Frailty Index: Mexican Health and Aging Study (MHAS)

Author

Kwabena Asare

Published

July 2, 2025

1 Introduction

PURPOSE: To create a frailty index for MHAS”. Use available variables and create a frailty index using the accumulation of deficits approach.

  • The deficit frailty index (FI) is defined as the proportion of deficits present in an observation out of the total number of age-related health variables considered.
  • The process that Charlotte used to generate the frailty index in Longitudinal Health and Ageing Study (LASI) is outlined in the do file entitled 2_LASI_frailty_indices.do located in the J drive.

APPROACH:

  • For MHAS, find out whether other studies have already created a frailty index and consider using the same approach
  • Could otherwise base our index on the Health and Retirement Study (HRS) index or similar, but might need to modify it if any of the relevant variables haven’t been collected in MHAS
  • Do some reading, come up with a plan and discuss and discuss with Charlotte before implementation.

BRIEF DESCRIPTION OF MHAS: - MHAS is a longitudinal population-based household survey in Mexico. We have data sets for 5 waves of data from 2001, 2003, 2012, 2015 and 2018. - Data sets are wide. All variables are named by wave, if not, then its not a time changing variable or not repeatedly measured, e.g race. - Household ID (cunicah), is the highest level of identification. - There is person number (np) which is the numerical order of persons in a household, so not unique per person in the survey, but unique per person within each household. - Each row of data therefore represents an observation in a household which is unique.

MHAS DATA SETS:

  • Harmonized MHAS: This is a cleaned, managed and research ready MHAS-data. It includes a selected set of variables and not exhaustive of all the variables in the MHAS study. It is a streamlined collection of variables derived from the Mexican Health and Aging Study (MHAS). The Harmonized MHAS data file incorporates data from the core interview data, the master follow-up file, household roster data, and next-of-kin data. It does not include any data which is not public release. for example, there is no data on infections exposure in the harmonized MHAS. CWG has already extracted the infections data and added to the harmonized MHAS data, file name H_MHAS_c2_infections. This data is already loaded in the setup code chunk.

  • Original MHAS: The full MHAS data set. Any variables not found in the harmonized data set can be found in the original/full MHAS data set. To extract a variable from the full MHAS data set, use the process CWG used.

2 Studies that developed and used the frailty index with MHAS data.

There are a number of studies that have created and validated the FI based on the accumulation of deficits approach. There is one by Garcia-Gonzales 2009 which was created in MHAS. Will use this study.

3 Brief overview of Garcia-Gonzales, etal, 2009 study:

3.1 Study design

  • Objectives: To analyze relationships between the frailty index, age and mortality (outcome) in a two year follow up study of Mexican elderly.
  • Population: 4082 adults 65+ years old, mean 73 years and 52.5% were women.
  • Data source: MHAS waves 1 and 2, exposures in wave 1 (age, frailty Index), mortality wave 2.
  • Weighted average frailty Index: 0.16, SD = 0.11
  • Categorical frailty: Five levels of frailty: .00-.07, .07-.14, .14-.21, .21-.35, .35-.65

3.2 Calculation of frailty Index

  • Garcia used 34 deficits (variables). But its technically 45 as they combined some deficits into cumulative deficits e.g depressive symptoms included 9 deficits and Difficulty with mobility included 4 deficits. The deficits and coding can be found here Table 1 Health deficits included in the frailty index and their coding..

  • Five sub categories of deficits

    • Health problems before age 10: will be excluded (6 deficits)
    • General Health
    • Medically diagnosed conditions
    • Medical symptoms during the past 2 years
    • Activities of daily living
    • Instrumental activities of daily living
  • Coding was on the scoring system suggested by Rockwood et. al.

  • Complete case use: 145 observations were excluded due to missing frailty. It is not clear what minimum number of missing deficits were allowed. Looks like zero. Thus missing frailty meant at least one deficit was missing.

  • Frailty index was defined as a proportion of the total number of deficits an observation has with respect to the 34 deficits included. The denominator was observation-specific: done by taking the mean of the total score for each observation.

3.3 Categorical frailty in the literature

4 Deficits to use in calculation

  • Will use the following categories of deficits, excluding 6 deficits before age 10, yielding 28 deficits (34 - 6).
    • General Health To be excluded
    • Medically diagnosed conditions
    • Medical symptoms during the past 2 years
    • Activities of daily living
    • Instrumental activities of daily living
  • The 28 deficits = 39 variables in the data set.
    • Depressive symptoms combines 9 deficits (depressed, unhappy, lonely, tired, sad, did not enjoy life, had no energy, restless sleep, or thought that everything did was an effort) and
    • Difficulty with mobility combines 4 deficits, difficulty (picking up a coin, dressing, walking several blocks and walking across the room)
    • For difficulty with mobility, each deficit will stand-alone in our calculation

5 What percentage of deficits are available and missing by wave in MHAS?

  • 36/39 deficits available in Harmonised MHAS dataset
  • 3 of remaining deficits in original MHAS
  • so 39 deficits available, but not all are available in all waves

