9 Writing the guideline

During development of the guideline, the developer and committee will write:

  • the recommendations

  • recommendations for research

  • the rationale for the recommendations, and their likely impact on practice

  • the context for the guideline – such as the need for the guideline, or the reason for updating an existing guideline

  • summaries of evidence supporting shared decision-making, if there are preference-sensitive decision points in the guideline

  • information about changes to published recommendations (if the guideline is an update)

  • structured summaries of the committee's discussions

  • summaries of the evidence – with details of analysis and any modelling

  • the methods used for guideline development – highlighting the reasons for options taken, and any deviations from the methods and processes described in this manual.

For publication, the recommendations, recommendations for research, rationale for the recommendations and likely impact, context, and any summaries of evidence supporting shared decision-making are presented in a 'web version'. The committee's discussion and evidence are presented as PDFs in separate evidence reviews, along with the methods used during guideline development. Exactly how we organise and present the information is likely to change in future.

9.1 Interpreting the evidence to make recommendations

Assessment and interpretation of the evidence to inform guideline recommendations is at the heart of the work of the committee. Recommendations are developed using a range of scientific evidence (see section 4.4) and other evidence – such as expert testimony, views of stakeholders, people using services and practitioners, and the committee's discussions and debate (see chapter 3). Sometimes recommendations are made because of a legal duty or the consequences of not following a course of action are extremely serious. If there is a legal duty, the recommendation should refer to supporting documents and should build on the law or statutory guidance rather than simply repeating it.

The committee must use its judgement to decide what the evidence means in the context of the guideline referral and decide what recommendations can be made to practitioners, commissioners of services and others. The strength and quality of the evidence is assessed for both internal and external validity, but also requires interpretation. Evidence also needs to be assessed in light of any conceptual framework.

As soon as the committee has discussed the evidence, they should start drafting recommendations. They should decide what action to recommend and keep in mind which sectors (including which practitioners or commissioners within those sectors) should act on the recommendations. The record of the committee's discussion should explain clearly how they moved from the evidence to each recommendation, and document how any issues influenced their decision-making. In line with the GRADE principles on 'evidence to decisions', summaries of the discussions should describe the relative value placed on outcomes, benefits and harms, resource use, and the overall quality of the evidence, as well as other considerations (see Alonso-Coello et al. 2016, Schünemann 2016).

Findings from several evidence reviews may be integrated into a single summary of the committee's discussions if they relate to the same recommendation or group of recommendations.

For each group of recommendations, the committee should briefly explain their rationale for making the recommendations and record their views on any likely impact of the recommendations on practice or services.

Quality of the evidence

The committee is presented with GRADE tables, GRADE-CERQual tables or (if GRADE or GRADE-CERQual is not used) evidence statements. These describe the number, type and quality of the studies for each review question and provide an overall rating of confidence (high, moderate, low or very low) in estimates of effect for each outcome. The committee should agree that the reviews are a fair summary of the evidence and should discuss any uncertainty, including the presence, likely magnitude and direction of potential biases.

GRADE tables, GRADE-CERQual tables or the summary of the committee's discussion should include the committee's view of the directness (or applicability) of the evidence to people affected by the guideline and the setting.

Trade‑off between benefits and harms of an intervention

A key stage in moving from evidence to recommendations is weighing up the magnitude and importance of the benefits and harms of an intervention, and the potential for unintended consequences. This may be done qualitatively (for example, 'the evidence of a reduction in medicines errors in care homes outweighed a small increase in staff workload and resources') or quantitatively using a decision model.

The committee should assess the extent to which the available evidence is about efficacy (the extent to which an intervention produces a beneficial result under controlled experimental conditions), effectiveness (the extent to which a specific intervention, when used under 'real world' circumstances, does what it is intended to do) or both. Often the distinction between the 2 is not clear; this may be of particular importance in reports of complex interventions, because these are often evaluated only in pragmatic studies.

If several possible interventions are being considered, it is useful to include details of the committee's discussion of the relative position of interventions within a pathway of care or service model.

The committee should also assess whether, when recommending an intervention, they are able to recommend stopping other interventions because they have been superseded by the new one.

The committee should also assess the extent to which the recommendations may impact on health inequalities. This needs to be made clear, regardless of whether the recommendation is aimed at the whole population, specific subgroups or a combination of both.

