Katz ADL Index

Katz ADL Index in Ageing Research and Economic Evaluation

The Katz ADL Index, published in 1963 as the Index of ADL and often called the Katz Index of Independence in Activities of Daily Living, rates whether an older or chronically ill person can manage six basic functions without help. Health economists meet it as a trial outcome in care for older people and as a basis for dependency states in models. Its judgements of independence carry no preference weights, so a Katz result has to be linked to utilities or costs before it can enter a cost-utility analysis. In a Dutch trial of nurse-led care reported by Suijker and colleagues, the maximum probability that the programme was cost-effective was 0.14 at a willingness to pay of EUR 50,000 per point of a modified Katz score and 0.04 at EUR 50,000 per QALY, which shows the index working both as a natural unit and beside the QALY.

Six functions and the original grades

Katz, Ford, Moskowitz, Jackson and Jaffe developed the Index of ADL to study the results of treatment and prognosis in the elderly and chronically ill. It covers six functions, namely bathing, dressing, toileting, transferring, continence and feeding. The original paper summarised a person's overall performance across the six as a letter grade. Grade A marks independence in all six functions, and the grouping used in later studies treats the final letters as the most dependent, as when Asberg compared grades A to E with grades F and G.

The functions behave as a cumulative scale. In a Swedish study of 100 older patients, Brorsson and Asberg found that the activities ranked on a cumulative scale, meaning that dependence tends to build up across the functions in a consistent order. This ordering is what lets a single letter grade or count stand for a person's pattern of dependence.

A simpler scoring is now common. The Hartford Institute for Geriatric Nursing's Try This guide rates each of the six functions yes or no for independence and adds them up, with 6 indicating full function, 4 moderate impairment and 2 or less severe functional impairment. Other studies use a longer modified index that also covers instrumental activities.

The direction of the scales differs. A reader who assumes that a higher total is better on both the 0 to 6 count and the 15-point modified index will read a deterioration on the modified index as an improvement.

Reliability and prediction in hospital care

Evidence on the index in short-term care links Katz status to outcomes that drive costs. Brorsson and Asberg had enrolled nurses assess 100 aged patients independently in a department of internal medicine in Sweden and found the index reliable. Independent patients had shorter stays and went home more often, and most dependent patients were dead or in institutions a year later, which led the authors to recommend the index as a basic functional measure in short-term care.

Asberg then compared Katz grades with physicians' judgements in 129 patients aged 65 or over on an internal medicine ward, testing both against survival, length of stay and type of discharge. Predictions were similar for a stay under 10 days and for discharge home. For deaths, predictions based on grades F and G, or on grade G alone, had higher sensitivity than physicians' ratings with similar specificity.

For a modeller these findings matter because length of stay, discharge to an institution and death are events that carry large costs and health consequences in models of acute care for older people. A Katz grade recorded at admission is therefore a candidate stratifier for transition probabilities and resource use.

Dependency in economic evaluation and population health

Dependence in the six Katz activities of daily living drives the use of formal and informal care, but Katz status carries no cost or utility of its own. Each economic use of the index therefore brings those values from other evidence.

Trial outcome beside the QALY

Suijker and colleagues ran a cost-effectiveness and cost-utility analysis alongside a cluster randomised trial of nurse-led multifactorial care for people aged 70 or over at increased risk of functional decline, with 1,209 participants in the intervention group and 1,074 in usual care. The modified Katz-ADL index, scored 0 to 15 with higher scores indicating more dependence, was the primary outcome, and QALYs were also measured. After one year there were no significant differences in Katz score or QALYs. Total mean costs were EUR 6,518 in the intervention group against EUR 5,214 under usual care, an adjusted mean difference of EUR 1,457 (95% CI 572 to 2,537), and the authors concluded that implementing the model should not be recommended. A cost per Katz point has no external threshold, so the decision rested on how much a payer would pay for a point of independence.

Dependency states in a model

A Markov model of ageing or chronic illness can define its states by Katz status, for example independent in all six functions, dependent in one or more, institutionalised and dead. Each state then receives a utility and an annual care cost from other studies, because the index supplies neither. The cut point between states is a choice, and moving it changes how many people a treatment keeps out of the dependent state as well as the utility gap between states.

Active life expectancy

Katz and colleagues used life-table methods in 1983 to estimate the expected remaining years of functional well-being, in terms of the activities of daily living, among non-institutionalised older people in Massachusetts in 1974. Active life expectancy was 10 years for those aged 65 to 70 and 2.9 years for those aged 85 or older. It was shorter for the poor than for others, and women had a longer average expected period of dependence than men. The method is close to the life-table approach described under DFLE calculation, with ADL dependence in place of a disabling condition.

