Bathing water guide

This page is a knowledge crystal of subtype method. The page identifier is Op_en7961
Moderator:Nobody (see all) Click here to sign up.

Upload data {{#opasnet_base_link:Op_en7961}}

Question

How to asses the microbiological risks of natural bathing waters, and the health effects they have? Data on the water and bathing behaviour must be possible to give to the model as input.

Answer

The model below is already functional. However, there is still work to do to ensure that the input data the model uses is accurate.

The contamination classes used in the model are not comparable to the bathing water classification used by the EU. To see the microbe concentrations for each of the contamination classes, see page Pathogen concentrations in raw water.

+ Show code - Hide code

# This is code "Op_en7961/bathingwaterguide" on page [[Bathing water guide]]
library(OpasnetUtils)
library(ggplot2)

objects.latest("Op_en7961", code_name="initialization") # [[Bathing water guide]] fetches the variables needed in the model

# Create variables from input data
RawClass <- Ovariable("RawClass", data=data.frame(RawWaterClass = rawclass, Result=1))

# If microbe concentrations are given by hand, they're put into an ovariable
if (rawclass == "Hand input") {
  RawConcentration <- Ovariable(
    "RawConcentration",
    data=data.frame(
      Pathogen=c(
        "campylobacter",
        "E.coli O157:H7",
        "rotavirus",
        "norovirus",
        "cryptosporidium", 
        "giardia"),
      Result=c(
        Campylo,
        E.coli,
        Rota,
        Noro,
        Crypto,
        Giardia)
    )
  )
}

population <- Ovariable(
  "population",
  dependencies=data.frame(
    Name=c("total_population", "distribution"),
    Ident=c(NA, "Op_en7961/age_distribution")
  ),
  formula=function (...) {
    return(distribution * total_population)
  }
)

#oprint(EvalOutput(population)@output)

#exposure <- EvalOutput(exposure)

#objects.latest("Op_en2261", code_name="BoDattr2") # [[Health impact assessment]] BoDattr

BoDattr <- EvalOutput(BoDattr, forceEval=TRUE)

cat("Microbe concentrations in bathing water, microbes/liter. \n")
oprint(RawConcentration@output[,c("Pathogen", "RawConcentrationResult")])
cat("Exposure: how many of each microbe people of different ages swallow with bathing water.\n")
oprint(summary(exposure))

ggplot(
  oapply(unkeep(BoDattr, sources=TRUE), NULL, mean, "Iter")@output,
  aes(x=Exposure_agent, weight=BoDattrResult, fill=Response)) +
  geom_bar() +
  coord_flip() +
  theme_gray(base_size=24)+
  labs(
    title="Burden of disease of bathing water microbes per day",
    subtitle=paste("on a beach of", sum(result(population)), "bathers daily"),
    y = "Burden of disease (DALY/day)",
    x = "Microbe"
  )

# number of cases
cases <- BoDattr/case_burden

ggplot(
  oapply(unkeep(cases, sources=TRUE), NULL, mean, "Iter")@output,
  aes(x=Exposure_agent, weight=Result, fill=Response)) +
  geom_bar() +
  coord_flip() +
  theme_gray(base_size=24)+
  labs(
    title="Cases of illness caused daily by bathing water microbes",
    subtitle=paste("on a beach of", sum(result(population)), "bathers daily"),
    y = "Cases per day",
    x = "Microbe"
  )


plotdata <- oapply(unkeep(cases, sources=TRUE), NULL, mean, "Iter")@output
plotdata$Age <- factor(plotdata$Age, ordered=T, levels=c("0-4", "5-9", "10-14", "15-24", "25-64", "65-79", "80+"))

cat("The cases of each disease are calculated as a sum of cases of that disease caused by each pathogen. Because it is possible for an individual to be ill as a result of diseases from multiple pathogens, in the figure below the numbers of sick individuals is higher than it would be in reality. For example, all six pathogens included in the model cause gastroenteritis, so if an individual is ill with gastroenteritis caused by all six pathogens, they would be counted in the gastroenteritis cases six times. In these cases the gastroenteritis this individual has would be worse than if ill from only one pathogen. It is also possible for an individual to get one disease from one pathogen and a different disease at the same time from a different one, in which case the individual is counted multiple, once in each disease they have. \n")

ggplot(
  plotdata,
  aes(x=Age, weight=Result, fill=Response)) +
  geom_bar() +
#  coord_flip() +
  theme_gray(base_size=24)+
  labs(
    title="Cases of illness caused daily by bathing water microbes in different age groups",
    subtitle=paste("on a beach of", sum(result(population)), "bathers daily"),
    y = "Cases of illness daily",
    x = "Age"
  ) +
  theme(axis.text.x = element_text(angle = 45, hjust=1))

