You are on page 1of 14

13th National Convention on Statistics (NCS)

EDSA Shangri-La Hotel, Mandaluyong City


October 3-4, 2016

ESTIMATING THE PREVALENCE OF YOUNG ADULT OBESITY IN PROVINCES


AND HIGHLY URBANIZED CITIES IN THE PHILIPPINES USING POISSON
REGRESSION WITH SPATIALLY AUTOCORRELATED RANDOM EFFECTS

by

Marie Joy F. Lopez

For additional information, please contact:

Author’s name Marie Joy F. Lopez


Designation Instructor 7
Affiliation Institute of Statistics (INSTAT), University of the Philippines Los Baños
Address 3rd Flr. Physical Sciences Bldg., UPLB, College, Los Baños, Laguna
Tel. no. (049) 536-2381
E-mail mflopez2@up.edu.ph

Page 1 of 14
ESTIMATING THE PREVALENCE OF YOUNG ADULT OBESITY IN PROVINCES
AND HIGHLY URBANIZED CITIES IN THE PHILIPPINES USING POISSON
REGRESSION WITH SPATIALLY AUTOCORRELATED RANDOM EFFECTS1

by

Marie Joy F. Lopez2

ABSTRACT

This paper presents the results of a study on using the model-based approach in
estimating the prevalence of young adult obesity in provinces and highly urbanized cities
(HUCs) in the Philippines. Specifically, this demonstrates the use of a Poisson regression
with spatially autocorrelated random effects as an estimation procedure for count data which
is the nature of most health-related outcomes. The model also accounts for residual spatial
dependence which may be present when the variable of interest is measured in areas that
are geographically proximate. The resulting model identified provincial count of commercial
establishments, provincial count of men who have at least college education and provincial
count of barangay with health centers as determinants of young adult obesity accounting
for about 70% of the total variation in the variable under study. Furthermore, residual spatial
dependence was present within 100 kilometer radius. Using the model, 95 out of 112
(84.82%) of provinces and HUCs have reliable estimates and 17 provinces are unreliable
but still with acceptable measures of reliability.

Keywords: small area estimation, model-based approach, count data

I. INTRODUCTION

The government, together with international partners, civil society, private sectors and
communities play a vital role in changing the policies and practices that shape behavior around
diet and well-being. As basis for their decisions, they usually rely on data and statistics to come
up with efficient solutions on the alarming issues of the society. As a common practice, national
surveys are conducted to produce official statistics. Important indicators were derived from these
surveys to inform the public of the present status of the nation as well as to assess the impact of
existing government policies. For instance, the National Nutrition Survey (NNS) conducted by
the Food, Nutrition and Research Institute (FNRI) is the basis of the Philippine government to
evaluate the food and nutrition situation of the country and the nutritional status of the population.

Most of these surveys, such as that of the NNS, employ the multistage probability
sampling design where the primary sampling units are frequently constructed from enumeration
areas and the secondary sampling units as well as the ultimate sampling units are the
households and the individuals, respectively. This design is being carried out in most surveys
primarily for frame development and for clustering interviews in order to reduce cost (United
Nations, 2005). However, one of the limitations of this design is that it only produced precise and
reliable estimates at the national and subnational levels such as administrative regions. Coming
up with precise estimates for the domain of interest (e.g. smaller domains) requires adequate
sample sizes to be allocated to the domain.

The increasing demand from policy makers and administrators for reliable estimates at
small area level has led to the development of several small area estimation techniques. One
of which is the use of a direct estimator—an estimator which uses values of the variable of
interest only from the sample units in the domain of interest. Though commonly used, one major
disadvantage of using this estimator is the unacceptably large standard errors which may be a
result of small sample sizes within the domain (Hidiroglou, M., 2007). An alternative is to use an

1 Part of the ongoing master’s thesis in the University of the Philippines Los Baños
2 Instructor, Institute of Statistics, College of Arts and Sciences, University of the Philippines Los Baños

Page 2 of 14
indirect estimator which borrows strength across areas by modelling the dependent variable on
the independent variables. The use of these estimators has been proven efficient in producing
reliable estimates in the domain of interest.

