Sage Journals: Discover world-class research

Abstract

Background:

The Sequential Multiple-Assignment Randomised Trial (SMART) design is considered the gold standard for developing adaptive interventions, which tailor treatments to individual patient characteristics and responses. While SMART offers a rigorous framework aligned with real-world clinical decision-making, it is often complex, time-consuming, and costly. As interest in SMART design grows, there is increasing recognition for the need to improve its implementation through more explicit guidance and best practices. Efficiency gains may also be possible by incorporating external data to inform their design, conduct, and analysis. This review aimed to identify all published trials using the SMART design, summarise their design, conduct, and reporting practices and evaluate the use of external data in their implementation.

Methods:

We searched PubMed, Medline, PsycINFO, Scopus, and Web of Science databases for all SMART up to June 30, 2024. External data were defined as non-simulated individual patient data collected outside the main SMART to supplement or inform the main trial.

Results:

We included 80 SMART, of which 35 (44%) were completed and 45 (56%) were ongoing. Most trials reported two phases of randomisation (93%), with the primary aim focusing on evaluating main effects (81%) of interventions at the first stage of randomisation. There was inadequate reporting of several key aspects, including sample size estimation, statistical analysis software, allocation concealment, data missingness, multiple testing, sensitivity analysis, and the use of SMART in the title. Seventeen (21%) SMART (4-completed trials and 13-trial protocols) referred to the use of external data from electronic health records (n = 12) and registries (n = 5). External data was used for recruitment (n = 11), outcome measures (n = 6), and to provide baseline covariate information (n = 1).

Conclusion:

SMART designs are increasingly used to develop adaptive interventions across diverse clinical contexts, yet key methodological features and basic components remain inconsistently reported. This limits transparency, reproducibility, and potential for translation into routine care. Although external data are widely used in standard randomised controlled trials, their use in the SMART is still limited, likely due to methodological and infrastructural challenges and the absence of tailored reporting standards. To improve the efficiency and generalisability of SMART designs, expert-led extensions of CONSORT and SPIRIT guidelines are needed, including specific recommendations for reporting external data use. Future research should explore optimal external data sources for informing SMART components and promote interdisciplinary collaboration and training to support high-quality implementation.

Keywords

Sequential Multiple-Assignment Randomised Trial external data routine data adaptive interventions systematic review

Introduction

Adaptive interventions, also known as personalised medicine, describe the principle of tailoring treatments to individual patient characteristics and response to optimise care.¹ This is formalised as a sequence of decision rules, at given decision points, that establishes treatments for patients, or the next course of action, based on up-to-date information. Adaptive interventions, therefore, reflect the ongoing multi-stage decision-making by clinicians in practice.

Finding the best adaptive interventions would be a way to apply evidence-based personalised medicine in the real world.² To develop and refine adaptive interventions, an experimental design known as the Sequential Multiple Assignment Randomised Trial (SMART) was proposed.³ SMART re-randomises participants at multiple stages based on pre-specified rules.³ Adaptive interventions are embedded in a SMART. Patients are randomised to initial and subsequent treatments based on treatment response and individual characteristics. Each randomisation stage in SMART corresponds to a critical decision point in adaptive interventions. This enables researchers to build or refine adaptive interventions within a SMART, making it a robust tool for personalised medicine.

Although SMART designs provide a rigorous framework for developing adaptive interventions in chronic diseases and other health-related conditions,⁴ their implementation continues to encounter methodological and practical challenges, including limited expertise and their inherent complexity.^5–11 To mitigate these challenges, SMART studies must be carefully designed with clear objectives. Like standard trials, their design, conduct, and reporting are guided by the CONSORT statement and International Conference on Harmonisation E9.^12,13 Systematic reviews of SMART design, conduct, and reporting are crucial for closing the knowledge gap, improving transparency and interpretability, informing guidelines, promoting best practices, and enhancing utilisation. Recent reviews by Bigirumurame et al.⁷ and Lorenzoni et al.⁸ highlighted the need for transparency and comprehensive reporting in SMART. With the growing interest in personalised medicine, SMART designs are likely to gain a firm foothold in clinical trials to aid in developing adaptive interventions,¹¹ thus necessitating their continuous updates in design, conduct, and reporting.

Despite their alignment with real-world clinical practice,³ SMART designs can be time-consuming and costly. These burdens may be reduced by using external data to inform key aspects of their design. External data – defined as individual patient data collected outside the main trial – is increasingly used to support randomised control trial’s (RCT) design and conduct.^14,15 External data sources like electronic health records (EHRs), including Clinical Practice Research Datalink,¹⁶ and disease registries can guide SMART’s planning and implementation.¹⁷ Specifically, they can help define decision rules and decision point(s),¹⁸ identify at-risk populations (tailoring variables), assess feasibility, and support sample size estimation.^19,20 The growing availability of EHRs and registries has led regulators and industry to recognise their value in practice and drug development,^21,22 as reflected in the CONSORT extension for reporting external data in RCTs²³ and the ongoing development of the SPIRIT guidelines.²⁴ Establishing similar standards for SMART designs is crucial to improve transparency and reproducibility.

Herein, we present a review that evaluates the design, conduct, and reporting of SMART studies and assesses their use of external data in published literature.

Methods

The review protocol is available on the Open Science Framework.²⁵

Search strategy

We expanded the search terms from Bigirumurame et al.⁷ to include additional SMART-related terms (Supplemental Appendix S1). A focused search was conducted in Scopus, Medline (Ovid), Web of Science, PubMed and PsycINFO for articles published between 2020 and June 30, 2024. The 2020 start date was chosen to build on Bigirumurame et al’.s⁷ prior review, which covered literature up to 2019, to ensure continuity and avoid duplication. We also screened reference lists and eligible articles from their full-text review.

Eligibility criteria

We included all interventional studies using a SMART design, published in English, with full-text reports. SMART design was defined as involving ≥ 2 stages of randomisation based on pre-defined rules.³ External data was defined as non-simulated individual patient data collected outside the main SMART study to supplement or inform the trial. We excluded book chapters, dissertations, commentaries, editorials, mixed-methods articles, systematic reviews, conference papers, and feasibility studies.

Article screening

All articles were imported into EndNoteX20²⁶ and duplicates removed. Two reviewers (IE and CTS) independently screened titles, abstracts, and keywords using Rayyan.²⁷ For full-text screening, both reviewers assessed all articles independently, noting reasons for exclusion. Discrepancies were resolved by consensus following discussion between reviewers.

Data extraction

Data were extracted using a standardised Microsoft Excel spreadsheet,²⁸ piloted on 12 completed trials from Bigirumurame et al.⁷ Extracted data covered study characteristics, SMART design and conduct, aims, analysis, reporting, and external data use. The first reviewer (IE) extracted all data, which was cross-verified by an independent reviewer (CTS).

Data analysis

All analyses were performed in R version 4.3.0²⁹ using descriptive statistics to report frequencies and proportions of study characteristics and trial types.

Quality assessment

Quality assessment was not performed as the focus was on SMART study design, conduct, reporting, and use of external data.

Results

Search results

Figure 1 presents the PRISMA flowchart. Of the 3685 records identified, 2150 duplicates were removed, leaving 1535 for title and abstract screening. Backward citation searching added 82 records. After excluding 1339 records, 192 full-text articles from the updated search and 82 from the citation search were assessed. Four records lacking full-texts were excluded. In total, 431 full-text articles were screened, including 274 from the updated search and 157 from the previous review.⁷ Unlike the earlier review,⁷ we excluded reviews and methodological papers to focus on randomised trials. The final sample included 80 independent SMART; 35 (44.0%) completed trials^30–64 and 45 (56.0%) protocols.^65–109

Figure 1.

