Covering the Annex V items and settling what they leave open
Annex V of the EU Clinical Trials Regulation fixes ten items that a lay summary of trial results has to cover. It says nothing about format, length, reading level or language, so each sponsor settles these, along with how to handle each item and which results to report.
The ten required items and their order
Annex V of the regulation requires the summary to contain information on ten elements: (1) clinical trial identification, including the title of the trial, the protocol number, the EU trial number and other identifiers; (2) the name and contact details of the sponsor; (3) general information about the trial, including where and when it was conducted, its main objectives and an explanation of the reasons for conducting it; (4) the population of subjects, including the number of subjects included in the member state concerned, in the Union and in third countries, the age group and gender breakdowns, and the inclusion and exclusion criteria; (5) the investigational medicinal products used; (6) a description of adverse reactions and their frequency; (7) the overall results of the trial; (8) comments on the outcome of the trial; (9) an indication of whether follow-up trials are foreseen; and (10) an indication of where additional information could be found. The order of the ten items is not binding. One writer highly recommends adapting it to improve readability, because some of the items are not very lay-friendly. According to the European Patients' Forum, the ten items were added by the European Council at the last stage of negotiations, without consultation with patient groups.
Annex V as guidance or as regulation
One reading holds that the list in Annex V can hardly be considered a guidance document, since its items are stated without explanatory instructions; each item needs interpretation, and many important aspects of a lay summary are missing. These limitations could be intentional, to give sponsors leeway in fulfilling the requirements, or could indicate that the thinking on the topic had not been finalised.
A second reading holds that Annex V is not guidance but part of a regulation, a distinction that matters in the EU, where regulations, directives and guidelines differ. Regulations are binding, inflexible legislative acts that must be applied in their entirety across the Union and leave no room for interpretation. Directives set out a goal that every member state must achieve, each through laws of its own. Guidelines are the most flexible: they represent an agency's current thinking on a topic in more detail, are not mandatory and can be deviated from with justification. On this reading, Annex V consists of ten mandatory items that legally must be included in the lay summary of any trial taking place in at least one member state. It is not intended as guidance or a template. It is sparse precisely because it is a regulation. Too much legislation would limit the flexibility such documentation needs.
What the annex leaves to the sponsor
The guidance the regulation itself provides is scant. Annex V gives no explanations or instructions on the format, structure or overall length of the summary, so each sponsor has to make reasonable assumptions and develop its own approach within the broad limits the regulation sets. Because the summary is meant for non-specialists, anything beyond two pages seems inappropriate. The regulation specifies no target reading level, often expressed as a Flesch–Kincaid grade or a reading-ease score, so each sponsor sets its own, and content and style can vary considerably between companies as a result. The regulation does not specify the language of the summary either.
Because detailed guidance on their content is lacking, the summaries are very likely to vary in quality and content with each sponsor's interpretation of the regulation. The EU's guidance document lists the elements a summary must include and provides a template, but the format, language level and design of summaries vary across the organisations that produce them. This gives sponsors, writers and designers both freedom and room for uncertainty.
Language
The key documents of a trial are usually written in English, which makes an English summary seem the straightforward choice. Although proficiency in English is high in certain member states and in some age or professional groups, a summary provided only in English would exclude many citizens of the Union, which has 23 other official languages. If the objective of increased transparency is taken seriously, the summary will have to be provided in all EU languages. The summary should also be translated into the language of the place where the study was conducted, so that the patients included and the local public are informed.
Audience and reading level
The recommendations of an expert group of stakeholders, developed after the regulation was published in 2014, name the general public as the primary audience of a lay summary and say that readers should not be assumed to have any prior knowledge of medical terminology, of clinical research or of the specific context of the study. A summary needs to be understandable to people with low literacy skills, and literacy in the general population is typically at level 2 to 3 of the five levels of the International Adult Literacy Survey, level 3 roughly corresponding to the level attained on completing secondary school. Under these recommendations, a summary should be written at a proficiency level of 2 to 3, roughly a sixth- to eighth-grade reading level.
