Quick start guide

The objective of this package is to compute rates adjusted by a reference population or other rate. This is a very common procedure in epidemiology, allowing the comparison of rates of a event (like mortality) among groups that have different age distributions.

Some packages like the epitools compute these adjusted rates. This package functions wraps the epitools functions in a tidy way, allowing the computation of age adjusted rates for several groups using key variables, like year and regions for example.

Setup

Installing the package

devtools::install_github("rfsaldanha/tidyrates")

Load the package

library(tidyrates)

Direct adjusted rate

Events and population data

Let’s use the Fleiss dataset, quoted by the epitools package (Fleiss, 1981, p. 249 ).

population <- c(230061, 329449, 114920, 39487, 14208, 3052,
72202, 326701, 208667, 83228, 28466, 5375, 15050, 175702,
207081, 117300, 45026, 8660, 2293, 68800, 132424, 98301, 
46075, 9834, 327, 30666, 123419, 149919, 104088, 34392, 
319933, 931318, 786511, 488235, 237863, 61313)

population <- matrix(population, 6, 6, 
dimnames = list(c("Under 20", "20-24", "25-29", "30-34", "35-39",
"40 and over"), c("1", "2", "3", "4", "5+", "Total")))

count <- c(107, 141, 60, 40, 39, 25, 25, 150, 110, 84, 82, 39,
3, 71, 114, 103, 108, 75, 1, 26, 64, 89, 137, 96, 0, 8, 63, 112,
262, 295, 136, 396, 411, 428, 628, 530)
count <- matrix(count, 6, 6, 
dimnames = list(c("Under 20", "20-24", "25-29", "30-34", "35-39",
"40 and over"), c("1", "2", "3", "4", "5+", "Total")))

population
#>                  1      2      3      4     5+  Total
#> Under 20    230061  72202  15050   2293    327 319933
#> 20-24       329449 326701 175702  68800  30666 931318
#> 25-29       114920 208667 207081 132424 123419 786511
#> 30-34        39487  83228 117300  98301 149919 488235
#> 35-39        14208  28466  45026  46075 104088 237863
#> 40 and over   3052   5375   8660   9834  34392  61313

count
#>               1   2   3   4  5+ Total
#> Under 20    107  25   3   1   0   136
#> 20-24       141 150  71  26   8   396
#> 25-29        60 110 114  64  63   411
#> 30-34        40  84 103  89 112   428
#> 35-39        39  82 108 137 262   628
#> 40 and over  25  39  75  96 295   530

The Fleiss data present events (count object) and population (population object) for six age groups on five different groups (from 1 to 5+).

The tidyrates package present the same Fleiss data in a tidy way, with a tibble in long format.

fleiss_data
#>       key   age_group       name  value
#> 1      k1    Under 20 population 230061
#> 2      k1    Under 20     events    107
#> 3      k1       20-24 population 329449
#> 4      k1       20-24     events    141
#> 5      k1       25-29 population 114920
#> 6      k1       25-29     events     60
#> 7      k1       30-34 population  39487
#> 8      k1       30-34     events     40
#> 9      k1       35-39 population  14208
#> 10     k1       35-39     events     39
#> 11     k1 40 and over population   3052
#> 12     k1 40 and over     events     25
#> 13     k2    Under 20 population  72202
#> 14     k2    Under 20     events     25
#> 15     k2       20-24 population 326701
#> 16     k2       20-24     events    150
#> 17     k2       25-29 population 208667
#> 18     k2       25-29     events    110
#> 19     k2       30-34 population  83228
#> 20     k2       30-34     events     84
#> 21     k2       35-39 population  28466
#> 22     k2       35-39     events     82
#> 23     k2 40 and over population   5375
#> 24     k2 40 and over     events     39
#> 25     k3    Under 20 population  15050
#> 26     k3    Under 20     events      3
#> 27     k3       20-24 population 175702
#> 28     k3       20-24     events     71
#> 29     k3       25-29 population 207081
#> 30     k3       25-29     events    114
#> 31     k3       30-34 population 117300
#> 32     k3       30-34     events    103
#> 33     k3       35-39 population  45026
#> 34     k3       35-39     events    108
#> 35     k3 40 and over population   8660
#> 36     k3 40 and over     events     75
#> 37     k4    Under 20 population   2293
#> 38     k4    Under 20     events      1
#> 39     k4       20-24 population  68800
#> 40     k4       20-24     events     26
#> 41     k4       25-29 population 132424
#> 42     k4       25-29     events     64
#> 43     k4       30-34 population  98301
#> 44     k4       30-34     events     89
#> 45     k4       35-39 population  46075
#> 46     k4       35-39     events    137
#> 47     k4 40 and over population   9834
#> 48     k4 40 and over     events     96
#> 49 k5plus    Under 20 population    327
#> 50 k5plus    Under 20     events      0
#> 51 k5plus       20-24 population  30666
#> 52 k5plus       20-24     events      8
#> 53 k5plus       25-29 population 123419
#> 54 k5plus       25-29     events     63
#> 55 k5plus       30-34 population 149919
#> 56 k5plus       30-34     events    112
#> 57 k5plus       35-39 population 104088
#> 58 k5plus       35-39     events    262
#> 59 k5plus 40 and over population  34392
#> 60 k5plus 40 and over     events    295

The key variable refers to the groups, age_group to the age groups, name separates the values into events and population.

