UnirDatos
Javiera
16/4/2020
df=dfWalking %>% left_join(dfBasal) %>% left_join(dfLongitudinal) %>% left_join(dfGIS) %>% select(all_of(names(dfBasal)), everything()) %>% mutate(Nodo=str_sub(Codigo,3,end = 4)) %>% select(Codigo,Nodo,everything()) %>%
filter(Nodo %in% c("01","18"))## Joining with `by = join_by(Codigo)`
## Joining with `by = join_by(Codigo, Tiempo)`
## Joining with `by = join_by(Codigo)`La base de datos en la que se basa este informe está en el siguiente fichero:
df %>% haven::write_sav("salida/datosAbstract-Malaga.sav")Descriptiva de la base de datos
El total de participantes en el estudio se almacena en esta base de datos:
dfTransversal=dfBasal %>% left_join(dfGIS) %>%
inner_join(df %>% group_by(Codigo,Nodo) %>% summarise(numTiempos=n()) %>% ungroup()) %>%
select(Nodo,Codigo,numTiempos,everything())## Joining with `by = join_by(Codigo)`
## `summarise()` has grouped output by 'Codigo'. You can override using the
## `.groups` argument.
## Joining with `by = join_by(Codigo)`dfTransversal %>% haven::write_sav("salida/datosAbstract-Malaga-Transversal.sav")
dfTransversal## # A tibble: 427 × 22
## Nodo Codigo numTiempos grupo_int sexo_s1 diab_prev_s1 HOMA_IR escolarizacion
## <chr> <chr> <int> <fct> <fct> <fct> <dbl> <fct>
## 1 01 P2014… 4 B Hombre No NA Escuela secun…
## 2 01 P2014… 5 B Mujer SÃ NA Escuela secun…
## 3 01 P2012… 7 B Hombre No 3.45 Escuela prima…
## 4 01 P2013… 7 B Hombre SÃ 4.20 Escuela prima…
## 5 01 P2015… 6 A Hombre SÃ 8.36 Escuela secun…
## 6 01 P2010… 7 B Hombre SÃ 4.07 Escuela prima…
## 7 01 P2011… 1 B Hombre SÃ 4.43 Escuela secun…
## 8 01 P2012… 2 B Hombre Sà 6.55 Técnico escu…
## 9 01 P2013… 3 A Mujer SÃ NA Escuela secun…
## 10 01 P2010… 6 B Hombre SÃ NA Escuela secun…
## # ℹ 417 more rows
## # ℹ 14 more variables: estadocivil <fct>, local_nh_p <dbl>, local_nh_i <dbl>,
## # local_dail <dbl>, local_walk <dbl>, NDVI_2015 <dbl>, NDVI_2016 <dbl>,
## # NDVI_2017 <dbl>, NDVI_2018 <dbl>, NDVI_2019 <dbl>, NDVI_2020 <dbl>,
## # NDVI_2021 <dbl>, NDVI_2022 <dbl>, NDVI_2023 <dbl>Usamos tableone para describir las variables más relevantes de esta base de datos:
tableone::CreateTableOne(vars = c("sexo_s1","diab_prev_s1","HOMA_IR","escolarizacion","estadocivil","local_walk"), data = dfTransversal, factorVars = c("sexo_s1","diab_prev_s1","escolarizacion","estadocivil"))##
## Overall
## n 427
## sexo_s1 = Mujer (%) 214 (50.1)
## diab_prev_s1 = SÃ (%) 167 (39.1)
## HOMA_IR (mean (SD)) 5.03 (3.46)
## escolarizacion (%)
## Titulado superior o similares 25 ( 5.9)
## Técnico escuela universitaria 49 (11.5)
## Escuela secundaria o bachiller 147 (34.4)
## Escuela primaria 198 (46.4)
## No sabe leer ni escribir 8 ( 1.9)
## estadocivil (%)
## Soltero/a 29 ( 6.8)
## Casado/a 303 (71.0)
## Viudo/a 43 (10.1)
## Divorciado/a 34 ( 8.0)
## Separado/a 18 ( 4.2)
## local_walk (mean (SD)) 0.74 (2.01)tableone::CreateTableOne(vars = c("Nodo", "diab_prev_s1","HOMA_IR","escolarizacion","estadocivil","local_walk"), data = dfTransversal, factorVars = c("sexo_s1","diab_prev_s1","escolarizacion","estadocivil"),strata = "sexo_s1")## Stratified by sexo_s1
## Hombre Mujer p test
## n 213 214
## Nodo = 18 (%) 52 (24.4) 57 (26.6) 0.678
## diab_prev_s1 = SÃ (%) 96 (45.1) 71 (33.2) 0.016
## HOMA_IR (mean (SD)) 5.52 (4.24) 4.55 (2.40) 0.007
## escolarizacion (%) 0.062
## Titulado superior o similares 13 ( 6.1) 12 ( 5.6)
## Técnico escuela universitaria 30 (14.1) 19 ( 8.9)
## Escuela secundaria o bachiller 81 (38.0) 66 (30.8)
## Escuela primaria 87 (40.8) 111 (51.9)
## No sabe leer ni escribir 2 ( 0.9) 6 ( 2.8)
## estadocivil (%) <0.001
## Soltero/a 13 ( 6.1) 16 ( 7.5)
## Casado/a 171 (80.3) 132 (61.7)
## Viudo/a 6 ( 2.8) 37 (17.3)
## Divorciado/a 14 ( 6.6) 20 ( 9.3)
## Separado/a 9 ( 4.2) 9 ( 4.2)
## local_walk (mean (SD)) 0.71 (1.92) 0.77 (2.09) 0.790LINEAR MIXED MODEL:
Exploración.