Guidance in the United States and in the United Kingdom recommends writing information for patients at a sixth- to eighth-grade reading level. A plain language summary of trial results is generally written to the literacy level of a 12-year-old. Other advice sets the text at a grade 6 to 7 reading level. Once a target reading level is agreed, writers need to develop criteria to assess whether a text meets it, which may involve developing thesauruses of clinical research terms with their lay translations and using software tools to measure readability.
Wording, layout and numbers
Sentences should not be too long, and technical terms should be replaced by plain-language words or phrases. Where a long sentence cannot be avoided, as when a technical term has to be explained, an illustration may help. Plain replacements for technical terms include high blood sugar for hyperglycaemia, high blood pressure for hypertension, blood cells that fight infection for leucocytes, chest pain for angina, spread of cancer for metastasis, side effect for adverse drug reaction and swelling for inflammation. The active voice should be used, because it engages the reader and is most effective at communicating the information.
A lay summary should not be simplified so far that it loses the opportunity to educate patients or a lay audience about a disease area or an intervention. The writer needs to find the middle ground between that loss of educational value and a summary not simple enough for a patient to digest without the help of a qualified medical professional.
Most of the key guiding principles for developing lay materials can be grouped under three headings:
- Format: keep documents as short as possible, to improve accessibility and make readers more likely to read the whole document; leave sufficient white space rather than squeezing content onto pages; and graphics can break up the text and improve the document's aesthetic appeal.
- Word choice: choose words that make the text readable and understandable to people from the age of 12, and tools such as the Flesch-Kincaid test can be used to assess readability; short sentences can also aid comprehension.
- Numeracy: follow numeracy principles, for example presenting percentages rather than risk ratios or odds ratios, using whole numbers rather than decimals, and stating the numerator and denominator when reporting a percentage.
Following these principles helps produce good lay materials, but challenges remain. Perhaps the biggest is the tension between keeping documents short and the many words it often takes to explain a single medical term in lay language. Someone with a basic knowledge of oncology will recognise the word "metastatic", but someone with no medical knowledge may need several words to understand it, such as "a cancer tumour that has spread to other parts of the body", so the length of a document can increase substantially. Other challenges include avoiding generalisations that may be perceived as promotional, communicating necessarily complex terms such as lists of adverse events, and choosing graph or chart types that non-specialists can understand.
Avoiding promotion
Generalisations that could be perceived as promotional must be avoided, and communicating specific findings in lay terms will be challenging, as will keeping the document to a manageable size for readability.
Because a lay summary is not a regulatory document and is intended for patients, particular care is needed in stating conflicts of interest and in refraining from being promotional in both content and tone. At one company, more steps were needed to make sure that a summary was not promotional, so it took longer to finalise than any regulatory document.
The information in a lay summary needs to be presented objectively, and to keep lay summaries credible with the public anything promotional must be strictly avoided, a point the expert recommendations strongly emphasise. The text must not be too promotional, so as not to mislead the reader. A lay summary should be written in a neutral, nonpromotional tone, and in clinical research a neutral, factual tone is essential to convey findings accurately and to uphold public trust.
Handling each item
The requirements appear straightforward at first reading, but only items 2 and 10 can be readily implemented, and all the others need further specification. Sponsors, whether from academia or the pharmaceutical industry, have to make reasonable assumptions about implementing the requirements.
For items 1 and 2, the protocol number, the EU trial number and the sponsor's name and contact details are easily given, but the full protocol title is unlikely to help a lay audience. Protocol titles reflect the scientific and medical content of a trial and are intended for a medical audience, so they are often long and rich in technical terms and abbreviations. They typically carry dosages, study design features and descriptions of the patient population that a layperson usually does not easily understand. A shorter, simplified lay title is proposed instead, and the challenge is to make it succinct without being misleading or inaccurate.
Item 3 seems straightforward, since its only difficulty is a clear and brief explanation of the trial's rationale. Issues may arise where the rationale can be made understandable only with details of a disease, such as the current treatment options and unmet medical needs for patients with a particular severity of a condition, stage IV chronic obstructive lung disease for example. Medical information such as severity gradings is often not useful for laypersons but might nevertheless be needed. The study rationale should be explained in plain language, with background information about the disease and the drug studied.
