Abstract
Background:
Whether prolonged and excessive alcohol consumption contributes to dilated cardiomyopathy (DCM) remains uncertain. This study aimed to describe the prevalence of alcohol use in patients with DCM and their first-degree relatives (FDRs) and determine if cumulative alcohol exposure associates with DCM/partial DCM or modifies the association of DCM with DCM-relevant rare variants.
Methods:
All probands had DCM; FDRs were classified as with or without DCM or partial DCM. Alcohol exposure was measured with the AUDIT-C questionnaire and years of drinking. Rare variants in 36 DCM genes were classified as pathogenic, likely pathogenic or variants of uncertain significance (P, LP, VUS). Generalized linear mixed models were used to assess the association of DCM/partial DCM with alcohol use among FDRs.
Results:
DCM/partial DCM was found in 21.8% of 1373 FDRs of 1148 DCM probands. Prevalence of former or current alcohol use was 68% for probands and 70% for FDRs. About 30% of probands and 37% of FDRs had positive AUDIT-C scores indicating moderate or heavy drinking. Among FDRs, DCM/partial DCM was associated with the presence of P/LP variants in DCM genes (OR=3.51, 95% CI: 2.33 – 5.29) but not with alcohol exposure. Cumulative alcohol exposure was not found to modify the association between DCM/partial DCM and these variants (p=0.55).
Conclusion:
Alcohol use was frequent among probands and FDRs. This study did not provide evidence supporting an association of cumulative alcohol exposure with DCM/partial DCM or a modifying effect of alcohol use on the association of DCM with DCM-relevant rare variants.
Clinical Trial Registration:
URL: http://www.clinicaltrials.gov. Unique identifier: NCT03037632.
Keywords: Dilated cardiomyopathy, alcohol use, genetics
Introduction
Excessive alcohol intake has been shown to have major health and social impact and is a significant cause of morbidity and mortality worldwide.1 Understanding the association of alcohol consumption with cardiovascular disease is complex and challenging. Meta-analyses have demonstrated an association between higher levels of alcohol intake and greater risk for developing atrial fibrillation,2 stroke and heart failure,3, 4 although increased risk of heart failure was not observed in the Framingham study5 or a more recent Mendelian randomization study conducted in Sweden.6 Whether alcohol use is associated with risk of dilated cardiomyopathy (DCM), a condition characterized by left ventricular dilatation and systolic dysfunction, has not been characterized in large representative populations. A 2016 scientific statement indicated alcohol as a leading cause of DCM and mentioned genetic susceptibility, but the data supporting this are inconclusive.7
Understanding the role of alcohol in DCM etiology is crucial for clinical practice and for investigating the causality of DCM. The multisite DCM Consortium Study recruited a large number of patients with DCM to examine its rare variant genetic architecture in a geographically representative, multi-racial and ethnic cohort while also collecting key related medical and social information.8–10 This report aimed to: 1) describe the epidemiologic profile of alcohol use among patients with DCM and their FDRs; 2) assess the association between cumulative alcohol exposure and prevalent DCM/partial DCM among FDRs while controlling for potential confounding factors such as genetic risk, demographics, and comorbidities; and 3) examine whether cumulative alcohol exposure modifies the association of DCM-related rare variants with prevalent DCM/partial DCM among FDRs.
Methods
The DCM Precision Medicine Study
Access to the data from this analysis can be made upon reasonable request. Genetic data from the DCM Precision Medicine Study are available at dbGaP (the database of Genotypes and Phenotypes) and can be accessed at www.ncbi.nlm.nih.gov/gap under accession phs002641. Complete methods can be found in Supplemental Material. The Institutional Review Boards (IRB) at the Ohio State University (OSU) and all clinical sites approved the initial study, followed by single IRB oversight at the University of Pennsylvania. All participants gave written informed consent.9
Results
A total of 1148 probands with DCM and 1373 of their FDRs were included in the analysis. Among FDRs, 299 (21.8%) had DCM/partial DCM (133 DCM and 166 partial DCM); of them, 165 (55%) were identified through the study clinical screening. Table 1 shows the distributions of demographic characteristics and lifestyle factors of probands and their FDRs with and without DCM/partial DCM. Probands and FDRs with DCM or partial DCM tended to be older (51.8 years and 49.6 years), compared to FDRs without DCM or partial DCM (44.7 years). The majority in the 3 groups were White participants, followed by African American participants. Most of the participants in the 3 groups had more than 12 years of education and were married or living together with a partner. The percentage of patients who were current smokers or who quit were higher among probands and FDRs with DCM or partial DCM, compared with FDRs without DCM or partial DCM. Most of the participants in the 3 groups never used amphetamines or cocaine. Frequencies with which probands and FDRs harbored variants (P/LP or VUS) in each gene are shown in Table S1.
Table 1.
Social-demographic characteristics of probands and their FDRs
| Variable | Probands (N = 1148) |
FDRs with DCM/partial DCM (N = 299) |
FDRs without DCM/partial DCM (N = 1074) |
|---|---|---|---|
| Age at enrollment, Mean (SD) | 51.8 (13.7) | 49.6 (15.9) | 44.7 (17.1) |
| Female, N (%) | 510 (44.4) | 180 (60.2) | 667 (62.1) |
| Race * , N (%) | |||
| Black | 498 (43.4) | 102 (34.1) | 299 (27.9) |
| White | 648 (56.4) | 195 (65.2) | 752 (70.1) |
| Other | 2 (0.2) | 2 (0.7) | 21 (2.0) |
| Missing | 0 | 0 | 2 |
| Hispanic, Yes, N (%) | 99 (8.6) | 28 (9.4) | 110 (10.3) |
| Missing | 0 | 2 | 1 |
| Education, N (%) | |||
| ≤12 yrs | 421 (38.8) | 111 (39.8) | 370 (35.7) |
| >12 yrs | 665 (61.2) | 168 (60.2) | 666 (64.3) |
| Missing | 62 | 20 | 38 |
| Marital status, N (%) | |||
| Married or live together | 657 (59.7) | 160 (55.4) | 557 (53.6) |
| Divorced/widowed/separated/ never married | 443 (40.3) | 129 (44.6) | 483 (46.4) |
| Missing | 48 | 10 | 34 |
| Tobacco use, N (%) | |||
| Never | 693 (60.5) | 202 (68.2) | 757 (70.8) |
| Quit | 389 (34.0) | 65 (22.0) | 199 (18.6) |
| Current | 63 (5.5) | 29 (9.8) | 113 (10.6) |
| Missing | 3 | 3 | 5 |
| Amphetamine use, N (%) | |||
| Never | 1,086 (94.7) | 291 (97.3) | 1,026 (95.9) |
| Quit | 41 (3.6) | 7 (2.3) | 20 (1.9) |
| Current | 20 (1.7) | 1 (0.3) | 24 (2.2) |
| Missing | 1 | 0 | 4 |
| Cocaine use, N (%) | |||
| Never | 1,071 (93.5) | 288 (96.3) | 1,034 (96.4) |
| Quit | 67 (5.8) | 8 (2.7) | 32 (3.0) |
| Current | 8 (0.7) | 3 (1.0) | 7 (0.7) |
| Missing | 2 | 0 | 1 |
| Residential location, N (%) | |||
| Northeast | 136 (11.8) | 32 (10.7) | 108 (10.1) |
| Midwest | 363 (31.6) | 111 (37.1) | 397 (37.0) |
| South | 476 (41.5) | 98 (32.8) | 382 (35.6) |
| West | 173 (15.1) | 58 (19.4) | 185 (17.3) |
| Missing | 0 | 0 | 2 |
Other race group includes 5 Native American or Alaskan Native, and 20 participants with more than 1 race.
