Skip to main content

Table 3 Individual study diagnostic test results

From: How accurate is the ‘Surprise Question’ at identifying patients at the end of life? A systematic review and meta-analysis

First author SQ time frame Total (n) SQ responses Diagnostic test results
SQ n Died (n) Survived (n) Sensitivity Specificity PPV NPV c-statistic
% (95% CI)
SQ up to 12 months
Amro [18] 12 months 201 No 50 22 28 56.4 82.7 44 88.7 0.696
    Yes 151 17 134 (39.6–72.2) (76–88.2) (30–58.7) (82.6–93.3) (0.612–0.78)
Carmen [19] 12 months 49 No 20 7 13 77.8 67.5 35 93.1 0.726
    Yes 29 2 27 (40–97.2) (50.9–81.4) (15.4–59.2) (77.2–99.2) (0.565–0.888)
Feyi [21] 6–12 months 178 No 58 37 21 72.5 83.5 63.8 88.3 0.78
    Yes 120 14 106 (58.3–84.1) (75.8–89.5) (50.1–76) (81.2–93.5) (0.71–0.85)
Halbe [22] 12 months 651 No 139 71 68 65.7 87.5 51.1 92.8 0.766
    Yes 512 37 475 (56–74.6) (84.4–90.1) (42.5–59.6) (90.2–94.9) (0.719–0.813)
Moroni [25] 12 months 231 No 126 87 39 83.7 69.3 69 83.8 0.765
    Yes 105 17 88 (75.1–90.2) (60.5–77.2) (60.2–77) (75.3–90.3) (0.711–0.819)
Moss [26] 12 months 147 No 34 10 24 45.5 80.8 29.4 89.4 0.631
    Yes 113 12 101 (24.4–67.8) (72.8–87.3) (15.1–47.5) (82.2–94.4) (0.519–0.743)
Moss [27] 12 months 826 No 131 53 78 74.6 89.7 40.5 97.4 0.822
    Yes 695 18 677 (62.9–84.2) (87.3–91.7) (32–49.4) (95.9–98.5) (0.769–0.874)
O'Callaghan [28] 12 months 501 No 99 67 32 62.6 91.9 67.7 90 0.772
    Yes 402 40 362 (52.7–71.8) (88.7–94.4) (57.5–76.7) (86.7–92.8) (0.724–0.82)
Pang [29] 12 months 367 No 109 27 82 61.4 74.6 24.8 93.4 0.68
    Yes 258 17 241 (45.5–75.6) (69.5–79.3) (17–34) (89.7–96.1) (0.603–0.756)
Thiagarajan [33] 12 months 130 No 83 47 36 75.8 47.1 56.6 68.1 0.614
    Yes 47 15 32 (63.3–85.8) (34.8–59.6) (45.3–67.5) (52.9–80.9) (0.534–0.695)
Vick [34] 12 months 4617 No 796 374 422 58.3 89.4 47 93 0.739
    Yes 3821 267 3554 (54.4–62.2) (88.4–90.3) (43.5–50.5) (92.2–93.8) (0.719–0.758)
Fenning [30] 6–12 months 172 No 38 6 32 35.3 79.4 15.8 91.8 0.573
    Yes 134 11 123 (14.2–61.7) (72.1–85.4) (6.02–31.3) (85.8–95.8) (.452,– .695)
Barnes [35] 12 months 231 No 14 11 3 11.6 97.8 78.6 61.3 0.547
    Yes 217 84 133 (5.92–19.8) (93.7–99.5) (49.2–95.3) (54.5–67.8) (0.512–0.581)
South [32] 6–12 months 199 No 96 14 82 93.3 55.4 14.6 99 0.744
    Yes 103 1 102 (68.1–99.8) (47.9–62.7) (8.21–23.3) (94.7–100) (0.669–0.818)
Haga [15] 6–12 months 138 No 120 39 81 88.6 13.8 32.5 72.2 0.512
    Yes 18 5 13 (75.4–96.2) (7.57–22.5) (24.2–41.7) (46.5–90.3) (0.453–0.571)
Da Silva [40] 12 months 3896 No 938 281 657 49.6 80.3 30 90.4 0.65
    Yes 2958 285 2673 (45.5–53.8) (78.9–81.6) (27–33) (89.2–91.4) (0.628–0.671)
SQ up to 6 months
Cohen [20] 6 months 450 No 71 39 32 37.9 90.8 54.9 83.1 0.643
    Yes 379 64 315 (28.5–48) (87.2–93.6) (42.7–66.8) (79–86.7) (0.594–0.693)
Khan [44] 6 months 500 No 238 148 90 82.2 71.9 62.2 87.8 0.77
    Yes 262 32 230 (75.8–87.5) (66.6–76.7) (55.7–68.4) (83.2–91.5) (0.733–0.808)
O'Callaghan [28] 6 months 501 No 99 56 43 72.7 89.9 56.6 94.8 0.813
    Yes 402 21 381 (61.4–82.3) (86.6–92.6) (46.2–66.5) (92.1–96.7) (0.761–0.865)
Gopinathan [31] 6 months 39 No 19 5 14 83.3 57.6 26.3 95 0.705
    Yes 20 1 19 (35.9–99.6) (39.2–74.5) (9.15–51.2) (75.1–99.9) (0.52–0.889)
SQ imminent
Gibbins [41] Admission 6642 No 327 215 112 57 98.2 65.7 97.4 0.776
    Yes 6315 162 6153 (51.9–62.1) (97.9–98.5) (60.3–70.9) (97–97.8) (0.751–0.801)
Hamano [23] 7 days 2361 No 931 282 649 84.7 68 30.3 96.4 0.763
    Yes 1430 51 1379 (80.4–88.4) (65.9–70) (27.4–33.4) (95.3–97.3) (0.742–0.785)
Hamano [23] 30 days 2361 No 1851 1066 785 95.6 37 57.6 90.4 0.663
    Yes 510 49 461 (94.2–96.7) (34.3–39.7) (55.3–59.9) (87.5–92.8) (0.648–0.678)
Strout [42] 30 days 330 No 108 15 93 48.4 68.9 13.9 92.8 0.586
    Yes 222 16 206 (30.2–66.9) (63.3–74.1) (7.99–21.9) (88.6–95.8) (0.493–0.68)
  1. Sensitivity (the ability to recognise those who were dying, e.g. 15/31 for study by Strout [42])
  2. Specificity (the ability to recognise those who were not dying, e.g. 206/299 for study by Strout [42])
  3. PPV positive predictive value (the proportion of patients who died when the clinician predicted dying, e.g. 15/108 for study by Strout [42]), NPV negative predictive value (the proportion of patients who survived when the clinician predicted survival, e.g. 206/222 for study by Strout [42])