Number of new HIV/AIDS infections (WHO)
- Rows
- 3,759
- Columns
- 4
- Data range
- 1990-01-01 to 2024-01-01
- Auto-refresh
- Re-checks its source every 60 days
- Last refreshed with new data
- 2026-10-08
New infections time-series from WHO sources, country-level annual data.
AI analysis
Dataset Analysis: New HIV/AIDS Infections (WHO)
1. Dataset Overview & Value This dataset, sourced from Our World in Data ([ourworldindata.org](https://ourworldindata.org/grapher/number-of-new-hivaids-infections-who.csv?v=1&csvType=full&useColumnShortNames=false)), captures annual new HIV/AIDS infection counts at the country level across 148 entities and 34 years, yielding 3,573 rows of longitudinal public health data. Our World in Data is a well-regarded publisher that aggregates WHO source data, lending credibility and methodological consistency to the figures. The time-series structure makes this dataset particularly valuable for trend analysis, policy impact studies, and cross-country epidemiological comparisons — especially for examining whether healthcare interventions, antiretroviral therapy (ART) availability, or socioeconomic shifts correlate with changes in infection rates over time.
2. Data Quality Observations Data quality is notably strong for a public health dataset of this scope. Total null cells across all four columns are zero, which is exceptional and suggests either thorough imputation by the WHO/OWID pipeline or careful curation of only well-documented country-year pairs. No type mismatches are flagged. The duplicate row count is pending Phase D recomputation, which should be verified before any aggregation analysis — even a small number of duplicates on a (Entity, Year) composite key could distort time-series totals. The 148 distinct values in both Entity and Code columns align perfectly, suggesting no orphaned or mismatched country codes, a common data quality pitfall in multi-source global datasets.
3. Key Column Distributions The Number of new HIV infections column tells a striking epidemiological story through its statistics alone. The mean of ~55,425 dwarfs the median of 1,700, immediately signaling severe right skew — confirmed by a skewness coefficient of 7.74. The interquartile range (Q1=500, Q3=9,700) shows that the bulk of country-year observations represent relatively small absolute infection counts, while a small number of high-burden nations (likely Sub-Saharan African countries at epidemic peaks) drive the distribution toward its maximum of 3.3 million. The 565 statistical outliers (~15.8% of all rows) are unusually numerous and represent genuinely extreme public health situations rather than data errors — they deserve their own analytical treatment rather than removal. The Year column spans 34 distinct years, enabling robust longitudinal modeling.
4. Recommended Join Key Columns The natural composite join key is (Entity + Year) or equivalently (Code + Year). Code — which appears to be ISO 3166-1 alpha-3 country codes given its 1:1 mapping with Entity — is the preferred technical join key for cross-dataset merges, as it avoids entity name inconsistencies (e.g., "United States" vs. "USA"). Year should be treated as a range-bounded integer join (noting the 34-year span) when linking to datasets with different temporal resolutions. Analysts should confirm whether Entity includes any non-country aggregates (e.g., "World," regional groupings) that would need filtering before country-level joins.
5. Recommended Companion Datasets for Correlation Analysis Several dataset pairings would unlock high-value correlations:
- Antiretroviral therapy (ART) coverage rates — the most direct policy lever; testing whether ART scale-up correlates with declining new infections is a core epidemiological question - Healthcare expenditure per capita (World Bank) — to explore whether health system investment predicts infection trajectory - GDP per capita / poverty rates — socioeconomic determinants of HIV vulnerability are well-documented; quantifying this relationship cross-nationally adds analytical weight - Condom use / sexual health education metrics — behavioral intervention datasets from UNAIDS or DHS surveys would support causal pathway analysis - HIV/AIDS mortality or prevalence data (also available via OWID/WHO) — pairing incidence with mortality enables case fatality and treatment effectiveness modeling across the same Entity/Year keys
Columns
- Year (date)
- Entity (string)
- Code (string)
- Number of new HIV infections (integer)