Factors Associated with Outcomes of Inpatient Severe Malaria Cases in the Ashanti Region, Ghana: An Analytic Cross-sectional Study using Routine Surveillance Data, 2018 to 2022.

This analytic cross-sectional study of 54,544 severe malaria admissions in Ghana's Ashanti Region (2018–2022) identifies that mortality is significantly influenced by age, comorbidities, insurance status, and facility type, highlighting the need for targeted interventions such as improved NHIS enrollment, extended inpatient monitoring, and enhanced comorbidity management to reduce preventable deaths.

The Gist

Imagine the Ashanti Region of Ghana as a bustling, busy city where a dangerous, invisible storm called Malaria frequently hits. While the city has a strong defense system (hospitals and health workers), sometimes the storm is so fierce that it overwhelms people, forcing them into the hospital.

This study is like a detective report written by health investigators. They looked at the records of 54,544 people who were hospitalized with severe malaria between 2018 and 2022. Their goal? To figure out who survived the storm and who didn't, and why.

Here is the breakdown of their findings, translated into everyday language:

1. The Big Picture: A Rare but Deadly Storm

Out of every 1,000 people admitted to the hospital with severe malaria, only 4 died. That sounds low, but in the world of health, that's still a significant number of lives lost. The researchers wanted to know: What makes the difference between a patient walking out of the hospital and a patient not making it?

2. The "Survival Cheat Codes" (What Helped People Live)

The study found several "superpowers" or protective factors that increased a patient's chances of survival:

• The "Insurance Shield" (NHIS): Think of Ghana's National Health Insurance Scheme (NHIS) as a golden ticket. Patients with active insurance were 67% less likely to die than those without. Why? Because the ticket removes the fear of cost. It means they can get to the hospital faster, stay longer if needed, and get the best medicine without worrying about the bill.
• The "Age Advantage": Children under 5 are like tender saplings; their immune systems are still growing, making them very vulnerable. However, kids aged 5 to 17 were much tougher. They had developed some "armor" (immunity) and were better at telling adults when they felt sick, leading to earlier treatment.
• The "Faith-Based Fortress": Surprisingly, patients treated in Faith-Based hospitals (run by churches or religious groups) had much lower death rates than those in government hospitals. It's as if these facilities had a secret recipe for care—perhaps better organization, more attentive staff, or a different way of handling patients.
• The "Student Status": Students had a much better survival rate than unemployed people. This isn't because being a student is a magic cure, but likely because students are often supported by families or have better access to information and care compared to those struggling financially.

3. The "Danger Zones" (What Made Things Worse)

Some factors acted like speed bumps or traps that made survival harder:

• The "Comorbidity Backpack": Imagine a patient carrying a heavy backpack full of other health problems (like anemia, diabetes, or infections) on top of their malaria. These patients were twice as likely to die. The malaria was fighting the body, but the other diseases were fighting it too, leaving the patient exhausted and overwhelmed.
• The "Medical Ward Maze": Patients admitted to the Medical Ward or the Casualty (Emergency) room had higher death rates than those in the Pediatric (Children's) Ward. This doesn't mean the doctors in the Medical Ward were bad; rather, it suggests that the people ending up there were often sicker to begin with, or arrived later in the game when the disease had already done more damage.
• The "Short Stay Trap": Patients who stayed in the hospital for less than 3 days had higher death rates. Warning: This is tricky! It doesn't mean "staying longer cures you." It's more like a reverse causality: If a patient is going to die, they often die quickly (within 1 or 2 days). So, a short stay is often a sign of a very severe case that ended tragically, not a sign that they were sent home too soon.

4. The "Gender Twist"

The study found that men were slightly more likely to die than women. The researchers suspect this might be because men are sometimes "tough guys" who wait too long to go to the doctor, or perhaps they are exposed to mosquitoes in different ways (like working late at night in fields).

The Takeaway: How to Build a Better Defense

The researchers aren't just pointing out problems; they are handing the government a blueprint for saving lives:

1. Fix the Insurance: Make sure everyone has that "golden ticket" (NHIS). If people can't afford care, they wait too long, and the storm gets too strong.
2. Check the Backpack: When a patient comes in with malaria, doctors must immediately check for other diseases (like anemia or infections) and treat them at the same time. You can't win a battle if you are fighting on two fronts without help.
3. Learn from the Best: The government should send teams to study the Faith-Based hospitals and Pediatric wards to see what they are doing right, and then teach those tricks to everyone else.
4. Watch the First Few Days: Since many deaths happen quickly, the first 3 days in the hospital are critical. Patients need to be watched like hawks during this time.

In a nutshell: Malaria is a fierce enemy, but with the right insurance, early detection, treating all health issues at once, and learning from the best hospitals, the Ashanti Region can turn the tide and save many more lives.

Technical Summary

1. Problem Statement

Despite intensified malaria control efforts in Ghana, the Ashanti Region bears a disproportionate burden of severe malaria, accounting for 24% of national hospital admissions in 2021 (compared to a national average of 19%). While general malaria incidence is monitored, the specific determinants of inpatient mortality for severe malaria cases in this region remain poorly understood. Previous studies have identified broad factors (age, education, insurance), but context-specific drivers within the Ashanti Region's health system, particularly regarding facility types, admission departments, and comorbidities, require deeper investigation to inform targeted interventions.

