The answer is 17 years, what is the question: understanding time lags in translational research

The answer is 17 years, what is the question: understanding time lags in translational research Zoë Slote Morris Steven Wooding Jonathan Grant

Tag layer of the DHSS Meta-Tagging Project. The article text is reproduced from the staged copy named in sourceDesc and is not altered; the tags are held separately in standOff so that overlapping and nested spans are preserved.

The answer is 17 years, what is the question: understanding time lags in translational research 2011 Journal of the Royal Society of Medicine 10.1258/jrsm.2011.110180 morris2011-the-answer-is-17-years-what-is-the morris2011-the-answer-is-17-years-what-is-the.txt, the project's staged copy 0348f9d5ab4f521c885261cdedf43220ae88fd73 cc-by medicine

the paper's own words

the tag layer; a sentence written by the tagger about the paper, always anchored to a quotation but not itself a quotation

0 spans located by searching the stored text for the quote's exact words, allowing any whitespace between them A NUMBER IN UNIVERSAL CIRCULATION, WORKED BACKWARDS TO THE DEFINITIONS UNDERNEATH IT - '17 years' for research to reach clinical practice, reviewed across 23 studies that measure different starts, different ends and different things Quotation withheld because this record is not cleared for redistribution. what was actually measured to produce the time lags that health research policy is built on, and whether the figures can be compared or added at all Quotation withheld because this record is not cleared for redistribution. a conceptual model of the translation pipeline with every reviewed study's start point and end point mapped onto it, and the finding that the studies are not measuring the same quantity Quotation withheld because this record is not cleared for redistribution. critique Quotation withheld because this record is not cleared for redistribution. a literature that converges on one figure for a quantity it has never defined, and a policy apparatus that spends against that figure Quotation withheld because this record is not cleared for redistribution. review review Quotation withheld because this record is not cleared for redistribution. THREE STUDIES AGREE ON 17 YEARS, AND THE EXPLANATION OFFERED IS ABOUT THE INSTRUMENT RATHER THAN THE WORLD: `Using different endpoints, different domains and different approaches, Balas and Bohen and Grant et al. both estimate the time lag ... BEING 17 YEARS ... It is surprising that 17 is the answer to several related but differing questions. Is this coincidence or not? One possible reason for the convergence is the difficulty of measuring longer lags ... which provides a ceiling to such estimates.` Agreement between measurements is read as a property of the ruler, which is the strongest form of `an instrument's first output is itself` Quotation withheld because this record is not cleared for redistribution. AND THE CANONICAL 17 IS WRONG BY SIX YEARS ONCE ITS ASSEMBLY IS ACCOUNTED FOR: `Balas and Bohen calculate an average of 17 years from original research to practice formed from adding together a number of single studies of different phases including one that estimates a lag of 6-13 years. Accounting for this changes their estimate of the time lag between journal submission to use in practice from between 17 years to 23 years.` The most-quoted figure in translational research, restated as 23 Quotation withheld because this record is not cleared for redistribution. AND THE NUMBER SETS A NATIONAL RATE OF RETURN: `This level of benefit was calculated using an estimated lag of 17 years. Varying the lag time from 10 to 25 years produced rates of return of 13% and 6%, respectively.` An internal rate of return on thirty years of UK cardiovascular research moves by a factor of two on the choice of a definition nobody in the literature shares. This is the clearest case in the corpus of a definitional choice with a monetary consequence Quotation withheld because this record is not cleared for redistribution. AND THE STUDIES ARE NOT MEASURING ONE QUANTITY: `Much of the literature on lags focuses on dissemination and publication in peer-review journals in particular as these are the most measureable ... the variation in choice of proxy measures means that studies are almost never measuring the same thing, making valid aggregation and generalization difficult.` The proxy was chosen for being measurable, not for being the thing - and the consequence is that the figures cannot be added Quotation withheld because this record is not cleared for redistribution. AND A STATISTIC THAT CHANGED ITS NAME IN TRANSIT: `Decullier et al. report `MEAN` LAGS, and Dwan, in reporting Decullier et al., in their review refer to `MEDIAN` LAGS. NEITHER REPORTS DISTRIBUTIONS. Ranges - or even interquartile ranges as large as 221 YEARS - are seldom reported.