# cases as percent of beachgoers
cases_percent <- cases*100/total_population

ggplot(
  oapply(unkeep(cases_percent, sources=TRUE), NULL, mean, "Iter")@output,
  aes(x=Exposure_agent, weight=Result, fill=Response)) +
  geom_bar() +
  coord_flip() +
  theme_gray(base_size=24)+
  labs(
    title="Proportion of bathers who get ill from bathing water microbes",
    subtitle=paste("on a beach of", sum(result(population)), "bathers daily"),
    y = "Percent of bathers",
    x = "Microbe"
  )

"Run model" -button opens a new tab, on which the results will appear after the model is done running. Current running time is ~40 seconds.

Rationale

Dependencies

**Pages containing the data and codes used by the model**
Page	What data or code the page has	What is it used for	Other observations
Op_en7961 Bathing water guide (this page)	Data on time spent in water and amount of water swallowed during bathing, as well as the age distribution of beach-goers.	The amount of water swallowed per hour is multiplied by time spent in water to find the total water swallowed per bathing event. The age distribution is used, because people of different ages swallow different amounts of water and spend different lengths of time in water. Also susceptibility to diseases and their severity changes with age, so burden of disease and cases of illness are calculated for each age group.
Op_en7953 Pathogen concentrations in raw water	The amounts of pathogens in different (raw) water classes	If the user has chosen a water contamination class, the amounts of each pathogen in water of that contamination class is found here. If the user has given the amounts of pahtogens by hand, the data on this page isn't used and the given values are used instead.	The data on this page is also used in Water guide, and it is better suited to be used as data on rawa water for drinking water than for bathing water. This is alos why the classification does not match the bathing water classification used by the EU.
Op_en7947 Case burden of waterborne microbes	Case burdens of different pathogens, or how severe and long-lasting the diseases caused by them are	This data is used to calculate the health detriments caused by the pathogens in bathing water.
Op_en7957 Health impacts of waterborne microbes	Doesn't contain data, only code
Op_en7948 ERF of waterborne microbes	Describes the exposure-response functions for the pathogens	These functions tell how likely an individual is to get sick with different exposures. This and the exposure are used to estimate the cases of illness.
Op_en2261 HIA	Code that combines all of the above	The code on the page uses exposure, exposure-response functions and the case burdens to calculate the total burdens caused by pathogens in bathing water. From here it is also possible to calculate the number of people getting ill from bathing water. These are the final results of the model.

Data

Amount of water swallowed, ml/h

Reference	Adults	Children	Notes
Dufour et al. 2006 ^[1]	0 - 70,67 (avg 21,33)	0 - 205,33 (avg 49,33)	They measured cleaning chemicals added to the pool water from the bathers' urine
Dufour et al. 2017 ^[2]	12,4 (CI95 11-14)	6-10-year-olds: 23,9 (CI95 17-33) 11-15 vuotiaat: 23,7 (CI95 19-30)	Same as previous, but with a bigger sample size
Suppes et al. 2013 ^[3]	0 - 50,9 (avg 3,5)	0,9 - 105,5 (avg 25,7)	Also measuring chemicals in urine
Suppes et al. 2016 ^[4]	0 - 60,6 (avg 6,3)	0 - 105,5 (avg 24,2)	Same methods as the previous one
Schets et al. 2011 ^[5]	20 - 30	28	Questionnaire. These values are for fresh water, but the study also looked at values for sea and pool water. Values are averages (20 for women, 30 for men).
DeFlorio-Barker et al. 2018 ^[6]	n. 0 - 114 (keskiarvo n. 16)	n. 0.7 - 144 (keskiarvo n. 27)	Quistionnaire, simulations of amounts of swallowed water based on behavior and Dufour (2017) values.

