{smcl} {* Last modified 11th October 2005}{...} {hline} help for {hi:intcens} {hline} {title:Interval-censored survival analysis} {p 8 16 2}{cmd:intcens}{space 2}{it:depvar1} {it:depvar2} [{it:indepvars}] [{it:weight}] [{cmd:if} {it:exp}] [{cmd:in} {it:range}] , {cmdab:d:istribution:(}{it:distname}{cmd:)} [{cmd:time} {cmd:odds} {cmd:eform} {cmdab:r:obust} {cmdab:cl:uster:(}{it:varname}{cmd:)} {cmd:cwien(}{it:varlist}{cmd:)} {cmd:small(}{it:#}{cmd:)} {cmdab:l:evel:(}{it:#}{cmd:)} {cmd:init(}{it:matname}{cmd:)} {it:maximize_options}] {title:Description} {p 4 4 2} {cmd:intcens} fits various distributions to a non-negative outcome {it:y} which is known to be in the interval [{it:depvar1},{it:depvar2}] and which depends on {it:indepvars}. {it:y} for each observation is either point data, interval-censored, left-censored or right-censored. {it:depvar1} and {it:depvar2} should have the following form: {p_end} {col 12}Type of data {col 42} {it:depvar1} {it:depvar2} {col 12}{hline 46} {col 12}point data {col 32}{it:a} = [{it:a},{it:a}]{col 46}{it:a}{space 7}{it:a} {col 12}interval-censored{col 36}[{it:a},{it:b}]{col 46}{it:a}{space 3}<{space 3}{it:b} {col 12}left-censored{col 36}(0,{it:b}]{col 46}{cmd:.}{space 7}{it:b}{space 5}or {col 12}{col 46}{cmd:0}{space 7}{it:b} {col 12}right-censored{col 36}[{it:a},inf){col 46}{it:a}{space 7}{cmd:.} {col 12}{hline 46} {p 4 4 2}Any {cmd:st} settings are ignored.{p_end} {title:Options} {p 4 8 2}{cmd:distribution(}{it:distname}{cmd:)} is not optional. Available distributions are:{p_end} {col 12}Distribution{col 33}syntax{col 45}remarks {col 12}{hline 40} {col 12}Exponential{col 33}{cmd:exp}{col 45} {col 12}Weibull{col 33}{cmd:weib}{col 45} {col 12}Gompertz{col 33}{cmd:gomp}{col 45} {col 12}Log-logistic{col 33}{cmd:logl} or {cmd:llo}{col 45} {col 12}Log-normal{col 33}{cmd:logn} or {cmd:ln}{col 45} {col 12}2 parameter gamma{col 33}{cmd:gam}{col 45}note difference from {cmd:streg} {col 12}Generalized gamma{col 33}{cmd:gen}{col 45}3 parameter gamma {col 12}Inverse Gaussian{col 33}{cmd:invg}{col 45} {col 12}Inverse Gaussian{col 33}{cmd:wien}{col 45}Alternative parameterisation {col 12}Time to endpoint of{col 33}{cmd:wienran}{col 45}See below for parameterisations {col 14}Wiener process {col 14}with random drift {p 4 8 2}{cmd:time} requests that exponential or Weibull results be reported in accelerated failure form.{p_end} {p 4 8 2}{cmd:odds} requests that log-logistic estimates be presented as log odds ratios. The default is accelerated failure form.{p_end} {p 4 8 2}If {cmd:eform} is specified then exponentiated coefficients are reported. The default is to report log hazard ratios etc.{p_end} {p 4 8 2}{cmd:cwien(}{it:varlist}{cmd:)} is only valid with distributions {cmd:wien} or {cmd:wienran}. It allows the initial distance from the endpoint of the Wiener process to be modelled as a function of covariates (see below).{p_end} {p 4 8 2}{cmd:robust} specifies that the sandwich estimator of variance is to be used in place of the conventional MLE variance estimator. {cmd:robust} combined with {cmd:cluster()} further allows observations which are not independent within cluster (although they must be independent between clusters).{p_end} {p 4 8 2}{cmd:cluster(}{it:varname}{cmd:)} specifies that the observations are independent across groups (clusters) but not necessarily independent within groups. {it:varname} specifies to which group each observation belongs. {cmd:cluster()} implies {cmd:robust}; that is, specifying {cmd:robust cluster()} is equivalent to typing {cmd:cluster()} by itself. {p 4 8 2}{cmd:small(}{it:#}{cmd:)} specifies a tolerance within which interval-censored observations are treated as point data. The default is 1E-6.