Determinants of Children s Health Services Utilization in the Philippines

Transcription

1 Determinants of Children s Health Services Utilization in the Philippines by Amardeep Thind and Anne Marie Cruz Department of Health Services, UCLA School of Public Health, Los Angeles, USA Summary Respiratory illness and diarrhea are the two most important causes of death in children under the age of 5 in the Philippines. The government has accorded high priority to reducing the number of deaths caused by these illnesses as it attempts to reform its healthcare system. In order to re-design health systems, policy makers need to have a good understanding not only of overall health services utilization, but also of the use of public and private sectors. We used the 1998 Philippines National Demographic and Health Survey dataset to understand the service utilization patterns of children under 5 with diarrhea and/or respiratory illness. Using the Andersen Model as the conceptual framework, a nested logit regression approach was used to determine predictors of health services use, and of public vs. private use in this population. Our results indicate that maternal education and number of illnesses determine the decision to seek care. Once the decision has been made to seek care, the choice between a public or private provider is affected by the family s economic status and size of the household. Policy makers can use such information for future health planning and reform, in order to increase access to healthcare for this vulnerable population in the Philippines. Introduction The under-fives mortality rate in the Philippines is 35 per 1000 live births. 1 Respiratory illness and diarrhea are the two most important causes of deaths in this age group; it is estimated that nearly children under five die from diarrhea alone each year. 1,2 The Philippine government has accorded high priority to reducing deaths caused by these illnesses as it attempts to reform its healthcare system. In order to re-design health systems, policy makers should have a good understanding of the overall health services utilization and of the public and private sectors. An important question to ask is, who uses each system? In this paper, this question will be explored from the perspective of a vulnerable population, i.e. children under 5 years of age (underfives) with diarrhea and/or respiratory illness. In particular, predictors of use and non-use of healthcare services and use of public and private healthcare services will be assessed. This information will be critical when restructuring health systems in order to distribute scarce resources more efficiently. Correspondence: Amardeep Thind, MD, PhD., Assistant Professor, Department of Health Services, UCLA School of Public Health, Room CHS, Box Los Angeles, CA USA. Tel ; Fax <amardeep@ucla.edu>. Research Methods Research questions 1. What are the predictors of health services utilization for under-fives with respiratory illness and/or diarrhea? 2. Among the users, what are the predictors of public vs. private sector utilization for underfives who suffer from diarrhea and/or respiratory illness? Data source The data for this study were obtained from the 1998 Philippines National Demographic and Health Survey (NDHS). This was a national survey of women aged years in households. 3 Our study population was under-fives who had diarrhea and/or respiratory illness in the 2 weeks preceding the survey; there were 1501 such children in the survey. Conceptual model We used the Andersen Behavioral Model as a basis for understanding the determinants of health services utilization. This model has been validated by studies conducted both in the US and internationally. 4 9 The Model states that health services utilization is a function of three categories: (a) predisposing factors such as age, gender, marital status, educational status and health beliefs; (b) enabling factors such as income, health insurance and regular sources of care; Journal of Tropical Pediatrics, Vol. 49, No. 5 Oxford University Press 2003; all rights reserved 269

2 and (c) need, which is the most proximate cause of health services utilization. 4 Variable specification Dependent variables. There were two dependent variables in the study: (1) Users/non-users of health services: under-fives who had an episode of diarrhea and/or respiratory illness in the preceding 2 weeks and sought consultation from a provider (either public or private) were defined as users ; children not seeking care were categorized as non-users : (2) Public/private sector users: among users of health services, care received at a government hospital or subcenter, public dispensary or rural clinic was categorized as public ; care received from private clinics, dispensaries, hospitals or pharmacies was categorized as private. Independent variables. Among the predisposing variables, age of the child was measured in years, while sex was a dichotomous categorical variable. Maternal education was coded into three categories based on the highest level completed (primary, secondary, or higher than secondary). was a binary variable (Catholic and non-catholic). Household size was dichotomised into households with less than six members and households with six or more members. The two enabling variables used were asset index quartile and location. As the NDHS survey did not have information on household income, an asset index was computed, based on the method by Filmer & Pritchett. 