Summary Estimates Derived from a Multi-state Non-Markov Framework to Characterize the Course of Heart Disease

Ming Ding, Feng-Chang Lin, Michelle L. Meyer
doi: https://doi.org/10.1101/2024.09.18.24313882

ABSTRACT

Multi-state Markov models have been used to model the course of chronic disease. However, they are unsuitable to chronic disease where past and present states interplay and affect future states, and the estimated transition probabilities are time-specific which are not straightforward for public health interpretation. We have proposed a multi-state non-Markov framework that splits disease states into substates conditioning on past states. As the substates track past states and indicate multimorbidity, the estimated transition rates can be used to derive two summary estimates: Disease path which shows path of state transition, and multimorbidity-adjusted life year (MALY) which represents the adjusted life year in full health. In this paper, we showed the derivation of the two summary estimates and applied them to characterize the course of heart disease using data from the Atherosclerosis Risk in Communities Study (ARIC) study. The course of heart disease was modeled in five states, namely, healthy, at metabolic risk, coronary heart disease (CHD), heart failure, and mortality. In this mid- to old-age population, the estimated MALY was 24.13 (95% CI: 16.55, 32.06) years. For healthy participants at baseline, the most likely disease paths were: “Healthy → at metabolic risk → mortality” (37%), “Healthy → mortality” (21%), “Healthy → at metabolic risk → heart failure → mortality” (19%), and “Healthy → at metabolic risk → CHD → mortality” (8%). The MALY was higher among women than men and higher among Whites than Blacks. The distribution of disease path was similar across sex and race subgroups. In summary, MALY and disease path characterize the disease course in a summary manner and have potential use in chronic disease prevention.

1. INTRODUCTION

Multi-state models, which fit one survival model to each state, have been used to describe course of chronic disease. Most of the existing methods are Markov models, which assume that present and future states are independent of past history.1, 2 Markov models are unsuitable to describe the course of chronic disease for several reasons. First, for chronic disease, past and present states interplay and affect future states which violates the Markov assumption. Second, by assuming present state is independent of past history, Markov models do not allow for multimorbidity, a common condition for chronic disease.3–5 Finally, the transition probabilities estimated from Markov models are time-specific, and these high-dimensional matrices are not straightforward for public health interpretation.

Non-Markov models are a largely unexplored field, with only several non-parametric methods proposed,6–12 such as the landmark Aalen-Johansen (LMAJ) estimator.6 We have proposed a multi-state non-Markov regression framework to investigate the course of chronic disease.13 By conditioning on past disease history, disease states are divided into substates to convert non-Markov to Markov process, where cause-specific Cox models are used to estimate transition rate between substates. As substates memorize disease history and track multimorbidity, the transition rates between substates can be synthesized into summary estimates to characterize the whole disease course: Disease path, and multimorbidity-adjusted life year (MALY) which represents the adjusted life year in full health. The two summary estimates are easy to interpret and can be used for Markov process (as a special case for non-Markov process), which may advance multi-state modeling in disease prevention. In this paper, we showed the derivation of the two summary estimates from our non-Markov framework and applied them to characterize the course of heart disease in a large cohort study.

2. METHODS

2.1. An introduction to the multi-state non-Markov framework

We developed a non-Markov regression framework that allows past states to affect transition rates of current states.13 While it can be used for chronic disease of any states, we assume five states S0-S4 for illustration purpose (Figure 1a). We assume that transition occurs between any two states, while our method works if there is no transition between two states. The details of our framework are shown below.

Figure 1.
Figure 1. Convert non-Markov to Markov process by splitting disease states into substates.13

Convert a non-Markov to Markov process

Our framework allows for non-Markov process, where past disease states can affect current and future states. However, the Aalen-Johansen (A-J) estimator is systematically biased in non-Markov process, causal difficulty in estimation of transition probabilities.10 Thus, we divide disease states into substates by conditioning on past disease history to convert the non-Markov process to Markov process (Figure 1b). As the substates are independent of past disease states, the converted process in Figure 1b satisfies the Markov assumption.13

Construct transition rate matrix between substates 𝐐(𝐭)

The matrix of transition rates can be constructed Based on Figure 1b (Figure 2). Each row and each column stand for a disease substate, and 𝐐(𝐭)[a,b] stands for the transition rate from the substate in row a to the substate in column b. If there is no transition between two substates, the transition rate is 0.

Figure 2.
Figure 2. Transition rate matrix Q(t).13

Estimate transition rates 𝐐(𝐭)

The transition rate matrix is estimated by fitting multiple cause-specific Cox (CSC) models, which was described in detail in our previous paper.14 In brief, we subset the data by disease states and change the data from wide to long structure by age. CSC model is fitted to each subset to estimate transition rates.13 As the data is changed into long structure by age, the predicted risk at the end of each time interval is approximate to age-specific hazard in discrete time, which can be estimated using cumulative incidence function (CIF) to account for competing risks.13

Estimate state occupational probabilities 𝐏(t). 𝐏(t) =

(p0(t), p1(t), p2|0(t), p2|0_1(t), p3|0(t), p3|0_1(t), p3|0_2(t), p3|0_1_2(t), p4(t)), where p(t) is the probability of participants in a particular state. Based on the discrete-time A-J estimator, the change in 𝐏(t) from t to t+1 can be estimated as 𝐏’(t) = 𝐏(t) ∗ 𝐐(t). Thus, 𝐏(t2 + 1) = 𝐏(t = t1) ∗ Embedded Image, where 𝐐(t) is the transition rate matrix and I9 an identity matrix. In this paper, based on the age-specific transition rates 𝐐(𝐭) and state occupational probabilities 𝐏(𝐭), we derive two summary estimates, MALY and disease path.

2.2. Estimate MALY

To estimate MALY, we first estimate disease-state specific life year (SSLY), which is calculated as sum of state occupation probabilities in each state from start age until death. Specifically, SSLY in substate k from 𝑡1 to 𝑡2 can be estimated as Embedded Image, where 𝐏(t)[1, 𝑘] is the kth column of 𝐏(t). The substate k ranges from 1 to 8 refers to substates S0, S1, S2|0, S2|0_1, S3|0, S3|0_1, S3|0_2, and S3|0_1_2, respectively. We stop estimating 𝐏(t) at age t2 if 𝐏(t)[1,9]>0.95, as we assume state occupational probability for death>0.95 indicates a participant’s death.

