Drafting the protocol

Drafting a clinical trial protocol starts from the decision the study has to inform and works through the rationale, objectives, population, design, control group and statistics, keeping those sections in agreement with one another. The protocol also has to organise information that unavoidably repeats itself, rest on a case report form that carries its observations into the database, protect the participants, and be worded so that no site can read an instruction two ways.

What a protocol is

The protocol is an action document, the recipe for performing a study, whereas the study report is an outcome document that describes the results of the study, so the two contain the same basic elements. ¹ It is also a stand-alone, comprehensive recipe or set of instructions for conducting the study, prepared by the sponsor and directed particularly at the investigators and study staff, the ethics committee and the regulatory authorities. ² Clinical trial protocols for trials evaluating pharmacological products are complex documents that describe the medical, ethical and regulatory foundations of the trial. ³ The database links the protocol to the study report. ⁴

The sections of a protocol

The sections of a typical study protocol are: ⁵

• introduction ⁵
• rationale, objectives ⁵
• design ⁵
• subjects: entry criteria, treatments ⁵
• study procedures, particularly the schedule of assessments ⁵
• consent, ethics ⁵
• adverse event management ⁵
• withdrawals ⁵
• data management ⁵
• abbreviations, references, appendices (methodology) ⁵

A protocol requires full consistency between the rationale, the proposed objectives, the endpoints and the statistical analysis, so that it answers the questions the study poses. ⁶ The population, the schedule of assessments, randomisation, blinding or both, the control group, the sample size and the regulatory and operational requirements must also be aligned with the methods proposed. ^{6, 7} A change to a single one of these may have a profound impact on all the others, as in a game of Jenga, which makes protocols particularly sensitive documents. ^{6, 8} In one case, a seemingly tiny oversight resulted in the study being redesigned. ⁸

Planning before drafting

Clinical studies drive both internal decision-making, on how to proceed with a development programme, and external decision-making, on regulatory approval and labelling. ¹ An optimal development programme depends on a high-quality and unambiguous target profile based on nonclinical, clinical, regulatory and marketing input, which describes the properties the new drug should have to meet a medical need and to be a marketing success, and which determines the clinical studies to be conducted. ¹ The target profile should not be a wish list with constantly changing goalposts: it should stay fixed unless new scientific, medical or marketing evidence necessitates a change, as when a new competitor might require a more stringent target profile or a competitor's withdrawal from the market might allow a relaxed one. ¹

Facing a typical protocol template with its many headings and subheadings is a daunting task, and its sections should not be rushed into in sequence. ⁹ It is worthwhile to stand back and look at the big picture and the key issues: the beginning of a protocol is not the beginning of its template. ⁹ Successful medical and scientific writing is always outcome-oriented, so the recommended start is to consider what the study is to achieve and then take these steps in order: ⁹

• Consider what the results of the study are to achieve. ⁹
• Based on that, define the key messages the outcome of the study should ideally carry. ⁹
• Based on those key messages, define the objectives of the study. ⁹
• Identify the study population. ⁹
• Create the schedule of assessments. ⁹

The rest follows logically, and it is advisable to complete the synopsis before writing the full protocol. ⁹

Most pharmaceutical companies and contract research organisations have internal protocol guidelines and templates, and many protocol templates are available on the internet. ⁴ A template for planning a study protocol sets out the following: ⁴

Planning element	Content
Guidelines	ICH 6, FDA, company-internal guidance, templates
Outcome	A well-performed, high-quality study giving a clear answer to the questions posed; appropriate internal decision-making; approval (ethics committee, regulatory authorities)
Target audience	Peers and company management; regulatory authorities; ethics committee (usually includes lay people); investigator(s) and staff; monitors
Key messages	Clear rationale and objective; appropriate and rational design; safety, efficacy and other drug features adequately addressed
Information sources	Literature; competitive intelligence; internal data (galenics, nonclinical, clinical)

⁴

Rationale

The rationale for doing a study is the reason for doing it, that is, the particular outcome it looks for. ¹⁰ It is usually based on the current state of knowledge and the requirements stipulated in the target profile, and it depends on the nature of the drug, the condition to be treated, the stage of development and the overall drug development strategy. ¹⁰ For a first study in humans with single ascending doses, the rationale argues that there are enough nonclinical data on safety and efficacy to justify a study in humans scientifically and ethically. ¹⁰ For a pivotal trial in a specific disease, it should argue that there are enough nonclinical and clinical safety and efficacy data to justify the dose selection and to expect a positive outcome. ¹⁰ Any unusual design features, particularly ones that deviate from standard practice and regulatory guidelines, should be mentioned upfront and justified, and it can be valuable to discuss such deviations with regulatory authorities before the study starts. ¹⁰

