WIN and HHI association with PACC/MOCA
semere
2025-01-21
1 Introduction
Research Question Is there an association between cognitive function (PACC and MoCA scores) and: Hearing Handicap Inventory (HHI) scores? Words in Noise (WIN) test results? Does hearing aid use moderate this association, accounting for age, sex, and education? ## Statistical Analysis plan: 1.Descriptive Statistics Perform descriptive statistics for:
Hearing Handicap Inventory (HHI) scores. Words in Noise (WIN) test results (average raw scores and thresholds). Cognitive function (PACC and MoCA scores). Demographics: Age, sex, education, and hearing aid use status. PACC=β0+β1(HHI)+β2(WIN Raw)+β3(HearingAid)+β4(HHI×HearingAid)+β5(WIN Raw×HearingAid)+Covariates+ϵ MOCA=β0+β1(HHI)+β2(WIN Raw)+β3(HearingAid)+β4(HHI×HearingAid)+β5(WIN Raw×HearingAid)+Covariates+ϵ
2 Data visualization
For Word in noise data: There are two extremely low results for WIN threshold values which were further investigated. 100077: the raw scores avg (1.5) are accurate with low score on follow up too. 100117: the raw scores avg (0.5) are also accurate with low follow up score too.
Category of WIN threshold was created by taking those with 10 and lower as normal and above 10 threshold to be impaired for hearing based on the guideline provided by the NIH toolbox.
2.1 Scatter plots for continuous variables
2.2 Histograms for continuous variables
## Histograms: Normal hearing Vs Impaired hearing based on WIN
# Table 1 Baseline Characteristics for the 420 participants who have non
missing data for Words in noise data at baseline.
| Total (N=420) |
Impaired hearing (N=179) |
Normal hearing (N=241) |
P-value | |
|---|---|---|---|---|
| age | ||||
| Mean (SD) | 72.5 (5.15) | 73.9 (5.86) | 71.4 (4.24) | <0.001 |
| Median [Min, Max] | 71.6 [65.1, 94.7] | 73.0 [65.4, 94.7] | 70.5 [65.1, 87.2] | |
| gender | ||||
| Female | 222 (52.9%) | 69 (38.5%) | 153 (63.5%) | <0.001 |
| Male | 198 (47.1%) | 110 (61.5%) | 88 (36.5%) | |
| educ | ||||
| Mean (SD) | 16.3 (2.48) | 16.1 (2.43) | 16.4 (2.51) | 0.248 |
| Median [Min, Max] | 16.0 [10.0, 25.0] | 16.0 [10.0, 24.0] | 16.0 [10.0, 25.0] | |
| Missing | 7 (1.7%) | 4 (2.2%) | 3 (1.2%) | |
| mocatots | ||||
| Mean (SD) | 24.9 (3.40) | 23.9 (3.67) | 25.7 (2.95) | <0.001 |
| Median [Min, Max] | 26.0 [12.0, 30.0] | 24.0 [12.0, 30.0] | 26.0 [15.0, 30.0] | |
| Missing | 85 (20.2%) | 35 (19.6%) | 50 (20.7%) | |
| PACC_BL | ||||
| Mean (SD) | -0.144 (0.758) | -0.350 (0.852) | 0.00901 (0.641) | <0.001 |
| Median [Min, Max] | -0.0570 [-3.14, 1.42] | -0.242 [-3.14, 1.42] | 0.0487 [-2.38, 1.34] | |
| Missing | 107 (25.5%) | 46 (25.7%) | 61 (25.3%) | |
| ADI_NATRANK | ||||
| Mean (SD) | 48.5 (24.8) | 47.6 (22.7) | 49.1 (26.2) | 0.518 |
| Median [Min, Max] | 48.0 [3.00, 100] | 48.0 [4.00, 98.0] | 47.0 [3.00, 100] | |
| Missing | 4 (1.0%) | 2 (1.1%) | 2 (0.8%) | |
| wrat_rawword | ||||
| Mean (SD) | 63.1 (5.75) | 62.4 (5.69) | 63.7 (5.75) | 0.0258 |
| Median [Min, Max] | 64.0 [31.0, 70.0] | 63.0 [40.0, 70.0] | 65.0 [31.0, 70.0] | |
| phq_totscore | ||||
| Mean (SD) | 4.89 (5.07) | 4.97 (5.52) | 4.82 (4.66) | 0.819 |
| Median [Min, Max] | 3.00 [0, 22.0] | 3.00 [0, 22.0] | 4.00 [0, 21.0] | |
| Missing | 158 (37.6%) | 57 (31.8%) | 101 (41.9%) | |
| race | ||||
| Black or African American | 66 (15.7%) | 18 (10.1%) | 48 (19.9%) | 0.00904 |
| White | 354 (84.3%) | 161 (89.9%) | 193 (80.1%) | |
| PACC_class_sick | ||||
| high PACC | 259 (61.7%) | 101 (56.4%) | 158 (65.6%) | 0.00963 |
| low PACC | 54 (12.9%) | 32 (17.9%) | 22 (9.1%) | |
| Missing | 107 (25.5%) | 46 (25.7%) | 61 (25.3%) | |
| noise_censusblock2020_mean | ||||
| Mean (SD) | 52.3 (2.55) | 52.4 (2.48) | 52.3 (2.61) | 0.803 |
| Median [Min, Max] | 52.4 [46.0, 59.3] | 52.6 [46.0, 59.3] | 52.2 [46.2, 58.4] | |
| Missing | 2 (0.5%) | 1 (0.6%) | 1 (0.4%) | |
| Hearing Aid Use | ||||
| No | 322 (76.7%) | 102 (57.0%) | 220 (91.3%) | <0.001 |
| Yes | 69 (16.4%) | 62 (34.6%) | 7 (2.9%) | |
| Missing | 29 (6.9%) | 15 (8.4%) | 14 (5.8%) | |
