Agreement between self-reported diseases from health surveys and national health registry data: a Danish nationwide study

Identifying the prevalence of Parkinson's disease in Denmark using healthcare registries and self-reported survey data

INTRODUCTION

Existing estimates of PD prevalence in Denmark are lower than those in the rest of Europe and are based on identification via single registries. Hence, are aim was to use a combined registry/self-report survey approach to identify people with PD and also investigate whether using different registry methods led to differences in the accuracy, completeness and characteristics of the identified cohorts.

METHODS

This study had a cross-sectional design using routinely collected health registry data to identify adults, ≥18 years of age and resident in Denmark, with PD from either the Danish National Patient (DNP) registry or Danish Prescription Medicines (DPM) registry. Those identified were asked to confirm their PD diagnosis using a national self-report survey.

RESULTS

13,433 people were identified potentially as having PD via the DNP or DPM registry and sent a survey. Of these, 9094 responded (68 %) of which 85 % confirmed they had PD (n = 7763; 194/100,000; 95%CI:7650-7876). When adjusting for non-respondents, assuming an equal rate of confirmation in respondents and non-respondents, estimated Danish PD population was 11,467 (198.4/100,000; 95 % CI:197.2-199.6). Identification of people using those found in both registries led to 98 % confirming they had PD versus using one registry: DNP 93 % and DPM 88 %. No clear differences in sociodemographic characteristics were found between different registry identification methods.

CONCLUSIONS

Estimated PD population in Denmark was significantly higher than previous Danish estimates and close to existing estimates in other European countries. The most accurate PD population was identified when including those found in both the DNP and DPM registries.

Cardiovascular Disease Events in Adults with a History of State Care in Childhood: Pooling of Unpublished Results from 9 Cohort Studies

Background

Individuals who were separated from their biological family and placed into the care of the state during childhood (out-of-home care) are more prone to developing selected physical and mental health problems in adulthood, however, their risk of cardiovascular disease (CVD) is uncertain. Accordingly, we pooled published and unpublished results from cohort studies of childhood care and adult CVD.

Methods

We used two approaches to identifying relevant data on childhood care and adult CVD (PROSPERO registration CRD42021254665). First, to locate published studies, we searched PubMed (Medline) until November 2023. Second, with the aim of identifying unpublished studies with the potential to address the present research question, we scrutinised retrieved reviews of the impact of childhood state care on related adult health outcomes. All included studies were required to have prospective measurement of state care in childhood and a follow-up of CVD events in adulthood as the primary outcome (incident coronary heart disease and/or stroke). Collaborating investigators provided study-specific estimates which were aggregated using random-effects meta-analysis. The Newcastle-Ottawa Scale was used to assess individual study quality.

Findings

Thirteen studies (2 published, 11 unpublished) met the inclusion criteria, and investigators from nine provided viable results, including updated analyses of the published studies. Studies comprised 611,601 individuals (301,129 women) from the US, UK, Sweden, Finland, and Australia. Relative to the unexposed, individuals with a care placement during childhood had a 50% greater risk of CVD in adulthood (summary rate ratio after basic adjustment [95% confidence interval]: 1.50 [1.22, 1.84]); range of study-specific estimates: 1.28 to 2.06; I2 = 69%, p = 0.001). This association was attenuated but persisted after multivariable adjustment for socioeconomic status in childhood (8 studies; 1.41 [1.15, 1.72]) and adulthood (9 studies, 1.28 [1.10, 1.50]). There was a suggestion of a stronger state care-CVD association in women.

Interpretation

Our findings show that individuals with experience of state care in childhood have a moderately raised risk of CVD in adulthood. For timely prevention, clinicians and policy makers should be aware that people with a care history may need additional attention in risk factor management.

Variations in the agreement of self-reported cancer: A Danish nationwide study

Previous studies show that the agreement between self-reported and registry-documented diseases varies across diseases. Few studies have addressed these challenges across site-specific cancer diagnoses. The present study aimed to examine the sensitivity and negative predictive value (NPV) of self-reported cancer in a Danish nationwide survey among adults aged ≥16 years, using registry data as the criterion standard. Moreover, the influence of sociodemographic variables and time since diagnosis on sensitivity was explored using multiple logistic regression models. Self-reported data on cancer history of any site were derived from the Danish National Health Survey 2017 (n = 183 372). Individual-level survey data were linked to data from the Danish Cancer Registry on 10 site-specific cancer diagnoses. NPV was consistently high ≥99.5% across the included cancer diagnoses. In contrast, sensitivity varied greatly and was lowest for cancer in brain/central nervous system (CNS) among both men (25.6%) and women (23.9%) and highest for rectal cancer among men (96.9%) and for breast cancer among women (98.9%). Sensitivity was also relatively low for nonmelanoma skin cancer (41.4% among men; 44.6% among women) and urinary tract cancer (60.0% among men; 60.4% among women). When restricting diagnostic definitions for cancer in brain/CNS and urinary tract cancer to include only malignant neoplasms, sensitivity increased. For several cancer diagnoses, sensitivity decreased with increasing age and lower educational level, whereas conflicting results were observed for time from diagnosis to self-report. Future studies are encouraged to use self-reported cancer history data with caution and for example, include questions on only site-specific cancer diagnoses with high sensitivity.