Clinical trials assess the efficacy, safety or pharmacological characteristics of medicinal products in human participants, and the protocol must present a rationale that identifies the primary purpose of the trial. ¹¹ At trial conception, the protocol development team should consider whether the rationale is clear, whether the trial is clinically relevant and feasible, and whether it would pose an unnecessary risk or burden on participants. ^{11, 12}

Objectives

Trial objectives are the actions proposed to fulfil the trial's rationale. ¹³ They should be conceived by the protocol development team and written by the medical writer according to the SMART principles. ¹³ They fall into three categories: a primary objective, typically one, that aims to answer the primary purpose of the trial directly; secondary objectives, other actions relevant to or indirectly associated with the rationale; and exploratory objectives, which generate hypotheses that can be confirmed in dedicated studies. ¹⁴ An example of a SMART objective is to assess the efficacy of paracetamol 1,000 mg by mouth against placebo in adult patients with fever of unknown origin (body temperature above 37.5 °C) 2 hours after administration. ¹² This objective is specific in identifying the action, the medicines, the population and the indication; measurable, since temperature is a measurable variable; achievable, in being feasible and not burdensome to the participant, which the trial medical expert is to confirm; clinically relevant, which the trial medical expert is also to confirm; and time-based, in specifying a 2-hour timeframe. ¹³

An objective should in the first instance be clear and achievable, and multiple objectives should not conflict with each other. ¹⁵ The objective is often closely linked to a target-profile requirement on galenics (the physical properties and formulation of the drug), pharmacokinetics, efficacy or safety. ¹⁵ Depending on the stage of development, it might be to obtain the data for deciding whether to continue or discontinue the project, a go/no-go decision, and, if the project continues, how to proceed. ¹⁵ The main objectives of a first study in humans could be: ¹⁵

• to show that a certain range of single doses, including the anticipated therapeutic dose with a reasonable safety margin beyond it, is safe and well tolerated in healthy volunteers ¹⁵
• to show that the drug has an appropriate pharmacokinetic profile, such as the once-daily dosing and absence of a food effect on absorption that a target profile might require, which could be a go/no-go decision ¹⁵
• to fine-tune the design of a multiple ascending-dose study, if repeated dosing will be required in the clinical setting ¹⁵
• to establish a concentration–effect relationship for an appropriate pharmacodynamic marker, a marker of drug effects, to identify potential therapeutic doses ¹⁵
• to decide whether the programme should continue and whether, based on the results, any plans should be modified ¹⁵

A pivotal study mainly aims to show that the drug is safe and effective in the target disease and population and that introducing it to the market would add value to the currently available treatment options. ¹⁵ It aims to obtain compelling evidence that will convince regulatory authorities that the drug has a favourable benefit–risk ratio, justifying approval of specified claims. ¹⁵ As in other trials, its objectives should be clearly defined, for example to demonstrate that a drug reduces 2-year mortality by 15% with a p-value below 0.05 compared with standard therapy. ¹⁵

Study population and eligibility

The protocol should briefly define the target population, the set of people throughout the world to whom the trial results may be generalised. ¹⁴ A detailed list of inclusion and exclusion criteria should follow, specifying demographic characteristics such as age, sex and body mass index; the medical indication under study; acceptable and prohibited comorbidities and concomitant medications; ethical requirements for participation such as informed consent; and exclusion criteria for anything that may bias the interpretation of results or pose an unnecessary risk to the participant. ^{14, 13}

