Needle autopsy to establish the cause of death in HIV-infected hospitalized adults in Uganda: a comparison to complete autopsy.

INTRODUCTION: Minimal invasive but accurate methods to establish the cause of death in HIV-infected patients are needed. We studied the agreement in cause of death between blind and ultrasound-guided needle autopsy and complete autopsy in HIV-infected patients in Uganda.
METHODS: We subsequently performed a blind and ultrasound-guided needle autopsy followed by a complete autopsy in HIV-infected adults who died during hospitalization. Two teams of pathologists reviewed the tissue from either the needle autopsies or the complete autopsy and formulated the major diagnoses, that is, diseases directly contributing to death. The primary outcome was concordance in major diagnosis between needle and complete autopsies.
RESULTS: We performed 96 blind needle and complete autopsies and 95 ultrasound-guided needle autopsies. Concordance in major diagnosis between blind needle and complete autopsy was 50%. For the main major diagnosis, tuberculosis (TB) concordance was higher (71%; P < 0.01). Blind needle autopsy identified at least 1 major diagnosis in 60% of patients; and in 46%, there was complete concordance for all major diagnoses. The main reason for discordance was sampling error of the lesion. Concordance with the addition of ultrasound guidance was 52% for all major diagnoses and 79% for TB. Major diagnoses were mainly identified in tissue cores from the liver (76%) and the spleen (82%). DISCUSSION: Blind needle autopsy identified half of the major diagnosis. The addition of ultrasound guidance did not significantly improve the performance of needle autopsy. Needle autopsy is a valuable method to confirm causes of death in HIV-infected patients, especially for highly prevalent diseases like TB.