A practical guide for teams working on hypothyroidism, hyperthyroidism and thyroid nodules: digital health products, clinics and laboratories, pharma-supported programmes, insurers and employer health programmes.
Thyroid disease looks simple from the outside. A single blood test, thyroid-stimulating hormone (TSH), is checked millions of times a year, and a single tablet treats most underactive thyroids. Yet thyroid data is full of traps: reference ranges that shift with the laboratory and with pregnancy, supplements that distort results, medicines that change thyroid function, and imaging that can find far more than it should.
These conditions are common, and far more common in women. A global review found overt hypothyroidism (an underactive thyroid) in 0.2% to 5.3% of people in Europe and 0.3% to 3.7% in the United States, and overt hyperthyroidism (an overactive thyroid) in 0.2% to 1.3% in regions with enough iodine. Hypothyroidism is about ten times more common in women than in men.1 Iodine still matters globally: 21 countries had insufficient iodine intake in 2021, down from 113 in 1993.2
Sources: Taylor PN et al., Nature Reviews Endocrinology, 2018; Iodine Global Network scorecard.
This guide draws mainly on the National Institute for Health and Care Excellence (NICE) guideline on thyroid disease (NG145), American Thyroid Association (ATA) guidance and regulatory safety advice. We wrote it for four groups of readers. If you are building a digital health product, it sets out the data decisions to get right before you scale. If you run a clinic or laboratory, it covers what it takes to make results comparable and safe to act on. If you lead a pharma-supported programme, it explains what your data needs in order to show outcomes. If you are an insurer or run an employer health programme, it shows why more testing is not always better. Thyroid cancer treatment and childhood thyroid disease have their own considerations and are outside the scope of this guide.
A person with thyroid disease may have blood tests from several laboratories, each with its own reference range. Their ultrasound and biopsy reports sit as PDFs, their levothyroxine prescriptions change dose and sometimes brand, and the supplements and other medicines that affect their results are often recorded nowhere at all. Their GP (general practitioner), endocrinologist and, during pregnancy, their obstetric team each keep their own notes.
Each source holds part of the story. The value comes from bringing them together in a way that stays accurate over years of care.
The twelve points below are grouped into four themes: getting the clinical picture right, making the data usable, connecting the longer story, and building for trust and scale.
NICE advises starting with TSH alone when pituitary disease is not suspected, adding free thyroxine (FT4) if TSH is above the reference range, and both FT4 and free triiodothyronine (FT3) if TSH is below it. It also advises against testing during an acute illness unless the illness itself may be caused by thyroid dysfunction, because illness can affect the results.3
Simplified from NICE guideline NG145, recommendation 1.2.
For a data product, this means a result needs its context: which tests were requested and why, and whether the person was unwell at the time.
What good looks like: each result stored with the reason for testing and the clinical context, logic that follows the testing sequence rather than reacting to any single value, and results taken during acute illness flagged.
Matters most to: Startups, Clinics and laboratories
There is no single "normal" TSH. Reference ranges differ between laboratories and assay platforms. A reference method now exists for FT4, but commercial assays are not yet uniformly recalibrated to it, and TSH can only be harmonised statistically.4, 5 Pregnancy changes the picture further: the ATA's 2017 pregnancy guideline recommends population-based, trimester-specific TSH ranges, with an upper limit of about 4.0 mU/L suggested in early pregnancy where local ranges are not available.6
What good looks like: every result stored with the reference range printed by its laboratory, the assay platform, and pregnancy status and trimester where relevant, so results are never compared against a single universal range.
Matters most to: Clinics and laboratories, Startups, Pharma programmes
Some results are wrong for reasons that have nothing to do with the thyroid. The US Food and Drug Administration (FDA) warned in 2017, and again in 2019, that biotin supplements can cause falsely high or low results on some laboratory tests.7 In thyroid testing, biotin can lower TSH and raise FT4 on some platforms, mimicking an overactive thyroid, and the size of the effect depends on the analyser used.8
Some medicines genuinely change thyroid function. Lithium causes overt hypothyroidism in 5% to 15% of people treated, amiodarone can cause both over- and underactivity, and immune checkpoint inhibitors used in cancer treatment are a newer cause of thyroid dysfunction.1
What good looks like: supplements and medicines captured alongside every result, biotin use flagged before testing, and unexpected results checked for these causes before they trigger alerts or changes.
Matters most to: Clinics and laboratories, Startups
FT4 is reported in pmol/L in some countries and ng/dL in others, and TSH appears as mU/L or µIU/mL, which are numerically equivalent. Because assays differ, the same blood sample can produce different numbers on different platforms, and work on standardising FT4 has shown that recalibration could shift results substantially for some assays.5 When results from several laboratories sit in one record without their units, platforms and ranges, trends become unreliable.
What good looks like: every result stored with its original value, unit, assay platform and reference range, converted to a consistent unit for analysis, and traceable to its source laboratory.
Matters most to: Startups, Clinics and laboratories, Pharma programmes
"Hypothyroidism" or "hyperthyroidism" says what is happening, not why, and the cause changes treatment and follow-up. NICE recommends measuring TSH receptor antibodies (TRAbs) to confirm Graves' disease, the most common cause of an overactive thyroid.3 Other causes, from thyroiditis to medicine-induced disease, need to be recorded with the evidence behind them.
What good looks like: diagnoses coded to their cause where known, antibody results stored with their assay and date, and the evidence for each diagnosis traceable in the record.
