Diarrheal Diseases: GBD 2019

Disease Description

Diarrhea is commonly defined as three or more loose stools in a 24-hour period. [GBD-2019-Capstone-Appendix-Diarrhea]

Diarrhea has various etiologies, with infectious diarrhea accounting for the vast majority of global diarrheal disease burden. The top pathogens responsible for diarrhea include norovirus, rotavirus, E. Coli, Camplyobacter, and Salmonella. Bacterial infections, and specifically species of Shigella, account for the majority of bloody diarrhea.

Infection most commonly occurs via feces-contamined water, and can also spread via contamined food and person-to-person contact. ([WHO-Diarrhea])

The global prevalence of diarrhea thus varies considerably accoring to resource access. In particular, resource-limited countries have a “baseline frequency… superimposed with epidemic cases of diarrhea” ([UpToDate_1-Diarrhea]). The top risk factors for diarrheal diseases thus include crowding (such as living in refugee camps) and poor sanitation, in addition to immune system-compromising conditions, such as living with HIV.

The most significant outcomes of a nonfatal diarrhea episode are dehydration and the loss of nutrition. In particular, in low-income countries, the high prevalence of diarrhea is a major cause of child malnutrition ([WHO-Diarrhea]), which in turn makes such children more susceptible to future diarrheal episodes and other negative sequelae.

The WHO-recommended measures for diarrhea prevention include:

  • Access to safe drinking water;

  • Use of improved sanitation;

  • Hand washing with soap;

  • Exclusive breastfeeding for the first six months of life;

  • Good personal and food hygiene;

  • Health education about how infections spread; and

  • Rotavirus vaccination.

Noninfectious diarrhea etiologies are far less common, but are more likely among chronic cases of diarrhea. Causes of noninfectious diarrhea include ischemic colitis, inflammatory bowl disease, among others ([UpToDate_2-Diarrhea]).

Other info: [CDC-Diarrhea], [Wikipedia-Diarrhea].

Modeling Diarrheal Diseases in GBD 2019

The GBD diarrheal diseases model includes a cause of death (CoD) model and a nonfatal model, each of which is split into diarrhea etiologies.

The CoD model used data from vital registration systems, surveillance systems, and verbal autopsy. Deaths were modelled separately for males and females and for under 5 year olds vs. the rest of the population using CODEm models with country-level covariates.

The nonfatal model used diarrhea incidence and prevalence data in community and hospital settings in addition to hospital and claims data tagged ICD9 codes 001-009.9 and ICD10 codes A00-A09. All hospital and claims data was converted to prevalence (or point prevalence) data using a mean duration of 4.3 (4.2-4.4) days. Data was then adjusted for seasonality, and all non-reference case data was adjusted to fit the reference category of “community-based diarrhea episodes”. Newly in GBD 2019, EMR priors were estimated using MR-BRT instad of DisMod. Finally, this data is fed into a DisMod model.

A systematic review of diarrhea severity was conducted, from which simple severity splits were obtained and applied across all cases:

Diarrhea severity splits

Severity level

Definition

Lay description

Disability weight

Proporiton

Mild

Diarrhea cases that did not seek medical care

Has diarrhea defined as 3 or more loose stools in a 24-hour period with no dehydration

0.074 (0.049-0.104)

64.8%

Moderate

Diarrhea cases that sought medical care but did not have severe dehydration or bloody stool

Has diarrhea defined as 3 or more loose stools in a 24-hour period with painful cramps and feeling thirsty and any dehydration

0.188 (0.125-0.264)

28.9%

Severe

Diarrhea cases that sought medical care with severe dehydration or bloody stool

Has diarrhea defined as 3 or more loose stools in a 24-hour period with painful cramps and is very thirsty or feels nauseated or tired and/or severely dehydrated

0.247 (0.164-0.348)

6.9%

GBD then modelled diarrhea etiologies, for which PAFs of fatal and nonfatal diarrhea were calculated. Etiologies include enteric adenovirus, Aeromonas, Entamoeba histolytica (amoebiasis), Campylobacter, Cryptosporidium, typical enteropathogenic Escherichia coli (t-EPEC), enterotoxigenic Escherichia coli (ETEC), norovirus, non-typhoidal salmonella infections, rotavirus, Shigella, Vibrio cholerae and Clostridium difficile.

