Common Non-Cardiovascular Multimorbidity Groupings and Clinical Outcomes in Older Adults with Major Cardiovascular Disease

Abstract

Background:

Among older adults, non-cardiovascular multimorbidity often coexists with cardiovascular disease (CVD) but their clinical significance is uncertain. We identified common non-cardiovascular comorbidity patterns and their association with clinical outcomes in Medicare fee-for-service beneficiaries with acute myocardial infarction (AMI), congestive heart failure (CHF), or atrial fibrillation (AF).

Methods:

Using 2015–2016 Medicare data, we took 1% random sample to create 3 cohorts of beneficiaries diagnosed with AMI (n=24,808), CHF (n=57,285), and AF (n=36,277) prior to 1/1/2016. Within each cohort, we applied latent class analysis to classify beneficiaries based on 9 non-cardiovascular comorbidities (anemia, cancer, chronic kidney disease, chronic lung disease, dementia, depression, diabetes, hypothyroidism, and musculoskeletal disease). Mortality, cardiovascular and non-cardiovascular hospitalizations, and home time lost over a 1-year follow-up period were compared across non-cardiovascular multimorbidity classes.

Results:

Similar non-cardiovascular multimorbidity classes emerged from the 3 CVD cohorts: 1) minimal, 2) depression-lung, 3) CKD-diabetes, and 4) multi-system class. Across CVD cohorts, multi-system class had the highest risk of mortality (hazard ratio [HR], 2.7 to 3.9), cardiovascular hospitalization (HR, 1.6 to 3.3), non-cardiovascular hospitalization (HR, 3.1 to 7.2), and home time lost (rate ratio, 2.7 to 5.4). Among those with AMI, the CKD-diabetes class was more strongly associated with all the adverse outcomes than the depression-lung class. In CHF and AF, differences in risk between the depression-lung and CKD-diabetes classes varied per outcome; and the depression-lung and multi-system classes had double the rates of non-cardiovascular hospitalizations compared to cardiovascular hospitalizations.

Conclusion:

Four non-cardiovascular multimorbidity patterns were found among Medicare beneficiaries with CHF, AMI or AF. Compared to minimal class, the multi-system, CKD-diabetes, and depression-lung classes were associated with worse outcomes. Identification of these classes offers insight into specific segments of the population that may benefit from more than the usual cardiovascular care.

INTRODUCTION

Multimorbidity, defined as having two or more conditions, is associated with increased mortality, hospitalizations, health care costs, decreased functional status, and quality of life.^1–3 More than half of older adults with major cardiovascular disease (CVD), such as ischemic heart disease, heart failure, and atrial fibrillation, also have non-cardiovascular comorbidities;^4,5 thus, they meet the criteria for multimorbidity. Previous studies have investigated the effect of non-cardiovascular disease on individuals with CVD and multimorbidity and found that specific individual^6–10 or dyads¹¹ of non-cardiovascular conditions are linked to a higher risk of mortality and hospitalizations. However, examining a single or dyad of non-cardiovascular conditions at a time offers limited utility because it ignores the fact that age-related chronic conditions occur in aggregate (i.e., 3 or more conditions).⁷ It also limits its ability to inform care since it is impractical to investigate and target all possible combinations of cardiovascular and non-cardiovascular conditions.

The latest research has applied data-driven approaches, such as latent class analysis, to define multimorbidity patterns of patients with CVD and their associated clinical significance. Most of the studies were on congestive heart failure,^12,13 while there were a few studies on acute myocardial infarction¹⁴ and atrial fibrillation.^15,16 However, most of the studies included adults under 65 years of age, and cardiovascular and non-cardiovascular comorbidities were analyzed together; thus, determining the impact of non-cardiovascular multimorbidity was not possible. In addition, similarities or differences between non-cardiovascular multimorbidity patterns across major CVDs have yet to be determined.

