A semiparametric method for the analysis of outcomes during a gap in HIV care under incomplete outcome ascertainment
-
Giorgos Bakoyannis
,
Lameck Diero
Abstract
Objectives
Estimation of the cascade of HIV care is essential for evaluating care and treatment programs, informing policy makers and assessing targets such as 90-90-90. A challenge to estimating the cascade based on electronic health record concerns patients “churning” in and out of care. Correctly estimating this dynamic phenomenon in resource-limited settings, such as those found in sub-Saharan Africa, is challenging because of the significant death under-reporting. An approach to partially recover information on the unobserved deaths is a double-sampling design, where a small subset of individuals with a missed clinic visit is intensively outreached in the community to actively ascertain their vital status. This approach has been adopted in several programs within the East Africa regional IeDEA consortium, the context of our motivating study. The objective of this paper is to propose a semiparametric method for the analysis of competing risks data with incomplete outcome ascertainment.
Methods
Based on data from double-sampling designs, we propose a semiparametric inverse probability weighted estimator of key outcomes during a gap in care, which are crucial pieces of the care cascade puzzle.
Results
Simulation studies suggest that the proposed estimators provide valid estimates in settings with incomplete outcome ascertainment under a set of realistic assumptions. These studies also illustrate that a naïve complete-case analysis can provide seriously biased estimates. The methodology is applied to electronic health record data from the East Africa IeDEA Consortium to estimate death and return to care during a gap in care.
Conclusions
The proposed methodology provides a robust approach for valid inferences about return to care and death during a gap in care, in settings with death under-reporting. Ultimately, the resulting estimates will have significant consequences on program construction, resource allocation, policy and decision making at the highest levels.
Funding source: NIAID 10.13039/100000060
Award Identifier / Grant number: U01AI069911 and R21AI145662
Funding source: PEPFAR 10.13039/100009054
Award Identifier / Grant number: AID-623-A-12-0001
Acknowledgments
The authors thank the two anonymous referees for their insightful comments that led to a significant improvement of this manuscript. Research reported in this publication was supported by the National Institute Of Allergy And Infectious Diseases (NIAID), Eunice Kennedy Shriver National Institute Of Child Health & Human Development (NICHD), National Institute On Drug Abuse (NIDA), National Cancer Institute (NCI), and the National Institute of Mental Health (NIMH), in accordance with the regulatory requirements of the National Institutes of Health under Award Numbers U01AI069911 and R21AI145662. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This research has also been supported by the President’s Emergency Plan for AIDS Relief (PEPFAR) through USAID under the terms of Cooperative Agreement No. AID-623-A-12-0001 it is made possible through joint support of the United States Agency for International Development (USAID). The contents of this journal article are the sole responsibility of AMPATH and do not necessarily reflect the views of USAID or the United States Government.
Research funding: NIAID Award Numbers U01AI069911 and R21AI145662. PEPFAR Cooperative Agreement No. AID-623-A-12-0001.
Author contributions: All authors have accepted responsibility for the entire content of this manuscript and approved its submission.
Competing interests: Authors state no conflict of interest.
Informed consent: Informed consent was obtained from all individuals included in this study.
Ethical approval: The local Institutional Review Board deemed the study exempt from review.
Appendix: Analysis of the hazards of a first gap in care and death
In this Appendix, we provide the analysis of the hazards of death while in care and of a gap in care after ART initiation (i.e. the remaining hazards in the multi-state churn model depicted in Figure 2). Here, we focus on the first occuring event (death or gap in care) after ART initiation and, thus, the analysis can be based on methods for competing risks data (Bakoyannis and Touloumi 2012; Putter, Fiocco, and Geskus 2007). To account for the missing event types (i.e. death or gap in care) due to death under-reporting among the non-outreached lost patients, we use appropriate pseudolikelihood methods (Bakoyannis, Zhang, and Yiannoutsos 2019, 2020). In this analysis we include 38,490 patients who initiated ART in one of the clinics in the AMPATH program. These patients are a superset of the 18,892 patients who were identified as lost to clinic and analyzed in the main text of this manuscript. Characteristics of the 38,490 patients are shown in Table A1.