For item 4, the numbers of subjects in the member state concerned, in the Union and in third countries had not commonly been reported but are easily given, and the age group and gender breakdowns are self-explanatory and pose no problem. The protocol's lists of inclusion and exclusion criteria, however, might be too long for a lay audience and may contain criteria relevant only to specialised readers, so it is suggested that they be limited to the most important, and a total of five or fewer might be desirable. It may be useful to mention criteria that a layperson can observe personally or is likely to be familiar with. Some technical terms used to define a patient population, such as a forced vital capacity below 50% of the predicted value, are not informative for a lay audience and could be omitted.
Item 5, the investigational medicinal products, is easily fulfilled, but a study early in clinical development may be able to give only the sponsor's internal compound code, which is completely uninformative. At later stages the international nonproprietary name becomes available, and giving both the code and the name then seems advisable, because it lets a reader link the summary to previous studies. For a marketed product with several trade names across the Union, all identifiers, the trade names included, should ideally be provided. Every comparator, placebo included, should be given the same information, because the regulation counts comparators as investigational medicinal products.
A lay summary should use words familiar to a lay audience, such as the brand names of medications. Some guidelines suggest listing both the generic and the brand name where possible, and where the brand name is not or cannot be used, the summary needs to signpost where the reader can find it. The same applies to medical terminology: listing an adverse event as high blood pressure, with hypertension in parentheses, describes it better than hypertension alone, and keeps the document understandable while helping to educate and improve health literacy.
Item 6, adverse reactions, requires several decisions. The regulation defines an adverse reaction as "a response to a medicinal product which is noxious and unintended", which is the concept of drug-related adverse events: events for which a causal relationship with the product has been established or cannot be ruled out. Both concepts might be challenging for a lay audience, so a sponsor could consider reporting adverse events whether or not they were deemed drug-related. Adverse events are commonly categorised with the Medical Dictionary for Regulatory Activities (MedDRA), many of whose terms are not easily understood by a layperson. Clinical study reports commonly present adverse events by preferred term and system organ class, and for the lay summary it has to be decided whether MedDRA terms are used at all and, if so, at which level of granularity. A translation of the terms into lay language may often be necessary, and since writers of patient information leaflets face the same problem, it is advisable to use the thesauruses they have developed. One view holds that the section should also cover deaths, serious adverse events and adverse events leading to discontinuation, although the annex does not mention them. The sponsor also has to decide whether clinical laboratory findings and vital signs are included.
Patients and the public tend to assume that all adverse events are caused by the study treatment, even when it is clearly explained that this is not the case, and this can give the impression that the treatment is less safe than it is. For that reason, one writer of plain-language summaries of clinical trials recommends the opposite of reporting adverse events whether or not they were deemed drug-related: writers should consider including only possibly related adverse events above a thoughtfully determined frequency threshold.
The scope of item 7, the overall results, is not clear. In most cases the overall results of a trial would include both efficacy and safety results. Safety results are largely covered by item 6, so providing only efficacy data in this section is suggested. The primary and the key secondary endpoints should always be reported, and where applicable data on endpoints related to quality of life can be included, as they might be of particular relevance for the patient. Protocols phrase endpoints in a detailed, technical way so that they can be evaluated statistically and medically. The writer's challenge is to rephrase the results for these endpoints so that the description is adequate and accessible for lay readers.
Item 8, comments on the outcome, is potentially the most problematic requirement. The word comments leaves a wide spectrum of interpretation, taking in making a statement, expressing an opinion and discussing the meaning. It can be understood as the EU regulators' wish for a section that presents the results at an aggregate level and gives conclusions. Summarising needs generalising terms, such as that the treatment showed efficacy or that no critical safety issues could be identified. Such terms are more comprehensive and hence open to misinterpretation, and they could be perceived as promotional, with legal issues as a possible result. That may be why most of the lay summaries available on the internet in November 2014 contained no summarising or concluding statement. The writer's task is a high-level summary that does not overstate the results, so the extent of the comments has to be considered carefully.