The substates identified from our non-Markov framework inform multimorbidity. For example, S3|0_1 shows a participant in state S3 with a history of S0 and S1 and indicates multimorbidity of S1 and S3. MALY is calculated as an average of SSLY across substates weighted by multimorbidity, where the weight is calculated using a multiplicative approach.15, 16 Particularly, we assign a disability weight (DW) to substates without multimorbidity (i.e., S0, S1, S2|0, S3|0), which measures the severity of health state from disease or injury. DW has a value of 0-1, with 0 indicating full health and 1 indicating death.15 We use a multiplicative approach to assign DW to states with multimorbidity (i.e., S2|0_1, S3|0_1, S3|0_2, and S3|0_1_2).16 For example, for multimorbidity with two diseases (e.g., S2|0_1), DW2|0_1 1 − (1 − DW2)(1 − DW1), and for multimorbidity with three diseases (S3|0_1_2), DW3|0_1_2 1 − (1 − DW1)(1 − DW2) (1 − DW3). As all substates are mutually exclusive, MALY ∑𝑘 DWk ∗ Lk, which is as a weighted sum of SSLY across substates.

2.3. Estimate disease path

Path of disease can be estimated by decomposing death into probabilities through each disease path. We use state occupation probabilities multiplied by transition rates to estimate age-specific probability of death through each path, and then sum up the probabilities over time to obtain probability of each path. By applying 𝐏’(t) = 𝐏(t) ∗ 𝐐(t), change in 𝐏(t + 1) from 𝐏(t) can be calculated for each substate. Particularly, Embedded Image, which is change in 𝐏(t + 1) from 𝐏(t) for substate k, can be estimated as 𝐏(t) ∗ (𝐐(t)[, k]). Thus, the probability of increase in state of death from time t to t + 1 is 𝐏(t) ∗ (𝐐(t)[, 9]); the probability of this increase in death transited from substate k is (𝐏(t)[1, k]) ∗ (𝐐(t)[𝑘, 9]); and the probability of death from time 𝑡1to 𝑡2 transited from substate k are Embedded Image ∗ (𝐐(t)[𝑘, 9]). Specifically, for k 1 to 9, Embedded Image ∗ (𝐐(t)[𝑘, 9]) calculates probabilities of death transited from S0, S0→S1, S0→S2, S0→S1→S2, S0→S3, S0→S1→S3, S0→S2→S3, S0→S1→S2→S3, and S4, respectively.

Use bootstrap to obtain 95% CIs

We adopted a parametric bootstrapping approach to repeatedly draw transition rates from their corresponding distributions to estimate the variability of the parameter estimate. We repeated the resampling process 100 times in each estimation by the CSC model. Statistically, the resulting 100 parameter estimates represent the empirical distribution. MALY are summarized with median values and 95% CI defined by 2.5th to 97.5th percentiles of the empirical distribution, and probabilities of each disease path is calculated over the 100 estimates.

3. APPLICATION

3.1. Study population

The Atherosclerosis Risk in Communities Study (ARIC) study was designed to investigate the causes of atherosclerosis and its clinical outcomes, as well as variations in cardiovascular risk factors and disease by sex and race.17 The enrolled participants ARIC underwent a phone interview and clinic visit at baseline and were followed up by telephone calls and re-examinations from 1987-2019. Participants were contacted periodically by phone and interviewed about interim hospital admissions, cardiovascular outpatient diagnoses, and deaths. Participants who reported CVD-related events were asked to provide medical records that were reviewed by physicians.18 For each event, the month and year of diagnosis were recorded as the diagnosis date. Heart failure was ascertained by surveillance calls or clinic visits and was verified from death certificates, medical records, and outpatient diagnoses. Deaths were identified from systematic searches of the vital records of states and of the National Death Index, supplemented by reports from family members and postal authorities.19 We obtained ARIC data through Biologic Specimen and Data Repository Information Coordinating Center (BioLINCC), an open repository created by the National Heart, Lung, and Blood Institute (NHLBI), with the Institutional Review Board (IRB) deemed exempt by the University of North Carolina (UNC)-Chapel Hill Review Board.

3.2. Methods

We modeled the course of heart disease in five states: Healthy, at metabolic risk, coronary heart disease (CHD), heart failure, and mortality. At metabolic risk state was defined as development of hypertension, hyperlipidemia, or diabetes. Hypertension was defined as blood pressure ≥140/90 mmHg or a history of hypertension or use of blood pressure medications.20 Hyperlipidemia included primary hypertriglyceridemia (≥175 mg/dL) or primary hypercholesterolemia (LDL-c 160–189 mg/dL, and/or non-HDL-c 190–219 mg/dL).21 Diabetes was defined as fasting glucose ≥7.7mmol/L.22 We calculated the time of incidence of at-risk state as the earliest time that any of the risk factors were documented. For participants entering CHD and mortality states at the same time (i.e., incident fatal CHD), we assigned the next state as CHDW. Similarly, we assigned the next state as heart failure for incident fatal heart failure. For participants entering CHD and heart failure states at the same time (i.e., incident CHD combined with heart failure), we assigned the next state as CHD.

We applied our non-Markov framework to examine the course of heart disease. In particular, S0 to S4 states referred to healthy, at-metabolic risk, CHD, heart failure, and mortality states, respectively. By conditioning on past disease history, S2|0 and S2|0_1 referred to CHD without metabolic risk factors and CHD with metabolic risk factors, respectively, and S3|0, S3|0_1, S3|0_2, S3|0_1_2 referred to heart failure with no metabolic risk factors or CHD, heart failure only with metabolic risk factors, heart failure only with CHD, and heart failure with metabolic risk factors and CHD, respectively. We divided the data into subsets 1-4, starting from healthy, at-risk, CHD, and heart failure states, respectively. Within each subset, we changed the data into a long format by dividing them into small intervals by age and applied CSC models. We modeled age flexibly in cubic terms and included past disease history and the interaction terms with age as covariates in subsets 3 and 4. To estimate MALY, we used disability weights estimated from the Global Burden of Disease Study 2019,23 where the disability weights were 0.049, 0.072, and 0.074, for at-metabolic-risk (S1|0), CHD (S2|0), and heart failure (S3|0) states, respectively. For states with multimorbidity, a multiplicative approach was used to assign disability weights. For example, for CHD transited from healthy and at-metabolic-risk states, DW2|0_1 1 − (1 − 0.049)(1 − 0.072).

