Health Policy and Planning Advance Access originally published online on August 29, 2008
Health Policy and Planning 2008 23(6):443-451; doi:10.1093/heapol/czn032
Modelling prenatal health care utilization in Tajikistan using a two-stage approach: implications for policy and research
1School of Social Work, University of Windsor, Windsor, Ontario, Canada.
2School of Social Work, Lakehead University, Thunder Bay, Ontario, Canada. E-mail: lfan{at}ucalgary.ca
* Corresponding author. School of Social Work, University of Windsor, Room 2164 Chrysler Hall North, 401 Sunset Avenue, Windsor, Ontario, N9B 3P4, Canada. E-mail: nnh{at}uwindsor.ca
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionConclusion and implicationsEndnotesReferences |
|---|
Since the transition from a centrally planned to a market economy, Tajikistan has witnessed a high rate of child and maternal mortality, a decline in the birth rate and a significant drop in public expenditures on health care. Against this backdrop, this paper analyses the determinants of prenatal health care utilization using Andersen's behavioural model, which has been modified to the context of Tajikistan. We applied a two-stage sequential model to data drawn from a nationally representative survey. Binary logit regression is used to predict and explain the probability of using prenatal health care services, while negative binomial regression is used to predict and explain the frequency of using these services. Findings suggest that higher educational attainment increases the utilization of prenatal care. Conversely, poverty, limited knowledge about matters related to sex, low quality of health care service, lack of public infrastructure, as well as absence of or long distance of travel to the nearest health facility, all reduce the utilization of prenatal health care. Health policy and research implications are presented and discussed.
Key Words: Two-part model, reproductive health, maternal care, family planning, Central Asia, Tajikistan
KEY MESSAGES
|
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusion and implicationsEndnotesReferences |
|---|
Several recent studies have raised concerns about the accessibility of health services in the countries of Central Asia and the Caucasus (Sari and Langenbrunner 2001
; Cashin et al. 2002; Balabanova et al. 2004; Gotsadze et al. 2005). In the past, the countries of this region enjoyed universal free-of-charge health care provision for their entire populations in the framework of the Soviet model of health care delivery. The dramatic economic transition from a centrally planned to a market economy, the disintegration of economic and social co-operation between newly independent countries, and ethnic and civil unrest all led to a prolonged and profound economic recession that has negatively impacted the health care systems in these countries (Bonilla-Chacin et al. 2005). During this period, government expenditures on health care dropped sharply as a share of total government expenditures and as a share of Gross Domestic Product (GDP). Moreover, total government expenditures and GDP also declined considerably. In contrast, the proportion of health care expenditures for private, mostly informal out-of-pocket payments rose significantly. For instance, the average proportion of all patients making informal payments in the region reached 60% (Lewis 2001).
Against this background, we focus on Tajikistan, one of the low-income countries in Central Asia. Even before independence from the former Soviet Union, Tajikistan had one of the lowest levels of economic output and highest poverty rates among all the Soviet Republics (Falkingham 2004). The independence achieved in 1991 from the Soviet Union led to disintegration of economic cooperation with other republics of the former Union, withdrawal of subsidies from Moscow, and prolonged a civil war that ended with the signing of a peace accord only in 1997. These factors have made transitional processes in Tajikistan more prolonged and laborious than in other republics. For instance, the GDP in US dollars per capita plummeted sharply from 501 in 1989 to 139 in 1996 (UNICEF 2007). Even in 2005, this indicator did not reach even half of the pre-transitional level. As a result, the ability of the state to maintain a universal health care system, inherited from the country's socialist past, was severely undermined. In addition, the brutal civil war caused great damage to the country's infrastructure including health facilities.
In this context, we concentrate in this paper on investigating the determinants of prenatal health care utilization in Tajikistan, for three major reasons. First, poor pregnancy outcomes have been high in the country during transition. Official statistics show an increase in child mortality from 43 per 1000 live births in 1989 to 48 in 1997, followed by a decline to 43 in 2003 (UNICEF 2007). However, these figures come from vital registration data which are widely acknowledged to be incomplete and the resulting estimates too low due to under-reporting. Aleshina and Redmond (2003) provide alternative estimations of child mortality based on several surveys conducted in Tajikistan in various years. According to the authors, child mortality in Tajikistan fluctuated from 89 per 1000 births in 1993 to 79 per 1000 births for the years 1995–99, although no clear trend over time can be identified because of the differences between the surveys. Second, the high level of child mortality was accompanied by a considerable decline in the birth rate (Falkingham 2000). In fact, Tajikistan experienced the steepest drop in birth rates among all the countries of Central Asia. In 1990, children under five constituted 18% of Tajikistan's population. By 1998, this proportion dropped to 13.5%. Finally, maternal mortality is also high in Tajikistan, at about 100 per 100 000 births in 2000 (WHO 2006).
