Modeling the HIV Cascade of Care Using Routinely Collected Clinical Data to Guide Programmatic Interventions and Policy Decisions.

BACKGROUND: The HIV care cascade is a framework to examine effectiveness of HIV programs and progress toward global targets to end the epidemic but has been conceptualized as a unidirectional process that ignores cyclical care patterns. We present a dynamic cascade that accounts for patient "churn" and apply novel analytic techniques to readily available clinical data to robustly estimate program outcomes and efficiently assess progress toward global targets. METHODS: Data were assessed for 35,649 people living with HIV and receiving care at 78 clinics in East Africa between 2014 and 2020. Patients were aged ≥15 years and had ≥1 viral load measurements. We used multi-state models to estimate the probability of being in 1 of 5 states of a dynamic HIV cascade: (1) in HIV care but not on antiretroviral therapy (ART), (2) on ART, (3) virally suppressed, (4) in a gap-in-care, and (5) deceased and compared these among subgroups. To assess progress toward global targets, we summed those probabilities across patients and generated population-level proportions of patients on ART and virally suppressed in mid-2020. RESULTS: One year after enrollment, 2.8% of patients had not initiated ART, 86.7% were receiving ART, 57.4% were virally suppressed, 10.2% were disengaged from care, and 0.3% had died. At 5 years, the proportion on ART remained steady but viral suppression increased to 77.2%. Of those aged 15-25, >20% had disengaged from care and <60% were virally suppressed. In mid-2020, 90.1% of the cohort was on ART, 90.7% of whom had suppressed virus. CONCLUSIONS: Novel analytic approaches can characterize patient movement through a dynamic HIV cascade and, importantly, by capitalizing on readily available data from clinical cohorts, offer an efficient approach to estimate population-level proportions of patients on ART and virally suppressed. Significant progress toward global targets was observed in our cohort but challenges remain among younger patients.

Sieve estimation of a class of partially linear transformation models with interval-censored competing risks data.

In this paper, we consider a class of partially linear transformation models with interval-censored competing risks data. Under a semiparametric generalized odds rate specification for the cause-specific cumulative incidence function, we obtain optimal estimators of the large number of parametric and nonparametric model components via maximizing the likelihood function over a joint B-spline and Bernstein polynomial spanned sieve space. Our specification considers a relatively simpler finite-dimensional parameter space, approximating the infinite-dimensional parameter space as n → ∞, thereby allowing us to study the almost sure consistency, and rate of convergence for all parameters, and the asymptotic distributions and efficiency of the finite-dimensional components. We study the finite sample performance of our method through simulation studies under a variety of scenarios. Furthermore, we illustrate our methodology via application to a dataset on HIV-infected individuals from sub-Saharan Africa.

Estimating optimal individualized treatment rules with multistate processes.

Multistate process data are common in studies of chronic diseases such as cancer. These data are ideal for precision medicine purposes as they can be leveraged to improve more refined health outcomes, compared to standard survival outcomes, as well as incorporate patient preferences regarding quantity versus quality of life. However, there are currently no methods for the estimation of optimal individualized treatment rules with such data. In this paper, we propose a nonparametric outcome weighted learning approach for this problem in randomized clinical trial settings. The theoretical properties of the proposed methods, including Fisher consistency and asymptotic normality of the estimated expected outcome under the estimated optimal individualized treatment rule, are rigorously established. A consistent closed-form variance estimator is provided and methodology for the calculation of simultaneous confidence intervals is proposed. Simulation studies show that the proposed methodology and inference procedures work well even with small-sample sizes and high rates of right censoring. The methodology is illustrated using data from a randomized clinical trial on the treatment of metastatic squamous-cell carcinoma of the head and neck.

POLST recall, concordance, and decision quality outcomes among nursing home residents and surrogate decision-makers.

