‘Cough officer’ nurses in a general medical clinic successfully detect drug-susceptible and -resistant tuberculosis

Abstract

Setting:

Intensive case finding (ICF) for tuberculosis (TB) is recommended by the World Health Organization among known human immunodeficiency virus (HIV) patients. However, ICF may also be appropriate in generalized patient populations.

Objective:

To evaluate the yield of ICF in a general medical clinic in a high HIV prevalence setting.

Methods:

A nurse designated as a ‘cough officer’ identified clinic attendees with cough of >2 weeks and collected sputum for evaluation at the hospital and provincial referral laboratories. We retrospectively evaluated the number and proportion of patients with microbiologically confirmed TB identified in 2007–2008.

Results:

Among 56 207 clinic attendees, 1442 (2.6%) TB suspects were identified and 122 (8.5%) were sputum Ziehl-Neelsen (ZN) positive. Of 389 available results, 72 (18.5%) were auramine-positive and 99 (25.4%) were culture-positive; multidrug-resistant and extensively drug-resistant TB were identified in 16 (16.2%). The number needed to screen was 11.8 patients to identify one ZN-positive case and 3.9 to identify one culture-positive case.

Conclusions:

A nurse-facilitated cough officer program successfully identified TB suspects and drug-susceptible and drug-resistant TB. Culture was more sensitive for TB screening and critical for identifying drug resistance. ICF is operationally feasible, and should be expanded to general medical clinics in high HIV and TB prevalence, resource-limited settings.

The World Health Organization (WHO) strategy for tuberculosis (TB) control among human immunodeficiency virus (HIV) infected populations promotes the ‘3 Is’: intensive case finding (ICF), infection control and isoniazid preventive therapy.¹ ICF differs from passive case finding, the traditional method of case detection of TB, where patients self-refer into care. In contrast, in the ICF model, health care personnel conduct face-to-face screening for TB to enhance case detection. Although much progress has been made in clinics for the prevention of mother-to-child transmission and HIV, widespread implementation of ICF remains a goal, and appropriate non-HIV target groups need to be clarified. Consistent with the current WHO guidelines, ICF efforts focus on previously defined high-risk groups, such as HIV-infected patients, miners and prisoners.^1–6 However, broader ICF implementation may be appropriate in any sub-Saharan population with high rates of HIV and TB.⁷ The paucity of data regarding the utility of ICF strategies may contribute to its lack of widespread implementation.

In KwaZulu-Natal Province, TB incidence rates (1100 per 100 000 population) are the highest in South Africa and among the highest globally, and antenatal HIV prevalence is approximately 39%.^8,9 Drug-resistant TB has emerged as an additional serious public health challenge. Approximately 10% of all TB cases reported at the Church of Scotland Hospital in rural KwaZulu-Natal have multidrug-resistant (MDR) or extensively drug-resistant (XDR) TB, and high rates have also been reported throughout the province and country.^10,11 Given the recognition of nosocomial transmission of MDR/XDR-TB in 2006,¹² a comprehensive infection control strategy was implemented. In addition to instituting strategies to reduce transmission in in-patient wards and the HIV clinic, attention was also given to case detection in the general medical clinic by implementing a nurse-based ‘cough officer’ ICF program. At the time of this initiative, screening for TB symptoms focused solely on cough; WHO guidelines subsequently recommended the use of a combination of symptoms (fever, night sweats, weight loss, cough) for screening.^2,13

In the present study, we discuss the overall yield of this strategy, traditionally recommended for high-risk HIV groups and now deployed in the general medical clinic in a high HIV prevalence region.

STUDY POPULATION, DESIGN AND METHODS

The Church of Scotland Hospital, Tugela Ferry, is a 350-bed, district-level government hospital serving 180 000 traditional Zulu people. The general medical clinic is part of the main hospital building contiguous with the casualty, pharmacy and file room departments. After registration, patients wait on benches in a large open congregate waiting room area and are seen by clinicians in private consulting rooms. We performed a retrospective evaluation of this ICF effort to determine the yield from screening clinic attendees for cough with subsequent sputum acid-fast bacilli (AFB) smear and culture. We sought to determine the number and proportion of patients identified with microbiologically confirmed TB. For the purpose of review, TB suspects were defined as all patients with cough of >2 weeks submitting sputum specimens. All TB suspects recorded in the TB registers from 1 January 2007 to 31 December 2008 were included for review.

