--- title: "Using chlorpromazineR to calculate chlorpromazine-equivalent doses" author: "Eric Brown, Parita Shah, Julia Kim" date: "`r Sys.Date()`" output: rmarkdown::html_vignette vignette: > %\VignetteIndexEntry{Using chlorpromazineR to calculate chlorpromazine-equivalent doses} %\VignetteEngine{knitr::rmarkdown} %\VignetteEncoding{UTF-8} --- ```{r setup, include = FALSE} library(chlorpromazineR) knitr::opts_chunk$set( collapse = TRUE, comment = "#>" ) ``` ## Background Studies investigating or controlling for the impact of antipsychotic medications often need to quantify the amount of medication to which an individual is or has been exposed. As different antipsychotics have different potencies, the task is more complicated than using each medication's daily dosage in milligrams, for example. Further complicating the matter is that antipsychotic medications have different formulations including oral medications, and injectable formulations that are either short- or long-acting (also called depots). A commonly used strategy to account for the differing potencies is to calculate the equivalent dose of each medication in terms of a reference medication. Chlorpromazine (CPZ) is an antipsychotic medication most often used for this purpose. A CPZ-equivalent dose is typically defined as a dose of antipsychotic that is comparable to 100 mg of CPZ.^[Davis, J. M. (1974). Dose equivalence of the antipsychotic drugs. Journal of Psychiatric Research, 1, 65–69.] The total daily dose of a medication expressed in milligrams of CPZ per day is the daily-dose equivalent, and is commonly utilized in both clinical and research settings.^[Danivas, V., & Venkatasubramanian, G. (2013). Current perspectives on chlorpromazine equivalents: Comparing apples and oranges! Indian Journal of Psychiatry, 55(2), 207–208. ] Both strategies allow the dose of one medication, such as quetiapine, to be expressed in equivalents of chlorpromazine. There are various approaches to defining dose equivalents, including the "defined daily doses" (Methodology WCCfDS, 2013), and Delphie methods.^[Dalkey, N. (1969). The Delphi Method: An Experimental Study of Group Opinion | RAND. Retrieved from ] Defined daily doses, which are produced by the World Health Organization, provide an estimate of the average maintenance antipsychotic daily dose required for a 70-kg adult with psychosis.^[Leucht, S., Samara, M., Heres, S., & Davis, J. M. (2016). Dose Equivalents for Antipsychotic Drugs: The DDD Method. Schizophrenia Bulletin, 42(suppl_1), S90–S94. ] On the other hand, the Delphi method is an approach to reach a consensus to antipsychotic dose equivalents based on a two-stage, questionnaire-based survey of experts.^[Hasson, F., Keeney, S., & McKenna, H. (2000). Research guidelines for the Delphi survey technique. Journal of Advanced Nursing, 32(4), 1008–1015.] Using the doses defined through these methods, equivalency ratio for each antipsychotic can be calculated, with CPZ as a reference medication (i.e. x mg/day of antipsychotic A is equal to 1 mg/day CPZ). While there is no single standard method to defining dose equivalence, the International Consensus Study by Gardner et al. (2010), which used the Delphi method,^[Gardner, D. M., Murphy, A. L., O’Donnell, H., Centorrino, F., & Baldessarini, R. J. (2010). International consensus study of antipsychotic dosing. The American Journal of Psychiatry, 167(6), 686–693. ] is a widely used resource to calculate equivalency ratios. ## chlorpromazineR We present here an easy-to-use R package to calculate CPZ daily dose equivalents for common oral and injectable antipsychotic medications. We do not propose to suggest which conversion factors are appropriate to use, or how to interpret the converted data. All users should also refer to the papers from which the conversion factor data originates to determine whether the use of such data is appropriate for their study. With this package, we hope: * to improve transparency and consistency in calculating CPZ equivalents, * to reduce human error and improve accuracy, * and to simplify workflows for large datasets, as from chart reviews of electronic health records. We use the data from the International Consensus Study (Gardner et al. 2010) as the default conversion "key". We also include keys generated from the most widely cited papers on CPZ equivalence. The user can also create their own keys (and contribute them to this package), which are `list` objects as described below. While the default key is `gardner2010`, derived from the 2010 paper by Gardner et al, no equivalency was provided for short acting vs. oral chlorpromazine, so `gardner2010` only includes oral and long acting injectable conversion factors. However, the SAI equivalencies are available in terms of chlorpromazine SAI equivalents, so to use that data knowing this limitation, the key `gardner2010_withsai` is available if needed. The SAI values calculated could then be converted using an external reference value). ### Input data structure The data is