This study demonstrates a model-based estimation technique in estimating the


prevalence of young adult obesity in provinces and highly urbanized cities in the Philippines. For
the past years, there has been a growing concern on the increasing rate of obesity in the country
as the number of overweight and obese adults already doubled from 16.6% in 1993 to 31.1% in
2013 (National Nutrition Council, 2015). Nevertheless, this study aims to illustrate the advantage
of using a Poisson regression with spatially autocorrelated random effects as an estimation
procedure for count data (which is the nature of most health outcomes) and in addition,
accounting for residual spatial dependence which may be present when the variable of interest
is measured in areas that are geographically proximate.

II. METHODOLOGY

2.1. Data Sources

The data sets that were used in the study includes the data from the Anthropometric
component of the 7th National Nutrition Survey conducted last 2008 by the Food and Nutrition
Research Institute (FNRI), the 2007 Census of Population and Housing (CPH), and the 2008 List
of Establishments (LE), all gathered by the National Statistics Office (NSO).

The National Nutrition Survey (NNS) is conducted every five years in the country to
assess the food and nutritional status of the Filipinos and various population groups. The survey
has four components namely: Dietary, Anthropometric, Biochemical and Clinical of which the
BMI is obtained in the anthropometric component of the survey. The 7th NNS adopted the Master
Sample of the National Statistics Office from the first quarter Labor Force Survey and is based
on a multi-stage stratified sampling design covering all regions and provinces in the country. The
primary sampling units (PSUs) are contiguous barangays with at least 500 households within
provinces while the secondary sampling units or enumeration areas are the contiguous areas in
a barangay within PSUs, and the ultimate sampling units are the households. In addition, the
NNS Anthropometric survey component utilized the four replicates of the master sample to come
up with provincial estimates which covered 3,377 enumeration areas and 36,348 sampled
households in which all members in the sampled households were enumerated. Among the total
survey sample, 71, 808 adults were included (Patalen-Dasco et al., 2008).

The auxiliary variables that were utilized for the model-based estimation came from 2007
Census of Population of Housing (CPH). The CPH provides the information on the size and
distribution of total population as well as characteristics of individuals and households such as
household size and highest educational attainment of the household head. It also provides
information on supply, characteristics and facilities of housing units as well as characteristics of
barangays such as number of schools, number of service centers and number of health facilities
(PSA, 2013).

2.2. Estimation of the Prevalence of Young Adult Obesity

The prevalence of young adult obesity in provinces and highly urbanized cities was
determined by obtaining the ratio of the number of obese young adults and total number of young
adults in the given province/ highly urbanized city. The variances and coefficients of variation of
these estimates were also computed to assess the precision and reliability of the estimates.

The proposed model-based estimation procedure in estimating the count of obese young
adults is a generalized mixed model composed of a fixed effect (effects associated with auxiliary
count) and spatially autocorrelated random effects which will account for residual spatial
dependence that may be present when the variable of interest is measured in areas that are

Page 3 of 14
geographically proximate. The said model is a Poisson model with spatial random effect given
by:
 
log Yi SP  z T * T
i β + ui γ

where 𝑧𝑖𝑇is a vector of auxiliary count data, 𝛽 ∗ is a vector of regression parameters, 𝑢𝑖𝑇
is a vector of random effects and 𝛾 is a vector of parameters for the random effects. The spatial
correlation in the random effects will be incorporated in its variance-covariance structure denoted
as σ 2f(dij ) .
Unlike the residual variance-covariance structure of a usual generalized linear model
𝜎12 ⋯ 0
Σ=( ⋮ ⋱ ⋮ ) which assumes errors that are independent (hence covariance is zero),
0 ⋯ 𝜎𝑝2
𝜎12 ⋯ 𝑓(𝑑𝑖𝑗 )
2
σ f(dij ) has Σ = ( ⋮ ⋱ ⋮ ) where dij represents the Euclidian distance between
𝑓(𝑑𝑖𝑗 ) ⋯ 𝜎𝑝2
province i and province j measured based on their geographic coordinates (latitude and
longitude) that reflect the geographical centroids of each province and f (dij) is the covariance
between two provinces as a function of their distance.

