Hice una prueba logica para baja adherencia (<4). O sea que False=Buena adherencia y True=Baja adherencia
| Characteristic | FALSE, N = 361 | TRUE, N = 271 | p-value2 |
|---|---|---|---|
| edad | 34 (30, 53) | 42 (29, 54) | >0.9 |
| sexo | 0.5 | ||
| Femenino (0) | 20 (67%) | 13 (76%) | |
| Masculino (1) | 10 (33%) | 4 (24%) | |
| em_tipo | 0.7 | ||
| EMPP (2) | 2 (6.7%) | 0 (0%) | |
| EMRR (1) | 27 (90%) | 17 (100%) | |
| EMSP (3) | 1 (3.3%) | 0 (0%) | |
| Duración de la enfermedad (años) | 4.0 (2.0, 6.0) | 4.0 (4.0, 7.8) | 0.4 |
| tto_actual | 0.7 | ||
| Alemtuzumab (9) | 1 (2.9%) | 0 (0%) | |
| Anti CD20 (8) | 6 (17%) | 1 (4.5%) | |
| Cetato de glatiramer (2) | 1 (2.9%) | 0 (0%) | |
| Cladribine (6) | 3 (8.6%) | 1 (4.5%) | |
| Dimetilfumarato (4) | 12 (34%) | 13 (59%) | |
| Fingolimod (5) | 3 (8.6%) | 3 (14%) | |
| Interferon (1) | 3 (8.6%) | 2 (9.1%) | |
| Natalizumab (7) | 4 (11%) | 1 (4.5%) | |
| Teriflunomida (3) | 2 (5.7%) | 1 (4.5%) | |
| hads_ans | 8.0 (5.2, 10.8) | 8.0 (4.5, 11.0) | >0.9 |
| hads_dep | 3.00 (1.25, 5.75) | 5.00 (2.00, 8.00) | 0.2 |
| moca_pz | 0.13 (-0.30, 0.52) | 0.19 (-0.32, 0.67) | 0.7 |
| 1 Median (IQR); n (%) | |||
| 2 Wilcoxon rank sum test; Pearson's Chi-squared test; Fisher's exact test | |||
La variable que mas correlaciona con la adherencia al tratamiento es el MOS global (0.76), el efecto de las funciones ejecutivas y la personalidad parece significativamente despreciable
| Characteristic | FALSE, N = 361 | TRUE, N = 271 | p-value2 | ES (90% CI)3 |
|---|---|---|---|---|
| neuroticismo | 22 (13, 28) | 23 (17, 32) | 0.4 | -0.19 (-0.62, 0.24) |
| extraversi_n | 27 (22, 34) | 26 (23, 31) | 0.5 | 0.15 (-0.28, 0.58) |
| apertura | 29 (24, 32) | 32 (28, 34) | 0.062 | -0.47 (-0.91, -0.04) |
| amabilidad | 31.5 (29.0, 35.0) | 28.0 (27.0, 36.0) | 0.8 | 0.06 (-0.37, 0.49) |
| responsabilidad | 33 (30, 36) | 28 (26, 34) | 0.048 | 0.53 (0.09, 1.0) |
| Executive_function_composite | -0.38 (-0.71, 0.26) | -0.07 (-0.72, 0.38) | 0.8 | 0.06 (-0.39, 0.51) |
| 1 Median (IQR) | ||||
| 2 Welch Two Sample t-test | ||||
| 3 Cohen's D (90% CI) | ||||
| Characteristic | FALSE, N = 361 | TRUE, N = 271 | p-value2 | ES (90% CI)3 |
|---|---|---|---|---|
| mos_indice_global | 83 (73, 95) | 77 (54, 88) | 0.016 | 0.71 (0.28, 1.1) |
| mos_apoyo_afectivo | 14.5 (13.0, 15.0) | 14.0 (8.0, 15.0) | 0.020 | 0.70 (0.26, 1.1) |
| mos_emocional | 35 (28, 40) | 32 (20, 37) | 0.025 | 0.64 (0.21, 1.1) |
| mos_instrumental | 18 (16, 20) | 15 (9, 17) | 0.003 | 0.89 (0.45, 1.3) |
| mos_social_positiva | 18 (14, 20) | 17 (11, 19) | 0.066 | 0.52 (0.09, 0.95) |
| 1 Median (IQR) | ||||
| 2 Welch Two Sample t-test | ||||
| 3 Cohen's D (90% CI) | ||||
Como se puede ver en ambas tablas, el efecto de las funciones ejecutivas y la personalidad es bastante bajo (y muchas veces no signiticativo). Como planteaba el analisis de las correlaciones el MOS global es la variable que mas diferencia los grupos con un tamaño de efecto moderado (D cohen: efecto pequeño (0.2-0.5), mediano (0.5-0.8) y grande (>0.8)), el resto de efectos son malisimos
Basados en las correlaciones y la interpretacion clinica cosntrui dos modelos
En el modelo 1 decidi que el MOS puede que sea el factor desencadenante de la adherencia pero este puede estar unfluenciado por las funciones cognitivas (interpretacion falopa, suponiendo que de algun modo una persona con dificultades ejecutivas tiene mas dificultades para establecer una red de apoyo o viceversa)
