Introduction

Participating Dialysis Centers
In a national quality improvement initiative, dialysis centers from ESRD Networks throughout the United States were invited to participate in a one day Explore Transplant (ET) educational training. Only dialysis centers that participated in the ET trainings are considered for this study. The inclusion criterion for dialysis centers is to have sent a staff representative to one of 78 ET trainings over a four year period between 3/20/2011 and 3/18/2015. Representatives from 1989 unique dialysis centers attended trainings, though centers were excluded due to only providing acute dialysis (19 centers), serving only pediatric patients (17 centers), refusal to participate in the study (23 centers), and not responding to a survey given during the trainings (38 centers). Further, X centers did not initiate dialysis with at least 1 new patient during wait-listing and transplant rate follow-up periods (see below). This resulted in a sample of X unique dialysis centers.

Patient Selection
Patients from participating dialysis center will be selected to calculate transplant wait-listing and transplantation rates in two cohorts: a pre-ET cohort and a post-ET cohort. For the pre-ET cohort, incident dialysis patients who initialized dialysis treatment between 21 and 12 months prior to their center’s ET training will be sampled. For the post-ET cohort, incident dialysis patients who initialized dialysis between 3 months before the ET training date and 6 months after will be sampled. Patients in the post-ET cohort who started dialysis during this time are most likely to have been exposed ET from their dialysis staff, while patients sampled for the pre-ET cohort would not have any exposure to ET.

Calculation of Wait-listing and Transplant Rates Annualized 12 month wait-listing and transplant rates for each dialysis center will be calculated for before and after the ET trainings. Each rate will be defined as the number of patients who have the event of interest (wait-list or receive a transplant) divided by the total days patients are at risk during the follow-up period, then multiplied by 365.25. Patients will be censored at death. To calculate the pre-ET 12-month annualized rates, incident dialysis patients between 21 and 12 months prior to the ET training will be followed for 12 months. The follow-up for the last patients included in this cohort would end at the ET training date. For the post-ET 12 month rates, incident dialysis patients between 3 months prior to the ET training date and 6 months after the training date will be followed for 12 months. The follow-up for the last patients in this cohort will end 18 months after the ET date. This gives an equal sample period (9 months) and equal follow-up period (12 months) for both the pre- and post-ET rates. The figure below shows the sampling and follow-up of patients for the wait-listing and transplant rates.

Aims

HRSA 4&5 Combined Analysis Plan
Aims – Paper 1
1. Characteristics of dialysis center providers educating about transplant including level of preparedness (slides exist) [is this its own paper?]
2. Did trainings increase dialysis providers’ knowledge about deceased and living donor transplantation?
3. Did trainings motivate dialysis providers to use Explore Transplant with their own Spanish- and English-speaking patients in the next six months?

Statistical methods

  • The proportion of dialysis centers sending at least one dialysis educator to the in-person or eLearning trainings will then be assessed. The total number of dialysis providers trained and the average number of providers trained per center will be calculated using descriptive statistics including proportions and means.
  • We will assess whether providers’ deceased and living donation knowledge significantly increased following the trainings using paired t-tests or McNemar tests.
  • The proportion of providers willing to take each action, including educating 5 patients, and the total number of patients planning to be educated will be calculated to assess the scope of the planned education to occur in the next six months by providers across all Networks.
  • Difference-In-Difference analysis will be used to compare WL and LDKT rates pre and post Explore Transplant education.

Data Preparation

## No duplicate combinations found of: provider

Main educators

The top 3 job titiles for main educators are social workers, dieticians, and nurses.
Main educators are defined as self-selected “direct” educator AND who educated at least 1 English/Spanish Speaking patient.

Table 1. Demograghics of Educators.

Main educators are mostly Black women, ~ 45 yrs old, mostly social workers or nurses.

