Efficient estimation of optimal regimes under a no direct effect assumption

Andrea Rotnitzky; James M. Robins; Lin Liu; Zach Shahn

REVIEW 3 major objections 5 minor 37 references

Efficient estimation of optimal regimes under a no direct effect assumption

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Under the no-direct-effect-of-testing assumption, projecting g-estimating functions onto mean-zero testing-treatment residuals yields more efficient, doubly robust estimators of optimal testing and treatment regimes.

desk verdict The core projection-based variance reduction is real and well proven, but the paper's headline near-optimality claim for continuous outcomes rests on an unproved convergence step that should be flagged in the review. read the letter →

arxiv 1908.10448 v2 pith:BRSQVAX3 submitted 2019-08-27 stat.ME stat.AP

Lin Liu , Zach Shahn , James M. Robins , Andrea Rotnitzky This is my paper · ORCID

classification stat.MEstat.AP MSC 62G0562G20

keywords optimaldynamictreatmentregimesstructuralnestedmeanmodelsg-estimationnodirecteffectassumptionsemiparametricefficiencydoublyrobustestimationvalueofinformationscreeningtests

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper establishes that the no-direct-effect (NDE) assumption—that a diagnostic test affects outcomes only through the treatment it triggers—can be converted into a source of statistical precision, not just an identifying restriction. The authors construct estimators for optimal joint testing-and-treatment regimes by taking the usual g-estimating function and subtracting its projection onto a space of mean-zero test-treatment residual variables implied by NDE. The resulting estimators are regular asymptotically linear, never less efficient than standard g-estimators, and strictly more efficient whenever the residual correlation is nonzero, while remaining doubly robust. This matters for cost-benefit analyses and for estimating the value of information of expensive tests, because the efficiency gain can translate into dramatically smaller required sample sizes.

What carries the argument

The central object is the ortho-complement $\Lambda_{\mathrm{NDE}}^{\perp}$ of the tangent space of the NDE model. The paper represents it as $\{T_b = \sum_{t=0}^{K} T_{b,t}\}$, where $T_{b,t}$ is the residual from projecting $D_{b,t} = b_t(\bar H_t,\bar S_t,Y^d)\,W_{t+1}^{-1}(A_t - \mathrm{E}[A_t\mid \bar H_t,S_t])$ onto the space of testing and treatment scores. Here $W_{t+1}$ is the product of treatment probabilities, and the NDE restriction makes every $T_b$ have mean zero. Subtracting the population least-squares projection $c_{\mathrm{OLS}}T_b$ of an influence function $U$ onto this space yields a new influence function, and the Pythagorean theorem guarantees the variance reduction. For feasibility, the paper projects onto a large subspace $\Omega$ built from $b_t^\ast = (\phi_1(Y^d)I_t^{\top},\ldots,\phi_\xi(Y^d)I_t^{\top})^{\top}$, with $I_t$ the vector of indicators of full treatment histories; Theorem 4 supplies recursive least-squares coefficients for the closed-form projection.

What would settle it

Estimate $\mathrm{E}[D_{b,t}]$ from data with variation in testing and treatment, where $D_{b,t} = b_t(\bar H_t,\bar S_t,Y^d)\,W_{t+1}^{-1}(A_t - \mathrm{E}[A_t\mid \bar H_t,S_t])$; a nonzero mean for some $b_t$, or a difference in mean outcome between randomized testing arms with treatment held fixed, refutes the NDE assumption and implies the efficiency gains rest on a false premise.

Watch

Extended reading notes

Core claim

Under the NDE assumption, the paper constructs estimators $\tilde\Psi(q,b)$ that solve $0 = \hat U(q,b,\Psi)$, where $\hat U(q,b,\Psi)$ is the residual from projecting the influence function of the usual opt-SNMM estimating function $\hat U(q,\Psi)$ onto the space $T_b$ of mean-zero random variables implied by NDE. Theorem 3 shows each such estimator is regular asymptotically linear with asymptotic variance $V^{\mathrm{oracle}}(q,b) = J^{-1}\{\operatorname{var}[U(q,\Psi)] - c_{\mathrm{OLS}}\mathrm{E}[T_b T_b^\top]c_{\mathrm{OLS}}^\top\}J^{-\top}$, which is no larger, and strictly smaller whenever $\mathrm{E}[U(q,\Psi^\ast)T_b] \neq 0$, than the variance of the standard g-estimator. The projected estimators remain doubly robust, and the paper provides a closed-form feasible construction based on a large subspace spanned by basis functions of the health outcome and treatment-history indicators, whose efficiency approaches the intractable optimal projection as the basis grows. Simulations and an HIV monitoring application indicate gains that can amount to a roughly 50-fold reduction in variance.

