Decision Making in Neonatology

class: title-slide, bottom, left
background-image: url(images/img-baby-in-nicu1.jpg)
background-size: cover

.pull-left[
## Decision Making in Neonatology

### What can we learn from Health Economics?
Tim Disher, PhD, RN | slido.com #425115]

---

class: middle, center

# About Me

![:scale 20%](images/tim.png)

### Tim Disher, PhD, RN
#### Director - Evidence Synthesis and Data Analytics
EVERSANA

<a href="http://twitter.com/halifaxtim">  
<svg aria-hidden="true" role="img" viewBox="0 0 512 512" style="height:1em;width:1em;vertical-align:-0.125em;margin-left:auto;margin-right:auto;font-size:inherit;fill:currentColor;overflow:visible;position:relative;"><path d="M459.37 151.716c.325 4.548.325 9.097.325 13.645 0 138.72-105.583 298.558-298.558 298.558-59.452 0-114.68-17.219-161.137-47.106 8.447.974 16.568 1.299 25.34 1.299 49.055 0 94.213-16.568 130.274-44.832-46.132-.975-84.792-31.188-98.112-72.772 6.498.974 12.995 1.624 19.818 1.624 9.421 0 18.843-1.3 27.614-3.573-48.081-9.747-84.143-51.98-84.143-102.985v-1.299c13.969 7.797 30.214 12.67 47.431 13.319-28.264-18.843-46.781-51.005-46.781-87.391 0-19.492 5.197-37.36 14.294-52.954 51.655 63.675 129.3 105.258 216.365 109.807-1.624-7.797-2.599-15.918-2.599-24.04 0-57.828 46.782-104.934 104.934-104.934 30.213 0 57.502 12.67 76.67 33.137 23.715-4.548 46.456-13.32 66.599-25.34-7.798 24.366-24.366 44.833-46.132 57.827 21.117-2.273 41.584-8.122 60.426-16.243-14.292 20.791-32.161 39.308-52.628 54.253z"/></svg> @halifaxtim</a> |  <a href="https://github.com/timdisher"><svg aria-hidden="true" role="img" viewBox="0 0 496 512" style="height:1em;width:0.97em;vertical-align:-0.125em;margin-left:auto;margin-right:auto;font-size:inherit;fill:currentColor;overflow:visible;position:relative;"><path d="M165.9 397.4c0 2-2.3 3.6-5.2 3.6-3.3.3-5.6-1.3-5.6-3.6 0-2 2.3-3.6 5.2-3.6 3-.3 5.6 1.3 5.6 3.6zm-31.1-4.5c-.7 2 1.3 4.3 4.3 4.9 2.6 1 5.6 0 6.2-2s-1.3-4.3-4.3-5.2c-2.6-.7-5.5.3-6.2 2.3zm44.2-1.7c-2.9.7-4.9 2.6-4.6 4.9.3 2 2.9 3.3 5.9 2.6 2.9-.7 4.9-2.6 4.6-4.6-.3-1.9-3-3.2-5.9-2.9zM244.8 8C106.1 8 0 113.3 0 252c0 110.9 69.8 205.8 169.5 239.2 12.8 2.3 17.3-5.6 17.3-12.1 0-6.2-.3-40.4-.3-61.4 0 0-70 15-84.7-29.8 0 0-11.4-29.1-27.8-36.6 0 0-22.9-15.7 1.6-15.4 0 0 24.9 2 38.6 25.8 21.9 38.6 58.6 27.5 72.9 20.9 2.3-16 8.8-27.1 16-33.7-55.9-6.2-112.3-14.3-112.3-110.5 0-27.5 7.6-41.3 23.6-58.9-2.6-6.5-11.1-33.3 2.6-67.9 20.9-6.5 69 27 69 27 20-5.6 41.5-8.5 62.8-8.5s42.8 2.9 62.8 8.5c0 0 48.1-33.6 69-27 13.7 34.7 5.2 61.4 2.6 67.9 16 17.7 25.8 31.5 25.8 58.9 0 96.5-58.9 104.2-114.8 110.5 9.2 7.9 17 22.9 17 46.4 0 33.7-.3 75.4-.3 83.6 0 6.5 4.6 14.4 17.3 12.1C428.2 457.8 496 362.9 496 252 496 113.3 383.5 8 244.8 8zM97.2 352.9c-1.3 1-1 3.3.7 5.2 1.6 1.6 3.9 2.3 5.2 1 1.3-1 1-3.3-.7-5.2-1.6-1.6-3.9-2.3-5.2-1zm-10.8-8.1c-.7 1.3.3 2.9 2.3 3.9 1.6 1 3.6.7 4.3-.7.7-1.3-.3-2.9-2.3-3.9-2-.6-3.6-.3-4.3.7zm32.4 35.6c-1.6 1.3-1 4.3 1.3 6.2 2.3 2.3 5.2 2.6 6.5 1 1.3-1.3.7-4.3-1.3-6.2-2.2-2.3-5.2-2.6-6.5-1zm-11.4-14.7c-1.6 1-1.6 3.6 0 5.9 1.6 2.3 4.3 3.3 5.6 2.3 1.6-1.3 1.6-3.9 0-6.2-1.4-2.3-4-3.3-5.6-2z"/></svg> @timdisher</a>