You may use this same structure for your use case data.

Reference population data

The Fleiss example uses the average population as standard population reference.

standard<-apply(population[,-6], 1, mean)
standard
#>    Under 20       20-24       25-29       30-34       35-39 40 and over 
#>     63986.6    186263.6    157302.2     97647.0     47572.6     12262.6

Using tidyrates, we must supply a tibble with two variables: age group and population.

standard_pop <- tibble::tibble(
    age_group = c("Under 20", "20-24", "25-29", "30-34", "35-39", "40 and over"),
    population = c(63986.6, 186263.6, 157302.2, 97647.0, 47572.6, 12262.6)
  )

Rate computation

To use the direct adjustment procedure, tidyrate present the rate_adj_direct function. The .data argument must be a tibble with the events and population data, and the .std argument must be standard population tibble. The .keys argument must point to grouping variables on the .data tibble, if available.

The rate_adj_direct will compute the crude rate, adjusted rate and exact confidence intervals for each group.

rate_adj_direct(fleiss_data, .std = standard_pop, .keys = "key")
#> # A tibble: 5 × 5
#>   key    crude.rate adj.rate      lci      uci
#>   <chr>       <dbl>    <dbl>    <dbl>    <dbl>
#> 1 k1       0.000563 0.000923 0.000804 0.00106 
#> 2 k2       0.000676 0.000912 0.000824 0.00101 
#> 3 k3       0.000833 0.000851 0.000772 0.000942
#> 4 k4       0.00115  0.000927 0.000800 0.00115 
#> 5 k5plus   0.00167  0.000755 0.000677 0.00188

Indirect adjusted rate

Events and population data

Let’s use the Selvin dataset, quoted by the epitools package (Selvin, 2004).

dth40 <- c(45, 201, 320, 670, 1126, 3160, 9723, 17935,
22179, 13461, 2238)

pop40 <- c(906897, 3794573, 10003544, 10629526, 9465330,
8249558, 7294330, 5022499, 2920220, 1019504, 142532)

The tidyrates present the same dataset in a tidy way.

selvin_data_1940
#> # A tibble: 22 × 3
#>    age_group name   value
#>    <chr>     <chr>  <dbl>
#>  1 <1        events    45
#>  2 1-4       events   201
#>  3 5-14      events   320
#>  4 15-24     events   670
#>  5 25-34     events  1126
#>  6 35-44     events  3160
#>  7 45-54     events  9723
#>  8 55-64     events 17935
#>  9 65-74     events 22179
#> 10 75-84     events 13461
#> # ℹ 12 more rows

Events and population reference data

dth60 <- c(141, 926, 1253, 1080, 1869, 4891, 14956, 30888,
41725, 26501, 5928)

pop60 <- c(1784033, 7065148, 15658730, 10482916, 9939972,
10563872, 9114202, 6850263, 4702482, 1874619, 330915)

The tidyrates present the same dataset in a tidy way.

selvin_data_1960
#> # A tibble: 22 × 3
#>    age_group name   value
#>    <chr>     <chr>  <dbl>
#>  1 <1        events   141
#>  2 1-4       events   926
#>  3 5-14      events  1253
#>  4 15-24     events  1080
#>  5 25-34     events  1869
#>  6 35-44     events  4891
#>  7 45-54     events 14956
#>  8 55-64     events 30888
#>  9 65-74     events 41725
#> 10 75-84     events 26501
#> # ℹ 12 more rows

Rate computation

To use the indirect adjustment procedure, tidyrate present the rate_adj_indirect function. The .data argument must be a tibble with the events and population data, and the .std argument must be also a tibble with the events and population data. The .keys argument must point to grouping variables on the .data tibble, if available.

The rate_adj_indirect will compute the crude rate, adjusted rate and exact confidence intervals for each group.

rate_adj_indirect(selvin_data_1940, selvin_data_1960)
#> # A tibble: 1 × 4
#>   crude.rate adj.rate     lci     uci
#>        <dbl>    <dbl>   <dbl>   <dbl>
#> 1    0.00120  0.00120 0.00119 0.00120