HHCOVID_metaanalysis/metaregression notes.Rmd at master · BenSCooper/HHCOVID_metaanalysis · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
---
title: "Meta-regression analysis of the impact of hand hygiene"
output: html_notebook
---
Random effects metaregression on all the data.
This allows for random intercepts and also for effectiveness of interventions (hand hygiene and maks use) to vary between the studies (but not between arms within the studies). This seems reasonable as hand hygiene implementation (soap or sanitizer and training) and study populations vary considerably.

Analysis of full data

```{r}
source("metareg_define_data.R")    # define data
require(rstan)

fit1 <- stan(
  file = "metareg0.3.stan",  # Stan program
  data = hh_trial_data,    # named list of data
  chains = 2,             # number of Markov chains
  warmup = 1000,          # number of warmup iterations per chain
  iter = 100000,            # total number of iterations per chain
  cores = 2,              # number of cores (could use one per chain)
  refresh = 1000,             # no progress shown
  control = list(max_treedepth = 15,adapt_delta=0.8)
)
summary(fit1)
plot(fit1, pars=c("c0", "b0", "b[1]","b[2]","b[3]","b[4]","b[5]","b[6]","c[1]","c[2]","c[3]","c[4]","c[5]" ))
```

```{r}
plot(fit1, pars=c("c0", "b0", "b[1]","b[2]","b[3]","b[4]","b[5]","b[6]","c[1]","c[2]","c[3]","c[4]","c[5]" ))
```


Analysis of full data excluding Simmerman study (trial 2), which seems to be something of an outlier.

```{r}
source("metareg_define_data.R")    # define data
require(rstan)

fit2 <- stan(
  file = "metareg0.3.no.study.2.stan",  # Stan program
  data = hh_trial_data,    # named list of data
  chains = 2,             # number of Markov chains
  warmup = 1000,          # number of warmup iterations per chain
  iter = 100000,            # total number of iterations per chain
  cores = 2,              # number of cores (could use one per chain)
  refresh = 1000,             # no progress shown
  control = list(max_treedepth = 15,adapt_delta=0.8)
)
summary(fit2)

```

```{r}
plot(fit2, pars=c("c0", "b0", "b[1]","b[2]","b[3]","b[4]","b[5]","b[6]","c[1]","c[2]","c[3]","c[4]","c[5]" ))
```
Next run model metareg0.4.stan that uses a log link for the binomial regression so that everything is on the same scale and we
get risk ratios rather than odds ratios


```{r}
fit4 <- stan(
  file = "metareg0.4.stan",  # Stan program
  data = hh_trial_data,    # named list of data
  chains = 2,             # number of Markov chains
  warmup = 1000,          # number of warmup iterations per chain
  iter = 100000,            # total number of iterations per chain
  cores = 2,              # number of cores (could use one per chain)
  refresh = 1000,             # no progress shown
  control = list(max_treedepth = 15,adapt_delta=0.8)
)
summary(fit4)
```

```{r}
plot(fit4, pars=c("c0", "b0", "b[1]","b[2]","b[3]","b[4]","b[5]","b[6]","c[1]","c[2]","c[3]","c[4]","c[5]" ))
```