Propensity Score Weighting for PSsurvival Package

Description

Functions for calculating propensity score weights with optional trimming for binary and multiple treatment groups. Calculate Inverse Probability Weights (IPW)

Calculates inverse probability weights for binary or multiple treatment groups. For ATE (Average Treatment Effect), weights are 1/e_j(X) for each treatment j. For ATT (Average Treatment Effect on the Treated), weights target a specific treatment group.

Usage

calculate_ipw_weights(
  ps_result,
  data,
  treatment_var,
  estimand = "ATE",
  att_group = NULL
)

Arguments

Details

For ATE with multiple treatments, weights are:

w_j(X_i) = \frac{1}{e_j(X_i)} \text{ for individual i in treatment j}

For ATT targeting treatment k, weights are:

w_j(X_i) = \begin{cases} 1 & \text{if } j = k \\ \frac{e_k(X_i)}{e_j(X_i)} & \text{if } j \neq k \end{cases}

Value

A numeric vector of IPW weights with length equal to nrow(data).

Calculate Overlap Weights

Description

Calculates overlap weights using the Li & Li (2019) formula for binary or multiple treatment groups. Overlap weights target the population with the most equipoise and are bounded between 0 and 1.

Usage

calculate_overlap_weights(ps_result, data, treatment_var)

Arguments

Details

For multiple treatments, overlap weights are calculated as:

w_j(X_i) = \frac{1/e_j(X_i)}{\sum_{l=1}^{J} 1/e_l(X_i)}

For binary treatment (J=2), this reduces to:

w(X) = 1 - P(Z = observed | X)

which equals the probability of receiving the opposite treatment.

Value

A numeric vector of overlap weights with length equal to nrow(data).

References

Li, F., & Li, F. (2019). Propensity score weighting for causal inference with multiple treatments. The Annals of Applied Statistics, 13(4), 2389-2415.

Check Data Structure

Description

Validates that the data structure is appropriate for model fitting. Only stops if models cannot be fit numerically.

Usage

check_data_structure(data, treatment_var, ps_formula, censor_formula)

Arguments

Value

Invisible NULL if checks pass; otherwise throws an error.

Check if Variables Exist in Data

Description

Verifies that all variables referenced in formulas exist in the dataset.

Usage

check_variables_exist(data, treatment_var, ps_formula, censor_formula)

Arguments

Value

Invisible NULL if all checks pass; otherwise throws an error.

Survival Effect Estimation with Weibull Censoring Scores

Description

Implements Estimator I (AJE 2022, Eq. 2) combining propensity score weighting and inverse probability of censoring weighting to estimate counterfactual survival functions and treatment effects. Compute Etau Normalization Constant

Usage

compute_etau(
  estimand,
  method,
  ps_matrix,
  n_levels,
  att_group = NULL,
  treatment_levels = NULL
)

Arguments

Value

Numeric scalar Etau.

Estimate Censoring Scores Using Cox Regression

Description

Estimate Censoring Scores Using Cox Regression

Usage

estimate_censoring_score_cox(
  data,
  time_var,
  treatment_var,
  formula,
  control = list(),
  ties = "efron"
)

Arguments

Details

Fits Cox models within each treatment group. Censoring scores computed as:

K_c^{(j)}(t, X) = \exp(-H_0^{(j)}(t) \cdot \exp(\beta_j' X))

where H_0^{(j)}(t) is cumulative baseline hazard. Baseline hazards evaluated at nearest time point for each individual.

Value

List with class "censoring_score_cox":

Censoring Score Estimation

Description

Estimate censoring scores P(C >= T | X) using Weibull or Cox models fit separately within each treatment group. Estimate Censoring Scores Using Weibull Regression

Usage

estimate_censoring_score_weibull(
  data,
  time_var,
  treatment_var,
  formula,
  control = list(maxiter = 350)
)

Arguments

Details

Fits Weibull models within each treatment group. Censoring scores computed as:

K_c^{(j)}(t, X) = \exp(-\exp(X'\theta_j) \cdot t^{\gamma_j})

where \theta_j = -\beta_j/\sigma_j, \gamma_j = 1/\sigma_j.

