KP_Analysis_10222021_murata
Stephen Murata
10/22/2021
Re-organizing dataframe variables
KP_df<-readr::read_csv("./batchcontrol_newdf_20210916.csv")##
## -- Column specification --------------------------------------------------------
## cols(
## .default = col_double(),
## group = col_character()
## )
## i Use `spec()` for the full column specifications.demo_group_vars<-c( "group",
"sex_male",
"race_white",
"age",
"bmi" )
med_vars<-c("fluoxetine",
"escitalopram",
"hydroxyzine",
"trazodone",
"aripiprazole",
"venlafaxine" ,
"lamotrigine" ,
"lithium",
"bupropion",
"suboxone")
SUD_vars = c("alcohol_30days",
"amphetamines.stims.uppers_30days",
"cocaine.crack_30days",
"otc.meds_30days",
"heroine.morphine.opiates_30days",
"presx.painkillers_30days",
"barbituates_30days",
"tranquilizers_30days",
"lsd.hallucinogens_30days",
"ecstasy_30days",
"marijuana_30days",
"smoke.tobacco_30days",
"chew.tobacco_30days")
KP_data<-c("ido1",
"ido2",
"tdo2",
"kmo",
"kynu",
"afmid",
"aadat",
"ccbl1",
"ccbl2",
"got2",
"haao",
"acmsd")
psychometric_vars<-c( "phq15_tot",
"gad_tot",
"phq_tot",
"bhs_tot",
"ctq_physical_tot",
"ctq_sexual_tot" ,
"ctq_tot",
"pcl_tot" ,
"bps_tot",
"als_tot" ,
"apathy_tot" ,
"rfl_tot" ,
"dusib_tot",
"asiq_tot" ,
"bis_tot" ,
"mspss_tot" ,
"srrs_tot",
"pss_tot" ,
"shaps_tot" ,
"asr_total",
"crs")
diagnosis_vars<-c( "scid_ehr_mood_disorder" ,
"scid_ehr_psychotic" ,
"scid_ehr_aud_sud" ,
"scid_ehr_anxiety" ,
"scid_ehr_ptsd")
df<-KP_df %>% dplyr::select(all_of(demo_group_vars),
all_of(diagnosis_vars),
all_of(med_vars),
all_of(SUD_vars),
all_of(psychometric_vars),
all_of(KP_data))
df[c("group", "sex_male","race_white")]<-lapply(df[c("group", "sex_male","race_white")], as.factor)
df[diagnosis_vars]<-lapply(df[diagnosis_vars], as.factor)
df[SUD_vars]<-lapply(df[SUD_vars], as.factor)
df$group<-relevel(df$group, ref="HC")
df$phq_severe<-ifelse(df$phq_tot>=15, 1, 0)Note: “group” variable is coded as unordered factor with HC as reference group (assumption - not ordinal)
Sample description by variable
##
## --------Summary descriptives table ---------
##
## ________________________________________________
## [ALL] N
## N=152
## ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
## group: 148
## HC 32 (21.6%)
## PC 23 (15.5%)
## SA 45 (30.4%)
## SI 48 (32.4%)
## sex_male 68 (51.5%) 132
## race_white 101 (76.5%) 132
## age 24.3 (3.79) 132
## bmi 25.6 (5.77) 114
## scid_ehr_mood_disorder 101 (67.3%) 150
## scid_ehr_psychotic 11 (7.33%) 150
## scid_ehr_aud_sud 62 (41.3%) 150
## scid_ehr_anxiety 49 (32.7%) 150
## scid_ehr_ptsd 32 (21.3%) 150
## fluoxetine 0.23 (0.42) 111
## escitalopram 0.18 (0.39) 111
## hydroxyzine 0.15 (0.36) 111
## trazodone 0.15 (0.36) 111
## aripiprazole 0.09 (0.29) 111
## venlafaxine 0.08 (0.27) 111
## lamotrigine 0.09 (0.29) 111
## lithium 0.09 (0.29) 111
## bupropion 0.05 (0.21) 111
## suboxone 0.05 (0.21) 111
## alcohol_30days 108 (76.6%) 141
## amphetamines.stims.uppers_30days 11 (7.86%) 140
## cocaine.crack_30days 20 (14.2%) 141
## otc.meds_30days 47 (33.6%) 140
## heroine.morphine.opiates_30days 18 (12.9%) 140
## presx.painkillers_30days 15 (10.7%) 140
## barbituates_30days 5 (3.57%) 140
## tranquilizers_30days 4 (2.86%) 140
## lsd.hallucinogens_30days 4 (2.84%) 141
## ecstasy_30days 8 (5.71%) 140
## marijuana_30days 72 (51.4%) 140
## smoke.tobacco_30days 58 (41.4%) 140
## chew.tobacco_30days 11 (7.86%) 140
## phq15_tot 7.77 (5.82) 148
## gad_tot 9.21 (6.67) 148
## phq_tot 12.2 (8.73) 148
## bhs_tot 9.86 (2.51) 148
## ctq_physical_tot 8.01 (4.08) 147
## ctq_sexual_tot 7.55 (5.44) 146
## ctq_tot 58.9 (10.3) 147
## pcl_tot 32.1 (23.3) 136
## bps_tot 0.48 (0.16) 143
## als_tot 1.29 (0.87) 142
## apathy_tot 38.8 (8.61) 129
## rfl_tot 173 (42.7) 139
## dusib_tot 33.2 (33.4) 141
## asiq_tot 57.1 (49.5) 148
## bis_tot 65.6 (14.8) 127
## mspss_tot 5.28 (1.42) 129
## srrs_tot 11.4 (7.30) 129
## pss_tot 22.1 (10.4) 128
## shaps_tot 3.21 (3.05) 128
## asr_total 81.1 (49.3) 139
## crs 2.68 (1.30) 123
## ido1 5.35 (1.29) 152
## ido2 3.93 (0.40) 152
## tdo2 5.21 (0.60) 152
## kmo 6.20 (0.83) 152
## kynu 5.91 (1.00) 152
## afmid 3.64 (0.61) 152
## aadat 3.37 (0.51) 152
## ccbl1 6.17 (0.64) 152
## ccbl2 8.41 (0.57) 152
## got2 8.13 (0.63) 152
## haao 4.72 (0.70) 152
## acmsd 5.92 (0.42) 152
## phq_severe 0.49 (0.50) 148
## ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯Visual inspection: normality of residuals (note: individual biomarkers modelled by demo + group)
vars_KP<-df %>% dplyr::select(all_of(KP_data)) %>% names()
for (x in vars_KP) {
LM1 <- lm(substitute(i ~ sex_male+age+bmi+race_white+group, list(i = as.name(x))), data = df)
graphics::plot(LM1, which=c(2), title(x))
}
### Note: in the previous iteration we inspected residuals after univariate modelling of biomarker by age (arbitrary). Here I fitted each variable model according to demographics and group, to try to minimize transformations that are premature or out of context to the final model.