Reduces to

Deficit availabilitys (variables) by wave (HARM MHAS)
rowid label_2 w1 w2 w3 w4 w5
1 R Self-report of health 1 1 1 1 1
2 R Ever had high blood pressure 1 1 1 1 1
3 R Ever had diabetes 1 1 1 1 1
4 R Ever had cancer 1 1 1 1 1
5 R Ever had respiratory disease, incl asthma 1 1 1 1 1
6 R Ever had heart attack 1 1 1 1 1
7 R Ever had stroke 1 1 1 1 1
8 R Ever had arthritis 1 1 1 1 1
9 R Self-rated eyesight 1 1 1 1 1
10 R Self-rated hearing 1 1 1 1 1
11 R Fallen down last 2 years 1 1 1 1 1
12 R Ever fractured a bone (including hip) since age 50 1 1 1 1 NA
13 R Leaks urine when coughing (last 2 yrs) NA NA 1 1 1
14 R Difficulty breathing 1 1 1 1 1
15 R Severe fatigue 1 1 1 1 1
16 R Frequent problems with pain 1 1 1 1 1
17 R CESD-Felt depressed 1 1 1 1 1
18 R CESD-Everything an effort 1 1 1 1 1
19 R CESD-Sleep was restless 1 1 1 1 1
20 R CESD-Felt happy 1 1 1 1 1
21 R CESD-Felt lonely 1 1 1 1 1
22 R CESD-Felt sad 1 1 1 1 1
23 R CESD-Felt tired 1 1 1 1 1
24 R CESD-Had a lot of energy 1 1 1 1 1
25 R Difficulty-Dressing 1 1 1 1 1
26 R Difficulty-Walking across room 1 1 1 1 1
27 R Difficulty-Walking several blocks 1 1 1 1 1
28 R Difficulty-Picking up a coin 1 1 1 1 1
29 R Difficulty-Bathing or showering 1 1 1 1 1
30 R Difficulty-Eating 1 1 1 1 1
31 R Difficulty-Getting in/out of bed 1 1 1 1 1
32 R Difficulty-Using the toilet 1 1 1 1 1
33 R Difficulty-Managing money 1 1 1 1 1
34 R Difficulty-Taking medications 1 1 1 1 1
35 R Difficulty-Shopping for groceries 1 1 1 1 1
36 R Difficulty-Preparing hot meals 1 1 1 1 1

5.1 3 deficits NOT available in Harmonised MHAS dataset

First checked if available in MHAS survey questionnaires and found all

  • not found, Leg pain on walking (c68h)

  • not found, Stomach pain, indigestion or diarrhea (c68f_wave)

  • not found, c49_6_wave: did not enjoy life

  • Extract these variables from the original MHAS data set

  • Check from which waves these variables are available

Deficit availabilitys (variables) by wave (full MHAS)
label w1 w2 w3 w4 w5
R Stomach pain, indigestion, diarrhea 1 1 1 1 1
R Pain in lower limbs while (or after) walking 1 1 NA NA NA
R enjoyed life 1 1 1 1 1
R Leaks urine when coughing (last 2 yrs) 1 1 1 1 1
R Ever fractured a bone (including hip) since age 50 1 1 1 1 1

5.2 Number of unavailable deficits by wave:

wave 1: 0

wave 2: 0

wave 3: 1

wave 4: 1

wave 5: 1

R Pain in lower limbs while (or after) walking

6 How many deficits (distribution) are available by observation in each wave

  • Responses coded as .s are skipped questions, which means that people who have previously responded ‘no’ to a question like ‘do you have difficulty walking several blocks?’ will not be asked whether they have difficulty walking across a room.

  • In this case, we can impute their response to be ‘no’ to difficulty walking across a room.

  • These were largely the ADL questions.

  • Can see the criteria for skipped questions in the documentation of the section H functionality questionnaire (attached).

  • Also if Can’t Do or Don’t Do for the ADLs and IADLs, update to Yes (difficulty)

  • Which deficits have > 5% missing: Criteria for including a deficit/variable

  • Self-rated hearing is over 90% missing in wave 3: Although question has a skip condition, it is difficult to impute from the related questions. To exclude.

  • Leg pain on walking not available in waves 3-5: To exclude.

Frailty deficits (variables) availability-missingness by Wave
Characteristic1 Wave_1 Wave_2 Wave_3 Wave_4 Wave_5
N = 14,1541 N = 12,5261 N = 14,4481 N = 13,8501 N = 15,7861
1. R Self-report of health




    1.Excellent 271 (1.9%) 189 (1.5%) 363 (2.5%) 342 (2.5%) 445 (2.8%)
    2.Very good 627 (4.4%) 396 (3.2%) 646 (4.5%) 465 (3.4%) 796 (5.0%)
    3.Good 4,495 (32%) 3,559 (28%) 4,303 (30%) 3,710 (27%) 4,670 (30%)
    4.Fair 6,585 (47%) 6,204 (50%) 7,316 (51%) 7,347 (53%) 8,381 (53%)
    5.Poor 2,169 (15%) 2,173 (17%) 1,817 (13%) 1,983 (14%) 1,491 (9.4%)
    Missing 7 (<0.1%) 5 (<0.1%) 3 (<0.1%) 3 (<0.1%) 3 (<0.1%)
2. R Ever had high blood pressure




    0.no 8,552 (60%) 6,583 (53%) 7,126 (49%) 5,787 (42%) 7,279 (46%)
    1.yes 5,176 (37%) 5,915 (47%) 7,292 (50%) 8,049 (58%) 8,496 (54%)
    Missing 426 (3.0%) 28 (0.2%) 30 (0.2%) 14 (0.1%) 11 (<0.1%)
3. R Ever had diabetes