In determining the best spatial structure when spatial information is treated as random
effects, an empirical semivariogram is constructed (Cressie, 1991). An empirical semivariogram
is a plot showing the variability of the observations in pairs of locations separated by distance h
wherein plotted in the x-axis are the distances and on the y-axis is the semivariance at distance
h computed as:
2
𝑆𝑒𝑚𝑖𝑣𝑎𝑟𝑖𝑎𝑛𝑐𝑒(𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒ℎ ) = 0.5 ∗ 𝑎𝑣𝑒𝑟𝑎𝑔𝑒((𝑦𝑖 − 𝑦𝑗 ) ))

Based from the given formula, this suggests that in pairs of locations that are closer to each
other, the average squared difference will be lower compared to the pairs of locations that are
farther apart since nearby locations tend to have similar values than those father apart (Tobler’s
First Law Geography). However, in most cases, the semivariogram already levels off to a
constant value called sill at a separation distance, called range(α). Beyond this distance,
observations are spatially uncorrelated (Waller and Gotway, 2004).

The empirical semivariogram provides information on the spatial autocorrelation of datasets but
it does not provide information for all possible directions and distances, hence, similar to the
regression analysis, a continuous line or curve must be fitted to the data points.

The three commonly used covariance structures and their corresponding covariance
structures are: exponential - σ2exp  -dij / α  , Gaussian - σ2exp  -dij / α  and spherical -
 3d 1  d ij  
3
2 
σ exp 1 -
ij
+  
 2α 2  α  
 .

III. RESULTS AND DISCUSSION

3.1. Correlates of Local Level Obesity Prevalence

Prior to indirect estimation, pairwise correlation analysis was conducted to identify the
relationship between the estimated provincial and city count of obese young adults and some
auxiliary variables which characterize a province or a city. Out of 56 auxiliary variables
considered in the study 49 variables were found to be significantly associated with the estimated
provincial and city count of obese young adults. These variables represent the demographic
characteristics and composition of the household and its members (e.g. gender, marital status

Page 4 of 14
and educational attainment) and the economic characteristics of a province or a city (e.g.
infrastructures and availability of services).

Among the demographic characteristics, the provincial count of household heads who
have at least high school education and the provincial count who at least college education have
the strongest associations. These variables have a positive correlation coefficient which
indicates a direct association to the variable of interest. It can be further noticed that some
variables pertaining to education are significantly associated to the estimated count of obese
young adults which suggests that education is a significant factor in determining the count of
obese adults.

Another dimension investigated in the study is the economic characteristics of the


province or the city. In particular, the presence of commercial infrastructures in the province or
city were examined since based from the literature this factor significantly contributes to the rising
cases of obesity. True enough that this variable was found to be strongly and positively
associated with the provincial count of obese young adults, along with provincial count of
restaurants/cafes/fastfood which is also directly associated to the provincial count of obese
young adults.

3.2. Estimation of the Prevalence of Young Adult Obesity using Poisson Regression with
Spatially Autocorrelated Random Effects

Initially, a Poisson model was constructed to determine the variables which are dependent
to the estimated count of obese young adults. Among the variables which are associated to the
estimated count of obese young adults, the provincial count of men with at least college
education, the provincial count of commercial establishment and the provincial count of
barangays with puericulture or barangay health centers are the variables which entered the
model. Afterwhich, spatial autocorrelation and spatial dependencies were examined by obtaining
the Global Moran’s I coefficient and constructing the empirical semivariogram of the residuals
from the Poisson model. The correlogram and empirical semivariogram are shown in Figures 1
and 2 respectively.
0.35
0.288
0.3

0.25
Moran's I

0.2

0.15

0.1

0.05

0
50 60 70 80 90 100 110 120 130 140 150
Distance

Figure 1. Spatial correlogram of the total number Figure 2. Empirical semivariogram of the residuals
of obese young adults. from the Poisson model

Based from the spatial correlogram, the Moran’s I coefficient is at its maximum at 90
kilometers threshold distance. Further increase in the threshold distance, results to a decrease
in the Moran’s I coefficient. The obtained Moran’s I coefficient (I=0.288, p-value: <0.0001)
indicates spatial clustering of similar values (high-high or low-low) of the total number of obese
young adults across the different provinces and highly urbanized cities. In addition, the significant
spatial autocorrelation implies that in provinces and highly urbanized cities within approximately
90-kilometer distance, there exists spatial correlation in the count of obese young adults.