| MOCA Category | ||||
| high moca | 178 (42.4%) | 57 (31.8%) | 121 (50.2%) | <0.001 |
| low moca | 157 (37.4%) | 87 (48.6%) | 70 (29.0%) | |
| Missing | 85 (20.2%) | 35 (19.6%) | 50 (20.7%) |
3 Bivariate analysis
3.1 t-test results
| Test | Mean_Group_1 | Mean_Group_2 | MeanDiff | CI | t | df | p.value |
|---|---|---|---|---|---|---|---|
| HHI by PACC_class | 13.2509363 | 15.3703704 | -2.12 | [-8.04, 3.8] | -0.72 | 64.50 | 0.476835 |
| WIN by PACC_class | 10.1374517 | 11.9629630 | -1.83 | [-3.09, -0.56] | -2.88 | 70.21 | 0.005343 |
| HHI by MOCA_CAT | 12.1318681 | 15.3416149 | -3.21 | [-6.66, 0.24] | -1.83 | 298.52 | 0.067949 |
| WIN by MOCA_CAT | 9.4719101 | 11.6713376 | -2.20 | [-3.04, -1.36] | -5.18 | 293.45 | < 1e-04 |
| WIN by Race | 8.8606061 | 10.7186441 | -1.86 | [-2.61, -1.1] | -4.87 | 133.47 | < 1e-04 |
| PACC by WIN_CAT | -0.3499298 | 0.0090126 | -0.36 | [-0.53, -0.19] | -4.08 | 235.29 | < 1e-04 |
| MOCA by WIN_CAT | 23.8541667 | 25.6910995 | -1.84 | [-2.57, -1.1] | -4.92 | 268.42 | < 1e-04 |
| PACC by Hearing Aids | 0.0100037 | -0.0518199 | 0.06 | [-0.11, 0.23] | 0.72 | 115.28 | 0.473102 |
| MOCA by Hearing Aids | 25.0594406 | 24.3968254 | 0.66 | [-0.24, 1.56] | 1.46 | 94.57 | 0.146916 |
| WIN by hhi category | 9.4897436 | 13.1333333 | -3.64 | [-4.57, -2.72] | -7.76 | 144.89 | < 1e-04 |
| PACC by_HHI_cat | -0.1311867 | -0.1584278 | 0.03 | [-0.16, 0.22] | 0.29 | 150.71 | 0.775454 |
| WIN by Hearing Aids | 9.4757764 | 14.7478261 | -5.27 | [-6.27, -4.27] | -10.48 | 88.63 | < 1e-04 |
3.2 Correlation analysis between HHI/WIN and PACC:
Significant correlation in the right directions for HHI, WINRAW and WIN threshold with PACC
##
## Pearson's product-moment correlation
##
## data: Drives_WIN_HHI$hhi_total_score and Drives_WIN_HHI$PACC_BL
## t = -2.0301, df = 319, p-value = 0.04318
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## -0.219687809 -0.003509403
## sample estimates:
## cor
## -0.1129349##
## Pearson's product-moment correlation
##
## data: Drives_WIN_HHI$WIN_Threshold_Avg and Drives_WIN_HHI$PACC_BL
## t = -4.5284, df = 311, p-value = 8.477e-06
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## -0.3499268 -0.1417622
## sample estimates:
## cor
## -0.2487143##
## Pearson's product-moment correlation
##
## data: Drives_WIN_HHI$noise_censusblock2020_mean and Drives_WIN_HHI$ADI_NATRANK
## t = -3.2745, df = 584, p-value = 0.001121
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## -0.2129507 -0.0538612
## sample estimates:
## cor
## -0.1342714 Multivariate analysis: MLR
4.1 Multiple linear regression models with PACC score as the outcome
For the following models, demographic variables of age, race, educ, and gender are controlled. ADI is add to model 2 (Rt and Lt side WIN are almost the same, so ADI would be compared if it has any effect), 6 and 7.| Model | term | estimate | std.error | statistic | p.value |
|---|---|---|---|---|---|
| Model 1: WIN_PACC | (Intercept) | 0.883 | 0.650 | 1.36 | 0.175296 |
| Model 1: WIN_PACC | WIN_Threshold_Avg | -0.037 | 0.012 | -3.02 | 0.002755 |
| Model 1: WIN_PACC | age | -0.020 | 0.008 | -2.33 | 0.020474 |
| Model 1: WIN_PACC | educ | 0.025 | 0.017 | 1.46 | 0.145347 |
| Model 1: WIN_PACC | raceWhite | 0.357 | 0.118 | 3.03 | 0.002687 |
| Model 1: WIN_PACC | gender2 | 0.154 | 0.090 | 1.71 | 0.088244 |
| Model 2: RT_WIN_PACC | (Intercept) | 0.894 | 0.723 | 1.24 | 0.216954 |
| Model 2: RT_WIN_PACC | nihtb_words_in_noise_threshold_right_ear | -0.028 | 0.011 | -2.52 | 0.012244 |
| Model 2: RT_WIN_PACC | age | -0.022 | 0.009 | -2.52 | 0.012112 |
| Model 2: RT_WIN_PACC | ADI_NATRANK | 0.000 | 0.002 | -0.06 | 0.952481 |
| Model 2: RT_WIN_PACC | educ | 0.028 | 0.018 | 1.56 | 0.119485 |
| Model 2: RT_WIN_PACC | raceWhite | 0.324 | 0.127 | 2.55 | 0.011190 |
| Model 2: RT_WIN_PACC | gender2 | 0.176 | 0.091 | 1.95 | 0.052619 |
| Model 3: LT_WIN_PACC | (Intercept) | 0.901 | 0.652 | 1.38 | 0.168109 |
| Model 3: LT_WIN_PACC | nihtb_words_in_noise_threshold_left_ear | -0.029 | 0.011 | -2.70 | 0.007268 |
