Spatio-temporal variation in the uptake of the Human Papilloma Virus (HPV) vaccine among Malawian girls between 2019 and 2024

This study analyzes spatio-temporal HPV vaccination data from 2019 to 2024 in Malawi, revealing that coverage for both doses remains significantly below WHO targets with notable regional disparities and a counterintuitive negative association with urbanization, thereby highlighting the urgent need for decentralized, targeted interventions to accelerate cervical cancer elimination.

The Gist

Imagine Malawi as a large, vibrant garden where the goal is to protect every young girl from a dangerous weed called cervical cancer. The best way to stop this weed is to spray a special protective mist called the HPV vaccine.

This paper is like a gardener's report card from 2019 to 2024. It asks two big questions:

1. How many girls got the mist? (Did we spray enough?)
2. Why are some parts of the garden getting sprayed while others are dry? (Where are the gaps?)

Here is the story of their findings, broken down into simple parts:

1. The Big Picture: A Garden That Needs More Water

The researchers looked at data from all 28 districts (regions) of Malawi. They found that while they started strong, the spraying has been inconsistent.

• The Goal: The World Health Organization (WHO) wants 90% of girls to get the vaccine.
• The Reality: On average, only about 47% of girls got the first spray, and only 32% got the second one.
• The Analogy: Imagine trying to water a field with a hose, but you only manage to wet about one-third of the grass. Some patches are soaking wet, but many are still dry and vulnerable to the weeds.

2. The Rollercoaster Ride (Time)

The coverage didn't stay steady; it went up and down like a rollercoaster.

• 2019–2020: The gardeners were enthusiastic! They sprayed a lot, and coverage was high.
• 2021–2022: Then came the "Great Pause" (the COVID-19 pandemic). Schools closed, and the spraying stopped. The coverage dropped sharply, like a bucket with a hole in it.
• 2023–2024: They tried to pick up the pace again, but they haven't quite reached the high levels they had before the pause.

3. The Surprising Clues (Why is it happening?)

The researchers used a special "detective lens" (statistical models) to see what factors made the spraying better or worse. Here are the most interesting clues they found:

• The "City vs. Country" Twist: Usually, we think cities have better access to medicine. But in Malawi, the opposite happened.

  • The Analogy: Think of Nighttime Lights as a measure of how "busy" and urban an area is. The study found that brighter, busier cities actually had lower vaccination rates.
  • Why? Maybe the school-based programs (where most girls get the shot) don't reach girls in crowded city neighborhoods as well as they reach girls in quiet villages. Or perhaps city parents are more skeptical about the vaccine.

• The "Road" Paradox: You might think being far from a road makes it harder to get a vaccine.

  • The Finding: Surprisingly, areas farther from roads had higher second-dose coverage.
  • Why? This suggests that in remote areas, the health workers might be going out of their way to do "mobile outreach" (bringing the vaccine to the people), whereas in easy-to-reach areas, people might just assume they will get it later and forget.

• The "Health Crisis" Connection:

  • Areas with higher HIV rates actually had better HPV vaccination.
  • The Analogy: It's like a neighborhood that is already fighting a big fire (HIV) has better fire drills and emergency plans. Because they are already working hard to treat HIV, they are also better at catching girls for the HPV vaccine.
  • Conversely, areas with more teenage pregnancies had lower vaccination. This suggests that girls who are already pregnant or sexually active might be missed by the school-based programs or feel excluded.

• The Weather Factor: Heavy rain, high heat, and thick vegetation seemed to make it harder to get the vaccine to girls.

  • The Analogy: If the roads are muddy or the fields are too overgrown, the "vaccine trucks" can't get through, or the health workers can't reach the schools.

4. The Takeaway: How to Fix the Garden

The authors conclude that we can't just spray the whole country the same way and hope for the best. We need a smarter strategy:

1. Go Where the Dry Patches Are: Don't just spray the easy parts. We need to find the specific districts that are consistently dry (like Balaka, Karonga, and Nsanje) and focus extra effort there.
2. Fix the City Problem: Since cities are lagging, we need new ways to reach girls there, not just relying on schools. Maybe we need mobile clinics that drive right into the neighborhoods.
3. Keep the Momentum: We need to make sure that when things get busy (like during a pandemic or bad weather), we have a backup plan to keep the spraying going.
4. Talk to the Community: We need to explain to parents in cities and in areas with high teenage pregnancy rates why this vaccine is important, so they don't miss out.

In short: Malawi has a great vaccine, but the delivery system is like a leaky hose. By understanding where and why the leaks are happening (cities, remote areas, bad weather), the health workers can patch the holes and finally water the whole garden, protecting all girls from cancer.

Technical Summary

1. Problem Statement

Malawi faces a severe public health crisis regarding cervical cancer, which is caused by the Human Papillomavirus (HPV). The country holds the world's highest age-standardized incidence rate (67.9 per 100,000 women) and mortality rate for cervical cancer. While the World Health Organization (WHO) has set a target of 90% HPV vaccination coverage for adolescent girls by age 15, Malawi's national program, launched in January 2019, has struggled to meet this goal.

Previous research in Malawi has been limited to sub-national case studies (specific districts like Dowa or Zomba), lacking a comprehensive, nationally representative analysis. There is a critical gap in understanding the spatio-temporal heterogeneity of vaccine uptake across all 28 districts and identifying the specific environmental, epidemiological, and temporal covariates driving these variations.

2. Methodology

Data Sources and Study Population:

• Population: Female adolescents aged 9–14 years across all 28 districts of Malawi.
• Timeframe: 2019 to 2024 (covering pre-pandemic, pandemic, and post-pandemic periods).
• Data: Routinely collected administrative data from the Malawi Ministry of Health (MOH) regarding first and second HPV vaccine doses.
• Data Cleaning: Coverage values exceeding 100% (observed in 6% of first-dose and 19% of second-dose records) were capped at 99% to correct for reporting errors common in routine surveillance.

Covariates (Predictors):
The study integrated a diverse set of geospatial covariates to explain coverage patterns:

• Socio-economic/Access: Nighttime light intensity (NTL) as a proxy for urbanization/development; distance to the nearest road.
• Climate/Environmental: Annual mean precipitation, maximum temperature, vegetation index (NDVI), and elevation.
• Epidemiological/Behavioral: District-level HIV prevalence and proportion of teenage pregnancies (sourced from DHS and MPHIA surveys).
• Temporal: Categorized into Initial (2019–2020), Intermediate (2021–2022), and Late (2023–2024) periods.

Statistical Modeling:

• Framework: Spatio-temporal Bayesian models implemented using R-INLA (Integrated Nested Laplace Approximation).
• Model Structure: The number of vaccinated girls ($y_{it}$) was modeled as a Binomial random variable with a logit link function:
  $$\text{logit}(p_{it}) = d(x_{it})\beta + u_i + \delta_{it} + f_t$$
  Where $\beta$ represents fixed effects, $u_i$ is an unstructured district random effect, $\delta_{it}$ is a district-year interaction, and $f_t$ is the temporal effect.
• Temporal Functional Forms: Several non-linear temporal priors were tested to capture complex trends:
  • Broken-stick (First-order Random Walk, RW1).
  • Penalized Spline (Second-order Random Walk, RW2).
  • Cubic Smoothing Spline with Penalized Complexity (PC) priors.
• Model Selection: Models were compared using the Widely Applicable Information Criterion (WAIC).
• Spatial Autocorrelation: Assessed via the Empirical Bayes Index (EBI). Results indicated negligible residual spatial correlation, leading to the exclusion of structured spatial random effects (e.g., ICAR/BYM) in the final models.

3. Key Contributions

• First National Analysis: This is the first study to provide a nationally representative, district-level assessment of HPV vaccine uptake in Malawi since the program's inception in 2019.
• Methodological Rigor: It employs advanced spatio-temporal Bayesian modeling to disentangle the effects of environmental, climatic, and temporal factors on vaccination coverage, moving beyond simple descriptive statistics.
• Counter-Intuitive Findings: The study challenges the assumption that urbanization always correlates with better health service access, revealing specific negative associations in the Malawian context.

4. Key Results

Overall Coverage:

• First Dose: Average uptake was 46.83% (95% CrI: 46.52%–47.21%).
• Second Dose: Average uptake was significantly lower at 32.44% (95% CrI: 32.09%–32.96%).
• Trend: Coverage peaked in 2019–2020, dropped sharply during the 2021–2022 pandemic period, and showed partial recovery in 2023–2024, though it did not return to initial levels. No district consistently met the 90% WHO target.

Significant Covariates (Posterior Means):

• Nighttime Lights (Urbanization): A strong negative association with uptake. Higher NTL (urban areas) correlated with lower vaccination rates for both doses (Dose 1: -0.599; Dose 2: -2.164).
• Distance to Roads: Surprisingly, greater distance to the nearest road was associated with higher uptake for the second dose (1.179), suggesting rural areas may have been better reached by specific outreach campaigns than urban townships.
• HIV Prevalence: Higher district-level HIV prevalence was positively associated with second-dose uptake (2.440), likely due to the integration of HPV services into HIV care programs.
• Teenage Pregnancy: Higher rates of teenage pregnancy were negatively associated with second-dose uptake (-1.589), potentially due to girls being perceived as ineligible or excluded from school-based campaigns.
• Climate: Increased precipitation and vegetation index were negatively associated with uptake, likely due to accessibility issues during rainy seasons.

Model Performance:

• The best-fitting model for the first dose included categorical time effects; for the second dose, a model without an explicit temporal function performed best.
• Validation showed high predictive accuracy (Correlation > 0.99) with low error rates (RMSE < 0.035).

5. Significance and Implications

Policy Implications:

• Decentralized Planning: The significant subnational heterogeneity necessitates district-specific strategies rather than a "one-size-fits-all" national approach.
• Urban Focus: The negative correlation with urbanization suggests that urban townships are "coldspots" for vaccination. Interventions must target these areas specifically, potentially addressing vaccine hesitancy or logistical barriers in dense urban settings.
• Integration: The positive link between HIV prevalence and uptake supports the strategy of integrating HPV vaccination into existing HIV and reproductive health services.
• Pandemic Recovery: The data confirms a severe disruption during the pandemic, highlighting the need for robust catch-up campaigns and mobile outreach to recover lost coverage, particularly for the second dose.

Conclusion:
The study concludes that while Malawi has made progress, HPV vaccination coverage remains critically low. Achieving the WHO 90% target requires targeted, context-specific interventions that address the unique barriers in urban centers, leverage successful integration with HIV programs, and utilize mobile outreach to overcome climatic and geographic challenges.