Value

List with class "censoring_score_weibull":

Propensity Score Estimation for PSsurvival Package

Description

Functions for estimating propensity scores for binary and multiple treatment groups. Estimate Propensity Scores

Fits a propensity score model and extracts propensity scores for binary or multiple treatment groups. For binary treatments, uses binomial logistic regression. For multiple treatments (>2 levels), uses multinomial logistic regression to estimate generalized propensity scores.

Usage

estimate_ps(data, treatment_var, ps_formula, ps_control = list())

Arguments

Details

Propensity Score Definition: Returns P(Z = observed | X) for each individual, not P(Z=1|X) for all (as in Rosenbaum & Rubin 1983). This definition enables direct use in IPW and extends naturally to multiple treatments.

Binary Treatments (2 levels): Fits binomial logistic regression via glm(). Treatment is factorized with levels sorted by sort(): numerically for numeric, alphabetically for character, by factor level order for factor. Returns P(Z = observed | X).

Multiple Treatments (>2 levels): Fits multinomial logistic regression via nnet::multinom(). Returns P(Z = observed | X) for each individual from the generalized PS matrix.

Control Parameters (ps_control):

• Binary: glm.control() parameters (default: epsilon=1e-08, maxit=25)

• Multiple: multinom() parameters (default: MaxNWts=10000, maxit=100, trace=FALSE)

Value

A list with the following components:

Examples

# Example 1: Binary treatment
data(simdata_bin)
ps_bin <- estimate_ps(
  data = simdata_bin,
  treatment_var = "Z",
  ps_formula = Z ~ X1 + X2 + X3 + B1 + B2
)
summary(ps_bin$ps)
table(simdata_bin$Z)

# Example 2: Multiple treatments
data(simdata_multi)
ps_multi <- estimate_ps(
  data = simdata_multi,
  treatment_var = "Z",
  ps_formula = Z ~ X1 + X2 + X3 + B1 + B2
)
head(ps_multi$ps_matrix)

Estimate Propensity Score Weights

Description

Calculates propensity score weights for causal inference with optional trimming. Supports ATE, ATT, and overlap population estimands for binary and multiple treatment groups.

Usage

estimate_weights(
  ps_result,
  data,
  treatment_var,
  estimand = "ATE",
  att_group = NULL,
  trim = NULL,
  delta = NULL,
  alpha = NULL
)

Arguments

Details

Trimming Workflow: When trimming is requested, the function: (1) identifies observations to trim using PS from full data, (2) re-estimates PS on trimmed data, (3) calculates weights from re-estimated PS. This ensures trimming uses the original covariate distribution while weights reflect the overlapping population.

Overlap weights do not support trimming (already bounded in [0,1]).

Value

A list containing:

References

Li, F., & Li, F. (2019). Propensity score weighting for causal inference with multiple treatments. The Annals of Applied Statistics, 13(4), 2389-2415.

Yoshida, K., et al. (2019). Multinomial extension of propensity score trimming methods: A simulation study. American Journal of Epidemiology, 188(3), 609-616.

Crump, R. K., et al. (2009). Dealing with limited overlap in estimation of average treatment effects. Biometrika, 96(1), 187-199.

Examples

# Example 1: Overlap weighting for binary treatment
data(simdata_bin)
ps_bin <- estimate_ps(
  data = simdata_bin,
  treatment_var = "Z",
  ps_formula = Z ~ X1 + X2 + X3 + B1 + B2
)
weights_ow <- estimate_weights(
  ps_result = ps_bin,
  data = simdata_bin,
  treatment_var = "Z",
  estimand = "overlap"
)
summary(weights_ow$weights)

# Example 2: ATT with multiple treatments
data(simdata_multi)
ps_multi <- estimate_ps(
  data = simdata_multi,
  treatment_var = "Z",
  ps_formula = Z ~ X1 + X2 + X3 + B1 + B2
)
weights_att <- estimate_weights(
  ps_result = ps_multi,
  data = simdata_multi,
  treatment_var = "Z",
  estimand = "ATT",
  att_group = "C"
)
summary(weights_att$weights)

Marginal Cox Model Estimation with Propensity Score Weighting

Description

Fits a weighted marginal Cox proportional hazards model to estimate marginal hazard ratios between treatment groups using propensity score weights. Fit Weighted Marginal Cox Model

Fits a marginal Cox model 'Surv(time, event) ~ treatment' with propensity score weights to estimate marginal hazard ratios. Automatically handles zero weights (from trimming) by subsetting data before fitting.

