Evidence and measurement · 3 min read

Measuring caregiver burden: what the instruments miss

Every serious instrument for caregiver strain shares one property: it is a snapshot. For a programme director deciding where to spend a limited budget, that property matters more than any of the differences between them.

Evidence and measurement

The instruments are good. The cadence is the problem.

The Zarit Burden Interview has been the reference for caregiver burden since 1980. Twenty-two items, validated across languages and care contexts, and it does what it claims. The PHQ-9 does the same for depressive symptoms. Neither is in question here.

What is in question is how often they are administered. In most programmes the answer is once at intake, and then perhaps annually. A caregiver whose situation deteriorates in March is measured again in November.

That is not a criticism of the instruments. A twenty-two item questionnaire cannot be administered weekly without the response quality collapsing, and nobody has the staff to do it. The cadence is a consequence of the format, not a design flaw.

Evidence and measurement

Why the gap costs more than it looks

Two findings from the literature explain why the interval matters.

Schulz and Beach followed spousal caregivers and found a materially higher four-year mortality risk among those reporting strain. The finding has been reproduced and refined many times since, and the mechanism is not mysterious — chronic stress, disrupted sleep, delayed care for one's own conditions.

Rodakowski and colleagues showed that meaningful caregiver involvement in discharge planning is associated with a substantial reduction in readmission.

Put those together and the shape of the problem appears. The caregiver's condition predicts both their own health outcomes and the patient's. And the standard measurement cadence means a programme learns about deterioration one to two quarters after it began.

Evidence and measurement

What a programme can actually do about it

Three things, in rough order of cost.

  • Shorten the interval for the highest-risk cohort rather than for everyone. Re-administering the ZBI quarterly to the twenty percent of families with a recent hospitalisation costs far less than an annual sweep of everybody, and it catches more.
  • Treat the intake score as a baseline, not as a category. A caregiver who scored 40 at intake and 48 six months later is a different case from one who scored 48 both times, even though a threshold-based system files them identically.
  • Watch the operational traces you already have. Missed appointments, a sudden change in who calls the programme, a family that stops responding — these are not clinical measures and should never be treated as such, but they arrive continuously and they arrive free.

Evidence and measurement

The honest limitation

None of the above replaces a validated instrument, and no programme should present it as though it does. What continuous signals can do is tell you when to administer the instrument — which is a scheduling question, not a clinical one, and a legitimate use of operational data.

The distinction matters for anyone building or buying in this space. A tool that claims to replace the ZBI is making a clinical claim and should be asked for clinical evidence. A tool that helps you decide who to re-assess this month is making an operational claim, and the bar is different.

A measure administered twice a year is a good measure. It is not an early warning system, and the two should not be confused.

Evidence and measurement

Where to read more

The primary sources are worth reading directly rather than through summaries. Zarit's original 1980 paper on burden among caregivers of people with dementia sets out the instrument's reasoning. Schulz and Beach's 1999 work in JAMA is the mortality finding. For programme design specifically, the Administration for Community Living's caregiver support materials describe how respite and assessment are structured under the National Family Caregiver Support Program.