According to hospital protocol, a staff nurse interviewed all patients attending the out-patient clinic on weekdays. Patients identified as TB suspects were given sputum bottles and directed outdoors to produce sputum specimens for examination. Those patients unable to produce sputum were not recorded. Sputum was sent for Ziehl-Neelsen (ZN) microscopy to the hospital laboratory. A second sputum specimen was sent to the provincial TB laboratory in Durban for auramine microscopy, culture and drug susceptibility testing (DST). Culture was performed using the automated BACTEC™ MGIT™ system (BD, Sparks, MD, USA), which is monitored for 6 weeks. Positive cultures were confirmed by niacin and nitrate reductase tests. DST was performed using the 1% proportional method on Middlebrook 7H11 agar at the following critical concentrations: isoniazid (INH) 0.2 μg/ml, rifampin (RMP) 1 μg/ml, ethambutol 7.5 μg/ml, ofloxacin 2 μg/ml, kanamycin 6 μg/ml and streptomycin (2 μg/ml). Patients whose sputum was AFB-positive at the hospital laboratory were immediately directed to the TB DOTS office to initiate anti-tuberculosis treatment, while all TB suspect patients who were AFB-negative were directed to the medical officer for clinical evaluation. Patients known to be HIV-infected were referred to the HIV clinic, opposite the general medical clinic, on the hospital grounds. This report is limited to case finding in the general medical clinic.

TB registers were reviewed for age, sex and AFB smear results. Auramine smear and culture results were retrieved through a review of laboratory reports. Results from discrete patient encounters were used; duplicate results from the same patient visit were excluded. We calculated simple frequencies of AFB smear and culture among clinic attendees to determine a minimum estimate of the yield of the ICF strategy. The number needed to screen (NNS) to find one TB suspect was calculated using all clinic attendees as a denominator, while the NNS to find one smear- or culture-positive case was calculated using TB suspects submitting sputum as a denominator.

Ethical approval was obtained from the University of KwaZulu-Natal Nelson R Mandela School of Medicine and the Yale University School of Medicine.

RESULTS

Over the course of 2 years, 56 207 patients were seen in the general medical clinic of the district government hospital. Sputum was collected at 1442 (2.6%) patient encounters with TB suspects over this period. Half (51%) of the TB suspects were women and the mean age was 38 years (Table 1). Among all TB suspects, 122 (8.5%) were AFB smear-positive (Figure). Auramine, culture and DST results were available for only 389 (27.4%) TB suspects. There was no difference in age (P = 0.65) or sex (P = 0.22) between all TB suspects and those with a culture result available. Among these, 72 (18.5%) were auramine-positive and 99 (25.4%) yielded positive cultures; 73 (73.7%) were drug-susceptible. Monoresistance to INH or RMP was seen in six culture-positive patients (6.1%), while MDR- and XDR-TB were diagnosed in 16 (16.2%) patients. In two culture-positive specimens, laboratory testing failed to determine drug susceptibility; for one of these patients, a specimen from a subsequent hospitalization revealed MDR-TB.

TABLE 1.

Yield of nurse-driven cough officer screening in a general medical clinic

	Total n (%)
Total screened, n	56 207
Total TB suspects	1 442 (2.6)
Women	736 (51)
Mean age, years	38.3
Ziehl-Neelsen-positive	122 (8.5)
Specimen result available from referral laboratory	389 (26.9)
Auramine-positive	72 (18.5)
M. tuberculosis culture-positive	99 (25.4)
Drug-susceptible	73 (73.7)
Drug-resistant	22 (22.2)
Monoresistant to isoniazid	5 (5.1)
Monoresistant to rifampin	1 (1)
Multidrug-resistant	9 (9.1)
Extensively drug-resistant	7 (7.1)
Resistance status unknown	4 (4.0)

TB = tuberculosis; DST = drug susceptibility testing.

FIGURE.