Usage

fit_marginal_cox(
  data,
  treatment_var,
  time_var,
  event_var,
  weights,
  treatment_levels,
  reference_level,
  robust = TRUE,
  functionality = "main"
)

Arguments

Details

Functionality Modes:

**"main" mode (point estimation):** - If reference group is missing or has no events after trimming: throws error - If non-reference group is missing: sets its HR and SE to NA, continues - If non-reference group has no events: sets its HR and SE to NA if coxph fails - Does NOT suppress warnings/messages from coxph

**"boot" mode (bootstrap):** - If reference group is missing or has no events: returns hr_estimates with all NA, no error - If non-reference group is missing: sets its HR and SE to NA - Does NOT suppress warnings (this is handled in bootstrap wrapper function)

Model Formula: Fits Surv(time, event) ~ treatment where treatment is a factor with k levels. Cox regression automatically creates k-1 dummy variables relative to the reference level.

Zero Weights: coxph does not accept zero or negative weights. Observations with weight <= 0 are excluded before model fitting. This handles trimming automatically.

Coefficients: Coefficients represent log hazard ratios. To get hazard ratios, use exp(hr_estimates). A positive coefficient means higher hazard (worse survival) compared to reference group.

Robust Variance: When robust = TRUE, the sandwich variance estimator accounts for weighting and provides more conservative standard errors. This is recommended for propensity score weighted analyses.

Value

A list containing:

Generate Bootstrap Sample Indices

Description

Helper function to generate bootstrap sample indices for either full sample resampling (observational studies) or stratified resampling within treatment groups (randomized controlled trials with fixed allocation).

Usage

generate_boot_indices(data, treatment_var, boot_level = "full")

Arguments

Value

Integer vector of bootstrap indices of length nrow(data).

Marginal Cox Model with Propensity Score Weighting

Description

Main user interface for estimating marginal hazard ratios using propensity score weighting. Supports binary and multiple treatment groups with various weighting schemes (ATE, ATT, overlap) and optional trimming. Variance can be estimated via bootstrap or robust sandwich estimator.

Usage

marCoxph(
  data,
  ps_formula,
  time_var,
  event_var,
  reference_level,
  estimand = "ATE",
  att_group = NULL,
  trim = NULL,
  delta = NULL,
  alpha = NULL,
  variance_method = "bootstrap",
  boot_level = "full",
  B = 100,
  parallel = FALSE,
  mc.cores = 2,
  seed = NULL,
  ps_control = list(),
  robust = TRUE
)

Arguments

Details

**Analysis Workflow:** 1. Extract treatment variable from ps_formula. 2. Estimate propensity scores using multinomial logistic regression (or logistic for binary treatment). 3. Calculate propensity score weights based on estimand and optional trim. 4. Fit marginal Cox model Surv(time, event) ~ treatment with weights. 5. Estimate variance via bootstrap (resampling full pipeline) or robust sandwich estimator.

**Variance Estimation:** - bootstrap: Resamples data (full or stratified), re-estimates PS and weights, re-fits Cox model. Provides bootstrap SE for log hazard ratios. - robust: Uses robust sandwich variance from coxph() directly. No bootstrap performed (faster but may be less accurate with extreme weights).

**Trimming:** - Symmetric: Crump extension for multiple treatments (Yoshida et al., 2019). - Asymmetric: Sturmer extension for multiple treatments (Yoshida et al., 2019). - Not supported with overlap weights (already bounded [0,1]).

Value

Object of class "marCoxph" containing:

References

Li, F., & Li, F. (2019). Propensity score weighting for causal inference with multiple treatments. The Annals of Applied Statistics, 13(4), 2389-2415.