Excluding Vibrio cholerae and Clostridium difficile, all etiologies were modelled by calculating the proportion of severe diarrhea cases that tested positive for each etiology, with hospitalized diarrhea cases serving as a proxy for severe cases. Note that as pathogens can co-infect, this yields PAFs that sum to greater than 100% of diarrhea cases. Vibrio cholerae and Clostridium difficile cases were each modelled directly using DisMod.

GBD Hierarchy

../../../../_images/DD_cause_hierarchy1.svg

Vivarium Modeling Strategy

Cause Model Diagram

../../../../_images/DD_cause_model1.svg

S: Susceptible to diarrheal diseases

I: Infected and currently experiencing a diarrheal disease bout

Model Assumptions and Limitations

Note that GBD has done extensive work to divide up diarrhea cases into their respective etiologies. For now, we omit this complexity. Further, GBD incorporates seasonality into the diarrhea model. Our model currently does not.

Regarding severity: the GBD model splits nonfatal diarrhea estimates into three severity categories, using a ratio applied across all estimates. This ratio might be expected to vary from location to location, or perhaps across time, and thus we assume this is a limitation of the GBD model.

Todo

Verify the simple severity split approach is indeed a limitation. I.e., the verify that the modelers expect a more complex pattern.

Data Descriptions

State Definitions

State

State name

Definition

S

Susceptible

Simulant does not currently have diarrheal disease

I

Infected

Simulant currently has diarrheal disease
State Data

State

Measure

Value

Notes

I

prevalence

For early neonatal age group: (birth_prevalence_I + (incidence_rate_c302 * duration_c302))/2. For all other age groups: incidence_rate_c302 * duration_c302

Early neonatal age group exception due to non-steady state dynamics in this age group given birth prevalence of zero causes increasing prevalence within age group and short duration of age group. Citation on these dynamics and approximations here for reference.

I

birth prevalence

0

I

excess mortality rate

\(\frac{\text{deaths\_c302}}{\text{population} \,\times\, \text{prevalence\_I}}\)

Use prevalence calculated for the I state in the row above

I

disability weight

\(\displaystyle{\sum_{s\in \text{sequelae\_c302}}} \scriptstyle{\text{disability\_weight}_s \,\times\, \text{prevalence}_s}\)

S

prevalence

1-prevalence_I

Use prevalence calculated for the I state in the first row

S

birth prevalence

1

S

emr

0

S

disability weight

0

All

cause-specific mortality rate

\(\frac{\text{deaths\_c302}}{\text{population}}\)
Transition Data

Transition

Source State

Sink State

Value

Notes

i

S

I

\(\frac{\text{incidence\_rate\_c302}}{1-\text{incidence\_rate\_c302}*(\text{duration\_c302} / 365)}\)

We transform incidence to be a rate within the susceptible population under the assumption that prevalence ~= incidence * duration.

r: USED IN CIFF AND IV IRON SIMULATIONS AS WELL AS MODELS 1-11 OF NUTRITION OPTIMIZATION SIMULATION

I

S

(-1/time_step)*log(1-time_step/duration_c302)

Where time_step is the simulation time_step in years. See notes below on adjusted duration. Use np.log() function. The above is equivalent to 1/adjusted_duration_c302.

r: FOR USE IN NUTRITION OPTIMIZATION SIMULATION AFTER IMPLEMENTATION OF VARIABLE TIMESTEPS

I

S

1/duration_c302

Note

We are using a custom remission rate for diarrheal diseases based on the estimated duration of disease because after scaling to the total population using the estimated prevalence of diarrheal diseases, the remission rate was greater than the incidence rate for children under five in Ethiopia, which is implausible. We assume an average duration of a diarrheal disease episode of 4.3 days, as estimated by [Troeger-et-al-2018-Diarrhea].

