One size does not fit all: an application of stochastic modeling to estimating primary healthcare needs in Ethiopia at the sub-national level

doi:10.21203/rs.3.rs-2613600/v1

Download PDF

Research

One size does not fit all: an application of stochastic modeling to estimating primary healthcare needs in Ethiopia at the sub-national level

https://doi.org/10.21203/rs.3.rs-2613600/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background: Primary healthcare systems require adequate staffing to meet the needs of their local population. Guidelines typically use population ratio targets for healthcare workers, such as Ethiopia’s goal of two health extension workers for every five thousand people. However, fixed ratios do not reflect local demographics, fertility rates, disease burden (e.g., malaria endemicity), or trends in these values. Recognizing this, we set out to estimate the clinical workload to meet the primary healthcare needs in Ethiopia by region.

Methods: We utilize the open-source modeling package PACE-HRH for our analysis. This is a stochastic Monte Carlo simulation model, which samples annually from distributions for fertility, mortality, disease burden, and the trends in these rates. Inputs were drawn from literature, DHS, and UN population estimates. We model seven regions and two charted cities of Ethiopia, based on data availability and the anticipated reliability of historical trends into the future.

Results: All regions and charted cities are expected to experience increased workload between 2021 and 2035 for a starting catchment of five thousand people. The expected (mean) clinical workload varied from 2,930 hours (Addis) to 3,752 (Gambela) and increased by 19-28% over fifteen years. This results from a decline in per capita workload (due to declines in fertility and infectious diseases), overpowered by total population growth. Pregnancy, non-communicable diseases, sick child care, and nutrition remain the largest service categories, but their priority shifts substantially in some regions by 2035. Sensitivity analysis shows that fertility assumptions have major implications for workload. We incorporate seasonality and estimate monthly variation of up to 8.9% (Somali), though most services with high variability are declining.

Conclusions: Regional variation in demographics, fertility, seasonality, and disease trends all affect the workload estimates. This results in differences in expected clinical workload, the level of uncertainty in those estimates, and relative priorities between service categories. By showing these differences, we demonstrate the inadequacy of a fixed population ratio for staffing allocation. Policy-makers and regulators need to consider these factors in designing their healthcare systems, or they risk sub-optimally allocating workforce and creating inequitable access to care.

human resources for health

workload

Ethiopia

sub-national

Monte Carlo

stochastic

PACE-HRH

primary healthcare

The design of healthcare systems is variable across countries depending on the local context, affected by the resources available, health priorities, and cultural preferences. To meet the needs of a population, primary healthcare systems require a properly staffed healthcare workforce(1). Building a fit-for-purpose workforce requires planning, training, certification, placement, and continuing education, as outlined by the World Health Organization (WHO) in their six building blocks for primary healthcare framework(2).

To estimate the number of trained workers that are needed, primary healthcare systems typically use health-workers-to-population ratio-based targets. For example, the WHO established a global threshold of 4.5 doctors and nurses/midwives per 1,000 population to achieve human resources for health goals of 80% coverage across 12 health indicators. The indicators selected are linked to the health sustainable development goals, which are based on empirical analysis of historical trends across specific geographies ². This is an update from the previous threshold of 2.3 physicians and nurses/midwives per 1,000 population identified in the 2006 World Health Report (3) to attain 80% coverage of skilled birth attendance. As demonstrated by these updates, static values are inadequate to account for evolving health priorities, coverage targets, and disease burden.

These ratios have been useful for aggregate planning (4,5), but since they are globally derived values, they do not reflect the differences in demographics, socio-economic access to care, or disease burden between geographies and populations. Recognizing the need for more tailored staffing models, many countries have established similar, population ratio-based targets. For example, India set a target ratio of one Accredited Social Health Activist (ASHA) per 1,000 population (6) and Nigeria established a goal of one primary healthcare center (PHC) per ward of 10,000 to 30,000 people (7). The methods by which these national standards are set are not often well-documented in the literature, but even if they originally came from a bottom-up process of estimating workload and the needed number of healthcare workers, once the ratios are enshrined as targets, the flexibility to adapt them locally or update based on trends over time stagnates.