Abbreviations: DCM, dilated cardiomyopathy; FDR, first-degree relative; N, number.
Table 2 shows the prevalence and patterns of alcohol use in these groups. Because the DCM Precision Medicine Study did not exclude DCM patients based on high alcohol consumption, more than two-thirds of probands were current or former alcohol users, and 30.1% had a positive AUDIT-C score. The prevalence of current alcohol use was 45.6% for probands, 53.8% for FDRs with DCM or partial DCM, and 65.8% for FDRs without DCM or partial DCM. Among probands, about 23% reported quitting drinking at the time of interview; 3.7% reported consuming 7 drinks or more per occasion; 9.6% reported having more than 5 drinks weekly or daily; the average years of drinking was 25.2. For FDRs with and without DCM/partial DCM, 9.0% and 5.8% of them reported quitting drinking; 2.4% and 2.2% reported consuming 7 drinks or more per occasion; 5.3% and 6.0% reported having more than 5 drinks weekly or daily; the average years of drinking was 23.4 for those with DCM/partial DCM and 20.7 for those without. For probands, the percentage of those with severe AUDIT-C risk score was 7.5; the median high/severe drinking years was 160.0. For FDRs with and without DCM or partial DCM, 3.3% and 4.2% had severe AUDIT-C risk score; the median high/severe drinking years was 154.0 and 102.0, respectively.
Table 2.
Prevalence and patterns of alcohol use among adult probands and their FDRs.
| Alcohol use | Probands (N = 1148) |
FDRs with DCM/Partial DCM (N = 299) |
FDRs without DCM/partial DCM (N = 1074) |
|---|---|---|---|
| Alcohol use status, N (%) | |||
| Never | 366 (31.9) | 111 (37.1) | 305 (28.4) |
| Quit | 259 (22.6) | 27 (9.0) | 62 (5.8) |
| Current | 523 (45.6) | 161 (53.8) | 707 (65.8) |
| How many times per week? N (%) | |||
| Never | 366 (31.9) | 111 (37.1) | 305 (28.4) |
| Monthly or less | 244 (21.3) | 67 (22.4) | 219 (20.4) |
| 2–4 times/month | 250 (21.8) | 46 (15.4) | 242 (22.5) |
| 2–3 times/week | 160 (13.9) | 42 (14.0) | 187 (17.4) |
| 4 or more times/week | 128 (11.1) | 33 (11.0) | 121 (11.3) |
| How many drinks per occasion? N (%) | |||
| Never or <3 drinks | 903 (78.7) | 241 (80.6) | 831 (77.4) |
| 3–4 drinks | 156 (13.6) | 43 (14.4) | 168 (15.6) |
| 5–6 drinks | 46 (4.0) | 8 (2.7) | 51 (4.7) |
| 7–9 drinks | 21 (1.8) | 5 (1.7) | 14 (1.3) |
| 10 or more | 22 (1.9) | 2 (0.7) | 10 (0.9) |
| How often does the participant have >5 drinks? N (%) | |||
| Never | 829 (72.2) | 222 (74.2) | 735 (68.4) |
| Less than monthly | 152 (13.2) | 41 (13.7) | 183 (17.0) |
| Monthly | 57 (5.0) | 20 (6.7) | 92 (8.6) |
| Weekly | 79 (6.9) | 9 (3.0) | 47 (4.4) |
| Daily | 31 (2.7) | 7 (2.3) | 17 (1.6) |
| Average years of drinking * , mean (SD) | 25.2 (14.6) | 23.4 (16.5) | 20.7 (15.2) |
| Positive AUDIT-C score † , N (%) | 345 (30.1) | 96 (32.1) | 406 (37.8) |
| AUDIT-C score level, N (%) | |||
| Low | 803 (69.9) | 203 (67.9) | 668 (62.2) |
| Moderate | 192 (16.7) | 65 (21.7) | 280 (26.1) |
| High | 67 (5.8) | 21 (7.0) | 81 (7.5) |
| Severe | 86 (7.5) | 10 (3.3) | 45 (4.2) |
| Moderate/High/Severe drinking years, Median (IQR) § | 120.0 (125.0) | 100.0 (128.5) | 79.0 (95.0) |
| High/Severe drinking years, Median (IQR) § | 160.0 (150.0) | 154.0 (177.0) | 102.0 (141.8) |
Excluded those who responded ‘No’ to alcohol use question. There are 21 missing values for probands, 11 missing values for FDRs with DCM/partial DCM, and 20 missing values for FDRs without DCM/partial DCM.
A positive AUDIT-C score was defined by a total of AUDIT-C score ≥4 for men and ≥3 for women.
For those with moderate, high, or severe AUDIT-C scores, this variable is the individual’s alcohol drinking years multiplied by his or her AUDIT-C score.
For those with high or severe AUDIT-C scores, this variable is the individual’s alcohol drinking years multiplied by his or her AUDIT-C score.
Abbreviations: AUDIT-C, alcohol use disorder identification test-C; DCM, dilated cardiomyopathy; FDR, first-degree relative; N, number; IQR, interquartile range; SD, standard deviation.
Table 3 shows differences in cumulative alcohol exposure by proband characteristics; differences in alcohol use intensity (AUDIT-C score level) are shown in Table S2. After controlling for site, declining alcohol use intensity with age did not achieve statistical significance, but cumulative alcohol exposure did increase with proband’s age, likely due to its relation to drinking years. Higher alcohol use intensity and cumulative exposure were found in men, White individuals, and less educated individuals as compared with their reference groups. Differences in use intensity or cumulative exposure among the three genetic risk groups were small and did not achieve statistical significance.
Table 3.
