Vitamin B12 deficiency in over 16s: diagnosis and management

Why the committee made the recommendations

The committee agreed it was important to give people information about vitamin B12 deficiency that is tailored to their individual circumstances, including the condition's various symptoms and causes, and what they can expect from investigations and treatment. They also highlighted difficulties in diagnosing vitamin B12 deficiency because its symptoms vary between individuals and are also linked to many other conditions. Based on their experience and expertise, the committee also agreed that the limitations of diagnostic tests should be explained to people with suspected vitamin B12 deficiency. In addition, they highlighted that some people would need further diagnostic tests, such as those with initial test results that are difficult to interpret because they are indeterminate. Having these discussions would mean people know what to expect from testing and could prevent any distress or concern over unclear test results.

Some causes of vitamin B12 deficiency are irreversible and therefore treatment is needed for life. The committee also wanted to reassure people with an irreversible cause of deficiency that vitamin B12 replacement will not be stopped in the future.

The committee also agreed that people starting treatment should be told that the dosage and form of vitamin B12 replacement could be adjusted or changed if their symptoms do not disappear, get worse or improve.

How the recommendations might affect practice

Information and support are already provided to people with suspected or confirmed vitamin B12 deficiency so these recommendations are unlikely to lead to a resource impact.

Return to recommendations

Why the committee made the recommendations

The committee recognised that vitamin B12 deficiency can be difficult to diagnose based on symptoms or signs alone. This is because symptoms and signs, such as unexplained fatigue, can vary from person to person and can also be indicative of other conditions.

People who have undergone a total gastrectomy or complete terminal ileal resection will need vitamin B12 replacement. This is usually planned for at the time of surgery. However, there is a risk that some people may not continue with, or have problems accessing, treatment (for example, if they lived abroad when they had the surgery). Therefore, the committee highlighted the need to continue or restart treatment in these groups of people.

The committee also made a recommendation to help ensure vitamin B12 deficiency is not missed in people who do not have anaemia or macrocytosis, as there is a common misconception that you cannot have a deficiency unless either, or both, of these signs are present.

They also highlighted an association between mental health problems and vitamin B12 deficiency, but acknowledged that mental health problems are also associated with many other conditions. In light of this, and in the absence of evidence to suggest a strong association between mental health problems and vitamin B12 deficiency, they did not recommend testing based on the presence of mental health problems alone.

How the recommendations might affect practice

The symptoms, signs and risk factors included in the recommendations are widely regarded as indications of vitamin B12 deficiency. However, this guideline could lead to a greater awareness of the condition. This may have a resource impact as it could lead to more testing and treatment.

Return to recommendations

Why the committee made the recommendations

The committee noted that there is currently no 'gold standard' test for diagnosing vitamin B12 deficiency. This means that, while tests can be a diagnostic aid, they cannot be relied on completely to confirm or rule out deficiency.

In the absence of high-quality evidence, the committee agreed that either total B12 (serum cobalamin) or active B12 (serum holotranscobalamin) should be used as the initial diagnostic test for most people. They noted that, while total B12 is the more commonly used test, in their experience active B12 is more accurate. This is because active B12 measures the form of vitamin B12 that can be taken up and used by the body, whereas total B12 measures a combination of both active and inactive forms of the vitamin. However, the active B12 test is also significantly more costly. Recommending it over total B12 would therefore lead to a notable change in practice and would be difficult to justify without evidence of cost effectiveness. The committee did acknowledge that active B12 is a more reliable test during pregnancy, when total B12 concentrations in the body fall even when there is no deficiency, so they recommended active B12 as the initial test for this group based on expert witness advice.

Total and active B12 are not suitable tests for people who use nitrous oxide recreationally because the substance inactivates the vitamin B12 molecule, so the person's total or active concentrations may therefore appear to be normal even when they have a deficiency. The committee acknowledged that in this group of people, both plasma homocysteine and serum methylmalonic acid (MMA) are used as diagnostic tests in current practice. Based on their experience, the committee agreed that plasma homocysteine was more reliable because elevated homocysteine in the body is seen before a rise in MMA. However, they agreed that both tests are likely to be suitable for users of nitrous oxide. The committee also recognised that serum MMA, where available, may be more practical to use in primary care. If plasma homocysteine is used, then the test will need to be done in a secondary care phlebotomy unit because the blood sample needs to be put on ice and promptly transported to the laboratory, which would be challenging in primary care.

