Modeling trends of health and health related indicators in Ethiopia ( 1995-2008 ) : a time-series study

Background: The Federal Ministry of Health of Ethiopia has been publishing Health and Health related indicators of the country annually since 1987 E.C. These indicators have been of high importance in indicating the status of health in the country in those years. However, the trends/ patterns of these indicators and the factors related to the trends have not yet been investigated in a systematic manner. In addition, there were minimal efforts to develop a model for predicting future values of Health and Health related indicators based on the current trend. Objectives: The overall aim of this study was to analyze trends of and develop model for prediction of Health and Health related indicators. More specifically, it described the trends of Health and Health related indicators, identified determinants of mortality and morbidity indicators and developed model for predicting future values of MDG indicators. Methods: This study was conducted on Health and Health related indicators of Ethiopia from the year 1987 E.C to 2000 E.C. Key indicators of Mortality and Morbidity, Health service coverage, Health systems resources, Demographic and socio-economic, and Risk factor indicators were extracted and analyzed. The trends in these indicators were established using trend analysis techniques. The determinants of the established trends were identified using ARIMA models in STATA. The trend-line equations were then used to predict future values of the indicators. Results: Among the mortality indicators considered in this study, it was only Maternal Mortality Ratio that showed statistically significant decrement within the study period. The trends of Total Fertility Rate, physician per 100,000 population, skilled birth attendance and postnatal care coverage were found to have significant association with Maternal Mortality Ratio trend. There was a reversal of malaria parasite prevalence in 1999 E.C from Plasmodium Falciparum to Plasmodium Vivax. Based on the prediction from the current trend, the Millennium Development Goal target for under-five mortality rate and proportion of people having access to basic sanitation can be achieved. Conclusion: The current trend indicates the need to accelerate the progress of the indicators to achieve MDGs at or before 2015, particularly for Maternal Health and access to safe water supply. Published: 13 December 2009 Health Research Policy and Systems 2009, 7:29 doi:10.1186/1478-4505-7-29 Received: 21 September 2009 Accepted: 13 December 2009 This article is available from: http://www.health-policy-systems.com/content/7/1/29 © 2009 Abraha and Nigatu; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


Introduction
There is no single "standard" measurement of health status for individuals or population groups. Individual health status may be measured by an observer (e.g., a physician), who performs an examination and rates the individual along any of several dimensions, including presence or absence of life-threatening illness, risk factors for premature death, severity of disease, and overall health. Individual health status may also be assessed by asking the person to report his/her health perceptions in the domains of interest, such as physical functioning, emotional well-being, pain or discomfort, and overall perception of health [1].
The Health of an entire population is determined by aggregating data collected on individuals. The health of an individual is easier to define than the health of a population. Once the definition of optimum health for the individual is agreed upon, health status can be placed along a continuum from perfect health to death. No comparable scale exists for whole populations. What is the population-level equivalent of death? (Keep in mind that it is unusual for entire populations to die.) In the absence of comprehensive or absolute measures of the health of a population, the average lifespan, the prevalence of preventable diseases or deaths, and availability of health services serve as indicators of health status [2]. Hence, judgments regarding the level of health of a particular population are usually made by comparing one population to another, or by studying the trends in a health indicator within a population over time. There are two main goals of time-series analysis (a) identifying the nature of the phenomenon represented by the sequence of observations, and (b) forecasting (predicting future values of the time series variable). Both of these goals require that the pattern of observed time series data is identified and more or less formally described. Once the pattern is established, we can interpret and integrate it with other data [3].
One of the hallmarks of epidemiologic analysis is the understanding that health outcomes in a population can only be fully understood if their frequency and distribution is examined in terms of person, place, and time. Trend analysis is one leg of this analytic triangle, and is used for public health surveillance and monitoring, for forecasting, for program evaluation, for policy analysis, and for etiologic analysis, investigation of potentially causal relationships between risk factors and outcomes [4].
The Federal Ministry of Health of Ethiopia has been publishing Health and Health related indicators as a standalone document since 1987 E.C. Through this document, which is published on annual basis, the annual status of several Health and Health related indicators is reported. Mortality and morbidity indicators, Health service coverage indicators, Health system resource indicators and socio-economic indicators are the major ones [5].
Though the Ministry has been producing such useful indicators the trend of those important indicators hasn't been systematically investigated. This doesn't however mean that there are no efforts in analyzing the trends. There are efforts in presenting trends through graphs and tables within the indicator documents. But as to the existing knowledge, there are no studies that tried to fit the trends in modeling equations and predict future values based on the current trends for the majority of the indicators. Besides, the determinants of changes in key health status indicators like Maternal Mortality Ratio and Infant Mortality Rate hasn't been established through modeling exercises and regression models in a time-series dataset of the indicators [6].
In light of this, the envisaged study was designed to describe the overall pattern of Health and Health related indicators overtime by comparing one time-period to another time-period in the series. The study had also developed trend-line equations/models that can be used to predict future values based on the current trend. Moreover, these equations/models were used to predict MDG indicator values for 2015 which were then compared with their target values. Among the purposes was also identification of determinants of changes/improvements in mortality and morbidity indicators during the study period. This study therefore has identified determinants of mortality and morbidity indicators in Ethiopia during the period of 1987 E.C to 2000 E.C (1995 to 2008) [7].
Health and Health related indicators in Ethiopia for 14 years period (1987-2000 E.C). Accordingly, the study indicators were selected Health and Health related indicators in Ethiopia.