---
# About EVERSANA
.large[
- Primarily consult for pharmaceutical and medical device companies
 - Some ad-hoc work with academia/health technology assessment agencies
- Health Economics and Outcomes Research (HEOR) group responsible for
 - Development of global health economic models
 - Adaptation of models for local decision makers
 - Trial/Claims/EHR analysis to support reimbursement]
 
---
layout: true
class: split-four

.row.bg1[.content.center[
#.white[Decisions are Local]
####.white[Background event rates, practices, and availability of resources]
]]

.row.bg2[.content.center[
#.white[Decisions are Personal]
####.white[Within a location the right decision depends on patient preferences]
]]

.row.bg3[.content.center[
#.white[Decision-making benefits from bespoke methods]
####.white[Many to choose from - MCDA and Ordinal models today]
]]

.row.bg4[.content.center[
#.white[Decision-making benefits from open data]
####.white[But truly open data is far away so we need a temporary solution]
]]

---
class: fade-row2 fade-row3 fade-row4
---
class: fade-row1 fade-row3 fade-row4
---
class: fade-row1 fade-row2 fade-row4
---
class: fade-row1 fade-row2 fade-row3
---
layout: false

# Objectives 
.large[.center[### Upon completion of this activity, participants will be able to]

1.	Describe the purpose of decision theoretic approaches
2.	Give examples of ways in which neonatal studies could include components of decision theory
3.	Identify potential roadblocks to generalization of the results of neonatal trials for decision making
4.	Develop analysis plans that allow for flexible adaptations of results to new locations
]

---
# Presentation Flow

### Approaches you can use based on published data
  - Neyman-Pearson hypothesis testing as a decision theoretic approach (you can change `\(\alpha\)` and `\(\beta\)`)
  - Using MCDA to make decisions based on commonly reported summaries

### Approaches that need individual participant data (real or **synthethic**)
  - A simple decision theory approach to RCTs (Ordinal models)
  - A package to make this portable

---
<iframe src="https://wall.sli.do/event/eqmqpzmh?section=927ba679-43c6-40b2-9024-282a10569e8d" style="position:fixed; top:0; left:0; bottom:0; right:0; width:100%; height:100%; border:none; margin:0; padding:0; overflow:hidden; z-index:999999;"></iframe>
---
## What I Love about Neonatal Research

.pull-left[
### Robust and Collaborative
- Commitment to randomized controlled trials
- Thinking with an eye to meta-analyses
- Neonatal networks  
- Multidiscplinary

### Family Integrated
- Care paradigms centered around families (eg, fiCARE)
- Increased call for family preferences in guidelines
]

.pull-right[
![](https://upload.wikimedia.org/wikipedia/en/thumb/3/3c/Cochrane_logo_stacked.svg/1200px-Cochrane_logo_stacked.svg.png)]

