Stata 15 help for stcrreg

[ST] stcrreg -- Competing-risks regression


stcrreg [indepvars] [if] [in], compete(crvar[==numlist]) [options]

options Description ------------------------------------------------------------------------- Model * compete(crvar[==numlist]) specify competing-risks event(s) tvc(tvarlist) time-varying covariates texp(exp) multiplier for time-varying covariates; default is texp(_t) offset(varname) include varname in model with coefficient constrained to 1 constraints(constraints) apply specified linear constraints collinear keep collinear variables

SE/Robust vce(vcetype) vcetype may be robust, cluster clustvar, bootstrap, or jackknife noadjust do not use standard degree-of-freedom adjustment

Reporting level(#) set confidence level; default is level(95) noshr report coefficients, not subhazard ratios noshow do not show st setting information noheader suppress header from coefficient table notable suppress coefficient table nodisplay suppress output; iteration log is still displayed nocnsreport do not display constraints display_options control columns and column formats, row spacing, line width, display of omitted variables and base and empty cells, and factor-variable labeling

Maximization maximize_options control the maximization process; seldom used

coeflegend display legend instead of statistics ------------------------------------------------------------------------- * compete(crvar[==numlist]) is required. You must stset your data before using stcrreg; see [ST] stset. varlist and tvarlist may contain factor variables; see fvvarlist. bootstrap, by, fp, jackknife, mfp, mi estimate, nestreg, statsby, and stepwise are allowed; see prefix. vce(bootstrap) and vce(jackknife) are not allowed with the mi estimate prefix. Weights are not allowed with the bootstrap prefix. fweights, iweights, and pweights may be specified using stset; see [ST] stset. In multiple-record data, weights are applied to subjects as a whole, not to individual observations. iweights are treated as fweights that can be noninteger, but not negative. coeflegend does not appear in the dialog box. See [ST] stcrreg postestimation for features available after estimation.


Statistics > Survival analysis > Regression models > Competing-risks regression


stcrreg fits, via maximum likelihood, competing-risks regression models on st data, according to the method of Fine and Gray (1999). Competing-risks regression posits a model for the subhazard function of a failure event of primary interest. In the presence of competing failure events that impede the event of interest, a standard analysis using Cox regression (see stcox) is able to produce incidence-rate curves that either 1) are appropriate only for a hypothetical universe where competing events do not occur or 2) are appropriate for the data at hand, yet the effects of covariates on these curves are not easily quantified. Competing-risks regression, as performed using stcrreg, provides an alternative model that can produce incidence curves that represent the observed data and for which describing covariate effects is straightforward.

stcrreg can be used with single- or multiple-record data. stcrreg cannot be used when you have multiple failures per subject.


+-------+ ----+ Model +------------------------------------------------------------

compete(crvar[==numlist]) is required and specifies the events that are associated with failure due to competing risks.

If compete(crvar) is specified, crvar is interpreted as an indicator variable; any nonzero, nonmissing values are interpreted as representing competing events.

If compete(crvar==numlist) is specified, records with crvar taking on any of the values in numlist are assumed to be competing events.

The syntax for compete() is the same as that for stset's failure() option. Use stset, failure() to specify the failure event of interest, that is, the failure event you wish to model using stcox, streg, stcrreg, or whatever. Use stcrreg, compete() to specify the event or events that compete with the failure event of interest. Competing events, because they are not the failure event of primary interest, must be stset as censored.

If you have multiple records per subject, only the value of crvar for the last chronological record for each subject is used to determine the event type for that subject.

tvc(tvarlist) specifies those variables that vary continuously with respect to time, that is, time-varying covariates. These variables are multiplied by the function of time specified in texp().

texp(exp) is used in conjunction with tvc(tvarlist) to specify the function of analysis time that should be multiplied by the time-varying covariates. For example, specifying texp(ln(_t)) would cause the time-varying covariates to be multiplied by the logarithm of analysis time. If tvc(tvarlist) is used without texp(exp), Stata understands that you mean texp(_t), and thus multiplies the time-varying covariates by the analysis time.

Both tvc(tvarlist) and texp(exp) are explained more in Option tvc() and testing the proportional-subhazards assumption in [ST] stcrreg.

offset(varname), constraints(constraints), collinear; see [R] estimation options.

+------------+ ----+ SE/Robust +-------------------------------------------------------

vce(vcetype) specifies the type of standard error reported, which includes types that are robust to some kinds of misspecification (robust), that allow for intragroup correlation (cluster clustvar), and that use bootstrap or jackknife methods (bootstrap, jackknife); see [R] vce_option. vce(robust) is the default in single-record-per-subject st data. For multiple-record st data, vce(cluster idvar) is the default, where idvar is the ID variable previously stset.