People who are severely ill with symptoms or signs of vitamin B12 deficiency may need urgent treatment. Therefore, the committee agreed that people with suspected megaloblastic anaemia and neurological symptoms (especially when related to sub-acute combined degeneration of the spinal cord) need immediate vitamin B12 replacement to prevent worsening outcomes, so the committee recommended this treatment should start without waiting for any test results to come back.

People who are taking a medicine that is known to cause vitamin B12 deficiency may also need to start vitamin B12 replacement while waiting for their test results. This is because, in the committee's experience, the presence of symptoms combined with the use of the medicine suggests the person is highly likely to have a medicine-induced deficiency and they will therefore need vitamin B12 replacement.

The committee emphasised the need to take blood samples from people before they start treatment with vitamin B12 replacement because this could affect their test results. The committee also agreed that people should be asked which over-the-counter preparations containing vitamin B12 they are taking (if any), including different forms like injections or transdermal patches, and at what dosage. This could include some vitamin B12 preparations that are not licensed medicines available to prescribe by healthcare professionals. The committee agreed it was important to have these discussions because over-the-counter preparations can raise vitamin B12 concentrations in the blood and potentially mask a deficiency, depending on how much vitamin B12 they contain.

How the recommendations might affect practice

It is unlikely that these recommendations will lead to a substantial increase in the use of total or active B12 tests because these are already used as initial tests in current practice. If more centres decide to use active B12, there will be an increased cost to the NHS because the test is significantly more expensive. However, this could be offset because, in the committee's view, active B12 is a more accurate test. Overall, it is not anticipated that these recommendations will lead to a significant change in practice.

The recommendation to use plasma homocysteine or serum MMA as the initial test for people who use nitrous oxide recreationally reflects current practice, in most places, and is therefore unlikely to have a big resource impact.

Return to recommendations

Why the committee made the recommendation

The committee were aware that some test results need to be interpreted with caution but that some healthcare professionals may be unaware of this. These include the total or active B12 test results of people who are already taking over-the-counter preparations containing vitamin B12, because these can raise concentrations of the vitamin in the body and therefore mask a deficiency that has not been properly addressed. Total B12 test results may also be affected by hormonal changes in the body caused by the combined oral contraceptive pill.

How the recommendation might affect practice

The recommendation reflects current practice and is therefore unlikely to have a big resource impact.

Return to recommendation

Why the committee made the recommendations

The committee noted that there is currently no 'gold standard' test for diagnosing vitamin B12 deficiency. This means that while tests can be a diagnostic aid, they cannot be completely relied on to confirm or rule out deficiency (for example, a person with an indeterminate test result could still have a deficiency). For this reason, the committee made a recommendation for research into which diagnostic testing strategies are best for vitamin B12 deficiency.

In the absence of evidence, the committee used their experience and expertise to define clear thresholds for test results. These will help healthcare professionals decide when to diagnose deficiency and what to do if results are indeterminate or suggest deficiency is unlikely. The committee recognised that the cut-offs they specified for test results may increase the risk of false positives. However, these thresholds are likely to be used for people who are already symptomatic and are likely to need treatment, including those with test results near the cut-off for diagnosis. They also noted that treatment is inexpensive, not known to be harmful and could improve the person's health and quality of life.

There are 4 main manufacturers of the total B12 tests. The committee noted that, while 3 out of the 4 have thresholds that are similar to those recommended in the guideline, 1 manufacturer does not. Therefore, the guideline's thresholds should be applicable to most tests, but sometimes a laboratory's own different validated thresholds for total B12 will need to be used instead.

There are no generally accepted reference ranges for serum MMA and plasma homocysteine testing. Therefore, the committee agreed that the laboratory reference ranges would need to be used when interpreting results for these tests. They also highlighted that other factors can affect plasma homocysteine test results and should therefore be taken into account.

There was no evidence to show that using serum MMA to test everyone with an indeterminate result would be clinically and cost effective. It is also an expensive test. However, based on their expertise and experience, the committee acknowledged that a follow-up test with serum MMA could potentially lead to an earlier diagnosis. For this reason, they agreed that this test could be an option for people who had an indeterminate total or active B12 test result and symptoms or signs of deficiency.