In light of the poor birth outcomes and declining birth rates in Tajikistan, utilization of prenatal services has become an important issue that requires careful investigation. Prenatal health care is instrumental in achieving positive birth outcomes, since a direct positive relationship between utilization of prenatal services and low child mortality is well documented (Ivanov and Flynn 1999).1 Hence, the objective of this study is to explain and predict the use of prenatal care in the country. We seek to provide policy makers, health planners and administrators in Tajikistan with up-to-date information so that they can develop effective prenatal health care services. The following research questions are asked in this study:
- What are the characteristics that explain and predict the probability of using prenatal health care services?
- What are the characteristics that explain the frequency of use of prenatal health care services?
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusion and implicationsEndnotesReferences |
|---|
Data set
The data in this study are taken from the Tajikistan Living Standards Survey of 2003 (henceforth the TLSS). The TLSS is a cross-sectional, multi-topic, nationally representative survey that was designed and conducted with the technical assistance of the World Bank to measure living standards in Tajikistan (World Bank 2003). It covers 4156 households. The sampling frame of the TLSS is based on the recent 1999 national census. The survey employs a two-stage sample, stratified by regions and rural/urban settlements with a probability proportionate to its size, and includes 208 population points as primary sampling units.
Several advantages of using the TLSS should be highlighted. The TLSS contains detailed information about demographics, education, income, consumption, population point data and other information pertinent to the assessment of living standards. It permits an exploration of levels of material poverty for households, their demographics and educational attainments, as well as the specific socio-economic situation for each population point. More important for our study, the survey contains a separate questionnaire that was administered to all female members of households between the ages of 15 and 49 years old. This questionnaire collected information about the birth history of each woman, their utilization of prenatal health care services and their knowledge about topics related to sexual matters. The above-mentioned features make the TLSS a rich source of information that permits us to combine individual, household and community data to allow us to perform an in-depth analysis of the determinants of prenatal health care utilization.
However, the TLSS has two important limitations. First, it does not collect data about the perceived quality of prenatal care. Second, it does not collect information about postnatal health care utilization. These limitations prevent us from analysing two vital dimensions of maternal care.
Since the TLSS used a complex multi-stage, stratified, clustered sampling design, there is unequal selection of probabilities for the surveyed households and heteroscedasticity of standard error by primary sampling units (Lee and Forthofer 2006). In order to compute unbiased estimations from the TLSS, we used the svy family of commands available through the STATA software package that was especially designed for this complex survey design (STATA 2005). In particular, the Taylor linearization procedure was employed to adjust for a complex survey design and to achieve accurate, nationally representative results for all the estimations.
Conceptual framework
The Andersen behavioural model provides a conceptual framework for this study (Andersen and Newman 1973; Andersen 1995). First introduced in 1968, this model has been extensively tested to study the factors predicting health care utilization (Thind and Cruz 2003). The main assumption of the model is that some individuals have a higher propensity for health service utilization than others. The propensity to use the services varies according to one's personal, household and socio-economic characteristics, which can be classified into three broad groups: predisposing, enabling and need.
Predisposing characteristics are not the direct reasons for an individual to use health care services but reflect the probability of certain individuals seeking health care. For instance, having a higher level of educational attainment does not cause individuals to be sick more than others. However, individuals with higher educational attainment can have a higher predisposition to utilizing health care compared with individuals with lower educational attainment. By contrast, enabling characteristics reflect the resources available that influence health care utilization. According to Andersen and Newman (1973, p. 109), an enabling factor can be defined as ‘a condition which permits a family to act on a value or satisfy a need regarding health service use’. For example, having a higher household income can enable household members to use health care more frequently. Finally, the need for health care is the last group of characteristics, and it refers to perceived health status. Both perceived illness and the probability of its occurrence will influence an individuals’ decision to seek health care.