Time spent in water (min)

Reference	Adults	Children	Notes
Dufour et al. 2017	50,3 (CI95 49-52)	6-10-year.olds: 95,9 (CI95 88-104) 11-15-year-olds: 55,8 (CI95 55-59)	The study participants were asked to stay in the water for approximately one hour
Suppes et al. 2016	20 - 240 (avg 72)	30 - 480 (avg 114)
Schets et al. 2011	54 (CI95 6-220)	79
DeFlorio-Barker et al. 2018	1-360 (avg 45,4-47)	0-3-year-olds: 2 - 300 (avg 56-66,7) 4-18-year-olds: 1 - 360 (avg 64-93)

The data for the tables below is from DeFlorio-Barker et al. 2018, and is the data the model uses.

Swallowed water(l/h)
Obs	Age	Result
1	0-4	0.024(0.0005-0.087)
2	5-9	0.024(0.0005-0.087)
3	10-14	0.027(0.0005-0.113)
4	15-24	0.020(0.001-0.112)
5	25-64	0.014(0.0004-0.100)
6	65-79	0.013(0.0004-0.115)
7	80+	0.013(0.0004-0.115)

Bathing time(h)
Obs	Age	Result
1	0-4	0.5-1.5
2	5-9	0.75-2
3	10-14	0.75-2
4	15-24	0.5-2
5	25-64	0.5-1.5
6	65-79	0.5-1.5
7	80+	0.5-1.5

The table below shows an estimate of the age distribution of beachgoers. The data in the table is provided by Finnish Swimming Teaching and Lifesaving Federation, and is a very, very rough estimate of the age distribution. No data is currently collected of the age distribution of beachgoers.

Some arguments for this distribution:

0-4-year-olds only go to the beach with 25-64-year-olds (also to a much lesser extent with the 65-79-year-olds).
25-64-year-olds cover 40 years, so even though this is a group of people working and spending time at summer cottages, the total proportion is probably 0.2
Due to the lack of any real statistics, there is nothing to suggest differences in proportions between the age groups covering 5-64-year-olds.
The children and young adults between ages 5-24 might well cover 60% of the beachgoers, also depending on any possible activities on the beach, such as a tower with diving platforms.

Beachgoers' age distribution(-)
Obs	Age	Result
1	0-4	0.1
2	5-9	0.2
3	10-14	0.2
4	15-24	0.2
5	25-64	0.2
6	65-79	0.08
7	80+	0.02

Calculations

+ Show code - Hide code

# This is code "Op_en7961/swallowed_water" on page [[Bathing water guide]]
library(OpasnetUtils)

swallowedwater <- Ovariable("swallowedwater", ddata="Op_en7961", subset="Swallowed water")
bathingtime <- Ovariable("bathingtime", ddata="Op_en7961", subset="Bathing time")

Consumption <- Ovariable(
  "Consumption",
  Dependencies=data.frame(
    Names=c("bathingtime", "swallowedwater")),
  formula=function(...){
   return(bathingtime*swallowedwater) 
  }
)

objects.store(Consumption, swallowedwater, bathingtime)
#oprint(EvalOutput(Consumption)@output, show_all=TRUE)
cat("Ovariables Consumption, swallowedwater and bathingtime saved. \n")

+ Show code - Hide code

# This is code "Op_en7961/age_distibution" on page [[Bathing water guide]]
library(OpasnetUtils)

distribution <- Ovariable("distribution", ddata="Op_en7961", subset="Beachgoers' age distribution")

objects.store(distribution)
cat("Ovariable distribution saved. \n")

+ Show code - Hide code

# This is code Op_en7961/initialization on page [[Bathing water guide]]
library(OpasnetUtils)
library(ggplot2)

openv.setN(1)

# Create ovariables from user input data
RawClass <- Ovariable("RawClass", data=data.frame(RawWaterClass = 'Surface water - low contamination', Result=1))

population <- Ovariable("population", data=data.frame(Result=100))
# population <- EvalOutput(population)

objects.latest("Op_en7947", code_name="caseburden")
#case_burden <- EvalOutput(case_burden)

incidence <- 1 # Number per 1py # Potentially everyone can get the disease. PAF is the fraction of this that actually occurs

exposure <- Ovariable(
  "exposure",
  dependencies=data.frame(
    Name=c("Consumption", "RawConcentration"),
    Ident=c("Op_en7961/swallowed_water", "Op_en7953/variable")
  ),
  formula = function(...) {
    out <- RawConcentration * Consumption
    colnames(out@output)[colnames(out@output)=="Pathogen"] <- "Exposure_agent"
    return(out)
  }
)

