Professional Documents
Culture Documents
1986
Copyright
0X7-9536
56 S3 00 + 0.00
( 1986 Pergamon Press Ltd
S.
AKIS.
CHARLES
and
Carolina
Population
Center.
University
C.
BARRY M.
POPKIS
of Sorth Carolina,
NC 27514, U.S.A.
University
Square
300A. Chapel
Hill.
.Gstract-The
absence of demand analysis for primary health care services has hampered efforts to finance
these services and to make them permanent parts of Third World medical systems. This paper introduces
a demand model for adult outpatient
services, describes the types of data required for estimating it, and
presents the results of a preliminary
estimation
using data from a poor rural region of the Philippines.
The results indicate that prices and distance are not nearly as important
as determinants
of demand in
this sample as has usually been assumed by planners. There appears to be considerable
room for full or
partial financing of outpatient
services from user fees.
I. ISTRODUCTION
Anthropologists,
sociologists and geographers
have had a considerable amount of success finding
negative correlations between medical service use and
so-called barriers to utilization. These barriers include such items as physical distance from households to facilities, cultural distance between patients
and providers, the unavailability of drugs, the length
of time spent waiting at facilities and the unavailability of transportation.
Building on this work, we have developed a demand model under the assumption that a sick individual faces the choice of self-treatment as one alternative, and professional treatment from a traditional
practitioner, a government clinic or a private physician as another. The choice of practitioner depends
on household constraints, such as income, the number of residents, the authority structure of the household and its assets. The choice is also determined
partially by the barriers to use mentioned earlier,
*The authors wish to emphasize the preliminary
nature of
the analysis done in this paper. We have since completed
a much more complete improved analysis, the report of
which has been published as a book, Tk Demand for
Primary Healrh Care in rhe Third World [I], in the fall
of 1984.
which have important economic significance. Distance is translated into the opportunity cost, stated in
terms of both time and money, ot getting to a
practitioner. Similarly, there are time and pecuniary
costs associated with waiting to be seen. purchasing
drugs and making return visits to continue treatment.
Variables must also be included to account for the
effects of education, sex. rural/urban residence and
practitioner supply on medical service use. Figure 1
presents these factors. which are reveiwed in more
detail elsewhere [I].
111.DATA
Data requirements for such a model are extensive.
Community data are required to supply the prices,
travel time and waiting time associated with each type
of facility or practitioner. To this end, it is necessary
to inventory the practitioners used by residents from
each community and to visit each practitioner to
collect facility-level data. Household data must supply information on assets, household organization,
household location, income and time constraints.
Individual data on education, illnesses, pregnancies
and medical practitioner choices must be collected for
all people living in the survey households.
Estimating the demand model described in the
previous section is accomplished in this paper using
two sets of data: a 1978 household survey conducted
in one of the poorest regions of the Philippines, the
Bicol Multipurpose Survey (BMS78) [2], and a 1982
medical facility survey (BMSSSI) collected during
visits to 518 traditional and modern facilities (or lone
practitioners) serving the 100 communities included
in the 1978 survey [3].
From the 1978 survey, household information was
collected on the condition of the dwelling, ownership,
assets and a wide variety of other economic and
demographic variables. Individual information on
about 17,000 household members includes extensive
work, income and home time-allocation data; health,
child-rearing and pregnancy-related information: and
a massive compilation of demographic variables.
Respondents were asked whether anyone in the
household had been sick in the month previous to the
survey and, if so, what was done for the illness,
whether it was serious, how much was spent, how
long physician visits took, and so on. One problem
with this survey was that only cash- and timeexpenditure data reported by people who used medical services were available to use as proxies for
medical visit prices.
In the medical service demand literature, the use of
expenditures instead of prices is a common approach.
but it is to be avoided for several reasons. First,
expenditures are simply the number of visits multiplied by the average price of each visit. What a
demand model explains, however, is the number of
visits; so it is a serious mismeasurement of the price
variable to include elements of the dependent variable
in it. Second, expenditures are likely to include more
than just the price of the visit; drugs are the most
obvious addition. Third, an individual will not report
expenditures for services not used. For example, if
someone chooses a private physician, he or she will
have missing values for expenditures at traditional
Demand
for adult
outpatient
services
in the Philippines
323
PRICE
cost
rransportar1on
rime
waltlng
money cost
co~nwrance
OTHER
cash price
/soc~aI
securlry
PRICES
subst!rJtes
INCOME
level
sources
fYpes
assefs
TIME
nature
wealth
ALLOCATION
of work
felt
FOR
HEALTH
CARE
modern
public
modern
prlvare
trodiflonal
accupaflon
HEALTrl
physlologtcal
DEMAND
PRIMARY
prtvote
NEEDS
/real
household
stze
KNDWLEDGE
culrural
INFORMATION
Issues
education
SEASONALITY
healfh
effecrs
cost
effecrs
E>lPIRICAL
STUDY
JOHN S. AKIS
37-l
source of disagreement
with the approach used here
is the matching of 1978 household data (BMS78) and
1981 barangay-level
prices from the facility survey
(BMSS8l). It should be clear at the outset. however.
that in order to isolate demand equations for medical
services from other goods categories. the assumption
is already made that the relative prices of the goods
are stable. which is the same assumption necessary to
validate the matching of 1978 and 1981 surveys. We
are not assuming that in 1978 the sample faced the
same prices as were collected in 1981, only that they
faced the same price structure. If private clinics were
twice as expensive as traditional healers in I98 1, they
are assumed to have been twice as expensive in 1978.