Recent evidence from large cohort studies, based on samples that are representative of the UK population, suggests that people of Black ethnicity may have a higher reference range for serum concentrations of vitamin B12 in their blood than people of White or Asian ethnicity. However, the same reference range is currently used for people of all ethnicity. This means people of Black ethnicity may need treatment even if their blood test results are above one of the cut-offs used in the guideline to determine vitamin B12 deficiency. This will need to be taken into account together with symptoms, signs and risk factors when interpreting test results.

The committee identified groups of people who may need treatment if they have an indeterminate test result because a deficiency could adversely affect their health or quality of life. These groups may also need treatment without waiting for a serum MMA test, if one is being ordered. This was supported by economic analyses which suggested that, in some cases, starting treatment without waiting for a serum MMA test result could be more cost effective than offering treatment afterwards. The committee also agreed that a serum MMA test may still be the best indicator of a deficiency and may be helpful when deciding if the person needs vitamin B12 replacement in the long term. Therefore, the test may still be useful even if treatment is offered.

The committee looked at options for people who have an indeterminate total or active B12 test result but no symptoms or signs. This can happen because of incidental findings, such as the test results of people who do not have a suspected vitamin B12 deficiency but whose vitamin B12 concentrations were tested as part of routine blood investigations (for example, during a preoperative assessment or general health check). The committee agreed this group would not need treatment unless they developed symptoms or signs of vitamin B12 deficiency.

The committee agreed that it was important to investigate other causes of symptoms in people whose initial test results suggested deficiency was unlikely. However, those who still have symptoms 3 to 6 months on may need a repeat of the initial test in case the original result was a false negative.

How the recommendations might affect practice

Adopting the recommended thresholds for total or active B12 may change current practice for some laboratories but is unlikely to have a big resource impact. There are 4 main assays used for assessing total B12 concentrations. Three of the 4 use thresholds for total B12 that closely reflect the recommended thresholds for a diagnosis of vitamin B12 deficiency. If laboratories that use the fourth assay adopt the guideline thresholds instead of the locally validated thresholds, this may lead to a resource impact.

Currently, serum MMA testing is not routinely used in primary care as a follow-up test to aid the diagnosis of vitamin B12 deficiency. The recommendations on further testing will lead to a greater awareness of serum MMA and are likely to increase its use. Not all laboratories currently provide MMA testing and therefore this recommendation is likely to lead to a significant resource impact. However, increased use of serum MMA testing is also likely to lead to faster diagnosis and treatment which, in some cases, could avoid unnecessary investigations or a referral to secondary care.

Return to recommendations

Why the committee made the recommendations

The committee agreed it was important to test people if they have suspected autoimmune gastritis because this is an irreversible condition and they will need lifelong vitamin B12 replacement. Evidence found the anti-intrinsic factor antibody test to be the best initial option for diagnosing autoimmune gastritis. However, the committee agreed it should only be an option for people with suspected autoimmune gastritis because testing everyone believed to have the condition would be very expensive and may not influence treatment (because the person is likely to be receiving vitamin B12 replacement already). They also agreed the test was unnecessary for people who have already had a positive anti-intrinsic factor antibody test (because the result would be the same) or for those who have had a total gastrectomy or complete terminal ileal resection (because the surgery would be the likely cause of deficiency).

Based on expert witness advice and their expertise and experience, the committee recommended anti-intrinsic factor antibody testing for anyone who has suspected autoimmune gastritis and is pregnant or breastfeeding, unless they have previously had a positive test or have undergone certain operations. This is to ensure their health and that of their baby. For this reason, the committee also agreed that intramuscular vitamin B12 replacement should be started without waiting for test results.

The committee agreed that laboratory guidance would need to be followed when interpreting anti-intrinsic factor antibody test results. Based on their expertise and experience, they acknowledged that these test results can often be difficult to interpret, so laboratories would need to provide additional information to aid diagnosis.