Andersen's model has been commonly accepted and extensively used since it was developed. However, it was developed specifically for use within the context of the US, a developed country. Hence, we modified the model to make it applicable to the context of Tajikistan, a country in many ways very different from the US.
We commence with predisposing characteristics. In this study, we consider two characteristics predisposing women to use prenatal health care: formal educational attainments and knowledge about topics related to sexual matters. In transitional countries, higher levels of educational attainment are positively correlated to the utilization of health care services (Ivanov and Flynn 1999). Therefore, we expect that in Tajikistan, higher educational attainment will increase prenatal care utilization. In addition to formal education, another important factor is knowledge about pregnancy (Ivanov 2000). Having a better understanding of the processes related to pregnancy increases the likelihood of utilizing prenatal health care services. The TLSS did not ask any direct questions regarding knowledge about pregnancy. However, it did ask women about their major sources of knowledge about topics related to sexual matters. Possible answers included family members, relatives, friends, co-workers, health care personnel or the mass-media. Women are most likely to have limited knowledge about pregnancy if their primary information about topics related to sex comes mainly from family members. We hypothesize that limited knowledge will reduce utilization of prenatal health care services.
Turning to enabling characteristics, we begin with another factor, specific year of childbirth. As was outlined in the preceding sections of this paper, prenatal health care in independent Tajikistan has deteriorated significantly when compared with the Soviet period. Thus, we hypothesize the effect of year of childbirth on health care utilization to be negative. The closer the year of childbirth is to the Soviet period, the higher will be the likelihood of prenatal health care utilization. This result may be affected by recall measurement error linked with the long period since the birth of the last child. However, previous studies show that recall bias cannot account for expected significant year effect (Falkingham 2003).
The next enabling factor to consider is the economic status of the household. The economic transition led to a decrease in government spending on health care, which includes spending on salaries for medical personnel, who are known to have one of the lowest paid occupations in the country (European Observatory on Health Care Systems 2000). Furthermore, salaries within the public sector, which includes health care, are consistently paid in arrears, with it not being uncommon for salaries to be paid several months late. Under these circumstances, unofficial fees for health services become an important source of income for health sector personnel. Despite the fact that prenatal services are officially free-of-charge in Tajikistan, evidence suggests that widespread unofficial out-of-pocket fees are given in the form of cash or in-kind exchanges (Falkingham 2004). The burden of informal payments has become a major deterrent for women from poor households in seeking health care. As a result, in 1999, nearly half of the women who did not seek prenatal care reported that they could not afford it. Taking these issues into consideration, we expect that the poverty status of a household will have a negative effect on the utilization of prenatal health care.
The presence of health facilities is an important predictor of health care utilization in both developed and developing countries (Khan et al. 2005). Tajikistan, like other former republics of the Soviet Union, inherited a wide network of health care facilities from its Soviet past (Walters and Suhrcke 2005). This network had a relatively egalitarian distribution to ensure equal access to health care for the entire population. However, during the economic crisis caused by transition, many of these facilities were closed down. In addition, during the civil war and subsequent periods of civil unrest, a significant share of the country's health infrastructure was damaged or destroyed. The quality of service offered in other facilities also deteriorated due to lack of medical personnel or as a result of limited or nonexistent access to utilities such as electricity, water and sanitation. Thus, we investigate the effects of the availability of health care facilities on health care utilization. More specifically, we expect that physical absence of health care facilities and a deteriorated quality of service will reduce health care utilization. Likewise, we expect that a longer distance or longer travel time to reach those facilities will also reduce utilization.
Finally, the literature (Gwatkin et al. 2000) has highlighted the importance of broader enabling resources within the community in the form of public infrastructure such as water and electricity supplies and sewage systems. Consequently, we examine the association between the availability of public infrastructure and health care utilization. Our expectation is that a better infrastructure is associated with higher levels of utilization.