The objective of this study was to identify patterns of non-cardiovascular comorbidities in Medicare fee-for-service beneficiaries with either of the three major CVDs, namely, acute myocardial infarction (AMI), congestive heart failure (CHF), and atrial fibrillation (AF) and determine associated clinical significance. We postulated that there would be clinically meaningful clustering of non-cardiovascular multimorbidity that define subgroups with different prognoses among patients with AMI, CHF, and AF, respectively. The classes identified by the latent class analysis and their association with adverse outcomes may inform health systems planning by providing specific subgroups for which development of innovative care models or clinical pathways to improve care may be beneficial.

METHODS

Data sources and study population

This retrospective cohort study using Medicare claims data (January 1, 2015 to December 31, 2016) was approved by the Institutional Review Board at Brigham and Women’s Hospital. A waiver of informed consent was obtained. We analyzed the master beneficiary summary file, chronic condition segment, inpatient, outpatient, skilled nursing facility, home health, carrier, and durable medical equipment claims files from an 8% Medicare random sample dataset. Because our analytical method is computation-intensive, we defined 3 major CVD cohorts by taking a 1% random sample from this dataset that satisfies the following criteria: 1) incident AMI cohort included beneficiaries who were discharged from an acute-care hospital with the primary diagnosis of AMI from January 1, 2015 to December 31, 2015 (cohort entry date); 2) prevalent CHF cohort included beneficiaries diagnosed with CHF anytime prior to January 1, 2016 (cohort entry date); and 3) prevalent AF cohort included beneficiaries diagnosed with AF anytime prior to January 1, 2016 (cohort entry date). The cohorts were not mutually exclusive. Each CVD was defined using the International Classification of Diseases 9^th and 10^th revision diagnosis codes in the Chronic Condition Warehouse algorithm.¹⁷ In each cohort, beneficiaries without continuous fee-for-service enrollment in Medicare part A (inpatient) and B (outpatient enrollment) 365 days prior to the cohort entry date were excluded to ensure measurement of pre-existing comorbidities. The final sample included 24,808 beneficiaries with AMI, 57,285 with CHF, and 36,277 with AF.

Measurement of non-cardiovascular comorbidities and other characteristics

Non-cardiovascular comorbidities included in the study were selected from the CMS chronic condition segment file because validated algorithms for these conditions and dates when these algorithms were first satisfied are readily available to researchers to measure and monitor health status of Medicare beneficiaries. We chose comorbidities that occur in both sexes and are prevalent in no more than 80% of the three CVD cohorts. These included Alzheimer’s disease and related dementia (ADRD), anemia, cancer (colorectal, endometrial, breast, lung, or prostate cancer), chronic kidney disease (CKD), chronic lung disease (asthma or chronic obstructive pulmonary disease), depression, diabetes, hypothyroidism, and musculoskeletal disorders (osteoarthritis, rheumatoid arthritis, osteoporosis, or hip/pelvic fracture). Non-cardiovascular comorbidities were considered present if the algorithm for each condition was satisfied any time before the cohort entry date. Presence of any of the conditions classified under one category of comorbidity was considered as having the said category. We then assessed for history of CVD including AMI, CHF, AF, and transient ischemic attack or stroke. We also measured age, sex, and race from the master beneficiary summary file; and Gagne combined comorbidity index (CCI) to quantify the overall comorbidity burden.^18,19

Outcomes

We assessed all-cause mortality from the master beneficiary summary file and hospitalizations from inpatient claims file over 12 months from the cohort entry date. They were followed until death, disenrollment, or end of 12-month follow-up period. We classified hospitalizations into cardiovascular (International Classification of Diseases 10^th revision diagnosis codes I00-I99 in the primary position) and non-cardiovascular hospitalizations (diagnosis codes other than I00-I99 in the primary position). We calculated the number of days alive and spent out of hospitals (dates obtained from inpatient claims file) and skilled nursing facilities (dates obtained from skilled nursing facility claims file), also known as home time. Home time calculated from administrative claims data is strongly correlated with self-rated health, mobility, depression, difficulty in self-care, and social activity limitations.²⁰ More than 15 days of lost home time is considered clinically meaningful.²⁰

