Writing a protocol for investigator-led research

A research protocol written by an investigator for an ethics committee or institutional review board, and perhaps for a funder, is the plan of the whole study: it states the question and the reasoning behind it, how the study will be run, overseen and reported, and what supports it, in enough detail that another investigator could carry it out. Writing it well starts with a research team and a common set of elements, runs through the summary, the methods and the supporting information, and ends with a review for the problems common in such protocols.

What the protocol is and what it serves

A research protocol describes the reasoning behind a research project and its process: what the investigators hope to accomplish and how they will get it done. ¹ It imposes a discipline, because the team has to anticipate every step of the project before the first subject is recruited. ¹ Its body more or less follows the IMRAD model of introduction, methods, results and discussion. ¹

A research proposal comes before the protocol and is brief. ¹ It describes the idea for a project with emphasis on the research question, and tells how the question might be answered and how the data will be obtained, with a focused literature search and perhaps the results of a pilot project. ¹ It helps the investigator crystallize their thoughts and is useful when recruiting coinvestigators. ¹ The protocol is the next stage, a detailed road map of how the project will be completed. ¹

The protocol will almost certainly be submitted to a human subjects committee, called an institutional review board (IRB) at many universities, and perhaps also to a funding agency if grant support is sought. ¹ Some IRBs have a research protocol template that the investigator is required to follow. ¹ The protocol itself may not go to a funder, which receives a grant proposal instead, and its greatest value to the investigator is in clarifying their thinking before the research begins. ¹ It deserves a lot of time and energy, because it is likely to be a key to IRB approval and grant funding and because it is the plan the team will follow in the months to come. ¹

A good research protocol gives enough detail that another investigator could do the study and arrive at comparable conclusions. ² The protocol offers a start-to-finish rationale for the proposed research: a documented introduction to the topic area, an account of current research and thought on the topic, the proposed project presented as a logical progression of research in the field, the intended methodology explaining how the project will be conducted, and the references that support statements of fact and document any statistics used to define the topic area. ^{3, 4} It accompanies the other forms submitted to the IRB, which either approves the project or refuses to let it move forward. ⁴

The protocol is often written quickly, even hastily, in the rush to receive IRB approval, and in the rush to complete it as an IRB requirement researchers often find themselves later having to recreate the documentation that originally supported their project. ⁴ A well-written protocol can serve as the basis for the IRB submission, represent three-fifths of the final manuscript submitted to a peer-reviewed journal, and serve as the springboard for poster and slide presentations. ⁴ If it answers every conceivable question on the IRB submission forms, filling out those forms can be a simple cut-and-paste process. ⁴ It lacks only the results and conclusions that come from the gathered data and their statistical analysis, and from the completed manuscript researchers can more easily abstract discussions for other ways of disseminating the findings. ⁴ The protocol should be the primary building block of the project, the document that gains its approval and the foundation for disseminating its findings. ⁴

The protocol can also be treated as the framework for the paper that will report the findings, and an early exercise is to prepare the format of the tables that will go in that report. ¹ For a study of the safety of live influenza virus vaccine in children with asthma, the report might have a table of subjects by age, gender, ethnicity and other characteristics, a table of adverse outcomes such as asthma attacks, bronchitis or pneumonia in patients who did and did not receive the vaccine, and a table of the incidence of influenza, with and without complications, in both groups. ¹ The tables are then already conceived and await only the data. ¹

The research team

Two steps come before writing a research protocol: identifying a researchable question and assembling the research team. ⁵ A research team needs specialized skills to plan and carry out the study, and where the person with the research question is the principal investigator, they can consider including one or more of these members: ⁵

• A clinical epidemiologist, who guides the planning of the study design so that the team uses the best method to answer the research question. ⁵
• A biostatistician, who can determine the sample sizes needed and the statistical tests best suited to the project; one investigator holds, while noting that some differ, that the biostatistician should be included from the very start. ⁵
• A senior investigator with more experience than the principal investigator, who serves as a respected arbitrator if differences arise on the team and can be a big help when writing the results. ⁵
• An aspiring investigator with less experience, who brings energy and learns from the experience. ⁵
• A project manager, who plans and manages the budgets, the time commitments of the investigators and all the records that will eventually be submitted to the funding agency. ⁵

