Regression Calibration Using the Deattenuation Factor Method
Description
Corrects for measurement error in continuous exposures (and covariates) and
returns corrected coefficients, standard errors, p-values, and
variance-covariance matrices. Users may supply their own uncorrected
estimates (e.g., from logistic regression model) instead of using the built-in
outcome model. Only externalvalidation study design is
supported. Based on Rosner, Spiegelman & Willett (1989, 1990) and
Spiegelman, McDermott & Rosner (1997).
Usage
RegCalibDF(
supplyEstimates = FALSE,
ms,
vs,
sur,
exp,
covCalib = NULL,
covOutcomePlus = NULL,
outcome = NA,
method = "lm",
family = NA,
link = NA,
external = TRUE,
pointEstimates = NA,
vcovEstimates = NA
)
RegCalibDF(
supplyEstimates = FALSE,
ms,
vs,
sur,
exp,
covCalib = NULL,
covOutcomePlus = NULL,
outcome = NA,
method = "lm",
family = NA,
link = NA,
external = TRUE,
pointEstimates = NA,
vcovEstimates = NA
)
Arguments
supplyEstimates |
Logical. If TRUE, uncorrected estimates are
supplied by the user via pointEstimates and vcovEstimates,
and ms is optional. Standard regression results are not returned
in this case. Default FALSE.
|
ms |
Main study data frame. Must contain all variables specified in
sur, covCalib, and covOutcomePlus. Required when
supplyEstimates = FALSE.
|
vs |
Internal or external validation study data frame. Must contain
all variables specified in exp, sur, and covCalib.
|
sur |
Character vector of mismeasured exposure(s)/covariate(s)
(surrogates) in the main study dataset.
|
exp |
Character vector of correctly-measured exposure(s)/covariate(s)
in the validation dataset. Must correspond one-to-one with sur and
have the same length.
|
covCalib |
Character vector of correctly-measured covariates to adjust
for in both the calibration model and the outcome model. Default
NULL.
|
covOutcomePlus |
Character vector of correctly-measured risk factors for
the outcome that are not associated with the exposure or surrogate.
These are included in the outcome model only and must not overlap with
covCalib. Default NULL.
|
outcome |
Character. Name of the outcome variable. Required when
method is "lm" or "glm".
|
method |
Character. Outcome modelling method: "lm",
"glm". Default "lm".
|
family |
Family function for glm (e.g., binomial). Not
a character string. Required when method = "glm".
|
link |
Character. Link function for glm (e.g.,
"logit"). Required when method = "glm".
|
external |
Logical. TRUE (default) for an external validation
study; FALSE for an internal validation study. When
external = FALSE, vs must contain the outcome variable and
supplyEstimates must be FALSE.
|
pointEstimates |
Named numeric vector of uncorrected point estimates
from standard regression (intercept excluded). Names must match the
(expanded) covariate names from covCalib followed by
covOutcomePlus. Required when supplyEstimates = TRUE.
|
vcovEstimates |
Named square matrix of uncorrected variance-covariance
estimates (intercept excluded). Column names must match those of
pointEstimates. Required when supplyEstimates = TRUE.
|
Value
A named list containing:
- correctedCoefTable
-
Data frame of corrected estimates, standard
errors, Z-values, p-values, and 95% confidence intervals.
- correctedVCOV
-
Variance-covariance matrix of the corrected
estimates.
- standardCoefTable
-
(When supplyEstimates = FALSE) Results
from the uncorrected standard regression.
- standardVCOV
-
(When supplyEstimates = FALSE) Variance-
covariance matrix from the uncorrected standard regression.
- calibrationModelCoefTable
-
Calibration model slope estimates.
- calibrationModelVCOV
-
Residual variance-covariance matrix from
the calibration model.
References
Rosner B, Willett WC, Spiegelman D (1989). Correction of logistic relative
risk estimates and confidence intervals for systematic within-person
measurement error. Statistics in Medicine 8:1051–1069.
Rosner B, Spiegelman D, Willett WC (1990). Correction of logistic
regression relative risk estimates and confidence intervals for measurement
error: the case of multiple covariates measured with error.
American Journal of Epidemiology 132:734–745.
Spiegelman D, McDermott A, Rosner B (1997). The many uses of the
'regression calibration' method for measurement error bias correction in
nutritional epidemiology. American Journal of Clinical Nutrition
65:1179S–1186S.
Spiegelman D, Carroll RJ, Kipnis V (2001). Efficient regression calibration
for logistic regression in main study/internal validation study designs
with an imperfect reference instrument. Statistics in Medicine
20:139–160.
Examples
data("main_data_sim", package = "RegCalib")
data("valid_data_sim", package = "RegCalib")
set.seed(123)
selected_rows <- sample(
seq_len(nrow(main_data_sim)),
1000
)
main_example <- main_data_sim[
selected_rows,
,
drop = FALSE
]
result <- RegCalibDF(
ms = main_example,
vs = valid_data_sim,
sur = c("fqtfatinc", "fqcalinc", "fqalcinc"),
exp = c("drtfatinc", "drcalinc", "dralcinc"),
covCalib = "agec",
outcome = "case",
method = "glm",
family = binomial,
link = "logit",
external = TRUE
)
result$correctedCoefTable
data("main_data_sim", package = "RegCalib")
data("valid_data_sim", package = "RegCalib")
set.seed(123)
selected_rows <- sample(
seq_len(nrow(main_data_sim)),
1000
)
main_example <- main_data_sim[
selected_rows,
,
drop = FALSE
]
result <- RegCalibDF(
ms = main_example,
vs = valid_data_sim,
sur = c("fqtfatinc", "fqcalinc", "fqalcinc"),
exp = c("drtfatinc", "drcalinc", "dralcinc"),
covCalib = "agec",
outcome = "case",
method = "glm",
family = binomial,
link = "logit",
external = TRUE
)
result$correctedCoefTable