| Model 3: LT_WIN_PACC | age | -0.022 | 0.008 | -2.59 | 0.010139 |
| Model 3: LT_WIN_PACC | educ | 0.027 | 0.017 | 1.58 | 0.114912 |
| Model 3: LT_WIN_PACC | raceWhite | 0.355 | 0.119 | 2.99 | 0.003006 |
| Model 3: LT_WIN_PACC | gender2 | 0.181 | 0.088 | 2.05 | 0.041400 |
| Model 4: HHI_PACC | (Intercept) | 0.944 | 0.642 | 1.47 | 0.142470 |
| Model 4: HHI_PACC | hhi_total_score | -0.005 | 0.003 | -1.81 | 0.071381 |
| Model 4: HHI_PACC | age | -0.027 | 0.008 | -3.37 | 0.000847 |
| Model 4: HHI_PACC | educ | 0.033 | 0.017 | 1.97 | 0.049408 |
| Model 4: HHI_PACC | raceWhite | 0.328 | 0.117 | 2.81 | 0.005255 |
| Model 4: HHI_PACC | gender2 | 0.222 | 0.084 | 2.64 | 0.008799 |
| Model 5: WIN+HHI_PACC | (Intercept) | 0.908 | 0.653 | 1.39 | 0.164948 |
| Model 5: WIN+HHI_PACC | WIN_Threshold_Avg | -0.034 | 0.014 | -2.49 | 0.013173 |
| Model 5: WIN+HHI_PACC | hhi_total_score | -0.002 | 0.003 | -0.51 | 0.608809 |
| Model 5: WIN+HHI_PACC | age | -0.020 | 0.009 | -2.38 | 0.018164 |
| Model 5: WIN+HHI_PACC | raceWhite | 0.366 | 0.119 | 3.06 | 0.002373 |
| Model 5: WIN+HHI_PACC | educ | 0.025 | 0.017 | 1.46 | 0.145582 |
| Model 5: WIN+HHI_PACC | gender2 | 0.153 | 0.090 | 1.69 | 0.091551 |
| Model 6: HA+WIN+HHI_PACC | (Intercept) | 0.997 | 0.647 | 1.54 | 0.124097 |
| Model 6: HA+WIN+HHI_PACC | WIN_Threshold_Avg | -0.049 | 0.014 | -3.41 | 0.000727 |
| Model 6: HA+WIN+HHI_PACC | hhi_total_score | -0.004 | 0.003 | -1.46 | 0.145839 |
| Model 6: HA+WIN+HHI_PACC | hearingaids11 | 0.391 | 0.129 | 3.04 | 0.002575 |
| Model 6: HA+WIN+HHI_PACC | age | -0.020 | 0.009 | -2.38 | 0.018062 |
| Model 6: HA+WIN+HHI_PACC | raceWhite | 0.377 | 0.118 | 3.20 | 0.001532 |
| Model 6: HA+WIN+HHI_PACC | educ | 0.026 | 0.017 | 1.53 | 0.127439 |
| Model 6: HA+WIN+HHI_PACC | gender2 | 0.156 | 0.090 | 1.74 | 0.083055 |
| Model 7: Full (WIN+HHI+…) | (Intercept) | 1.116 | 1.058 | 1.05 | 0.292419 |
| Model 7: Full (WIN+HHI+…) | hhi_total_score | -0.005 | 0.003 | -1.50 | 0.133468 |
| Model 7: Full (WIN+HHI+…) | WIN_Threshold_Avg | -0.048 | 0.014 | -3.32 | 0.001015 |
| Model 7: Full (WIN+HHI+…) | age | -0.020 | 0.009 | -2.25 | 0.025442 |
| Model 7: Full (WIN+HHI+…) | noise_censusblock2020_mean | -0.004 | 0.016 | -0.28 | 0.781158 |
| Model 7: Full (WIN+HHI+…) | raceWhite | 0.387 | 0.128 | 3.02 | 0.002710 |
| Model 7: Full (WIN+HHI+…) | ADI_NATRANK | 0.000 | 0.002 | 0.25 | 0.800933 |
| Model 7: Full (WIN+HHI+…) | hearingaids11 | 0.390 | 0.131 | 2.98 | 0.003147 |
| Model 7: Full (WIN+HHI+…) | educ | 0.028 | 0.018 | 1.57 | 0.117594 |
| Model 7: Full (WIN+HHI+…) | gender2 | 0.159 | 0.091 | 1.74 | 0.082215 |
| Model | r.squared | adj.r.squared | statistic | p.value | df | df.residual | sigma | AIC | BIC |
|---|---|---|---|---|---|---|---|---|---|
| Model 1: WIN_PACC | 0.115 | 0.101 | 8.01 | <1e-04 | 5 | 307 | 0.719 | 689.5 | 715.7 |
| Model 2: RT_WIN_PACC | 0.107 | 0.089 | 6.08 | <1e-04 | 6 | 304 | 0.725 | 691.6 | 721.5 |
| Model 3: LT_WIN_PACC | 0.110 | 0.096 | 7.61 | <1e-04 | 5 | 307 | 0.721 | 691.3 | 717.5 |
| Model 4: HHI_PACC | 0.096 | 0.082 | 6.70 | <1e-04 | 5 | 315 | 0.718 | 706.5 | 732.9 |
| Model 5: WIN+HHI_PACC | 0.116 | 0.099 | 6.70 | <1e-04 | 6 | 306 | 0.720 | 691.2 | 721.2 |
| Model 6: HA+WIN+HHI_PACC | 0.143 | 0.123 | 7.23 | <1e-04 | 7 | 304 | 0.710 | 681.9 | 715.6 |
| Model 7: Full (WIN+HHI+…) | 0.142 | 0.116 | 5.51 | <1e-04 | 9 | 299 | 0.716 | 682.0 | 723.0 |
4.1.1 Check multicollinearity for the last MLR model
## WIN_Threshold_Avg hhi_total_score age race
## 1.704128 1.352889 1.156904 1.081590
## educ gender
## 1.090904 1.2321454.1.2 MLR with hhi, WIN, hearing aid use and PACC interpretation:
After adjusting for noise, age, gender and education: hearing aid use and Threshold WIN are significant. Average Threshold WIN of both ears yielded a higher estimate and lower p value than each ear individually. Thus, average threshold WIN values will be used to model the remaining models. The last model adjusts for noise and demographics.