Cumulative alcohol exposure among probands by demographic, clinical and genetic characteristics
| Variable | Cumulative alcohol exposure* | P value† | |||||
|---|---|---|---|---|---|---|---|
| Total (N = 1148) |
Low Risk (N = 803) |
1st Quartile (0 – 63) (N = 86) |
2nd Quartile (64 – 120) (N = 94) |
3rd Quartile (125 – 189) (N = 79) |
4th Quartile (192 – 540) (N = 86) |
||
| N (Row %) | |||||||
| Enrollment age, years | 0.003 0.03 (trend +) |
||||||
| <45 | 355 | 239 (67.3) | 59 (16.6) | 37 (10.4) | 15 (4.2) | 5 (1.4) | |
| 45–64 | 575 | 408 (71.0) | 20 (3.5) | 46 (8.0) | 51 (8.9) | 50 (8.7) | |
| ≥65 | 218 | 156 (71.6) | 7 (3.2) | 11 (5.0) | 13 (6.0) | 31 (14.2) | |
| Sex | 0.01 0.002 (means -) |
||||||
| Women | 510 | 390 (76.5) | 42 (8.2) | 39 (7.6) | 25 (4.9) | 14 (2.7) | |
| Men | 638 | 413 (64.7) | 44 (6.9) | 55 (8.6) | 54 (8.5) | 72 (11.3) | |
| Race§ | 0.047 0.004 (means -) |
||||||
| African American | 498 | 377 (75.7) | 33 (6.6) | 32 (6.4) | 30 (6.0) | 26 (5.2) | |
| White | 648 | 424 (65.4) | 53 (8.2) | 62 (9.6) | 49 (7.6) | 60 (9.3) | |
| Hispanic | 0.14 | ||||||
| Yes | 99 | 76 (76.8) | 6 (6.1) | 4 (4.0) | 7 (7.1) | 6 (6.1) | |
| No | 1049 | 727 (69.3) | 80 (7.6) | 90 (8.6) | 72 (6.9) | 80 (7.6) | |
| Marital status | 0.01 0.14 (means +) |
||||||
| Married/live together | 657 | 457 (69.6) | 38 (5.8) | 53 (8.1) | 50 (7.6) | 59 (9.0) | |
| Divorced/widowed/ separated/never married | 443 | 316 (71.3) | 46 (10.4) | 34 (7.7) | 24 (5.4) | 23 (5.2) | |
| Missing | 48 | 30 | 2 | 7 | 5 | 4 | |
| Education | 0.006 0.53 (means +) |
||||||
| ≤12 years | 421 | 293 (69.6) | 43 (10.2) | 25 (5.9) | 24 (5.7) | 36 (8.6) | |
| >12 years | 665 | 467 (70.2) | 40 (6.0) | 63 (9.5) | 50 (7.5) | 45 (6.8) | |
| Missing | 62 | 43 | 3 | 6 | 5 | 5 | |
| Comorbidities | |||||||
| Hypertension | 0.01 0.40 (means +) |
||||||
| Yes | 612 | 437 (71.4) | 41 (6.7) | 38 (6.2) | 41 (6.7) | 55 (9.0) | |
| No | 536 | 366 (68.3) | 45 (8.4) | 56 (10.4) | 38 (7.1) | 31 (5.8) | |
| Medical history | |||||||
| Arrhythmia/conduction system disease | 0.05 0.19 (means +) |
||||||
| Yes | 559 | 392 (70.1) | 37 (6.6) | 42 (7.5) | 31 (5.5) | 57 (10.2) | |
| No | 589 | 411 (69.8) | 49 (8.3) | 52 (8.8) | 48 (8.1) | 29 (4.9) | |
| Stroke | 0.18 | ||||||
| Yes | 32 | 26 (81.2) | 2 (6.2) | 3 (9.4) | 0 (0.0) | 1 (3.1) | |
| No | 1116 | 777 (69.6) | 84 (7.5) | 91 (8.2) | 79 (7.1) | 85 (7.6) | |
| Sudden cardiac death | 0.57 | ||||||
| Yes | 55 | 32 (58.2) | 5 (9.1) | 8 (14.5) | 3 (5.5) | 7 (12.7) | |
| No | 1093 | 771 (70.5) | 81 (7.4) | 86 (7.9) | 76 (7.0) | 79 (7.2) | |
| P/LP/VUS, Any | 0.08 | ||||||
| Yes | 704 | 481 (68.3) | 60 (8.5) | 68 (9.7) | 41 (5.8) | 54 (7.7) | |
| No | 444 | 322 (72.5) | 26 (5.9) | 26 (5.9) | 38 (8.6) | 32 (7.2) | |
| Most deleterious variant | 0.25 | ||||||
| P/LP | 210 | 138 (65.7) | 23 (11.0) | 21 (10.0) | 14 (6.7) | 14 (6.7) | |
| VUS | 494 | 343 (69.4) | 37 (7.5) | 47 (9.5) | 27 (5.5) | 40 (8.1) | |
| Negative | 444 | 322 (72.5) | 26 (5.9) | 26 (5.9) | 38 (8.6) | 32 (7.2) | |
Note: Percentages are row percentages defining the distribution among cumulative alcohol exposure categories for a group having the denominator given in the total column for that row.
Moderate-severe drinking years were obtained by multiplying participants’ AUDIT-C scores by alcohol drinking years among those with moderate to severe AUDIT-C scores. Cumulative alcohol exposure was classified into one of 5 categories: low risk or 1st, 2nd, 3rd, or 4th quartile of moderate-severe drinking years among those with moderate to severe AUDIT-C scores.
Two-sided p-values from a generalized Cochran-Mantel-Haenszel test for correlated data with proband enrollment sites as the primary sampling units. The p-value for the general association statistic, which is designed to detect within-site differences in the distribution of the column variable across rows that are consistent across sites, is always reported and was used as a gatekeeper test. If the null was rejected with this p-value at α = 0.05, the p-value for a second test treating cumulative alcohol exposure category as ordinal and the other variable as nominal (means) or ordinal (trend), as appropriate, is also reported to facilitate interpretation. The direction of the association for the means statistic (for two-row comparisons only) and trend statistic was determined by examining the underlying G statistic and noted next to the p-value. For means statistic p-values for two-row comparisons, the notation has the form (means *), with each * taking the value + (-) to indicate that the first row has a higher (lower) mean cumulative alcohol exposure category score than the second row consistently across sites. For trend statistic p-values, the notation has the form (trend *), with * taking the value + (-) to indicate that categories in lower rows were associated with higher (lower) cumulative alcohol exposure category scores consistently across sites.
Excluded 2 probands with “other” races.
Abbreviations: N, number; P, pathogenic; LP, likely pathogenic; VUS, variant of uncertain significance.
