Midterm review of national health plans: an example from the United Republic of Tanzania

Abstract In the health sector, planning and resource allocation at country level are mainly guided by national plans. For each such plan, a midterm review of progress is important for policy-makers since the review can inform the second half of the plan’s implementation and provide a situation analysis on which the subsequent plan can be based. The review should include a comprehensive analysis using recent data – from surveys, facility and administrative databases – and global health estimates. Any midterm analysis of progress is best conducted by a team comprising representatives of government agencies, independent national institutions and global health organizations. Here we present an example of such a review, done in 2013 in the United Republic of Tanzania. Compared to similar countries, the results of this midterm review showed good progress in all health indicators except skilled birth attendance.


Introduction
In the health sector, regular reviews of progress and performance are critical for good planning and resource allocation. Most countries have a national strategic plan for the health sector that outlines major strategies and sets targets. Such plans often cover a period of five years and usually include regular monitoring of core indicators and the progress being made towards set targets. The results of progress reviews are typically published in annual progress reports [1][2][3][4][5] and discussed at annual meetings of the relevant stakeholders. The stakeholders may then decide how -and if -the implementation of the strategic plan needs to be adjusted. In a recent investigation of 75 countries, it was found that 58 (77%) countries were conducting annual reviews of their health sectors. 6 During the implementation of a strategic plan, such annual reviews are often supplemented with more extensive midterm andsometimes -final reviews. The World Health Organization (WHO) has specified the key characteristics of a monitoring and evaluation platform for supporting regular reviews of health sector strategy. 7 Compared with annual reviews, midterm reviews are usually broader in process and scope. They may be conducted in conjunction with a regular annual review or they may replace it. 6,7 They take a multi-year perspective and pull together all of the available relevant data to assess the progress and performance of the health sector as a whole. Ideally, they should pay special attention to data quality, long-term trends, contextual changes and equity, and compare performance at national level with that in other similar countries. They should also form the basis for the situation analysis for the next strategic plan.
At present, most midterm reviews are conducted by teams of international and local consultants and the analysis and synthesis of evidence are usually quite rapid. 6 Ideally, any midterm review should begin with a thorough and systematic analysis and synthesis of all of the relevant data, to provide a comprehensive picture of progress and performance. It should be conducted by national research and public health institutions, in close collaboration with the national health ministry, the national statistical authority and international partners.
As an example, we present a midterm review done for the United Republic of Tanzania in 2013. A systematic approach was used to analyse and synthesize data from multiple sources.

Context of the review
The Ministry of Health and Social Welfare of the United Republic of Tanzania is currently implementing its third national plan, which covers the period 2009-2015. 8 This plan represents the key government document that provides specific guidance to the health sector. The main aim of the plan is improving access to those health services that are relevant to the Millennium Development Goals -at community, facility and district levels. The plan pays special attention to equity and gender issues and the general improvement of service quality. There are baseline values available -for 2008-2009 or earlier -for most of the 42 indicators used to track progress with the plan's implementation and corresponding target values for 2015.
The results of household surveys and longitudinal community studies have indicated that the United Republic of Tanzania has made major progress towards the achievement of its health goals since the mid-1990s. 9 In particular, the child mortality rate declined substantially between 1995 and 2005. Since 1999, health sector reviews have been conducted on an annual basis and a health sector performance profile has been produced. 3 Our review -done in the context of general midterm policy review -included several sessions to report midterm progress to the Ministry of Health and Social Welfare and relevant development partners.

Monitoring health sector performance
Leonard EG Mboera et al.