4.2 Multiple linear regression models with MOCA score as the outcome
MLR with hhi, win, hearing aid use and MOCA
| Model | term | estimate | std.error | statistic | p.value |
|---|---|---|---|---|---|
| Model 1: WIN_PACC | (Intercept) | 0.883 | 0.650 | 1.36 | 0.175296 |
| Model 1: WIN_PACC | WIN_Threshold_Avg | -0.037 | 0.012 | -3.02 | 0.002755 |
| Model 1: WIN_PACC | age | -0.020 | 0.008 | -2.33 | 0.020474 |
| Model 1: WIN_PACC | educ | 0.025 | 0.017 | 1.46 | 0.145347 |
| Model 1: WIN_PACC | raceWhite | 0.357 | 0.118 | 3.03 | 0.002687 |
| Model 1: WIN_PACC | gender2 | 0.154 | 0.090 | 1.71 | 0.088244 |
| Model 2: RT_WIN_PACC | (Intercept) | 0.894 | 0.723 | 1.24 | 0.216954 |
| Model 2: RT_WIN_PACC | nihtb_words_in_noise_threshold_right_ear | -0.028 | 0.011 | -2.52 | 0.012244 |
| Model 2: RT_WIN_PACC | age | -0.022 | 0.009 | -2.52 | 0.012112 |
| Model 2: RT_WIN_PACC | ADI_NATRANK | 0.000 | 0.002 | -0.06 | 0.952481 |
| Model 2: RT_WIN_PACC | educ | 0.028 | 0.018 | 1.56 | 0.119485 |
| Model 2: RT_WIN_PACC | raceWhite | 0.324 | 0.127 | 2.55 | 0.011190 |
| Model 2: RT_WIN_PACC | gender2 | 0.176 | 0.091 | 1.95 | 0.052619 |
| Model 3: LT_WIN_PACC | (Intercept) | 0.901 | 0.652 | 1.38 | 0.168109 |
| Model 3: LT_WIN_PACC | nihtb_words_in_noise_threshold_left_ear | -0.029 | 0.011 | -2.70 | 0.007268 |
| Model 3: LT_WIN_PACC | age | -0.022 | 0.008 | -2.59 | 0.010139 |
| Model 3: LT_WIN_PACC | educ | 0.027 | 0.017 | 1.58 | 0.114912 |
| Model 3: LT_WIN_PACC | raceWhite | 0.355 | 0.119 | 2.99 | 0.003006 |
| Model 3: LT_WIN_PACC | gender2 | 0.181 | 0.088 | 2.05 | 0.041400 |
| Model 4: HHI_PACC | (Intercept) | 0.944 | 0.642 | 1.47 | 0.142470 |
| Model 4: HHI_PACC | hhi_total_score | -0.005 | 0.003 | -1.81 | 0.071381 |
| Model 4: HHI_PACC | age | -0.027 | 0.008 | -3.37 | 0.000847 |
| Model 4: HHI_PACC | educ | 0.033 | 0.017 | 1.97 | 0.049408 |
| Model 4: HHI_PACC | raceWhite | 0.328 | 0.117 | 2.81 | 0.005255 |
| Model 4: HHI_PACC | gender2 | 0.222 | 0.084 | 2.64 | 0.008799 |
| Model 5: WIN+HHI_PACC | (Intercept) | 0.908 | 0.653 | 1.39 | 0.164948 |
| Model 5: WIN+HHI_PACC | WIN_Threshold_Avg | -0.034 | 0.014 | -2.49 | 0.013173 |
| Model 5: WIN+HHI_PACC | hhi_total_score | -0.002 | 0.003 | -0.51 | 0.608809 |
| Model 5: WIN+HHI_PACC | age | -0.020 | 0.009 | -2.38 | 0.018164 |
| Model 5: WIN+HHI_PACC | raceWhite | 0.366 | 0.119 | 3.06 | 0.002373 |
| Model 5: WIN+HHI_PACC | educ | 0.025 | 0.017 | 1.46 | 0.145582 |
| Model 5: WIN+HHI_PACC | gender2 | 0.153 | 0.090 | 1.69 | 0.091551 |
| Model 6: HA+WIN+HHI_PACC | (Intercept) | 0.997 | 0.647 | 1.54 | 0.124097 |
| Model 6: HA+WIN+HHI_PACC | WIN_Threshold_Avg | -0.049 | 0.014 | -3.41 | 0.000727 |
| Model 6: HA+WIN+HHI_PACC | hhi_total_score | -0.004 | 0.003 | -1.46 | 0.145839 |
| Model 6: HA+WIN+HHI_PACC | hearingaids11 | 0.391 | 0.129 | 3.04 | 0.002575 |
| Model 6: HA+WIN+HHI_PACC | age | -0.020 | 0.009 | -2.38 | 0.018062 |
| Model 6: HA+WIN+HHI_PACC | raceWhite | 0.377 | 0.118 | 3.20 | 0.001532 |
| Model 6: HA+WIN+HHI_PACC | educ | 0.026 | 0.017 | 1.53 | 0.127439 |
| Model 6: HA+WIN+HHI_PACC | gender2 | 0.156 | 0.090 | 1.74 | 0.083055 |
| Model 7: Full (WIN+HHI+…) | (Intercept) | 1.116 | 1.058 | 1.05 | 0.292419 |
| Model 7: Full (WIN+HHI+…) | hhi_total_score | -0.005 | 0.003 | -1.50 | 0.133468 |
| Model 7: Full (WIN+HHI+…) | WIN_Threshold_Avg | -0.048 | 0.014 | -3.32 | 0.001015 |
| Model 7: Full (WIN+HHI+…) | age | -0.020 | 0.009 | -2.25 | 0.025442 |
| Model 7: Full (WIN+HHI+…) | noise_censusblock2020_mean | -0.004 | 0.016 | -0.28 | 0.781158 |
| Model 7: Full (WIN+HHI+…) | raceWhite | 0.387 | 0.128 | 3.02 | 0.002710 |
| Model 7: Full (WIN+HHI+…) | ADI_NATRANK | 0.000 | 0.002 | 0.25 | 0.800933 |
| Model 7: Full (WIN+HHI+…) | hearingaids11 | 0.390 | 0.131 | 2.98 | 0.003147 |
| Model 7: Full (WIN+HHI+…) | educ | 0.028 | 0.018 | 1.57 | 0.117594 |
