Electronic Supplementary Material (ESM)

TABLE OF CONTENTS

ESM Table 1 Deficits of the frailty index at baseline and follow-ups ahead of
multiple imputation Page 1-2

ESM Figure 1 Relationship of frailty index and age at ESTHER baseline contact Page 3

ESM Figure 2 Mean frailty index with 95% confidence interval bands at numbers of fulfilled frailty items of the frailty definition of Fried et al. (Weight loss, weak grip strength, slow gait speed, self-reported exhaustion and low energy expenditure) in participants of the ESTHER 8-year follow-up home visit Page 4

ESM Table 2 Comparison of the Fried frailty definition and the frailty index after application of thresholds for the frailty index to define non-frail, pre-frail and frail subjects in the ESTHER 8-year follow-up home visit population Page 5

ESM Figure 3 Dose-response relationship of the frailty index at ESTHER baseline Page 6 contact and all-cause mortality

ESM Table 3 Proportions of imputed missing values in the total ESTHER study

population at baseline and follow-ups Page 7

ESM Table 4 Association of changes in risk factors for premature mortality with
the change in season-standardized 25-hydroxyvitamin D level
z-scores during about 5 years of follow-up in male participants of
the ESTHER study Page 8

ESM Text Details of multiple imputation Page 9-10

ESM Table 5 ICD-10 codes used to group the underlying causes of deaths Page 11

ESM Table 1. Deficits of the frailty index at baseline and follow-ups ahead of multiple imputation