As a result, these targets still face the challenges of being static in time, not being specific to the needs of sub-populations, and not reflecting the practicalities of local delivery systems such as population density. In addition, the rate at which disease burden and fertility are changing are heterogeneous at a sub-national level (8)^,so the geographical and temporal variation in healthcare needs are not reflected in a fixed ratio.

Even with a constant package of services, local incidence and demography impact the necessary healthcare worker capacity. Given this context, we set out to quantify the magnitude of sub-national and temporal variation in primary healthcare workload faced by healthcare workers to understand the implications of fixed-ratio planning. In doing so, we question whether fixed staffing ratios are adequate to ensure that a healthcare system can meet a population’s dynamic health needs.

Ethiopia’s health extension program (HEP) has been widely viewed as successful and the government has recently completed a planning process for the program over the next fifteen years (9), making it an interesting potential case study. We use this as the basis for assessing the primary healthcare workload across multiple regions of the country, projecting forward from 2020 to 2035 to align with their strategic roadmap.

We use a previously published, bottom-up time accounting approach (10) to estimate the level of workload for each primary healthcare service included in a comprehensive service package. The list of healthcare tasks (“services”) is based on the Ethiopian HEP optimization roadmap for 2021-2035 (9) (complete list in Additional file 1, table 1). This is a general primary healthcare service package with a broad range of services including immunization, antenatal care, family planning, nutrition, testing and treatment for both infectious and non-communicable diseases (NCDs), and routine preventive care and counseling. The analysis is run in R, based on the publicly available PACE-HRH software package (11).

Model structure

The detailed model structure is outlined in a different publication (10) and we summarize it here. The model calculates the total amount of direct clinical contact time (in minutes) required to meet all of the healthcare needs for a baseline population. We start with a population of 5,000 in 2020, which is the target population size that each primary healthcare facility (Health Post) is supposed to serve (9), although the actual catchment population varies depending on spatial distribution, varying fertility, and internal migration.

For each service, we break out the relevant tasks (e.g. testing, treatment, counseling), specify the relevant population to which this service applies (e.g. children under 5, pregnant women), the number of contacts with the health system to fulfill the service (e.g. 1 test per suspected malaria case), the amount of time to complete the task, and lastly, the incidence or prevalence rate to reflect the proportion of the relevant population in need of care. The relevant population and the incidence rate are used to calculate the number of people expected to receive the service each year, which is multiplied by number of contacts and the time per contact to get the total service time for each PHC service.

Administrative tasks, in-service trainings, community engagement, public health surveillance, travel, and related tasks can be optionally included in the model. For this study, we have chosen to focus on the population-based drivers of workload, so only include direct clinical care tasks.

We utilize a Monte Carlo simulation to estimate the expected time required for each task, and then sum all tasks to calculate the total clinical workload. The model includes both parameter uncertainty (the initial rate) and the stochasticity (year-over-year change) inherent in estimating future trends. The sampling occurs for parameters including fertility and mortality rates by five-year age groups, disease incidence rates, and time per contact (details of distributions in Additional File 1, Table 2).

We account for birth seasonality and time-varying conditions including malnutrition, tuberculosis (TB), malaria, and diarrhea by applying condition-specific seasonality curves to each relevant service. For example, birth seasonality is applied to antenatal care (ANC) visits, which requires 4 contacts with the health system. The seasonality curves are detailed in Additional File 1, Table 3 and offsets in in Additional File 1, Table 4.

To demonstrate the calculation method, Table 1 lists data inputs required by the model to compute total expected service time for a clinical task, using treatment for moderate acute malnutrition in under-five children as an example.

