streg — Parametric survival models - Stata

1 Parametric survival modelsDescriptionQuick startMenuSyntaxOptionsRemarks and examplesStored resultsMethods and formulasReferencesAlso seeDescriptionstregperforms maximum likelihood estimation for Parametric regression survival -time be used with single- or multiple-record or single- or multiple-failure st data. Survivalmodels currently supported are exponential, Weibull, Gompertz, lognormal, loglogistic, and generalizedgamma. Parametric frailty models and shared-frailty models are also fit see [ST]stcoxfor proportional hazards startWeibull survival model with covariatesx1andx2usingstsetdatastreg x1 x2, distribution(weibull)Use accelerated failure-time metric instead of proportional-hazards parameterizationstreg x1 x2, distribution(weibull) timeDifferent intercepts and ancillary parameters for strata identified bysvarstreg x1 x2, distribution(weibull) strata(svar)Lognormal survival modelstreg x1 x2, distribution(lognormal)

2 As above, but also model frailty using the gamma distributionstreg x1 x2, distribution(lognormal) frailty(gamma)Specify shared frailty within groups identified bygvarstreg x1 x2, distribution(lognormal) frailty(gamma) shared(gvar)MenuStatistics> survival analysis>Regression models > Parametric survival models12 streg Parametric survival modelsSyntaxstreg[indepvars] [if] [in] [,options]optionsDescriptionModelnoconst antsuppress constant termdistribution(exponential)exponential survival distributiondistribution(gompertz)Gomper tz survival distributiondistribution(loglogistic)log logistic survival distributiondistribution(llogistic)synon ym fordistribution(loglogistic)distribution (weibull)Weibull survival distributiondistribution(lognormal)logno rmal survival distributiondistribution(lnormal)synonym fordistribution(lognormal)distribution(g gamma)generalized gamma survival distributionfrailty(gamma)gamma frailty distributionfrailty(invgaussian)

3 Inverse-Gaussian distributiontimeuse accelerated failure-time metricModel 2strata(varname)strataIDvariableoffset(v arname)includevarnamein model with coefficient constrained to 1shared(varname)shared frailtyIDvariableancillary(varlist)useva rlistto model the first ancillary parameteranc2(varlist)usevarlistto model the second ancillary parameterconstraints(constraints)apply specified linear constraintsSE/Robustvce(vcetype)vcetypem ay beoim,robust,clusterclustvar,opg,bootstr ap, orjackknifeReportinglevel(#)set confidence level; default islevel(95)nohrdo not report hazard ratiostratioreport time ratiosnoshowdo not show st setting informationnoheadersuppress header from coefficient tablenolrtestdo not perform likelihood-ratio testnocnsreportdo not display constraintsdisplayoptionscontrol columns and column formats, row spacing, line width,display of omitted variables and base and empty cells, andfactor-variable labelingMaximizationmaximizeoptionscontr ol the maximization process.

4 Seldom usedcollinearkeep collinear variablescoeflegenddisplay legend instead of statisticsstreg Parametric survival models 3 You muststsetyour data before usingstreg; see [ST] contain factor variables; see[U] Factor ,bootstrap,by,collect,fmm,fp,jackknife,m fp,mi estimate,nestreg,statsby,stepwise, andsvyare allowed; see[U] Prefix commands. For more details, see [BAYES]bayes: stregand [FMM]fmm: (bootstrap)andvce(jackknife)are not allowed with themi estimateprefix; see [MI]mi (),vce(), andnoheaderare not allowed with thesvyprefix; see [SVY] ,iweights, andpweights may be specified usingstset; see [ST]stset.

5 However, weights may not bespecified if you are using thebootstrapprefix with not appear in the dialog [U] 20 Estimation and postestimation commandsfor more capabilities of estimation model noconstant; see [R]Estimation (distname)specifies the survival model to be fit. A specifieddistribution()isremembered from one estimation to the next whendistribution()is not instance, typingstreg x1 x2, distribution(weibull)fits a Weibull model . Subsequently,you do not need to specifydistribution(weibull)to fit other Weibull regression Stata estimation commands, includingstreg, redisplay results when you type the commandname without arguments.

6 To fit a model with no explanatory variables, typestreg, distribu-tion(distname)..frailty(gamma|i nvgaussian)specifies the assumed distribution of the frailty, or estimation results, in addition to the standard parameter estimates, will contain an estimate ofthe variance of the frailties and a likelihood-ratio test of the null hypothesis that this variance iszero. When this null hypothesis is true, the model reduces to the model withfrailty(distname)not specifiedfrailty()is remembered from one estimation to the next whendistribution()is not specified.

7 When you specifydistribution(), the previously remembered specification offrailty()is that the model be fit in the accelerated failure-time metric rather than in the logrelative-hazard metric. This option is valid only for the exponential and Weibull models becausethese are the only models that have both a proportional hazards and an accelerated failure-timeparameterization. Regardless of metric, the likelihood function is the same, and models are equallyappropriate viewed in either metric; it is just a matter of changing the be specified at estimation.

8 model 2 strata(varname)specifies the stratificationIDvariable. Observations with equal values of thevariable are assumed to be in the same stratum. Stratified estimates (with equal coefficients acrossstrata but intercepts and ancillary parameters distinct for each stratum) are then obtained. Thisoption is not available iffrailty(distname)is (varname); see [R]Estimation (varname)is valid withfrailty()and specifies a variable defining those groups over whichthe frailty is shared, analogous to a random-effects model for panel data wherevarnamedefines ()specified withoutshared()treats the frailties as occurring at the streg Parametric survival modelsA specifiedshared()is remembered from one estimation to the next whendistribution()is not specified.

9 When you specifydistribution(), the previously remembered specification ofshared()is ()may not be used withdistribution(ggamma),vce(robust),vce (clusterclust-var),vce(opg), thesvyprefix, or in the presence of delayed entries or ()is specified withoutfrailty()and there is no rememberedfrailty()from theprevious estimation,frailty(gamma)is assumed to provide behavior analogous tostcox; see[ST] (varlist)specifies that the ancillary parameter for the Weibull, lognormal, Gompertz, andloglogistic distributions and that the first ancillary parameter (sigma) of the generalized log-gammadistribution be estimated as a linear combination ofvarlist.

10 This option may not be used withfrailty(distname).When an ancillary parameter is constrained to be strictly positive, the logarithm of the ancillaryparameter is modeled as a linear combination (varlist)specifies that the second ancillary parameter (kappa) for the generalized log-gammadistribution be estimated as a linear combination ofvarlist. This option may not be used withfrailty(distname).constraints(constr aints); see [R]Estimation options. SE/Robust vce(vcetype)specifies the type of standard error reported, which includes types that are derived fromasymptotic theory (oim,opg), that are robust to some kinds of misspecification (robust), thatallow for intragroup correlation (clusterclustvar), and that use bootstrap or jackknife methods(bootstrap,jackknife); see [R]vceoption.