Task shifting routine inpatient pediatric HIV testing improves program outcomes in urban Malawi: A retrospective observational study

Background: This study evaluated two models of routine HIV testing of hospitalized children in a high HIV-prevalence resource-constrained African setting. Both models incorporated “task shifting,” or the allocation of tasks to the least-costly, capable health worker. Methods and Findings: Two models were piloted for three months each within the pediatric department of a referral hospital in Lilongwe, Malawi between January 1 and June 30, 2008. Model 1 utilized lay counselors for HIV testing instead of nurses and clinicians. Model 2 further shifted program flow and advocacy responsibilities from counselors to volunteer parents of HIV-infected children, called “patient escorts.” A retrospective review of data from 6318 hospitalized children offered HIV testing between January-December 2008 was conducted. The pilot quarters of Model 1 and Model 2 were compared, with Model 2 selected to continue after the pilot period. There was a 2-fold increase in patients offered HIV testing with Model 2 compared with Model 1 (43.1% vs 19.9%, p<0.001). Furthermore, patients in Model 2 were younger (17.3 vs 26.7 months, p<0.001) and tested sooner after admission (1.77 vs 2.44 days, p<0.001). There were no differences in test acceptance or enrollment rates into HIV care, and the program trends continued 6 months after the pilot period. Overall, 10244 HIV antibody tests (4779 maternal; 5465 child) and 453 DNA-PCR tests were completed, with 97.8% accepting testing. 19.6% of all mothers (n = 1112) and 8.5% of all children (n = 525) were HIV-infected. Furthermore, 6.5% of children were HIV-exposed (n = 405). Cumulatively, 72.9% (n = 678) of eligible children were evaluated in the hospital by a HIVtrained clinician, and 68.3% (n = 387) successfully enrolled into outpatient HIV care. Conclusions/Significance:The strategy presented here, task shifting from lay counselors alone to lay counselors and patient escorts, greatly improved program outcomes while only marginally increasing operational costs. The wider implementation of this strategy could accelerate pediatric HIV care access in high-prevalence settings. © 2010 McCollum et al. In 2007 a multidisciplinary task force at KCH created a PITC program strategy, program procedural manual [Text S1], and confidential program register [Text S2]. The task force also revised the pediatric department admissions form to include a standing written order for HIV testing as well as a section for documentation of test results. Based on low nurse and clinician staffing levels, high rates of evening admissions when staffing was even lower, and undeveloped triage systems, the task force concluded that HIV testing at admission by nurses and clinicians could not achieve the program objectives and could delay emergent care. Furthermore, the task force decided that a bedside testing model was potentially coercive and inadequate for confidentiality. Instead, the group concluded that a model built around private testing by counselors during inpatient ward rounds, with or without patient escorts, could be immediately piloted. A program algorithm divided PITC into eight steps [Text S3], and the confidential patient register facilitated referrals and monitoring [Text S2]. All PITC staff worked from 7:30am to 4:00pm Monday through Friday. HIV testing was not available on weekends. Government approval was obtained, and funding was acquired from the Baylor International Pediatric AIDS Initiative. Baylor International Pediatric AIDS Initiative receives primary funding from Bristol Myers Squibb, the Abbott Fund, Texas Children's Hospital, The United Nations Children's Fund, and the Malawi Ministry of Health. Two rooms within the pediatric department were equipped with partitions and materials to create four private testing spaces. Four nationally certified HIV counselors employed by the COE and Lighthouse Trust were assigned to the program, with one counselor per room. Prior to implementing each model, two half-day departmental orientations took place each preceding month [Text S4, S5, S6]. As a part of the orientation, clinicians and nurses were trained to order HIV testing on all patients during wards rounds, irrespective of clinical symptoms, physical examination, admission diagnosis, or ward location. Counselors also received guidance regarding how to conduct pretest group counseling and opt-out testing. In both models, counselors followed a PITC opt-out protocol and the Malawi National HIV Counseling and Testing guidelines. National guidelines recommended investigating maternal HIV status before testing the child. Eligible