.footnote[Mitra, Dorling, and Johnston (2021)]
---
## What I Would Like to See More of

.pull-left[### Open Data

- Research will be more flexible and more portable if it's more open
  - Most common example is prediction tools
- What is preventing us from sharing the data itself?
  - Privacy rules (may not be the barrier we think it is)
  - Ownership/competition
]
.pull-right[### Decision Theory

- A set of methods/approach to analysis and interpretation of data that maximizes
"goodness"
  - Health economics, MCDA, win-ratio, and others
- We are already using pseudo decision theory by basing decisions off of statistical significance
  - Can we do better?
]
.footnote[Drummond, Sculpher, Claxton, Stoddart, and Torrance (2015); Lakens, Adolfi, Albers, Anvari, Apps, Argamon, Baguley, Becker, Benning, Bradford, and others (2018)]
---

## We Usually Can't Rely on Guidelines to Make Decisions for Us

.pull-left[ 
 ### Local Variability
- Variation in patient populations
- Variation in practices
- Variation in patient outcomes

]

.pull-right[
### Preferences are Personal
- All the methods we might be interested in require us to weight outcomes in
some way
  - Win-ratio and ordinal models: Ordinal ranking
  - MCDA: Ordinal or swing weighting
  - Cost-effectiveness/utility: Willingness to pay and valuing health states
  - Decision curves: Thresholds for treatment
 ]
---
class: inverse, middle, center

# Simple Tools | Synthetic Data | Standard Code

---
# Example Data

- The following sections of this presentation will use some fake data to walk through
potential approaches.

- Results were simulated for a hypothetical two-arm parallel randomized controlled
trial of very pre-term infants (N = 1350).

- Outcomes include:
  1. Mortality
  2. Sepsis (any)
  3. Severe IVH 
  4. NEC
  5. Bronchopulmonary Dysplasia

- Baseline rates were simulated using available Canadian event rates where possible
with a treatment that improves mortality, severe IVH, and sepsis but increases the
risk of bronchopulmonary dysplasia and NEC.

---
# Example Data | Results

---
# Simple Tools | Can we change `\(\alpha\)` and `\(\beta\)`?

- Trials designed under a Neyman-Pearson decision framework provide a framework
for action based on the results of a significance test. Most commonly we are testing
that a treatment effect is different than zero. We design the trial to:
 - Have 1 - `\(\beta\)` (power) probability of detecting an effect
 - No more than `\(\alpha\)` of claiming an effect exists when the true difference is zero.
 
- These are typically set at `\(\beta\)` = 0.8 and `\(\alpha\)` = 0.05 which suggests that
is it 4 times worse to claim an effect exists when it doesn't.

>This .80 desired power convention is offered with the hope that it will be ignored whenever an investigator can find a basis in his substantive concerns in his specific research investigation to choose a value ad hoc  
 - Cohen (1998)

---
# Simple Tools | Changing `\(\alpha\)` and `\(\beta\)`

- Imagine that our simulated trial was powered to find a decrease in sepsis. Maybe
locally we are also concerned that the mechanism of action of the treatment is such
that we're concerned about an increase in BPD.
  - At our center we have very good outcomes for neonates with sepsis but have
  found BPD is associated with poor developmental outcomes

- We decide that we won't evaluate a treatment further if there is treatment
increases BPD by more than 8.5%. Based on an expected baseline event rate of 36% in
this population this would equate to an odds ratio of 1.43.

- Instead of a 4:1 ratio of the importance of false negative/positives we are
willing to risk 2:1. Since we can't change the n we can just find the threshold
for `\(\alpha\)` where the ratio is 2:1.
  - This is 0.08 which when compared to trial analysis p-value of 0.059 would lead us to "act as if"
  there were harm and not implement the intervention.

---
# Simple Tools | MCDA Overview

- Trials typically provide us with estimates of effects on multiple outcomes that
are relevant for decision making.

