Free scoping read. A written assessment of your manuscript in two business days, at no cost.

Start one

Home/Insights/Methodology reference/Secondary research methods

Secondary research methods

Using data someone else collected is fast and cheap, and the constraint is permanent: you cannot add the variable that was not recorded.

Foundations

What it covers

Analysis of existing datasets — registries, cohorts, administrative and claims data, electronic health records, trial data shared for reuse — and evidence synthesis, where the units of analysis are published studies.

The structural limitations

The variables are fixed. A confounder not recorded cannot be adjusted for, and no amount of statistical sophistication recovers it.

The data were collected for another purpose. Administrative data exist for billing; coding reflects reimbursement incentives rather than clinical precision. Diagnostic codes are a proxy for diagnosis, not a record of it.

Quality is inherited. Measurement error, missingness and selection in the original collection pass straight into the reanalysis.

Multiplicity is easy and invisible. A large dataset permits many analyses at no cost, which makes a pre-specified analysis plan more important here than in primary research, not less.

Doing it defensibly

Name the dataset, its version and the extraction date. Give the exact inclusion criteria and the code lists used to define exposures and outcomes — a diagnosis defined by a code list is only reproducible if the list is published.

State the data access approvals. Register the analysis plan where possible. RECORD extends STROBE specifically for routinely collected health data and is the checklist to write to.

In practice

Fix, flag or query.

The distinction that keeps an edit defensible: what an editor may change silently, what must be shown to the author, and what only the author can answer.

  • Cannot be remedied: a confounder the original collection did not record.
  • Publish: the code lists defining every exposure and outcome.
  • State: dataset version and extraction date, since databases are revised.

Reporting standards

  • RECORD Statement, for routinely collected health data
  • STROBE Statement, for observational designs

Reporting guidelines named here are hosted by the EQUATOR Network. Writing to the checklist for your design before submission is the cheapest available improvement to a methods section.

Get a free quote

Regulations

Check your project against this rule

  1. 1Your work
  2. 2Your details
  3. 3Send
What do you need?

Material you send is held under signed confidentiality, with access limited to the assigned team and encrypted transfer. It is not used to train anything and is not passed to third-party tools. Ask for the agreement before you send anything you would rather cover first.