PURPOSE
Thrombocytopenia, which is common in critically ill patients, can increase bleeding risk
or be a clinical sign of heparin-induced thrombocytopenia (HIT). This investigation: 1)
estimated the incidence of, identified explanatory variables at and after admission, and evaluated
the performance of predictive models for thrombocytopenia, and 2) estimated the incidence of
HIT, and evaluated the predictive performance of a heparin-PF4 enzyme-linked immunosorbent
assay (ELISA) for diagnosing HIT.
METHODS
Logistic regression was used to identify predictors of thrombocytopenia (platelet count <
150 x 10⁹/L; < 100 x 10⁹/L) for 792 patients admitted to a community hospital combined
intensive and coronary care unit (ICU/CCU). ICU/CCU and ICU admission and post-admission
models were developed and validated internally using bootstrap re-sampling techniques.
Admission models were validated externally using data from 572 ICU patients admitted to a
different hospital.
HIT diagnosis was based on clinical criteria and a positive ¹⁴C-serotonin release assay
(SRA). Specificity and predictive values for the ELISA were estimated in patients who met the
clinical criteria for HIT.
RESULTS
One hundred and twenty-two (17.3%) ICU/CCU patients developed thrombocytopenia
(two consecutive counts < 150 x 10⁹/L). Specific predictors were consistently identified for the
admission models (APACHE II score, admission diagnosis, and admission platelet count) and
post-admission models (admission diagnosis, APACHE II score, admission platelet count, fresh
frozen plasma and packed red blood cell transfusions). These models demonstrated excellent
discriminating ability that was supported by internal validation. On external validation, the
admission models demonstrated acceptable to excellent discriminating ability, but there was a
tendency to over-predict at higher probabilities.
HIT incidence was 0.39% (95% CI, 0.01% to 2.1%). Positive and negative predictive
values (PPV and NPV) ofthe ELISA were 10% and 100%, respectively.
CONCLUSIONS
Thrombocytopenia, in critically ill patients, appears to be multi-factorial. Logistic
regression models could be used to identify patients at higher risk of bleeding, and to rule out
HIT prior to diagnostic testing in a subset of ICU/CCU patients who meet the clinical criteria for
this syndrome. ELISA testing could then be used to rule out HIT in the remaining patients who
meet the clinical criteria, thus reducing the proportion of patients requiring a change in therapy. / Pharmaceutical Sciences, Faculty of / Graduate
| Identifer | oai:union.ndltd.org:UBC/oai:circle.library.ubc.ca:2429/16968 |
| Date | 05 1900 |
| Creators | Verma, Arun Kumar |
| Source Sets | University of British Columbia |
| Language | English |
| Detected Language | English |
| Type | Text, Thesis/Dissertation |
| Rights | For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use. |
Page generated in 0.0015 seconds