Data Sources and Definitions

Value

Source

Description

Notes

prevalence_c302

como

Prevalence of diarrheal diseases

deaths_c302

codcorrect

Deaths from diarrheal diseases

duration_c302

(4.3 days; 95% CI: 4.2, 4.4; normal distribution of uncertainty)/365

Mean duration of diarrheal disease episode (in years). Obtained from [Troeger-et-al-2018-Diarrhea] and the GBD YLD appendix.

This value should not vary by age group

adjusted_duration_c302

4.04485 (95% CI: 3.94472, 4.144975), assume normal distribution of uncertainty

Average duration of a diarrheal disease episode in days among children under five (defined in the note column) TRANSFORMED to accomodate a short timestep of 0.5 days, as discussed in this slack thread. See the note below for more information.

This value does not necessarily need to be stored – included here for reference.

incidence_rate_c302

como

Incidence of diarrheal disease within the entire population

population

demography

Mid-year population for given age/sex/year/location

sequelae_c302

gbd_mapping

List of 4 sequelae for diarrheal diseases

Note Guillain-Barre due to diarrheal diseases is included in sequelae.

prevalence_s{sid}

como

Prevalence of sequela with id sid

disability_weight_s{sid}

YLD appendix

Disability weight of sequela with id sid

Note

We implemented a remission rate of diarrheal diseases equal to 1/the average duration of diarrheal diseases = 1/4.3 days. However, the remission rate output from our simulation was slower than the artifact value, approximating 1/4.55 days.

As identified by Nathaniel, this appeared to be due to the fact that the product of the remission rate r=1/4.3 times the time step dt=0.5 was too large for the approximation 1-exp(-r*dt) ~= r*dt to be sufficiently good for the rates to match.

What’s going on is that we’re thinking of the duration of diarrhea as a continuous random variable, exponentially distributed with rate r=1/4.3, but in Vivarium this random variable gets discretized into a geometric random variable, I believe with parameter p=1-exp(-r*dt) . The mean of the exponential random variable is 1/r = 4.3 days, whereas the mean of the geometric random variable, converted from time steps back to days, is dt/p ~= 4.55 days . This same issue will arise whenever we have a transition rate that is large relative to the simulation time step. You could always solve it by making the time steps even smaller, but of course that adds a lot of computation time.

To deal with this, solved for the mean rate (in days) to input to vivarium that would produce the desired output of the a remission rate equal to 1/4.3 days using the following equation for r’

\[ \begin{align}\begin{aligned}r' = (-1/dt)*log(1-dt*r)\\ = (-1/0.5)*log(1-0.5/4.3)\\ = 0.24722791193435328\\1 / r' = 4.044850729740949 days\end{aligned}\end{align} \]

We then also solved for the upper and lower bound estimates using the same methodology.

See the Choosing an Appropriate Time Step page page for more information.

Restrictions

Restriction type

Value

Notes

Male only

False

Female only

False

YLL only

False

YLD only

False

YLL age group start

Post neonatal (age group ID 4, 1 month to 1 year)

GBD age group start is early neonatal (age group ID 2, 0-6 days)

YLL age group end

95 plus

age_group_id = 235; 95 years +

YLD age group start

Post neonatal (age group ID 4, 1 month to 1 year)

GBD age group start is early neonatal (age group ID 2, 0-6 days)

YLD age group end

95 plus

age_group_id = 235; 95 years +

Note

A note on the diarrheal diseases age start parameter:

This Vivarium modeling strategy sets the diarrheal diseases cause model age start to the post neonatal age group (1 month to 1 year) despite the GBD age start parameter being the early neonatal age group (0 to 6 days). We exclude the early and late neonatal age groups from the diarrheal diseases cause model as a strategy that allows us to increase the timestep of our cause models.

