Parameter Sourcing, Reference Cost Lists and Utility Catalogues When Local Data Are Sparse

อ่านฉบับภาษาไทย (Thai version)

Abstract

An economic model needs inputs that answer its own decision question. A published number becomes useful only after its population, outcome, unit and time frame match that question. This article shows how to define parameters before searching and how to select evidence by parameter type. Clinical effects need credible comparative evidence, while baseline risks and resource use need locally relevant populations. Reference cost lists and utility catalogues help locate candidates, but cannot certify their suitability. A transparent evidence ledger records the search, source, applicability, uncertainty and any transformation. An illustrative visit calculation separates changing a price year from changing resource quantities. We then distinguish evidence that is unavailable, inapplicable or imprecise. These problems call for different responses, including alternative sources, justified scenarios and structured elicitation. Research priorities depend on whether resolving uncertainty could change the decision. The aim is a defensible input trail that another analyst can inspect, update and challenge.

🔍 Click to enlarge Click anywhere to close

Write the parameter definition before searching

A parameter is a quantity that controls the model calculation. Define its estimand: the population quantity you want to estimate. Specify population, comparator, outcome, time horizon and analysis perspective before selecting a source. A precise estimate of another quantity is still the wrong input.

For a transition, specify the starting state, destination and cycle length. Distinguish an incidence rate from a risk or prevalent count. For costs, separate resource use from its unit price. For utility, specify the health state, instrument, respondents and country value set.

Choose evidence by the parameter it must support

There is no universal evidence pyramid for every input [1]. Comparative treatment effects need systematic assessment of causal evidence and its certainty. Local baseline risks need representative populations with the correct denominator and follow-up. Local invoices alone cannot establish comparative effectiveness.

For resource use, examine pathways, capacity and practice variation. For unit prices, inspect the included cost components, setting and price year. Utility catalogues require checks of instrument version, health-state definition and valuation population [3]. A foreign tariff imports preferences; a similar diagnosis does not prove transportability.

Keep an evidence ledger that can change the base case

An evidence ledger is a record of each input and why it was chosen. Record the search query, databases or catalogues, search date, source locator and version. Then record the parameter definition, estimate, uncertainty and transformations. Keep source provenance in the analytic record, not merely a bibliography.

Appraise bias and applicability separately. Record units, included overhead, price year and currency for costs. Record instrument, valuation set and population for utilities. For uncertainty, record bounds or a distribution and any dependence on shared data [2].

Mark a compatible source for the base case, meaning the principal analysis. Send a plausible but uncertain transport assumption to a named scenario. Mark an unsupported essential input as research needed. Keep the reason visible so another analyst can revisit it.

Separate price and quantity arithmetic

# Writes: stdout only; coordinator captures results.
source_cost <- 1200
source_cpi <- 100
target_cpi <- 115
visits <- 3
restated <- source_cost * target_cpi / source_cpi
cat("Restated THB per visit:", restated, "\n")
cat("Base-case visit total THB:", restated * visits, "\n")
cat("Two-visit scenario THB:", restated * 2, "\n")
cat("Four-visit scenario THB:", restated * 4, "\n")

During startup - Warning messages:
1: Setting LC_COLLATE=C.UTF-8 failed
2: Setting LC_CTYPE=C.UTF-8 failed
3: Setting LC_MONETARY=C.UTF-8 failed
4: Setting LC_TIME=C.UTF-8 failed
Restated THB per visit: 1380
Base-case visit total THB: 4140
Two-visit scenario THB: 2760
Four-visit scenario THB: 5520

Diagnose the evidence gap before filling it

Unavailable evidence was not located after a documented search. Inapplicable evidence answers a different question or comes from an unsuitable setting. Imprecise evidence targets the right quantity but leaves a wide plausible range. More decimal places fix none of these problems.

Use structured expert elicitation only with a justified protocol and an explicit uncertainty distribution. It is expert judgment, not a newly observed effect. Test plausible alternatives and show whether they reverse the preferred decision. Prioritize research where better information could change action, considering the cost and feasibility of collecting it.

Review incidence in [[incidence-prevalence-and-the-right-denominator]] and certainty in [[evidence-pyramid-versus-grade-certainty]]. Costs and utilities are developed in [[costing-perspective-inflation-discounting]] and [[qaly-daly-and-utility-elicitation]]. Carry uncertainty into [[decision-trees-markov-models-and-psa]]; A14 uses these inputs for conditional headroom pricing.

The series, in learning order

1. Health Economics Series Guide: Read the Evidence Before the Costs [[health-economics-series-guide]]
2. Incidence, Prevalence and the Right Denominator: The Epidemiology Beneath Cost-of-Illness and Markov Models [[incidence-prevalence-and-the-right-denominator]]
3. Design, Model and Estimand: What Each Study Design Can Actually Estimate [[study-design-estimand-and-effect-measures]]
4. Risk Ratio, Odds Ratio, Risk Difference and NNT: One 2×2 Table, Four Answers [[risk-ratio-odds-ratio-risk-difference-nnt]]
5. The Hazard Ratio Is Not a Risk Ratio: Proportional Hazards, Restricted Mean Survival Time and Model Inputs [[hazard-ratio-proportional-hazards-rmst]]
6. Chance, Bias and Confounding: Why an Association Is Not Yet a Treatment Effect [[chance-bias-and-confounding]]
7. Beyond the Evidence Pyramid: GRADE Certainty and Choosing Effectiveness Inputs for Economic Models [[evidence-pyramid-versus-grade-certainty]]
8. From Sensitivity to Cost per Case Detected: Predictive Values, Prevalence and the Economics of Screening [[diagnostic-accuracy-predictive-values-screening]]
9. Whose Cost, in Which Year? Perspective, Opportunity Cost, Inflation and Discounting [[costing-perspective-inflation-discounting]]
10. QALY, DALY and the Utilities Behind Them: What Is Actually Being Measured [[qaly-daly-and-utility-elicitation]]
11. ICER, Dominance and the Cost-Effectiveness Plane: Reading the Thai Threshold of 160,000 Baht per QALY [[icer-dominance-ce-plane-and-thresholds]]
12. Decision Trees, Markov Cohorts and Probabilistic Sensitivity Analysis: How a Model Turns Evidence into a Decision [[decision-trees-markov-models-and-psa]]
13. Cost-Effective Is Not Affordable, and Efficient Is Not Fair: Budget Impact and Equity [[budget-impact-equity-and-limits-of-qaly]]
14. Parameter Sourcing, Reference Cost Lists and Utility Catalogues When Local Data Are Sparse (this article)
15. Early Health Technology Assessment, Headroom Pricing and Benefit-Package Decisions
16. Communicating Health Economic Evidence: From Model Outputs to Policy Briefs

Related in the wiki: Incidence, Prevalence and the Right Denominator: The Epidemiology Beneath Cost-of-Illness and Markov Models · Beyond the Evidence Pyramid: GRADE Certainty and Choosing Effectiveness Inputs for Economic Models · Whose Cost, in Which Year? Perspective, Opportunity Cost, Inflation and Discounting · QALY, DALY and the Utilities Behind Them: What Is Actually Being Measured · Decision Trees, Markov Cohorts and Probabilistic Sensitivity Analysis: How a Model Turns Evidence into a Decision