Standard Hessian-based standard errors -- vcetype oim -- are not statistically appropriate for this model and thus are not allowed.

noadjust is for use with vce(robust) or vce(cluster clustvar). noadjust prevents the estimated variance matrix from being multiplied by N/(N-1) or g/(g-1), where g is the number of clusters. The default adjustment is somewhat arbitrary because it is not always clear how to count observations or clusters. In such cases, however, the adjustment is likely to be biased toward 1, so we would still recommend making it.

+------------+ ----+ Reporting +-------------------------------------------------------

level(#); see [R] estimation options.

noshr specifies that coefficients be displayed rather than exponentiated coefficients or subhazard ratios. This option affects only how results are displayed and not how they are estimated. noshr may be specified at estimation time or when redisplaying previously estimated results (which you do by typing stcrreg without a variable list).

noshow prevents stcrreg from showing the key st variables. This option is seldom used because most people type stset, show or stset, noshow to set whether they want to see these variables mentioned at the top of the output of every st command; see [ST] stset.

noheader suppresses the header information from the output. The coefficient table is still displayed. noheader may be specified at estimation time or when redisplaying previously estimated results.

notable suppresses the table of coefficients from the output. The header information is still displayed. notable may be specified at estimation time or when redisplaying previously estimated results.

nodisplay suppresses the output. The iteration log is still displayed.

nocnsreport; see [R] estimation options.

display_options: noci, nopvalues, noomitted, vsquish, noemptycells, baselevels, allbaselevels, nofvlabel, fvwrap(#), fvwrapon(style), cformat(%fmt), pformat(%fmt), sformat(%fmt), and nolstretch; see [R] estimation options.

+--------------+ ----+ Maximization +----------------------------------------------------- maximize_options: difficult, technique(algorithm_spec), iterate(#), [no]log, trace, gradient, showstep, hessian, showtolerance, tolerance(#), ltolerance(#), nrtolerance(#), nonrtolerance, and from(init_specs); see [R] maximize. These options are seldom used.

The following option is available with stcrreg but is not shown in the dialog box:

coeflegend; see [R] estimation options.

Examples Setup . webuse hypoxia

Declare data to be survival-time data and declare the failure event of interest, that is, the event to be modeled . stset dftime, failure(failtype==1)

Fit competing-risks model with failtype==2 as the competing event . stcrreg ifp tumsize pelnode, compete(failtype==2)

Replay results, but show coefficients rather than subhazard ratios . stcrreg, noshr

Stored results

stcrreg stores the following in e():

Scalars e(N) number of observations e(N_sub) number of subjects e(N_fail) number of failures e(N_compete) number of competing events e(N_censor) number of censored subjects e(k) number of parameters e(k_eq) number of equations in e(b) e(k_eq_model) number of equations in overall model test e(k_dv) number of dependent variables e(df_m) model degrees of freedom e(ll) log pseudolikelihood e(N_clust) number of clusters e(chi2) chi-squared e(p) p-value for model test e(rank) rank of e(V) e(fmult) 1 if > 1 failure events, 0 otherwise e(crmult) 1 if > 1 competing events, 0 otherwise e(fnz) 1 if nonzero indicates failure, 0 otherwise e(crnz) 1 if nonzero indicates competing, 0 otherwise e(ic) number of iterations e(rc) return code e(converged) 1 if converged, 0 otherwise

Macros e(cmd) stcrreg e(cmdline) command as typed e(depvar) name of dependent variable e(mainvars) variables in main equation e(tvc) time-varying covariates e(texp) function used for time-varying covariates e(fevent) failure event(s) in estimation output e(crevent) competing event(s) in estimation output e(compete) competing event(s) as typed e(wtype) weight type e(wexp) weight expression e(title) title in estimation output e(clustvar) name of cluster variable e(offset1) offset e(chi2type) Wald; type of model chi-squared test e(vce) vcetype specified in vce() e(vcetype) title used to label Std. Err. e(opt) type of optimization e(which) max or min; whether optimizer is to perform maximization or minimization e(ml_method) type of ml method e(user) name of likelihood-evaluator program e(technique) maximization technique e(properties) b V e(predict) program used to implement predict e(marginsnotok) predictions disallowed by margins e(asbalanced) factor variables fvset as asbalanced e(asobserved) factor variables fvset as asobserved

Matrices e(b) coefficient vector e(Cns) constraints matrix e(ilog) iteration log e(gradient) gradient vector e(V) variance-covariance matrix of the estimators e(V_modelbased) model-based variance

Functions e(sample) marks estimation sample


Fine, J. P., and R. J. Gray. 1999. A proportional hazards model for the subdistribution of a competing risk. Journal of the American Statistical Association 94: 496-509.

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index