Page 5 of 14
Since spatial autocorrelation is present, spatial dependencies might exist in the model
residuals. This was examined through an empirical semivariogram of the residuals shown in
Figure 9. At 100-km threshold distance, the semivariogram started to level off. This means that
within approximately 100-km, the count of obese young adults is spatially correlated and beyond
this distance, the counts are no longer correlated. Further, the Moran’s I coefficient at this
threshold distance (I=0.0644, p-value: 0.0185) also indicates presence of spatial autocorrelation.

To determine the appropriate covariance structure, three widely-used theoretical


covariance structure were fitted in the empirical semivariogram. Figure 3 shows that the spherical
covariance structure fits well the empirical semivariogram.

Figure 3. Empirical and fitted theoretical semivariogram of the residuals from the Poisson model

Incorporating the spherical covariance structure in random error component of the


Poisson Regression produces estimates with high precision and good measures of reliability.
The significant predictors which accounts for about 70% of the total variation in the obesity counts
are shown in Table 1.

Table 1. Estimated regression coefficients of the spatial Poisson regression model for the
provincial and highly urbanized cities count of young adult obesity.

INCIDENCE
PREDICTOR COEFFICIENT RATE STANDARD
RATIO ERROR
Provincial count of commercial establishments 0.000109 1.0001 2.51e-05
Provincial count of barangays with puericulture
or barangay health center -0.000775 0.9992 2.51e-04
Provincial count of men with at least college
education 0.000002 1.000002 8.20e-07
Constant 8.138706 3424.4837 0.0964

The positive coefficients of the provincial count of commercial establishments and the
provincial count of men with at least college implies that for every unit increase in these variables,
the log expected counts of obese young adults increases by 0.000109 and 0.000002,
respectively. On the other hand, the provincial count of barangays with puericulture or barangay
health center decreases the rate of the provincial count obese young adults by 0.08%.

Page 6 of 14
Given the model, 95 provinces and highly urbanized cities have reliable estimates and
the remaining 17 of which are unreliable but still with acceptable measures of reliability (See
Appendix, Table 1A). Quezon City, Cavite and Cebu are the top 3 provinces with the highest
count of obese young adults and Batanes, Camiguin and Siquijor are the top provinces with the
highest prevalence of young adult obesity. The high prevalence rate of obesity in these provinces
may not be due to high number of young adults who are obese but may be attributed to low
number of young adults in these provinces, thus, resulting to large proportion of obesity cases.
Similarly, Apayao which is the province next in rank has the same case. Among the provinces/
HUCs with high prevalence of young adult obesity are Iligan City, Biliran, Mountain Province,
San Juan City and Davao del Sur. Their corresponding CV’s are given in Table 2.

Table 2. Top 10 provinces/HUCs with high estimates of young adult obesity prevalence.

Prevalence of Young
Province/Highly CV
Rank Region Adult Obesity
Urbanized City (%)
(%)

1 BATANES II 78.50 9.42


2 SIQUIJOR VII 56.13 8.99
3 CAMIGUIN X 17.61 8.94
4 APAYAO CAR 11.55 8.97
5 ILIGAN CITY X 11.00 7.94
6 BILIRAN VIII 10.57 8.53
7 MOUNTAIN PROVINCE CAR 9.51 8.66
8 SAN JUAN CITY NCR 8.64 8.75
9 DAVAO DEL SUR XI 8.35 5.72
10 CATANDUANES V 8.18 7.91

The distribution of the count and the prevalence of young adult obesity across the country
are shown in the choropleth map given in Figures 4 and 5.

NCR NCR

Figure 4. Choropleth map of the spatial Poisson Figure 5. Choropleth map of the spatial Poisson
estimates of the count of young adult estimates of the prevalence of young
obesity in the Philippines. adult obesity in the Philippines.
Page 7 of 14
Based from the map, nearly half of the provinces and cities have estimates ranging from
0% to 4% which suggests that majority of the provinces have prevalence rate lower than the
national estimate (which is 4.5% of the total young adult population). However, there are few
provinces with noticeably high prevalence of young adult obesity. Such provinces are distributed
across the country, however, most of them can be found in Luzon. The distribution of the
estimates can be further explained using the boxplot in Figure 6 which illustrates a highly skewed
distribution.

