www. helance .com/public-heal h Vol 10 May 2025
e380
A icles
Ciga e e axa ion and socioeconomic inequali ies in
unde -5 mo ali y ac oss 94 low-income and middle-income
coun ies: a longi udinal ecological s udy
Oli ia S Bannon, Jaspe V Been, Sam Ha pe , An hony A La e y, Ch is ophe Mille , F ank J an Len he, Filippos T Filippidis, Má a K Radó
Summa y
Backg ound Al hough inc easing ciga e e axes is known o imp o e child su i al, he e a e ew da a on hei e ec
on socioeconomic inequali ies in child mo ali y. We in es iga ed he associa ion be ween ciga e e axa ion and
socioeconomic inequali ies in mo ali y in child en younge han 5 yea s (he ea e e e ed o as unde -5 mo ali y)
in low-income and middle-income coun ies (LMICs).
Me hods This was a longi udinal ecological s udy. We linked coun y-le el annual da a on 94 LMICs, as de ined by he
Wo ld Bank, and annual da a on unde -5 mo ali y by weal h quin ile om he UN In e -agency G oup o Child
Mo ali y Es ima ion om 2008 o 2020. We used ixed-e ec panel eg ession models o assess he associa ion o
ciga e e axes wi h absolu e and ela i e inequali ies in unde -5 mo ali y by weal h quin ile.
Findings Inc easing o al ciga e e ax by 10-pe cen age-poin s was associa ed wi h educed unde -5 mo ali y a es in
all weal h quin iles. Raising o al ciga e e ax om 0·0–24·9% o 25·0–74·9% and 75·0% o mo e o hei o al e ail
alue was associa ed wi h 3·8% (95% CI 0·2 o 7·3) and 7·6% (1·4 o 13·4) dec eases in absolu e inequali y in
unde -5 mo ali y, espec i ely. This inding was mainly a ibu able o speci ic ax, which was associa ed wi h a 1·4%
(0·3 o 2·6) educ ion in absolu e inequali y o each 10-pe cen age-poin inc ease. We es ima ed ha aising o al
ciga e e axes o 75·0% o mo e in all 94 LMICs could ha e a e ed 281 017 (196 916 o 362 301) unde -5 dea hs in
2021.
In e p e a ion High ciga e e axes a e associa ed wi h a la ge dec ease in absolu e inequali y in child mo ali y in
LMICs. These indings suppo aising ciga e e axes o he WHO- ecommended 75% o mo e o he e ail alue o
p o ec he poo es child en.
Funding Swedish Resea ch Council o Heal h, Wo king Li e, and Wel a e; S i elsen Riksbankens Jubileums ond;
Eu opean Union’s Ho izon 2020 Resea ch and Inno a ion; and UK Na ional Ins i u e o Heal h and Ca e Resea ch.
Copy igh © 2025 The Au ho (s). Published by Else ie L d. This is an Open Access a icle unde he CC BY 4.0
license.
In oduc ion
Second-hand smoke exposu e om obacco smoke in
u e o o du ing childhood is an en i ely p e en able
con ibu o o mo bidi y and p ema u e mo ali y,
causing an es ima ed 5·6 million disabili y-adjus ed li e-
yea s and app oxima ely 200 000 annual dea hs in
child en younge han 5 yea s (he ea e e e ed o as
unde -5 mo ali y) globally.1–3 Smoking p e alence and
child en’s exposu e o second-hand smoke a e o en
highe among people o lowe socioeconomic s a us and
inc ease he isk o ad e se child heal h ou comes,
including mo ali y.4–8 This is ue o ma e nal smoking
du ing p egnancy, as well as an ena al and pos na al
second-hand smoke exposu e (appendix pp 1–2).4–8 Low-
income and middle-income coun ies (LMICs) bea he
g ea es bu den o child mo ali y, wi h an es ima ed
90% o global unde -5 mo ali y occu ing in hese
se ings.9,10 Mo eo e , he majo i y o smoke s eside in
LMICs, u he emphasising he c ucial ole o obacco
con ol policies in achie ing he wo UN Sus ainable
De elopmen Goals: goal 3.2.1, aiming o imp o e
unde -5 mo ali y, and goal 10, aiming o dec ease heal h
inequali ies.11,12
The e is s ong e idence suppo ing he e ec i eness
o obacco con ol policies in imp o ing o e all child
heal h and su i al.7,8,13,14 The MPOWER measu es
(moni o obacco use; p o ec people om second-hand
smoke; o e help o qui obacco use; wa n abou he
dange s o obacco; en o ce bans on obacco ad e ising,
p omo ion, and sponso ship; and aise axes on obacco)
a e key policies ecommended by WHO as pa o he
F amewo k Con en ion on Tobacco Con ol o educe
demand o obacco p oduc s.15 Al hough all hese
policies a e impo an and e ec i e, acco ding o WHO,
aising axes on obacco is he mos e ec i e measu e o
educing obacco use.16,17 Some s udies conduc ed in bo h
LMICs and high-income coun ies ha e shown hei
posi i e e ec on o e all child su i al.13,18–20 Despi e his
compelling e idence and widesp ead a i ica ion o he
F amewo k Con en ion on Tobacco Con ol (including
Lance Public Heal h 2025;
10: e380–90
See Commen page e352
Depa men o Medical
Epidemiology and Bios a is ics,
Ka olinska Ins i u e ,
S ockholm, Sweden
(O S Bannon MMSc,
M K Radó PhD); The Ins i u e o
Analy ical Sociology, Linköping
Uni e si y, No köping, Sweden
(O S Bannon, M K Radó); Di ision
o Neona ology, Depa men o
Neona al and Paedia ic
In ensi e Ca e, E asmus MC
Sophia Child en’s Hospi al,
Uni e si y Medical Cen e
Ro e dam, Ro e dam,
Ne he lands (J V Been PhD);
Depa men o Public Heal h,
E asmus MC, Uni e si y Medical
Cen e Ro e dam, Ro e dam,
Ne he lands (J V Been,
P o F J an Len he PhD);
Depa men o Obs e ics and
Gynaecology, E asmus MC
Sophia Child en’s Hospi al,
Uni e si y Medical Cen e
Ro e dam, Ro e dam,
Ne he lands (J V Been);
Depa men o Epidemiology,
Bios a is ics and Occupa ional
Heal h, McGill Uni e si y,
Mon éal, QC, Canada
(S Ha pe PhD); Public Heal h
Policy E alua ion Uni , School
o Public Heal h, Impe ial
College London, London, UK
(A A La e y PhD,
P o C Mille PhD,
F T Filippidis PhD); NOVA
Na ional School o Public
Heal h, Public Heal h Resea ch
Cen e, Comp ehensi e Heal h
Resea ch Cen e , CHRC, NOVA
Uni e si y Lisbon, Lisbon,
Po ugal (P o C Mille ); Cen e
U banisa ion Cul u e Socié é,
Ins i u Na ional de la
Reche che Scien i ique,
Mon éal, QC, Canada
(M K Radó)
Co espondence o:
D Má a K Radó, Depa men o
Medical Epidemiology and
Bios a is ics, Ka olinska Ins i u e,
SE-171 77 S ockholm, Sweden
[email p o ec ed]
A icles
e381
www. helance .com/public-heal h Vol 10 May 2025
See Online o appendix
182 coun ies, co e ing mo e han 90% o he wo ld’s
popula ion), he majo i y o coun ies ha e no
implemen ed hese policies, pa icula ly in e ms o
le ying su icien ly high axes.1 By 2022, only 41 coun ies
(21% o all 195 coun ies globally)—16 o which a e
LMICs—had aised o al ciga e e axes o he WHO-
ecommended le el o a leas 75% sha e o he e ail
alue o ciga e es.1
The equi y e ec o ciga e e axa ion on child su i al
is unknown.1,21,22 Exis ing esea ch on ciga e e axa ion
and social inequali ies has ocused on socioeconomic
di e ences in he smoking beha iou o adul s.23–29
Al hough mos s udies ound a posi i e equi y e ec on
adul s, some s udies ha e aised conce ns ha he
posi i e heal h e ec s associa ed wi h ciga e e axa ion
migh no ha e e ec i ely eached lowe socioeconomic
g oups in he long un because hey swi ch o
cheape p oduc s (eg, oll-you -own ciga e es) o canno
sus ain qui ing.7,13,25,27–30 Fu he mo e, he majo i y o
hese s udies we e om high-income coun ies and
he e o e migh no be gene alisable o LMICs.23,24,29 Mos
impo an ly, a di ec e alua ion o he e ec o ciga e e
axa ion on social inequali y in child su i al is needed
because s udies ocusing on social inequali ies in adul
smoking beha iou migh no be gene alisable o child
su i al due o complex causal pa hways be ween he
wo ac o s.13,14,18,20 I is c ucial o unde s and he equi y
e ec o obacco con ol in e en ions o p o ec
ulne able child en om he obacco indus y, which
has shi ed i s ocus o expand ma ke s in low-income
popula ions.30,31–33 Because low-income popula ions end
o disp opo iona ely bea he bu den o obacco- ela ed
mo bidi y and mo ali y, disco e ing whe he o no
obacco con ol measu es each he mos ulne able
g oups is c ucial o educe obacco- ela ed heal h
dispa i ies.
