Keywords
human milk, diet, neonatal, nutrition, cost-effectiveness, budget impact, preterm infants
This article is included in the Agriculture, Food and Nutrition gateway.
human milk, diet, neonatal, nutrition, cost-effectiveness, budget impact, preterm infants
Babies born before 30 weeks’ gestation are at increased risk of major clinical complications including necrotising enterocolitis (NEC), sepsis, and mortality1. The clinical management of preterm babies is complicated by their having greater nutritional requirements than full-term babies. In many cases, the mothers’ own milk (MOM) is not sufficient – in volume or nutritional content – to meet preterm babies’ needs. Consequently, both preterm formulas and milk fortifiers are used to feed preterm infants.
In England, routinely used fortifiers and formulas are derived from cow milk. The use of cow milk-derived fortifier (CMDF) in the diet of preterm infants has been shown to be associated with several adverse health outcomes2. Clinical trials have demonstrated that an exclusive human milk diet (EHMD), based on a MOM alongside fortifiers and formulas manufactured from donor human milk, may be clinically beneficial3,4. An EHMD has been associated with reduced risk of negative sequelae such as NEC, sepsis, neurodevelopmental problems, and lung disease5,6.
A randomised controlled trial was recently completed in England, sponsored by Newcastle Hospitals NHS Foundation Trust. The Interactions between the diet and gut microbes and metabolism in preterm infants (INDIGO) study sought to evaluate EHMDs in the English setting in terms of its impact on gut bacteria and body composition7. The INDIGO trial also recorded data relating to health care resource use and clinical endpoints.
An EHMD, where human milk-derived fortifier (HMDF) and formula are provided (where MOM is insufficient for the preterm infant’s nutritional needs), is likely to be associated with higher upfront costs for the provision of nutrition. However, the major cost of neonatal care in England is attributable to time spent in a neonatal unit (NNU). If an EHMD reduced the time spent in the NNU, it could reduce costs overall.
Previous studies have evaluated the cost-effectiveness of an EHMD for low birth weight babies in the United States and found that it is likely to reduce mortality and reduce costs by reducing adverse clinical events8–10. However, there are important differences between the United States and the National Health Service (NHS) context in England, which mean that the findings may not be applicable. No previous studies have estimated the cost-effectiveness of an EHMD for low birth weight babies in England.
The aim of this analysis is to estimate the expected cost-effectiveness of an EHMD for preterm babies in England, and the budget impact of adopting its use in practice. The analysis will use a modelling approach based on the most relevant data available.
The population will be babies born in England before 30 weeks’ gestation, which aligns with the inclusion criteria used in the INDIGO trial. The population will represent a complete cohort of babies admitted to NNUs in England within one year.
Babies in the intervention arm are fed with MOM, supplemented with HMDFs (Humavant®+6 human milk fortifier [human, pasteurized], Prolacta Bioscience) with or without human milk-derived ready-to-feed preterm formula (Humavant® RTF 26 human milk-based premature infant formula, Prolacta Bioscience). The intervention arm is henceforth referred to as EHMD.
Babies in the comparator arm are fed with MOM, supplemented with CMDFs with or without cow milk-derived ready-to-feed formula. This comparator is intended to represent usual care in England, though usual care can vary between hospitals.
The cost-effectiveness analysis will estimate the cost per life-year associated with the intervention and comparator, using the best available evidence. If an EHMD is associated with improved outcomes and greater costs, its cost-effectiveness will be estimated as the cost per life year gained. This analysis will be conducted from the perspective of the NHS in England.
A secondary analysis will consider disaggregated outcomes in the form of a cost-consequence analysis. These outcomes will include counts of key events including death and several diagnostic indicators as described below.
As with the cost-effectiveness analysis, the budget impact of an EHMD will be estimated from the perspective of the NHS in England. This will be summarised as the total incremental cost based on health care costs associated with nutritional provision, and complications that incur service use. Costs will also be presented in a disaggregated form to guide decision-making at different levels (e.g. national and local).
