Technology evaluation methods support document: Health inequalities

Social groups should be stratified by Index of Multiple Deprivation (IMD) quintiles.  

Some inequalities may be concentrated in specific social groups. For example, certain ethnic or inclusion health groups (see NHS England's Equality, diversity and health inequalities section on inclusion health groups). In these cases, other approaches to stratification may be considered if there is evidence-based rationale. In some circumstances, it may be appropriate for a condition to be stratified by 2 characteristics. However, this can lower the quality and interpretability of the data. 

Supplementary indices of the IMD, such as for children or older people, should not be used. The IMD can be used for these groups.  

A full probabilistic sensitivity analysis is not needed for the health inequality analysis. However, sensitivity and scenario analyses should be done for the main sources of uncertainty that determine the health inequality impacts. These will likely include the distributions of disease prevalence and incremental health benefits. Sources of uncertainty should be specified in the submission, whether or not they are measured. A statement should also be included on the likelihood that the technology will increase or decrease the health inequality gap in the general population. 

Health inequalities can occur because of differences in access to care or in health-seeking behaviour. Equal uptake should be assumed across all groups, unless there is evidence to support an alternative distribution. Differences in uptake across groups or technologies should be justified. This should be based on the best available evidence on the technology under evaluation or similar treatments for the same condition. All data sources must be explicitly stated, quality assessed and justified. 

DCEAs should assume that the new technology has no impact on uptake, unless evidence comparing it with existing treatment suggests it would improve access or adherence for certain groups. Impact on uptake should be justified by robust evidence, for example, from prescribing data for similar treatments or conditions.    

The results of the DCEA should not weight the costs or benefits of a technology differently based on the recipients' social characteristics. This includes the use of mathematical functions that use health inequality aversion parameters to reflect social preferences around health inequalities.

Calculating the distribution of net health benefits requires an estimate of the distribution of health opportunity cost. This shows the differences between social groups of: 

• forgone health benefits from displaced interventions when a technology or test is cost increasing, or 

• health benefits from freed-up resources when a technology or test is cost lowering.  

Health opportunity costs should be equally distributed between social groups in the base-case analysis. This should be accompanied by scenario analyses of light and moderate gradients that reflect a higher proportion of displacement in more disadvantaged groups to demonstrate how changing assumptions impacts the results. Alternative gradients should be justified with evidence where possible. The opportunity costs used for all gradients should be expressed in QALYs and presented in a summary table.   

The following outputs should be presented:  

• The distribution of the technology or test's population health benefits across social groups, excluding any health opportunity costs. Evidence on cost-effectiveness for individual subgroups defined by social characteristics will not be considered. 

• Net health benefit, which shows the distribution of health benefits across social groups taking into account health opportunity costs. Net health benefit can be either: 

  • negative, which suggests that a technology's health benefits do not outweigh the health losses caused by displacing other healthcare to fund the technology, or 

  • positive, which suggests that the new technology will increase population health. 

• Descriptive inequality metrics. These can include simple gaps and ratios between the top and bottom of the distribution, or the results of simple regressions that can incorporate information on the net health benefits of groups in the middle of the distribution.