Ou s udy aims o b idge his knowledge gap by
es ima ing he associa ion be ween changes in ciga e e
ax le el and s uc u e, and absolu e and ela i e
inequali ies in unde -5 mo ali y wi hin and be ween
coun ies. Wi h unique annual da a on unde -5
mo ali y a es by socioeconomic g oups ac oss a ious
coun ies and mul iple yea s, we pe o med he i s
analysis o he ela ionship be ween aising ciga e e
axes on socioeconomic inequali ies in child su i al in
LMICs.
Resea ch in con ex
E idence be o e his s udy
Exposu e o second-hand smoke in u e o o in ea ly childhood
is a known isk ac o o ad e se child heal h ou comes,
including unde -5 mo ali y, and is a subs an ial con ibu o o
socioeconomic inequali ies in child heal h and su i al. Raising
axes on ciga e es is an es ablished measu e o educe smoking
p e alence and child en’s exposu e o second-hand smoke, and
consequen ly imp o e o e all child heal h and su i al.
Two p e ious sys ema ic e iews ha e highligh ed he e ec o
aising ciga e e axes on inequali ies in adul heal h and smoking
beha iou , showing po en ial o educing hese dispa i ies in he
sho e m and long e m. To explo e whe he simila e ec s
migh exis o child su i al, we conduc ed a li e a u e sea ch on
PubMed on Aug 22, 2024, using he e ms: (“ obacco p ice*”,
“ obacco ax*”, “ciga e e p ice*”, OR “ciga e e ax*”) AND
(“ineq*” o “socio*”) AND (“in an mo ali y” OR “neona al
mo ali y” OR “child mo ali y” OR “unde - i e mo ali y”), wi h
no limi s on da es o language. This sea ch iden i ied i e s udies,
including single-coun y and mul i-coun y assessmen s o links
be ween changes in ciga e e p ices and axes and o e all
childhood su i al, all o which iden i ied obus gains in child
su i al associa ed wi h inc eased ciga e e axes. Howe e , none
o he s udies speci ically examined he associa ion be ween
ciga e e p ices o axes and inequali ies in child su i al.
The e o e, i is unclea whe he he bene i s obse ed in adul
ou comes ex end o educing inequali ies in child su i al.
Added alue o his s udy
In his s udy, we used unde -5 mo ali y da a by socioeconomic
income g oup om he UN In e -agency G oup o Child
Mo ali y Es ima ion (UN IGME) on 94 low-income and
middle-income coun ies (LMICs) om 2008 o 2020 o
examine associa ions o o al ciga e e axes and di e en
ypes o ciga e e axes wi h inequali ies in child su i al. This
is he i s analysis linking changes in ciga e e axes and
s uc u e o inequali ies in child su i al. The use o UN IGME
da a allowed us o go beyond simply examining o e all child
su i al and o inco po a e socioeconomic a ia ions in
unde -5 mo ali y. The da a con ained in o ma ion on LMICs,
whe e obacco indus y in e e ence and child mo ali y end
o be high, and he e is ewe esea ch. Ou s udy iden i ied
s a is ically signi ican educ ions in socioeconomic
inequali ies in child mo ali y associa ed wi h aising ciga e e
axes subs an ially (a leas 75% o hei o al e ail alue), as
well as educ ions in unde -5 mo ali y a es ac oss all weal h
quin iles. This inding was mainly a ibu able o inc eases in
speci ic ax.
Implica ions o all he a ailable e idence
Building on he s udies ha ha e iden i ied associa ions
be ween aising ciga e e axes and imp o ed o e all child
su i al, ou s udy indica es ha aising ciga e e axes has he
po en ial o educe inequali ies in child su i al, especially i
aised subs an ially, as well as imp o e child su i al ac oss all
weal h quin iles. These indings emphasise he impo ance o
mee ing he WHO- ecommended 75% minimum ax le el o he
o al e ail alue h eshold. Raising axes on ciga e es could
con ibu e o he UN’s Sus ainable De elopmen Goal 3.2.1 o
imp o e unde -5 mo ali y, and goal 10 o dec ease heal h
inequali ies.
A icles
www. helance .com/public-heal h Vol 10 May 2025
e382
Me hods
S udy design and da a sou ces
Coun y-le el da a on 132 LMICs, as de ined by he
Wo ld Bank, we e ga he ed be ween 2008 and 2020 o
his longi udinal ecological s udy.34 Da a on unde -5
mo ali y by weal h quin ile we e sou ced om he
UN In e -agency G oup o Child Mo ali y Es ima ion
(UN IGME) and we e a ailable annually om 2008 o
2020 h ough he WHO Heal h Inequali y Da a
Reposi o y o all LMICs.35,36 UN IGME da a we e
compiled om all a ailable sou ces, including censuses,
i al egis a ion da a, and household su eys, and we e
modelled using a Bayesian spline eg ession model.35,36
Biennial da a on MPOWER measu es and ciga e e
axes we e a ailable om 2008 o 2020 a he coun y
le el and we e ex ac ed om he WHO Global Heal h
Obse a o y.3 Da a on all o he co a ia es we e ob ained
om he Wo ld Bank and we e a ailable annually
be ween 2008 and 2020.37 This s udy used anonymised
agg ega ed da a om publicly a ailable esou ces,
he e o e e hical app o al was no equi ed.
