Answering your questions on implementing delta check standards, testing for hypoglycemia, the proper workload for phlebotomists, and patients obtaining lab results via the Internet.

Delta check standards

Q We wish to implement delta checking in our lab. We would like to begin by using delta check values that other places have used. Where can we find such information?

A Delta checks involve comparison of current results on a patient to the previous results. While they can be performed manually, the only practical application involves computer-based comparison algorithms. Delta checks were introduced to detect analytical or labeling problems. In one of the first studies of delta checks, by Ladenson in 1975, the most common cause of errors (74%) was erroneous manual data entry, while use of the wrong specimen and specimen mislabeling caused 13% of errors each. [1] Even with the less automated systems used at that time, most delta checks were "false positive," in that repeat analysis gave the same results. In Ladenson's study, only 22% of 998 delta check alerts indicated errors. As computer and laboratory systems have improved, the frequency of delta checks indicating specimen errors has declined; for example, in two more recent studies, the predictive value of delta checks for specimen errors was 0.4% and 6%. [2,3] The predictive value was higher for those specimens in which four or more results on a given specimen failed delta check alerts. [4]

With such a low predictive value, one might question the utility of delta checks; in fact, many laboratories do not use the delta check capabilities in their laboratory information systems. We have found that the majority of delta checks represent changes in patient condition, and not sample errors. In about 20% of cases, investigation of a delta check alert led to detection of a clinical change that was not apparent to the patient's physician, and led to changes in patient treatment. [3] We thus consider delta checks to be a quality improvement tool in our institution, and have a policy that any specimen with four or more delta checks must be reviewed by a pathologist and, in most cases, discussed with the clinician before results can be reported.

If delta checks are utilized, what rules should be used for computer comparison? Tests with little day-to-day variation (total protein, albumin, urea, creatinine, alkaline phosphatase, total bilirubin, mean cell volume, and red cell distribution width) are the best to utilize; electrolytes are also of use since they are frequently performed. Tests that often show significant changes over short periods of time, or that are infrequently measured, are not good choices for delta alerts. There are several publications that list suggested delta check ranges that can be used in a computer based system. [1,5,6,7] The delta check alert can be an absolute change (e.g., 5 mmol/L), a percentage change (e.g., 10%), or a rate change, as suggested by Lacher and Connelly [8]; this has the advantage of reducing the frequency of delta alerts and slightly increasing the positive predictive value. [2] In our laboratory, we limit delta checks to the most recent specimen, and do not perform delta checks if there is greater than a seven-day difference between the time of collection of the two specimens.

D. Robert Dufour

Chief of Pathology

Veterans Affairs Medical Center

Washington, DC

References

(1.) Ladenson JH: Patients as their own controls: use of the computer to identify "laboratory error." Clin Chem 1975;27:1648-1653.

(2.) Kim JW, Kim JO, Kim SI: Differential application of rate and delta check on selected clinical chemistry tests. J Korean Med Sci 1990;5:189-195.

(3.) Dufour DR, Cruser, DL, Buttolph, T, Nasir, A, Aoun, P: The clinical significance of delta checks. Am J Clin Pathol 1998;110:531.

(4.) Rheem I, Lee KN: The multi-item univariate delta check method: anew approach. Medinfo 1998;9:859-863.

(5.) Houwen B: Random errors in haematology tests: a process control approach. Clin Lab Haematol 1990;12 (Suppl):157-168.

(6.) Westgard JO, Klee GG: Quality management in burtis CA, Ashwood ER (eds.): Tietz Textbook of Clinical Chemistry. 3rd Edition. Philadelphia: W.B Saunders, 1999, p. 407.

(7.) Dufour DR: Control of pre-analytic Variation in laboratory testing. In Kaplan L, Pesce A (Eds.): Clinical Chemistry: Theory and Practice, 3rd Edition. St. Louis: C. V. Mosby, 1996, p. 80.

(8.) Lacher DA, Connelly DP: Rate and delta checks compared for selected chemistry tests. Clin Chem 1988;34:1966-1970.

