survfit adjust

2025

3 min read

·

28-03-2025

9

0

Share

The survfit function in R's survival package is a cornerstone for analyzing survival data. While it's straightforward to generate basic survival curves, understanding and interpreting adjusted survival curves, particularly when accounting for covariates, is crucial for drawing accurate conclusions. This article will guide you through the process, explaining how to create and interpret adjusted survival curves using survfit in R. We will focus on the application of this function, providing practical examples and interpretation strategies.

Creating Adjusted Survival Curves with survfit

Adjusted survival curves show the predicted survival probabilities after controlling for the influence of other variables. This allows us to isolate the effect of a specific treatment or risk factor, even if confounded by other factors.

To create adjusted curves, you need to include a survfit formula that includes your predictor variable(s) of interest along with other covariates you want to adjust for. The basic syntax is:

survfit(Surv(time, status) ~ treatment + covariate1 + covariate2, data = your_data)

• Surv(time, status): This specifies the survival object. time represents the follow-up time, and status indicates event occurrence (1 = event, 0 = censored).

• treatment: This is your primary variable of interest—the effect you're trying to isolate.

• covariate1, covariate2: These are the covariates you wish to adjust for in your analysis.

• data = your_data: Specifies the data frame containing your variables.

Example:

Let's say you're investigating the effect of a new drug (treatment) on survival, while adjusting for age (age) and gender (gender). Your code might look like this:

library(survival)

# Sample data (replace with your actual data)
data <- data.frame(
  time = c(10, 15, 20, 25, 30, 12, 18, 22, 28, 35),
  status = c(1, 1, 0, 1, 0, 1, 0, 1, 1, 0),
  treatment = factor(c(1, 1, 0, 0, 1, 0, 1, 0, 1, 0)),
  age = c(60, 55, 70, 65, 50, 75, 62, 58, 68, 72),
  gender = factor(c("M", "F", "M", "F", "M", "F", "M", "F", "M", "F"))
)

# Fit the adjusted survival model
fit <- survfit(Surv(time, status) ~ treatment + age + gender, data = data)

# Plot the adjusted survival curves
plot(fit, col = c("blue", "red"), lty = 1, xlab = "Time", ylab = "Survival Probability",
     main = "Adjusted Survival Curves")
legend("topright", legend = levels(data$treatment), col = c("blue", "red"), lty = 1)

This code generates adjusted survival curves for the treatment groups, accounting for age and gender. The curves will show the estimated survival probabilities for each treatment group after controlling for the effects of age and gender.

Interpreting Adjusted Survival Curves

Interpreting the adjusted survival curves involves comparing the curves visually and potentially using statistical tests like the log-rank test (on the unadjusted data for overall difference) to assess if the differences between curves are statistically significant.

• Visual Inspection: Look at the separation between the curves. Larger separations suggest a stronger effect of the treatment, even after adjusting for covariates.

• Statistical Tests: While the survfit function itself doesn't directly perform comparisons between adjusted curves, you can use the survdiff function to compare the survival distributions for the unadjusted data (before considering covariates). This gives an overall comparison but doesn't represent the adjusted difference. To get fully adjusted p-values, you need to use a Cox Proportional Hazards model.

• Confidence Intervals: Pay attention to confidence intervals around the survival curves. Wider intervals indicate more uncertainty in the estimates.

Considering Model Assumptions

Remember that the validity of adjusted survival curves depends on the assumptions of the underlying model (often a Cox proportional hazards model, which is used to calculate adjusted survival probabilities). Ensure that the proportional hazards assumption is met. If it's violated, alternative methods or model modifications might be needed.

Beyond Basic Adjustments: Stratification and Interactions

survfit can handle more complex scenarios:

• Stratification: You can stratify your analysis by adding a strata() term to the formula. This creates separate adjusted curves within each stratum. This can be useful to explore how the treatment effect varies across different subgroups.

• Interactions: Adding interaction terms (e.g., treatment:age) allows you to examine whether the treatment effect differs across levels of other variables.

Conclusion

The survfit function in R is a versatile tool for analyzing survival data, allowing the creation and interpretation of adjusted survival curves. This enables researchers to draw more nuanced conclusions about the impact of a particular variable, while controlling for confounding factors. Remember to consider the assumptions of the underlying model, and explore techniques like stratification and interaction terms to perform more sophisticated analyses. Always remember to report both the visual comparison and appropriate statistical tests to support the interpretation of your findings.