# Master Rally Template *This template is to be filled out for each rally. It is meant to be worked on collaboratively by all rally participants throughout the course of the rally.* *All rally outputs (such as the rally analysis plan, rally summary, and initial drafts of report-out presentations / documents) will be automatically generated from this content, so please take care in filling the entries out clearly.* *Each template entry provides you with a description of what type of content should be provided, as well as an assignee and a due date. In some cases, text already exists for you to edit.* ## Title `[id: title]` `[description: the title of the rally - be descriptive]` `[assignee: rally master]` `[due: rally kickoff]` Incidence of Wasting Rally 4a ## Previous Rally `[id: previous]` `[description: Is this a continuation of a previous rally, if so, provide the OSF link.]` `[assignee: rally master]` `[due: rally kickoff]` None: This was the first rally in this series. ## Tags `[id: tags]` `[description: a comma-separated list of terms that describe the rally]` `[assignee: rally master]` `[due: rally kickoff]` wasting incidence; wasting prevalence; risk factors for wasting; recovery ## Timeline `[id: timeline]` `[description: Rally start and end dates. Fill in {} below.]` `[assignee: rally master]` `[due: rally kickoff]` - Start: 07/21/17 - End: 08/07/17 ## Participants `[id: participants]` `[description: List of rally participants and affiliations / roles. Fill in {} below and repeat the line for each participant.]` `[assignee: rally master]` `[due: rally kickoff]` - Ken Brown, Gates Foundation, Domain Expert - Alan Hubbard, UC Berkeley, LAND surge team, Data Scientist - Andrew Mertens, UC Berkeley, LAND surge team, Data Scientist - Ben Arnold, UC Berkeley, Data Scientist - David Benkeser, UC Berkeley, Data Scientist - Jack Colford, UC Berkeley, Data Scientist and Domain Expert - Mark van der Laan, UC Berkeley, Data Scientist - Oleg Sofrygin, UC Berkeley, Data Scientist - Wenjing Zheng, UC Berkeley, Data Scientist ## Problem Statement `[id: problem]` `[description: Fill in the {} below appropriately.]` `[assignee: rally master]` `[due: rally kickoff]` This rally will {primary objective} for {target audience} so that {end user} can {primary use case} thereby {Foundation / PST objective}. ## Background `[id: background]` `[description: What has been done before? What are the deficiencies that are prompting the primary objective in this rally?]` `[assignee: rally master]` `[due: rally kickoff]` The Nutrition team is developing a strategy to identify and implement preventative interventions for wasting. Wasting is a significant problem world wide, however most interventions are currently focused on treating already exisiting cases of wasting and not preventing new cases. Expanding the knowledge of what kids are at risk for wasting and what the main drivers of both wasting episodes and recovery will allow researchers and funders to invest in the preventions most likely to have impact. ## Background Detail `[id: bg_detail]` `[description: Provide bullet points about additional detail, and/or links to figures stored on OSF that illustrate background.]` `[assignee: rally master or domain expert]` `[due: end of rally]` CONTENT HERE Include figures with the following: Figure: - {Link to figure} - {Informative title} - {Explanation of why figure is important} ## Motivation `[id: motivation]` `[description: What is the context for this rally and why is this rally expected to be important? Elaborate on the primary use case from the Problem Statement.]` `[assignee: rally master]` `[due: rally kickoff]` There currently exists good data on the prevalence of wasting in LMICs, there is minimal data on the incidence of wasting, especially age-specific incidence, severity rates, and information on the duration and recovery from wasting episodes. Knowing both the incidence of wasting episodes and the conditions that precede the episode and contribute to the recovery will allow for focused and targeted interventions to prevent wasting. ## HBGD 5 Questions `[id: hbgd5]` `[description: Choose below which of the HBGD 5 Questions this rally is most related to (delete the others).]` `[assignee: rally master]` `[due: rally kickoff]` - To what extent is growth faltering explained by pre- vs postnatal insults? - Are there disproportionately large contributions on growth faltering from certain pathwyas, and can we rank-order risk factors? ## Deliverables `[id: deliverables]` `[description: In addition to the final data story, what other deliverables are planned? This may be an algorithm, a publication manuscript, a Grand Challenges presentation.]` `[assignee: rally master]` `[due: rally kickoff]` Descriptive epidemiology of wasting using as many of the datasets in GHAP as possible. This will include age-specific incidence by season, by geography and be severity. Severity will be defined as WHZ <-3, WHZ >-3 to <-2. Completed preliminary model on risk factors for incidence and recovery of wasting at 12 months. ## Rally Questions/Goals `[id: question]` `[description: Has to be refined by rally team so that it represents the problem that you are working on.]` `[assignee: rally master]` `[due: rally kickoff]` 1. Decriptive epidemiology of wasting using longitudinal data including: Age-specific incidence / severity / duration and recovery / seasonality / geographic differences 2. Completion of ongoing work on a risk factor analysis for wasting at 12 months ## Dataset List `[id: data_list]` `[description: List the IDs of the studies used for this rally. You can find a list of the IDs (here)[http://bit.ly/hbgd-data].]` `[assignee: data scientist]` `[due: rally kickoff]` ## Data Description `[id: data_desc]` `[description: This is a descriptive paragraph helping to understand the context of the datasets that will be used in the in analysis.]