Overall
n 1765
age (mean (SD)) 45.60 (11.43)
gender2 = Female (%) 1231 (69.9)
race8 (%)
White 1201 (68.6)
African American or Black 236 (13.5)
Hispanic or Latino 113 ( 6.4)
American Indian or Alaskan Native 19 ( 1.1)
Asian 149 ( 8.5)
Native Hawaiian or other Pacifc Islander 3 ( 0.2)
Multiracial 19 ( 1.1)
Other 12 ( 0.7)
jobs (%)
Medical Director/Physician 142 ( 8.1)
Nurse Manager/Facility Administrator 273 (15.5)
Nurse (RN/LPN/MSN/APN) 81 ( 4.6)
Dietician 1021 (57.9)
Social Worker (MSW) 168 ( 9.5)
Dialysis Technician 77 ( 4.4)
totknowpre (mean (SD)) 5.34 (2.07)
op_priord = Yes (%) 1606 (91.8)
op_knowd = Yes (%) 1046 (60.0)
op_abilityd = Yes (%) 976 (56.4)
op_txmatd = Yes (%) 810 (46.5)

Are providers prepared enough in terms of…

Transplant knowledge

Mean = 5.34, SD = 2.07, Median = 5, range 0-12.
n mean sd IQR p00 p50 p100
1762 5.3 2.1 3 0 5 12

Priority?

91.8% main educators consider patient education an important priority for them

## Warning: Grouping rowwise data frame strips rowwise nature

Knowledgable?

But only 60% educators feel sufficiently knowledgable about transplant to answer most of patients’ questions

## Warning: Grouping rowwise data frame strips rowwise nature

Confident?

Only 56.4% educators are confident in their ability as a transplant educator

## Warning: Grouping rowwise data frame strips rowwise nature

Enough Materials?

Only 46.5% educators have excellent transplant education materials available at their dialysis center for patients.

## Warning: Grouping rowwise data frame strips rowwise nature

Dialysis providers’ knowledge - pre and post ET

Paired t-test results suggest providers’ knowledges after ET class are significantly increased by 5.6 points on average than knowledge before ET class, p < 0.001 (Paired t test).

## Warning: Removed 68 rows containing non-finite values (stat_boxplot).

Did ET Trainings motivated Providers to use ET materials in the next 6 months

nat_pro2$postedu5 n percent
Precontemplation 188 0.09
Contemplation 627 0.31
Preparation 927 0.46
Action 99 0.05
Maintenance 42 0.02
NA 120 0.06
Total 2003 1.00

Comparisons, Pre and Post ET

Generally speaking, after ET training, providers are more likely to “consider patient edu as a priority”,“feel knowledgable”, “confident in their ability to educate”, “have excellent materials available”.

1 = agree, 0 = disagree

Priority?

## 
##    Cell Contents
## |-------------------------|
## |                   Count |
## | Chi-square contribution |
## |             Row Percent |
## |          Column Percent |
## |           Total Percent |
## |-------------------------|
## 
## Total Observations in Table:  1888 
## 
##                    | nat_pro2$op_priorpstd 
## nat_pro2$op_priord |        0  |        1  | Row Total | 
## -------------------|-----------|-----------|-----------|
##                 No |       25  |      146  |      171  | 
##                    |   70.444  |    2.312  |           | 
##                    |   14.620% |   85.380% |    9.057% | 
##                    |   41.667% |    7.987% |           | 
##                    |    1.324% |    7.733% |           | 
## -------------------|-----------|-----------|-----------|
##                Yes |       35  |     1682  |     1717  | 
##                    |    7.016  |    0.230  |           | 
##                    |    2.038% |   97.962% |   90.943% | 
##                    |   58.333% |   92.013% |           | 
##                    |    1.854% |   89.089% |           | 
## -------------------|-----------|-----------|-----------|
##       Column Total |       60  |     1828  |     1888  | 
##                    |    3.178% |   96.822% |           | 
## -------------------|-----------|-----------|-----------|
## 
##  
## McNemar's Chi-squared test 
## ------------------------------------------------------------
## Chi^2 =  68     d.f. =  1     p =  1.6e-16 
## 
## McNemar's Chi-squared test with continuity correction 
## ------------------------------------------------------------
## Chi^2 =  67     d.f. =  1     p =  2.9e-16 
## 
##  
##        Minimum expected frequency: 5.4

Knowledgable?

