Analysis of stroke type and pre-stroke modified Rankin Scale by age

The proportion of stroke types (haemorrhagic, nlvo ischaemic, lvo ischaemic) are shown by age.

NIHSS of 0-10, and 11+, are taken as surrogates of nlvo and lvo respectively.

The average mRS (modified Rankin Scale) is shown by age.

# import libraries
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd

# Change default colour scheme:
plt.style.use('seaborn-colorblind')

# import data
data = pd.read_csv(
    './../data/2019-11-04-HQIP303-Exeter_MA.csv', low_memory=False)

Get occurance of ischaemic and haemorrgahic by age

ages = []
for group in data.S1AgeOnArrival.values:
    minage, maxage = group.split(',')
    
    minage = int(''.join(list(minage)[1:]))
    maxage = int(''.join(list(maxage)[:-1]))
    
    ages.append(np.median([minage,maxage]))
    
data['Age_midpoint'] = ages

# Censor data to 40-100
mask = (data['Age_midpoint'] > 50) & (data['Age_midpoint'] <100)
data = data[mask]

# Remove unknown stroke stype or NIHSS
data.dropna(subset=['S2StrokeType'], inplace=True)
data.dropna(subset=['S2NihssArrival'], inplace=True)

# Add row for counting
data['count'] = 1

results = pd.DataFrame()
results.index.name='Age midpoint'

# Count all by age 
results['all'] = data.groupby('Age_midpoint').count()['count']

# Get ischaemic strokes
mask = data['S2StrokeType'] == 'Infarction'
results['infarction'] = data[mask].groupby('Age_midpoint').count()['count']

# Get haemorrhagic stroke
mask = data['S2StrokeType'] == 'Primary Intracerebral Haemorrhage'
results['haemorrhage'] = data[mask].groupby('Age_midpoint').count()['count']

# Calculate proportion haemorrhagic
results['prop_haemorrhage'] = results['haemorrhage'] / results['all']

# Show results
results

	all	infarction	haemorrhage	prop_haemorrhage
Age midpoint
52.5	9334	8346	988	0.105850
57.5	12394	11235	1159	0.093513
62.5	15967	14359	1608	0.100708
67.5	21406	19130	2276	0.106325
72.5	28421	25049	3372	0.118645
77.5	33288	29138	4150	0.124670
82.5	37404	32638	4766	0.127420
87.5	32585	28577	4008	0.123001
92.5	18260	16298	1962	0.107448
97.5	5489	5014	475	0.086537

Add breakdown of ischaemic strokes by nLVO (NIHSS 0-10) and LVO (NIHSS 11+)

# NIHSS 0-10 is a surrogate for non large vessel occlusions
mask = (data['S2StrokeType'] == 'Infarction') & (data['S2NihssArrival'] < 11)
results['nlvo'] = data[mask].groupby('Age_midpoint').count()['count']

# NIHSS 11+ is a surrogate for non large vessel occlusions
mask = (data['S2StrokeType'] == 'Infarction') & (data['S2NihssArrival'] >10)
results['lvo'] = data[mask].groupby('Age_midpoint').count()['count']

# Calculate proportions
results['prop_nlvo'] = results['nlvo'] / results['all']
results['prop_lvo'] = results['lvo'] / results['all']

# Check proportions add up
results['checksum'] = \
    results['prop_haemorrhage'] + results['prop_nlvo'] + results['prop_lvo']

# Show results
results

	all	infarction	haemorrhage	prop_haemorrhage	nlvo	lvo	prop_nlvo	prop_lvo	checksum
Age midpoint
52.5	9334	8346	988	0.105850	7188	1158	0.770088	0.124063	1.0
57.5	12394	11235	1159	0.093513	9668	1567	0.780055	0.126432	1.0
62.5	15967	14359	1608	0.100708	12197	2162	0.763888	0.135404	1.0
67.5	21406	19130	2276	0.106325	16022	3108	0.748482	0.145193	1.0
72.5	28421	25049	3372	0.118645	20452	4597	0.719609	0.161747	1.0
77.5	33288	29138	4150	0.124670	22826	6312	0.685713	0.189618	1.0
82.5	37404	32638	4766	0.127420	24391	8247	0.652096	0.220484	1.0
87.5	32585	28577	4008	0.123001	19777	8800	0.606936	0.270063	1.0
92.5	18260	16298	1962	0.107448	10305	5993	0.564348	0.328204	1.0
97.5	5489	5014	475	0.086537	2873	2141	0.523410	0.390053	1.0

Plot results

fig = plt.figure(figsize=(5,5))
ax = fig.add_subplot()
ax.plot(results['prop_haemorrhage'], marker='o', label='haemorrhage')
ax.plot(results['prop_nlvo'], marker='s', label='infarction: nlvo (NIHSS 0-10)')
ax.plot(results['prop_lvo'], marker = '^', label='infarction: lvo (NIHSS 11+)')
ax.set_ylabel('Proportion of patients')
ax.set_xlabel('Age')
ax.set_ylim(0, 1)
ax.grid()
plt.legend()
plt.savefig('./output/stroke_type_by_age.jpg', dpi=300)
plt.show()

Show mRS by age

mrs_results = pd.DataFrame()
mrs_results.index.name='Age midpoint'
# Get prestroke mRS for all patients
mrs_results['all'] = data.groupby('Age_midpoint').mean()['S2RankinBeforeStroke']

# Get prestroke mRS for haemorrgagic patients
mask = data['S2StrokeType'] == 'Primary Intracerebral Haemorrhage'
mrs_results['haemorrhage'] = \
    data[mask].groupby('Age_midpoint').mean()['S2RankinBeforeStroke']

# Get prestroke mRS for nlvo patients
mask = (data['S2StrokeType'] == 'Infarction') & (data['S2NihssArrival'] < 11)
mrs_results['nlvo'] = \
    data[mask].groupby('Age_midpoint').mean()['S2RankinBeforeStroke']

# Get prestroke mRS for lvo patients
mask = (data['S2StrokeType'] == 'Infarction') & (data['S2NihssArrival'] >10)
mrs_results['lvo'] = \
    data[mask].groupby('Age_midpoint').mean()['S2RankinBeforeStroke']

fig = plt.figure(figsize=(5,5))
ax = fig.add_subplot()
ax.plot(mrs_results['haemorrhage'], marker = 'o', label='haemorrhagic')
ax.plot(mrs_results['nlvo'], marker = 's', label='infarction: nlvo (NIHSS 0-10)')
ax.plot(mrs_results['lvo'], marker = '^', label='infarction: lvo (NIHSS 10+)')
ax.set_ylabel('Mean mRS before stroke')
ax.set_xlabel('Age')
ax.set_ylim(0, 3)
ax.legend()
ax.grid()
plt.savefig('./output/mrs_by_age.jpg', dpi=300)
plt.show()