1.1.3.1. Included studies

A total of 3,435 RCTs and systematic reviews were screened at title and abstract stage after deduplication.

Following title and abstract screening, 288 studies were included for full text screening to see if they were relevant to any of the CGM questions that were included in this update (CGM for adults with type 1 diabetes, CGM for adults with type 2 diabetes and CGM for children and young people with type 1 diabetes).

Of the 288 included studies, 70 were potentially relevant for the type 1 diabetes children and young people question. The other 218 were assessed for relevance for the other CGM questions (for more information on the included studies for the other questions see Evidence review X: CGM for type 1 diabetes and Evidence review X: CGM for type 2 diabetes).

The 70 studies were reviewed against the inclusion criteria as described in the review protocol (Appendix A). Overall, 6 studies were included, along with 8 systematic reviews that were checked for additional references. No additional studies were identified from the systematic reviews.

Most studies compared rtCGM against SMBG but some compared isCGM to SMBG. The number of studies for each comparison is outlined in Table 2. Further information about these studies is shown in Table 3.

Table 2

List of comparisons and associated studies/trials.

Regarding rtCGM vs isCGM, a check for observational studies and propensity matched cohort studies was carried out and nothing was identified. The committee therefore felt they had enough evidence to make recommendations.

See Appendix E for evidence tables and the reference list in section 1.1.8 References – included studies.

1.1.3.2. Excluded studies

Overall, 56 studies were excluded. See Appendix K for the list of excluded studies with reasons for their exclusion.

Evidence in meta-analysis

Table 4

Summary of GRADE: rtCGM vs SMBG.

Table 5

Summary of GRADE: isCGM vs SMBG.

1.1.6.1. Included studies

A systematic literature search was undertaken to identify published health economic evidence relevant to the review questions. Studies were identified by searching EconLit, Embase, CRD NHS EED, International HTA database, MEDLINE, PsycINFO and NHS EED. All searches were updated on 5th May 2021, and no papers published after this date were considered. This returned 3,040 references (see appendix C for the literature search strategy). After deduplication and title and abstract screening against the review protocol, 3,021 references were excluded, and 19 references were ordered for screening based on their full texts.

Of the 19 references screened as full texts, 2 were systematic reviews. Both were investigated as a source of references, from which one more study was added (Healthcare Improvement Scotland 2018). In total, there were 14 primary studies that contained cost-utility analyses evaluating some of the following methods of glucose monitoring to improve glycaemic control: 1) rtCGM; 2) isCGM; 3) intermittent capillary blood glucose monitoring. However, none of these studies were in a population of children and young people with type 1 diabetes, and therefore all these studies were excluded from the review. The health economic evidence study selection is presented as a flowchart in appendix H.

1.1.6.2. Excluded studies

Studies excluded in the full text review, together with reasons for exclusion, are listed in appendix K.

1.1.6.2. Economic model

No economic modelling was undertaken for this review question. However, the committee did consider the results of the modelling undertaken for adults with type 1 diabetes when making recommendations for children and young people.

The outcomes that matter most

The committee agreed that outcomes such as HbA1c and time in range were important for measuring a person’s blood sugar levels over time. HbA1c is limited by it reflecting the previous 3 months of therapy, whereas time in range is a measurement over a shorter time period. The committee considered time in range to be a better measure than HbA1c as it captures variation over time and can be used to highlight hypoglycaemia and hyperglycaemia, whereas HbA1c gives an average value and does not indicate how often hypoglycaemia or hyperglycaemia occurs. The committee thought that time in range was an important measure when assessing the clinical effectiveness of CGM interventions.

Hypoglycaemia events, severe hypoglycaemia events, and nocturnal hypoglycaemia were also considered to be important outcomes. These are often highlighted by people living with type 1 diabetes as key due to the fear these events generate and the impact they can have on quality of life. Therefore, a reduction in hypoglycaemia events results in significant improvements to quality of life. Outcomes relating to hypoglycaemic events and quality of life were therefore both considered important.

The committee highlighted that fear of hypoglycaemia was a key quality of life outcome, due to the severity this fear has on children and young people and their parents and carers.

Other key outcomes can be seen in the review protocol in Appendix A.