Biomarker transformations to log, sqrt and box-cox versions
vars_KP<-df %>% dplyr::select(all_of(KP_data)) %>% names()
df[paste(vars_KP,sep="_", "log")] <- log(df[vars_KP])
df[paste(vars_KP,sep="_", "sqrt")] <- sqrt(df[vars_KP])
df[paste(vars_KP,sep="_", "inverse")] <- 1/(df[vars_KP])
#Box Cox transformation: IDO2
bc<-MASS::boxcox(ido2~sex_male+age+bmi+race_white+group, data=df)
(bc.power<-bc$x[which.max(bc$y)]) ## [1] -2BCTransform <- function(y, lambda=0) {
if (lambda == 0L) { log(y) } else { (y^lambda - 1) / lambda }}
df$ido2_bc <- BCTransform(df$ido2, bc.power)
#Box Cox transformation: TDO2
bc<-MASS::boxcox(tdo2~sex_male+age+bmi+race_white+group, data=df)
(bc.power<-bc$x[which.max(bc$y)]) ## [1] -0.2222222BCTransform <- function(y, lambda=0) {
if (lambda == 0L) { log(y) } else { (y^lambda - 1) / lambda }}
df$tdo2_bc <- BCTransform(df$tdo2, bc.power)
#Box Cox transformation: AADAT
bc<-MASS::boxcox(aadat~sex_male+age+bmi+race_white+group, data=df)
(bc.power<-bc$x[which.max(bc$y)]) ## [1] -2BCTransform <- function(y, lambda=0) {
if (lambda == 0L) { log(y) } else { (y^lambda - 1) / lambda }}
df$aadat_bc <- BCTransform(df$aadat, bc.power)
#Box Cox transformation: acmsd
bc<-MASS::boxcox(acmsd~sex_male+age+bmi+race_white+group, data=df)
(bc.power<-bc$x[which.max(bc$y)]) ## [1] -2BCTransform <- function(y, lambda=0) {
if (lambda == 0L) { log(y) } else { (y^lambda - 1) / lambda }}
df$acmsd_bc <- BCTransform(df$acmsd, bc.power)
#Box Cox transformation: kynu
bc<-MASS::boxcox(kynu~sex_male+age+bmi+race_white+group, data=df)
(bc.power<-bc$x[which.max(bc$y)]) ## [1] 1.555556BCTransform <- function(y, lambda=0) {
if (lambda == 0L) { log(y) } else { (y^lambda - 1) / lambda }}
df$kynu_bc <- BCTransform(df$kynu, bc.power)
#Box Cox transformation: afmid
bc<-MASS::boxcox(afmid~sex_male+age+bmi+race_white+group, data=df)
(bc.power<-bc$x[which.max(bc$y)]) ## [1] -1.191919BCTransform <- function(y, lambda=0) {
if (lambda == 0L) { log(y) } else { (y^lambda - 1) / lambda }}
df$afmid_bc <- BCTransform(df$afmid, bc.power)
#Visualize
vars_KP_transformed<-df %>% dplyr::select(all_of(vars_KP), contains("log"), contains("sqrt"), contains("inverse")) %>% names() %>% sort()
#final selection
bx_transformed<-c("aadat_bc", "acmsd_bc", "ido2_bc", "tdo2_bc", "haao_log")
bx_final_names<-df %>% dplyr::select(all_of(vars_KP), all_of(bx_transformed), -aadat, -acmsd, -ido2, -tdo2, -haao) %>% names()
for (x in bx_final_names) {
LM1 <- lm(substitute(i ~ sex_male+age+bmi+race_white+group, list(i = as.name(x))), data = df)
graphics::plot(LM1, which=c(2), title(x))
}
### Note: Unlike prior iteration, it looks like IDO1 (raw) looks slightly better than IDO1 (log).
Comparing residuals for IDO1 by model
### Note: When IDO1 residuals are modelled by age, then log(IDO1) transform looks better (prior iteration). When IDO1 residuals are modelled in hypothesis driven manner (group, adjusted by demographics), then untransformed IDO1 looks better. I’m going to continue with hypothesis-driven models for each biomarker for this iteration. source: https://towardsdatascience.com/how-to-detect-unusual-observations-on-your-regression-model-with-r-de0eaa38bc5b
Comparison of biomarkers models according to group (adjusted for demographics)
| Beta_coeff | T_test | Std_error | p_value | R_coeff |
|---|---|---|---|---|
| 0.33 | 2.64 | 0.13 | 0.01 | 0.09 |
| -0.20 | -1.63 | 0.12 | 0.11 | 0.09 |
| 0.25 | 1.61 | 0.16 | 0.11 | 0.15 |
| -0.30 | -1.17 | 0.26 | 0.24 | 0.11 |
| 0.00 | -0.87 | 0.00 | 0.39 | 0.02 |
| 0.00 | -0.76 | 0.00 | 0.45 | 0.06 |
| 0.16 | 0.72 | 0.22 | 0.47 | 0.03 |
| 0.02 | 0.56 | 0.03 | 0.57 | 0.09 |
| 0.00 | 0.49 | 0.00 | 0.62 | 0.04 |
| 0.01 | 0.44 | 0.02 | 0.66 | 0.02 |
| 0.04 | 0.35 | 0.12 | 0.72 | 0.04 |
| -0.04 | -0.30 | 0.13 | 0.77 | 0.07 |
Note: This table is meant to be a “univariate-ish” version of the univariate analysis of each biomarker, in which I controlled for the entire set of hypothesis-driven IV’s (group adjusted by demo).
Leverage plots (residuals by cooks & hat values)
### Finding: observation 52 may be influential across multiple variables. See potential outliers and high leverage points per variable. May do sensitivity analysis.