- The benefit portion of a decision theoretic model should bring all relevant
outcomes onto a weighted scale that helps us to decide whether one treatment
is better than the other over all
  - In health economic models this is often health utilities

- If preferences are linear and additive
  - Weights can be explicitly elicited, ranked ordinaly, or we can use random
  samples from all feasible weights.
  - Calculate the probability that a treatment is "best" given estimates of event
  rates and their uncertainty alongside weights (with or without uncertainty)
  - Can also calculate the vector of weights needed to prefer one treatment over
  another. This can be a nice tool to logic check your implied preferences.

.footnote[Tervonen, Naci, van
Valkenhoef, Ades, Angelis, Hillege, and Postmus (2015)]

---

# Simple Tools | MCDA applied

When using a "preference free" model treatment has the highest first rank acceptability.
Central weights show that choosing treatment suggests lower weight placed on
BPD and NEC. Also shows that we can confidently choose BPD but requires us to care
more about BPD than any other outcome
 - Consider how decisions would change as baseline events change

#cypbsflccf .gt_table {
  display: table;
  border-collapse: collapse;
  margin-left: auto;
  margin-right: auto;
  color: #333333;
  font-size: 16px;
  font-weight: normal;
  font-style: normal;
  background-color: #FFFFFF;
  width: auto;
  border-top-style: solid;
  border-top-width: 2px;
  border-top-color: #A8A8A8;
  border-right-style: none;
  border-right-width: 2px;
  border-right-color: #D3D3D3;
  border-bottom-style: solid;
  border-bottom-width: 2px;
  border-bottom-color: #A8A8A8;
  border-left-style: none;
  border-left-width: 2px;
  border-left-color: #D3D3D3;
}

#cypbsflccf .gt_heading {
  background-color: #FFFFFF;
  text-align: center;
  border-bottom-color: #FFFFFF;
  border-left-style: none;
  border-left-width: 1px;
  border-left-color: #D3D3D3;
  border-right-style: none;
  border-right-width: 1px;
  border-right-color: #D3D3D3;
}

#cypbsflccf .gt_title {
  color: #333333;
  font-size: 125%;
  font-weight: initial;
  padding-top: 4px;
  padding-bottom: 4px;
  border-bottom-color: #FFFFFF;
  border-bottom-width: 0;
}

#cypbsflccf .gt_subtitle {
  color: #333333;
  font-size: 85%;
  font-weight: initial;
  padding-top: 0;
  padding-bottom: 4px;
  border-top-color: #FFFFFF;
  border-top-width: 0;
}

#cypbsflccf .gt_bottom_border {
  border-bottom-style: solid;
  border-bottom-width: 2px;
  border-bottom-color: #D3D3D3;
}

#cypbsflccf .gt_col_headings {
  border-top-style: solid;
  border-top-width: 2px;
  border-top-color: #D3D3D3;
  border-bottom-style: solid;
  border-bottom-width: 2px;
  border-bottom-color: #D3D3D3;
  border-left-style: none;
  border-left-width: 1px;
  border-left-color: #D3D3D3;
  border-right-style: none;
  border-right-width: 1px;
  border-right-color: #D3D3D3;
}

#cypbsflccf .gt_col_heading {
  color: #333333;
  background-color: #FFFFFF;
  font-size: 100%;
  font-weight: normal;
  text-transform: inherit;
  border-left-style: none;
  border-left-width: 1px;
  border-left-color: #D3D3D3;
  border-right-style: none;
  border-right-width: 1px;
  border-right-color: #D3D3D3;
  vertical-align: bottom;
  padding-top: 5px;
  padding-bottom: 6px;
  padding-left: 5px;
  padding-right: 5px;
  overflow-x: hidden;
}

#cypbsflccf .gt_column_spanner_outer {
  color: #333333;
  background-color: #FFFFFF;
  font-size: 100%;
  font-weight: normal;
  text-transform: inherit;
  padding-top: 0;
  padding-bottom: 0;
  padding-left: 4px;
  padding-right: 4px;
}

