Screening for diabetes mellitus among tuberculosis patients in Southern Nigeria: a multi-centre implementation study under programme settings

Implementation studies are recommended to assess the feasibility and effectiveness of programmes. In Nigeria, little is known about the burden of diabetes mellitus (DM) among tuberculosis (TB) patients. The objective of this study was to determine screening efficacy, prevalence of DM and determinants of DM among TB patients. We report on a multi-centre implementation study carried-out in 13 health facilities in six States of Southern Nigeria. All newly diagnosed TB patients registered from March to October 2015 were screened for DM using current World Health Organisation guidelines. Overall, 2094 TB patients were evaluated, 196 (9.4%) were found to have DM. The prevalence of newly diagnosed DM was 5.5% (115/2094). DM prevalence varied according to age group; occurring in 2.2% of patients aged ≤ 25 years and 16.9% in patients aged (56–65) years. The additional yield of DM was 59% while the number needed to screen to detect a new case of DM was 18. Factors associated with DM were; age >40 years (aOR2.8, CI 2.1–3.9), rural residence (aOR2.3, 1.6–3.2), private health facility care (aOR2.0, 1.4–2.7), and having an occupation that engages in vigorous activity (aOR0.6, 0.4–0.9). The burden of DM among TB patients is high. Prioritization of DM screening for TB patients is indicated.

In recognition of the burden of DM and TB, the World Health Organization (WHO) and the International Union against Tuberculosis and Lung Disease (The Union) launched the Collaborative Framework for Care and Control of Tuberculosis and Diabetes to guide policy makers and programme managers in combating the TB-DM epidemic 9 . However, recent studies have shown that DM prevalence among TB cases is variable. It ranges from about 50% reported by some studies in India, about 30% in Mexico, less than 10% in the United States, to even lower values in certain African countries, Spain and China [10][11][12][13][14][15][16] . As reported levels of co-morbidity vary highly, National TB Control Programmes (NTPs) will need to decide whether such DM screening is needed. Where necessary, the most cost-effective approach for each setting, for example targeted screening or screening of all patients should be determined 17 .
Nigeria ranks third among 30 countries with estimated highest burden of TB in the world as well as third among African countries with highest burden of DM 1,2 . Considering that two thirds of DM patients in Africa are not aware that they have the disease, screening for DM among TB patients may contribute to early diagnosis 2 . In Nigeria, some research has been conducted on the prevalence of DM among TB patients 18,19 . However, as these studies were limited to hospitals located in a single urban area, it is unlikely to reflect the disease burden in rural areas or other geographical and cultural settings in the country. As WHO/The Union have called for operational and clinical research to build up and strengthen local evidence base for action in combating the dual epidemic 17 , quality data to guide this process in Nigeria is crucial but lacking. The aim of this study was to investigate the epidemiology of DM among TB patients in Nigeria. The specific objectives were; to determine 1) the proportion of TB patients with DM, 2) the additional yield of newly diagnosed DM cases, 3) the number needed to screen (NNS) to find a new case of DM, and 4) the determinants of DM among TB patients in Southern Nigeria.

Results
General characteristics. A total 2132 new and previously-treated TB patients (aged ≥ 15 years) were registered for treatment during the study period. However, 38 (1.8%) who were previously-treated TB patients were excluded from the study because they were not eligible. All eligible 2094 newly diagnosed TB patients completed the screening. The socio-demographic characteristics of the patients are shown in Table 1 , were found to be independently associated with increased risk of DM among TB patients. However, having an occupation that engages in vigorous activity (aOR 0.6, 0.4-0.9) was found to be independently associated with decreased risk of DM among TB patients.