Va iables
The ou come a iables in his s udy we e: (1) unde -5
mo ali y—de ined as he es ima ed numbe o child en
who died be o e he age o 5 yea s pe 1000 li ebi hs—by
weal h quin iles one o i e (quin ile one being he poo es
and quin ile i e being he weal hies ); (2) he absolu e
di e ence in unde -5 mo ali y be ween he weal hies
and poo es weal h quin iles wi hin coun ies; and (3) he
ela i e inequali y in unde -5 mo ali y—namely, he
p opo ional di e ence in unde -5 mo ali y be ween
he weal hies and poo es weal h quin iles ( able 1).
Measu ing bo h absolu e and ela i e inequali ies in
unde -5 mo ali y is impo an because each highligh s
di e en aspec s o inequali ies. The ela i e measu e
ocuses on equali y in i sel because i is independen o
he a es in he lowes and highes quin iles, whe eas he
absolu e measu e emphasises he size o he di e ence
in he a es be ween he lowes and highes quin iles.
These measu es can mo e in di e en di ec ions— o
example, absolu e inequali ies migh dec ease while
ela i e inequali ies inc ease i absolu e declines in
Da a sou ce De ini ion Pe iod F equency
Ou comes
Unde -5 mo ali y a e by
economic s a us (weal h
quin ile)
WHO The numbe o child en younge han 5 yea s who die pe 1000
li ebi hs, by weal h quin ile
2008–20 Annual
Absolu e inequali ies in
unde -5 mo ali y
WHO The absolu e di e ence be ween he poo es (quin ile 1) and weal hies
(quin ile 5) weal h quin iles e e s o he di e ence in unde -5 mo ali y
a es be ween hese wo g oups calcula ed o each yea be ween 2008
and 2020, exp essed as dea hs pe 1000 li ebi hs
2008–20 Annual
Rela i e inequali ies in
unde -5 mo ali y
WHO The ela i e di e ence be ween he poo es (quin ile 1) and weal hies
(quin ile 5) weal h quin iles e e s o he p opo ional di e ence in
unde -5 mo ali y a es be ween hese wo g oups calcula ed o each
yea be ween 2008 and 2020, exp essed as dea hs pe 1000 li ebi hs, as
shown in his equa ion: (U5 mo ali y Q1– U5 mo ali y Q5)/(U5
mo ali y Q5)
2008–20 Annual
Exposu e a iables
To al ax WHO To al ciga e e ax as a pe cen age o he e ail alue (pe 10%), including
speci ic ax plus ad alo em plus o he axes
2008–20 Biennial
Speci ic ax WHO Speci ic excise ax (ie, ixed amoun pe ciga e e o pe weigh o each
ciga e e) as a pe cen age o he e ail p ice (pe 10%)
2008–20 Biennial
Ad alo em WHO Ad alo em excise ax (ie, a pe cen age o he ac o y p ice o e ail alue)
as a pe cen age o he e ail alue (pe 10%)
2008–20 Biennial
O he axes (impo du ies,
alue added ax, and o he
axes)
WHO Value added ax o sales (ie, gene al ax on consump ion), impo du ies
(ie, a ax on impo ed goods ha a e des ined o domes ic
consump ion), and o he axes (ie, di e en ly named axes) as
a pe cen age o he e ail alue (pe 10%)
2008–20 Biennial
MPOWER measu es
P o ec people WHO Fi e-poin scale based on WHO e alua ion abou p o ec ing people om
obacco smoke
2008–20 Biennial
O e help o qui WHO Fi e-poin scale based on WHO e alua ion abou o e ing help o qui
obacco use
2008–20 Biennial
Wa ning abou dange s:
heal h wa nings
WHO Fi e-poin scale based on WHO e alua ion abou heal h wa nings abou
obacco
2008–20 Biennial
Wa ning abou dange s: mass
media campaigns
WHO Fi e-poin scale based on WHO e alua ion abou mass media campaigns
agains obacco
2008–20 Biennial
En o ce bans WHO Fi e-poin scale based on WHO e alua ion abou en o cing bans on
obacco ad e ising, p omo ion, and sponso ship
2008–20 Biennial
(Table 1 con inues on nex page)
A icles
e383
www. helance .com/public-heal h Vol 10 May 2025
unde -5 mo ali y a e la ge bu ela i e declines a e
smalle in he mo e disad an aged compa ed wi h he
mos ad an aged g oups.38
Ou main exposu e a iable was o al ciga e e ax as
a pe cen age o he a e age e ail alue o a 20-pack o
ciga e es in each coun y as a con inuous a iable. We
also conside ed o al ciga e e ax as a ca ego ical a iable
(0·0–24·9%; 25·0–74·9%; and ≥75·0% o he e ail alue
[ie, he WHO- ecommended le el o axes]). Finally,
ins ead o o al ax, we used h ee a iables measu ing
he di e en ypes o ax (speci ic ax [ie, ax cha ged
pe quan i y, such as pe pack, ega dless o p ice];
ad alo em [ie, ax cha ged as a pe cen age o he alue o
he p oduc ]; and o he axes, including impo du ies
and alue added ax) as con inuous a iables sepa a ely.
We an sepa a e models o each sepa a e exposu e, as
well as o each ou come a iable (weal h quin iles 1–5,
and absolu e and ela i e inequali y).
The co a ia es included in his s udy we e: MPOWER
obacco con ol measu e sco es ha indica ed he
exis ence o he measu es based on a i e-poin scale
(excluding moni o ing obacco use due o he i ele ance
o he ou come, and aising axes on obacco due o
edundancy, gi en ou exposu e a iables also measu e
axes; appendix p 3), ime (calenda yea as a ca ego ical
a iable), g oss domes ic p oduc pa i y pu chasing
powe pe pe son, coun ies’ e ili y a es (a e age
numbe o child en bo n o a woman each yea ),
pe cen age o he popula ion li ing in u al a eas,
pe cen age o he popula ion wi h access o basic
d inking wa e , heal h expendi u e pe pe son, emale
p ima y educa ion comple ion a e, pe cen age o he
popula ion wi h access o clean uels and echnologies
o cooking, and na ional CO2 emissions. All con ol
a iables we e chosen based on exis ing li e a u e ha
had shown hei e ec on child su i al.13,14,39,40
All ou come a iables, aside om ela i e inequali y in
unde -5 mo ali y, we e ans o med o he log scale
because hey we e no no mally dis ibu ed.
Consequen ly, he esul s om he panel eg ession
models a e p esen ed as exponen ia ed β coe icien s
(excluding ela i e inequali y in unde -5 mo ali y,
because i is no on he log scale) and should be
in e p e ed as he pe cen change in he ou come
associa ed wi h one-uni changes in he exposu e
a iables.