Trade‑off between economic considerations and resource use

As noted in section 7.7, the committee should discuss cost effectiveness at the same time as effectiveness when formulating recommendations.

The guideline should include an explanation of how the implications of costs, resource use and economic considerations were taken into account in determining the cost effectiveness of an intervention. This may be informal, or may be more formal and include economic modelling (see chapter 7).

If several possible interventions are being considered, it is useful to include the committee's discussion of the relative position of an intervention in a care pathway or service model, based on cost effectiveness.

Use of indirect evidence

Sometimes when there is no evidence directly relevant to a specific population or setting, indirect evidence from other populations or settings may be considered. For example, a review of systems for managing medicines in care homes for people with dementia may identify good practice that is relevant in other care home settings. The use of indirect evidence must be considered carefully by the committee, with explicit consideration of the features of the condition or interventions that allow extrapolation to a different context or population. This also applies when extrapolating findings from evidence in different care settings (for example, between primary and secondary care). The committee should consider and document any similarities in case mix, staffing, facilities and processes, and any limitations.

Availability of evidence to support implementation (including evidence from practice)

The committee should also judge to what extent it will be feasible to put the recommendations into practice. They can use expert oral or written testimony, the experience of committee members or results from other approaches (see chapter 10 and appendix B) if these have been used. They may also be able to draw on qualitative studies or other forms of evidence relating to organisational and political processes where appropriate.

The committee should consider the extent of change in practice that will be needed to implement a recommendation, staff training needs, policy levers and funding streams, and the possible need for carefully controlled implementation with, for example, training programmes. This should be documented in the guideline and in any resources to support implementation.

Size of effect and potential impact on population health

The committee should consider whether it is possible to anticipate effect sizes at the population level, if this is appropriate for the topic. If this is the case, it is important to consider effect sizes along the whole causal chain, not just at the endpoints.

Wider basis for making recommendations

The committee should take into account a range of issues (including any ethical issues, social value judgements, equity considerations and inequalities in outcomes) and policy imperatives, as well as equality legislation (see chapter 1) to ensure that the guideline recommendations are ethical, practical and specific. There are no hard-and-fast rules or mechanisms for doing this: the committee should make conscious and explicit use of its members' skills and expertise. All evidence needs interpretation: evidence alone cannot determine the content of a recommendation. The development of evidence-based recommendations involves inductive or deductive reasoning:

  • inductive because it involves using what is known (the evidence) while accepting that there is uncertainty about what is reasonably expected to happen as a consequence of implementing a recommendation

  • deductive when it is drawn from theory or methodological principles that are generally held to be true so that any logical conclusion will be true.

NICE's principles on social value judgements explicitly acknowledge that non-scientific values are brought to bear, and all of NICE's advisory committees are encouraged to take account of (and to make explicit) the value judgements they make. The committee may also draw on the principles outlined in the report on ethical issues in public health by the Nuffield Council on Bioethics when making its judgements.

Conceptual framework or logic model

When the committee is developing its recommendations, it should consider any conceptual frameworks or logic models that have been used to inform the guideline because these may help to identify any practical issues involved with a recommendation that will change practice.

Equality considerations

The guideline should also document how the committee's responsibilities under equality legislation and NICE's equality policy have been discharged in reaching the recommendations (see section 1.4), and how the recommendations address areas highlighted in the equality impact assessment. The committee needs to consider whether:

  • the evidence review has addressed areas identified in the scope as needing specific attention with regard to equality issues

  • criteria for access to an intervention might be discriminatory (for example, through membership of a particular group, or by using an assessment tool that might discriminate unlawfully)

  • any groups of people might find it impossible or unreasonably difficult to receive or access an intervention

  • recommendations can be formulated to advance equality (for example, by making access more likely for certain groups, or by tailoring the intervention to specific groups).

Insufficient evidence

If evidence of efficacy or effectiveness for an intervention is either lacking or too low quality for firm conclusions to be reached, the committee has several options. It may:

  • make a 'consider' recommendation based on the limited evidence (see section 9.2)

  • decide not to make a recommendation, and make a recommendation for research (see section 9.4)

  • recommend that the intervention is used only in the context of research

  • recommend not to offer the intervention.

