04: Generating Estimates: Age-standardization

Overview

In the previous vignette, we discussed the model setup process in-depth. But how do we get our estimates once we’ve run our model? In this vignette, we discuss extracting estimates from our model object with the get_estimates() function, and how to age-standardize those estimates with age_standardize().

The get_estimates() function

In the RSTr introductory vignette, we generated age-standardized estimates for lambda based on our example Michigan dataset. To extract rates from an RSTr object, we can simply run get_estimates():

mod_mst <- mstcar(name = "my_test_model", data = miheart, adjacency = miadj)
#> Starting sampler on Batch 1 at Fri Dec 19 00:08:27

#> Generating estimates...
#> Model finished at Fri Dec 19 00:08:53

estimates <- get_estimates(mod_mst, rates_per = 1e5)
head(estimates)
#>   county group year  medians ci_lower ci_upper rel_prec events population
#> 1  26001 35-44 1979 45.07149 31.48676 59.17320 1.627926      1        964
#> 2  26003 35-44 1979 56.47176 35.01048 85.83469 1.111119      1       1011
#> 3  26005 35-44 1979 24.35092 17.47599 30.61988 1.852641      0       9110
#> 4  26007 35-44 1979 35.13386 18.28414 48.08389 1.178999      0       3650
#> 5  26009 35-44 1979 34.24613 19.40691 45.97550 1.288970      0       1763
#> 6  26011 35-44 1979 46.13104 28.16870 65.56977 1.233415      0       1470

The age_standardization() function

In many cases, we will want to age-standardize our estimates based on some (or all) age groups in our dataset. In our Michigan dataset, we have six ten-year age groups over which we can standardize; let’s age-standardize from ages 35-64. For RSTr objects, age_standardize() takes in four arguments:

• RSTr_obj: The RSTr model object created with *car();

• std_pop: A vector of standard populations associated with the age groups of interest. Since our Michigan data is from 1979-1988, we can use 1980 standard populations from NIH. It is recommended that you use the standard population that is most closely associated with your dataset;

• new_name: The name of your new standard population group; and

• groups: A vector of names matching each group of interest. To age-standardize by all groups in a dataset, leave this argument blank.

Once we have our std_pop vector, we can age-standardize our estimates:

std_pop <- c(113154, 100640, 95799)
mod_mst <- age_standardize(mod_mst, std_pop, new_name = "35-64", groups = c("35-44", "45-54", "55-64"))
mod_mst
#> RSTr object:
#> 
#> Model name: my_test_model 
#> Model type: MSTCAR 
#> Data likelihood: binomial 
#> Estimate Credible Interval: 95% 
#> Number of geographic units: 83 
#> Number of samples: 6000 
#> Estimates age-standardized: Yes 
#> Age-standardized groups: 35-64 
#> Estimates suppressed: No

Notice now that the mod_mst object indicates we have age-standardized our estimates and the names of our age-standardized group. We can also add on to our list of age-standardized estimates by simply specifying a different group:

std_pop <- c(68775, 34116, 9888)
mod_mst <- age_standardize(mod_mst, std_pop, new_name = "65up", groups = c("65-74", "75-84", "85+"))
mod_mst
#> RSTr object:
#> 
#> Model name: my_test_model 
#> Model type: MSTCAR 
#> Data likelihood: binomial 
#> Estimate Credible Interval: 95% 
#> Number of geographic units: 83 
#> Number of samples: 6000 
#> Estimates age-standardized: Yes 
#> Age-standardized groups: 35-64 65up 
#> Estimates suppressed: No

If we want to generate estimates for all groups, i.e. 35 and up, we can omit the groups argument and expand std_pop to include all of our populations:

std_pop <- c(113154, 100640, 95799, 68775, 34116, 9888)
mod_mst <- age_standardize(mod_mst, std_pop, new_name = "35up")
mod_mst
#> RSTr object:
#> 
#> Model name: my_test_model 
#> Model type: MSTCAR 
#> Data likelihood: binomial 
#> Estimate Credible Interval: 95% 
#> Number of geographic units: 83 
#> Number of samples: 6000 
#> Estimates age-standardized: Yes 
#> Age-standardized groups: 35-64 65up 35up 
#> Estimates suppressed: No
mst_estimates_as <- get_estimates(mod_mst)
head(mst_estimates_as)
#>   county group year  medians ci_lower ci_upper rel_prec events population
#> 1  26001 35-64 1979 203.3206 157.7320 241.9712 2.413612      7       3353
#> 2  26003 35-64 1979 290.8983 223.9136 384.6817 1.809428     12       3105
#> 3  26005 35-64 1979 124.6186 103.7021 142.7059 3.195035     27      23926
#> 4  26007 35-64 1979 174.7536 138.9969 204.4177 2.671222     15      10000
#> 5  26009 35-64 1979 171.6994 134.1164 209.4447 2.279347     11       5152
#> 6  26011 35-64 1979 223.7704 183.4417 276.1411 2.413933      8       4517

Now, get_estimates(mod_mst) shows the age-standardized estimates as opposed to our non-standardized estimates. Should you want to see the non-standardized estimates instead, you can set the argument standardized = FALSE.

Final thoughts

In this vignette, we explored the get_estimates() function and investigated age-standardization with the age_standardize() function. Age-standardization is one of the most important features of the RSTr package; using just a few arguments, we can easily generate estimates across our population groups.