| Model 7: Full (WIN+HHI+…) | gender2 | 0.159 | 0.091 | 1.74 | 0.082215 |
| Model | r.squared | adj.r.squared | statistic | p.value | df | df.residual | sigma | AIC | BIC |
|---|---|---|---|---|---|---|---|---|---|
| Model 1: WIN_MOCA | 0.106 | 0.095 | 9.75 | <1e-04 | 4 | 330 | 3.235 | 1744.2 | 1767.1 |
| Model 2: RT_WIN_MOCA | 0.095 | 0.084 | 8.61 | <1e-04 | 4 | 330 | 3.255 | 1748.4 | 1771.3 |
| Model 3: LT_WIN_MOCA | 0.105 | 0.094 | 9.69 | <1e-04 | 4 | 330 | 3.236 | 1744.5 | 1767.4 |
| Model 4: HHI_MOCA | 0.078 | 0.067 | 7.10 | <1e-04 | 4 | 338 | 3.263 | 1791.7 | 1814.7 |
| Model 5: WIN+HHI_MOCA | 0.109 | 0.095 | 8.03 | <1e-04 | 5 | 329 | 3.234 | 1745.1 | 1771.8 |
| Model 6: HA+WIN+HHI_MOCA | 0.112 | 0.096 | 6.90 | <1e-04 | 6 | 327 | 3.236 | 1741.3 | 1771.8 |
| Model 7: Full (WIN+HHI+…) | 0.129 | 0.108 | 6.02 | <1e-04 | 8 | 324 | 3.219 | 1734.6 | 1772.7 |
4.3 Interaction terms assessment: Outcome = PACC score
Creating glm models with interaction terms with PACC score as the outcome. The following interactions were assessed: hearing aid use and WIN threshold, race and WIN threshold, race and hearing aid use. None of the interaction terms were significant. They were put individually and also assessed on a model with other variables.
##
## Call: glm(formula = PACC_BL ~ race_releveled * WIN_Threshold_Avg, family = gaussian(),
## data = Drives_WIN_HHI_glm)
##
## Coefficients:
## (Intercept)
## 0.45555
## race_releveledBlack or African American
## 0.15162
## WIN_Threshold_Avg
## -0.05204
## race_releveledBlack or African American:WIN_Threshold_Avg
## -0.05926
##
## Degrees of Freedom: 311 Total (i.e. Null); 308 Residual
## (79 observations deleted due to missingness)
## Null Deviance: 178.9
## Residual Deviance: 161.6 AIC: 690.2##
## Call: glm(formula = PACC_BL ~ hearingaids1 * WIN_Threshold_Avg, family = gaussian(),
## data = Drives_WIN_HHI_glm)
##
## Coefficients:
## (Intercept) hearingaids11
## 0.55572 -0.09148
## WIN_Threshold_Avg hearingaids11:WIN_Threshold_Avg
## -0.07303 0.03155
##
## Degrees of Freedom: 311 Total (i.e. Null); 308 Residual
## (79 observations deleted due to missingness)
## Null Deviance: 178.9
## Residual Deviance: 163.1 AIC: 693.1##
## Call: glm(formula = PACC_BL ~ hearingaids1 * race_releveled, family = gaussian(),
## data = Drives_WIN_HHI_glm)
##
## Coefficients:
## (Intercept)
## -0.09834
## hearingaids11
## -0.03876
## race_releveledBlack or African American
## -0.25083
## hearingaids11:race_releveledBlack or African American
## 0.07110
##
## Degrees of Freedom: 311 Total (i.e. Null); 308 Residual
## (79 observations deleted due to missingness)
## Null Deviance: 178.9
## Residual Deviance: 176.6 AIC: 717.8| Model | term | estimate | std.error | statistic | p.value |
|---|---|---|---|---|---|
| Model 1: Race*WIN | (Intercept) | 0.456 | 0.128 | 3.56 | 0.00043 |
| Model 1: Race*WIN | race_releveledBlack or African American | 0.152 | 0.403 | 0.38 | 0.70724 |
| Model 1: Race*WIN | WIN_Threshold_Avg | -0.052 | 0.011 | -4.68 | < 1e-04 |
| Model 1: Race*WIN | race_releveledBlack or African American:WIN_Threshold_Avg | -0.059 | 0.044 | -1.34 | 0.18157 |
| Model 2: HA*WIN | (Intercept) | 0.556 | 0.141 | 3.95 | < 1e-04 |
| Model 2: HA*WIN | hearingaids11 | -0.091 | 0.424 | -0.22 | 0.82919 |
| Model 2: HA*WIN | WIN_Threshold_Avg | -0.073 | 0.014 | -5.23 | < 1e-04 |
| Model 2: HA*WIN | hearingaids11:WIN_Threshold_Avg | 0.032 | 0.030 | 1.06 | 0.29044 |
| Model 3: HA*Race | (Intercept) | -0.098 | 0.052 | -1.90 | 0.05896 |
| Model 3: HA*Race | hearingaids11 | -0.039 | 0.116 | -0.33 | 0.73897 |
| Model 3: HA*Race | race_releveledBlack or African American | -0.251 | 0.127 | -1.98 | 0.04843 |
| Model 3: HA*Race | hearingaids11:race_releveledBlack or African American | 0.071 | 0.467 | 0.15 | 0.87906 |
| Model | deviance | df.residual | AIC | BIC |
|---|---|---|---|---|