Methods
The review team organized the work into four main elements (Box 1). To assess progress systematically, the national plan's indicators: six input, five output and 14 intervention coverage -including reductions in risk behaviours -and 17 impact or health status measures (Table 1) were put together in a stepwise logical results framework. 7 This framework was designed to provide a general overview of where investments had been successful and whether more investments were likely to lead to better results. We used a mix of health facility, administrative and survey data and the results of relevant research studies (Box 2) to estimate progress made since the beginning of the strategic plan.
The United Republic of Tanzania has had frequent population-based demographic and health surveys 10 as well as topic-specific surveys that covered human immunodeficiency virus (HIV), tuberculosis, malaria or other health issues. 11,12 We analysed data collected in recent surveys of these types, two na-tionwide socioeconomic panel surveys 13 and two health facility surveys. 14 Health surveys were conducted in 2008-2009 15 and 2012, 16 to assess the services available and the readiness of those services in terms of staffing, equipment, medicines and diagnostics. The readiness score was computed for different intervention areas, such as child health or malaria, by averaging the availability of a set of essential items.
The United Republic of Tanzania has a national system for the routine collection of data from health facilities and this system provided regional data on many of the indicators of interest for the period 2009-2012. Although we also had access to crude data in some very recent health facility reports, we were cautious in using them because of the potentially low quality of the numerators -the number of individuals covered by an intervention -and denominators -the estimated size of the population in need of that interventionthat had been used. Wherever possible, we assessed the reliability of these reports against the results of relevant surveys.
For our analysis, we used data from the 2012 national census to estimate denominators. 17 Coverages for antiretroviral therapy (ART) and prevention of mother-to-child transmission of HIV were estimated using the methods and data of the national acquired immunodeficiency syndrome (AIDS) control programme, the 2012 national census and the 2011 HIV/AIDS indicator survey. 11 Data on tuberculosis indicators were obtained from the national tuberculosis and leprosy control programme and a recent national survey. 12 As the results from the 2013 national health accounts exercise were not yet available at the time of our analysis, we used the corresponding WHO estimates. 18 Health workforce data were obtained from the human resources for health information system of the Ministry of Health and Social Welfare.
The strategic plan has no explicit targets for equity. However, wherever possible -and mainly using household survey data 10 -we investigated trends in inequalities by sex, age group, wealth quintile, region of residence and urban/ rural setting.
We compared the time trends in selected indicators and in health sector efficiency -assessed by comparing selected inputs and results -that we recorded for the mainland health sector of the United Republic of Tanzania with the corresponding values for the following countries; Burundi, the Democratic Republic of the Congo, Ethiopia, Kenya, Malawi, Mozambique, Rwanda, Uganda, Zambia and Zimbabwe. We also made between-country comparisons of mortality and health financing indicators using global estimates produced by WHO and other United Nations agencies. We used data from health surveys that had been implemented between 2009 and 2012 to compare progress on coverage indicators between these countries.
The preparatory stage of the midterm analysis took several months and included meetings to ensure broad participation. The actual analytical work was done over a period of four months, with different partners focusing on specific topics and data sources. The analytical team consisted of representatives of the Ministry of Health and Social Welfare, the National Institute for Medical Research, the Ifakara Health Institute and WHO.
All analyses were conducted in Excel (Microsoft, Redmond, United States of America) or Stata version 12 (StataCorp. LP, College Station, USA).

Core indicators and targets
• Map the core indicators of the health sector's strategic plan on to the logical results framework. • Update baseline data where new data are available.

Data sources
• Use all relevant data, including routine health facility data and those from household, special population and facility surveys and administrative data sources. • Assess data quality by considering trends over time and internal consistency of levels and trends at regional -or other, subnational -levels and by comparing results from surveys with facility-based data. • Make transparent adjustments, where needed, to correct improbable values for indicators that affect the overall analysis. • Archive and make the dataset used for the synthesis available.

Analysis and synthesis
• Wherever the relevant data are available, assess progress in terms of each of the core indicators, focusing on the first half of the period covering the plan's implementation. • Where relevant and possible, assess inequalities in demographic, socioeconomic and geographical stratifiers. • Compare national progress and performance on selected indicators with those of countries that have comparable data and estimates.

Communication of results
• Create (i) a comprehensive report organized by the key areas of the strategic plan, (ii) a set of statistical profiles for all major subnational areas, and (iii) a policy brief that summarizes the main results. • Present the results to -and discuss them with -the midterm review team and Ministry of Health • Present the main findings at the plan's next annual review.

Results
A detailed report of the midterm analytical review was published, 19 and we summarize progress made from 2009 to 2012 and compare this to the 2015 targets in Table 1.

Health of children
According to United Nations projections, 20 the target child mortality set in the national strategic plan had already been reached by 2012. Levels of child mortality showed reduction of the gaps between urban and rural children and between children in the poorest and those in the wealthiest families. Children receiving pentavalent and measles vaccines and nutritional status had also already reached the targets set in the national plan.

Maternal health
The Although HIV transmission appeared to decrease gradually during the first half of the plan's implementation, the number of people living with HIV remained the same -due to increases in both the population and in survival following treatment. By 2012, the percentage of HIV-positive pregnant women receiving antiretroviral prophylaxis was 71%, 9 percentage points from the 2015 target. The percentage of HIV-positive adults receiving ART in 2012 was already higher than the 60% target for those in need of such treatment, while 48% children in need of ART received the treatment.
In almost every region, case notification rates for tuberculosis decreased between 2008 and 2011. However, the tuberculosis survey of 2012 revealed a higher overall prevalence -295 cases per 100 000 population. This survey also showed that the frequency of treatment success for tuberculosis was high and already above the 2015 target value. 12

Health services
Rates of outpatient department utilization, which are considered as indica-tors of general access to health services showed no increase between 2009 and 2012. The facility assessments revealed improvements in the proportion of facilities offering integrated child health services, family planning, ART and malaria treatment -but a decline in care for women giving birth over this period (Table 2). There was a slight improvement in general service readiness -as measured by the availability in health facilities of tracer indicators such as diagnostics and medicines -but the percentage of facilities stocking the drugs used in the first-line treatments of HIV infection and tuberculosis fell.