Univariate analysis: sample characteristics according to ido1 (alpha=0.05)
| Characteristic | N | Beta | 95% CI1 | p-value |
|---|---|---|---|---|
| group | 148 | |||
| HC | — | — | ||
| PC | 0.147 | -0.546, 0.839 | 0.7 | |
| SA | 0.316 | -0.270, 0.902 | 0.3 | |
| SI | 0.769 | 0.190, 1.35 | 0.010 | |
| sex_male | 132 | -0.118 | -0.563, 0.326 | 0.6 |
| race_white | 132 | -0.388 | -0.908, 0.132 | 0.14 |
| age | 132 | 0.023 | -0.036, 0.082 | 0.4 |
| bmi | 114 | -0.004 | -0.046, 0.038 | 0.9 |
| scid_ehr_mood_disorder | 150 | 0.148 | -0.298, 0.594 | 0.5 |
| scid_ehr_psychotic | 150 | 0.796 | 0.002, 1.59 | 0.049 |
| scid_ehr_aud_sud | 150 | 0.234 | -0.190, 0.657 | 0.3 |
| scid_ehr_anxiety | 150 | 0.136 | -0.310, 0.582 | 0.5 |
| scid_ehr_ptsd | 150 | 0.038 | -0.474, 0.549 | 0.9 |
| phq15_tot | 148 | 0.004 | -0.032, 0.040 | 0.8 |
| gad_tot | 148 | 0.020 | -0.011, 0.051 | 0.2 |
| phq_tot | 148 | 0.020 | -0.003, 0.044 | 0.093 |
| bhs_tot | 148 | 0.077 | -0.006, 0.160 | 0.068 |
| ctq_physical_tot | 147 | 0.007 | -0.045, 0.058 | 0.8 |
| ctq_sexual_tot | 146 | -0.007 | -0.046, 0.031 | 0.7 |
| ctq_tot | 147 | 0.004 | -0.016, 0.025 | 0.7 |
| pcl_tot | 136 | 0.002 | -0.008, 0.011 | 0.7 |
| bps_tot | 143 | 0.825 | -0.477, 2.13 | 0.2 |
| als_tot | 142 | 0.182 | -0.064, 0.428 | 0.15 |
| apathy_tot | 129 | 0.017 | -0.009, 0.044 | 0.2 |
| rfl_tot | 139 | 0.001 | -0.005, 0.006 | 0.8 |
| dusib_tot | 141 | 0.002 | -0.004, 0.009 | 0.5 |
| asiq_tot | 148 | 0.005 | 0.001, 0.009 | 0.021 |
| bis_tot | 127 | 0.013 | -0.003, 0.028 | 0.11 |
| mspss_tot | 129 | -0.099 | -0.258, 0.061 | 0.2 |
| srrs_tot | 129 | 0.042 | 0.012, 0.072 | 0.007 |
| pss_tot | 128 | 0.031 | 0.010, 0.053 | 0.005 |
| shaps_tot | 128 | 0.130 | 0.058, 0.201 | <0.001 |
| asr_total | 139 | 0.003 | -0.002, 0.007 | 0.2 |
| crs | 123 | 0.180 | 0.001, 0.359 | 0.049 |
| fluoxetine | 111 | -0.234 | -0.840, 0.373 | 0.4 |
| escitalopram | 111 | 0.023 | -0.638, 0.683 | >0.9 |
| hydroxyzine | 111 | -0.181 | -0.885, 0.523 | 0.6 |
| trazodone | 111 | -0.622 | -1.32, 0.073 | 0.079 |
| aripiprazole | 111 | -0.003 | -0.890, 0.884 | >0.9 |
| venlafaxine | 111 | 0.527 | -0.399, 1.45 | 0.3 |
| lamotrigine | 111 | 0.165 | -0.722, 1.05 | 0.7 |
| lithium | 111 | 0.171 | -0.715, 1.06 | 0.7 |
| bupropion | 111 | -0.020 | -1.24, 1.20 | >0.9 |
| suboxone | 111 | 0.902 | -0.311, 2.11 | 0.14 |
| alcohol_30days | 141 | -0.252 | -0.763, 0.259 | 0.3 |
| amphetamines.stims.uppers_30days | 140 | -0.495 | -1.30, 0.312 | 0.2 |
| cocaine.crack_30days | 141 | 0.238 | -0.383, 0.859 | 0.4 |
| otc.meds_30days | 140 | 0.348 | -0.110, 0.807 | 0.14 |
| heroine.morphine.opiates_30days | 140 | 0.255 | -0.396, 0.906 | 0.4 |
| presx.painkillers_30days | 140 | 0.673 | -0.024, 1.37 | 0.058 |
| barbituates_30days | 140 | 0.582 | -0.591, 1.75 | 0.3 |
| tranquilizers_30days | 140 | -0.126 | -1.44, 1.19 | 0.8 |
| lsd.hallucinogens_30days | 141 | 0.747 | -0.555, 2.05 | 0.3 |
| ecstasy_30days | 140 | 0.043 | -0.898, 0.98 | >0.9 |
| marijuana_30days | 140 | 0.110 | -0.326, 0.547 | 0.6 |
| smoke.tobacco_30days | 140 | 0.059 | -0.384, 0.502 | 0.8 |
| chew.tobacco_30days | 140 | 0.120 | -0.692, 0.931 | 0.8 |
| kmo | 152 | 0.458 | 0.218, 0.698 | <0.001 |
| kynu | 152 | 0.226 | 0.019, 0.432 | 0.032 |
| afmid | 152 | -0.331 | -0.668, 0.006 | 0.054 |
| ccbl1 | 152 | -0.026 | -0.354, 0.302 | 0.9 |
| ccbl2 | 152 | 0.838 | 0.500, 1.18 | <0.001 |
| got2 | 152 | -0.308 | -0.635, 0.020 | 0.065 |
| aadat_bc | 152 | -1.76 | -26.3, 22.7 | 0.9 |
| acmsd_bc | 152 | -91.9 | -181, -2.31 | 0.044 |
| ido2_bc | 152 | -59.3 | -92.1, -26.6 | <0.001 |
| tdo2_bc | 152 | 2.41 | -0.200, 5.01 | 0.070 |
| haao_log | 152 | 1.11 | -0.278, 2.50 | 0.12 |
|
1
CI = Confidence Interval
|
||||
Results: IDO1 significantly distinguishes SI from HC (p=0.01, ref=HC). This significance level is less than Emily’s version. Few covariates overall: scid_ehr_psychotic, asiq_tot, srrs_tot, pss_tot, shaps_tot, kmo, kynu, ccbl2, acmsd_bc, ido2_bc.