    0.no 11,609 (82%) 10,112 (81%) 11,024 (76%) 10,095 (73%) 11,516 (73%)
    1.yes 2,129 (15%) 2,368 (19%) 3,395 (23%) 3,739 (27%) 4,245 (27%)
    Missing 416 (2.9%) 46 (0.4%) 29 (0.2%) 16 (0.1%) 25 (0.2%)
4. R Ever had cancer




    0.no 13,483 (95%) 12,208 (97%) 13,996 (97%) 13,312 (96%) 15,160 (96%)
    1.yes 266 (1.9%) 290 (2.3%) 429 (3.0%) 525 (3.8%) 614 (3.9%)
    Missing 405 (2.9%) 28 (0.2%) 23 (0.2%) 13 (<0.1%) 12 (<0.1%)
5. R Ever had respiratory disease, incl asthma




    0.no 12,915 (91%) 11,437 (91%) 13,165 (91%) 12,337 (89%) 14,127 (89%)
    1.yes 842 (5.9%) 1,056 (8.4%) 1,259 (8.7%) 1,507 (11%) 1,647 (10%)
    Missing 397 (2.8%) 33 (0.3%) 24 (0.2%) 6 (<0.1%) 12 (<0.1%)
6. R Ever had heart attack




    0.no 13,312 (94%) 11,937 (95%) 13,711 (95%) 12,964 (94%) 14,750 (93%)
    1.yes 439 (3.1%) 573 (4.6%) 716 (5.0%) 879 (6.3%) 1,028 (6.5%)
    Missing 403 (2.8%) 16 (0.1%) 21 (0.1%) 7 (<0.1%) 8 (<0.1%)
7. R Ever had stroke




    0.no 13,418 (95%) 12,169 (97%) 14,010 (97%) 13,334 (96%) 15,224 (96%)
    1.yes 325 (2.3%) 350 (2.8%) 424 (2.9%) 504 (3.6%) 554 (3.5%)
    Missing 411 (2.9%) 7 (<0.1%) 14 (<0.1%) 12 (<0.1%) 8 (<0.1%)
8. R Ever had arthritis




    0.no 11,039 (78%) 9,042 (72%) 11,092 (77%) 10,090 (73%) 12,101 (77%)
    1.yes 2,704 (19%) 3,469 (28%) 3,333 (23%) 3,744 (27%) 3,668 (23%)
    Missing 411 (2.9%) 15 (0.1%) 23 (0.2%) 16 (0.1%) 17 (0.1%)
9. R Fallen down last 2 years




    0.no 9,133 (65%) 7,948 (63%) 8,819 (61%) 7,716 (56%) 9,327 (59%)
    1.yes 5,004 (35%) 4,568 (36%) 5,623 (39%) 6,124 (44%) 6,456 (41%)
    Missing 17 (0.1%) 10 (<0.1%) 6 (<0.1%) 10 (<0.1%) 3 (<0.1%)
10. R Ever fractured a bone (including hip) since age 50




    0.no 10,858 (77%) 3,518 (28%) 11,667 (81%) 10,711 (77%) 453 (2.9%)
    1.yes 1,595 (11%) 859 (6.9%) 2,357 (16%) 2,994 (22%) 14,416 (91%)
    Missing 1,701 (12%) 8,149 (65%) 424 (2.9%) 145 (1.0%) 917 (5.8%)
11. R Self-rated eyesight




    1.Excellent 427 (3.0%) 443 (3.5%) 604 (4.2%) 580 (4.2%) 910 (5.8%)
    2.Very good 1,363 (9.6%) 893 (7.1%) 1,168 (8.1%) 909 (6.6%) 1,366 (8.7%)
    3.Good 6,029 (43%) 5,577 (45%) 6,203 (43%) 5,815 (42%) 5,879 (37%)
    4.Fair 4,891 (35%) 4,309 (34%) 5,071 (35%) 5,507 (40%) 6,447 (41%)
    5.Poor 1,295 (9.1%) 1,171 (9.3%) 1,014 (7.0%) 979 (7.1%) 1,097 (6.9%)
    6.Legally Blind 44 (0.3%) 53 (0.4%) 41 (0.3%) 47 (0.3%) 49 (0.3%)
    Missing 105 (0.7%) 80 (0.6%) 347 (2.4%) 13 (<0.1%) 38 (0.2%)
12. R Self-rated hearing




    1.Excellent 802 (5.7%) 589 (4.7%) 7 (<0.1%) 673 (4.9%) 1,215 (7.7%)
    2.Very good 1,961 (14%) 1,435 (11%) 23 (0.2%) 1,331 (9.6%) 1,861 (12%)
    3.Good 7,747 (55%) 7,052 (56%) 47 (0.3%) 7,101 (51%) 7,310 (46%)
    4.Fair 2,774 (20%) 2,631 (21%) 67 (0.5%) 4,121 (30%) 4,757 (30%)
    5.Poor 643 (4.5%) 637 (5.1%) 19 (0.1%) 576 (4.2%) 619 (3.9%)
    6.Legally Deaf 13 (<0.1%) 15 (0.1%) 1 (<0.1%) 3 (<0.1%) 5 (<0.1%)
    Missing 214 (1.5%) 167 (1.3%) 14,284 (99%) 45 (0.3%) 19 (0.1%)
13. R Severe fatigue




    0.no 10,315 (73%) 9,294 (74%) 11,282 (78%) 10,928 (79%) 12,777 (81%)
    1.yes 3,803 (27%) 3,229 (26%) 3,158 (22%) 2,910 (21%) 2,996 (19%)
    Missing 36 (0.3%) 3 (<0.1%) 8 (<0.1%) 12 (<0.1%) 13 (<0.1%)
14. R Difficulty breathing