Summary of case finding program. * May be duplicated with provincial laboratory culture results.

Of all TB suspects with sputum culture results, an additional three patients (0.8%) were found to have non-tuberculous mycobacteria (NTM) on sputum culture; all three were initially AFB-positive and initiated on first-line anti-tuberculosis treatment.

The median time to receipt of results for the 16 patients diagnosed with MDR/XDR-TB was 65.5 days (interquartile range 54.5–87.3). Among these, only 7/16 (43.8%) were still alive when the results of the sputum culture were received by the hospital. All seven were placed on standardized treatment for MDR/XDR-TB as per provincial protocol.

Among clinic attendees, the NNS to detect a TB suspect was 39 (Table 2), while among TB suspects the NNS was 12 for a positive AFB smear result. Among TB suspects with results available from the provincial referral laboratory (n = 389), the NNS to detect a positive auramine result, a culture-positive result and a drug-resistant TB case were respectively 5, 4 and 25.

TABLE 2.

NNS among all clinic attendees and TB suspects

	NNS to identify 1 case
	TB suspect	ZN-positive	Auramine-positive	Culture-positive	Drug-resistant
All general medical clinic attendees (n = 56 207)	39
TB suspects identified by cough officer, with AFB sputum smear result (n = 1442)		12	20	15	90
TB suspects identified by cough officer with sputum specimen result available from referral laboratory (n = 389)			5	4	24

NNS = number needed to screen; TB = tuberculosis; ZN = Ziehl-Neelsen; AFB = acid-fast bacilli.

DISCUSSION

Using a strategically placed nurse cough officer, TB ICF, traditionally recommended for known HIV patients, carried out in the general medical clinic of a rural district hospital in a high prevalence HIV region, identified a large number of TB suspects. Among these, a large number of smear- and culture-positive patients were found, including a substantial proportion with drug-resistant TB. To explore this finding operationally, we calculated the NNS to find a positive result at a number of points in the screening cascade. NNS has been used as a measure of screening programs, but cross-study comparison is limited by different methods of calculating the NNS. Some reports have used all patients eligible for screening as the denominator, while others used patients who participated in screening, based on narrower criteria such as TB-related symptoms.^7,14–16 In our general medical clinic, using patients identified by a simple screen performed by a nurse cough officer as the denominator, the case detection rate and the NNS to find one smear-positive patient were respectively 8.5% and 12. This yield is similar to findings in clinics offering antiretroviral treatment in a recent meta-analysis,⁷ and better than that found in another recent general medical clinic ICF initiative in Botswana.¹⁴ The incomplete information from the referral TB laboratory suggests that the use of auramine and culture with DST not only reduced the NNS but provided the added benefit of identifying those with drug-resistant TB in the general medical clinic setting.

Few studies in the HIV era have evaluated ICF for TB in non-HIV patient populations in high-prevalence regions. ICF has traditionally focused on HIV clinics and hospital wards, where nosocomial transmission has been well-documented.¹⁷ The findings from this programmatic evaluation highlight the value of ICF for pulmonary TB in a high-risk setting that is not traditionally recommended, i.e., a general medical clinic in a high-prevalence HIV and TB area, using symptom screen. In this setting, there was a worryingly high number of coughing patients in a crowded, poorly ventilated out-patient clinic with potentially unrecognized mycobacterial-laden sputum, with a risk of transmission to other patients.

In contrast to the recent lower yield and more complex screening protocol described in a general medical clinic in Botswana,¹⁴ the screening algorithm in our program used a junior nurse and a simple protocol inquiring about cough only. Patients were requested to submit sputum immediately and were not burdened with additional clinic visits. Using a nurse without further evaluation by a doctor allowed this pathway to be more widely implementable and more likely to be decentralized to clinics where doctors are in short supply, in keeping with global task-shifting priorities. These factors may have contributed to the success of the program. Future ICF scale-up should consider patient ease of use and available human resources to optimize the design and implementation of feasible patient pathways to diagnosis and treatment.

Clinical outcomes could not be determined in this retrospective study. Of note, however, only a minority of MDR/XDR-TB patients were alive by the time culture and DST results became available. This distressing result is consistent with previous reports.^10,12 The long delay resulting from conventional culture and DST contribute to high mortality rates and supports the need for rapid TB diagnostics and earlier case detection.