Some a iables we e ans o med o imp o e he
in e p e a ion o he esul s. All ypes o ciga e e ax
( o al ax, speci ic ax, ad alo em, and o he axes),
u al popula ion, access o clean uels and echnologies
o cooking, and access o basic d inking wa e we e
ecoded o e lec a 10-pe cen age-poin change in hei
alues. Addi ionally, g oss domes ic p oduc pa i y
pu chasing powe and heal h expendi u e we e
Da a sou ce De ini ion Pe iod F equency
(Con inued om p e ious page)
Addi ional co a ia es o unde -5 mo ali y
Time No applicable Calende yea 2008–20 Annual
G oss domes ic p oduc Wo ld Bank G oss domes ic p oduc pe pe son, pe 1000 pa i y pu chasing powe
(cu en in e na ional dolla s)
2008–20 Annual
Fe ili y a e Wo ld Bank The a e age numbe o child en bo n o a woman (gi en women su i e
he childbea ing age and e ili y is in line wi h age-speci ic e ili y a es
o he speci ied yea )
2008–20 Annual
Ru al popula ion Wo ld Bank The p opo ion o he popula ion li ing in u al a eas, as de ined by
na ional s a is ical o ices (pe 10% o he popula ion)
2008–20 Annual
D inking wa e Wo ld Bank The p opo ion o he popula ion wi h access o basic d inking wa e
(ie, collec ion ime <30 min; pe 10% o he popula ion)
2008–20 Annual
Heal h expendi u e Wo ld Bank Cu en heal h expendi u e pe pe son, pe 1000 pa i y pu chasing
powe (cu en in e na ional dolla s)
2008–20 Annual
Female p ima y educa ion
comple ion a e
Wo ld Bank The a io o he numbe o new emale en an s in he las g ade o
p ima y educa ion ( ega dless o age) and he numbe o gi ls a he
en ance age o he las g ade o p ima y educa ion (pe 10% o he
popula ion)
2008–20 Annual
Clean cooking Wo ld Bank The p opo ion o he popula ion wi h access o clean uels and
echnologies ( hose ha a ain he ine pa icula e ma e [PM2·5] and
ca bon monoxide [CO] le els ecommended in he 2021 WHO global ai
quali y guidelines) o cooking (pe 10% o he popula ion)
2008–20 Annual
CO2 emissions, kilo on Wo ld Bank Ca bon dioxide emissions a e hose s emming om he bu ning o ossil
uels and he manu ac u e o cemen . They include ca bon dioxide
p oduced du ing consump ion o solid, liquid, and gas uels and gas
la ing
2008–20 Annual
MPOWER measu es a e u he de ailed in he appendix (p 3). Cu en in e na ional dolla s e e s o alues exp essed in in e na ional dolla s o each espec i e yea in ou
da ase .
Table 1: Full de ini ions, da a sou ces, pe iods, and epo ing equencies o all included a iables
A icles
www. helance .com/public-heal h Vol 10 May 2025
e384
ans o med o e lec he pe cen change pe 1000
pu chasing powe pa i y-adjus ed inc ease (in cu en
in e na ional dolla s, which e e s o alues exp essed
in in e na ional dolla s o each espec i e yea in ou
da ase ). The MPOWER measu e W consis s o
wo scales (wa n abou he dange s o obacco: heal h
wa nings and mass media campaigns) ha we e
a e aged o ob ain one singula W sco e. To simpli y
he model, he P, O, W, and E sco es we e summed in o
one single composi e sco e.
Missing da a
Missing obse a ions o yea s whe e da a we e no
epo ed we e impu ed h ough linea in e pola ion
be ween yea s wi h a ailable da a. Coun ies we e
au oma ically excluded om he models by he s a is ical
so wa e i he e we e no obse a ions o all yea s o
any o he a iables (N=34).
Da a analysis
We used ixed-e ec s panel eg ession models o es he
associa ion be ween ciga e e axes and unde -5 mo ali y
by weal h quin ile. Panel eg ession models cap u e
changes o e ime and accoun o he clus e ed da a
(coun y-le el obse a ions) collec ed epea edly o e
ime.13,14,20,41 The ixed-e ec speci ica ion uses dummy
a iables o accoun o ime-in a ian , coun y-le el
po en ial con ounding a iables ha a e no possible o
easible o measu e.42 The choice o he ixed-e ec
speci ica ion o e he less es ic i e andom-e ec
speci ica ion was suppo ed by Hausman es s.42
Non-lagged models we e chosen as he main model,
simila o p e ious s udies on he associa ion be ween
ciga e e axes and child su i al.13,20 Ne e heless, we
es ed an up o 3-yea ime lag o accoun o any delayed
e ec s o cumula i e e ec s o axa ion on unde -5
mo ali y using Akaike in o ma ion c i e ion and
Bayesian in o ma ion c i e ion; howe e , he lags did no
imp o e he i o he model (appendix p 4).
We es ima ed he unde -5 mo ali y ha could ha e
been a oided in 2021 i each coun y included in he
analysis would aise axes on ciga e es o he minimum
WHO ecommenda ion o 75% o hei e ail alue
(appendix p 5). Fo each coun y ha did no ye each
he ecommended le el pe quin ile, we applied quin ile-
speci ic e ec es ima es (ie, exp essing changes in he
ou come by a one-pe cen age-poin inc ease in axa ion
as a pe cen age o he e ail alue o ciga e es) om ou
ixed e ec model de ining axes as a con inuous
measu e o he di e ence be ween each coun y’s
ciga e e ax and he 75% h eshold, and o he coun y’s
es ima ed numbe o unde -5 dea hs disagg ega ed by
weal h quin ile. Each coun y’s es ima ed dea hs a e ed
we e hen summed up pe quin ile.
We pe o med se e al sensi i i y analyses wi h o al ax
as he exposu e a iable whe e we es ed: (1) a 1-yea ime
lag o check o any po en ial delayed policy e ec a e
implemen a ion; (2) excluding 2020 da a o exclude any
po en ial e ec om he COVID-19 pandemic; (3) no
adjus men s o co a ia es o see how hey a ec ed he
esul s; (4) no con olling o ime; (5) no missing
obse a ion impu a ion o see whe he linea in e pola ion
changed he esul s; (6) including he P, O, W, and E
measu es as indi idual sco es o ensu e he composi e
sco e had he same e ec ; (7) including he MPOWER
measu e M (moni o obacco use) o see i i s exclusion
a ec ed he esul s; (8) including o al ax ca ego ised in o
ou equal pe cen age g oups as opposed o h ee; (9) no
adjus men o co a ia es wi h o al ax as a ca ego ical
a iable; and (10) no con olling o ime wi h o al ax as
a ca ego ical a iable. All analyses we e pe o med wi h
S a a ( e sion 17.0).
Role o he unding sou ce
The unde s o he s udy had no ole in s udy design,
da a collec ion, da a analysis, da a in e p e a ion, o
w i ing o he epo .