A study in humans has three options for selecting its subjects: ¹⁶

• Healthy volunteers. Early studies emphasise safety, tolerability and pharmacokinetics, and patients, with their disease symptoms and co-medication, have a clearly greater potential for interindividual variability than healthy volunteers, whose homogeneity can be enhanced further by criteria such as age, gender and body mass index. Where the risk in healthy volunteers may be too high, as in testing chemotherapeutic agents, the first studies are done in cancer patients, who have at least some potential therapeutic benefit. The most relevant information on a drug comes from the patient population for which it is intended, so it is useful to match the healthy volunteers in a first clinical study as closely as possible to the target population in age and gender: for a drug to treat osteoporosis in postmenopausal women, a study in healthy men aged 20 to 40 makes no sense. ¹⁶
• Patients. Dose-finding studies and pivotal therapeutic studies should be done in patients, and as a general rule patients should be studied as early as possible in the development programme; after a single ascending-dose study in healthy volunteers, it might be more informative to test multiple doses in patients, particularly if a therapeutic response can be measured. It can also make sense to study the effect of a disease other than the intended indication, such as impaired renal or hepatic function. ¹⁶
• Healthy volunteers and patients. It often makes sense to combine the two in the same study. For a drug intended for rheumatoid arthritis, ascending doses might be given until the potential therapeutic dose is nearly reached and then continued in patients with rheumatoid arthritis, or a complete single ascending-dose study in healthy volunteers might add one cohort of patients at one of the doses to check whether the pharmacokinetic profiles and tolerability are similar in the two groups. ¹⁶

Deciding which subjects to select and which to exclude serves several purposes, such as reducing variability, optimising the chance of a response or minimising the risk of a particular adverse event, and the issues around inclusion and exclusion criteria should ideally be settled before pivotal trials begin. ¹⁷ What is studied is what reaches the label: if a pivotal trial studies only a subgroup of the disease population, the label will cover only that subpopulation, and a subpopulation that is excluded will be excluded from the label. ¹⁷ Many diseases of adults, such as type II diabetes, rheumatoid arthritis and hypertension, can also affect children, and because children are often excluded from studies, it should be remembered that there are regulatory incentives for conducting studies in paediatric populations. ¹⁷

Concepts in paediatric research design are unique and need special consideration from anyone first conducting clinical research with children as subjects or facing a first paediatric protocol. ¹⁸ The protocol allows the study design to be acceptable to the referring physician, the investigator, the legal guardian and the child. ¹⁸

Endpoints and estimands

Variables are the parameters to be measured, such as sociodemographic, clinical efficacy and safety, or laboratory and imaging parameters; they can be independent, potentially influencing health outcomes as treatment, age, sex or smoking history do, or dependent, being health outcomes such as vital signs or laboratory parameters. ¹⁹ An endpoint is an indicator measured in a patient or a biological sample to assess safety, efficacy or another trial objective. ¹⁹ The protocol development team is responsible for defining the endpoints, and the medical writer should ensure that they are clearly written. ¹³ Endpoints are classified in accordance with their objectives as primary, secondary or exploratory; the primary endpoint should provide the most clinically relevant and convincing evidence directly related to the primary objective, and its selection should reflect the accepted norms and standards of the field. ¹³

Each objective must be matched with at least one endpoint, to ensure that the right variables are captured and that every question the trial poses is addressed. ¹⁹ For the paracetamol objective, two endpoints can be proposed: the mean body temperature 2 hours after the test treatment, and the proportion of patients with a body temperature above 37.5 °C at that time. ^{19, 13} The better of the two should be selected with the protocol development team on clinical relevance and feasibility. ¹⁹ Endpoints should be measurable and concisely described in the protocol, while the details of how they are captured can be given in one or more dedicated protocol sections. ¹⁹

Estimands, introduced in ICH E9(R1), describe the treatment effect with consideration of specified post-randomisation events and of what the outcome would be under different conditions. ¹⁹ Four interrelated attributes are considered for this purpose: the population of participants targeted by the trial; the variable or endpoint; the post-randomisation events that happen to participants and may affect results, such as death, treatment discontinuation or use of rescue medication; and the population-level summary statistic for the endpoint, which is the basis for treatment comparisons. ¹⁹

Trial design and the schedule of assessments

A protocol must provide a clear trial design that meets the study objectives. ²⁰ Writers must understand trial design concepts, which are often complex, and be able to communicate them, and they must be ready to review and discuss aspects of the trial design with the protocol development team in order to write a clear and accurate protocol. ^{20, 3} The protocol defines the procedures by which a clinical trial is conducted, and writing it requires an assessment of all the concepts surrounding trial design. ^{21, 19} Medical writers must understand these concepts and work with the protocol development team to communicate them clearly, which protects the scientific integrity of the trial and the safety of the participants. ²¹