#cypbsflccf .gt_column_spanner_outer:first-child {
  padding-left: 0;
}

#cypbsflccf .gt_column_spanner_outer:last-child {
  padding-right: 0;
}

#cypbsflccf .gt_column_spanner {
  border-bottom-style: solid;
  border-bottom-width: 2px;
  border-bottom-color: #D3D3D3;
  vertical-align: bottom;
  padding-top: 5px;
  padding-bottom: 6px;
  overflow-x: hidden;
  display: inline-block;
  width: 100%;
}

#cypbsflccf .gt_group_heading {
  padding: 8px;
  color: #333333;
  background-color: #FFFFFF;
  font-size: 100%;
  font-weight: initial;
  text-transform: inherit;
  border-top-style: solid;
  border-top-width: 2px;
  border-top-color: #D3D3D3;
  border-bottom-style: solid;
  border-bottom-width: 2px;
  border-bottom-color: #D3D3D3;
  border-left-style: none;
  border-left-width: 1px;
  border-left-color: #D3D3D3;
  border-right-style: none;
  border-right-width: 1px;
  border-right-color: #D3D3D3;
  vertical-align: middle;
}

#cypbsflccf .gt_empty_group_heading {
  padding: 0.5px;
  color: #333333;
  background-color: #FFFFFF;
  font-size: 100%;
  font-weight: initial;
  border-top-style: solid;
  border-top-width: 2px;
  border-top-color: #D3D3D3;
  border-bottom-style: solid;
  border-bottom-width: 2px;
  border-bottom-color: #D3D3D3;
  vertical-align: middle;
}

#cypbsflccf .gt_from_md > :first-child {
  margin-top: 0;
}

#cypbsflccf .gt_from_md > :last-child {
  margin-bottom: 0;
}

#cypbsflccf .gt_row {
  padding-top: 8px;
  padding-bottom: 8px;
  padding-left: 5px;
  padding-right: 5px;
  margin: 10px;
  border-top-style: solid;
  border-top-width: 1px;
  border-top-color: #D3D3D3;
  border-left-style: none;
  border-left-width: 1px;
  border-left-color: #D3D3D3;
  border-right-style: none;
  border-right-width: 1px;
  border-right-color: #D3D3D3;
  vertical-align: middle;
  overflow-x: hidden;
}

#cypbsflccf .gt_stub {
  color: #333333;
  background-color: #FFFFFF;
  font-size: 100%;
  font-weight: initial;
  text-transform: inherit;
  border-right-style: solid;
  border-right-width: 2px;
  border-right-color: #D3D3D3;
  padding-left: 12px;
}

#cypbsflccf .gt_summary_row {
  color: #333333;
  background-color: #FFFFFF;
  text-transform: inherit;
  padding-top: 8px;
  padding-bottom: 8px;
  padding-left: 5px;
  padding-right: 5px;
}

#cypbsflccf .gt_first_summary_row {
  padding-top: 8px;
  padding-bottom: 8px;
  padding-left: 5px;
  padding-right: 5px;
  border-top-style: solid;
  border-top-width: 2px;
  border-top-color: #D3D3D3;
}

#cypbsflccf .gt_grand_summary_row {
  color: #333333;
  background-color: #FFFFFF;
  text-transform: inherit;
  padding-top: 8px;
  padding-bottom: 8px;
  padding-left: 5px;
  padding-right: 5px;
}

#cypbsflccf .gt_first_grand_summary_row {
  padding-top: 8px;
  padding-bottom: 8px;
  padding-left: 5px;
  padding-right: 5px;
  border-top-style: double;
  border-top-width: 6px;
  border-top-color: #D3D3D3;
}

#cypbsflccf .gt_striped {
  background-color: rgba(128, 128, 128, 0.05);
}

#cypbsflccf .gt_table_body {
  border-top-style: solid;
  border-top-width: 2px;
  border-top-color: #D3D3D3;
  border-bottom-style: solid;
  border-bottom-width: 2px;
  border-bottom-color: #D3D3D3;
}

#cypbsflccf .gt_footnotes {
  color: #333333;
  background-color: #FFFFFF;
  border-bottom-style: none;
  border-bottom-width: 2px;
  border-bottom-color: #D3D3D3;
  border-left-style: none;
  border-left-width: 2px;
  border-left-color: #D3D3D3;
  border-right-style: none;
  border-right-width: 2px;
  border-right-color: #D3D3D3;
}