Table 4 and Table S3 show differences in cumulative exposure and use intensity by FDR characteristics. After controlling for site, alcohol use intensity decreased with increasing age, but cumulative alcohol exposure showed an increasing trend with age outside of the low risk group. Both use intensity and cumulative exposure were higher in men and White individuals compared with their reference groups.
Table 4.
Cumulative alcohol exposure among FDRs of DCM patients by demographic, clinical and genetic characteristics
| Variable | Cumulative alcohol exposure* | P value† | |||||
|---|---|---|---|---|---|---|---|
| Total (N = 1373) |
Low Risk (N = 871) |
1st Quartile (3 – 40) (N = 129) |
2nd Quartile (42 – 78) (N = 115) |
3rd Quartile (80 – 140) (N = 132) |
4th Quartile (144 – 480) (N = 126) |
||
| N (Row %) | |||||||
| Enrollment age, years | 0.06 | ||||||
| <45 | 674 | 395 (58.6) | 111 (16.5) | 101 (15.0) | 59 (8.8) | 8 (1.2) | |
| 45–64 | 484 | 317 (65.5) | 14 (2.9) | 13 (2.7) | 67 (13.8) | 73 (15.1) | |
| ≥65 | 215 | 159 (74.0) | 4 (1.9) | 1 (0.5) | 6 (2.8) | 45 (20.9) | |
| Sex | 0.01 <0.001 (means -) |
||||||
| Women | 847 | 573 (67.7) | 72 (8.5) | 81 (9.6) | 81 (9.6) | 40 (4.7) | |
| Men | 526 | 298 (56.7) | 57 (10.8) | 34 (6.5) | 51 (9.7) | 86 (16.3) | |
| Race ‡ | 0.03 0.004 (means -) |
||||||
| African American | 401 | 296 (73.8) | 40 (10.0) | 30 (7.5) | 17 (4.2) | 18 (4.5) | |
| White | 947 | 556 (58.7) | 88 (9.3) | 83 (8.8) | 114 (12.0) | 106 (11.2) | |
| Missing | 2 | 1 | 1 | 0 | 0 | 0 | |
| Hispanic | 0.93 | ||||||
| Yes | 138 | 109 (79.0) | 12 (8.7) | 8 (5.8) | 6 (4.3) | 3 (2.2) | |
| No | 1,232 | 761 (61.8) | 116 (9.4) | 107 (8.7) | 126 (10.2) | 122 (9.9) | |
| Missing | 3 | 1 | 1 | 0 | 0 | 1 | |
| Marital status | 0.009 0.002 (means +) |
||||||
| Married/live together | 717 | 445 (62.1) | 37 (5.2) | 57 (7.9) | 84 (11.7) | 94 (13.1) | |
| Divorced/widowed/ separated/never married | 612 | 393 (64.2) | 89 (14.5) | 54 (8.8) | 45 (7.4) | 31 (5.1) | |
| Missing | 44 | 33 | 3 | 4 | 3 | 1 | |
| Education | 0.02 0.15 (means -) |
||||||
| ≤12 years | 481 | 319 (66.3) | 53 (11.0) | 32 (6.7) | 29 (6.0) | 48 (10.0) | |
| >12 years | 834 | 509 (61.0) | 71 (8.5) | 78 (9.4) | 100 (12.0) | 76 (9.1) | |
| Missing | 58 | 43 | 5 | 5 | 3 | 2 | |
| Comorbidities | |||||||
| Hypertension | 0.04 0.43 (means +) |
||||||
| Yes | 397 | 273 (68.8) | 15 (3.8) | 13 (3.3) | 30 (7.6) | 66 (16.6) | |
| No | 976 | 598 (61.3) | 114 (11.7) | 102 (10.5) | 102 (10.5) | 60 (6.1) | |
| Medical history | |||||||
| Arrhythmia/conduction system disease | 0.13 | ||||||
| Yes | 118 | 70 (59.3) | 6 (5.1) | 7 (5.9) | 13 (11.0) | 22 (18.6) | |
| No | 1,255 | 801 (63.8) | 123 (9.8) | 108 (8.6) | 119 (9.5) | 104 (8.3) | |
| Stroke | 0.21 | ||||||
| Yes | 13 | 9 (69.2) | 0 (0.0) | 2 (15.4) | 0 (0.0) | 2 (15.4) | |
| No | 1,360 | 862 (63.4) | 129 (9.5) | 113 (8.3) | 132 (9.7) | 124 (9.1) | |
| Sudden cardiac death | 0.35 | ||||||
| Yes | 5 | 4 (80.0) | 0 (0.0) | 1 (20.0) | 0 (0.0) | 0 (0.0) | |
| No | 1,368 | 867 (63.4) | 129 (9.4) | 114 (8.3) | 132 (9.6) | 126 (9.2) | |
| P/LP/VUS, Any | 0.61 | ||||||
| Yes | 521 | 328 (63.0) | 52 (10.0) | 48 (9.2) | 44 (8.4) | 49 (9.4) | |
| No | 852 | 543 (63.7) | 77 (9.0) | 67 (7.9) | 88 (10.3) | 77 (9.0) | |
| Most deleterious of proband’s variants harbored | 0.74 | ||||||
| P/LP | 154 | 94 (61.0) | 16 (10.4) | 15 (9.7) | 17 (11.0) | 12 (7.8) | |
| VUS | 367 | 234 (63.8) | 36 (9.8) | 33 (9.0) | 27 (7.4) | 37 (10.1) | |
| Negative | 852 | 543 (63.7) | 77 (9.0) | 67 (7.9) | 88 (10.3) | 77 (9.0) | |
Note: Percentages are row percentages defining the distribution among cumulative alcohol exposure categories for a group having the denominator given in the total column for that row.
Moderate-severe drinking years were obtained by multiplying participants’ AUDIT-C scores by alcohol drinking years among those with moderate to severe AUDIT-C scores. Cumulative alcohol exposure was classified into one of 5 categories: low risk or 1st, 2nd, 3rd, or 4th quartile of moderate-severe drinking years among those with moderate to severe AUDIT-C scores.
Two-sided p-values from a generalized Cochran-Mantel-Haenszel test for correlated data with proband enrollment sites as the primary sampling units. The p-value for the general association statistic, which is designed to detect within-site differences in the distribution of the column variable across rows that are consistent across sites, is always reported and was used as a gatekeeper test. If the null was rejected with this p-value at α = 0.05, the p-value for a second test treating cumulative alcohol exposure category as ordinal and the other variable as nominal (means) or ordinal (trend), as appropriate, is also reported to facilitate interpretation. The direction of the association for the means statistic (for two-row comparisons only) and trend statistic was determined by examining the underlying G statistic and noted next to the p-value. For means statistic p-values for two-row comparisons, the notation has the form (means *), with each * taking the value + (-) to indicate that the first row has a higher (lower) mean cumulative alcohol exposure category score than the second row consistently across sites. For trend statistic p-values, the notation has the form (trend *), with * taking the value + (-) to indicate that categories in lower rows were associated with higher (lower) cumulative alcohol exposure category scores consistently across sites.