The protocol's project management section states the specific role and duties of each team member, including the project manager. ⁶ Reaching consensus on these roles can prove vital to success, and misunderstandings about who is responsible for what can spell trouble, so this section should be the subject of informed agreement by everyone on the team. ⁶

The elements of a protocol

There is no single, universally accepted format for a research protocol. ⁷ An expanded list of its possible elements, adapted from the World Health Organization Recommended Format for a Research Protocol, is: ⁷

• Project title ⁷
• Primary investigator (PI) and coinvestigators and their institutions ⁷
• Project summary: a short overview to orient the reader ⁷
• Background and rationale: analogous to the Introduction in a research report ⁷
• Study goals and objectives: what the study hopes to accomplish ⁷
• Study design: what type of study is proposed ⁷
• Study description and methods: who the subjects will be and what will be done to them ⁷
• Safety considerations: what the risks to subjects are and what will be done about them ⁷
• Follow-up: what happens after the study is complete ⁷
• Data management: how the data obtained will be dealt with ⁷
• Quality assurance: how the study will assure that what is done reflects high standards ⁷
• Expected outcomes: what might be found, and why it might matter ⁷
• Dissemination of results: who will be told about the findings ⁷
• Duration of the project: how long each phase of the study is expected to take ⁷
• Problems that may occur: potential difficulties and possible solutions ⁷
• Project management: who is in charge of what ⁷
• Ethical considerations: what the ethical concerns might be and how subjects will be informed of them ⁷
• References: prior studies pertinent to the project ⁷
• Budget: what funding will be needed and how it will be used ⁷
• Informed consent forms: forms to be read and signed by research subjects ⁷
• Other support: any funding sources that should be disclosed ⁷
• Collaborations planned: whether the team will work with colleagues outside it ⁷
• Links to other projects: whether the proposal is the next step in a series of investigations ⁷
• Curriculum vitae of investigators ⁷
• Other research activity of investigators ⁷
• Clinical trial registration status ⁷
• Appendix ⁷

Not every protocol contains every item, and most have fewer topics or combine some of them; separate discussions of safety considerations, follow-up and quality assurance, for example, may not be relevant to a study. ⁷ Alternatively, a protocol can list every topic in this order and mark those not pertinent to the study as "NA". ⁷ The order can differ, but not with too much creativity; the format is, more or less, what IRB reviewers and grant funding agencies expect to see. ⁷ The list can also serve as a check that the protocol is complete before it leaves the investigator's desk. ⁷

A worked case runs through the sections that follow: a case-control study of whether patients who have gastric surgery for weight control are at risk of developing serum copper deficiency, which mirrors a study published in 2009. ^{7, 8}

The title

Most investigators turn to the project title once they have decided on the research topic, question or hypothesis. ⁹ An outcome-neutral title tells what will be studied, as in "The Incidence of Low Serum Copper Levels after Gastric Surgery". ⁹ "Hypocupremia after Gastric Surgery" presumes that the outcome will be lower copper levels; it may suit the final paper, but it is presumptive as the title of a protocol unless a question mark is added. ⁹

It helps to append a descriptor of the method used, such as randomized controlled trial, case-control, cross-sectional study or meta-analysis. ⁹ The worked case compares two groups, those who had weight-loss gastric surgery and those who did not, so its working title is "The Incidence of Low Serum Copper Levels after Gastric Surgery: A Case Control Study". ⁹ An acronym was considered for it and rejected, because the acronyms its candidate titles yielded, ILSCLAGS and HAGS, were not compelling. ⁹

The project summary and the research question

The project summary states the research question or hypothesis explicitly, so that anyone reading it has a clear idea of what the study hopes to learn. ¹⁰ A good research question is short, focused and unequivocal. ¹⁰ Most research questions describe something that can be answered with quantitative data, although there are an increasing number of reports of qualitative research on topics that defy quantification. ¹⁰ Careful wordsmithing of the question is vital, and everything hinges on its quality. ¹⁰