BACKGROUND: POLST orders are actionable in an emergency, so it is important that the decisions be of high quality and concordant with current preferences. The goal of this study is to determine the relationship between concordance and decision quality outcomes, including decision satisfaction and decisional conflict, among nursing facility residents and surrogates who recall POLST. METHODS: We completed structured interviews in 29 nursing facilities with 275 participants who had previously signed a POLST form. This included residents who were still making their own medical decisions (n = 123) and surrogate decision-makers for residents without decisional capacity (n = 152). POLST recall was defined as remembering talking about and/or completing the POLST form previously signed by the participant. Concordance was determined by comparing preferences elicited during a standardized interview with the POLST form on file. Decisional conflict, decision satisfaction, and conversation quality were assessed with standardized tools. RESULTS: Half of participants (50%) remembered talking about or completing the POLST form, but recall was not associated with the length of time since POLST completion or concordance with existing preferences. In multivariable analyses, there was no association between POLST recall, concordance, and decision quality outcomes, though satisfaction was associated with conversation quality. CONCLUSIONS: Half of the residents and surrogates in this study recalled the POLST they previously signed. Neither the age of the form nor the ability to recall the POLST conversation should be considered indicators of whether existing POLST orders match current preferences. Findings confirm a relationship between POLST conversation quality and satisfaction, underscoring the importance of POLST completion as a communication process.

Parental Engagement in Consent Processes for Enrollment in Biomedical HIV Prevention Trials: Implications for Minor Adolescents' Willingness to Participate.

PURPOSE: Minor adolescents are often excluded from HIV prevention clinical trials due to unresolved ethical issues. Their under-representation in research leads to delayed access to new HIV prevention approaches. We examine the relationship between consent procedures, trial features, demographic and social characteristics, and minor adolescents' willingness to participate (WTP) in biomedical HIV prevention research. METHODS: We recruited 14-17-year-olds at risk of HIV for this quasi-experimental study. Adolescents were randomly assigned to (1) self-consent, (2) adult permission required, or (3) parental permission required and underwent simulated consent procedures for two types of HIV prevention trials. They rated likelihood of participating in each study if offered the opportunity and completed a survey with demographic, social, and behavioral measures. RESULTS: One hundred and twenty nine adolescents with diverse identities and socioeconomic status enrolled. Among the 58% of participants who identified as lesbian, gay, bisexual, transgender, or queer (LGBTQ), 76% were out to at least one parent/guardian (outness). Mean WTP was 3.6 (of 5; 5 = definitely would participate) across all participants and both trial types. We found no evidence of an association between WTP and consent condition, LGBTQ identity, or outness. However, medical mistrust, communication with parents, and concern about HIV were associated with WTP. DISCUSSION: Our results suggest adolescents are willing to participate in HIV prevention trials and parental involvement in the consent process may not be the most important deciding factor. However, variation in WTP within consent groups, and variation in other significant variables, underscores the need for individualized approaches to recruitment and consent for these trials.

Individual factors predict substance use treatment course patterns among patients in community-based substance use disorder treatment.

BACKGROUND AND OBJECTIVES: Substance use disorders (SUDs) usually involve a complex natural trajectory of recovery alternating with symptom reoccurrence. This study examined treatment course patterns over time in a community SUD clinic. We examined depressive symptoms level, primary SUD assigned at each admission, and lifetime misuse of multiple substances as potential risk factors for premature treatment termination and subsequent treatment readmission. METHODS: De-identified longitudinal data were extracted from charts of 542 patients from an SUD treatment center. Survival analysis methods were applied to predict two time-to-event outcomes: premature treatment termination and treatment readmission. RESULTS: Primary opioid (vs alcohol) use disorder diagnosis at admission was associated with higher hazard of premature termination (HR = 1.91, p<0.001). The interaction between depressive symptoms level and substance use status (multiple vs single use) on treatment readmission was significant (p = 0.024), such that higher depressive symptoms level was predictive of readmission only among those with a history of single substance use (marginally significant effect). Lifetime use of multiple (vs single) substances (HR = 1.55, p = 0.002) and age (HR = 1.01, p = 0.019) predicted increased hazard of readmission. CONCLUSIONS: Findings did not support a universal role for depressive symptoms level in treatment course patterns. Primary SUD diagnosis, age, and history of substance misuse can be easily assessed and incorporated into treatment planning to support SUD patients and families. This study is the first to our knowledge that afforded a stringent test of these relationships and their interactions in a time-dependent, recurrent event, competing risks survival analysis examining both termination and readmission patterns utilizing a real-world clinic-based sample.

