College of American Pathologists

  Feature Story


cap today

Getting blood gas laboratories into tip-top shape

July 2003
Karen Titus

The latest updates to the Laboratory Accreditation Program’s
blood gas lab inspection recall a certain French adage: There’s change (Plus ça change), and, yes, much has stayed the same (plus c’est la même chose). But in a slight twist that conquers this Gallic inertia, there are improvements to be found throughout the checklists as well.

The grande change (and improvement): On March 31, the CAP accreditation program ushered in the era of custom checklists. The beauty of such checklists, said C. Robert Baisden, MD, is that labs will receive only checklist questions relevant to them. Custom checklists also mean “the laboratory does not see the same question over again in multiple checklists—for example, in the automated general chemistry checklist and the toxicology checklist and the specific chemistry checklist, and in the blood gas list,” said Dr. Baisden, Gulf regional commissioner for the Commission on Laboratory Accreditation. Eliminating this redundancy will save labs time when preparing for inspections and cut paperwork for inspectors.

Custom checklists will be sent to laboratories at the re-application and self-evaluation portion of the cycle, about 12 months after an inspection. “So many of you will begin seeing your custom checklists if this is the off-year for the on-site inspection,” Dr. Baisden said. The checklists will be used during inspections starting this September, he predicted.

March 31 also marked a number of chemistry-related checklist changes and refinements, which Dr. Baisden explored in an accreditation program audioconference May 14.

The chemistry checklists have been combined into a single chemistry and toxicology checklist, abbreviated as CHM. This single checklist contains questions from the former checklists, including automated general chemistry, toxicology, special chemistry, and blood gas laboratory. The toxicology, special chemistry, and blood gas checklists are no longer separate entities.

“All chemistry questions, including blood gas questions, will now be found only in the new combined chemistry and toxicology checklist,” Dr. Baisden said, adding that the disappearance of the blood gas checklist “will be transparent to laboratories performing only blood gas testing. The cover of the checklist will say ‘chemistry and toxicology.’ Questions inside the checklist will be the same as those in the former blood gas lab checklist, with updates.” Labs that perform arterial blood gas analysis—including special function labs, stand-alone blood gas labs, and those that do arterial blood gases in the main lab—will receive a customized chemistry and toxicology checklist that includes general chemistry and arterial blood gas-related questions.

A much-anticipated final ruling on quality control and personnel was published in the Federal Register on Jan. 24. The accreditation program and CAP’s scientific resource committees reviewed the ruling, which does not contain any changes to the CLIA QC requirements for blood gas testing. The Chemistry Resource Committee has recommended that the checklist requirements for blood gases be changed to match CLIA’s. “Therefore, beginning with the checklist edition effective March 31, 2003, the QC requirement will be: one level of QC each eight hours of patient testing for pH, pCO2, and pO2, and a low and a high level of QC must be run each 24 hours of patient testing,” said Dr. Baisden.

Blood gas question (BGL) .27465, concerning numeric QC data, no longer contains the term “Gaussian” when asking whether QC statistics are calculated at least monthly to define analytic imprecision. The term “was not technically correct in all situations,” Dr. Baisden explained. Moreover, not all laboratorians are familiar with the term, he said, which caused confusion in blood gas laboratories.

BGL.29780 asks whether at least one control or calibrator sample is included each time patient specimens are tested, unless automated instruments internally calibrate at least once every 30 minutes of use. The question has been updated to clarify that this applies to pH, pCO2, and pO2.

Change isn’t the only source of confusion when it comes to
checklists. Dr. Baisden expounded on common problems that plague labs, which are noted during blood gas inspections, including specimen labeling, QC requirements, and calibration verification requirements.

Arterial blood gas specimens must be labeled, Dr. Baisden said. No exceptions. “The fact that specimens require immediate analysis and the specimen is unstable does not eliminate the need—or the requirement—to label those specimens.” Inspectors may glance in the discarded specimen container to see if labels are present, he noted. And “lining up unlabeled specimens on gauze 4x4s ‘in order’ is a very dangerous—and common—practice.”