` A mean becomes a median between a paper and its reviewer, and neither publishes the distribution that would settle it Quotation withheld because this record is not cleared for redistribution. AND AN AVERAGE THAT HIDES A RANGE LARGER THAN ITSELF: `clinical trials with null or negative results `ON AVERAGE` took `JUST OVER A YEAR LONGER` to be published than those with positive results. This average is associated with A range of six to eight years for studies with a negative or null result, compared with four to five years for those with positive results. Comparing the slowest negative publication with the fastest positive publication MAKES A potential difference of four years - half of the maximum lag.` The headline is one year; the spread is four Quotation withheld because this record is not cleared for redistribution. AND THE FIELD'S HABIT IS NAMED: `There is a clear trend in the literature to seek a single answer to a single question through the calculation of an average. The variation found in the literature suggests that this is not possible (or even desirable), and variation matters. Moreover, many of the published `averages` are derived from adding an empirically derived mean duration for one section of journey from one point in time, in one topic, and adding it to other parts without reflection.` A summary statistic assembled from incommensurable parts Quotation withheld because this record is not cleared for redistribution. AND THE TERM ITSELF MEANS DIFFERENT THINGS TO DIFFERENT PEOPLE, ON THE RECORD: `Woolf has argued that `TRANSLATION RESEARCH MEANS DIFFERENT THINGS TO DIFFERENT PEOPLE` and this is reflected in the various models and definitions found in the literature. However, as translational research also `SEEMS IMPORTANT TO ALMOST EVERYONE` there would seem to be benefit in trying to unify models and definitions.` The two halves of the problem in one sentence: maximum importance, minimum agreement Quotation withheld because this record is not cleared for redistribution. AND ONE OF THE TWO PHASE NAMES IS FLAGGED BY ITS OWN USERS AS CONFUSING: `Type 1 translation, also somewhat confusingly called `BENCH TO BEDSIDE`, refers to the conversion of knowledge from basic science research into a potential clinical product for testing on human subjects` - while Type 2, `research into practice`, is the one that actually reaches the bedside. The metaphor names the wrong end of the pipeline Quotation withheld because this record is not cleared for redistribution. AND THE GAPS THE POLICY TARGETS ARE THE LEAST WELL DEFINED PART: `Conceptual models typically include `TRANSLATIONAL GAPS`, which describe the movement from one phase of research to another. Each of these is also associated with delays, although precisely what and where these gaps are, and how long they are, is again not consistent in the literature. Policy measures to expedite the translation process typically focus on these gaps.` Intervention is aimed at the part with no agreed boundaries Quotation withheld because this record is not cleared for redistribution. AND THE VARIATION SURVIVES INSIDE A SINGLE TOPIC: `examining research relating to advances in neonatal care, Grant et al. traced research papers back through four `GENERATIONS` of publication. They found `the overall time between generations 1 to 4 RANGES FROM 13 YEARS (for artificial surfactant) TO 21 YEARS (for parenteral nutrition). The other three advances TOOK 17 YEARS to develop through four generations of citations`.` One method, one clinical area, five interventions, and an eight-year spread Quotation withheld because this record is not cleared for redistribution. AND THE LAG RUNS IN BOTH DIRECTIONS: `it took six years for a review of evidence supporting the use of thrombolytic drugs to result in a standard recommendation, whereas prophylactic lidocaine was used widely in practice for 25 years based on no evidence of effectiveness.` A field whose central complaint is that evidence reaches practice too slowly, with a case of practice arriving twenty-five years AHEAD of any evidence at all Quotation withheld because this record is not cleared for redistribution. ONLY ONE STUDY MEASURES THE THING EVERYONE QUOTES THE NUMBER FOR: `Table 1 also points to two substantive gaps in knowledge: the time lag involved in and between discovery and development (T1), and the time lag between publication to practice. Only one study has `IMPLEMENTATION` into practice as its endpoint.` The `17 years to reach clinical practice` figure rests on a single measurement of reaching clinical practice Quotation withheld because this record is not cleared for redistribution. AND THE AUTHORS PUT THEIR OWN SEARCH'S WEAKNESS FIRST: `Despite our attempts to be comprehensive, however, we are aware that studies of time lags in health research are widely distributed and not easily identified using formal literature searches and we may have failed to capture relevant studies.