3 Considerations

This section describes the factors and issues the Expert Reference Group (ERG) considered when examining the cost effectiveness of moving the Healthy Start vitamin programme from the current targeted offering to a universal offering. For further details see the modelling report.

Model inputs

3.1 The cost effectiveness review of the vitamins contained in the Healthy Start supplements only identified 9 studies. Most were local evaluations and only 2 studies – 1 on folic acid and 1 on vitamin D – were formal economic evaluations. As a result, it was necessary to build a de novo economic model to answer the questions outlined in section 1. The methods and data used in the model are discussed in detail in the modelling report.

Limitations in the data: outcome measures

3.2 The review did not identify any data on the cost effectiveness of multivitamins of the same composition as those provided by the Healthy Start scheme. So the cost effectiveness of providing the relevant population groups with each vitamin contained in the supplements had to be modelled.

3.3 The ERG provisionally agreed 3 outcome measures for each nutrient. These related to: the lower reference nutrient intake (LRNI), nutritional status and a functional outcome measure. The Group agreed the functional outcome measure should reflect the original rationale for including these nutrients in Healthy Start supplements (see Scientific review of the Welfare Food Scheme Department of Health). (Members noted that supplements containing these nutrients may also provide additional benefits.) It was not possible to identify all 3 outcome measures for each nutrient, however. Most vitamin D, for example, is produced through skin synthesis and no LRNI has been set. Measures of nutritional status could not be used because the national survey data are not analysed according to whether supplements were or were not taken. It was not possible to carry out meaningful additional analyses because the numbers of participants in each of these groups (takers and non‑takers of supplements) who provided a blood sample were too small. In addition, there were no suitable functional outcomes for vitamins A or C: being generally unwell, having poor immunity or 'growth faltering' (failure to thrive, for more detail see the introduction to section 4) may be associated with poor nutritional status, but this is not only attributable to a poor vitamin A or C status. It would also be difficult to assign costs or QALYs to these states. So the outcome for vitamins A and C focused on the proportion of the population with intakes below the LRNI[6].

3.4 There are concerns about rickets re‑occurring in the UK (Diagnosis and management of vitamin D deficiency Pearce and Cheetham 2010) and the occurrence of hypocalcaemic fits due to vitamin D deficiency. But these conditions are still relatively rare. For example, the incidence of hypocalcaemic fits in children aged 0–15 years in the UK and Ireland has been reported as 3.49 per million (Incidence of hypocalcemic seizures due to vitamin D deficiency in children in the United Kingdom and Ireland Basatemur and Sutcliffe 2014). Symptomatic vitamin D deficiency[7] is more common. For example, a survey in the West Midlands found an annual incidence of 7.5 per 100,000 in children under 5. Because costs can be assigned to symptomatic vitamin D deficiency, this was selected as the functional outcome for vitamin D. But due to a lack of data, it was not possible to assign quality‑adjusted life‑year (QALY) data for this outcome in the main analysis. The hypothetical effect of adding quality of life benefits for the effects of vitamin D was tested in a separate sensitivity analysis. Neural tube defects were used as the outcome for folic acid, because this was added to the supplements to reduce the risk of conditions such as spina bifida and anencephaly. In addition, it is possible to assign both costs and quality of life measures to these conditions.

3.5 A major limitation of the model is that, with the exception of folic acid, it was not possible to assign QALYs to the outcome measures. It was only possible to consider a cost–consequence analysis of vitamins A, C and D. This means folic acid has a strong impact on the results.

Limitations of the data: model inputs

3.6 The age range of the children being modelled and the 'life‑stage' of the women did not always correspond exactly with the ages and life stages reported in national surveys. 'Life stage' here refers to: planning a pregnancy, being more or less than 10 weeks pregnant or having a child aged under 1 year. The ERG agreed that the model should use the closest possible match. See pages 13–15 of the modelling report.

3.7 The ERG had to make assumptions if surveys reported that supplements were taken but didn't specify what they contained. For example, the Infant Feeding Survey does not report specifically on vitamin C supplements, but it does collect data on the use of 'multivitamins' and these have been assumed to contain vitamin C.

Limitations of the data: costs

3.8 The Department of Health provided the central administration costs of the vitamin aspects of the scheme in confidence. A survey was commissioned to estimate the local costs of running the scheme (see pages 10–11 of the modelling report). Only 8 partially completed responses were received and there was enormous variation in the costs.

3.9 The ERG recognised that there is more than 1 distribution route for universal provision. For example, the supplements could be distributed by health professionals, in community pharmacies and in supermarkets. But the costs of these approaches are unknown.

3.10 Healthy Start leads in 2 specific geographical areas provided data to estimate the costs of a targeted and universal scheme. Data from one that provided supplements via an electronic card system was used in the main analysis because the data were more robust. But the ERG was aware that these data may not be representative of other areas where different approaches may be used. They also noted that the approach and costs may vary from rural to urban areas. The modellers therefore carried out a separate sensitivity analysis (see pages 43 and 44 of the modelling report) to determine the effect of varying the costs.