3.3. Results

Our study included 15,027 participants at baseline, including 6831 men and 8196 women and 3917 Blacks and 11,110 Whites. Of the 15,027 participants, 4619 participants were healthy, and 10,408 participants were at metabolic risk at baseline. During a median of 27 years follow up, 2057 participants developed incidence of CHD, 3283 participants developed incidence of heart failure, and 7677 participants died. Figure 3 shows the number of participants transited through the disease course over the follow-up period. The participants’ transition between states by sex and race is shown in Figures S1-S4.

Figure S1.
Figure S1. Participants’ transition through the disease course by sex in the Atherosclerosis Risk in Communities Study (ARIC) study starting from 1987.
Figure S2.
Figure S2. Participants’ transition through the disease course by race in the Atherosclerosis Risk in Communities Study (ARIC) study starting from 1987.
Figure 3.
Figure 3. Participants’ transition through the disease course in the Atherosclerosis Risk in Communities Study (ARIC) study from 1987-2019 (n=15027).

Table 1 shows the projected expected life years spent in each disease state. From the start of follow up, the expected life years were longest in at-metabolic-risk state (median: 18.25 years), followed by healthy state (median: 12.26 years). By summing up all disease states, the total expected life years was 25.69 (95% CI: 18.02, 34.16) years, which led to an estimated life expectancy of 79.87 (95% CI: 78.56, 82.06) years. By assigning a disability weight to account for multimorbidity in each state, the estimated MALY was 24.13 (95% CI: 16.55, 32.06) years, and estimated multimorbidity-adjusted life expectancy was 78.23 (95% CI: 75.95, 80.96) years. The multimorbidity-adjusted life expectancy was higher among women (79.29 [95% CI: 77.42, 81.57] years) than men (77.07 [95% CI: 74.37, 80.05] years) (Table 2), and was higher among Whites (79.04 [95% CI: 77.05, 81.31] years) than Blacks (75.49 [95% CI: 73.19, 79.30] years) (Table 3).

View this table:
Table 1. Estimated disease state-specific life year and multimorbidity-adjusted life year (MALY) in the Atherosclerosis Risk in Communities Study (ARIC) study starting from 1987 (n=15027).
View this table:
Table 2. Estimated disease state-specific life year and multimorbidity-adjusted life year (MALY) by sex i in the Atherosclerosis Risk in Communities Study (ARIC) study starting from 1987.
View this table:
Table 3. Estimated disease state-specific life year and multimorbidity-adjusted life year (MALY) by race in the Atherosclerosis Risk in Communities Study (ARIC) study starting from 1987.

For healthy participants at baseline, the most likely disease paths estimated from our framework are: “Healthy → at metabolic risk → mortality” (37%), “Healthy → mortality” (21%), “Healthy → at metabolic risk → heart failure → mortality” (19%), and “Healthy → at metabolic risk → CHD → mortality” (8%) (Table 4). For participants at metabolic risk at baseline, 51% of participants were estimated to transit through the path “At metabolic risk → mortality”, 28% of participants were through the path “At metabolic risk → heart failure → mortality”, and 13% of participants were through the path “At metabolic risk → CHD → mortality”. The estimated disease paths were similar between women and men (Table 5), and similar between Whites and Blacks (Table 6).

View this table:
Table 4. Estimated disease path in the Atherosclerosis Risk in Communities Study (ARIC) study starting from 1987 (n=15027).
View this table:
Table 5. Estimated disease path by sex in the Atherosclerosis Risk in Communities Study (ARIC) study starting from 1987.
View this table:
Table 6. Estimated disease path by race in the Atherosclerosis Risk in Communities Study (ARIC) study starting from 1987.

4. DISCUSSION

In this study, we derived two summary estimates to characterize disease course, MALY which accounts for multimorbidity and estimates the adjusted life year in full health, and disease path which estimates path of state transition over disease course. We used the two estimates to characterize the course of heart disease and showed that they have potential to be used in chronic disease prevention.

From a methodology perspective, the summary estimates derived from our study may advance multi-state modeling in disease prevention. Multi-state Markov models have been used to investigate course of chronic disease, which estimate hazard ratio of exposures with disease state transition.1, 2 Semi-Markov model models relax the Markov assumption by allowing future states to depend on duration in current state.24, 25 However, these models may be inappropriate to model the course of chronic disease, as past states of chronic disease can interplay in affecting future states which violates the Markov assumption. Most importantly, transition probabilities estimated from these models are time-specific and are not straightforward for public health interpretation. Our non-Markov framework conditions on past states to divide disease states into substates to convert a non-Markov to Markov process. As the substates memorize past disease history and inform multimorbidity, the transition rates between substates can be used to generate MALY and disease path. These summary estimates characterize the whole disease course, are easy to interpret, can be used for Markov process (as a special case of non-Markov process), and will potentially advance multi-state modeling in comparative effectiveness of intervention, which compares effect of intervention to inform decision-making. Focusing on one disease endpoint may not allow to fully assess benefits and harms of intervention, which leads to inaccurate estimation of intervention effect and misidentify optimal intervention. MALY estimates life year across the disease course accounting for multimorbidity and can be used by health policy makers to better evaluate intervention effects to facilitate policy decision-making. Disease paths characterize long-term disease course and can facilitate medical resources planning for effective early prevention of disease progression.

Heart disease has been the leading cause of mortality in the U.S.,26 and imposes a heavy economic burden to families and society.27 Previous epidemiological studies have generated fruitful knowledge about heart disease prevention,28–32 however, most of them have focused on one single endpoint, such as hypertension, CHD, or heart failure, or a composite endpoint such as CVD.21, 33–36 However, the development of heart disease is a multi-state process that is biologically inseparable: Hypertension and hyperlipidemia alter vascular elasticity, narrow vascular lumen, increase cardiac work, and result in heart disease.37, 38 Although a few studies applied multi-Markov models to investigate heart disease prevention,39, 40 the inconsistent associations of exposures with state transitions make it difficult to draw an conclusion. For example, the China Kadoorie Biobank showed a healthy lifestyle played different roles in states transitions, which was associated with a lower rate of transition of healthy to cardiometabolic disease (CMD) but not associated with transition from CMD to mortality.39 The MALY derived from our study synthesizes transition rates across disease states into one estimate, which can be used to evaluate the effects of prevention across disease course.