Statistical Analysis

Within each CVD cohort, latent class analysis was performed using poLCA package in R Statistical Software v4.1.1²¹ to classify beneficiaries based on nine non-cardiovascular comorbidities (ADRD, anemia, cancer, CKD, chronic lung disease, depression, diabetes, hypothyroidism, and musculoskeletal disorders). After considering 2 to 7 latent classes, the final number of non-cardiovascular latent classes was the one with the lowest possible Akaike Information Criterion and Bayesian Information Criterion that fulfills other criteria: 1) prevalence of the smallest class of at least 10% of the cohort sample, 2) average posterior probability of class membership of at least 0.7, and 3) clinical interpretation of the classes. Each latent class was named after the two most prevalent chronic conditions that distinguish the class from the others. Conditions which repeatedly emerged as most prevalent across classes were not used for naming to avoid redundancy. We then summarized the demographic and clinical characteristics of beneficiaries by latent classes. To examine their clinical associations, we determined the rates of death and hospitalizations (cardiovascular or non-cardiovascular hospitalizations) across identified classes. For this purpose, we fitted Cox proportional hazards model to calculate the hazard ratio (HR) and 95% confidence intervals (CI) of events across classes after adjusting for age, sex, race, and prevalent CVDs other than the index CVD for each cohort. We also compared the mean home time lost due to death, hospitalizations, or skilled nursing facility stays in 365 days among the latent classes. To account for variable follow-up time, we fitted Poisson regression to estimate the rate ratio (RR) and 95% CI after adjusting for age, sex, race, and prevalent CVDs, with log observation time as an offset.

RESULTS

Characteristics of CVD cohorts

Clinical characteristics of each CVD cohort are shown in Table 1. The mean age of the cohorts was from 77.9 (standard deviation [SD], 8.2) years for the AMI cohort to 79.8 (7.8) years for the AF cohort. Approximately 44.4% to 50.6% were female and 84.2% to 91.1% were White. A majority of the AMI (65.7%) and AF (58.1%) cohorts had concomitant CHF, while 13.1% and 37.7% of the CHF cohort had AMI and AF, respectively. History of transient ischemic attack or stroke was present in approximately 27% of individuals in all cohorts. The most common non-cardiovascular comorbidities in all cohorts were anemia and musculoskeletal disease. The AMI cohort had a higher mean CCI of 5.9 (SD, 3.8) than the CHF cohort (2.0 [3.5]) and the AF cohort (1.7 [3.3]).

Table 1.

Characteristics of Medicare Beneficiaries with Acute Myocardial Infarction, Congestive Heart Failure, and Atrial Fibrillation

Characteristics	AMI	CHF	AF
Sample size	24808	57285	36277
Age, years, mean (SD)	77.9 (8.2)	79.7 (8.0)	79.8 (7.8)
Male, n (%)	12556 (50.6)	25414 (44.4)	18209 (50.2)
Race, n (%)
White	21156 (85.3)	48250 (84.2)	33038 (91.1)
Black	2043 (8.2)	5410 (9.4)	1637 (4.5)
Hispanic	454 (1.8)	1071 (1.9)	306 (0.8)
Asian	471 (1.9)	1065 (1.9)	438 (1.2)
Other	684 (2.8)	1489 (2.6)	858 (2.3)
Combined comorbidity index, mean (SD)	5.9 (3.8)	2.0 (3.5)	1.7 (3.3)
Cardiovascular comorbidities, n (%)
AMI	24808 (100)	7530 (13.1)	3811 (10.5)
AF	7040 (28.4)	21622 (37.7)	36277 (100)
CHF	16287 (65.7)	57285 (100)	21064 (58.1)
Stroke or TIA	6759 (27.2)	15467 (27.0)	9887 (27.3)
Non-cardiovascular comorbidities, n (%)
ADRD	6370 (25.7)	16018 (28.0)	8619 (23.8)
Anemia	18175 (73.3)	45013 (78.6)	26262 (72.4)
Cancer	4570 (18.4)	11786 (20.6)	7669 (21.1)
CKD	16420 (66.2)	33912 (59.2)	18801 (51.8)
Chronic lung disease	12508 (50.4)	30243 (52.8)	16398 (45.2)
Depression	10494 (42.3)	26516 (46.3)	14187 (39.1)
Diabetes	14619 (58.9)	32306 (56.4)	16982 (46.8)
Hypothyroidism	8282 (33.4)	22586 (39.4)	13432 (37.0)
Musculoskeletal disease	18053 (72.8)	45334 (79.1)	27819 (76.7)

