Advertisement

Laboratory predictors of bleeding and the effect of platelet and RBC transfusions on bleeding outcomes in the PLADO trial

Lynne Uhl, Susan F. Assmann, Taye H. Hamza, Ryan W. Harrison, Terry Gernsheimer and Sherrill J. Slichter

Data supplements

Article Figures & Data

Figures

  • Figure 1.

    Relationship between morning platelet count and patient-days with bleeding outcomes. (A) Unadjusted percentages of patient-days (95% CIs) with grade ≥2A (grade 2A+) bleeding. (B) ORs (95% CIs) for grade ≥2A bleeding compared with the reference category of ≥81 × 109/L, taking into account within-person correlation. The 16 df test for any association between morning platelet count category and grade ≥2A bleeding had P < .001. (C) Unadjusted percentages of patient-days (95% CIs) with grade ≥3 (grade 3+) bleeding. (D) ORs (95% CIs) for grade ≥3 bleeding compared with the reference category of ≥81 × 109/L, taking into account within-person correlation. The 16 df test for any association between morning platelet count category and grade ≥3 bleeding had P = .85.

  • Figure 2.

    Association between morning platelet count and grade ≥2A bleeding by stratum. (A) Unadjusted percentages of patient-days with grade ≥2A bleeding by stratum. (B-D) ORs (95% CIs), taking into account the within-patient correlation, comparing morning platelet count categories to the reference category of ≥81 × 109/L for (B) ALLO, (C) AUTO, and (D) CHEMO strata.

  • Figure 3.

    Relationship between morning hematocrit and percentage of patient-days with bleeding outcomes. (A) Unadjusted percentages of patient-days (95% CIs) with grade ≥2A bleeding. (B) ORs (95% CIs) for grade ≥2A bleeding compared with the reference category of hematocrit >29%, taking into account within-person correlation. The 2 df test for any association between morning hematocrit category and grade ≥2A bleeding had P = .002. (C) Unadjusted percentages of patient-days (95% CIs) with grade ≥3 bleeding. (D) ORs (95% CIs) for grade ≥3 bleeding compared with the reference category of hematocrit >29%, taking into account within-person correlation. The 2 df test for any association between morning hematocrit category and grade ≥3 bleeding had P < .001.

  • Figure 4.

    Relationship between bleeding grade and availability of fibrinogen, aPTT, and INR data. The y-axis indicates percentage of patient days with laboratory test performed among all patient days within specified bleeding grade.

  • Figure 5.

    Relationship between aPTT category and percentage of days with bleeding outcomes. (A) Unadjusted percentages of patient-days (95% CIs) with grade ≥2A bleeding. (B) ORs (95% CIs) for grade ≥2A bleeding compared with the reference category of aPTT ≤30, taking into account within-person correlation. The 2 df test for any association between aPTT category and grade ≥2A bleeding had P = .002. (C) Unadjusted percentages of patient-days (95% CIs) with grade ≥3 bleeding. (D) ORs (95% CIs) for grade ≥3 bleeding compared with the reference category of aPTT ≤30, taking into account within-person correlation. The 2 df test for any association between aPTT category and grade ≥3 bleeding had P = .40.

  • Figure 6.

    Relationship between INR category and percentage of days with bleeding outcomes. (A) Unadjusted percentages of patient-days (95% CIs) with grade ≥2A bleeding. (B) ORs (95% CIs) for grade ≥2A bleeding compared with the reference category of INR ≤1.2, taking into account within-person correlation. The 2 df test for any association between INR category and grade ≥2A bleeding had P < .001. (C) Unadjusted percentages of patient-days (95% CIs) with grade ≥3 bleeding. (D) ORs (95% CIs) for grade ≥3 bleeding compared with the reference category of INR ≤1.2, taking into account within-person correlation. The 2 df test for any association between INR category and grade ≥3 bleeding had P = .04.

Tables

  • Table 1.

    Baseline characteristics by stratum

    CharacteristicStratumP*
    AUTO (n = 378)ALLO (n = 413)CHEMO (n = 286)
    No.%MeanSDNo.%MeanSDNo.%MeanSD
    Male sex238632526115554.06
    Age, y531346125415<.001
    Body surface area, m21.950.231.950.231.910.23.02
    Platelet count, ×109/L473650383725<.001
    Hematocrit, %293294284.002
    INR1.050.111.070.111.100.13<.001
    aPTT, s3072953011.34
    Fibrinogen, mg/dL394155389155425214.31
    • * P values were calculated by using Fisher’s exact test for categorical variables and Kruskal-Wallis test for continuous variables.

  • Table 2.

    Daily bleeding outcomes by stratum

    Stratum
    AUTO (n = 3442)ALLO (n = 7143)CHEMO (n = 5735)P* (2 df)
    No.%No.%No.%
    Grade ≥2A<.001
     Yes3411014542164311
     No307090561079502489
     Missing data313168
    Grade ≥3.88
     Yes301721571
     No340699706599566599
     Missing data6613
    Grade 4.34
     Yes4<118<17<1
     No3435>997117>995717>99
     Missing data3311
    • The percentages expressed are based on the total days with non-missing data.

    • * P values take into account within-person correlation.

  • Table 3.