The PICO model is a useful way to state a research question, and in the worked case it reads: ¹⁰

• P, population or patients: patients who had weight-loss gastric surgery ¹⁰
• I, indicator or intervention: serum copper levels ¹⁰
• C, comparator or control: patients who did not have gastric surgery ¹⁰
• O, outcome: differences in serum copper levels ¹⁰

Many requests for proposals specify a maximum number of pages, and even where length is not prescribed the summary should be short, 250 words or less and no more than one page. ¹⁰ It starts with the study title and then briefly gives the rationale and objectives of the project, the research question, the methods and subjects, the duration of the study and the anticipated outcome. ¹⁰ Its language should make sense to a reader who is not fully familiar with the area of inquiry. ¹⁰ This page is arguably the most important of the protocol: it must give a crisp and memorable overview, and it must stand on its own, without sending the reader to search for items such as abbreviations found later in the document. ¹⁰

Background, objectives, design and expected outcomes

The background and rationale section, much like the Introduction of a research report, tells why the proposed research is pertinent in the context of current knowledge. ¹¹ It may cover: ¹¹

• the importance of the topic in the context of current knowledge ¹¹
• an overview of the approach to the question ¹¹
• any results already obtained that indicate that the question can be answered by that approach ¹¹
• how the study will advance medical practice and the health of humankind ¹¹

In the worked case, the study is pertinent because undetected and uncorrected hypocupremia can cause health problems including anemia, neutropenia, optic neuropathy, myopathy and myelopathy with a spastic gait. ¹¹ Current references can be listed in this section or in a separate references section. ¹¹

The World Health Organization's recommended format describes goals as broad statements of what the proposal hopes to accomplish, which create a setting for it. ¹² Specific objectives are statements of the research questions, and they should be simple, specific and stated in advance rather than after the research is done. ¹² Objectives should also be specific, measurable, achievable, relevant and time based, which gives SMART objectives. ¹² The objective of the worked case might be to determine whether patients who have had weight-loss gastric surgery are more likely later to have lower serum copper levels than people who have not. ¹²

The study design section describes the choice of design, such as a randomized trial, case-control study, cohort study or cross-sectional study, and why it is the best way to answer the research question. ⁸

The expected outcomes section says what might be found and why it might matter. ⁷ The ideal clinical study will have an impact on some segment of the population. ¹³ In the worked case, the results will either help alert physicians to a potential adverse outcome of an increasingly common procedure or serve as a caution to patients considering gastric surgery for weight control or for other reasons. ¹³

The study description and analysis

The study description is best written in the active voice, as in "We will draw blood samples", which is stronger prose, and in the future tense, because it describes something still to be done. ¹⁴ The people in the study are described as subjects, not patients, because the investigator is not typically in a traditional physician-patient relationship with them. ¹⁴

The section often begins with the subjects in the study and control groups: the sample size, who will be included and who excluded and why, with the possibility of exclusion bias in mind, and how the number of subjects in each group was decided. ¹⁴ A sample that is too small may not yield a statistically satisfying answer to the research question, and the study will be criticized as underpowered. ¹⁴ It then says how subjects will be found, such as through a newspaper advertisement, a search of existing hospital records or another method. ¹⁴ It describes each step of the study, beginning with enrollment and continuing through the last follow-up event, and, if subjects are to be randomized, the method to be used. ¹⁴

The section names the specific statistical tests that will be used to answer the research questions or test the hypotheses, and for an interventional study it may be useful to identify the dependent and independent variables. ¹⁴ The published study the worked case mirrors used independent t tests for continuous variables and Fisher exact tests for categorical variables. ¹⁴ One investigator holds that the diversity and complexity of the tests that might be used shows the importance of a team member with statistical expertise. ¹⁴