| Model 1: Race*WIN | 161.61 | 308 | 690.19 | 708.90 |
| Model 2: HA*WIN | 163.11 | 308 | 693.07 | 711.78 |
| Model 3: HA*Race | 176.58 | 308 | 717.82 | 736.54 |
4.4 Stratified analysis: Based on Race
Stratified analyses revealed intriguing differences in the association between hearing aid use and cognitive function (PACC scores) across racial groups. Specifically:
Black or African American Participants: WIN_Threshold_Avg: Marginally significant (p = 0.0506) but a higher estimate -0.089123 (though P value is low because of sample size, this is right as hearing impairment has a higher negative effect on cognitive funtion in blacks than Whites). Hearing Aid Use (hearingaids1): Not significant (p = 0.5344) Gender: Not significant (p = 0.8524)
White Participants: WIN_Threshold_Avg: Highly significant (p < 0.001) but with a lower estimate of -0.050213. Hearing Aid Use (hearingaids1): Significant (p = 0.0139) Gender: Marginally significant (p = 0.0487) These findings suggest that hearing aid use is positively associated with cognitive function among White participants but not among Black or African American participants.
| Model | term | estimate | std.error | statistic | p.value |
|---|---|---|---|---|---|
| Black or African American | (Intercept) | 4.803 | 2.803 | 1.71 | 0.094933 |
| Black or African American | WIN_Threshold_Avg | -0.089 | 0.044 | -2.02 | 0.050648 |
| Black or African American | hearingaids11 | 0.253 | 0.403 | 0.63 | 0.534432 |
| Black or African American | age | -0.023 | 0.025 | -0.92 | 0.362307 |
| Black or African American | educ | 0.041 | 0.037 | 1.12 | 0.271276 |
| Black or African American | gender2 | 0.042 | 0.222 | 0.19 | 0.852369 |
| Black or African American | ADI_NATRANK | -0.006 | 0.004 | -1.38 | 0.175511 |
| Black or African American | noise_censusblock2020_mean | -0.058 | 0.033 | -1.73 | 0.091837 |
| White | (Intercept) | 0.948 | 1.147 | 0.83 | 0.409025 |
| White | WIN_Threshold_Avg | -0.050 | 0.015 | -3.39 | 0.000797 |
| White | hearingaids11 | 0.332 | 0.134 | 2.48 | 0.013880 |
| White | age | -0.018 | 0.009 | -1.92 | 0.055475 |
| White | educ | 0.021 | 0.020 | 1.02 | 0.309402 |
| White | gender2 | 0.200 | 0.101 | 1.98 | 0.048734 |
| White | ADI_NATRANK | 0.001 | 0.002 | 0.63 | 0.529402 |
| White | noise_censusblock2020_mean | 0.004 | 0.018 | 0.24 | 0.812365 |
| Full Model with Interaction | (Intercept) | 1.492 | 1.052 | 1.42 | 0.156959 |
| Full Model with Interaction | race_releveledBlack or African American | 0.066 | 0.335 | 0.20 | 0.844503 |
| Full Model with Interaction | ADI_NATRANK | 0.002 | 0.002 | 0.79 | 0.430543 |
| Full Model with Interaction | WIN_Threshold_Avg | -0.054 | 0.014 | -3.94 | 0.000101 |
| Full Model with Interaction | hearingaids11 | 0.345 | 0.125 | 2.76 | 0.006151 |
| Full Model with Interaction | age | -0.018 | 0.009 | -2.05 | 0.040856 |
| Full Model with Interaction | educ | 0.028 | 0.018 | 1.58 | 0.114583 |
| Full Model with Interaction | gender2 | 0.171 | 0.091 | 1.87 | 0.061923 |
| Full Model with Interaction | noise_censusblock2020_mean | -0.008 | 0.016 | -0.49 | 0.627341 |
| Full Model with Interaction | race_releveledBlack or African American:ADI_NATRANK | -0.006 | 0.005 | -1.38 | 0.169443 |
| Model | deviance | df.residual | AIC | BIC | logLik |
|---|---|---|---|---|---|
| Black or African American | 14.25 | 37 | 93.97 | 110.23 | -37.98 |
| White | 136.79 | 256 | 593.61 | 625.80 | -287.81 |
| Full Model with Interaction | 153.32 | 299 | 682.34 | 723.41 | -330.17 |
