the dual purpose of meeting competency certification requirements of the Joint Commission on Accreditation of Healthcare Organizations (JCAHO).
Not only must nurses enter the date, the time, and the results of QC testing, they must also judge whether the result is in the appropriate quality control range. If it is not, they must show what remedial action was taken.
Remedial action codes at the bottom of the page provide a guide. Also at the bottom of each page is a signature line for the unit supervisor to certify that the nurses entering QC results on that page are competent to continue testing.
* Meeting JCAHO rules. The JCAHO requires hospitals to test on a quarterly basis the competency of every nurse who performs bedside testing. Fortunately, JCAHO allows successful QC testing to serve as certification of competency. Signing the bottom of each quality control test result page testifies to a nurse's continued competency, thereby fulfilling the JCAHO requirement. The laboratory's ancillary testing QA coordinator also signs the log, to show that he or she also reviewed the data.
In addition to providing a monthly report to each unit supervisor, Eyberg gives monthly reports to the director of the nursing service and to the chief of laboratory service. All meters are inventoried by the hospital's biomedical engineering department, which services or replaces defective ones and keeps maintenance records.
* Deficiencies. We recognize deficiencies in the VAMC program but, at the present time, have decided not to attempt a more ambitious approach. So doing would require the introduction of prohibitively expensive, new technologies or procedures. Known deficiencies of our program are:
* Patient name and other data are not entered into the testing log.
* Only one concentration of QC material is tested.
* Instruments sometimes record an insufficient patient specimen as being a QC result that registers in the very low range.
* The OHSU program. In 1988 an ad hoc committee recommended that bedside testing continue only if a quality improvement program was initiated and an implementation group established. The implementation group--the bedside testing committee--included representatives from nursing service, nursing quality assurance (QA), nursing education, and the clinical laboratory. A designated laboratory bedside testing coordinator (author Peterson) was identified and initial training sessions for our new meter system began in December 1990.
* Early disaster. The 21 wards and clinics, involving approximately 900 nurses, were trained over a 6-month period. At the onset, training was performed by representatives from the manufacturers of two instruments. Then nurses were hired from an outside agency to replace the company representatives and continue the training process. This proved to be a bad move; since the agency staff did not fully understand the instrument, they were unqualified to answer specific questions and communicated misinformation to the entire nursing staff.
The ramifications were enormous. Many retraining sessions had to be conducted by the bedside testing coordinator and the original company trainers. The early results obtained by the inadequately trained staff were abysmal and a skeptical attitude about the test system flourished throughout the OHSU nursing staff.
* Testing skills. After the problematic introductory hands-on training demonstration, each nurse was assigned a unique operator number. Since our meters store only 250 such numbers, it was necessary to identify users first by location, then by individual. Every operator was then required to perform an initial proficiency test that consisted of 10 unknown specimens (commercial ones provided by the laboratory).
The results were returned to Peterson, who evaluated the 10 results against the test criteria limits--|+ or -~25% TABULAR DATA OMITTED from the mean value. Those who had 3 or more of the 10 results outside this range repeated the exercise. The training documentation was kept by the coordinator.
* Key operator. In December 1991, 1 year after implementation, a "key operator" was identified for each testing unit or area. This individual (a technically oriented person other than a department director) became the local expert, overseeing instrument maintenance and inventory as well as ordering reagents and reviewing the monthly quality control reports generated by the bedside testing coordinator.
The key operator notes any action to be taken on the QC report, signs it, and sends a copy to the nursing QA coordinator, who, in turn, sends a quarterly report to the hospital quality management department. This feedback loop is critical to monitoring effectively each nursing unit's compliance with the program's QA criteria.
* Building ownership. The addition of a key operator accelerated the flow of information and improved communication channels between the coordinator, the departments, and the individual staff members. Pride of ownership in the program blossomed cautiously in many of the key operators and extended slowly throughout the staff.
To increase the nurses' level of confidence in the system and assuage their anxiety at treating a patient based on what they perceived to be an inaccurate test result, a non-billed "glucose check test" was introduced. In this program, nurses draw paired specimens to be analyzed on the reflectance meter and sent to the lab for a comparison value. The lab technologist then enters the meter and lab value in the computer and calculates the percent difference between the two. In this way we achieve relatively immediate comparison feedback. (The laboratory instrument chosen for the comparison test analyzed the specimen by a whole blood method, thereby maintaining specimen consistency with the bedside method.)
