Project 1: Deaths by tuberculosis

by Michel Wermelinger and Alexandre Campos, 14 July 2015, edited 5 April 2016, updated 18 October, 20 December 2017 and 20 November 2018.

This is the project notebook for the first part of The Open University's Learn to code for Data Analysis course.

In 2000, the United Nations set eight Millenium Development Goals (MDGs) to reduce poverty and diseases, improve gender equality and environmental sustainability, etc. Each goal is quantified and time-bound, to be achieved by the end of 2015. Goal 6 is to have halted and started reversing the spread of HIV, malaria and tuberculosis (TB). TB doesn't make headlines like Ebola, SARS (severe acute respiratory syndrome) and other epidemics, but is far deadlier. For more information, see the World Health Organisation (WHO) page http://www.who.int/gho/tb/en/.

Given the population and number of deaths due to TB in some countries during one year, the following questions will be answered:

What is the total, maximum, minimum and average number of deaths in that year?
Which countries have the most and the least deaths?
What is the death rate (deaths per 100,000 inhabitants) for each country?
Which countries have the lowest and highest death rate?

The death rate allows for a better comparison of countries with widely different population sizes.

The data

The data consists of total population and total number of deaths due to TB (excluding HIV) in 2013 in each of the South America (Argentina, Bolivia, Brazil, Chile, Colombia, Ecuador, Guyana, Paraguay, Peru, Suriname, Uruguay, Venezuela) countries.

The data was taken in July 2015 from http://apps.who.int/gho/data/node.main.POP107?lang=en (population) and http://apps.who.int/gho/data/node.main.1317?lang=en (deaths). The uncertainty bounds of the number of deaths were ignored.

The data was collected into an Excel file which should be in the same folder as this notebook.

In [1]:

import warnings
warnings.simplefilter('ignore', FutureWarning)

from pandas import *
all_data = read_excel('WHO POP TB all.xls')

# subselecting only South America countries, not including:
# - French Guiana (France) [https://en.wikipedia.org/wiki/French_Guiana]
# - Falkland Islands (UK) [https://en.wikipedia.org/wiki/Falkland_Islands]

# reference: 
# https://cmdlinetips.com/2018/02/how-to-subset-pandas-dataframe-based-on-values-of-a-column/
# https://pandas.pydata.org/pandas-docs/stable/generated/pandas.DataFrame.copy.html
# https://pandas.pydata.org/pandas-docs/stable/generated/pandas.DataFrame.reset_index.html
countries = [
    'Argentina',
    'Bolivia (Plurinational State of)', # Bolivia
    'Brazil',
    'Chile',
    'Colombia',
    'Ecuador',
    'Guyana',
    'Paraguay',
    'Peru',
    'Suriname',
    'Uruguay',
    'Venezuela (Bolivarian Republic of)' # Venezuela
]
data = all_data[all_data['Country'].isin(countries)].reset_index(drop=True).copy()
data

	Country	Population (1000s)	TB deaths
0	Argentina	41446	570.0
1	Bolivia (Plurinational State of)	10671	430.0
2	Brazil	200362	4400.0
3	Chile	17620	220.0
4	Colombia	48321	770.0
5	Ecuador	15738	320.0
6	Guyana	800	130.0
7	Paraguay	6802	200.0
8	Peru	30376	2300.0
9	Suriname	539	12.0
10	Uruguay	3407	40.0
11	Venezuela (Bolivarian Republic of)	30405	480.0

The range of the problem

The column of interest is the last one.

In [2]:

tbColumn = data['TB deaths']

The total number of deaths in 2013 is:

In [3]:

tbColumn.sum()

9872.0

The largest and smallest number of deaths in a single country are:

In [4]:

tbColumn.max()

4400.0

In [5]:

tbColumn.min()

12.0

From 12 to 4400 deaths is a large range. The average number of deaths, over all countries in the selected data, can give a better idea of the seriousness of the problem in each country. The average can be computed as the mean or the median. Given the wide range of deaths, the median is probably a more sensible average measure.

In [6]:

tbColumn.mean()

822.6666666666666

In [7]:

tbColumn.median()

375.0

The median is lower than the mean. This indicates that some of the countries had a high number of TB deaths in 2013, pushing the value of the mean up.

The most affected

To see the most affected countries, the table is sorted in ascending order by the last column, which puts those countries in the last rows.

In [8]:

data.sort_values('TB deaths')

	Country	Population (1000s)	TB deaths
9	Suriname	539	12.0
10	Uruguay	3407	40.0
6	Guyana	800	130.0
7	Paraguay	6802	200.0
3	Chile	17620	220.0
5	Ecuador	15738	320.0
1	Bolivia (Plurinational State of)	10671	430.0
11	Venezuela (Bolivarian Republic of)	30405	480.0
0	Argentina	41446	570.0
4	Colombia	48321	770.0
8	Peru	30376	2300.0
2	Brazil	200362	4400.0

The table raises the possibility that a large number of deaths may be partly due to a large population. To compare the countries on an equal footing, the death rate per 100,000 inhabitants is computed.

In [10]:

populationColumn = data['Population (1000s)']
data['TB deaths (per 100,000)'] = tbColumn * 100 / populationColumn
data.sort_values('TB deaths (per 100,000)')

	Country	Population (1000s)	TB deaths	TB deaths (per 100,000)
10	Uruguay	3407	40.0	1.174053
3	Chile	17620	220.0	1.248581
0	Argentina	41446	570.0	1.375284
11	Venezuela (Bolivarian Republic of)	30405	480.0	1.578688
4	Colombia	48321	770.0	1.593510
5	Ecuador	15738	320.0	2.033295
2	Brazil	200362	4400.0	2.196025
9	Suriname	539	12.0	2.226345
7	Paraguay	6802	200.0	2.940312
1	Bolivia (Plurinational State of)	10671	430.0	4.029613
8	Peru	30376	2300.0	7.571767
6	Guyana	800	130.0	16.250000

Conclusions

The South America countries had a total of 9872 deaths due to TB in 2013. The median shows that half of these coutries had fewer than 375 deaths. The mean (over 822) indicates that some countries had a high number. The least affected were Suriname and Uruguay, with 12 and 40 deaths respectively, and the most affected were Peru and Brazil with 2300 and 4400 deaths in a single year. However, taking the population size into account, the least affected were Uruguay and Chile with less than 1.3 deaths per 100 thousand inhabitants, and the most affected were Peru and Guyana with over 7.5 and 16 deaths respectively per 100,000 inhabitants.

One should not forget that most values are estimates, and that the chosen countries are a small sample of all the world's countries. Nevertheless, they convey the message that TB is still a major cause of fatalities, and that there is a huge disparity between countries, with several ones being highly affected.