` The Methods say the same before the results: `Our search was not intended to be exhaustive` Quotation withheld because this record is not cleared for redistribution. AND THE MODEL THEY BUILD IS LINEAR AND THEY SAY SO: `Clearly this model can be critiqued for being linear and we acknowledge the considerable literature that challenges this notion and accept that research translation is a messy, iterative and complex process ... At the same time, we would argue that for the purposes of understanding and conceptualising time lags the model is appropriate.` A simplification defended by its purpose rather than its accuracy Quotation withheld because this record is not cleared for redistribution. AND THE LAGS ARE NOT STABLE OVER TIME, SO A FIGURE ALSO HAS A DATE: `Pulido noted a difference of 0.9 years or 0.3 years from acceptance to publication in 1992 and 1982, respectively. DiMasi reported a slight shortening of the approval process between 1991 and 2003. Tsuiji and Tsutani reported reduced lags in the drug approval process in Japan following a change in policy to try and expedite it.` A quantity that moves when policy moves, quoted as a constant Quotation withheld because this record is not cleared for redistribution. the 17-year translation lag is a figure produced by an economic calculation of the returns to medical research, including spillovers to the wider economy Quotation withheld because this record is not cleared for redistribution. mapping each reviewed study's start point and end point onto one conceptual model before comparing any figures eliminates: adding or averaging intervals that do not denote the same quantity · outcome: held Quotation withheld because this record is not cleared for redistribution. combining formal search with backward and forward citation tracking and targeted journal searches, after 'a formal search yielded few relevant papers' eliminates: missing a scattered literature that database indexing does not group · outcome: partial Quotation withheld because this record is not cleared for redistribution. reporting the range and the distribution alongside any average eliminates: a headline figure that conceals a spread larger than itself · outcome: failed Quotation withheld because this record is not cleared for redistribution. time lags time lags Quotation withheld because this record is not cleared for redistribution. durations of activities and waits durations of activities and waits Quotation withheld because this record is not cleared for redistribution. Type 1 translation Type 1 translation Quotation withheld because this record is not cleared for redistribution. bench to bedside bench to bedside Quotation withheld because this record is not cleared for redistribution. difficulty of measuring longer lags difficulty of measuring longer lags Quotation withheld because this record is not cleared for redistribution. convergence of average lags convergence of average lags Quotation withheld because this record is not cleared for redistribution. publication to guideline publication to guideline Quotation withheld because this record is not cleared for redistribution. development to commercialization development to commercialization Quotation withheld because this record is not cleared for redistribution. positive results positive results Quotation withheld because this record is not cleared for redistribution. `translational research` - `means different things to different people` (Woolf) and `seems important to almost everyone`; the paper's response is to build a model rather than to pick a definition Quotation withheld because this record is not cleared for redistribution. `Type 1 translation` / `bench to bedside` - basic science to a clinical product FOR TESTING; the metaphor points at the bedside and the phase stops well short of it Quotation withheld because this record is not cleared for redistribution. `time lag` (as used across the 23 studies) - the interval between ANY two markers a study chose: ethical approval to publication, publication to guideline, chemical synthesis to FDA approval, trial registration to publication Quotation withheld because this record is not cleared for redistribution. `17 years` - three separate estimates of three different intervals, whose agreement the authors suggest may be A CEILING EFFECT of citation indexes rather than a fact about research Quotation withheld because this record is not cleared for redistribution. critiques → Balas and Bohen's 17-year estimate, assembled by adding single studies of different phases; corrected here to 23 years Quotation