Deficits / Baseline
(nalive = 9,949;
nresponse = 9,949) / 2-year follow-up
(nalive = 9,746;
nresponse = 9,358; 96.0%) / 5-year follow-up
(nalive = 9,402;
nresponse = 8,271; 88.0%) / 8-year follow-up
(nalive = 8,926;
nresponse = 7,012; 78.6%)a
No. / Variable / Coding with assigned
value for frailty index /
Freq (%) / Proportion
of missing
values (%) / Freq (%) / Proportion
of missing
values (%) b / Freq (%) / Proportion
of missing
values (%) b / Freq (%) / Proportion
of missing
values (%) b
Chronic diseases
1 / Coronary artery disease / “Yes”=1; “No”=0 / 1,449 (15.1) / 0.1 / 1,539 (16.5) / 4.1 / 1,532 (18.5) / 12.0 / 1,541 (25.4) / 32.0
2 / Heart failure / “Yes”=1; “No”=0 / 1,111 (11.3) / 0.8 / 1,124 (12.6) / 8.6 / 1,066 (13.2) / 14.2 / 1,009 (16.8) / 32.6
3 / Diabetes / “Yes”=1; “No”=0 / 1,429 (14.4) / 0.2 / 1,548 (16.7) / 4.9 / 1,654 (20.1) / 12.3 / 1,737 (25.0) / 22.1
4 / Cancer / “Yes”=1; “No”=0 / 774 (7.8) / 0.0 / 903 (9.8) / 5.1 / 1,023 (12.4) / 12.1 / 1,085 (15.6) / 22.2
5 / Glaucoma / “Yes”=1; “No”=0 / 410 (4.1) / 0.5 / 495 (5.4) / 5.1 / 556 (6.7) / 12.3 / 524 (8.5) / 30.9
6 / Cataract / “Yes”=1; “No”=0 / 1,038 (10.5) / 0.4 / 1,445 (15.6) / 4.8 / 1,764 (21.4) / 12.4 / 1,679 (28.0) / 32.8
7 / Morbus Parkinson / “Yes”=1; “No”=0 / 56 (0.6) / 1.1 / 112 (1.2) / 4.0 / 156 (1.9) / 12.3 / 178 (2.9) / 31.3
History of acute disease events
8 / Myocardial infarction / “Yes”=1; “No”=0 / 538 (5.6) / 3.1 / 533 (5.7) / 4.6 / 522 (6.3) / 12.4 / 484 (7.0) / 22.1
9 / Stroke / “Yes”=1; “No”=0 / 336 (3.5) / 3.3 / 409 (4.4) / 4.6 / 458 (5.6) / 12.4 / 551 (7.9) / 22.1
10 / Femur fracture / “Yes”=1; “No”=0 / 77 (0.8) / 3.5 / 94 (1.0) / 4.8 / 111 (1.3) / 12.2 / 135 (1.9) / 22.1
11 / Joint replacement / “Yes”=1; “No”=0 / 288 (2.9) / 0.5 / 411 (4.5) / 5.4 / 574 (7.0) / 12.5 / 681 (11.0) / 30.9
Use of drugs
12 / Anti-hypertensives / “Yes”=1; “No”=0 / 4,314 (43.9) / 1.1 / 4,885 (52.2) / 4.0 / 4,948 (59.8) / 12.0 / 3,926 (67.8) / 35.1
13 / Lipid lowering drugs / “Yes”=1; “No”=0 / 1,109 (11.3) / 1.1 / 1,595 (17.0) / 4.0 / 1,870 (22.6) / 12.0 / 1,898 (32.8) / 35.1
14 / Drugs against gout / “Yes”=1; “No”=0 / 690 (7.0) / 1.1 / 791 (8.5) / 4.0 / 779 (9.4) / 12.0 / 630 (10.9) / 35.1
15 / Vasodilatators / “Yes”=1; “No”=0 / 586 (6.0) / 1.1 / 590 (6.3) / 4.0 / 499 (6.0) / 12.0 / 360 (6.2) / 35.1
16 / Cardiac glycosides / “Yes”=1; “No”=0 / 249 (2.5) / 1.1 / 294 (2.7) / 4.0 / 200 (2.4) / 12.0 / 123 (2.1) / 35.1
17 / Acetylsalicylic acid / “Yes”=1; “No”=0 / 1,146 (11.6) / 1.1 / 1,747 (18.7) / 4.0 / 1,809 (21.9) / 12.0 / 1,751 (30.3) / 35.1
18 / Anti-osteoporotic drugs / “Yes”=1; “No”=0 / 100 (1.0) / 1.1 / 182 (1.9) / 4.0 / 300 (3.6) / 12.0 / 364 (6.3) / 35.1
19 / Anxiolytics / “Yes”=1; “No”=0 / 203 (2.1) / 1.1 / 181 (1.9) / 4.0 / 174 (2.1) / 12.0 / 183 (3.2) / 35.1
20 / Sedatives / “Yes”=1; “No”=0 / 152 (1.6) / 1.1 / 181 (1.9) / 4.0 / 128 (1.6) / 12.0 / 191 (3.3) / 35.1
21 / Anti-dementives / “Yes”=1; “No”=0 / 96 (1.0) / 1.1 / 165 (1.8) / 4.0 / 148 (1.8) / 12.0 / 178 (3.1) / 35.1
22 / Drugs against prostatic
hyperplasia or incontinence / “Yes”=1; “No”=0 / 327 (3.3) / 1.1 / 537 (5.7) / 4.0 / 520 (6.3) / 12.0 / 510 (8.8) / 35.1
Impairments or disabilities
23 / Problems with averagely
difficult tasks / “Yes, severely”=1; / 788 (8.5) / 6.5 / 761 (8.4) / 6.6 / 834 (10.2) / 13.2 / 619 (11.0) / 36.8
“Yes, a little”=0.5; / 3,311 (35.6) / 2,911 (32.0) / 2,999 (36.7) / 2,166 (38.4)
“No, none”=0 / 5,199 (55.9) / 5,431 (59.7) / 4,330 (53.0) / 2,854 (50.6)
24 / Problems climbing several
flights of stairs / “Yes, severely”=1; / 1,287 (13.5) / 4.5 / 1,159 (12.6) / 5.7 / 1,044 (12.8) / 13.3 / 786 (13.7) / 35.7
“Yes, a little”=0.5; / 3,906 (41.1) / 3,447 (37.5) / 3,190 (39.1) / 2,288 (39.9)
“No, none”=0 / 4,311 (45.4) / 4,580 (49.9) / 3,922 (48.1) / 2,664 (46.4)

ESM Table 1 continues

Abbreviations: Freq, frequency;

a Response by participant’s questionnaire (n=6,061), general practitioner’s questionnaire (n=5,997) or home visit (n=3,124).

b Calculated by no. of participants with data divided by no. of participants alive at follow-up contact.

