Optimizing the use of surveillance data for monitoring the care status of persons recently diagnosed with HIV in NYC.

BACKGROUND: Comprehensive laboratory reporting of CD4 and viral load (VL) tests to surveillance has been used to assess HIV care-related outcomes at the population level, but their validity for this purpose has not been comprehensively evaluated.
OBJECTIVE: Assess performance characteristics and validity of surveillance-based measures of linkage to and establishment of HIV primary care among HIV-infected persons in the first 12 months after diagnosis using medical record (MR) data on outpatient HIV primary care visits as the gold standard.
METHODS: All patients diagnosed with HIV in 2009 at 24 New York City high-volume, HIV diagnostic and treatment facilities who linked to care within 12 months at the same site as defined by the presence of ≥1 CD4/VL report received by surveillance were selected for MR review to confirm linkage to outpatient HIV primary care within the first year. All HIV care visit dates were abstracted and considered associated with a surveillance laboratory report, if within 14 days of a care visit. The proportion linking to care according to the MR was compared with the proportion linking per CD4/VL tests reported to surveillance. Four measures of the establishment of outpatient HIV primary care in the first year were assessed: (1) sustained care (first visit within 3 months; second visit, 3-9 months later), (2) continuous care (2 visits at least 90 days apart), (3) trimester visits (visit in each 4-month period), and (4) visit constancy (visit in each 3-month period). The validity of surveillance data for measuring this outcome was assessed by comparing results for each of the 4 measures calculated using surveillance data to those calculated using MR data.
RESULTS: Of the 782 patients selected, 20% (N = 157) of patients did not link to outpatient HIV primary care at the co-located care facility within 12 months of diagnosis. Half (48.5%) of patients' care visits after linkage did not have an associated CD4/VL reported to surveillance. Of the 4 establishment measures, sustained and continuous care had the highest agreement with MR (86.6% and 88.8%, respectively) as compared with the trimester visits and visit constancy (77.8% and 72.8%, respectively).
CONCLUSIONS: Surveillance data overestimated linkage rates but underestimated the frequency of HIV care in the first year after HIV diagnosis. Of the 4 measures of establishment of HIV care evaluated, "sustained care" is best suited for measurement using surveillance data because of its high level of agreement with MR data and close alignment with national standards for timely linkage and flexible follow-up.