Load-bearing premise

The load-bearing premise is that testing has no direct effect on the health outcome, meaning a test changes outcomes only through the treatment it triggers; if testing affects outcomes directly, the mean-zero property of the residual variables fails and the projected estimators become biased.

Editorial extensions

If this is right

The projected estimators are never less efficient than standard opt-SNMM g-estimators, and strictly more efficient whenever the initial estimating function correlates with the NDE-implied residuals.
The efficiency gain translates into a sample-size reduction: in the HIV monitoring application cited by the paper, exploiting NDE was associated with a roughly 50-fold variance reduction.
The same projection recipe applies to other doubly robust RAL estimators, including estimators of dynamic marginal structural models, so the improvement is not tied to opt-SNMMs.
A feasible, closed-form implementation exists by projecting onto a basis-expanded subspace; its efficiency approaches the semiparametrically optimal projected estimator as the basis dimension grows.
Under stronger NDE variants, such as no direct effect on covariates or on latent test results, larger projection spaces are available, so the estimators are at least as efficient and typically more efficient than under NDE on the health outcome alone.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

An immediate extension is to use the same projected quantities as a specification test: the NDE assumption implies infinitely many mean-zero restrictions, so a data-driven check of whether empirical projections are near zero could precede the efficiency-gaining analysis.
The construction suggests a general principle for causal inference: any domain assumption that generates extra mean-zero variables can be converted into precision by projecting influence functions onto the ortho-complement of the implied tangent space, not just the NDE assumption considered here.
In cost-benefit decisions near a value-of-information threshold, the variance reduction could change conclusions: the paper's simulation shows the projected estimator raising the empirical rejection rate of 'screening is not cost-effective' from around 25 percent to 100 percent, so policy conclusions may flip.
At the boundary where everyone is tested, the NDE assumption identifies parameters that are otherwise unidentified; whether the optimal regime can still be computed there is left open, so a natural extension is to study computation and estimation under such positivity failures.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

First, what to know: this paper earns its keep on Theorem 3. The idea is simple and clever: under the no direct effect assumption, the estimating function from opt-SNMM g-estimation can be projected onto a space of mean-zero variables implied by NDE, and the leftover estimating function is more efficient. That theorem is proved, and the algebra is sound. The simulation results back it up, and the authors ship R code. The NDE-IPW estimator from Robins et al. 2008 is contextualized nicely; the paper clearly shows what is new: no one had used the NDE assumption inside opt-SNMM estimation via influence-function projection before. For discrete outcomes, the closed-form projection in Theorem 4 is a genuine contribution, and the double robustness preservation is real. The HIV example and the value-of-information framing are useful for decision-science readers, not just methodologists.

Now the soft spots, in proportion. The main one is Remark 4, and the stress-test note lands. For continuous Y, the practical estimator projects onto a finite-dimensional basis space of dimension xi. The paper asserts, without proof, that as xi grows, the projection converges to the projection onto the full ortho-complement, and that xi can be chosen to grow slowly with n so the estimators become asymptotically equivalent. The text literally says 'should converge' and 'should converge' again. That is an omitted proof, not a technical aside. If the closure of the union of these finite-dimensional subspaces is a proper subspace of the ortho-complement—say, because the conditional expectations in T_b smooth out high-frequency components—then the practical estimator has a positive variance gap from the oracle, and the 'nearly optimal' selling point is unsupported. The fixed-b variance reduction of Theorem 3 still holds; the paper is just overclaiming the asymptotic equivalence part. This should be fixed by either providing conditions and a proof, or softly rephrasing the claim to 'conjectured to be nearly optimal.'