PRISMA flowchart.

SMART study characteristics

Table 1 summarises characteristics of the 80 SMART. Studies were conducted in 12 countries, with the majority published in the United States (n = 58, 72.5%) (Figure 2). Studies were published between 1992 and 2024, with a notable increase in the number of studies published after 2018. Since the 2020 review,⁷ completed SMART nearly tripled and protocols doubled (Figure 3).

Table 1.

Characteristics of SMART.

Characteristics	Completed trials (n = 35, %)	Trial protocols (n = 45, %)	Overall (N = 80, %)
Trial characteristics
Field of health research
Cancer	10 (28.6)	1(2.2)	11 (13.8)
Infectious diseases	1 (2.9)	9 (20.0)	10 (12.5)
Pain and physical health disorder	5 (14.3)	8 (17.8)	13 (16.3)
Neurological disorder	1 (2.9)	3 (6.7)	4 (5.0)
Psychiatric disorder	10 (28.6)	17 (37.8)	27 (33.8)
Substance use disorder	8 (22.9)	7 (15.6)	15 (18.8)
Population
Children and adolescents	7 (20.0)	10 (22.2)	17 (21.3)
Adults	27 (77.1)	31 (68.9)	58 (72.5)
All age	1 (2.9)	4 (8.9)	5 (6.3)
Study setting
Hospital	20 (57.1)	20 (44.4)	40 (50.0)
Community	9 (25.7)	15 (33.3)	24 (30.0)
Hospital and community	6 (17.1)	10 (22.2)	16 (20.0)
Funding
Private	2 (5.7)	3 (6.7)	5 (6.3)
Public	23 (65.7)	41 (91.1)	64 (80.0)
Private and public	6 (17.1)	1 (2.2)	7 (8.8)
No funding	1 (2.9)		1 (1.3)
No information	3 (8.6)		3 (3.8)
NIH funding	18 (51.4)	31 (68.9)	49 (61.2)
Multicentre trial
No	13 (37.1)	21 (46.7)	34 (42.5)
Yes	22 (62.9)	24 (53.3)	46 (57.5)
Reporting
Study title identified as SMART
No	16 (45.7)	7 (15.6)	23 (28.8)
Yes	19 (54.3)	38 (84.4)	57 (71.2)
Trial registration details
No	5 (14.3)	2 (4.4)	7 (8.8)
Yes	30 (85.7)	43 (95.6)	73 (91.2)
Allocation concealment reporting
No	17 (48.6)	15 (33.3)	32 (40.0)
Yes	18 (51.4)	30 (66.7)	48 (60.0)
Sample size estimation reporting
No	7 (20.0)		7 (8.8)
Yes	28 (80.0)	45 (100)	73 (91.2)
Reported software for sample size estimation
No	33 (94.3)	35 (77.8)	68 (85.0)
Yes	2 (5.71)	10 (22.2)	12 (15.0)
Reported missingness
No	19 (54.3)	21 (46.7)	40 (50.0)
Yes	16 (45.7)	24 (53.3)	40 (50.0)
Pilot/feasibility study informing trial reported
No	1 (2.9)	11 (24.4)	12 (15.0)
Yes	34 (97.1)	34 (75.6)	68 (85.0)
Blinding reported
No	5 (14.3)	1 (2.2)	6 (7.5)
Yes	30 (85.7)	44 (98.8)	74 (92.5)
Analysis software reported
No	20 (57.1)	30 (66.7)	50 (62.5)
Yes	15 (42.9)	15 (33.3)	30 (37.5)
Design and conduct
Randomisation phases
Two	33 (94.3)	41 (91.1)	74 (92.5)
Three	3 (2.9)	4 (8.9)	5 (6.3)
Four	1 (2.8)		1 (1.2)
Sample size
Median (IQR)	360 (160-491)	400 (280-800)	377 (194-671)
Minimum-maximum	61-2876	36-6000	36-6000
Attrition rate, median % (IQR)
First stage	11.3 (6.2-14.8)
Second stage	17.5 (13.0-30)
SMART model
Prototypical	24 (68.6)	38 (84.4)	62 (77.5)
Responsive intervention	7 (20.0)	1 (2.2)	8 (10.0)
Unrestricted SMART	4 (11.4)	6 (13.3)	10 (12.5)
Study duration in months, median (IQR)	12 (6, 24)
Number of initial treatments
1	5 (14.3)	7 (15.6)	12 (15.0)
2	24 (68.6)	34 (75.6)	58 (72.5)
3	4 (11.4)	3 (6.7)	7 (8.8)
4	1 (2.9)		1 (1.3)
5	1 (15.6)	1 (2.2)	2 (2.5)
SMART aims considerations
Randomisation stage of the primary aim
First stage	29 (82.9)	28 (62.2)	57 (71.2)
Second stage	2 (5.7)	3 (6.67)	5 (6.3)
Both	4 (11.4)	14 (31.1)	18 (22.5)
Embedded adaptive interventions comparisons
No	17 (48.6)	17 (37.8)	34 (42.5)
Yes	18 (51.4)	28 (62.2)	46 (57.5)
Selecting deeply embedded adaptive interventions
No	32 (91.4)	27 (60.0)	59 (73.8)
Yes	3 (8.6)	18 (40.0)	21 (26.2)
SMART aim considerations
Single aim	21 (60.0)	14 (31.1)	35 (43.7)
Two aims	13 (37.1)	22 (48.9)	35 (43.7)
Three aims	1 (2.9)	9 (20.0)	10 (12.5)
Main effect aim consideration as the primary aim
No	2 (5.7)	13 (28.9)	15 (18.8)
Yes	33 (94.3)	32 (71.1)	65 (81.2)
Embedded adaptive intervention consideration as the primary aim
No	32 (91.4)	34 (75.6)	66 (82.5)
Yes	3 (8.6)	11 (24.4)	14 (17.5)
Optimisation aims consideration as the primary aim
No	34 (97.1)	43 (95.6)	78 (97.5)
Yes	1 (2.9)	2 (4.4)	3 (2.5)
Analysis considerations
Use of multiple testing
No	31 (88.6)	35 (77.8)	66 (82.5)
Yes	4 (11.4)	10 (22.2)	14 (17.5)
Analysis method for primary aim
ANOVA	1 (2.9)
Generalised estimating equation	3 (8.6)
Generalised linear model	12 (34.3)
Linear mixed-effect model	7 (20.0)
T-test	1 (2.9)
Survival	11 (31.4)
Analysis type
Intention to treat analysis	29 (82.9)	34 (75.6)	63 (78.7)
Not specified	6 (17.1)	11 (24.4)	17 (21.3)

SMART: Sequential Multiple-Assignment Randomised Trial; IQR: interquartile range; ANOVA: analysis of variance.

Figure 2.

Included study publications by countries.

Figure 3.

SMART studies published over the years.