Specimen labeling is so important, Dr. Baisden said, that it pops up in two checklist questions. General checklist question GEN.40950 asks whether specimens are uniquely identified to avoid patient mixups and mislabelings; BGL.21070 asks for a documented procedure describing methods for specimen collection and labeling, among other tasks.

Labs have also been known to stumble over QC frequency mandates, perhaps because they don’t realize CAP demands more than CLIA. “CAP-accredited labs must adhere to the requirements of the CAP checklist, even when CLIA requires less,” Dr. Baisden said.

While CAP now requires the aforementioned one level of QC every eight hours of patient testing for pH, pCO2, and pO2, this requirement does not apply to calculated values, co-oximetry analytes, or other measured analytes, such as sodium, potassium, or glucose. For these, the CAP requires two levels of QC each day. “Some labs are doing more QC than required, and others fail to do the minimum required,” Dr. Baisden observed.

The March 31 CHM checklist also revises the requirement calling for control materials to cover the analyzer’s entire analytical measurement range. The requirement now asks, For pH, pCO2, and pO2, do control materials represent high and low values on each day of patient testing?

Calculating in-house QC acceptance ranges trips up many labs, which must calculate their own ranges—relying on manufacturers’ stated ranges for assayed controls is unacceptable. Sounding like a weary cop who’s heard every excuse from speeders, Dr. Baisden recounted oft-used reasons why labs fail to do so:

  • “If they do 30 days of testing to calculate a range before putting a new lot number in use, they will exhaust their control.”
  • “Controls outdate fairly quickly, and after performing a 30-day study the lab does not get enough mileage out of the lot before it expires.”
  • “Most manufacturers make control materials to the same target value from lot to lot, striving for manufacturing uniformity.”

Acknowledging labs’ concerns about limited advance time to establish a range and their desire to curb use of material in determining a range, Dr. Baisden offered a suggestion: “The lab can initially verify the new lot as it comes from the manufacturer using the manufacturer’s range, then create a temporary QC range prior to use by applying the historical standard deviation from previous lot controls to calculate an approximate new range that is plus or minus 2.5 to 3 standard deviations, then calculate a new range after 20 or more data points have been obtained.”

In the March 31 combined chemistry and toxicology checklist, calibration verification and reportable range requirements were clarified. The latter, in fact, is now referred to as analytical measurement range, or AMR.

Labs are required to calibrate or to perform calibration verification at least every six months; when recommended by the manufacturer; after major maintenance or service; at complete changes of reagent (unless the lab has data showing lot uniformity); and at a change of critical reagents (as defined by the laboratory). If the laboratory calibrates at least every six months, a separate calibration verification is not required. “However, the lab must validate analytical measurement range, that is, the reportable range, every six months,” said Dr. Baisden.

AMR, simply defined, means the analyte values that can be directly measured in a system. Labs must establish AMR initially when a test is introduced and validate it every six months. They must use materials—usually linearity materials, though any matrix-appropriate material is acceptable—that challenge the low, mid, and high points of the AMR; linearity calculations are not required. Acceptance criteria are based on historical instrument performance; a typical goal is 10 percent recovery of the high target and 50 percent recovery of the low target.

“A frequent complaint the CAP hears about the AMR requirement for blood gas testing is that there’s very little commercial material available at the extreme ends of the AMR for the pH, pCO2, and pO2,” Dr. Baisden said. His suggestion: “The lab should ask the manufacturer to make available materials at extreme high and low values for use in the AMR validation.” The lab can report results only within the AMR that it can validate, regardless of what the manufacturer claims.

Another frequently asked question: Does the daily use of controls constitute AMR validation? “It does not,” Dr. Baisden said.