assumed to be stored in a `data.frame` where each row represents a participant, and with the required variables stored in columns. The two variables always required are the antipsychotic names and the dose to convert. The generic name of the medication must be used. There are differing conventions for drug name abbreviations, and different brand names around the world, so brand names are not built into the package. For oral medications or short-acting injectables, the dose is converted directly, i.e. not accounting for dosing frequencies, which must be done separately. The exception is for long-acting injectables (LAIs), which converts to daily equivalent oral CPZ dose by dividing by the frequency of administration in days. Thus when using LAIs, a third variable, the administration frequency in days, must be available, unless the stored doses have already been divided by the administration frequency. ### Key data structure The conversion factors are stored in "keys". A key is a `list` object containing 3 named sub-`lists`: `oral`, `sai` and `lai`. ```{r} names(gardner2010) ``` Each sub-`list` is itself a named list where the names are the antipsychotic names, and the values are the conversion factors. ```{r} names(gardner2010$oral)[1:5] ``` The conversion factor is what a given dose needs to be multiplied by to get the CPZ-equivalent. ```{r} gardner2010$oral$aripiprazole ``` For example, with aripiprazole, the value is 20, meaning that the CPZ-equivalent dose of aripiprazole 5 mg is `5 * 20 = 100`. ## Tutorial ### Example for one route only To demonstrate with a simple example using only oral medications, consider the following data: ```{r} participant_ID <- c("P01", "P02", "P03", "P04") age <- c(42, 29, 30, 60) antipsychotic <- c("olanzapine", "olanzapine", "quetiapine", "ziprasidone") dose <- c(10, 12.5, 300, 60) example_oral <- data.frame(participant_ID, age, antipsychotic, dose, stringsAsFactors = FALSE) example_oral ``` In this case we have `example_oral`, a `data.frame`, with one row per participant, and a variable for the antipsychotic name, labeled `antipsychotic` and a variable for the dose, named `dose`. To calculate CPZ-equivalent doses, use the `to_cpz()` function: ```{r} to_cpz(example_oral, ap_label = "antipsychotic", dose_label = "dose", route = "oral") ``` Notice that 2 new columns appear, which are named based on the default arguments in `to_cpz()`. The CPZ-equivalent doses are in the column labeled `cpz_eq`. ### Example for data with oral and injectable routes The default key from the study by Gardner et al. included dosing equivalents for oral medications, short acting injectables and long-acting injectables. If the original data includes more than one of these routes, the `route` argument can be set to "mixed". In this case, more information is required to calculate the equivalent doses. Specifically, a variable to indicate the route ("oral", "sai", or "lai") is required, and if long-acting injectables are used, then the injection frequency is also required.* *If the doses have already manually been divided by the frequency, then setting the `q` parameter to `1` will forgo the need to specify a variable storing the frequency information. Here is some example data: ```{r} example_lai <- example_oral example_lai$participant_ID <- c("P05", "P06", "P07", "P08") example_lai$antipsychotic <- c("zuclopenthixol decanoate", "zuclopenthixol decanoate", "perphenazine enanthate", "fluspirilene") example_lai$q <- c(14, 21, 14, 14) example_lai$dose <- c(200, 200, 50, 6) example_sai <- example_oral example_sai$participant_ID <- c("P09", "P10", "P11", "P12") example_sai$antipsychotic <- c("Chlorpromazine HCl", "Loxapine HCl", "Olanzapine tartrate", "Olanzapine tartrate") example_sai$dose <- c(100, 50, 10, 5) example_oral$q <- NA example_sai$q <- NA example_oral$route <- "oral" example_lai$route <- "lai" example_sai$route <- "sai" example_mixed <- rbind(example_oral, example_sai, example_lai) example_mixed ``` The function `to_cpz()` is used similarly as above, with the extra variable names (route_label, q) specified: ```{r} to_cpz(example_mixed, ap_label = "antipsychotic", dose_label = "dose", route = "mixed", route_label = "route", q_label = "q", key=gardner2010_withsai) ``` ### Participants on multiple medications If the included participants are on multiple medications, then the name and dose of the antipsychotic should be specified as separate variables in the `data.frame`. In this case, `to_cpz()` can be used multiple times, each time specifying the appropriate variables. Run it first specifying the name and dose of the first medication, and then run it again specifying the name and dose of the second medication. Be sure to change the name of the `eq_label` where the result will be stored. ## Checking for inclusion and matching of antipsychotic names The software depends on an exact match (except for case) of the generic names of antipsychotics. To check if there are rows that do not match, `check_ap()` can be used. Here is an example of