10 This composite measure is a sum of seven household possessions, i.e. electricity, radio, television, refrigerator, bicycle, motorcycle, and car, weighted using principal component analysis. Based on this weighted score, households were categorized into four quartiles. Location was dichotomized as urban/rural. Need was captured by noting the number of illnesses (children having either diarrhea or respiratory illness were coded as 1 and those having both diarrhea and respiratory illness were coded as 2 ). Data analysis Stata version 7.0 was used to carry out the data analysis. The unit of analysis was a child born in the previous 5 years ( ). Bivariate analyses were conducted and a multivariate nested logit model was used to ascertain relationships between the dependent and independent variables. Results Descriptive analyses Users vs. non-users of services. Out of the 1501 children who had respiratory/respiratory illness in the past 2 weeks, 897 (59.8 per cent) sought care (either from a public, private, or traditional TABLE 1 Characteristics of users vs. non-users Non-users Users (n = 604) (n = 897) (a) Continuous variables (mean ± SD) Age (years) 1.80 ± ± 1.32 (b) Categorical variables n (%) Sex* Male 301 (49.8) 496 (55.3) Female 303 (50.2) 401 (44.7) Maternal education** Primary 324 (53.6) 330 (36.8) Secondary 188 (33.1) 365 (40.7) Higher 92 (13.3) 202 (22.5) Catholic 468 (77.5) 699 (77.9) Non-Catholic 136 (22.5) 198 (22.1) Household Size Less than (55.8) 511 (57.0) 6 or more 267 (44.2) 386 (43.0) Location** Urban 171 (28.3) 328 (36.6) Rural 433 (71.7) 569 (63.4) Asset index quartile** (48.1) 333 (37.9) 2 62 (10.4) 70 (8.0) (23.0) 218 (24.8) (18.5) 257 (29.3) Number of illnesses** (88.3) 687 (76.6) 2 71 (11.7) 210 (23.4) ** p < 0.05; * p < 0.1 Note: numbers may not add to 100% due to rounding provider) (Table 1). Males were more likely to be users, and users mothers were more likely to have higher education and belong to a higher asset index quartile. Children living in urban locations were more likely to use services. Children having two illnesses were more likely to seek care compared with children having only one illness. Age, religion and household size were not statistically significant predictors of health services utilization. Public vs. private sector users. Among children taken to a provider for healthcare, slightly more (56 per cent) used the public sector (Table 2). Being a male, having a more highly educated mother, urban residence, and a higher asset index quartile were significantly associated with the use of private providers. However, having more than one illness is associated with a greater use of a public provider. Age, religion and household size were not significant determinants of public vs. private sector use. 270 Journal of Tropical Pediatrics Vol. 49, No. 5

3 TABLE 2 Characteristics of private vs. public sector users Public Private (n = 466) (n = 373) (a) Continuous variables (mean ± SD) Age (years) 1.57 ± ± 1.36 (b) Categorical variables n (%) Sex* Male 248 (53.2) 220 (59.0) Female 218 (46.8) 153 (41.0) Maternal education** Primary 208 (44.6) 98 (26.3) Secondary 205 (44.0) 139 (37.3) Higher 53 (11.4) 136 (36.4) Catholic 362 (77.7) 290 (77.8) Non-Catholic 104 (22.3) 83 (22.2) Household Size Less than (55.4) 223 (59.8) 6 or more 208 (44.6) 150 (40.2) Location** Urban 147 (31.4) 161 (43.2) Rural 319 (68.5) 212 (56.8) Asset index quartile** (52.1) 73 (20.1) 2 42 (9.2) 23 (6.3) (27.0) 78 (21.5) 4 54 (11.1) 189 (52.1) Number of illnesses** (75.1) 302 (81.0) (24.9) 71 (19.0) ** p < 0.05; *p < 0.1. Note: numbers may not add to 100% due to rounding. Variable TABLE 3 Nested logit model results Odds ratio Use vs. non-use Public vs. private use (a) Predisposing factors Age Sex Male Female Catholic Non-Catholic Maternal education Primary Secondary 2.03** 0.73 Higher 1.50** 1.37 Household size Less than 6 6 or more ** (b) Enabling factors Asset index quartile ** ** ** Location Urban Rural (c) Need Number of illnesses ** 0.83 Inclusive value 1.69 Multivariate analyses A two-stage nested logit model was used to determine predictors of health care utilization (Table 3). Statistically significant predictors of the first stage (use/non-use of services) were maternal education and number of illnesses. Compared with mothers with primary education or lower, children of mothers with secondary education had twice the odds of using health services, and children of mothers having higher than secondary education had 1.5 times the odds of using health services. Children with two illnesses had more than twice the odds of seeking care compared with children with one illness. Age, sex, religion, household size, location