Background: Investigating changes in prediagnostic healthcare utilisation can help identify how much earlier conditions could be diagnosed. Such 'diagnostic windows' are established for cancer but remain relatively unexplored for non-neoplastic conditions.
Aim: To extract evidence on the presence and length of diagnostic windows for non-neoplastic conditions.
Design and setting: A systematic review of studies of prediagnostic healthcare utilisation was carried out.
Method: A search strategy was developed to identify relevant studies from PubMed and Connected Papers. Data were extracted on prediagnostic healthcare use, and evidence of diagnostic window presence and length was assessed.
Results: Of 4340 studies screened, 27 were included, covering 17 non-neoplastic conditions, including both chronic (for example, Parkinson's disease) and acute conditions (for example, stroke). Prediagnostic healthcare events included primary care encounters and presentations with relevant symptoms. For 10 conditions, sufficient evidence to determine diagnostic window presence and length was available, ranging from 28 days (herpes simplex encephalitis) to 9 years (ulcerative colitis). For the remaining conditions, diagnostic windows were likely to be present, but insufficient study duration was often a barrier to robustly determining their length, meaning that diagnostic window length may exceed 10 years for coeliac disease, for example.
Conclusion: Evidence of changing healthcare use before diagnosis exists for many non-neoplastic conditions, establishing that early diagnosis is possible, in principle. In particular, some conditions may be detectable many years earlier than they are currently diagnosed. Further research is required to accurately estimate diagnostic windows and to determine how much earlier diagnosis may be possible, and how this might be achieved.
Keywords: diagnosis; electronic health records; primary health care.
© The Authors.