Combining Register and Radiological Visits Data Allows to Reliably Identify Incident Wrist Fractures

Figures & data

Figure 1 Flowchart for establishing the gold standard of wrist fractures for the participants of the OSTPRE study cohort from 2011 to 2021. The Venn diagram in the bottom box illustrates the overlap between different sources of fracture events in the gold standard.

Table 1 Coverage and False Discovery Rates for Different Data Sources and Algorithms

	Full Cohort (1016 Fractures)				North Savo Subcohort (892 Fractures)
	True Positives	Gold Standard Coverage	False Positives	False Discovery Rate	True Positives	Gold Standard Coverage	False Positives	False Discovery Rate
Hilmo register	449	44.2%	18	3.9%	380	42.6%	12	3.1%
Avohilmo register	663	65.3%	47	6.6%	592	66.4%	39	6.2%
Registers combined (Hilmo + Avohilmo)	823	81.0%	59	6.7%	719	80.6%	47	6.1%
Wrist radiograph in PACS	943	92.8%	903	48.9%	885	99.2%	834	48.5%
PACS 2+ algorithm	910	89.6%	101	10.0%	855	95.9%	94	9.9%
PACS 3+ algorithm	824	81.1%	10	1.2%	775	86.9%	8	1.0%
PACS&register algorithm A	882	86.8%	18	2.0%	828	92.8%	16	1.9%
PACS&register algorithm B	926	91.1%	23	2.4%	838	93.9%	20	2.3%

Figure 2 The yearly register coverage of gold standard fractures. Since 2014, almost all the fractures have some contact in Hilmo or Avohilmo within 14 days from the index date but the recording rate for diagnostic codes is incomplete.