Behaviour Change Technique

Behaviour Change Technique Coding with BCT Taxonomy v1, the BCT Ontology and Costed Active Ingredients

A behaviour change technique (BCT) is one named, defined ingredient of a behaviour change programme, such as setting a behaviour goal, self-monitoring or feedback on behaviour, coded from what is delivered rather than from the programme's label. The standard code lists are the Behaviour Change Technique Taxonomy v1 (BCTTv1), with 93 techniques in 16 groups, and its successor, the Behaviour Change Technique Ontology (BCTO). For health economists, coding techniques shows which content was paid for and which content the comparator already contained.

What counts as a technique

Marques and colleagues define BCTs as the smallest parts of an intervention's content that are observable, replicable and on their own have the potential to bring about behaviour change. NICE guideline PH49 (2014) on individual approaches uses the same idea. A technique must meet specified criteria so that it can be identified, delivered and reliably replicated, and it should be observable and irreducible. Techniques can be used alone or in combination.

Intervention functions are often mistaken for techniques. They are the broad routes by which an intervention changes behaviour; the Behaviour Change Wheel lists nine, from education to enablement, and each function is put into practice through one or more techniques. Mechanisms of action are also kept apart from techniques, because one technique can work through more than one mechanism depending on context and delivery. The Behaviour Change Intervention Ontology (BCIO) records mode of delivery, source, schedule, style, setting and mechanisms of action as separate entities, so an app or a counselling session is a delivery mode, never a technique.

BCT Taxonomy v1 with 93 techniques in 16 groups

In building BCTTv1, Michie and colleagues ran a Delphi-type exercise in which 14 experts rated the labels and definitions of 124 techniques drawn from six earlier classification systems, and another 18 experts grouped them by similarity of active ingredients in an open-sort task. The result was 93 techniques clustered into 16 groups. Of the 26 techniques that occurred at least five times when six researchers coded 85 intervention descriptions, 23 reached adjusted kappas of 0.60 or above.

The fuller NIHR Health Technology Assessment report involved 400 participants from 12 countries. Good intercoder reliability was found for 80 of the 93 techniques. Training doubled the proportion of coders reaching competence but did not improve intercoder reliability, and usefulness for reporting observed interventions was mixed.

By 2024 BCTTv1 had been reported in more than 5,000 published studies.

The BCT Ontology as a computer-readable successor

User feedback showed that BCTTv1 needed sharper definitions, more techniques and a formal structure, and Marques and colleagues collated 282 proposed changes. Nineteen techniques were split; "Goal setting (behaviour)", for example, became "Set behaviour goal BCT" and "Agree behaviour goal BCT". The first published version of the BCTO contains 281 techniques in 20 higher-level groups over five hierarchical levels. Annotators testing the version before the last 34 techniques were added reached inter-rater reliability of 0.82 (familiar with it) and 0.79 (unfamiliar).

Each class carries a unique identifier (the set behaviour goal technique is BCIO:007003), and the ontology is released in the Web Ontology Language (OWL). The BCTO sits inside the BCIO, whose upper level has 42 entities. The authors state that no ontology is ever finished. A review pooling studies coded with BCTTv1 and with the BCTO therefore needs a documented crosswalk; the second version of the BCTO paper maps BCTO techniques to BCTTv1.

Coding an intervention and its comparator

Coding means reading intervention materials and recording each technique as present or absent, usually by two trained coders working independently. It can only find what is written down. Beard and colleagues coded a technique only when the materials gave unequivocal evidence of its use, and noted that many intervention reports are poorly specified.

In 104 comparator groups from smoking cessation trials, Black and colleagues found between 0 and 45 smoking cessation techniques (mean 15.97), and groups labelled usual care also ranged from 0 to 45. In HIV antiretroviral adherence trials, de Bruin and colleagues found that care given to control groups predicted outcomes and that interventions became less effective as standard care improved. In a cost-effectiveness analysis, the incremental effect therefore belongs to the techniques that differ between arms, so the comparator has to be coded as carefully as the new intervention.