Inequalities
The data from household surveys indicated that, in the first half of the plan's implementation, there were reductions in inequalities -by sex, urban/rural residence and socioeconomic position -in several indicators, including child mortality and immunization and malaria intervention coverages. For several anthropometric indicators in children and skilled birth attendance, however, large inequalities persisted (Table 1).
Combined analysis of the health facility and survey data indicated that, in general, the health services in western regions were relatively weak whereas those in eastern and northern regions were relatively strong. Further analysis of the regional data showed that some regions performed markedly better than might have been predicted from their level of socioeconomic development. 19

Between-country comparisons
In 2012, among the 11 countries we included in our comparison, only Kenya and Zambia had higher gross domestic products per capita than the United Republic of Tanzania

Discussion
Overall, the data collected on the progress made during the first half of the implementation of the United Republic of Tanzania  For any national plan that is to be implemented over several years, accountability and implementation need to be guided by up-to-date data on the relevant long-and short-term trends. Our analysis illustrates how a range of data sources can be used to obtain a comprehensive picture of the progress being made towards the targets of such a plan. Recent population-based health surveys were used to assess trends in several indicators such as mortality, intervention coverage and health behaviours -according to equity stratifiers -and to verify statistics derived from studies based in health facilities. Health facility data played a major role in our analysis because they were available for the year preceding the midterm review, allowed considerable geographical disaggregation and included all of the relevant information available on tuberculosis and HIV treatment. In addition, recent surveys of health facilities allowed us to assess service readiness and, therefore, the performance of the service delivery system. 14 Reliable administrative data were needed to assess trends in the financial and health workforce indicators.
Comparisons with similar countries can provide a useful perspective on the performance of a particular country's health system. In our choice comparison countries, we aimed to avoid an excessive preoccupation with country rankings. 23 Gaps in the data meant that we had to rely on the predictions of WHO or another United Nations agency for the between-country comparisons of several financial and health status indicators. Although the United Republic of Tanzania is relatively data-rich, we struggled to find enough data -or, at least, enough data of adequate quality -on several topics of interest. Such gaps are likely to be more prominent in countries with less frequent national health surveys or with poorly functioning routine health management information systems. The quality of the data collected routinely from health facilities is often very variable and such data need to be assessed carefully. A change in the system used to collect facility data, from a paper-based system to a web-based system, 24 should facilitate the systematic management, quality assessment and analysis of facility data.
A system of national health accounts and a comprehensive system for the registration of health workers are both necessary for the implementation of any national health plan. Although the indicators and targets that we investigated may not have been sufficient to provide a comprehensive picture of progress and performance, we made no attempt to gather data on topics that were beyond the scope of the plan. Some of these areas, such as noncommunicable diseases, may nonetheless have major and growing impacts on the health of Tanzanians. 25 In any midterm analysis, inequalities as the result of demographic, geographical or socioeconomic char- acteristics should be considered. A lack of district-level data of adequate quality meant that we were obliged to focus on regional differences in our analysis of geographical inequalities, 19 even though analysis at district level -i.e. the level at which United Republic of Tanzanian resources for health are allocated -may have been preferable. The regional data we used came from large samples and from population surveys and health facilities. Although this is not the case in the United Republic of Tanzania, the districts of some countries, such as South Africa, have such large populations that district-level analysis of data from health facilities is both feasible and useful. 26 This analysis should help the United Republic of Tanzania develop a single strong monitoring and evaluation platform for its national strategic plans and major health programmes. 7 Such a platform, if adequately funded, could collect and disseminate useful data of good quality and transparency and form the basis for all global reporting to donors. ■

Monitoring health sector performance
Leonard EG Mboera et al.

Revisión intermedio de los planes de salud nacionales: un ejemplo de la República Unidad de Tanzanía
En el sector de la salud, la planificación y la asignación de recursos a nivel nacional se guían principalmente por planes nacionales. Para los planificadores de políticas es importante contar una revisión intermedia de los progresos de cada plan, ya que puede proporcionar información sobre la segunda mitad de la implementación del plan y un análisis de la situación en la que basar planes posteriores. La revisión debe incluir un análisis exhaustivo con datos recientes -a partir de encuestas y bases de datos administrativas y de los centros -y estimaciones mundiales de salud. La mejor manera de realizar cualquier análisis intermedio de los progresos es a través de un equipo integrado por representantes de organismos gubernamentales, instituciones nacionales independientes y organizaciones mundiales de salud. Aquí presentamos un ejemplo de ese tipo de revisiones, realizada en 2012 en Tanzania. En comparación con países similares, los resultados de este examen intermedio mostraron un progreso correcto en todos los indicadores de salud, excepto en la prestación de atención especializada en el parto.