Psychiatric disorders were most studied (n = 27, 33.8%), followed by substance use disorders (n = 15, 18.8%), pain/physical disorders (n = 13, 16.3%), cancer (n = 11, 13.8%), and infectious diseases (n = 10, 12.5%) and neurological disorders (n = 4, 5.0%). The psychiatric disorders reported included depression,^{58,59,66,67,74,84,85,88} anxiety,^59,96 post-traumatic stress disorder,^{33,88,100,101} attention-deficit hyperactivity disorder,^50,69 sleep disorders,^47,74,76 suicide,^90,93 schizophrenia,^61,89,92 bipolar disorder,³⁶ obsessive-compulsive disorder,³² eating disorder,³⁸ autism,⁴¹ and intimate partner violence.⁹⁹ Pain/physical disorders included chronic pain,^{34,60,78,80,82} weight loss/obesity or diabetes,^56,79,81,86 osteoarthritis,⁴⁰ and burns.⁸³ Infectious diseases were mainly HIV/AIDS,^{37,65,70,73,91,102,103} malaria,¹⁰⁸ and COVID-19.^98,106 Neurological disorders included dementia,^87,105 stroke,⁷² and learning disability.⁴² Substance use disorders include alcohol use disorder,^{45,46,49,94,104} smoking,^{31,35,54,75,107} and cocaine and opioids.^52,55,95,97

Most trials focused on adults (≥18 years) (n = 58, 72.5%), followed by children/adolescents (n = 17, 21.3%), and all-age groups (n = 6, 6.3%). Half of the trials used hospital/clinic samples; about one-fifth combined community and hospital samples, and one-third sampled from the community alone. Most studies were funded by public funding (n = 71, 88.7%), with the National Institute of Health (NIH) being the main funder (61.3%) for publicly funded studies.^{65,67,68,70,73–75,79,97,100,105,109}

Design characteristics of SMART

Almost all trials (n = 74, 92.5%) reported two randomisation phases. In the first randomisation stage, 58 (72.5%) randomised between two interventions, seven between three, one between four, and two between five. Twelve studies (15.0%) used a run-in intervention model, where participants received the same treatment before first randomisation.^{34,45,46,54,58,68,82,84,85,91,92,97} Over three-quarters of trials (n = 62, 77.5%) were prototypical SMART, where re-randomisation occurs for non-responders of stage 1 interventions. Ten trials (12.5%) were unrestricted SMART (re-randomisation occurs regardless of response status), and eight (10.0%) were responsive SMART (re-randomisation occurs to responders to initial treatment). Most trials (n = 74, 92.5%) randomised at individual level, while six (7.5%) randomised at the cluster level. The median sample for completed trials was 360 participants (interquartile range (IQR), 160–491), with the smallest and largest trial recruiting 61 and 2876 participants, respectively. Follow-up duration ranged from 2 days (in an emergency setting) to 96 months, with a median of 12 months (IQR 6–24).

Aims considerations in SMART

SMART design evaluates aims for main effect, embedded adaptive interventions, and optimisation. The main effect aims to identify the best initial intervention or effective option for responders/non-responders. Embedded adaptive intervention aims to compare or/and identify effective adaptive interventions, while the optimisation aim, often referred to as hypothesis generating, explores deeply tailored adaptive interventions.⁶ Trials considered all three aims in 10 (12.5%) and at least two in 35 (43.7%) trials. Over three-quarters (n = 65) of trials considered the main effect aim as the primary aim, while embedded adaptive intervention and optimisation were considered in 14 (17.5%) and 2 (2.5%), respectively. One trial considered the main effect and embedded adaptive intervention as the primary aim.⁵² Over two-thirds of trials (n = 57) focused on the primary aim in first-stage randomised interventions, five in second-stage, and 18 across all stages. Forty-six trials compared multiple embedded adaptive interventions, while 21 evaluated a single adaptive strategy that optimises outcomes.

Analysis considerations in SMART

Completed trials used analysis of variance (ANOVA) (n = 1), t-test (n = 1), generalised estimation equation (GEE) (n = 3), generalised linear model (GLM) (n = 11), survival analysis (n = 11), and mixed effects model (n = 8) analytical techniques to address the primary research questions. Unlike completed trials, protocols described the analysis plans in a high-level summary, often mentioning broad methodological approaches in SMART without providing specific details. For example, protocols referenced potential analysis techniques in SMART,^83,95 or mentioned specific techniques such as mixed-effect regression⁷⁶ and longitudinal models.⁸² Only 22 (48.9%) protocols reported a detailed analysis plan for the primary research question, including GEE (n = 5), mixed effects model (n = 9), GLM (n = 6), marginal mean model (n = 1), and targeted maximum likelihood (n = 1). Q-learning was intended/used for secondary or exploratory analysis in 14 trials.^{49,59,66,70,78,80,83,84,88,91,96,97,101,108}

Multiple comparisons of adaptive interventions in SMART require pre-specified or reported adjustments to control for Type 1 error.^110–112 Only four completed trials reported doing so, using the Bonferroni method,^48,58 Hochberg adjustment,⁶¹ or discussing the impact of multiple testing on effect estimate.⁴² Ten protocols intended to assess multiple testing in the analysis plan, including the use of a composite outcome strategy to avoid multiplicity.⁷⁷

Reporting of SMART

One-third (28.8%, n = 23) of the studies did not mention SMART design in the title, and about half of the completed trial reports omitted it. Trial registration was reported in 73 trials (91.2%), but only 35 trials (43.8%) reported the assignment model as sequential. Other assignment models reported included parallel (n = 26), factorial (n = 9), cross-over (n = 1), and single group (n = 1), while eight studies (10.0%) did not specify a model. SMART design features, such as tailoring variables, decision rules, and critical decision points, were exclusively reported in 17 (21.3%), 8 (10.0%), and 3 (3.8%) studies, respectively.

Seventy-four trials (92.5%) reported blinding status, which included: no blinding (n = 24), single (n = 35), double (n = 10), triple (n = 1) or quadruple blinding (n = 4). Allocation concealment was exclusively reported in 48 (60.0%) trials. Sample size estimation details were reported in 73 studies (91.2%), though software used was reported in 12 (15.0%), including SAS, G Power, PASS, R, PS, and STPLAN. Pilot or feasibility studies before the main SMART or development of the SMART protocol were reported in 12 (15.0%) studies. Statistical software was reported in 30 trials (37.5%), including East Software Cytel (n = 1), MPLUS (n = 1), SPLUS (n = 1), SPSS (n = 3), SPSS and SAS (n = 1), SAS and R (n = 3), SAS (n = 8), R (n = 10), STATA and R (n = 1), and STATA and SAS (n = 1). Missing data or plans to address it were reported in 40 trials (50.0%).

External data in SMART

External data use was reported in 17 trials (21.2%); 4 completed trials^40,48,55,60 and 13 protocols^{65,67,68,70,73–75,79,97,98,100,105,109} (Table 2). It was mostly reported in pain and physical disorder, infectious diseases and psychiatric disorders in four studies each (23.5%). Almost all the studies that used external data were publicly funded (94.1%), of which 75.0% were NIH funded. Most studies were from the United States (82.4%), involved multisite trials (58.8%) and were published beyond 2020 (70.5%). The majority used EHRs (n = 12) than registries (n = 5). Each study reported a single external data source. External data usage ranged from minimal to substantial across trials and protocols. All studies used external data to supplement other data sources, rather than relying exclusively on it for primary purposes (Table 2).

Table 2.

External data in SMART.

	Completed trial (n = 35, %)	Trial protocols (n = 45, %)	Overall (N = 80, %)
Use of external data
No	31(88.6)	32 (71.1)	63 (78.8)
Yes	4 (11.4)	13 (28.9)	17 (21.2)
External data type
Registry	2 (5.7)	3 (6.7)	5 (6.3)
Electronic health record	2 (5.7)	10 (22.2)	12 (15.0)
Purpose of external data
Recruitment	4 (11.4)	7 (15.6)	11 (13.8)
Outcome measure		6 (13.3)	6 (7.5)
Baseline covariate measure		1 (2.2)	1 (1.3)

Completed trials

Of the four completed trials that used external data: two used data from EHRs and two used patient registries. Completed trials used external data to supplement recruitment or enrolment of participants. Medical records were reviewed alongside self-referral, direct provider referral, and community inquiries to identify potentially eligible participants in EHRs.^48,60 For registries: a university research registry to supplement primary care in the recruitment of participants⁴⁰ and a smoking registry to prompt potential participants on their willingness to participate in the trial or invite them to participate⁵⁴ were used.