## 
##    Cell Contents
## |-------------------------|
## |                   Count |
## | Chi-square contribution |
## |             Row Percent |
## |          Column Percent |
## |           Total Percent |
## |-------------------------|
## 
## Total Observations in Table:  1871 
## 
##                   | nat_pro2$op_knowpstd 
## nat_pro2$op_knowd |        0  |        1  | Row Total | 
## ------------------|-----------|-----------|-----------|
##                No |      100  |      703  |      803  | 
##                   |   15.280  |    1.409  |           | 
##                   |   12.453% |   87.547% |   42.918% | 
##                   |   63.291% |   41.039% |           | 
##                   |    5.345% |   37.573% |           | 
## ------------------|-----------|-----------|-----------|
##               Yes |       58  |     1010  |     1068  | 
##                   |   11.489  |    1.060  |           | 
##                   |    5.431% |   94.569% |   57.082% | 
##                   |   36.709% |   58.961% |           | 
##                   |    3.100% |   53.982% |           | 
## ------------------|-----------|-----------|-----------|
##      Column Total |      158  |     1713  |     1871  | 
##                   |    8.445% |   91.555% |           | 
## ------------------|-----------|-----------|-----------|
## 
##  
## McNemar's Chi-squared test 
## ------------------------------------------------------------
## Chi^2 =  547     d.f. =  1     p =  6.6e-121 
## 
## McNemar's Chi-squared test with continuity correction 
## ------------------------------------------------------------
## Chi^2 =  545     d.f. =  1     p =  1.5e-120 
## 
##  
##        Minimum expected frequency: 68

Confident?

## 
##    Cell Contents
## |-------------------------|
## |                   Count |
## | Chi-square contribution |
## |             Row Percent |
## |          Column Percent |
## |           Total Percent |
## |-------------------------|
## 
## Total Observations in Table:  1856 
## 
##                      | nat_pro2$op_abilitypstd 
## nat_pro2$op_abilityd |        0  |        1  | Row Total | 
## ---------------------|-----------|-----------|-----------|
##                   No |      138  |      741  |      879  | 
##                      |   27.596  |    3.092  |           | 
##                      |   15.700% |   84.300% |   47.360% | 
##                      |   73.797% |   44.398% |           | 
##                      |    7.435% |   39.925% |           | 
## ---------------------|-----------|-----------|-----------|
##                  Yes |       49  |      928  |      977  | 
##                      |   24.828  |    2.782  |           | 
##                      |    5.015% |   94.985% |   52.640% | 
##                      |   26.203% |   55.602% |           | 
##                      |    2.640% |   50.000% |           | 
## ---------------------|-----------|-----------|-----------|
##         Column Total |      187  |     1669  |     1856  | 
##                      |   10.075% |   89.925% |           | 
## ---------------------|-----------|-----------|-----------|
## 
##  
## McNemar's Chi-squared test 
## ------------------------------------------------------------
## Chi^2 =  606     d.f. =  1     p =  7.7e-134 
## 
## McNemar's Chi-squared test with continuity correction 
## ------------------------------------------------------------
## Chi^2 =  604     d.f. =  1     p =  1.8e-133 
## 
##  
##        Minimum expected frequency: 89

Enough Materials?

## 
##    Cell Contents
## |-------------------------|
## |                   Count |
## | Chi-square contribution |
## |             Row Percent |
## |          Column Percent |
## |           Total Percent |
## |-------------------------|
## 
## Total Observations in Table:  1851 
## 
##                    | nat_pro2$op_txmatpstd 
## nat_pro2$op_txmatd |        0  |        1  | Row Total | 
## -------------------|-----------|-----------|-----------|
##                 No |      286  |      734  |     1020  | 
##                    |   26.156  |    6.847  |           | 
##                    |   28.039% |   71.961% |   55.105% | 
##                    |   74.479% |   50.034% |           | 
##                    |   15.451% |   39.654% |           | 
## -------------------|-----------|-----------|-----------|
##                Yes |       98  |      733  |      831  | 
##                    |   32.105  |    8.404  |           | 
##                    |   11.793% |   88.207% |   44.895% | 
##                    |   25.521% |   49.966% |           | 
##                    |    5.294% |   39.600% |           | 
## -------------------|-----------|-----------|-----------|
##       Column Total |      384  |     1467  |     1851  | 
##                    |   20.746% |   79.254% |           | 
## -------------------|-----------|-----------|-----------|
## 
##  
## McNemar's Chi-squared test 
## ------------------------------------------------------------
## Chi^2 =  486     d.f. =  1     p =  9.7e-108 
## 
## McNemar's Chi-squared test with continuity correction 
## ------------------------------------------------------------
## Chi^2 =  485     d.f. =  1     p =  2.1e-107 
## 
##  
##        Minimum expected frequency: 172