Semiparametric marginal regression for clustered competing risks data with missing cause of failure.

Clustered competing risks data are commonly encountered in multicenter studies. The analysis of such data is often complicated due to informative cluster size (ICS), a situation where the outcomes under study are associated with the size of the cluster. In addition, the cause of failure is frequently incompletely observed in real-world settings. To the best of our knowledge, there is no methodology for population-averaged analysis with clustered competing risks data with an ICS and missing causes of failure. To address this problem, we consider the semiparametric marginal proportional cause-specific hazards model and propose a maximum partial pseudolikelihood estimator under a missing at random assumption. To make the latter assumption more plausible in practice, we allow for auxiliary variables that may be related to the probability of missingness. The proposed method does not impose assumptions regarding the within-cluster dependence and allows for ICS. The asymptotic properties of the proposed estimators for both regression coefficients and infinite-dimensional parameters, such as the marginal cumulative incidence functions, are rigorously established. Simulation studies show that the proposed method performs well and that methods that ignore the within-cluster dependence and the ICS lead to invalid inferences. The proposed method is applied to competing risks data from a large multicenter HIV study in sub-Saharan Africa where a significant portion of causes of failure is missing.

COVID-19 Diagnosis and Risk of Death Among Adults With Cancer in Indiana: Retrospective Cohort Study.

BACKGROUND: Prior studies, generally conducted at single centers with small sample sizes, found that individuals with cancer experience more severe outcomes due to COVID-19, caused by SARS-CoV-2 infection. Although early examinations revealed greater risk of severe outcomes for patients with cancer, the magnitude of the increased risk remains unclear. Furthermore, prior studies were not typically performed using population-level data, especially those in the United States. Given robust prevention measures (eg, vaccines) are available for populations, examining the increased risk of patients with cancer due to SARS-CoV-2 infection using robust population-level analyses of electronic medical records is warranted. OBJECTIVE: The aim of this paper is to evaluate the association between SARS-CoV-2 infection and all-cause mortality among recently diagnosed adults with cancer. METHODS: We conducted a retrospective cohort study of newly diagnosed adults with cancer between January 1, 2019, and December 31, 2020, using electronic health records linked to a statewide SARS-CoV-2 testing database. The primary outcome was all-cause mortality. We used the Kaplan-Meier estimator to estimate survival during the COVID-19 period (January 15, 2020, to December 31, 2020). We further modeled SARS-CoV-2 infection as a time-dependent exposure (immortal time bias) in a multivariable Cox proportional hazards model adjusting for clinical and demographic variables to estimate the hazard ratios (HRs) among newly diagnosed adults with cancer. Sensitivity analyses were conducted using the above methods among individuals with cancer-staging information. RESULTS: During the study period, 41,924 adults were identified with newly diagnosed cancer, of which 2894 (6.9%) tested positive for SARS-CoV-2. The population consisted of White (n=32,867, 78.4%), Black (n=2671, 6.4%), Hispanic (n=832, 2.0%), and other (n=5554, 13.2%) racial backgrounds, with both male (n=21,354, 50.9%) and female (n=20,570, 49.1%) individuals. In the COVID-19 period analysis, after adjusting for age, sex, race or ethnicity, comorbidities, cancer type, and region, the risk of death increased by 91% (adjusted HR 1.91; 95% CI 1.76-2.09) compared to the pre-COVID-19 period (January 1, 2019, to January 14, 2020) after adjusting for other covariates. In the adjusted time-dependent analysis, SARS-CoV-2 infection was associated with an increase in all-cause mortality (adjusted HR 6.91; 95% CI 6.06-7.89). Mortality increased 2.5 times among adults aged 65 years and older (adjusted HR 2.74; 95% CI 2.26-3.31) compared to adults 18-44 years old, among male (adjusted HR 1.23; 95% CI 1.14-1.32) compared to female individuals, and those with ≥2 chronic conditions (adjusted HR 2.12; 95% CI 1.94-2.31) compared to those with no comorbidities. Risk of mortality was 9% higher in the rural population (adjusted HR 1.09; 95% CI 1.01-1.18) compared to adult urban residents. CONCLUSIONS: The findings highlight increased risk of death is associated with SARS-CoV-2 infection among patients with a recent diagnosis of cancer. Elevated risk underscores the importance of adhering to social distancing, mask adherence, vaccination, and regular testing among the adult cancer population.