Resul s
O he 132 LMICs we ga he ed da a om, da a om
94 coun ies we e included in he panel eg ession
Figu e 1: Median unde -5 mo ali y a es (pe 1000 li ebi hs) by weal h quin ile om 2008 o 2020 in
94 low-income and middle-income coun ies
Quin ile 1 (poo es )
Quin ile 2
Quin ile 3
Quin ile 4
Quin ile 5 (weal hies )
2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020
0
10
20
30
40
50
60
70
80
Median unde -5 mo ali y
Yea
To al ax (IQR) Speci ic ax (IQR) Ad alo em (IQR) O he axes including
impo du ies, alue-
added ax, and o he
axes (IQR)
2008 39·0% (25·0–52·0) 0·0% (0·0–25·5) 3·0% (0·0–18·0) 13·0% (9·5–16·0)
2010 39·0% (25·0–55·0) 5·0% (0·0–30·0) 2·0% (0·0–17·0) 14·0% (10·0–16·0)
2012 36·0% (28·0–52·0) 8·0% (0·0–24·0) 2·0% (0·0–17·0) 14·0% (9·0–16·0)
2014 38·0% (28·0–54·0) 12·0% (0·0–29·0) 0·0% (0·0–16·0) 14·0% (11·0–16·0)
2016 39·0% (28·0–54·0) 14·0% (0·0–33·0) 2·0% (0·0–15·0) 14·0% (10·0–16·0)
2018 43·0% (31·5–54·5) 16·5% (0·0–35·5) 2·5% (0·0–15·0) 14·0% (9·5–17·0)
2020 44·0% (34·0–58·0) 18·0% (0·0–36·0) 3·0% (0·0–16·0) 14·0% (11·0–17·0)
Table 2: Median o al ciga e e axes (as a pe cen age o he e ail alue) o each ype o ax biennially
om 2008 o 2020 in 94 low-income and middle-income coun ies
A icles
e385
www. helance .com/public-heal h Vol 10 May 2025
analysis om 2008 o 2020 (appendix p 6). Median
unde -5 mo ali y a es dec eased be ween 2008 and 2020
ac oss all weal h quin iles ( igu e 1; appendix p 7).
Median unde -5 mo ali y a es we e highes in
he poo es quin ile (quin ile 1), anging be ween
71·5 (IQR 37·6–116·2) dea hs pe 1000 li ebi hs in 2008
and 47·6 (25·3–75·8) dea hs pe 1000 li ebi hs in 2020.
This was subs an ially highe han in he weal hies
quin ile, whe e median unde -5 mo ali y a es anged
be ween 36·9 (17·6–72·0) dea hs pe 1000 li ebi hs in
2008 and 24·0 (12·0–42·3) dea hs pe 1000 li ebi hs in
2020.
Median ax pe cen ages gene ally inc eased o e he
s udy pe iod, wi h some a ia ion depending on he ype
o ax ( able 2). Median o al ax ose om 39·0%
(IQR 25·0–52·0) in 2008 o 44·0% (34·0–58·0) in 2020.
Median speci ic ax also inc eased du ing he s udy pe iod,
om 0·0% (0·0–25·5) in 2008 o 18·0% (0·0–36·0) in
2020. Median ad alo em ax was consis en , wi h 3·0%
(0·0–18·0) in 2008 and 3·0% (0·0–16·0) in 2020, al hough
he e was mino a ia ion be ween yea s. Median o he
axes we e also ai ly s able, inc easing om 13·0%
(9·5–16·0) in 2008 o 14·0% (11·0–17·0) in 2020. O all he
o al ax obse a ions, 16·5% we e equal o o mo e han
he WHO- ecommended h eshold o 75% o he o al
e ail alue.
When examining he associa ion be ween changes in
ciga e e axes and unde -5 mo ali y, we ound
signi ican bene i s o o e all child su i al in all models
and weal h quin iles and a small, al hough s a is ically
insigni ican , dec ease in absolu e inequali ies in
unde -5 mo ali y (0·7%; 95% CI –0·3 o 1·8; able 3). In
he model whe e we in e p e ed o al ax as a ca ego ical
a iable, we ound ha inc easing o al ax om he
0·0–24·9% ca ego y o he 25·0–74·9% ca ego y was
associa ed wi h consis en dec eases in unde -5 mo ali y
in all quin iles, wi h he s onges e ec in he second
poo es quin ile (2·8%; 0·6 o 5·0) and he weakes
e ec in he weal hies (2·6%; 0·4 o 4·8). Inc easing
o al ax om he 0·0–24·9% ca ego y o he 75·0% o
highe ca ego y was associa ed wi h obus dec eases in
unde -5 mo ali y, spanning om a 6·3% (2·5 o 9·9)
dec ease in he second poo es quin ile o a 5·8%
(2·1 o 9·4) dec ease in he weal hies . S a is ically
signi ican educ ions in absolu e inequali y we e ound
when aising he o al ax o bo h he 25·0–74·9% and
he 75% o mo e ca ego ies, wi h a 3·8% (0·2 o 7·3)
dec ease in he 25·0–74·9% ca ego y and a 7·6%
(1·4 o 13·4) dec ease in he 75% o mo e ca ego y.
Raising ciga e e axes was no associa ed wi h
a s a is ically signi ican dec ease in ela i e inequali y
o any o he ax ca ego ies examined.
A 10-pe cen age-poin inc ease in o al ciga e e ax was
associa ed wi h nea ly iden ical educ ions in unde -5
mo ali y ac oss all weal h quin iles, wi h a dec ease o
2·0% (95% CI 1·4–2·6) in he poo es quin ile, and
a 2·1% (quin ile 2: 1·5–2·7; quin ile 3: 1·4–2·7; quin ile 4
and 5: 1·5–2·7) dec ease in all o he quin iles. We did no
ind e idence o s a is ically signi ican declines in
absolu e o ela i e inequali ies in his model.
In he model wi h sepa a e ax elemen s, a 10-pe cen age-
poin inc ease in speci ic ax was associa ed wi h a 1·4%
(95% CI 0·3–2·6) dec ease in absolu e inequali y in
unde -5 mo ali y as well as consis en declines in unde -5
Unde -5 mo ali y Inequali y in unde -5 mo ali y
Quin ile 1 (poo es ) Quin ile 2 Quin ile 3 Quin ile 4 Quin ile 5 (weal hies ) Absolu e inequali y Rela i e inequali y
10-pe cen age-poin
inc ease in o al ax
(% o e ail alue)
2·0 (1·4 o 2·6) 2·1 (1·5 o 2·7) 2·1 (1·4 o 2·7) 2·1 (1·5 o 2·7) 2·1 (1·5 o 2·7) 0·7 (–0·3 o 1·8) 0·0 (–0·2 o 0·2)
N 826 826 826 826 826 826 826
To al ax (% o e ail alue), ca ego ical
0·0–24·9% Re e ence Re e ence Re e ence Re e ence Re e ence Re e ence Re e ence
25·0–74·9% 2·7 (0·6 o 4·8) 2·8 (0·6 o 5·0) 2·7 (0·5 o 4·9) 2·7 (0·5 o 4·8) 2·6 (0·4 o 4·8) 3·8 (0·2 o 7·3) 0·0 (–0·8 o 1·4)
≥75·0 6·0 (2·4 o 9·5) 6·3 (2·5 o 9·9) 6·1 (2·3 o 9·7) 6·1 (2·3 o 9·7) 5·8 (2·1 o 9·4) 7·6 (1·4 o 13·4) 0·1 (–1·3 o 0·2)
N 826 826 826 826 826 826 826
10-pe cen age-poin inc ease in sepa a e ypes o axes
Speci ic ax 2·6 (1·9 o 3·2) 2·7 (2·0 o 3·4) 2·6 (1·9 o 3·3) 2·6 (1·9 o 3·3) 2·6 (2·0 o 3·3) 1·4 (0·3 o 2·6) 0·1 (–0·2 o 0·3)
Ad alo em 1·2 (0·4 o 1·9) 1·1 (0·3 o 1·9) 1·2 (0·4 o 2·0) 1·3 (0·5 o 2·1) 1·3 (0·5 o 2·1) –0·5 (–1·9 o 0·9) –0·2 (–0·4 o 0·1)
O he ax 3·0 (2·0 o 3·9) 3·0 (2·0 o 4·0) 3·1 (2·1 o 4·1) 3·2 (2·2 o 4·2) 3·1 (2·1 o 4·1) 1·4 (–0·4 o 3·1) –0·1 (–0·5 o 0·2)
N 826 826 826 826 826 826 826
Da a a e ou come a iables (dec ease in pe cen age poin and 95% CIs). Da a a e dec ease in pe cen age poin and 95% CIs. Nega i e alues indica e an inc ease in pe cen age poin . This able con ains only he
coe icien s o he main exposu e a iables (ciga e e axes). All models adjus o POWE obacco con ol measu es, ime, g oss domes ic p oduc pa i y pu chasing powe pe pe son, pe cen age o he
popula ion li ing in u al a eas, coun ies’ e ili y a es, pe cen age o he popula ion wi h access o basic d inking wa e , heal h expendi u e pe pe son, emale p ima y educa ion comple ion a e, pe cen age o
he popula ion wi h access o clean cooking, and CO2 emissions. POWE=WHO MPOWER measu es (p o ec people om second-hand smoke, o e help o qui obacco use, wa n abou he dange s o obacco,
and en o ce bans on obacco ad e ising, p omo ion, and sponso ship).
