How to conduct AMR analysis

Matthijs S. Berends

10 November 2019

Note: values on this page will change with every website update since they are based on randomly created values and the page was written in R Markdown. However, the methodology remains unchanged. This page was generated on 10 November 2019.

Introduction

For this tutorial, we will create fake demonstration data to work with.

You can skip to Cleaning the data if you already have your own data ready. If you start your analysis, try to make the structure of your data generally look like this:

date patient_id mo AMX CIP
2019-11-10 abcd Escherichia coli S S
2019-11-10 abcd Escherichia coli S R
2019-11-10 efgh Escherichia coli R S

Needed R packages

As with many uses in R, we need some additional packages for AMR analysis. Our package works closely together with the tidyverse packages dplyr and ggplot2 by Dr Hadley Wickham. The tidyverse tremendously improves the way we conduct data science - it allows for a very natural way of writing syntaxes and creating beautiful plots in R.

Our AMR package depends on these packages and even extends their use and functions.

Creation of data

We will create some fake example data to use for analysis. For antimicrobial resistance analysis, we need at least: a patient ID, name or code of a microorganism, a date and antimicrobial results (an antibiogram). It could also include a specimen type (e.g. to filter on blood or urine), the ward type (e.g. to filter on ICUs).

With additional columns (like a hospital name, the patients gender of even [well-defined] clinical properties) you can do a comparative analysis, as this tutorial will demonstrate too.

Patients

To start with patients, we need a unique list of patients.

patients <- unlist(lapply(LETTERS, paste0, 1:10))

The LETTERS object is available in R - it’s a vector with 26 characters: A to Z. The patients object we just created is now a vector of length 260, with values (patient IDs) varying from A1 to Z10. Now we we also set the gender of our patients, by putting the ID and the gender in a table:

patients_table <- data.frame(patient_id = patients,
                             gender = c(rep("M", 135),
                                        rep("F", 125)))

The first 135 patient IDs are now male, the other 125 are female.

Dates

Let’s pretend that our data consists of blood cultures isolates from between 1 January 2010 and 1 January 2018.

dates <- seq(as.Date("2010-01-01"), as.Date("2018-01-01"), by = "day")

This dates object now contains all days in our date range.

Microorganisms

For this tutorial, we will uses four different microorganisms: Escherichia coli, Staphylococcus aureus, Streptococcus pneumoniae, and Klebsiella pneumoniae:

bacteria <- c("Escherichia coli", "Staphylococcus aureus",
              "Streptococcus pneumoniae", "Klebsiella pneumoniae")

Other variables

For completeness, we can also add the hospital where the patients was admitted and we need to define valid antibmicrobial results for our randomisation:

hospitals <- c("Hospital A", "Hospital B", "Hospital C", "Hospital D")
ab_interpretations <- c("S", "I", "R")

Put everything together

Using the sample() function, we can randomly select items from all objects we defined earlier. To let our fake data reflect reality a bit, we will also approximately define the probabilities of bacteria and the antibiotic results with the prob parameter.

sample_size <- 20000
data <- data.frame(date = sample(dates, size = sample_size, replace = TRUE),
                   patient_id = sample(patients, size = sample_size, replace = TRUE),
                   hospital = sample(hospitals, size = sample_size, replace = TRUE,
                                     prob = c(0.30, 0.35, 0.15, 0.20)),
                   bacteria = sample(bacteria, size = sample_size, replace = TRUE,
                                     prob = c(0.50, 0.25, 0.15, 0.10)),
                   AMX = sample(ab_interpretations, size = sample_size, replace = TRUE,
                                 prob = c(0.60, 0.05, 0.35)),
                   AMC = sample(ab_interpretations, size = sample_size, replace = TRUE,
                                 prob = c(0.75, 0.10, 0.15)),
                   CIP = sample(ab_interpretations, size = sample_size, replace = TRUE,
                                 prob = c(0.80, 0.00, 0.20)),
                   GEN = sample(ab_interpretations, size = sample_size, replace = TRUE,
                                 prob = c(0.92, 0.00, 0.08)))

Using the left_join() function from the dplyr package, we can ‘map’ the gender to the patient ID using the patients_table object we created earlier:

The resulting data set contains 20,000 blood culture isolates. With the head() function we can preview the first 6 values of this data set:

head(data)
date patient_id hospital bacteria AMX AMC CIP GEN gender
2015-05-08 P3 Hospital A Escherichia coli R I S S F
2017-11-03 Y8 Hospital C Escherichia coli R S S S F
2013-09-06 U9 Hospital B Escherichia coli R S S S F
2015-11-16 E7 Hospital B Escherichia coli I S S S M
2011-04-18 F4 Hospital B Streptococcus pneumoniae S I S S M
2010-04-22 L4 Hospital D Escherichia coli S R S S M

Now, let’s start the cleaning and the analysis!

Cleaning the data

We also created a package dedicated to data cleaning and checking, called the cleaner package. It gets automatically installed with the AMR package. For its freq() function to create frequency tables, you don’t even need to load it yourself as it is available through the AMR package as well.

For example, for the gender variable:

data %>% freq(gender) # this would be the same: freq(data$gender)
# Frequency table 
# 
# Class:   factor (numeric)
# Length:  20,000 (of which NA: 0 = 0%)
# Levels:  2: F, M
# Unique:  2
# 
#      Item     Count   Percent   Cum. Count   Cum. Percent
# ---  -----  -------  --------  -----------  -------------
# 1    M       10,417    52.09%       10,417         52.09%
# 2    F        9,583    47.92%       20,000        100.00%

So, we can draw at least two conclusions immediately. From a data scientists perspective, the data looks clean: only values M and F. From a researchers perspective: there are slightly more men. Nothing we didn’t already know.

The data is already quite clean, but we still need to transform some variables. The bacteria column now consists of text, and we want to add more variables based on microbial IDs later on. So, we will transform this column to valid IDs. The mutate() function of the dplyr package makes this really easy:

data <- data %>%
  mutate(bacteria = as.mo(bacteria))

We also want to transform the antibiotics, because in real life data we don’t know if they are really clean. The as.rsi() function ensures reliability and reproducibility in these kind of variables. The mutate_at() will run the as.rsi() function on defined variables:

Finally, we will apply EUCAST rules on our antimicrobial results. In Europe, most medical microbiological laboratories already apply these rules. Our package features their latest insights on intrinsic resistance and exceptional phenotypes. Moreover, the eucast_rules() function can also apply additional rules, like forcing ampicillin = R when amoxicillin/clavulanic acid = R.

Because the amoxicillin (column AMX) and amoxicillin/clavulanic acid (column AMC) in our data were generated randomly, some rows will undoubtedly contain AMX = S and AMC = R, which is technically impossible. The eucast_rules() fixes this:

data <- eucast_rules(data, col_mo = "bacteria")
# 
# Rules by the European Committee on Antimicrobial Susceptibility Testing (EUCAST)
# http://eucast.org/
# 
# EUCAST Clinical Breakpoints (v9.0, 2019)
# Aerococcus sanguinicola (no changes)
# Aerococcus urinae (no changes)
# Anaerobic Gram-negatives (no changes)
# Anaerobic Gram-positives (no changes)
# Campylobacter coli (no changes)
# Campylobacter jejuni (no changes)
# Enterobacterales (Order) (no changes)
# Enterococcus (no changes)
# Haemophilus influenzae (no changes)
# Kingella kingae (no changes)
# Moraxella catarrhalis (no changes)
# Pasteurella multocida (no changes)
# Staphylococcus (no changes)
# Streptococcus groups A, B, C, G (no changes)
# Streptococcus pneumoniae (1,545 values changed)
# Viridans group streptococci (no changes)
# 
# EUCAST Expert Rules, Intrinsic Resistance and Exceptional Phenotypes (v3.1, 2016)
# Table 01: Intrinsic resistance in Enterobacteriaceae (1,309 values changed)
# Table 02: Intrinsic resistance in non-fermentative Gram-negative bacteria (no changes)
# Table 03: Intrinsic resistance in other Gram-negative bacteria (no changes)
# Table 04: Intrinsic resistance in Gram-positive bacteria (2,733 values changed)
# Table 08: Interpretive rules for B-lactam agents and Gram-positive cocci (no changes)
# Table 09: Interpretive rules for B-lactam agents and Gram-negative rods (no changes)
# Table 11: Interpretive rules for macrolides, lincosamides, and streptogramins (no changes)
# Table 12: Interpretive rules for aminoglycosides (no changes)
# Table 13: Interpretive rules for quinolones (no changes)
# 
# Other rules
# Non-EUCAST: amoxicillin/clav acid = S where ampicillin = S (2,194 values changed)
# Non-EUCAST: ampicillin = R where amoxicillin/clav acid = R (121 values changed)
# Non-EUCAST: piperacillin = R where piperacillin/tazobactam = R (no changes)
# Non-EUCAST: piperacillin/tazobactam = S where piperacillin = S (no changes)
# Non-EUCAST: trimethoprim = R where trimethoprim/sulfa = R (no changes)
# Non-EUCAST: trimethoprim/sulfa = S where trimethoprim = S (no changes)
# 
# --------------------------------------------------------------------------
# EUCAST rules affected 6,489 out of 20,000 rows, making a total of 7,902 edits
# => added 0 test results
# 
# => changed 7,902 test results
#    - 118 test results changed from S to I
#    - 4,776 test results changed from S to R
#    - 1,063 test results changed from I to S
#    - 318 test results changed from I to R
#    - 1,603 test results changed from R to S
#    - 24 test results changed from R to I
# --------------------------------------------------------------------------
# 
# Use eucast_rules(..., verbose = TRUE) (on your original data) to get a data.frame with all specified edits instead.

Adding new variables

Now that we have the microbial ID, we can add some taxonomic properties:

First isolates

We also need to know which isolates we can actually use for analysis.

To conduct an analysis of antimicrobial resistance, you must only include the first isolate of every patient per episode (Hindler et al., Clin Infect Dis. 2007). If you would not do this, you could easily get an overestimate or underestimate of the resistance of an antibiotic. Imagine that a patient was admitted with an MRSA and that it was found in 5 different blood cultures the following weeks (yes, some countries like the Netherlands have these blood drawing policies). The resistance percentage of oxacillin of all isolates would be overestimated, because you included this MRSA more than once. It would clearly be selection bias.

The Clinical and Laboratory Standards Institute (CLSI) appoints this as follows:

(…) When preparing a cumulative antibiogram to guide clinical decisions about empirical antimicrobial therapy of initial infections, only the first isolate of a given species per patient, per analysis period (eg, one year) should be included, irrespective of body site, antimicrobial susceptibility profile, or other phenotypical characteristics (eg, biotype). The first isolate is easily identified, and cumulative antimicrobial susceptibility test data prepared using the first isolate are generally comparable to cumulative antimicrobial susceptibility test data calculated by other methods, providing duplicate isolates are excluded.
M39-A4 Analysis and Presentation of Cumulative Antimicrobial Susceptibility Test Data, 4th Edition. CLSI, 2014. Chapter 6.4

This AMR package includes this methodology with the first_isolate() function. It adopts the episode of a year (can be changed by user) and it starts counting days after every selected isolate. This new variable can easily be added to our data:

So only 28.5% is suitable for resistance analysis! We can now filter on it with the filter() function, also from the dplyr package:

For future use, the above two syntaxes can be shortened with the filter_first_isolate() function:

First weighted isolates

We made a slight twist to the CLSI algorithm, to take into account the antimicrobial susceptibility profile. Have a look at all isolates of patient P1, sorted on date:

isolate date patient_id bacteria AMX AMC CIP GEN first
1 2010-01-26 P1 B_ESCHR_COLI I S S R TRUE
2 2010-04-19 P1 B_ESCHR_COLI S S S S FALSE
3 2010-04-24 P1 B_ESCHR_COLI R R R S FALSE
4 2010-06-11 P1 B_ESCHR_COLI S S S R FALSE
5 2010-11-24 P1 B_ESCHR_COLI S S S S FALSE
6 2010-12-11 P1 B_ESCHR_COLI R S S S FALSE
7 2010-12-23 P1 B_ESCHR_COLI S S R S FALSE
8 2011-01-14 P1 B_ESCHR_COLI R S S S FALSE
9 2011-01-19 P1 B_ESCHR_COLI R R S S FALSE
10 2011-01-26 P1 B_ESCHR_COLI R I S S TRUE