The evidence also showed that, while a positive anti-intrinsic factor antibody test strongly suggests autoimmune gastritis, a negative test is less reliable so cannot be used to rule out the condition. Therefore, some people may need further investigations. Based on their experience and expertise, the committee agreed to a list of some options for this but there was insufficient evidence to recommend 1 investigation over the other. The choice would also depend on availability of the investigation and whether there were suitably trained and skilled healthcare staff available to carry it out. Gastroscopy can only be carried out by a specialist. The committee also acknowledged that this procedure, if it included a gastric body biopsy, would normally determine if someone had autoimmune gastritis. However, it is also an invasive and expensive procedure. The committee also acknowledged that none of the other tests listed in the recommendation can completely rule out autoimmune gastritis. They also agreed that more research was needed to determine the most effective investigations for people with negative anti-intrinsic factor antibody tests and so made a recommendation for research on the use of pepsinogen, gastrin, parietal cell antibodies and CobaSorb for identifying the cause of vitamin B12 deficiency.

It was unclear from the evidence whether it is more clinically and cost effective to do reflex testing (laboratories automatically testing) for anti-intrinsic factor antibodies when a low vitamin B12 concentration is detected, or for clinicians to request testing based on their own clinical judgement. Therefore, the committee also made a recommendation for research to compare the use of reflex testing with clinician-requested training.

The committee also highlighted the need to test for coeliac disease when further investigations have not identified a cause of vitamin B12 deficiency. This is in line with NICE's guideline on coeliac disease.

How the recommendations might affect practice

An anti-intrinsic factor antibody test is often requested when autoimmune gastritis is suspected, therefore the recommendations are unlikely to have a big impact. Further investigations are also used in some cases depending on their availability. However, the recommendation may raise awareness of these investigations and potentially lead to an increase in their use, which might require referral to hospital in some cases.

Return to recommendations

Why the committee made the recommendations

Evidence on the effectiveness of vitamin B12 replacement was mainly gathered from studies that were based on blood test results and made little reference to quality of life or other patient-reported outcomes, such as an improvement in symptoms. The aim of treatment is to increase vitamin B12 concentrations. However, based on their experience and expertise, the committee agreed that management of deficiency should also be based on both its cause and factors such as symptoms. They also highlighted a need for further research to identify the optimal treatment strategies for different causes of deficiency, particularly focusing on patient-reported outcomes. Therefore, they made a recommendation for research on the use of vitamin B12 replacement.

Based on their experience and expertise, the committee recommended that all people with vitamin B12 deficiency should be given information on what to expect from treatment and be advised that they may need to seek help if their symptoms are not improving as expected.

Based on expert witness advice, the committee agreed that vitamin B12 replacement should continue if it was started before pregnancy or breastfeeding and its use should be reviewed at a later date. This is because there is no known harm associated with vitamin B12 replacement and stopping treatment may lead to a return of, or worsening, symptoms, potentially leading to harm.

How the recommendations might affect practice

The recommendations reflect current practice and are therefore unlikely to have a big resource impact.

Return to recommendations

Why the committee made the recommendations

The committee agreed that anyone with confirmed vitamin B12 deficiency caused by autoimmune gastritis should receive lifelong intramuscular vitamin B12 replacement. This is because their bodies cannot adequately absorb the vitamin through the gastrointestinal tract, often making oral replacement ineffective. Intramuscular injections are also cheaper than oral replacement if used for 6 months or more, based on the guideline's costing analysis. Autoimmune gastritis is often difficult to diagnose, so the committee agreed that people who are likely to have the condition should have the same treatment as those with a diagnosis. This could prevent the effects of vitamin B12 deficiency becoming permanent.

People who have undergone a total gastrectomy or a complete terminal ileal resection need lifelong intramuscular injections because these types of surgery cause permanent malabsorption.

The committee agreed that people who have malabsorption for other reasons also need vitamin B12 replacement. This includes people who have coeliac disease or have had a partial gastrectomy, partial terminal ileal resection or some forms of bariatric surgery. These groups of people may not need lifelong treatment because, if the malabsorption is managed, the body may be able to absorb vitamin B12. This means the deficiency could potentially be reversed in the long term (for example, in coeliac disease by following a gluten-free diet). Intramuscular treatment was found to be more cost effective than oral replacement when used for 6 months or more. In the committee's experience, injections could be the better option for these groups of people because it can be difficult to judge how much of an oral dose will be absorbed by the body, so injections will help ensure they are getting enough of the vitamin. However, the committee did not rule out the use of oral vitamin B12 replacement, because there was no evidence on its use in these groups of people and therefore there was nothing to suggest it would be ineffective.