The quality of the evidence

All outcomes other than mortality were captured in at least 1 comparison in the data extracted. There was no time in range or glycaemic variability data available for isCGM vs SMBG. Time in range is harder to record in isCGM as this does not continuously capture glycaemic levels in the same way as a rtCGM device.

The committee acknowledged that there was no evidence directly comparing rtCGM and isCGM in children and young people, and found this unsurprising considering the small amount of evidence in the adult population for the same comparison. The committee judged that for type 1 diabetes they had enough evidence to justify the superiority of rtCGM over isCGM in this population, and as a result did not consider there was need for a research recommendation. The committee did note that due to the increasing incidence of type 2 diabetes in children and young people, they should make a research recommendation into clinical effectiveness for this group (see Appendix L.1.1).

The committee also noted that much of the outcome evidence for diabetes in children and young people is now available in routinely collected real-world data, rather than clinical trials. They therefore made another research recommendation to determine effectiveness and cost effectiveness of CGM devices in children and young people using this evidence base (see Appendix L.1.2).

Evidence for rtCGM vs SMBG ranged from very low to high quality and all but one of the studies (Laffel 2020) were directly applicable to the review question. Laffel (2020) was considered partially applicable to the review because it included people with an age range of 14 – 24 years. However, the 14-<19 population made up 64.9% - 67% of the study, and so it still met the criteria in the protocol for >50% of included people being paediatric cases. As the study was at low risk of bias and presented many outcomes, the committee thought it was important to consider as part of the review, The quality of outcome data from some of the other studies were downgraded for risk of bias, mostly due to limited information about randomisation and allocation concealment methods. The committee pointed out that no study was based entirely in the UK. The SWITCH trial (Hommel 2014) had the majority of its centres in the UK, but only reported 2 quality of life outcomes, meaning there was no information to directly show the clinical effectiveness of CGM in UK practice. The committee noted availability and cost of devices would vary considerably across other healthcare systems and data from other countries and this had to be taken into account when making recommendations. The committee did highlight that hypoglycaemia fear survey outcomes were of moderate quality and did show an effect in rtCGM vs SMBG, indicating the effectiveness of rtCGM in this key quality of life outcome.

Only 2 studies compared the use of isCGM to SMBG, and one of these (Boucher 2020) had an inclusion criteria age limit of 13–20 that was only partially applicable to this review. However, as the mean age was within the inclusion criteria for this review, the committee considered that this was still acceptable for inclusion in the analysis. The other study (Xu 2021) was graded as high risk of bias due to limited information about the type of analysis used, and so outcomes containing this study where it was weighted >33.3% were downgraded for very serious risk of bias.. Due to reasons outlined above, the committee did not have full confidence in this evidence alone and used a combination of the evidence, and their clinical knowledge and experience to inform recommendations for the use of CGM for children and young people with type 1 diabetes.

Benefits and harms

The committee considered that the results showing a decrease in HbA1c and an increase in time in target glucose range in rtCGM vs SMBG were promising outcomes, and reflected their experience from clinical practice. They specifically noted that time in range increased by more than the minimal important difference, they interpreted the increase in time in range to be clinically meaningful (>5%). They acknowledged that only dichotomous HbA1c outcomes were effective, rather than the continuous HbA1c outcomes. However they consider these to be acceptable from their experience and were included in the study protocol as relevant outcomes. The fact that these results were also supported by a reduction of time in hyperglycemia and a reduction in concerns reported in the hypoglycemia fear survey (an important and a moderate quality outcome), both of which the committee noted had clinical importance gave them confidence that the effects shown in the meta-analysis were supportive of rtCGM use in children and young people. The committee did not consider any other quality of life measures to be as important in decision making.

For isCGM, the committee noted that none of the outcomes they deemed informative showed an effect greater than the minimally important difference (MID), and most of the outcomes showed no meaningful difference, or could not differentiate between isCGM or SMBG. There was an effect of an increased number of glucose checks in isCGM users vs SMBG. However the committee considered that this outcome was not informative regarding the effectiveness of isCGM and it did not answer the review question. They explained there were a number of reasons unrelated to its isCGM effectiveness as to why glucose check numbers might increase, particularly in a clinical study where people were likely reminded of the importance of recording data.