Table 1. Input parameters for task time calculations. Every clinical task in the model requires a set of inputs that are used in the time calculation. One example, for Amhara region, the task of treating moderate acute malnutrition in under-five children, is shown here. Incidence rate or prevalence rate are used as appropriate for the task being described. Annual change rates are calculated as a ratio to the prior year (i.e., 1.0 = no change)

Task time input parameter	Example value
Name of task	Treatment for moderate acute malnutrition
Relevant population	Children aged 0 to 59 months
Incidence/prevalence rate in the population	41.5% of under-five children
Number of contacts with the health system	2 contacts for malnutrition treatment
Time per contact with the health system	5 minutes per contact
Annual change in incidence/prevalence rate	0.98
Applicable seasonality curve	Malnutrition
Seasonality offset (months)	0 for the first contact, +1 for the second contact

Data collection and assumptions

We collected data for subnational modeling from a range of sources, depending on availability and quality. The model accounts for differences between regions across each of the key input variables described in Table 1 and uses the most localized data possible.

We utilized United Nations (UN) population estimates(12) for regions of Ethiopia by 5-year age bands, interpolated to 1-year cohorts and modeled future population pyramids based on trends in fertility and mortality. Fertility rates are based on DHS-reported regional fertility rates from 2019 (13). Mortality rates are sourced from the World Development Indicators (WDI) from 2020 (14). We calculated the annualized rate of year-over-year change from the most recent pair of consecutive observations. In the cases where recent trends conflict with long-term historical patterns or when we were uncertain about whether a rapid change can persist, we include additional historical data in calculating the annualized change rate to get a more stable trend.

Because population trends are a key driver of the model, we conducted a sensitivity analysis that compared the trend of annual change rates going back at least 10 years and up to 20 years.

We searched Google and Google Scholar for published research articles, national prevalence surveys, and surveillance surveys to gather data on regional disease incidence rates, annual change rates in disease incidence, and regional seasonality patterns for disease incidence. In cases where regional data were not published, but the disease is known to exist, we backfilled with national averages, based on the expectation that the national average is the next-best estimate for the region. Where the data are inconsistent, we use the most recent available and prioritize estimates found in population-based surveys and meta-analyses with random sampling techniques. Due to gaps in the literature, it was difficult to find reliable estimates for annual change rates in regional disease incidence. Accounting for local disease eradication efforts and long-term trends in disease prevalence, we assumed a moderate decline of two percent per year for the relevant infectious diseases. Assumptions for number of contacts are based on common practice, and minutes per contact are educated guesses, which were also validated by experts.

Detailed data sources for all model inputs are listed in Additional File 1, Table 5.

Model application

We forecast PHC clinical workloads at the regional level from 2021 to 2035. Ethiopia is composed of eleven National Regional States (regions) – Tigray, Afar, Amhara, Oromia, Somali, Benishangul-Gumuz, Southern Nations, Nationalities and Peoples’ Region (SNNPR), Sidama, South West Ethiopia Peoples’ Region, Gambela, and Harari – and two chartered cities – Addis Ababa and Dire Dawa. The Sidama region and the South West Ethiopia Peoples’ Region were split off from the SNNPR in 2019 and 2021 respectively. Due to the recency of these changes and the age of our data sources, we treat the Sidama region and the South West Ethiopia Peoples’ Region as part of SNNPR for expediency purposes; this should not be construed as an opinion on the changes to administrative borders.

We exclude Tigray and Afar in predicted clinical hours analyses because Tigray and parts of Afar have been affected by civil conflict from November 2020 to November 2022 (15,16). Displacement caused by the conflict has great implications for food security, healthcare needs and access to healthcare for the affected population. Prior to the onset of conflict, Afar had the highest overall fertility rates in the country and the fertility rates were increasing for the youngest age groups. The conflict could affect the age-group specific fertility and mortality rates in ways that are hard to predict, so making future estimates is inappropriate until the situation has stabilized and new data has been gathered.

