Professional Documents
Culture Documents
M
aterials and Methods the CRT as well as in the intralaboratory cording to thefirstformula will be 100,
The hematology laboratory report regarding the problem encoun- regardless of the order in which the re-
computer data management tered, eg, "delta Hgb failed," and the pro- sults were obtained. If the second
system was developed in-house in 1982 hibition of the vertification of the test re- formula is used, the outcome of the delta
and has been expanded in modules. The sult until the delta check problem is re- percentage is determined by the order in
delta check software has been integrated solved. which results are obtained. If thefirstre-
into the hematology laboratory data Two levels of flagging are applied. The sult (Rt) shows a platelet count of 40 X
management system (CALM™ Software, first level indicated by an "A" reflects a 109/L and the second result (R2) shows a
Sharecom Industries Ltd, Calgary, Alber- clinically significant but medically plausi- platelet count of 80 X 109/L, then the
ta). The system can operate on a Micro- ble change from the previous test result. A delta percentage will be 50. However, if
vax™ Computer (Digital Equipment Cor- delta check result outside level "B" is not the results are obtained in reverse order,
poration, Boston) or a Macintosh™ mi- likely to be caused by a medical cause, at the delta percentage will be 100. Exam-
croprocessor (Apple Inc, Mountain least not within seven days, and a random ples of changes in parameter levels and
The sample identity is confirmed if both MCV and MCH are within limit A and platelets are at least within limit B.
• The sample identity cannot be established if either MCV or MCH are outside limit B, or if both are outside limit A, but within limit B.
• A change in Hgb within limit B but outside limit A, while the sample identity is intact, would result in an automatic comment on the report: "note
change in Hgb." A change in platelet count outside limit A but within limit B while both MCV and MCH are within limit A would result in a "delta
fail Pit count," followed by a check of the platelet count by the technologist, and if confirmed by a comment on the report: "note change in
platelet count." If the platelet count had been within limit A, then the result would have been verified without further checks regardless of the
acutal platelet count.
• MCV abnormal but within delta check limits: accept and do not perform red cell morphology check. The computer system will verify
automatically.
• Platelets are abnormally low but within delta check limits: accept and do not perform manual platelet count or platelet check on film. The
computer system will autoverify, and the result is'reported.
• Platelets are low, no delta check sample available. Check sample identification; if ok check sample for problems such as clots, clumping; if none
are detected, rerun and perform manual platelet count or check platelets on film. If platelet counts concur, verify the result and report the result.
• Platelets are outside delta check level A. Check sample identification; if ok, check for sample problems such as clots, if not detected, rerun,
perform manual platelet count or platelet check on film. If the counts concur, verify and report the result. The following comment will be printed
automatically on the report form: "note change in platelet count."
• Platelets are outside delta check level B, check results as above; if the results concur and no sample problems are detected, contact referring unit
or physician for clinical information. If no satisfactory explanation is obtained, suggest that the sample be recollected. Results are reported only if
this is specifically requested with a statement: "delta check for platelets failed, recollection of sample is suggested."
platelets within delta limits will not be de- of bloodfilms,reticulocyte preparations, problems through mistakes at the site of
tected as having sample identification ESRs, etc) in the laboratory. All work sta- sample collection, usually by staff other
problems, since Hgb and WBC changes, tions are equipped with a bar code reader, than the blood procurement team. The
unless extreme, will usually be ascribed to while the cell counters used in the labora- detection of such errors prior to release of
medically caused changes without further tory (Coulter S+V, STKR, Sysmex E- results or of verification is important, be-
checking. The small number of specimen 5000) are equipped with bar code readers cause it often involves critically ill pa-
identification problems in our laboratory and are on line with the laboratory com- tients, in whom blood is drawn from in-
is much lower than we expected and than puter system. Transcription errors are fur- dwelling intravenous catheters, and on
quoted in the literature,12 possibly be- ther reduced by on-line, real-time differ- whose test results immediate medical de-
cause the majority of causes for preana- ential counting on the computer termi- cisions may depend. In these patients an
lytic or postanalytic errors already may nal's modified numerical keypad and by identification mix-up or improperly col-
have been eliminated by an extensive limiting transcription of data to a mini- lected sample may delay or, possibly
"random error prevention" program. mum, for example, in the case of nonau- worse, change treatment-related deci-
This program includes a bar-code label tomated procedures such as entry of ESR sions. Without a delta check system in the
system, with labels constituting an inte- data. All these measures contribute to re- laboratory there is little chance that a ran-
gral part of the requisition form (R.L. duce the incidence of preanalytical and dom error will be detected before the re-
Crayn, Crayn Laboratory Forms; Calgary, postanalytical errors. port is sent out. The error may or may not
Alberta, Canada). The labels are affixed Chart review in cases of sample identi- be noted by the clinician, but it is hoped
on the sample at the site of collection and fication problems has brought to light that it will cause recollection and repeat
are also used throughout all sample han- that the majority of such problems have of testing. Such random errors, however,
dling and sample splitting (eg, preparation been caused by specimen mix-up or other