A second, smaller point: the exceptional-law exclusion is explicit and reasonable, but it means the method does not cover the full optimal-regime space. The NDE assumption itself is substantive and, as the paper acknowledges in Section 7, can fail in ways that bias the projected estimators; that is a modeling assumption, not a math error.

Who should read this: anyone doing causal inference for dynamic treatment regimes, especially with testing-and-treatment decisions. It deserves serious peer review: the central variance-reduction result is correct, the discrete-outcome case is clean, and the simulation evidence is transparent. My clear recommendation: send it to a referee, but instruct them to focus on Remark 4. If that step is not provable, the near-optimality language must be softened before publication.

Referee Report

3 major / 5 minor

Summary. The paper develops semiparametric estimators of the parameters of an optimal regime structural nested mean model (opt-SNMM) under the assumption that a diagnostic test has no direct effect on the health outcome except through treatment choice (the NDE assumption). The main construction subtracts from a standard g-estimating function its projection onto the ortho-complement of the tangent space of the NDE model, thereby producing estimating equations with smaller asymptotic variance. The paper characterizes that ortho-complement (Theorem 1), proves a variance-reduction result for any user-specified projection subspace (Theorem 3), and provides a closed-form projection formula for a finite-dimensional subspace spanned by treatment-history indicators and basis functions of the outcome (Theorem 4 and Corollary 1). For discrete outcomes the subspace can be taken to be the full ortho-complement, while for continuous outcomes the authors propose a finite-basis subspace and assert in Remark 4 that the resulting estimator is asymptotically equivalent to the optimal estimator as the basis dimension grows. The paper also develops a cross-fitted doubly robust feasible estimator and studies the NDE-IPW estimator as a special case, with simulations illustrating large efficiency gains in an HIV monitoring setting.

Significance. If the central results stand, the paper makes a substantive contribution to causal inference for dynamic treatment and testing regimes. Theorem 3 gives a clean projection argument that, for fixed b, any estimator solving the projected estimating equation is RAL and dominates the standard opt-SNMM g-estimator whenever the projection is nonzero, while retaining double robustness. Theorem 1 and Theorem 4 are proved in sufficient detail and provide a workable route to constructing the projection, including a closed form for discrete outcomes. The explicit connection to NDE-IPW estimation and the simulation studies, which show large efficiency gains and improved regime selection, are valuable and are based on reproducible code in the supplementary materials. The main weakness is that the recommended near-optimal estimator for continuous outcomes relies on an unproved convergence assertion in Remark 4; as written, the paper proves improved efficiency over g-estimation for any finite basis but does not prove the advertised approach to the efficiency bound.

major comments (3)

[Sec. 5.2, Remark 4] The claim that \hat\Psi(q,bsub) is asymptotically equivalent to \hat\Psi(q,bopt) for continuous outcomes is supported only by the sentence that the projection 'should converge' to \Pi[U|\Lambda^\perp_NDE]. No theorem establishes L2(P)-density of \cup_\xi \Omega_\xi in \Lambda^\perp_NDE, no rates for \xi(n) are given, and no conditions are stated under which the map b \mapsto T_b preserves L2 density. This statement is load-bearing: it is the basis for recommending \hat\Psi(q,bsub) as a near-optimal estimator. If the closure of \cup_\xi \Omega_\xi is a proper subspace of \Lambda^\perp_NDE, the asymptotic variance of \hat\Psi(q,bsub) has a strictly positive gap from V_oracle(q,bopt), and the paper's central practical message that relative efficiency can be made arbitrarily close to optimal would be unsupported. The fixed-b variance reduction of Theorem 3 would remain valid, but the near-optimality claim requires a proof or explicit sufficient conditions.
[Sec. 4, Eq. (8) and Sec. 7] The entire projection construction uses that E[T_b]=0, which follows from Eq. (8), a consequence of the NDE(Yd) assumption. The manuscript acknowledges in Section 7 that NDE(Yd) can fail in realistic settings (for example, through ancillary care), but it does not quantify the bias of the adjusted estimators under such violations or provide a sensitivity analysis. This is not an internal inconsistency, but it is a substantive limitation of the practical recommendation: the efficiency gains are conditional on an untestable assumption, and the paper would be strengthened by an explicit statement of the resulting bias-variance trade-off or a small sensitivity analysis.
[Sec. 6, Theorem 5] The feasible estimator is shown to be RAL under high-level conditions E[\hat U(q,\Psi^*)|Nu] = op(n^{-1/2}) and \sum_t E[\hat T_{b,t}|Nu] = op(n^{-1/2}). These conditions are stated as sufficient and are standard in the double/debiased machine learning literature, but for the recommended near-optimal estimator with estimated bsub they are combined with the unproved density claim of Remark 4. The paper should make explicit that the practical guarantee for \hat\Psi(q,\hat bsub) is therefore only the fixed-b variance reduction of Theorem 3 unless Remark 4 is upgraded to a theorem.