Test for hypoglycemia 

Q Are there new laboratory screening guidelines for testing for hypoglycemia? The only test we are aware of is the five-hour tolerance test.

A Two excellent articles on hypoglycemia were recently published in Endocrinology and Metabolism Clinics in September of 1999 [1,2] In these, Service outlines the approaches and recommended diagnostic tests for otherwise healthy individuals with suspected hypoglycemia.

Hypoglycemia is defined by the presence of Whipple's triad: documented hypoglycemia (plasma glucose [less than]40 mg/dL), symptoms of hypoglycemia (neurologic abnormalities, nervousness, anxiety, sweating), and relief of symptoms with administration of glucose. Many individuals develop symptoms of "hypoglycemia" such as nervousness and sweating, usually after meals, but are not actually hypoglycemic. Fingerstick glucose levels obtained at home are not reliable for documentation of hypoglycemia. [1]

One of the simplest ways to evaluate such individuals is with the mixed meal test. The person is instructed to ingest a meal similar to the one that produces symptoms, and then come to a place where blood can be drawn; when symptoms develop, a sample is obtained for measurement of plasma glucose. It is critical that serum be separated from cells within one hour of collection or that glycolytic inhibitors such as sodium fluoride be used to prevent artifactual decreases in plasma glucose. If plasma glucose is not below 50 mg/dL, then the person can be told that they do not have hypoglycemia. Another, more complicated screening procedure is the C-peptide suppression test, in which insulin is infused over a period of one hour to reduce plasma glucose. Blood for C-peptide is drawn before the insulin is given and again two hours later.

Results must be interpreted from a nomogram or table, based on the person's age and weight. The five-hour glucose tolerance test is not recommended for diagnosis of hypoglycemia, as at least 10% of healthy individuals have glucose concentration [less than] 50 mg/dL during this procedure. [3] Either of the previous two tests could be used to screen for hypoglycemia in place of the five-hour glucose tolerance test in the writer's institution.

In persons with abnormal results on the screening test, the formal approach to differential diagnosis is to perform a prolonged fast. If an individual comes after an overnight fast and has borderline low (40-60 mg/dL) glucose values, they can be observed with frequent (every two-hour) measurements of plasma glucose, and often will become absolutely hypoglycemic during a six- to eight-hour period of observation. If not, the patient is admitted either that day or on another day for a prolonged fast. The majority of otherwise healthy hypoglycemic individuals will be found to have insulin-mediated hypoglycemia, either due to inappropriate insulin production, insulin injection, or ingestion of oral hypoglycemic agents. With either the shorter term fast or the prolonged fast, samples are collected at the time of true hypoglycemia for insulin, C-peptide, proinsulin, and oral hypoglycemic agents. Alternatively, urine can be used for measurement of most oral hypoglycemic agents; the advantage of urine testing is that concentrations are generally higher and remain detectable for longer. Some also measure insulin antibodies; persons injecting insulin or with autoimmune hypoglycemia may have detectable antibodies to insulin. Insulin-mediated hypoglycemia is recognized if insulin is detectable ([greater than] 6 mU/mL with RIA. methods or [greater than] 3 mU/mL with chemiluminescent methods) at the time of hypoglycemia. Interpretation of results is shown in the table below.

D. Robert Dufour

Chief of Pathology

Veterans Affairs Medical Center

Washington, DC

References:

(1.) Service FJ: Classification of hypoglycemic disorders. Endacrinol Metab Clin North Am 1999;28:501-517.

(2.) Service FJ: Diagnostic approach to adults with hypoglycemic disorders. Endocrinol Metab Clin North Am 1999;28:501-517.

(3.) Lev-Ran A, Anderson RW: The diagnosis of postprandial hypoglycemia. Diabetes 30:996-999, 1981.

Workload for phlebotomists

Q Our laboratory is trying to increase the number of phlebotomists for the volume of patients that we currently service. Can you give me a recommendation/reference as to how many minutes it should take to perform a "normal" phlebotomy in both an inpatient and outpatient setting?

A In the past, workload units were used to standardize procedures. Every billable task was assigned a unit based on its degree of complexity and the amount of time expected to carry out the task under normal circumstances. Some facilities used "time and motion" studies to make the same assessments. Both of these techniques are now out of favor and seldom used.