Sample and sampling
The number of time-periods being examined for each health and health related indicator was 14 (1987-2000 E.C), that is since the time at which Health and Health related indicators were obtained. The number of Health and Health related indicators considered in this study was selected in the following way. First, the Health and Health related indicators were categorized in to five major categories based on the WHO 2008 health statistics.
The categories and the number of indicators in each category were Mortality and Morbidity indicators (14 indicators), Health service coverage indicators (10 indicators), Health systems resources indicators (10 indicators), Demographic and socio-economic indicators (10 indicators), and Risk factors indicators (6 indicators). These all make the total number of indicators included in the study to be 50.
From each category of Health & Health related indicator, study indicators were selected purposively. The criteria that were used to select the indicators were: Public Health importance/widely used, data availability for the study period, measurement at national, relevance to MDGs and presence/absence of significant measurement change during the study period.
A total of 90 indicator documents were reviewed of which 45 national and the rest 45 were international documents. Indicator values were cross-evaluated across and within these sources.
For the purpose of this study, three instruments were employed: Documents collection checklists, indicator collection format and indicator definition. In order to collect the documents from which study indicators and/or their determinant obtained a document collection checklist was used to facilitate the collection of indicator documents. An excel sheet to which the values of study indicators were filled was used to collect the relevant indicators of the study across the study duration. This definition was based on the definition of Ministry of Health and other sectors. The indicator definition was prepared (adopted) to assist the interpretation of the findings.
Six Public Health Professionals were oriented about the study and its data requirements based on the objective of the study. They collected the available documents (in soft and/or hard copies) that bear indicator values and they referred for the rest of the documents to departments of the Ministries (Authorities) and libraries. The principal investigator reviewed the collected documents and extracted the indicators.

Data analysis and interpretation
All the indicator values were entered in to a spreadsheet in Ms-excel. The entered data were checked once again for accuracy of data entry. Errors during the entry were cleaned and analysis started. The analysis of the data followed the following steps:

Describing Trends
The first step in the analysis looked in to trend in each Health and Health related indicator. The patterns of each Health and Health related indicator were described numerically and graphically. The changes of the indicators were also described using numerical summaries and line graphs. Changes of mortality and morbidity indicators were additionally described by confidence intervals using STATA 8.0 and only those with statistically significant changes, non-overlapping confidence intervals were considered for Auto Regressive Iintegarted Moving Average (ARIMA) model.

Fitting trend-line equations
The trend was plotted using a line graph by connecting the points on the X-Y axis, where X is the time in year and Y is value of the selected indicator for each year. Then an effort was made to identify a trend-line equation that best fits the line using coefficient of determination (R 2 ). Data transformation and smoothing was made to change the rate and flatten the series of rates without changing the shape of the trend (e.g. MMR≅ log MMR). The rate of annual change of the indicator values (R) was estimated by equating the indicator (I) and time (T): I = B0 + R (T), when linearity was assumed. If linearity was not assumed, other trend-line options (trend/regression types) including polynomial, logarithmic, exponential and power were considered

Identifying determinants
Here morbidity and mortality indicators were considered as outcome variables and any indicator that affected the outcome variables was taken as independent variables. For the given period of 14 years, the morbidity (burden of disease) and mortality indicators which showed statistically significant change were assessed for their association with possible determinants (based on literature evidence) which had statistically significant change across time. First correlation of variables was observed in a correlation matrix. Then all potential determinants of a dependent variable (indicator) were fitted in to ARIMA (Auto Regression Integrated Moving Average) model in STATA 8.0. For all cases, P-values < 0.05 was considered to be statistically significant.