    0.no 12,101 (85%) 10,737 (86%) 11,855 (82%) 11,160 (81%) 13,007 (82%)
    1.yes 2,026 (14%) 1,784 (14%) 2,588 (18%) 2,681 (19%) 2,771 (18%)
    Missing 27 (0.2%) 5 (<0.1%) 5 (<0.1%) 9 (<0.1%) 8 (<0.1%)
15. R Pain in lower limbs while (or after) walking




    0.no 8,179 (58%) 7,139 (57%)


    1.yes 5,942 (42%) 5,221 (42%)


    Missing 33 (0.2%) 166 (1.3%)


16. R Stomach pain, indigestion, diarrhea




    0.no 11,389 (80%) 9,901 (79%) 11,312 (78%) 10,703 (77%) 12,532 (79%)
    1.yes 2,734 (19%) 2,459 (20%) 3,129 (22%) 3,129 (23%) 3,245 (21%)
    Missing 31 (0.2%) 166 (1.3%) 7 (<0.1%) 18 (0.1%) 9 (<0.1%)
17. R Frequent problems with pain




    0.no 8,298 (59%) 7,594 (61%) 8,935 (62%) 8,532 (62%) 9,561 (61%)
    1.yes 5,838 (41%) 4,929 (39%) 5,508 (38%) 5,314 (38%) 6,218 (39%)
    Missing 18 (0.1%) 3 (<0.1%) 5 (<0.1%) 4 (<0.1%) 7 (<0.1%)
18. R CESD-Felt depressed




    0.No 8,723 (62%) 7,596 (61%) 9,416 (65%) 9,302 (67%) 10,705 (68%)
    1.Yes 5,284 (37%) 4,907 (39%) 5,011 (35%) 4,528 (33%) 5,065 (32%)
    Missing 147 (1.0%) 23 (0.2%) 21 (0.1%) 20 (0.1%) 16 (0.1%)
19. R CESD-Everything an effort




    0.No 8,896 (63%) 7,942 (63%) 9,304 (64%) 8,886 (64%) 10,580 (67%)
    1.Yes 5,088 (36%) 4,568 (36%) 5,122 (35%) 4,947 (36%) 5,181 (33%)
    Missing 170 (1.2%) 16 (0.1%) 22 (0.2%) 17 (0.1%) 25 (0.2%)
20. R CESD-Sleep was restless




    0.No 8,898 (63%) 7,620 (61%) 8,391 (58%) 8,002 (58%) 9,218 (58%)
    1.Yes 5,136 (36%) 4,893 (39%) 6,042 (42%) 5,841 (42%) 6,551 (41%)
    Missing 120 (0.8%) 13 (0.1%) 15 (0.1%) 7 (<0.1%) 17 (0.1%)
21. R CESD-Felt happy




    0.No 3,437 (24%) 3,382 (27%) 2,923 (20%) 2,606 (19%) 3,064 (19%)
    1.Yes 10,559 (75%) 9,095 (73%) 11,492 (80%) 11,208 (81%) 12,678 (80%)
    Missing 158 (1.1%) 49 (0.4%) 33 (0.2%) 36 (0.3%) 44 (0.3%)
22. R CESD-Felt lonely




    0.No 9,407 (66%) 8,409 (67%) 10,086 (70%) 9,648 (70%) 11,153 (71%)
    1.Yes 4,605 (33%) 4,096 (33%) 4,351 (30%) 4,191 (30%) 4,606 (29%)
    Missing 142 (1.0%) 21 (0.2%) 11 (<0.1%) 11 (<0.1%) 27 (0.2%)
23. R CESD-Felt sad




    0.No 8,423 (60%) 7,423 (59%) 8,594 (59%) 8,370 (60%) 9,691 (61%)
    1.Yes 5,583 (39%) 5,091 (41%) 5,839 (40%) 5,465 (39%) 6,069 (38%)
    Missing 148 (1.0%) 12 (<0.1%) 15 (0.1%) 15 (0.1%) 26 (0.2%)
24. R enjoyed life




    0.no 3,975 (28%) 3,917 (31%) 3,296 (23%) 2,888 (21%) 3,429 (22%)
    1.yes 9,943 (70%) 8,384 (67%) 11,117 (77%) 10,913 (79%) 12,306 (78%)
    Missing 236 (1.7%) 225 (1.8%) 35 (0.2%) 49 (0.4%) 51 (0.3%)
25. R CESD-Felt tired




    0.No 5,665 (40%) 5,331 (43%) 5,906 (41%) 5,536 (40%) 6,602 (42%)
    1.Yes 8,347 (59%) 7,184 (57%) 8,534 (59%) 8,308 (60%) 9,165 (58%)
    Missing 142 (1.0%) 11 (<0.1%) 8 (<0.1%) 6 (<0.1%) 19 (0.1%)
26. R CESD-Had a lot of energy




    0.No 7,770 (55%) 7,512 (60%) 7,479 (52%) 7,298 (53%) 8,433 (53%)
    1.Yes 6,168 (44%) 4,958 (40%) 6,940 (48%) 6,531 (47%) 7,305 (46%)
    Missing 216 (1.5%) 56 (0.4%) 29 (0.2%) 21 (0.2%) 48 (0.3%)
27. R Difficulty-Picking up a coin




