We examined whether case classification in notifiable disease records varies by disease, level of care, and year, and how restricting analyses to confirmed-case records changes the distribution of notifications. Using surveillance records from Afyonkarahisar, Türkiye (January 2017–June 2024), we analyzed 13,806 notifications covering 42 diagnoses, excluding potential rabies exposures and Group D notifications. Each record was the latest notification for a case; level of care referred to its notifying facility. Mixed-effects logistic regression compared confirmed with suspected or probable records, adjusting for year, age group, and sex, with crossed random intercepts for diagnosis and district. Overall, 50.1% were recorded as confirmed. Compared with family medicine, adjusted odds ratios were 15.33 (95% CI 13.08–17.97) for secondary-care hospitals and 11.79 (9.75–14.25) for the university hospital; restricted to Group A diseases, they were 10.39 and 8.40. The association varied substantially between diagnoses. Year differences persisted after adjustment. Family medicine accounted for 26.6% of all notifications but 3.5% of confirmed-case records. Differences across levels of care may partly reflect how notifications are recorded and consolidated, and do not indicate differences in diagnostic accuracy or quality of care. Restricting analyses to confirmed-case records changes the distribution of notifications; studies should state which case categories are included.