Safety, follow-up, data management and quality assurance

The safety considerations section is about risks to the subjects, which are especially pertinent when subjects will take drugs or undergo procedures. ¹⁵ It states the risks, including possible drug side effects, allergic reactions or procedural complications, then what will be done to reduce risk and what will happen if adverse events occur. ¹⁵ The systematic evaluation of research risks (SERR), a framework proposed by Rid and colleagues for assessing the risks of a research project, has four steps: ¹⁵

1. Identifying the potential harms associated with the project. ¹⁵
2. Categorizing the magnitude of those potential harms. ¹⁵
3. Quantifying or estimating the likelihood of each potential harm occurring. ¹⁵
4. Comparing the likelihood of each potential intervention-related harm with what might occur with a comparable activity. ¹⁵

In the worked case, the chief risks are judged to be those of obtaining blood samples; the surgery in the study cohort is completed before the study begins. ¹⁵

The follow-up section tells what, if anything, will happen to subjects after the study is completed. ¹⁶ In the worked case, patients found to have significant hypocupremia will be referred to their personal physicians for further evaluation and management. ¹⁶

The data management section covers the data the study accumulates, such as information about subjects, consent forms, laboratory reports and tables of results. ¹⁷ It says how the data will be coded and entered into computer files and how their accuracy and completeness will be assured, and, especially when the study involves human subjects, how their security will be maintained. ¹⁷ It identifies who will do these tasks, ideally the team's project manager. ¹⁷

The quality assurance section tells how the study will attend to the fundamentals of good clinical practice, which is especially important when human subjects are involved. ¹⁸ It says how best practice in the conduct of the study will be determined and whether there will be an independent oversight committee. ¹⁸

Duration, anticipated problems and dissemination

The duration section presents a timeline, which may help identify possible bottlenecks, such as finding fewer qualifying subjects than anticipated, and may be useful later when writing a grant application. ¹⁹ For the worked case, an 18-month project that is not very complicated, the timeline is: ^{19, 20}

Project activity	Months 1-3	Months 4-6	Months 7-9	Months 10-12	Months 13-15	Months 16-18
Identify study subjects	X
Identify control subjects	X
Assure matching study/control subjects	X
Recruit study and control subjects		X
Collect serum samples from study and control subjects			X
Analyze data; determine conclusions				X
Plan report of study					X
Report study at national meeting						X
Submit final report for publication						X

²⁰

In the worked case, the subjects are people with a history of partial gastric resection, and the controls, who have no such history, are matched to them for age, gender and concurrent illnesses. ²⁰

The problems section anticipates what could go wrong and what would be done to remedy it. ²¹ In the worked case, one potential problem is discovering that the surgery subjects have a much higher incidence of a concurrent disease, such as inflammatory bowel disease, than the controls. ²¹

The dissemination section says who will be told of the findings: whether a specific journal or clinical specialty is targeted as the eventual reader of the report, whether the results will be presented to a specialty group or at a scientific meeting, and whether study participants will be informed of the outcome. ²⁰ The World Health Organization's research protocol format advises discussing here who will take the lead in publication and who will be acknowledged in publications. ²⁰ In the worked case, the findings would be especially pertinent to surgeons, notably bariatric surgeons, and to generalists such as general internists and family physicians who see these patients after surgery, and whether to aim for a surgery or a generalist journal is a decision for the research team. ²⁰

Investigators, resources and ethics

The protocol clearly identifies the principal investigator and all coinvestigators, with titles, institutional affiliation, mailing addresses, telephone numbers, fax numbers and e-mail addresses, especially for the principal investigator. ²² The contact information matters because someone may want to get in touch about collaboration in this study or the next, or even to offer grant support. ²²

The references section lists the published reports that serve as background to the proposal. ²³

The budget section describes the funding needed to complete the project and justifies how every dollar will be spent. ²⁴ Where a request for proposals is involved, it has to be read carefully for which expenditures it allows, such as administrative assistance, student research workers or travel to present findings at a scientific meeting. ²⁴ The budget may be of only passing interest to the IRB, but it will be a key part of the eventual grant proposal. ²⁴ Its figures should be realistic and accurate: a budget with not nearly enough money to do the job is as bad as an underpowered study, and budget preparation is no time for guesswork. ²⁴