Empirical method
The research questions discussed earlier consider the utilization of prenatal health care services as a two-stage process where decisions are sequential (see Figure 1). In the first stage, the decision is made by women regarding whether or not to use prenatal health care services. In the second stage, another decision is made regarding the frequency of the usage. For both of the stages, the decision is affected by the factors described in the conceptual framework. However, the effect of these factors on decisions can be different at different stages. Hence, we apply a two-stage model to investigate the determinants of utilization of prenatal health care services.
|
In the first stage, we predict the probability of women utilizing prenatal medical services. Our dependent variable is binary, and therefore the probability is modelled using a binary logit regression. In the second stage, we predict the frequency of prenatal health care utilization. Our dependent variable here is a count variable which reflects the number of times that women utilized health services. Although the Poisson model is generally considered appropriate for count dependent variables, it often suffers from overdispersion (Long and Freese 2006
). In our case, data do show the problem of overdispersion.2 Therefore, in the second stage, instead of Poisson regression, we use an alternative method, a negative binomial regression, to estimate the frequency of women utilizing prenatal health care.
It must be noted that two-stage models with count variables are widely used in the analysis of health care utilization. For instance, Pohlmeier and Ulrich (1995) used a two-stage model with negative binomial regression to analyse ambulatory service demand in Germany. Similarly, Deb and Trivedi (1997) evaluated health care utilization by seniors in the US by employing a two-stage model with negative binomial regression. Most recently, Lahiri and Xing (2003, 2004) investigated veterans’ health care utilization in the US by estimating the two-stage models with Poisson and negative binomial regressions.
Variables and descriptive statistics
Since this study uses a two-stage model, we employ two distinctive dependent variables. For the first stage, the dependent variable is whether a woman attended a prenatal consultation when she was pregnant with her last child. This variable is a dummy variable and takes a value of 1 if a woman did attend. Conversely, it takes a value of 0 if a woman did not attend. Distribution of this variable shows that approximately 85% of women had a prenatal consultation, while a significant share, approximately 15%, did not have any prenatal consultation (see Figure 1).
For the second stage, the dependent variable is a count variable indicating the frequency of utilization—the number of prenatal consultations attended. This variable is continuous and varies from 1 to 9 in the survey. The distribution of this variable demonstrates that the highest proportion of women had only 2–3 consultations during the whole period of pregnancy, while a considerable share, approximately 11%, did not have even one prenatal consultation (see Figure 1).
The definitions of variables and descriptive statistics are shown in Table 1. Note that not all of the independent variables are included in both regression models due to collinearity. Initially, we began our analysis with a larger number of independent variables widely known to affect prenatal health care utilization, such as parity, woman's age at first birth, woman's age at marriage, time elapsed between births, rural/urban and regional variables. However, we found strong collinearity between these variables and other variables in the models.3 In fact, the collinearity was so strong that it led to a change in the sign of a regression coefficient to the opposite direction and we had to drop these highly collinear variables. Thus, the independent variables that were dropped due to collinearity are not listed in Table 1 or in the subsequent tables.
|
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusion and implicationsEndnotesReferences |
|---|
First stage: probability of prenatal care utilization
Results of binary logit estimations for probability of prenatal care utilization are shown in Table 2. Overall, the model demonstrates a good fit as suggested by the F statistics. The signs of coefficients for all independent variables in the model are also as expected. Six independent variables are strongly significant at the P < 0.001 level and two independent variables are significant at the P < 0.05 level.
|
The results indicate that the year of last birth has a negative impact on the likelihood of using prenatal care, which confirms that a deterioration of access to prenatal services happened during the transition. As predicted, the closer to the Soviet period is the year of childbirth, the higher is the likelihood of prenatal health care utilization. This trend is clearly illustrated by Figure 2 which demonstrates the proportion of women not attending a prenatal consultation by year. As we can see from the Figure, the proportion of women not utilizing prenatal care steadily decreased from 1988 to 1991, before the Soviet Union collapsed and transition began. This percentage then increased dramatically in 1992, when the Soviet Union collapsed and civil war began in Tajikistan. After 1992, the percentage of women not attending any prenatal consultation has fluctuated from year to year, reaching the highest level in 2002. The percentage has never returned to the level of pre-transitional years.4
|
Education is another strong predictor of prenatal health care utilization. An increase in years of schooling improves the likelihood of utilizing prenatal care. Likewise, women who have limited knowledge about matters related to sex are less likely to utilize prenatal care than women with better knowledge about sex-related matters. Poverty appears to be a major barrier for prenatal care utilization. Likelihood of care utilization is lower for women living in poor households compared with women from wealthier households. Four population point characteristics are significant predictors of prenatal health care utilization. The absence of a polyclinic at a population point, and living a longer distance from a health care facility reduce the likelihood of utilizing health care. Similarly, low quality of health services and limited access to the water pipe grid at the population point are associated with a reduced propensity to utilize prenatal health care services.
