Bidirectional screening and treatment outcomes of diabetes mellitus (DM) and Tuberculosis (TB) patients in hospitals with measures to integrate care of DM and TB and those without integration measures in Malawi

Studies have shown that the risk of developing active tuberculosis (TB) is three times higher in persons with Diabetes mellitus (DM); RR 3.0, 95%CI (2.3, 4.3), than in persons without DM [1‐4]. According to World Health Organisation (WHO) TB screening guidelines, uncontrolled DM doubles the risk of TB treatment failure, relapse and death [1], and both DM type 1 and DM type 2 (T2DM), are risks factors for the development of active TB with T2DM accounting for more than 90% of TB cases attributed to DM [5]. Although studies are still being conducted to better understand the immunological mechanisms of susceptibility of TB among people with Diabetes mellitus (PWD), several factors [6] in relation to complex natural history of TB progression [7] have already been endorsed that would increase the risk of TB among PWDs.

With wide variations between countries and regions that ranges from 1.9% to 45%, the average global DM prevalence is estimated at 16% [8, 9]. Further on TB and DM comorbidity, a meta-analysis that was conducted in 2018 found 5.13% (95% CI 4.34–5.92) pooled prevalence of TB among DM patients in Africa [10]. On the other hand, Peer et al. 2017 [11] reasoned that DM prevalence in the continent might be underestimated since the prevalence of undiagnosed DM is between 50.7 and 75.1% in LMICs.

The higher risk of developing TB in DM patients and the challenges of diabetic TB patients in the treatment and care of both diseases have been well recognized, hence integrated approach for TB/DM health care is recommended by WHO, The International Union against Tuberculosis and Lung Diseases (IUATLD) and The International Diabetes Federation (IDF) [12‐14]. Integration is aimed at improving the service in relation to efficiency and quality, thereby maximising use of resources and opportunities through; promoting access to care for NCD patients, maximizing efficiency given the severe human resource shortages, and replicate strong infectious diseases outcomes for patients with other chronic conditions [15]. However, there is limited evidence on the effects of TB/DM integrated healthcare services especially in poor resource settings.

Due to increased burden of both NCDs and Infectious diseases, a few hospitals in Malawi started providing NCD care services between 2009 and 2017 with assistance from health development partners. However, NCDs were formally included in Health Sector Strategic Plan 2011–2016 in Malawi, and an NCDs Action Plan 2012–2016 was developed in the Ministry Of Health (MoH) [16] to facilitate adoption of the integrated approach to delivery of NCD services [16‐18] in secondary level care hospitals in the country. This small landlocked country with approximately 19 million people [19‐21] has 5.6% DM prevalence, 33% prevalence of hypertension among the adults aged between 25 to 64 years [22], and an estimated 41% of the identified PWDs were undiagnosed [23]. Furthermore, the country is ranked among top 20 global countries with the highest estimated numbers of incident TB cases among people living with HIV; 88 (55–129) per 100,000, and has an estimated TB incidence of 181 (113–265) per 100,000 in general population [24]. However, in Malawi, like other TB high incidence settings, evidence of the actual effect of integrated NCD/TB programs on patients is still missing. Therefore, this study was conducted in Malawi to investigate bidirectional screening and treatment outcomes of DM and TB patients in hospitals with measures to integrate care of DM and TB as compared with those without integration measures. In this study, integration of TB and DM health care was defined as a variety of managerial or operational changes to health systems to bring together inputs, delivery, management and organisation of TB and DM service functions [25] as well as different kinds of DM and TB services or operational programs or activities joined together to ensure and perhaps maximize collective outcomes for both DM and TB Patients. In Malawi, the integration efforts of DM and TB has been well described and categorized within the complex model where NCDs and HIV and TB services are incorporated in one clinic, offered by the same healthcare workers under one roof [15].

