Cluster-level dynamic treatment regimens can be used to guide sequential treatment decision-making at the cluster level in order to improve outcomes at the individual or patient-level. In a cluster-level dynamic treatment regimen, the treatment is potentially adapted and re-adapted over time based on changes in the cluster that could be impacted by prior intervention, including aggregate measures of the individuals or patients that compose it. Cluster-randomized sequential multiple assignment randomized trials can be used to answer multiple open questions preventing scientists from developing high-quality cluster-level dynamic treatment regimens. In a cluster-randomized sequential multiple assignment randomized trial, sequential randomizations occur at the cluster level and outcomes are observed at the individual level. This manuscript makes two contributions to the design and analysis of cluster-randomized sequential multiple assignment randomized trials. First, a weighted least squares regression approach is proposed for comparing the mean of a patient-level outcome between the cluster-level dynamic treatment regimens embedded in a sequential multiple assignment randomized trial. The regression approach facilitates the use of baseline covariates which is often critical in the analysis of cluster-level trials. Second, sample size calculators are derived for two common cluster-randomized sequential multiple assignment randomized trial designs for use when the primary aim is a between-dynamic treatment regimen comparison of the mean of a continuous patient-level outcome. The methods are motivated by the Adaptive Implementation of Effective Programs Trial which is, to our knowledge, the first-ever cluster-randomized sequential multiple assignment randomized trial in psychiatry.
MurrayDM. Design and analysis of group-randomized trials1998; Vol. 29, USA: Oxford University Press.
2.
RaudenbushSWBrykAS. Hierarchical linear models: applications and data analysis methods2002; Vol. 1, Thousand Oaks, California: Sage.
3.
DonnerAKlarN. Design and analysis of cluster randomization trials in health research2010; Vol. 1, New York, New York: Wiley.
4.
KilbourneAMAbrahamKMGoodrichDE, et al.Cluster randomized adaptive implementation trial comparing a standard versus enhanced implementation intervention to improve uptake of an effective re-engagement program for patients with serious mental illness. Implement Sci2013; 8: 1–14.
5.
MurphyS. An experimental design for the development of adaptive treatment strategies. Stat Med2005; 24: 1455–1481.
6.
LeiHNahum-ShaniILynchK, et al.A SMART design for building individualized treatment sequences. Ann Rev Clin Psychol2012; 8: 21–48.
7.
ChakrabortyBMoodieE. Statistical methods for dynamic treatment regimes, New York, New York: Springer, 2013.
8.
LavoriPWDawsonR. Introduction to dynamic treatment strategies and sequential multiple assignment randomization. Clin Trials2014; 11: 393–399.
9.
KosorokMRMoodieEE. Adaptive treatment strategies in practice: planning trials and analyzing data for personalized medicine, Philadelphia, Pennsylvania: SIAM, 2015.
KilbourneAMAlmirallDEisenbergD, 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 Sci2014; 9: 132–132.
12.
Nahum-ShaniIQianMAlmirallD, et al.Experimental design and primary data analysis methods for comparing adaptive interventions. Psychol Meth2012; 17: 457–477.
13.
OrellanaLRotnitzkyARobinsJ. Dynamic regime marginal structural mean models for estimating optimal dynamic treatment regimes, part I: main content. Int J Biostat2010; 6: Article 8.
14.
OettingALevyJWeissR, et al.Statistical methodology for a SMART design in the development of adaptive treatment strategies. In: ShroutPKeyesKOrnsteinK (eds). Causality and psychopathology: finding the determinants of disorders and their cures, Arlington, VA: Oxford University Press, 2010, pp. 179–205.
15.
LiZMurphySA. Sample size formulae for two-stage randomized trials with survival outcomes. Biometrika2011; 98: 503–518.
GhoshPCheungYChakrabortyBSample size calculations for clustered SMART designs. In: KosorokMRMoodieEE (eds). Adaptive treatment strategies in practice: planning trials and analyzing data for personalized medicine, chapter 5, Alexandria, Virginia: SIAM, 2015, pp. 55–70.
18.
LavoriPDawsonD. A design for testing clinical strategies: biased individually tailored within-subject randomization. J Royal Stat Soc Ser A2000; 163: 29–38.
19.
KilbourneAMGoodrichDELaiZ, et al.Life goals collaborative care for patients with bipolar disorder and cardiovascular disease risk. Psychiat Serv2012; 63: 1234–1238.
20.
KasariCKaiserAGoodsK, et al.Communication interventions for minimally verbal children with autism: sequential multiple assignment randomized trial. J Am Acad Child Adoles Psychiatr2014; 53: 635–646.
21.
AlmirallDDiStefanoCChangYC, et al.Longitudinal effects of adaptive interventions with a speech-generating device in minimally verbal children with ASD. J Clin Child Adoles Psychol2016; 45: 442–456.
22.
Chronis-TuscanoAWangCHStricklandJ, et al.Personalized treatment of mothers with adhd and their young at-risk children: a SMART pilot. J Clin Child Adoles Psychol2016; 45: 510–521.
23.
LuXLynchKGOslinDW, et al.Comparing treatment policies with assistance from the structural nested mean model. Biometrics2016; 72: 10–19.
24.
PelhamWEFabianoGAWaxmonskyJG, et al.Treatment sequencing for childhood adhd: a multiple-randomization study of adaptive medication and behavioral interventions. J Clin Child Adoles Psychol2016; 45: 396–415.
25.
Gunlicks-StoesselMMufsonLWesterveltA, et al.A pilot SMART for developing an adaptive treatment strategy for adolescent depression. J Clin Child Adoles Psychol2016; 45: 480–494.
26.