Finally, the variables not significant in a strict statistical sense also have the predicted signs. Thus, deteriorating health care services and limited access to electricity at a population point are associated with decreased chances of utilizing prenatal health care services.
Second stage: frequency of prenatal care utilization
Results of the negative binomial model for the frequency of prenatal care utilization are shown in Table 3. Again, overall, the model fits well, as suggested by the F statistics. The coefficient signs for all independent variables are also as expected. Three independent variables are strongly significant at the P < 0.001 level and two others are significant at P < 0.05.
|
The results show that giving birth in the years close to the year of the survey reduces the frequency of health care utilization. In other words, the closer to the Soviet period is the year of childbirth, the higher is the frequency of prenatal health care utilization. Higher educational attainments also increase the frequency of prenatal health care usage. The last significant personal characteristic, having a limited knowledge about matters related to sex, reduces the frequency of using prenatal health care. Two population point characteristics are significant. Living at a population point without a polyclinic, and living at a greater distance from a health facility, both reduce the frequency of utilization.
Again, the variables not significant in a strict statistical sense also have the predicted signs. Being poor and living in a population point without a centralized sewage system both reduce the frequency of prenatal health care utilization.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusion and implicationsEndnotesReferences |
|---|
There are several interesting findings which should be highlighted. First, the combination of economic crisis, political instability and civil war has had a strong negative impact on the utilization of prenatal health care services. Both models exhibit a strong negative effect of year of last birth on prenatal health care utilization; the likelihood of using health care services declined for births later in the transition in comparison with births during the Soviet period. This finding is in line with the findings of previous reports in Tajikistan based on TLSS 1999 (Falkingham 2003
), but shows no positive changes from 1999 to 2003, indicating a continuous deterioration in access to prenatal health care. The direct implication of this finding is that there has been no or very little progress in health care reform in Tajikistan during the period in question. Second, quality of health care appears to be an important determinant of the usage of prenatal health services. A perception of a low quality of health care is associated with a low propensity of using services. However, it is interesting to note that this variable is not significant for the frequency of health care utilization. This demonstrates that the decision regarding whether or not to use health services and the decision regarding the frequency of use may be based on a sequential decision-making process. This renders the two-stage model pertinent to an analysis of the utilization of prenatal services. In the case of our study, the first decision of whether or not to use the services is affected by the perception of health care quality. However, after the decision is made to use the services, this may implicitly imply a satisfactory level of service quality. Hence, perceived quality does not affect the second stage decision about the frequency of using the services.
Third, the same sequential mechanism appears to explain the effect of poverty. On the one hand, living in a poor household is associated with a decrease in the probability of using prenatal health care. Formally, maternal health care remains free in Tajikistan but our findings indicate a wide gap in utilization between the poor and wealthy. This finding is in line with previous findings that point to a growing burden of informal payments and in-kind gifts from patients to health care personnel (Falkingham 2004). The rising costs of informal payments serve as a significant deterrent to health care utilization. However, in contrast with using the service or not, poverty does not show a significant effect regarding the frequency of utilization. This implies that after a decision has been made to use health care in the first stage, poverty becomes less significant in determining the frequency of visits during the second stage.
Fourth, women's education and their knowledge about sexual matters are significant in determining both the probability and the frequency of health care use. Having a higher educational attainment and having knowledge about sexual matters that has been gained outside of the immediate family are both strongly associated with decisions to use health care, and also the subsequent frequency of its use. Unfortunately, a significant gender gap exists in Tajikistan's educational system (World Bank 2000). Girls’ enrolment lags behind that of boys at primary level and the gender gap is even wider at higher levels of education, where male students far outnumber female students.
Fifth, variables regarding health care infrastructure also demonstrate a significant effect on health care utilization. Absence of health facilities at a population point negatively affects the likelihood of using prenatal health services, while longer time in reaching a health facility decreases the frequency of service use.