Item 9 can be met by a simple statement of whether additional trials are ongoing or planned. Item 10 can be met by a link to the sponsor's homepage, where further information such as the synopsis of the clinical study report may be available. Item 10 might become easier still to comply with, because under the regulation the EU clinical trial results database will enable hyperlinking of the summary of results, the layperson's summary, the protocol and the clinical study report of a trial and link to data from other trials of the same investigational medicinal product.
The items other than 2 and 10 each raise an issue with a proposed handling:
| Annex V item | Issue | Proposed handling |
|---|---|---|
| 1. Clinical trial identification, including the title of the trial | The trial title is usually written for a specialist audience in technical language. | Devise an additional lay title, shorter and simpler than the full trial title, and give it along with the full title. |
| 3. Main objectives and rationale of the trial | The objectives and rationale are usually described in the protocol for a specialist audience in technical language. | Give a simplified description that avoids specialist terms but keeps important medical terminology, such as disease stages, to maintain specificity. |
| 4. Inclusion and exclusion criteria | The protocol usually contains many inclusion and exclusion criteria written for a specialist audience in technical language. | Reduce the protocol's lists to the most important criteria, such as age, body mass index and indication-specific criteria. |
| 5. Investigational medicinal product | Depending on the stage of product development, different identifiers are usually available, such as the internal compound code, the international nonproprietary name or trade names, and trade names often differ among countries or regions. | Give the compound code for early trials and all available identifiers for later stages, give all available identifiers for comparators if feasible, and say whether a placebo was used. |
| 6. Description of adverse reactions and their frequency | The term adverse reactions refers by definition to drug-related adverse events, a concept that might not be appropriate especially in early drug development programmes, and it would be reasonable to report all adverse events, and adverse events and their frequencies can be reported at several levels of granularity, such as MedDRA preferred terms and system organ class. | To keep consistency with other sources, give adverse events by MedDRA preferred term as the default and at system organ class level only if useful, with an additional explanation of the medical terms in lay terms where needed; give the frequencies of all adverse events, any deaths and the adverse events leading to trial discontinuation; and give clinical laboratory data only if considered useful for the reader. |
| 7. Overall results of the trial | Because trials usually have several endpoints, primary, secondary and further, it is not clear whether the section should contain all efficacy and safety data or to what extent numerical data should be presented, and quality-of-life data might be of special interest to patients but are often not included as primary or secondary endpoints. | Focus on the primary and the key secondary endpoints, give numerical results so that the data can be compared with other resources, such as clinical trial reports, publications and trial results databases, and include quality-of-life data if relevant results were obtained in the trial. |
| 8. Comments on the outcome of the trial | The item might refer to the trial objective or the primary endpoint, but it is not clear what the sponsor is supposed to comment on; because the results are already reported under an earlier item, it might require qualitative statements, and qualitative summary statements are easily perceived as promotional. | Give a high-level factual statement on whether the trial fulfilled its objective. |
| 9. Follow-up trials | The regulation defines trial and study precisely but not follow-up trial; all planned trials of the same product, only trials already recruiting, or any planned future trials could be reported, but this might be perceived as advertising future studies by the same sponsor. | Report only true extension trials related to the trial in question. |
Statistical detail and consistency with other public sources
It can be assumed that a lay summary is not expected to include everything in the technical summary. What it can leave out raises the question whether a lay summary has to explain a mean plus or minus a standard deviation, or a 95% confidence interval, or whether it is sufficient to state that one treatment was better than the other, or was not.
A major issue for sponsors is keeping the summary consistent with their other channels for sharing trial data. To achieve full transparency, the data in a lay summary need to be linked to the more technical summaries on ClinicalTrials.gov and in the EU database. Describing results for laypersons has to allow for the level of numerical literacy in the general population, so details of the statistical analyses will likely have to be omitted, since odds ratios, P values and confidence intervals will not be informative for a lay audience with a sixth- to eighth-grade reading level. A lay reader might then find it difficult to relate the summary to the detailed data in the corresponding technical summaries. Sponsors have to find a balance between the possibly low numerical literacy of lay readers and consistency across the different public sources.