Abbreviations: AMI, acute myocardial infarction; ADRD, Alzheimer’s disease or related dementia; AF, atrial fibrillation; CHF, congestive heart failure; CKD, chronic kidney disease; SD, standard deviation; TIA, transient ischemic attack.

Latent class analysis of non-cardiovascular comorbidities

For each CVD cohort, four classes were chosen as the final model. The four latent class model had the lowest AIC and BIC and fulfilled the minimum required average posterior probability of class membership (0.72 to 0.74) and proportion of the smallest class (11.9% to 19.0%) (Table 2). The resulting classes were similar across all CVD cohorts, with the prevalence of each class ranging from 11.9% to 19.0% for the minimal class, 18.0% to 21.3% for depression-lung class, 31.8% to 35.4% for the CKD-diabetes class, and 27.9% to 34.8% for the multi-system class (Supplementary Tables S1, S2, and S3).

Table 2.

Latent Class Analysis Results of Non-Cardiovascular Comorbidities

Number of classes	AIC	BIC	Smallest class size (%)	Posterior probability, mean (SD)
AMI
2 Classes	263179.1	263333.4	38.2	0.88 (0.14)
3 Classes	261205.8	261441.3	25.4	0.80 (0.17)
4 Classes	260286.1	260602.7	13.5	0.74 (0.17)
5 Classes	259981.0	260378.8	14.8	0.71 (0.19)
6 Classes	259886.5	260365.5	13.1	0.64 (0.19)
7 Classes	259807.7	260367.9	8.6	0.62 (0.18)
CHF
2 Classes	615345.3	615515.4	43.1	0.85 (0.15)
3 Classes	612116.0	612375.7	24.3	0.78 (0.17)
4 Classes	610428.7	610034.1	11.9	0.72 (0.16)
5 Classes	609683.6	610122.4	14.7	0.68 (0.18)
6 Classes	609505.7	610034.1	12.6	0.69 (0.18)
7 Classes	609312.6	609930.6	11.1	0.62 (0.16)
AF
2 Classes	389126.1	389287.6	43.8	0.87 (0.14)
3 Classes	387299.1	387545.5	24.0	0.79 (0.17)
4 Classes	385996.1	386327.5	19.0	0.72 (0.16)
5 Classes	385540.5	385957	12.2	0.68 (0.19)
6 Classes	385378.2	385879.7	11.4	0.64 (0.18)
7 Classes	385272.7	385859.1	4.0	0.61 (0.17)

Abbreviations: AIC, Akaike information criterion; AMI, acute myocardial infarction; AF, atrial fibrillation; BIC, Bayesian Information Criterion; CHF, congestive heart failure; SD, standard deviation.