Yoshida, K., et al. (2019). Multinomial extension of propensity score trimming methods: A simulation study. American Journal of Epidemiology, 188(3), 609-616.

Examples

# Example 1: Binary treatment with overlap weighting
data(simdata_bin)
result1 <- marCoxph(
  data = simdata_bin,
  ps_formula = Z ~ X1 + X2 + X3 + B1 + B2,
  time_var = "time",
  event_var = "event",
  reference_level = "A",
  estimand = "overlap"
)
summary(result1)

# Example 2: Multiple treatments with ATT and robust variance
data(simdata_multi)
result2 <- marCoxph(
  data = simdata_multi,
  ps_formula = Z ~ X1 + X2 + X3 + B1 + B2,
  time_var = "time",
  event_var = "event",
  reference_level = "C",
  estimand = "ATT",
  att_group = "C",
  variance_method = "robust"
)
summary(result2)

Plot Method for surveff Objects

Description

Plot Method for surveff Objects

Usage

## S3 method for class 'surveff'
plot(
  x,
  type = "surv",
  max_time = NULL,
  strata_to_plot = NULL,
  strata_colors = NULL,
  conf_level = 0.95,
  include_CI = TRUE,
  curve_width = 1,
  CI_alpha = 0.3,
  legend_position = "right",
  legend_title = NULL,
  plot_title = NULL,
  ...
)

Arguments

Details

Creates publication-ready plots of survival curves or treatment effects over time.

For type = "surv": Plots estimated survival functions with optional confidence intervals. Y-axis ranges from 0 to 1.

For type = "survdiff": Plots estimated treatment effects (survival differences) with optional confidence intervals. Y-axis is not constrained to [0,1].

Value

A ggplot2 object.

Print Method for marCoxph Objects

Description

Print Method for marCoxph Objects

Usage

## S3 method for class 'marCoxph'
print(x, max.len = 10, round.digits = 4, ...)

Arguments

Value

Invisibly returns the input object x.

Print Method for surveff Objects

Description

Print Method for surveff Objects

Usage

## S3 method for class 'surveff'
print(x, max.len = 6, round.digits = 4, ...)

Arguments

Value

Invisibly returns the input object x.

Simulated Survival Data with Binary Treatment

Description

A simulated dataset for demonstrating propensity score weighting methods in survival analysis with a binary treatment.

Usage

simdata_bin

Format

A data frame with 1000 observations and 8 variables:

X1

Continuous covariate (standard normal).

X2

Continuous covariate (standard normal).

X3

Continuous covariate (standard normal).

B1

Binary covariate (0/1).

B2

Binary covariate (0/1).

Z

Treatment group: "A" or "B". Distribution is approximately 40:60.

time

Observed follow-up time (event or censoring), range 0-20.

event

Event indicator: 1 = event observed, 0 = censored.

Details

The data were generated with the following characteristics:

• Treatment assignment depends on X1, X2, and B1 via logistic model.

• Survival times follow Weibull distributions with group-specific scales (group A has better survival than group B).

• Censoring times follow an exponential distribution depending on X1 and B1.

• Administrative censoring occurs at time 20.

• Overall censoring rate is approximately 30

See Also

simdata_multi for a dataset with 4 treatment groups.

Examples

data(simdata_bin)
head(simdata_bin)
table(simdata_bin$Z)

Simulated Survival Data with Multiple Treatments

Description

A simulated dataset for demonstrating propensity score weighting methods in survival analysis with four treatment groups.

Usage

simdata_multi

Format

A data frame with 1000 observations and 8 variables:

X1

Continuous covariate (standard normal).

X2

Continuous covariate (standard normal).

X3

Continuous covariate (standard normal).

B1

Binary covariate (0/1).

B2

Binary covariate (0/1).

Z

Treatment group: "A", "B", "C", or "D". Distribution is approximately 20:20:20:35.

time

Observed follow-up time (event or censoring), range 0-20.

event

Event indicator: 1 = event observed, 0 = censored.

Details

The data were generated with the following characteristics:

• Treatment assignment depends on X1, X2, X3, B1, and B2 via multinomial logistic model.