Illustrative worked example of Katz dependency states in a model

All numbers here are illustrative. A cohort of 1,000 people who are independent in all six Katz functions receives a falls and function programme costing GBP 300 per person. Over the first year, 100 people under usual care and 80 under the programme become dependent in one or more functions. The analyst follows the extra dependent people for one further year, ignoring deaths and discounting, with a utility of 0.75 when independent, 0.45 when dependent and a care cost of GBP 12,000 a year for each dependent person.

$$\text{ICER} = \frac{nc-dk}{d(u_I-u_D)}$$

where $n$ is the cohort size, $c$ is the programme cost per person, $d$ is the number of people kept out of dependence, $k$ is the annual care cost of a dependent person, $u_I$ is the utility of independence and $u_D$ is the utility of dependence.

1. Cases avoided. The programme avoids 100 − 80 = 20 transitions into dependence.

2. Costs. The programme costs 1,000 × 300 = 300,000, and care savings are 20 × 12,000 = 240,000, so the net cost is 300,000 − 240,000 = 60,000, or GBP 60,000.

3. QALYs. Each case avoided gains 0.75 − 0.45 = 0.30, so the cohort gains 20 × 0.30 = 6 QALYs.

4. Ratio. The ICER is 60,000 ÷ 6 = 10,000, or GBP 10,000 per QALY.

Suppose instead that the dependent state carries a utility of 0.55, as it might if the state were defined by milder dependence, with the 20 avoided cases and the care cost held fixed for illustration. The gain per case falls to 0.75 − 0.55 = 0.20, the cohort gains 20 × 0.20 = 4 QALYs, and the ICER rises to 60,000 ÷ 4 = 15,000, or GBP 15,000 per QALY. In a real model a milder definition would also change the number of transitions and the care cost, so both would need re-estimating. The Katz cut point used to define the state, and the source of the utility attached to the state, therefore move the cost-utility analysis result as much as the programme's effect does.

Where the Katz index misleads

The index was designed for older and chronically ill people with real care needs. Several errors recur when it is used outside that setting or fed into economic models, and each can change the result of an evaluation.

• Only basic self-care. The six functions stop at feeding, continence and similar tasks, so a person can keep a perfect Katz count while losing harder abilities, a ceiling effect that the 15-point modified index used by Suijker and colleagues addresses by adding instrumental activities.
• All or nothing judgements. Each function is either independent or dependent, so a person who needs a little help with bathing counts the same as one who needs total care, and a trial can miss real gains short of full independence.
• Counts read as utilities. A count of 4 out of 6 is not two thirds of full health, and losing bathing does not weigh the same as losing continence. Any QALY derived from Katz data depends on a separate valuation or a mapping function.
• Missing domains. The six functions say nothing about cognition, pain or mood, which carry much of the burden in conditions such as dementia.

Katz index, Barthel Index and instrumental activities

The Katz index, the Barthel Index and measures of instrumental activities answer different questions about daily function, and choosing between them decides which changes a trial can detect. All three describe what a person does, not how the person values it.

The Barthel Index also measures basic activities of daily living, but it weights ten items, includes walking and stair climbing, and records partial help. The Katz count, with six yes or no judgements, maps naturally onto a two-state independent or dependent model, while the Barthel gives finer steps within dependence. Instrumental activities, such as handling money or using transport, are what the modified Katz index adds, which makes it a different instrument from the six-function original. The modified Rankin Scale gives a single global grade of disability after stroke rather than a profile of functions.

Sources

• Asberg KH. Disability as a predictor of outcome for the elderly in a department of internal medicine: a comparison of predictions based on index of ADL and physician predictions. Scandinavian Journal of Social Medicine. 1987. PMID 3685928. https://pubmed.ncbi.nlm.nih.gov/3685928/
• Brorsson B, Asberg KH. Katz index of independence in ADL: reliability and validity in short-term care. Scandinavian Journal of Rehabilitation Medicine. 1984;16(3):125-132. PMID 6494836. https://pubmed.ncbi.nlm.nih.gov/6494836/
• Hartford Institute for Geriatric Nursing. Katz Index of Independence in Activities of Daily Living (ADL). Try This series, General Assessment Issue 2, consulted 2026. https://hign.org/consultgeri/try-this-series/katz-index-independence-activities-daily-living-adl
• Katz S, Ford AB, Moskowitz RW, Jackson BA, Jaffe MW. Studies of illness in the aged. The index of ADL: a standardized measure of biological and psychosocial function. JAMA. 1963;185(12):914-919. doi:10.1001/jama.1963.03060120024016
• Katz S, Branch LG, Branson MH, Papsidero JA, Beck JC, Greer DS. Active life expectancy. New England Journal of Medicine. 1983;309(20):1218-1224. doi:10.1056/NEJM198311173092005
• Suijker JJ, MacNeil-Vroomen JL, van Rijn M, Buurman BM, de Rooij SE, Moll van Charante EP, Bosmans JE. Cost-effectiveness of nurse-led multifactorial care to prevent or postpone new disabilities in community-living older people: results of a cluster randomized trial. PLoS One. 2017;12(4):e0175272. doi:10.1371/journal.pone.0175272