Aim: To analyse the rising prevalence of cerebral palsy (CP) in children born preterm in Denmark.Method: We included all live‐born children born preterm in Denmark from 1997 to 2013. The prevalence of CP in children born preterm was categorized by gestational age and correlated with neonatal mortality and changes in clinical factors.Results: Among 70 876 children, 824 (1.2%) had CP. The overall CP prevalence in children born preterm decreased substantially until 2001, from when it increased annually by 2.8% (95% confidence interval 0.6–5.0). When categorized, the prevalence only increased significantly in children born very preterm (gestational weeks 28–31). Neonatal mortality rates decreased steadily at all gestational ages during the entire study period. Clinical factors that changed during the study period were increasing numbers of high‐risk pregnancies, maternal obesity, emergency caesarean sections, neonatal admissions, and usage of assisted ventilation.Interpretation: The increasing prevalence of CP in children born preterm was driven by the subgroup born very preterm and matched their decrease in neonatal mortality. In similar population studies, decreased mortality was not followed by increased CP prevalence. An increase in clinical risk factors was unlikely to explain our findings, but more active neonatal life support may have played a role.

Aims: To determine the quality of prospectively collected data from the highly specialized Danish Cerebral Palsy Follow-up Program (CPOP), and to establish the validity of a reported cerebral palsy (CP) diagnosis in the Danish National Patient Registry (NPR), regularly used as a proxy for neurodevelopmental disorders in epidemiological research. Methods: We compared data from the two registries on children with registered CP, born in Denmark between 2008 and 2009, with information from medical records verified by two experienced physicians specializing in pediatric neurology. Data accuracy was estimated by completeness, correctness, and reliability. Completeness was calculated as the number of cases with correctly registered CP diagnoses divided by the total number of true CP diagnoses (similar to sensitivity). Correctness was calculated as the number of cases with correct registrations divided by the total number of cases (similar to positive predictive value). Reliability was estimated using kappa statistics. Results: Registered CP diagnoses in the CPOP had high accuracy, with 94% correctness and 91% completeness. Furthermore, most key variables in the CPOP showed excellent reliability, especially variables defining the severity of the condition. In the Danish NPR, only 225 of 348 children with a noted CP diagnosis fulfilled the diagnostic criteria for CP, resulting in 65% correctness. Conclusions: Danish CPOP data are a valid source for epidemiological research. Conversely, a noted CP diagnosis in the Danish NPR was, at best, correct in only two out of three patients.