##
## Call:
## glm(formula = mocatots ~ race_releveled * WIN_Threshold_Avg,
## family = gaussian(), data = Drives_WIN_HHI_glm)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -12.3531 -1.6588 0.2869 2.3169 6.2985
##
## Coefficients:
## Estimate Std. Error
## (Intercept) 27.65610 0.54631
## race_releveledBlack or African American 1.12653 1.73082
## WIN_Threshold_Avg -0.24287 0.04713
## race_releveledBlack or African American:WIN_Threshold_Avg -0.28449 0.18983
## t value Pr(>|t|)
## (Intercept) 50.624 < 2e-16 ***
## race_releveledBlack or African American 0.651 0.516
## WIN_Threshold_Avg -5.153 4.42e-07 ***
## race_releveledBlack or African American:WIN_Threshold_Avg -1.499 0.135
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 10.50904)
##
## Null deviance: 3858.1 on 333 degrees of freedom
## Residual deviance: 3468.0 on 330 degrees of freedom
## (57 observations deleted due to missingness)
## AIC: 1739.5
##
## Number of Fisher Scoring iterations: 2##
## Call:
## glm(formula = mocatots ~ hearingaids1 * WIN_Threshold_Avg, family = gaussian(),
## data = Drives_WIN_HHI_glm)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -11.8260 -1.7872 0.4444 2.5149 5.8042
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 27.85532 0.60986 45.675 < 2e-16 ***
## hearingaids11 -0.37588 1.76447 -0.213 0.831
## WIN_Threshold_Avg -0.29628 0.06047 -4.900 1.51e-06 ***
## hearingaids11:WIN_Threshold_Avg 0.08629 0.12364 0.698 0.486
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 10.72064)
##
## Null deviance: 3858.1 on 333 degrees of freedom
## Residual deviance: 3537.8 on 330 degrees of freedom
## (57 observations deleted due to missingness)
## AIC: 1746.1
##
## Number of Fisher Scoring iterations: 2##
## Call:
## glm(formula = mocatots ~ hearingaids1 * race_releveled, family = gaussian(),
## data = Drives_WIN_HHI_glm)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -13.1903 -2.1903 0.8097 2.7447 5.7447
##
## Coefficients:
## Estimate Std. Error
## (Intercept) 25.1903 0.2258
## hearingaids11 -0.8282 0.4996
## race_releveledBlack or African American -0.9349 0.5441
## hearingaids11:race_releveledBlack or African American 0.9062 2.0821
## t value Pr(>|t|)
## (Intercept) 111.570 <2e-16 ***
## hearingaids11 -1.658 0.0983 .
## race_releveledBlack or African American -1.718 0.0867 .
## hearingaids11:race_releveledBlack or African American 0.435 0.6637
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 11.52066)
##
## Null deviance: 3858.1 on 333 degrees of freedom
## Residual deviance: 3801.8 on 330 degrees of freedom
## (57 observations deleted due to missingness)
## AIC: 1770.2
##
## Number of Fisher Scoring iterations: 2##
## Call:
## glm(formula = mocatots ~ hearingaids1 * WIN_Threshold_Avg + race_releveled *
## WIN_Threshold_Avg, family = gaussian(), data = Drives_WIN_HHI_glm)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -12.054 -1.685 0.300 2.416 5.851
##
## Coefficients:
## Estimate Std. Error
## (Intercept) 28.05146 0.66101
## hearingaids11 -0.11929 1.77336
## WIN_Threshold_Avg -0.29393 0.06396
## race_releveledBlack or African American 0.79063 1.77233
## hearingaids11:WIN_Threshold_Avg 0.05947 0.12511
## WIN_Threshold_Avg:race_releveledBlack or African American -0.24396 0.19477
## t value Pr(>|t|)
## (Intercept) 42.437 < 2e-16 ***
## hearingaids11 -0.067 0.946
## WIN_Threshold_Avg -4.596 6.16e-06 ***
## race_releveledBlack or African American 0.446 0.656
## hearingaids11:WIN_Threshold_Avg 0.475 0.635
## WIN_Threshold_Avg:race_releveledBlack or African American -1.253 0.211
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 10.5143)
##
## Null deviance: 3858.1 on 333 degrees of freedom
## Residual deviance: 3448.7 on 328 degrees of freedom
## (57 observations deleted due to missingness)
## AIC: 1741.6
##
## Number of Fisher Scoring iterations: 2##
## Call:
## glm(formula = mocatots ~ hearingaids1 * WIN_Threshold_Avg + race_releveled *
## WIN_Threshold_Avg + age + hhi_total_score + race_releveled +
## gender + hearingaids1 + educ + WIN_Threshold_Avg + ADI_NATRANK +
## noise_censusblock2020_mean, family = gaussian(), data = Drives_WIN_HHI_glm)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -11.8895 -1.6815 0.3715 2.3258 5.7370