    0.No 13,324 (94%) 11,785 (94%) 13,391 (93%) 12,796 (92%) 14,469 (92%)
    1.Yes 767 (5.4%) 716 (5.7%) 1,054 (7.3%) 1,006 (7.3%) 1,290 (8.2%)
    Missing 63 (0.4%) 25 (0.2%) 3 (<0.1%) 48 (0.3%) 27 (0.2%)
28. R Difficulty-Dressing




    0.No 13,147 (93%) 11,745 (94%) 13,103 (91%) 12,296 (89%) 14,151 (90%)
    1.Yes 901 (6.4%) 756 (6.0%) 1,340 (9.3%) 1,506 (11%) 1,608 (10%)
    Missing 106 (0.7%) 25 (0.2%) 5 (<0.1%) 48 (0.3%) 27 (0.2%)
29. R Difficulty-Walking several blocks




    0.No 10,634 (75%) 9,382 (75%) 10,487 (73%) 9,638 (70%) 11,576 (73%)
    1.Yes 3,464 (24%) 3,118 (25%) 3,957 (27%) 4,166 (30%) 4,187 (27%)
    Missing 56 (0.4%) 26 (0.2%) 4 (<0.1%) 46 (0.3%) 23 (0.1%)
30. R Difficulty-Walking across room




    0.No 13,339 (94%) 11,896 (95%) 13,621 (94%) 12,873 (93%) 14,756 (93%)
    1.Yes 601 (4.2%) 581 (4.6%) 790 (5.5%) 928 (6.7%) 982 (6.2%)
    Missing 214 (1.5%) 49 (0.4%) 37 (0.3%) 49 (0.4%) 48 (0.3%)
31. R Difficulty-Bathing or showering




    0.No 13,529 (96%) 12,110 (97%) 13,946 (97%) 13,142 (95%) 15,134 (96%)
    1.Yes 411 (2.9%) 367 (2.9%) 466 (3.2%) 659 (4.8%) 605 (3.8%)
    Missing 214 (1.5%) 49 (0.4%) 36 (0.2%) 49 (0.4%) 47 (0.3%)
32. R Difficulty-Eating




    0.No 13,729 (97%) 12,302 (98%) 14,070 (97%) 13,433 (97%) 15,452 (98%)
    1.Yes 211 (1.5%) 175 (1.4%) 342 (2.4%) 368 (2.7%) 287 (1.8%)
    Missing 214 (1.5%) 49 (0.4%) 36 (0.2%) 49 (0.4%) 47 (0.3%)
33. R Difficulty-Getting in/out of bed




    0.No 13,289 (94%) 11,943 (95%) 13,420 (93%) 12,632 (91%) 14,774 (94%)
    1.Yes 651 (4.6%) 534 (4.3%) 990 (6.9%) 1,169 (8.4%) 963 (6.1%)
    Missing 214 (1.5%) 49 (0.4%) 38 (0.3%) 49 (0.4%) 49 (0.3%)
34. R Difficulty-Using the toilet




    0.No 13,481 (95%) 12,113 (97%) 13,715 (95%) 13,099 (95%) 14,924 (95%)
    1.Yes 444 (3.1%) 364 (2.9%) 695 (4.8%) 702 (5.1%) 812 (5.1%)
    Missing 229 (1.6%) 49 (0.4%) 38 (0.3%) 49 (0.4%) 50 (0.3%)
35. R Difficulty-Preparing hot meals




    0.No 12,841 (91%) 11,404 (91%) 13,190 (91%) 12,560 (91%) 14,701 (93%)
    1.Yes 1,160 (8.2%) 1,094 (8.7%) 1,250 (8.7%) 1,241 (9.0%) 739 (4.7%)
    Missing 153 (1.1%) 28 (0.2%) 8 (<0.1%) 49 (0.4%) 346 (2.2%)
36. R Difficulty-Shopping for groceries




    0.No 12,858 (91%) 11,419 (91%) 12,924 (89%) 12,119 (88%) 14,362 (91%)
    1.Yes 1,158 (8.2%) 1,080 (8.6%) 1,516 (10%) 1,682 (12%) 1,392 (8.8%)
    Missing 138 (1.0%) 27 (0.2%) 8 (<0.1%) 49 (0.4%) 32 (0.2%)
37. R Difficulty-Taking medications




    0.No 13,692 (97%) 12,177 (97%) 13,944 (97%) 13,143 (95%) 15,328 (97%)
    1.Yes 364 (2.6%) 323 (2.6%) 501 (3.5%) 659 (4.8%) 425 (2.7%)
    Missing 98 (0.7%) 26 (0.2%) 3 (<0.1%) 48 (0.3%) 33 (0.2%)
38. R Difficulty-Managing money




    0.No 13,693 (97%) 12,188 (97%) 14,056 (97%) 13,321 (96%) 15,350 (97%)
    1.Yes 359 (2.5%) 311 (2.5%) 383 (2.7%) 480 (3.5%) 404 (2.6%)
    Missing 102 (0.7%) 27 (0.2%) 9 (<0.1%) 49 (0.4%) 32 (0.2%)
39. R Leaks urine when coughing (last 2 yrs)




    0.no 13,023 (92%) 11,256 (90%) 12,203 (84%) 11,204 (81%) 13,088 (83%)
    1.yes 1,098 (7.8%) 1,098 (8.8%) 2,234 (15%) 2,633 (19%) 2,682 (17%)
    Missing 33 (0.2%) 172 (1.4%) 11 (<0.1%) 13 (<0.1%) 16 (0.1%)
1 Deficit 19 to 27 combines into one deficit of CESD-depressive symptoms so total deficits is 31

7 Explore distribution of number of available deficits per observation by wave

Note: maximum number of deficits are actually 31 since 9 deficits combines into CESD-depressive symptoms.