Protocols

Thirteen SMART protocols used external data from EHRs (n = 10) and registries (n = 3). Seven studies planned using EHRs or registry to identify potential participants,^{74,75,100,105,109} enrol participants,⁶⁷ and send invitation messages to potential participants.⁷⁹ Six protocols intended to use external data to measure outcomes, including blood pressure, laboratory tests (glycated haemoglobin, HIV viral load, urinalysis and lipid profile), adherence, and suicide screening uptake evaluation.^{65,68,70,73,97,109} One protocol planned using immunisation status data from EHR as a baseline covariate information to analyse the study outcome.⁹⁸ One protocol planned using data for both recruitment and outcome measure.¹⁰⁹

Discussion

This review identified 80 SMART (35 completed trials and 45 protocols), twice as many as in the previous review in a span of 4 years. Most trials were published in the United States, recruited adults from hospitals, conducted in multicentre settings, publicly funded, and focused on psychiatric disorders.

SMART design features

Most SMART began with two treatment options and two phases of randomisation, consistent with prior studies.^7,8 Although SMART designs can address three main scientific aims,¹¹³ only 12.5% of the trials explored all. This suboptimal use may reflect gaps in methodological expertise, complexity of multi-aim designs, funding constraints^6,20 or the field’s early focus on narrow objectives. Encouragingly, training resources have expanded in recent years. Notable contributions include annual online SMART-workshops,¹¹⁴ the d3centre webpage,¹¹⁵ and Statistical Horizons,¹¹⁶ which provide valuable guidance on SMART. However, many are prohibitively expensive for individual researchers, and expanding their access by making them affordable may increase SMART’s utilisation.

Re-randomisation, a key feature of SMART design, enables tailoring treatment to individual characteristics and response. In this review, 77.5% of studies re-randomised non-responders in the second phase, aligning with SMART’s deviation from a ‘one-size-fits-all’ approach.⁴ A few trials used a run-in phase before stage 1, potentially reducing resource use and treatment burden by restricting further interventions to non-responders. However, this may limit generalisability and introduce selection bias, as only those with a severe form of the disease are likely to be included.

Reporting in SMART

The CONSORT¹¹² and SPIRIT¹¹¹ statements stipulate reporting guidelines for any trials and trial protocols, respectively. However, reporting was inadequate in identifying trials as SMART, allocation concealment, sample size estimation, analysis software, data missingness, and pilot studies informing the main trial. While registration details, blinding, and sample size estimation were adequately reported, over half incorrectly registered SMART as non-sequential assignment models, suggesting limited knowledge of the design. These reporting gaps are consistent with a previous finding.⁷

Although the CONSORT and SPIRIT reporting guidelines apply to most trials, SMART has features that these guidelines may not capture. We found an unclear reporting of key SMART features such as decision rules, tailoring variables, and critical decision points. Failing to objectively report these features hinders replication by other researchers and undermines the transparency of SMART. While some authors may argue that there are no clear reporting guidelines for reporting SMART features. Hampton and colleagues recently recommended guidelines for reporting SMART features, which include reporting demographic characteristics, attrition, missing data for every randomisation stage, tailoring variables, and effect size for each embedded adaptive intervention.¹¹⁷ These recommendations may be relevant and a good starting point. However, developing reporting guidelines should be systematic, thorough and involve an expert consensus meeting and/or Delphi process.^118,119

External data in SMART

Given the novelty of SMART in health research, it is not surprising that only 21% of SMART used external data. While the cost and time-saving potentials are well documented in standard trials,^120,121 no prior review, to the best of our knowledge, has examined its use in SMART. Standard RCTs commonly use external data from registries, EHRs, cohorts and administrative databases,^{14,15,122–125} supported by reporting guidelines.²³ One explanation is SMART’s methodological complexity, which complicates integrating observation data, especially when detailed treatment adaptations are needed to inform decision rules.⁸ Investigators may also lack experience and guidance on leveraging external data within SMART. The absence of methodological and reporting guidelines further hinders this integration.

Beyond design complexity, barriers common to standard RCTs may also apply to SMART. These include limited access to external data due to bureaucracy, data security, and privacy concerns¹²⁶ and infrastructure gaps such as weak data governance frameworks and IT support.¹²¹ In addition, statistical challenges – such as handling missing data, selection bias, and immortal time bias – require sophisticated methods,¹²⁷ not always available to researchers.

Nevertheless, the minimal use of observation data in SMART is a missed opportunity. External data have only supplemented participant enrolment, outcome, and baseline covariate measurements. Because SMART design mirrors routine clinical practice, particularly in most chronic and recurrent diseases, rich longitudinal datasets from EHRs or registries hold potential for feasibility and efficiency.⁵ Such data could inform estimates of response rates, decision points, tailoring variables, and improve sample representativeness.¹²⁸ External data can also help identify adaptive interventions. For example, Krakow et al.¹²⁹ used a registry and Q-learning,¹³⁰ to derive adaptive interventions that may then be prospectively validated in a SMART.

Strengths and limitations

This review provides the most comprehensive synthesis of completed SMART and protocols. Our search identified 80 studies, double the number found in the previous review. We systematically evaluated trial characteristics, reporting practices and use of external data, offering new insights into methodological gaps and under-utilisation of real-world data in SMART. This is the first review to examine external data in SMARTs, highlighting an important design consideration. Furthermore, we offer a detailed record of SMART features such as re-randomisation, decision rules and tailoring variables, contributing to the need for tailored reporting guidelines.

This review has limitations. We excluded non-English publications and unpublished protocols, which may have introduced selection and publication bias. The focus on published trials may also overrepresent trials with favourable results or clearer reporting practices. While we identified whether external data were used, we did not assess their quality, completeness, or analytic contribution of these sources to trial design or outcomes, which remains an important area for future investigation. In addition, the interpretation of SMART features was limited by the inconsistent or incomplete reporting, underscoring the need for standardised methodological guidance and reporting standards.

Conclusion

The SMART design is an innovative and increasingly used methodology in clinical research, with application across diverse conditions. However, key methodological features unique to SMART remain inconsistently reported and underreported, limiting transparency, reproducibility, and translation. Although external data such as EHRs, disease registries, and cohort studies are common in traditional RCTs, their use in SMART remains limited. This gap likely reflects methodological challenges, including data access and safety concerns, infrastructural hurdles, and a lack of reporting guidelines. Addressing these gaps will improve efficiency, enhance generalisability, reduce duplications and support translation of adaptive interventions.

To advance methodological quality, expert-led and consensus-driven reporting standards are needed, including extensions of CONSORT and SPIRIT. In parallel, guidance for reporting external data in SMART designs is required. Future research should examine which external data best support SMART, not only to inform tailoring variables, decision rules, and decision points, but also to identify candidate adaptive interventions. Equally important is the need for continuous affordable training programmes on SMART methodologies and interdisciplinary collaboration, particularly between biostatisticians and data scientists. Finally, given the complexity of SMART designs, journals could offer extra space or structured templates to support comprehensive reporting.

