Table 2 The leading principal component-based syndromes for mortality prediction in terms of p values.

From: Using syndrome mining with the Health and Retirement Study to identify the deadliest and least deadly frailty syndromes

Ranking

Regression coefficients

Log(p)

Mean

SD

PC

Numbers of variables

Variables

1 to 59

0.19 to 0.46

Log(0)

<0.001

1.26 to 3.28

1

9, 11, 15 to 71

r7mobila, r7bathcat, r7bath, …

60 to 70

0.33 to 1.34

−322 to −223

<0.001

0.36 to 1.77

1

2 to 8, 10, 12, to 14

r7mobila, r7bathcat, r7bath, …

71 to 77

−2.87 to −3.04 ×1014

−206 to −222

<0.001

1.62 to 1.75 ×10−15

68

39 to 45

r7sleepr, r7sleeprcat, r7diabs, …

78

−0.89

−203.82

<0.001

0.53

40

1

r7mobila

79

2.16

−203.82

<0.001

0.22

1

1

r7mobila

80

−2.93 ×1014

−202.39

<0.001

<0.001

68

38

r7sleepr, r7sleeprcat, r7diabs, …

  1. Positive regression coefficients suggesting the syndrome positively correlated with mortality and negative ones suggesting negatively correlated with mortality. r7bath = Problems with bathing; r7bathcat = Dummy: Problems with bathing; r7diabs = History of diabetes mellitus; r7mobila = Impaired mobility; r7sleepr = Sleep was restless; r7sleeprcat = Dummy: Sleep was restless. PC = principal component.