In answering other perennial questions, Dr. Baisden offered other pearls of wisdom:

  • Blood gas instrument correlation and method comparison are required every six months for instruments with the same CLIA or CAP number; between identical instruments; and between different instrument makes and models.
  • Correlation is required between whole blood analyzers (those with electrolytes and arterial blood gases) and serum analyzers. “It is important to demonstrate that the differences between serial results from different instruments are of patient origin and not due to inter-instrument bias.”
  • Correlation is also required between whole blood analyzers and hematology analyzers for hemoglobin. For measured values—but not calculations—correlation should challenge points spanning the AMR.
  • Correlation of blood gas instruments requires using at least some patient specimens. Labs can use control/linearity material to challenge high and low ends of the reportable range—as Dr. Baisden noted, it can be difficult to obtain stable patient specimens at these concentrations. Labs do not have to correlate all instruments on the same day, and they can correlate instruments in subsets.
  • When comparing methods using fresh whole blood patient samples, which can be unstable compared with serum samples, keep the samples constantly iced. Try to analyze them within 30 minutes of collection, and avoid introducing air into the samples. If handled carefully, samples usually maintain integrity until 1 to 2 mLs remain.
  • Sample retention guidelines (available at and in the Laboratory Accreditation Manual, appendix G) specify that serum/CSF/body fluids (excluding urine) must be retained for 48 hours. This applies to blood gas specimens—which “leads to the following issue: An arterial sample is so unstable that the integrity is compromised within an hour or less of collection,” Dr. Baisden said. “A sample 48 hours old cannot be rerun to obtain valid results in cases where patient identity is in question.”
  • But, he noted, retaining an unstable specimen has its pluses: postanalytical confirmation for patient ID or specimen labeling; for specimen collection and transport; and for turnaround times. “The lab director may establish a different blood gas specimen retention time. This would require documentation of that policy in writing, stating the reason for less than 48 hours’ retention.”

  • BGL.21290 asks whether, in the absence of on-site supervisors, test results are reviewed by the lab director or general supervisor within 24 hours. “This applies only to high-complexity tests,” Dr. Baisden explained. “Most arterial blood gas testing methods are moderate complexity.” The CAP does not require supervisory review of all test results before or after reporting to patient records.

    This 24-hour rule comes into play only when high-complexity testing is done by trained high school graduates qualified under 42CFR493. 1489(b)(5) when a qualified general supervisor is not present.

  • New reagent lots must be validated before they’re put in use (BGL.26116). For blood gas testing, this typically involves calibration and running two levels of QC.
  • Labs must evaluate common interferences for all analytes measured with each reagent system or rely on credible information (BGL. 26132). Manufacturers’ data are acceptable. If interferences are not listed in the instrument manual or package insert, consult with the manufacturer to make sure studies have been done and ask for a statement or summary of the studies. “There is no need to re-create interference studies in your own lab,” Dr. Baisden said.
  • BGL.29770, which addresses whether control materials cover the analyzer’s AMR, was revised March 31. Labs are now asked, Do control materials for pH, pCO2, and pO2 represent both high and low values on each day of patient testing? “If you perform blood gas testing on all three shifts,” Dr. Baisden said, “the minimum requirement is three controls per day, one per shift, one of which must be high and one of which must be a low control.”
  • Calibration materials for pH, CO2, and O2 sensors must be traceable to NIST standard reference materials or conform to manufacturers’ specifications. This checklist item, BGL. 29720, refers to gas purity, Dr. Baisden said. “It also applies to institutions that make their own pH buffers.” Liquid reagents supplied by instrument manufacturers normally meet this requirement.
  • BGL.29740 was revised March 31 to read: “Is the calibration rechecked periodically using barometric pressure if appropriate?” Internal barometers are now common in instruments, Dr. Baisden observed, and labs can check barometric accuracy by calling the local airport for a barometric pressure reading and comparing it to the instrument’s.
  • Is it really necessary to save QC printouts for two years (GEN.41480)? “Yes,” said Dr. Baisden. “Even—and especially if—QC is transcribed onto log sheets, since transcription itself can introduce data errors.”

Karen Titus is CAP TODAY contributing editor and co-managing editor.