Table 3: The associa ion be ween o al ciga e e ax (% o e ail alue) and dec eases in unde -5 mo ali y in di e en weal h quin iles, and inequali ies in unde -5 mo ali y om 2008 o
2020 in 94 low-income and middle-income coun ies in he main models
A icles
www. helance .com/public-heal h Vol 10 May 2025
e386
mo ali y ac oss all weal h quin iles, wi h a sligh ly la ge
dec ease in he second poo es quin ile (2·7%; 2·0–3·4).
Raising ad alo em ax by 10-pe cen age-poin s was
associa ed wi h declines in unde -5 mo ali y ha anged
om 1·3% (0·5–2·1) in he weal hies and second
weal hies quin iles o 1·1% (0·3–1·9) in he second
poo es quin ile. A 10-pe cen age-poin inc ease in o he
axes was associa ed wi h he mos obus dec eases in
unde -5 mo ali y o all he sepa a e ax elemen s,
spanning om a 3·2% (2·2–4·2) educ ion in he second
weal hies quin ile o a 3·0% (2·0–3·9) educ ion in he
poo es quin ile. We did no ind e idence o changes in
ela i e inequali y o sepa a e ax elemen s.
O he 94 coun ies in ou analysis, 84 had no ye aised
ciga e e axes o he WHO- ecommended minimum o
75% o hei e ail alue by he end o he s udy pe iod.
Based on ou models, aising ciga e e axes o his le el
ac oss hese 84 coun ies could ha e a e ed an es ima ed
69 596 (95% CI 48 717–90 474) unde -5 dea hs in he poo es
weal h quin ile e sus 36 156 (25 826–46 486) in he
weal hies quin ile in 2021 ( igu e 2). Cumula i ely ac oss
all weal h quin iles, an es ima ed 281 017 (196 916–362 301)
unde -5 dea hs migh ha e been a oided o he
4 721 418 epo ed unde -5 dea hs in he 94 coun ies in
2021.
Ou sensi i i y analyses p esen ed esul s la gely in
line wi h ou main indings (appendix pp 8–10).
Associa ions we e s a is ically signi ican ac oss mos
models and di ec ionally simila o he main models in
ha unde -5 mo ali y was associa ed wi h aising
ciga e e axes by 10-pe cen age-poin s in all weal h
quin iles, al hough a s a is ically signi ican equi y e ec
was only de ec ed in he model wi hou co a ia es and
he model whe e we did no con ol o ime.
Discussion
In ou panel analysis o 94 LMICs om 2008 o 2020,
a ma ginal 10-pe cen age-poin ciga e e ax inc ease was
associa ed wi h s a is ically signi ican educ ions in
unde -5 mo ali y ac oss all weal h quin iles. S a is ically
signi ican educ ions in absolu e inequali y we e ound
o ca ego ically la ge ax inc eases, pa icula ly when
aised o a leas 75% o hei o al e ail alue. In absolu e
e ms, aising ciga e e axes o a leas 75% o hei e ail
alue in he 84 coun ies in ou analysis ha did no ye
each his le el could ha e a oided an es ima ed 281 017
(95% CI 196 916–362 301) unde -5 dea hs in 2021, wi h
g ea e bene i s acc uing o he poo es quin ile
compa ed wi h he weal hies one.
To ou knowledge, his is he i s examina ion o he
ela ionship be ween ciga e e axa ion and unde -5
mo ali y by socioeconomic g oup ac oss mul iple
coun ies o e ime, going beyond he p e iously
obse ed o e all coun y-le el e ec .7,13,18,35 Ou s udy
con i med he indings o p e ious s udies ha ciga e e
axes a e bene icial o o e all child su i al,13,18–20 and
ad ances his wo k by showing ha his o e all posi i e
e ec can be ound o all socioeconomic g oups and has
he po en ial o dec ease inequali ies in unde -5 mo ali y.
The main s eng h o ou s udy is he use o no el UN
IGME da a on child su i al disagg ega ed by socio-
economic g oup, which a e compa able ac oss coun ies
and o e ime.43
Se e al mechanisms migh explain why ciga e e axes
a ec child en’s mo ali y isk di e en ly ac oss socio-
economic g oups. Fi s , ciga e e axes can educe
smoking p e alence une enly, wi h lowe socioeconomic
g oups being mo e likely o qui smoking due o highe
p ice elas ici y (ie, hey a e mo e sensi i e o p ice
inc eases).44 Howe e , small ax inc eases migh no ha e
a meaning ul e ec on hese g oups because he obacco
indus y can abso b some cos s, keeping ciga e es
a o dable.31 A subs an ial ax hike migh mo e e ec i ely
educe smoking in lowe socioeconomic popula ions,
helping o close he gap in smoking p e alence be ween
poo e and weal hie indi iduals.33,45 This educ ion
would likely dec ease inequali ies in neona al exposu e
o second-hand smoke and ma e nal smoking, po en ially
lowe ing a es o in an mo ali y ia imp o ing ou comes
such as s illbi h, sudden in an dea h synd ome, and
complica ions such as p ema u e bi h o low bi h-
weigh ,4–8 o educing child mo ali y om as hma o
espi a o y illnesses.46–48 Some p e ious esea ch has
ques ioned whe he highe ciga e e axes can lead o
long- e m imp o emen s in socioeconomic inequali ies
in smoking p e alence because poo e indi iduals o en
ha e highe a es o elapse due o g ea e nico ine
dependence, lowe sel -e icacy, and educed qui ing
success compa ed wi h hei weal hie coun e pa s.29,44,49
E en i smoking a es decline simila ly ac oss all
socioeconomic g oups, child en om lowe socio-
economic s a us backg ounds migh s ill bene i mo e
because hey end o ha e highe exposu e o obacco
smoke in public o wo kplace en i onmen s, bo h in
u e o and a e bi h.4,50,51 Finally, al hough highe
ciga e e axes o p ices can educe inequali ies in
Figu e 2: The es ima ed absolu e numbe o unde -5 dea hs (and 95% CIs) a oided by aising ciga e e axes
o 75% o hei e ail alue in 84 low-income and middle-income coun ies in 2021
Quin ile 1
(poo es )
Quin ile 2 Quin ile 3 Quin ile 4 Quin ile 5
(weal hies )
Numbe o unde -5 dea hs a oided
69
596
(48
717–90
474) 66
697
(47
641–85
753) 59
127
(39
418–76
020)
49
442
(35
315–63
568)
36
156
(25
826–46
486)
A icles
e387
www. helance .com/public-heal h Vol 10 May 2025
smoking p e alence, hey migh cause inancial ha dship
among low socioeconomic s a us households ha
con inue o smoke, which migh nega i ely a ec hei
abili y o a o d heal h-essen ial goods such as ood o
heal h ca e.52–54 Some o hese mechanisms migh bene i
he poo es people in socie y mo e, whe eas o he
mechanisms do no , esul ing in socioeconomic
di e ences in child en’s mo ali y isk. Al hough his is
ou side he scope o he cu en s udy, u u e esea ch is
wa an ed o explo e he link be ween ciga e e axes and
he po en ial o unin ended consequences, such as ood
insecu i y, ac oss income g oups.