Linking techniques to effect sizes

In meta-regression, each study's effect size is regressed on indicators for the techniques it contains, to see which account for between-study heterogeneity. Michie and colleagues pooled 122 healthy eating and physical activity evaluations (N = 44,747), finding an overall effect size of 0.31 (95% CI 0.26 to 0.36, I² = 69%). Self-monitoring explained the most heterogeneity (13%), and interventions combining self-monitoring with at least one other technique derived from control theory were more effective than the rest (0.42 against 0.26).

In smoking cessation, West and colleagues found nine techniques in service manuals, including providing rewards contingent on abstinence and measuring carbon monoxide, that were associated with both self-reported and carbon monoxide verified four-week quit rates. Black and colleagues estimated that each additional smoking cessation technique in a comparator group raised the logit of the cessation rate by 0.020 (95% CI 0.008 to 0.032); modelled cessation was 23% in the most intensively supported comparator groups against 8% in the least.

Peters and colleagues explain why these are associations rather than technique-level effects. Techniques co-occur, so a coefficient for self-monitoring may pick up its usual companions. A technique can be present yet applied without the conditions it needs to work, which they call parameters for effectiveness. Control groups may receive the same technique, and sample, design and study quality can confound the link. They recommend coding only techniques delivered differentially to the intervention arm and modelling co-occurrence and context.

Specifying interventions for economic evaluation

NICE PH49 (published 2014, links updated March 2025) asks that the manual for a behaviour change intervention include a clear definition of the techniques used so that each component can be replicated, for example by using a taxonomy (recommendation 6). Recommendation 16 asks for evaluation designs that can provide new evidence of effectiveness and, ideally, cost effectiveness, with assessment of intervention fidelity. The TIDieR checklist complements technique coding with 12 items covering materials, procedures, provider, mode, location, timing and amount, tailoring, modifications and fidelity.

A technique list gives economic analysis a defined unit of content. Staff time, training and materials can be attached to named components in a micro-costing. The list also shows whether a trial result transfers, because if a local version drops techniques the trial delivered, the trial's effect and ICER no longer describe what will be bought. Where a decision turns on one component, a factorial design randomising techniques separately gives a direct estimate.

Beard and colleagues coded 79 interventions judged cost-effective in NICE public health guidance. Education was the most common intervention function (82.3%) and shaping knowledge the most common technique group (79.7%); smoking interventions contained the most techniques and were the most cost-effective. The reward and threat group was associated with better cost-effectiveness estimates, but the authors state that cost-effectiveness cannot be causally attributed to specific techniques, since interventions contained many techniques and the analyses were likely underpowered.

Worked example pricing three added techniques

All numbers in this example are illustrative. A stop smoking service treats 1,000 people a year with a six-month quit rate of 15%. A redesign adds three techniques (self-monitoring of smoking, feedback on behaviour and strengthening ex-smoker identity) at an extra GBP 30 of adviser time per person. A meta-regression supplies a coefficient of 0.06 on the log-odds scale per added technique.

$$p_1 = \frac{1}{1 + \exp\left(-\left[\ln\frac{p_0}{1 - p_0} + k\beta\right]\right)}$$

where $p_1$ is the predicted quit probability after the change, $p_0$ is the current quit probability, $k$ is the number of techniques that differ from the current service and $\beta$ is the log-odds coefficient per technique.

Step 1. Current log-odds. ln(0.15 / 0.85) = ln(0.1765) = −1.7346.

Step 2. Add the techniques. With k = 3, the log-odds become −1.7346 + 3 × 0.06 = −1.5546, so p₁ = 1 / (1 + e^1.5546) = 1 / (1 + 4.7332) = 0.17442.

Step 3. Cost per additional quitter. The gain is 0.17442 − 0.15 = 0.02442, or 24.42 extra quitters per 1,000. The extra cost is 1,000 × 30 = GBP 30,000, so the cost per additional quitter is 30,000 / 24.42, about GBP 1,230.

Step 4. Check the comparator. Suppose coding the current manual shows two of the three techniques are already delivered, so k = 1 and the log-odds become −1.7346 + 0.06 = −1.6746, p₁ = 1 / (1 + 5.3367) = 0.15781, and the gain is 0.00781, or 7.81 extra quitters per 1,000. If the redesign still costs GBP 30 per person, the cost per additional quitter is 30,000 / 7.81, about GBP 3,840, roughly three times the first estimate.