minor comments (5)

[Sec. 2, paragraph after notation] The sentence 'We let \bar H_m be the sample space of the random vector \bar H_m' uses the same symbol for the random vector and its sample space; this is confusing and should be rephrased, for example by using a script or calligraphic letter for the sample space.
[Sec. 5.2, Remark 4] Remark 4 refers to \phi(Y) while Corollary 1 defines b*_t in terms of \phi(Y_d). Since the total utility Y includes the known testing cost and the NDE assumption concerns Y_d, the paper should clarify which variable is used in the basis functions and why the cost-adjusted version is appropriate or not.
[Introduction, Section 1] The 50-fold efficiency gain attributed to Caniglia et al. [3] concerns an NDE-IPW estimator in a dyn-MSM analysis, not the opt-SNMM estimators developed here; the text should make this distinction explicit to avoid overstating the simulation evidence for the proposed estimators.
[Sec. 3.2, paragraph after Eq. (4)] The paper excludes exceptional laws but does not define them in the main text; a one-sentence definition or a more precise pointer to Robins [27] would make the exclusion self-contained.
[Sec. 5.1, Theorem 3] Theorem 3 assumes \hat\Psi(q,b) is RAL rather than stating conditions under which the estimator solving 0 = \hat U(q,b,\Psi) is RAL; the regularity conditions in Section 6 and Appendix A.5 are stated for the cross-fitted version, and it would help the reader if the theorem explicitly noted that these conditions are being assumed.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: efficiency gain is a direct projection variance identity, NDE is an input, and the Remark 4 gap is a correctness risk, not circularity.

full rationale

The paper's central claim is that projecting the influence function of a baseline RAL estimator onto the space T_b of mean-zero variables implied by the NDE assumption yields an estimator with variance Voracle(q,b) = J^-1(var[U] - c_OLS E[T_b T_b^T] c_OLS^T)J^-T. This is a mathematical identity following from the definition of c_OLS as the population least-squares coefficient; it is not a fitted quantity renamed as a prediction. The target Psi* and the baseline estimating function U(q,Psi) are defined from the opt-SNMM and ID assumptions independently of the projection, so the improvement is not self-definitional. The NDE assumption (eq. 7) is an explicit substantive input, and its observed-data consequence (eq. 8) is used to give every T_b mean zero; the paper proves the key characterization Lambda_perp_NDE = T_0+...+T_K as Theorem 1 in Appendix A.1 rather than importing it as an unverified postulate. Citations to Robins [25,27] supply the baseline opt-SNMM and the M1 nuisance tangent space; these are parameter-free published characterizations with stated assumptions that do not include the paper's efficiency result, so they do not constitute load-bearing self-citation. The one passage requiring explicit flagging is Remark 4: for continuous Y, the paper asserts that Pi[U|Omega] 'should converge' to Pi[U|Lambda_perp_NDE] as xi->infinity and that choosing xi slowly with n makes the estimators asymptotically equivalent. No proof or rate condition is given, so the near-optimality of the recommended finite-basis estimator is not established; however, this is an omitted convergence proof, not a circularity, because the target projection is defined independently of the approximating subspace. The simulation studies compare the proposed estimators with standard g-estimation and IPW benchmarks and do not redefine those benchmarks in terms of the proposed method. Overall, I find no step in which a claimed derivation reduces by construction to its own inputs.