Number of available deficits per observation by wave
Characteristic Wave_5 Wave_4 Wave_3 Wave_2 Wave_1
N = 15,7861 N = 13,8501 N = 14,4481 N = 12,5261 N = 14,1541
Number of available deficits




    31


4,200 (34%) 11,407 (81%)
    30 14,404 (91%) 13,514 (98%) 157 (1.1%) 7,889 (63%) 1,895 (13%)
    29 1,282 (8.1%) 265 (1.9%) 13,354 (92%) 166 (1.3%) 173 (1.2%)
    28 31 (0.2%) 17 (0.1%) 850 (5.9%) 85 (0.7%) 42 (0.3%)
    27 5 (<0.1%) 3 (<0.1%) 36 (0.2%) 125 (1.0%) 27 (0.2%)
    26 8 (<0.1%) 1 (<0.1%) 3 (<0.1%) 12 (<0.1%) 141 (1.0%)
    25 22 (0.1%) 4 (<0.1%) 3 (<0.1%) 22 (0.2%) 35 (0.2%)
    24 4 (<0.1%)
33 (0.2%) 2 (<0.1%) 308 (2.2%)
    23 1 (<0.1%)
6 (<0.1%)
62 (0.4%)
    22 1 (<0.1%)
4 (<0.1%)
4 (<0.1%)
    21 1 (<0.1%)


3 (<0.1%)
    20



1 (<0.1%)
    19 4 (<0.1%)

5 (<0.1%) 40 (0.3%)
    18 19 (0.1%) 35 (0.3%)
14 (0.1%) 7 (<0.1%)
    17 2 (<0.1%) 2 (<0.1%) 1 (<0.1%) 2 (<0.1%) 2 (<0.1%)
    16
1 (<0.1%)
2 (<0.1%)
    15 1 (<0.1%)
1 (<0.1%) 1 (<0.1%)
    14
5 (<0.1%)

2 (<0.1%)
    12
1 (<0.1%)


    11



1 (<0.1%)
    9
1 (<0.1%)


    5 1 (<0.1%)



    3
1 (<0.1%)


    1


1 (<0.1%) 4 (<0.1%)
Categories




    <30 1,382 (8.8%) 336 (2.4%) 14,291 (99%) 437 (3.5%) 852 (6.0%)
    >=30 14,404 (91%) 13,514 (98%) 157 (1.1%) 12,089 (97%) 13,302 (94%)
1 n (%)

8 Create Frailty Index

  • Sum frailty deficit codes for each observation (i.e, for all deficits)

  • Then divide total score per observation by total number of included deficits (by wave)

  • Frailty Index Versions: based on number of deficits included due to deficit exceeding % missing threshold (i.e proportion of the deficit missing)

    • Version 5% missing:
      • In waves 1, 2, 3, 4, 5 … 1, 1, 1, 0, 1 deficits had >= 5% missing. Plus R Pain in lower limbs while (or after) walking not available in waves 3 to 5
      • Thus 30/31, 30/31, 29/31, 30/31, 29/31 deficits available in waves 1, 2, 3, 4, 5
    • Version 10% missing: same stats as with 5% missing cut-off
  • For cut-off by number of deficits missing per observation,

    • Each observation must not have more than 20% of deficits missing
    • Thus if >= 20% of deficits are missing by observation, Frailty Index = missing
    • 20% of 31=5.8, 30=5.2,21=4.2, to round down to the nearest whole number

8.1 Check the specific deficits to exclude by wave based on missing thresholds

  • Note: 5% and 10% missingness cut-off yields the same exclusions since variable missingness is >= 10%.
Wave 1, 10% cut-off
No. deficit percent_missing
1 r1hip50e 12.0178
Wave 2, 10% cut-off
No. deficit percent_missing
1 r2hip50e 65.05668
Wave 3, 10% cut-off
No. deficit percent_missing
1 r3hearing 98.86489
2 r3walkpain 100.00000
Wave 4, 10% cut-off
No. deficit percent_missing
1 r4walkpain 100
Wave 5, 10% cut-off
No. deficit percent_missing
1 r5walkpain 100
Wave 1, 5% cut-off
No. deficit percent_missing
1 r1hip50e 12.0178
Wave 2, 5% cut-off
No. deficit percent_missing
1 r2hip50e 65.05668
Wave 3, 5% cut-off
No. deficit percent_missing
1 r3hearing 98.86489
2 r3walkpain 100.00000
Wave 4, 5% cut-off
No. deficit percent_missing
1 r4walkpain 100
Wave 5, 5% cut-off
No. deficit percent_missing
1 r5hip50e 5.808945
2 r5walkpain 100.000000

8.2 Create Frailty Index: version 10% and 5% deficit missing cut-off

9 Distribution of frailty score by wave

  • Row 1 is 10% deficit missing cut-off
  • Row 2 is 5% deficit missing cut-off

Continuous and categorical frailty (5% and 10% missig deficit cut-off) by wave
Characteristic Wave1 Wave2 Wave3 Wave4
N = 14,1541 N = 12,5261 N = 14,4481 N = 13,8501
frailty_10