Finally, a lack of public infrastructure also decreases utilization and its frequency in both models. Having no centralized water grid and limited access to electricity reduces the probability of health care utilization, while having no centralized sewage system reduces the frequency of utilization. These findings are not surprising. The lack of basic services forces women to spend more time at home taking care of their family's needs. For example, in Azerbaijan, another transitional country, the corrosion of services such as water pipes and heating forces women, particularly in rural areas, to spend more time obtaining, purchasing, transporting and preparing water for drinking and wood for heating (ADB 2005).
|
Conclusion and implications |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusion and implicationsEndnotesReferences |
|---|
Based on the 2003 Tajikistan Living Standards Survey, we analysed the determinants of women's prenatal health care utilization. The analysis was conceptualized using Andersen's behavioural model which was adjusted to the conditions of Tajikistan as a low-income transitional country. A two-stage statistical model was used to separately assess the probability of utilization and the frequency of utilization. Empirical evidence presented in this paper points to several health policy and research implications.
Policy implications
Since the transition from a centrally planned to a market economy, Tajikistan has witnessed a high rate of child and maternal mortality, a decline in the birth rate and a significant drop in public expenditures on health care. In response, the Government of Tajikistan has taken serious measures to improve reproductive health within the framework of a broad poverty reduction strategy (Government of Tajikistan 2005). The government has already adopted a ten-year strategic plan for reproductive health services and has taken several practical measures to put a number of laws, regulations and presidential degrees into practice regarding reproductive health. For example, the national Primary Health Centre (PHC) training centre has been inaugurated in the capital, Dushanbe, and training for trainers was provided to doctors and nurses for the regional-level PHC training centres. In addition, the Ministry of Health has conducted a series of information events on gender issues, engaging the mass media and educational institutions.
Nevertheless, the situation with regard to reproductive health does not seem to show signs of improvement. Empirical evidence presented in this paper demonstrates that the situation with regard to prenatal health care utilization continues to deteriorate in terms of both the probability and the frequency of service use. Our findings suggest that urgent measures are needed to reverse this negative trend.
The major challenge to improving prenatal health care utilization in Tajikistan is chronic under-funding in the health sector. On average for the period between 1999 and 2004, the public health budget of Tajikistan constituted approximately only 1% of GDP (Government of Tajikistan 2005). This low level of public spending has led to a low quality of services and a high rate of private out-of-pocket expenditures which affect the poor in particular. International experience shows that one of the ways to improve health service accessibility is to develop a basic package of benefits which are made available for free to eligible groups. A successful example of this is JPS-BK, an Indonesian safety-net health sector programme introduced in 1997 following the Indonesian economic crisis (Suci 2006).
In addition, the findings presented show that the determinants of health care utilization may lie beyond the direct domain of health care. Increasing the levels of educational attainment for women and improving the public infrastructure are primary goals of broader socio-economic policies. However, as shown in this paper, in the context of transitional countries, the education of women and the development of a general public infrastructure play a considerable role in determining health care utilization. Therefore, progress towards a broader improvement in these fields will have a great impact on health care utilization.
Research implications
Findings from this study highlight the importance of adjusting the Andersen model of health care utilization to conditions in transitional countries. This study demonstrates the importance of looking beyond the individual, household and health care characteristics stressed in the original model to a broader set of population point/community characteristics in explaining health care utilization. In our particular study on Tajikistan, we identified access to public infrastructure, namely, the lack of access to the water pipe grid within a population point, as a significant determinant of prenatal health care utilization. Thus, researchers should be encouraged to explore the community determinants of health care utilization that are pertinent in their countries.
Methodologically, our findings confirm the findings of other studies which suggest that health care utilization may have a distinctive characteristic (Sari and Langenbrunner 2001). The decision to utilize health care and the further decision of how frequently to utilize it can be made separately and sequentially at two different stages. Different sets of explanatory factors may affect the decision-making process at each stage. This makes the two-stage model pertinent in explaining and predicting prenatal health care utilization.