Although some outcomes showed no meaningful difference, or could not differentiate between rtCGM or SMBG, where there was an effect it consistently favoured the use of rtCGM.

As the evidence showed key outcomes favoured rtCGM over SMBG, the committee recommended rtCGM use first in all children and young people with type 1 diabetes, only offering isCGM if rtCGM is not preferred or contraindicated. The committee highlighted that the active component of isCGM, of having to “swipe to take a reading” although easier than doing a blood test may be part of the reason adherence may not be as good in some young people more than adults as it requires them to undertake an action. They also highlighted that currently the isCGM device Freestyle Libre does not have a license for children aged under 4 so could not be used in that age group. They also highlighted that the function of sharing readings with parents or carers and is available for both rtCGM and isCGM. This function is important for young children but also for older children as they become more independent at school and start to make their own decisions about meals and insulin doses. Feedback from a CGM device that is provided to both a child and their parents or carers can help to provide remote support and early identification of hypoglycaemia.The committee highlighted that the individual choice element of different CGM devices would be a benefit to children and young people and their parents or carers, as the ‘best’ device for each individual would depend on their preferences, needs and characteristics. They therefore included a summary table in the recommendations outlining the factors to consider when choosing a CGM device. This was adapted from a list of factors that the committee had already decided were important for adults with type 1 diabetes (see evidence review on continuous glucose monitoring in adults with type 1 diabetes). Changes to this list were made based on the committee’s clinical knowledge and experience. They agreed it was important to acknowledge the role of the parent or carer in the decision-making process when deciding on the best method of glucose monitoring for children and young people, so this was added to the list of factors. The recommendations for adults for type 1 diabetes indicates that the ease of use should be considered when choosing the best method of blood glucose monitoring, considering factors such as whether someone has limited dexterity. The committee discussed how other factors should also be considered for children and young people, such as their age and abilities and how this might affect the best choice of monitor. They thought it was also important to consider whether other people would have to take recordings from the device, such as teachers or other people who temporarily care for the child or young person. An additional consideration is how unpredictable their activity patterns are and whether they take part in sport and exercise. The committee noted that children and young people often have less predictable activity patterns than adults and so it is important for them to be aware of their changing blood glucose levels in response to any changes in activity.

The committee clarified that the child or young person and their families or carers should consult with a member of the diabetes care team with expertise in the use of CGM. Furthermore, children and young people using CGM who have language difficulties or physical or learning disabilities would also benefit from this team’s support.

The committee agreed that the recommendations should also highlight the importance of children and young people and their parents or carers being given education about CGM. This will help them understand how CGM works and the benefits it can provide. Improving understanding of CGM will increase the likelihood that it will be used correctly, such as scanning frequently and reporting the results so that no important data is missed. This will help children and young people gain the greatest benefit from the use of this technology and be able to manage their diabetes effectively. Extra effort should be made to ensure that the training is accessible to families where English is not their first language by use of interpreters and providing information in different languages.

The committee highlighted that it was important to use the device consistently to ensure a more positive effect. They therefore made a recommendation for the device to be worn 70% of the time, and for education and support to be provided if this wasn’t the case. This recommendation was also made to avoid ongoing prescription of devices that aren’t being used, and to give providers an opportunity to address any barriers that may be reducing someone’s ability to use the device effectively. The committee justified this 70% figure as this was reported in the JDRF study (2008, 2010) which showed CGM use of on average >= 6 days a week was predictive of positive outcomes. This usage figure from the JDRF study was also considered in the Chase (2010) study, which showed the 17 subjects using CGM >=6 days/week had substantially greater improvement from baseline in HbA1c than did the 63 subjects using CGM <6 days/week. The committee acknowledged that 80% is a high threshold, and that in clinical practice the more lenient threshold of 70% is used, they therefore deferred to current practice and their clinical experience for this value.