Choosing which endpoints to report
The guidance is clear on the primary endpoint and ambiguous beyond it. The expert group's recommendations say that the results section should describe the primary endpoints and additional safety data important to the overall results, and should reference the complete list of outcomes available in the technical results summary, including patient-relevant secondary endpoints. The European Commission's questions-and-answers document says that the summary should reflect at a minimum the primary endpoints and patient-relevant secondary endpoints. Neither defines a patient-relevant secondary endpoint. That leaves sponsors to decide which endpoints beyond the primary, if any, may be appropriate for inclusion, and how to select them.
A patient-relevant secondary endpoint can be defined as one prespecified as secondary in the protocol, with its statistical analysis described in the protocol or the statistical analysis plan, that represents something of particular importance or value to patients. Endpoints of interest to patients are those that reflect their experiences, perspectives, needs and priorities related to the symptoms of their condition and its natural history, the impact on their functioning and quality of life, or their experience with treatments. "Patient relevant" cannot be defined in any narrow or strict sense.
It is scientifically and ethically most appropriate to present the results of the primary endpoint as the key efficacy outcome. The primary endpoint is the assessment the study was designed for and powered to show differences in, in a confirmatory way, and for that reason the expert recommendations require primary endpoint data always to be shown. Secondary endpoints are evaluated in an exploratory way, and studies commonly evaluate large numbers of them. Within the limited space of a graphical abstract, including secondary or tertiary endpoints amounts to a selection and may lead to presenting only the results favourable to the drug. That is ethically and scientifically questionable, because studies are usually not powered to show differences in secondary endpoints. Multiple testing and the adjustment of significance levels also need to be considered before such results are reported. Where a secondary endpoint captures a patient-relevant observation, a solution is to include the data in the lay summary under a separate heading, with a statement that no firm conclusions can be drawn.
One writer of plain-language summaries of clinical trials holds that, beside the balance between scientific accuracy and clarity, a summary also has to balance complete transparency against thoughtful data selection. In that writer's view, it is important not to cherry-pick positive data, but also not to overwhelm the reader with copious amounts of data that might be misinterpreted.
In another writer's view, a summary can generally focus on the primary endpoint results. Some sponsors also include secondary endpoint results, which that writer finds tricky, because the summary should not cherry-pick only the good results. In the same writer's view, the protocol ideally already defines which endpoints are patient relevant, and the sponsor then decides upfront which information the summary will include.
Although the expert group's recommendation is to include only primary endpoint results, trial participants may want to know more. Including more than the primary endpoint, or a subset of secondary endpoints, presents the challenge of conveying the additional information concisely, in a non-biased, fair and balanced way that patients can easily understand. It is each sponsor's responsibility to interpret and apply the regulatory guidance and to determine which endpoints the summary includes and how they are presented. Selecting a subset of patient-relevant endpoints is subjective even with patient input, so to reduce the risk of perceived bias sponsors should predefine a systematic, transparent approach to selection that can be applied across the clinical trials of their organisation. A sponsor that includes more than the primary endpoint should select only patient-relevant secondary endpoints that have otherwise been publicly disclosed.
The considerations for deciding whether more than the primary endpoint is disclosed include whether the result has been publicly described elsewhere, whether the endpoint represents something of particular importance or value to patients, and whether giving the additional information creates a risk of misinterpretation. Secondary results could confuse the reader about the primary endpoint and be given more weight than is appropriate, and the importance of the secondary endpoints relative to the primary endpoint should not be overstated. The results should also be meaningful to patients. For sponsors that include patient-relevant secondary endpoints, emerging practice suggests a predefined, systematic approach to selecting and summarising them, so that summaries are consistently meaningful to patients, with fair, clear and balanced content:
- Summarise only patient-relevant secondary endpoints.