The study design includes the definition of the endpoints, also called study variables, which can be primary or secondary, the construction of the database, and the data analysis, statistical analyses included. ²² The construction of the database and the methods of data analysis should be defined in the protocol, and in pivotal studies the endpoints should be clearly defined. ²³ The design should facilitate data analysis and presentation in the study report. ²³ The trial design dictates, among other aspects, how participants are treated and followed up, the number of treatment groups and how data are collected. ²⁴

In the clinical programme of a new drug, most studies are randomised and double-blind, and they tend to be either placebo-controlled or compared with an approved drug, in parallel groups or with a crossover design. ²² Probably the simplest pivotal trial is a bioequivalence study comparing the plasma concentrations of a generic drug formulation with those of an approved drug's formulation. ²²

In single-group trials, variables are compared in the same participant before and a certain time after exposure to the test treatment; these are typically early-phase trials or trials with limited participant pools, and their results are usually preliminary because the treatment cannot be blinded. ²⁵

Comparative trials compare two or more test treatments. ^{26, 27} In a parallel design, participants keep the same treatment until trial completion or discontinuation, and variables are compared between groups at defined time points. ²⁷ In a crossover design, participants receive all treatments in a randomised sequence, with the treatments separated by a washout period without active treatment so that the previous treatment's effect does not bias the next, and because the same participant receives different treatments, within-participant analyses are possible. ²⁷ Crossover trials are useful for stable variables and for conditions that respond during treatment and return to near-baseline levels during washout. ²⁷ Factorial designs compare treatments given alone or in combination, creating a group for each possibility, four groups for two products; they assess possible interactions and complementary effects, but can be complex if a high number of groups is required. ^{27, 28}

Comparative trials typically include measures to minimise bias and make the groups comparable. ²⁷ Randomisation, in which participants are randomly assigned to the study treatments, is the standard method of obtaining treatment groups with similar baseline characteristics, such as age and sex; it increases the likelihood that baseline characteristics that could confound treatment effects are distributed equally among the groups. ²⁷ Randomisation lists of participant codes and treatment allocations are generated by different methods: ²⁷

• Simple randomisation, a single sequence of random assignments like tossing a coin, useful for large samples but liable to create unequal groups in smaller ones. ²⁷
• Block randomisation, in which blocks of equal size containing all possible treatment orders are defined and picked randomly; in one example, twelve participants are allocated to treatment A or B through three randomly selected blocks of four, each with a unique sequence. ²⁷
• Stratified randomisation, in which the sample is stratified by key baseline characteristics such as sex and age and participants are randomised so that these characteristics are distributed equally between the groups. ²⁷

Another allocation method is adaptive randomisation, in which the first participant is randomised and later participants are allocated non-randomly to minimise group imbalances in key baseline variables. ²⁷ Once the groups are comparable, it is important to confirm that participant follow-up and outcome assessments are not biased, and the key trial procedure for that is blinding, or masking, which ensures that the investigator, the participant or both are unaware of the treatment assigned. ²⁷ In double blinding neither knows, which is preferred because it avoids biased assessment of outcomes by either; in single blinding only one of them is aware of the treatment; and in an open-label trial both are. ²⁷ Double blinding requires indistinguishable test treatments in shape, colour, smell and taste, which is sometimes impossible, as when two active treatments with different formulations are compared; a double-dummy trial overcomes this by giving each treatment a matched placebo, so that each participant receives one active treatment and the placebo version of the comparator. ²⁷

The schedule of assessments is a key component of the protocol, reflecting the various observations and their timing. ²² All endpoints and variables must be captured in the schedule of assessments, which identifies which variables are measured at each trial visit and gives the site teams summarised procedural information for quick reference. ¹⁹ In a simple example, its rows are the activities and its columns the times at which they take place: ²²

Activity/measurement	Day 1 07 h00	Day 1 08 h00	Day 1 14 h00	Day 2 07 h00	Day 2 08 h00	Day 3 07 h00	Day 3 14 h00
Blood pressure measurement	✓			✓		✓
Electrocardiogram (ECG)	✓			✓		✓
PK* blood sample		✓	✓		✓		✓
Urinary albumin		✓
Blood glucose		✓
Drug administration	✓			✓		✓