This is not, in fact, an outlandish
assumption
over
such a short time period for services that are close
substitutes for each other.
The adult outpatient model is a modification of the
following general demand system:
where
Q,, = whether medical service i is used by the jth
individual. where i = (public. private. traditional. or no care) and j = (all sick or
pregnant individuals, the sample depending
on the model)
PU, = public clinic or hospital
serving the jth
individual
PR, = private clinic or hospital serving the jth
individual
TR, = traditional healer or midwife serving thejth
individual
P = vector of cash prices paid for each service
(including visit cost, drug costs and transport costs)
T = vector of time costs associated with each
facility and service (waiting time, transport
time)
Y, = household assets for the jth individual
Z, = a vector of social, demographic,
and biological control variables for thejth individual.
I. List of variables
Dependent
The choice
of practitioner
variable
for adult
Frequency
Visit:
Traditional
Public
Private
60
73
124
142
NOM
Independent
er al.
variables
to USSl.00
0.49 pesos*
14.49 pesos
3. IO pesos
13.72 km
9.41 km
0.31 km
63.24
100.32
I I.68
26.39
0.1 I
on the relative
outpatient
?/,
15
IS
31
36
Mean
Opporiunit~ cm0
Cash prices for one adult outpatient visit
Public
Private
Traditional
Distance to closest facility of practitioner
Public
Private
Traditional
Waiting time
Public
Private
Traditional
Amount spent on drugs
Whether covered by insurance (0 = no; I = yes)
Household msers and income
Number of rooms in house
Annual household income from all sources
Number of individuals
in household
Sanitary raring of water source
Sanitary quality of toilet facilities
Demographic
Male (0 = female; I = male)
Urban (0 = rural; I = urban)
Model
Type of household
0 = extended family
I = nuclear with other residents
3 = nuclear or single
Education
Highest grade completed
Perceprion
Perceived quality of life-family
health and physical condition:
I = dissatisfied
7 = very satisfied
Perceived quality of life--availability
of health services:
I = dissatisfied
7 = very satisfied
depends
SD
2.00
4.05
1.66
13.37
10.45
0.53
min
min
min
pesos
84.32
101.67
36.3 I
70.81
0.3 I
I .43
1955.83 pesos
6.79
1.56
1.82
0.94
10426.23
2.78
1.04
1.33
0.52
0.23
0.50
0.42
2.44
I .08
6.57
3.80
I .40
3.09
I.21
3.4
wits,
or
No Visit = no professional
sought.
clinic or to a
consultation
P (Traditional
log
P (Traditional
P (Private
log P
Visit)
Visit)
= JYPublic
Visit)
Visit)
= WLXe
P (No Visit)
(Traditional
Visit)
= UN0 ,W.
signs for
demand
in Table
variables
315
326
Table
2.
Expected
signs
far
economic
variables
\s
distance.
waiting
logrt
1s
traditional
Public
L
?
Private
coberqe
No
No
VIJlt
vs
traditional
>
estrmates
PrrKite
pribate
public
VlSli
vs
\s
pubhc
time
Traditional
Insurance
outpatient
visit
No
rs
traditional
Pnces.
in adult
Pn\ste
Pubix
+
_
Income--assets
Sumber
of
Income--all
T.ible
rooms
in house
3.
Multiple
logit
results:
demand
for
adult
outpatient
so~~rces
services,
Philippines.
1978.