Studies comparing oral vitamin B12 replacement with intramuscular injections used an oral dosage of 1 mg a day. The committee agreed that it was important to use the same dosage because there is nothing to suggest a lower dosage is effective in people with suspected or confirmed malabsorption. It is also standard practice to offer intramuscular injections instead of oral treatment when malabsorption is suspected or confirmed.

How the recommendations might affect practice

The recommendations largely reflect current practice and are therefore unlikely to have a big resource impact.

Return to recommendations

Why the committee made the recommendations

There was no evidence to suggest that either oral or intramuscular vitamin B12 replacement was better for treating medicine-induced deficiency. So, based on their experience and expertise, the committee agreed that either could be offered depending on clinical judgement and the person's preference. Vitamin B12 replacement should also continue for as long as the person remains on the medicine, because the deficiency is unlikely to be resolved otherwise.

If appropriate, a medicine that can cause vitamin B12 deficiency should be stopped or changed. The committee agreed that if the medicine is stopped or changed and the person no longer has symptoms of vitamin B12 deficiency, the need for vitamin B12 replacement should be reviewed. This is because the cause of deficiency is likely to have been removed.

How the recommendations might affect practice

The recommendations for treating medicine-induced vitamin B12 deficiency reflect current practice and are unlikely to have a resource impact.

Return to recommendations

Why the committee made the recommendations

Recreational nitrous oxide is known to cause vitamin B12 deficiency by inactivating the vitamin in the body, but its longer-term effects are unknown. There was no evidence to suggest either oral or intramuscular vitamin B12 replacement was a better treatment for deficiency of this cause, so the committee agreed that either should be offered, depending on clinical judgement and the person's preference. They also agreed that people should be advised to stop using nitrous oxide recreationally to prevent the deficiency from getting worse.

The committee agreed that, if the person stops using nitrous oxide recreationally and they no longer have symptoms of vitamin B12 deficiency, the need for vitamin B12 replacement should be reviewed. This is because the cause of deficiency would have been removed. However, the person may need to continue with vitamin B12 replacement even after they have stopped using nitrous oxide, because it is unclear how long it will take for the deficiency to fully resolve.

How the recommendations might affect practice

The recommendations for managing deficiency caused by recreational nitrous oxide use reflect current practice. They are therefore unlikely to have a resource impact.

Return to recommendations

Why the committee made the recommendations

Deficiency caused by diet is potentially reversible. Based on their experience and expertise, the committee agreed it was important to talk to people who have a suspected or confirmed dietary deficiency about what they eat and drink, as well as symptoms, signs and risk factors, to establish if this is the cause of deficiency. They also noted that diet can incorrectly be assumed to be a cause of deficiency in people who get enough of the vitamin through their diet and may have a deficiency for other reasons. This includes people who are vegetarian or vegan who may get vitamin B12 from sources other than meat, fish and dairy products like yeast extract spreads and foods and drinks fortified with vitamin B12, such as some breakfast cereals. Misconceptions about some diets could potentially lead to under-investigation of other causes of deficiency, such as autoimmune gastritis, which can have serious long-term implications if left undiagnosed.

The committee also wanted to raise awareness of the wide variation in forms of oral over-the-counter supplements containing the vitamin. While some can effectively treat deficiency, others do not contain enough or the right type of the vitamin. This could include pseudo supplements or vitamin B12 analogues which are not appropriate for treating a deficiency. Therefore, the committee agreed to list the types of vitamin B12 people should look out for, should they wish to buy oral supplements.

Based on their experience and expertise, the committee agreed deficiency could be reversed in some people if they changed their diet, without the need for treatment. However, there was a lack of evidence in this area, so they also recommended oral vitamin B12 replacement for some people with a suspected or confirmed dietary deficiency so they get quick and effective treatment if they need it.

Based on expert witness advice, the committee agreed that a dosage of 1 mg may be needed if oral vitamin B12 replacement is prescribed during pregnancy or breastfeeding, when the body can need more vitamin B12. This minimum dosage could help ensure that enough vitamin is being absorbed and protect the health of those taking the treatment, and the health of their baby.