Finally, the findings presented in this paper demonstrate, through the example of the TLSS, that a cross-sectional multi-topic survey can be used to provide important information about health care utilization. Surveys similar to the TLSS are regularly conducted in transitional countries of Central Asia and the Caucasus in an attempt to monitor poverty reduction. In most of these countries, the region's poverty monitoring surveys are conducted on an annual basis. By adding a separate section to the questionnaire, as was done in the TLSS, it would be possible to collect rich and relevant data about the health needs of the population and the situation in the health care sector. Analysing collected data will help policy makers and health planners to make informed decisions on specific directions of reforms in health care.
|
Endnotes |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusion and implicationsEndnotesReferences |
|---|
1 The standards of prenatal health care in Tajikistan have not changed since the dissolution of the former Soviet Union. These standards include more than a dozen visits during pregnancy for uncomplicated cases. However, for complicated cases more visits could be recommended. The purpose of prenatal care is to yield information regarding the health of the mother and foetus, and to carefully and regularly monitor the progress of the pregnancy in order to detect any potential problems early. In Tajikistan, prenatal care usually consists of collecting the mother's medical history, measuring the mother's blood pressure, recording her height and weight, testing the mother's blood and urine as well as ultrasound scans. In addition, routine monitoring provides an opportunity to discuss with the mother a wide range of issues related to changes during pregnancy, such as nutritional requirements, exercise and vitamin intake, as well as to address any psychosocial problems associated with pregnancy.
Literature has positively emphasized the high number of visits before the transition; for instance, the World Bank (1999) stated that during the Soviet period, comprehensive maternal care in Tajikistan included at least 14 prenatal visits to a health facility. Subsequently, several studies have expressed concern with the fact that the number of visits declined in Tajikistan during the transition; see, for example, Government of Tajikistan (2005), UNICEF (2002), Falkingham (2003, 2004) and World Bank (1999). 
2 We tested the overdispersion of Poisson regression using the likelihood ratio (Long and Freeze 2006). The results strongly suggest an overdispersion in the Poisson regression model (G2 = 128.94, Prob 
G2 = 0.000). In addition, the estimated measure of dispersion (alpha) in the negative binomial model is positive 0.080148 (0.014753). This also confirms overdispersion in the Poisson model. Because of overdispersion, we moved to the alternative method of negative binomial regression.
3 For instance, parity, woman's age at first birth, woman's age at marriage, and time elapsed between births are highly correlated to poverty, lower educational attainment and limited knowledge about matters related to sex. Thus, poor women with low education attainment and limited knowledge about sexual matters usually have more children, marry and have children earlier, and have shorter time periods between births. Likewise, urban/rural and regional variables are highly correlated with the availability of health care facilities and sewerage, time to reach a health care facility and quality of health care.
4 In addition, the data presented in Figure 2 confirm that the changes in prenatal health care utilization were gradual. Therefore, a continuous variable ‘year of last birth’ used in this study reflects the actual situation better than a dummy variable. A dummy variable indicates that the year of childbirth took place before or after dissolution of the former Soviet Union.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusion and implicationsEndnotesReferences |
|---|
ADB. Azerbaijan. Country gender assessment (2005) Manila: Asian Development Bank (ADB).
Aleshina N, Redmond. How high is infant mortality in Central and Eastern Europe and the CIS (2003) Innocenti Working Paper No. 76. Florence: UNICEF Innocenti Research Centre.
Andersen R. Revisiting the behavioral model and access to medical care: Does it matter? Journal of Health and Social Behavior (1995) 36:1–10.[CrossRef][Web of Science][Medline]
Andersen R, Newman J. Societal and individual determinants of medical care utilization in United States. Milbank Quarterly (1973) 51:95–124.[Web of Science]
Balabanova D, McKee M, Pomerleau J, Rose R, Haerpfer C. Health services utilization in the former Soviet Union: evidence from eight countries. Health Services Research (2004) 39:1927–49.[CrossRef][Web of Science][Medline]
Bonilla-Chacin ME, Murrugarra E, Temourov M. Health care during transition and health system reform: evidence from the poorest CIS countries. Social Policy and Administration (2005) 39:381–408.[CrossRef]
Cashin CE, Borowitz M, Zuess O. The gender gap in primary health care resource utilization in Central Asia. Health Policy and Planning (2002) 17:264–72.
Deb P, Trivedi PK. Demand for medical care for the elderly: a finite mixture approach. Journal of Applied Econometrics (1997) 12:313–36.[CrossRef][Web of Science]
European Observatory on Health Care Systems. Health care systems in transition: Tajikistan (2000) Copenhagen: European Observatory on Health Care Systems, WHO Regional Office for Europe.
