Demographic Table


The table below details the size of the sample, the demographic breakdown of the sample, and a summary of the key variables of interest (frailty, total recommendations, recommendation subtypes, matters most primary grouping, matters most specificity)


Frailty breakdown


The table below summarizes the demographics, recommendations, recommendations subcategories, matters most groups, and matters most specificity by frailty group (based on VAFI)


The frail and pre-frail groups are slightly older than the non-frail group (mean (SD) age = 80 (8) for frail, 80 (7) for pre-frail, 77 (7) for non-frail), but the groups are similar with respect to sex and race. The frail group has a higher proportion of subjects who are from urban areas (72%) compared to the non-frail group (52%).


Non-frail
(N=21)		 Pre-frail
(N=45)		 Frail
(N=61)		 P-value
vafi_va
Mean (SD) 0.0783 (0.0281) 0.158 (0.0274) 0.331 (0.103) <0.001
Median [Min, Max] 0.0968 [0, 0.0968] 0.161 [0.129, 0.194] 0.290 [0.226, 0.645]
Age
Mean (SD) 77.0 (6.50) 80.4 (6.95) 79.9 (7.59) 0.191
Median [Min, Max] 78.0 [64.0, 92.0] 80.0 [65.0, 94.0] 79.0 [64.0, 98.0]
Sex
Female 1 (4.8%) 4 (8.9%) 4 (6.6%) 0.81
Male 20 (95.2%) 41 (91.1%) 57 (93.4%)
Race
Asian 1 (4.8%) 2 (4.4%) 0 (0%) 0.691
Pacific Islander 1 (4.8%) 1 (2.2%) 2 (3.3%)
White 18 (85.7%) 38 (84.4%) 50 (82.0%)
Black or African American 0 (0%) 2 (4.4%) 4 (6.6%)
American Indian 0 (0%) 0 (0%) 1 (1.6%)
Missing 1 (4.8%) 2 (4.4%) 4 (6.6%)
Rurality
Rural 9 (42.9%) 14 (31.1%) 16 (26.2%) 0.309
Urban 11 (52.4%) 31 (68.9%) 44 (72.1%)
Missing 1 (4.8%) 0 (0%) 1 (1.6%)
TotalRecs
Mean (SD) 3.24 (1.79) 4.31 (3.01) 4.21 (2.30) 0.236
Median [Min, Max] 3.00 [1.00, 7.00] 4.00 [1.00, 14.0] 4.00 [1.00, 11.0]
Mobility_bi
No 8 (38.1%) 15 (33.3%) 19 (31.1%) 0.842
Yes 13 (61.9%) 30 (66.7%) 42 (68.9%)
Mentation_bi
No 14 (66.7%) 27 (60.0%) 33 (54.1%) 0.577
Yes 7 (33.3%) 18 (40.0%) 28 (45.9%)
SSS_bi
No 11 (52.4%) 27 (60.0%) 36 (59.0%) 0.832
Yes 10 (47.6%) 18 (40.0%) 25 (41.0%)
Meds_bi
No 8 (38.1%) 13 (28.9%) 9 (14.8%) 0.055
Yes 13 (61.9%) 32 (71.1%) 52 (85.2%)
AllOther_bi
No 19 (90.5%) 38 (84.4%) 53 (86.9%) 0.796
Yes 2 (9.5%) 7 (15.6%) 8 (13.1%)
rec_types_total
Mean (SD) 2.14 (0.910) 2.33 (1.11) 2.54 (1.07) 0.292
Median [Min, Max] 2.00 [1.00, 4.00] 2.00 [1.00, 5.00] 3.00 [1.00, 5.00]
subcat1_group
Connecting 3 (14.3%) 9 (20.0%) 12 (19.7%) 0.7
Functioning 12 (57.1%) 18 (40.0%) 28 (45.9%)
Managing health 4 (19.0%) 10 (22.2%) 11 (18.0%)
Enjoying life 0 (0%) 4 (8.9%) 7 (11.5%)
Missing 2 (9.5%) 4 (8.9%) 3 (4.9%)
as.factor(Specificity)
0 2 (9.5%) 1 (2.2%) 1 (1.6%) 0.123
1 11 (52.4%) 16 (35.6%) 20 (32.8%)
2 7 (33.3%) 27 (60.0%) 37 (60.7%)
Missing 1 (4.8%) 1 (2.2%) 3 (4.9%)
rx_count
Mean (SD) 5.62 (3.91) 7.72 (4.77) 10.3 (6.61) 0.012
Median [Min, Max] 6.00 [1.00, 12.0] 7.00 [1.00, 19.0] 10.0 [1.00, 28.0]
Missing 8 (38.1%) 6 (13.3%) 6 (9.8%)