* PK, pharmacokinetic. ²²

In pivotal studies the design is usually fixed, although it is possible, for example, to schedule an interim analysis to select one of two treatments to be continued. ²² Such variants need to be discussed with regulatory authorities and often incur a statistical penalty, such as a more stringent requirement for claiming statistical significance. ²² In early studies, particularly single and multiple ascending-dose studies, flexibility is allowed provided it is well described and motivated, and such flexibility can avoid the delays caused by protocol amendments. ²² Although allometric scaling based on animal data can predict human pharmacokinetics, the half-life might turn out shorter or longer than expected, so it is important to allow for changing the timing or number of blood samples in a typical single ascending-dose study. ²² The protocol can specify the number and timing of blood samples for pharmacokinetic analysis and the maximum volume of blood that can be collected for pharmacokinetic and safety assessments; if the half-life after the first dose then differs from the expected one, the timing or number of samples can be changed, provided the protocol made provision for it and the blood taken stays within the specified volume limits. ²²

Control groups

The choice of the control group in a comparative trial should consider the rationale and objectives along with regulatory, operational and ethical aspects. ²⁹ Placebos are formulations without any active pharmaceutical ingredient, used in double-blind trials, and they must be indistinguishable from the test treatment in packaging, labelling, size, shape, opacity, coatings, viscosity, colour, smell, flavour and route of administration. ³⁰ They help minimise bias and distinguish the real effects of the test treatment from noise such as normal physiology and spontaneous disease fluctuations, and from external factors that alter the participant's response, such as liver transaminase levels rising after a long stay in a clinical trial unit because of a strict diet, lack of exercise or other lifestyle changes. ³⁰ Placebo groups may not be appropriate in some settings, such as serious diseases with available therapeutic alternatives, as in oncology or infectious diseases. ^{30, 31}

Against an active comparator, several kinds of trial can be considered. ³² A superiority trial aims to show the test treatment superior to the comparator on the primary endpoint, which it is considered to be if the difference between them and its 95% confidence interval lie above 0. ³² A non-inferiority trial can be considered when a test treatment may not be more effective but offers other advantages, such as a better toxicity profile or more convenient administration; it evaluates whether the test treatment is as good as the reference or worse only to an acceptable degree, defined by a non-inferiority margin. ³² The test treatment is considered non-inferior if the difference and its 95% confidence interval for the primary endpoint lie above the negative margin, and if the interval lies entirely above 0 there is evidence of superiority at the two-sided 5% significance level. ³² A justification for the non-inferiority margin, typically provided by the trial statistician, should be included in the protocol. ³² A bioequivalence trial compares two products with the same active pharmaceutical ingredient, such as different formulations of the same product or a generic and its comparator; they are considered bioequivalent if they produce the same plasma concentration–time profiles, that is, if the 90% confidence intervals for the geometric mean ratios of the area under the curve from time 0 to the last measurable concentration and of the maximum plasma concentration lie entirely within 80% to 125%. ³²

Other potential control groups include: ³³

• No treatment, which cannot be blinded and can be unethical unless it is confirmed that participants will not be subjected to an unacceptable risk. ³³
• Standard of care, also called treatment as usual or routine care, which compares the test treatment with existing practice, though standard of care can vary between study sites and countries, hampering an objective definition. ³²
• Active placebo, a placebo that mimics the adverse effects of the test treatment, which can be useful when the risk of unblinding through characteristic adverse events is high. ³²

Statistical considerations

Statistician support is needed when developing the statistical sections of the protocol, and where applicable other experts should also be consulted, such as a pharmacokinetics and pharmacodynamics expert or a quality-of-life expert. ³⁴ The protocol should present a rationale for the calculated sample size, with the necessary statistical and clinical assumptions, based on the primary endpoint. ³⁴

Checking the key sections

Writing tips for the key sections of a protocol advise the following checks: ³⁵

• Introduction. Include a literature review covering the indication, the available therapeutic options and the test treatments, and present the scientific rationale of the trial, identifying its primary purpose, what it aims to achieve and its importance; the rationale is usually presented in this section. ^{36, 11}
• Objectives. Confirm that the objectives are consistent with the rationale, and write them as SMART objectives. ^{37, 20}
• Population. Define the target population, present a list of inclusion and exclusion criteria, and confirm that the criteria are consistent with the objectives. ³⁸
• Endpoints. Write short descriptions of the endpoints, which should be measurable; confirm consistency between the objectives and the endpoints, confirm that all variables are captured in the schedule of assessments, and confirm consistency between the defined estimands, objectives, endpoints and analyses. ^{39, 19}
• Trial design. Confirm that the design is clear and consistent with the objectives, endpoints and schedule of assessments, and that the methods of minimising bias are presented; where a comparative trial does not use randomisation, blinding or both, justify it. ⁴⁰
• Control groups. Confirm that the control group is clearly identified, justified and aligned with the design and objectives, and that all test treatments, placebo included, are clearly identified and characterised. ⁴¹
• Statistical considerations. Confirm consistency between the statistical considerations and the trial endpoints, and that all analysed variables are collected at the appropriate times in the schedule of assessments; summarise the statistical analysis methods for all endpoints without too much detail, referring to a separate statistical analysis plan for the full details. ^{35, 34}

