8  Smoking

8.1 Workflow

flowchart TD
A[Data] --> B[population data]
A --> C[smoking data]
C --> D[[reclassify directorate names to regions]] --> E[[calculate age bands]] --> F[[calculate smoking counts]]
B --> G[[calculate age bands]] --> H[[calculate population counts]]
H --> I[[join datasets]]
F --> I
I --> J[calculate rates and ci]

Figure 8.1: Proposed workflow for calculating smoking rates

8.1.1 Load census population and create age bands

## create age band

pops <- setDT(pops)[, `:=` (`18-44`  = dplyr::between(age, 18, 44), `15+` = age >= 15, `80+` = age >= 80, age_band = cut(age, seq(0, 110, 5), right = FALSE))][]

pop_age <- pops[, .(n = .N, sumpops = sum(Population, na.rm = TRUE)), by = .(Region, age_band,  Gender)][order(age_band)]

pop1844F <- pops[, .(n = .N, sumpops = sum(Population, na.rm = TRUE)), by = .(Region, `18-44`,  Gender)][`18-44` == "TRUE" & Gender == "Female",]

pop15F <- pops[, .(n = .N, sumpops = sum(Population, na.rm = TRUE)), by = .(Region, `15+`,  Gender)][`15+` == "TRUE" & Gender == "Female",]

8.1.2 Load smoking data and calculate age bands

Note: it is not clear if the dummy data is a random sample of clinic attendance data.

smok_age <- smoking[, `:=` (`18-44`  = dplyr::between(age, 18, 44), `15+` = age >= 15, `80+` = age >= 80, age_band = cut(age, seq(0, 110, 5), right = FALSE))][]

smok_age_bands <- smok_age[, .(n = .N, smokers = sum(n, na.rm = TRUE)), by = .(Region, age_band,  Gender)][order(age_band)]

smok1844F <- smok_age[, .(n = .N, smokers = sum(n, na.rm = TRUE)), by = .(Region, `18-44`,  Gender)][`18-44` == "TRUE" & Gender == "female",]

smok15F <- smok_age[, .(n = .N, smokers = sum(n, na.rm = TRUE)), by = .(Region, `15+`,  Gender)][`15+` == "TRUE" & Gender == "female",]

8.1.3 Join datasets and calculate rates

This step uses Byar’s method for confidence interval for rates

smok_rates <- pop1844F |>
    left_join(smok1844F, by = "Region") |> 
    phe_rate(x = smokers, n = sumpops) 

smok_rates |>
    dplyr::select(Region, smokers, population = sumpops, value, lowercl, uppercl) |>
    flextable::flextable()

Region	smokers	population	value	lowercl	uppercl
'Asir		366,324
Al Bahah	28	61,969	45.183882	30.017729	65.30573
Al Hudud ash Shamaliyah	4	70,149	5.702148	1.553643	14.59976
Al Jawf	23	103,536	22.214495	14.077587	33.33412
Al Madinah al Munawwarah		387,944
Al Mintaqah ash Sharqiyah		877,403
Al Qasim		249,081
Ar Riyadh		1,609,493
Ha'il	22	136,642	16.100467	10.086571	24.37741
Jazan		267,342
Makkah al Mukarramah		1,507,301
Najran	82	105,915	77.420573	61.572876	96.10050
Tabuk	291	163,453	178.032829	158.162585	199.70848

Figure 8.2: Smoking clinic attendance rates (18-44)

smok_rates |>
    write_csv("data/smok_rates.csv")
   
smok_rates |>
    ggplot() +
    geom_col(aes(reorder(Region, value), value)) +
    geom_linerange(aes(x = Region, ymin = lowercl, ymax = uppercl)) +
    coord_flip() +
    labs(x = " ",
         y = "Rate per 100.000")

Figure 8.3: Smoking clinic attendance rates (18-44)

8.1.4 Age-specific attendance rates

smok_age_agg <- smok_age_bands[, .(tot_smokers = sum(smokers)), by = .(age_band, Gender)][]

options(scipen = 999, digitds = 2)

## m:f as smoking ratios

smok_rates_gender <- pop_age |>
    dplyr::select(-Region) |>
    mutate(Gender = recode(Gender, "Female" = "female", "Male" = "male")) |>
    left_join(smok_age_agg, by = c("age_band", "Gender")) |>
    group_by(age_band, Gender) |>
    reframe(nat_smok = sum(tot_smokers, na.rm = TRUE), 
           nat_pop = sum(sumpops)) |>
    phe_rate(x = nat_smok, n = nat_pop) |> 
    mutate(value = ifelse(Gender == "female", -value, value)) |>
    ggplot() +
    geom_col(aes(age_band, value, fill = Gender)) +
    coord_flip() +
    scale_fill_manual(values = c("blue", "red")) +
    labs(x = "",  y = "Attendance rate per 100000")

smok_rates_gender    

smok_gender_ratio <- pop_age |>
    dplyr::select(-Region) |>
    mutate(Gender = recode(Gender, "Female" = "female", "Male" = "male")) |>
    left_join(smok_age_agg, by = c("age_band", "Gender")) |>
    group_by(age_band, Gender) |>
    reframe(nat_smok = sum(tot_smokers, na.rm = TRUE), 
           nat_pop = sum(sumpops)) |>
    phe_rate(x = nat_smok, n = nat_pop) |>
    slice(7:36) |>
    group_by(age_band) |>
    reframe(ratio = value[2] / value[1]) |>
    mutate(mean_ratio = mean(ratio))

The relative attendance rate for males is 17.32 times that of females. Smoking clinic attendance rates could be used as a proxy for population smoking rates if the relationship between attendance and prevalence is known . For example, survey data suggests that KSA male smoking rates are around 20% which implies a female smoking rate based on 1.15%. In fact, available survey data (Algabbani et al. (2018)), and estimates from the Global Burden of Disease ((“Prevalence of Smoking,” n.d.; Xiang et al. 2023)), suggest a smoking prevalence in females over 15 of 2%,

Algabbani, Aljoharah Mohammed, Rasha Almubark, Nora Althumiri, Amani Alqahtani, and Nasser BinDhim. 2018. “The Prevalence of Cigarette Smoking in Saudi Arabia in 2018.” Food and Drug Regulatory Science Journal 1 (1): 1. https://doi.org/10.32868/rsj.v1i1.22.

“Prevalence of Smoking.” n.d. https://doi.org/10.1787/301887638251.

Xiang, Hong, Deshi Dong, Linlin Lv, and Xufeng Tao. 2023. “Global Burden of Disease Study 2019 Indicates That Smoking Gradually Becomes a Key Driver of the Burden of Pancreatic Cancer in Developing Regions.” In. IntechOpen. https://doi.org/10.5772/intechopen.1003616.