Ten guiding principles for model write-ups and economic sections of HTA submissions

The following pointers aim to provide practical guidance to strengthen the communication of your economic case, avoid the introduction of mistakes and make the deadline.

1. Validate your model first. Then get someone else to validate it.
2. Agree your base case assumptions and scenario analyses before you start. Iterative changes to analyses increase the likelihood of errors in the write-up.
3. Clarity. Health economists love their methods for technical modelling and statistical analysis. However, some reviewers are not health economists. The first test of a good model write-up is whether the non-health economist in the office can understand the model write up.
4. Where flow (that is the reader’s ability to follow the content) is interrupted by a particularly deep dive into the methods of a statistical analysis, consider moving the detail to an appendix.
5. Stick to your sections. Data should be in the data section, results in the results section … It is surprising how often text can sneak into the wrong place. Content in the wrong section makes a write up hard to follow.
6. All key data sources should be defined up front. As well as trial data, systematic reviews and meta-analysis, this should include expert interviews, advisory panels, questionnaires, and databases. Sufficient information should be provided to show that the methods of data collection were appropriate, unbiased and where required systematic and reproducible.
7. There needs to be a clear, robust rationale for every assumption. It’s prudent to document assumptions as they’re agreed, otherwise this can be the most time-consuming aspect of the write-up.
8. Key areas of uncertainty should be presented in simple scenario and threshold analyses to clarify the effect of the uncertainty. Make sure no key areas of uncertainty have been omitted or that any scenario analyses leave you none the wiser.
9. Be tight on review and feedback. In general, feedback should focus on content rather than style. It should be limited to a manageable number of rounds of review (two to four is typical); there is a point at which iterative rounds of review negatively affect the style, clarity and flow of the document. Reviewers in a single organisation should also consolidate their feedback. Unconsolidated feedback relies on the writer to guess whose feedback takes precedence.
10. Avoid last minute changes prior to an HTA submission. Some changes are unavoidable but if at all possible, changes should be avoided in the last week and certainly the last few days prior to submission. Models often have hundreds of variables, multiple settings for different analyses and the submission document itself is large. Unplanned, rapid changes run the risk of introducing errors.