Excluded 23 FDRs with other races.
Abbreviations: AUDIT-C, alcohol use disorder identification test-C; FDR, first-degree relative; N, number; P, pathogenic; LP, likely pathogenic; VUS, variant of uncertain significance.
Table 5 shows the adjusted associations between cumulative alcohol exposure and DCM/partial DCM among FDRs while controlling for proband enrollment site, age at enrollment, and various combinations of the presence of DCM-related rare variants, demographic characteristics, and clinical characteristics. With site and age adjustment only (Model 1), we were unable to find evidence supporting an association between cumulative alcohol exposure and DCM/partial DCM (p=0.69). In all models including both cumulative alcohol exposure and genetic risk categories, genetic risk was associated with DCM/partial DCM (p<0.001), with the presence of at least 1 P/LP variant being associated with a more than threefold increase in the odds of DCM/partial DCM. However, we were unable to find evidence to support a possible association between cumulative alcohol exposure and DCM/partial DCM (p≥0.68) in any of these models. We were also unable to find evidence supporting modification of the effect of genetic risk by cumulative alcohol exposure in FDRs for either the P/LP (p=0.55) or VUS (p=0.93) categories. A secondary analysis of FDRs that considered DCM and partial DCM separately yielded the same inferences (Table S4). The case-only conditional logistic regression analyses of probands also did not find evidence supporting modification of the effect of genetic risk by cumulative alcohol exposure for either the P/LP category (p=0.28; Table S5) or the VUS category (Table S5).
Table 5.
Association of cumulative alcohol exposure and genetic risk with prevalent DCM/partial DCM among FDRs
| Variable | OR | 95% CI | P value |
|---|---|---|---|
| Model 1. Controlling for proband enrollment site and enrollment age* | |||
| Cumulative alcohol exposure category | 0.69 | ||
| Low risk | 1.35 | (0.67, 2.74) | 0.40 |
| 1st Quartile (3 – 40) | Reference | - | - |
| 2nd Quartile (42 – 78) | 1.34 | (0.65, 2.76) | 0.43 |
| 3rd Quartile (80 – 140) | 0.95 | (0.40, 2.31) | 0.92 |
| 4th Quartile (144 – 480) | 1.34 | (0.60, 3.01) | 0.48 |
| Model 2. Controlling for proband enrollment site and enrollment age† | |||
| Cumulative alcohol exposure category | 0.68 | ||
| Low risk | 1.39 | (0.66, 2.94) | 0.38 |
| 1st Quartile (3 – 40) | Reference | - | - |
| 2nd Quartile (42 – 78) | 1.34 | (0.61, 2.91) | 0.47 |
| 3rd Quartile (80 – 140) | 0.97 | (0.38, 2.51) | 0.95 |
| 4th Quartile (144 – 480) | 1.42 | (0.59, 3.42) | 0.43 |
| Genetic risk category | <0.001 | ||
| P/LP | 3.59 | (2.39, 5.38) | <0.001 |
| VUS only | 1.40 | (0.95, 2.05) | 0.09 |
| Negative | Reference | - | - |
| Model 3. Controlling for proband enrollment site, enrollment age, and race‡ | |||
| Cumulative alcohol exposure category | 0.77 | ||
| Low risk | 1.38 | (0.67, 2.87) | 0.38 |
| 1st Quartile (3 – 40) | Reference | - | - |
| 2nd Quartile (42 – 78) | 1.36 | (0.62, 2.96) | 0.44 |
| 3rd Quartile (80 – 140) | 1.05 | (0.41, 2.68) | 0.92 |
| 4th Quartile (144 – 480) | 1.54 | (0.64, 3.72) | 0.33 |
| Genetic risk category | <0.001 | ||
| P/LP | 3.90 | (2.55, 5.95) | <0.001 |
| VUS only | 1.41 | (0.96, 2.07) | 0.08 |
| Negative | Reference | - | - |
| Model 4. Controlling for proband enrollment site, enrollment age, race, and comorbidities§ | |||
| Cumulative alcohol exposure category | 0.84 | ||
| Low risk | 1.32 | (0.61, 2.85) | 0.48 |
| 1st Quartile (3 – 40) | Reference | - | - |
| 2nd Quartile (42 – 78) | 1.31 | (0.60, 2.86) | 0.49 |
| 3rd Quartile (80 – 140) | 0.95 | (0.36, 2.53) | 0.92 |
| 4th Quartile (144 – 480) | 1.25 | (0.50, 3.13) | 0.64 |
| Genetic risk category | <0.001 | ||
| P/LP | 3.51 | (2.33, 5.29) | <0.001 |
| VUS only | 1.30 | (0.86, 1.96) | 0.21 |
| Negative | Reference | - | - |
For each model, a moments-based or generalized estimating equation–type generalized linear mixed model with the logit link was fit to the binary outcome of the presence of DCM/partial DCM among FDRs using residual subject-specific pseudolikelihood. The linear predictor included a random effect for proband enrollment site to account for site heterogeneity; inferences are thus conditional on site, and ORs compare randomly sampled FDRs of probands enrolled at the same site. Intra-familial correlation was accounted for within a generalized estimating equations framework by using a working independence correlation matrix and the Morel-Bokossa-Neerchal bias-corrected empirical covariance estimator with sites as independent units. All models included a fixed effect for enrollment age (<45, 45–64, ≥65) because DCM penetrance is age-dependent and alcohol exposure category is related to age through drinking years. For each categorical variable, a Wald p-value for the joint null hypothesis that the odds of DCM/partial DCM were identical across categories was calculated using the chi-square distribution with 4 (cumulative alcohol exposure) or 2 (DCM variant category) degrees of freedom and is reported next to the variable name. ORs comparing individual categories with the reference, their Wald 95% CIs, and two-sided p-values obtained with the standard normal distribution are also reported. Two subjects were removed from models 3 and 4 due to missing race.
Model 1 included the cumulative alcohol exposure category and enrollment age.
Model 2 additionally included the FDR’s measured genetic risk category (negative, VUS, P/LP), defined by the most deleterious of the proband’s P/LP/VUS variants carried.
Model 3 considered adjustment for potential demographic confounders such as FDR race (White, Black, or other), education (≤12 years, >12 years), ethnicity, and sex. Results with adjustment for FDR race only are reported; addition of education (≤12 yrs, >12 yrs), ethnicity, and sex to the model yielded identical inferences to the results presented but reduced the sample size due to missing values in these variables. A model using proband genomic ancestry group (African, Native American, or European) in lieu of FDR race also yielded identical inferences to the results presented. A Wald p-value for the null hypothesis of no interaction between genetic risk category and cumulative alcohol exposure variable was calculated separately for the P/LP (p=0.55) and VUS (p=0.93) categories using the chi-square distribution with 4 degrees of freedom.