The time horizon for the analysis will be two years from baseline, where baseline is initial admission to an NNU. Costs will be discounted at an annual rate of 3.5% for the cost-effectiveness analysis in accordance with methodological guidance published by the National Institute for Health and Care Excellence (NICE). Discounting will not be applied for the budget impact analysis.
The overall approach for the analysis will be a model-based cost-effectiveness analysis. We will construct an individual sampling model to simulate clinical pathways and disease events for individual babies. The study is informed by published methods and reporting guidance, as set out in principles of good practice in state-transition modelling, budget impact analysis, and reporting for economic evaluations of health interventions11–14. The model will be developed using Microsoft Excel (Microsoft 365 version).
We will develop a probabilistic discrete-time state-transition microsimulation. The cycle length for the model will be one day. We will conduct 10,000 Monte Carlo simulations for the purpose of probabilistic sensitivity analysis. Each simulation will count the occurrence of events and sum costs over the time horizon.
The state-based transition model will have seven states, made up of four levels of neonatal care – intensive, high dependency, special, and transitional – inpatient hospital care, home, and death, as shown in Figure 1. Each state will be associated with a per-cycle cost. Each day in a neonatal care state will also be associated with a cost of nutrition.
Informed by the modelling exercise reported by Seaton et al.15, we will assume that infants born before 30 weeks’ gestation are transferred to one of three levels of neonatal care: intensive care, high dependency care, or special care, and that subsequent transitions are to lower levels of dependency. While this may not always be the case in practice, the key driver of health care costs is likely to be length of stay, rather than the specific pathway, and so we do not anticipate that this simplifying assumption will introduce substantial bias to our cost estimates.
Transitions are modelled from any neonatal care state to any post-discharge state. An unpopulated transition matrix is shown in Table 1.
Black cells represent transitions with zero probability. White cells represent transitions with positive probability. Grey cells represent the probability of no transition.
To: From: | Intensive care | High dependency care | Special care | Transitional care | Inpatient care | Home | Dead |
---|---|---|---|---|---|---|---|
Intensive care | |||||||
High dependency care | |||||||
Special care | |||||||
Transitional care | |||||||
Inpatient care | |||||||
Home | |||||||
Dead |
A set of events can occur before a baby is discharged from neonatal care. Our model will include the following events:
Surgical treatment for NEC
Diagnosis of late-onset sepsis
Diagnosis of short bowel syndrome
Diagnosis of bronchopulmonary dysplasia (BPD)
Diagnosis of retinopathy of prematurity (ROP)
Diagnosis of neurosensory impairment
The probability of these events occurring will be assumed to be fixed across the different levels of care but to be potentially co-dependent on other events. For instance, the probability of short bowel syndrome and BPD will be associated with the occurrence and treatment of NEC. Stochastic occurrence of all possible events will be recorded within each cycle of each simulation. Each event will be associated with a cost, if relevant.
Table 2 shows the list of parameters that will be required by the model and their candidate sources. Transition probabilities, event probabilities, and diet-specific costs will depend on treatment allocation.