Mountain Province
CIty

Camiguin

Batanes
Apayao

Siquijor
Biliran
Iligan

0 20 40 60 80
Prevalence of Young Adult Obesity

Figure 6. Distribution of the estimates of the prevalence of young


adult obesity in the Philippines using the Poisson
regression with spatially correlated random effects

In terms of the precision and reliability of the estimates, the model produced low standard
errors and coefficients of variations within the prescribed measure of reliability. About 85% of
these estimates are reliable and approximately 15% are unreliable but within the acceptable
measure of reliability. (See Table 3).

Table 3. Distribution of the coefficient of variation of the estimates of prevalence of young adult
obesity in the different provinces and highly urbanized cities in the Philippines using the
spatial Poisson regression.

NO. OF
ABSOLUTE COEFFICIENT OF VARIATION PROVINCES PERCENTAGE
(%) AND HUCs (%)
0.00 – 10.00 95 84.82
10.01-20.00 17 15.18
20.01-30.00 0 0.00
> 30.01 0 0.00

IV. SUMMARY AND CONCLUSION

The Poisson regression with spatially autocorrelated random effects produced reliable
estimates of the prevalence of young adult obesity in each province and highly urbanized city.
Majority of the provinces and HUCs have estimates within 0% to 4% which is lower than the
obtained national estimate of 4.5%. The top 3 provinces with the highest prevalence of young
adult obesity are Batanes, Siquijor and Camiguin. The high prevalence rate of obesity in these
provinces may not be due to high number of young adults who are obese but may be attributed
to low number of young adults in these provinces, thus, resulting to large proportion of obesity
cases. Among the provinces with high estimates of count and prevalence of young adult obesity

Page 8 of 14
are Iligan City, Biliran, Mountain Province, San Juan City and Davao del Sur. Two of which are
highly urbanized cities which implies that high prevalence rate of obesity is already evident in
cities. This is in accordance with the previous studies that rising obesity cases is a result of rapid
economic transition and urbanization. Similarly, the provincial count of commercial
establishments are among the determinants of obesity, along with the provincial count of men
who have at least college education and provincial count of barangays with health center. The
other provinces with also high prevalence rates are parts of the regions with high percentage of
young adults who are obese. Further, these regions are still the top regions with high percentage
of young adult obesity from the recent NNS survey (NCR:10.1%, Northern Mindanao:8.7%,
CAR:7.5%) suggesting that there is a need to address the issue of obesity in these areas.

LITERATURE CITED

Cressie, N. A. C. (1991). Statistics for Spatial Data. New York: John Wiley & Sons, Inc.

Hidiroglou, M. (2007). Small Area Estimation: Theory and Practices. Section on Survey
Research Methods.

National Nutrition Council (2015). Talking Points: Timbang iwasto sa tamang nutrisyon at
ehersisyo. Retrieved from
http://www.nnc.gov.ph/~ziamelinbe/images/nutrition_month/2015_nm_talking_points
_final.pdf

Waller, L., and Gotway, C. (2004). Applied Spatial Statistics for Public Health Data. New York:
John Wiley and Sons.

United Nations (2005), Households Sample Surveys in Developing and Transition Countries.
United Nations Publication. E.05.XVII.6 retrieved from
http://unstats.un.org/unsd/hhsurveys/pdf/Household_surveys.pdf

ACKNOWLEDGEMENT

The author gratefully acknowledged the Food and Nutrition Research Institute (FNRI) for
providing the data used in this study.

Page 9 of 14
APPENDIX

Table 1A. Spatial Poisson estimates of the provincial and highly urbanized cities prevalence of young adult obesity
in the Philippines.