Model 4 additionally controlled for the presence of hypertension and arrhythmias/conduction system disease.
Abbreviations: OR = odds ratio; CI = confidence interval; DCM, dilated cardiomyopathy; FDR, first-degree relative; N, number; P, pathogenic; LP, likely pathogenic; VUS, variant of uncertain significance.
Discussion
Analysis of 1148 probands with DCM and 1373 of their FDRs from the multi-racial-ethnic and multi-center DCM Precision Medicine Study found that moderate to severe alcohol use was frequent in both probands and FDRs. Patients with DCM who drank heavily were more likely to be men, White, Hispanic, with an educational attainment of 12 years or less, or have comorbidities such as hypertension and history of arrhythmia and conduction disease. However, we were unable to identify evidence supporting an association between cumulative alcohol exposure and prevalent DCM/partial DCM among FDRs. In contrast, harboring DCM-relevant rare genetic P/LP variants was strongly associated with DCM/partial DCM among FDRs after adjusting for other confounding factors. Data were compatible with this association being unmodified by cumulative alcohol exposure in two separate analyses.
These findings vary from conventional wisdom and historical practice, both long established in the medical and cardiovascular literature, that prolonged and excessive alcohol consumption causes DCM and that risk of DCM can be mitigated through abstinence.11 This wisdom has evolved in part from anecdotal accounts of improved myocardial function upon cessation of alcohol use in patients with substantial exposure to alcohol, although usually with concomitant implementation of medical therapy. The underlying question is of causality – does alcohol exposure itself cause myocardial dysfunction that presents as formally defined DCM? Or does alcohol interact with genetic predisposition to hasten the development of DCM? Alternatively, might alcohol exposure in large amounts cause other organ dysfunction (e.g., chronic liver disease) and thereby facilitate an adverse biologic environment such as hypertension or nutritional deficiencies that in turn affect the health of the myocardium?11 Another possibility is that some alcohol products carry myocardial toxins, as shown from additives to beer such as cobalt or arsenic.11 The data derived from this study, based on rigorous clinical and genetic analyses of probands with DCM and their FDRs, suggest that genetics, not alcohol, more likely underlies the DCM found in patients who drink alcohol excessively. These findings suggest that a genetic evaluation should be undertaken regardless of a history of heavy alcohol use.
Alcohol is a known cellular and organ system toxin, but the mechanisms by which alcohol may damage the heart and cardiovascular system are not yet fully understood.12 Results from a population-based study showed an association between alcohol drinking and subclinical echocardiographic abnormalities in LV structure and function with a statistically significant decrement in LVEF, although the LVEF decrement was less than 2% and still within normal limits.13 An earlier study of extremely heavy alcohol use in 50 men showed concurrent skeletal and cardiac muscle dysfunction, including abnormal histology at biopsy.14 In agreement with the findings presented here, a 2018 study observed no difference in the fraction of patients with rare variants identified in the DCM groups with and without alcohol exposure.15 While the data presented here are compelling, a larger study with greater numbers of probands and relatives with substantial genetic diversity is needed to replicate these results. It is important to note that although the current study showed that excessive alcohol use should not rule out a genetic cause in a patient with DCM, it remains essential to educate DCM patients and their family members regarding the adverse health effects of excessive alcohol use, both for cardiovascular conditions beyond DCM and the many conditions beyond cardiovascular.1
Study Limitations
Alcohol use was assessed using self-report and was susceptible to under-reporting due to social desirability bias, which may result in an underestimated prevalence of alcohol use among probands with DCM and their FDRs. In addition, the original AUDIT-C was designed to measure current alcohol exposure by querying alcohol use in the past year, whereas the questionnaire used in this study queried lifetime alcohol use on average without a specific timeframe to investigate the association between cumulative alcohol exposure and DCM. Further, only about one-third of living FDRs enrolled in the DCM Precision Medicine Study,9 and it is possible that those enrolled into the study were more health-conscious. Such selection also could lead to underestimation of the prevalence of excessive alcohol use among FDRs but is unlikely to bias the association assessment because it would impact FDRs with and without DCM/partial DCM similarly. Because the study was cross-sectional, the associations evaluated were with prevalent disease and do not necessarily reflect the effect of prior cumulative alcohol exposure on risk of incident DCM/partial DCM. For FDRs with prior DCM diagnosis, if their drinking behavior changed to lower intake after the diagnosis, the association between DCM and alcohol use could be underestimated. However, the impact of this, if any, would be minimal as the alcohol assessment in the study was based on lifetime alcohol use on average. Also, while we were unable to find evidence supporting association or effect modification in the context of a large study with a clear genetic signal, we were also unable to find evidence supporting lack of association or effect modification due to the wide confidence intervals. Finally, the association assessment was based on DCM/partial DCM without usual clinical cause,9 and so these findings cannot be generalized to other cardiomyopathies.
Conclusions
One third of patients with DCM and their FDRs with DCM or partial DCM were moderate to severe alcohol users. After accounting for the age-dependent nature of DCM risk, we were unable to identify evidence supporting an association between cumulative alcohol exposure and DCM/partial DCM among FDRs of probands with DCM. However, this study provides strong evidence that harboring DCM-relevant rare variants was associated with DCM/partial DCM. We were unable to find evidence that the association between harboring DCM-relevant rare variants and DCM/partial DCM is modified by excessive alcohol use. While this rigorously conducted study did not provide evidence supporting or ruling out clinically relevant associations between alcohol use and DCM, untangling the variable effects of alcohol is challenging, and it is possible that other relevant factors beyond the available data were not considered.
Supplementary Material
Acknowledgments:
The investigators thank the families with DCM who have participated in this study, without whom this effort would not be possible. The DCM Precision Medicine Study was supported by computational infrastructure provided by The Ohio State University Division of Human Genetics Data Management Platform and the Ohio Supercomputer Center. Drs Kinnamon and Ni had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Sources of Funding:
Research reported in this publication was supported by a parent award from the National Heart, Lung, And Blood Institute of the National Institutes of Health under Award Number R01HL128857 to Dr. Hershberger, which included a supplement from the National Human Genome Research Institute. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Nonstandard Abbreviations and Acronyms.