• Survival times follow Weibull distributions with group-specific scales. Survival ordering (best to worst): C > A > B > D.

• Censoring times follow an exponential distribution depending on X1 and B1.

• Administrative censoring occurs at time 20.

• Overall censoring rate is approximately 30

See Also

simdata_bin for a dataset with binary treatment.

Examples

data(simdata_multi)
head(simdata_multi)
table(simdata_multi$Z)

Summary Method for marCoxph Objects

Description

Summary Method for marCoxph Objects

Usage

## S3 method for class 'marCoxph'
summary(object, conf_level = 0.95, round.digits = 4, style = "prints", ...)

Arguments

Details

Confidence intervals are Wald-type intervals calculated as:

• Log scale: logHR ± z_crit * SE

• Original scale: exp(logHR ± z_crit * SE)

The SE used depends on variance_method from the original marCoxph call:

• "robust": Uses logHR_se_robust from sandwich estimator.

• "bootstrap-full" or "bootstrap-strata": Uses logHR_se_bootstrap.

Value

If style = "prints", returns invisibly. If style = "returns", returns a list with:

Summary Method for surveff Objects

Description

Summary Method for surveff Objects

Usage

## S3 method for class 'surveff'
summary(
  object,
  conf_level = 0.95,
  max.len = 6,
  round.digits = 4,
  style = "prints",
  ...
)

Arguments

Value

If style = "prints", returns invisibly. If style = "returns", returns a list with:

Survival Effect Estimation with Cox Censoring Scores

Description

Implements Estimator I (AJE 2022, Eq. 2) combining propensity score weighting and inverse probability of censoring weighting to estimate counterfactual survival functions and treatment effects. Variance estimation uses bootstrap only. Estimate Counterfactual Survival Functions Using Cox Censoring Scores

Usage

surv_cox(
  data,
  time_var,
  event_var,
  treatment_var,
  eval_times = NULL,
  weight_result,
  censoring_formula,
  censoring_control = list(),
  ties = "efron"
)

Arguments

Details

Estimates counterfactual survival function for each treatment group j:

S^{(j)}_w(t) = 1 - \frac{\sum_i w_i I(A_i=j) \delta_i I(T_i \leq t) / K_c^{(j)}(T_i, X_i)}{\sum_i w_i I(A_i=j)}

Censoring scores are estimated using Cox proportional hazards models fit separately within each treatment group. Variance estimation for Cox-based survival functions is performed using bootstrap only (no analytical variance available).

Value

List containing:

Estimate Counterfactual Survival Functions Using Weibull Censoring Scores

Description

Estimate Counterfactual Survival Functions Using Weibull Censoring Scores

Usage

surv_weibull(
  data,
  time_var,
  event_var,
  treatment_var,
  eval_times = NULL,
  weight_result,
  censoring_formula,
  censoring_control = list(maxiter = 350)
)

Arguments

Details

Estimates counterfactual survival function for each treatment group j:

S^{(j)}_w(t) = 1 - \frac{\sum_i w_i I(A_i=j) \delta_i I(T_i \leq t) / K_c^{(j)}(T_i, X_i)}{\sum_i w_i I(A_i=j)}

Value

List containing:

Survival Effect Estimation with Propensity Score Weighting

Description

Main user interface for estimating counterfactual survival functions and treatment effects using propensity score weighting and inverse probability of censoring weighting. Supports binary and multiple treatment groups with various weighting schemes (ATE, ATT, overlap) and optional trimming.

Usage

surveff(
  data,
  ps_formula,
  censoring_formula,
  eval_times = NULL,
  estimand = "ATE",
  att_group = NULL,
  trim = NULL,
  delta = NULL,
  alpha = NULL,
  contrast_matrix = NULL,
  censoring_method = "weibull",
  variance_method = NULL,
  B = 100,
  parallel = FALSE,
  mc.cores = 2,
  seed = NULL,
  censoring_control = NULL,
  ties = "efron",
  ps_control = list(),
  boot_level = "full"
)

Arguments

Details

**Variance Estimation:** - Analytical: Binary treatment with Weibull censoring only (M-estimation). - Bootstrap: All settings (resamples entire pipeline). - Cox: Always uses bootstrap.

