docs: update api results, part of #19

docs/api/app.results.tex

@@ -33,47 +33,46 @@ \subsubsection{Distributions}
 \subsection{ART Prevention Impact}\label{aa:res:api}
 The following figures illustrate the projected ART prevention impact (Dataset~B),
 stratified by various factors of heterogeneity and intervention contexts (colours).
-The left panels show the relative reduction in HIV incidence rate;
-the right panels show the relative reduction in cumulative new HIV infections;
+Left panels show the relative reduction in HIV incidence rate;
+right panels show the relative reduction in cumulative new HIV infections;
 both as compared to a base-case scenario reflecting status quo.
+If any study included multiple scenarios of ART scale-up,
+then each scenario was included separately;
+if any scenario reported multiple time horizons,
+each time horizon was included separately.
 The number of studies (scenarios) reporting
 incidence reduction, cumulative infections averted, both, or either was:
 \x{n/n.a.api.inc}~(\x{n/n.s.api.inc}),
 \x{n/n.a.api.chi}~(\x{n/n.s.api.chi}),
 \x{n/n.a.api.both}~(\x{n/n.s.api.both}), and
 \x{n/n.a.api}~(\x{n/n.s.api}), respectively.
-If any study included multiple scenarios of ART scale-up,
-then each scenario was included separately,
-but the size of each data point was reduced
-in proportion to the number of scenarios;
-so studies with only one scenario have the largest data points.
-Some scenarios have multiple data points if multiple time horizons were reported.
 If any factor could not be quantified due to missing data or varying values,
-the data point is grey.
-A small random offset has been added to the data points to reduce overlap.
+it was omitted from that plot.
+In box plots, the numbers of unique scenario time-horizons
+contributing to each box are given above it.
 \begin{figure}[H]
   \begin{subfigure}{0.5\linewidth}
-    \includegraphics[width=\linewidth]{{inc.s.Risk.both}.pdf}
+     \includegraphics[width=\linewidth]{{inc.s.Risk.both}.pdf}
     \caption{Reduction in HIV incidence (\%)}
   \end{subfigure}%
   \begin{subfigure}{0.5\linewidth}
-    \includegraphics[width=\linewidth]{{chi.s.Risk.both}.pdf}
+     \includegraphics[width=\linewidth]{{chi.s.Risk.both}.pdf}
     \caption{Cumulative HIV infections averted (\%)}
   \end{subfigure}
   \caption{Projected ART prevention benefits,
     stratified by factors of risk heterogeneity: whether models considered
     differences in sexual activity, key populations, and
-    ART cascade prioritized to key populations.
-    Subset of studies reporting both outcomes.}
+    ART cascade prioritized to key populations
+    (subset of studies reporting both outcomes)}
   \label{fig:api:both}
 \end{figure}
 \foreach \var/\title in \plotlistapi{
   \begin{figure}[H]
     \begin{subfigure}{0.5\linewidth}
-      \includegraphics[width=\linewidth]{{inc.s.\var}.pdf}
+      \includegraphics[width=\linewidth]{{inc.\var}.pdf}
     \end{subfigure}%
     \begin{subfigure}{0.5\linewidth}
-      \includegraphics[width=\linewidth]{{chi.s.\var}.pdf}
+      \includegraphics[width=\linewidth]{{chi.\var}.pdf}
     \end{subfigure}
     \caption{\title}
     \label{fig:api:\var}

docs/api/iter.tex

@@ -2,7 +2,6 @@
 \def\bibidB{Salomon2005,Abbas2006,Granich2009,Hallett2009b,Bacaer2010,Pretorius2010,Metzger2011,Yusuf2012,Andrews2012,Granich2012,Wagner2012,Abbas2013,Long2013,Cremin2013,Alsallaq2013,Nichols2014,Nichols2014a,Alistar2014,Eaton2014a,Ying2015,Low2015,Khademi2015,Gilbert2015,Heaton2015,Rahman2016,Gilbert2016,Blaizot2016,Ying2016,Barnighausen2016,Heffernan2016,Maheu-Giroux2017a,Maheu-Giroux2017,Volz2017,Blaizot2017,Mukandavire2018,Guillon2018,Akudibillah2018,Stuart2018,deMontigny2018,Hauser2019}
 \def\bibidAxB{Johnson2006,Baggaley2006,Wilson2006,Bacaer2008,Chigidi2009,Williams2010,Nyabadza2011,Barnighausen2012,Wagner2013,Decker2013,Wirtz2013,Shafer2014,Hove-Musekwa2014,Braithwaite2014,Nichols2014b,Abu-Raddad2014,Anderson2014,Alistar2014a,Cori2014,Stover2014,Wirtz2014,Korenromp2015,Knight2015,Kerr2015,Fraser2015,Kassa2015,Bekker2015,Shannon2015,Blaizot2015,Smith2016,Atun2016,Shattock2016,McGillen2016,Johnson2016,Sharma2016,Akudibillah2017,Alsallaq2017,Anderson2017,Chiu2017,Johnson2017b,Stuart2017,McGillen2017,Cremin2017,Ross2018,Anderson2018,Anderson2018a,Omondi2018,Woods2018,Stevens2018,Stopard2019,Beacroft2019a,Reidy2019,Omondi2019,Maheu-Giroux2019}
 \def\plotlistdist{
-  api.prev.cat/HIV prevalence at $t_0$ (\%),
   api.prev/HIV prevalence at $t_0$ (\%),
   api.inc/HIV incidence at $t_0$ (per 1000 PY),
   api.phase/HIV epidemic phase,
@@ -14,9 +13,9 @@
   act.HRM.p/Proportion of men in the client or highest activity male group
 }
 \def\plotlistapi{%
-  api.prev/HIV prevalence at $t_0$ (\%),
+  api.prev.cat/HIV prevalence at $t_0$ (\%),
   art.cd4/CD4 initiation criteria,
   api.phase/HIV epidemic phase,
-  art.cov/ART intervention coverage target,
-  art.rbeta/Relative infectiousness on ART
+  art.cov.cat/ART intervention coverage target,
+  art.rbeta.cat/Relative infectiousness on ART
 }