Parameter | Anticipated source(s) |
---|---|
Baseline characteristics | |
Population size | NNRD |
Birth weight | NNRD |
Gestation length (in weeks) | NNRD |
Initial state | NNRD |
Transition probabilities (dependent on allocation | |
From intensive care | |
To high dependency care | INDIGO |
To special care | INDIGO |
To transitional care | INDIGO |
To inpatient care | INDIGO |
To home | INDIGO |
To dead | INDIGO, NNRD, literature15 |
From high dependency care | |
To special care | INDIGO |
To transitional care | INDIGO |
To inpatient care | INDIGO |
To home | INDIGO |
To dead | INDIGO, NNRD, literature15 |
From special care | |
To transitional care | INDIGO |
To inpatient care | INDIGO |
To home | INDIGO |
To dead | INDIGO, NNRD, literature15 |
From transitional care | |
To inpatient care | INDIGO |
To home | INDIGO |
To dead | INDIGO, NNRD, literature |
From inpatient care | |
To home | INDIGO, NNRD, literature |
To dead | INDIGO, NNRD, literature |
From home | |
To inpatient care | Literature |
To dead | Literature |
Event probabilities during NNU (dependent on allocation) | |
Surgical treatment of NEC | Literature2,16 |
Diagnosis of late onset sepsis | Literature2,6 |
Diagnosis of short bowel syndrome (following NEC) | Literature17 |
Diagnosis of BPD | Literature2,18 |
Diagnosis of ROP | Literature19,20 |
Diagnosis of neurosensory impairment | Literature21 |
Resource use (dependent on allocation) | |
Humavant+6 quantity per day | INDIGO |
Humavant RTF 26 quantity per day | INDIGO |
Formula quantity per day | INDIGO, literature2 |
Parenteral nutrition | INDIGO, literature |
Humavant+6 price | Provided by Prolacta Bioscience |
Humavant RTF 26 price | Provided by Prolacta Bioscience |
Intensive care day | INDIGO, NHS Reference Costs |
High dependency care day | INDIGO, NHS Reference Costs |
Special care day | INDIGO, NHS Reference Costs |
Transitional care day | INDIGO, NHS Reference Costs |
Inpatient care day | INDIGO, NHS Reference Costs |
Surgical interventions | INDIGO, NHS Reference Costs |
Abbreviations: NNRD – National Neonatal Research Database; INDIGO – Interactions between the diet and gut microbes and metabolism in preterm infants (study); NNU – neonatal unit; NHS – National Health Service; NEC – necrotising enterocolitis; BPD – bronchopulmonary dysplasia; ROP – retinopathy of prematurity; RTF – ready-to-feed
As part of the INDIGO trial, data were collected for participants, both directly and through the National Neonatal Research Database (NNRD). The variables available from the INDIGO trial are shown in Table 3.
Collection and analysis of variables as part of the INDIGO study was approved by the North East –Tyne & Wear South Research Ethics Committee (REC reference 17/NE/0169).
The key driver of total costs is likely to be the length of stay in the NNU. The INDIGO data will be used to estimate daily transition probabilities between different levels of care, assuming that babies are admitted to the highest level of care observed and are discharged from the lowest level of care observed, where intensive care > high dependency care > special care > transitional care. As described above, we do not anticipate that this assumption will introduce substantial bias to our cost estimates. Each transition probability will be derived from the rate at which babies leave each state.
The INDIGO data will also be used to estimate the cost of nutrition associated with each comparator, based on the quantity of Humavant+6 fortifier, Humavant RTF 26 premature infant formula, and other formula provided.
Key clinical inputs for this project will be sought through collaboration with clinical experts and from existing publications of previous research. Published sources used will include studies focusing on the prevalence and prognosis of complications associated with very premature babies (for example, (e.g. 21), as well as the outcomes of procedures (e.g. surgery) used to address these complications (e.g. 16). We will source papers that report estimates that most closely correspond to parameters required by our model, will use evidence from England wherever available, and will also prioritise more recent data over older data.
We will use NNRD data to define the population and to support external validation of our model. The extracted data items will be at the individual level, as described in Table 4.
The size of the population will be determined by the NNRD population, which we will assume to be equal to the number of eligible babies born in England for the one-year period from 1 January 2019 to 31 December 2019. Each baby simulated by the model will be attributed a birth weight and gestation length at birth, which will be used to determine the amount of feed required. The comparator group will be simulated to be of the same size and birth characteristics. The NNRD data will also define the proportion of babies allocated to intensive, high dependency, or special care at initial admission to the NNU.
We will compare our estimates with nationally representative data from NNRD to externally validate the estimates of our model with respect to clinical outcomes and resource use.
An application has been submitted to a national Research Ethics Committee for the use of NNRD data for the budget impact analysis. This study will involve analysis of data already collected by the NNRD, with no novel data collection or identifiable information used.
The time horizon for both the cost-effectiveness analysis and the budget impact analysis will be two years following admission to the NNU. Costs will be calculated from the perspective of the NHS using a combination of data from the INDIGO clinical trial and NHS Reference Costs.
The key outcome of the cost-effectiveness model will be the incremental cost per life-year gained for preterm babies fed with an EHMD, relative to those receiving standard care. Costs considered will include upfront costs associated with providing an EHMD, as well as costs of health care resource use associated with common clinical complications in preterm babies, including BPD and ROP. Only directly incurred costs associated with these clinical events will be included.
