Reference-based imputation is a multiple imputation method for missing endpoint values following treatment discontinuation or another relevant event. It does not supplement observed values by continuing the patient’s own treatment course, but rather derives the subsequent course from a pre-defined reference group. This explicitly incorporates a clinical assumption about the unobservable benefit after the event into the analysis. <\/p>\n
In repeated measurements, several complete datasets are first generated, analyzed, and then combined into an overall estimate. The information prior to discontinuation remains patient- and treatment-specific. For the period thereafter, however, the chosen reference scenario determines which mean development and which variability are used to generate the missing values. <\/p>\n
Copy Reference transfers the mean structure and covariance of the reference group to the entire course, thereby assuming that the affected person derived no benefit from the assigned treatment. Jump to Reference retains the expected values for the randomized group until the event and then immediately adopts the values of the reference group; thus, the treatment advantage achieved up to that point is lost with the event. Copy Increments in Reference retains the advantage achieved up to the event and subsequently extrapolates only the increments of the reference group, but assumes no further additional treatment benefit. The three variants thus answer different questions about the course after the event and are not arbitrarily interchangeable. The precise technical implementation must match the endpoint, visit schedule, covariates, and the intended analysis model. <\/p>\n
The reference group is often the control or placebo group, but can only be meaningful if it reflects the postulated situation after the event. Treatment discontinuation due to adverse effects may require a different assumption than discontinuation due to lack of efficacy. Therefore, the scenario must not be selected based on the results obtained. <\/p>\n
In the protocol and statistical analysis plan, the trigger for imputation, the reference group, the time points treated, and the analysis dataset must be described together. ICH E9(R1) distinguishes intercurrent events from data missing for the respective estimand. Reference-based imputation does not answer this conceptual question, but operationalizes an assumption for the subsequent estimation. <\/p>\n
The number of imputations, random initialization, and rules for intermittently missing values also belong in the technical specification. In the event of premature death or an endpoint that can no longer be meaningfully measured thereafter, it must first be decided whether imputation at all corresponds to the clinical objective. This decision arises from the estimand and not from a preferred software option. <\/p>\n
Unlike a standard imputation under the MAR assumption, this method does not extrapolate a course derived solely from observed data considering observed covariates. It deliberately makes a conservative assumption about what would have happened after discontinuation. The method is therefore not an automatic correction for missing data, but a specified assumption about unobserved treatment effects. <\/p>\n
The existing entry ‘imputation’ refers to the general term for supplementing missing values; ‘missing-data’ describes the problem area. Reference-based imputation is a narrower, scenario-based variant within multiple imputation. It replaces neither the most complete possible follow-up after treatment completion nor the justification for why the chosen reference is clinically plausible. <\/p>\n
The resulting datasets are auxiliary constructs for uncertainty propagation, not reconstructed patient records. Quality controls therefore check whether each imputed value arises only in the intended cases, occurs temporally after the triggering event, and follows the documented reference group. <\/p>\n
In studies with long-term symptom assessments or functional scales, discontinuations between visits can affect treatment groups differently. For evaluation, it is crucial whether values were still collected after treatment completion and which events trigger a change in assumption. Programming, data management, and biostatistics must therefore consistently integrate event data, visit status, and assignment to the imputation scenario. <\/p>\n
Full-service CROs like Mediconomics support the translation of clinical discontinuation constellations into estimands, the specification of reference-based imputation sets in the analysis plan, and the validated programming of imputation and evaluation steps. This also includes tables that comprehensibly present the results alongside a primary analysis and further sensitivity scenarios. <\/p>\n
Are all missing values replaced by a placebo value in Jump to Reference?<\/strong><\/p>\n No. The missing values are generated multiple times from a distribution whose expected course is linked to the defined reference group. Observed values of the individual patient before the event and the covariates considered in the model continue to be included. <\/p>\n Is Copy Reference always less stringent than Jump to Reference?<\/strong><\/p>\n The two scenarios address different clinical assumptions about the benefit after the event. Their results cannot therefore be generally classified as more stringent or more favorable; what is decisive is whether the assumed post-treatment pattern is justified. <\/p>\n Can the method replace data collection after treatment discontinuation?<\/strong><\/p>\n No. Actually collected endpoint data after discontinuation reduce the proportion of model-based assumptions. For confirmatory studies, the EMA emphasizes the continued collection of clinical endpoints, as far as possible. <\/p>\n Reference-based imputation is a multiple imputation method for missing endpoint values following treatment discontinuation or another relevant event. It does not supplement observed values by continuing the patient’s own treatment course, but rather derives the subsequent course from a pre-defined reference group. This explicitly incorporates a clinical assumption about the unobservable benefit after the event […]<\/p>\n","protected":false},"author":10,"featured_media":0,"parent":0,"template":"","meta":{"_acf_changed":false,"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"glossary-cat":[22],"class_list":["post-7747","glossary","type-glossary","status-publish","hentry","glossary-cat-biostatistik-methodik"],"acf":[],"related_terms":"","external_url":"","internal_reference_id":"","_links":{"self":[{"href":"https:\/\/mediconomics.com\/en\/wp-json\/wp\/v2\/glossary\/7747","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/mediconomics.com\/en\/wp-json\/wp\/v2\/glossary"}],"about":[{"href":"https:\/\/mediconomics.com\/en\/wp-json\/wp\/v2\/types\/glossary"}],"author":[{"embeddable":true,"href":"https:\/\/mediconomics.com\/en\/wp-json\/wp\/v2\/users\/10"}],"version-history":[{"count":0,"href":"https:\/\/mediconomics.com\/en\/wp-json\/wp\/v2\/glossary\/7747\/revisions"}],"wp:attachment":[{"href":"https:\/\/mediconomics.com\/en\/wp-json\/wp\/v2\/media?parent=7747"}],"wp:term":[{"taxonomy":"glossary-cat","embeddable":true,"href":"https:\/\/mediconomics.com\/en\/wp-json\/wp\/v2\/glossary-cat?post=7747"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}Regulatory References<\/h2>\n
\n