Assumptions & free parameters 1 free parameters · 6 assumptions · 1 invented entities

The core input assumptions are the standard causal identification conditions and, crucially, the NDE assumption. The paper does not fit any physical constants; the only hand-chosen numeric quantity relevant to the near-optimal estimator is the basis dimension ξ. The latent variable R*_t is introduced for stronger NDE variants but is not needed for the main NDE(Yd) result.

free parameters (1)

basis dimension ξ of the subspace Ω = 6 in simulations; grows to infinity in theory
The near-optimal continuous-outcome estimator \tilde Ψ(q,b_sub) projects onto the finite-dimensional subspace Ω spanned by ξ basis functions. The dimension is chosen by hand and must grow with the sample size for asymptotic equivalence to the optimal projection, but this convergence is not proved.

assumptions (6)

domain assumption ID assumptions: consistency, positivity, and sequential exchangeability (Section 2, assumptions 1-3).
These standard causal assumptions identify counterfactual means under regimes and are required for the g-formula and for the NDE restriction (8).
domain assumption NDE(Yd) assumption, equation (7): testing history has no direct effect on Yd except through treatment history.
This is the defining assumption of the paper. It implies the observed-data restriction (8) that gives T_b mean zero.
domain assumption Exclusion of exceptional laws (Robins 2004, page 219).
Under exceptional laws the opt-SNMM g-estimator is not RAL, and the paper explicitly excludes them so that \tilde Ψ(q) is RAL; this limits the scope of the central claim.
domain assumption Correct specification of the opt-SNMM in equation (5).
The parameters Ψ* are defined through the assumed parametric model for the treatment-effect contrasts; if misspecified, the estimated 'optimal regime' is not the true optimum.
standard math Regularity conditions for RAL and cross-fitted DR-ML estimators (Appendix A.5).
Theorem 5 requires nuisance estimates to converge in L2 and the conditional bias terms to be op(n^{-1/2}); these are standard but unverified conditions in practice.
ad hoc to paper For the near-optimal continuous-outcome estimator, convergence of the basis projection as ξ and n grow (Remark 4).
The asymptotic equivalence of \tilde Ψ(q,b_sub) and \tilde Ψ(q,b_opt) is asserted without proof; this is the main unsupported assumption.

invented entities (1)

Latent underlying test result R*_t
purpose: Introduced in Section 7 to define stronger NDE assumptions (NDE(\bar R*_K, Y_d)) and to prove Lemma 1 that V_nde-ipw,g is a statistic.
R*_t is a counterfactual/latent variable revealed only when a test is performed; it is not measured for untested subjects and carries no falsifiable prediction outside the model. It is a definitional device, not an explanatory mechanism, so its lack of independent evidence is not a threat to the main efficiency claim.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Efficient estimation of optimal regimes under a no direct effect assumption." pith.science (2026). https://pith.science/paper/BRSQVAX3

@misc{pith2026190810448,
  author       = {Pith},
  title        = {Pith review of: Efficient estimation of optimal regimes under a no direct effect assumption},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BRSQVAX3}},
  note         = {Machine review of arXiv:1908.10448}
}

read the original abstract

We derive new estimators of an optimal joint testing and treatment regime under the no direct effect (NDE) assumption that a given laboratory, diagnostic, or screening test has no effect on a patient's clinical outcomes except through the effect of the test results on the choice of treatment. We model the optimal joint strategy using an optimal regime structural nested mean model (opt-SNMM). The proposed estimators are more efficient than previous estimators of the parameters of an opt-SNMM because they efficiently leverage the `no direct effect (NDE) of testing' assumption. Our methods will be of importance to decision scientists who either perform cost-benefit analyses or are tasked with the estimation of the `value of information' supplied by an expensive diagnostic test (such as an MRI to screen for lung cancer).