Descriptive characteristics of the study sample for the analysis of the first gap in care and death prior to the first gap.
| Passively ascertained outcome | p-Value | |||
|---|---|---|---|---|
| In care | Death | LTCa | ||
| (n=17,619) | (n=1,979) | (n=18,892) | ||
| n(%) | n(%) | n(%) | ||
| Outreach | ||||
| Not attempted | 0 (−) | 0 (−) | 14,774 (78.2) | – |
| Not found | 0 (−) | 0 (−) | 1,580 (8.4) | |
| Found | 0 (−) | 0 (−) | 2,538 (13.4) | |
| True outcomeb | ||||
| Death | 0 (−) | 0 (−) | 491 (19.3) | – |
| Gap in care | 0 (−) | 0 (−) | 2,047 (80.7) | |
| Gender | ||||
| Female & non-pregnantc | 9,412 (58.9) | 726 (43.1) | 8,058 (51.8) | <0.001 |
| Female & pregnantc | 1,076 (6.7) | 32 (1.9) | 1,190 (7.6) | |
| Male | 5,488 (34.4) | 926 (55.0) | 6,320 (40.6) | |
| HIV status disclosed | ||||
| No | 6,269 (35.6) | 670 (33.9) | 6,972 (36.9) | 0.003 |
| Yes | 11,350 (64.4) | 1,309 (66.1) | 11,920 (63.1) | |
| Travel time to clinic | ||||
| <30′ | 4,570 (25.9) | 480 (24.3) | 4,752 (25.2) | <0.001 |
| 30–59′ | 6,153 (34.9) | 679 (34.3) | 5,936 (31.4) | |
| 1–2 h | 4,346 (24.7) | 482 (24.4) | 4,659 (24.7) | |
| 2 + h | 2,550 (14.5) | 338 (17.1) | 3,545 (18.8) | |
| Level of care | ||||
| Primary | 5,777 (32.8) | 649 (32.8) | 5,814 (30.8) | <0.001 |
| Secondary | 9,561 (54.3) | 1,176 (59.4) | 9,967 (52.8) | |
| Tertiary | 2,281 (12.9) | 154 (7.8) | 3,111 (16.5) | |
| Median (IQR) | Median (IQR) | Median (IQR) | p-Value | |
|---|---|---|---|---|
| Aged, years | 37.9 (32.0, 45.4) | 37.8 (31.7, 45.2) | 36.0 (30.3, 43.1) | <0.001 |
| CD4d, cells/μL | 186 (113, 263) | 106 (52, 179) | 155 (83, 234) | <0.001 |
| Outreach worker ratioe ( | 5.0 (3.6, 5.9) | 5.0 (4.0, 5.9) | 5.0 (4.0, 5.9) | <0.001 |
aLost to clinic. bAscertained through outreach. cAt or prior to ART initiation. dAt ART initiation. e# of outreach workers to total daily # of adult patients.
Of the 38,490 patients in our sample, 18,892 (49.1%) patients were identified as lost to clinic, 1,979 (5.1%) were reported as deceased without a prior gap in care, while the remaining 17,619 (45.8%) patients were alive and without a gap in care at the date of data request. In total, 2,538 (13.4%) lost patients were successfully traced by AMPATH outreach workers (Table A1). Of them, 491 (19.3%) were found to have died within two months from the next scheduled visit and this indicates a substantial death under-reporting issue. The potential predictors of interest included patient gender, pregnancy status at last clinic visit, age and CD4 count at ART initiation, HIV status disclosure, travel time to clinic, and the level of care of the clinic attended by each patient. To make the key MAR assumption more plausible, we also considered the ratio of the number of outreach workers to the average daily number of adult patients in the clinic as an auxiliary variable that could plausibly be related to the probability that a patient lost to program would be outreached (Table A1). The pseudolikelihood methods we use here require the specification of a (parametric) logistic model for the probability of an unreported death among the lost patients. For flexibility, we use cubic B-splines with three internal knots for the continuous covariates in this model (regression splines). Note that here, unlike the SIPW approach, the number of knots does not depend on the sample size n and thus the model involves only a finite-dimensional parameter (i.e. it is a parametric model). The overall estimated cumulative incidences of a first gap in care and death prior to the first gap in care are, based on the nonparametric maximum pseudolikelihood estimator by Bakoyannis et al. (2019), are given in Figure A1.
Cumulative incidence of death while in care and gap in care after ART initiation.