Nonparametric tests for multistate processes with clustered data.

In this work, we propose nonparametric two-sample tests for population-averaged transition and state occupation probabilities for continuous-time and finite state space processes with clustered, right-censored, and/or left-truncated data. We consider settings where the two groups under comparison are independent or dependent, with or without complete cluster structure. The proposed tests do not impose assumptions regarding the structure of the within-cluster dependence and are applicable to settings with informative cluster size and/or non-Markov processes. The asymptotic properties of the tests are rigorously established using empirical process theory. Simulation studies show that the proposed tests work well even with a small number of clusters, and that they can be substantially more powerful compared to the only, to the best of our knowledge, previously proposed test for this problem. The tests are illustrated using data from a multicenter randomized controlled trial on metastatic squamous-cell carcinoma of the head and neck.

Retention in care and viral suppression in the PMTCT continuum at a large referral facility in western Kenya.

Medical records of pregnant and postpartum women living with HIV and their infants attending a large referral facility in Kenya from 2015 to 2019 were analyzed to identify characteristics associated with retention in care and viral suppression. Women were stratified based on the timing of HIV care enrollment: known HIV-positive (KHP; enrolled pre-pregnancy) and newly HIV-positive (NHP; enrolled during pregnancy). Associations with retention at 18 months postpartum and viral suppression (< 1000 copies/mL) were determined. Among 856 women (20% NHP), retention was 83% for KHPs and 53% for NHPs. Viral suppression was 88% for KHPs and 93% for NHPs, but 19% of women were missing viral load results. In a competing risk model, viral suppression increased by 18% for each additional year of age but was not associated with other factors. Overall, 1.9% of 698 infants with ≥ 1 HIV test result were HIV-positive. Tailored interventions are needed to promote retention and viral load testing, particularly for NHPs, in the PMTCT continuum.

Semiparametric regression on cumulative incidence function with interval-censored competing risks data and missing event types.

Competing risk data are frequently interval-censored, that is, the exact event time is not observed but only known to lie between two examination time points such as clinic visits. In addition to interval censoring, another common complication is that the event type is missing for some study participants. In this article, we propose an augmented inverse probability weighted sieve maximum likelihood estimator for the analysis of interval-censored competing risk data in the presence of missing event types. The estimator imposes weaker than usual missing at random assumptions by allowing for the inclusion of auxiliary variables that are potentially associated with the probability of missingness. The proposed estimator is shown to be doubly robust, in the sense that it is consistent even if either the model for the probability of missingness or the model for the probability of the event type is misspecified. Extensive Monte Carlo simulation studies show good performance of the proposed method even under a large amount of missing event types. The method is illustrated using data from an HIV cohort study in sub-Saharan Africa, where a significant portion of events types is missing. The proposed method can be readily implemented using the new function ciregic_aipw in the R package intccr.

Social, economic, and health effects of the COVID-19 pandemic on adolescents retained in or recently disengaged from HIV care in Kenya.

