RES E AR C H Open Access
Impoverishing effects of catastrophic health
expenditures in Malawi
Martina Mchenga
1*
, Gowokani Chijere Chirwa
2
and Levison S. Chiwaula
2
Abstract
Background: Out of pocket (OOP) health spending can potentially expose households to risk of incurring large
medical bills, and this may impact on their welfare. This work investigates the effect of catastrophic OOP on the
incidence and depth of poverty in Malawi.
Methods: The paper is based on data that was collected from 12,271 households that were interviewed during the
third Malawi integrated household survey (IHS-3). The paper considered a household to have incurred a catastrophic
health expenditure if the share of health expenditure in the household’snon-foodexpenditurewasgreaterthana
given threshold ranging between 10 and 40%.
Results: As we increase the threshold from 10 to 40%, we found that OOP drives between 9.37 and 0.73% of households
into catastrophic health expenditure. The extent by which households exceed a given threshold (mean overshoot) drops
from 1.01% of expenditure to 0.08%, as the threshold increased. When OOP is accounted for in poverty estimation,
additional 0.93% of the population is considered poor and the poverty gap rises by almost 2.54%. Our analysis suggests
that people in rural areas and middle income households are at higher risk of facing catastrophic health expenditure.
Conclusion: We conclude that catastrophic health expenditure increases the incidence and depth of poverty in Malawi.
This calls for the introduction of social insurance system to minimize the incidence of catastrophic health expenditure
especially to the rural and middle income population.
Keywords: Catastrophic health expenditure, Out of pocket expenditure (OOP), Total health expenditure (THE),
Impoverishment, Health financing, Malawi
Background
Health is considered to be a basic human right, as
enshrined in the 1948, constitution of the world health
organization (WHO) [1]. One of the biggest challenges
to this right is cost of accessing care which may result in
some financial risk by those seeking health care. The
financial risks are in the form of out-of-pocket expend-
iture (OOP), total health expenditure (THE) as well as
catastrophic health expenditure (CHE) [2–5]. CHE
occurs when OOP payments for health services consume
large portion of a household’s income [6]. It is estimated
that 150 million people face some financial catastrophe
due to OOP and that 100 million are pushed into
poverty every year, as a result of CHE [7].
1
CHE are
heavily influenced by the meth od of financing of health
care by the health sector.
Health sector financing in Malawi is composed of
government financing, donor financing and private
financing. Government financing is through subvention
that is directed to public providers and other providers.
Donor financing is through donor support to government’s
development budget, commodity aid and direct support to
programs and other providers. Private financing is
comprised of household out of pocket expenditure, firms
and insurance providers of the three [8]. While government
contributions to total health expenditure have been falling,
for example from 22% in 2004 to 18% in 2 011, and
to 13% in 2012 , donor contributions have been rising
from 46 to 66% of total health expenditure between
2002/03 and 2008/09 [9]. Despite the rising donor contri-
bution, Malawi’s health system still faces absolute and
relative inadequacy of financing to adequately fund its free
* Correspondence: martinamchenga@gmail.com
1
Stellenbosch University, Stellenbosch, South Africa
Full list of author information is available at the end of the article
© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Mchenga et al. International Journal for Equity in Health (2017) 16:25
DOI 10.1186/s12939-017-0515-0
primary health care services [10]. Consequently, about
27% of total health expenditure comes from the private
sector, and from this, 53.4% is total OOP spending. This
means that it is important to study OOP because it re-
mains one of the commonest form of financing health not
only in Malawi, but also the wider African continent as
well [11, 12].
Malawi has no social medical insurance and an almost
nonexistent private medical insurance that plays a
marginal role as a source of health care finance. For in-
stance, private health insurance managed an average of
3% of total health spending between 2007 and 2009 [13].
This is unlikely to significantly change, because Malawi
has a small formal se ctor from which health insurance
premiums could be collected with relative ease. Besides,
the inf ormal sector is characterized by low wages and
salaries. Most households are therefore dependent on
uninsured medical expenditures to finance health care.