Waitlisting and LDKT rates

The zero inflated model with offset of “potential events” showed no significance of treatment on number (counts) of patients listed 12 month post ET. DiD (interaction between treatment and time) was also non-significant.

12 Month WL rates

## 
## Call:
## zeroinfl(formula = list12m ~ ET + time2 + did, data = df2, offset = log(DIAL2LIST_12M))
## 
## Pearson residuals:
##    Min     1Q Median     3Q    Max 
## -1.583 -0.781 -0.396  0.513 16.803 
## 
## Count model coefficients (poisson with log link):
##             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)  -8.0768     0.0304 -265.69   <2e-16 ***
## ET            0.0634     0.0410    1.55    0.122    
## time2        -0.0855     0.0458   -1.87    0.062 .  
## did          -0.0284     0.0619   -0.46    0.647    
## 
## Zero-inflation model coefficients (binomial with logit link):
##             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)  -1.5618     0.1384  -11.28   <2e-16 ***
## ET           -0.0670     0.1924   -0.35    0.727    
## time2         0.3869     0.1820    2.13    0.034 *  
## did           0.0593     0.2537    0.23    0.815    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 
## 
## Number of iterations in BFGS optimization: 17 
## Log-likelihood: -7.67e+03 on 8 Df
## [1] 1.2
## [1] 2.9
## 
## Call:
## zeroinfl(formula = list12m ~ ET + time2 + did, data = df2, offset = log(DIAL2LIST_12M), 
##     dist = "negbin")
## 
## Pearson residuals:
##    Min     1Q Median     3Q    Max 
## -1.145 -0.709 -0.370  0.426 16.282 
## 
## Count model coefficients (negbin with log link):
##             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)  -8.2455     0.0349 -236.53   <2e-16 ***
## ET            0.0628     0.0475    1.32     0.19    
## time2        -0.1027     0.0625   -1.64     0.10    
## did          -0.0608     0.0826   -0.74     0.46    
## Log(theta)    0.5774     0.0691    8.35   <2e-16 ***
## 
## Zero-inflation model coefficients (binomial with logit link):
##             Estimate Std. Error z value Pr(>|z|)
## (Intercept)   -12.07      70.54   -0.17     0.86
## ET             -3.81     354.10   -0.01     0.99
## time2           9.09      70.55    0.13     0.90
## did             3.30     354.10    0.01     0.99
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 
## 
## Theta = 1.781 
## Number of iterations in BFGS optimization: 39 
## Log-likelihood: -7.44e+03 on 9 Df

12M Listing rates, ET >= 5 vs. ET < 5

DiD (interaction between treatment and time) was non-significant.