ERF <- Ovariable(
  "ERF",
  dependencies=data.frame(Name="ERF_micr",Ident="Op_en7948/ERF_micr2"),
  formula = function(...) return(ERF_micr)
)

objects.latest("Op_en2261", code_name="BoDattr2") # [[Health impact assessment]] BoDattr
BoDattr <- EvalOutput(BoDattr, verbose=TRUE)

oprint(summary(BoDattr, marginals=c("Response","Exposure_agent")))

oprint(summary(exposure))

objects.store(list=setdiff(ls(), "wiki_username"))
cat("The whole model stored (", setdiff(ls(), "wiki_username"), ").\n")

ggplot(
  oapply(unkeep(BoDattr, sources=TRUE), NULL, mean, "Iter")@output,
  aes(x=Exposure_agent, weight=BoDattrResult, fill=Response)) +
  geom_bar() +
  coord_flip() +
  labs(
    title="Disease burden of waterborne microbes",
    subtitle=paste("in a population of", sum(result(population))),
    y = "Disease burden (DALY/incident)"
  )

if(FALSE) {
  ######################## Diagnostics
  
  objects.latest("Op_en6007", code_name="diagnostics")
  oprint(showLoctable())
  oprint(showind())
  
  ################ Insight network
  
  gr <- scrape(type="assessment")
  objects.latest("Op_en3861", "makeGraph") # [[Insight network]]
  gr <- makeGraph(gr)
  #export_graph(gr, "ruori.svg")
  #render_graph(gr)
}

Plotly graphs. Code doesn't work in Opasnet

+ Show code - Hide code


# This is code Op_en7961/plotly on page [[Bathing water guide]]

library(OpasnetUtils)
library(plotly)
library(stringr)

openv.setN(100)

## Datan valmistelu ----

objects.latest("Op_en7961", code_name="initialization") # [[Uimavesiopas]] hakee koko mallin muuttujat

rawclass <- c("Surface water - low contamination",
              "Surface water - medium contamination",
              "Surface water - high contamination")
total_population <- Ovariable("total_population",
                              data=data.frame(Pop = seq(50,500,50),
                                              Result = seq(50,500,50)))
labels <- c("Low contamination", "Medium contamination", "High contamination")

population <- Ovariable(
  "population",
  dependencies=data.frame(
    Name=c("total_population", "distribution"),
    Ident=c(NA, "Op_en7961/age_distribution")
  ),
  formula=function (...) {
    return(distribution * total_population)
  }
)

uimavesidata <- data.frame(matrix(ncol=7, nrow=0))
colnames(uimavesidata) <- c("Response", "Exposure_agent", "Pop", "RawWaterClass", "BoDattrResult", "Result", "label")

for(i in 1:length(rawclass)) {
  RawClass <- Ovariable("RawClass", data=data.frame(RawWaterClass = rawclass[i], Result=1))
  BoDattr <- EvalOutput(BoDattr, forceEval=TRUE)
  cases <- BoDattr/case_burden
  cases <- oapply(cases, c("Response","Exposure_agent","Pop","RawWaterClass", "Iter", "BoDattrResult"), sum)
  cases <- aggregate(cases@output[,c("BoDattrResult", "Result")], by=cases@output[,c("Response", "Exposure_agent", "Pop", "RawWaterClass")], FUN=mean)

  uimavesidata <- rbind(uimavesidata, data.frame(cases, label = labels[i]))
}


uimavesidata$proportion <- uimavesidata$Result/uimavesidata$Pop

uimavesidata$Pop <- as.factor(uimavesidata$Pop)
uimavesidata$visible <- F
uimavesidata$visible[uimavesidata$Pop==levels(uimavesidata$Pop)[1]] = T # smallest population visible

uimavesidata <- uimavesidata[uimavesidata$Response == "gastroenteritis",]
colnames(uimavesidata)[colnames(uimavesidata) == "BoDattrResult"] <- "DALY"
colnames(uimavesidata)[colnames(uimavesidata) == "Result"] <- "cases"