Only 2 isolates are marked as ‘first’ according to CLSI guideline. But when reviewing the antibiogram, it is obvious that some isolates are absolutely different strains and should be included too. This is why we weigh isolates, based on their antibiogram. The key_antibiotics() function adds a vector with 18 key antibiotics: 6 broad spectrum ones, 6 small spectrum for Gram negatives and 6 small spectrum for Gram positives. These can be defined by the user.

If a column exists with a name like ‘key(…)ab’ the first_isolate() function will automatically use it and determine the first weighted isolates. Mind the NOTEs in below output:

isolate date patient_id bacteria AMX AMC CIP GEN first first_weighted
1 2010-01-26 P1 B_ESCHR_COLI I S S R TRUE TRUE
2 2010-04-19 P1 B_ESCHR_COLI S S S S FALSE TRUE
3 2010-04-24 P1 B_ESCHR_COLI R R R S FALSE TRUE
4 2010-06-11 P1 B_ESCHR_COLI S S S R FALSE TRUE
5 2010-11-24 P1 B_ESCHR_COLI S S S S FALSE TRUE
6 2010-12-11 P1 B_ESCHR_COLI R S S S FALSE TRUE
7 2010-12-23 P1 B_ESCHR_COLI S S R S FALSE TRUE
8 2011-01-14 P1 B_ESCHR_COLI R S S S FALSE TRUE
9 2011-01-19 P1 B_ESCHR_COLI R R S S FALSE TRUE
10 2011-01-26 P1 B_ESCHR_COLI R I S S TRUE TRUE

Instead of 2, now 10 isolates are flagged. In total, 76.2% of all isolates are marked ‘first weighted’ - 47.7% more than when using the CLSI guideline. In real life, this novel algorithm will yield 5-10% more isolates than the classic CLSI guideline.

As with filter_first_isolate(), there’s a shortcut for this new algorithm too:

So we end up with 15,241 isolates for analysis.

We can remove unneeded columns:

Now our data looks like:

head(data_1st)
date patient_id hospital bacteria AMX AMC CIP GEN gender gramstain genus species first_weighted
1 2015-05-08 P3 Hospital A B_ESCHR_COLI R I S S F Gram-negative Escherichia coli TRUE
3 2013-09-06 U9 Hospital B B_ESCHR_COLI R S S S F Gram-negative Escherichia coli TRUE
5 2011-04-18 F4 Hospital B B_STRPT_PNMN S S S R M Gram-positive Streptococcus pneumoniae TRUE
7 2016-10-10 S10 Hospital D B_STPHY_AURS S S S S F Gram-positive Staphylococcus aureus TRUE
8 2010-01-21 R3 Hospital B B_STRPT_PNMN S S R R F Gram-positive Streptococcus pneumoniae TRUE
9 2011-11-23 B7 Hospital B B_STRPT_PNMN R R S R M Gram-positive Streptococcus pneumoniae TRUE

Time for the analysis!

Analysing the data

You might want to start by getting an idea of how the data is distributed. It’s an important start, because it also decides how you will continue your analysis. Although this package contains a convenient function to make frequency tables, exploratory data analysis (EDA) is not the primary scope of this package. Use a package like DataExplorer for that, or read the free online book Exploratory Data Analysis with R by Roger D. Peng.

Dispersion of species

To just get an idea how the species are distributed, create a frequency table with our freq() function. We created the genus and species column earlier based on the microbial ID. With paste(), we can concatenate them together.