We ran the model under two specifications. The first model specification imitates the reality in which the population continues to grow (12) and the fertility rates exhibit changes year over year consistent with historical trends. The second model specification fixes the population at 5,000 for all years in the simulation, to isolate the expected changes in clinical workload due to population structure shifts only. The analyses are based on simulated results from 100 trials for each model specification.

Even though fertility rate predictions are a key assumption in the model, it is uncertain whether future trends will mirror the recent history on which we have based our model parameters. Rates are known to change in response to a complex set of dynamics, including social norms, access to family planning, and girls’ education, and have been known to evolve in unexpected ways (17). To acknowledge this uncertainty about the future, we conducted sensitivity analysis to assess its impact on the predicted PHC workload. We reran the model using two different historical time periods for our parameters: 2011-19 and 2000-19, and a benchmark of moderate decline in fertility rates at 0.5% per year relative to baseline.

The analyses were completed using PACE-HRH (11) release version 1.0.2. The model was built in R Studio version 2022.07.2+576, using the R statistical programming language version 4.2.0. Code is publicly available on GitHub.

Regional variation in clinical workload

With population growth, the total clinical workload is predicted to rise in all regions in the 15 years from 2021 to 2035. Of the regions modeled, Addis Ababa had the lowest total clinical workload in 2021 at 2,930 hours per health post per year, increasing to 3,742 hours by 2035 (61-78 hours per week, assuming 48 workweeks per year). Gambela had the highest workload, increasing from 3,752 to 4,626 hours per health post per year. The gap between the regions with the highest and the lowest predicted clinical workload expands from 17 hours per week in 2021 to 19 hours per week in 2035 (28% and 25% of the lowest regional workload, respectively) (Figure 1, Panel A).

Across the regions, there are three general shapes to the workload curves, which reflect the combined effects of population growth and shifts in the population pyramid, partially mitigated by the anticipated decline in infectious diseases.

Addis Ababa displayed the most pronounced slowdown and workload was projected to stabilize by 2028.
Dire Dawa and Somali have decelerating growth rates and began to flatten in later years.
In the remaining regions, the workload predictions for Gambela, Harari, Oromia, and SNNPR rise at a modest rate in the first few years and then saw an acceleration of the growth in workload by 2035.

While the total increased, the workload on a per-capita basis is projected to decline in all regions. (Figure 1 Panel B) Within these per-capita declines, most regions decline consistently, except for Addis Ababa and Somali. In those two regions, the per-capita workload is predicted to rise initially in the first few years before eventually starting to decline. This reflects the combined effect of lower proportions of children under age 5 in year 2020 (8% in Addis, 11% in Somali, near 15% in other regions) and fertility rates that are either comparatively high or growing more rapidly, thus resulting in a near-term baby boom (Additional File 1, Table 1).

Overall, both the total level clinical workload and the shape of the growth curve varied by region. We explored the relative impact of the contributors to these differences, such as regional disease burden, initial demographics, and fertility rates, in the following sections.

Primary clinical service categories

To understand which services contributed most to the total clinical workload, we calculated the proportional contribution to the total by service category (a total of 12 categories, listed in Additional File 1, Table 5), by year (Figure 2). We find that pregnancy, NCDs and sick child care are the top three contributors in 2021 (except for in Somali) and remain in the top four in 2035. Local disease burden and demographics influence the variability of each service category’s rank across regions. For example, malaria ranks in the top six in the two regions (18) with the highest malaria incidence. Malnutrition is a higher priority concern in Somali due to food insecurity, compounded by a young age pyramid and high fertility.