The committee emphasised that use of less than 70% should trigger a discussion to assess whether the device is working for them, or steps could be taken to help make use easier. The committee stressed that this should be a positive discussion to encourage and support the use of the device. This could include initially introducing CGM over a trial period and explaining that the benefits will be assessed over the trial period to decide whether it is an appropriate longer-term option. This is useful to assess both clinical benefit, such as reduction in hypoglycaemic episodes, and benefit to the child or young person using it. For instance, while some will find CGM a helpful method to manage their diabetes, others may feel overwhelmed by the additional information it provides. The committee discussed how temporary, rather than permanent, use of CGM may actually be useful for some children and young people. Using CGM for a short period of time may help children and young people to understand when they have hypoglycaemic episodes, thereby helping them to develop a more effective treatment plan. By developing this understanding of their blood glucose patterns, children and young people can still benefit from CGM even if is decided that they do not want to use the monitor on a long-term basis. By making people aware from the outset that the effectiveness of CGM will be assessed based on discussions between clinicians and children or young people and their families and carers, mutual decisions can be made over whether to pause the use of CGM. This will avoid the risk of conflict that might be present if a clinician were to decide that the use of the device should be stopped without discussions with the child or young person.

The committee highlighted that one barrier to adherence to CGM that is a particular area of concern for children and young people is that some children develop skin reactions when wearing a CGM device due to the sensor adhesive. The committee therefore made a research recommendation (see Appendix L.1.3) to investigate strategies to reduce local skin reactions to promote ease of use of these devices.

Cost effectiveness and resource use

In the absence of any economic evidence specific to children and young people with type 1 diabetes, the committee considered whether the evidence from adults with type 1 diabetes could reasonably be extrapolated to the younger population. They agreed that, assuming the same clinical benefits for a technology were identified in children and young people as in adults, then the technology should be at least as cost-effective in children and young people as in adults. This was because there are some situations where the same outcomes would be expected in children and adults (for example, the direct quality of life impact of a hypoglycaemic event) and some where the benefit might be larger in children (for example fear of hypoglycaemia, where both the child and their parents/guardians may experience this fear), but nothing where the impact in children would be expected to be less. The committee agreed there would be limited value in additional modelling specific to children and young people because of the extra uncertainties in the CORE diabetes model for that population.

The committee agreed the clinical review showed similar benefits for rtCGM in children as in adults, and were therefore comfortable to extrapolate the cost-effectiveness results, concluding that rtCGM was cost-effective in this population. However, since the same clinical benefits were not found for isCGM in children as in adults, the committee agreed those cost-effectiveness findings could not be extrapolated, and therefore were not prepared to conclude that isCGM is a cost-effective technology. They therefore agreed the use of isCGM should be restricted to those people who are unable to or do not want to use rtCGM.

They agreed that this finding (a positive result for rtCGM but not for isCGM) was consistent with their experience of the technologies in children and young people. The committee highlighted the fact that rtCGM has better functionality that makes it more suitable for children and young people. Although some versions of isCGM also have active alerts/alarm functions that warn users of immediate or impending hypoglycaemic events, they still require users to consciously scan the sensor to obtain glucose data. rtCGM, on the other hand, automatically shows a continuous stream of real-time numerical and graphical information on the receiver, so is easier to manage for children and young people, or their parents/guardians. This could lead to a higher adherence rate for rtCGM compared with isCGM among the younger cohorts. They also agreed that if a child or young person expressed a clear preference for using isCGM over rtCGM, their adherence was then likely to be better, meaning the device would be beneficial, as adherence was felt to be the key reason for rtCGM being a more effective technology on average in children and young people.

The committee noted that although the new recommendations are an expansion of the use of rtCGM compared to the previous recommendations for children and young people, the resource impact will be relatively small compared with current practice, as in recent years there has already been a considerable expansion of it’s use in this population. Additionally, the population of children and young people with type 1 diabetes is much smaller than the population of adults with type 1 diabetes, and rtCGM is already being used in a considerable proportion of this the paediatric population, meaning the recommendations do not represent a considerable a change in practice as they do for adults. They also noted that there were a number of different rtCGM devices available with considerable overlap in functionality and features, and that therefore if there were multiple different devices available that would meet the person’s needs and preferences, the cheapest of those available devices should be used.

The recommendations on education, monitoring and support for people using rtCGM are not expected to require substantial additional resources. This is because education, monitoring and support are al already recommended for all children and young people with type 1 diabetes and would be necessary whether or not a person was using rtCGM. Group training sessions rather than individual training sessions will help reduce the extra resource that maybe required for this purpose.