In Figure 7, it appears that a large proportion of patients who initiate ART have a subsequent gap in care. The estimated cumulative incidence of a gap in care at 1, 2, and 5 years since ART initiation is 0.187, 0.314, and 0.505, respectively. The corresponding figures for the cumulative incidence of death while in care are 0.108, 0.131, and 0.170. Effect estimates for factors potentially associated with the hazards of death while in care and gap in care are provided in Tables A2, A3 respectively.
Factors associated with death while in care after ART initiation.
| CSHRa | 95% CI | p-Value | |
|---|---|---|---|
| Gender | |||
| Female & non-pregnant | 1.000 | – | – |
| Female & pregnant | 0.529 | (0.341, 0.820) | 0.004 |
| Male | 1.306 | (1.164, 1.465) | <0.001 |
| Ageb, per 10 years | 1.110 | (1.035, 1.192) | 0.004 |
| CD4b, per 100 cell/μL | 0.663 | (0.608, 0.723) | <0.001 |
| HIV status disclosed | 1.072 | (0.914, 1.257) | 0.395 |
| Travel time to clinic >30′ | 1.081 | (0.945, 1.235) | 0.256 |
| Level of care | |||
| Secondary/Tertiary | 1.000 | – | – |
| Primary | 0.804 | (0.652, 0.992) | 0.042 |
aCause-specific hazard ratio bAt ART initiation
Factors associated with a first gap in care after ART initiation.
| CSHRa | 95 | p-Value | |
|---|---|---|---|
| Gender | |||
| Female & non-pregnant | 1.000 | – | – |
| Female & pregnant | 1.169 | (1.072, 1.274) | <0.001 |
| Male | 1.108 | (1.042, 1.179) | 0.001 |
| Ageb, per 10 years | 0.769 | (0.742, 0.797) | <0.001 |
| CD4b, per 100 cell/μL | 0.981 | (0.960, 1.002) | 0.070 |
| HIV status disclosed | 0.927 | (0.869, 0.990) | 0.023 |
| Travel time to clinic >30′ | 1.038 | (0.987, 1.092) | 0.148 |
| Level of care | |||
| Secondary/Tertiary | 1.000 | – | – |
| Primary | 1.067 | (0.979, 1.163) | 0.142 |
aCause-specific hazard ratio bAt ART initiation
Factors associated with a decreased hazard of death while in care, include pregnancy status (pregnant women have generally less advanced disease), female gender, younger age, higher CD4 cell count and being treated at a primary clinic (Table A2). Factors associated with a higher rate of a gap in care after ART initiation includes pregnancy, male gender, younger age, and non-disclosure of the HIV status (Table 7).
References
Agresti, A. 2002. Categorical Data Analysis. New Jersey: John Wiley & Sons.10.1002/0471249688Search in Google Scholar
An, M., C. Frangakis, B. Musick, and C. Yiannoutsos. 2009. “The Need for Double-Sampling Designs in Survival Studies: an Application to Monitor Pepfar.” Biometrics 65: 301–6, https://doi.org/10.1111/j.1541-0420.2008.01043.x.Search in Google Scholar PubMed PubMed Central
Andersen, P. K., and N. Keiding. 2002. “Multi-state Models for Event Historyanalysis.” Statistical Methods in Medical Research 11: 91–115, https://doi.org/10.1191/0962280202sm276ra.Search in Google Scholar PubMed
Bakoyannis, G. 2020 In press. “Nonparametric Analysis of Nonhomogeneous Multistate Processes with Clustered Observations.” Biometrics 1–14, https://doi.org/10.1111/biom.13327.https://doi.org/10.1111/biom.13327Search in Google Scholar PubMed PubMed Central
Bakoyannis, G., and G. Touloumi. 2012. “Practical Methods for Competing Risks Data: a Review.” Statistical Methods in Medical Research 21: 257–72, https://doi.org/10.1177/0962280210394479.Search in Google Scholar PubMed
Bakoyannis, G., and C. T. Yiannoutsos. 2015. “Impact of and Correction for Outcome Misclassification in Cumulative Incidence Estimation.” PloS One 10: e0137454, https://doi.org/10.1371/journal.pone.0137454.Search in Google Scholar PubMed PubMed Central
Bakoyannis, G., Y. Zhang, and C. T. Yiannoutsos. 2019. “Nonparametric Inference for Markov Processes with Missing Absorbing State.” Statistica Sinica 29: 2083–104, https://doi.org/10.5705/ss.202017.0175.Search in Google Scholar PubMed PubMed Central