data where all the antipsychotic names match the key, showing that 0 is returned (the number of mismatches): ```{r} check_ap(example_oral, ap_label = "antipsychotic", route = "oral", key=gardner2010) ``` When there are mismatches, they are listed, and the number is returned: ```{r} example_sai <- example_sai check_ap(example_sai, ap_label = "antipsychotic", route = "sai", key=gardner2010) ``` As with `to_cpz()`, mixed routes can be used with `check_ap()`: ```{r} example_mixed_bad <- example_mixed example_mixed_bad[1,3] <- "olanzzzzapine" example_mixed_bad[4,3] <- "Seroquel" example_mixed_bad[5,3] <- "chlorpromazineHCl" example_mixed_bad[9,3] <- "zuclo decanoate" check_ap(example_mixed_bad, ap_label = "antipsychotic", route = "mixed", route_label = "route", key=gardner2010_withsai) ``` While the default `gardner2010` key uses the full antipsychotic name for SAI and LAI formulations (e.g. Chlorpromazine HCl) in order to be consistent with the publication and reduce the risk of errors due to ambiguity, the use can modify the key to "trim" the antipsychotic names to just 1 word (e.g. chlorpromazine). ```{r} names(gardner2010$lai) ``` To trim to 1-word antipsychotic names, we have made the `trim_key()` function, which takes a key and generates a new key from it with the 1-word antipsychotic labels. If you want to use the `gardner2010` key but want 1-word antipsychotic names, then you can use `trim_key` to generate that new key, which you can then use in `to_cpz()`. For example: ```{r} trimmed_gardner <- trim_key(gardner2010) names(trimmed_gardner$lai) ``` ## Using other keys The above examples have used the default CPZ conversion key, `gardner2010`, based on data extracted from Gardner et al. 2010. However, not all antipsychotic medications were included in that study. This package also includes a key extracted from the data included in the Leucht et al. 2016 publication based on the World Health Organization's Defined Daily Doses. To explicitly change the CPZ key, use the key parameter when calling `to_cpz()`. ```{r} to_cpz(example_oral, ap_label = "antipsychotic", dose_label = "dose", route = "oral", key = leucht2016) ``` However, the authors do not recommend using the DDD values when more scientific values are available. Therefore, it can be desirable to create a new key with data from the `gardner2010` key (or your own created key) combined with the `leucht2016` key, using data from `gardner2010` key where possible. For this purpose, we have created the `add_key()` function, which takes a preferred base key from which ALL available values are taken, and adds only the missing antipsychotic conversion values from the second key to be added. Further, there is a `trim` parameter, which invokes the `trim_key()` function on both keys, in the case as in `leucht2016` where only 1-word antipsychotic names are provided. ```{r} gardner_leucht <- add_key(base = gardner2010, add = leucht2016, trim = TRUE) names(gardner_leucht$sai) ``` Let's change a value in the `example_oral` data so that there is an antipsychotic that is not listed in the `gardner2010` key: asenapine. ```{r} other_oral <- example_oral other_oral[2,3] <- "asenapine" other_oral[2,4] <- 10 check_ap(other_oral, ap_label = "antipsychotic", route = "oral") ``` We can see that asenapine is not in the default key. We could use the `leucht2016` key: ```{r} to_cpz(other_oral, ap_label = "antipsychotic", dose_label = "dose", route = "oral", key = gardner_leucht) ``` Note that it used the `gardner2010` values where they were available (olanzapine, quetiapine, ziprasidone) and the `leucht2016` key for asenapine. To draw from values from additional keys (e.g. user-created keys), the `add_key()` function could be used iteratively. The first argument of `add_key()`, `base`, is always the key with values that take priority. ## List of included keys Use the built in help function to see the full reference for any key, e.g. `help(davis1974)`. The data may need to be loaded prior to using the help function (e.g. load via `chlorpromazineR::davis1974`, then use `help(davis1974)`. | Key name | Reference | Description | |----------|-----------|-------------| |davis1974 ||Classic paper on chlorpromazine equivalent doses.| |gardner2010||Consensus study.| |gardner2010_withsai||The included SAI data result in equivalent to SAI (parenteral) chlorpromazine, not oral.| |leucht2016||WHO DDD method.| |leucht2020||Dose-response meta-analysis.| |woods2003||| ## Converting to equivalents of an arbitrary antipsychotic Historically, CPZ has been used as the reference antipsychotic. However, using the same key data, any arbitrary reference antipsychotic could be used. Similar to `to_cpz()`, `to_ap()` can perform this conversion. The desired reference antipsychotic can be specified, e.g. with the parameters `convert_to_ap = "olanzapine", convert_to_route = "oral"`. ```{r} to_ap(other_oral, convert_to_ap = "olanzapine", convert_to_route = "oral", ap_label = "antipsychotic", dose_label = "dose", route = "oral", key = gardner_leucht) ```