    Mean (SD) 0.17 (0.12) 0.18 (0.12) 0.17 (0.12) 0.20 (0.13)
    Median (IQR) 0.14 (0.08, 0.23) 0.15 (0.09, 0.23) 0.14 (0.08, 0.22) 0.17 (0.11, 0.26)
    Range 0.00, 0.82 0.00, 0.81 0.00, 0.88 0.00, 0.84
    Missing 434 27 12 46
frailty_5



    Mean (SD) 0.17 (0.12) 0.18 (0.12) 0.17 (0.12) 0.20 (0.13)
    Median (IQR) 0.14 (0.08, 0.23) 0.15 (0.09, 0.23) 0.14 (0.08, 0.22) 0.17 (0.11, 0.26)
    Range 0.00, 0.82 0.00, 0.81 0.00, 0.88 0.00, 0.84
    Missing 434 27 12 46
frailty_10_cat



    Not Frail 10,982 (78%) 9,736 (78%) 11,468 (79%) 10,079 (73%)
    Frail 2,738 (19%) 2,763 (22%) 2,968 (21%) 3,725 (27%)
    Missing 434 (3.1%) 27 (0.2%) 12 (<0.1%) 46 (0.3%)
frailty_5_cat



    Not Frail 10,982 (78%) 9,736 (78%) 11,468 (79%) 10,079 (73%)
    Frail 2,738 (19%) 2,763 (22%) 2,968 (21%) 3,725 (27%)
    Missing 434 (3.1%) 27 (0.2%) 12 (<0.1%) 46 (0.3%)
1 n (%)

10 Frailty distribution vs Garcia-Gonzales, etal, 2009 study

The frailty index by Garcia-Gonzales, etal, 2009 had an average value of 0.16 in the target population (standard deviation = 0.11), ranging in the sample from 0 to 0.65 and showing a right-skewed distribution. Figure is below.

Differences with Garcia-Gonzales, etal, 2009:

  • 6 deficits excluded, medically diagnosed conditions before age 10, with likely very low prevalence. This reduces denominator so makes sense wave 1 score slightly (0.02 points) higher.
  • Includes all ages thus 40+, against 65+ by Garcia

Distribution of the fraility index in the Mexican elderly population (65+ years). Data from the Mexican Health and Ageing Study, 2001. Sample weighted estimates.

11 Codebook of frailty calculation

The table below summarizes the the methods and definitions for the frailty score estimation

Health deficits included in the frailty index and their coding
No. Deficit description Deficit coding Deficit availability (or > 5% not missing)
1 R Self-report of health Excellent = 0, Very good = 0.25, Good = 0.5, Fair = 0.75, Poor = 1 waves 1, 2, 3, 4, 5
2 R Ever had high blood pressure No = 0, Yes = 1 waves 1, 2, 3, 4, 5
3 R Ever had diabetes No = 0, Yes = 1 waves 1, 2, 3, 4, 5
4 R Ever had cancer No = 0, Yes = 1 waves 1, 2, 3, 4, 5
5 R Ever had respiratory disease, incl asthma No = 0, Yes = 1 waves 1, 2, 3, 4, 5
6 R Ever had heart attack No = 0, Yes = 1 waves 1, 2, 3, 4, 5
7 R Ever had stroke No = 0, Yes = 1 waves 1, 2, 3, 4, 5
8 R Ever had arthritis No = 0, Yes = 1 waves 1, 2, 3, 4, 5
9 R Fallen down last 2 years No = 0, Yes = 1 waves 1, 2, 3, 4, 5
10 R Ever fractured a bone (including hip) since age 50 No = 0, Yes = 1 waves 1, 2, 3, 4, 5
11 R Self-rated eyesight Excellent = 0, Very good = 0.2, Good = 0.4, Fair = 0.6, Poor = 0.8, Legally Blind = 1 waves 1, 2, 3, 4, 5
12 R Self-rated hearing Excellent = 0, Very good = 0.2, Good = 0.4, Fair = 0.6, Poor = 0.8, Legally Deaf = 1 waves 1, 2, 4, 5
13 R Severe fatigue No = 0, Yes = 1 waves 1, 2, 3, 4, 5
14 R Difficulty breathing No = 0, Yes = 1 waves 1, 2, 3, 4, 5
15 R Pain in lower limbs while (or after) walking No = 0, Yes = 1 waves 1, 2
16 R Stomach pain, indigestion, diarrhea No = 0, Yes = 1 waves 1, 2, 3, 4, 5
17 R Frequent problems with pain No = 0, Yes = 1 waves 1, 2, 3, 4, 5
18 R CESD-Felt depressed No = 0, Yes = 1/9 waves 1, 2, 3, 4, 5
19 R CESD-Everything an effort No = 0, Yes = 1/9 waves 1, 2, 3, 4, 5
20 R CESD-Sleep was restless No = 0, Yes = 1/9 waves 1, 2, 3, 4, 5
21 R CESD-Felt happy No = 1/9, Yes = 0 waves 1, 2, 3, 4, 5
22 R CESD-Felt lonely No = 0, Yes = 1/9 waves 1, 2, 3, 4, 5
23 R CESD-Felt sad No = 0, Yes = 1/9 waves 1, 2, 3, 4, 5
24 R enjoyed life No = 1/9, Yes = 0 waves 1, 2, 3, 4, 5
25 R CESD-Felt tired No = 0, Yes = 1/9 waves 1, 2, 3, 4, 5
26 R CESD-Had a lot of energy No = 1/9, Yes = 0 waves 1, 2, 3, 4, 5
27 R Difficulty-Picking up a coin No = 0, Yes = 1 waves 1, 2, 3, 4, 5
28 R Difficulty-Dressing No = 0, Yes = 1 waves 1, 2, 3, 4, 5
29 R Difficulty-Walking several blocks No = 0, Yes = 1 waves 1, 2, 3, 4, 5
30 R Difficulty-Walking across room No = 0, Yes = 1 waves 1, 2, 3, 4, 5
31 R Difficulty-Bathing or showering No = 0, Yes = 1 waves 1, 2, 3, 4, 5
32 R Difficulty-Eating No = 0, Yes = 1 waves 1, 2, 3, 4, 5
33 R Difficulty-Getting in/out of bed No = 0, Yes = 1 waves 1, 2, 3, 4, 5
34 R Difficulty-Using the toilet No = 0, Yes = 1 waves 1, 2, 3, 4, 5
35 R Difficulty-Preparing hot meals No = 0, Yes = 1 waves 1, 2, 3, 4, 5
36 R Difficulty-Shopping for groceries No = 0, Yes = 1 waves 1, 2, 3, 4, 5
37 R Difficulty-Taking medications No = 0, Yes = 1 waves 1, 2, 3, 4, 5
38 R Difficulty-Managing money No = 0, Yes = 1 waves 1, 2, 3, 4, 5
39 R Leaks urine when coughing (last 2 yrs) No = 0, Yes = 1 waves 1, 2, 3, 4, 5
Total number of deficits available: waves 1, 2 = 39, wave 3 = 37, waves 4, 5 = 38.
9 deficits coded each as Yes = 1/9 combines into one deficit = Depressive symptoms.
Thus total number of denominators (total deficits) in frailty score calculation is thus, 30 (waves 1 and 2), 29 (wave 3) and 30 (waves 4 and 5).
The final frailty score per observation in each wave = total number of deficits (sum) of that observation / denominator of that wave.