- To reduce the appearance of selection bias, decide and document which ones to include before database lock and before any knowledge of the trial results, including the results of interim analyses, and consider identifying them at protocol development.
- Do not include the results of post hoc analyses.
- Develop a selection process that can be applied consistently and transparently across trials and therapeutic areas.
- Support every endpoint in the summary with data in the clinical study report and with a description in the technical results summaries posted on public registries.
- Because of their exploratory nature, leave tertiary and exploratory endpoints, and secondary endpoints that are not patient relevant, out of the summary, since including them may cause misinterpretation and confuse readers without providing valuable information.
- Where such an endpoint is considered important to patients, the summary may instead acknowledge that it was assessed and comment on next steps or indicate where additional information may be found.
The sponsor is responsible for the final content, and the suggestion is to choose the patient-relevant secondary endpoints with input from patient-centric sources. Patient feedback is the most direct source, and it may be obtained through the sponsor's ongoing patient engagement activities, market research or other sources, and input from patient advocacy groups, carers and health care providers can supplement it. Clinical team members with experience in the disease area and direct involvement in protocol development and execution are a further source, and including at least one contributor with detailed knowledge of the trial's statistics to advise on summarising the data is worth considering. Clinical team input can be a valuable way of identifying potentially patient-relevant secondary endpoints, which patient-centric sources may then corroborate.
Patients can give input at stages that include the initial creation of the research objectives, the definition of the protocol's endpoints and their hierarchy, and the selection of the patient-relevant secondary endpoints for the summary. At the last of these stages, the selection can be done with qualitative and quantitative preference research. The patients selected are usually not participants in the trial but should be familiar with the concept of study endpoints and able to suggest how to describe them in plain language. They should be shown the primary and all secondary endpoints, to clarify the trial objectives and to make clear that other criteria will also be applied to the final selection. The endpoints, each listed with its attributes, such as efficacy, safety, duration of effect, duration of use, lifestyle aspects and other benefit-risk considerations, can then be presented to them for ranking, which shows their preference for each endpoint relative to the others.
Presenting the selected endpoints
How a selected endpoint is summarised should follow the data that are available, the statistical rigour applied to its analysis and how valuable its numerical results might be to a patient. The approaches fall into four tiers, which range from a presentation of numerical results with health-literate graphics to a description of the endpoints with no mention of results:
| Tier | Endpoint | Results available | Summarising the results | Example, for a pain score defined as a reduction from baseline in pain of at least 50% at Week X, or for plasma concentrations of Treatments A and B |
|---|---|---|---|---|
| Tier 1, quantitative summary | Statistically powered and predefined to confirm a hypothesis, or considered in determining the sample size of the study. | Statistical results, such as P values and 95% confidence intervals, were provided, and conclusions can be drawn from the data because of the statistical rigour applied to the analysis. | Consider presenting numerical results, such as an average or a range, with figures, tables or graphs that meet health literacy principles; comparative statements may be included if the results and the statistical rigour of the analysis support them. | "At Week X, more patients taking Treatment A (50 of 100 patients, 50%) had a lower pain score than patients taking Treatment B (35 of 100 patients, 35%)." The statistical rigour allows a comparison between the groups. |
| Tier 2, semi-quantitative summary | Not statistically powered to confirm a hypothesis but statistically analysed to investigate differences, such as between groups or from baseline. | Statistical results may have been provided, but conclusions should not be drawn from the data, as the study was not designed for this. | Consider presenting numerical results, but avoid graphical displays that could be read in isolation and misinterpreted without context, and avoid statements that suggest a conclusive finding. | "At Week X, 50 of 100 patients (50%) taking Treatment A and 35 of 100 patients (35%) taking Treatment B had lower pain scores." Where the results suggest a potential difference, a disclaimer such as "The results suggest there could be a difference between the 2 groups. Researchers cannot be certain whether the difference between the 2 groups was due to chance or due to the treatment. These results should be confirmed in another study."; where they suggest no potential difference, "The results suggest there may be no difference between the 2 groups. These results should be confirmed in another study." Numerical results are presented, but no comparison is made between the groups, since the study was not powered to confirm one, and a disclaimer about the limitations of the study is added. |