The budget impact will be calculated as the difference in total cost between a scenario where babies are fed an EHMD, and one in which CMDFs (with or without cow milk-derived read-to-feed formula) are used. Cost items included will be the same as those for the cost-effectiveness model.
The budget impact analysis will adopt a payer (NHS) perspective. The time horizon will be two years post-admission. The model will evaluate additional costs arising from the switch to a more expensive feeding regime against potential reductions in costs associated with lower health care resource use as a result of improved health outcomes and lower rates of complications (if observed).
The increase in costs associated with an EHMD consist of the additional (total) cost of human milk supplementation, which in turn will depend on the additional cost per day of human milk supplementation, the length of time supplementation is required, and the size of the target population. Cost reductions may arise from improved health outcomes for very preterm babies, with reductions in morbidity, surgical procedures (and associated complications), along with reduced length of stay in enhanced care facilities.
The overall budget impact will be presented as a net cost (or saving) to NHS England.
As a sensitivity analysis, we will conduct a within-trial analysis using only INDIGO trial data in combination with unit cost estimates.
Estimates generated by our model will be compared to estimates from NNRD as a means of externally validating our model. We will compare the following between NNRD and our model’s estimates for the usual care arm:
Findings of this study will be published in a peer-reviewed journal or other publishing platform.
Study status. The study is in the initial planning stages, with early development of the model. The researchers have not yet accessed any data to be analysed as part of the study. Funding for the study is secured and the study has provisional approval from the Neonatal Data Analysis Unit (subject to ethical approval) to access NNRD data.
This study is the first economic evaluation of EHMD use for very preterm babies in England. Given the potential for EHMD to reduce the incidence of health complications associated with significant costs to the health system – as shown in a previous evaluation for the United States – it may represent significantly reduced costs for the NHS and alleviate pressure on neonatal care resources. Beyond cost considerations, this intervention has the potential to bring about significant improvements in quality of life for preterm babies and, by association, their carers.
By using the results of a recent clinical trial for an EHMD in England, as well as costs specific to the English setting, the findings here will be highly relevant to decision-making about whether to use EHMD in the NHS. The inclusion of both a cost-consequence and budget impact analysis will allow us to illustrate a more comprehensive picture of the overall impact of an EHMD on the NHS.
We are grateful to Prof Nicholas Embleton for the expert opinion provided during the development of our study. Any errors or omissions are our own.
Views | Downloads | |
---|---|---|
F1000Research | - | - |
PubMed Central
Data from PMC are received and updated monthly.
|
- | - |
Is the rationale for, and objectives of, the study clearly described?
Yes
Is the study design appropriate for the research question?
Yes
Are sufficient details of the methods provided to allow replication by others?
Partly
Are the datasets clearly presented in a useable and accessible format?
Partly
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Nutritional Epidemiology
Is the rationale for, and objectives of, the study clearly described?
Yes
Is the study design appropriate for the research question?
Yes
Are sufficient details of the methods provided to allow replication by others?
Yes
Are the datasets clearly presented in a useable and accessible format?
Yes
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Health Economics
Alongside their report, reviewers assign a status to the article:
Invited Reviewers | ||
---|---|---|
1 | 2 | |
Version 1 13 Jan 21 |
read | read |
Provide sufficient details of any financial or non-financial competing interests to enable users to assess whether your comments might lead a reasonable person to question your impartiality. Consider the following examples, but note that this is not an exhaustive list:
Sign up for content alerts and receive a weekly or monthly email with all newly published articles
Already registered? Sign in
The email address should be the one you originally registered with F1000.
You registered with F1000 via Google, so we cannot reset your password.
To sign in, please click here.
If you still need help with your Google account password, please click here.
You registered with F1000 via Facebook, so we cannot reset your password.
To sign in, please click here.
If you still need help with your Facebook account password, please click here.
If your email address is registered with us, we will email you instructions to reset your password.
If you think you should have received this email but it has not arrived, please check your spam filters and/or contact for further assistance.
Comments on this article Comments (0)