Analysis of stroke admission numbers and types using NIHSS 11+ as a surrogate for large vessel occlusions

Aims:

• Analyse proportion of haemorrhagic vs ischaemic stroke by arrival time.

• Analyse proportion of ischaemic stroke by NIHSS 0-10 vs 11+ by arrival time.

# import libraries
import pandas as pd

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

Remove unknown stroke type or unknown NIHSS

mask = (data['S2StrokeType'].isnull() == False) & (data['S2NihssArrival'].isnull() == False)
proportion_removed = (data.shape[0] - mask.sum()) / data.shape[0]
data = data[mask]
all_arrivals_count = data.shape[0]
print(f'Percent data removed: {proportion_removed*100:0.1f}')

Percent data removed: 7.8

All admissions

Get proportion with haemorrhagic stroke:

prop_bleed = (data['S2StrokeType'] == 'Primary Intracerebral Haemorrhage').mean()
print (f'Percent haemorrhage: {prop_bleed * 100:0.1f}')

Percent haemorrhage: 11.5

Analyse infarction stroke:

mask = data['S2StrokeType'] == 'Infarction'
data_infaction = data[mask]
prop_nlvo = (data_infaction['S2NihssArrival'] < 11).mean()
prop_lvo = (data_infaction['S2NihssArrival'] > 10).mean()
print (f'Percent infarction with NIHSS 0-10: {prop_nlvo * 100:0.1f}')
print (f'Percent infarction with NIHSS 11+: {prop_lvo * 100:0.1f}')

Percent infarction with NIHSS 0-10: 77.2
Percent infarction with NIHSS 11+: 22.8

Restrict to admissions within 6 hours of known stroke onset

mask = (data['S1OnsetToArrival_min'] <= 360) & (data['S1OnsetToArrival_min'] >= 0)
prop_6_hour_arrival = mask.sum() / all_arrivals_count
print (f'Percent with known onset of up to 6 hours: {prop_6_hour_arrival * 100:0.1f}')
data = data[mask]

Percent with known onset of up to 6 hours: 42.9

Get proportion with haemorrhagic stroke:

prop_bleed = (data['S2StrokeType'] == 'Primary Intracerebral Haemorrhage').mean()
print (f'Percent haemorrhage: {prop_bleed * 100:0.1f}')

Percent haemorrhage: 13.6

Analyse infarction stroke:

mask = data['S2StrokeType'] == 'Infarction'
data_infaction = data[mask]
prop_nlvo = (data_infaction['S2NihssArrival'] < 11).mean()
prop_lvo = (data_infaction['S2NihssArrival'] > 10).mean()
print (f'Percent infarction with NIHSS 0-10: {prop_nlvo * 100:0.1f}')
print (f'Percent infarction with NIHSS 11+: {prop_lvo * 100:0.1f}')

Percent infarction with NIHSS 0-10: 70.5
Percent infarction with NIHSS 11+: 29.5

Restrict to admissions within 4 hours of known stroke onset

mask = (data['S1OnsetToArrival_min'] <= 240) & (data['S1OnsetToArrival_min'] >= 0)
prop_6_hour_arrival = mask.sum() / all_arrivals_count
print (f'Percent with known onset of up to 4 hours: {prop_6_hour_arrival * 100:0.1f}')
data = data[mask]

Percent with known onset of up to 4 hours: 37.1

Get proportion with haemorrhagic stroke:

prop_bleed = (data['S2StrokeType'] == 'Primary Intracerebral Haemorrhage').mean()
print (f'Percent haemorrhage: {prop_bleed * 100:0.1f}')

Percent haemorrhage: 14.1

Analyse infarction stroke:

mask = data['S2StrokeType'] == 'Infarction'
data_infaction = data[mask]
prop_nlvo = (data_infaction['S2NihssArrival'] < 11).mean()
prop_lvo = (data_infaction['S2NihssArrival'] > 10).mean()
print (f'Percent infarction with NIHSS 0-10: {prop_nlvo * 100:0.1f}')
print (f'Percent infarction with NIHSS 11+: {prop_lvo * 100:0.1f}')

Percent infarction with NIHSS 0-10: 68.9
Percent infarction with NIHSS 11+: 31.1

Summary

Admission type	All arrivals	Arrival within 6 hrs known onset	Arrival within 4 hrs known onset
Proportion all admissions	100	42.9	37.1
Proportion haemorrhagic	11.5	13.6	14.1
Proportion ischaemic	88.5	86.4	85.9
Proportion ischaemic with NIHSS 0-10	77.2	70.5	68.9
Proportion ischaemic with NIHSS 11+	22.8	29.5	31.1