#cypbsflccf .gt_footnote {
  margin: 0px;
  font-size: 90%;
  padding: 4px;
}

#cypbsflccf .gt_sourcenotes {
  color: #333333;
  background-color: #FFFFFF;
  border-bottom-style: none;
  border-bottom-width: 2px;
  border-bottom-color: #D3D3D3;
  border-left-style: none;
  border-left-width: 2px;
  border-left-color: #D3D3D3;
  border-right-style: none;
  border-right-width: 2px;
  border-right-color: #D3D3D3;
}

#cypbsflccf .gt_sourcenote {
  font-size: 90%;
  padding: 4px;
}

#cypbsflccf .gt_left {
  text-align: left;
}

#cypbsflccf .gt_center {
  text-align: center;
}

#cypbsflccf .gt_right {
  text-align: right;
  font-variant-numeric: tabular-nums;
}

#cypbsflccf .gt_font_normal {
  font-weight: normal;
}

#cypbsflccf .gt_font_bold {
  font-weight: bold;
}

#cypbsflccf .gt_font_italic {
  font-style: italic;
}

#cypbsflccf .gt_super {
  font-size: 65%;
}

#cypbsflccf .gt_footnote_marks {
  font-style: italic;
  font-size: 65%;
}
</style>
<div id="cypbsflccf" style="overflow-x:auto;overflow-y:auto;width:auto;height:auto;"><table class="gt_table">
  <thead class="gt_header">
    <tr>
      <th colspan="8" class="gt_heading gt_title gt_font_normal" style>MCDA Summaries</th>
    </tr>
    <tr>
      <th colspan="8" class="gt_heading gt_subtitle gt_font_normal gt_bottom_border" style></th>
    </tr>
  </thead>
  <thead class="gt_col_headings">
    <tr>
      <th class="gt_col_heading gt_center gt_columns_bottom_border" rowspan="2" colspan="1">Arm</th>
      <th class="gt_col_heading gt_center gt_columns_bottom_border" rowspan="2" colspan="1">First Rank Acceptability</th>
      <th class="gt_col_heading gt_center gt_columns_bottom_border" rowspan="2" colspan="1">Confidence Factor</th>
      <th class="gt_center gt_columns_top_border gt_column_spanner_outer" rowspan="1" colspan="5">
        <span class="gt_column_spanner">Central Weights</span>
      </th>
    </tr>
    <tr>
      <th class="gt_col_heading gt_columns_bottom_border gt_center" rowspan="1" colspan="1">Mortality</th>
      <th class="gt_col_heading gt_columns_bottom_border gt_center" rowspan="1" colspan="1">Severe IVH</th>
      <th class="gt_col_heading gt_columns_bottom_border gt_center" rowspan="1" colspan="1">Sepsis</th>
      <th class="gt_col_heading gt_columns_bottom_border gt_center" rowspan="1" colspan="1">BPD</th>
      <th class="gt_col_heading gt_columns_bottom_border gt_center" rowspan="1" colspan="1">NEC</th>
    </tr>
  </thead>
  <tbody class="gt_table_body">
    <tr>
      <td class="gt_row gt_left">Control</td>
      <td class="gt_row gt_right">0.24</td>
      <td class="gt_row gt_right">0.75</td>
      <td class="gt_row gt_right">0.16</td>
      <td class="gt_row gt_right">0.17</td>
      <td class="gt_row gt_right">0.07</td>
      <td class="gt_row gt_right">0.33</td>
      <td class="gt_row gt_right">0.27</td>
    </tr>
    <tr>
      <td class="gt_row gt_left">Treatment</td>
      <td class="gt_row gt_right">0.76</td>
      <td class="gt_row gt_right">1.00</td>
      <td class="gt_row gt_right">0.21</td>
      <td class="gt_row gt_right">0.21</td>
      <td class="gt_row gt_right">0.24</td>
      <td class="gt_row gt_right">0.16</td>
      <td class="gt_row gt_right">0.18</td>
    </tr>
  </tbody>
  