The case report form

The observations and measurements the protocol requires are documented in the case report form, from which they are entered into the database for analysis. ⁴ The case report form is the link between the protocol and the database. ⁴² A good protocol and a successful study can be nullified if the data are not recorded correctly and appropriately, so the case report form is an important accompanying document to the protocol. ⁴ The case report form records all relevant information, such as demographics, study procedures, measurements and observations. ⁴² It should be user-friendly, with a format that makes review and analysis of the data easy, and as far as possible it should consist of numbers or choices. ⁴² Lengthy text entries by investigators should be avoided as much as possible, particularly in large studies, but in early tolerability studies they cannot be avoided, because descriptions of adverse events and other tolerability information are not standardised. ⁴² To facilitate monitoring, its information should as far as possible be verifiable from source documents, and the data should be recorded in a format ready for data entry and analysis: a time after dosing of 1 hour and 15 minutes is recorded as 1.25, not 1.15. ⁴²

Organising the information and managing redundancy

ICH good clinical practice recommends what a trial protocol should include, and TransCelerate's Common Protocol Template and the SPIRIT statement are efforts to harmonise the structure of protocols, but no official guidance addresses the content and structure of a protocol in as much detail as ICH E3 does for the clinical study report, which creates an additional challenge for writers. ^{3, 43, 44} ICH E6 lists content items to be included in a protocol, but the list is far from exhaustive, is considerably less detailed than ICH E3's guidance for the corresponding report section on the investigational plan, and gives no advice on how the various pieces of information are best organised and arranged. ⁴⁴ The SPIRIT initiative has published a far more extensive list of content items, again without any guidance on how best to structure the document. ⁴⁴

As a consequence, how protocols are organised and structured varies hugely across the pharmaceutical industry, in sharp contrast to the relative homogeneity of report appearance shaped by ICH E3, and the quality of protocols in how they are organised and present their information can vary greatly. ⁴⁴ One experienced writer holds that the proportion of poorly written documents is therefore considerably higher among protocols than among reports. ⁴⁴ ICH E3 advises that in each report section describing the design and conduct of the study it is particularly important to clarify features of the study that are not well described in the protocol. ⁴⁴ The same writer takes this advice as a sign that the guideline's originators shared that impression, and as making clear that a report's methods description is not written by blindly pasting the protocol text and adapting its tense. ⁴⁴

It is easier to prepare a well-organised, well-written, user-friendly report than an equally high-standard protocol, and there is more behind this than the availability or absence of formal guidance. ⁴⁴ An important reason is the protocol's need for built-in redundancies: it has to describe many separate but interlinked aspects, such as the way examinations are conducted, which variables are collected at them, how the collected variables lead to derived variables, how these are statistically analysed, which conclusions may be drawn from the results, and how all of this relates to the objectives of the trial. ⁴⁴ Presenting all these inter-related aspects clearly, separately and logically, without confusing and tiring the reviewer with a mass of repetitive information, is a true challenge, and carefully managing redundancies in the text is a key success factor for a user-friendly protocol. ⁴⁴

The protocol's audience is also more diverse than the report's. ⁴⁴ The report almost exclusively targets reviewers in the regulatory agencies, whereas the protocol targets the investigators and their staff, its most important audience, and also external bodies such as ethics committees, independent data review committees and regulatory agencies. ⁴⁴ Each sub-audience is primarily interested in selected topics only, and none of them is likely to read the protocol from cover to cover, so user-friendliness requires thoughtful structuring with meaningful headings that let each sub-audience quickly locate the information relevant to it. ⁴⁴