Factors the committee should consider before making a 'do not offer' recommendation include:

  • The intervention should have no reasonable prospect of providing cost-effective benefits to people using services.

  • Stopping the intervention is not likely to cause harm for people currently receiving it.

Factors the committee should consider before issuing 'only in research' recommendations include:

  • The necessary research can realistically be set up or is already planned, or people using services are already being recruited for a study.

  • The intervention should have a reasonable prospect of providing cost-effective benefits to people using services.

  • There is a real prospect that the research will inform future NICE guidelines.

The principles for wording recommendations described in section 9.2 should be used, reflecting the strength of the evidence. The committee's discussion and the rationale should be documented fully.

Strength of recommendations

The concept of the 'strength' of a recommendation (Guyatt et al. 2003) is key to translating evidence into recommendations. This takes into account the quality of the evidence but is conceptually different.

Some recommendations are 'strong' in that the committee believes that the vast majority of practitioners or commissioners and people using services would choose a particular intervention if they considered the evidence in the same way as the committee. This is generally the case if the benefits clearly outweigh the harms for most people and the intervention is likely to be cost effective. Similarly, if the committee believes that the vast majority of practitioners or commissioners and people using services would not choose a particular intervention if they considered the evidence in the same way as the committee, a negative recommendation can be made (that is, 'Do not offer'). This is generally the case if the harms clearly outweigh the benefits for most people, or the intervention is not likely to be cost effective.

However, there is often a closer balance between benefits and harms, and some people would not choose an intervention whereas others would. This may happen, for example, if some people are particularly likely to benefit and others are not, or people have different preferences and values. In these circumstances, the recommendation is generally weaker, although it may be possible to make stronger recommendations for specific groups of people. The committee should also discuss making negative recommendations in such situations if they believe most people are particularly likely to experience no benefit or experience harm, but that there may be a benefit for some. Be as specific as possible about the circumstances (for example, 'Do not offer…, unless…') or population.

NICE reflects the strength of the recommendation in the wording (see section 9.2). NICE uses 'offer' (or words such as 'measure', 'advise', or 'refer') to reflect a strong recommendation, usually where there is clear evidence of benefit. NICE uses 'consider' to reflect a recommendation for which the evidence of benefit is less certain.

There might be little evidence of differences in cost effectiveness between interventions. However, interventions that are not considered cost effective should not usually be offered. This is because that course of action has been judged to be an inefficient use of limited resources, with the population as a whole gaining fewer benefits than from alternative (cost-effective) courses of action. In some cases though, considerations about equity may also affect the decision whether to recommend the intervention (see section 7.6).

The committee's view of the strength of a recommendation should be clear from its discussions, and reported in the guideline.

In most cases the committee reaches decisions through a process of informal consensus, but sometimes formal voting procedures are used. The proceedings should be recorded and a clear statement made about the factors considered and the methods used to achieve consensus. This ensures that the process is as transparent as possible. A structured summary of the generic and specific issues considered and the key deliberations should be included in the guideline.

Principles of person-centred care

All NICE guidelines advocate the principles of person-centred care: people using services and the wider public should be informed of their options and be involved in decisions about their care.

There are 3 NICE guidelines specifically on the experience of people using services: patient experience in adult NHS services, service user experience in adult mental health and people's experience in adult social care services. These include general recommendations on the principles of person-centred care, such as communication and providing information, which should not be restated in topic-specific guidelines.

However, recommendations on person-centred care can be included in topic-specific guidelines if there is evidence of specific need for the topic.

9.2 Wording the recommendations

This section gives the principles of writing recommendations. For more on NICE style, see the style guide and writing for NICE.

Focus on the action and what readers need to know

Recommendations should be clear about what needs to be done, without the reader having to read the committee's discussion of the evidence. When writing recommendations, keep in mind a reader asking, 'What does this mean for me?'

Include only one action per recommendation or bullet point, and use direct instructions wherever possible because these are easier to follow. Recommendations often start with a verb such as 'offer' (or 'do not'), 'consider', 'measure', 'advise', 'discuss', 'ask about'.

Exceptions to this principle include:

  • Recommendations that specify who should take action, or cover service organisation. For example, 'A multidisciplinary team should provide care'.

  • Recommendations that use 'must' or 'must not' (because of a legal duty or a very serious consequence of not following the recommendation).