Supplemental Material

sj-docx-1-ctj-10.1177_17407745251385535 – Supplemental material for A review of use of external data and update on reporting standards in Sequential Multiple-Assignment Randomised Trials

Supplemental material, sj-docx-1-ctj-10.1177_17407745251385535 for A review of use of external data and update on reporting standards in Sequential Multiple-Assignment Randomised Trials by Isaac J Egesa, Laura Bonnett, Richard Emsley, Anthony Marson and Catrin T Smith in Clinical Trials

Footnotes

Acknowledgements

We thank Théophile Bigirumurame and James Wason (Biostatistics Research Group, Population Health Sciences Institute, Newcastle University, UK) for sharing EndNote library search results from the prior review, which were used to screen eligible articles.

Author contributions

All authors contributed to the review’s conception and design. I.J.E. and C.T.S. developed the search strategy, conducted the literature search, and led the screening and data extraction process. I.J.E. performed data analysis with the support of C.T.S and L.B. I.E. prepared the initial manuscript draft, and all authors critically reviewed and revised the manuscript for important intellectual content. All authors have approved the final version of the manuscript.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: I.J.E. is funded by the MRC Trial Methodology Research Partnership (TMRP) Doctoral Training Partnership (Grant Number MR/W006049/1).

Ethical approval

None.

ORCID iDs

Isaac J Egesa

Laura Bonnett

Richard Emsley

Supplemental material

Supplemental material for this article is available online.

References

Collins

LM.

Introduction to adaptive interventions. In: Collins

(ed.) Optimization of behavioral, biobehavioral, and biomedical interventions: the multiphase optimization strategy (MOST). Cham: Springer International Publishing, 2018, pp. 267–287.

Kosorok

Laber

EB.

Precision medicine. Annu Rev Stat Appl 2019; 6: 263–286.

Murphy

SA.

An experimental design for the development of adaptive treatment strategies. Stat Med 2005; 24: 1455–1481.

Lavori

Dawson

Adaptive treatment strategies in chronic disease. Annu Rev Med 2008; 59: 443–453.

Almirall

Compton

Gunlicks-Stoessel

, et al. Designing a pilot sequential multiple assignment randomized trial for developing an adaptive treatment strategy. Stat Med 2012; 31: 1887–1902.

Almirall

Nahum-Shani

Sherwood

, et al. Introduction to SMART designs for the development of adaptive interventions: with application to weight loss research. Transl Behav Med 2014; 4(3): 260–274.

Bigirumurame

Uwimpuhwe

Wason

Sequential multiple assignment randomized trial studies should report all key components: a systematic review. J Clin Epidemiol 2022; 142: 152–160.

Lorenzoni

Petracci

Scarpi

, et al. Use of Sequential Multiple Assignment Randomized Trials (SMARTs) in oncology: systematic review of published studies. Br J Cancer 2023; 128(7): 1177–1188.

Pistorello

Jobes

Compton

, et al. Developing adaptive treatment strategies to address suicidal risk in college students: a pilot Sequential, Multiple Assignment, Randomized Trial (SMART). Arch Suicide Res 2018; 22: 644–664.

10.

Laber

Davidian

Dynamic treatment regimes, past, present, and future: a conversation with experts. Stat Methods Med Res 2017; 26(4): 1605–1610.

11.

Wallace

Moodie

EEM

Stephens

DA.

SMART thinking: a review of recent developments in Sequential Multiple Assignment Randomized Trials. Curr Epidemiol Rep 2016; 3: 225–232.

12.

Moher

Hopewell

Schulz

, et al. CONSORT 2010 explanation and elaboration: updated guidelines for reporting parallel group randomised trials. Int J Surg 2012; 10: 28–55.

13.

Lewis

JA.

Statistical principles for clinical trials (ICH E9): an introductory note on an international guideline. Stat Med 1999; 18: 1903–1942.

14.

Fitzpatrick

Perrier

Shakik

, et al. Assessment of long-term follow-up of randomized trial participants by linkage to routinely collected data: a scoping review and analysis. JAMA Netw Open 2018; 1: e186019.

15.

Xie

Sun

Ingram

, et al. Use of routine healthcare data in randomised implementation trials: a methodological mixed-methods systematic review. Implement Sci 2023; 18: 47.

16.

Gallagher

Kousoulis

Williams

, et al. Clinical Practice Research Datalink (CPRD). In: Sturkenboom

Schink

(eds) Databases for pharmacoepidemiological research. Cham: Springer International Publishing, 2021, pp. 57–65.

17.

Sydes

Barbachano

Bowman

, et al. Realising the full potential of data-enabled trials in the UK: a call for action. BMJ Open 2021; 11: e043906.

18.

Rugo

Olopade

DeMichele

, et al. Adaptive randomization of veliparib-carboplatin treatment in breast cancer. N Engl J Med 2016; 375: 23–34.

19.

Tam

Laurent

Gauthier

, et al. Prediction of cognitive decline for enrichment of Alzheimer’s disease clinical trials. J Prev Alzheimers Dis 2021; 9: 400–409.

20.

Lei

Nahum-Shani

Lynch

, et al. A ‘SMART’ design for building individualized treatment sequences. Annu Rev Clin Psychol 2012; 8: 21–48.

21.

Baumfeld Andre

Reynolds

Caubel

, et al. Trial designs using real-world data: the changing landscape of the regulatory approval process. Pharmacoepidemiol Drug Saf 2020; 29(10): 1201–1212.

22.

Khosla

White

Medina

, et al. Real world evidence (RWE) – a disruptive innovation or the quiet evolution of medical evidence generation? F1000res 2018; 7: 111.

23.

Kwakkenbos

Imran

McCall

, et al. CONSORT extension for the reporting of randomised controlled trials conducted using cohorts and routinely collected data (CONSORT-ROUTINE): checklist with explanation and elaboration. BMJ 2021; 373: n857.

24.

McCarthy

O’Keeffe

Williamson

, et al. A study protocol for the development of a SPIRIT extension for trials conducted using cohorts and routinely collected data (SPIRIT-ROUTINE). HRB Open Res 2021; 4: 82.

25.

Egesa

Bonnett

Emsley

, et al. How is external data utilised in Sequential Multiple Assignment Randomization Trials: a systematic review protocol. OSF Registrations; 2024. https://osf.io/k6zf4/ (accessed 10 February 2025).

26.

The EndNote Team. EndNote. 20th ed. Philadelphia, PA: Clarivate, 2013.

27.

Ouzzani

Hammady

Fedorowicz

, et al. Rayyan – a web and mobile app for systematic reviews. Syst Rev 2016; 5: 210.

28.

Microsoft Corporation. Microsoft Excel. Microsoft 365 ed. Redmond, WA: Microsoft Corporation, 2024. https://www.microsoft.com/microsoft-365/excel

29.

R Core Team. R: a language and environment for statistical computing. Vienna: R Foundation for Statistical Computing, 2023.

30.

Bianchi

Mosca

Dalla Volta

, et al. Maintenance versus discontinuation of androgen deprivation therapy during continuous or intermittent docetaxel administration in castration-resistant prostate cancer patients: a multicentre, randomised Phase III study by the Piemonte Oncology Network. Eur J Cancer 2021; 155: 127–135.

31.

Cinciripini

Green

Shete

, et al. Smoking cessation after initial treatment failure with varenicline or nicotine replacement: a randomized clinical trial. JAMA 2024; 331: 1722–1731.

32.

Fatori

De Bragança Pereira

Asbahr

, et al. Adaptive treatment strategies for children and adolescents with Obsessive-Compulsive Disorder: a sequential multiple assignment randomized trial. J Anxiety Disord 2018; 58: 42–50.