#
## RANNAN KOKO + RAW CLASS MUUTTUJINA ----


# create steps and plot all traces
beach_size <- list()
uimavesi_plot <- plot_ly()
for (i in 1:length(unique(uimavesidata$Pop))) {
  uimavesi_plot <- add_trace(uimavesi_plot, # add traces to uimavesi_plot
                            type = "scatter",
                            x=uimavesidata$cases[uimavesidata$Pop==levels(uimavesidata$Pop)[i]], # x (BoD) on x
                            y=uimavesidata$Exposure_agent[uimavesidata$Pop==levels(uimavesidata$Pop)[i]], # y (exposure agent) on y
                            visible = uimavesidata$visible[uimavesidata$Pop==levels(uimavesidata$Pop)[i]], # is this trace visible
                            mode="markers",
                            marker = list(size=10, # marker size
                                          colorscale = list(list(0, 'rgba(200,200,200,0.2)'), list(1, 'rgba(0,181,204,1)')), # sets the colors used when color=0 and color=1
                                          color = c(rep(1, 6), # the first 17 markers use color 1
                                                    rep(0, 6*2))),
                            hoverinfo = "x", # what is the info shown when mouse on top of marker
                            showlegend=F)
  
  bw <- list(args = list('visible', rep(FALSE, length(unique(uimavesidata$Pop)))), # set all steps to not visible
             method = 'restyle',
             label= levels(uimavesidata$Pop)[i]) # Sets the text label to appear on the menu 
  bw$args[[2]][i] = TRUE  # set data on i step to be visible when step i is chosen
  beach_size[[i]] = bw # make a list with all the steps
}  

water_qlty <- list(list(args = list('marker.color', # marker color is changed
                                   array(c(rep(1, 6), rep(0, 6*2)), dim=c(1, 18))), # these are the new markers with color 1 and 0
                       method = 'restyle', # restyling the markers
                       label = "Vähäinen kuormitus"),
#                       label = "Low contamination"), # label on the button
                  list(args = list('marker.color',
                                   array(c(rep(0, 6), rep(1, 6), rep(0, 6)), dim=c(1, 18))),
                       method = 'restyle',
                       label = "Keskisuuri kuormitus"),
#                       label = "Medium contamination"),
                  list(args = list('marker.color',
                                   array(c(rep(0, 6*2), rep(1, 6)), dim=c(1, 18))),
                       method = 'restyle',
                       label = "Suuri kuormitus")
#                       label = "High contamination")
)

# add slider control to plot
uimavesi_plot <- uimavesi_plot %>%
  layout(sliders = list(list(active = 0, #Determines which button (by index starting from 0) is active.
                             currentvalue = list(prefix = "Uimarannan kävijämäärä:"), #suomeksi
#                             currentvalue = list(prefix = "Daily beach goers: "), # in English
                             # the text under the graph that tells what the slider values mean
                             steps = beach_size)),
         updatemenus = list(list(active = 0,
                                 y = -0.15,
                                 x = -0.1,
                                 direction = "up",
                                 buttons = water_qlty)),
#         title = paste("Cases of illness caused by bathing water"), # in English
         title = paste("Uimaveden aiheuttamat tautitapaukset päivässä"), # suomeksi
         xaxis = list(title ="Tautitapaukset"),
         yaxis = list(title =""))

uimavesi_plot

#

## X:N YKSIKKÖ + RAW CLASS MUUTTUJINA ----


uimavesi_leikattu <- uimavesidata[uimavesidata$Pop == 150,]

nakyma <- list()
uimavesiplot <- plot_ly()
steps <- c("DALY", "cases", "proportion")
label <- c("DALY", "tautitapaukset", "sairastumistodennäköisyys")
visible <- c(T, F, F)
for (i in 1:3) {
  uimavesiplot <- add_trace(uimavesiplot, # add traces to uimavesi_plot
                             type = "scatter",
                             x=uimavesi_leikattu[,colnames(uimavesi_leikattu) == steps[i]],
                             y=uimavesi_leikattu$Exposure_agent, # y (exposure agent) on y
                             visible = visible[i], # is this trace visible
                             mode="markers",
                             marker = list(size=10, # marker size
                                           colorscale = list(list(0, 'rgba(200,200,200,0.2)'), list(1, 'rgba(0,181,204,1)')), # sets the colors used when color=0 and color=1
                                           color = c(rep(1, 6), # the first 17 markers use color 1
                                                     rep(0, 6*2))),
                             hoverinfo = "x", # what is the info shown when mouse on top of marker
                             showlegend=F)
  