The proportional contribution by service category changes over time based on expected trends in demographics and disease burden, resulting in shifts in the ranking. In Benishangul-Gumuz and Harari, malnutrition overtakes sick child care to become the third largest concern by 2035, primarily due to expected decline in incidence of childhood disease. Providing care for sick children was a large proportion of the workloads in Oromia and SNNPR in 2021, but its contribution dropped by a third by 2035. NCD-related workload rose in all regions, especially for Oromia and Dire Dawa, overtaking sick child and pregnancy-related work in some regions.

Impact of fertility trend assumptions

Fertility rates underpin the predictions for future population, and thus have a large impact on the results of the prediction model. At the national level, fertility rates are declining for all age groups, but this is not the case for individual regions. (Figure 3) Pastoral regions (e.g., Afar, Benishangul-Gumuz, and Somali) have the highest fertility rates. Addis Ababa, Afar, and Harari stand out for having an increasing trend in fertility rates for the most productive age groups.

Addis Ababa, the capital city, has a different fertility pattern than the rest of Ethiopia and the degree to which rates are changing has varied in recent years. Addis had the lowest current fertility rates in the country as of 2019 (e.g., 0.1 for women ages 20-24 (13)), but they rose in all age groups from 2000 to 2019 (19). Because of this unique combination, we reran the model with different plausible fertility trajectories (Figure 4).

Oromia has fertility rates significantly higher than Addis Ababa as of 2019 (e.g., 0.235 for women ages 20-24), but they decline in all age groups from 2000 to 2019. Oromia exhibits larger declines in fertility rates during the period 2011-2019.

We find that assumptions on fertility rates greatly affect our estimates of clinical workload in Addis, but less so in Oromia. If future trends correspond to those in the last 10 years, the total workload in Addis will be 50% higher by 2035 than if the trends correspond to the longer-term 20-year historical trend. The corresponding difference is 6% lower for Oromia.

Magnitude of seasonal variability in workload

We included seasonal variation in workload for births (19) (pregnancy-related services & immunization), diarrhea (20), malnutrition (21), malaria (22), and TB (23) and calculated the ratio of the highest month (maximum workload) to the average value. Figure 5 shows some regions should expect more variability than others; Somali had the most seasonal variation with the peak month 8.9% (90% CI 8.2%, 9.8%) above average in 2022. Most regions have declining demand for services affected by seasonality, which results in a slow decline in the calculated ratio. Additional File 1, Figure 2 displays variation in clinical workload by month of the year for each region.

Clinical hours for age-specific population

We examined clinical workload for age-specific population in the three most populous regions of Ethiopia: Oromia, Amhara, and SNNPR. For all three regions, infants and children under the age of 5 years are the groups predicted to accrue the highest proportion of clinical workload, accounting for a combined 57.2% of predicted clinical workload in Amhara, 62% in Oromia, and 64.5% in SNNPR (Figure 6, Panel A). The regions show slight differences in proportions of workload accrued by different age groups due to demographics and disease burden.

From 2021 to 2035, the per capita workload for children under 5 are predicted to decline due to expected decrease in incidence of childhood diseases (Figure 6, Panel B). During the reproductive ages from 15 to 49, the per capita workload is higher for female compared to male, explained by utilization of family planning services by females.

This analysis focused on primary healthcare in a low-income setting, which constrained the inclusion of clinical workload on the treatment of some leading causes of morbidity and mortality in population aged 65 and over (e.g. Alzheimer’s care, cancer treatments, and liver disease treatments).

In summary, we find that the total annual primary healthcare clinical workload varies by region by up to 28%, due to differences in population pyramid, fertility, and disease incidence. This will shift over time in regionally specific ways as demographics and disease burden evolve, with some regions stabilizing at a higher-than-current workload over 15 years, while others will continue to rise rapidly.

On net, our estimates show that for a given population in 2020, the workload will rise substantially across all regions by 2035. The increase in workload is predominantly due to the growing population (estimated to rise by 40% between 2020 and 2035, per UN estimates (12)).