**Treatment Effects:** - Binary: S1 - S0 (second level minus first). - Multiple groups: Requires contrast_matrix for pairwise comparisons.

Value

List containing:

Examples

# Example 1: Binary treatment with overlap weighting and Weibull censoring model
data(simdata_bin)
result1 <- surveff(
  data = simdata_bin,
  ps_formula = Z ~ X1 + X2 + X3 + B1 + B2,
  censoring_formula = survival::Surv(time, event) ~ X1 + B1,
  estimand = "overlap",
  censoring_method = "weibull"
)
summary(result1)
plot(result1)

# Example 2: Multiple treatments with ATE and Cox censoring model
data(simdata_multi)
result2 <- surveff(
  data = simdata_multi,
  ps_formula = Z ~ X1 + X2 + X3 + B1 + B2,
  censoring_formula = survival::Surv(time, event) ~ X1 + B1,
  estimand = "ATE",
  censoring_method = "cox",
  variance_method = "bootstrap",
  B = 100
)
summary(result2)

Wrapper for Cox Survival Effect Estimation with Variance

Description

High-level wrapper combining propensity score estimation, weighting (with optional trimming), survival function estimation, and variance estimation using bootstrap. Unlike Weibull-based estimation, Cox censoring scores do not support analytical variance; bootstrap is always used.

Usage

surveff_cox(
  data,
  treatment_var,
  ps_formula,
  time_var,
  event_var,
  censoring_formula,
  eval_times = NULL,
  estimand = "ATE",
  att_group = NULL,
  trim = NULL,
  delta = NULL,
  alpha = NULL,
  contrast_matrix = NULL,
  B = 100,
  parallel = FALSE,
  mc.cores = 2,
  seed = NULL,
  censoring_control = list(),
  ties = "efron",
  ps_control = list(),
  boot_level = "full"
)

Arguments

Details

This function implements the complete estimation workflow:

**Without trimming:** 1. Estimate propensity scores on full data 2. Estimate weights from PS 3. Estimate censoring scores on full data 4. Estimate survival functions 5. Estimate differences (if applicable) 6. Estimate variances using bootstrap

**With trimming:** 1. Estimate propensity scores on full data (PS_full) 2. Use PS_full to identify observations to trim 3. Re-estimate propensity scores on trimmed data (PS_trimmed) 4. Estimate weights from PS_trimmed 5. Estimate censoring scores on trimmed data 6. Estimate survival functions on trimmed data 7. Estimate differences (if applicable) 8. Estimate variances using bootstrap

**Treatment differences:** - Binary (2 groups): Always estimates S1 - S0 (second level minus first level), ignoring contrast_matrix even if provided. - Multiple groups (>2): Requires contrast_matrix to estimate differences. Returns NULL for difference_* if contrast_matrix not provided.

**Variance estimation for differences:** - Binary: Differences columns in each bootstrap sample, calculates sample variance across B trials. - Multiple groups: Always uses bootstrap method.

Value

List containing:

Wrapper for Weibull Survival Effect Estimation with Variance

Description

High-level wrapper combining propensity score estimation, weighting (with optional trimming), survival function estimation, and variance estimation (analytical or bootstrap).

Usage

surveff_weibull(
  data,
  treatment_var,
  ps_formula,
  time_var,
  event_var,
  censoring_formula,
  eval_times = NULL,
  estimand = "ATE",
  att_group = NULL,
  trim = NULL,
  delta = NULL,
  alpha = NULL,
  contrast_matrix = NULL,
  variance_method = NULL,
  B = 100,
  parallel = FALSE,
  mc.cores = 2,
  seed = NULL,
  censoring_control = list(maxiter = 350),
  ps_control = list(),
  boot_level = "full"
)

Arguments

Details

This function implements the complete estimation workflow:

**Without trimming:** 1. Estimate propensity scores on full data 2. Estimate weights from PS 3. Estimate censoring scores on full data 4. Estimate survival functions 5. Estimate differences (if applicable) 6. Estimate variances