95% Confidence
Total No. Prevalence Interval
Estimate of
of Young of Young
Province/ the Total Standard CV
Adults Adult
Highly Urbanized Cities No. of Error (%)
based on Obesity
Obese
2007 CPH (%)
Lower Upper
Limit Limit

CAR
ABRA 4,155 66,405 6.26 0.0051 0.0526 0.0725 8.13
APAYAO 3554 30,774 11.55 0.0104 0.0952 0.1358 8.97
BENGUET 7,065 118,810 5.95 0.0037 0.0522 0.0667 6.21
BAGUIO CITY 5,081 107,340 4.73 0.0038 0.0400 0.0547 7.94
IFUGAO 3,494 55,035 6.35 0.0059 0.0518 0.0751 9.36
KALINGA 3,767 51,487 7.32 0.0064 0.0606 0.0857 8.73
MOUNTAIN PROVINCE 3,733 39,245 9.51 0.0082 0.0790 0.1113 8.66
REGION I
ILOCO NORTE 5,640 170,179 3.31 0.0023 0.0286 0.0376 6.94
ILOCOS SUR 6,628 192,051 3.45 0.0027 0.0292 0.0398 7.84
LA UNION 7,208 219,680 3.28 0.0024 0.0282 0.0374 7.19
PANGASINAN 17,976 767,920 2.34 0.0030 0.0176 0.0292 12.64
REGION II
BATANES 3440 4,382 78.50 0.0739 0.6401 0.9299 9.42
CAGAYAN 7,729 312,372 2.47 0.0020 0.0209 0.0286 7.97
ISABELA 12,888 429,745 3.00 0.0036 0.0228 0.0371 12.17
NUEVA VIZCAYA 4,982 121,983 4.08 0.0030 0.0350 0.0467 7.27
QUIRINO 3,806 50,738 7.50 0.0064 0.0624 0.0876 8.55
REGION III
AURORA 3,734 50,693 7.37 0.0064 0.0611 0.0862 8.70
BATAAN 5,898 211,153 2.79 0.0018 0.0243 0.0315 6.60
BULACAN 40,117 907,808 4.42 0.0043 0.0357 0.0526 9.76
NUEVA ECIJA 10,349 567,123 1.82 0.0014 0.0156 0.0209 7.48
PAMPANGA 21,232 612,324 3.47 0.0024 0.0299 0.0394 6.96
ANGELES CITY 5,833 108,123 8.65 0.0047 0.0203 0.1528 5.49
TARLAC 8,079 376,756 2.14 0.0012 0.0191 0.0238 5.69
ZAMBALES 7,354 144,576 5.09 0.0037 0.0436 0.0581 7.25
OLONGAPO CITY 4,339 75,367 5.76 0.0049 -0.0269 0.0741 8.53

Page 10 of 14
Continuation…

95% Confidence
Total No. Prevalence Interval
Estimate of of Young of Young
Province/ Standard CV
the Total No. Adults Adult
Highly Urbanized Cities Error (%)
of Obese based on Obesity
2007 CPH (%)
Lower Upper
Limit Limit

NCR
MANILA CITY 28,248 600,038 4.71 0.0048 0.0376 0.0565 10.24
MANDALUYONG CITY 4,691 113,516 4.13 0.0035 0.0344 0.0482 8.51
MARIKINA CITY 6,066 148,205 4.09 0.0032 0.0346 0.0473 7.89
PASIG CITY 8,221 233,531 3.52 0.0028 0.0298 0.0406 7.87
QUEZONCITY 55,300 976,267 5.66 0.0084 0.0401 0.0731 14.87
SAN JUAN CITY 4,082 47,241 8.64 0.0076 0.0716 0.1012 8.75
CALOOCAN CITY 11,594 473,834 2.45 0.0021 0.0204 0.0285 8.43
MALABON CITY 4,424 124,460 3.55 0.0032 0.0293 0.0418 9.03
NAVOTAS CITY 3,859 81,642 4.73 0.0045 0.0384 0.0561 9.54
VALENZUELA CITY 6,739 209,611 3.22 0.0025 0.0273 0.0370 7.69
LAS PINAS CITY 5,617 193,917 2.90 0.0026 0.0239 0.0340 8.94
MAKATI CITY 9,112 209,350 4.35 0.0039 0.0359 0.0512 8.95
MUNTINLUPA CITY 5,101 169,292 3.01 0.0026 0.0250 0.0353 8.75
PARANAQUE CITY 7,688 206,582 3.72 0.0030 0.0313 0.0431 8.12
PASAY CITY 5,606 159,273 3.52 0.0027 0.0300 0.0404 7.60
TAGUIG CITY 5,652 225,005 2.51 0.0023 0.0207 0.0296 9.02
REGION IV-A
BATANGAS 13,801 707,018 1.95 0.0022 0.0153 0.0238 11.11
CAVITE 53,196 983,337 5.41 0.0062 0.0420 0.0662 11.37
LAGUNA 30,994 845,726 3.66 0.0028 0.0313 0.0420 7.51
QUEZON 11,489 457,528 2.51 0.0027 0.0198 0.0304 10.71
LUCENA CITY 4,503 60,726 7.42 0.0061 -0.0057 0.0098 8.28
RIZAL 23,974 761,494 3.15 0.0031 0.0254 0.0375 9.79
REGION IV-B
MARINDUQUE 3,969 53,421 7.43 0.0061 0.0623 0.0863 8.21
OCCIDENTAL MINDORO 4,338 119,360 3.63 0.0029 0.0306 0.0421 8.09
ORIENTAL MINDORO 5,135 192,831 2.66 0.0020 0.0226 0.0306 7.63
PALAWAN 6,110 175,742 3.48 0.0023 0.0302 0.0393 6.72
PUERTO PRINCESA CITY 4,343 64,949 6.69 0.0055 -0.0328 0.1060 8.29
ROMBLON 4,311 64,351 6.70 0.0052 0.0567 0.0773 7.81