  
</table></div>

---

# Synthetic Data

- If our decisions need to be based on combinations of events or our preferences
are not linearly additive then we need actual individual participant data
  - We usually will not have access or access could take months and/or be 
  prohibitively expensive

- Solution: Synthetic data

- Synthetic data that will generate approximately the same results can be generated from the original
  - Re-identification risk? See [this report](http://hummedia.manchester.ac.uk/institutes/cmist/archive-publications/reports/2015-02%20-Report%20on%20disclosure%20risk%20analysis%20of%20synthpop%20synthetic%20versions%20of%20LCF_%20final.pdf)
  - Alternative: Use the same software but only share the actual models. Simulate
  new datasets using local patient characteristics

.footnote[Nowok, Raab, Dibben, and others (2016)]
---
# Synthetic Data | Applied Example

Using our example data we create a new synthetic dataset based on fitting binomial
models first to mortality given treatemnt then IVH given mortality and treatment, etc..

---
# Synthetic Data | Ordinal Analysis

- If we order our outcomes from best (uncomplicated discharge) to worst (death)
we can create an ordinal outcome that acts as a quasi utility based on worst
outcome.
  - No events > Sepsis > NEC > CLD > IVH > Death
  - For simplicity we just compare to a single synthetic dataset. In real applications
  you may simulate multiple datasets and take estimates averaged over them all.
  
<img src="improving-neo-decisions_files/figure-html/unnamed-chunk-6-1.svg" width="100%" />

---

# Standard Code | Next Steps

- {neoDecision} is an in development R package to provide a simple interface to methods described
today (and more later). It is a wrapper for multiple existing packages.
- Goals (more or less in order)
  - Simple interface to relevant formulations of MCDAs (via {smaa} and {hitandrun})
  - Creation of synthetic data (via {synthpop}) with the option to share actual
  datasets or models used to create them
  - API to an open database of neonatal data (true IPD, synthetic data, models for synthetic data, results from a set of standard models)
  - Point and click interfaces to create data, add to database, and create relevant models

---
# Summary

- Decisions are local and personal and the same data can (and should!) lead to 
different decisions in different environments

- We made several different decisions from our simulated trial
  - Original analysis suggests effective for sepsis
  - Changing value of false positives/false negatives leads to rejection based on increase in BPD
  - MCDA suggests treatment is effective across all outcomes unless you value BPD
  much higher than mortality
  - Ordinal analysis point estimate suggests treatment may decrease utility (increases
  scores)

- We can start using basic methods today with an eye to making more flexible methods
available through synthetic data
---
<iframe src="https://wall.sli.do/event/eqmqpzmh?section=927ba679-43c6-40b2-9024-282a10569e8d" style="position:fixed; top:0; left:0; bottom:0; right:0; width:100%; height:100%; border:none; margin:0; padding:0; overflow:hidden; z-index:999999;"></iframe>
---
class: inverse, center, middle
# Questions?

---
# References
[1] M. F. Drummond, M. J. Sculpher, K. Claxton, et al. _Methods for the
economic evaluation of health care programmes_. Oxford university
press, 2015.

[2] D. Lakens, F. G. Adolfi, C. J. Albers, et al. "Justify your alpha".
In: _Nature Human Behaviour_ 2.3 (2018), pp. 168-171.

[3] S. Mitra, J. Dorling, and B. C. Johnston. "Optimizing practice
guidelines through incorporating patient and family values and
preferences". In: _Seminars in Fetal and Neonatal Medicine_. Elsevier.
2021, p. 101194.

[4] B. Nowok, G. M. Raab, C. Dibben, et al. "synthpop: Bespoke creation
of synthetic data in R". In: _Journal of statistical software_ 74.11
(2016), pp. 1-26.

[5] T. Tervonen, H. Naci, G. van Valkenhoef, et al. "Applying multiple
criteria decision analysis to comparative benefit-risk assessment:
choosing among statins in primary prevention". In: _Medical Decision
Making_ 35.7 (2015), pp. 859-871.