Bakoyannis, G., Y. Zhang, and C. T. Yiannoutsos. 2020. “Semiparametric Regression and Risk Prediction with Competing Risks Data under Missing Cause of Failure.” Lifetime Data Analysis 26 (4): 659–684, https://doi.org/10.1007/s10985-020-09494-1.Search in Google Scholar PubMed PubMed Central
Bärnighausen, T., J. Bor, S. Wandira-Kazibwe, and D. Canning. 2011. “Correcting HIV Prevalence Estimates for Survey Nonparticipation Using Heckman-type Selection Models.” Epidemiology 22 (1): 27–35, https://doi.org/10.1097/ede.0b013e3181ffa201.Search in Google Scholar
Brinkhof, M., B. Spycher, and C. Yiannoutsos. 2010. “Adjusting Mortality for Loss to Follow-Up: Analysis of Five ART Programmes in Sub-saharan Africa.” PloS One 5: e14149, https://doi.org/10.1371/journal.pone.0014149.Search in Google Scholar PubMed PubMed Central
Cheng, G., and J. Z. Huang. 2010. “Bootstrap Consistency for General Semiparametric M-Estimation.” Annals of Statistics 38: 2884–915, https://doi.org/10.1214/10-aos809.Search in Google Scholar
Field, C. A., and A. H. Welsh. 2007. “Bootstrapping Clustered Data.” Journal of the Royal Statistical Society: Series B (Statistical Methodology) 69: 369–90, https://doi.org/10.1111/j.1467-9868.2007.00593.x.Search in Google Scholar
Gardner, E. M., M. P. McLees, and J. F. Steiner. 2011. “The Spectrum of Engagement in Hiv Care and its Relevance to Test-And-Treat Strategies for Prevention of HIV Infection.” Clinical Infectious Diseases 52: 793–800, https://doi.org/10.1093/cid/ciq243.Search in Google Scholar
Geng, E., T. Odeny, and R. Lyamuya. 2015. “Estimation of Mortality Among HIV-Infected People on Antiretroviral Treatment in East Africa: a Sampling-Based Approach in an Observational, Multi-Site, Cohort Study.” Lancet HIV 2: e107–116, https://doi.org/10.1016/s2352-3018(15)00002-8.Search in Google Scholar
Geng, E., T. Odeny, and R. Lyamuya. 2016. “Retention in Care and Patient-Reported Reasons for Undocumented Transfer or Stopping Care Among HIV-Infected Patients on Antiretroviral Therapy in Eastern Africa: Application of a Sampling-Based Approach.” Clinical Infectious Diseases 62: 935–44, https://doi.org/10.1093/cid/civ1004.Search in Google Scholar
Gentleman, R. C., J. F. Lawless, J. C. Lindsey, and P. Yan. 1994. “Multi-state Markov Models for Analysing Incomplete Disease History Data with Illustrations for Hiv Disease.” Statistics in Medicine 13: 805–21, https://doi.org/10.1002/sim.4780130803.Search in Google Scholar
Gill, M., and H. Krentz. 2009. “Unappreciated Epidemiology: the Churn Effect in a Regional Hiv Care Programme.” International Journal of STD and AIDS 20: 540–4, https://doi.org/10.1258/ijsa.2008.008422.Search in Google Scholar
Graham, S. M., J. Raboud, R. S. McClelland, W. Jaoko, J. Ndinya-Achola, K. Mandaliya, J. Overbaugh, and A. M. Bayoumi. 2013. “Loss to Follow-Up as a Competing Risk in an Observational Study of Hiv-1 Incidence.” PloS One 8: e59480, https://doi.org/10.1371/journal.pone.0059480.Search in Google Scholar
Granich, R., C. Gilks, and C. Dye. 2009. “Universal Voluntary HIV Testing with Immediate Antiretroviral Therapy as a Strategy for Elimination of Hiv Transmission: a Mathematical Model.” Lancet 373: 48–57, https://doi.org/10.1016/s0140-6736(08)61697-9.Search in Google Scholar
Hastie, T., and R. Tibshirani. 1986. “Generalized Additive Models.” Statistical Science 1: 297–318, https://doi.org/10.1214/ss/1177013604.Search in Google Scholar
Lee, H., J. W. Hogan, B. L. Genberg, X. K. Wu, B. S. Musick, A. Mwangi, and P. Braitstein. 2018. “A State Transition Framework for Patient-Level Modeling of Engagement and Retention in Hiv Care Using Longitudinal Cohort Data.” Statistics in Medicine 37: 302–19, https://doi.org/10.1002/sim.7502.Search in Google Scholar
Li, L., C. Shen, X. Li, and J. M. Robins. 2013. “On Weighting Approaches for Missing Data.” Statistical Methods in Medical Research 22: 14–30, https://doi.org/10.1177/0962280211403597.Search in Google Scholar