Discussion
This study found a DM prevalence of 9.4% among newly diagnosed TB patients. The additional yield of DM was 59% while the NNS to detect a new case of DM was 18. The prevalence of DM among patients who were older than 40 years was far higher than among younger persons; sex on the other hand was not associated with the occurrence of DM. The additional yield of DM was greater among older patients, persons who sought care at a public facility, rural residents and HIV-negative persons. In all, factors favouring occurrence of DM among TB patients in the study group were older age (over 40 years), private facility care and rural residence. Conversely, patients engaged in occupations with vigorous activity were less likely have DM. The DM prevalence of 9.4% among newly diagnosed TB patients found in this study is much higher than was observed in a pilot study in a Nigerian city as well as in the general population, but was lower than a prevalence of 12.3% observed in some health facilities in the same city 2,18,19 . Our finding was also comparable with the results of studies carried-out in Ethiopia (8.3%), Uganda (8.5%) and USA (11.4%) 11,12,20 . However, DM prevalence among TB patients reported in this study was higher than the rates reported from China (6.3%), Spain (5.9%), Guinea-Bissau (2.8%) and Benin (1.9%); but lower than the rates reported from Taiwan (29.5%),  Southern-Mexico (29.3%) and India (25.3%) [13][14][15][16][21][22][23] . Reason/s for the differences in prevalence might be due to disparities in patient background and disease burden. In Guinea-Bissau and Benin where low DM prevalence was observed, the burden of TB in both countries is low 21,22 . In addition, the low DM prevalence observed in these settings may also be explained by the high burden of TB among younger persons, who in general have a lower incidence of DM than those aged (40 years or over) 24 . Furthermore, differences in dietary habits and the screening methods used for DM diagnosis might contribute to the variation in prevalence of DM among TB patients in various regions [13][14][15][16][21][22][23] . DM screening was conducted in large public and private facilities across Southern Nigeria under programme conditions with minimal additional training to the health facility workers. This suggests that the active screening strategy of TB patients for DM is feasible and acceptable. This finding may indicate the possibility of DM screening for TB patients at primary health care (PHC) facilities, and the need to establish its referral linkages in order to ensure quality DM services to patients diagnosed with DM. This would however require the establishment of a functional referral system to ensure access to quality DM services at appropriate secondary and tertiary health facilities.
The NNS to detect a new case of DM among TB patients was 18. This number is higher than the findings in India which showed large variations in NNS to detect DM among TB patients 16 . These differences may be due to the high prevalence of DM in India, the quality of screening test used and the countrywide standard procedure applied for DM screening in India 25 . In addition, we found that NNS needed to detect one case of DM among TB patients decreases as the age increases. Owing to the differences between the various studies, it is crucial to understand the relationship between TB and DM in different regions around the world in order to establish appropriate priorities and adapt the needs of National TB Control Programmes. In high TB burden countries with weak public health systems, the current recommendation is to rule out TB among diabetics and vice versa 15,17,26 . Given the low NNS noted in our setting to detect a DM case, and the need to sustain DM-TB screening programmes in Nigeria, the most cost-effective approach would be to prioritise older TB patients for targeted screening.
The strength of this study is that we were able to implement this screening programme in a routine programme setting across multiple regions and facilities with minimal additional costs and training. The study has a few limitations, however. First, it was carried out at high-end performing TB centres, therefore, the results may not be representative of all TB treatment settings e.g. PHC settings in Southern Nigeria. Secondly, we utilised FBG for diagnosis; FBG testing has low sensitivity and may fail to detect some persons with DM. However, FBG has been recommended as the initial DM screening test in resource-limited settings 5 . The HbA1c has also been recently recommended for the diagnosis of DM among TB patients; however, it is limited by being expensive, cumbersome to perform and inappropriate for screening individuals within routine general health services 17   excluded previously-treated TB patients in this study. Although it could be argued that previously-treated TB patients are a risk group for the development of co-morbid TB-DM, we excluded them because in some of the facilities where the survey was carried-out, these patients were still being investigated for drug-resistant TB. Also, because the optimal time for DM screening among TB patients is not clear 3 , we excluded previously-treated TB patients because it will be difficult to ascertain if the DM diagnosed in this group is as a result of persistent TB disease or due to previous exposures to TB medications both of these situations has been shown to cause persistent hyperglycemia 3,[27][28][29] . Besides, we wanted to establish the burden of DM in treatment-naive TB patients receiving first-line anti-TB treatment in order to provide evidence to advocate for systematic DM screening among the largest burden of adult TB patients in Nigeria. Finally, we did not undertake anthropometric measurements (e.g., body mass index, waist circumference, etc.) of the respondents. A recent study in Pakistan found anthropometric heterogeneity among TB patients with/without DM -those with known or newly-diagnosed DM being more likely to be obese compared with those without DM 30 . Thus, such anthropometric measurements need to be considered in future studies in our setting. This study has potential major policy implications. First, DM screening among TB patients is important for early diagnosis of DM -especially in Nigeria where two thirds of DM cases may be unaware of their status. The current screening strategy can improve early diagnosis of DM and facilitate necessary interventions. Second, it is especially important to screen older TB patients (≥ 40 years). Third, the study highlights the need for similar studies in PHC settings to estimate the burden of DM and the feasibility of decentralizing DM-TB services in rural and remote under-resourced settings.