  bw <- list(args = list('visible', rep(FALSE, 3)), # set all steps to not visible
             method = 'restyle',
             label= label[i]) # Sets the text label to appear on the menu 
  bw$args[[2]][i] = TRUE  # set data on i step to be visible when step i is chosen
  nakyma[[i]] = bw # make a list with all the steps
}  

water_qlty <- list(list(args = list('marker.color', # marker color is changed
                                    array(c(rep(1, 6), rep(0, 6*2)), dim=c(1, 18))), # these are the new markers with color 1 and 0
                        method = 'restyle', # restyling the markers
                        label = "Vähäinen kuormitus"),
                   #                       label = "Low contamination"), # label on the button
                   list(args = list('marker.color',
                                    array(c(rep(0, 6), rep(1, 6), rep(0, 6)), dim=c(1, 18))),
                        method = 'restyle',
                        label = "Keskisuuri kuormitus"),
                   #                       label = "Medium contamination"),
                   list(args = list('marker.color',
                                    array(c(rep(0, 6*2), rep(1, 6)), dim=c(1, 18))),
                        method = 'restyle',
                        label = "Suuri kuormitus")
                   #                       label = "High contamination")
)

# add slider control to plot
uimavesiplot <- uimavesiplot %>%
  layout(updatemenus = list(list(active = 0,
                                 y = 1,
                                 x = -0.2,
                                 buttons = water_qlty),
                            list(active = 0, #Determines which button (by index starting from 0) is active.
                                 y = 0.85,
                                 x = -0.2,
                                 buttons = nakyma)),
#         title = paste("Cases of illness caused by bathing water on a beach of 150 dialy visitors"), # in English
         title = "Uimaveden mikrobien aiheuttama terveyshaitta päivässä 150 kävijän rannalla", # suomeksi
#         xaxis = list(title ="Tautitapaukset"),
         yaxis = list(title =""))

uimavesiplot

Viitteet

↑ Dufour et al. 2006. Water ingestion during swimming activities in a pool: a pilot study. J Water Health (2006) 4 (4): 425-430. [1]
↑ Dufour et al. 2017. Ingestion of swimming pool water by recreational swimmers. J Water Health (2017) 15 (3): 429-437. [2]
↑ Suppes et al. 2013. Assessment of swimmer behaviors on pool water ingestion. J Water Health (2013) 12 (2): 269-279. [3]
↑ Suppes et al. 2016. Cryptosporidium risk from swimming pool exposures. International Journal of Hygiene and Environmental Health 219(8):915-919 [4]
↑ Schets et al. 2011. Exposure assessment for swimmers in bathing waters and swimming pools. Water Research 45:2392-2400. [5]
↑ DeFlorio-Barker et al. 2018. Child environmental exposures to water and sand at the beach: Findings from studies of over 68,000 subjects at 12 beaches. Journal of Exposure Science and Environmental Epidemiology 28:93–100 [6]

[1] Dufour et al. 2006. Water ingestion during swimming activities in a pool: a pilot study. J Water Health (2006) 4 (4): 425-430. [1]

[2] Dufour et al. 2017. Ingestion of swimming pool water by recreational swimmers. J Water Health (2017) 15 (3): 429-437. [2]

[3] Suppes et al. 2013. Assessment of swimmer behaviors on pool water ingestion. J Water Health (2013) 12 (2): 269-279. [3]

[4] Suppes et al. 2016. Cryptosporidium risk from swimming pool exposures. International Journal of Hygiene and Environmental Health 219(8):915-919 [4]

[5] Schets et al. 2011. Exposure assessment for swimmers in bathing waters and swimming pools. Water Research 45:2392-2400. [5]

[6] DeFlorio-Barker et al. 2018. Child environmental exposures to water and sand at the beach: Findings from studies of over 68,000 subjects at 12 beaches. Journal of Exposure Science and Environmental Epidemiology 28:93–100 [6]

[1]

[2]

[3]

[4]

[5]

[6]

Bathing water guide

Contents

Question

Answer

Rationale

Dependencies

Data

Calculations

Plotly graphs. Code doesn't work in Opasnet

See also

Viitteet

Navigation menu