Objective To analyze and compare the sources and the previous testing history of people living with human immunodeficiency virus (HIV), and their association with late diagnosis of HIV infection, so as to provide references for formulating HIV prevention and control strategies.

Methods From January to July 2024, a cross-sectional study was conducted using convenient sampling to recruit newly reported HIV cases from 2017 to 2023 in Pudong New Area, Jing’an District, and Jiading District in Shanghai, as well as Dehong Dai and Jingpo Autonomous Prefecture (hereinafter referred to as “Dehong Prefecture” in Yunan Province for face-to-face questionnaire surveys. Logistic regression analyses were used to analyze factors influencing late diagnosis in both regions.

Results In Shanghai, 69.08% of cases were diagnosed through medical institutions, followed by voluntary counseling and testing clinics of the center for disease control and prevention (21.54%). In Dehong Prefecture, 52.08% of cases were from medical institutions, followed by physical examinations for health certificates/driver’s licenses (13.43%) and free proactive HIV testing and counseling by rural medical workers (10.37%). Among participants in Shanghai, 28.08% had self-initiated testing experiences, 15.38% had three or more previous tests, and 26.78% had late diagnosis. In Dehong Prefecture, the corresponding proportions were 8.09%, 10.53%, and 24.67%, respectively. Multivariate analyses adjusted for demographic characteristics and pre-diagnosis behaviors revealed that having three or more previous tests and prior self-initiated testing experiences were negatively associated with late diagnosis among people infected with HIV in Shanghai, while case-finding through non-STD departments of medical institutions were positively associated with late diagnosis. In Dehong Prefecture, one and more than three previous tests were negatively associated with late diagnosis, whereas prior self-initiated testing experiences showed no significant association with late diagnosis.

Conclusion Significant differences existed in sample sources and prior testing experiences among people infected with HIV in Shanghai and Dehong Prefecture. Shanghai demonstrated superior performance in HIV case identification and late diagnosis reduction through self-initiated testing and STD clinic-based detection, while Dehong Prefecture relied primarily on passive testing despite having diverse testing channels. Multiple prior testing episodes were protective against late HIV diagnosis in both regions. These findings suggest that region-specific HIV testing strategies should be developed based on local epidemiological patterns and socio-structural contexts. Economically developed areas should leverage well-established medical networks to expand screening while capitalizing on self-testing advantages to promote active uptake, whereas border regions or areas with high population mobility required policy innovations to address limitations of traditional passive testing systems.