**With trimming:** 1. Estimate propensity scores on full data (PS_full) 2. Use PS_full to identify observations to trim 3. Re-estimate propensity scores on trimmed data (PS_trimmed) 4. Estimate weights from PS_trimmed 5. Estimate censoring scores on trimmed data 6. Estimate survival functions on trimmed data 7. Estimate differences (if applicable) 8. Estimate variances

**Treatment differences:** - Binary (2 groups): Always estimates S1 - S0 (second level minus first level), ignoring contrast_matrix even if provided. - Multiple groups (>2): Requires contrast_matrix to estimate differences. Returns NULL for difference_* if contrast_matrix not provided.

**Variance estimation for differences:** - Binary with analytical: Uses influence function approach with proper correlation structure (NOT sum of variances). - Binary with bootstrap: Differences columns in each bootstrap sample, calculates sample variance across B trials. - Multiple groups: Always uses bootstrap method.

Value

List containing:

Asymmetric Propensity Score Trimming (Sturmer Extension)

Description

Performs asymmetric (percentile-based) trimming using within-group percentile thresholds. Implements the Sturmer extension to multiple treatments as described in Yoshida et al. (2019).

Usage

trim_weights_asymmetric(ps_result, data, treatment_var, alpha = NULL)

Arguments

Details

The asymmetric trimming rule retains observation i if:

e_{ji} \geq F^{-1}_{e_{ji}|A_i=j}(\alpha|j) \text{ for all } j

where F^{-1}_{e_{ji}|A_i=j}(\alpha|j) is the \alpha-percentile of propensity scores e_{ji} among individuals who actually received treatment j.

Value

A logical vector of length n, where TRUE indicates the observation should be kept and FALSE indicates it should be trimmed.

References

Yoshida, K., et al. (2019). Multinomial extension of propensity score trimming methods: A simulation study. American Journal of Epidemiology, 188(3), 609-616.

Symmetric Propensity Score Trimming (Crump Extension)

Description

Performs symmetric trimming based on minimum propensity score across all treatment groups. Implements the Crump extension to multiple treatments as described in Yoshida et al. (2019).

Usage

trim_weights_symmetric(ps_result, data, treatment_var, delta = NULL)

Arguments

Details

The symmetric trimming rule retains observation i if:

\min_j\{e_{ji}\} \geq \delta

For binary treatment (J=2), this reduces to: e(X) \in [\delta, 1-\delta].

Value

A logical vector of length n, where TRUE indicates the observation should be kept and FALSE indicates it should be trimmed.

References

Yoshida, K., et al. (2019). Multinomial extension of propensity score trimming methods: A simulation study. American Journal of Epidemiology, 188(3), 609-616.

Validate All Inputs for PSsurvdiff

Description

Umbrella function that calls all validation functions and returns the cleaned dataset with complete cases ready for model fitting.

Usage

validate_PSsurvdiff_inputs(
  data,
  ps_formula,
  censor_formula,
  weight_method,
  censor_method,
  trim_alpha,
  trim_q,
  time_points,
  conf_level,
  ps_control,
  censor_control,
  bootstrap_control
)

Arguments

Value

A list containing:

Validate Censoring Formula

Description

Checks that the censoring formula is valid, uses Surv() notation, and extracts the time and event variable names.

Usage

validate_censor_formula(censor_formula)

Arguments

Value

A list containing:

Validate Data Variables

Description

Performs all variable-specific validation checks on treatment, time, event, and covariates. Only stops execution if critical errors are found.

Usage

validate_data_variables(data, treatment_var, ps_formula, censor_formula)

Arguments

Value

Invisible NULL if all checks pass; otherwise throws an error.

Validate Method Arguments

Description

Validates weight_method, censor_method, trimming parameters, time_points, and control lists. Synthesizes all method-specific checks.

Usage

validate_method_arguments(
  weight_method,
  censor_method,
  trim_alpha,
  trim_q,
  time_points,
  conf_level,
  ps_control,
  censor_control,
  bootstrap_control,
  max_time
)

Arguments

Value

Invisible NULL if checks pass; otherwise throws an error.