Univariate analysis: sample characterics according to group
| Variable | N | Overall, N = 1481 | Clinical Subgroup | p-value2 | |||
|---|---|---|---|---|---|---|---|
| HC, N = 321 | PC, N = 231 | SA, N = 451 | SI, N = 481 | ||||
| sex_male | 131 | 0.22 | |||||
| 0 | 63 (48%) | 17 (55%) | 5 (26%) | 21 (51%) | 20 (50%) | ||
| 1 | 68 (52%) | 14 (45%) | 14 (74%) | 20 (49%) | 20 (50%) | ||
| Unknown | 17 | 1 | 4 | 4 | 8 | ||
| race_white | 131 | 0.24 | |||||
| 0 | 31 (24%) | 5 (16%) | 5 (26%) | 14 (34%) | 7 (18%) | ||
| 1 | 100 (76%) | 26 (84%) | 14 (74%) | 27 (66%) | 33 (82%) | ||
| Unknown | 17 | 1 | 4 | 4 | 8 | ||
| age | 131 | 24.3 (3.8) | 24.9 (3.5) | 23.1 (3.2) | 24.4 (3.8) | 24.3 (4.3) | 0.40 |
| Unknown | 17 | 1 | 4 | 4 | 8 | ||
| bmi | 112 | 25.6 (5.8) | 26.3 (7.1) | 23.9 (2.9) | 25.9 (5.5) | 25.6 (5.9) | 0.59 |
| Unknown | 36 | 1 | 5 | 16 | 14 | ||
| scid_ehr_mood_disorder | 146 | <0.001 | |||||
| 0 | 47 (32%) | 30 (100%) | 5 (22%) | 7 (16%) | 5 (10%) | ||
| 1 | 99 (68%) | 0 (0%) | 18 (78%) | 38 (84%) | 43 (90%) | ||
| Unknown | 2 | 2 | 0 | 0 | 0 | ||
| scid_ehr_psychotic | 146 | 0.044 | |||||
| 0 | 136 (93%) | 30 (100%) | 20 (87%) | 44 (98%) | 42 (88%) | ||
| 1 | 10 (6.8%) | 0 (0%) | 3 (13%) | 1 (2.2%) | 6 (12%) | ||
| Unknown | 2 | 2 | 0 | 0 | 0 | ||
| scid_ehr_aud_sud | 146 | <0.001 | |||||
| 0 | 86 (59%) | 30 (100%) | 7 (30%) | 19 (42%) | 30 (62%) | ||
| 1 | 60 (41%) | 0 (0%) | 16 (70%) | 26 (58%) | 18 (38%) | ||
| Unknown | 2 | 2 | 0 | 0 | 0 | ||
| scid_ehr_anxiety | 146 | <0.001 | |||||
| 0 | 99 (68%) | 30 (100%) | 12 (52%) | 30 (67%) | 27 (56%) | ||
| 1 | 47 (32%) | 0 (0%) | 11 (48%) | 15 (33%) | 21 (44%) | ||
| Unknown | 2 | 2 | 0 | 0 | 0 | ||
| scid_ehr_ptsd | 146 | 0.003 | |||||
| 0 | 116 (79%) | 30 (100%) | 17 (74%) | 32 (71%) | 37 (77%) | ||
| 1 | 30 (21%) | 0 (0%) | 6 (26%) | 13 (29%) | 11 (23%) | ||
| Unknown | 2 | 2 | 0 | 0 | 0 | ||
| phq15_tot | 144 | 7.5 (5.6) | 1.6 (1.7) | 6.4 (4.3) | 9.0 (4.7) | 10.7 (5.5) | <0.001 |
| Unknown | 4 | 0 | 0 | 2 | 2 | ||
| gad_tot | 144 | 9.1 (6.7) | 0.5 (0.7) | 9.2 (5.3) | 11.4 (5.5) | 12.9 (5.1) | <0.001 |
| Unknown | 4 | 0 | 0 | 2 | 2 | ||
| phq_tot | 144 | 12 (9) | 0 (1) | 10 (7) | 16 (7) | 18 (6) | <0.001 |
| Unknown | 4 | 0 | 0 | 2 | 2 | ||
| bhs_tot | 144 | 9.83 (2.52) | 8.25 (1.93) | 10.35 (2.98) | 10.40 (2.59) | 10.13 (2.18) | <0.001 |
| Unknown | 4 | 0 | 0 | 2 | 2 | ||
| ctq_physical_tot | 143 | 8.0 (4.1) | 5.3 (0.7) | 8.3 (4.0) | 9.8 (5.0) | 8.0 (3.6) | <0.001 |
| Unknown | 5 | 0 | 0 | 2 | 3 | ||
| ctq_sexual_tot | 142 | 7.2 (5.0) | 5.0 (0.0) | 5.7 (2.5) | 9.6 (6.7) | 7.5 (5.1) | <0.001 |
| Unknown | 6 | 0 | 0 | 3 | 3 | ||
| ctq_tot | 143 | 58 (10) | 52 (3) | 58 (7) | 63 (12) | 59 (10) | <0.001 |
| Unknown | 5 | 0 | 0 | 2 | 3 | ||
| pcl_tot | 132 | 31 (23) | 1 (2) | 28 (19) | 44 (16) | 41 (18) | <0.001 |
| Unknown | 16 | 3 | 2 | 7 | 4 | ||
| bps_tot | 139 | 0.48 (0.16) | 0.31 (0.06) | 0.47 (0.14) | 0.54 (0.16) | 0.54 (0.14) | <0.001 |
| Unknown | 9 | 0 | 1 | 4 | 4 | ||
| als_tot | 138 | 1.27 (0.87) | 0.18 (0.20) | 1.26 (0.72) | 1.67 (0.70) | 1.71 (0.66) | <0.001 |
| Unknown | 10 | 0 | 1 | 4 | 5 | ||
| apathy_tot | 126 | 39 (9) | 31 (3) | 37 (7) | 42 (9) | 43 (8) | <0.001 |
| Unknown | 22 | 1 | 2 | 10 | 9 | ||
| rfl_tot | 136 | 173 (43) | 206 (23) | 202 (28) | 152 (47) | 155 (33) | <0.001 |
| Unknown | 12 | 0 | 2 | 4 | 6 | ||
| dusib_tot | 139 | 33 (34) | 6 (10) | 38 (30) | 46 (36) | 38 (34) | <0.001 |
| Unknown | 9 | 0 | 2 | 3 | 4 | ||
| asiq_tot | 144 | 57 (49) | 1 (2) | 21 (22) | 81 (42) | 91 (35) | <0.001 |
| Unknown | 4 | 0 | 0 | 2 | 2 | ||
| bis_tot | 124 | 65 (15) | 50 (8) | 64 (8) | 71 (13) | 72 (15) | <0.001 |
| Unknown | 24 | 1 | 4 | 10 | 9 | ||
| mspss_tot | 126 | 5.31 (1.37) | 6.58 (0.43) | 5.12 (1.37) | 4.83 (1.48) | 4.84 (1.15) | <0.001 |
| Unknown | 22 | 1 | 2 | 10 | 9 | ||
| srrs_tot | 126 | 11 (7) | 4 (3) | 12 (6) | 14 (6) | 14 (7) | <0.001 |
| Unknown | 22 | 1 | 2 | 10 | 9 | ||
| pss_tot | 125 | 22 (10) | 8 (5) | 22 (8) | 27 (7) | 29 (5) | <0.001 |
| Unknown | 23 | 1 | 3 | 10 | 9 | ||