ESM Figure 1. Relationship of frailty index and age at ESTHER baseline contact

ESM Figure 2. Mean frailty index with 95% confidence interval bands at numbers of fulfilled frailty items of the frailty definition of Fried et al. (Weight loss, weak grip strength, slow gait speed, self-reported exhaustion and low energy expenditure) in participants of the ESTHER 8-year follow-up home visit

ESM Table 2. Comparison of the Fried frailty definition and the frailty index after application of thresholds for the frailty index to define non-frail, pre-frail and frail subjects in the ESTHER 8-year follow-up home visit population

Modified Fried frailty definition a / Frailty index
No. of frailty
criteria met / n / Proportion
(%) / Cut-offs / n / Proportion
(%)
Non-frail / 0 / 1,030 / 33.1 / ≤ 0.11 / 1,041 / 33.5
Pre-frail / 1-2 / 1,806 / 58.0 / > 0.11 - ≤ 0.35 / 1,803 / 57.9
Frail / 3-5 / 276 / 8.9 / > 0.35 / 268 / 8.6

a The modified Fried frailty definition using population-independent cut-points comprises the following 5 frailty criteria:

1. Weight loss: Self-reported unintentional weight loss of 5 kg or more in the last year

2. Exhaustion: Self-reported Center for Epidemiologic Studies Depression (CES-D) Scale questions: “I felt everything I did was an effort,” or “I could not get going” answered with “3 to 4 days” or “5 to 7 days” in the last week

3. Grip strength: ≤20 kg (women), ≤30 kg (men), measured with Jamar hand dynamometer (Lafayette Instrument Company, Lafayette, IN)

4. Low physical activity: ≤ 9.4 points in total score of the modified version of the Physical Activity Questionnaire for the Elderly (PAQE)

5. Low gait speed: ≤0.8 m/s assessed at a 3-m walking distance

ESM Figure 3. Dose-response relationship of the frailty index at ESTHER baseline contact and all-cause mortality

The dose-response relationship was assessed with a restricted cubic splines SAS macro [Ref no. 31, main document]. The frailty index was adjusted for age and sex. Five knots were pre-defined knots at the 10th, 25th, 50th, 75th and 90th frailty index percentile and the frailty index value “0” was chosen as the reference.

ESM Table 3. Proportions of imputed missing values in the total ESTHER study population at baseline and follow-ups

Proportion of missing values (%) a
Characteristic / Baseline
(n = 9,949) / 2-year follow-up
(nalive = 9,746;
nresponse = 9,358 (96.0%)) / 5-year follow-up
(nalive = 9,402;
nresponse = 8,271 (88.0%)) / 8-year follow-up
(nalive = 8,926;
nresponse = 7,012 (78.6%)) b
Age / 0.0 / 4.0 / 12.0 / 21.4
Sex / 0.0 / n. a. / n. a. / n. a.
School education / 2.5 / n. a. / n. a. / n. a.
BMI / 0.2 / 5.3 / 12.5 / 32.7
Smoking / 2.8 / 7.3 / 13.0 / 32.5
Light physical activity / 6.3 c / 11.9 / 13.6 / 35.6
25(OH)D / 3.7 / 45.7 d
Self-rated health / 4.4 / 4.4 / 12.1 / 36.7

Abbreviations: 25(OH)D, 25-hydroxyvitamin D; BMI, body mass index; n. a., not assessed.

a Calculated by no. of participants with data divided by no. of participants alive at follow-up contact. Therefore, complete non-participation at the follow-up was the reason for 4%, 12%, and 21.4% of the missing values at the 2-, 5, and 8-year follow-up, respectively.

b Response by participant’s questionnaire (n=6,061), general practitioner’s questionnaire (n=5,997) or home visit (n=3,124).

c Baseline light physical activity did not have a missing at random pattern because, in contrast to the follow-up questionnaires, study participants were not explicitly asked to fill a “0” if they did not do any light exercise. The respective question was “How many hours per week do you on average perform light physical activity (at work and leisure time combined), like e.g. walking, bicycling or hiking”. The variable was therefore not imputed by multiple imputation and information on light physical activity at the 2-year follow-up was also used for the baseline assessment and only missing values at 2-year follow-up were filled up with information from the baseline questionnaire assuming that there was no change in light physical activity from baseline to 2-year follow-up.

d Missing values in 5-year follow-up 25(OH)D measurements were not imputed by multiple imputation because of too many missing values due to a lower number of collected 5-year follow-up blood samples (n=5,104) than questionnaires (n=8,271). Missings in 5-year follow-up 25(OH)D measurement were therefore filled with values from baseline assuming that they did not change from baseline to 5-year follow-up. This assumption can be made because season-adjusted 25(OH)D measurements are rather stable (Jorde R, et al. Am J Epidemiol 2010;171:903-908).