Financing health care through this method can pose a
major threat to living standards particularly in low and
middle income countries (LIMIC) with little formal
health insurance coverage [14–18].
Reliance on OOP in financing health care leaves
households exp osed to risk of incurring large medical
expenses should a household member fall sick. As such,
health shocks can push households into financial catas-
trophe resulting from health payments and lost earnings
due to inability to work [6, 15, 19–23]. It should be
noted that OOP spending is incurred only when sick
individuals seek care and this can increase if care seeking
is delayed. In poor countries like Malawi, this burden is
particularly disastrous because incomes are significantly
low with almost more than half of the population living
below the poverty line [24] and it is the poorer segments
of the population that are highly affected.
In LIMC context, many studies have concentrated on
assessing determinants of CHE and OOP and established
that poverty, gender, age, education, location have signifi-
cant influence [14, 16, 19, 21, 25]. Other empirical
evidence also suggest that insurance offers some protec-
tion against CHE and poverty [14, 26]. It is also reported
that CHE varies with the threshold (5%,10%,15%,20%,
40%) used [14, 19, 21, 27]. However only one study has
tried to assess the exit time from CHE using Malawian
data [28], and concluded that households have the poten-
tial if proper health insurance can be designed. Since
financial protection is an ultimate goal of every health
system [22, 29, 30] studying the incidence and depth of
poverty due to OOP is important as it gives a picture of
the extent of the financial risk and the level of protection
required from the same . This creates the need to under-
stand the effect of OOP spending on poverty, which may
generate interest among policy makers to consider design-
ing financial protection mechanisms.
The aim of the paper is to investigate effect of OOP
on the incidence and depth of poverty in Malawi. The
paper seeks to determine the extent of CHE due to OOP
and estimate the extent of impoverishment due to OOP.
Our paper adds value to literature in the sense that
firstly it quantifies the incidence and depth of poverty,
secondly it quantifies the extent of impoverishment due
to OOP, and thirdly it quantifies the ext ent of poverty
due to CHE, in the context where health care consump-
tion is free at point of use. This is different to other
studies cited earlier, where user fees are still used in
some places and a number insurance schemes exist. Our
results show that OOP pushes people into poverty even
when they are access ing health for free.
Methods
Data sources
The data used comes from the third round of the inte-
grated household survey (IHS3) conducted by the
Malawi National Statistical Office (NSO) [31], from
March 2010 to March 2011, with financial and technical
support from the Wo rld Bank [24]. IHS3 was a cross
sectional survey with a total sample of 12,271 house-
holds. The sample was drawn using a two-stage stratified
random sampling procedure. This dataset has extensive
information on household socio-economic characteris-
tics including geographic and demographic data, and a
health module. The IHS3 provides a benchmark for
poverty and vulnerability indicators to foster evidence-
based policy formulation, and monitor the progress of
meeting the Millennium Development Goals (MDGs)
(now the Sustainable Development Goals(SDGs) as well
as the Malawi Growth and Development Strategy
(MGDS). The survey is conducted every 5 years. The
IHS3 collected data on out of pocket health spending
over the past 4 weeks, and hospitalization or stay in a
traditional healer’s in the last 12 months.
Analytical methodology
The methods adopted are those proposed by Wagstaff and
Doorslaer [32]. These methods have also been recom-
mended by the World Health Organization [5] and used
in a number of studies [6, 12, 14–16, 21, 25, 33–35] to
investigate similar issues.