INTRODUCTION: Adolescents living with HIV (ALHIV, ages 10-19) experience complex challenges to adhere to antiretroviral therapy (ART) and remain in care, and may be vulnerable to wide-scale disruptions during the COVID-19 pandemic. We assessed for a range of effects of the pandemic on ALHIV in western Kenya, and whether effects were greater for ALHIV with recent histories of being lost to program (LTP). METHODS: ALHIV were recruited from an ongoing prospective study at 3 sites in western Kenya. The parent study enrolled participants from February 2019-September 2020, into groups of ALHIV either 1) retained in care or 2) LTP and traced in the community. Phone interviews from July 2020-January 2021 assessed effects of the pandemic on financial and food security, healthcare access and behaviors, and mental health. Responses were compared among the parent study groups. RESULTS: Phone surveys were completed with 334 ALHIV or their caregivers, including 275/308 (89.3%) in the retained group and 59/70 (84.3%) among those LTP at initial enrollment. During the pandemic, a greater proportion of LTP adolescents were no longer engaged in school (45.8% vs. 36.4%, p = 0.017). Over a third (120, 35.9%) of adolescents reported lost income for someone they relied on. In total, 135 (40.4%) did not have enough food either some (121, 36.2%) or most (14, 4.2%) of the time. More LTP adolescents (4/59, 6.8% vs. 2/275, 0.7%, p = 0.010) reported increased difficulties refilling ART. Adolescent PHQ-2 and GAD-2 scores were ≥3 for 5.6% and 5.2%, respectively. CONCLUSIONS: The COVID-19 pandemic has had devastating socioeconomic effects for Kenyan ALHIV and their households. ALHIV with recent care disengagement may be especially vulnerable. Meanwhile, sustained ART access and adherence potentially signal resilience and strengths of ALHIV and their care programs. Findings from this survey indicate the critical need for support to ALHIV during this crisis.

Factors associated with concordance between POLST orders and current treatment preferences.

BACKGROUND: POLST is widely used to document the treatment preferences of nursing facility residents as orders, but it is unknown how well previously completed POLST orders reflect current preferences (concordance) and what factors are associated with concordance. OBJECTIVES: To describe POLST preference concordance and identify factors associated with concordance. DESIGN: Chart reviews to document existing POLST orders and interviews to elicit current treatment preferences. SETTING: POLST-using nursing facilities (n = 29) in Indiana. PARTICIPANTS: Nursing facility residents (n = 123) and surrogates of residents without decisional capacity (n = 152). MEASUREMENTS: Concordance was determined by comparing existing POLST orders for resuscitation, medical interventions, and artificial nutrition with current treatment preferences. Comfort-focused POLSTs contained orders for do not resuscitate, comfort measures, and no artificial nutrition. RESULTS: Overall, 55.7% (123/221) of residents and 44.7% (152/340) of surrogates participated (total n = 275). POLST concordance was 44%, but concordance was higher for comfort-focused POLSTs (68%) than for non-comfort-focused POLSTs (27%) (p < 0.001). In the unadjusted analysis, increasing resident age (OR 1.04, 95% CI 1.01-1.07, p < 0.01), better cognitive functioning (OR 1.07, 95% CI 1.02-1.13, p < 0.01), surrogate as the decision-maker (OR 2.87, OR 1.73-4.75, p < 0.001), and comfort-focused POLSTs (OR 6.01, 95% CI 3.29-11.00, p < 0.01) were associated with concordance. In the adjusted multivariable model, only having an existing comfort-focused POLST was associated with higher odds of POLST concordance (OR 5.28, 95% CI 2.59-10.73, p < 0.01). CONCLUSIONS: Less than half of all POLST forms were concordant with current preferences, but POLST was over five times as likely to be concordant when orders reflected preferences for comfort-focused care. Findings suggest a clear need to improve the quality of POLST use in nursing facilities and focus its use among residents with stable, comfort-focused preferences.