3.11 The cost of expanding production for, and monitoring and evaluation of, a universal scheme was not included in the model. The model also excluded the possible costs that may be incurred if more manufacturers need to be licensed to produce the children's vitamin drops.

3.12 The main analysis focused on cost effectiveness from the public sector (local authority, central government and the NHS) perspective. A separate analysis was carried out to determine cost effectiveness from a wider societal perspective (including any costs to individuals). To inform the latter, an online survey aimed to ascertain the average price women pay for vitamins for themselves and their children (see pages 9, 52 and 53 of the modelling report). The supplements they bought appeared to be relatively expensive (around £6–£8 per pack), given that lower cost supplements are available. Members had concerns about how representative these data were but because they were the best available, this information was used.

3.13 Additional sensitivity and scenario analyses were carried out to test the effect of varying some of the inputs on cost effectiveness. For example, one scenario analysis on the uptake of folic acid in women planning a pregnancy was based on data from a study by Bestwick et al. 2014 ('Prevention of neural tube defects: a cross‑sectional study of the uptake of folic acid supplementation in nearly half a million women'), in place of data from the Infant Feeding Survey. The study sample was larger and consisted of pregnant women as opposed to new mothers, but was not nationally representative. A scenario analysis using data on supplement use from the 'Born in Bradford' study was carried out to include more data on black and minority ethnic groups: these groups are under‑represented in national dietary surveys. In both cases, the estimated ICER per QALY gained was slightly lowered, but using different data did not affect the overall findings of the model. See pages 60 and 61 of the modelling report.


3.14 The ERG noted that the important assumptions made in the model about delivery may have a crucial effect on the cost effectiveness findings. In particular, the model assumes that existing delivery mechanisms would be used for any new group included in the universal offering. This may be plausible for some groups, such as infants aged 0–6 months, because a delivery mechanism already exists. But for others, for example women planning a pregnancy and those in the first 10 weeks of pregnancy, different routes may be needed. The costs of these may differ and their impact on the overall cost effectiveness of a universal scheme is unknown.

3.15 The ERG noted that reaching women who are planning a pregnancy could be particularly difficult, because they may not be involved with health services. But these women and those less than 10 weeks pregnant, need to be included if the scheme is to be cost effective.

3.16 The ERG noted that including infants aged 0–6 months in a universal scheme may be feasible, because they are in regular contact with health professionals. This may have practical benefits too. A recent study showed that 27% of hypocalcaemic fits due to vitamin D deficiency occur in infants younger than 1 month ('Incidence of hypocalcemic seizures due to vitamin D deficiency in children in the United Kingdom and Ireland'). Offering Healthy Start supplements on a universal basis may help reduce this risk. However, no suitable data were identified to carry out a specific analysis on hypocalcaemic fits alone (see page 22 of the modelling report).

3.17 Currently, mothers who qualify receive Healthy Start supplements during pregnancy and infants receive them from age 6 months, so there is a 'break' in supplementation between 0 and 6 months. The ERG noted that avoiding this 'break' may encourage adherence to advice about vitamin D, because mothers would still be seeing health professionals. Having to resume contact with mothers at 6 months involves making a specific arrangement and that incurs a cost. The ERG noted that including children aged 4–5 years in a universal scheme may prove challenging, because they are no longer in contact with health visitors. The ERG discussed the feasibility of providing the supplements through the school nursing service. However, the model assumes that the costs of including this age group would be the same as for other children already engaged in the scheme. If it proves more expensive to reach this group then the overall cost effectiveness of the scheme will be affected.

3.18 The ERG noted that NICE's guideline on vitamin D makes recommendations on increasing uptake of supplements among pregnant and breastfeeding women and infants and children aged under 5. It also noted NICE's recommendations on encouraging the uptake of folic acid supplements among women planning and in the early stages of pregnancy (see NICE's guideline on maternal and child nutrition).

3.19 The model assumes that coupons would still be needed to get Healthy Start supplements. But the coupons have been a source of confusion (Healthy Start: understanding the use of vouchers and vitamins University of Dundee) and may not be needed for a universal scheme. A sensitivity analysis was carried out to test the impact on cost effectiveness of varying the cost of issuing coupons. This is discussed on page 46 of the modelling report.

3.20 The ERG stressed the importance of addressing gaps in the evidence to accurately judge the cost effectiveness of adding any new group to a universal Healthy Start vitamin scheme. Evidence is lacking on the most effective and cost‑effective routes for including women planning a pregnancy and those in the first 10 weeks of pregnancy. Evidence is also lacking on infants aged 0–6 months and children aged 4–5 years. The ERG agreed that pilot testing a universal scheme with these groups might help to fill these gaps in the data. Members also noted that various universal schemes are currently being tested but data were not available in time to inform the cost effectiveness model.