| Tier 3, qualitative summary | Not statistically powered or analysed, having perhaps been evaluated for its potential use as an outcome measure in future studies or representing a procedure that was particularly burdensome for patients, and so included for the patients' benefit. | Summary results may have been provided, although the data may be difficult for patients to understand without a lengthy explanation. | Present no results in any format, textual, numerical or graphical; describe the endpoint qualitatively and say why it was measured and, if applicable, what was learned from it. | "At Week X, both groups had lower pain scores.", with a disclaimer such as "This study was not designed to provide firm evidence of an effect on pain." or "This study was not designed to confirm if there was a true difference in pain reduction between patients receiving Treatment A or Treatment B. These results should be confirmed in another study."; or, instead of a summary of the results, "The researchers learned that measuring pain may be useful for studying drugs to treat X disease. These results should be confirmed in another study." No numerical results are presented; a textual description of what was observed in each treatment arm, or of what was learned, is given with a disclaimer about the limitations of the study. |
| Tier 4, aggregate description | Results would not be meaningful to patients, as with pharmacokinetic parameters or disease scores. | Statistical or descriptive results may have been provided, and the data may or may not be conclusive, depending on the statistical rigour applied; the data are likely not understandable to patients without training in the scientific discipline. | Present no results in any format; acknowledge in aggregate that other patient-relevant secondary endpoints were evaluated, rather than listing each one, and briefly describe in aggregate how the data were used. | "Researchers also measured the level of drug in patients' blood. Further details can be found at (source)." This acknowledges that pharmacokinetic endpoints were assessed but gives no results, and says where additional information can be found. |
An algorithm for choosing a tier asks three questions:
- Is summarising the results considered meaningful to patients? If not, consider Tier 4, with no summary of results and a reference to other sources of information.
- If it is, was the endpoint statistically powered to confirm a hypothesis? If so, consider Tier 1.
- If not, was the endpoint tested for significance, as with a P value or a confidence interval, but not designed to be confirmatory, as in a hypothesis-generating analysis or one without adjustment for multiplicity? If so, consider Tier 2, and if not, consider Tier 3.
The tiers are a guide to help sponsors think through the most appropriate way to describe the data, and their examples are illustrative only. The tiers are not absolute, and the same data may be represented in more than one way. In every case, the text should reflect the statistical rigour applied to the analysis and the data and conclusions in the clinical study report. Comparative statements should be avoided, particularly where no numerical results are presented. Endpoints in the results section should already have been explained or defined in plain language.
Patient-relevant secondary endpoints should be clearly separated from the primary endpoint, for example in separate, differently named sections, such as "main aim of the trial" and "other trial results". Their results should be presented in the context that the level of rigour applied to their statistical analysis calls for. Visual icons may help draw readers' attention to the main messages of the primary endpoint, and the language should say that these endpoints were not the main focus of the trial. The summary of these endpoints should be consistent with what has already been publicly disclosed, such as documents posted on public registries and scientific publications, and the suggestion is to keep it consistent with other patient and public communications about the trial, although presenting the information the same way in every case may not be possible.
Results of endpoints that were not statistically powered should come with a clear explanation that they are preliminary and non-confirmatory, may reflect chance findings and support no conclusions, with the statement adjusted to fit the data. To reduce the risk that readers overlook such disclaimers, emerging practice suggests keeping them concise and limiting their use throughout the summary. A statement and link showing where further information about the endpoint can be found are worth including. All information in the summary, patient-relevant secondary endpoints included, should be summarised in plain language that is simple, clear, fair, balanced and non-technical without diluting or changing the meaning or importance of the data. Health literacy and numeracy principles should be followed. In general, it is worth considering how meaningful numbers such as hospitalisations, blood transfusions or months of survival may be to a patient, against something more abstract, such as a decimal change in a clinical index, whose absolute changes may be less meaningful without a lengthy explanation or training in the scientific discipline.