Figures

Figures reproduced from arXiv: 1908.10448 by the authors.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 33 canonical work pages

[1]

Bang, H. and J. M. Robins (2005). Doubly robust estimation in missing data and causal inference models. Biometrics 61 (4), 962–973

work page 2005
[2]

Bellman, R. (1952). On the theory of dynamic programming. Proceedings of the National Academy of Sciences of the United States of America 38 (8), 716

work page 1952
[3]

Caniglia, E. C., J. M. Robins, L. E. Cain, C. Sabin, R. Logan, S. Abgrall, M. J. Mugavero, S. Hern´ andez-D´ ıaz, L. Meyer, R. Seng, et al. (2019). Emulating a trial of joint dynamic strategies: An application to monitoring and treatment of HIV-positive individuals. Statistics in Medicine 38 (13), 2428–2446

work page 2019
[4]

Chetverikov, M

Chernozhukov, V., D. Chetverikov, M. Demirer, E. Duﬂo, C. Hansen, W. Newey, and J. Robins (2018). Double/debiased machine learning for treatment and structural parameters. The Econometrics Journal 21 (1), C1–C68

work page 2018
[5]

Routine versus clinically driven laboratory monitoring of hiv antiretroviral therapy in Africa (DART): a randomised non-inferiority trial

DART Trial Team (2010). Routine versus clinically driven laboratory monitoring of hiv antiretroviral therapy in Africa (DART): a randomised non-inferiority trial. The Lancet 375 (9709), 123–131

work page 2010
[6]

Ford, D., J. M. Robins, M. L. Petersen, D. M. Gibb, C. F. Gilks, P. Mugyenyi, H. Grosskurth, J. Hakim, E. Katabira, A. G. Babiker, et al. (2015). The impact of diﬀerent cd4 cell-count monitoring and switching strategies on mortality in hiv-infected african adults on antiretroviral therapy: an application of dynamic marginal structural models. American Jou...

work page 2015
[7]

Gould, J. P. (1974). Risk, stochastic preference, and the value of information. Journal of Economic Theory 8 (1), 64–84

work page 1974
[8]

statistical issues arising in the Women’s Health Initiative

Hern´ an, M. A., J. M. Robins, and L. A. Garc´ ıa Rodr´ ıguez (2005). Discussion on 48 “statistical issues arising in the Women’s Health Initiative”. Biometrics 61 (4), 922– 930

work page 2005

Show all 37 references

[9]

Hilton, R. W. (1981). The determinants of information value: Synthesizing some general results. Management Science 27 (1), 57–64

1981
[10]

Mao, and M

Kallus, N., X. Mao, and M. Uehara (2019). Localized debiased machine learning: Eﬃcient estimation of quantile treatment eﬀects, conditional value at risk, and beyond. arXiv preprint arXiv:1912.12945

2019 arXiv
[11]

Krahn, M. D., J. E. Mahoney, M. H. Eckman, J. Trachtenberg, S. G. Pauker, and A. S. Detsky (1994). Screening for prostate cancer: a decision analytic view. Jama 272 (10), 773–780

1994
[12]

Sofrygin, J

Kreif, N., O. Sofrygin, J. A. Schmittdiel, A. S. Adams, R. W. Grant, Z. Zhu, M. J. van der Laan, and R. Neugebauer (2020). Exploiting nonsystematic covariate moni- toring to broaden the scope of evidence about the causal eﬀects of adaptive treatment strategies. Biometrics

2020
[13]

Lara, A. M., J. Kigozi, J. Amurwon, L. Muchabaiwa, B. N. Wakaholi, R. E. M. Mota, A. S. Walker, R. Kasirye, F. Ssali, A. Reid, et al. (2012). Cost eﬀectiveness analysis of clinically driven versus routine laboratory monitoring of antiretroviral therapy in Uganda and Zimbabwe. ...