Potential adverse effects

3.21 The model assumed that offering the Healthy Start supplements universally would do no harm. The ERG discussed the possibility of over‑supplementation with vitamins A and D in infants aged 0–6 months who are formula‑fed. They noted that the Committee on Toxicology has considered this for vitamin A. The Committee concluded that, given the relatively low dosage contained in Healthy Start supplements, if infants did exceed the tolerable upper limit, it would only be by very small amounts. (Statement on the potential risks from high levels of vitamin A in the infant diet Food Standards Agency Committee on Toxicity.) Members agreed that while the potential risk may be greater for vitamin A than vitamin D, Healthy Start supplements were unlikely to do any harm. Members noted the importance of parents being aware that they should not give other supplements in addition to Healthy Start vitamins.

Increasing uptake

3.22 Getting all subgroups in the model to take up the offer of supplements, in particular, those who are not already taking other supplements they have bought or been prescribed, is key to cost effectiveness. Current uptake is very low (around 3–10%) and has been linked to levels of awareness and distribution. (See section 4 and NICE's guidance on vitamin D: increasing supplement use among at-risk groups.)

3.23 It was not possible to model any 'spill over effect' whereby a universal scheme might increase uptake among the current target group. But the ERG noted that a universal scheme is likely to make the supplements more accessible. For example, it might mean health professionals are trained to routinely discuss the supplements with those who are eligible. In addition, a universal offering may involve more widespread publicity, so further raising awareness of the benefits of the supplements and their availability. It may also overcome any 'stigma' attached to the current offering by making use of the supplements the norm (rather than linking it to income). This, in turn, could increase uptake because it is the accepted thing to do.

Health inequalities

3.24 The ERG discussed whether or not a universal scheme would encourage those most in need to take up the offer of supplements, because it is generally accepted that the 'worried well' are most likely to take them. However, this did not appear to be the case in Birmingham where Healthy Start supplements were offered to all pregnant women and for 12 months after they had given birth. They were also offered to all infants and children aged from 2–4 weeks up to the age of 4 years. Uptake increased to 17% of those who were eligible, but the number of cases of symptomatic vitamin D deficiency in children under 5 dropped by 59%. This may not have been entirely due to uptake of Healthy Start supplements: the associated publicity may have increased uptake of other supplements too. But it does suggest that Healthy Start supplements were being taken by those in most need of them (Vitamin D supplementation: putting recommendations into practice McGee and Shaw 2013).

3.25 The ERG agreed that those in need would benefit the most from a universal scheme. This includes asylum seekers and people who are on a low income but who are not eligible for the current scheme. It also includes women and children from black and minority ethnic groups – because many of them are at increased risk of vitamin D deficiency but they do not all currently qualify for the supplements.

3.26 The Group noted that a universal scheme may encourage greater uptake by people whose first language is not English and among those with poor literacy skills. This is because it would not involve a complicated assessment process and may not involve redeeming coupons.

Other policy issues

3.27 The ERG noted that there may be other cost effective ways to improve the vitamin D status of at‑risk groups. This includes improving awareness of this issue among health professionals and the public. Members noted that NICE has published a guideline on vitamin D: increasing supplement use among at-risk groups.

3.28 The ERG was aware of other options for increasing the folate status of women planning, or in the early stages of, a pregnancy. For example, members noted SACN's proposal to fortify flour with folic acid in the UK supported by existing advice on supplementation ('SACN report to CMO on folic acid and colorectal cancer risk'). They agreed that it may be helpful to compare the cost effectiveness of other options with universal provision of Healthy Start supplements.

3.29 If flour was fortified with folic acid this would reduce the proportion of women at risk of having a pregnancy affected by a neural tube defect. This, in turn, would affect the cost effectiveness of the Healthy Start supplements as estimated in the model for this report.

3.30 Healthy Start supplements are only 1 aspect of the Healthy Start programme and uptake of the food vouchers is high (see section 4). In addition, the programme receives positive feedback ('Understanding the use of vouchers and vitamins'). Although the cost effectiveness of the wider programme was beyond the scope of this work, members felt its contribution to the health of women of child‑bearing age and young children from low income families should not be overlooked.

[6] The adequacy of vitamin or mineral intake can be expressed as the proportion of individuals with intakes below the lower reference nutrient intake (LRNI). This is the level considered likely to be sufficient for only 2.5% of the population.

[7] In the modelling report, symptomatic vitamin D deficiency is defined as a combination of conditions such as bow legs, rickety rosary, tetany, convulsions due to hypocalcaemia, radiological evidence, biochemistry results such as raised alkaline phosphatase (ALP) with or without high parathyroid hormone levels, or low levels (less than 25 nmol/litre) of 25‑hydroxycholecalciferol (25OHC). This is the definition used in the 2 papers (Moy et al. 2012 and Zipitis et al. 2006) from which effectiveness and cost data were obtained.