| shaps_tot | 125 | 3.18 (3.05) | 0.90 (1.62) | 2.05 (1.82) | 4.43 (3.45) | 4.46 (2.84) | <0.001 |
| Unknown | 23 | 1 | 3 | 10 | 9 | ||
| asr_total | 136 | 80 (49) | 15 (10) | 72 (32) | 103 (37) | 109 (34) | <0.001 |
| Unknown | 12 | 0 | 3 | 4 | 5 | ||
| crs | 123 | ||||||
| 1 | 34 (28%) | 30 (100%) | 1 (5.3%) | 1 (2.7%) | 2 (5.4%) | ||
| 2 | 16 (13%) | 0 (0%) | 11 (58%) | 4 (11%) | 1 (2.7%) | ||
| 3 | 38 (31%) | 0 (0%) | 6 (32%) | 12 (32%) | 20 (54%) | ||
| 4 | 25 (20%) | 0 (0%) | 1 (5.3%) | 12 (32%) | 12 (32%) | ||
| 5 | 10 (8.1%) | 0 (0%) | 0 (0%) | 8 (22%) | 2 (5.4%) | ||
| Unknown | 25 | 2 | 4 | 8 | 11 | ||
| fluoxetine | 110 | 22% (24) | 0% (0) | 12% (2) | 30% (11) | 31% (11) | 0.010 |
| Unknown | 38 | 11 | 7 | 8 | 12 | ||
| escitalopram | 110 | 18% (20) | 0% (0) | 12% (2) | 22% (8) | 28% (10) | 0.032 |
| Unknown | 38 | 11 | 7 | 8 | 12 | ||
| hydroxyzine | 110 | 15% (17) | 0% (0) | 12% (2) | 22% (8) | 19% (7) | 0.091 |
| Unknown | 38 | 11 | 7 | 8 | 12 | ||
| trazodone | 110 | 15% (17) | 0% (0) | 6.2% (1) | 30% (11) | 14% (5) | 0.012 |
| Unknown | 38 | 11 | 7 | 8 | 12 | ||
| aripiprazole | 110 | 8.2% (9) | 0% (0) | 12% (2) | 2.7% (1) | 17% (6) | 0.061 |
| Unknown | 38 | 11 | 7 | 8 | 12 | ||
| venlafaxine | 110 | 7.3% (8) | 0% (0) | 0% (0) | 11% (4) | 11% (4) | 0.27 |
| Unknown | 38 | 11 | 7 | 8 | 12 | ||
| lamotrigine | 110 | 9.1% (10) | 0% (0) | 0% (0) | 19% (7) | 8.3% (3) | 0.057 |
| Unknown | 38 | 11 | 7 | 8 | 12 | ||
| lithium | 110 | 9.1% (10) | 0% (0) | 25% (4) | 2.7% (1) | 14% (5) | 0.020 |
| Unknown | 38 | 11 | 7 | 8 | 12 | ||
| bupropion | 110 | 4.5% (5) | 0% (0) | 0% (0) | 5.4% (2) | 8.3% (3) | 0.53 |
| Unknown | 38 | 11 | 7 | 8 | 12 | ||
| suboxone | 110 | 4.5% (5) | 0% (0) | 6.2% (1) | 8.1% (3) | 2.8% (1) | 0.49 |
| Unknown | 38 | 11 | 7 | 8 | 12 | ||
| alcohol_30days | 139 | 0.24 | |||||
| 0 | 33 (24%) | 6 (19%) | 3 (14%) | 9 (21%) | 15 (34%) | ||
| 1 | 106 (76%) | 26 (81%) | 18 (86%) | 33 (79%) | 29 (66%) | ||
| Unknown | 9 | 0 | 2 | 3 | 4 | ||
| amphetamines.stims.uppers_30days | 138 | 0.12 | |||||
| 0 | 127 (92%) | 32 (100%) | 18 (86%) | 37 (88%) | 40 (93%) | ||
| 1 | 11 (8.0%) | 0 (0%) | 3 (14%) | 5 (12%) | 3 (7.0%) | ||
| Unknown | 10 | 0 | 2 | 3 | 5 | ||
| cocaine.crack_30days | 139 | 0.019 | |||||
| 0 | 119 (86%) | 32 (100%) | 16 (76%) | 35 (83%) | 36 (82%) | ||
| 1 | 20 (14%) | 0 (0%) | 5 (24%) | 7 (17%) | 8 (18%) | ||
| Unknown | 9 | 0 | 2 | 3 | 4 | ||
| otc.meds_30days | 138 | 0.19 | |||||
| 0 | 92 (67%) | 25 (78%) | 16 (76%) | 24 (57%) | 27 (63%) | ||
| 1 | 46 (33%) | 7 (22%) | 5 (24%) | 18 (43%) | 16 (37%) | ||
| Unknown | 10 | 0 | 2 | 3 | 5 | ||
| heroine.morphine.opiates_30days | 138 | 0.013 | |||||
| 0 | 120 (87%) | 32 (100%) | 18 (86%) | 32 (76%) | 38 (88%) | ||
| 1 | 18 (13%) | 0 (0%) | 3 (14%) | 10 (24%) | 5 (12%) | ||
| Unknown | 10 | 0 | 2 | 3 | 5 | ||
| presx.painkillers_30days | 138 | 0.017 | |||||
| 0 | 123 (89%) | 32 (100%) | 17 (81%) | 34 (81%) | 40 (93%) | ||
| 1 | 15 (11%) | 0 (0%) | 4 (19%) | 8 (19%) | 3 (7.0%) | ||
| Unknown | 10 | 0 | 2 | 3 | 5 | ||
| barbituates_30days | 138 | 0.62 | |||||
| 0 | 133 (96%) | 32 (100%) | 20 (95%) | 40 (95%) | 41 (95%) | ||
| 1 | 5 (3.6%) | 0 (0%) | 1 (4.8%) | 2 (4.8%) | 2 (4.7%) | ||
| Unknown | 10 | 0 | 2 | 3 | 5 | ||
| tranquilizers_30days | 138 | 0.63 | |||||
| 0 | 134 (97%) | 32 (100%) | 21 (100%) | 40 (95%) | 41 (95%) | ||
| 1 | 4 (2.9%) | 0 (0%) | 0 (0%) | 2 (4.8%) | 2 (4.7%) | ||
| Unknown | 10 | 0 | 2 | 3 | 5 | ||
| lsd.hallucinogens_30days | 139 | 0.63 | |||||
| 0 | 135 (97%) | 32 (100%) | 21 (100%) | 40 (95%) | 42 (95%) | ||
| 1 | 4 (2.9%) | 0 (0%) | 0 (0%) | 2 (4.8%) | 2 (4.5%) | ||
| Unknown | 9 | 0 | 2 | 3 | 4 | ||
| ecstasy_30days | 138 | 0.18 | |||||
| 0 | 130 (94%) | 32 (100%) | 20 (95%) | 37 (88%) | 41 (95%) | ||
| 1 | 8 (5.8%) | 0 (0%) | 1 (4.8%) | 5 (12%) | 2 (4.7%) | ||
| Unknown | 10 | 0 | 2 | 3 | 5 | ||
| marijuana_30days | 138 | <0.001 | |||||
| 0 | 68 (49%) | 25 (78%) | 4 (19%) | 14 (33%) | 25 (58%) | ||
| 1 | 70 (51%) | 7 (22%) | 17 (81%) | 28 (67%) | 18 (42%) | ||
| Unknown | 10 | 0 | 2 | 3 | 5 | ||
