Collecting sexually transmitted disease clinic chart data in multisite studies

Abstract

ion of biological data in support of the NIMH Multisite HIV Prevention Trial STD reinfection (i.e., incident STD in the follow-up period) is one of the clinical endpoints for the STD populations. Although STD data were collected for male and female clients at the STD clinics, there was concern that reinfection would not be detectable among female participants due to the asymptomatic nature of some diseases in women (e.g., C. trachomatis). For each follow-up period, clinical records were systematically checked to determine whether each participant had been treated at the clinic during the period, and if so, what clinical impressions, treatments, and laboratory test results were recorded. STD clients provided written consent for study personnel to obtain data from clinic records. Biologic endpoint abstraction form Even though laboratory tests were performed on many clients, because of missing data, the syndromic approach to STD identification was used in this trial for determining incident infection for members of the male and female STD populations. A biologic endpoint abstraction form was designed to obtain information for making syndromic evaluations of STD (Fig. 1). The abstraction form was developed by tr ial consultants and investigators and STD clinicians from three participating sites who reviewed a sample abstracting form, made recommendations for its expansion, and then reviewed the revised form. Consultants and investigators reviewed clinic charts from STD clinics at all sites and designed two versions of baseline and follow-up forms for use at all clinics, based on the data that were available in the charts. Both versions of the abstraction form collected demographic information, the reason for the current visit, information about prior STD, laboratory test results (both immediate tests and tests ordered), and clinical impressions. On the longer form, the abstractor also collected information about risk factors and HIV testing, and this form was used when such data were available. Follow-up forms were designed to be completed for each visit during the period between the baseline visit and 13 months after the end of the intervention. Follow-up forms were similar to the baseline forms, but information on referral for further services or treatment was omitted. Method of STD data collection The unit of analysis for abstraction was the patient visit. Thus, patients may have had none, one, or multiple visits over a subsequent year. In the tr ial’s pilot study, the number of return visits ranged from 0 to 14. Where possible, a trained nurse abstractor conducted the record abstraction; a study clinician at each site assisted in the determination of local charting practices, as well as in maintaining quality control of the data collection procedures. A second staff member reabstracted data for 20% of the cases randomly selected by the coordinating center so that inter-abstractor reliability could be assessed. Although it was recommended that a trained nurse was employed by the study to perform the abstracting, this was not always possible because of clinic regulations. Three data collection approaches were used in this tr ial: model 1, both persons who abstracted charts were employed by the study and were permitted to search for the charts and refile them; model 2, both individuals were employed by the study but the charts were retrieved from the files by clinic employees; and model 3, outside personnel were STD clinic chart data collection S61 Fig. 1. Biologic endpoint abstraction form (continued overleaf). AIDS 1997, Vol 11 (suppl 2) S62 not permitted in the clinic and so clinic staff were hired by the project to perform both the abstracting and the 20% reabstracting. Although model 1 was preferable because abstractors could search for charts using multiple strategies, model 2 was deemed only slightly less acceptable. Model 1 and 2 abstractors participated in the quality assurance telephone calls and a central monitor could audit the actual charts. Model 3 required special procedures in order to demonstrate reliability of the abstracting because the central monitor was not permitted to review the charts. In order to ascertain reliability, dummy charts were therefore prepared, for completion by both abstractors. Manuals detailing how to abstract information from the charts at each clinic were then developed. Centralized scoring of the abstracted STD charts was conducted with the input of STD clinician consultants who determined whether there was an STD case by using an algorithm. Determination of STD cases To diagnose gonorrhea, either the laboratory immediate results for smears from the urethra, cervix, or rectum were positive, or cultures from the urethra, cervix, rectum, or throat were positive. If a positive trichomonas test was reported, then it was recorded as a new STD. Syphilis was diagnosed if the immediate laboratory results for the dark field were positive, or if a VDRL/RPR test was reactive (1 : 1 or greater) and the fluorescent treponemal antibody absorption assay (FTA)/microhemagglutination assay for antibody to Treponema pallidum (MHA-TP) was positive. If a positive chlamydia test was reported, then a new STD was recorded. A clinical impression of mucopurulent cervicitis (MPC)/non-gonococcal urethritis (NGU) was recorded as a new STD. If a herpesvirus culture was positive, and there was no indication of disease history, it was recorded as an STD. To determine incident STD during the follow-up period, the above criteria were used to diagnose a positive case, but there were a few other issues to consider. For gonorrhea, chlamydia, and trichomonas, an incident case was recorded if (i) the participant had no history of the STD, (ii) there was documentation of treatment in the chart, or (iii) the next infection occurred more than 1 month later (adequate treatment was assumed for prior infection). For incident trichomonas, chlamydia, or MPC/ NGU infections, the definition used to diagnose a new case (descr ibed above) was used. Incident herpesvirus infections were recorded if there was no history reported in the participant’s chart. To detect incident syphilis cases, the process was more complex. The three algorithms used to assess an incident case of syphilis were if (i) the participant was VDRL/RPR-negative or positive and FTA/ MHA-TP-negative at a previous visit, and became VDRL/ RPR-positive and FTA/MHA-TP-positive, despite titer; (ii) the participant had a history of syphilis, with documented or assumed therapy, and the VDRL/RPR titer Fig. 1. Biologic endpoint abstraction form (continued). STD clinic chart data collection S63 Table 1. Percentage of charts recovered at sexually transmitted dis-