Conclusions
The burden of DM among TB patients in Nigeria is high. Screening can detect more than half of undiagnosed DM among newly diagnosed TB patients. In this and similar under-resourced settings where prioritization for DM screening is necessary, screening TB patients who are aged > 40 years may yield the greatest benefits. Furthermore, screening among TB patients residing in rural areas and those engaged in occupations with sedentary and moderate physical activity is expected to increase the additional yield of DM cases.

Materials
Study design. This was a cross-sectional implementation study conducted under programme conditions in six States in Southern Nigeria. The study was conducted over an eight-month period from March to October 2015.
Setting and study population. Nigeria is divided into two regions (north and south), and six geo-political zones (three per region) with each having between five to seven States. The study was conducted in the three zones in Southern Nigeria. Patients were recruited through a two-stage sampling technique. First, two States were selected from each of the three geo-political zones in southern Nigeria through simple random sampling. The study states include Edo and Cross River (south-south zone), Enugu and Ebonyi (south-east zone), and Ogun and Ondo (south-west zone). In each selected State, the top-two (three in Ondo State) health facilities notifying and treating the highest proportion of TB patients were identified and selected for the survey. In all, 13 health facilities which offer both TB and diabetes services were selected for the survey.
Eligibility. The inclusion criteria were all adults (≥ 15 years) with newly-diagnosed TB starting treatment.
Patients were ineligible if they had previously-treated TB.
Screening procedure. All newly-diagnosed adult TB patients aged (≥ 15 years) attending the study health facilities were interviewed on their first visit for commencement of TB treatment. DM status was assessed using patients self-reported history regarding DM. All eligible patients irrespective of known diagnosis of DM were investigated as follows: Random blood glucose (RBG) test was performed -if the level was less than 110 mg/dl, no further action was taken. If the RBG was ≥ 110 mg/dl, then the patient was asked to return for fasting blood glucose (FBG) test the next day. A glucometer (ACCU-CHEK Active Glucose Monitoring System, Roche Diabetes Care Limited, Burgess Hill, UK) was used for screening DM. Standard WHO diagnostic criteria were used for making a diagnosis of diabetes mellitus, i.e., fasting blood glucose (FBG) value of ≥ 126 mg/dl 19 . Patients who were found to be diabetic were referred to the diabetes clinic of their facility for further evaluation and management. Moreover, known DM patients were educated on the need to continue follow-up care for DM.
A standardized proforma was used to collect patients' socio-demographic and clinical variables from the TB registers as well as the screening results 17 . Physical activity was assessed based on patient's occupation and classified as 'sedentary' , 'moderate' and 'vigorous' according to WHO classification 31 . Diagnosis of TB was made in line with the WHO/NTP guidelines 32 . Patients found positive for Mycobacterium tuberculosis on smear microscopy or based on MTB RIF Xpert test were considered to have bacteriological-positive TB. Smear-negative TB was defined by negative smear (or Xpert) results with radiology, clinical and/or histology findings consistent with active TB, followed by a clinicians' decision to treat with a full course of anti-TB chemotherapy. Extra-pulmonary TB (EPTB) was based on histological or strong clinical evidence consistent with active EPTB, and a clinicians' decision to treat with a full course of TB chemotherapy. HIV counseling and testing was provided to all persons with presumed TB during submission of sputum specimen for microscopy. Statistical analysis. Data were double-entered, validated for consistency and analysed using Epi Info.
Dataset available as a supplementary information (SI). Patients' demographic and clinical characteristics were described in terms of frequencies and percentages. The main outcomes for analysis were the number and proportion of TB patients with known and/or newly diagnosed DM, further stratified by age, sex, type of TB, residence and physical activity. Bivariate analysis was performed to assess the association of these factors with DM prevalence. Odds ratios (ORs) with 95% confidence intervals (CIs) were used to describe associations between groups. Multivariable logistic regression analysis was performed to estimate adjusted ORs using the full model fits. NNS was calculated for all patients whose DM status was assessed. Additional yield of new cases of DM was calculated using the following formula: (newly diagnosed DM cases × 100)/(known DM cases + newly diagnosed DM cases).
Ethics approval. An informed verbal consent was obtained from each participant before enrolment in the study. We confirm that all blood sugar testing were performed in accordance with relevant WHO guidelines and regulations in use in clinical practice in Nigeria. The Ethics and Research Advisory Board of German Leprosy and TB Relief Association, Nigeria approved the study. Approval was also obtained from the State TB Control Programme in six states selected for the project.