Matters most to: Clinics and laboratories, Pharma programmes
Thyroid nodules are assessed with ultrasound risk scoring systems. Several are in use, known as Thyroid Imaging Reporting and Data Systems (TI-RADS), including the American College of Radiology system (ACR TI-RADS), the European EU-TIRADS, the Korean K-TIRADS and the Chinese C-TIRADS, which score nodules differently.9, 10 When a biopsy is taken, cytology is reported using the Bethesda System for Reporting Thyroid Cytopathology, whose third edition (2023) keeps six diagnostic categories.11
A nodule scored under one system cannot be compared directly with a score under another, and most of this detail arrives as free-text reports.
What good looks like: ultrasound and cytology reports structured into findings (nodule size, location, score, Bethesda category), with the scoring system and version recorded and the original report kept for reference.
Matters most to: Clinics and laboratories, Startups
Levothyroxine looks like a simple, stable medicine, but the details matter. It is best taken on an empty stomach, and calcium, iron, antacids and proton pump inhibitors can reduce its absorption.12 Some people notice symptoms after switching between products, and the UK Medicines and Healthcare products Regulatory Agency (MHRA) advises testing thyroid function if this happens and considering keeping them on one product they tolerate well.13
What good looks like: each prescription recorded with dose, product or brand, formulation and dates, mapped to medical coding standards, and linked to the results that followed each change and to medicines that affect absorption.
Matters most to: Clinics and laboratories, Pharma programmes
Subclinical hypothyroidism, where TSH is raised but FT4 is normal, is judged on repeated results rather than a single test. NICE suggests considering levothyroxine for adults with a TSH of 10 mU/L or higher on two separate occasions three months apart, and considering a six-month trial for adults under 65 with symptoms whose TSH is above the reference range but below 10 mU/L on two occasions three months apart.3
What good looks like: results stored with dates so repeat testing intervals are visible, logic that recognises a confirmed pattern rather than one abnormal value, and age and symptoms recorded alongside.
Matters most to: Startups, Insurers and employers, Clinics and laboratories
More testing is not always better. In South Korea, widespread ultrasound screening meant that by 2011 thyroid cancer diagnoses were 15 times the 1993 rate, while deaths from thyroid cancer stayed stable.14 Many of the cancers found would never have caused harm, but the people diagnosed still faced surgery, lifelong medication and anxiety.
This matters for employer health checks, insurer screening benefits and consumer testing products. Routine thyroid ultrasound or broad test panels in people without symptoms can generate findings that lead to more investigation without better outcomes.
What good looks like: testing offered for a clear clinical reason, outcomes tracked alongside diagnoses, and programme metrics that measure benefit to people rather than the number of conditions found.
Matters most to: Insurers and employers, Startups, Pharma programmes
A useful thyroid summary shows TSH and FT4 trends against their own laboratory ranges, the cause of disease where known, current medicines and doses with their change history, pregnancy status, and nodule findings with their scoring system.
Some situations need a clear route to action. Antithyroid medicines used for an overactive thyroid can rarely cause agranulocytosis, a dangerous fall in white blood cells, and people taking them are advised to stop the medicine and seek urgent medical advice if they develop a sore throat, fever, mouth ulcers or unexplained bruising.15, 3 Pregnancy, or planning a pregnancy, in someone with thyroid disease also needs prompt review.
What good looks like: an exportable clinician summary in a consistent format, escalation rules agreed with clinicians and tested, and a record that can move into a clinic's own system.
Matters most to: Clinics and laboratories, Startups
Software intended for a medical purpose, without being part of a hardware device, is described internationally as Software as a Medical Device.16 Thyroid tools cross into this territory when they interpret results to suggest a diagnosis, recommend levothyroxine dose changes, or score nodules for cancer risk. Regulators in Singapore, the UK, the European Union, Australia, the United States and many other markets each apply their own rules.
What good looks like: an intended purpose written down early, claims reviewed against the rules in each market you serve, and validation evidence gathered from the start rather than retrofitted.
Matters most to: Startups, Pharma programmes
Thyroid data is health data protected under laws such as the EU General Data Protection Regulation (GDPR), Singapore's Personal Data Protection Act (PDPA) and, where it applies, the Health Insurance Portability and Accountability Act (HIPAA) in the United States. When insurers and employers are involved, people need confidence that data shared for their care or wellbeing will not be used against them. Consent should be specific about how data will be used, and data used beyond direct care should be pseudonymised, with access controls and audit trails.
Thyroid disease is usually lifelong. Structuring data to interoperability standards such as FHIR (Fast Healthcare Interoperability Resources), with results and medications mapped to medical coding standards, keeps it useful when people change clinics, laboratories, insurers or countries.
What good looks like: granular consent recorded alongside the data, clear separation from employment or underwriting decisions, pseudonymisation for secondary use, and standard-format export with provenance intact.
Matters most to: Insurers and employers, Startups, Clinics and laboratories, Pharma programmes
You do not need to solve all twelve at once. The right starting point depends on what you are building and what your data needs to do.
Some of the twelve points are data problems, and some are questions of design, strategy and delivery. We support both, so you can come to us with the whole picture rather than splitting it across several partners.
Jonda Health is ISO 27001 certified, and our platform is designed to comply with HIPAA, GDPR and PDPA.
This guide is for teams designing and running digital health tools and programmes. It is not clinical advice for individuals.