Our study showed that the longest life year was from at-metabolic-risk state, providing evidence to formulate prevention strategies specifically targeting at-metabolic-risk population. The American Heart Association “Life’s Essential 8” should be adopted at early states to prevent the development of heart disease, which includes maintaining a normal weight, consuming a healthy diet, being physically active, no smoking, and adequate amount of sleep.41 The life expectancy in the U.S. has drastically increased over the past 50 years, primarily driven by a continuous increase in survival rate after CVD prognosis.42, 43 The estimated life year for states with CHD or heart failure allows to estimate the demand for health services and the costs for treatment more precisely. Moreover, the estimated life year of disease states with CHD or heart failure was short and suggested the importance of secondary prevention of heart disease. For example, adopting a healthy lifestyle together with statin use have shown long-term benefits for secondary prevention of heart disease and mortality.44, 45

The estimated path of heart disease is of potential public health significance in the early prevention of heart disease. Our study found that the most likely path of heart disease progression was “healthy → at metabolic risk → mortality” followed by “Healthy → at metabolic risk → heart failure → mortality”. This highlights the importance of early screening of heart failure among participants with metabolic risk factors. Heart failure is commonly considered a consequence of CHD, where CHD can damage heart muscle and cause left ventricular dysfunction and heart failure.46 However, the incidence of heart failure has substantially increased in the past decades, with etiologies differing from CHD.47 To identify heart failure at early stage, heart failure-specific biomarker, B-type natriuretic peptide (BNP), can be used to identify heart failure before structural and functional changes of heart muscle become obvious.48–51 Moreover, echocardiogram-based predictors, such as left ventricular ejection fraction, can detect left ventricular dysfunction and heart muscle damage.52

Our study found that the estimated MALY was higher among women than men and higher among Whites than Blacks. Consistently, previous studies have shown that biological race and sex influence risk of heart disease:53–57 The prevalence of heart disease is higher in men than women,54 and higher among Blacks than Whites.53 One reason for this is due to the different prevalence of healthy lifestyles. The prevalence of obesity and being physically inactivity is higher among Blacks than Whites, although the prevalence of smoking is higher among Whites.53 Men are more likely to smoke, although they are more active than women.55–57 The other reason may be due to the racial disparities in accessing healthcare for heart disease prevention and treatment. Studies have shown that the overall health care use was higher among Whites than Blacks, and this have persisted for 6 decades and widened in recent years.58, 59

In conclusion, we proposed two summary estimates derived from a multi-state non-Markov framework and applied it to characterize the course of heart disease. The two estimates describe disease course in a summary manner and have potential use in chronic disease prevention.

Data Availability

The data can be requested through Biologic Specimen and Data Repository Information Coordinating Center (BioLINCC).

