References (.csv)

Download original .csv file from GitHub

	Unnamed: 0	Unnamed: 1
0
1		“Life(style)” characteristics
2
3
4		urban - rural
5
6		baseline in 2018:
7
8		17% urban
9		83% rural
10
11		assume this is time fixed and is allocated at birth to new births, with same status as mother
12
13		source: DHS 2015
14
15		Comment: Malawi is urbanising I think, so perhaps increase urban and reduce rural by 1% per year? We could use the 2018 census data – should be available end of this year / early next year to see what it is in 2018 and compare with 2008 census and 2010 and 2015 DHS to see trend i.e. whether 1% increase per year seems about right.
16		Response: according to world bank the recent proportion urban is increasing by just 0.25% per year. I’m happy for us tp include this. https://data.worldbank.org/indicator/SP.RUR.TOTL.ZS?locations=MW
17
18		wealth level
19		(since recorded in DHS I suggest we use this instead of socio-economic status)
20
21		(wealth level is initially based on wealth quintile from 2015 DHS but distribution will not remain uniform if fertility and death rate differ by wealth level)
22
23		baseline in 2018 (assuming not changed since 2015 DHS):
24
25		if urban = 1 then
26		75% wealth_level = 1
27		16% wealth_level = 2
28		5% wealth_level = 3
29		2% wealth_level = 4
30		2% wealth_level = 5
31
32		if urban = 0 then
33		11% wealth_level = 1
34		21% wealth_level = 2
35		23% wealth_level = 3
36		23% wealth_level = 4
37		23% wealth_level = 5
38
39		source: DHS 2015
40
41		assume this is time fixed and allocated at birth with same status as mother.
42
43
44		tobacco use
45		(much tobacco use is not in fact cigarettes)
46
47		baseline in 2018:
48
49		if age < 20 and male then percent using tobacco = 1% x wealth_level
50		(i.e. 5 times higher in lowest level)
51		if 20 <= age < 40 and male then percent using tobacco = 4% x wealth_level
52		if 40 <= age and male then percent using tobacco = 6% x wealth_level
53		if female then percent using tobacco = 0.2% x wealth_level
54
55		source: DHS 2015
56
57		assume time varying with value allocted age 18 (1% x wealth level) but rate of initiating smoking for those assigned as non-smokers and rate of stopping for those assigned as tobacco users to initially be set at zero.
58
59
60		excess alcohol
61
62		baseline in 2018:
63
64		if age > 18 and male 15% drink excess_alcohol
65		if age > 18 and female 1% drink excess_alcohol
66
67		source: WHO 2014 report http://www.who.int/substance_abuse/publications/global_alcohol_report/msb_gsr_2014_2.pdf?ua=1
68
69		no impact of urban / rural or wealth level
70		(if we can contact authors we may be able to see if wealth level has an independent influence)
71
72		Msyamboza et al; 2012; WHO STEPS
73
74		assume time varying with value allocted age 18 (15% for men, 1% for women) but rate of initiating alcohol for those assigned as non-excess-drinkers and rate of stopping for those assigned as excess alcohol drinkers to initially be set at zero.
75
76
77		low exercise
78		(I suggest we create this variable for people aged 18 and over only)
79
80		baseline in 2018:
81
82		if urban and male 32% have low exercise
83		if urban and female 18% have low exercise
84		if rural and male 11% have low exercise
85		if rural and female 7% have low exercise
86
87		Msyamboza et al; 2011; WHO STEPS
88
89		(if we can contact authors we may be able to see if wealth level has an independent influence)
90
91		assume time varying with value allocted age 18 (25% for men, 9% for women) but rate of becoming low exercise for those assigned as having exercise and rate of starting exercise for those assigned as low exercise to initially be set at zero.
92
93
94		BMI
95
96		Informed by Price et al 2018 Prevalence of obesity, hypertension, and diabetes, and cascade of care in sub-Saharan Africa: a cross-sectional, population-based study in rural and urban Malawi Lancet Diabetes Endocrinol
97
98
99		marital status
100		DHS 2015/2016 chapter 4
101
102
103		education
104		DHS 2015/2016 chapter 2 p 14-17