Measuring catastrophic expenditures
In this paper, a catastrophic payment is defined based on
a household’s capacity to pay [7] which measures the
proportion of household income remaining after spend-
ing for basic subsistence needs [8]. Basing on this
definition, indicators of catastrophic health expenditure
can be measured as cata strophic head count (HC), cata-
strophic payment overshoot (O), and the mean positive
gap (MPG). Catastrophic head count estimates the share
Mchenga et al. International Journal for Equity in Health (2017) 16:25 Page 2 of 8
of households in the population, whose health care costs
expressed as a proportion of income exceeds a given
discretionary fract ion of their income Z. The cata-
strophic payment overshoot gives the average level by
which payments, as a proportion of income, exceed the
threshold Z, and the mean positive gap measures the
payments in excess of the threshold average over all
households.
Considering R
i
to be the share of health care expend-
iture in non-food expenditure for household i and Z be
the threshold beyond which household i incurs cata-
strophic expenditure, then household i is said to have
incurred catastrophic expenditure if R
i
> Z. There is no
clear guidance on the value of Z and we decided to use
four threshold (10%, 20%, 30% and 40%) based on the
existing literature [5, 6, 17, 19, 23, 27]. HC is then given
by:
HC ¼
1
n
X
n
i¼1
E
i
ð1Þ
Where N is the sample size and E is an indicator
measure that takes the value 1 if R
i
> Z. The average
overshoot is defined as:
O ¼
1
n
X
n
i¼1
O
i
ð2Þ
Where O
i
is the amount by which household i share of
health expenditure in non-food expenditure exceeds the
chosen threshold and is estimated as:
O
i
¼ E
i
R
i
− ZðÞ ð3Þ
The mean positive gap (MPG) is defined as the gap
over the;
MPG ¼
O
HC
ð4Þ
In all the three indicators, the extent of the problem
increases as the magnitude of the indicator increases.
Measuring impoverishing effects of OOP health spending
It is very common to use household per capita
consumption expenditure, as opposed to household per
capita income in estimating money-metric poverty indi-
cators. One of the disadvantages of consumption
expenditure as a welfare indicate is that its measurement
includes expenditures that are not welfare increasing but
rather prevent the deterioration of welfare, such as some
OOP health spending because spending is done when
there is a sickness [7]. Poverty indicators that are based
on con sumption expenditure can therefore underesti-
mate poverty le vels if these expenses are not discounted.
We discounted OOP health spending, from the per
capita consumption expenditure, to estimate the impov-
erishing effects of OOP health spending.
A household is considered to be impoverished by
OOP when its total per capita consumption spending
falls below the poverty line (Malawi poverty line is
K37002 per year
2
) after paying for health care. There-
fore, the difference in the poverty headcounts before and
after discounting OOP for health reflects the poverty
impact of OOP or what is called the impoverishing
impact of OOP.
Using the Foster–Greer–Thorbecke (FGT
) indicators,
P
α
[5], poverty is measured as:
P
α
¼
1
n
X
q
i¼1
PL−X
i
PL
α
ð8Þ
Where n, is the number of households in the sample,
α is some non-negative parame ter, PL is the poverty line,
X denotes per capita consumpti on expenditure, i repre-
sents individuals, and q is the number of households
with consumption below the poverty line. The head-
count (HC) index (α = 0), gives the share of the poor in
the total population. The poverty gap (α = 1), is the
average consumption shortfall of the population relative
to the poverty line. Finally, the severity of poverty is
measured by the normalized poverty gap or the squared
poverty gap (α = 2) [5]. Poverty indicators that have
netted out OOP health spending are derived as P
α
net
:
P
α
net
¼
1
n
X
q
i¼1
PL−X
i
−OOP
i
PL
α
ð9Þ
Where OOP is the per capita out of pocket health
spending. The impoverish ing effects P
α
effect
of OOP on
poverty is then derived as differ ences in the poverty
measures;
P
effect
α
¼ P
α
−P
α
net
ð10Þ
Results
Demographic and socioeconomic profile
This section provides the descrip tive statistics of the
households in our sample (see Table 1). About 84% of
the households that were surveyed live in rural areas,
and this is similar to the population as indicated by the
last national Population and Housing Census (NSO
2008). The average household size is four members
which is slightly lower than 4.6 reported in the most re-
cent Population and Housing Census (NSO 2008).