  • Recommendations to take different actions in different circumstances or for different populations. For example, it is often clearer to start with details of the population covered by the recommendation.

Think carefully about how much detail to include. Recommendations should be specific about interventions. But this needs to be balanced against the need to be clear and concise. Including a lot of detail can reduce the impact and make it harder for users to identify the most important actions.

Generally, avoid making recommendations about basic good practice unless there is evidence of poor practice or variation in practice.

Be clear about who should be offered an intervention, but don't spell this out in every recommendation if it is obvious from the context. Similarly, readers need to know which sections of the guideline they need to act on. Clear subheadings can help here, for example, by spelling out the setting where the recommendations apply (for example, 'first treatment', 'in primary care' or 'in care homes').

The content and measurability of any related NICE quality standard will be affected by the clarity and precision of recommendation wording. If possible, be clear when interventions should take place, and include the reasoning in the rationale and discussion sections.

Reflect the strength of the recommendation

The 'strength' of a recommendation (see section 9.1) should be reflected in the consistent wording of recommendations within and across guidelines.

In recommendations on activities or interventions that should (or should not) be offered, use directive language such as 'offer' (or 'do not offer'), 'advise', or 'ask about'. In keeping with the principles of shared decision-making, people may choose whether or not to accept what they are offered or advised.

If there is a closer balance between benefits and harms (activities or interventions that could be used), use 'consider'.

If there is a legal duty to apply a recommendation, or the consequences of not following a recommendation are extremely serious, the recommendation should use 'must' or 'must not' and be worded in the passive voice.

Box 9.1 Examples of recommendations made with 3 different levels of certainty

Recommendations for activities or interventions that must or must not be used

  • Ultra‑rapid detoxification under general anaesthesia or heavy sedation (where the airway needs to be supported) must not be used. This is because of the risk of serious adverse events, including death.

  • Patient group directions must be authorised only by an appropriate authorising body in line with legislation.

Recommendations for activities or interventions that should or should not be offered

  • Offer a trial of supervised pelvic floor muscle training of at least 3 months' duration as a first treatment to women with stress or mixed urinary incontinence.

  • If a smoker's attempt to quit is unsuccessful using NRT, varenicline or bupropion, do not offer a repeat prescription within 6 months, unless special circumstances have hampered the person's initial attempt to stop smoking, when it may be reasonable to try again sooner.

  • Record the person's blood pressure every 6 months.

  • Do not offer lamotrigine to treat mania.

  • Do not offer tigecycline to treat diabetic foot infections, unless other antibiotics are unsuitable.

Recommendations for activities or interventions that could be offered

  • Consider pelvic MRI to assess the extent of deep endometriosis involving the bowel, bladder or ureter.

  • Consider referring people to other behavioural support services within the local health and care network (for example, to voluntary or community services) for interventions that are not available in the pharmacy.

Use 'person-centred', precise, concise, clear English

Person-centred language

Use language that is person centred, and acknowledges the experience of people who are directly affected by the recommendations (and family members, carers or advocates), and their role in decision-making. Generally use verbs such as 'offer' and 'discuss' in recommendations, rather than 'prescribe' or 'give'. Use 'people' or 'people with [condition]' (or 'people using services', or 'patients') rather than 'individuals', 'service users', 'cases' or 'subjects'.

Clarity and consistency

Using clear, consistent wording is an important part of NICE's approach to presenting guidelines and other products, and is in line with gov.uk style.

Follow the principles of effective writing as described in writing for NICE and NICE's approach to consistency in language and terms across guidelines and other products. More information is available in the NICE style guide and our guide to writing recommendations.

Use bullet lists and tables if they help to make recommendations easier to follow. Make sure there is a clear link back to the evidence and the committee's discussion for all information in a table.

Recommendations on medicines, including off‑label use of licensed medicines

Do not give dosages routinely

Readers are expected to refer to a medicine's summary of product characteristics (SPC) for details of dosages for licensed indications. Include dosage information only if there is evidence that a particular medicine is often prescribed at the wrong dosage, or there is clear evidence about the effectiveness of different dose levels. Information about dosage can be found in the BNF or BNF for Children as well as the SPC. SPCs can be found in the Electronic Medicines Compendium.