2012
[14]

LaValle, I. H. (1968a). On cash equivalents and information evaluation in decisions under uncertainty Part I: Basic theory. Journal of the American Statistical Associa- tion 63 (321), 252–276

1968
[15]

LaValle, I. H. (1968b). On cash equivalents and information evaluation in decisions under uncertainty Part II: Incremental information decisions. Journal of the American Statistical Association 63 (321), 277–284. 49

1968
[16]

Luedtke, A. R., O. Sofrygin, M. J. van der Laan, and M. Carone (2017). Se- quential double robustness in right-censored longitudinal models. arXiv preprint arXiv:1705.02459

2017 arXiv
[17]

Murphy, S. A. (2003). Optimal dynamic treatment regimes. Journal of the Royal Statistical Society: Series B (Statistical Methodology) 65 (2), 331–355

2003
[18]

Mushlin, A. I. and L. Fintor (1992). Is screening for breast cancer cost-eﬀective? Cancer 69 (S7), 1957–1962

1992
[19]

Neugebauer, R., J. A. Schmittdiel, A. S. Adams, R. W. Grant, and M. J. van der Laan (2017). Identiﬁcation of the joint eﬀect of a dynamic treatment intervention and a stochastic monitoring intervention under the no direct eﬀect assumption. Journal of Causal Inference 5 (1)

2017
[20]

Rotnitzky, and J

Orellana, L., A. Rotnitzky, and J. M. Robins (2010a). Dynamic regime marginal structural mean models for estimation of optimal dynamic treatment regimes, Part I: main content. The International Journal of Biostatistics 6 (2)

2010
[21]

Rotnitzky, and J

Orellana, L., A. Rotnitzky, and J. M. Robins (2010b). Dynamic regime marginal structural mean models for estimation of optimal dynamic treatment regimes, Part II: proofs of results. The International Journal of Biostatistics 6 (2)

2010
[22]

Robins, J. (1986). A new approach to causal inference in mortality studies with a sustained exposure period -— application to control of the healthy worker survivor eﬀect. Mathematical Modelling 7 (9-12), 1393–1512

1986
[23]

Orellana, and A

Robins, J., L. Orellana, and A. Rotnitzky (2008). Estimation and extrapolation of optimal treatment and testing strategies. Statistics in Medicine 27 (23), 4678–4721

2008
[24]

A new approach to causal inference in mortality 50 studies with a sustained exposure period – application to control of the healthy worker survivor eﬀect

Robins, J. M. (1987). Addendum to “A new approach to causal inference in mortality 50 studies with a sustained exposure period – application to control of the healthy worker survivor eﬀect”. Computers & Mathematics with Applications 14 (9-12), 923–945

1987
[25]

Robins, J. M. (1999). Testing and estimation of direct eﬀects by reparameterizing directed acyclic graphs with structural nested models. In Computation, Causation, and Discovery (C. Glymour and G. Cooper, eds.) , pp. 349–405. AAAI Press, Menlo Park, CA

1999
[26]

Robins, J. M. (2000). Marginal structural models versus structural nested models as tools for causal inference. In Statistical models in epidemiology, the environment, and clinical trials, pp. 95–133. Springer

2000
[27]

Robins, J. M. (2004). Optimal structural nested models for optimal sequential de- cisions. In Proceedings of the Second Seattle Symposium in Biostatistics , pp. 189–326. Springer

2004
[28]

Robins, J. M. and A. Rotnitzky (1995). Semiparametric eﬃciency in multivariate regression models with missing data. Journal of the American Statistical Associa- tion 90 (429), 122–129

1995
[29]

Rotnitzky, A., J. M. Robins, and L. Babino (2017). On the multiply robust estimation of the mean of the g-functional. arXiv preprint arXiv:1705.08582

2017 arXiv
[30]

Rotnitzky, and J

Smucler, E., A. Rotnitzky, and J. M. Robins (2019). A unifying approach for doubly- robust 𝓁1 regularized estimation of causal contrasts. arXiv preprint arXiv:1904.03737

2019 arXiv
[31]

van der Laan, M. J. and M. L. Petersen (2007). Causal eﬀect models for realistic individualized treatment and intention to treat rules. The International Journal of Biostatistics 3 (1)

2007
[32]

van der Vaart, A. and J. Wellner (1996). Weak Convergence and Empirical Processes: with Applications to Statistics . Springer Science & Business Media. 51

1996
[33]

van der Vaart, A. W. (1998). Asymptotic statistics, Volume 3. Cambridge University Press

1998
[34]

Vansteelandt, S. and M. Joﬀe (2014). Structural nested models and G-estimation: The partially realized promise. Statistical Science 29 (4), 707–731

2014
[35]

Report of the commission on macroeconomics and health

World Health Organization (2001). Report of the commission on macroeconomics and health. 52 A Appendix The online supplementary materials include technical proofs and more details on the simulation studies omitted in the main text. A.1 Proof of Theorem 1 In this section, we pr...