withheld because this record is not cleared for redistribution. critiques → the practice of reporting a single average time lag at all - 'The variation found in the literature suggests that this is not possible (or even desirable)' Quotation withheld because this record is not cleared for redistribution. extends → Trochim et al.'s 'process marker model', which the authors endorse as the framework for the operational markers their review says are missing Quotation withheld because this record is not cleared for redistribution. time lags —operationalised-as→ durations of activities and waits time lags durations of activities and waits the relation itself is separately evidenced · strength of the relation: 0.95 · note: 'Understanding time lags requires…' The operationalisation is the whole argument: 23 studies each picked two markers off an unstated model, and the figures cannot be added because the models were never the same · the relation stated: a lag is only measurable once a conceptual model names the activities and the waits between them - which is why the paper builds the model before reviewing the figures Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but the span says a conceptual model is needed so that durations can be measured; operationalised-as is loose. Quotation withheld because this record is not cleared for redistribution. difficulty of measuring longer lags —causes→ convergence of average lags difficulty of measuring longer lags convergence of average lags the relation itself is separately evidenced · strength of the relation: 0.85 · note: 'ONE POSSIBLE…' Weight 0.85 because the authors offer it as 'One possible reason' and do not test it. But the shape is the finding: agreement between independent measurements read as a property of the instrument rather than of the world · the relation stated: the ceiling on observable lags - citation indexes, records, recollection - is offered as a reason three studies of three different intervals all report 17 years Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but the span offers this as 'one possible reason' for the convergence; hedged. Quotation withheld because this record is not cleared for redistribution. Type 1 translation —renamed-across-fields→ bench to bedside Type 1 translation bench to bedside the relation itself is separately evidenced · strength of the relation: 0.95 · note: `Type 1 translation, also somewhat confusingly called `BENCH TO BEDSIDE`, refers to the conversion of knowledge from basic science research into a potential clinical product for testing on human subjects.` Type 2 - `research into practice` - is `THE ONE CLOSER TO THE NOTION OF THE BEDSIDE`. A metaphor naming the wrong end of its own interval, which matters because the two names are used interchangeably in policy and the intervals they denote are different lengths · the relation stated: the same phase carries a technical name and a metaphor, and the paper flags the metaphor as confusing because the phase stops at testing rather than at the bedside Quotation withheld because this record is not cleared for redistribution. publication to guideline —contrasted-with→ development to commercialization publication to guideline development to commercialization the relation itself is separately evidenced · strength of the relation: 0.95 · note: 'SOME STUDIES REPORT…' A late sub-step of the pipeline exceeding a long earlier stretch of it - which is only possible because the two figures come from different studies of different things, and is exactly why the paper says they cannot be added · the relation stated: some studies report the publication-to-guideline step as LONGER than others report the whole development-to-commercialization stretch Quotation withheld because this record is not cleared for redistribution. three estimates converge on 17 years, possibly because longer lags are hard to measure, which caps such estimates three estimates converge on 17 years, possibly because longer lags are hard to measure, which caps such estimates Quotation withheld because this record is not cleared for redistribution. AND THE CANONICAL FIGURE IS WRONG BY SIX YEARS: `Accounting for this changes their estimate ... from between 17 years to 23 years` AND THE CANONICAL FIGURE IS WRONG BY SIX YEARS: `Accounting for this changes their estimate ... from between 17 years to 23 years` Quotation withheld because this record is not cleared for redistribution. AND THE FIGURE SETS A NATIONAL RATE OF RETURN: `Varying the lag time from 10 to 25 years produced rates of return of 13% and 6%, respectively` AND THE FIGURE SETS A NATIONAL RATE OF RETURN: `Varying the lag time from 10 to 25 years produced rates of return of 13% and 6%, respectively` Quotation withheld because this record is not cleared for redistribution. AND THE