Slightly over half of the households, have children less
than five years old, while nearly 20% of households have
at least one year aged member. Nearly 70% of the house-
holds have members with at least primary education,
and only 4.87% of the households have members with
tertiary education. The analysis also shows that majority
Mchenga et al. International Journal for Equity in Health (2017) 16:25 Page 3 of 8
(76%) of households are headed by males compared to
72% reported in the 2008 census data. Having a male
headed household would reduce the chances of being
poor [36] and influence resource allocation decision.
Finally, on average the nearest health facility wa s
reported to be 8.59KM away, this makes accessing health
care costly as they have to pay money for transport.
Largely, the demographic characteristics from the IHS3
are similar with the characteristics from the census data
which makes the IHS3 representative of the national
population.
Annual consumption expenditure
The data indicate that the average per capita annual
consumption expenditure was MK59699.80 (US$136.93),
of which almost 60% was spent on food. There are
significant differences in the mean values of per capita
consumption expenditure across quintiles (see Table 2).
The average household expenditure of the richest quin -
tile is nearly nine times of the poorest quintile. The aver-
age expenditure of the middle quintile is almost 32%
higher than the second quintile whereas the average ex-
penditure of the fourth quintile is 34% higher than the
middle quintile.
Catastrophic health expenditures
Table 3 presents the incidence and intensity of cata-
strophic health expenditures in Malawi.
The result s suggest that OOP drives between 0.73
and 9.37% of the households in Malawi to encounter
CHE, as we decrease the threshold from 40 to 10%.
If we increa se the threshold from 10 to 40%, the
mean overshoot drops from 1.01% of expenditure to
only 0.08%. Unlike the head count and the over-
shoot, the mean overshoot among those exceeding
the threshold (MPO) need not de cline as the thresh-
old is raised. Those spending more than 10% of
non-food expenditure, on average spent 20.76% on
health care whereas those spending more than 40%
of the non-food expenditure, on average spent
51.63% on health care.
Health expenditures and poverty
Under this se ction we look at the effect of health care
payments on the incidence and depth of poverty. We
compare the headcount and poverty gap measures
before and after expenditures on healthcare are taken
into consideration. Table 4 demonstrates the sensitivity
of poverty measures in Malawi to the treatment of
health payments.
The conventional methodology of measuring poverty
suggests that about 50.98% of the population is poor. If
OOP payments for health care are netted out of house-
hold consumption, this percentage rises to 51.91%. So
about 0.93% of the Malawian population is not counted
as living in poverty but would be considered poor if
spending on health care is discounted from household
resources. This represents a substantial rise of 1.82% in
the estimate of poverty. The estimate of the poverty gap
of the poor also rises by almost 2.54%, from 19% less
than the poverty line to 20%. Expressed as a percentage
of the poverty line, the poverty gap increases from 9.37
to 9.67% when health payments are netted out of house-
hold consumption.
Table 1 Descriptive statistics (n = 12271)
Variable IHS3 (%)
Residence
Rural 84.40
Urban 15.60
Household size
1 7.06
2 10.13
3 16.87
4 17.72
5+ 48.22
Household age composition
Children <5 years 59.16
Adults > 60 years 19.18
Schooling years per household
No education 0.01
Primary 60.95
Secondary 34.26
Tertiary 4.87
Household Headship
Male 76.01
Female 23.99
Employment status
Employed 21.83
Unemployed 78.17
Access variable
Distance to nearest facility 8.59KM
Source: Author based on NSO (2011) [24] data
Table 2 Expenditure on Health and Access to Health
Quintiles Percent Average annual per capita
consumption expenditure (MK)
Poorest 17.58 17,160.34
Poor 22.17 29,140.16
Middle 20.97 42,833.15
Wealth 21.57 64,973.40
Wealthiest 17.171 153,726.58
Source: Author based on NSO (2011) [24] data
Mchenga et al. International Journal for Equity in Health (2017) 16:25 Page 4 of 8
Impoverishment by location
Figure 1, compares the poverty impact of catastrophic
health expenditures between rural and urban residents.