33.

Fortney

Heagerty

Bauer

, et al. Study to promote innovation in rural integrated telepsychiatry (SPIRIT): rationale and design of a randomized comparative effectiveness trial of managing complex psychiatric disorders in rural primary care clinics. Contemp Clin Trials 2020; 90: 105873.

34.

Friedman

Chen

Campbell

, et al. A sequential, multiple-assignment, randomized trial of analgesic strategies for acute musculoskeletal pain. Am J Emerg Med 2024; 82: 15–20.

35.

Rothman

Vock

, et al. Optimizing longitudinal tobacco cessation treatment in lung cancer screening: a sequential, multiple assignment, randomized trial. JAMA Netw Open 2023; 6: e2329903.

36.

Gao

Arnold

Prihoda

, et al. Sequential Multiple Assignment Randomized Treatment (SMART) for bipolar disorder at any phase of illness and at least mild symptom severity. Psychopharmacol Bull 2020; 50: 8–25.

37.

Geng

Odeny

Montoya

, et al. Adaptive strategies for retention in care among persons living with HIV. NEJM Evid. Epub ahead of print 28 March 2023. DOI: 10.1056/EVIDoa2200076.

38.

Grilo

White

Masheb

, et al. Randomized controlled trial testing the effectiveness of adaptive ‘SMART’ stepped-care treatment for adults with binge-eating disorder co-morbid with obesity. Am Psychol 2020; 75(2): 204–218.

39.

Hammel

Huguet

Van Laethem

, et al. Effect of chemoradiotherapy vs chemotherapy on survival in patients with locally advanced pancreatic cancer controlled after 4 months of gemcitabine with or without erlotinib: the LAP07 randomized clinical trial. JAMA 2016; 315: 1844–1853.

40.

Karp

Zhang

Wahed

, et al. Improving patient reported outcomes and preventing depression and anxiety in older adults with knee osteoarthritis: results of a Sequenced Multiple Assignment Randomized Trial (SMART) study. Am J Geriatr Psychiatry 2019; 27(10): 1035–1045.

41.

Kasari

Kaiser

Goods

, et al. Communication interventions for minimally verbal children with autism: a sequential multiple assignment randomized trial. J Am Acad Child Adolesc Psychiatry 2014; 53(6): 635–646.

42.

Kim

Asher

Burkhauser

, et al. Using a Sequential Multiple Assignment Randomized Trial (SMART) to develop an adaptive K–2 literacy intervention with personalized print texts and app-based digital activities. AERA Open. Epub ahead of print 29 August 2019. DOI: 10.1177/2332858419872701.

43.

Lebeau

Chastang

Allard

, et al. Six vs twelve cycles for complete responders to chemotherapy in small cell lung cancer: definitive results of a randomized clinical trial. Eur Respir J 1992; 5(3): 286–290.

44.

Matthay

Reynolds

Seeger

, et al. Long-term results for children with high-risk neuroblastoma treated on a randomized trial of myeloablative therapy followed by 13-cis-retinoic acid: a children’s oncology group study. J Clin Oncol 2009; 27: 1007–1013.

45.

McKay

Drapkin

Van Horn

, et al. Effect of patient choice in an adaptive sequential randomization trial of treatment for alcohol and cocaine dependence. J Consult Clin Psychol 2015; 83(6): 1021–1032.

46.

Morgenstern

Kuerbis

Shao

, et al. An efficacy trial of adaptive interventions for alcohol use disorder. J Subst Abuse Treat 2021; 123: 108264.

47.

Morin

Edinger

Beaulieu-Bonneau

, et al. Effectiveness of sequential psychological and medication therapies for insomnia disorder: a randomized clinical trial. JAMA Psychiatry 2020; 77: 1107–1115.

48.

Naar-King

Ellis

Idalski Carcone

, et al. Sequential Multiple Assignment Randomized Trial (SMART) to construct weight loss interventions for African American adolescents. J Clin Child Adolesc Psychol 2016; 45(4): 428–441.

49.

Patrick

Lyden

Morrell

, et al. Main outcomes of M-bridge: a sequential multiple assignment randomized trial (SMART) for developing an adaptive preventive intervention for college drinking. J Consult Clin Psychol 2021; 89(7): 601–614.

50.

Pelham

Jr Fabiano

Waxmonsky

, et al. Treatment sequencing for childhood ADHD: a multiple-randomization study of adaptive medication and behavioral interventions. J Clin Child Adolesc Psychol 2016; 45: 396–415.

51.

Petracci

Scarpi

Passardi

, et al. Effectiveness of bevacizumab in first- and second-line treatment for metastatic colorectal cancer: ITACa randomized trial. Ther Adv Med Oncol. Epub ahead of print 23 July 2020. DOI: 10.1177/1758835920937427.

52.

Petry

Alessi

Rash

, et al. A randomized trial of contingency management reinforcing attendance at treatment: do duration and timing of reinforcement matter. J Consult Clin Psychol 2018; 86(10): 799–809.

53.

Powell

Moser

Stock

, et al. Arsenic trioxide improves event-free and overall survival for adults with acute promyelocytic leukemia: North American Leukemia Intergroup Study C9710. Blood 2010; 116: 3751–3757.

54.

Schlam

Baker

Piper

, et al. What to do after smoking relapse? A sequential multiple assignment randomized trial of chronic care smoking treatments. Addiction 2024; 119(5): 898–914.

55.

Schmitz

Stotts

Vujanovic

, et al. Contingency management plus acceptance and commitment therapy for initial cocaine abstinence: results of a sequential multiple assignment randomized trial (SMART). Drug Alcohol Depend 2024; 256: 111078.

56.

Sherwood

Crain

Seburg

, et al. BestFIT sequential multiple assignment randomized trial results: a SMART approach to developing individualized weight loss treatment sequences. Ann Behav Med 2022; 56: 291–304.

57.

Sikorskii

Badger

Segrin

, et al. A sequential multiple assignment randomized trial of symptom management after chemotherapy. J Pain Symptom Manage 2023; 65(6): 541.e2–552.e2.

58.

Sinyor

Schaffer

Levitt

The Sequenced Treatment Alternatives to Relieve Depression (STAR*D) trial: a review. Can J Psychiatry 2010; 55: 126–135.

59.

Smith

Almirall

Choi

, et al. Primary aim results of a clustered SMART for developing a school-level, adaptive implementation strategy to support CBT delivery at high schools in Michigan. Implement Sci 2022; 17: 42.

60.

Somers

Winger

Fisher

, et al. Behavioral cancer pain intervention dosing: results of a Sequential Multiple Assignment Randomized Trial. Pain 2023; 164: 1935–1941.

61.

Swartz

Stroup

McEvoy

, et al. What CATIE found: results from the schizophrenia trial. Psychiatr Serv 2008; 59: 500–506.

62.

Thall

Logothetis

Pagliaro

, et al. Adaptive therapy for androgen-independent prostate cancer: a randomized selection trial of four regimens. J Natl Cancer Inst 2007; 99: 1613–1622.

63.

Tummarello

Mari

Graziano

, et al. A randomized, controlled phase III study of cyclophosphamide, doxorubicin, and vincristine with etoposide (CAV-E) or teniposide (CAV-T), followed by recombinant interferon-alpha maintenance therapy or observation, in small cell lung carcinoma patients with complete responses. Cancer 1997; 80: 2222–2229.

64.