The freq() function can be used like the base R language was intended:

freq(paste(data_1st$genus, data_1st$species))

Or can be used like the dplyr way, which is easier readable:

data_1st %>% freq(genus, species)

Frequency table

Class: character
Length: 15,241 (of which NA: 0 = 0%)
Unique: 4

Shortest: 16
Longest: 24

Item Count Percent Cum. Count Cum. Percent
1 Escherichia coli 7,593 49.82% 7,593 49.82%
2 Staphylococcus aureus 3,734 24.50% 11,327 74.32%
3 Streptococcus pneumoniae 2,327 15.27% 13,654 89.59%
4 Klebsiella pneumoniae 1,587 10.41% 15,241 100.00%

Resistance percentages

The functions resistance() and susceptibility() can be used to calculate antimicrobial resistance or susceptibility. For more specific analyses, the functions proportion_S(), proportion_SI(), proportion_I(), proportion_IR() and proportion_R() can be used to determine the proportion of a specific antimicrobial outcome.

As per the EUCAST guideline of 2019, we calculate resistance as the proportion of R (proportion_R(), equal to resistance()) and susceptibility as the proportion of S and I (proportion_SI(), equal to susceptibility()). These functions can be used on their own:

Or can be used in conjuction with group_by() and summarise(), both from the dplyr package:

data_1st %>% 
  group_by(hospital) %>% 
  summarise(amoxicillin = resistance(AMX))
hospital amoxicillin
Hospital A 0.4637681
Hospital B 0.4776811
Hospital C 0.4811697
Hospital D 0.4694820

Of course it would be very convenient to know the number of isolates responsible for the percentages. For that purpose the n_rsi() can be used, which works exactly like n_distinct() from the dplyr package. It counts all isolates available for every group (i.e. values S, I or R):

data_1st %>% 
  group_by(hospital) %>% 
  summarise(amoxicillin = resistance(AMX),
            available = n_rsi(AMX))
hospital amoxicillin available
Hospital A 0.4637681 4554
Hospital B 0.4776811 5399
Hospital C 0.4811697 2257
Hospital D 0.4694820 3031

These functions can also be used to get the proportion of multiple antibiotics, to calculate empiric susceptibility of combination therapies very easily:

genus amoxiclav gentamicin amoxiclav_genta
Escherichia 0.9212433 0.8984591 0.9927565
Klebsiella 0.8298677 0.8897290 0.9836169
Staphylococcus 0.9290305 0.9258168 0.9946438
Streptococcus 0.6067899 0.0000000 0.6067899

To make a transition to the next part, let’s see how this difference could be plotted:

Plots

To show results in plots, most R users would nowadays use the ggplot2 package. This package lets you create plots in layers. You can read more about it on their website. A quick example would look like these syntaxes:

The AMR package contains functions to extend this ggplot2 package, for example geom_rsi(). It automatically transforms data with count_df() or proportion_df() and show results in stacked bars. Its simplest and shortest example:

ggplot(data_1st) +
  geom_rsi(translate_ab = FALSE)

Omit the translate_ab = FALSE to have the antibiotic codes (AMX, AMC, CIP, GEN) translated to official WHO names (amoxicillin, amoxicillin/clavulanic acid, ciprofloxacin, gentamicin).

If we group on e.g. the genus column and add some additional functions from our package, we can create this:

To simplify this, we also created the ggplot_rsi() function, which combines almost all above functions:

Independence test

The next example uses the included example_isolates, which is an anonymised data set containing 2,000 microbial blood culture isolates with their full antibiograms found in septic patients in 4 different hospitals in the Netherlands, between 2001 and 2017. This data.frame can be used to practice AMR analysis.

We will compare the resistance to fosfomycin (column FOS) in hospital A and D. The input for the fisher.test() can be retrieved with a transformation like this:

check_FOS <- example_isolates %>%
  filter(hospital_id %in% c("A", "D")) %>% # filter on only hospitals A and D
  select(hospital_id, FOS) %>%             # select the hospitals and fosfomycin
  group_by(hospital_id) %>%                # group on the hospitals
  count_df(combine_SI = TRUE) %>%          # count all isolates per group (hospital_id)
  tidyr::spread(hospital_id, value) %>%    # transform output so A and D are columns
  select(A, D) %>%                         # and select these only
  as.matrix()                              # transform to good old matrix for fisher.test()

check_FOS
#       A  D
# [1,] 25 77
# [2,] 24 33

We can apply the test now with:

As can be seen, the p value is 0.031, which means that the fosfomycin resistances found in hospital A and D are really different.