Many protocols are excessively long, complex, difficult to read and repetitive, probably because of fear of litigation, yet lay members of an ethics committee, for example, should be able to understand them, so particular effort should go into their summary sections. ¹ Crucial information lost in a long, complicated document that is difficult to read has the same impact as a document that is too brief and leaves out essential information. ¹

Presenting complex information in a well-structured manner, with carefully managed redundancies and organisation targeted at each user group, is a core competency of medical writing, and any protocol team is well advised to make use of this expertise. ⁴⁴ Its most efficient use goes a step further, because the principles of effective protocol organisation should already have been taken into account in the underlying protocol template and its associated guidance text. ⁴⁴ The absence of formal guidelines for protocol structure is an enormous opportunity to develop a general protocol template that lays the foundations for future quality protocols, and wherever medical writers see a chance to contribute to a protocol template, they should not hesitate to do so, as this can make a large difference to subsequent projects. ⁴⁴ Medical writers should get involved in setting up and maintaining the protocol template their organisation uses. ⁴⁵

The way a specific protocol presents its information can have a huge impact on the smoothness of trial conduct and thus on the quality of the data collected. ⁴⁴ The protocol also sets the stage for several documents further down the road, such as the statistical analysis plan, the report and the clinical summaries, so it should reflect very conscious and thoughtful decisions on the choice of terminology, the definition of terms and the way information is worded. ⁴⁴ The final protocol should be sound and consistent in all three. ⁴⁵

Participant protection

The first and foremost consideration in preparing a study protocol is the safety and well-being of the study subjects. ⁴ The protocol should be designed in line with the principles of the latest version of the Declaration of Helsinki, and it must say so. ⁴⁶ All subjects should be adequately informed, in written and verbal information, about all the relevant issues of the study, particularly the potential risks, and should give written consent before any study-related procedures begin. ^{46, 2}

Writers should look out for attempts, intentional or unintentional, to add study procedures or specimen collections that the study objectives do not warrant. ^{47, 48} Every study method and procedure included should contribute to fulfilling the study objectives, and exploratory tests should have a solid scientific rationale, particularly if they include an invasive procedure or an extra specimen collection. ⁴⁸ Nothing should be done to or taken from a subject without making sure that the subject understands and consents. ⁴⁸

Unambiguous wording

It is the writer's responsibility to make sure that both the expected and the prohibited study procedures are adequately defined in the protocol, and just one word can make a difference: in the phrase "chronic use of NSAIDs is prohibited", the simple undefined word chronic could lead to spurious data. ⁴⁷ Ambiguous or missing definitions provide an open invitation to deviate from or violate the protocol, cases of intentional protocol violation are well documented, and part of a regulatory writer's job is to do their best to close those windows of opportunity with a well-written protocol that leaves very little to chance or interpretation. ⁴⁷ What the protocol does not say can be as important as what it does say. ⁴⁸

A slash can mean either, or, or both, and if different readers, such as investigators reading instructions in a protocol, interpret it differently, they will record and produce different data. ⁴⁹ Ambiguous phrases such as within a time period should be avoided at all costs, yet one experienced writer has found a phrase equivalent to within a week of baseline in almost every protocol they have read, and it can mean a week before baseline, a week after it, or both. ⁴⁹ Many writers clarify it as within a week prior to baseline, which the writer finds understandable but dislikes, and within 4 weeks prior to the first study drug administration takes the writer a second or two to work out; in the week before baseline and in the 4 weeks before the first dose are simpler and more easily understood. ⁴⁹

The clumsy he/she construction can appear a dozen or more times on a single page of a protocol, and the same writer does their best to edit it out of documents. ⁵⁰ The writer holds that the best way to avoid it is the plural, as in investigators and their staff are to ensure … they should … and patients should be interviewed and their answers recorded, and sometimes a verb in the imperative gets around the problem, as in record the data directly on the CRF. ⁵⁰ Writers seem to shy away from the imperative in a protocol, possibly because they feel they are writing not only to the investigators but also to all the reviewers in their company and to health authorities and ethics committees; anyone very uncomfortable with it might write the imperative as a note, as in Note: Record any abnormal findings on page 14 of the CRF. ⁵⁰

Protocols are often strewn with cross-references, a great many of them to the schedule of assessments, but all readers of a protocol should know that the schedule of assessments is always provided and do not need to be sent to it at every mention of an assessment; one clear reference to it at the beginning of the procedures and variables section should suffice. ⁵¹