Predicting future values
The best fit equations of the trend-line of an indicator were used to predict the future values of the indicator for 2015. These predicted values were then compared with the targets for 2015 by the MDG. Then the findings were discussed in the perspective of achieving the MDGs.

Ethical consideration
Ethical clearance was obtained from the Ethical review committee of Faculty of Health, Jimma University. Official permission was requested from the different organizations for documents. For data obtained from publications and web pages, all the data sources were acknowledged. shows that it is only the Maternal Mortality that shown a statistically significant decrement in the study period. The rest of the mortality indicators showed a decrement but haven't shown any statistically significant reduction.

Description of trends
The decrease in the mortality rates (maternal mortality rate, infant mortality rate, and under-five mortality rates) during the study period are shown in the following graph. The higher change (decrement) of Infant mortality rate was occurred between years 1987 to 1993 however for Maternal Mortality Ratio, Under-five Mortality rate and Crude Death Rate greater change was shown between 1987 and 1993 E.C. The highest decrease in mortality was the under-five mortality rate followed by the infant mortality rate. The least decrement is observed in Crude Death Rate followed by Maternal Mortality Ratio.
In analysis of trends in the major mortality indicators, infant mortality rate, under-five mortality rate, and crude death rate were found to have linear decrement trend. The yearly decrement rates for these indicators were 2.97 per 100,000 live births, 3.45 per 100,000 live births and 0.38 per 1000 population respectively ( Figure 1). However, the Maternal Mortality Ratio followed the logarithmic trend with a yearly decrement rate of 672 per 100,000 live births ( Figure 2).
With regard to AIDS related mortality, there were an estimated total of 1.3 million AIDS deaths in the study period (1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000). This will make median AIDS related deaths per year to be 101,000. The maximum estimated number of AIDS deaths per year was 134 thousand which was observed in 1997. The trend of AIDS related deaths had an increasing trend but it showed a decrement after 1997 ( Figure 3).
As to Tuberculosis related mortality, there were 17,022 reported Tuberculosis deaths in the period 1992-2000. The median number of TB deaths per year was found to be 1993 with a range of 898 to 2548. The smallest number of TB deaths was observed in 1992 while the largest was in 1997. The pattern of TB deaths had an increasing trend till 1997 but a decreasing pattern after 1997.  trend for smear negative PTB and Extra-pulmonary TB.

Morbidity indicators
There is a decreasing in smear positive pulmonary TB (Figure 6).
With regard to malaria morbidity, there were about 7 million malaria positive cases in the study period (1987-2000 E.C). This will make an average of half a million malaria positive cases per year. There were two peaks in the number of malaria cases. The smaller peak was in 1991 and the higher peak was in 1997 E.C. Lower number cases were reported in 1992 and the lowest was in 2000. There was a dramatic decrease in malaria as of 1997 E.C. The median number of malaria positive cases per year was 773 per 100,000 population with a minimum of 457 in 2000 and a maximum of 1092 in 1997 E.C. The median number of P. Falciparum and P. Vivax cases per 100,000 population per year in the same period was 493 and 249 respectively ( Figure 7).
There was a reversal of malaria parasite prevalence in 1999 E.C from P. Falciparum to P. Vivax. P. Falciparum remained to be the major prevalent species till 1998. However, in 1999 this leading role was overtaken by P. Vivax. In 1987 E.C, the percentage of P. Falciparum and P. Vivax were 65.6 and 34.3 respectively. But in 1999 E.C, the percentage for P. Falciparum and P. Vivax were 36.3 and 63.7 respectively. This shows the reversal of the parasite type in Ethiopia ( Figure 8).
Regarding the percentage of mixed infection (P. Falciparum and P. Vivax) there was almost non/few mixed infections at 1987 (0.046%) but gradually the mixed infection had an increasing trend which reached peak in 1998 (3.4%) ( Figure 9).