Reference

  1. 1.↵
    Meira-Machado L, de Una-Alvarez J, Cadarso-Suarez C, Andersen PK. Multi-state models for the analysis of time-to-event data. Stat Methods Med Res. 2009;18(2):195–222. Epub 20080618. doi: 10.1177/0962280208092301. PubMed PMID: 18562394; PMCID: PMC2692556.
    OpenUrlCrossRefPubMed
  2. 2.↵
    de Wreede LC, Fiocco M, Putter H. The mstate package for estimation and prediction in non- and semi-parametric multi-state and competing risks models. Comput Methods Programs Biomed. 2010;99(3):261–74. Epub 2010/03/17. doi: 10.1016/j.cmpb.2010.01.001. PubMed PMID: 20227129.
    OpenUrlCrossRefPubMedWeb of Science
  3. 3.↵
    Drozd M, Relton SD, Walker AMN, Slater TA, Gierula J, Paton MF, Lowry J, Straw S, Koshy A, McGinlay M, Simms AD, Gatenby VK, Sapsford RJ, Witte KK, Kearney MT, Cubbon RM. Association of heart failure and its comorbidities with loss of life expectancy. Heart. 2021;107(17):1417–21. Epub 20201105. doi: 10.1136/heartjnl-2020-317833. PubMed PMID: 33153996; PMCID: PMC8372397.
    OpenUrlAbstract/FREE Full Text
  4. 4.
    Khan MS, Samman Tahhan A, Vaduganathan M, Greene SJ, Alrohaibani A, Anker SD, Vardeny O, Fonarow GC, Butler J. Trends in prevalence of comorbidities in heart failure clinical trials. Eur J Heart Fail. 2020;22(6):1032–42. Epub 2020/04/16. doi: 10.1002/ejhf.1818. PubMed PMID: 32293090; PMCID: PMC7906002.
    OpenUrlCrossRefPubMed
  5. 5.↵
    Tran J, Norton R, Conrad N, Rahimian F, Canoy D, Nazarzadeh M, Rahimi K. Patterns and temporal trends of comorbidity among adult patients with incident cardiovascular disease in the UK between 2000 and 2014: A population-based cohort study. PLoS Med. 2018;15(3):e1002513. Epub 20180306. doi: 10.1371/journal.pmed.1002513. PubMed PMID: 29509757; PMCID: PMC5839540.
    OpenUrlCrossRefPubMed
  6. 6.↵
    Putter H, Spitoni C. Non-parametric estimation of transition probabilities in non-Markov multi-state models: The landmark Aalen-Johansen estimator. Stat Methods Med Res. 2018;27(7):2081–92. Epub 2018/05/31. doi: 10.1177/0962280216674497. PubMed PMID: 29846146.
    OpenUrlCrossRefPubMed
  7. 7.
    de Una-Alvarez J, Meira-Machado L. Nonparametric estimation of transition probabilities in the non-Markov illness-death model: A comparative study. Biometrics. 2015;71(2):364–75. Epub 2015/03/05. doi: 10.1111/biom.12288. PubMed PMID: 25735883.
    OpenUrlCrossRefPubMed
  8. 8.
    Pepe MS. Inference for Events With Dependent Risks in Multiple Endpoint Studies. Journal of the American Statistical Association. 1991;86(415):770–8.
    OpenUrlCrossRefWeb of Science
  9. 9.
    Andersen PK, Wandall ENS, Pohar Perme M. Inference for transition probabilities in non-Markov multi-state models. Lifetime Data Anal. 2022;28(4):585–604. Epub 20220628. doi: 10.1007/s10985-022-09560-w. PubMed PMID: 35764854.
    OpenUrlCrossRefPubMed
  10. 10.↵
    Meira-Machado L, de Una-Alvarez J, Cadarso-Suarez C. Nonparametric estimation of transition probabilities in a non-Markov illness-death model. Lifetime Data Anal. 2006;12(3):325–44. Epub 2006/08/19. doi: 10.1007/s10985-006-9009-x. PubMed PMID: 16917736.
    OpenUrlCrossRefPubMedWeb of Science
  11. 11.
    Glidden DV. Robust inference for event probabilities with non-Markov event data. Biometrics. 2002;58(2):361–8. doi: 10.1111/j.0006-341x.2002.00361.x. PubMed PMID: 12071409.
    OpenUrlCrossRefPubMedWeb of Science
  12. 12.↵
    Meira-Machado L, Sestelo M. Estimation in the progressive illness-death model: A nonexhaustive review. Biom J. 2019;61(2):245–63. Epub 20181120. doi: 10.1002/bimj.201700200. PubMed PMID: 30457674.
    OpenUrlCrossRefPubMed
  13. 13.↵
    Ding M, Chen H, Lin F. A Multi-state Non-Markov Framework to Estimate Progression of Chronic Disease. MedRxiv doi: https://doiorg/101101/2024021524302901. 2024.
  14. 14.↵
    Lau B, Cole SR, Gange SJ. Competing risk regression models for epidemiologic data. Am J Epidemiol. 2009;170(2):244–56. Epub 20090603. doi: 10.1093/aje/kwp107. PubMed PMID: 19494242; PMCID: PMC2732996.
    OpenUrlCrossRefPubMedWeb of Science
  15. 15.↵
    Murray CJ. Quantifying the burden of disease: the technical basis for disability-adjusted life years. Bull World Health Organ. 1994;72(3):429–45. PubMed PMID: 8062401; PMCID: PMC2486718.
    OpenUrlCrossRefPubMedWeb of Science
  16. 16.↵
    Mathers CD, Sadana R, Salomon JA, Murray CJ, Lopez AD. Healthy life expectancy in 191 countries, 1999. Lancet. 2001;357(9269):1685-91. doi: 10.1016/S0140-6736(00)04824-8. PubMed PMID: 11425392.
    OpenUrlCrossRefPubMedWeb of Science
  17. 17.↵
    The Atherosclerosis Risk in Communities (ARIC) Study: design and objectives. The ARIC investigators. Am J Epidemiol. 1989;129(4):687–702. Epub 1989/04/01. PubMed PMID: 2646917.
    OpenUrlCrossRefPubMed
  18. 18.↵
    Rose GA. Cardiovascular survey methods. Geneva Albany, N.Y.: World Health Organization; WHO Publications Centre distributor 1982.
  19. 19.↵
    Fung TT, Chiuve SE, McCullough ML, Rexrode KM, Logroscino G, Hu FB. Adherence to a DASH-style diet and risk of coronary heart disease and stroke in women. Archives of internal medicine. 2008;168(7):713–20. Epub 2008/04/17. doi: 10.1001/archinte.168.7.713. PubMed PMID: 18413553.
    OpenUrlCrossRefPubMedWeb of Science
  20. 20.↵