AugustGJPiehlerTFBloomquistML. Being “smart” about adolescent conduct problems prevention: executing a SMART pilot study in a juvenile diversion agency. J Clin Child Adoles Psychol2016; 45: 495–509.
27.
SherwoodNEButrynMLFormanEM, et al.The bestfit trial: a smart approach to developing individualized weight loss treatments. Contem Clin Trials2016; 47: 209–216.
28.
Naar-KingSEllisDAIdalski CarconeA, et al.Sequential multiple assignment randomized trial (SMART) to construct weight loss interventions for African American adolescents. J Clin Child Adoles Psychol2016; 45: 428–441.
29.
McKayJRDrapkinMLVan HornDH, et al.Effect of patient choice in an adaptive sequential randomization trial of treatment for alcohol and cocaine dependence. J Consult Clin Psychol2015; 83: 1021–1021.
30.
AlmirallDNahum-ShaniISherwoodN, et al.Introduction to SMART designs for the development of adaptive interventions: with application to weight loss research. Transl Behav Med2014; 4: 260–274.
31.
NeymanJIwaszkiewiczKKolodziejczykS. Statistical problems in agricultural experimentation. Suppl J Royal Stat Soc1935; 2: 107–180.
32.
RubinDB. Bayesian inference for causal effects: the role of randomization. Ann Stat1978; 6: 34–58.
33.
RobinsJM. Association, causation, and marginal structural models. Synthese1999; 121: 151–179.
34.
HernánMBrumbackBRobinsJ. Marginal structural models to estimate the causal effect of zidovudine on the survival of hiv-positive men. Epidemiology2000; 11: 561–570.
35.
RobinsJMHernanMBrumbackB. Marginal structural models and causal inference. Epidemiology2000; 11: 550–560.
36.
MurphySAvan der LaanMJRobinsJM, et al.Marginal mean models for dynamic regimes. J Am Stat Assoc2001; 96: 1410–1423.
37.
LiangKYZegerSL. Longitudinal data analysis using generalized linear models. Biometrika1986; 73: 13–22.
38.
LiangKYZegerSL. Regression analysis for correlated data. Ann Rev Pub Health1993; 14: 43–68.
39.
RobinsJMRotnitzkyAZhaoLP. Analysis of semiparametric regression models for repeated outcomes in the presence of missing data. J Am Stat Assoc1995; 90: 106–121.
40.
HernanMBrumbackBRobinsJ. Estimating the causal effect of zidovudine on cd4 count with a marginal structural model for repeated measures. Stat Med2002; 21: 1689–1709.
41.
HiranoKImbensGWRidderG. Efficient estimation of average treatment effects using the estimated propensity score. Econometrica2003; 71: 1161–1189.
42.
BembomOvan der LaanM. Statistical methods for analyzing sequentially randomized trials. J Natl Cancer Inst2007; 99: 1577–1582.
43.
BrumbackBA. A note on using the estimated versus the known propensity score to estimate the average treatment effect. Stat Prob Lett2009; 79: 537–542.
44.
WilliamsonEJForbesAWhiteIR. Variance reduction in randomised trials by inverse probability weighting using the propensity score. Stat Med2014; 33: 721–737.
45.
CohenJ. Statistical power analysis for the behavioral sciences, 2nd ed. Hillsdale, NJ: Lawrence Earlbaum Associates, 1988.
46.
KerrySMBlandMJ. Unequal cluster sizes for trials in english and welsh general practice: implications for sample size calculations. Stat Med2001; 20: 377–390.
47.
EldridgeSMUkoumunneOCCarlinJB. The intra-cluster correlation coefficient in cluster randomized trials: a review of definitions. Int Stat Rev2009; 77: 378–394.
48.
HedekerDGibbonsRD. Longitudinal data analysis2006; Vol. 451, Hoboken, New Jersey: John Wiley & Sons.
49.
QianMMurphySA. Performance guarantees for individualized treatment rules. Ann Stat2011; 39: 1180–1180.
50.
LinnKALaberEBStefanskiLA. Interactive q-learning for quantiles. Journal of the American Statistical Association2017; 112: 638–649.
51.
MoodieEEDeanNSunYR. Q-learning: flexible learning about useful utilities. Stat Biosci2014; 6: 223–243.
52.
LaberEBZhaoY. Tree-based methods for individualized treatment regimes. Biometrika2015; 102: 501–514.
53.
ZhangYLaberEBTsiatisA, et al.Using decision lists to construct interpretable and parsimonious treatment regimes. Biometrics2015; 71: 895–904.
54.
ZhaoYQZengDLaberEB, et al.New statistical learning methods for estimating optimal dynamic treatment regimes. J Am Stat Assoc2015; 110: 583–598.
55.
Zhou X, Mayer-Hamblett N, Khan U, et al. Residual weighted learning for estimating individualized treatment rules. J Am Stat Assoc 2015, http://dx.doi.org/10.1080/01621459.2015.1093947. (accessed 20 April 2017).
56.
Kasari C, Gulsrud A and Almirall D. Getting SMART about social and academic engagement of elementary aged students with autism spectrum disorder, http://ies.ed.gov/funding/grantsearch/details.asp?ID=1758 (2016, accessed 20 April 2017).
57.
RobinsJCausal inference from complex longitudinal data. In: MaiaB (ed). Latent variable modeling and applications to causality. Lecture notes in statistics, New York: Springer-Verlag, 1997.
58.
HedgesLRhoadsC. Statistical power analysis in education research2009. National Center for Special Education Research, Institute of Education Sciences, U.S. Department of Education, 2009. NCSER 2010-3006.
59.
LuXNahum-ShaniIKasariC, et al.Comparing dynamic treatment regimes using repeated-measures outcomes: modeling considerations in SMART studies. Stat Med2016; 35: 1595–1615.
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.