and asset index quartile were not found to be significant determinants of the decision to seek care. Once the decision was made to seek care, the choice between public and private provider is predicated on household size and the asset index quartile. Compared with children in the lowest asset index quartile, children in the second and third quartiles had twice the odds of using private providers; those ** p < in the highest quartile had 12 times greater odds of using a private provider. The odds for use of private providers by children belonging to households with six or more members were 32 per cent lower compared with children from smaller households. Age, sex of the child, maternal education, location, and severity of illness were not found to be significant determinants of the choice between public and private providers. Discussion Our results indicate that health services utilization for children is a two-step process, with the first step being the decision to use or not use services, followed by the users subsequently making a choice to use a public or private provider. Need (as measured by number of illnesses) and predisposing (maternal education) factors determine the first step. Journal of Tropical Pediatrics Vol. 49, No

4 The second step is predicated on predisposing (household size) and enabling factors (asset index quartile). Maternal education is a predisposing factor that has been widely researched as a predictor of health services utilization, and our findings are consistent with the literature Research from the Philippines has shown that a low maternal education level is correlated with low prenatal care utilization. 14 Need is the proximate determinant of health services utilization, and as measured by the number of illness, it was a significant determinant of the decision to use health services. A possible explanation is that children having both diarrhea and respiratory illness were thought to be sicker or more severely ill by the mothers, thus acting as a stronger cue for action. Need, as conceptualized as severity, has been shown to be a significant determinant of health services utilization. 15 However, earlier work from the Philippines has not corroborated this link; this might be due to the inappropriate measures used to ascertain severity in that study. 16 Once the decision to use health services is made, the choice between which provider to use is determined by economic factors (a higher asset index quartile leads to greater odds of using private services) and the household size. This is consonant with previous work in the Philippines; for example, Wong found that increases in assets owned led to increased private sector visits and a reduction in the number of public visits; 14 Russo found that households in the highest income quartile were approximately twice as likely to utilize private hospital services vs. households in the lowest income quartile. 17 A larger household size is associated with less use of the private providers. Similar results have been reported previously from the Philippines. For example, Ching found that having more members in a family had a negative effect on the probability of utilizing government or private facilities. 16 He postulated that having more family members may mean more time and individuals to care for sick members at home, thus reducing the need to seek healthcare services in private or public institutions. Another possible explanation is that a larger family size effectively reduces the per head income of the family. Policy implications. The Philippines is a povertystricken country with a democratic government. Thirty-seven per cent of its population is below the poverty level. 18 Healthcare services are provided by a mixture of public and private institutions. 17 The government has attempted to close the gap in access to health services between the rich and the poor by a variety of measures. In 1991, the government began the decentralization of governance through the enactment of Republic Act No. 7160, known as the Local Government Code. 19 One of the main aims of decentralization is for local health boards, also known as local governing units (LGUs), to delegate decision-making of health services delivery to the local level. In 1995, the National Health Insurance Act, one of the first compulsory health insurance systems among developing countries, was passed. 