Exercise in Pregnancy
The background
There is a recent publication: Exercise in Pregnancy and Risk of Postpartum Depression: A Randomised Controlled Trial
The publication is available via Wiley.
The challenge
In the publication the results are presented in a table and described verbally. Create an effective visualisation to convey the positive outcome of the trial.
You may imagine being at a conference presenting the data or even presenting it in an online post or article.
A description of the challenge can also be found here.
A recording of the session can be found here.
Visualisation
Visualising the Saccone et al. RCT
Code
"""
Wonderful Wednesdays Entry: Exercise in Pregnancy & Postpartum Depression
Rendered with matplotlib (mirrors the R ggplot2 version)
Source: Saccone G et al. BJOG 2026;133:211-217
"""
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec
import matplotlib.patches as mpatches
from matplotlib import rcParams
from scipy.ndimage import gaussian_filter1d
# --- Configuration ---
rcParams['font.family'] = 'sans-serif'
rcParams['font.sans-serif'] = ['DejaVu Sans', 'Arial', 'Helvetica']
rcParams['font.size'] = 10
COL_EX = '#0072B2'
COL_CT = '#D55E00'
COL_BEN = '#009E73'
COL_NEU = '#999999'
COL_TXT = '#333333'
COL_BG = '#FAFAFA'
COL_GRID = '#E8E8E8'
fig = plt.figure(figsize=(18, 13), facecolor='white')
gs = gridspec.GridSpec(2, 2, hspace=0.32, wspace=0.30,
left=0.06, right=0.96, top=0.89, bottom=0.06)
# =============================================================================
# PANEL A: Icon array — primary outcome
# =============================================================================
ax_a = fig.add_subplot(gs[0, 0])
ax_a.set_facecolor(COL_BG)
def draw_icon_grid(ax, n_events, n_total, x_offset, ncols=20, color_event=COL_CT):
nrows = int(np.ceil(n_total / ncols))
xs, ys, colors = [], [], []
for i in range(n_total):
c = i % ncols
row = nrows - 1 - (i // ncols)
xs.append(c + x_offset)
ys.append(row)
colors.append(color_event if i < n_events else '#D5D5D5')
ax.scatter(xs, ys, c=colors, s=28, marker='s', edgecolors='none', zorder=3)
return nrows
nrows_ex = draw_icon_grid(ax_a, 12, 199, x_offset=0)
draw_icon_grid(ax_a, 40, 199, x_offset=24)
ax_a.text(9.5, nrows_ex + 0.4, 'Exercise (n = 199)', ha='center',
fontsize=10, fontweight='bold', color=COL_EX)
ax_a.text(33.5, nrows_ex + 0.4, 'Control (n = 199)', ha='center',
fontsize=10, fontweight='bold', color=COL_CT)
ax_a.text(9.5, -1.2, '12/199 (6.0%)', ha='center', fontsize=9,
fontweight='bold', color=COL_EX)
ax_a.text(33.5, -1.2, '40/199 (20.1%)', ha='center', fontsize=9,
fontweight='bold', color=COL_CT)
ax_a.axvline(x=21.5, color='#808080', linewidth=0.5, linestyle='-', alpha=0.4)
ax_a.set_xlim(-1, 45)
ax_a.set_ylim(-2.5, nrows_ex + 1.5)
ax_a.set_aspect('equal')
ax_a.axis('off')
ax_a.set_title('A. EPDS \u2265 12 at 3 months postpartum (primary outcome)\n'
'RR 0.30 (95% CI 0.16\u20130.55), p < 0.001',
fontsize=11, fontweight='bold', color=COL_TXT, loc='left', pad=8)
leg_event = mpatches.Patch(color=COL_CT, label='Screen-positive (EPDS \u2265 12)')
leg_no = mpatches.Patch(color='#D5D5D5', label='Screen-negative')
ax_a.legend(handles=[leg_event, leg_no], loc='lower center', ncol=2,
fontsize=8, frameon=False, bbox_to_anchor=(0.5, -0.08))
# =============================================================================
# PANEL B: Forest plot — y-tick labels (tight) + legend
# =============================================================================
ax_b = fig.add_subplot(gs[0, 1])
ax_b.set_facecolor(COL_BG)
outcomes = [
('EPDS \u2265 12 (primary)', 0.30, 0.16, 0.55, True, True),
('EPDS \u2265 9', 0.56, 0.39, 0.80, True, False),
('PPD (DSM-V)', 0.59, 0.31, 1.14, False, False),
('NICU admission', 0.27, 0.08, 0.96, True, False),
('PTB < 37 wks', 0.67, 0.33, 1.35, False, False),
('GDM', 0.89, 0.58, 1.37, False, False),
('Caesarean delivery', 0.96, 0.66, 1.38, False, False),