ESM Table 4. Association of changes in risk factors for premature mortality with the change in season-standardized 25-hydroxyvitamin D level z-scores during about 5 years of follow-up in male participants of the ESTHER study

Decile with highest decrease in 25(OH)D z-scores / ∆ 25(OH)D z-scores
Characteristic / Ntotal a / Ncases (%) / OR (95%CI) b / Mean (SD) / ß b / P
Total population / 2,407 / 241 (10.0) / n.a. / -0.002 (1.02) / n.a. / n.a.
∆ Age (years)
4.75; < 5.25 / 1,963 / 183 (8.5) / Ref / 0.066 (1.02) / Ref / Ref
5.25; < 5.75 / 381 / 58 (15.2) / 2.01 (1.44; 2.81) / -0.268 (0.98) / -0.327 / <0.001
5.75; < 6.75 / 63 / 16 (25.4) / 3.76 (2.03; 6.94) / -0.480 (1.06) / -0.548 / <0.001
∆ BMI (kg/m²)
< -2 / 196 / 29 (14.8) / 1.62 (1.01; 2.59) / -0.180 (1.03) / -0.199 / 0.014
≥ -2; < -1 / 280 / 30 (10.7) / 1.18 (0.76; 1.84) / -0.055 (1.00) / -0.084 / 0.215
≥ -1; < +1 / 1,356 / 120 (8.9) / Ref / 0.043 (1.01) / Ref / Ref
≥ +1; < +2 / 322 / 35 (10.9) / 1.21 (0.80; 1.82) / -0.045 (1.04) / -0.072 / 0.254
≥ +2 / 253 / 27 (10.7) / 1.17 (0.73; 1.87) / 0.011 (1.07) / -0.042 / 0.567
∆ Smoking
Quit / 126 / 13 (10.3) / 0.75 (0.37; 1.50) / 0.028 (1.14) / 0.131 / 0.253
Unchanged / 2,255 / 226 (10.0) / Ref / -0.003 (1.02) / Ref / Ref
Started / 26 / 2 (7.7) / 0.71 (0.16; 3.19) / -0.027 (0.96) / 0.036 / 0.864
∆ Physical activity
Down in category c / 588 / 71 (12.1) / 1.60 (1.10; 2.34) / -0.085 (1.03) / -0.129 / 0.029
Unchanged / 1,290 / 121 (9.4) / Ref / 0.035 (1.04) / Ref / Ref
Up in category c / 529 / 49 (9.3) / 0.91 (0.59; 1.41) / 0.002 (0.98) / -0.062 / 0.315
∆ Self-rated health
Down in category d / 378 / 40 (10.6) / 0.99 (0.65; 1.50) / -0.041 (1.06) / -0.032 / 0.614
Unchanged / 1,554 / 155 (10.0) / Ref / -0.009 (1.00) / Ref / Ref
Up in category d / 475 / 46 (9.7) / 1.11 (0.73; 1.70) / 0.055 (1.06) / 0.049 / 0.441
∆ Frailty index / 2,407 / 241 (10.0)
< 0 / 707 / 65 (9.2) / Ref / 0.036 (0.98) / Ref / Ref
0; ≤ +0.5 / 839 / 81 (9.7) / 1.09 (0.74; 1.61) / 0.013 (1.01) / -0.045 / 0.424
> +0.5 / 861 / 95 (11.0) / 1.29 (0.88; 1.90) / -0.046 (1.06) / -0.055 / 0.344

Abbreviations: 25(OH)D, 25-hydroxyvitamin D; ß, beta-coefficient; BMI, body mass index; n.a., not applicable; OR, odds ratio; STD, standard deviation.