2. Methodology

• Study Design: Analytic cross-sectional study utilizing secondary data.
• Data Source: Routine inpatient surveillance data from the District Health Information Management System 2 (DHIMS2) covering the period from January 1, 2018, to December 31, 2022.
• Study Setting: The Ashanti Region of Ghana, encompassing 43 Metropolitan, Municipal, and District Assemblies (MMDAs) and 65 health facilities.
• Population: All patients of all ages diagnosed with severe malaria (based on Ghana Standard Treatment Guidelines aligned with WHO criteria) and admitted to health facilities in the region.
• Inclusion/Exclusion:
  • Included: 54,544 records of confirmed severe malaria admissions with known outcomes.
  • Excluded: Records with missing outcomes, unspecified outcomes, absconded patients, or those transferred out (total exclusions: 825 from an initial 55,379).
• Variables:
  • Outcome: Admission outcome (Died vs. Did not die).
  • Predictors: Demographics (sex, age, education, occupation), Socio-economic factors (NHIS status, residence), Facility characteristics (ownership, level of care, admitting department), and Clinical variables (length of stay, comorbidities).
• Statistical Analysis:
  • Descriptive & Bivariate: Frequencies, percentages, and Chi-square tests.
  • Multivariable Modeling: Given the rare outcome (Case Fatality Rate < 5%), Firth's penalized likelihood logistic regression was employed to reduce bias and handle separation issues.
  • Survey Design Adjustment: Robust survey estimation (using svy commands in Stata) accounted for the clustering of patients within health facilities.
  • Sensitivity Analysis: Survey-adjusted logistic regression was performed to validate findings.
  • Model Diagnostics: Variance Inflation Factors (VIF) for multicollinearity, AIC/BIC for model fit, and ROC/AUC for discrimination.

3. Key Results

• Demographics & Descriptive Statistics:

  • Total analyzed cases: 54,544.
  • Case Fatality Rate (CFR): 0.4% (200 deaths).
  • Gender: 51.1% Female; 48.9% Male.
  • Age: 39.4% were children under 5; 31.4% were aged 5–17.
  • Insurance: 85.4% had active National Health Insurance Scheme (NHIS) membership.
  • Facility: 81.4% admitted to government facilities; 98.7% to primary-level facilities.
  • Comorbidities: 42.5% of patients presented with at least one comorbidity (most common: anaemia, septicaemia, respiratory disorders).

• Multivariable Analysis (Firth's Penalized Logistic Regression):
  Significant independent predictors of mortality included:

  • Sex: Males had 1.4 times higher odds of death compared to females (aOR = 1.39; 95% CI: 1.04–1.86).
  • Age: Patients aged 5–17 had 44% lower odds of death compared to children <5 (aOR = 0.56; 95% CI: 0.33–0.94).
  • Occupation: Students had 70% lower odds of death compared to the unemployed (aOR = 0.30; 95% CI: 0.16–0.58).
  • NHIS Status: Active membership was associated with 67% lower odds of death compared to inactive membership (aOR = 0.33; 95% CI: 0.25–0.45).
  • Facility Ownership: Admissions to faith-based facilities had 62% lower odds of death compared to government facilities (aOR = 0.38; 95% CI: 0.23–0.65).
  • Admitting Department:
    • Medical wards had 2.38 times higher odds of death compared to paediatric wards (aOR = 2.38; 95% CI: 1.48–3.84).
    • Casualty/Emergency had 5.46 times higher odds of death (aOR = 5.46; 95% CI: 2.48–11.99).
  • Length of Stay: Stays of 3–5 days were associated with 33% lower odds of death compared to stays <3 days (aOR = 0.67; 95% CI: 0.47–0.97). Note: Authors caution this may reflect reverse causality (early deaths have shorter stays).
  • Comorbidities: Presence of comorbidities doubled the odds of death (aOR = 2.30; 95% CI: 1.72–3.08).

• Model Performance:

  • The model showed good fit (AIC = 2295.48) and acceptable discrimination (AUC = 0.81).
  • Sensitivity analysis confirmed the direction and significance of most covariates, though sex and length of stay lost statistical significance in the survey-adjusted model.

4. Key Contributions

1. First Regional Analysis: This is the first study to utilize DHIMS2 data to specifically analyze inpatient severe malaria outcomes and mortality determinants within the Ashanti Region.
2. Methodological Rigor: The application of Firth's penalized logistic regression addresses the statistical challenge of rare events (low mortality rate), providing more stable and unbiased estimates than standard logistic regression.
3. Identification of Systemic Gaps: The study highlights that mortality is not just a clinical issue but is strongly influenced by facility type (faith-based vs. government), admission department (paediatric vs. medical/casualty), and insurance status.
4. Comorbidity Focus: It quantifies the significant impact of comorbidities (specifically anaemia and systemic infections) on mortality, suggesting a need for integrated care protocols.

5. Significance and Recommendations

The study provides critical evidence for policymakers and health administrators in Ghana to refine malaria control strategies:

• Policy Intervention: Strengthening NHIS enrollment and ensuring active coverage is vital, as it is a strong protective factor against mortality.
• Clinical Protocols:
  • Comorbidity Screening: Guidelines must integrate routine screening and management of comorbidities (e.g., anaemia, sepsis) for all severe malaria admissions.
  • Departmental Benchmarking: The superior outcomes in paediatric and faith-based facilities suggest that best practices from these units should be benchmarked and adopted by medical and casualty departments.
• Resource Allocation: Facilities with high volumes of older patients or complex comorbidities require targeted training, diagnostics, and supplies.
• Monitoring: Extending inpatient monitoring beyond the first 3 days is crucial, as early discharge or premature death skews data; close monitoring during the initial 3–5 days is essential for survival.

Limitations: The study is cross-sectional (limiting causal inference), relies on routine data which lacks granular clinical severity markers (e.g., parasite density, consciousness level), and excludes the major referral center (Komfo Anokye Teaching Hospital), potentially affecting generalizability.