The other support section reveals any funding, such as a pharmaceutical company grant, that might not otherwise be apparent in the protocol. ²⁵ Where the project is done with investigators outside the research team, perhaps at other institutions, letters of agreement from them will eventually be needed. ²⁶ A project may be related to another research project, which is not uncommon because academics tend to develop career topics, and statements of such links may document leadership in the topic or raise the suspicion that the project is a "chain letter" project. ²⁷

The curriculum vitae of each member of the research team is appended, following any instructions in the request for proposals on format and number of pages. ²⁸ The principal investigator and coinvestigators should briefly describe their current research projects, the time devoted to each, how long each will last and its sources of funding. ²⁹ The protocol should also mention the status of the study's clinical trial registration. ³⁰

Ethical issues are a chief concern of the IRB, whose scope includes hospital chart reviews, questionnaire studies and anything that involves human bodies, tissues, fluids or information. ³¹ The IRB weighs questions such as these: ³¹

• whether the potential outcome of the study justifies any discomfort or pain from a study intervention, such as the small discomfort of a blood draw ³¹
• whether the potential benefit justifies the risks, such as adverse reactions in any study involving drugs, and hazards that might not be readily apparent, such as the long-term risk in studies involving radiographic imaging ³¹
• whether any payment to participants is enough to compensate for their time without being so large that people volunteer who should not ³¹
• whether subjects can withdraw from the study without physical risk and without the threat of having subsequent care withdrawn ³¹

All of these issues, and perhaps more, should be discussed in a carefully worded informed consent document appended to the protocol. ³¹ The appendix should contain a copy of the informed consent form, and it can also list the literature reviewed, if it is not given earlier, and other items that do not fit logically elsewhere in the outline. ³²

Common problems and the final review

Common mistakes in research planning and protocol writing include: ³³

• An inadequate literature review, a quick survey when a systematic review is needed; the protocol should not propose to do what has already been done. ³³
• A vague research question, often one too broad or unanswerable that needs to be focused. ³³
• A flawed research design, a study type that will not answer the question, which wastes much time and effort; most research failures can be traced to the early planning stages. ³³
• A project too ambitious for the team's research skills or for the institutional resources available to it. ³³
• Too many objectives and research questions; one investigator holds that a study can answer up to three related research questions and that more suggests a lack of focus, and recalls a study with 57 research questions whose investigators never wrote a publishable paper. ³³
• Too few subjects to answer the question; the IRB cannot be counted on to flag this problem, which poses no risk to subjects even though it wastes the time of willing volunteers. ³³
• An inappropriate plan for statistical analysis, in which the statistician plans to use the wrong tests on the data. ³³
• A poorly written consent form, which the IRB will send back for rewriting if it is not comprehensive and legally sound. ³³
• A budget inadequate for the project, or even too high. ³³
• Failure to reach consensus on all aspects of the protocol; the discipline of working towards group agreement on the research plan will prove invaluable if disagreements arise later in the team. ³³
• Underestimating the time needed to complete all the steps, including peer review and institutional review, to get the grant out the door on time. ³³

The best research protocols have these characteristics in common: ²

• An early overview, the project summary: a brief but crystal-clear synopsis that orients the reader to what comes later. ²
• A logical progression of ideas, following the outline to lead the reader point by point to a clear understanding of what will be done and how. ²
• Carefully constructed headings, since the protocol is really an essay and headings announce how its ideas are grouped. ²
• Clear prose, because a reader such as a member of the IRB may not be familiar with the field at all. ²
• Completeness, anticipating the questions a reader might have and trying to answer them. ²

Once the protocol is written, and rewritten a few times, it should be reviewed by a critical reader colleague, who helps identify problems with organization, syntax and grammar that may detract from the merits of the idea. ³⁴ If that colleague is not research savvy, especially in the area of investigation, an experienced investigator should review it next. ³⁴ The protocol is then ready for consideration by the human subjects committee or IRB, and with its approval in hand the investigator is ready to work on the application for grant funding. ³⁴