* Outliers. While the feedback loop was meant to increase acceptance of the instrument, in actuality the occasional discrepant values increased the nurses' skepticism to the point where the close comparison values were overshadowed, in their minds, by the outliers. The experience, however, had the advantage of exposing the nursing staff to a common lab reality hereto unforeseen by our nurses--the implication of different test methodologies and instrument precision variability.
Nursing units that had previously followed a tight insulin sliding scale range were confused about dosage requirements since they expected the reflectance meter to match the precision and accuracy of a central laboratory instrument. On the whole, the comparison data has improved since this check test went into operation.
* Required QC. Continuous improvement prompted the bedside testing committee to recommend a decrease in required QC to one time per shift for nurses who are scheduled to perform patient testing.
Interestingly, after hearing the committee's recommendation to decrease QC, many key operators chose not to share the recommendation with their staffs. They believed it wise to maintain the program as is. Given the minimum level of compliance in performing QC, this was a prudent decision.
The coefficient of variation for high glucose control materials is 7.7% at VAMC and 12.9% at OHSU. These figures represent all QC test results from both institutions (excluding obvious outliers more than 100 mg/dl away from the mean value).
* Report formats. In our two hospitals, we have experience both with visually evaluated and computer evaluated reports for QC testing. Each format has its strengths and weaknesses. The visually evaluated reports are produced by a meter that is unable to identify individual nurses. When a problem arises, it is necessary to rely on a comparison between the electronically downloaded data and the manually produced log.
Such a comparison, however unwieldy, has the advantage of logging accurately every testing event, whether it be a patient or a QC specimen. From a comparison, it is possible to determine accurately whether there is compliance with the quality control testing requirement and whether unacceptable QC tests are (correctly) repeated before patient testing. We are unaware of any computer program that can assess such parameters.
Computer analysis of quality control testing, developed by author Belsey, depends on a system that identifies accurately both the location of the QC test and the nurse who performed it. The computer prepares sophisticated reports for the individual nurse, unit manager, and laboratory or nursing service QC coordinator. The program accurately identifies those nurses and units with good performance and those in need of additional training. Weaknesses of the system include:
* The requirement that the testing nurse enter a QC result in the computer's memory, using specific keystrokes, at the time of testing.
* Errors by nurses due to incorrectly identifying high, normal, and low controls.
* The inability to capture patient data.
Presently, the computer system does not automatically allow the quality control coordinator to assess whether there has been compliance with quality control testing prior to patient testing.
* The program's cost. Managing quality for bedside testing is expensive, both in terms of personnel and supplies. We estimate that the laboratory ancillary testing quality assurance coordinator spends about 3 days per month inspecting, downloading, and analyzing QC data. OHSU has recently begun limited proficiency testing (PT), which is terribly time-consuming.
The labor investment by nurses is also considerable. It is extremely time-consuming to train, test for competency, and document each of the nurses who performs bedside testing. We estimate that one to two nursing FTEs are required to run and document daily QC testing. Patient testing probably requires the time equivalent of three FTEs in each institution. In addition, supplies costing between $30,000 and $40,000, including reagent strips and control materials, are required for bedside glucose QC testing throughout each of our institutions per year. As an example of overall cost, OHSU spends about $60,000 each year on the quality control of bedside glucose testing. The equivalent annual expense for VMAC is about $50,000.
* CLIA '88. So far the accreditation requirements of JCAHO have been the major force in shaping our glucose testing quality management programs. It is still too early to know whether CLIA '88 will place new requirements on our programs.
OHSU recently enrolled in a PT program and for several months has sent out an informal proficiency testing specimen to nurses. The VAMC's bedside testing program has been inspected by CAP. Because of CAP's requirements, we will be instituting a PT program shortly. Not every whole blood glucose PT material is suitable for specific glucose instruments. Be careful when selecting a PT program to insure that it is compatible with your test system.
* Differing complexity. The uncertainty about the impact of CLIA '88 is partly due to the differing complexity classifications of the various glucose testing systems in use in hospitals. For example, the system in use at VAMC is approved for home use by patients and, therefore, is in the wavered complexity category. There are no CLIA '88 requirements that apply other than to follow the manufacturer's instructions (although the VA has more stringent quality management requirements that we follow).