Esto es solo para tener una idea de la relación entre las variables. No utilizar para nada que no sea tener algo de intuición de qué esperar:
dg=df %>% select(Codigo,sexo_s1,imc,peso,starts_with("n"), starts_with("dur"), starts_with("ENMO"), starts_with("local")) %>%
group_by(Codigo,sexo_s1) %>%
summarise_all(mean,na.rm=T) %>% ungroup()## Warning: There were 854 warnings in `summarise()`.
## The first warning was:
## ℹ In argument: `Nodo = (new("standardGeneric", .Data = function (x, ...) ...`.
## ℹ In group 1: `Codigo = "P201001"` and `sexo_s1 = Hombre`.
## Caused by warning in `mean.default()`:
## ! argument is not numeric or logical: returning NA
## ℹ Run `dplyr::last_dplyr_warnings()` to see the 853 remaining warnings.dg %>% ggplot( aes(y=n_wb5_01_90_3m_i2m_i25_daily, x = local_walk, color=sexo_s1)) + geom_point(alpha=0.5)+geom_smooth(method="lm")## `geom_smooth()` using formula = 'y ~ x'## Warning: Removed 39 rows containing non-finite values (`stat_smooth()`).## Warning: Removed 39 rows containing missing values (`geom_point()`).
dg %>% ggplot( aes(y=dur_wb5_01_90_3m_i2m_i25_daily, x = local_walk, color=sexo_s1)) + geom_point(alpha=0.5)+geom_smooth(method="lm")## `geom_smooth()` using formula = 'y ~ x'## Warning: Removed 39 rows containing non-finite values (`stat_smooth()`).## Warning: Removed 39 rows containing missing values (`geom_point()`).
dg %>% ggplot( aes(y=ENMO_mean_wb5_01_90_3m_i2m_i25_daily, x = local_walk, color=sexo_s1)) + geom_point(alpha=0.5)+geom_smooth(method="lm")## `geom_smooth()` using formula = 'y ~ x'## Warning: Removed 40 rows containing non-finite values (`stat_smooth()`).## Warning: Removed 40 rows containing missing values (`geom_point()`).
## Base longitudinal de solo la base de datos con datos longitudinales
recogidos
Esta base longitudinal la creamos y almacenamos para chequear que no hay errores. Se imputan los valores de SF36Dim_SF mediante LOCF por que parece que no se mide siempre.
dfLongNoMissing=
df %>% select(Codigo,sexo_s1,grupo_int,Tiempo, n_wb5_01_90_3m_i2m_i25_daily,local_walk, imc,peso , SF36Dim_SF, dur_wb5_01_90_3m_i2m_i25_daily,ENMO_mean_wb5_01_90_3m_i2m_i25_daily) %>%
arrange(Codigo,Tiempo) %>%
#quedarsos solo con las filas completas
filter(complete.cases(.[,1:8])) %>%
mutate(TiempoCuali=as.factor(Tiempo))
dfLongNoMissing=dfLongNoMissing %>% group_by(Codigo) %>% mutate(SF36Dim_SF=zoo::na.locf0(SF36Dim_SF)) %>% ungroup()
dfLongNoMissing %>% haven::write_sav("salida/abstract-Longitudinal-NoMissing.sav")Vamos a describirla con tableone:
tableone::CreateTableOne(vars = c("Tiempo", "TiempoCuali", "n_wb5_01_90_3m_i2m_i25_daily", "dur_wb5_01_90_3m_i2m_i25_daily","imc","SF36Dim_SF", "local_walk"),
data = dfLongNoMissing)##