Health service coverage indicators
To describe the Health service coverage, Antenatal care (ANC), Skilled birth attendance (SBA), Postnatal care (PNC), Contraceptive prevalence rate (CPR), Expanded Program for Immunization coverage (EPI), and Potential Health Service coverage from the government facilities (PHS) were used. There is a general exponentially increasing trend in all these Health service coverage indicators during the study period. The total increase and the average increase per year are shown in Table 2.
As shown in the above table, the highest increase in service coverage was on EPI coverage followed by CPR and Potential Health Service coverage. The lowest increment among the six Health service coverage indicators was on skilled birth attendance and postnatal care ( Figure 10).

Demographic and socioeconomic indicators
During the study period, the population of Ethiopia has increased from 53-73 million. The average increment rate per year was 1.54 million (2.91%). The total fertility rate in 1987 was 6.4 then it decreased to 5.5 in 1992 and to 5.4 in 1997. This is equivalent to a decrease of 1 child per woman during ten year period. The highest decline in TFR was in the period 1987-1992 E.C which is 0.9 as compared to 0.1 in the period 1993-1997 E.C. Crude birth rate was 44.2,39.9 and 35.7 per 1000 population in 1987, The percentage contribution of P. Falciparum and P. Vivax to all malaria cases  Table 3.

Risk factor indicators
The average (SD) increment rates of access to safe water, piped water and basic sanitation were 20.2 (8.6), 5(1.4) and 15(7.1) respectively from 1987 to 1997. The pattern of these environmental health indicators is shown in Table 4.
Prevalence of underweight and stunting of under-five had a decreasing trend with an average (SD) increment rate of 6.3(3.3) and 8.78(4.6) per five year respectively. In annual basis, the prevalence of underweight had a decrement rate of 1.26. However prevalence of wasting showed an increment of 2.5 from 1987 to 1992 and had no change for the subsequent five years (Table 5).

Determinants of Trends of Mortality and Morbidity Indicators
Maternal Mortality Ratio (MMR) Twelve variables were found to have statistically significant correlation with Maternal Mortality Ratio. The varia-  bles, their correlation coefficients and significance level are shown in (Table 6).
The crude and adjusted beta-coefficients obtained using Autoregressive integrated moving average (ARIMA) model for the association of the above covariates with Maternal Mortality Ratio (MMR) are shown in (Table 7). As there were equal laggs on all the variables ARIMA is used to control for third variables that could explain the association.
In the final model, physician per 100,000 populations, skilled birth attendant and postnatal care coverage were found the only predictor time series variables that had statistically significant association with Maternal Mortality Ratio ( Table 8).
As it mention in

Prediction of Health and Health Related Indicators
Based on the time-trend of the mortality and morbidity indicators in the past 14 years, the following equations with their respective R 2 were obtained. Using those equations the values of the indicators for 2015 (MDG target year) were predicted. As it is shown in the following

Mortality and morbidity indicators
There was generally a decreasing trend of all mortality indicators during the study period. As indicated in the results section, all mortality indicators except the maternal mortality rate haven't shown a statistically significant decrease between 1987 and 1998 E.C. This finding is a result of confidence intervals comparison that depends on the sample size considered. However, a decrease of 33 infant deaths per 1000 live births, a reduction of 42 under-five deaths per 1000 children under-five years of age, and a decrement of 5 deaths per 1000 population are of great practical significance for a health system in a developing country. These were in line with the Health policy directions of the Ethiopian government [5][6][7] There was high decrement of AIDS related mortality since 1997 E.C as shown in the AIDS mortality graph. This time was also associated with an increment in the number of people living with HIV/AIDS. These changes are most probably due to the introduction of free Anti-retroviral treatment (ART) at that period in time which had resulted in the decrement of number of AIDS related deaths and increment of HIV+ people.
The magnitude of deaths due to Tuberculosis and Malaria had shown a dramatically decreasing trend since 1996 E.C. That decrement trend was higher for malaria than for Tuberculosis. National efforts for malaria and Tuberculosis control programs through the support from the Global fund are most likely the possible explanations for such decrement. The ministry of health, as part of the Millennium Development Goals targets, has given the greatest attention to availing two insecticide treated nets per household and improving the case detection and decreasing the default rates of TB treatment [6].
There was an increasing trend in the number of outpatient visits and inpatient admissions during the study period.
Although it is expected to have a decreasing trend the number of new and total Tuberculosis cases had also an increasing trend. It is true that there is an expansion of Health facilities and increasing access to health facilities and an improvement in the health management information system (HMIS) during the study period. It would most likely be due to these factors, rather than a true increment in morbidity that the trend has shown an increasing pattern.
The peak in the number of malaria cases per 100,000 population were in 1991 and 1996 E.C. These peaking effects in the number of malaria cases were due to the malaria epidemics in those years. Plasmodium Falciparum remained to be the dominant malaria species among the diagnosed cases contributing up to 65% of the malaria cases till 1998 E.C. However, that scenario was reversed in 1998 E.C when Plasmodium Vivax diagnosed cases contributed 65% of all malaria diagnosed cases. What has caused this reversal of the two malaria species needs further study. On the other hand the prevalence of mixed infections has increased during the last three years of the study period.