Today, applying an expected time of performance for phlebotomy is difficult because the number of variables involved are facility-dependent. I know of no standards that can be universally applied. It may be best, and more accurate, to establish your own allocation for phlebotomy by conducting a small study that will take into account those variables that are unique to your facility and can measure exactly the specifics of the procedure you want to include. Time your personnel in the performance of phlebotomies according to the parameters you want included. Questions to ask in designing a study include: When does the procedure begin? At the time the lab receives the order, at the time the phlebotomist leaves the lab, or arrives at the bedside? When does it end? Should capillary punctures and heelsticks be studied separately from venipunctures? Should the study measure the time it takes veteran phlebotomists to perform punctures, or should it include those new to the procedure as well?

I suggest timing "phlebotomies" as you define them randomly over a two-week period and incorporate all the variables you think should be considered in determining your standard. Make sure you include enough measurements to make the sampling data a fair representation of the performance of the procedure in your facility.

Dennis Ernst

Director

Center for Phlebotomy Education

Ramsey, IL

Giving lab results to patient via Internet

Q In the Feb. 2000 MLO (page 9) there was a small article about a website that gave test results to the patient. How can they do this?

A The company mentioned in the news brief in MLO is LabDat Inc. (www.labdat.com). They provide an intermediate service between laboratories and doctors, as well as between doctors and their patients. If the doctor orders tests from a laboratory that is a client of LabDat, the doctor can order the tests through the LabDat interface and have the results reported in the same way. The interface provides a listing of ICD-9 and CPT codes to facilitate reimbursement. When results are available, the system alerts the doctor to review the results.

A key feature of the program is the ability for the doctor to allow a patient to view their own results through the Internet. Once the doctor has looked at the results, he or she can authorize viewing for the patient, and create a password to allow access to the patient. It is also possible for the physician to attach an explanatory note to the patient, for example, with instructions on the meaning of the results or further evaluation or treatment indicated. The site itself has a knowledge base of information on laboratory tests, as well as graphical presentation of panel test results (a "wheel" of results that graphically represents whether each test result is normal or abnormal).

I visited the site via a 56K modem connection. I found the site easy to navigate. Access is protected by 128-bit encryption, the highest level of security. On my modem, the wait to change screens was lengthy, with some screens taking 30 seconds or more to change. The information in the knowledge base was inconsistent. For example, iron only had a description of uses of the test, but no interpretive information. Zinc protoporphyrin, however, had much more extensive interpretive information. Some of the links produced puzzling connections. For example, if I clicked on "iron" on the list of test results, it also brought up information on total bilirubin and total triiodothyronine (T3).

There are several other companies that have similar, but more limited, services available via telephone. With these, the doctor can record a message on the test results for a patient.

Currently, these services are only available to patients through their physician's efforts. Both LabDat and the telephone services are marketed to physicians as ways to improve test reporting to their patients. It is not currently possible for patients to get results for tests their doctor ordered unless the doctor authorizes it. Howerer, there are companies that allow patients to request tests directly, and receive the results directly. These are not generally reimbursed by insurance companies, and the one site I visited online did not provide any interpretive information other than "normal" or "abnormal," advising the individual to consult his own physician.

D. Robert Dufour

Chief of Pathology

veterans Affairs Medical Center

Washington. DC

Daniel M. Baer is professor emeritus of laboratory madicine at Oregon Health Sciences University in Portland, OR, and a member of MLO's editorial advisory board.

Interpretation of test results in persons with true hypoglycemia
([less than] 40 mg/dL)
                              Insulin   C-Peptide  Proinsulin
Insulinoma                   Increased  Increased  Increased
Insulin Injection            Increased  Decreased  Decreased
Oral hypoglycemic ingestion  Increased  Increased  Increased
Non-insulin medicated        Decreased  Decreased  Decreased
                             Hypoglycemic Agents
Insulinoma                        Negative
Insulin Injection                 Negative
Oral hypoglycemic ingestion       Positive
Non-insulin medicated             Negative