Lu, K., and A. A. Tsiatis. 2001. “Multiple Imputation Methods for Estimating Regression Coefficients in the Competing Risks Model with Missing Cause of Failure.” Biometrics 57: 1191–7, https://doi.org/10.1111/j.0006-341x.2001.01191.x.Search in Google Scholar
Meira-Machado, L., J. de Uña Álvarez, and C. Cadarso-Suárez. 2009. “Multi-state Models for the Analysis of Time-To-Event Data.” Statistics in Medicine 18: 195–222, https://doi.org/10.1177/0962280208092301.Search in Google Scholar
Meng, X.-L. 1994. “Multiple-imputation Inferences with Uncongenial Sources of Input.” Statistical Science 9 (4): 538–58, https://doi.org/10.1214/ss/1177010269.Search in Google Scholar
Mussini, C., G. Touloumi, G. Bakoyannis, C. Sabin, A. Castagna, L. Sighinolfi, L. E. Erikson, G. Bratt, V. Borghi, and A. Lazzarin. 2009. “Magnitude and Determinants of Cd4 Recovery after Haart Resumption after 1 Cycle of Treatment Interruption.” JAIDS Journal of Acquired Immune Deficiency Syndromes 52: 588–94, https://doi.org/10.1097/qai.0b013e3181b9e94d.Search in Google Scholar
Nsanzimana, S., A. Binagwaho, S. Kanters, and E. Mills. 2014. “Churning in and Out of HIV Care.” Lancet HIV 2: e58–9, https://doi.org/10.1016/s2352-3018(14)70028-1.Search in Google Scholar
Putter, H., M. Fiocco, and R. B. Geskus. 2007. “Tutorial in Biostatistics: Competing Risks and Multi-State Models.” Statistics in Medicine 26: 2389–430, https://doi.org/10.1002/sim.2712.Search in Google Scholar PubMed
Rachlis, B., G. Bakoyannis, P. Easterbrook, B. Genberg, R. S. Braithwaite, C. R. Cohen, E. A. Bukusi, A. Kambugu, M. B. Bwana, and G. R. Somi. 2016. “Facility-level Factors Influencing Retention of Patients in Hiv Care in East Africa.” PloS One 11: e0159994, https://doi.org/10.1371/journal.pone.0159994.Search in Google Scholar PubMed PubMed Central
Rebeiro, P., G. Bakoyannis, and B. Musick. 2017. “Observational Study of the Effect of Patient Outreach on Return to Care: The Earlier the Better.” Journal of Acquired Immune Deficiency Syndromes 76: 141–8, https://doi.org/10.1097/qai.0000000000001474.Search in Google Scholar
Robins, J. M., and N. Wang. 2000. “Inference for Imputation Estimators.” Biometrika 87: 113–24, https://doi.org/10.1093/biomet/87.1.113.Search in Google Scholar
Rubin, D. B. 1996. “Multiple Imputation after 18+ Years.” Journal of the American Statistical Association 91: 473–89, https://doi.org/10.1080/01621459.1996.10476908.Search in Google Scholar
Schafer, J. L. 1999. “Multiple Imputation: a Primer.” Statistical Methods in Medical Research 8: 3–15, https://doi.org/10.1177/096228029900800102.Search in Google Scholar PubMed
Scharfstein, D. O., and R. A. Irizarry. 2003. “Generalized Additive Selection Models for the Analysis of Studies with Potentially Nonignorable Missing Outcome Data.” Biometrics 59: 601–13, https://doi.org/10.1111/1541-0420.00070.Search in Google Scholar PubMed
Schöni-Affolter, F., O. Keiser, A. Mwango, J. Stringer, B. Ledergerber, L. Mulenga, H. C. Bucher, A. O. Westfall, A. Calmy, and A. Boulle. 2011. “Estimating Loss to Follow-Up in Hiv-Infected Patients on Antiretroviral Therapy: The Effect of the Competing Risk of Death in zambia and switzerland.” PloS One 6: e27919, https://doi.org/10.1371/journal.pone.0027919.Search in Google Scholar PubMed PubMed Central
Shen, X., and W. Wong. 1994. “Convergence Rate of Sieve Estimates.” Annals of Statistics 22: 580–615.10.1214/aos/1176325486Search in Google Scholar
Spiekerman, C. F., and D. Lin. 1998. “Marginal Regression Models for Multivariate Failure Time Data.” Journal of the American Statistical Association 93: 1164–75, https://doi.org/10.1080/01621459.1998.10473777.Search in Google Scholar