The test system in use at the OHSU is designed for hospitals and has not been approved for use by patients at home. It is classified as a moderately complex system, so CLIA '88 rules apply. We expect that, in the future, variations in the complexity for similar test systems will disappear as manufacturers and the Federal government work to change system configurations, the approval process, and the complexity model. (JCAHO has indicated that its requirements hold for all test systems regardless of the CLIA complexity level.)
* Future wish lists. At the present, no instrument on the market provides all of the features that we believe are necessary to assist in managing the quality of bedside capillary blood glucose testing. Here are some of the hardware features we would like to see:
* A single-step analytic system that automatically times the reaction (starting with the application of the blood specimen) and does not require wiping off the specimen. The test system should sense accurately when an inadequate specimen has been applied, when the reagent strip has been inserted improperly, and when a deteriorated reagent strip is used.
* Automatic recognition of QC specimens as being different from patient specimens. The instrument should sense whether the quality control specimen is high, low, or normal. The meter should accurately flag the QC specimen and segregate that reading in memory.
* Operator identification without the need to key in a code number. If the operator must enter a code, the button sequence must be simple and a required function at the time the instrument is switched on. In fact, the instrument should be inoperative without the operator code.
* Patient ID so that test results can be downloaded to a computer and a bill can be generated.
* Sufficient memory to allow downloading at relatively infrequent intervals (such as once a month). In our experience, a 1,000-test memory is probably required. We would prefer to have an instrument capable of storing the current and previous month's data.
* Software dreams. We also have several items on our software wish list. Download and analytic software features need to be developed. We have found that it is better to bring a laptop computer to the instruments than to bring the instruments to a central location for downloading. For one thing, doing so eliminates meter down time. Download software should have the following characteristics:
* Simple and rapid downloading.
* A chronologic printout of all events so the reviewer can see the relationship between quality control and patient testing.
* Control tests segregated in different ranges, allowing easy statistical review.
* Analysis by operator, shift, and location. In that way we can better identify those users with both adequate and inadequate performance.
* Identification of shifts during which patients were tested but no QC was performed. This software feature must be sufficiently flexible to accommodate unconventional shifts (including 10- or 12-hour workdays).
* Isolation of instances where out-of-range quality control results were not followed by within-range quality control results prior to patient testing.
* Preparation of reports for nursing supervisors.
* Continued reliability. The current generation of blood glucose analyzers and quality management software has improved our ability to produce accurate, timely, reliable capillary blood glucose test results and to manage effectively the quality TABULAR DATA OMITTED of capillary blood glucose testing.
These instruments help us develop documentation and analytic tools to demonstrate that non-laboratory personnel are following good quality management practices. Analytic instruments and quality management computer software still need further development and refinement to make the process less labor intensive and more effective, however. The ultimate goal is continued collaboration between nursing and the laboratory.
1. Belsey R, Baer DM. Protocols for bedside testing, Part 1. MLO. February 1988; 20(2):37-39.
2. Belsey R, Baer, DM. Protocols for bedside testing, Part 2. MLO. March 1988; 20(3): 63-72.
3. Belsey R, Baer DM. Protocols for bedside testing, Part 3. MLO. April 1988; 20(4): 45-52.
4. Mazze RS, Shamoun H, Passmanter R, et al. Reliability of blood glucose monitoring by patients with diabetes mellitus. Am J Med. 1984; 77: 211-217.
5. Zaloga GP. Evaluation of bedside testing options for the critical care unit. Chest. 1990; 97: 185s-190s.
6. Baer DM, Belsey RE. Managing quality and risk of bedside testing. Perspec Healthcare Risk Management. Winter 1990; 10(1): 3-7.
Dr. Baer is chief of laboratory service and Eyberg is ancillary testing quality assurance coordinator at the Veterans Affairs Medical Center in Portland, Ore. Dr. Belsey is professor emeritus of pathology and Petersen is laboratory bedside testing coordinator at the Oregon Health Sciences University, also in Portland.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Bedside Testing, part 3|
|Author:||Baer, Daniel M.; Petersen, Juanita; Belsey, Richard E.; Eyberg, Richard|
|Publication:||Medical Laboratory Observer|
|Date:||Jun 1, 1993|
|Previous Article:||How to cope with being fired.|
|Next Article:||Comprehensive orientation extends a warm welcome.|