Health service coverage indicators
Among the Health service coverage indicators EPI and CPR had shown the highest increment during the study period as compared to the other health service coverage indicators considered in this study. This could possibly be due to the several well coordinated immunization campaign coordinated among the ministry of health, World Health Organization, UNICEF and other governmental and non-governmental stakeholders.
On the other hand, the lowest increments among the health service coverage indicators were on skilled birth  attendant and postnatal care services coverage. This has been the case in several other cross-sectional studies which claimed cultural factors and perceived quality of the services to be the major contributing factors.
The increments in most health service coverage indicators were smooth till 1998 E.C. However, in 1999 and 2000 E.C there were sharp increments especially in CPR and EPI coverage. This might be due to the deployment of more Health Extension workers to rural health posts who are primarily focusing on promotive and preventive health care services.

Health system resource indicators
The hospital to population ratio in Ethiopia which is 1:518.948 is higher when compared to the previous val-ues but still lower than that of most African countries. The current Health centre to population ratio which is 1:91,726 is much lower than the expected which is 1:25,000. Based on this expected health centre to population ratio additional 2200 health centres are needed. But the current Health post to population ratio (1: 5,198) is very close to the standard (1:5,000).
The Ethiopian Health system has currently a physician to population ratio of 1:22,198 in 2000 E.C. The WHO standard is 1:10,000. Hence the country needs to double the current number of physicians to meet the standard. In contrary to the physician to population ration, the current nurse to population ratio (1:4,519) now exceeds the WHO standard (1:5,000). The trend in nurse to population ratio has showed a decreasing tip in 1999 and 2000  E.C, which may be due to their enrolment in to the accelerated health officers training program. While the expected physician to nurse ratio is 1:2, the current Ethiopian physician to nurse ratio is 1:5.
Had it not been for the non-uniform distribution of physicians in different areas of the country, each hospital would have about 19 physicians, if the currently available physicians were to be equally distributed to all hospitals. Similarly, if all the Health officers are to be equally distributed to all the health centres, each health centre will have at least one health officer. With regard to the health centre to hospital referral, each hospital will have about 6 referring health centres.
During the study period the average increment in total health budget and total Health Expenditure per capita per year were 1.33ETB and 1.1ETB respectively. During the same period, the average increment in GDP per capita per year was 202ETB. Hence one can see that these rates of increment are far apart.

Determinants of maternal mortality
Total fertility rate, physician per 100,000 population, skilled birth attendance and post-natal care service coverage were found to be determinants of Maternal Mortality Ratio. Among those determinants total fertility rate and physician per 100,000 population have positive association with Maternal Mortality Ratio. However skilled birth attendant and post-natal care coverage associated negatively with Maternal Mortality Ratio.
The decrement of Maternal Mortality Ratio has strong association with Total fertility with β coefficient of 723.6. The decrement in total fertility rate in turn might be due to the increment in contraceptive prevalence rate which is the one which has significant increment among all health service coverage indicators in the study and female literacy rate.
When physician per 100,000 population is correlated with Maternal Mortality Ratio, there was a negative association. However, when other variables were controlled, the negative association is reversed to positive association. This is contrary to the expectation that an increment in physician per 100,000 population is related with a decrement in Maternal Mortality Ratio. What so ever the case, this might be due to the distribution of physicians more in urban areas which might result a decrement of Maternal Mortality Ratio in urban areas only. This might need disaggregated analysis across urban and rural areas.
Skilled birth attendance rate and postnatal care coverage had negative association with Maternal Mortality Ratio. But these are the health service coverage indicators with lowest increment during the study period. To reduce Maternal Mortality Ratio to a greater extent, skilled birth attendance rate and postnatal care coverage should get increased.