    Unger T, Borghi C, Charchar F, Khan NA, Poulter NR, Prabhakaran D, Ramirez A, Schlaich M, Stergiou GS, Tomaszewski M, Wainford RD, Williams B, Schutte AE. 2020 International Society of Hypertension Global Hypertension Practice Guidelines. Hypertension. 2020;75(6):1334-57. Epub 20200506. doi: 10.1161/HYPERTENSIONAHA.120.15026. PubMed PMID: 32370572.
    OpenUrlCrossRefPubMed
  21. 21.↵
    Grundy SM, Stone NJ, Bailey AL, Beam C, Birtcher KK, Blumenthal RS, Braun LT, de Ferranti S, Faiella-Tommasino J, Forman DE, Goldberg R, Heidenreich PA, Hlatky MA, Jones DW, Lloyd-Jones D, Lopez-Pajares N, Ndumele CE, Orringer CE, Peralta CA, Saseen JJ, Smith SC, Jr., Sperling L, Virani SS, Yeboah J. 2018 AHA/ACC/AACVPR/AAPA/ABC/ACPM/ADA/AGS/APhA/ASPC/NLA/PCNA Guideline on the Management of Blood Cholesterol: Executive Summary: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. Circulation. 2019;139(25):e1046–e81. Epub 20181110. doi: 10.1161/CIR.0000000000000624. PubMed PMID: 30565953.
    OpenUrlCrossRefPubMed
  22. 22.↵
    American Diabetes A. Diagnosis and classification of diabetes mellitus. Diabetes Care. 2010;33 Suppl 1(Suppl 1):S62–9. doi: 10.2337/dc10-S062. PubMed PMID: 20042775; PMCID: PMC2797383.
    OpenUrlCrossRefPubMed
  23. 23.↵
    Diseases GBD, Injuries C. Global burden of 369 diseases and injuries in 204 countries and territories, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet. 2020;396(10258):1204–22. doi: 10.1016/S0140-6736(20)30925-9. PubMed PMID: 33069326; PMCID: PMC7567026.
    OpenUrlCrossRefPubMed
  24. 24.↵
    Król A, Saint-Pierre P. SemiMarkov: An R Package for Parametric Estimation in Multi-State Semi-Markov Models. Journal of Statistical Software. 2015;66(6):1–16.
    OpenUrl
  25. 25.↵
    Asanjarani A, Liquet B, Nazarathy Y. Estimation of semi-Markov multi-state models: a comparison of the sojourn times and transition intensities approaches. Int J Biostat. 2021;18(1):243–62. Epub 20210106. doi: 10.1515/ijb-2020-0083. PubMed PMID: 35641138.
    OpenUrlCrossRefPubMed
  26. 26.↵
    Tsao CW, Aday AW, Almarzooq ZI, Alonso A, Beaton AZ, Bittencourt MS, Boehme AK, Buxton AE, Carson AP, Commodore-Mensah Y, Elkind MSV, Evenson KR, Eze-Nliam C, Ferguson JF, Generoso G, Ho JE, Kalani R, Khan SS, Kissela BM, Knutson KL, Levine DA, Lewis TT, Liu J, Loop MS, Ma J, Mussolino ME, Navaneethan SD, Perak AM, Poudel R, Rezk-Hanna M, Roth GA, Schroeder EB, Shah SH, Thacker EL, VanWagner LB, Virani SS, Voecks JH, Wang NY, Yaffe K, Martin SS. Heart Disease and Stroke Statistics-2022 Update: A Report From the American Heart Association. Circulation. 2022;145(8):e153–e639. Epub 20220126. doi: 10.1161/CIR.0000000000001052. PubMed PMID: 35078371.
    OpenUrlCrossRefPubMed
  27. 27.↵
    Agency for Healthcare Research and Quality. Medical Expenditure Panel Survey (MEPS): household component summary tables: medical conditions, United States. Accessed August 30, 2022.
  28. 28.↵
    Colditz GA, Philpott SE, Hankinson SE. The Impact of the Nurses’ Health Study on Population Health: Prevention, Translation, and Control. Am J Public Health. 2016;106(9):1540–5. Epub 20160726. doi: 10.2105/AJPH.2016.303343. PubMed PMID: 27459441; PMCID: PMC4981811.
    OpenUrlCrossRefPubMed
  29. 29.
    Mahmood SS, Levy D, Vasan RS, Wang TJ. The Framingham Heart Study and the epidemiology of cardiovascular disease: a historical perspective. Lancet. 2014;383(9921):999-1008. Epub 20130929. doi: 10.1016/S0140-6736(13)61752-3. PubMed PMID: 24084292; PMCID: PMC4159698.
    OpenUrlCrossRefPubMed
  30. 30.
    Wright JD, Folsom AR, Coresh J, Sharrett AR, Couper D, Wagenknecht LE, Mosley TH, Jr., Ballantyne CM, Boerwinkle EA, Rosamond WD, Heiss G. The ARIC (Atherosclerosis Risk In Communities) Study: JACC Focus Seminar 3/8. J Am Coll Cardiol. 2021;77(23):2939–59. doi: 10.1016/j.jacc.2021.04.035. PubMed PMID: 34112321; PMCID: PMC8667593.
    OpenUrlCrossRefPubMed
  31. 31.
    Blaha MJ, DeFilippis AP. Multi-Ethnic Study of Atherosclerosis (MESA): JACC Focus Seminar 5/8. J Am Coll Cardiol. 2021;77(25):3195–216. doi: 10.1016/j.jacc.2021.05.006. PubMed PMID: 34167645; PMCID: PMC8091185.
    OpenUrlCrossRefPubMed
  32. 32.↵
    Lloyd-Jones DM, Lewis CE, Schreiner PJ, Shikany JM, Sidney S, Reis JP. The Coronary Artery Risk Development In Young Adults (CARDIA) Study: JACC Focus Seminar 8/8. J Am Coll Cardiol. 2021;78(3):260–77. doi: 10.1016/j.jacc.2021.05.022. PubMed PMID: 34266580; PMCID: PMC8285563.
    OpenUrlCrossRefPubMed
  33. 33.↵
    Virani SS, Newby LK, Arnold SV, Bittner V, Brewer LC, Demeter SH, Dixon DL, Fearon WF, Hess B, Johnson HM, Kazi DS, Kolte D, Kumbhani DJ, LoFaso J, Mahtta D, Mark DB, Minissian M, Navar AM, Patel AR, Piano MR, Rodriguez F, Talbot AW, Taqueti VR, Thomas RJ, van Diepen S, Wiggins B, Williams MS. 2023 AHA/ACC/ACCP/ASPC/NLA/PCNA Guideline for the Management of Patients With Chronic Coronary Disease: A Report of the American Heart Association/American College of Cardiology Joint Committee on Clinical Practice Guidelines. Circulation. 2023;148(9):e9-e119. Epub 20230720. doi: 10.1161/CIR.0000000000001168. PubMed PMID: 37471501.
    OpenUrlCrossRefPubMed
  34. 34.