Page 11 of 14
Continuation…

95% Confidence
Total No. Prevalence Interval
Estimate
of Young of Young
Province/ of the Standard CV
Adults Adult
Highly Urbanized Cities Total No. Error (%)
based on Obesity
of Obese
2007 CPH (%)
Lower Upper
Limit Limit

REGION V
ALBAY 11,840 314,500 3.76 0.0040 0.0297 0.0456 10.74
CAMARINES NORTE 5,978 134,869 4.43 0.0035 0.0374 0.0512 7.94
CAMARINES SUR 9,250 439,104 2.11 0.0021 0.0169 0.0252 10.12
CATANDUANES 4,592 56,155 8.18 0.0065 0.0691 0.0944 7.91
MASBATE 6,318 177,671 3.56 0.0026 0.0304 0.0407 7.35
SORSOGON 4,923 175,057 2.81 0.0023 0.0235 0.0327 8.34
REGION VI
AKLAN 3,906 135,200 2.89 0.0029 0.0233 0.0345 9.94
ANTIQUE 4,524 128,615 3.52 0.0028 0.0296 0.0407 8.02
CAPIZ 4,922 179,546 2.74 0.0022 0.0231 0.0318 8.07
GUIMARAS 3,629 44,614 8.13 0.0073 0.0671 0.0956 8.93
ILOILO 17,004 472,493 3.60 0.0044 0.0273 0.0446 12.25
ILOILO CITY 5,885 137,771 4.27 0.0031 -0.0120 0.1264 7.21
NEGROS OCCIDENTAL 22,927 638,601 3.59 0.0023 0.0313 0.0405 6.50
BACOLOD CITY 6,824 158,928 4.29 0.0034 -0.0205 0.1226 8.02
REGION VII
BOHOL 8,956 320,336 2.80 0.0029 0.0222 0.0337 10.54
CEBU 43,470 696,716 6.24 0.0111 0.0406 0.0842 17.79
CEBU CITY 7,933 275,776 2.88 0.0022 0.0061 0.0568 7.48
LAPU-LAPU CITY 4,217 105,590 3.99 0.0036 0.0025 0.0581 9.02
MANDAUE CITY 4,946 116,383 4.25 0.0034 -0.0058 0.1165 8.11
NEGROS ORIENTAL 7,295 329,292 2.22 0.0015 0.0192 0.0251 6.76
SIQUIJOR 3,634 6,474 56.13 0.0504 0.4625 0.6602 8.99
REGION VIII
BILIRAN 3,803 35,965 10.57 0.0090 0.0881 0.1234 8.53
EASTERN SAMAR 5,389 100,601 5.36 0.0042 0.0453 0.0619 7.91
LEYTE 15,576 454,650 3.43 0.0044 0.0256 0.0429 12.94
TACLOBAN CITY 4,205 65,298 6.44 0.0055 -0.0085 0.0260 8.51
NORTHERN SAMAR 6,058 134,462 4.51 0.0033 0.0385 0.0516 7.40
SAMAR 7,070 174,077 4.06 0.0030 0.0346 0.0466 7.51
SOUTHERN LEYTE 4,898 99,057 4.94 0.0040 0.0416 0.0573 8.06