STUDIES ARE NOT MEASURING ONE THING: `studies are almost never measuring the same thing, making valid aggregation and generalization difficult` - the proxy was picked for being `the most measureable` AND THE STUDIES ARE NOT MEASURING ONE THING: `studies are almost never measuring the same thing, making valid aggregation and generalization difficult` - the proxy was picked for being `the most measureable` Quotation withheld because this record is not cleared for redistribution. AND A MEAN BECAME A MEDIAN IN CITATION: `Decullier et al. report `MEAN` LAGS, and Dwan, in reporting Decullier et al., ... REFER TO `MEDIAN` lags. Neither reports distributions` AND A MEAN BECAME A MEDIAN IN CITATION: `Decullier et al. report `MEAN` LAGS, and Dwan, in reporting Decullier et al., ... REFER TO `MEDIAN` lags. Neither reports distributions` Quotation withheld because this record is not cleared for redistribution. AND ONLY ONE STUDY MEASURES WHAT THE NUMBER IS QUOTED FOR: `Only one study has `IMPLEMENTATION` into practice as its endpoint` AND ONLY ONE STUDY MEASURES WHAT THE NUMBER IS QUOTED FOR: `Only one study has `IMPLEMENTATION` into practice as its endpoint` Quotation withheld because this record is not cleared for redistribution. AND THE LAG RUNS BOTH WAYS: thrombolytics took `six years` to reach a recommendation, `whereas prophylactic lidocaine was used widely in practice for 25 years based on no evidence of effectiveness` AND THE LAG RUNS BOTH WAYS: thrombolytics took `six years` to reach a recommendation, `whereas prophylactic lidocaine was used widely in practice for 25 years based on no evidence of effectiveness` Quotation withheld because this record is not cleared for redistribution. the interval between any two markers a study chose - `Decullier et al. and Stern and Simes measure time between ethical approval and date of first publication; Grant et al. and HERG et al. look at PUBLICATION TO GUIDELINE; Sternitzke looked at ... `CHEMICAL SYNTHESIS` TO FDA APPROVAL; Ioannidis ... TRIAL REGISTRATION and several milestones to publication` time lag Quotation withheld because this record is not cleared for redistribution. `Type 1 ... refers to the conversion of knowledge from basic science research into a potential clinical product for testing on human subjects. Type 2 translation, `RESEARCH INTO PRACTICE`, tends to refer to the process of converting promising interventions in clinical research into healthcare practice` Type 1 translation Quotation withheld because this record is not cleared for redistribution. `Type 1 ... refers to the conversion of knowledge from basic science research into a potential clinical product for testing on human subjects. Type 2 translation, `RESEARCH INTO PRACTICE`, tends to refer to the process of converting promising interventions in clinical research into healthcare practice` Type 2 translation Quotation withheld because this record is not cleared for redistribution. `formed from adding together a number of single studies of different phases` the canonical figure, and its correction Quotation withheld because this record is not cleared for redistribution. the same thirty years of UK cardiovascular research, valued twice over on the choice of one undefined interval sensitivity of the reported rate of return to the assumed lag Quotation withheld because this record is not cleared for redistribution. `Ranges - or even interquartile ranges as large as 221 YEARS - are seldom reported` largest interquartile range reported in the reviewed literature Quotation withheld because this record is not cleared for redistribution. one method, one topic, and an eight-year range under a single average spread inside one clinical area Quotation withheld because this record is not cleared for redistribution. `clinical trials with null or negative results `ON AVERAGE` took `JUST OVER A YEAR LONGER` to be published ... a range of six to eight years ... compared with four to five years ... makes a potential difference of four years - half of the maximum lag`. The reported quantity is 1 year; the quantity it is drawn from spans 4 an average whose spread is four times the effect it reports Quotation withheld because this record is not cleared for redistribution. the two free parameters of every figure in this literature - `the start point and endpoint of the time lag measured in each study were mapped onto specific stages in the conceptual model`, which is the only way the 23 studies could be compared at all start point and endpoint Quotation withheld because this record is not cleared for redistribution.

The full text of this article is not redistributable, so it is not included here. The evidence sentences are withheld with it; the tag layer and its grounding status remain. The article itself is at https://doi.org/10.1258/jrsm.2011.110180.