20 As the country recognizes the importance of increasing community participation through devolution and providing more access to quality healthcare especially to the indigent, knowing the determinants of use/non-use of services and of private public service utilization becomes crucial to successful implementation of policies. In the 2 weeks preceding the survey, approximately 40.2 per cent of children with diarrhea or respiratory infection did not seek healthcare from a healthcare provider. While this is a relatively high proportion, our study indicates that the children with the greatest need (i.e. children with both diarrhea and respiratory illness) were more likely to be taken to a provider. Our study indicates that one way the LGUs could increase health services utilization is by strengthening the educational system and improving women s access to education. A mother with more education may have greater decision-making power in the household and understand the importance of early treatment and prevention, leading to an increased likelihood of utilizing healthcare services. Once the decision to use services is made, the choice between using a public or a private provider is predominantly dependent on the family s asset index quartile. Given the unequal distribution of wealth in the country, this raises issues regarding equitable access to care. In the Philippines, payments for healthcare are still predominantly out-ofpocket. 21 Poor people are more price-sensitive than the rich and would generally opt for less expensive (i.e. public) providers. The Philippine government needs to ensure that the poor are provided the necessary wherewithal under the National Health Insurance Act to ensure equitable access to health providers. It is important to understand the limitations of our study. The NDHS data rely on self-reported answers, and are thus subject to recall and reporting bias. Our predictive model does not control for health beliefs and quality of care, which may play a role in determining health services utilization. Our categorizations of the public and private sectors are not homogenous and include a diverse array of providers. Future research needs to improve upon these limitations. In conclusion, our study indicates that for underfives having diarrhea and/or respiratory illness, maternal education and number of illnesses determine the decision to seek care. Once the decision has been made to seek care, the choice between a public or private provider is affected by the family s economic status and the size of the household. Policy makers can use such information for future health 272 Journal of Tropical Pediatrics Vol. 49, No. 5

5 planning and reform, in order to increase access to healthcare for this vulnerable population in the Philippines. References 1. Department of Health Manila. Ten Leading Causes of Child Mortality by Age Group & Sex. data_stat/html/child_10htm. (accessed 27 August 2002). 2. Easton A. Philippine plan to cut diarrhea deaths. BMJ 1999; 319: National Statistics Office, Department of Health (Philippines). National Demographic and Health Survey 1998: Manila. Calverton, MD, National Statistics Office, Macro International, Andersen R, Rice TH, Kominski GF. Changing the US Health Care System: Key Issues in Health Services, Policy, and Management, 2nd edn. San Francisco, Jossey-Bass, Berkanovic E, Hurwicz ML, Batchlor E. Rheumatoid arthritis and the use of medical services. Arthritis Care Res 1991; 4: Fosu GB. Access to health care in urban areas of developing societies. J Hlth Soc Behav 1989; 30: Glik DC, Ward WB, Gordon A, Haba F. Malaria treatment practices among mothers in Guinea. J Hlth Soc Behav 1989; 30: Shah NM, Shah MA, Behbehani J. Predictors of non-urgent utilization of hospital emergency services in Kuwait. Soc Sci Med 1996; 42: Subedi J. Modern health services and health care behavior: a survey in Kathmandu, Nepal. J Hlth Soc Behav 1989; 30: Filmer D, Pritchett LH. Estimating wealth effects without expenditure data or tears: an application to educational enrollments in states of India. Demography 2001; 38: Becker S, Peters DH, Gray RH, Gultiano C, Black RE. The determinants of use of maternal and child health services in Metro Cebu, the Philippines. Hlth Transit Rev 1993; 3: Pebley AR, Goldman N, Rodriguez G. Prenatal and delivery care and childhood immunization in Guatemala: do family and community matter? Demography 1996; 33: Cleland JG, Van Ginneken JK. Maternal education and child survival in developing countries: the search for pathways of influence. Soc Sci Med 1988; 27: Wong EL, Popkin BM, Guilkey DK, Akin JS. Accessibility, quality of care and prenatal care use in the Philippines. Soc Sci Med 1987; 24: Thind AS, Andersen RA. Respiratory illness in the Dominican Republic: What are the predictors for health services utilization of young children? Soc Sci Med 2003; 56: Ching P. User fees, demand for children s health care and access across income groups: the Philippine case. Soc Sci Med 1995; 41: Russo G, Herrin AN, Pons MC. Household health care facility utilization in the Philippines. Asia Pac J Public Health 1996; 9: World Bank. Philippines at a glance. worldbank.org/eap/eap.nsf (accessed 27 August 2002). 19. Ramiro LS, Castillo FA, Tan-Torres T, et al. Community participation in local health boards in a decentralized setting: cases from the Philippines. Health Policy Plan 2001; 16 (Suppl 2): Bossert TJ, Beauvais JC. Decentralization of health systems in Ghana, Zambia, Uganda and the Philippines: a comparative analysis of decision space. Hlth Policy Plan 2002; 17: World Bank. Economic Update Philippines Development Policy Review: An opportunity for renewed poverty reduction report. (accessed 28 August 2002). Journal of Tropical Pediatrics Vol. 49, No