Provider initiated tuberculosis case finding in outpatient departments of health care facilities in Ghana: yield by screening strategy and target group.

BACKGROUND: Meticulous identification and investigation of patients presenting with tuberculosis (TB) suggestive symptoms rarely happen in crowded outpatient departments (OPDs). Making health providers in OPDs diligently follow screening procedures may help increase TB case detection. From July 2010 to December 2013, two symptom based TB screening approaches of varying cough duration were used to screen and test for TB among general outpatients, PLHIV, diabetics and contacts in Accra, Ghana.

METHODS: This study was a retrospective analysis comparing the yield of TB cases using two different screening approaches, allocated to selected public health facilities. In the first approach, the conventional 2 weeks cough duration with or without other TB suggestive symptoms was the criterion to test for TB in attendants of 7 general OPDs. In the second approach the screening criteria cough of >24 hours, as well as a history of at least one of the following symptoms: fever, weight loss and drenching night sweats were used to screen and test for TB among attendants of 3 general OPDs, 7 HIV clinics and 2 diabetes clinics. Contact investigation was initiated for index TB patients. The facilities documented the number of patients verbally screened, with presumptive TB, tested using smear microscopy and those diagnosed with TB in order to calculate the yield and number needed to screen (NNS) to find one TB case. Case notification trends in Accra were compared to those of a control area.

RESULTS: In the approach using >24-hour cough, significantly more presumptive TB cases were identified among outpatients (0.82% versus 0.63%), more were tested (90.1% versus 86.7%), but less smear positive patients were identified among those tested (8.0% versus 9.4%). Overall, all forms of TB cases identified per 100,000 screened were significantly higher in the >24-hour cough approach at OPD (92.7 for cough >24 hour versus 82.7 for cough >2 weeks ), and even higher in diabetics (364), among contacts (693) and PLHIV (995). NNS (95% Confidence Interval) varied from 100 (93-109) for PLHIV, 144 (112-202) for contacts, 275 (197-451) for diabetics and 1144 (1101-1190) for OPD attendants. About 80% of the TB cases were detected in general OPDs. Despite the intervention, notifications trends were similar in the intervention and control areas.

CONCLUSION: The >24-hour cough approach yielded more TB cases though required TB testing for a larger number of patients. The yield of TB cases per 100,000 population screened was highest among PLHIV, contacts, and diabetics, but the majority of cases were detected in general OPDs. The intervention had no discernible impact on general case notification.