Wyatt

Lehto

Guha-Niyogi

, et al. Reflexology and meditative practices for symptom management among people with cancer: results from a sequential multiple assignment randomized trial. Res Nurs Health 2021; 44(5): 796–810.

65.

Abuogi

Kulzer

Akama

, et al. Adapt for Adolescents: protocol for a sequential multiple assignment randomized trial to improve retention and viral suppression among adolescents and young adults living with HIV in Kenya. Contemp Clin Trials 2023; 127: 107123.

66.

Arean

Hull

Pullmann

, et al. Protocol for a sequential, multiple assignment, randomised trial to test the effectiveness of message-based psychotherapy for depression compared with telepsychotherapy. BMJ Open 2021; 11: e046958.

67.

Auyeung

Long

Royster

, et al. Sequential multiple-assignment randomized trial design of neurobehavioral treatment for patients with metastatic malignant melanoma undergoing high-dose interferon-alpha therapy. Clin Trials 2009; 6(5): 480–490.

68.

Bahraini

Matarazzo

Barry

, et al. Protocol: examining the effectiveness of an adaptive implementation intervention to improve uptake of the VA suicide risk identification strategy: a sequential multiple assignment randomized trial. Implement Sci 2020; 15: 58.

69.

Becker

Banaschewski

Brandeis

, et al. Individualised stepwise adaptive treatment for 3-6-year-old preschool children impaired by attention-deficit/hyperactivity disorder (ESCApreschool): study protocol of an adaptive intervention study including two randomised controlled trials within the consortium ESCAlife. Trials 2020; 21: 56.

70.

Belzer

MacDonell

Ghosh

, et al. Adaptive antiretroviral therapy adherence interventions for youth living with HIV through text message and cell phone support with and without incentives: protocol for a sequential multiple assignment randomized trial (SMART). JMIR Res Protoc 2018; 7: e11183.

71.

Buchholz

Wilbur

Halloway

, et al. Study protocol for a sequential multiple assignment randomized trial (SMART) to improve physical activity in employed women. Contemp Clin Trials 2020; 89: 105921.

72.

Carr

Whiston

O’Reilly

, et al. Sequential multiple assignment randomised trial to develop an adaptive mobile health intervention to increase physical activity in people poststroke in the community setting in Ireland: TAPAS trial protocol. BMJ Open 2024; 14: e072811.

73.

Comins

Schwartz

Phetlhu

, et al. Siyaphambili protocol: an evaluation of randomized, nurse-led adaptive HIV treatment interventions for cisgender female sex workers living with HIV in Durban, South Africa. Res Nurs Health 2019; 42(2): 107–118.

74.

Drake

Kalmbach

Cheng

, et al. Sleep to Reduce Incident Depression Effectively (STRIDE): study protocol for a randomized controlled trial comparing stepped-care cognitive-behavioral therapy for insomnia versus sleep education control to prevent major depression. Trials 2022; 23: 967.

75.

Edelman

Dziura

Deng

, et al. A SMARTTT approach to Treating Tobacco use disorder in persons with HIV (SMARTTT): rationale and design for a hybrid type 1 effectiveness-implementation study. Contemp Clin Trials 2021; 110: 106379.

76.

Eldridge-Smith

Manber

Tsai

, et al. Stepped care management of insomnia co-occurring with sleep apnea: the AIR study protocol. Trials 2022; 23: 806.

77.

Fernandez

Schlechter

Del Fiol

, et al. QuitSMART Utah: an implementation study protocol for a cluster-randomized, multi-level sequential multiple assignment randomized trial to increase reach and impact of tobacco cessation treatment in community health centers. Implement Sci 2020; 15: 9.

78.

Flynn

Eaton

Langford

, et al. A SMART design to determine the optimal treatment of chronic pain among military personnel. Contemp Clin Trials 2018; 73: 68–74.

79.

Fox

Vock

Sherwood

, et al. SMART use of medications for the treatment of adolescent severe obesity: a sequential multiple assignment randomized trial protocol. Contemp Clin Trials 2024; 138: 107444.

80.

Fritz

Rhon

Teyhen

, et al. A Sequential Multiple-Assignment Randomized Trial (SMART) for stepped care management of low back pain in the military health system: a trial protocol. Pain Med 2020; 21: S73–S82.

81.

Germeroth

Benno

Kolko Conlon

, et al. Trial design and methodology for a non-restricted sequential multiple assignment randomized trial to evaluate combinations of perinatal interventions to optimize women’s health. Contemp Clin Trials 2019; 79: 111–121.

82.

Hassett

Williams

Harris

, et al. An interventional response phenotyping study in chronic low back pain: protocol for a mechanistic randomized controlled trial. Pain Med 2023; 24: S126–S138.

83.

Hibbard

Friedstat

Thomas

, et al. LIBERTI: a SMART study in plastic surgery. Clin Trials 2018; 15(3): 286–293.

84.

Johnson

Wiltsey-Stirman

Sikorskii

, et al. Protocol for the ROSE sustainment (ROSES) study, a sequential multiple assignment randomized trial to determine the minimum necessary intervention to maintain a postpartum depression prevention program in prenatal clinics serving low-income women. Implement Sci 2018; 13: 115.

85.

Kilbourne

Almirall

Eisenberg

, et al. Protocol: Adaptive Implementation of Effective Programs Trial (ADEPT): cluster randomized SMART trial comparing a standard versus enhanced implementation strategy to improve outcomes of a mood disorders program. Implement Sci 2014; 9: 132.

86.

Kopelowicz

Nandy

Ruiz

, et al. Improving self-management of Type 2 diabetes in Latinx patients: protocol for a sequential multiple assignment randomized trial involving community health workers, registered nurses, and family members. JMIR Res Protoc 2023; 12: e44793.

87.

Kor

PPK

Chou

Zarit

, et al. Sequential multiple assignment randomised controlled trial protocol for developing an adaptive intervention to improve depressive symptoms among family caregivers of people with dementia. BMJ Open 2023; 13: e072410.

88.

Levy

Mathai

Chatterjee

, et al. Implementation research for public sector mental health care scale-up (SMART-DAPPER): a sequential multiple, assignment randomized trial (SMART) of non-specialist-delivered psychotherapy and/or medication for major depressive disorder and posttraumatic stress disorder (DAPPER) integrated with outpatient care clinics at a county hospital in Kenya. BMC Psychiatry 2019; 19: 424.

89.

Guo

Fan

, et al. Sequential Multiple-Assignment Randomized Trials to Compare Antipsychotic Treatments (SMART-CAT) in first-episode schizophrenia patients: rationale and trial design. Schizophr Res 2021; 230: 87–94.

90.

Liu

Chen

, et al. Sequential multiple assignment randomised trial of a brief contact intervention for suicide risk management among discharged psychiatric patients: an implementation study protocol. BMJ Open 2021; 11: e054131.

91.

Mustanski

Moskowitz

Moran

, et al. Evaluation of a stepped-care eHealth HIV prevention program for diverse adolescent men who have sex with men: protocol for a hybrid type 1 effectiveness implementation trial of SMART. JMIR Res Protoc 2020; 9: e19701.

92.

Nelson

Amminger

Yuen

, et al. Staged treatment in early psychosis: a sequential multiple assignment randomised trial of interventions for ultra high risk of psychosis patients. Early Interv Psychiatry 2018; 12(3): 292–306.

93.

O’Keefe

Haroz

Goklish

, et al. Employing a sequential multiple assignment randomized trial (SMART) to evaluate the impact of brief risk and protective factor prevention interventions for American Indian Youth Suicide. BMC Public Health 2019; 19: 1675.

94.