docs/api/results.tex

@@ -37,7 +37,7 @@ \subsection{Epidemic Context}
 \par
 ART prevention impacts were most often modelled in
 high-prevalence ({$>$10\%}) epidemics (\x{t0/n.prev.High} studies) and
-medium-prevalence ({1-10\%}) epidemics (\x{t0/n.prev.Mid}) (Figure~\ref{fig:api.prev.cat}).
+medium-prevalence ({1-10\%}) epidemics (\x{t0/n.prev.Mid}) (Figure~\ref{fig:api.prev}).
 No studies reported overall HIV prevalence of {$<$1\%} at time of intervention,
 although for \x{t0/n.prev.NA} studies, HIV prevalence was
 not reported or varied across simulated contexts/scenarios.
@@ -195,7 +195,7 @@ \subsection{ART Prevention Impact}
 Relative incidence reduction with ART scale-up
 as compared to status quo
 was reported in \x{n/n.a.api.inc} studies (\x{n/n.s.api.inc} scenarios);
-the proportion of cumulative infections averted due to ART scale-up
+proportion of cumulative infections averted due to ART scale-up
 was reported in \x{n/n.a.api.chi} (\x{n/n.s.api.chi});
 and \x{n/n.a.api.both} (\x{n/n.s.api.both}) reported both.
 Some scenarios reported these outcomes on multiple time horizons.
@@ -227,34 +227,31 @@ \subsection{ART Prevention Impact}
 \begin{figure}
   \begin{subfigure}{0.5\textwidth}
     \centering
-    \includegraphics[width=\textwidth]{{inc.s.Risk}.pdf}
+    \includegraphics[width=\textwidth]{{inc.Risk}.pdf}
     \caption{Reduction in HIV incidence}
     \label{fig:api:inc}
   \end{subfigure}%
   \begin{subfigure}{0.5\textwidth}
     \centering
-    \includegraphics[width=\textwidth]{{chi.s.Risk}.pdf}
+    \includegraphics[width=\textwidth]{{chi.Risk}.pdf}
     \caption{Cumulative HIV infections averted}
     \label{fig:api:chi}
   \end{subfigure}
-  \caption{Projected ART prevention benefits,
-    stratified by factors of risk heterogeneity: whether models considered
+  \caption{Projected ART prevention benefits over different time horizons,
+    stratified by levels of modelled risk heterogeneity: whether models considered
     differences in sexual activity, key populations, and
-    ART cascade prioritized to key populations}
+    ART cascade prioritized to key populations;
+    no scenarios considered lower cascade among key populations.}
   \label{fig:api}
   \floatfoot{
+    The numbers of unique scenario time-horizons
+    contributing to each box are given above it.
     The number of studies (scenarios) reporting
-    incidence reduction, cumulative infections averted, both, or either was:
+    incidence reduction, cumulative infections averted, both, and either was:
     \x{n/n.a.api.inc}~(\x{n/n.s.api.inc}),
     \x{n/n.a.api.chi}~(\x{n/n.s.api.chi}),
     \x{n/n.a.api.both}~(\x{n/n.s.api.both}), and
     \x{n/n.a.api}~(\x{n/n.s.api}), respectively (Dataset~B).
-    If any study included multiple scenarios of ART scale-up,
-    then each scenario was included as a separate data point,
-    but the size of each data point was reduced
-    in proportion to the number of scenarios in the study.
-    Some scenarios have multiple data points if multiple time horizons were reported.
-    A small random offset was added to all data points to reduce overlap.
     KP: key populations;
     priority: cascade transitions were faster for at least one step among KP vs overall;
     same: cascade transitions were assumed the same speed in KP as overall;
@@ -263,9 +260,9 @@ \subsection{ART Prevention Impact}
 \par
 ART prevention impacts were larger with
 longer time horizon,
-lower HIV prevalence (Figure~\ref{fig:api:api.prev}), and
+lower HIV prevalence (Figure~\ref{fig:api:api.prev.cat}), and
 greater ART eligibility (Figure~\ref{fig:api:art.cd4});
 however patterns were not consistent across
 epidemic phase (Figure~\ref{fig:api:api.phase}),
-ART coverage targets (Figure~\ref{fig:api:art.cov}), or
-relative infectiousness on ART (Figure~\ref{fig:api:art.rbeta}).
+ART coverage targets (Figure~\ref{fig:api:art.cov.cat}), or
+relative infectiousness on ART (Figure~\ref{fig:api:art.rbeta.cat}).