children were tested with a HIV antibody test and/or DNA-PCR test depending upon their age [29]. In some select cases, however, the child was tested first, and the mother was either tested after the child or not at all. If the caregiver produced valid documentation, such as the national health passport, indicating that the child's HIV status was known, then testing information was documented on the patient's admission form and the PITC register was updated. Patients that disclosed their HIV status during ward rounds would not go to the counseling room for testing. Two COE clinicians evaluated all hospitalized HIV+ and HIV-E children using the confidential program register to locate patients, and provided program supervision. Supervision consisted of a weekly review of program registers and monthly PITC meetings within the pediatric department. The meetings provided monitoring feedback to staff and addressed programmatic or staff issues. The Model 1 pilot period began January 1, 2008 and continued until March 31, 2008. In addition to opt-out HIV testing, Model 1 expected counselors to perform group counseling, team up with providers during ward rounds, interact with patients hospitalized on the ward, and accompany caregivers to the testing rooms. Four patient escorts were identified for Model 2 and introduced at the Model 2 orientation. The escorts were recruited from a group of pro-active parents of HIV+ children that were attending the COE clinic. Each escort was openly HIV+, compliant with ART, demonstrated proficient oral and written English and Chichewa language skills, and produced documentation confirming completion of primary schooling. All escorts received a half-day orientation [Text S4, S5, S6], as well as pediatric HIV diagnosis [Text S7, S8, S9, S10, S11, S12] and job-related training [Text S13, Text S14], followed by one week of on-the-job supervision by a COE pediatrician. Each volunteer patient escort earned a stipend for work-related transportation and food ($2/day). The Model 2 pilot period began April 1, 2008 and continued until June 30, 2008. Each escort was assigned to a counselor and KCH clinician for ward rounds, and accompanied patients to the testing room after a provider advised the caregiver about the routine HIV test. If the providers were unavailable or did not advise the caregiver appropriately, escorts would still mentor parents regarding the benefits of pediatric HIV testing and chaperone willing caregivers to the counseling room. Counselors continued to perform group counseling and opt-out HIV testing. The PITC task force evaluated the pilot models, and selected Model 2 for the programmatic period that started July 1, 2008. The study was approved by the Malawi National Health Sciences Research Committee and Baylor College of Medicine institutional review boards, respectively. Verbal consent for HIV testing was obtained by nationally certified HIV counselors and documented in patient files in accordance with Malawi National HIV Counseling and Testing guidelines [29]. Written consent was not required by study participants since data was collected as a part of routine program monitoring and evaluation. A retrospective review of data from 6318 hospitalized children offered HIV testing from January-December 2008 was conducted. The review was separated into two periods, pilot and programmatic. Disaggregated pre-PITC pediatric ward HIV testing data was unavailable for comparison. A data officer collected maternal and child study data from the PITC register, ward admissions book, and COE electronic medical record. Outcomes data included the proportion of admissions offered HIV testing, the proportion accepting testing, age in months of PITC recipients, days elapsed from admission to testing, maternal HIV antibody test results, child HIV antibody and DNA-PCR test results, mother and child HIV status, and successful enrollment into inpatient and outpatient pediatric HIV care. Continuous variables were evaluated with the Mann-Whitney U test. Age and time from hospital admission to HIV test were expressed as median months and mean days, respectively, both with interquartile range. For categorical parameters, data were reported as raw value and percentage of the respective group. Pearson's chi-square test was used to determine global significance between Model 1 and Model 2 pilot period data, and post-hoc analyses were performed using pair-wise chi-square tests or Fisher exact tests, with the Bonferroni correction applied to adjust α for multiple variable levels. Only p values smaller than the corrected α were considered statistically significant. All statistical analyses were performed using SPSS software (version 17.0; SPSS Inc., Chicago, IL).