| lhs | op | rhs | label | est | se | z | pvalue | ci.lower | ci.upper | std.lv | std.all | std.nox |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| mos_indice_global | ~ | Executive_function_composite | a | 3.4275685 | 2.5397144 | 1.3495881 | 0.1771481 | -2.1241232 | 7.8234304 | 3.4275685 | 0.1766130 | 0.1766130 |
| adh_tto_ | ~ | mos_indice_global | b | 0.0086713 | 0.0074978 | 1.1565229 | 0.2474673 | -0.0056448 | 0.0234367 | 0.0086713 | 0.1780292 | 0.1780292 |
| Executive_function_composite | ~ | Executive_function_composite | cp | 0.2777970 | 0.2650156 | 1.0482289 | 0.2945332 | 0.1230697 | 1.4301675 | 0.2777970 | 0.2777970 | 0.2777970 |
| mos_indice_global | ~~ | mos_indice_global | 188.0316180 | 36.7335254 | 5.1188013 | 0.0000003 | 122.9024455 | 264.8868684 | 188.0316180 | 0.9688078 | 0.9688078 | |
| adh_tto_ | ~~ | adh_tto_ | 0.4458557 | 0.0767252 | 5.8110714 | 0.0000000 | 0.3110899 | 0.6112363 | 0.4458557 | 0.9683056 | 0.9683056 | |
| Executive_function_composite | ~~ | Executive_function_composite | 0.5153073 | 0.1129797 | 4.5610618 | 0.0000051 | 0.3345722 | 0.7880460 | 0.5153073 | 1.0000000 | 1.0000000 | |
| ab | := | a*b | ab | 0.0297216 | 0.0411875 | 0.7216163 | 0.4705304 | -0.0122538 | 0.1650981 | 0.0297216 | 0.0314423 | 0.0314423 |
| total | := | cp+ab | total | 0.3075186 | 0.2838602 | 1.0833452 | 0.2786552 | 0.1231160 | 1.5017905 | 0.3075186 | 0.3092393 | 0.3092393 |
##
## Mediation/Moderation Analysis
## Call: mediate(y = adh_tto_ ~ Executive_function_composite + mos_indice_global,
## data = base, n.iter = 10000)
##
## The DV (Y) was adh_tto_ . The IV (X) was Executive_function_composite mos_indice_global . The mediating variable(s) = .Call: mediate(y = adh_tto_ ~ Executive_function_composite + mos_indice_global,
## data = base, n.iter = 10000)
##
## No mediator specified leads to traditional regression
## adh_tto_ se t df Prob
## Intercept 0.74 0.30 2.44 60 1.77e-02
## Executive_function_composite 0.04 0.14 0.31 60 7.60e-01
## mos_indice_global 0.03 0.00 8.92 60 1.36e-12
##
## R = 0.76 R2 = 0.57 F = 40.46 on 2 and 60 DF p-value: 7.52e-12
En este modelo inclui los rasgos de personalidad porq me parecio que pueden interaccionar con el MOS tambien, vean que en realidad las interacciones (y pedorras) son mas frecuentes entre rasgos que otra cosa
##
## Mediation/Moderation Analysis
## Call: mediate(y = adh_tto_ ~ Executive_function_composite + mos_indice_global +
## neuroticismo + extraversi_n + amabilidad + responsabilidad,
## data = base, n.iter = 10000)
##
## The DV (Y) was adh_tto_ . The IV (X) was Executive_function_composite mos_indice_global neuroticismo extraversi_n amabilidad responsabilidad . The mediating variable(s) = .Call: mediate(y = adh_tto_ ~ Executive_function_composite + mos_indice_global +
## neuroticismo + extraversi_n + amabilidad + responsabilidad,
## data = base, n.iter = 10000)
##
## No mediator specified leads to traditional regression
## adh_tto_ se t df Prob
## Intercept 0.76 0.87 0.88 56 3.82e-01
## Executive_function_composite 0.01 0.15 0.07 56 9.47e-01
## mos_indice_global 0.03 0.00 8.48 56 1.24e-11
## neuroticismo 0.00 0.01 -0.28 56 7.82e-01
## extraversi_n -0.01 0.01 -0.74 56 4.64e-01
## amabilidad 0.00 0.02 0.09 56 9.27e-01
## responsabilidad 0.01 0.02 0.53 56 6.00e-01
##
## R = 0.76 R2 = 0.58 F = 12.91 on 6 and 56 DF p-value: 4.24e-09