    Multipredictor models of laboratory predictors of grade ≥2A bleeding, taking into account within-person correlation

    PredictorModel 1*Model 2
    OR95% CIPOR95% CIP
    Stratum<.001<.001
     ALLO1.00Ref.1.00Ref.
     AUTO0.440.34-0.56<.0010.380.26-0.55<.001
     CHEMO0.550.43-0.71<.0010.550.39-0.77<.001
    Platelet count, ×109/L<.001.55
     1-52.961.93-4.54<.0011.790.86-3.75.12
     6-102.021.41-2.87<.0011.620.98-2.68.06
     11-152.341.65-3.33<.0011.610.95-2.72.08
     16-202.211.56-3.13<.0011.620.94-2.78.08
     21-251.961.37-2.79<.0011.520.91-2.54.11
     26-301.891.33-2.68<.0011.861.12-3.09.02
     31-352.061.45-2.94<.0011.680.96-2.94.07
     36-401.821.26-2.63.0011.500.86-2.61.16
     41-451.911.33-2.74<.0011.570.89-2.75.12
     46-502.011.33-3.04<.0011.810.97-3.37.06
     51-552.101.38-3.19<.0011.760.96-3.23.07
     56-601.821.18-2.80.011.560.78-3.12.21
     61-651.250.74-2.10.410.910.41-2.01.82
     66-702.541.60-4.02<.0012.280.97-5.34.06
     71-751.640.91-2.98.100.960.41-2.25.93
     76-801.670.91-3.04.101.460.60-3.55.40
     81+1.00Ref.1.00Ref.
    Hematocrit, %.02.35
     ≤251.201.03-1.39.021.080.84-1.38.56
     26-291.060.94-1.19.360.930.74-1.18.57
     >291.00Ref.1.00Ref.
    aPTT.04
     ≤301.00Ref.
     31-501.170.90-1.52.24
     >501.901.24-2.91.003
    INR.01
     ≤1.21.00Ref.
     1.3-1.51.351.06-1.71.01
     >1.51.791.28-2.52<.001
    • OR, odds ratio; Ref., reference.

    • * Grade ≥2A bleeding days/total days = 2403/15 979 = 15%.

    • Grade ≥2A bleeding days/total days = 612/3,604 = 17%.

  • Table 4.

    Multipredictor model of laboratory predictors of grade ≥3 bleeding, taking into account within-person correlation

    PredictorModel 1*
    OR95% CIP
    Stratum.91
     ALLO1.00Ref.
     AUTO1.090.62-1.91.76
     CHEMO1.110.66-1.88.69
    Platelet count, ×109/L.85
     1-51.100.35-3.44.86
     6-100.670.30-1.50.33
     11-150.700.31-1.56.38
     16-200.770.36-1.64.51
     21-250.870.37-2.03.75
     26-300.940.41-2.17.89
     31-350.390.12-1.30.13
     36-400.360.09-1.37.13
     41-450.840.31-2.31.74
     46-500.580.15-2.24.43
     51-551.010.33-3.07.98
     56-601.270.46-3.51.65
     61-651.090.27-4.34.90
     66-700.980.22-4.29.97
     71-751.850.31-11.07.50
     76-800.700.06-8.75.78
     81+1.00Ref.
    Hematocrit, %<.001
     ≤255.092.65-9.79<.001
     26-291.550.80-3.01.20
     >291.00Ref.
    • * Grade ≥3 bleeding days/total days = 155/16 129 = 1%.

  • Table 5.

    Association between platelet transfusion on the day of grade 2A bleeding and grade ≥2A bleeding the following day, within morning platelet count subgroups, taking into account within-person correlation

    Morning platelet countDays with grade 2A bleedingDays with platelet transfusion on the day of grade 2A bleedingNext day bleeding grade ≥2A after days with platelet transfusionDays without platelet transfusion on the day of grade 2A bleedingNext day bleeding grade ≥2A after days without platelet transfusionOR*95% CIP
    No.%No.%No.%No.%
    All platelet counts, ×109/L21811327617906085439445521.281.06-1.54.01
     1-1042040296228571848442.100.74-5.91.16
     11-20666410622245525638123481.200.86-1.69.28
     21-3037920153126631784793521.450.93-2.25.10
     31-402451134677681325472551.610.90-2.89.11
     41-5017192546368794645570.980.47-2.06.96
     51-6012559474373665338581.620.78-3.39.20
     61-706321331571426725602.100.80-5.47.13
     71-80381334538256613520.530.12-2.27.39
     81+671218650558226471.330.42-4.16.63
    • The interaction between the 9 morning platelet count categories and platelet transfusion was not significant (P = .34).

    • * Comparing platelet transfusion to no platelet transfusion.

    • Seven days had missing data for morning platelet count and are not included in the rows below.

  • Table 6.

    Association between RBC transfusion on the day of grade 2A bleeding and grade ≥2A bleeding the following day, within hematocrit subgroups, taking into account within-person correlation

    Hematocrit categoryDays with hematocrit data on day of grade 2A bleedingDays with RBC transfusion on the day of grade 2A bleedingNext day bleeding grade ≥2A after days with RBC transfusionDays without RBC transfusion on the day of grade 2A bleedingNext day bleeding grade ≥2A after days without RBC transfusionOR*95% CIP
    %No.%No.%No.%No.
    All hematocrit values21814552127661172679959561.200.98-1.47.08
     ≤25524296561826122844144631.020.74-1.41.91
     25-29104912312705792688513551.030.71-1.49.88
     >29585305196355595296531.790.81-3.96.15
    • The interaction between the three hematocrit categories and RBC transfusion was not significant (P = .60).

    • * Comparing RBC transfusion to no RBC transfusion.

    • Twenty-three days had missing hematocrit and are not included in the rows below.