As seen in the table above and the figure below, the non-frail group seems to have a slightly lower number of total recommendations per person compared to the pre-frail and frail groups (mean (SD) recommendations = 3.2 (1.8) for non-frail, 4.3 (3) for pre-frail, 4.2 (2) for frail). We’ve talked about looking into this before: comparing the types of recommendations that non-frail vs. frail people get, specifcally we can compare low frailty + high recs vs. high frailty + high recs, etc. This would be to see how the number and types of recs compare for people with low vs. high frailty. I’ve shared IDs to look into in the past





Next we can breakdown where the total recommendations are coming from for each frailty grouping. There is a pattern where the more frail individuals seem slightly more likely to get a recommendation in the medications, mentation, and mobility subcategories. But, it seems like this pattern does not exists for the SSS category, where the non-frail group has the highest proportion of subjects receiving a recommendation.





Recs by the full spectrum of VAFI



Matters most




The most common matters most group was “functioning”. Note that this figure includes “double dippers”, counting those subjects twice.




For non-double dippers





Below is a table with demographic, frailty, recommendation, and specificity breakdowns for the sample by matters most group. Note that this table does NOT include “double dippers”, subjects are only counted for their primary group. The matters most groups are similar demographically, though the average age of the “enjoying life” group is slightly lower than the others. In terms of frailty, all 3 frailty measures seem to suggest that the “managing health” group has a slightly lower level of frailty compared to the other 3 groups.