This s udy also ex ends he li e a u e on he e ec o
obacco con ol policies on socioeconomic inequali ies in
smoking beha iou o he e ec o hese policies on
inequali ies in child su i al.23–26,29,55,56 One sys ema ic
e iew ound ha he e ec o aising ciga e e p ices on
educing smoking p e alence was g ea e among lowe
socioeconomic g oups in he UK, USA, Canada, and
Taiwan.25 In con as , ano he s udy conduc ed in he UK
ound no socioeconomic di e ences in qui ing, despi e
g ea e p ice sensi i i y among lowe socioeconomic
g oups, p obably due o swi ching o cheape p oduc s o
an inabili y o sus ain qui a emp s.24 Mo eo e , in
Aus alia, he e ec o aising ciga e e axes on smoking
p e alence had a la ge immedia e e ec in lowe
socioeconomic g oups compa ed wi h highe
socioeconomic g oups, bu his posi i e e ec was no
sus ained in he long un.52 Ou indings mi o ed his
complexi y; we de ec ed a s a is ically insigni ican small
dec ease in absolu e inequali ies in unde -5 mo ali y
ollowed by a ma ginal inc ease in o e all axes, and
a la ge s a is ically signi ican dec ease in absolu e
inequali ies when we examined highe le els o o al ax
wi h o al ax as a ca ego ical a iable. Ou indings
he e o e sugges ha ciga e e axes should be aised
a subs an ial amoun o ob ain a s ong equi y e ec on
child su i al, which addi ionally emphasises he
impo ance o aising ciga e e axes o he WHO-
ecommended 75% minimum o he o al e ail alue
h eshold.
Al hough la ge ax inc eases we e associa ed wi h
s a is ically signi ican declines in absolu e inequali ies,
associa ions wi h educed ela i e inequali ies we e no
de ec ed. This sugges s ha , al hough axes we e
associa ed wi h educed mo ali y a es ac oss all
socioeconomic s a us g oups—wi h he g ea es absolu e
educ ions in he lowe socioeconomic s a us g oups—
he ela i e educ ion in unde -5 mo ali y was no
g ea e in lowe han in highe socioeconomic s a us
g oups. Howe e , in LMICs, whe e o e all unde -5
mo ali y is apidly declining, i is a e o ind a policy
ha also educes ela i e inequali ies in mo ali y.38
When examining he e ec o ypes o axes indi idually,
we ound ha aising speci ic axes was s a is ically
signi ican ly associa ed wi h educed inequali ies in
unde -5 mo ali y. Speci ic axes a e speci ic o ciga e es,
and aising hem is he mos ecommended s a egy by
WHO o educe smoking.16,57 Raising speci ic axes is
mo e s aigh o wa d han ad alo em and o he axes
because i does no equi e changes o en i e ax sys ems,
and is consequen ly less complica ed o implemen .16,57
Mo eo e , i educes he obacco indus y’s abili y o
manipula e p ices o main ain budge ciga e es on he
ma ke . Ou indings add o exis ing e idence ha aising
speci ic axes is mos bene icial o popula ion heal h,
indica ing he po en ial o an equi y e ec on unde -5
mo ali y. Al hough ou analyses indica e ha each ype
o ciga e e ax was associa ed wi h la ge bene i s in
o e all child su i al, only speci ic axes we e associa ed
wi h a dec ease in absolu e inequali ies. This is pa ly
explained by o he axes (including impo du ies, alue
added ax, and o he axes) gene ally a ec ing o he ypes
o heal h beha iou s as well, and as such should no be
neglec ed as a po en ial measu e o imp o e child
su i al.13
Ou s udy has some limi a ions. Despi e he use o
ixed-e ec models ha accoun o unobse ed ime-
in a iable ac o s, and he inclusion o addi ional
ime- a ying co a ia es, we we e unable o con ol o
some ac o s, such as he olun a y adop ion o smoke-
ee homes, o o how co a ia es migh a y be ween
di e en socioeconomic s a us s a a, he e o e, some
esidual con ounding canno be uled ou . Unde -5
mo ali y as an ou come measu e is no pe ec , because
mo e ac o s—such as acciden s o inju ies—could occu
in he i s 5 yea s o li e, whe eas a measu e such as
in an mo ali y ( he numbe o child en who die be o e
he age o 1 yea ) is less suscep ible o his limi a ion.
Ne e heless, unde -5 mo ali y p ima ily consis s o
in an mo ali y (app oxima ely 75% o unde -5 mo ali y
is in an mo ali y in LMICs), hus, he igu es a ailable
o unde -5 mo ali y could con iden ly be conside ed as
a p oxy o in an mo ali y igu es, o which da a
disagg ega ed by socioeconomic s a us a e no a ailable.37
To add ess he limi a ion o po en ial bias om he
possible a iable lag be ween exposu e and ou come
when using unde -5 mo ali y as an ou come measu e,
we es ed a 1-yea ime lag in ou sensi i i y analyses,
which yielded di ec ionally simila esul s as he main
model. Da a on unde -5 mo ali y a es we e necessa ily
modelled by he UN IGME and include unce ain y,
which is no cap u ed in ou es ima es. Ne e heless, his
is he sole sou ce o unde -5 mo ali y panel da a ac oss
mul iple coun ies disagg ega ed by socioeconomic
g oup and conside ed o be he mos eliable sou ce in
he absence o eal-wo ld obse a ion, wi h simila
modelled da a o en used in ela ed impac analyses.14,58–61
We we e no able o include da a on axes on o he o ms
o obacco, such as oll-you -own ciga e es o smokeless
obacco. As such, he o e all e ec o obacco axa ion is
p obably unde es ima ed in ou s udy, due o i ocusing
only on ciga e e ax. Ou s udy p ima ily used da a
agg ega ed a he coun y le el, wi h he ou come
A icles
www. helance .com/public-heal h Vol 10 May 2025
e388
a iables a ailable disagg ega ed by income le el only.
