03-results-8-treatment.Rmd

\newpage
## Effectiveness & Economic
```{r, include=FALSE}
if (file.exists("output/r-treatment.rds")) {
    r_treatment <- readRDS("output/r-treatment.rds")
} else {
    source("src/r-treatment.R", local = knitr::knit_global())
    saveRDS(r_treatment, "output/r-treatment.rds")
}
```
<!--
The total observed loss rate over the three survey years was in 2018/19 `r round(r_treatment$year_loss$year_loss[1], 1)`% (95% CI: `r round(r_treatment$year_loss$loss_lower_ci[1], 1)`-`r round(r_treatment$year_loss$loss_upper_ci[1], 1)`%), in 2019/20 `r round(r_treatment$year_loss$year_loss[2], 1)`% (95% CI: `r round(r_treatment$year_loss$loss_lower_ci[2], 1)`-`r round(r_treatment$year_loss$loss_upper_ci[2], 1)`%) and in 2020/21 `r round(r_treatment$year_loss$year_loss[3], 1)`% (95% CI: `r round(r_treatment$year_loss$loss_lower_ci[3], 1)`-`r round(r_treatment$year_loss$loss_upper_ci[3], 1)`%) (Fig.&nbsp;\@ref(fig:efficient-economic), blue dashed vertical lines). Over the three survey years there were 17 different treatment method combinations (without drone brood removal) which had in sum at least 30 participants employing it (Table&nbsp;\@ref(tab:ee-table)). The most common treatment combination is long term formic acid in summer followed by oxalic acid trickling in winter (**A** 'SU-Fa-LT & WI-Ox-pure'). It showed in each year on mean a lower loss rate compared to the total loss rate but only in one year a low expenses category. In comparison the application of short term formic acid treatment in summer and oxalic acid sublimation in summer and winter (**K** 'SU-Fa-ST & SU/WI-Ox-Sub') was observed with a low loss rate category in each year and low expense category in two out of three years. A total of three combinations showed in all three years in mean a higher loss rate than the annual total loss rate, which are formic acid short term in summer followed by oxalic acid trickling pure or ready-made mixture in winter (**E** 'SU-Fa-ST & WI-Ox-pure'; **L** 'SU-Fa-ST & WI-Ox-mix') and the sole application of formic acid long term method in summer, without winter treatment, which also showed high expenses in each year (**M** 'SU-Fa-LT'). The three combinations with biotechnical method (without drone brood removal or hyperthermia) in summer showed low expenses in each year (**G** 'SU-Biot. & SU-Ox-sub & WI-Ox-sub'; **P** 'SU-Biot. & SU-Fa-LT & WI-Ox-pure'; **Q** 'SU-Biot. & SU-Fa-ST & WI-Ox-pure'). Additionally the treatment method of oxalic acid sublimation in summer and winter resulted in low mean expenses each year (**F** 'SU-Ox-sub & WI-Ox-sub') (Fig.&nbsp;\@ref(fig:efficient-economic)).
-->

The intermediate observed mean loss rate of the treatment combinations, which is the inner quantile range of 33-67%, was `r fr(r_treatment$quantile[1,2][[1]])`-`r fr(r_treatment$quantile[1,3][[1]])`% in 2018/19, `r fr(r_treatment$quantile[2,2][[1]])`-`r fr(r_treatment$quantile[2,3][[1]])`% in 2019/20 and `r fr(r_treatment$quantile[3,2][[1]])`-`r fr(r_treatment$quantile[3,3][[1]])`% in 2020/21 (Fig.&nbsp;\@ref(fig:efficient-economic), blue dashed vertical lines). The same range for the mean expenses of the combinations was EUR&nbsp;`r fr(r_treatment$quantile[1,4][[1]])`-`r fr(r_treatment$quantile[1,5][[1]])` in 2018/19, EUR&nbsp;`r fr(r_treatment$quantile[2,4][[1]])`-`r fr(r_treatment$quantile[2,5][[1]])` in 2019/20 and EUR&nbsp;`r fr(r_treatment$quantile[3,4][[1]])`-`r fr(r_treatment$quantile[3,5][[1]])` in 2020/21 (Fig.&nbsp;\@ref(fig:efficient-economic), blue dashed horizontal lines).

Over the three survey years there were 17 different treatment method combinations (without drone brood removal) which had in sum at least 30 participants employing it (Table&nbsp;\@ref(tab:ee-table)). The most common treatment combination was long term formic acid in summer followed by oxalic acid trickling in winter (**A** 'SU-Fa-LT & WI-Ox-pure'). It showed in one year an intermediate loss rate and in the first two years a lower loss rate compared to the intermediate range. As for expenses the combination was categorized in two years as high expense category and only in one year in the intermediate category. In comparison the same combination but with formic acid short term application (**E** 'SU-Fa-ST & WI-Ox-pure') showed a higher than intermediate loss rate in the last two years and an intermediate categorized loss rate in the first year. The expenses category was different in each year. The sole application of formic acid long term method in summer, without winter treatment, was categorized as high expense in each year and in the first two years the loss rate was higher than the intermediate range (**M** 'SU-Fa-LT').