Dosage information for off‑label use of a licensed medicine will usually not be included in the SPC (see below). If off‑label use is being recommended, check whether there is any relevant dosage information in the BNF or BNF for Children for the particular population or indication it is being recommended for. If there is none, NICE staff with responsibility for guideline quality assurance will liaise with the BNF.

Off‑label use of licensed medicines

Using a UK licensed medicine outside the terms of its marketing authorisation is classed as off‑label use.

Recommendations are usually about the uses of medicines for which the regulatory authority has granted a marketing authorisation, either in the UK or under the European centralised authorisation procedure (often referred to as the licensed indications). However, there are clinical situations in which the off‑label use of a medicine may be judged by the prescriber to be in the best clinical interests of the patient. For example, off‑label use may be recommended if the clinical need cannot be met by a licensed product and there is sufficient evidence and/or experience of using the medicine to demonstrate its safety and efficacy to support this.

Off‑label prescribing is particularly common in pregnant women and in children and young people because these groups have often been excluded from clinical trials during medicine development. When prescribing a medicine off label, the prescriber should follow relevant professional guidance (for example, the General Medical Council's Good practice in prescribing medicines – guidance for doctors) and make a clinical judgement, taking full responsibility for the decision for the patient under his or her direct care. In addition, the patient (or those with authority to give consent on their behalf) should be made fully aware of these factors and provide informed consent, which should be documented by the prescriber.

The guideline developer and committee should check recommended uses of a medicine against the licensed indications listed in the SPC. If the medicine does not have a UK marketing authorisation for the use being recommended (off‑label use), add the following standard footnote to the recommendation:

At the time of publication ([month year]), [name of drug or drug class] did not have a UK marketing authorisation for this indication. The prescriber should follow relevant professional guidance, taking full responsibility for the decision. Informed consent should be obtained and documented. See the General Medical Council's Good practice in prescribing medicines – guidance for doctors for further information.

Recommendations on medical devices, including off‑label use

Using a CE‑marked device outside its instructions for use is classed as off‑label use.

Recommendations are usually about the use of devices within the terms of the instructions for their use. However, there are clinical situations in which the off‑label use of a device may be judged by the prescriber and patient to be in the best interests of the patient. For example, off‑label use may be recommended when using a device outside the time period specified in the instructions for use.

The Medicines and Healthcare products Regulatory Agency (MHRA) has issued guidance on the off‑label use of medical devices. The guideline developer and committee should check the instructions for use of a device and if the device is being recommended for off‑label use, include this information in this standard footnote to the recommendation:

At the time of publication (Month year), not all devices with a CE mark for [intended purpose] are intended by the manufacturer for use as recommended here. The healthcare professional should follow relevant professional guidance, taking full responsibility for the decision. Informed consent should be obtained and documented. MHRA's advice remains to use CE‑marked devices for their intended purpose where possible. See guidance on off-label use of a medical device for more information.

9.3 Supporting shared decision-making

Identify preference-sensitive decision points

Guidelines should include information to support shared decision-making between people and their health or social care practitioners (see NICE's recommendations on supporting people to make decisions about their care in patient experience in adult NHS services, service user experience in adult mental health and people's experience in adult social care services). The committee should identify any recommendations where someone's values and preferences are likely to be particularly important in their decision about the best course of action for them.

When identifying these recommendations committees should take account of people's autonomy and right to make individual decisions, in line with NICE's social value judgements. They should, however, ensure that their recommendations represent an effective use of health and social care resources and so would not disadvantage other people using the services (see section 4.3 of NICE's social value judgements).

These 'preference-sensitive decision points' occur when the committee recommends 2 or more options for investigation, treatment or care that deliver similar outcomes but:

  • they have different types of harms and benefits which people may value differently or

  • the likelihood of the harms or benefits may differ or

  • the practicalities of the options are different (for example, the choice is between medicine and surgery, or the requirements for monitoring differ) or

  • some people may consider the overall risks of harms for any of the options outweigh the overall benefits compared with no treatment.

Alternatively, a preference-sensitive decision point may occur if the choice between an investigation, treatment or care option and 'no treatment' is finely balanced.

When to identify preference-sensitive decision points

In some cases, preference-sensitive decision points might be apparent at the scoping stage of a guideline. In other cases, the committee will identify them when reviewing the evidence. This is the same evidence used to make the recommendations, and no additional evidence searches or syntheses are needed.