## 
## Call:
## zeroinfl(formula = list12m ~ ed5 + time2 + did, data = df2, offset = log(DIAL2LIST_12M))
## 
## Pearson residuals:
##    Min     1Q Median     3Q    Max 
## -1.682 -0.778 -0.404  0.484 10.317 
## 
## Count model coefficients (poisson with log link):
##             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)  -8.0400     0.0435 -184.64   <2e-16 ***
## ed5           0.0468     0.0561    0.83    0.405    
## time2        -0.1403     0.0657   -2.13    0.033 *  
## did           0.0458     0.0850    0.54    0.590    
## 
## Zero-inflation model coefficients (binomial with logit link):
##             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)   -1.424      0.183   -7.79  6.7e-15 ***
## ed5           -0.404      0.269   -1.50     0.13    
## time2          0.238      0.260    0.91     0.36    
## did            0.397      0.359    1.11     0.27    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 
## 
## Number of iterations in BFGS optimization: 16 
## Log-likelihood: -4.02e+03 on 8 Df
## [1] 1.2
## [1] 2.9
## 
## Call:
## zeroinfl(formula = list12m ~ ed5 + time2 + did, data = df2, offset = log(DIAL2LIST_12M), 
##     dist = "negbin")
## 
## Pearson residuals:
##    Min     1Q Median     3Q    Max 
## -1.194 -0.713 -0.369  0.415 10.114 
## 
## Count model coefficients (negbin with log link):
##             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)  -8.2358     0.0685 -120.29  < 2e-16 ***
## ed5           0.0978     0.0803    1.22    0.223    
## time2        -0.1574     0.0943   -1.67    0.095 .  
## did           0.0209     0.1209    0.17    0.863    
## Log(theta)    0.6754     0.1151    5.87  4.5e-09 ***
## 
## Zero-inflation model coefficients (binomial with logit link):
##             Estimate Std. Error z value Pr(>|z|)
## (Intercept)    -5.48      11.70   -0.47     0.64
## ed5            -5.66      42.21   -0.13     0.89
## time2           2.14      11.43    0.19     0.85
## did             6.23      42.20    0.15     0.88
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 
## 
## Theta = 1.965 
## Number of iterations in BFGS optimization: 53 
## Log-likelihood: -3.92e+03 on 9 Df

18M WL rates

Treatment was not a significant predictor ofnumber (counts) of patients listed 18 month post ET. DiD (interaction between treatment and time) was also non-significant. Only “time” was significant in both the count and zero inflated model.

## 
## Call:
## zeroinfl(formula = list18m ~ ET + time2 + did, data = df2, offset = log(DIAL2LIST_18M))
## 
## Pearson residuals:
##    Min     1Q Median     3Q    Max 
## -1.955 -0.811 -0.347  0.581 16.806 
## 
## Count model coefficients (poisson with log link):
##             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)  -8.1061     0.0249 -325.06   <2e-16 ***
## ET            0.0461     0.0340    1.35    0.176    
## time2        -0.0904     0.0383   -2.36    0.018 *  
## did           0.0155     0.0520    0.30    0.765    
## 
## Zero-inflation model coefficients (binomial with logit link):
##             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)  -2.2189     0.1742  -12.74  < 2e-16 ***
## ET           -0.0453     0.2470   -0.18     0.85    
## time2         0.9282     0.2055    4.52  6.3e-06 ***
## did           0.0878     0.2892    0.30     0.76    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 
## 
## Number of iterations in BFGS optimization: 18 
## Log-likelihood: -8.62e+03 on 8 Df
## [1] 1.5
## [1] 3.9
## 
## Call:
## zeroinfl(formula = list18m ~ ET + time2 + did, data = df2, offset = log(DIAL2LIST_18M), 
##     dist = "negbin")
## 
## Pearson residuals:
##    Min     1Q Median     3Q    Max 
## -1.338 -0.738 -0.320  0.481 16.267 
## 
## Count model coefficients (negbin with log link):
##              Estimate Std. Error z value Pr(>|z|)    
## (Intercept) -8.190588   0.029918 -273.77   <2e-16 ***
## ET           0.040582   0.040869    0.99    0.321    
## time2       -0.103011   0.051673   -1.99    0.046 *  
## did          0.000731   0.069270    0.01    0.992    
## Log(theta)   0.911103   0.066804   13.64   <2e-16 ***
## 
## Zero-inflation model coefficients (binomial with logit link):
##             Estimate Std. Error z value Pr(>|z|)
## (Intercept)   -12.82      72.06   -0.18     0.86
## ET             -3.18     286.81   -0.01     0.99
## time2          10.63      72.06    0.15     0.88
## did             3.17     286.81    0.01     0.99
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 
## 
## Theta = 2.487 
## Number of iterations in BFGS optimization: 39 
## Log-likelihood: -8.36e+03 on 9 Df

18M WL rates, Ed5 = 1 vs. ed5 = 0;

DiD (interaction between treatment and time) was also non-significant.