2001
[36]

Notice thatyk+1,η† k+1 ( ¯Hk+1) coincides with E [ η† k+1( ¯Hk+1,Sk+1) p(Sk+1| ¯Hk+1) ⏐⏐⏐⏐ ¯Hk+1 ]

Given arbitrary functions η† k( ¯Hk,Sk), for k = 0,...,K , deﬁne yk+1,η† k+1 ( ¯Hk+1) ≡ ∑ sk+1 η† k+1( ¯Hk+1,sk+1). Notice thatyk+1,η† k+1 ( ¯Hk+1) coincides with E [ η† k+1( ¯Hk+1,Sk+1) p(Sk+1| ¯Hk+1) ⏐⏐⏐⏐ ¯Hk+1 ] . Given a ﬁxed timet and a functionbt( ¯Ht, ¯St,Y d) deﬁneηK( ...
[37]

always treat

So, we can write, ~b1,0(R∗ 1, ¯H1,C 2,R∗ 2, (S1,S 2),Y d) =~b1,0( ¯H1,C 2, (S1,S 2),Y d). Next, evaluating at A0 = 1,A 1 = 0 we conclude that − ~b1,1(R∗ 1, ¯H1,C 2,R∗ 2, (S1,S 2),Y d) p(S2| ¯H2) 1 Π0 − ~b1,0( ¯H1,C 2, (S1,S 2),Y d) p(S2| ¯H2) ( 1 Π0 − 1 ) cannot depend on R∗ 2...

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

Bang, H. and J. M. Robins (2005). Doubly robust estimation in missing data and causal inference models. Biometrics 61 (4), 962–973

work page 2005

[2] [2]

Bellman, R. (1952). On the theory of dynamic programming. Proceedings of the National Academy of Sciences of the United States of America 38 (8), 716

work page 1952

[3] [3]

Caniglia, E. C., J. M. Robins, L. E. Cain, C. Sabin, R. Logan, S. Abgrall, M. J. Mugavero, S. Hern´ andez-D´ ıaz, L. Meyer, R. Seng, et al. (2019). Emulating a trial of joint dynamic strategies: An application to monitoring and treatment of HIV-positive individuals. Statistics in Medicine 38 (13), 2428–2446

work page 2019

[4] [4]

Chetverikov, M

Chernozhukov, V., D. Chetverikov, M. Demirer, E. Duﬂo, C. Hansen, W. Newey, and J. Robins (2018). Double/debiased machine learning for treatment and structural parameters. The Econometrics Journal 21 (1), C1–C68

work page 2018

[5] [5]

Routine versus clinically driven laboratory monitoring of hiv antiretroviral therapy in Africa (DART): a randomised non-inferiority trial

DART Trial Team (2010). Routine versus clinically driven laboratory monitoring of hiv antiretroviral therapy in Africa (DART): a randomised non-inferiority trial. The Lancet 375 (9709), 123–131

work page 2010

[6] [6]

Ford, D., J. M. Robins, M. L. Petersen, D. M. Gibb, C. F. Gilks, P. Mugyenyi, H. Grosskurth, J. Hakim, E. Katabira, A. G. Babiker, et al. (2015). The impact of diﬀerent cd4 cell-count monitoring and switching strategies on mortality in hiv-infected african adults on antiretroviral therapy: an application of dynamic marginal structural models. American Jou...

work page 2015

[7] [7]

Gould, J. P. (1974). Risk, stochastic preference, and the value of information. Journal of Economic Theory 8 (1), 64–84

work page 1974

[8] [8]

statistical issues arising in the Women’s Health Initiative

Hern´ an, M. A., J. M. Robins, and L. A. Garc´ ıa Rodr´ ıguez (2005). Discussion on 48 “statistical issues arising in the Women’s Health Initiative”. Biometrics 61 (4), 922– 930

work page 2005