    Arnett DK, Blumenthal RS, Albert MA, Buroker AB, Goldberger ZD, Hahn EJ, Himmelfarb CD, Khera A, Lloyd-Jones D, McEvoy JW, Michos ED, Miedema MD, Munoz D, Smith SC, Jr., Virani SS, Williams KA, Sr., Yeboah J, Ziaeian B. 2019 ACC/AHA Guideline on the Primary Prevention of Cardiovascular Disease: Executive Summary: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. Circulation. 2019;140(11):e563–e95. Epub 20190317. doi: 10.1161/CIR.0000000000000677. PubMed PMID: 30879339; PMCID: PMC8351755.
    OpenUrlCrossRefPubMed
  35. 35.
    Whelton PK, Carey RM, Aronow WS, Casey DE, Jr., Collins KJ, Dennison Himmelfarb C, DePalma SM, Gidding S, Jamerson KA, Jones DW, MacLaughlin EJ, Muntner P, Ovbiagele B, Smith SC, Jr., Spencer CC, Stafford RS, Taler SJ, Thomas RJ, Williams KA, Sr., Williamson JD, Wright JT, Jr.. 2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA Guideline for the Prevention, Detection, Evaluation, and Management of High Blood Pressure in Adults: Executive Summary: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. Circulation. 2018;138(17):e426-e83. doi: 10.1161/CIR.0000000000000597. PubMed PMID: 30354655.
    OpenUrlCrossRefPubMed
  36. 36.↵
    Heidenreich PA, Bozkurt B, Aguilar D, Allen LA, Byun JJ, Colvin MM, Deswal A, Drazner MH, Dunlay SM, Evers LR, Fang JC, Fedson SE, Fonarow GC, Hayek SS, Hernandez AF, Khazanie P, Kittleson MM, Lee CS, Link MS, Milano CA, Nnacheta LC, Sandhu AT, Stevenson LW, Vardeny O, Vest AR, Yancy CW. 2022 AHA/ACC/HFSA Guideline for the Management of Heart Failure: Executive Summary: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. J Am Coll Cardiol. 2022;79(17):1757-80. Epub 20220401. doi: 10.1016/j.jacc.2021.12.011. PubMed PMID: 35379504.
    OpenUrlCrossRefPubMed
  37. 37.↵
    Olafiranye O, Zizi F, Brimah P, Jean-Louis G, Makaryus AN, McFarlane S, Ogedegbe G. Management of Hypertension among Patients with Coronary Heart Disease. Int J Hypertens. 2011;2011:653903. Epub 20110713. doi: 10.4061/2011/653903. PubMed PMID: 21785704; PMCID: PMC3139133.
    OpenUrlCrossRefPubMed
  38. 38.↵
    Viau DM, Sala-Mercado JA, Spranger MD, O’Leary DS, Levy PD. The pathophysiology of hypertensive acute heart failure. Heart. 2015;101(23):1861–7. Epub 20150629. doi: 10.1136/heartjnl-2015-307461. PubMed PMID: 26123135.
    OpenUrlAbstract/FREE Full Text
  39. 39.↵
    Han Y, Hu Y, Yu C, Guo Y, Pei P, Yang L, Chen Y, Du H, Sun D, Pang Y, Chen N, Clarke R, Chen J, Chen Z, Li L, Lv J, China Kadoorie Biobank Collaborative G. Lifestyle, cardiometabolic disease, and multimorbidity in a prospective Chinese study. Eur Heart J. 2021;42(34):3374–84. doi: 10.1093/eurheartj/ehab413. PubMed PMID: 34333624; PMCID: PMC8423468.
    OpenUrlCrossRefPubMed
  40. 40.↵
    Chen G, Qian ZM, Zhang J, Zhang S, Zhang Z, Vaughn MG, Aaron HE, Wang C, Lip GY, Lin H. Regular use of fish oil supplements and course of cardiovascular diseases: prospective cohort study. BMJ Med. 2024;3(1):e000451. Epub 20240521. doi: 10.1136/bmjmed-2022-000451. PubMed PMID: 38800667; PMCID: PMC11116879.
    OpenUrlAbstract/FREE Full Text
  41. 41.↵
    Lloyd-Jones DM, Hong Y, Labarthe D, Mozaffarian D, Appel LJ, Van Horn L, Greenlund K, Daniels S, Nichol G, Tomaselli GF, Arnett DK, Fonarow GC, Ho PM, Lauer MS, Masoudi FA, Robertson RM, Roger V, Schwamm LH, Sorlie P, Yancy CW, Rosamond WD, American Heart Association Strategic Planning Task F, Statistics C. Defining and setting national goals for cardiovascular health promotion and disease reduction: the American Heart Association’s strategic Impact Goal through 2020 and beyond. Circulation. 2010;121(4):586-613. Epub 20100120. doi: 10.1161/CIRCULATIONAHA.109.192703. PubMed PMID: 20089546.
    OpenUrlAbstract/FREE Full Text
  42. 42.↵
    Collaborators GBDCoD. Global, regional, and national age-sex specific mortality for 264 causes of death, 1980-2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet. 2017;390(10100):1151–210. Epub 2017/09/19. doi: 10.1016/S0140-6736(17)32152-9. PubMed PMID: 28919116; PMCID: PMC5605883.
    OpenUrlCrossRefPubMed
  43. 43.↵
    Disease GBD, Injury I, Prevalence C. Global, regional, and national incidence, prevalence, and years lived with disability for 354 diseases and injuries for 195 countries and territories, 1990-2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet. 2018;392(10159):1789–858. Epub 2018/11/30. doi: 10.1016/S0140-6736(18)32279-7. PubMed PMID: 30496104; PMCID: PMC6227754.
    OpenUrlCrossRefPubMed
  44. 44.↵
    Ding M, Fitzmaurice GM, Arvizu M, Willett WC, Manson JE, Rexrode KM, Hu FB, Chavarro JE. Associations between patterns of modifiable risk factors in mid-life to late life and longevity: 36 year prospective cohort study. BMJ Med. 2022;1(1):e000098. Epub 20220926. doi: 10.1136/bmjmed-2021-000098. PubMed PMID: 36936601; PMCID: PMC9978682.
    OpenUrlAbstract/FREE Full Text
  45. 45.↵
    Cholesterol Treatment Trialists C. Efficacy and safety of statin therapy in older people: a meta-analysis of individual participant data from 28 randomised controlled trials. Lancet. 2019;393(10170):407–15. doi: 10.1016/S0140-6736(18)31942-1. PubMed PMID: 30712900; PMCID: PMC6429627.
    OpenUrlCrossRefPubMed
  46. 46.↵
    Stout KK, Broberg CS, Book WM, Cecchin F, Chen JM, Dimopoulos K, Everitt MD, Gatzoulis M, Harris L, Hsu DT, Kuvin JT, Law Y, Martin CM, Murphy AM, Ross HJ, Singh G, Spray TL, American Heart Association Council on Clinical Cardiology CoFG, Translational B, Council on Cardiovascular R, Imaging. Chronic Heart Failure in Congenital Heart Disease: A Scientific Statement From the American Heart Association. Circulation. 2016;133(8):770–801. Epub 20160119. doi: 10.1161/CIR.0000000000000352. PubMed PMID: 26787728.