The clinical workload is projected to decline on a per-capita basis because of the combined effect of the young population aging into their healthiest years and expected reductions in disease rates. However, this is dependent on continued declines in net fertility rates and decreasing levels of infectious diseases, which are not guaranteed to materialize.

The top three service categories were pregnancy-related services, non-communicable diseases, and sick child care, with sexual health, malnutrition and malaria being important as well in region-specific ways.

Several of these categories are affected by seasonal variability and planning based on average monthly clinical workload would underestimate staffing need by up to 10%. Underestimation for workload due to seasonality is more severe in regions with higher fertility rates (Somali) and regions with higher burdens in conditions subjected to seasonality (Benishangul-Gumuz, Somali, and SNNPR). Given this level of variation, we conclude that seasonality is an important element to consider during the healthcare workforce planning process, even if not all PHC services display a seasonal pattern. Additionally, in resource-constrained settings, tasks that are less time sensitive (e.g., community health education), could be intentionally shifted to lower-workload times of year (e.g., malaria off-season) in a strategic way to reduce the impact of seasonality on overwhelmed health workers.

Based on these analyses, we make three key observations. First, the regional PHC needs are closely linked to the local context of disease incidence rates and demographic characteristics and the combined effects of both contribute to non-trivial cross-region differences in both amounts of clinical time required and areas of attention. Second, the regional differences do not dissipate over time and attention needs to be paid to both the shared trends (e.g., the increasing burden in NCDs), and the localized trends (e.g., the increasing burden in malnutrition for Benishangul-Gumuz and Harari). Third, even if a declining per-capita workload does materialize, the population growth may make it difficult for the country to ‘keep up with’ demand for services.

As with any study, there are limitations. Primary here is that this exercise was intended to examine future trend and regional differences, but numeric values should be used with caution due to limited data availability, especially the lack of time and motion studies that reported data at the clinical tasks level. Additionally, data on regional disease incidence rates was lacking for health areas that have not historically been prioritized by Ethiopia’s health sector (24) or that go under-reported, such as leishmaniasis, cancer treatment, gender-based violence, and smoking cessation. Another challenge is projecting trends, especially for fertility and disease incidence, over the fifteen-year modeled time horizon. However, while future trends are never certain, this should not limit us from anticipating expected changes and planning ahead to meet those needs.

Future work is needed to refine and expand upon these results. First is that improved data on incidence rates, either through new collection or by making use of local surveillance records, would substantially improve the reliability of estimates for local planning. Second is the need to consider anticipated fertility trends and update models routinely as the future unfolds. It is important to have accurate census data on populations and the modeling points to the potential use of admin data on family planning use to track changes in more real-time. Third is that this modeling estimates needed capacity in order to inform capacity planning, but it does not report out what are the health or operational consequences when there is inadequate capacity available. For example, modeling could be used to estimate user waiting times, impact on care seeking rates, health worker burnout risk, and ultimately the health impact of missed opportunities to provide health services.

There are a few recommendations for policy makers who would like to learn from this study. Based on these analyses, we conclude that the regional variation in workload, which compounds over time, means that staffing equally based on the national average would result in inequitable access to care. We also conclude that while normative staffing ratios are useful for aggregated goal setting, they are inadequate for localized planning because they do not adequately consider the contextual differences in health needs, and how those evolve over time. Thus, health systems would be better able to ensure adequate resourcing if planning is based on population-aware workload projections.

WHO = World Health Organization

HRH = Human Resources for Health

PACE-HRH = Population-aware capacity estimator for human resources for health

ASHA = Accredited Social Health Activist

PHC = Primary Healthcare Center

HEP = Health Extension Program

NCDs = Non-Communicable Diseases

ANC = Antenatal Care

TB = Tuberculosis

UN = United Nations

WDI = World Development Indicators

SNNPR = Southern Nations, Nationalities and Peoples’ Region

DHS = Demographics and Health Surveys

HIV = Human Immunodeficiency Virus

BMGF = Bill & Melinda Gates Foundation

Ethics approval: Not applicable.

