Restate your problem in two different ways; the first using GNU Datamash, and the second with a awk.

The example file I took it from is this one I called suicidios_final.csv:

country,year,sex,age,suicides_no,population,suicides/100k pop,country-year,HDI for year, gdp_for_year ($) ,gdp_per_capita ($),generation
Albania,1987,male,15-24 years,21,312900,6.71,Albania1987,,2156624900,796,Generation X
Albania,1988,male,35-54 years,16,308000,5.19,Albania1987,,2156624900,796,Silent
Albania,1988,female,15-24 years,14,289700,4.83,Albania1987,,2156624900,796,Generation X
Albania,1987,male,75+ years,1,21800,4.59,Albania1987,,2156624900,796,G.I. Generation
Albania,1987,male,25-34 years,9,274300,3.28,Albania1987,,2156624900,796,Boomers
Albania,1987,female,75+ years,1,35600,2.81,Albania1987,,2156624900,796,G.I. Generation
Albania,1989,female,35-54 years,6,278800,2.15,Albania1987,,2156624900,796,Silent
Albania,1987,female,25-34 years,4,257200,1.56,Albania1987,,2156624900,796,Boomers
Albania,1987,male,55-74 years,1,137500,0.73,Albania1987,,2156624900,796,G.I. Generation
México,1987,female,5-14 years,0,311000,0,México1987,,2156624900,796,Generation X
México,1989,female,55-74 years,0,144600,0,México1987,,2156624900,796,G.I. Generation
México,1989,male,5-14 years,0,338200,0,México1987,,2156624900,796,Generation X
México,1988,female,75+ years,2,36400,5.49,México1988,,2126000000,769,G.I. Generation
México,1988,male,15-24 years,17,319200,5.33,México1988,,2126000000,769,Generation X
México,1988,male,15-24 years,17,319200,5.33,México1988,,2126000000,769,Generation X
Colombia,1988,male,15-24 years,17,319200,5.33,Colombia1988,,2126000000,769,Generation X
Colombia,1987,female,5-14 years,0,311000,0,Colombia1987,,2156624900,796,Generation X
Colombia,1967,female,55-74 years,0,144600,0,Colombia1987,,2156624900,796,G.I. Generation
Colombia,1957,male,5-14 years,0,338200,0,Colombia1987,,2156624900,796,Generation X
Colombia,1988,female,75+ years,2,36400,5.49,Colombia1988,,2126000000,769,G.I. Generation
Colombia,1988,male,15-24 years,17,319200,5.33,Colombia1988,,2126000000,769,Generation X
Colombia,1988,male,15-24 years,17,319200,5.33,Colombia1988,,2126000000,769,Generation X
Colombia,1988,male,15-24 years,17,319200,5.33,Colombia1988,,2126000000,769,Generation X

Withdatamash

In a single line:

$ datamash --sort -t , -H -g 1,2,3 sum 5 < suicidios_final.csv | column -t -s ,

Where I ask datamash to group by fields 1,2,3, and then add field 5 using the "," character as a separator.

Resulting in:

GroupBy(country)  GroupBy(year)  GroupBy(sex)  sum(suicides_no)
Albania           1987           female        5
Albania           1987           male          32
Albania           1988           female        14
Albania           1988           male          16
Albania           1989           female        6
Colombia          1957           male          0
Colombia          1967           female        0
Colombia          1987           female        0
Colombia          1988           female        2
Colombia          1988           male          68
México            1987           female        0
México            1988           female        2
México            1988           male          34
México            1989           female        0
México            1989           male          0

If you don't have datamash, install it with sudo apt install datamash.

Wearingawk

Here I just mixed this script that comes in the official documentation to display ("walk" through) the content of a multidimensional array.

Then I used the respective fields assigning them as keys of the array. Thus, awkit is in charge of doing all the work, since there cannot be repeated keys in an array, so the grouping is done automatically due to the nature of the array keys, and the sum is only indicated with the operator +=on the fifth field.

In a file called main.awk, we put the following content:

#!/usr/bin/awk

# 1. Country
# 2. Year
# 3. Sex
# 4. Age
# 5. Suicides No

function walk_array(arr, name, i) {
    for (i in arr) {
        if (isarray(arr[i]))
            walk_array(arr[i], (name "[" i "]"))
        else
            printf("%s[%s] = %s\n", name, i, arr[i])
    }
}
# Aquí nos saltamos el primer renglón
NR!=1{
    reporte[$1][$2][$3]+=$5
}
END {       
    walk_array(reporte,"Reporte: ")
}

And in the terminal we run:

$ awk -f main.awk -F , suicidios_final.csv

Getting:

Reporte: [México][1987][female] = 0
Reporte: [México][1988][male] = 34
Reporte: [México][1988][female] = 2
Reporte: [México][1989][male] = 0
Reporte: [México][1989][female] = 0
Reporte: [Albania][1987][female] = 5
Reporte: [Albania][1987][male] = 32
Reporte: [Albania][1988][female] = 14
Reporte: [Albania][1988][male] = 16
Reporte: [Albania][1989][female] = 6
Reporte: [Colombia][1957][male] = 0
Reporte: [Colombia][1967][female] = 0
Reporte: [Colombia][1987][female] = 0
Reporte: [Colombia][1988][female] = 2
Reporte: [Colombia][1988][male] = 68

Note:

Using a large mix of GNU/Linux tools can be tempting in the first few years, but it's highly inefficient (and highly unattractive), as each program opens file descriptors, then closes them, they can create temporary programs that you later delete. Keep in mind that Bash cannot be used lightly as if it were a programming language, but rather a great tool for orchestrating programs.

For this reason, it is better to use only utilities dedicated to the task that we want or powerful languages ​​such as sed, awk, python, perl, etc.