Based on their experience and expertise, the committee agreed that intramuscular vitamin B12 replacement could be the best option for some people. This includes those in whom treatment needs to work quickly because they are at risk of rapid deterioration that could significantly affect their quality of life. Injections could also be a better option if there are concerns about adherence to oral replacement. This could apply to older people who are in or have recently been in hospital, and also have complex comorbidity, or have frailty linked to undernutrition, dementia or decompensation. These groups of people are likely to be prescribed a few different medicines to take on a daily basis, so this would mean they do not have to take an additional medicine. Intramuscular injections may also be more effective in managing the vitamin B12 deficiency. Adherence may also be an issue for some people who are not in hospital but who may find it difficult, or may be unable, to collect, store or regularly take their medicine.

How the recommendations might affect practice

Most of the recommendations for managing dietary deficiency reflect current practice. However, the use of intramuscular instead of oral replacement when there are concerns about adherence will be a change in practice. This is unlikely to have a significant resource impact if no loading doses are needed because intramuscular treatment will always be cost saving compared with oral treatment. If loading doses are needed, then intramuscular injections will be cost effective if treatment is continued for longer than 6 months.

Return to recommendations

Why the committee made the recommendation

In the absence of evidence, the committee used their experience and expertise to recommend vitamin B12 replacement for anyone presenting with a deficiency that does not have a known or suspected cause, and where malabsorption is not suspected. They also agreed that oral vitamin B12 replacement should be considered instead of intramuscular injections, because this group of people should have had any necessary investigations to help rule out malabsorption, and oral treatment should therefore correct the deficiency. However, the committee agreed that oral treatment should be reviewed at the first follow up to make sure it is effective.

How the recommendation might affect practice

The recommendation for managing deficiency of an unknown cause, when malabsorption is not suspected, reflects current practice and is therefore unlikely to impact on resources.

Return to recommendation

Why we made no recommendations

No evidence comparing self-administered intramuscular or subcutaneous vitamin B12 injections with those administered by healthcare professionals was identified. The committee recognised that self-administration could be preferred by some people with vitamin B12 deficiency and that a precedent was set for it during the COVID-19 pandemic. It also has potential cost savings for the NHS. However, the committee did not recommend self-administration because there was no evidence regarding either the effectiveness or safety of this approach, and the use of subcutaneous vitamin B12 would be off-label. Instead, they made a recommendation for research on self-administration to inform future guidance.

Why the committee made the recommendations

There was no evidence on the most effective ongoing care and follow-up strategies for people with vitamin B12 deficiency, so the committee made recommendations based on their expertise and experience. For most people who have started treatment, a follow up at 3 months would give enough time to ensure it is working. However, some people with severe symptoms may need an earlier appointment so that treatment can be switched or adjusted if needed. Anyone who is pregnant or breastfeeding should also be followed up sooner to make sure they are getting the treatment they need to protect both their health and that of their baby.

Discussing symptoms at follow up would give an indication of how well treatment is working. The committee agreed that it is important to ask people about changes to their symptoms, especially symptoms that are so severe that they impact on normal daily activities.

The committee also made a recommendation for research into which components of follow-up reviews lead to the best outcomes for people receiving either oral or intramuscular vitamin B12 replacement. In particular, they agreed further research was needed into:

How the recommendations might affect practice

The recommendations largely reflect current practice. Usually, people are asked to seek medical help if treatment does not work as well as it should so the recommendations will not have a big resource impact. Offering everyone a follow-up appointment at 3 months or earlier depending on severity of symptoms (or 1 month during pregnancy or breastfeeding) will also help ensure the person has the right diagnosis and treatment. This will help prevent unnecessary treatment, which will outweigh any additional costs related to follow-up appointments.

Return to recommendations

Why the committee made the recommendations

There was no evidence demonstrating the effectiveness of monitoring people with autoimmune gastritis for gastric cancer. However, based on their experience and expertise, the committee agreed it was important to highlight the increased incidence of gastric adenocarcinoma and gastric neuroendocrine tumours in people with suspected or confirmed autoimmune gastritis. Raising awareness of this may mean people are more likely to report any new or worsening gastrointestinal symptoms to a healthcare professional. They also stressed that people may need referral for gastrointestinal endoscopy if they have new upper gastrointestinal symptoms (for example, dyspepsia, nausea or vomiting) as these could suggest the presence of cancer.

The committee also made a recommendation for research on monitoring for gastric cancer in people with autoimmune gastritis.

How the recommendations might affect practice

These recommendations reflect current practice and are unlikely to have a resource impact.

Return to recommendations