Only two combinations were categorized in each year with low loss rates (**J** SU-Fa-ST & SU-Ox-sub & Wi-Ox-sub; **O** SU-Fa-ST & SU-Fa-LT & SU-Ox-sub & WI-Ox-sub). Both have in common that they use oxalic acid sublimation in addition to an formic acid treatment in summer and the use of oxalic acid sublimation as winter treatment. The first variant (**J** SU-Fa-ST & SU-Ox-sub & Wi-Ox-sub) did show intermediate to low expenses and the second variant (**O** SU-Fa-ST & SU-Fa-LT & SU-Ox-sub & WI-Ox-sub) was in the low expense category in the first year and in the high expense category the last two years. 

The treatment combination of oxalic acid sublimation in summer and winter (**F** SU-Ox-sub & WI-Ox-sub) was categorized as intermediate loss rate in the first two years and as high category in the last year. In addition the expenses were lower than the intermediate range in each year.

All three combinations with another biotechnical method (without drone brood removal or hyperthermia) in summer were categorized as low expenses in close to all survey years (**G** 'SU-Biot. & SU-Ox-sub & WI-Ox-sub'; **P** 'SU-Biot. & SU-Fa-LT & WI-Ox-pure'; **Q** 'SU-Biot. & SU-Fa-ST & WI-Ox-pure'). But loss rates did vary between the high and low category (Fig.&nbsp;\@ref(fig:efficient-economic)).


(ref:efficient-economic) Effectiveness and expense plot for treatment method combinations with at least 30 subsumed over the three survey years. Letters correspond to different treatment methods (Table \@ref(tab:ee-table)). Vertical axis shows the mean expense for each treatment combination with standard error (SE). Horizontal axis shows the mean observed winter colony loss rates for each treatment combination with 95% confidence interval (CI). The blue dashed line indicate the intermediate range (quantile 33-67%) of the treatment combination expenses (horizontal) and the observed winter colony loss rates (vertical) per year. Categorizing of treatment combinations is based on location in relation to the blue dashed lines (Fig.&nbsp;\@ref(fig:efficient-economic-mm)). The nine categories are colour coded as seen in the plot legend.

<!-- The blue dashed line indicate mean total expense (horizontal) and total observed winter colony loss rates (vertical) per year. Categorizing of treatment combinations based on location in relation to the blue dashed lines (Fig.&nbsp;\@ref(fig:efficient-economic-mm)). The four categories are colour coded as seen in the plot legend. -->

```{r efficient-economic, include=T, fig.cap="(ref:efficient-economic)"}
include_custom("output/figs/efficient-economic.png")
```

(ref:ee-table) Treatment combinations with at least 30 participants subsumed over the three survey years. Ordered by number of participants (#) applying the combinations. 'Effectiveness' and 'Expenses' are colour coded based on intermediate (grey), low (teal) and high (orange). Categorized based on their mean expense and mean loss rate compared to the other combinations (Fig.&nbsp;\@ref(fig:efficient-economic)). Abbreviations: SP/SU/WI = Spring/Summer/Winter || Fa-LT/ST = Formic acid - long/short term || Ox-sub/pure/mix = Oxalic acid - sublimation, pure or mixture || Biot = Another biotechnical method. 

```{r ee-table, include=T}
cn <- c(" ", "Treatment", "Year", "\\#", "Median", "Mean", "SD", "Mean", "Lower CI", "Upper CI", "Effectiveness", "Expenses")
cell_colors_price <- c("high-expense" = "cred!80", "low-expense" = "cgreen!80")
cell_colors_loss <- c("high-loss" = "cred!80", "low-loss" = "cgreen!80")

tab <- kbl(
    r_treatment$tab %>%
        select(-quantile_efficient, -quantile_economical) %>%
        mutate(
            c_short_od = str_replace_all(c_short_od, "&", " \\\\& "),
            across(letter:c_short_od, ~ ifelse(lag(as.character(.x), default = "0") == .x, "", .x)),
            across(observed_costs:observed_costs_sd, ~ ifelse(lag(.x, default = 0) == .x, "", round(.x, 2)))
        ) %>%
        mutate(
            letter = cell_spec(letter, bold = TRUE),
            efficient = cell_spec(
                efficient,
                background = ifelse(efficient != "", cell_colors_loss[efficient], "cwhite!80")
            ),
            economical = cell_spec(
                economical,
                background = ifelse(economical != "", cell_colors_price[economical], "cwhite!80")
            )
        ),
    linesep = c("", "", "\\addlinespace"),
    booktabs = TRUE,
    col.names = cn,
    digits = 1,
    align = c("c", "l", "c", rep("r", 6), "c", "c"),
    caption = "(ref:ee-table)",
    escape = FALSE
) %>%
    add_header_above(c(" " = 4, "Expenses [EUR]" = 3, "Observed Loss Rate [%]" = 3, " " = 2)) %>%
    # landscape()
    kable_styling(latex_option = "scale_down")
tab
rm(cn, c, tab)
```