| smoke.tobacco_30days | 138 | 0.001 | |||||
| 0 | 82 (59%) | 28 (88%) | 8 (38%) | 22 (52%) | 24 (56%) | ||
| 1 | 56 (41%) | 4 (12%) | 13 (62%) | 20 (48%) | 19 (44%) | ||
| Unknown | 10 | 0 | 2 | 3 | 5 | ||
| chew.tobacco_30days | 138 | 0.50 | |||||
| 0 | 127 (92%) | 31 (97%) | 18 (86%) | 39 (93%) | 39 (91%) | ||
| 1 | 11 (8.0%) | 1 (3.1%) | 3 (14%) | 3 (7.1%) | 4 (9.3%) | ||
| Unknown | 10 | 0 | 2 | 3 | 5 | ||
| ido1 | 148 | 5.37 (1.30) | 5.00 (1.11) | 5.14 (0.93) | 5.31 (1.31) | 5.77 (1.49) | 0.048 |
| kmo | 148 | 6.21 (0.84) | 6.39 (0.74) | 5.86 (0.70) | 6.11 (1.01) | 6.36 (0.73) | 0.052 |
| kynu | 148 | 5.89 (1.00) | 5.85 (1.08) | 5.97 (1.08) | 5.76 (0.94) | 6.02 (0.98) | 0.63 |
| afmid | 148 | 3.63 (0.60) | 3.74 (0.64) | 3.58 (0.54) | 3.66 (0.65) | 3.54 (0.56) | 0.49 |
| ccbl1 | 148 | 6.17 (0.64) | 6.15 (0.63) | 6.35 (0.59) | 6.04 (0.63) | 6.22 (0.66) | 0.26 |
| ccbl2 | 148 | 8.41 (0.58) | 8.33 (0.44) | 8.57 (0.56) | 8.35 (0.70) | 8.43 (0.54) | 0.41 |
| got2 | 148 | 8.13 (0.64) | 8.17 (0.47) | 8.09 (0.64) | 8.07 (0.75) | 8.17 (0.63) | 0.84 |
| aadat_bc | 148 | 0.454 (0.009) | 0.454 (0.007) | 0.455 (0.007) | 0.454 (0.009) | 0.455 (0.010) | 0.88 |
| acmsd_bc | 148 | 0.4855 (0.0023) | 0.4854 (0.0022) | 0.4852 (0.0016) | 0.4853 (0.0032) | 0.4858 (0.0017) | 0.70 |
| ido2_bc | 148 | 0.467 (0.006) | 0.466 (0.006) | 0.467 (0.007) | 0.467 (0.006) | 0.466 (0.006) | 0.98 |
| tdo2_bc | 148 | 1.38 (0.08) | 1.38 (0.08) | 1.38 (0.08) | 1.38 (0.08) | 1.37 (0.08) | 0.93 |
| haao_log | 148 | 1.54 (0.15) | 1.55 (0.15) | 1.54 (0.13) | 1.53 (0.15) | 1.55 (0.15) | 0.85 |
|
1
Mean (SD) or n (%)
2
Pearson's Chi-squared test; Fisher's exact test; One-way ANOVA
|
|||||||
Note: see supplementary table (excel file) for pairwise comparisons
Correlational analysis of covariates
#Correlation matrix of significant covariates from univariate analyses
covariate_df<-df %>% dplyr::select(ido1, asiq_tot, srrs_tot, pss_tot, shaps_tot, kmo, kynu, ccbl2, acmsd_bc, ido2_bc) %>% na.omit() %>% stats::cor()
p.mat<-ggcorrplot::cor_pmat(covariate_df)
ggcorrplot::ggcorrplot(covariate_df,
hc.order = TRUE,
type = "lower",
outline.color = "black",
lab = TRUE,
lab_size = 3,
p.mat = p.mat, sig.level=0.05, insig="blank",
digits=1,
tl.srt=50,
tl.cex=10,
title="Pearson correlation matrix (alpha=0.05) - significant covariates")
### CCBL2 and KMO are not correlated here
Linear model #1: IDO1 by group only
## Anova Table (Type II tests)
##
## Response: ido1
## Sum Sq Df F value Pr(>F)
## group 13.294 3 2.6956 0.04823 *
## Residuals 236.731 144
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1## contrast estimate SE df t.ratio p.value
## HC - PC -0.147 0.351 144 -0.418 0.9753
## HC - SA -0.316 0.296 144 -1.067 0.7100
## HC - SI -0.769 0.293 144 -2.628 0.0465
## PC - SA -0.170 0.329 144 -0.517 0.9549
## PC - SI -0.622 0.325 144 -1.914 0.2268
## SA - SI -0.452 0.266 144 -1.701 0.3271
##
## P value adjustment: tukey method for comparing a family of 4 estimates## contrast effect.size SE df lower.CL upper.CL
## HC - PC -0.114 0.273 144 -0.655 0.4262
## HC - SA -0.247 0.232 144 -0.705 0.2112
## HC - SI -0.600 0.231 144 -1.056 -0.1432
## PC - SA -0.132 0.256 144 -0.639 0.3744
## PC - SI -0.485 0.255 144 -0.990 0.0191
## SA - SI -0.353 0.209 144 -0.765 0.0593
##
## sigma used for effect sizes: 1.282
## Confidence level used: 0.95##
## Call:
## lm(formula = ido1 ~ group, data = df)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.4467 -1.0243 -0.1311 0.9608 2.9333
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 4.9978 0.2267 22.050 < 2e-16 ***
## groupPC 0.1465 0.3505 0.418 0.67652
## groupSA 0.3164 0.2965 1.067 0.28767
## groupSI 0.7689 0.2926 2.628 0.00953 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.282 on 144 degrees of freedom
## (4 observations deleted due to missingness)
## Multiple R-squared: 0.05317, Adjusted R-squared: 0.03345
## F-statistic: 2.696 on 3 and 144 DF, p-value: 0.04823Findings: IDO1 distinguished group on ANOVA type II (p=0.048). IDO1 was associated with group level SI relative to HC (p=0.009, ES=0.60).