('Excl. breastfeeding', 0.91, 0.62, 1.33, False, False),
]
n_out = len(outcomes)
# Category shading (depression outcomes = top 3)
ax_b.axhspan(n_out - 2.5, n_out + 0.5, color=COL_BEN, alpha=0.05, zorder=0)
# Reference line at RR=1
ax_b.axvline(x=1, color='#808080', linewidth=0.7, linestyle='--', zorder=1)
y_positions = []
for i, (label, rr, lo, hi, sig, prim) in enumerate(outcomes):
y = n_out - i
y_positions.append(y)
col = COL_BEN if sig else COL_NEU
marker = 'D' if prim else ('s' if i < 3 else 'o')
ms = 10 if prim else 7
ax_b.plot([lo, hi], [y, y], color='#4D4D4D', linewidth=1.2, zorder=2)
ax_b.plot(rr, y, marker=marker, color=col, markersize=ms, zorder=3,
markeredgecolor='white', markeredgewidth=0.5)
# RR label positioned to right of the plot data area
rr_text = f'{rr:.2f} ({lo:.2f}\u2013{hi:.2f})'
ax_b.text(1.7, y, rr_text, va='center', ha='left', fontsize=8,
fontweight='bold' if sig else 'normal', color=COL_TXT)
# Y-axis: outcome names as tick labels (tight against plot edge)
ax_b.set_yticks(y_positions)
ax_b.set_yticklabels([o[0] for o in outcomes], fontsize=9)
for tl in ax_b.get_yticklabels():
if 'primary' in tl.get_text():
tl.set_fontweight('bold')
ax_b.set_xscale('log')
ax_b.set_xlim(0.06, 2.2)
ax_b.set_ylim(0.3, n_out + 0.7)
ax_b.set_xticks([0.1, 0.25, 0.5, 1.0, 1.5])
ax_b.set_xticklabels(['0.1', '0.25', '0.5', '1.0', '1.5'])
ax_b.set_xlabel('Relative Risk (95% CI)', fontsize=9)
ax_b.spines['top'].set_visible(False)
ax_b.spines['right'].set_visible(False)
ax_b.spines['left'].set_visible(False)
ax_b.tick_params(axis='y', length=0, pad=4)
ax_b.grid(axis='x', color=COL_GRID, linewidth=0.3, zorder=0)
# Direction annotations
# Direction annotations - position below last outcome row
ax_b.text(0.15, 0.55, '\u2190 Favours exercise', fontsize=7.5, color='#808080')
ax_b.text(1.15, 0.55, 'Favours control \u2192', fontsize=7.5, color='#808080')
# Legend - well below the plot
leg_dia = plt.Line2D([0], [0], marker='D', color='w', markerfacecolor=COL_BEN,
markersize=8, label='Primary')
leg_sq = plt.Line2D([0], [0], marker='s', color='w', markerfacecolor=COL_NEU,
markersize=7, label='Secondary (depression)')
leg_ci = plt.Line2D([0], [0], marker='o', color='w', markerfacecolor=COL_NEU,
markersize=7, label='Maternal/perinatal')
leg_sig = mpatches.Patch(color=COL_BEN, alpha=0.3, label='Significant (p < 0.05)')
ax_b.legend(handles=[leg_dia, leg_sq, leg_ci, leg_sig], loc='lower center',
fontsize=7, frameon=True, framealpha=0.9, edgecolor=COL_GRID, ncol=4,
bbox_to_anchor=(0.5, -0.18))
ax_b.set_title('B. Forest plot: all trial outcomes\n'
'Green = statistically significant',
fontsize=11, fontweight='bold', color=COL_TXT, loc='left', pad=8)
# =============================================================================
# PANEL C: EPDS score distribution (simulated from mean +/- SD)
# =============================================================================
ax_c = fig.add_subplot(gs[1, 0])
ax_c.set_facecolor(COL_BG)
np.random.seed(2026)
sim_ex = np.clip(np.random.normal(5.1, 3.7, 199), 0, 30)
sim_ct = np.clip(np.random.normal(7.1, 5.2, 199), 0, 30)
# Set axis limits FIRST so annotations use correct coordinates
YMAX = 0.16
ax_c.set_ylim(0, YMAX)
ax_c.set_xlim(0, 25)
# Histograms
hist_bins = np.arange(0, 26, 1)
ax_c.hist(sim_ex, bins=hist_bins, density=True, alpha=0.35, color=COL_EX,
edgecolor='white', linewidth=0.3, label='Exercise', zorder=2)
ax_c.hist(sim_ct, bins=hist_bins, density=True, alpha=0.35, color=COL_CT,
edgecolor='white', linewidth=0.3, label='Control', zorder=2)
# Smoothed density curves
for data, col in [(sim_ex, COL_EX), (sim_ct, COL_CT)]:
counts, edges = np.histogram(data, bins=np.arange(0, 26, 0.5), density=True)
centres = (edges[:-1] + edges[1:]) / 2
smooth = gaussian_filter1d(counts, sigma=2)
ax_c.plot(centres, smooth, color=col, linewidth=2, zorder=3)
# Threshold lines
ax_c.axvline(x=9, color='#808080', linestyle=':', linewidth=1, zorder=1)