## Overall
## n 2091
## Tiempo (mean (SD)) 3.05 (2.31)
## TiempoCuali (%)
## 0 291 (13.9)
## 0.5 182 ( 8.7)
## 1 224 (10.7)
## 2 268 (12.8)
## 3 238 (11.4)
## 4 232 (11.1)
## 5 247 (11.8)
## 6 216 (10.3)
## 7 193 ( 9.2)
## n_wb5_01_90_3m_i2m_i25_daily (mean (SD)) 1.77 (1.25)
## dur_wb5_01_90_3m_i2m_i25_daily (mean (SD)) 48.22 (44.26)
## imc (mean (SD)) 31.84 (3.84)
## SF36Dim_SF (mean (SD)) 86.14 (21.99)
## local_walk (mean (SD)) 0.54 (1.95)Los resultados usandomodelos lineales mixtos donde además de una fuente de error aleatorio por visita existe otra fuente de error aleatorio debido al propio paciente (1|Codigo):
variable="n_wb5_01_90_3m_i2m_i25_daily"
modelotxt=sprintf("%s ~ local_walk + SF36Dim_SF + imc + sexo_s1+ Tiempo + grupo_int+(1|Codigo)",variable)
cat (modelotxt)## n_wb5_01_90_3m_i2m_i25_daily ~ local_walk + SF36Dim_SF + imc + sexo_s1+ Tiempo + grupo_int+(1|Codigo)modelo=lmer(formula(modelotxt),data=dfLongNoMissing)
modelsummary(modelo, stars = TRUE,conf_level = 0.95,statistic = c("p.value","conf.int"))| (1) | |
|---|---|
| + p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001 | |
| (Intercept) | 2.730*** |
| (<0.001) | |
| [2.065, 3.394] | |
| local_walk | 0.071** |
| (0.004) | |
| [0.023, 0.119] | |
| SF36Dim_SF | 0.005*** |
| (<0.001) | |
| [0.003, 0.008] | |
| imc | -0.038*** |
| (<0.001) | |
| [-0.057, -0.019] | |
| sexo_s1Mujer | -0.446*** |
| (<0.001) | |
| [-0.638, -0.253] | |
| Tiempo | -0.052*** |
| (<0.001) | |
| [-0.068, -0.036] | |
| grupo_intB | 0.151 |
| (0.124) | |
| [-0.041, 0.344] | |
| SD (Intercept Codigo) | 0.881 |
| SD (Observations) | 0.746 |
| Num.Obs. | 2015 |
| R2 Marg. | 0.098 |
| R2 Cond. | 0.623 |
| AIC | 5363.3 |
| BIC | 5413.7 |
| ICC | 0.6 |
| RMSE | 0.68 |
variable="dur_wb5_01_90_3m_i2m_i25_daily"
modelotxt=sprintf("%s ~ local_walk + SF36Dim_SF + imc + sexo_s1+ Tiempo + grupo_int+(1|Codigo)",variable)
cat (modelotxt)## dur_wb5_01_90_3m_i2m_i25_daily ~ local_walk + SF36Dim_SF + imc + sexo_s1+ Tiempo + grupo_int+(1|Codigo)modelo=lmer(formula(modelotxt),data=dfLongNoMissing)
modelsummary(modelo, stars = TRUE,conf_level = 0.95,statistic = c("p.value","conf.int"))| (1) | |
|---|---|
| + p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001 | |
| (Intercept) | 99.726*** |
| (<0.001) | |
| [76.671, 122.781] | |
| local_walk | 2.359** |
| (0.007) | |
| [0.631, 4.088] | |
| SF36Dim_SF | 0.114** |
| (0.004) | |
| [0.037, 0.191] | |
| imc | -1.823*** |
| (<0.001) | |
| [-2.470, -1.177] | |
| sexo_s1Mujer | -17.239*** |
| (<0.001) | |
| [-24.141, -10.338] | |
| Tiempo | -0.351 |
| (0.200) | |
| [-0.889, 0.186] | |
| grupo_intB | 6.817+ |
| (0.053) | |
| [-0.079, 13.714] | |
| SD (Intercept Codigo) | 31.832 |
| SD (Observations) | 25.188 |
| Num.Obs. | 2015 |
| R2 Marg. | 0.098 |
| R2 Cond. | 0.653 |
| AIC | 19542.4 |
| BIC | 19592.9 |
| ICC | 0.6 |
| RMSE | 23.01 |
variable="ENMO_mean_wb5_01_90_3m_i2m_i25_daily"
modelotxt=sprintf("%s ~ local_walk + SF36Dim_SF + imc + sexo_s1+ Tiempo + grupo_int+(1|Codigo)",variable)
cat (modelotxt)## ENMO_mean_wb5_01_90_3m_i2m_i25_daily ~ local_walk + SF36Dim_SF + imc + sexo_s1+ Tiempo + grupo_int+(1|Codigo)modelo=lmer(formula(modelotxt),data=dfLongNoMissing)
modelsummary(modelo, stars = TRUE,conf_level = 0.95,statistic = c("p.value","conf.int"))| (1) | |