- AUDIT-C
Alcohol Use Disorder Identification Test-Consumption
- ACC
American College of Cardiology
- ACMG
American College of Medical Genetics
- AHA
American Heart Association
- DCM
Dilated cardiomyopathy
- ECG
Electrocardiogram
- FDR
First degree relative
- GEE
Generalized estimating-equation
- LVE
Left ventricular enlargement
- LVEF
Left ventricular ejection fraction
- LVIDD
Left ventricular internal dimension in diastole
- LVSD
Left ventricular systolic dysfunction
- P
Pathogenic
- LP
Likely pathogenic
- TTE
Transthoracic echocardiogram
- VUS
Variant of uncertain significance
Footnotes
Disclosures: None
References:
- 1.GBD Alcohol Collaborators. Alcohol use and burden for 195 countries and territories, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet. 2018;392:1015–1035. DOI: 10.1016/S0140-6736(18)31310-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Larsson SC, Drca N, Wolk A. Alcohol consumption and risk of atrial fibrillation: a prospective study and dose-response meta-analysis. J Am Coll Cardiol. 2014;64:281–9. DOI: 10.1016/j.jacc.2014.03.048. [DOI] [PubMed] [Google Scholar]
- 3.Wood AM, Kaptoge S, Butterworth AS, Willeit P, Warnakula S, Bolton T, Paige E, Paul DS, Sweeting M, Burgess S, et al. Risk thresholds for alcohol consumption: combined analysis of individual-participant data for 599 912 current drinkers in 83 prospective studies. Lancet. 2018;391:1513–1523. DOI: 10.1016/S0140-6736(18)30134-X. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Larsson SC, Wallin A, Wolk A, Markus HS. Differing association of alcohol consumption with different stroke types: a systematic review and meta-analysis. BMC Med. 2016;14:178. DOI: 10.1186/s12916-016-0721-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Walsh CR, Larson MG, Evans JC, Djousse L, Ellison RC, Vasan RS, Levy D. Alcohol consumption and risk for congestive heart failure in the Framingham Heart Study. Ann Intern Med. 2002;136:181–91. [DOI] [PubMed] [Google Scholar]
- 6.Larsson SC, Burgess S, Mason AM, Michaelsson K. Alcohol Consumption and Cardiovascular Disease: A Mendelian Randomization Study. Circ Genom Precis Med. 2020;13:e002814. DOI: 10.1161/CIRCGEN.119.002814. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Bozkurt B, Colvin M, Cook J, Cooper LT, Deswal A, Fonarow GC, Francis GS, Lenihan D, Lewis EF, McNamara DM, et al. Current Diagnostic and Treatment Strategies for Specific Dilated Cardiomyopathies: A Scientific Statement From the American Heart Association. Circulation. 2016;134:e579–e646. DOI: 10.1161/CIR.0000000000000455. [DOI] [PubMed] [Google Scholar]
- 8.Kinnamon DD, Morales A, Bowen DJ, Burke W, Hershberger RE, for the DCM Consortium. Toward Genetics-Driven Early Intervention in Dilated Cardiomyopathy: Design and Implementation of the DCM Precision Medicine Study. Circ Cardiovasc Genet. 2017;10:e001826. DOI: 10.1161/CIRCGENETICS.117.001826. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Huggins GS, Kinnamon DD, Haas GJ, Jordan E, Hofmeyer M, Kransdorf E, Ewald GA, Morris AA, Owens A, Lowes B, et al. Prevalence and Cumulative Risk of Familial Idiopathic Dilated Cardiomyopathy. JAMA. 2022;327:454–463. DOI: 10.1001/jama.2021.24674. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Jordan E, Kinnamon DD, Haas GJ, Hofmeyer M, Kransdorf E, Ewald GA, Morris AA, Owens A, Lowes B, Stoller D, et al. Genetic Architecture of Dilated Cardiomyopathy in Individuals of African and European Ancestry. JAMA. 2023;330:432–441. DOI: 10.1001/jama.2023.11970. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Maisch B. Alcoholic cardiomyopathy : The result of dosage and individual predisposition. Herz. 2016;41:484–93. DOI: 10.1007/s00059-016-4469-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Andersson C, Schou M, Gustafsson F, Torp-Pedersen C. Alcohol Intake in Patients With Cardiomyopathy and Heart Failure: Consensus and Controversy. Circ Heart Fail. 2022;15:e009459. DOI: 10.1161/CIRCHEARTFAILURE.121.009459. [DOI] [PubMed] [Google Scholar]
- 13.Iakunchykova O, Averina M, Kudryavtsev AV, Wilsgaard T, Soloviev A, Schirmer H, Cook S, Leon DA. Evidence for a Direct Harmful Effect of Alcohol on Myocardial Health: A Large Cross-Sectional Study of Consumption Patterns and Cardiovascular Disease Risk Biomarkers From Northwest Russia, 2015 to 2017. J Am Heart Assoc. 2020;9:e014491. DOI: 10.1161/JAHA.119.014491. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Urbano-Marquez A, Estruch R, Navarro-Lopez F, Grau JM, Mont L, Rubin E. The effects of alcoholism on skeletal and cardiac muscle. N Engl J Med. 1989;320:409–15. DOI: 10.1056/NEJM198902163200701. [DOI] [PubMed] [Google Scholar]
- 15.Ware JS, Amor-Salamanca A, Tayal U, Govind R, Serrano I, Salazar-Mendiguchia J, Garcia-Pinilla JM, Pascual-Figal DA, Nunez J, Guzzo-Merello G, et al. Genetic Etiology for Alcohol-Induced Cardiac Toxicity. J Am Coll Cardiol. 2018;71:2293–2302. DOI: 10.1016/j.jacc.2018.03.462. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Haas GJ, Zareba KM, Ni H, Bello-Pardo E, Huggins GS, Hershberger RE. Validating an Idiopathic Dilated Cardiomyopathy Diagnosis Using Cardiovascular Magnetic Resonance: The Dilated Cardiomyopathy Precision Medicine Study. Circ Heart Fail. 2022;15:e008877. DOI: 10.1161/CIRCHEARTFAILURE.121.008877. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Mauro M, Allen DS, Dauda B, Molina SJ, Neale BM, Lewis ACF. A scoping review of guidelines for the use of race, ethnicity, and ancestry reveals widespread consensus but also points of ongoing disagreement. Am J Hum Genet. 2022;109:2110–2125. DOI: 10.1016/j.ajhg.2022.11.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Brothers KB, Bennett RL, Cho MK. Taking an antiracist posture in scientific publications in human genetics and genomics. Genet Med. 2021;23:1004–1007. DOI: 10.1038/s41436-021-01109-w. [DOI] [PubMed] [Google Scholar]
- 19.Flanagin A, Frey T, Christiansen SL, Committee AMAMoS. Updated Guidance on the Reporting of Race and Ethnicity in Medical and Science Journals. JAMA. 2021;326:621–627. DOI: 10.1001/jama.2021.13304. [DOI] [PubMed] [Google Scholar]
- 20.Mudd-Martin G, Cirino AL, Barcelona V, Fox K, Hudson M, Sun YV, Taylor JY, Cameron VA, American Heart Association Council on G, Precision, et al. Considerations for Cardiovascular Genetic and Genomic Research With Marginalized Racial and Ethnic Groups and Indigenous Peoples: A Scientific Statement From the American Heart Association. Circ Genom Precis Med. 2021;14:e000084. DOI: 10.1161/HCG.0000000000000084. [DOI] [PubMed] [Google Scholar]
- 21.Breathett K, Spatz ES, Kramer DB, Essien UR, Wadhera RK, Peterson PN, Ho PM, Nallamothu BK. The Groundwater of Racial and Ethnic Disparities Research: A Statement From Circulation: Cardiovascular Quality and Outcomes. Circ Cardiovasc Qual Outcomes. 2021;14:e007868. DOI: 10.1161/CIRCOUTCOMES.121.007868. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Bush K, Kivlahan DR, McDonell MB, Fihn SD, Bradley KA. The AUDIT alcohol consumption questions (AUDIT-C): an effective brief screening test for problem drinking. Ambulatory Care Quality Improvement Project (ACQUIP). Alcohol Use Disorders Identification Test. Arch Intern Med. 1998;158:1789–95. DOI: 10.1001/archinte.158.16.1789. [DOI] [PubMed] [Google Scholar]
- 23.Alcohol Use Disorders Identification Test. [cited 2023 October 22, 2023]; Available from: https://www.hepatitis.va.gov/alcohol/treatment/audit-c.asp.