ax_c.axvline(x=12, color='#666666', linestyle='--', linewidth=1.2, zorder=1)
# Threshold annotations (explicit y positions)
ax_c.text(9.4, YMAX * 0.88, 'EPDS = 9\n(follow-up)',
fontsize=7.5, color='#808080', va='top')
ax_c.text(12.4, YMAX * 0.88, 'EPDS = 12\n(primary)',
fontsize=7.5, color='#666666', fontweight='bold', va='top')
# Mean markers
ax_c.annotate('mean\n5.1', xy=(5.1, 0), xytext=(5.1, YMAX * 0.92),
fontsize=8, fontweight='bold', color=COL_EX, ha='center',
arrowprops=dict(arrowstyle='-', color=COL_EX, lw=1))
ax_c.annotate('mean\n7.1', xy=(7.1, 0), xytext=(7.1, YMAX * 0.78),
fontsize=8, fontweight='bold', color=COL_CT, ha='center',
arrowprops=dict(arrowstyle='-', color=COL_CT, lw=1))
ax_c.set_xlabel('EPDS Score (3 months postpartum)', fontsize=9)
ax_c.set_ylabel('Density', fontsize=9)
ax_c.legend(fontsize=9, frameon=False, loc='upper right')
ax_c.spines['top'].set_visible(False)
ax_c.spines['right'].set_visible(False)
ax_c.grid(axis='y', color=COL_GRID, linewidth=0.3)
ax_c.set_title('C. EPDS score distributions: shift towards lower depression\n'
'Mean difference: \u22122.0 points (95% CI \u22122.89 to \u22121.11)',
fontsize=11, fontweight='bold', color=COL_TXT, loc='left', pad=8)
# =============================================================================
# PANEL D: Absolute risk differences + NNT
# =============================================================================
ax_d = fig.add_subplot(gs[1, 1])
ax_d.set_facecolor(COL_BG)
ard_outcomes = [
('EPDS \u2265 12\n(Primary)', 12/199, 40/199, True),
('EPDS \u2265 9', 35/199, 63/199, False),
('PPD\n(DSM-V)', 13/199, 22/199, False),
]
x_pos = np.arange(len(ard_outcomes))
ards, nnts, colors = [], [], []
for label, p_ex, p_ct, is_primary in ard_outcomes:
ard = (p_ct - p_ex) * 100
nnt = int(np.ceil(1 / (p_ct - p_ex)))
ards.append(ard)
nnts.append(nnt)
colors.append(COL_BEN if is_primary else COL_EX)
bars = ax_d.bar(x_pos, ards, width=0.55, color=colors, edgecolor='white',
linewidth=0.5, zorder=2)
for bar, ard, nnt in zip(bars, ards, nnts):
ax_d.text(bar.get_x() + bar.get_width()/2, bar.get_height() + 0.4,
f'ARD: {ard:.1f}%\nNNT: {nnt}',
ha='center', va='bottom', fontsize=10, fontweight='bold', color=COL_TXT)
ax_d.set_xticks(x_pos)
ax_d.set_xticklabels([o[0] for o in ard_outcomes], fontsize=9)
ax_d.set_ylabel('Absolute Risk Difference (%)', fontsize=9)
ax_d.set_ylim(0, max(ards) * 1.45)
ax_d.spines['top'].set_visible(False)
ax_d.spines['right'].set_visible(False)
ax_d.grid(axis='y', color=COL_GRID, linewidth=0.3)
ax_d.set_title('D. Clinical impact: absolute risk reduction & NNT\n'
'Number needed to exercise to prevent one case',
fontsize=11, fontweight='bold', color=COL_TXT, loc='left', pad=8)
leg_pri = mpatches.Patch(color=COL_BEN, label='Primary outcome')
leg_sec = mpatches.Patch(color=COL_EX, label='Secondary outcomes')
ax_d.legend(handles=[leg_pri, leg_sec], loc='upper right', fontsize=8, frameon=False)
# =============================================================================
# OVERALL TITLE & CAPTION
# =============================================================================
fig.suptitle(
'Exercise in Pregnancy Reduces Postpartum Depression: Visualising the Saccone et al. RCT',
fontsize=16, fontweight='bold', color=COL_TXT, x=0.06, ha='left', y=0.97
)
fig.text(0.06, 0.935,
'Saccone G et al. BJOG 2026;133:211\u2013217 | '
'Single-centre RCT, N = 398 low-risk singleton pregnancies | '
'Structured aerobic exercise (3\u00D760 min/wk, 1st trimester \u2192 35 wks) vs standard care',
fontsize=9.5, color='#737373', ha='left')
fig.text(0.06, 0.015,
'Source: Saccone et al. BJOG 2026; DOI: 10.1111/1471-0528.70010 | '
'Wonderful Wednesdays entry, Feb 2026 | Lovemore Gakava',
fontsize=8, color='#808080', ha='left')
plt.savefig('/home/claude/ww_submission/saccone_exercise_ppd.png',
dpi=200, bbox_inches='tight', facecolor='white')
plt.close()
print("Saved: saccone_exercise_ppd.png")