For updates, the committee only needs to identify decision points in the areas of the guideline being updated.

Summarise information to support decisions

When a preference-sensitive decision point is identified, the committee should create a summary of the evidence that will appear in the guideline. This is to make it easy for professionals and practitioners to compare the options and discuss them with the person.

This summary should set out information about the options clearly and simply. It should usually include the option of no treatment or investigation. This could include brief information, for each option, about:

  • efficacy or effectiveness: how well something is likely to work for particular outcomes or circumstances and for particular subgroups of people

  • safety: the risk of short- or long-term harms, and any contraindications

  • practical factors relevant to people using services: people's experience of treatment or care – for example, ease of use (route of administration for medicines), monitoring requirements, side effects, or impact on quality of life

  • quality and certainty of the evidence: an overall judgement about the quality of the underlying evidence for each option.

In some cases, NICE will develop additional decision aids (see chapter 12).

9.4 Formulating research recommendations

The committee is likely to identify areas in which there are uncertainties or in which robust evidence is lacking. NICE has published a Research recommendations process and methods guide, which details the approach to be used to identify key uncertainties and associated recommendations for research.

For guidelines where there could be many hundreds of uncertainties, it will not be possible to document every uncertainty in detail. Similarly, although committees could write research recommendations for dealing with each uncertainty, this is not likely to be feasible. The committee should select up to 5 key recommendations for research that are likely to inform future decision-making (based on a systematic assessment of gaps in the current evidence base). Methods such as value of information analyses can be useful in this process. They can also make other recommendations for research. These will be listed in the guideline after the key recommendations for research but will be of lower priority. Further information about how research recommendations should be derived can be found in the research recommendation process and methods guide.

9.5 Incorporating the guideline recommendations into NICE Pathways

The committee and developer should refer to both the guideline scope and the NICE Pathway outline when developing the guideline. This includes taking account of the links to other NICE Pathways and the guidance identified as related to the guideline topic at the scoping stage. The committee and developer should aim for the guideline structure to be as similar as possible to the NICE Pathway. They should also consider the links with existing pathways to help integrate the new topic into NICE Pathways.

9.6 References and further reading

Alonso-Coello P, Oxman AD, Moberg J et al. for the GRADE working group (2016) GRADE Evidence to Decision (EtD) frameworks: a systematic and transparent approach to making well informed healthcare choices. 2: Clinical practice guidelines. BMJ 353: i2089

Claxton K, Sculpher MJ (2006) Using value of information analysis to prioritise health research: some lessons from recent UK experience. Pharmacoeconomics 24: 1055–68

Glasziou P, Del Mar C, Salisbury J (2003) Evidence-based medicine workbook. London: British Medical Journal Books

Guyatt GH, Oxman AD, Vist GE et al. for the GRADE working group (2008) GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ 336: 924 (see also the GRADE website)

Joint Royal College of Paediatrics and Child Health/Neonatal and Paediatric Pharmacists Group Standing Committee on Medicines (2013) The use of unlicensed medicines or licensed medicines for unlicensed applications in paediatric practice

Kelly MP, Moore TA (2012) The judgement process in evidence-based medicine and health technology assessment. Social Theory and Health 10:1–19

Michie S, Johnston M (2004) Changing clinical behaviour by making guidelines specific. British Medical Journal 328: 343–5

Nuffield Council on Bioethics (2007) Public health: ethical issues. London: Nuffield Council on Bioethics

Sackett DL, Straus SE, Richardson WS (2000) Evidence-based medicine: how to practice and teach EBM. Edinburgh: Churchill Livingstone

Schünemann HJ, Mustafa R, Brozek J et al. for the GRADE working group (2016) GRADE Guidelines: 16. GRADE evidence to decision frameworks for tests in clinical practice and public health. Journal of Clinical Epidemiology 76: 89–98

Scottish Intercollegiate Guidelines Network (2015) SIGN 50. A guideline developer's handbook, revised edition. Edinburgh: Scottish Intercollegiate Guidelines Network

Tannahill A (2008) Beyond evidence – to ethics: a decision making framework for health promotion, public health and health improvement. Health Promotion International 23: 380–90

Weightman A, Ellis S, Cullum A et al. (2005) Grading evidence and recommendations for public health interventions: developing and piloting a framework. London: Health Development Agency