The prevalence of non-cardiovascular comorbidities for each class and CVD cohort is shown in Figure 1. In addition, the minimal class had the lowest prevalence of CVD comorbidities (except for prevalence of MI for the CHF cohort) and the lowest mean CCI and age (Supplementary Tables S1, S2 and S3). Opposite of the minimal class was the multi-system class, which had the highest prevalence of ADRD, CVD comorbidities and the highest mean CCI and age. The CKD-diabetes class had the second highest prevalence of CVDs (except for prevalence of stroke or TIA in CHF and AF cohorts). It also had a higher mean CCI compared to the depression-lung class for the AMI and CHF cohorts. The mean CCI for both classes was the same for the AF cohort. The depression-lung class had the highest proportion of White, second highest mean age, and the second highest prevalence of ADRD.

Figure 1. Non-Cardiovascular Multimorbidity Classes in Medicare Beneficiaries with Acute Myocardial Infarction, Congestive Heart Failure, and Atrial Fibrillation.

Within each cardiovascular disease cohort, Medicare beneficiaries were classified based on the presence of nine non-cardiovascular comorbidities. This figure illustrates the four classes found for each cardiovascular disease cohort and the prevalence (in percentage) of each non-cardiovascular condition for each class. Across cardiovascular diseases, the distribution of non-cardiac comorbidities between the same classes was similar.

Abbreviations: ADRD, Alzheimer’s disease or related dementia; AF, atrial fibrillation; AMI, acute myocardial infarction; CHF, congestive heart failure, CKD, chronic kidney disease; MSK, musculoskeletal.

Clinical outcomes associated with non-cardiovascular multimorbidity classes in each cardiovascular disease cohort

Among the three CVD cohorts, the AMI cohort had the worst outcomes, followed by the CHF and AF cohorts: mortality (AMI 36.5, CHF 12.8, and AF 11.8 per 100 person-years), cardiovascular hospitalizations (60.2, 6.2, and 5.8 per 100 person-years), non-cardiovascular hospitalizations (57.2, 14.7, and 12.0 per 100 person-years), and mean (SD) days at home lost (102.8 [141.5], 34.7 [84.1], and 30.7 [80.5] days) over 1 year (Supplementary Table S4).

Across all CVD cohorts, the minimal class had the lowest rates of mortality, cardiovascular hospitalizations, non-cardiovascular hospitalizations, and days at home lost, whereas the multi-system class had the highest rates (Supplementary Table S4). The adjusted HR (95% CI) comparing the multi-system class with the minimal class (Figure 2) was 2.7 (2.4–3.0) to 3.9 (3.3–4.5) for mortality; 1.6 (1.5–1.7) to 3.3 (2.7–4.1) for cardiovascular hospitalization; and 3.1 (2.8–3.4) to 7.2 (6.1–8.5) for non-cardiovascular hospitalization. The adjusted RR (95% CI) for days at home lost ranged from 2.7 (2.4–2.9) to 5.4 (4.6–6.2).

Figure 2. Non-Cardiovascular Multimorbidity Classes and Associated Outcomes by Cardiovascular Disease.

Overall, the AMI cohort had higher rates of death, cardiovascular and non-cardiovascular hospitalization, and lost home time than the CHF and AF cohorts (see the text and Supplementary Table S4). This figure shows the adjusted risk (hazard ratio [HR] or rate ratio [RR]) of mortality, cardiovascular hospitalization, non-cardiovascular hospitalization, and lost home time associated with each multimorbidity class by cardiovascular disease. HR or RR was adjusted for age, sex, race, and cardiovascular comorbidities (acute myocardial infarction, atrial fibrillation, heart failure, and stroke or transient ischemic attack) other than the index cardiovascular disease. With minimal class as the reference group, our findings showed a significant but variable increase in the risk of poor clinical outcomes for the remaining classes across all cohorts. All other pairwise comparisons were statistically significant except for the following outcomes between the depression-lung and CKD-diabetes classes, as indicated in the figure with an asterisk (*): 1) mortality in the CHF cohort; 2) non-cardiovascular hospitalization in the AF cohort; 3) days at home lost in the CHF and AF cohorts.