Linear model #2: IDO1 by demographics + group
## Anova Table (Type II tests)
##
## Response: ido1
## Sum Sq Df F value Pr(>F)
## sex_male 2.173 1 1.3722 0.24431
## age 0.377 1 0.2382 0.62663
## race_white 5.780 1 3.6502 0.05902 .
## bmi 0.088 1 0.0558 0.81372
## group 13.376 3 2.8159 0.04318 *
## Residuals 153.587 97
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1##
## Call:
## lm(formula = ido1 ~ sex_male + age + race_white + bmi + group,
## data = df)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.36633 -0.90789 -0.08619 0.87714 2.93644
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 5.352757 0.997693 5.365 5.5e-07 ***
## sex_male1 -0.299849 0.255976 -1.171 0.2443
## age 0.016343 0.033488 0.488 0.6266
## race_white1 -0.609398 0.318965 -1.911 0.0590 .
## bmi -0.005081 0.021504 -0.236 0.8137
## groupPC 0.285290 0.399975 0.713 0.4774
## groupSA 0.011214 0.338416 0.033 0.9736
## groupSI 0.840957 0.325303 2.585 0.0112 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.258 on 97 degrees of freedom
## (47 observations deleted due to missingness)
## Multiple R-squared: 0.1088, Adjusted R-squared: 0.0445
## F-statistic: 1.692 on 7 and 97 DF, p-value: 0.1199Findings: IDO1 retains significance to group (distinguishes SI vs. HC)
Linear model #3: IDO1 by group and demo, adding covariates
## Anova Table (Type II tests)
##
## Response: ido1
## Sum Sq Df F value Pr(>F)
## sex_male 1.691 1 1.2860 0.259635
## age 0.699 1 0.5314 0.467818
## race_white 3.156 1 2.4007 0.124607
## bmi 1.439 1 1.0946 0.298116
## group 9.756 3 2.4735 0.066314 .
## kmo 8.093 1 6.1560 0.014855 *
## ccbl2 12.403 1 9.4337 0.002779 **
## Residuals 124.899 95
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1##
## Call:
## lm(formula = ido1 ~ sex_male + +age + race_white + bmi + group +
## kmo + ccbl2, data = df)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.42775 -0.84368 0.01071 0.78327 2.47520
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -2.26562 1.90352 -1.190 0.23692
## sex_male1 -0.27441 0.24198 -1.134 0.25963
## age 0.02278 0.03126 0.729 0.46782
## race_white1 -0.45484 0.29356 -1.549 0.12461
## bmi -0.02088 0.01996 -1.046 0.29812
## groupPC 0.28233 0.39871 0.708 0.48062
## groupSA 0.11182 0.31274 0.358 0.72147
## groupSI 0.76085 0.29841 2.550 0.01238 *
## kmo 0.39069 0.15747 2.481 0.01486 *
## ccbl2 0.63105 0.20546 3.071 0.00278 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.147 on 95 degrees of freedom
## (47 observations deleted due to missingness)
## Multiple R-squared: 0.2753, Adjusted R-squared: 0.2066
## F-statistic: 4.009 on 9 and 95 DF, p-value: 0.0002268
## GVIF Df GVIF^(1/(2*Df))
## sex_male 1.169000 1 1.081203
## age 1.144853 1 1.069978
## race_white 1.101179 1 1.049371
## bmi 1.088614 1 1.043367
## group 1.455192 3 1.064519
## kmo 1.281507 1 1.132037
## ccbl2 1.204981 1 1.097716Note: manual backwards regression
Stepwise model (forward and backward, automated)
# Stepwise regression model as second opinion for variable selection (keep kynu or not)
step_df<-df %>% dplyr::select(c("ido1", "sex_male", "age", "bmi", "race_white", "group", "kmo", "ido2_bc", "ccbl2", "kynu")) %>% na.omit()
res.lm <- lm(ido1 ~ ., data = step_df)
step <- MASS::stepAIC(res.lm, direction = "both", trace = FALSE)
car::Anova(step, type = "II")## Anova Table (Type II tests)
##
## Response: ido1
## Sum Sq Df F value Pr(>F)
## race_white 3.752 1 2.9555 0.08878 .
## group 9.782 3 2.5683 0.05880 .
## kmo 5.182 1 4.0821 0.04610 *
## ido2_bc 4.899 1 3.8588 0.05235 .
## ccbl2 9.846 1 7.7553 0.00644 **
## Residuals 123.144 97
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1summary(step)##
## Call:
## lm(formula = ido1 ~ race_white + group + kmo + ido2_bc + ccbl2,
## data = step_df)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.87188 -0.85287 -0.06194 0.83522 2.71497
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 15.0920 9.0402 1.669 0.09826 .
## race_white1 -0.4980 0.2897 -1.719 0.08878 .
## groupPC 0.1767 0.3650 0.484 0.62934
## groupSA 0.1167 0.3069 0.380 0.70454
## groupSI 0.7497 0.2906 2.580 0.01139 *
## kmo 0.2994 0.1482 2.020 0.04610 *
## ido2_bc -34.9755 17.8049 -1.964 0.05235 .
## ccbl2 0.5670 0.2036 2.785 0.00644 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.127 on 97 degrees of freedom
## Multiple R-squared: 0.2855, Adjusted R-squared: 0.2339
## F-statistic: 5.536 on 7 and 97 DF, p-value: 2.217e-05Note: looks like KMO and CCBL2 at least, should be included in model
Linear model #4: IDO1 according to demo, group, and kmo:group interaction
## Anova Table (Type II tests)
##
## Response: ido1
## Sum Sq Df F value Pr(>F)
## sex_male 4.020 1 2.9168 0.0910003 .
## age 1.483 1 1.0757 0.3023424
## race_white 3.142 1 2.2800 0.1344416
## bmi 0.709 1 0.5145 0.4749909
## group 13.194 3 3.1910 0.0272358 *
## kmo 16.285 1 11.8165 0.0008798 ***
## group:kmo 9.129 3 2.2080 0.0923473 .
## Residuals 128.173 93
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1##
## Call:
## lm(formula = ido1 ~ sex_male + +age + race_white + bmi + group *
## kmo, data = df)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.7421 -0.7958 -0.0463 0.6967 2.6218
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 4.40966 2.15410 2.047 0.0435 *