Falkingham J. From security to uncertainty: The impact of economic change on child welfare in Central Asia (2000) Innocenti Working Paper No. 76. Florence: UNICEF Innocenti Research Centre.
Falkingham J. Inequality and changes in woman's; use of maternal health-care services in Tajikistan. Studies in Family Planning (2003) 34:32–43.[CrossRef][Web of Science][Medline]
Falkingham J. Poverty, out-of-pocket payments and access to health care: evidence from Tajikistan. Social Science and Medicine (2004) 58:247–58.[CrossRef][Web of Science][Medline]
Gotsadze G, Bennett S, Ranson K, Gzirishvili D. Health care-seeking behaviour and out-of-pocket payments in Tbilisi, Georgia. Health Policy and Planning (2005) 20:232–42.
Government of Tajikistan. Poverty reduction strategy paper. Second progress report (2005) Dushanbe, Tajikistan.
Gwatkin D, Rustein S, Johnson K, Pande R, Wagstaff A. Socio-economic differences in health, nutrition and population (2000) Washington, DC: World Bank.
Ivanov LL, Flynn B. Utilization and satisfaction with prenatal care services. Western journal of Nursing Research (1999) 21:372–86.
Ivanov LL. Use of a Western theoretical model to investigate relationship among characteristics of pregnant women, utilization and satisfaction with prenatal care services in St. Petersburg, Russia. Public Health Nursing (2000) 17:111–20.[CrossRef][Web of Science][Medline]
Khan M, Hotchkiss D, Berruti A, Hutchinson P. Geographical aspects of poverty and health in Tanzania: does living in the poor area matter? Health Policy and Planning (2005) 21:110–22.[CrossRef][Web of Science][Medline]
Lahiri K, Xing G. Modeling Medicare-eligible veterans’ demand for outpatient services: a two-stage approach. Health Services & Outcomes Research Methodology (2003) 4:221–40.[CrossRef]
Lahiri K, Xing G. An econometric analysis of veteran's; health care utilization using two-part methods. Empirical Economics (2004) 29:431–49.[CrossRef]
Lee ES, Forthofer RN. Analyzing complex survey data (2006) London: Sage.
Lewis M. Who is paying for health care in Europe and Central Asia? (2001) Washington, DC: World Bank.
Long S, Freese J. Regression models for categorical dependent variables using Stata (2006) College Station, TX: STATA Press Publication.
Pohlmeier W, Ulrich V. An econometric model of the two-part decision-making process in the demand for health care. Journal of Human Resources (1995) XXX:339–61.
Sari N, Langenbrunner J. Consumer out-of-pocket spending for pharmaceuticals in Kazakhstan: implications for sectoral reform. Health Policy and Planning (2001) 16:428–34.
STATA. Survey Data Reference Manual (2005) College Station, TX: STATA Corp.
Suci E. Child access to health services during the economic crisis: an Indonesian experience of the safety net program. Social Science and Medicine (2006) 63:2912–25.[CrossRef][Web of Science][Medline]
Thind A, Cruz AM. Determinants of children's health care utilization in the Philippines. Journal of Tropical Pediatrics (2003) 49:269–73.
UNICEF. The status of women and children: Tajikistan, 2000. Multiple indicator cluster survey (2002) Dushanbe, Tajikistan: UNICEF.
UNICEF. TransMONEE database 2007 (2007) Florence, Italy: UNICEF Innocenti Research Centre.
Walters S, Suhrcke M. Socioeconomic inequalities in health and health care access in Central and Eastern Europe and the CIS: a review of the recent literature (2005) Working Paper 2005/1. Venice: WHO European Office for Investment for Health and Development.
WHO. Mortality country fact sheet 2006. (2006) accessed 10 June 2007. Online at: http://www.who.int/whosis/mort/profiles/mort_euro_tjk_tajikistan.pdf.
World Bank. Project Appraisal Document: Tajikistan Primary Health Care Project (1999) Washington, DC: The World Bank.
World Bank. Republic of Tajikistan poverty assessment (2000) Washington, DC: World Bank.
World Bank. Tajikistan Living Standards Survey, 2003: Notes for users of the data (2003) Washington, DC: World Bank.
Accepted for publication 12 May 2008.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||