Consent for publication: Not applicable.

Availability of data and materials: All code used to run the models is publicly available on GitHub at https://institutefordiseasemodeling.github.io/PACE-HRH/articles/pacehrh.html and model input and results csv files are available upon request.

Competing interests: The authors declare that they have no competing interests.

Funding: This work was completed with non-specific funding from the Bill & Melinda Gates Foundation (BMGF). All authors were paid by the BMGF for the duration of the modeling, analysis, and manuscript writing.

Authors’ contributions: BH led the design of the study, results analysis, and the writing of the manuscript. RH was primarily responsible for collecting the input data and running the models. KM contributed to the formulation of the study and was a key editor of the manuscript. All authors read and approved the final manuscript.

Acknowledgements: The authors would like to thank Charles Eliot and Meikang Wu for their continued support of the code base for PACE-HRH that was used for this study. We would like to thank Mandy Izzo for her editorial review of the manuscript. We thank Jeremy Cooper for aggregating the population data from the UN to initialize the model. We would like to thank Nicholas Leydon and Susna De for their support and for asking the hard questions that led to the initiation of a sub-national analysis.

World Health Organization. Everybody’s business — Strengthening health systems to improve health outcomes: WHO’s framework for action. [Internet]. Geneva; 2007 [cited 2022 Nov 16]. Available from: https://apps.who.int/iris/handle/10665/43918
World Health Organization. Monitoring the building blocks of health systems: a handbook of indicators and their measurement strategies [Internet]. 2010 [cited 2023 Jan 31]. Available from: https://apps.who.int/iris/bitstream/handle/10665/258734/9789241564052-eng.pdf
World Health Organization. The world health report 2006: working together for health. 2006 [cited 2022 Nov 20]; Available from: https://apps.who.int/iris/handle/10665/43432
Kingdom of Swaziland Ministry of Health. Human Resources for Health Strategic Plan 2012-2017. 2012 Oct.
The Republic of Rwanda Human Resource for Health Secretariat. 10-Year Government Program: National Strategy for Health Professions Development 2020-2030. Kigali; 2020 Dec.
Government of India Ministry of Health. Guidelines on the Accredited Social Health Activists [Internet]. 2005 [cited 2022 Oct 15]. Available from: https://nhm.gov.in/images/pdf/communitisation/task-group-reports/guidelines-on-asha.pdf
World Health Organization. Primary health care systems (‎primasys)‎: case study from Nigeria [Internet]. 2017 [cited 2022 Oct 15]. Available from: https://apps.who.int/iris/handle/10665/341137
Misganaw A, Naghavi M, Walker A, Mirkuzie AH, Giref AZ, Berheto TM, et al. Progress in health among regions of Ethiopia, 1990–2019: a subnational country analysis for the Global Burden of Disease Study 2019. The Lancet. 2022 Apr 2;399(10332):1322–35.
Federal Ministry of Health of Ethiopia. Realizing universal health coverage through primary health care: A roadmap for optimizing the Ethiopian Health Extension Program, 2020-2035, 1st edition [Internet]. Addis Ababa; 2020 Jul [cited 2022 Oct 12]. Available from: https://e-library.moh.gov.et/library/wp-content/uploads/2021/07/1.-HEP-Roadmap_Main-Document_Final.pdf
Hagedorn B, Han R, Eliot C, Wu M, Schripsema J, McCarthy K. Implementation and applications of PACE-HRH, a stochastic workload model of human resources for health [Internet]. 2023 [cited 2023 Feb 13]. Available from: https://doi.org/10.21203/rs.3.rs-2515066/v1