` `[assignee: data scientist]` `[due: rally kickoff]` 13 studies encompassing 23 different cohorts. The studies included in this rally are longitudinal, have monthly anthropometric measures and are community studies with no inclusion criteria for specific conditions such as diarrhea. ## Outcome Variables `[id: data_outcomes]` `[description: What are the outcome variables in the datasets being used that we are interested in for this rally?]` `[assignee: data scientist]` `[due: rally kickoff / update throughout]` Incident wasting - change in WHZ from >-2 z to <-2 z Incident severe wasting - a change in WHZ from >-3z to <-3z ## Predictor Variables `[id: data_predictor]` `[description: List the predictor variables available in the data that will be used for the analysis.]` `[assignee: data scientist]` `[due: rally kickoff / update throughout]` This rally is descriptive epidemiology so no predictive variables. ## Methods `[id: methods]` `[description: A high-level one or two sentence description of the methods used in this rally.]` `[assignee: data scientist]` `[due: rally kickoff / update throughout]` - Systematically narrow down a list of studies to use in the analysis of wasting incidence and timing - Develop analysis pipeline to visualize and tabulate wasting in populations ## Methods Detail `[id: methods_detail]` `[description: Provide bullet points about additional methods detail, and/or links to figures stored on OSF that illustrate methods.]` `[assignee: data scientist]` `[due: end of rally]` - List of studies - Longitudinal studies of low- and middle-income countries - Measure children from 0-24 months - Monthly or higher frequency of measurements of child anthropometry - "Representative" cohorts (not exclusively enrolling extremely ill children) - Analysis Pipeline - Calculate incidence and prevalence of wasting - Calculate incidence of spontaneous recovery from wasting - Estimate duration of wasting episodes - Look at age-specific incidence rates of wasting and recovery from wasting - Apply to all studies selected Recovery from wasting : - A change in WHZ to >-2z - Child needed to maintain WHZ >-2z for 2 months to be considered "recovered" - Children only "at risk" for incident wasting episodes, and thus contribute person-time at risk for wasting, if they are not currently wasted and are classified as recovered Wasting duration: - Assumed that the episode started or ended at the midpoint between measurements. For example, if a child was not wasted at age 40 days, wasted at age 70 days and not wasted at age 100 and 130 days, the duration of the wasting episode was (70-40)/2 + (100-70)/2 = 30 days. Include figures with the following: Figure: - {Link to figure} - {Informative title} - {Explanation of why figure is important} ## Key Finding(s) `[id: findings]` `[description: Key findings are the final result, why we should care about it, and what the potential for future transformation arises from the result and learning.]` `[assignee: data scientist]` `[due: end of rally]` 1. There are clear differences across populations with the S. Asian cohorts having higher incidence of wasting and longer duration of wasting episodes compared with Latin American and African cohorts 2. Wasting is a transient condition, with episodes that last on average 1-4 months 3. Spontaneous recovery from wasting appears to be common, particularly in the 0-6 month window. 4. In high burden populations there are clearly children who recover and those who do not ## Value `[id: value]` `[description: What is the impact value of this work? What does it mean for-to-day operations?.]` `[assignee: data scientist]` `[due: end of rally]` - This is the first systematic description of wasting incidence across a large number of contemporary cohorts - These results also provide an example of how GHAP creates unprecedented opportunities to answer basic (and complex) epidemiologic questions across globally and historically important cohorts. - The descriptive epidemiology from this rally has provided information on what wasting looks like in different geographies, including times (0-6 months) and areas (S. Asia) to focus on for intervention ## Results Detail `[id: results_detail]` `[description: Provide bullet points about additional results detail, and/or links to figures stored on OSF that illustrate results.]` `[assignee: data scientist]` `[due: end of rally]` CONTENT HERE Include figures with the following: Figure: - {Link to figure} - {Informative title} - {Explanation of why figure is important} ## Future Plans `[id: plans]` `[description: Future plans are brief points on what is next; either further analysis needed, or how to begin to implement the findings into actionable knowledge.]` `[assignee: rally master or domain expert]` `[due: end of rally]` 1. Further refine the descriptive epidemiology of wasting report 2. finalize estimates across all cohorts in GHAP with monthly measurements 3. Study fixed characteristics associated with incident wasting (and recovery) such as birth characteristics, maternal characteristics and household characteristics 4. Study dynamic (time-varying) characteristics associated with incident wasting (and recovery). This requires a longitudinal causal model due to the reciprocal relationships between many exposures and WHZ/wasting. ## To be Continued? `[id: tbc]` `[description: Is this rally to be continued? (yes/no).]` `[assignee: rally master]` `[due: end of rally]` Yes. Next rally kicked off 8/14/17 ## Working Group Member Assessment `[id: wg_assess]` `[description: What value did this rally generate? Is the result earth-shattering or ho-hum? How was the process for you? What would you do different?]` `[assignee: rally master or domain expert]` `[due: end of rally]` CONTENT HERE ## Data Scientist Assessment `[id: ds_assess]` `[description: From the data scientist's perspective, why did / didn't this rally succeed?]` `[assignee: data scientist]` `[due: end of rally]` CONTENT HERE