## sex_male1 -0.41964 0.24571 -1.708 0.0910 .
## age 0.03337 0.03218 1.037 0.3023
## race_white1 -0.46288 0.30655 -1.510 0.1344
## bmi -0.01484 0.02068 -0.717 0.4750
## groupPC 1.10987 3.23521 0.343 0.7323
## groupSA -5.11637 2.44399 -2.093 0.0390 *
## groupSI -1.99113 2.64449 -0.753 0.4534
## kmo 0.11067 0.30332 0.365 0.7160
## groupPC:kmo -0.12260 0.54409 -0.225 0.8222
## groupSA:kmo 0.85983 0.39060 2.201 0.0302 *
## groupSI:kmo 0.44935 0.41587 1.080 0.2827
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.174 on 93 degrees of freedom
## (47 observations deleted due to missingness)
## Multiple R-squared: 0.2563, Adjusted R-squared: 0.1683
## F-statistic: 2.913 on 11 and 93 DF, p-value: 0.00242## (Intercept) sex_male1 age race_white1 bmi groupPC groupSA groupSI kmo groupPC:kmo groupSA:kmo groupSI:kmo
## 0.200 -0.167 0.101 -0.147 -0.070 0.033 -0.204 -0.073 0.036 -0.022 0.215 0.105## GVIF Df GVIF^(1/(2*Df))
## sex_male 1.149842e+00 1 1.072307
## age 1.157429e+00 1 1.075839
## race_white 1.145496e+00 1 1.070278
## bmi 1.115393e+00 1 1.056122
## group 3.437613e+05 3 8.369692
## kmo 4.535924e+00 1 2.129771
## group:kmo 3.435284e+05 3 8.368747
## StudRes Hat CookD
## 17 2.4000769 0.08986137 0.04508734
## 52 -0.6278348 0.35703624 0.01836004
## 131 1.7665466 0.29158933 0.10465584
## 147 -2.6573749 0.17704511 0.11885289Linear model #5: IDO1 according to demo, group, and CCBL2:group interaction
##
## Call:
## lm(formula = ido1 ~ sex_male + +age + race_white + bmi + group *
## ccbl2, data = df)
##
## Residuals:
## Min 1Q Median 3Q Max
## -3.08913 -0.79039 0.01909 0.78736 2.63842
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 4.660277 4.091826 1.139 0.2577
## sex_male1 -0.151306 0.239757 -0.631 0.5295
## age 0.007795 0.031033 0.251 0.8022
## race_white1 -0.567889 0.302691 -1.876 0.0638 .
## bmi -0.015866 0.019870 -0.798 0.4266
## groupPC -7.398578 5.957462 -1.242 0.2174
## groupSA -2.297322 4.715773 -0.487 0.6273
## groupSI -11.148783 4.993524 -2.233 0.0280 *
## ccbl2 0.130735 0.474341 0.276 0.7835
## groupPC:ccbl2 0.871082 0.699649 1.245 0.2163
## groupSA:ccbl2 0.274239 0.564387 0.486 0.6282
## groupSI:ccbl2 1.415644 0.595451 2.377 0.0195 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.147 on 93 degrees of freedom
## (47 observations deleted due to missingness)
## Multiple R-squared: 0.2901, Adjusted R-squared: 0.2061
## F-statistic: 3.455 on 11 and 93 DF, p-value: 0.0004625## (Intercept) sex_male1 age race_white1 bmi groupPC groupSA groupSI ccbl2 groupPC:ccbl2 groupSA:ccbl2 groupSI:ccbl2
## 0.111 -0.062 0.025 -0.183 -0.078 -0.121 -0.048 -0.218 0.027 0.122 0.047 0.232## GVIF Df GVIF^(1/(2*Df))
## sex_male 1.146917e+00 1 1.070942
## age 1.127871e+00 1 1.062013
## race_white 1.170057e+00 1 1.081692
## bmi 1.078559e+00 1 1.038537
## group 1.298655e+07 3 15.331416
## ccbl2 6.418779e+00 1 2.533531
## group:ccbl2 1.372637e+07 3 15.473644
## StudRes Hat CookD
## 12 -2.9966061 0.1228728 0.096542902
## 24 -0.6928566 0.3446842 0.021159782
## 31 2.5533881 0.1402044 0.083633102
## 102 0.1979037 0.4028169 0.002224524
## 118 1.8966409 0.2884065 0.118195159Note: there were no interactions between group and demo vars
Exploratory model #1: ordinal logistic regression
## Call:
## MASS::polr(formula = group_ordinal ~ sex_male + age + bmi + race_white +
## ido1, data = df, Hess = TRUE)
##
## Coefficients:
## Value Std. Error t value
## sex_male1 -0.02861 0.3644 -0.0785
## age 0.01699 0.0465 0.3654
## bmi 0.00151 0.0324 0.0466
## race_white1 -0.51885 0.4695 -1.1051
## ido1 0.07579 0.1362 0.5565
##
## Intercepts:
## Value Std. Error t value
## HC|PC -0.510 1.601 -0.319
## PC|SI 0.201 1.603 0.125
## SI|SA 1.491 1.615 0.923
##
## Residual Deviance: 284.1552
## AIC: 300.1552
## (47 observations deleted due to missingness)## Waiting for profiling to be done...## OR 2.5 % 97.5 %
## sex_male1 0.9717995 0.4749505 1.988987
## age 1.0171328 0.9284528 1.114674
## bmi 1.0015085 0.9401703 1.068182
## race_white1 0.5952055 0.2334536 1.486624
## ido1 1.0787394 0.8249346 1.410544## sex_male1 age bmi race_white1 ido1
## -0.028605798 0.016987700 0.001507405 -0.518848547 0.075793167
## --------------------------------------------
## Test for X2 df probability
## --------------------------------------------
## Omnibus 84 10 0
## sex_male1 6.55 2 0.04
## age 5.07 2 0.08
## bmi 4.57 2 0.1
## race_white1 10.5 2 0.01
## ido1 8 2 0.02
## --------------------------------------------
##
## H0: Parallel Regression Assumption holdsExploratory model #2: multinomial regression
## # weights: 28 (18 variable)
## initial value 145.560908
## iter 10 value 133.769710
## iter 20 value 129.494224
## final value 129.467553
## converged## (Intercept) sex_male1 age race_white1 bmi ido1
## PC 0.3083239 0.02052843 0.1094183 0.8602799 0.2121117 0.45489127
## SA 0.9267028 0.74341920 0.7279717 0.1413323 0.7158462 0.94325537
## SI 0.4774108 0.48535501 0.1918654 0.2028456 0.9498060 0.01312516