Hagedorn B, Eliot C. PACE-HRH: Population-Aware Capacity Estimator for Human Resources for Health [Internet]. 2022 [cited 2022 Oct 27]. Available from: https://github.com/InstituteforDiseaseModeling/PACE-HRH
Population Division - United Nations. World Population Prospects [Internet]. [cited 2022 Oct 12]. Available from: https://population.un.org/wpp/
Ethiopian Public Health Institute (EPHI) [Ethiopia], ICF. Ethiopia Mini Demographic and Health Survey 2019 [Internet]. 2021 [cited 2022 Oct 12]. Available from: https://dhsprogram.com/publications/publication-FR363-DHS-Final-Reports.cfm
World Bank. World Development Indicators [Internet]. 2021 [cited 2022 Oct 27]. Available from: https://data.worldbank.org/indicator/
GEOGLAM. Northern Ethiopia: Conflict and Food Insecurity [Internet]. 2022 Sep [cited 2022 Oct 19]. Available from: https://reliefweb.int/report/ethiopia/northern-ethiopia-conflict-and-food-insecurity-updated-september-27th-2022
Magome M, Anna C. Ethiopian govt, Tigray agree to end fighting after 2 years. AP News [Internet]. 2022 Nov 2 [cited 2022 Nov 21]; Available from: https://apnews.com/article/africa-south-ethiopia-african-union-70fb0d185aaccb668b2fabb4f6e45b9f
Bohk-Ewald C, Li P, Myrskylä M. Forecast accuracy hardly improves with method complexity when completing cohort fertility. Proceedings of the National Academy of Sciences [Internet]. 2018 Sep 11 [cited 2022 Nov 22];115(37):9187–92. Available from: https://www.pnas.org/doi/abs/10.1073/pnas.1722364115
Tadesse Y, Irish SR, Chibsa S, Dugassa S, Lorenz LM, Gebreyohannes A, et al. Malaria prevention and treatment in migrant agricultural workers in Dangur district, Benishangul-Gumuz, Ethiopia: social and behavioural aspects. Malar J [Internet]. 2021 Dec 1 [cited 2023 Feb 16];20(1):1–18. Available from: https://malariajournal.biomedcentral.com/articles/10.1186/s12936-021-03766-3
USAID Demographic and Health Surveys Program. The DHS Program - Ethiopia [Internet]. [cited 2022 Oct 12]. Available from: https://dhsprogram.com/Countries/Country-Main.cfm?ctry_id=65&c=Ethiopia
Beyene H, Deressa W, Kumie A, Grace D. Spatial, temporal, and spatiotemporal analysis of under-five diarrhea in Southern Ethiopia. Trop Med Health [Internet]. 2018 Jun 4 [cited 2022 Nov 14];46(1):1–12. Available from: https://tropmedhealth.biomedcentral.com/articles/10.1186/s41182-018-0101-1
Miller R. Childhood Health and Prenatal Exposure to Seasonal Food Scarcity in Ethiopia. World Dev. 2017 Nov 1;99:350–76.
Midekisa A, Beyene B, Mihretie A, Bayabil E, Wimberly MC. Seasonal associations of climatic drivers and malaria in the highlands of Ethiopia. 2015;
Gashu ZI, Jerene D, Datiko DG, Hiruy N, Negash S, Melkieneh K, et al. Seasonal patterns of tuberculosis case notification in the tropics of Africa: A six-year trend analysis in Ethiopia. 2018 [cited 2022 Nov 14]; Available from: https://doi.org/10.1371/journal.pone.0207552
Federal Ministry of Health of Ethiopia. Essential Health Services Package for Ethiopia. Addis Ababa; 2005 Aug.
Federal Ministry of Health of Ethiopia. An Implementation Manual for Optimizing Health Extension Program. Addis Ababa; 2021 Mar.

Additionalfile1.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

One size does not fit all: an application of stochastic modeling to estimating primary healthcare needs in Ethiopia at the sub-national level

Status:

Version 1

Abstract

Figures

Introduction

Methods

Results

Discussion

Conclusion

Abbreviations

Declarations

References

Supplementary Files

Status:

Version 1