Other factors the committee took into account

The committee considered extending this recommendation to all children and young people with type 1 diabetes would help remove the observed discrepancies in clinical practice and address known inequalities in access. For example, those from lower socioeconomic groups or those from black, Asian and minority ethnic minority groups who from their clinical experience have been less likely to be prescribed these devices. Despite the positive recommendation for the use of CGM in children and young people with type 1 diabetes, the committee were concerned that inequalities may still occur with uptake of CGM being lower in certain groups. To address this the committee added a recommendation outlining actions to address this. The committee also agreed that capillary blood glucose monitoring is still needed as a back-up in situations such as when blood glucose levels are changing quickly or due to technology failure.

Recommendations supported by this evidence review

This evidence review supports the updated recommendations in NG18: 1.1.2 – 1.1.12 and research recommendations 7 –9.

Quality assessment

Individual RCTs were quality assessed using the Cochrane Risk of Bias Tool 2.0. Cohort studies were quality assessed using the ROBINS-I tool. Each individual study was classified into one of the following groups:

• Low risk of bias – The true effect size for the study is likely to be close to the estimated effect size.
• Moderate risk of bias – There is a possibility the true effect size for the study is substantially different to the estimated effect size.
• High risk of bias – It is likely the true effect size for the study is substantially different to the estimated effect size.
• Critical risk of bias (ROBINS-I only) - It is very likely the true effect size for the study is substantially different to the estimated effect size.

Each individual study was also classified into one of three groups for directness, based on if there were concerns about the population, intervention, comparator and/or outcomes in the study and how directly these variables could address the specified review question. Studies were rated as follows:

• Direct – No important deviations from the protocol in population, intervention, comparator and/or outcomes.
• Partially indirect – Important deviations from the protocol in one of the following areas: population, intervention, comparator and/or outcomes.
• Indirect – Important deviations from the protocol in at least two of the following areas: population, intervention, comparator and/or outcomes.

Methods for combining intervention evidence

Meta-analyses of interventional data were conducted with reference to the Cochrane Handbook for Systematic Reviews of Interventions (Higgins et al. 2011).

Where different studies presented continuous data measuring the same outcome but using different numerical scales (e.g. a 0–10 and a 0–100 visual analogue scale), these outcomes were all converted to the same scale before meta-analysis was conducted on the mean differences.

A pooled relative risk was calculated for dichotomous outcomes (using the Mantel–Haenszel method) reporting numbers of people having an event, and a pooled incidence rate ratio was calculated for dichotomous outcomes reporting total numbers of events. Both relative and absolute risks were presented, with absolute risks calculated by applying the relative risk to the risk in the comparator arm of the meta-analysis (calculated as the total number events in the comparator arms of studies in the meta-analysis divided by the total number of participants in the comparator arms of studies in the meta-analysis).

Fixed-effects models were the preferred choice to report, but in situations where the assumption of a shared mean for fixed-effects model were clearly not met, even after appropriate pre-specified subgroup analyses were conducted, random-effects results are presented. Fixed-effects models were deemed to be inappropriate if one or both of the following conditions was met:

• Significant between study heterogeneity in methodology, population, intervention or comparator was identified by the reviewer in advance of data analysis. This decision was made and recorded before any data analysis was undertaken.
• The presence of significant statistical heterogeneity in the meta-analysis, defined as I²≥50%.

However, in cases where the results from individual pre-specified subgroup analyses are less heterogeneous (with I² < 50%) the results from these subgroups will be reported using fixed effects models. This may lead to situations where pooled results are reported from random-effects models and subgroup results are reported from fixed-effects models.

In situations where subgroup analyses were conducted, pooled results and results for the individual subgroups are reported when there was evidence of between group heterogeneity, defined as a statistically significant test for subgroup interactions (at the 95% confidence level). Where no such evidence as identified, only pooled results are presented.

In any meta-analyses where some (but not all) of the data came from studies at critical or high risk of bias, a sensitivity analysis was conducted, excluding those studies from the analysis. Results from both the full and restricted meta-analyses are reported. Similarly, in any meta-analyses where some (but not all) of the data came from indirect studies, a sensitivity analysis was conducted, excluding those studies from the analysis.