Interpretation. Crediting the redesign with techniques the comparator already had overstates the effect threefold. Cost per quitter is an intermediate result that a decision model must still convert into QALYs, and the coefficient carries every caution Peters and colleagues raise.

Techniques, interventions and neighbouring ideas

A technique is a single coded ingredient, while a behaviour change intervention is the whole package with its delivery, dose and setting. Goal setting and feedback interventions are programmes built mainly around one technique, and audit and feedback applies the feedback technique to health professionals. A choice architecture intervention, such as a nudge, works mainly through the environmental restructuring function. Habit formation is a process that several techniques aim to support.

Limitations and common misreadings

Coding improves description. Several errors recur when the codes are used as evidence.

1. "More techniques make a better intervention." Black's finding is a between-group association, and most cost-effective interventions in Beard's sample used relatively few techniques.
2. "A coded technique was delivered." Manual coding records design; PH49 treats fidelity, delivery as intended, separately.
3. "A meta-regression coefficient is the effect of adding a technique." It is a between-study association open to co-occurrence and confounding.
4. "Usual care is an empty comparator." Comparator groups can contain dozens of techniques.
5. "BCTTv1 and BCTO codes are interchangeable." Split and new classes need an explicit mapping.
6. "Evidence transfers across behaviours." The 2009 meta-regression covered healthy eating and physical activity only.

Sources

• Beard E, West R, Lorencatto F, et al. What do cost-effective health behaviour-change interventions contain? A comparison of six domains. PLoS One. 2019;14(4):e0213983. https://doi.org/10.1371/journal.pone.0213983
• Black N, Eisma MC, Viechtbauer W, et al. Variability and effectiveness of comparator group interventions in smoking cessation trials: a systematic review and meta-analysis. Addiction. 2020;115(9):1607-1617. https://doi.org/10.1111/add.14969
• de Bruin M, Viechtbauer W, Schaalma HP, et al. Standard care impact on effects of highly active antiretroviral therapy adherence interventions: a meta-analysis of randomized controlled trials. Archives of Internal Medicine. 2010;170(3):240-250. https://doi.org/10.1001/archinternmed.2009.536
• Hoffmann TC, Glasziou PP, Boutron I, et al. Better reporting of interventions: template for intervention description and replication (TIDieR) checklist and guide. BMJ. 2014;348:g1687. https://doi.org/10.1136/bmj.g1687
• Marques MM, Wright AJ, Corker E, et al. The Behaviour Change Technique Ontology: transforming the Behaviour Change Technique Taxonomy v1. Wellcome Open Research. 2023;8:308 (version 2, 2024). https://doi.org/10.12688/wellcomeopenres.19363.2
• Michie S, Abraham C, Whittington C, et al. Effective techniques in healthy eating and physical activity interventions: a meta-regression. Health Psychology. 2009;28(6):690-701. https://doi.org/10.1037/a0016136
• Michie S, Richardson M, Johnston M, et al. The behavior change technique taxonomy (v1) of 93 hierarchically clustered techniques: building an international consensus for the reporting of behavior change interventions. Annals of Behavioral Medicine. 2013;46(1):81-95. https://doi.org/10.1007/s12160-013-9486-6
• Michie S, Wood CE, Johnston M, et al. Behaviour change techniques: the development and evaluation of a taxonomic method for reporting and describing behaviour change interventions. Health Technology Assessment. 2015;19(99):1-188. https://doi.org/10.3310/hta19990
• National Institute for Health and Care Excellence. Behaviour change: individual approaches (NICE guideline PH49). London: NICE; 2014, links updated March 2025. Recommendations 6 and 16. https://www.nice.org.uk/guidance/ph49
• Peters GJY, de Bruin M, Crutzen R. Everything should be as simple as possible, but no simpler: towards a protocol for accumulating evidence regarding the active content of health behaviour change interventions. Health Psychology Review. 2015;9(1):1-14. https://doi.org/10.1080/17437199.2013.848409
• West R, Walia A, Hyder N, et al. Behavior change techniques used by the English Stop Smoking Services and their associations with short-term quit outcomes. Nicotine and Tobacco Research. 2010;12(7):742-747. https://doi.org/10.1093/ntr/ntq074