{p_end} {p 4 8 2}{cmd:level(}{it:#}{cmd:)} specifies the confidence level, in percent, for the confidence intervals of the coefficients; see help {help level}.{p_end} {p 4 8 2}{it:maximize_options} control the maximization process; see help {help maximize}.{p_end} {p 4 8 2}{cmd:init(}{it:matname}{cmd:)} supplies a matrix of inital values. They must be in the correct order but the columns do not need to be labelled. Use this option rather than from(), which is one of the {it:maximize_options}.{p_end} {title:Parameterisations} {p 4 4 2}The exponential, Weibull, Gompertz, log-logistic, log-normal and generalized gamma distributions are parameterised as in {cmd:streg}.{break} The log-likelihoods reported for the distributions which are special cases of the generalized gamma are calculated so as to be comparable to the generalized gamma using likelihood ratio tests.{p_end} {p 4 4 2}The log-logistic results can also be shown as odds ratios, in which case the survivor function is:{break} S(t) = 1/(1+lambda*t^(1/gamma)), lambda=exp(X*B){p_end} {p 4 4 2}The other densities are:{p_end} {p 4 4 2}Two parameter gamma {cmd:gam}{break} f(t) = (lambda*alpha)^alpha*t^(alpha-1)*exp(-lambda*alpha*t)/gamma(alpha), lambda = exp(-X*B){p_end} {p 4 4 2}Inverse Gaussian {cmd:invg}{break} f(t) = 1/sqrt(2*pi*phi*t^3)*exp(-(t-eta)^2/(2*eta^2*phi*t)), eta = exp(X*B){p_end} {p 4 4 2}Inverse Gaussian {cmd:wien}{break} f(t) = c/sqrt(2*pi*t^3)*exp(-(mu*t-c)^2/(2*t)), mu = exp(X*B), c = exp(X1*B1){p_end} {p 4 4 2}This is the time to first reach a distance c from the starting point for a Wiener process with drift rate mu and variance parameter 1.{break} The two parameterisations are related by eta = c/mu, phi = 1/c^2.{p_end} {p 4 4 2}Allowing the drift rate to have a normal distribution between subjects with mean mu and standard deviation tau gives the following density for time to reach c (not a proper density because the event may never happen). There may be convergence problems with this model:{p_end} {p 4 4 2}{cmd:wienran}{break} f(t) = c/(sqrt(2*pi)*t*sqrt(t^2*tau^2+t))*exp(-(mu*t-c)^2/(2*(t^2*tau^2+t))), mu = exp(X*B), c = exp(X1*B1){p_end} {p 4 4 2}With the distributions {cmd:wien} or {cmd:wienran}, X is {it:indepvars}, the covariates listed before the comma, and X1 is the covariates in the option {cmd:cwien( )}. X would be factors which affect the rate of movement towards the endpoint, while X1 would be factors which reflect the initial distance from the endpoint.{p_end} {p 4 9 2}See: Aalen O.O. & Gjessing H.K. (2001) Understanding the shape of the hazard rate: A process point of view. Statistical Science, 16, 1.{p_end} {title:Examples} {p 4 8 2}{cmd:. intcens t0 t1 age trt, dist(weib)} {p_end} {p 4 8 2}{cmd:. intcens t0 t1 age trt, dist(weib) time} {p_end} {p 4 8 2}{cmd:. intcens t0 t1 age trt, dist(logl) odds}{p_end} {p 4 8 2}{cmd:. intcens t0 t1 age trt, dist(invg)}{p_end} {p 4 8 2}{cmd:. intcens t0 t1 trt, dist(wien) cwien(age) }{p_end} {title:Author} {p 12 12 2}Jamie Griffin{break} Infectious Disease Epidemiology Unit{break} London School of Hygiene and Tropical Medicine{break} jamie.griffin@lshtm.ac.uk{p_end} {p 12 12 2}Updated 11th October 2005{p_end} {p 4 4 2}Manual: {hi:[ST] streg}{break} Online: help for {help streg} {p_end}