From the perspective of implementation science, this evaluation demonstrates that not only was the use of culture and DST as a screening tool more sensitive than AFB smear alone, it also identified a substantial number of patients with drug-resistant TB. Tests for detecting drug-resistant Mycobacterium tuberculosis are associated with substantial delay and are not widely available in most resource-limited settings.¹⁸ Case-finding efforts are dependent on the quality of laboratory services and TB diagnostics; even introducing auramine microscopy at all health facilities would increase the yield of case finding and improve TB diagnosis. While we experienced a low NNS in this retrospective evaluation using a simple cough screen algorithm and conventional mycobacterial diagnostic methods among those with a positive cough screen, there remained a substantial delay in confirming TB diagnoses (median 65.5 days for culture and DST). This likely contributed to the high mortality among those diagnosed with drug-resistant TB. Rapid diagnostic tests could improve sensitivity, rapidity of identification and overall case-finding efforts, potentially reduce high mortality rates and interrupt transmission. Several new technologies have shown promise, including urinary lipoarabinomannan for HIV-positive patients with advanced immunosuppression and the GeneXpert^® system (Cepheid, Sunnyvale, CA, USA) which is sensitive in both HIV-positive and -negative populations and may be appropriate for screening in populations with high HIV prevalence.^18–20 As new technologies such as GeneXpert are widely implemented, as is currently being done in South Africa, there may be substantial increases in TB case finding.^21,22

To reduce mortality and ongoing transmission, the optimum utility of newer rapid diagnostics still requires identification of patients early in the course of their TB and HIV disease. The high mortality rates strongly argue that ICF efforts should move beyond the health facility to the community level. While recent ICF studies have demonstrated feasibility and success in reducing TB prevalence using a community-based strategy,^23,24 additional prospective studies are needed to document the benefit of community-based ICF in a high-prevalence HIV setting, particularly assessment of patient outcomes; such studies are currently underway. Furthermore, as the present study indicates, expanding ICF implementation beyond screening of known HIV patients to more general medical populations in health care facilities in areas of high HIV and TB prevalence should also be strongly considered as part of an expanded ICF strategy.⁷

As a programmatic endeavor implemented in routine clinical settings, there are several limitations to this evaluation. This ICF program was not designed as a research study, and it was evaluated retrospectively. Data were not collected on all patients screened, staffing was not uniform or comprehensive and there was no comparison population. In this retrospective evaluation, the absence of selection bias cannot be confirmed. HIV status was not recorded, although in general known HIV patients were directed to the HIV clinic for evaluation. Patients who could not produce sputum were not recorded, leading to an underestimate of the yield of this strategy for detecting pulmonary and extra-pulmonary TB patients without cough. Furthermore, the reason for attending the clinic was not recorded; therefore, unlike other studies where a substantial proportion of patients were not planning to see the doctor for their cough,^14,25 we cannot comment on patient motivation in our general medical clinic setting. Some patients may have attended to be seen for TB symptoms. Finally, culture results were not available for all TB suspects, reflective of practice in a routine clinical setting. It is likely that only one specimen was obtained in many patients and this would have been used for ZN staining at the local hospital laboratory according to program practice. Nevertheless, the fact that the program was carried out in a routine clinical setting, where the results are likely to be widely applied, lends generalizability to the findings as representative of what might be expected in the complicated, not always controllable, but critical real world setting. Findings presented here thus represent only a minimum estimate of the true number of cases seen in a general medical clinic in a community with high HIV and drug-susceptible and -resistant TB prevalence during the period of review. The study thus strongly supports the rationale and utility of implementing ICF in a general medical clinic setting.

In summary, implementation of a relatively low-cost and simple nurse-driven cough officer program in the general medical clinic of a busy district hospital in a high HIV prevalence setting identified a substantial number of patients with cough, of whom a large proportion were smear- and culture-positive, and many also had MDR/XDR-TB. This is a readily applicable and low intensity ICF effort, which, in concert with improved laboratory services, may be of similar benefit in improving case detection in other resource-limited settings.