The aim of the study was to determine the effects of integrated care on bidirectional screening and treatment outcomes for both TB patients and people with DM recruited from 2016 to 2019 in eight Malawian hospitals through retrospective chart review analysis. In this study, we investigated bidirectional screening coverage, proportion of DM cases among TB patients and proportion of TB cases among people with DM, described effect of integration on TB treatment success, diabetic complications, treatment loss to follow-up for both diseases, retention into DM care and mortality of TB patients and people with DM.

As stated on the objectives above, this study assessed the following variables: TB screening coverage among people with DM and vice versa, PWDs retention on care and treatment adherence, TB treatment outcomes as defined by WHO [26], and DM complications related to uncontrolled DM by comparing an integrating hospital with non-integrating hospitals.

We conducted a retrospective chart review analysis for patients’ data from an integrated chronic care clinic at Neno Hospital [15], and seven other hospitals that are operating TB and DM care services on the traditional approach. The study was conducted at secondary level healthcare in a poor country where 84% of the population lives in the rural areas as compared to 16% in urban centres [27]. This study was implemented within a quasi-experimental study and policy analysis research that is being conducted in six government district hospitals namely, Kasungu, Dowa, Bwaila (Lilongwe), Ntcheu, Neno and Mzuzu Health Center; and two Christian Health Association of Malawi (CHAM) hospitals namely, David Gordon Memorial (DGM) in Rumphi district and Embangweni Mission in Mzimba district. Retrospective data of patients aged ≥ 15 years old between January 2016 to August 2019 were collected by using hypertension and DM and TB treatment records.

Two sample proportions commands were applied in Stata Version 15.1 to calculate a sample size which was stratified by integrated and non-integrated care on a ratio of 1:7. We assumed 0.05 significant level, 0.90 power, 40% and 45% bidirectional screening coverages in non-integrated care setting for DM and TB respectively, and 60% for both TB and DM in integrated care. The overall sample size for DM patients was 584 and for TB patients was 1048. As shown on Fig. 1, patients were selected through simple random sampling method from the list of the available records. This method was deemed appropriate since the list of the participants were available at each study site. When data was missing for a randomly selected patient identification, a replacement was obtained from the remaining records following the same process. However, due to limited available records and missing data, more records of both TB patients and PWDs who were still on treatment were included.

Data were collected in August 2019—September 2019 by using a checklist that were developed from the data elements adopted on the DM and hypertension treatment cards and TB treatment cards that are used in hospitals in Malawi. However, data on diagnostics tests that were performed to confirming the cases was not collected since such information is not recorded on the cards, and high proportion of data from patients who were still on treatment were collected due to missing data or lack of data on the old records from respective sites.

After development of data collection tools, research assistants were selected and trained and coordinated by the principal investigator to facilitate data collection exercise. These trained research assistants worked collaboratively with laboratory managers, Health Management Information System Officers, TB officers and NCDs coordinators in respective study sites. Android tablets with the digital version of Open Data Kit (ODK) were used to collect and organise the survey data, allowing for immediate data validation in the field [28].

Descriptive statistics in univariate analysis were performed and thereafter bivariate and multivariate analyses were performed to analyse the risk differences (RD) in variables between the two settings. Frequency tables, proportions and risk differences of the key findings were performed. After descriptive data analysis, age and sex were assessed for confounding by applying multinomial logistics regression (mlogit) with survey data setting in Stata 15.1 [29, 30]. Regressions were performed to assess the effects of sex and age on the association between the treatment outcomes and complication on their relationship with integrated care. Thereafter, calculated coefficients in the performed models were transformed to risk ratios and later converted into risk differences by direct commands in Stata 15.1.

The study followed the ethical principles as outlined in the World Medical Association (WMA) declaration [31, 32]. The study was approved by both Ministry of Health in Malawi through National Health Sciences Research Committee (NHSRC) (approval number 2187), and by the ethical committee of the University of Freiburg (approval reference number 29/20). Data were stored in an encrypted online database, from where they were exported via excel file to Stata version 15.1 for statistical analysis. Data were anonymised before analysis.