Coefficient
estrmates
(asymptotic
r-value
in
parentheses)
Probability
of
visit
Private
Public
vs 8
No
traditional
traditional
No
Private
VS
-.S
visitt
visit
\&it
traditional
No
visit
VS
private
public
public
Prices
Traditional
0.1513
0.0897
(1.136)
Public
(0.763)
0.0499
Prirate
(0.625)
0.047s
0.009
(0.090)
-0.036
(-0.137)
(-0.166)
-0.0782
- 0.069
(-0.860)
(-0.617)
-0.055
(-0.754)
-0.0166
-0.078
(-0.663)
-0.019
-0.0067
(O.SS5)
-0.062
(-0.524)
(-0.199)
0.0592
(0.4397)
0.073
(0.653)
-0.0295
-0.084
(-1.18)
(-
(-0.752)
1.75)
DiciUtlCtT
Traditional
Public
Pri\
1.212
0.881
ate
(I ,696)
0.00 I2
(2.496)
(0.0669)
(0.582)
0.0094
-0.0234
(-
0.607
(1.08)
0.0137
(-1.39)
-0.605
(-0.790)
0.0082
(0.887)
-0.0253
1.061)
-0.275
0.330
(1.27)
-0.0126
(0.587)
-0.01
-0.002
(-0.973)
( - 0.098)
(-2.09)
0.0044
(0.879)
(0.355)
0.0061
0.008
(0.247)
(0.494)
Waiting rime
Traditional
-0.0195*
-0.0107*
(-
Public
2.10)
(-
-0.008
-0.0032
Drug
expenditure
(0.032)
0.0181*
0.01879
Insurance
I.22
1.378
(1.497)
(I ,827)
Income
a.sW,S
Number
of
rooms
in
house
-O.OS-ll
Income--all
sources
0.242
Household
size-people
o.OvOao3
(0.208)
(-0.777)
(-0.3%)
Quality
of
water
0.4-l5*
0.4232
(1.95)
Quality
of
toilet
-0.023
(-0.0136)
(-
-0.027
(0.345)
(-0.445)
(-
0.943
1.39)
(-0.250)
0.0778
0.0506
(1.378)
(0.785)
-3.91
-0.412
1.097)
(-
1.94)
- 0.000007
(-1.14)
(-2.07)
(-0.197)
(0.649)
-0.366
(-
-0.CGO3
(-0.021)
(0.172)
-0.219
-0.0402
(-1.04)
0.233
-0.977
(-1.97)
(-0.0184)
-0.00@02
0.0321
(2.07)
-0.820
0.326*
(-1.02)
-0.0545
-0.0619*
(-5.33)
(-1.37)
(1.66)
- 0.00003
(-0.899)
-0.0615*
0. I57
-0.124
0.00014
(0.081)
(-5.28)
(0.280)
(-0.527)
-0.00002
-0.0272
0.00035
0.40 I
(1.04)
(-0.325)
0.00339
(1.47)
(0.192)
(0.517)
0.0038
(1.55)
3.24)
(1.52)
(-3.44)
coverage
-0.0067
(-
-0.0032
-0.0432.
(2.74)
(2.672)
I*
(-4.14)
O.OOQ? I
(0.1033)
-0.00007
(1.57)
(0.321)
-0.01
(0.539)
0.0104
0.0016
(-1.28)
-0.0012
(-0.923)
(-1.25)
-0.0088
(-2.154)
-0.0025
(3.22)
Private
-0.0091*
2.923)
(-
2.35)
O.lli
1.307)
0.314
(2.44)
(0.803)
Demographic
(I = male,
Male
0 = female)
0.463
-0.0149
tirban
(I
Scale
= urban,
of
0 = rural)
wad.-nuclear
HH
0.22
(1.29)
(-0.376)
0.310
0.406
0.370
(0.491)
(0.719)
(0.676)
-0.336
-0.0812
Education
0.0955
-0.0354
0.0601
0.0659
(-0.0873)
0.0375
-0.0285
(-0.205)
(0.239)
0.023
0.0437
0.607
(0.455)
(1.39)
(-0.833)
(0.120)
(0.427)
0.076
(0.558)
- 0.242
0.371
(1.10)
(0.195)
1.94)
(-
0.0325
0.0092
(0.149)
0.0618
(1.80)
-0.363
(-1.720)
(-0.329)
(0.657)
(0.942)
(1.33)
Perceptions
Satisfaction
with
family
health
-0.00421
-0.0625
(-0.413)
Satisfaction
with
health
services
-0.074
Consmnt
lr+alues
+The
numbers
first
relative
at
reported
entry
in the
or
are
first
traditional
visit
abobe
the
column
to a traditional
to a traditional
the
(-
-0.451
is signiricant
visit
goes
practitioner.
0.10
level.
logarithms
of
is positive
goes
up.
down-people
the
In contrast.
are
more
(-
which
that
indicates
1s traditional
likely
to
that
relative
as the
waiting
use
tradrtronal
time
goes
rather
probabihty
price
up (-0.0107*),
than
pubhc
0.0172
(0.800)
(0.125)
3.00
I.41
I .59
to the
(-0.426)
-0.122
(-0.937)
traditional
-0.0486
(-0.213)
-0.140
1.269)
2.55
occurs
-0.0281
(0.162)
-0.1964
1.38)
I.144
probability
(O.l513).
(-0.686)
-0.2136
( - 0.429)
0.0204
-0.0907
(-0.310)
that
increases.
the
facilities
occurx
the
probabdit)
if they
For
example.
probability
the
of
a public
of a public
relative
must
wait
longer
for
variables
The demographic
327
males to improve the quality of their care: alternatively. if males in the sample suffer from more serious
problems, it may be simply a need-oriented rather
than a behavioral phenomenon.
In only one case does the education variable even
approach statistical significance. This is in the cell for
the probability of no visit relative to a traditional
visit. The sign is positive, a finding consistent with
many simple correlations done by other authors,
which show that educated people are more willing
and confident than others to engage in home treatment.
The perception variables are included to control
for perceptions and acculturation
with no expectations about the signs.
V. CONCLUSION
JOHN S. AKIN er al
328
REFERENCES