12 Extract MHAS data for the different MSC projects

12.1 Aleixo’s project: Frailty trajectories in MHAS

12.1.1 Generate cognitive function variable by wave

Background - Cognitive function variable is not readily available but the MHAS group established the classification of cognitive function status using the combination of different conditions. There is a corresponding documentation with STATA code also available The Mexican Health and Aging Study:Cognitive Function Measures Scoring and Classification Across Waves. - The algorithm classifies an individual into one of 3 cognitive function categories; 1. normal 2. cognitive impairment no-dementia (CIND) and 3. dementia - the algorithms are specific to type of interviews but we are using those for direct interviews since this data extraction includes only direct interviews. - Cognitive function has been assessed in the MHAS using the screening portion of the Cross-Cultural Cognitive Examination (CCCE) for direct interviews (Glosser et al. 1993). - The CCCE has been modified over the MHAS waves to include a more comprehensive cognitive evaluation. - In 2001 (wave1), five tasks were included in the MHAS. - In 2003 (wave2), an additional task was included to measure orientation (marked in green in the list below). - In 2012 (wave3), semantic verbal fluency and backward counting tasks were added (marked in purple in the list below). - In 2015 (wave4), an additional task – serial 7 – was added (marked in red in the list below). - The tasks/variables used in calculating cognitive function are below.

  • Constructional Praxis (visuo-constructional). measured by presenting two geometrical figures and asking respondents to copy the figures in 90 seconds.
  • Verbal Fluency. added in wave 3. measured by asking respondents to list all the animals they can think of in the next 60 seconds.
  • Serial 7s.added in wave 4. measured by asking the respondent to sequentially subtract 7 starting from 100, until they complete five successive subtractions.
  • Verbal learning. measured by asking respondents to listen to a list of eight words and to repeat them.
  • Visual scanning. measured by asking respondents to circle all figures that are identical to a specific stimulus shown previously within an array of different stimuli in 60 seconds.
  • Backward Counting. added in wave 3. measured by asking respondents to count backwards from 20 to 0 as fast as possible in 60 seconds.
  • Constructional Praxis Recall. measured by asking respondents to remember the figures they copied in 1 and to draw them from memory on a blank piece of paper in 3 minutes.
  • Delayed Verbal recall. measured by asking respondents to repeat as many of the words they remember from the list provided in the verbal learning task.
  • Day, Month and Year. added in wave 2. This task asks respondents to indicate the day, month and year of the interview.
  • NOTE: If study aims to compare cognitive function across waves, it is recommended that only the first 5 tasks in wave 1 is used across waves.

Approach: Direct interviews

  1. Generate Z-score for each task.
  2. Define impairment for each task, based on Z score,
  3. Generate total number of tasks with impairment. Individual should complete at least 2/5 of tasks included in wave1 else cognitive function set to missing.
  4. Generate total number of difficulties with Instrumental Activities of Daily Living (IADL). Difficulty with meal preparation, shopping, taking medication and managing money.
  5. Classify cognitive function status as:
  • Normal; No. of tasks with impairment = 0-1 and No. of difficulties with IADL = 0.
  • CIND; No. of tasks with impairment = 2+ and No. of difficulties with IADL = 0.
  • Dementia; No. of tasks with impairment = 2+ and No. of difficulties with IADL = 1-4.

cognitive function variable has been generated accordingly by MHAS and available for download here

Constructed Data