Prediction of indicators for 2015
The MDG targets related to mortality for 2015 include reduction of under-five mortality by two-third and reducing three quarters of Maternal Mortality Ratio. The MDG targets related to morbidity for 2015 are to have halted and begun to reverse the spread of HIV/AIDS; to have halted and begun to reverse the spread of malaria, and to halve the prevalence of tuberculosis. The MDG target related to risk factors for 2015 is to halve the number of people without sustainable access to safe drinking water and basic sanitation (Table 11). These targets has taken in to account the baseline population in 1990 [8].
Although all MDG indicators are targeted for 2015 as an end date to measure success clearly earmarked target values for 2015 were obtained only for the four major indicators. These were the two mortality indicators (Maternal Mortality Ratio, under-five mortality rate) and two risk factor indicators (proportion of people with safe drinking water and proportion of people with basic sanitation). Table 11, the under-five mortality rate in 1990 was 204. The MDG was meant to decrease this value by two-third. This is equivalent to reducing it to 68 (52.4, 83.6). Using the prediction equation obtained in this study, the predicted value for 2015 was 92 (74,110). As the confidence intervals of the target and predicted values are overlapping, it can be assumed that the MDG target for under-five mortality can be achived provided that the current rates of reducing under-five mortality are maintained  Access to safe water supply seems on good progress. In 1990 access to safe water supply in Ethiopia was only 25%. Correspondigly, 75% of the population didn't have any access to safe water supply. The MDG target for 2015 was to halve the number of people who didn't have access to safe water supply. It was to increase the access to safe water supply to 62.5% (53%, 72%). The predicted value using the model obtained in this research was 41% (31.4%, 50.6%). A slightest acceleration in the rate of increament could result in the achievement of the MDG target.

As indicated in
More encouraging results were obtained on access to basic sanitation. Access to basic sanitation in 1990 was only 8% showing that about 92% of the population didn't have any access to basic sanitation. The MDG target was to halve this figure i.e. to reduce this 92% to 46%. This was equivalent to increasing access to basic sanitation to 54% by 2015. The good news is that access to basic sanitation will reach 100% (universal access) by 2011 based on the prediction in this study assuming that the current rate of growth is at least maintained.
Using the confidence intervals for projected values and target values, it is clear that the confidence intervals of under-five mortality rate and access to safe drinking water for target and projected values overlap indicating possibil-ity of reaching the targets. However, the confidence interval of Maternal Mortality Ratio for targets and projected values are not overlapping indicating that achieving the target for Maternal Mortality Ratio is less likely provided that the current trend continues. This analysis is based on the measurement methods of the MDG [11].
These findings have several policy implications. Firstly, the goverment should at least maintain the rate at which under-five mortality and access to safe water are improving in order to reach the MDG targets for this indicators. Secondly, the goverment of Ethiopia is less likely to achieve the MDG target for Maternal Mortality Ratio if the rate of decreament is not changed. Hence, there is a need to innovate some mechanisms to abruptly reduce the Maternal Mortality Ratio. This was actually similar to the trends of some of the developing countries [12].

Conclusions
Based on the findings of the study, the following conclusion are made: 1. Among the mortality indicators considered in this study, only Maternal Mortality Ratio has shown statistically significant change during the study period. Though statistical significance differs from practical significance, it indicates the need to decrease the mortality indicators to a greater extent.
2. Total Fertility Rate, physician per 100,000 population, skilled birth attendance and post-natal care service coverages were found to be determinants of the decrease in Maternal Mortality Ratio. While TFR and Physician per 100,000 population are positively associated, skilled birth attendance and PNC are negatively associated with MMR.
3. Plasmodium Falciparum was the leading malaria species till 1998. But there was a reversal of malaria parasite prevalence in 1999 E.C from P. Falciparum to P. Vivax. Mixed infection of both P. Falciparum and P. Vivax also shows an increasing trend.
4. Health Service coverage indicators showed a remarkable improvement during the study period. Despite they