Development and Testing of the Dysmenorrhea Symptom Interference (DSI) Scale.

Dysmenorrhea affects most reproductive-age women and increases the risk of future pain. To evaluate dysmenorrhea interventions, validated outcome measures are needed. In this two-phase study, we developed and tested the dysmenorrhea symptom interference scale. During the scale-development phase (n = 30), we created a nine-item scale based on qualitative data from cognitive interviews. During the scale-testing phase (n = 686), we evaluated reliability, validity, and responsiveness to change. The scale measures how dysmenorrhea symptoms interfere with physical, mental, and social activities. Internal consistency was strong with Cronbach's α > 0.9. Test-retest reliability was acceptable (r = 0.8). The scale showed satisfactory content validity, construct validity (supported by confirmatory factor analysis), concurrent validity, and responsiveness to change. The minimally important difference was 0.3 points on a scale with a possible total score ranging from 1 to 5. This new psychometrically sound scale can be used in research and clinical practice to facilitate the measurement and management of dysmenorrhea.

Nonparametric analysis of nonhomogeneous multistate processes with clustered observations.

Frequently, clinical trials and observational studies involve complex event history data with multiple events. When the observations are independent, the analysis of such studies can be based on standard methods for multistate models. However, the independence assumption is often violated, such as in multicenter studies, which makes standard methods improper. This work addresses the issue of nonparametric estimation and two-sample testing for the population-averaged transition and state occupation probabilities under general multistate models with cluster-correlated, right-censored, and/or left-truncated observations. The proposed methods do not impose assumptions regarding the within-cluster dependence, allow for informative cluster size, and are applicable to both Markov and non-Markov processes. Using empirical process theory, the estimators are shown to be uniformly consistent and to converge weakly to tight Gaussian processes. Closed-form variance estimators are derived, rigorous methodology for the calculation of simultaneous confidence bands is proposed, and the asymptotic properties of the nonparametric tests are established. Furthermore, I provide theoretical arguments for the validity of the nonparametric cluster bootstrap, which can be readily implemented in practice regardless of how complex the underlying multistate model is. Simulation studies show that the performance of the proposed methods is good, and that methods that ignore the within-cluster dependence can lead to invalid inferences. Finally, the methods are illustrated using data from a multicenter randomized controlled trial.

Do Life-sustaining Treatment Orders Match Patient and Surrogate Preferences? The Role of POLST.

BACKGROUND: It is essential to high-quality medical care that life-sustaining treatment orders match the current, values-based preferences of patients or their surrogate decision-makers. It is unknown whether concordance between orders and current preferences is higher when a POLST form is used compared to standard documentation practices. OBJECTIVE: To assess concordance between existing orders and current preferences for nursing facility residents with and without POLST forms. DESIGN: Chart review and interviews. SETTING: Forty Indiana nursing facilities (29 where POLST is used and 11 where POLST is not in use). PARTICIPANTS: One hundred sixty-one residents able to provide consent and 197 surrogate decision-makers of incapacitated residents with and without POLST forms. MAIN MEASUREMENTS: Concordance was measured by comparing life-sustaining treatment orders in the medical record (e.g., orders about resuscitation, intubation, and hospitalization) with current preferences. Concordance was analyzed using population-averaged binary logistic regression. Inverse probability weighting techniques were used to account for non-response. We hypothesized that concordance would be higher in residents with POLST (n = 275) in comparison to residents without POLST (n = 83). KEY RESULTS: Concordance was higher for residents with POLST than without POLST (59.3% versus 34.9%). In a model adjusted for resident, surrogate, and facility characteristics, the odds were 3.05 times higher that residents with POLST had orders for life-sustaining treatment match current preferences in comparison to residents without POLST (OR 3.05 95% CI 1.67-5.58, p < 0.001). No other variables were significantly associated with concordance. CONCLUSIONS: Nursing facility residents with POLST are significantly more likely than residents without POLST to have concordance between orders in their medical records and current preferences for life-sustaining treatments, increasing the likelihood that their treatment preferences will be known and honored. However, findings indicate further systems change and clinical training are needed to improve POLST concordance.