- 24.Campbell CE, Maisto SA. Validity of the AUDIT-C screen for at-risk drinking among students utilizing university primary care. J Am Coll Health. 2018;66:774–782. DOI: 10.1080/07448481.2018.1453514. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.van Gils Y, Franck E, Dierckx E, van Alphen SPJ, Saunders JB, Dom G. Validation of the AUDIT and AUDIT-C for Hazardous Drinking in Community-Dwelling Older Adults. Int J Environ Res Public Health. 2021;18. DOI: 10.3390/ijerph18179266. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Morales A, Kinnamon DD, Jordan E, Platt J, Vatta M, Dorschner MO, Starkey CA, Mead JO, Ai T, Burke W, et al. Variant Interpretation for Dilated Cardiomyopathy: Refinement of the American College of Medical Genetics and Genomics/ClinGen Guidelines for the DCM Precision Medicine Study. Circ Genom Precis Med. 2020;13:e002480. DOI: 10.1161/CIRCGEN.119.002480. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Jordan E, Peterson L, Ai T, Asatryan B, Bronicki L, Brown E, Celeghin R, Edwards M, Fan J, Ingles J, et al. Evidence-Based Assessment of Genes in Dilated Cardiomyopathy. Circulation. 2021;144:7–19. DOI: 10.1161/CIRCULATIONAHA.120.053033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Zhang J, Boos DD. Generalized Cochran-Mantel-Haenszel test statistics for correlated categorical data. Communications in Statistics-Theory and Methods. 1997;26:1813–1837. DOI: 10.1080/03610929708832016. [DOI] [Google Scholar]
- 29.Dobbins TA, Simpson JM. Comparison of tests for categorical data from a stratified cluster randomized trial. Stat Med. 2002;21:3835–46. DOI: 10.1002/sim.1256. [DOI] [PubMed] [Google Scholar]
- 30.Vonesh EF, Generalized Linear and Nonlinear Models for Correlated Data: Theory and Applications Using SAS. 1st ed. 2012, Cary, NC: SAS Institute. [Google Scholar]
- 31.Stroup W, Generalized linear mixed models: modern concepts, methods and applications. 2012, New York: CRC Press. [Google Scholar]
- 32.Morel JG, Bokossa MC, Neerchal NK. Small sample correction for the variance of GEE estimators. Biometrical Journal. 2003;45:395–409. DOI: 10.1002/bimj.200390021. [DOI] [Google Scholar]
- 33.Burkett EL, Hershberger RE. Clinical and genetic issues in familial dilated cardiomyopathy. J Am Coll Cardiol. 2005;45:969–81. DOI: 10.1016/j.jacc.2004.11.066. [DOI] [PubMed] [Google Scholar]
- 34.Piegorsch WW, Weinberg CR, Taylor JA. Non-hierarchical logistic models and case-only designs for assessing susceptibility in population-based case-control studies. Stat Med. 1994;13:153–62. DOI: 10.1002/sim.4780130206. [DOI] [PubMed] [Google Scholar]
- 35.Begg CB, Zhang ZF. Statistical analysis of molecular epidemiology studies employing case-series. Cancer Epidemiol Biomarkers Prev. 1994;3:173–5. [PubMed] [Google Scholar]
- 36.Albert PS, Ratnasinghe D, Tangrea J, Wacholder S. Limitations of the case-only design for identifying gene-environment interactions. Am J Epidemiol. 2001;154:687–93. DOI: 10.1093/aje/154.8.687. [DOI] [PubMed] [Google Scholar]
- 37.Yang QH, Khoury MJ, Sun FZ, Flanders WD. Case only design to measure gene-gene interaction. Epidemiology. 1999;10:167–170. DOI: Doi 10.1097/00001648-199903000-00014. [DOI] [PubMed] [Google Scholar]
- 38.Neuhaus JM, Kalbfleisch JD, Hauck WW. A Comparison of Cluster-Specific and Population-Averaged Approaches for Analyzing Correlated Binary Data. International Statistical Review. 1991;59:25–35. DOI: 10.2307/1403572. [DOI] [Google Scholar]
- 39.Molenberghs G, Verbeke G, Models for discrete longitudinal data. 2005, New York: Springer. [Google Scholar]
- 40.Tuerlinckx F, Rijmen F, Verbeke G, De Boeck P. Statistical inference in generalized linear mixed models: a review. Br J Math Stat Psychol. 2006;59:225–55. DOI: 10.1348/000711005X79857. [DOI] [PubMed] [Google Scholar]
- 41.Begg CB, Gray R. Calculation of Polychotomous Logistic-Regression Parameters Using Individualized Regressions. Biometrika. 1984;71:11–18. DOI: 10.2307/2336391. [DOI] [Google Scholar]
- 42.Glymour MM, Greenland S, Causal Diagrams, in Modern epidemiology. 2008, Wolters Kluwer Health/Lippincott Williams & Wilkins: Philadelphia. p. 183–209. [Google Scholar]
- 43.Greene WH, Econometric analysis. 5th ed. 2003, Upper Saddle River, N.J.: Prentice Hall. [Google Scholar]
- 44.Chamberlain G. Analysis of Covariance with Qualitative Data. Review of Economic Studies. 1980;47:225–238. DOI: 10.2307/2297110. [DOI] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.