Connecting
(N=24)		 Enjoying life
(N=12)		 Functioning
(N=58)		 Managing health
(N=25)		 P-value
Age
Mean (SD) 79.6 (7.51) 76.3 (7.27) 80.2 (7.05) 79.8 (7.60) 0.42
Median [Min, Max] 80.5 [66.0, 93.0] 76.0 [64.0, 91.0] 80.0 [64.0, 97.0] 78.0 [65.0, 98.0]
Sex
Male 24 (100%) 11 (91.7%) 53 (91.4%) 22 (88.0%) 0.428
Female 0 (0%) 1 (8.3%) 5 (8.6%) 3 (12.0%)
Race
Asian 1 (4.2%) 0 (0%) 1 (1.7%) 1 (4.0%) 0.385
Black or African American 3 (12.5%) 0 (0%) 2 (3.4%) 1 (4.0%)
White 19 (79.2%) 8 (66.7%) 52 (89.7%) 21 (84.0%)
Pacific Islander 0 (0%) 1 (8.3%) 0 (0%) 2 (8.0%)
American Indian 0 (0%) 0 (0%) 1 (1.7%) 0 (0%)
Missing 1 (4.2%) 3 (25.0%) 2 (3.4%) 0 (0%)
cMort_90d
Mean (SD) 81.8 (14.5) 79.5 (16.8) 82.1 (13.8) 76.9 (18.2) 0.53
Median [Min, Max] 87.5 [50.0, 98.0] 80.0 [40.0, 99.0] 85.0 [45.0, 98.0] 85.0 [35.0, 99.0]
Missing 0 (0%) 1 (8.3%) 0 (0%) 0 (0%)
JFI
Mean (SD) 5.63 (2.53) 5.91 (1.70) 5.45 (1.59) 5.20 (1.68) 0.722
Median [Min, Max] 6.00 [2.00, 10.0] 7.00 [4.00, 8.00] 5.50 [2.00, 9.00] 5.00 [2.00, 9.00]
Missing 0 (0%) 1 (8.3%) 0 (0%) 0 (0%)
vafi_va
Mean (SD) 0.249 (0.153) 0.264 (0.0966) 0.219 (0.112) 0.228 (0.147) 0.639
Median [Min, Max] 0.210 [0.0645, 0.581] 0.290 [0.129, 0.419] 0.194 [0.0323, 0.548] 0.194 [0, 0.645]
Missing 0 (0%) 1 (8.3%) 0 (0%) 0 (0%)
vafi_group3
Non-frail 3 (12.5%) 0 (0%) 12 (20.7%) 4 (16.0%) 0.7
Pre-frail 9 (37.5%) 4 (33.3%) 18 (31.0%) 10 (40.0%)
Frail 12 (50.0%) 7 (58.3%) 28 (48.3%) 11 (44.0%)
Missing 0 (0%) 1 (8.3%) 0 (0%) 0 (0%)
TotalRecs
Mean (SD) 4.00 (2.15) 3.92 (2.27) 4.10 (2.65) 4.52 (2.60) 0.861
Median [Min, Max] 3.00 [1.00, 9.00] 3.50 [1.00, 7.00] 4.00 [1.00, 14.0] 4.00 [1.00, 13.0]
Mobility_bi
No 8 (33.3%) 2 (16.7%) 20 (34.5%) 9 (36.0%) 0.657
Yes 16 (66.7%) 10 (83.3%) 38 (65.5%) 16 (64.0%)
Mentation_bi
No 15 (62.5%) 6 (50.0%) 34 (58.6%) 14 (56.0%) 0.904
Yes 9 (37.5%) 6 (50.0%) 24 (41.4%) 11 (44.0%)
SSS_bi
No 10 (41.7%) 8 (66.7%) 35 (60.3%) 14 (56.0%) 0.392
Yes 14 (58.3%) 4 (33.3%) 23 (39.7%) 11 (44.0%)
Meds_bi
No 5 (20.8%) 3 (25.0%) 14 (24.1%) 6 (24.0%) 0.988
Yes 19 (79.2%) 9 (75.0%) 44 (75.9%) 19 (76.0%)
AllOther_bi
No 20 (83.3%) 10 (83.3%) 54 (93.1%) 19 (76.0%) 0.183
Yes 4 (16.7%) 2 (16.7%) 4 (6.9%) 6 (24.0%)
as.factor(Specificity)
0 2 (8.3%) 0 (0%) 0 (0%) 0 (0%) 0.025
1 8 (33.3%) 1 (8.3%) 27 (46.6%) 11 (44.0%)
2 14 (58.3%) 11 (91.7%) 31 (53.4%) 14 (56.0%)