##
## Coefficients:
## Estimate Std. Error
## (Intercept) 30.222736 4.711892
## hearingaids11 -0.278063 1.801771
## WIN_Threshold_Avg -0.222859 0.072372
## race_releveledBlack or African American 0.957628 1.823212
## age -0.093728 0.038393
## hhi_total_score 0.002231 0.013607
## gender2 0.638448 0.399645
## educ 0.136798 0.074996
## ADI_NATRANK -0.001786 0.008429
## noise_censusblock2020_mean 0.025185 0.070553
## hearingaids11:WIN_Threshold_Avg 0.068309 0.126447
## WIN_Threshold_Avg:race_releveledBlack or African American -0.251443 0.199930
## t value Pr(>|t|)
## (Intercept) 6.414 5.11e-10 ***
## hearingaids11 -0.154 0.87745
## WIN_Threshold_Avg -3.079 0.00225 **
## race_releveledBlack or African American 0.525 0.59978
## age -2.441 0.01518 *
## hhi_total_score 0.164 0.86986
## gender2 1.598 0.11114
## educ 1.824 0.06908 .
## ADI_NATRANK -0.212 0.83231
## noise_censusblock2020_mean 0.357 0.72135
## hearingaids11:WIN_Threshold_Avg 0.540 0.58943
## WIN_Threshold_Avg:race_releveledBlack or African American -1.258 0.20944
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 10.43781)
##
## Null deviance: 3852.1 on 330 degrees of freedom
## Residual deviance: 3329.7 on 319 degrees of freedom
## (60 observations deleted due to missingness)
## AIC: 1729.5
##
## Number of Fisher Scoring iterations: 2##
## Call:
## glm(formula = mocatots ~ race_releveled + WIN_Threshold_Avg +
## hearingaids1 + ADI_NATRANK + noise_censusblock2020_mean +
## educ + age + gender, family = gaussian(), data = Drives_WIN_HHI_glm)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -11.9650 -1.6994 0.4013 2.2965 6.1042
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 30.429728 4.647622 6.547 2.31e-10
## race_releveledBlack or African American -1.212266 0.549470 -2.206 0.028073
## WIN_Threshold_Avg -0.222883 0.058622 -3.802 0.000172
## hearingaids11 0.735565 0.540447 1.361 0.174456
## ADI_NATRANK -0.002030 0.008368 -0.243 0.808472
## noise_censusblock2020_mean 0.025313 0.070436 0.359 0.719555
## educ 0.138153 0.074902 1.844 0.066036
## age -0.095958 0.037834 -2.536 0.011675
## gender2 0.587882 0.395624 1.486 0.138267
##
## (Intercept) ***
## race_releveledBlack or African American *
## WIN_Threshold_Avg ***
## hearingaids11
## ADI_NATRANK
## noise_censusblock2020_mean
## educ .
## age *
## gender2
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 10.41554)
##
## Null deviance: 3852.1 on 330 degrees of freedom
## Residual deviance: 3353.8 on 322 degrees of freedom
## (60 observations deleted due to missingness)
## AIC: 1725.8
##
## Number of Fisher Scoring iterations: 2##
## Call:
## glm(formula = mocatots ~ race_releveled + WIN_Threshold_Avg +
## hearingaids1, family = gaussian(), data = Drives_WIN_HHI_glm)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -12.049 -1.633 0.296 2.418 6.226
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 28.09826 0.56136 50.054 < 2e-16
## race_releveledBlack or African American -1.32981 0.50839 -2.616 0.00931
## WIN_Threshold_Avg -0.29774 0.05292 -5.626 3.94e-08
## hearingaids11 0.75002 0.53698 1.397 0.16343
##
## (Intercept) ***
## race_releveledBlack or African American **
## WIN_Threshold_Avg ***
## hearingaids11
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 10.51838)
##
## Null deviance: 3858.1 on 333 degrees of freedom
## Residual deviance: 3471.1 on 330 degrees of freedom
## (57 observations deleted due to missingness)
## AIC: 1739.8
##
## Number of Fisher Scoring iterations: 2##
## Call:
## glm(formula = mocatots ~ race_releveled + WIN_Threshold_Avg +
## hearingaids1 + age + educ + gender, family = gaussian(),
## data = Drives_WIN_HHI_glm)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -12.1747 -1.7359 0.4048 2.2836 6.0570
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 31.46422 2.84195 11.071 < 2e-16
## race_releveledBlack or African American -1.29064 0.50604 -2.550 0.011214
## WIN_Threshold_Avg -0.22612 0.05784 -3.909 0.000113
## hearingaids11 0.76783 0.53114 1.446 0.149244
## age -0.09260 0.03692 -2.509 0.012607
## educ 0.13920 0.07180 1.939 0.053396
## gender2 0.54921 0.39005 1.408 0.160065
##
## (Intercept) ***
## race_releveledBlack or African American *
## WIN_Threshold_Avg ***
## hearingaids11
## age *
## educ .
## gender2
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 10.28102)
##
## Null deviance: 3858.1 on 333 degrees of freedom
## Residual deviance: 3361.9 on 327 degrees of freedom
## (57 observations deleted due to missingness)
## AIC: 1735.1
##
## Number of Fisher Scoring iterations: 2## Analysis of Deviance Table
##
## Model: gaussian, link: identity
##
## Response: mocatots
##
## Terms added sequentially (first to last)
##
##
## Df Deviance Resid. Df Resid. Dev F Pr(>F)
## NULL 333 3858.1
## race_releveled 1 24.63 332 3833.5 2.3956 0.12264
## WIN_Threshold_Avg 1 341.91 331 3491.6 33.2562 1.871e-08 ***
## hearingaids1 1 20.52 330 3471.1 1.9959 0.15867
## age 1 59.34 329 3411.7 5.7723 0.01684 *
## educ 1 29.44 328 3382.3 2.8639 0.09154 .
## gender 1 20.38 327 3361.9 1.9826 0.16007
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1