## 
## Call:
## zeroinfl(formula = list18m ~ ed5 + time2 + did, data = df2, offset = log(DIAL2LIST_18M))
## 
## Pearson residuals:
##    Min     1Q Median     3Q    Max 
## -2.016 -0.815 -0.344  0.547  9.308 
## 
## Count model coefficients (poisson with log link):
##             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)  -8.1289     0.0367 -221.33   <2e-16 ***
## ed5           0.1181     0.0473    2.50    0.013 *  
## time2        -0.0362     0.0547   -0.66    0.509    
## did          -0.0646     0.0714   -0.90    0.366    
## 
## Zero-inflation model coefficients (binomial with logit link):
##             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)   -2.229      0.268   -8.33   <2e-16 ***
## ed5           -0.105      0.359   -0.29    0.770    
## time2          0.967      0.312    3.10    0.002 ** 
## did            0.122      0.416    0.29    0.769    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 
## 
## Number of iterations in BFGS optimization: 17 
## Log-likelihood: -4.52e+03 on 8 Df
## [1] 1.5
## [1] 3.9
## 
## Call:
## zeroinfl(formula = list18m ~ ed5 + time2 + did, data = df2, offset = log(DIAL2LIST_18M), 
##     dist = "negbin")
## 
## Pearson residuals:
##    Min     1Q Median     3Q    Max 
## -1.388 -0.748 -0.323  0.467  9.073 
## 
## Count model coefficients (negbin with log link):
##             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)  -8.2179     0.0417 -197.24   <2e-16 ***
## ed5           0.1203     0.0553    2.17     0.03 *  
## time2        -0.0665     0.0717   -0.93     0.35    
## did          -0.0448     0.0934   -0.48     0.63    
## Log(theta)    0.9918     0.0950   10.44   <2e-16 ***
## 
## Zero-inflation model coefficients (binomial with logit link):
##             Estimate Std. Error z value Pr(>|z|)
## (Intercept)   -12.99     104.37   -0.12     0.90
## ed5            -3.13     463.59   -0.01     0.99
## time2          10.77     104.37    0.10     0.92
## did             3.34     463.59    0.01     0.99
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 
## 
## Theta = 2.696 
## Number of iterations in BFGS optimization: 39 
## Log-likelihood: -4.4e+03 on 9 Df

12M LDKT rates

DiD (interaction between treatment and time) was non-significant. 

## # A tibble: 4 x 5
##   term         estimate std.error statistic p.value
##   <chr>           <dbl>     <dbl>     <dbl>   <dbl>
## 1 (Intercept)  0.107       0.0361    2.96   0.00306
## 2 ET          -0.000643    0.0497   -0.0129 0.990  
## 3 time2        0.0104      0.0511    0.204  0.838  
## 4 did          0.0464      0.0703    0.659  0.510
## [1] 0.12
## [1] 1.6
## [1] 0.15

12M LDKT rates, ET >=5 vs. ET < 5

DiD (interaction between treatment and time) was non-significant.

## # A tibble: 4 x 5
##   term        estimate std.error statistic p.value
##   <chr>          <dbl>     <dbl>     <dbl>   <dbl>
## 1 (Intercept)   0.100     0.0685     1.46    0.144
## 2 ed5           0.0115    0.0933     0.123   0.902
## 3 time2         0.135     0.0969     1.39    0.164
## 4 did          -0.145     0.132     -1.10    0.273
## [1] 0.12
## [1] 1.6
## [1] 0.15

18M LDKT rates comparing ET >=5 vs. ET < 5

DiD (interaction between treatment and time) was non-significant.

## # A tibble: 4 x 5
##   term        estimate std.error statistic p.value
##   <chr>          <dbl>     <dbl>     <dbl>   <dbl>
## 1 (Intercept)   0.104     0.0685     1.52    0.129
## 2 ed5           0.0131    0.0932     0.141   0.888
## 3 time2         0.136     0.0969     1.41    0.160
## 4 did          -0.148     0.132     -1.12    0.263
## [1] 0.13
## [1] 1.6
## [1] 0.15