Disagg ega ed da a by sex o gende and ace o e hnici y
we e no a ailable, hus subg oup analysis based on
hese ac o s was no possible. Da a on MPOWER
measu es—included in ou models as co a ia es—
desc ibe he exis ence o obacco con ol policies bu do
no con ey hei deg ee o implemen a ion, which likely
a ies, and his should be conside ed as a limi a ion.
Imp ecision migh ha e been in oduced om missing
da a impu a ion on he exposu e a iables, al hough his
was likely minimal because ou sensi i i y analysis
wi hou i indica ed di ec ionally simila indings. When
in e p e ing he es ima ions o po en ial unde -5 dea hs
a oided in 2021, i should be conside ed ha he g adien
shown in pa e lec s he la ge numbe o absolu e
unde -5 dea hs ha occu in poo e socioeconomic
g oups in gene al. Ou es ima es e lec many di e se
coun ies and canno be applied o indi idual coun ies
o egions wi hou cau ion. Fu u e s udies should explo e
egional a ia ion in obacco con ol policy imple-
men a ion and he po en ial acili a o s and ba ie s o
hei e ec i e implemen a ion.
Fu u e esea ch should also conside how he o he
MPOWER measu es a ec socioeconomic inequali ies in
unde -5 mo ali y because hey ha e also been ecognised
o ha e a posi i e e ec on child heal h and su i al,
al hough hei e ec on socioeconomic inequali ies on
child heal h is unclea .8,14,27,50,56 Conside ing he obacco
indus y’s in e e ence and endency o a ge ulne able
g oups,30,31,32 i is pa icula ly p uden o gain a clea e
unde s anding o he capaci y o educe mo ali y among
he mos ulne able child en h ough an a ay o obacco
con ol measu es.
Al hough child mo ali y a es a e declining globally,
LMICs con inue o ha e an unaccep ably high
bu den o child mo ali y, pa icula ly among mo e socio-
economically disad an aged g oups.35 Ou indings
sugges ha aising ciga e e axes migh aid in mee ing
he aims o Sus ainable De elopmen Goals 3.2.1 and
10 by imp o ing child su i al and educing inequali y in
child mo ali y by socioeconomic g oup. In addi ion, axes
gene a e go e nmen e enue, which means mo e can be
spen on heal h p og ammes and heal h-ca e se ices.57
These in es men s can u he educe smoking p e alence
and enhance child su i al a es. Raising ciga e e axes is
he mos a o dable and e ec i e obacco con ol measu e
and has been shown o be e ec i e in LMICs, ye he la es
da a om WHO show ha i is he leas -implemen ed
MPOWER measu e.13,62 Ou s udy adds o he exis ing
body o e idence ad oca ing o mo e coun ies o
implemen he WHO- ecommended 75% minimum
h eshold and p o ides compelling new e idence ha
doing so migh po en ially educe wi hin-coun y
inequali y in child mo ali y in LMICs. This is
un o una ely no a simple ask because in e e ence om
he obacco indus y in LMICs is a majo impedimen
o implemen ing ciga e e ax policy. Sp eading
mis in o ma ion, lobbying, exagge a ing hei impo ance
o local economies, h ea ening li iga ion, and using p ice-
educing p omo ions o o se ax inc eases a e jus a ew
o he ac ics used by he obacco indus y o p e en
obacco con ol p og ess.30,31,32 I is, he e o e, c ucial o
global heal h o ganisa ions o ad oca e and suppo
coun ies in hei obacco con ol e o s.
In ou panel analysis o 94 LMICs om 2008 o 2020,
aising o al axes on ciga e es by 10-pe cen age-poin s
was associa ed wi h s a is ically signi ican o e all
dec eases in unde -5 mo ali y ac oss all weal h quin iles.
Ca ego ically la ge inc eases in o al ax we e associa ed
wi h s a is ically signi ican dec eases in absolu e
inequali ies in unde -5 mo ali y, wi h he g ea es e ec
seen when aised o he WHO- ecommended le el o
≥75%. These indings indica e he po en ial child su i al
gains and educed inequali y esul ing om inc easing
ciga e e axes, and unde sco e he impo ance o eaching
he WHO- ecommended minimum le el o ciga e e
axa ion, equa ing o a leas 75% o hei o al e ail alue.
All child en, ega dless o hei socio economic s a us,
should be u gen ly p o ec ed om obacco smoke.
Con ibu o s
OSB: o mal analysis (lead), da a cu a ion (lead), w i ing o he o iginal
d a (lead), w i ing, me hodology, e iew, and edi ing o he manusc ip .
MKR: concep ualisa ion (lead), o mal analysis, w i ing o he o iginal
d a , e iew, and edi ing o he manusc ip . JVB: concep ualisa ion,
o mal analysis, w i ing, me hodology, e iew, and edi ing o he
manusc ip . SH, AAL, and FTF: o mal analysis, and w i ing, e iew,
and edi ing o he manusc ip . CM and FJ L: w i ing, e iew, and edi ing
o he manusc ip . All au ho s had access o he da a used in his s udy.
OSB and MKR di ec ly accessed and e i ied he unde lying da a.
MKR was esponsible o he decision o submi he manusc ip o
publica ion.
Decla a ion o in e es s
We decla e no compe ing in e es s.
Da a sha ing
This s udy used publicly a ailable da a ha can be ound online a
h ps://da a.unice .o g/ opic/child-su i al/unde - i e-mo ali y/#da a,
h ps://da a.wo ldbank.o g/, and h ps://www.who.in /da a/gho.
Acknowledgmen s
MKR is suppo ed by he Swedish Resea ch Council o Heal h, Wo king
Li e, and Wel a e (g an numbe 2023–00678) and S i elsen Riksbankens
Jubileums ond (g an numbe P23–0640). OSB was suppo ed by he
Eu opean Union’s Ho izon 2020 Resea ch and Inno a ion p og amme
unde he Ma ie Skłodowska-Cu ie g an (ag eemen numbe 101008139).
CM is suppo ed by he UK Na ional Ins i u e o Heal h and Ca e
Resea ch (g an numbe NIHR133252) using UK aid om he UK
Go e nmen o suppo global heal h esea ch.
Re e ences
1 WHO. WHO epo on he global obacco epidemic, 2023: p o ec
people om obacco smoke. July 31, 2023. h ps://www.who.in /
publica ions-de ail- edi ec /9789240077164 (accessed Sep 11, 2023).
2 IHME. Global Bu den o Disease S udy 2019 (GBD 2019) da a
esou ces. h ps://ghdx.heal hda a.o g/gbd-2019 (accessed
No 7, 2023).
3 WHO. The Global Heal h Obse a o y. h ps://www.who.in /da a/
gho (accessed Feb 8, 2023).
4 Naza GP, Lee JT, A o a M, Mille C. Socioeconomic inequali ies in
secondhand smoke exposu e a home and a wo k in 15 low- and
middle-income coun ies. Nico ine Tob Res 2016; 18: 1230–39.
5 Hiscock R, Bauld L, Amos A, Fidle JA, Muna ò M. Socioeconomic
s a us and smoking: a e iew. Ann N Y Acad Sci 2012; 1248: 107–23.