Data Validation Functions for PSsurvival Package

Description

Data Validation Functions for PSsurvival Package

Usage

validate_ps_formula(ps_formula)

Arguments

Value

A list containing:

Bootstrap Variance Estimation for Marginal Cox Model

Description

Estimates variance of marginal hazard ratios via bootstrap resampling. Bootstrap Variance for Marginal Cox Model

Performs bootstrap resampling to estimate variance of log hazard ratios from weighted marginal Cox model. Supports full (unstratified) and stratified bootstrap by treatment group.

Usage

var_marginalcox_bootstrap(
  data,
  treatment_var,
  time_var,
  event_var,
  ps_formula,
  treatment_levels,
  reference_level,
  estimand = "ATE",
  att_group = NULL,
  trim = NULL,
  delta = NULL,
  alpha = NULL,
  boot_level = "full",
  B = 100,
  robust = TRUE,
  parallel = FALSE,
  mc.cores = 2,
  seed = NULL
)

Arguments

Details

Bootstrap Workflow: For each bootstrap iteration:

1. Resample data with replacement (full or stratified by treatment)

2. Estimate propensity scores on bootstrap sample

3. Calculate weights (with optional trimming)

4. Fit marginal Cox model using fit_marginal_cox with functionality="boot"

5. Record estimates, sample sizes, and event counts

Parallel Processing: Uses parallel::mclapply for parallel bootstrap. Set ncores > 1 to enable. Note: mclapply uses forking (not available on Windows).

Error Handling: Uses fit_marginal_cox(..., functionality="boot") which returns NA for failed estimates instead of throwing errors. This ensures all B trials complete.

Value

A list containing:

Bootstrap Variance Estimation for Cox Survival Functions

Description

Estimates bootstrap variance for counterfactual survival functions from surv_cox(). Supports binary and multiple treatment groups.

Usage

var_surv_cox_bootstrap(
  data,
  surv_result,
  treatment_var,
  time_var,
  event_var,
  censoring_formula,
  censoring_control = list(),
  ties = "efron",
  B = 100,
  parallel = FALSE,
  mc.cores = 2,
  seed = NULL,
  boot_level = "full"
)

Arguments

Details

Each bootstrap iteration resamples observations with replacement and re-estimates propensity scores, weights, and survival functions using the same specifications as the original analysis. Treatment groups may fail independently if not sampled or if models fail to converge.

Value

List containing:

Compute Analytical M-Estimation Variance for Binary Treatment Survival Functions

Description

Computes analytical variance estimates using M-estimation for binary treatment. Calculates variances for S^(0)(t), S^(1)(t), and their difference S^(1)(t) - S^(0)(t).

Usage

var_surv_weibull_analytical(surv_result)

Arguments

Details

Implements M-estimation variance for binary treatment survival functions. For each group j:

I_j = \frac{1}{E_\tau}(I_{\tau,j} + I_{\theta_j} + I_{\gamma_j} + I_{\beta,j})

Var(S^{(j)}) = \sum I_j^2 / n^2

For the difference:

I_{diff} = \frac{1}{E_\tau}(I_{\tau,diff} + I_{\theta_1} - I_{\theta_0} + I_{\gamma_1} - I_{\gamma_0} + I_{\beta,diff})

Var(S^{(1)} - S^{(0)}) = \sum I_{diff}^2 / n^2

Value

List containing:

Bootstrap Variance Estimation for Weibull Survival Functions

Description

Estimates bootstrap variance for counterfactual survival functions from surv_weibull(). Supports binary and multiple treatment groups.

Usage

var_surv_weibull_bootstrap(
  data,
  surv_result,
  treatment_var,
  time_var,
  event_var,
  censoring_formula,
  censoring_control = list(maxiter = 350),
  B = 100,
  parallel = FALSE,
  mc.cores = 2,
  seed = NULL,
  boot_level = "full"
)

Arguments

Details

Each bootstrap iteration resamples observations with replacement and re-estimates propensity scores, weights, and survival functions using the same specifications as the original analysis. Treatment groups may fail independently if not sampled or if models fail to converge.

Value

List containing: