**Background:** Hip fracture is the most severe fragility fracture, associated with high morbidity, mortality, and societal costs. Accurate estimation of hip fracture burden is essential for healthcare planning and risk prediction. Previous projections have forecasted steep increases in hip fracture numbers, but these may be biased by overcounting due to readmissions, transfers, and complications in national databases. This study aimed to estimate the true incidence of hip fractures in Sweden from 1998 to 2019 using a validated algorithm, assess time trends, and project future burden to 2050.
**Methods:** The study used the Swedish National Patient Register (NPR) covering all inpatient care from 1998 to 2019, linked to the Cause of Death Register and population data from Statistics Sweden. Incident hip fractures were identified using a previously validated prediction model (96% sensitivity and specificity) that distinguishes new fractures from readmissions, transfers, and complications based on age, sex, admission type, time interval, diagnosis codes, and department type. The algorithm was applied to individuals > 50 years. Two comparison methods were used: a naïve approach counting all admissions with hip fracture codes (ICD-10 S720, S721, S722), and official Swedish statistics (3-year moving averages). Age-standardized rates were calculated using direct standardization to the 2019 population. Projections to 2050 were made under three scenarios: (1) continued annual decline at the 1998–2019 average rate; (2) constant rates from 2019; (3) decline through 2029 then no further change.
**Key Results:** Over 22 years, the algorithm identified 335,399 incident hip fractures (234,475 women, 100,924 men), compared to 539,947 by naïve analysis (56% overestimate in women, 72% in men) and 367,697 by official statistics (9–12% higher). The absolute number of hip fractures declined from 16,180 in 1998 to 13,929 in 2019 (14% decrease), driven entirely by a 21% reduction in women; men had a slight increase. Age-standardized incidence declined by 29.2% (95% CI 28.1–30.2%) in women and 29.3% (95% CI 27.5–30.7%) in men, corresponding to about 6,000 fewer fractures in 2019 than if the 1998 age-sex structure had persisted. The 10-year cumulative incidence of a second hip fracture was 11% (vs. 35% by naïve analysis), and 80% of patients had died within 10 years. Projections: Under scenario 1 (continued decline), hip fractures would fall to ~12,000 in 2050; under scenario 2 (constant rates), they would rise to ~22,000; under scenario 3 (decline to 2029 then stable), ~18,500 in 2050. For 2034, estimates were 14,380, 19,200, and 16,300 respectively.
**Clinical Implications:** The study demonstrates that hip fracture incidence in Sweden has declined substantially over two decades, more than offsetting population aging. This contradicts earlier projections of rising burden. The overestimation of hip fractures in naïve analyses and official statistics has important consequences: the widely used FRAX tool may overestimate 10-year hip fracture risk (e.g., 20–22% for an 80–85-year-old woman with prior fracture, vs. actual ~10% second fracture risk). Accurate incidence data are critical for fracture risk calculators, health economic analyses, and resource allocation. The findings also highlight the need for validated algorithms to identify incident cases in administrative databases, not only for hip fractures but also for other conditions like myocardial infarction and stroke. The decline in hip fracture rates may be attributable to healthier lifestyles, reduced smoking, increased body weight, and improved cardiovascular health, rather than bone-specific drug use (which remained low). The study's limitations include potential residual misclassification (though <0.5% had >2 fractures) and the inability to identify specific causal factors. Overall, the results call for updated, validated incidence data to inform clinical practice and public health policy.