The rationale behind this modeling decision is related to the Relationship between timesteps and modeled rates in Vivarium as described on the Choosing an Appropriate Time Step page that is exacerbated by the inclusion of the low birth weight and short gestation risk factor in the model. Essentially, because the LBWSG risk factor affects diarrheal diseases excess mortality rates in our models during the neonatal age groups and the LBWSG relative risk values for the highest risk categories are quite large (up to 700!), the inclusion of the LBWSG risk effects on diarrheal diseases causes individual-level diarrheal diseases excess mortality rates to be too large to accurately approximate in our models without a very small timestep, which leads to underestimation of neonatal diarrheal diseases mortality rates with a timestep on the order of 0.5 days.

Therefore, we employ the following strategy:

  • Model the diarrheal diseases SI cause model as described in this document for ages older than late neonatal only, and

  • Include diarrheal diseases as an unmodeled cause that is affected by the LBWSG risk factor (see the LBSWG risk effects page for details). This will allow us to model diarrheal diseases CSMR rather than EMR among the neonatal age groups, which is lower in magnitude and therefore less easier to approximate at larger simulation timesteps. Notably, this strategy does not allow us to model years lived with disability due to diarrheal diseases among the neonatal age groups.

This strategy allowed us to increase the simulation timestep to 4 days and still meet verification criteria.

Verification and Validation Criteria

Verification:

  • We should replicate the following parameters:

    • GBD incidence rates among ages older than the late neonatal age group

    • The custom input remission rate (~1/4.3 days) among ages older than the late neonatal age group

    • GBD cause-specific mortality rates among all modeled ages

Validation:

  • We should compare our estimates of diarrheal diseases prevalence to GBD estimates of diarrheal diseases prevalence (among age groups greater than the late neonatal age group). Our modeled estimates may deviate from the GBD estimates for this parameter given that we have chosen to prioritize estimates of incident and fatal cases of diarrheal diseases rather than prevalent cases.

References

[GBD-2019-Capstone-Appendix-Diarrhea]

Appendix to: GBD 2019 Diseases and Injuries Collaborators. Global burden of 369 diseases and injuries in 204 countries and territories, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019. The Lancet. 17 Oct 2020;396:1204-1222

[Troeger-et-al-2018-Diarrhea] (1,2)

Troeger C, Colombara DV, Rao PC, Khalil IA, Brown A, Brewer TG, Guerrant RL, Houpt ER, Kotloff KL, Misra K, Petri WA Jr, Platts-Mills J, Riddle MS, Swartz SJ, Forouzanfar MH, Reiner RC Jr, Hay SI, Mokdad AH. Global disability-adjusted life-year estimates of long-term health burden and undernutrition attributable to diarrhoeal diseases in children younger than 5 years. Lancet Glob Health. 2018 Mar;6(3):e255-e269. doi: 10.1016/S2214-109X(18)30045-7. PMID: 29433665; PMCID: PMC5861379. Troeger et al 2018 available here

[WHO-Diarrhea] (1,2)

Diarrheal disease Fact Sheet. World Health Organization, 2 May 2019. Retrieved 14 Nov 2019. https://www.who.int/news-room/fact-sheets/detail/diarrhoeal-disease

[UpToDate_1-Diarrhea]

Approach to the adult with acute diarrhea in resource-limited countries Retrieved 26 Dec 2019. https://www.uptodate.com/contents/approach-to-the-adult-with-acute-diarrhea-in-resource-limited-countries

[UpToDate_2-Diarrhea]

Approach to the adult with acute diarrhea in resource-rich countries Retrieved 26 Dec 2019. https://www.uptodate.com/contents/approach-to-the-adult-with-acute-diarrhea-in-resource-rich-settings

[Wikipedia-Diarrhea]

Diarrhea. From Wikipedia, the Free Encyclopedia. Retrieved 14 Nov 2019. https://en.wikipedia.org/wiki/Diarrhea