    OpenUrlFREE Full Text
  47. 47.↵
    Savarese G, Lund LH. Global Public Health Burden of Heart Failure. Card Fail Rev. 2017;3(1):7–11. doi: 10.15420/cfr.2016:25:2. PubMed PMID: 28785469; PMCID: PMC5494150.
    OpenUrlCrossRefPubMed
  48. 48.↵
    Gallagher J, Watson C, Campbell P, Ledwidge M, McDonald K. Natriuretic Peptide-based Screening and Prevention of Heart Failure. Card Fail Rev. 2017;3(2):83–5. doi: 10.15420/cfr.2017:20:1. PubMed PMID: 29387458; PMCID: PMC5789219.
    OpenUrlCrossRefPubMed
  49. 49.
    Velagaleti RS, Gona P, Larson MG, Wang TJ, Levy D, Benjamin EJ, Selhub J, Jacques PF, Meigs JB, Tofler GH, Vasan RS. Multimarker approach for the prediction of heart failure incidence in the community. Circulation. 2010;122(17):1700–6. Epub 20101011. doi: 10.1161/CIRCULATIONAHA.109.929661. PubMed PMID: 20937976; PMCID: PMC2993157.
    OpenUrlAbstract/FREE Full Text
  50. 50.
    Wells QS, Gupta DK, Smith JG, Collins SP, Storrow AB, Ferguson J, Smith ML, Pulley JM, Collier S, Wang X, Roden DM, Gerszten RE, Wang TJ. Accelerating Biomarker Discovery Through Electronic Health Records, Automated Biobanking, and Proteomics. J Am Coll Cardiol. 2019;73(17):2195–205. doi: 10.1016/j.jacc.2019.01.074. PubMed PMID: 31047008; PMCID: PMC6501811.
    OpenUrlFREE Full Text
  51. 51.↵
    Segar MW, Khan MS, Patel KV, Vaduganathan M, Kannan V, Willett D, Peterson E, Tang WHW, Butler J, Everett BM, Fonarow GC, Wang TJ, McGuire DK, Pandey A. Incorporation of natriuretic peptides with clinical risk scores to predict heart failure among individuals with dysglycaemia. Eur J Heart Fail. 2022;24(1):169–80. Epub 20211123. doi: 10.1002/ejhf.2375. PubMed PMID: 34730265; PMCID: PMC10535364.
    OpenUrlCrossRefPubMed
  52. 52.↵
    Obokata M, Reddy YNV, Borlaug BA. The Role of Echocardiography in Heart Failure with Preserved Ejection Fraction: What Do We Want from Imaging? Heart Fail Clin. 2019;15(2):241–56. Epub 20190202. doi: 10.1016/j.hfc.2018.12.004. PubMed PMID: 30832815; PMCID: PMC6402610.
    OpenUrlCrossRefPubMed
  53. 53.↵
    Virani SS, Alonso A, Aparicio HJ, Benjamin EJ, Bittencourt MS, Callaway CW, Carson AP, Chamberlain AM, Cheng S, Delling FN, Elkind MSV, Evenson KR, Ferguson JF, Gupta DK, Khan SS, Kissela BM, Knutson KL, Lee CD, Lewis TT, Liu J, Loop MS, Lutsey PL, Ma J, Mackey J, Martin SS, Matchar DB, Mussolino ME, Navaneethan SD, Perak AM, Roth GA, Samad Z, Satou GM, Schroeder EB, Shah SH, Shay CM, Stokes A, VanWagner LB, Wang NY, Tsao CW, American Heart Association Council on E, Prevention Statistics C, Stroke Statistics S. Heart Disease and Stroke Statistics-2021 Update: A Report From the American Heart Association. Circulation. 2021;143(8):e254–e743. Epub 2021/01/28. doi: 10.1161/CIR.0000000000000950. PubMed PMID: 33501848.
    OpenUrlCrossRefPubMed
  54. 54.↵
    Bots SH, Peters SAE, Woodward M. Sex differences in coronary heart disease and stroke mortality: a global assessment of the effect of ageing between 1980 and 2010. BMJ Glob Health. 2017;2(2):e000298. Epub 2017/06/08. doi: 10.1136/bmjgh-2017-000298. PubMed PMID: 28589033; PMCID: PMC5435266.
    OpenUrlAbstract/FREE Full Text
  55. 55.↵
    Flegal KM. Body mass index of healthy men compared with healthy women in the United States. Int J Obes (Lond). 2006;30(2):374–9. Epub 2005/09/29. doi: 10.1038/sj.ijo.0803117. PubMed PMID: 16189499.
    OpenUrlCrossRefPubMedWeb of Science
  56. 56.
    Azevedo MR, Araujo CL, Reichert FF, Siqueira FV, da Silva MC, Hallal PC. Gender differences in leisure-time physical activity. International journal of public health. 2007;52(1):8–15. Epub 2007/10/31. PubMed PMID: 17966815; PMCID: 2778720.
    OpenUrlCrossRefPubMedWeb of Science
  57. 57.↵
    Peters SA, Huxley RR, Woodward M. Do smoking habits differ between women and men in contemporary Western populations? Evidence from half a million people in the UK Biobank study. BMJ open. 2014;4(12):e005663. Epub 2015/01/01. doi: 10.1136/bmjopen-2014-005663. PubMed PMID: 25550291; PMCID: 4281541.
    OpenUrlAbstract/FREE Full Text
  58. 58.↵
    Dickman SL, Gaffney A, McGregor A, Himmelstein DU, McCormick D, Bor DH, Woolhandler S. Trends in Health Care Use Among Black and White Persons in the US, 1963-2019. JAMA Netw Open. 2022;5(6):e2217383. Epub 20220601. doi: 10.1001/jamanetworkopen.2022.17383. PubMed PMID: 35699954; PMCID: PMC9198752.
    OpenUrlCrossRefPubMed
  59. 59.↵
    Wallace J, Jiang K, Goldsmith-Pinkham P, Song Z. Changes in Racial and Ethnic Disparities in Access to Care and Health Among US Adults at Age 65 Years. JAMA Intern Med. 2021;181(9):1207–15. doi: 10.1001/jamainternmed.2021.3922. PubMed PMID: 34309621; PMCID: PMC8314180.
    OpenUrlCrossRefPubMed
Back to top
Posted September 19, 2024.
Download PDF
Data/Code
Email
Summary Estimates Derived from a Multi-state Non-Markov Framework to Characterize the Course of Heart Disease
Ming Ding, Feng-Chang Lin, Michelle L. Meyer
medRxiv 2024.09.18.24313882; doi: https://doi.org/10.1101/2024.09.18.24313882
Twitter logo Facebook logo LinkedIn logo Mendeley logo
Citation Tools

Subject Area