A pseudo-likelihood method for estimating misclassification probabilities in competing-risks settings when true-event data are partially observed.

Outcome misclassification occurs frequently in binary-outcome studies and can result in biased estimation of quantities such as the incidence, prevalence, cause-specific hazards, cumulative incidence functions, and so forth. A number of remedies have been proposed to address the potential misclassification of the outcomes in such data. The majority of these remedies lie in the estimation of misclassification probabilities, which are in turn used to adjust analyses for outcome misclassification. A number of authors advocate using a gold-standard procedure on a sample internal to the study to learn about the extent of the misclassification. With this type of internal validation, the problem of quantifying the misclassification also becomes a missing data problem as, by design, the true outcomes are only ascertained on a subset of the entire study sample. Although, the process of estimating misclassification probabilities appears simple conceptually, the estimation methods proposed so far have several methodological and practical shortcomings. Most methods rely on missing outcome data to be missing completely at random (MCAR), a rather stringent assumption which is unlikely to hold in practice. Some of the existing methods also tend to be computationally-intensive. To address these issues, we propose a computationally-efficient, easy-to-implement, pseudo-likelihood estimator of the misclassification probabilities under a missing at random (MAR) assumption, in studies with an available internal-validation sample. We present the estimator through the lens of studies with competing-risks outcomes, though the estimator extends beyond this setting. We describe the consistency and asymptotic distributional properties of the resulting estimator, and derive a closed-form estimator of its variance. The finite-sample performance of this estimator is evaluated via simulations. Using data from a real-world study with competing-risks outcomes, we illustrate how the proposed method can be used to estimate misclassification probabilities. We also show how the estimated misclassification probabilities can be used in an external study to adjust for possible misclassification bias when modeling cumulative incidence functions.

On shared gamma-frailty conditional Markov model for semicompeting risks data.

Semicompeting risks data are a mixture of competing risks data and progressive state data. This type of data occurs when a nonterminal event is subject to truncation by a well-defined terminal event, but not vice versa. The shared gamma-frailty conditional Markov model (GFCMM) has been used to analyze semicompeting risks data because of its flexibility. There are two versions of this model: the restricted and the unrestricted model. Maximum likelihood estimation methodology has been proposed in the literature. However, we found through numerical experiments that the unrestricted model sometimes yields nonparametrically biased estimation. In this article, we provide a practical guideline for using the GFCMM in the analysis of semicompeting risk data that includes: (a) a score test to assess if the restricted model, which does not exhibit estimation problems, is reasonable under a proportional hazards assumption, and (b) a graphical illustration to justify whether the unrestricted model yields nonparametric estimation with substantial bias for cases where the test provides a statistical significant result against the restricted model. This guideline was applied to the Indianapolis-Ibadan Dementia Project data as an illustration to explore how dementia occurrence changes mortality risk.

Nonparametric tests for transition probabilities in nonhomogeneous Markov processes.

This paper proposes nonparametric two-sample tests for the direct comparison of the probabilities of a particular transition between states of a continuous time non-homogeneous Markov process with a finite state space. The proposed tests are a linear nonparametric test, an L 2-norm-based test and a Kolmogorov-Smirnov-type test. Significance level assessment is based on rigorous procedures, which are justified through the use of modern empirical process theory. Moreover, the L 2-norm and the Kolmogorov-Smirnov-type tests are shown to be consistent for every fixed alternative hypothesis. The proposed tests are also extended to more complex situations such as cases with incompletely observed absorbing states and non-Markov processes. Simulation studies show that the test statistics perform well even with small sample sizes. Finally, the proposed tests are applied to data on the treatment of early breast cancer from the European Organization for Research and Treatment of Cancer (EORTC) trial 10854, under an illness-death model.