Continuation…

Page 12 of 14
95% Confidence
Total No. of Prevalence Interval
Estimate
Young of Young
Province/ of the Standard CV
Adults Adult
Highly Urbanized Cities Total No. Error (%)
based on Obesity
of Obese
2007 CPH (%)
Lower Upper
Limit Limit

REGION IX
ZAMBOANGA DEL NORTE 5,722 249,606 2.29 0.0022 0.0185 0.0273 9.81
ZAMBOANGA DEL SUR 6,262 248,907 2.52 0.0019 0.0215 0.0289 7.52
ZAMBOANGA CITY 4,440 247,335 1.80 0.0015 0.0038 0.0526 8.08
ZAMBOANGA SIBUGAY 10,752 151,534 7.10 0.0054 0.0603 0.0816 7.65
REGION X
BUKIDNON 10,262 333,546 3.08 0.0028 0.0253 0.0362 8.99
CAMIGUIN 3,679 20,895 17.61 0.0157 0.1452 0.2069 8.94
LANAO DEL NORTE 6,633 153,975 4.31 0.0027 0.0379 0.0483 6.18
ILIGAN CITY 5,031 45,740 11.00 0.0087 -0.0953 0.4433 7.94
MISAMIS OCCIDENTAL 3,682 146,358 2.52 0.0029 0.0195 0.0308 11.46
MISAMIS ORIENTAL 9,116 209,579 4.35 0.0029 0.0379 0.0491 6.58
CAGAYAN DE ORO CITY 6,033 182,241 3.31 0.0025 0.0151 0.1220 7.66
REGION XI
COMPOSTELA VALLEY 6,411 190,104 3.37 0.0027 0.0285 0.0389 7.89
DAVAO DEL NORTE 11,100 261,722 4.24 0.0054 0.0318 0.0530 12.78
DAVAO DEL SUR 20,073 240,373 8.35 0.0048 0.0741 0.0929 5.72
DAVAO CITY 13,241 454,212 2.92 0.0019 0.0459 0.1620 6.53
DAVAO ORIENTAL 4,195 131,342 3.19 0.0027 0.0267 0.0371 8.30
REGION XII
NORTH COTABATO 8,411 335,657 2.51 0.0020 0.0212 0.0289 7.83
SARANGANI 4,455 140,374 3.17 0.0025 0.0269 0.0366 7.84
SOUTH COTABATO 10,564 232,839 4.54 0.0027 0.0400 0.0507 6.01
GENERAL SANTOS CITY 5,739 175,596 3.27 0.0026 -0.0186 0.0887 7.83
SULTAN KUDARAT 4,907 207,947 2.36 0.0018 0.0201 0.0271 7.61
ARMM
BASILAN 3,835 121,288 3.16 0.0031 0.0255 0.0378 9.94
LANAO DEL SUR 5,898 318,469 1.85 0.0023 0.0140 0.0231 12.53
MAGUINDANAO 5,476 401,229 1.36 0.0016 0.0105 0.0168 11.68
SULU 5,819 258,885 2.25 0.0015 0.0195 0.0254 6.70
TAWI-TAWI 4,430 135,544 3.27 0.0029 0.0271 0.0383 8.73

Continuation…
Total No. Prevalence
Province/ Estimate Standard 95% Confidence CV
of Young of Young
Highly Urbanized Cities of the Error Interval (%)
Adults Adult

Page 13 of 14
Total No. based on Obesity
of Obese 2007 CPH (%)

Lower Upper
Limit Limit

REGION XIII
AGUSAN DEL NORTE 4,522 83,058 5.44 0.0049 0.0448 0.0641 9.05
BUTUAN CITY 4,106 85,762 4.79 0.0042 0.0047 0.0713 8.77
AGUSAN DEL SUR 5,409 166,163 3.26 0.0023 0.0280 0.0371 7.07
SURIGAO DEL NORTE 3,983 110,391 3.61 0.0035 0.0291 0.0430 9.81
SURIGAO DEL SUR 5,268 141,255 3.73 0.0027 0.0321 0.0425 7.11

Page 14 of 14

You might also like