Meta-analyses were performed in Cochrane Review Manager V5.3, with the exception of incidence rate ratio analyses which were carried out in R version 3.3.4.

Minimal clinically important differences (MIDs)

The Core Outcome Measures in Effectiveness Trials (COMET) database was searched to identify published minimal clinically important difference thresholds relevant to this guideline. Identified MIDs were assessed to ensure they had been developed and validated in a methodologically rigorous way, and were applicable to the populations, interventions and outcomes specified in this guideline.

In addition, the Guideline Committee were asked to prospectively specify any outcomes where they felt a consensus MID could be defined from their experience. In particular, any questions looking to evaluate non-inferiority (that one treatment is not meaningfully worse than another) required an MID to be defined to act as a non-inferiority margin.

MIDs found through this process and used to assess imprecision in the guideline are given in Table 2. For other continuous outcomes not specified in the table below, no MID was defined.

Table 6. Identified MIDs (PDF, 107K)

For continuous outcomes expressed as a mean difference where no other MID was available, an MID of 0.5 of the median standard deviations of the comparison group arms was used (Norman et al. 2003). For relative risks where no other MID was available, default MIDS of 0.8,1.25 were used.

When decisions were made in situations where MIDs were not available, the ‘Evidence to Recommendations’ section of that review makes explicit the committee’s view of the expected clinical importance and relevance of the findings. In particular, this includes consideration of whether the whole effect of a treatment (which may be felt across multiple independent outcome domains) would be likely to be clinically meaningful, rather than simply whether each individual sub outcome might be meaningful in isolation.

GRADE for pairwise meta-analyses of interventional evidence

GRADE was used to assess the quality of evidence for the selected outcomes as specified in ‘Developing NICE guidelines: the manual (2014)’. Data from randomised controlled trials, non-randomised controlled trials and cohort studies were initially rated as high quality while data from other study types were originally rated as low quality. The quality of the evidence for each outcome was downgraded or not from this initial point, based on the criteria given in Table 3.

Table 7. Rationale for downgrading quality of evidence for intervention studies (PDF, 130K)

Summary of evidence is presented in section 1.1.6. This summarises the effect size, quality of evidence and interpretation of the evidence in relation to the significance of the data.

Evidence was also identified for which GRADE could not be applied as the evidence was presented in the form of median and interquartile range. This evidence is presented in Appendix G. This evidence has been summarised narratively in section 1.1.10.

Search strategies

Download PDF (298K)

L.1.1.1. Why this is important

There is some evidence on the effectiveness and cost-effectiveness of CGM devices to improve glycaemic control in children and young people with type 1 diabetes. However, there is none for people in this age group who have type 2 diabetes. Evidence is therefore needed to see whether children and young people with type 2 diabetes could gain similar benefits from the use of CGM devices as those who have type 1 diabetes. This may make it possible to recommend CGM for use with this group in future.

L.1.1.2. Rationale for research recommendation

Download PDF (134K)

L.1.1.3. Modified PICO table

Download PDF (142K)

L.1.2.1. Why this is important

There is currently no evidence on the effectiveness and cost-effectiveness of CGM devices to improve glycaemic control in children and young people with type 2 diabetes, and only RCT evidence for children and young people with type 1 diabetes. While RCT evidence is useful, it does not necessarily provide the same evaluation of how well these devices work on a daily basis in normal life as real-world data. By using real-world data, it will be possible to identify how effective different CGM devices are to a wide range of children and young people from different backgrounds. This may lead to an increased understanding of CGM devices and make it possible to produce recommendations about their use for children and young people in future.

L.1.2.2. Rationale for research recommendation

Download PDF (143K)

L.1.2.3. Modified PICO table

Download PDF (132K)

L.1.3.1. Why this is important

One of the factors which affects the use of CGM devices in children and young people is sensitivities to the device, such as reactions to the adhesive used for the sensors. More research will help to determine which adhesives are least likely to result in these sensitivities, therefore potentially increasing adherence to the use of CGM devices.

L.1.3.2. Rationale for research recommendation

Download PDF (134K)

L.1.3.3. Modified PICO table

Download PDF (132K)