|---|---|
| + p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001 | |
| (Intercept) | 112.469*** |
| (<0.001) | |
| [85.115, 139.823] | |
| local_walk | 1.802* |
| (0.019) | |
| [0.294, 3.310] | |
| SF36Dim_SF | 0.020 |
| (0.737) | |
| [-0.095, 0.135] | |
| imc | -0.497 |
| (0.190) | |
| [-1.239, 0.246] | |
| sexo_s1Mujer | -10.645*** |
| (<0.001) | |
| [-16.657, -4.633] | |
| Tiempo | -0.995* |
| (0.037) | |
| [-1.931, -0.059] | |
| grupo_intB | 1.924 |
| (0.526) | |
| [-4.019, 7.867] | |
| SD (Intercept Codigo) | 20.446 |
| SD (Observations) | 44.995 |
| Num.Obs. | 1935 |
| R2 Marg. | 0.021 |
| R2 Cond. | 0.189 |
| AIC | 20487.3 |
| BIC | 20537.4 |
| ICC | 0.2 |
| RMSE | 42.76 |
Vamos a ver el efecto de walking sobre el IMC y el peso
modelotxt=sprintf("imc ~ %s + dur_wb5_01_90_3m_i2m_i25_daily+n_wb5_01_90_3m_i2m_i25_daily+ ENMO_mean_wb5_01_90_3m_i2m_i25_daily+ sexo_s1+ Tiempo+grupo_int+(1|Codigo)",variable)
modelo=lmer(formula(modelotxt),data=dfLongNoMissing)
modelsummary(modelo, stars = TRUE,conf_level = 0.95,statistic = c("p.value","conf.int"))| (1) | |
|---|---|
| + p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001 | |
| (Intercept) | 32.733*** |
| (<0.001) | |
| [32.062, 33.403] | |
| ENMO_mean_wb5_01_90_3m_i2m_i25_daily | -0.002* |
| (0.045) | |
| [-0.003, 0.000] | |
| dur_wb5_01_90_3m_i2m_i25_daily | -0.007*** |
| (<0.001) | |
| [-0.010, -0.003] | |
| n_wb5_01_90_3m_i2m_i25_daily | 0.042 |
| (0.484) | |
| [-0.075, 0.158] | |
| sexo_s1Mujer | 0.010 |
| (0.978) | |
| [-0.703, 0.723] | |
| Tiempo | -0.040** |
| (0.009) | |
| [-0.071, -0.010] | |
| grupo_intB | -0.258 |
| (0.477) | |
| [-0.970, 0.454] | |
| SD (Intercept Codigo) | 3.490 |
| SD (Observations) | 1.376 |
| Num.Obs. | 2009 |
| R2 Marg. | 0.007 |
| R2 Cond. | 0.866 |
| AIC | 8344.8 |
| BIC | 8395.2 |
| ICC | 0.9 |
| RMSE | 1.24 |
modelotxt=sprintf("peso ~ %s + dur_wb5_01_90_3m_i2m_i25_daily+n_wb5_01_90_3m_i2m_i25_daily+ ENMO_mean_wb5_01_90_3m_i2m_i25_daily+ sexo_s1+ Tiempo+grupo_int+(1|Codigo)",variable)
modelo=lmer(formula(modelotxt),data=dfLongNoMissing)
modelsummary(modelo, stars = TRUE,conf_level = 0.95,statistic = c("p.value","conf.int"))| (1) | |
|---|---|
| + p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001 | |
| (Intercept) | 92.746*** |
| (<0.001) | |
| [90.685, 94.807] | |
| ENMO_mean_wb5_01_90_3m_i2m_i25_daily | -0.003+ |
| (0.093) | |
| [-0.007, 0.001] | |
| dur_wb5_01_90_3m_i2m_i25_daily | -0.018*** |
| (<0.001) | |
| [-0.026, -0.009] | |
| n_wb5_01_90_3m_i2m_i25_daily | 0.138 |
| (0.351) | |
| [-0.152, 0.427] | |
| sexo_s1Mujer | -12.689*** |
| (<0.001) | |
| [-14.941, -10.437] | |
| Tiempo | -0.237*** |
| (<0.001) | |
| [-0.312, -0.162] | |
| grupo_intB | -1.400 |
| (0.223) | |
| [-3.651, 0.851] | |
| SD (Intercept Codigo) | 11.119 |
| SD (Observations) | 3.398 |
| Num.Obs. | 2009 |
| R2 Marg. | 0.227 |
| R2 Cond. | 0.934 |
| AIC | 12153.0 |
| BIC | 12203.5 |
| ICC | 0.9 |
| RMSE | 3.06 |