Abbreviations: AMI, acute myocardial infarction; AF, atrial fibrillation; CHF, congestive heart failure; HR, hazard ratio; PY, person-years; RR, rate ratio.

The outcomes of the CKD-diabetes and depression-lung classes varied by CVD cohort (Figure 2, Supplementary Table S4). In the AMI cohort, the CKD-diabetes class had worse outcomes than the depression-lung class (p-value <0.001 for all outcomes). In the CHF and AF cohorts, the CKD-diabetes class had higher rates of mortality (except for CHF cohort, p-value = 0.10) and cardiovascular hospitalizations than the depression-lung class (p-value <0.01). For non-cardiovascular hospitalizations, the depression-lung class had similar or higher rates compared to CKD-diabetes (AF p-value = 0.13; CHF p-value <0.01). In addition, rates of non-cardiovascular hospitalization in the depression-lung and multi-system classes were twice the rates of cardiovascular hospitalization. There was no difference in days at home lost between the CKD-diabetes and depression-lung classes for CHF (p-value = 0.94) and AF cohorts (p-value = 0.84).

DISCUSSION

In a nationally representative sample of older adults, we found four classes of non-cardiovascular multimorbidity patterns that were similar across Medicare beneficiaries with AMI, CHF, or AF. Except for the minimal class, the depression-lung, CKD-diabetes, and multi-disease classes were all associated with increased risk of mortality, cardiovascular and non-cardiovascular hospitalization, and lost home time. Across all CVD cohorts, the multi-system class showed the worst clinical outcomes. Among beneficiaries with AMI, the CKD-diabetes class was more strongly associated with all the adverse clinical outcomes than the depression-lung class. For beneficiaries with CHF or AF, differences in risks between the depression-lung and CKD-diabetes classes varied per outcome; and depression-lung and multi-system classes had double the rates of non-cardiovascular hospitalizations compared to cardiovascular hospitalizations.

Data-driven clustering methods, such as latent class analysis, can uncover population patterns that may not be readily apparent by only examining individual comorbidities. Although not fully comparable due to differences in the list of conditions included in our analysis, our results had similarities and differences with other CVD studies that used data-driven clustering methods. Similar to other studies,^13,14,16,22 we identified a multi-system class characterized as having a high burden of comorbidities and associated with the worst outcomes. We also found the CKD-diabetes class, which was not surprising since it is known to share common risk factors with CVD.²³ Majority of previous studies factor in both cardiovascular and non-cardiovascular comorbidities in their analysis; thus, resulting classes are more cardiovascular in nature (e.g. metabolic and ischemic groups).^{13–16,22,24–26} CKD and diabetes were usually classified with the metabolic group, but they often co-occur with other more prevalent comorbidities or characteristics, making their clinical significance less apparent.^13,14,24,25 Only Zheng et al. found a very similar class which was associated with the highest risk of mortality and hospitalizations compared to other classes in their study.²⁶ Unique to our study was the depression-lung class, which had risks of poor outcomes comparable to or sometimes exceeded the CKD-diabetes class. Although lung disease and depression are known to be linked with CVD,^27,28 their impact on CVD outcomes has received less attention. By separating and focusing on non-cardiovascular comorbidities, our results highlight the impact of non-cardiovascular diseases, which tend to be overshadowed by CVD in most studies that include both types of comorbidities in the analysis.

Since we used Medicare data, our results are representative of the older adult population in the United States. Thus far, none of the studies have focused primarily on older adults except for the study by Nakamura et al. on Japanese older adults with CHF.¹³ Moreover, while previous similar studies have focused on one CVD, we conducted the same analysis on three separate CVD cohorts and determined that classes were very similar across cohorts. It can be argued that the overlap in cardiovascular conditions may have influenced the similarity in pattern. However, each cohort was separately created through random selection, which means the overlap is representative of the Medicare population encountered in clinical settings. Coexistent CVD comorbidities were also adjusted for in the outcome analysis. The findings of this study can thus inform care planning for a broader range of older adults with CVDs in the United States.