Tchetgen Tchetgen, E. J., and K. E. Wirth. 2017. “A General Instrumental Variable Framework for Regression Analysis with Outcome Missing Not at Random.” Biometrics 73: 1123–31, https://doi.org/10.1111/biom.12670.Search in Google Scholar PubMed PubMed Central
Touloumi, G., N. Pantazis, A. Antoniou, H. A. Stirnadel, S. A. Walker, K. Porter, and C. Collaboration. 2006. “Highly Active Antiretroviral Therapy Interruption: Predictors and Virological and Immunologic Consequences.” JAIDS Journal of Acquired Immune Deficiency Syndromes 42: 554–61, https://doi.org/10.1097/01.qai.0000230321.85911.db.Search in Google Scholar PubMed
UNAIDS. 2014. 90-90-90. An Ambitious Treatment Target to Help End the AIDS Epidemic. Technical Report: Joint United Nations Programme on HIV/AIDS (UNAIDS). URL https://www.unaids.org/sites/default/files/media_asset/90-90-90_en.pdf.Search in Google Scholar
White, I. R., P. Royston, and A. M. Wood. 2011. “Multiple Imputation Using Chained Equations: Issues and Guidance for Practice.” Statistics in Medicine 30: 377–99, https://doi.org/10.1002/sim.4067.Search in Google Scholar PubMed
World Health Organization. 2015. Guideline on when to Start Antiretroviral Therapy and on Pre-exposure Prophylaxis for HIV: World Health Organization.Search in Google Scholar
Zhang, Y., L. Hua, and J. Huang. 2010. “A Spline-Based Semiparametric Maximum Likelihood Estimation Method for the Cox Model with Interval-Censored Data.” Scandinavian Journal of Statistics 37: 338–54, https://doi.org/10.1111/j.1467-9469.2009.00680.x.Search in Google Scholar
Supplementary Material
The online version of this article offers supplementary material (https://doi.org/10.1515/scid-2019-0013).
© 2020 Walter de Gruyter GmbH, Berlin/Boston
Articles in the same Issue
- Editorial
- Forward
- Research Articles
- Pre-selected class-level testing of longitudinal biomarkers reduces required multiple testing corrections to yield novel insights in longitudinal small sample human studies
- Joint modeling of time-varying HIV exposure and infection for estimation of per-act efficacy in HIV prevention trials
- Comparison of empirical and dynamic models for HIV viral load rebound after treatment interruption
- Comparison of machine learning methods for predicting viral failure: a case study using electronic health record data
- A semiparametric method for the analysis of outcomes during a gap in HIV care under incomplete outcome ascertainment
- Errors in multiple variables in human immunodeficiency virus (HIV) cohort and electronic health record data: statistical challenges and opportunities
- Challenges in evaluating the use of viral sequence data to identify HIV transmission networks for public health
- Evaluating the relative contribution of data sources in a Bayesian analysis with the application of estimating the size of hard to reach populations
Articles in the same Issue
- Editorial
- Forward
- Research Articles
- Pre-selected class-level testing of longitudinal biomarkers reduces required multiple testing corrections to yield novel insights in longitudinal small sample human studies
- Joint modeling of time-varying HIV exposure and infection for estimation of per-act efficacy in HIV prevention trials
- Comparison of empirical and dynamic models for HIV viral load rebound after treatment interruption
- Comparison of machine learning methods for predicting viral failure: a case study using electronic health record data
- A semiparametric method for the analysis of outcomes during a gap in HIV care under incomplete outcome ascertainment
- Errors in multiple variables in human immunodeficiency virus (HIV) cohort and electronic health record data: statistical challenges and opportunities
- Challenges in evaluating the use of viral sequence data to identify HIV transmission networks for public health
- Evaluating the relative contribution of data sources in a Bayesian analysis with the application of estimating the size of hard to reach populations