As shown on Fig. 1, records for 557 DM patients and 987 TB Patients were analysed in this study. 10.6% (n = 105) of the total 987 TB patients, and 11.5% (n = 64) of the total 557 PWDs were treated in the integrated clinic. There was a male predominance in the TB patient group (69.5%; n = 686) and a female predominance in the PWDs (58.0%; n = 323), which was similar in both the integrated and non-integrated setting.

TB patients were younger than PWDs (mean age 41 and 54 years, respectively) as shown on Table 1 and in Additional file 1: Table S1. 107/557 (19.2%) of the PWDs had Type 1 diabetes, 39/987 (4.0%) of the TB patients had MDRTB; see Table 2. 259/557 (46.5%) PWDs were hypertensive, 9.2% of PWDs had cardiovascular diseases whereas renal diseases were the least prevalent condition among PWDs (1.4%). On risk factors, 8/463 (1.7%) of the PWDs were smokers, 15/455 (3.3%) consumed alcohol, 89/198 (45.0%) consumed fruits and vegetables less than two times a week and 91/221 (41.2%) make less than two 30-min exercise sessions per week, respectively. Differences between integrated and non-integrated hospitals on PWDs performing daily 30 min of exercises involving planned activities such as jogging or walking sessions performed in the week as defined by WHO [33] or consumption of at least three or more times of fruits and vegetable portions per week were not significant, of course, with high percentages of missing records for these variables as seen on Table 2.

On Table 3, 10/105 (9.5%) TB patients were screened for DM in integrated care while 43/882 (4.9%) were screened for DM in non-integrated care setting, RD 4.6, 95%CI (− 1.1, 10.4), P = 0.065 respectively. Of the total TB patients screened, 5/53 (9.4%) were diagnosed with DM. Among PWDs, 36/64 (56.3%) were screened for TB in integrated care setting as compared to 5/493 (1.0%) in the traditional non-integrated care delivery system care RD 55.2%, (95%CI 43.0, 67.4), P < 0.001, and 5/41 (12.2%) of the screened DM patients were diagnosed with TB.

As shown on Table 4, 71/508 (14.0%) TB patients were lost to follow-up in non-integrated care hospitals, while in an integrated care, none were lost to follow-up in integrated care; RD -14.0%, (95%CI: -17.0,-11.0), p < 0.001, and TB treatment success was 92.5% in integrated cares, while in non-integrated care the treatment success was 78.5%, RD 14, (95%CI 7.6, 20.4); P = 0.001.

Among people with DM, 2/64 (3.1%) experienced loss to follow up in integrated care and 40/493 (8.1%) in non-integrated care, RD -5.0%, (95%CI: -10.0, -0.0%); P = 0.046.

On retention into care by PWD, 52/64 (81.3%) PWD in integrated care were in care for ≤ 3 years while in traditional non- integrating setting 387/493 (78.5%), RD 2.8%, 95%CI (-7.5, 13.0), P = 0.630 were recorded. Furthermore, aggregated data on Table 4 shows that after ≥ 2 years of follow up, 62.5% PWDs in integrated and 41.8% PWDs in non-integrated care were retained in care, RD 20.7, (95%CI: 8.1, 33.4), P = 0.001.

On blood glucose monitoring, 27/64 (42.2%) PWDs were checked for fasting blood glucose (FBG) assessment at 6 months at an integrating hospital, and 126/493 (25.6%) were checked at non-integrating hospitals RD 16.6%, (95%CI: 4.7,28.5), P = 0.005, while at 12 months; 32/64 (50.0%) were checked at integrating hospital while 106/493 (21.5%) were checked at non-integrating hospitals, 95%CI (15.3,41.7), P < 0.001, respectively (table 4). Additionally, 423/493 (85.8%) of PWD had their weights checked after 6 months into care at an integrated care hospital, while 28/64 (43.8%) were checked during the same period in non-integrated care hospitals.