The use of latent class analysis can inform population health management because it identifies segments of the CVD older adult population that can be targeted for interventions and subsequently monitored for evaluation. For instance, the depression-lung and multi-disease classes constituted 18.0–21.3% and 27.9–34.8% of the cardiovascular cohorts, respectively, representing a significant proportion of the population suitable for further investigation. In CHF and AF cohorts, these classes exhibited double the rates of non-cardiovascular hospitalizations compared to cardiovascular hospitalizations. This finding underscores the limitations of solely optimizing CVD such that despite advancements in CVD treatment, high utilization rates persist. It also suggests that comprehensive or multidisciplinary care models may be needed to address non-cardiovascular diseases such as depression and ADRD, which are more prevalent in depression-lung and multi-system classes (Figure 1) and often overlooked in CVD-focused care compared to conditions like CKD and diabetes.^29–31 The higher prevalence of ADRD in both classes is also a reminder that geriatric syndromes are likely to be present in a substantial proportion of older adults with CVD and may impact outcomes. Designing interventions need to take into consideration the complexities of dementia, polypharmacy, and other geriatric syndromes, which are known to often accompany or result from multimorbidity.^29,32,33

Home time is a patient-centered outcome,²⁰ which has not been studied in previous similar studies. All classes across all CVD cohorts (except minimal class) were associated with an increased risk of home time loss, reflecting the burden of repeated or prolonged hospitalization or institutional stays.³⁴ Evaluating home time provides insight on how to address the needs of these classes. For example, creating care models that involve asking what matters most to patients can promote individualized care and reduce unwanted or burdensome treatment.³⁵ Transition care programs may also be advantageous because multiple transfers to different settings create a cycle of worse outcomes due to disruption in care. Transition care programs can provide the needed continuity and potentially avoid readmissions.³⁶ Moreover, all classes across all CVD cohorts (except minimal class) reached the clinically meaningful threshold of 15 days of home time loss, which is associated with a sharp increase in the risk of mobility impairment, depression, and difficulty in self-care.²⁰ This suggests that these subgroups of older adults have additional unique needs, such as support from physical therapy, occupational therapy, social work, care manager, and mental health, further highlighting the need for multidisciplinary care for older adults with CVD and non-cardiovascular multimorbidity.

Our study has some limitations worth mentioning. First, using claims data could result in under-ascertainment of certain diagnoses (e.g., ADRD), potentially leading to misclassification of the latent classes and inaccuracies in the prevalence of diseases. Second, other common non-cardiovascular conditions (e.g., gastrointestinal diseases) were not included because these were not available in the chronic condition segment file. While adding other non-cardiovascular comorbidities could potentially change the patterns and the associated outcomes, it does not invalidate our results. Finally, our results identified comorbidity patterns and did not determine whether treatment will have an impact on outcomes. Further studies are needed to replicate our findings in an independent cohort and explore if and what care models and treatment options can mitigate the risk associated with the presence of non-cardiovascular multimorbidity patterns.

In conclusion, four patterns of non-cardiovascular multimorbidity were found in Medicare beneficiaries with CHF, AMI, or AF. Compared to minimal class, the multi-system, CKD-diabetes, and depression-lung classes were associated with increased risk of mortality, cardiovascular and non-cardiovascular hospitalizations, and clinically meaningful reduction in home time. Identification of these classes provide valuable insights into segments of the older adult population with major CVD who may require more than the usual cardiovascular care.

Key points:

• Four non-cardiovascular patterns were identified in older adults with major cardiovascular diseases.

• Depression-lung, CKD-diabetes, and multi-system classes were associated with worse outcomes.

Why does it matter?

The findings inform care planning by demonstrating specific subgroups for which non-traditional cardiovascular care models may be beneficial.