Matters most: one group vs. double dippers




Single dippers vs. double dippers



One domain
(N=51)		 Two domains
(N=68)		 P-value
vafi_va
Mean (SD) 0.216 (0.138) 0.243 (0.118) 0.247
Median [Min, Max] 0.194 [0, 0.645] 0.226 [0.0645, 0.581]
Missing 0 (0%) 1 (1.5%)
Age
Mean (SD) 81.1 (7.44) 78.5 (7.01) 0.055
Median [Min, Max] 80.0 [64.0, 98.0] 78.5 [64.0, 96.0]
Sex
Female 2 (3.9%) 7 (10.3%) 0.342
Male 49 (96.1%) 61 (89.7%)
Race
Asian 3 (5.9%) 0 (0%) 0.302
Black or African American 3 (5.9%) 3 (4.4%)
Pacific Islander 1 (2.0%) 2 (2.9%)
White 43 (84.3%) 57 (83.8%)
American Indian 0 (0%) 1 (1.5%)
Missing 1 (2.0%) 5 (7.4%)
Rurality
Rural 14 (27.5%) 20 (29.4%) 0.99
Urban 36 (70.6%) 47 (69.1%)
Missing 1 (2.0%) 1 (1.5%)
TotalRecs
Mean (SD) 4.47 (2.63) 3.91 (2.37) 0.227
Median [Min, Max] 4.00 [1.00, 13.0] 4.00 [1.00, 14.0]
Mobility_bi
No 14 (27.5%) 25 (36.8%) 0.382
Yes 37 (72.5%) 43 (63.2%)
Mentation_bi
No 36 (70.6%) 33 (48.5%) 0.026
Yes 15 (29.4%) 35 (51.5%)
SSS_bi
No 25 (49.0%) 42 (61.8%) 0.23
Yes 26 (51.0%) 26 (38.2%)
Meds_bi
No 12 (23.5%) 16 (23.5%) 1
Yes 39 (76.5%) 52 (76.5%)
AllOther_bi
No 40 (78.4%) 63 (92.6%) 0.048
Yes 11 (21.6%) 5 (7.4%)
rec_types_total
Mean (SD) 2.51 (1.07) 2.37 (1.02) 0.462
Median [Min, Max] 2.00 [1.00, 5.00] 2.00 [1.00, 4.00]
as.factor(Specificity)
0 2 (3.9%) 0 (0%) 0.175
1 22 (43.1%) 25 (36.8%)
2 27 (52.9%) 43 (63.2%)
rx_count
Mean (SD) 8.29 (5.91) 9.58 (5.98) 0.28
Median [Min, Max] 8.00 [1.00, 24.0] 9.00 [1.00, 28.0]
Missing 6 (11.8%) 11 (16.2%)







The number of recommendations given to those in each group is similar





The next step is took look at what proportion of people within each matters most group got each recommendation type. Most of the groups have similar proportions for each recommendation type, but it seems like the “Enjoying life” group has a slighly higher proportion of subjects getting mentation and mobility recommendations. The “Connecting” group has the highest proportion of subjects receiving an SSS recommendation.





On specificity


The only subjects who had a matters most specificity of 0 had statements in the “Connecting” category. This would be a good thing to look into: let’s check out the MM statements of those with 0 specificity and ask if theres something about “connecting” that makes these vague



The proportions of people in the 1 and 2 specificity groups was pretty even across the matters most groups, except for the “Enjoying life” group, where there was a much higher proportion of people in the “2” group. The sample size is relatively low for this group (n=1 with specificity=1, n=11 with specificity=2) This would be a another good thing to look into: let’s check out the MM statements of those in the “Enjoying life” group, why were their MM statements so much more specific?





Specificity



It’s also worth considering how matters most specificity relates to the number of recommendations subjects get. The figure below shows that those with a specificity of 0 had the least total recommendations (though there are only 4 people with a specificity of 0, so don’t read into this). Interestingly, there is no difference between the number of recommendations that people with specificity = 1 and specificity = 2 get. This would be a another good thing to look into: lets pick out IDs for some people who have a specificity of 1 and a specificity of 2. Maybe the number of recs these people get isn’t different, but are the TYPES of recs different? Are those with specificity=2 getting recommendations that are better tailored to them? It may be best to pick subjects who have the same MM group and same frailty group, but different specificity values.





RX count vs. medication recommendations


## `geom_smooth()` using formula = 'y ~ x'
## Warning: Removed 21 rows containing non-finite outside the scale range
## (`stat_smooth()`).
## Warning: Removed 21 rows containing missing values or values outside the scale range
## (`geom_point()`).

## `geom_smooth()` using formula = 'y ~ x'
## Warning: Removed 21 rows containing non-finite outside the scale range
## (`stat_smooth()`).
## Removed 21 rows containing missing values or values outside the scale range
## (`geom_point()`).

## `geom_smooth()` using formula = 'y ~ x'
## Warning: Removed 20 rows containing non-finite outside the scale range
## (`stat_smooth()`).
## Warning: Removed 20 rows containing missing values or values outside the scale range
## (`geom_point()`).

## 
##  Pearson's product-moment correlation
## 
## data:  ppc3$Meds and ppc3$rx_count
## t = 2.1957, df = 105, p-value = 0.03032
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  0.02047728 0.38410027
## sample estimates:
##       cor 
## 0.2095211