Osilla

Gore

Pedersen

, et al. Study protocol for a sequential multiple assignment randomized trial to reduce risky drinking among service members and their partners. Contemp Clin Trials 2023; 133: 107324.

95.

Peter

Murphy

Witkiewitz

, et al. Use of a sequential multiple assignment randomized trial to test contingency management and an integrated behavioral economic and mindfulness intervention for buprenorphine-naloxone medication adherence for opioid use disorder. Trials 2023; 24: 237.

96.

Peterson

West

Weisz

, et al. A sequential Multiple Assignment Randomized Trial (SMART) study of medication and CBT sequencing in the treatment of pediatric anxiety disorders. BMC Psychiatry 2021; 21: 323.

97.

Quanbeck

Almirall

Jacobson

, et al. The Balanced Opioid Initiative: protocol for a clustered, sequential, multiple-assignment randomized trial to construct an adaptive implementation strategy to improve guideline-concordant opioid prescribing in primary care. Implement Sci 2020; 15: 26.

98.

Rabin

Cain

Watson

Jr , et al. Scaling and sustaining COVID-19 vaccination through meaningful community engagement and care coordination for underserved communities: hybrid type 3 effectiveness-implementation sequential multiple assignment randomized trial. Implement Sci 2023; 18: 28.

99.

Sabri

Glass

Murray

, et al. A technology-based intervention to improve safety, mental health and empowerment outcomes for immigrant women with intimate partner violence experiences: it’s weWomen plus sequential multiple assignment randomized trial (SMART) protocol. BMC Public Health 2021; 21: 1956.

100.

Smith

Somers

Kuhn

, et al. A SMART approach to optimizing delivery of an mHealth intervention among cancer survivors with posttraumatic stress symptoms. Contemp Clin Trials 2021; 110: 106569.

101.

Sripada

Smith

Walters

, et al. Testing adaptive interventions to improve PTSD treatment outcomes in Federally Qualified Health Centers: protocol for a randomized clinical trial. Contemp Clin Trials 2023; 129: 107182.

102.

Van Heerden

Szpiro

Ntinga

, et al. A Sequential Multiple Assignment Randomized Trial of scalable interventions for ART delivery in South Africa: the SMART ART study. Trials 2023; 24: 32.

103.

Velloza

Poovan

Ndlovu

, et al. Adaptive HIV pre-exposure prophylaxis adherence interventions for young South African women: study protocol for a sequential multiple assignment randomized trial. PLoS One 2022; 17(4): e0266665.

104.

Walton

Carter

Seewald

, et al. Adaptive interventions for alcohol misuse and violent behaviors among adolescents and emerging adults in the emergency department: a sequential multiple assignment randomized controlled trial protocol. Contemp Clin Trials 2023; 130: 107218.

105.

Wan

Cai

Liao

, et al. Smartphone-based versus traditional face-to-face collaborative care for community-dwelling older adults living with dementia in China: protocol for an implementation science-based sequential multiple assignment randomised trial. BMJ Open 2023; 13: e067406.

106.

Windsor

Benoit

Pinto

, et al. Optimization of a new adaptive intervention using the SMART Design to increase COVID-19 testing among people at high risk in an urban community. Trials 2022; 23: 310.

107.

Zhao

Weng

Luk

, et al. Adaptive interventions to optimise the mobile phone-based smoking cessation support: study protocol for a sequential, multiple assignment, randomised trial (SMART). Trials 2022; 23: 681.

108.

Zhou

Lee

M-C

Atieli

, et al. Adaptive interventions for optimizing malaria control: an implementation study protocol for a block-cluster randomized, sequential multiple assignment trial. Trials 2020; 21: 665.

109.

Zullig

Shahsahebi

Neely

, et al. Low-touch, team-based care for co-morbidity management in cancer patients: the ONE TEAM randomized controlled trial. BMC Fam Pract 2021; 22: 234.

110.

Shaffer

JP.

Multiple hypothesis testing. Annu Rev Psychol 1995; 46: 561–584.

111.

Chan

Tetzlaff

Altman

, et al. SPIRIT 2013 statement: defining standard protocol items for clinical trials. Ann Intern Med 2013; 158: 200–207.

112.

Schulz

Altman

Moher

, et al. CONSORT 2010 statement: updated guidelines for reporting parallel group randomised trials. BMC Med 2010; 8: 18.

113.

Nahum-Shani

Qian

Almirall

, et al. Experimental design and primary data analysis methods for comparing adaptive interventions. Psychol Methods 2012; 17(4): 457–477.

114.

Kidwell

SMART design workshop information, 2025, https://smart-workshops.com/smart-design-info

115.

d3center. Module 2: introduction to sequential, multiple assignment, randomized trial (SMART) designs, 2025, https://d3c.isr.umich.edu/online-courses/module-2-introduction-to-sequential-multiple-assignment-randomized-trial-smart-designs/

116.

Seewald

. SMART designs for developing adaptive interventions, in press.

117.

Hampton

Chow

Bhat

, et al. Reporting and design considerations for SMART behavioral science research. Educ Psychol Rev 2023; 35: 117.

118.

Humphrey-Murto

Varpio

Wood

, et al. The use of the Delphi and other consensus group methods in medical education research: a review. Acad Med 2017; 92(10): 1491–1498.

119.

Moher

Schulz

Simera

, et al. Guidance for developers of health research reporting guidelines. PLoS Med 2010; 7: e1000217.

120.

Lewsey

Leyland

Murray

, et al. Using routine data to complement and enhance the results of randomised controlled trials. Health Technol Assess 2000; 4(22): 1–55.

121.

Raman

Qualls

Hammill

, et al. Optimizing data integration in trials that use EHR data: lessons learned from a multi-center randomized clinical trial. Trials 2023; 24: 566.

122.

Goudie

Sutton

Jones

, et al. Empirical assessment suggests that existing evidence could be used more fully in designing randomized controlled trials. J Clin Epidemiol 2010; 63(9): 983–991.

123.

Cake

Ogburn

Pinches

, et al. Correction: development and evaluation of rapid data-enabled access to routine clinical information to enhance early recruitment to the national clinical platform trial of COVID-19 community treatments. Trials 2023; 24: 598.

124.

Llewellyn-Bennett

Edwards

Roberts

, et al. Post-trial follow-up methodology in large randomised controlled trials: a systematic review. Trials 2018; 19: 298.

125.

NajafZadeh

Fernández Oromendia

Burcu

, et al. Linkage of clinical trial data to routinely collected data sources: a scoping review. JAMA Netw Open 2025; 8: e257797.

126.

NIH Big Data to Knowledge (BD2K) Think Tank. EHR data methodologies in clinical research: perspectives from the field (meeting summary). NIH Big Data to Knowledge (BD2K) Think Tank, 2015, https://datascience.nih.gov/sites/default/files/bd2k/docs/BD2K__EHR_Data_Methods_for_Research_TT_summary_4-17-2015.pdf

127.

Chevret

Timsit

J-F

Biard

Challenges of using external data in clinical trials – an illustration in patients with COVID-19. BMC Med Res Methodol 2022; 22: 321.

128.

Liu

Wang

Zeng

Sequential multiple assignment randomization trials with enrichment design. Biometrics 2016; 73: 378–390.

129.

Krakow

Hemmer

Wang

, et al. Tools for the precision medicine era: how to develop highly personalized treatment recommendations from cohort and registry data using Q-learning. Am J Epidemiol 2017; 186: 160–172.

130.

Bellman

Dynamic programming. Garden City, NY: Dover Publications, 2013.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.21 MB