From the figure, it can be seen that the proportion of
impoverishment in the rural area is relatively larger than
in the urban area. This difference can be attributed to a
lot of factors, one of which is access to health facility
and also that most people in the rural areas rely on
agriculture and majority of them are poor and cannot
afford private insurance. As such whatever little money
they spend in trying to access healthcare leaves them
worse off than before.
Impoverishment by expenditure quintile
Figure 2, compares the poverty impact of catastrophic
health expenditures across income quintiles. It can be seen
that the proportion of impoverishment at poorest quintile
is negligible, as households in the poorest quintile already
live below the poverty line. However, the impact of health
payments on household welfare reaches to the middle
quintile, which has the highest proportion of households
being pushed into poverty due to health care payments.
Negligible amount of the households at fourth and richest
quintiles are impoverished by health payments.
Discussion
We have determined the extent of catastrophic health
expenditure and its impoverishing effects in Malawi where
care is free at point of use in public hospitals and where
no social insurance scheme exist. This is an area of policy
interest given the ongoing implementation of the essential
health care package and the intention to introduce user
fees in some public health facilities such as central
hospitals (tertiary care). Our study results have shown that
OOP drives a substantial proportion of households in
Malawi to encounter financial catastrophe, and that up to
0.93% of households fall below the poverty line after
accounting for health care expenditure. Although the esti-
mated incidence of catastrophic OOP expenditure may
appear to be very low, the effects of the same on the
households that are affected may not be undermined.
These estimates also fall within the ranges that have been
found by other studies [14, 15, 19, 25, 27, 33].
The results show that most of the impoverishing
impact of out of pocket health expenditures is felt by
people who mostly resides in the rural areas and middle
income households. Not captured in this study is the
effect of inability of the poor to spend on health. The in-
ability of the poor to pay for health care limits access
and this may prolong a sickness which may limit their
ability to work reduce their net income. While our
analyses are not without their limitations, there is no
doubt tha t health expenditure contributes substantially
to the impoverishment of households in Malawi, in-
creasing the incidence of poverty and pushing poor
households into deeper poverty. These results are
consistent with earlier stud ies, in whic h poor households
were less able to cope with any given level of health ex-
penditure than richer households [11].
Both the incidence and intensity of poverty, are higher
at lower thresholds, and in all cases, as thresholds in-
crease, the mean positive poverty overshoot increases.
Much of the inc rease in the MPO is due to a modest de-
cline in the mean gap, relative to the headcount as the
threshold is raised. This means that the ‘catastrophic’
effect of health costs was mostly felt through increase in
Table 3 Incidence and Intensity of Catastrophic Health Payments in Malawi
Catastrophic payments measures Threshold budget share, z
OOP health spending as share of non-food expenditure 10% 20% 30% 40%
Head Count (H) 9.37% 3.41% 1.61% 0.73%
Standard Error 0.26% 0.16% 0.11% 0.08%
Overshoot (O) 1.01% 0.43% 0.20% 0.08%
Standard Error 0.04% 0.03% 0.02% 0.01%
Mean Positive Gap (MPG) 10.76% 12.64% 12.15% 11.63%
Source: Author based on NSO (2011) [24] data
Table 4 Measures of Poverty Before and After Netting Out Spending on Health Care, Malawi 2011
Gross of health Net of health Difference
Payments (1) Payments (2) Absolute Relative
(3) = (2)-(1) [(3)/(1)]*100
Poverty Head count 50.98% 51.91% 0.93% 1.82%
Poverty Gap 0.19 0.20 0.48 2.54%
Normalized Poverty gap 9.37 9.67 0.30 3.22%
Source: Author based on NSO (2011) [24] data
Mchenga et al. International Journal for Equity in Health (2017) 16:25 Page 5 of 8