Integrating tuberculosis and noncommunicable diseases care in low- and middle-income countries (LMICs): A systematic review

Background Low- and middle-income countries (LMICs) are facing a combined affliction from both tuberculosis (TB) and noncommunicable diseases (NCDs), which threatens population health and further strains the already stressed health systems. Integrating services for TB and NCDs is advantageous in tackling this joint burden of diseases effectively. Therefore, this systematic review explores the mechanisms for service integration for TB and NCDs and elucidates the facilitators and barriers for implementing integrated service models in LMIC settings. Methods and findings A systematic search was conducted in the Cochrane Library, MEDLINE, Embase, PubMed, Bibliography of Asian Studies, and the Global Index Medicus from database inception to November 4, 2021. For our search strategy, the terms “tuberculosis” AND “NCDs” (and their synonyms) AND (“delivery of healthcare, integrated” OR a range of other terms representing integration) were used. Articles were included if they were descriptions or evaluations of a management or organisational change strategy made within LMICs, which aim to increase integration between TB and NCD management at the service delivery level. We performed a comparative analysis of key themes from these studies and organised the themes based on integration of service delivery options for TB and NCD services. Subsequently, these themes were used to reconfigure and update an existing framework for integration of TB and HIV services by Legido-Quigley and colleagues, which categorises the levels of integration according to types of services and location where services were offered. Additionally, we developed themes on the facilitators and barriers facing integrated service delivery models and mapped them to the World Health Organization’s (WHO) health systems framework, which comprises the building blocks of service delivery, human resources, medical products, sustainable financing and social protection, information, and leadership and governance. A total of 22 articles published between 2011 and 2021 were used, out of which 13 were cross-sectional studies, 3 cohort studies, 1 case–control study, 1 prospective interventional study, and 4 were mixed methods studies. The studies were conducted in 15 LMICs in Asia, Africa, and the Americas. Our synthesised framework explicates the different levels of service integration of TB and NCD services. We categorised them into 3 levels with entry into the health system based on either TB or NCDs, with level 1 integration offering only testing services for either TB or NCDs, level 2 integration offering testing and referral services to linked care, and level 3 integration providing testing and treatment services at one location. Some facilitators of integrated service include improved accessibility to integrated services, motivated and engaged providers, and low to no cost for additional services for patients. A few barriers identified were poor public awareness of the diseases leading to poor uptake of services, lack of programmatic budget and resources, and additional stress on providers due to increased workload. The limitations include the dearth of data that explores the experiences of patients and providers and evaluates programme effectiveness. Conclusions Integration of TB and NCD services encourages the improvement of health service delivery across disease conditions and levels of care to address the combined burden of diseases in LMICs. This review not only offers recommendations for policy implementation and improvements for similar integrated programmes but also highlights the need for more high-quality TB–NCD research.


Methods and findings
A systematic search was conducted in the Cochrane Library, MEDLINE, Embase, PubMed, Bibliography of Asian Studies, and the Global Index Medicus from database inception to November 4, 2021. For our search strategy, the terms "tuberculosis" AND "NCDs" (and their synonyms) AND ("delivery of healthcare, integrated" OR a range of other terms representing integration) were used. Articles were included if they were descriptions or evaluations of a management or organisational change strategy made within LMICs, which aim to increase integration between TB and NCD management at the service delivery level. We performed a comparative analysis of key themes from these studies and organised the themes based on integration of service delivery options for TB and NCD services. Subsequently, these themes were used to reconfigure and update an existing framework for integration of TB and HIV services by Legido-Quigley and colleagues, which categorises the levels of integration according to types of services and location where services were offered. Additionally, we developed themes on the facilitators and barriers facing integrated service delivery Introduction Tuberculosis (TB), an infectious disease derived from various species of the genus Mycobacteria, remains a global threat to public health [1]. Although preventable and treatable, with 1.5 million deaths every year, TB is still on the world's top list of infectious killer diseases [1]. Furthermore, with the advent of multidrug-resistant tuberculosis (MDR-TB), first-line drugs have been rendered ineffective [2,3]. An estimated 10 million active TB cases were recorded in 2020, which consisted of 5.6 million men, 3.3 million women, and 1.1 million children [4]. More than 95% of TB deaths occurred in low-and middle-income countries (LMICs), with 8 LMICs accounting for two-thirds of the total cases, with India having the highest number of cases, followed by Indonesia, China, the Philippines, Pakistan, Nigeria, Bangladesh, and South Africa [1].
As the global population ages, LMICs will face an increasing brunt derived from noncommunicable diseases (NCDs). As defined by the World Health Organization (WHO), NCDs are chronic diseases lasting for a long duration and are a result of a combination of genetic, physiological, environmental, and behavioural factors [5]. NCDs, such as diabetes mellitus (DM), cardiovascular disease (CVD), chronic lung disease, cancer, and mental health illness (MHI) among others, are increasing in prevalence [6]. Each year, 41 million people die from NCDs, with 85% of these deaths taking place in LMICs [5]. In particular, CVDs account for 17.9 million deaths, followed by cancer at 9.3 million and DM at 1.5 million [5].
Global population health data posit that progress towards meeting TB control targets, including the expansion of TB intervention programmes, is hindered by the synergistic relationship between TB and NCDs [7,8]. This is strongly attributed to NCDs and their associated risk factors such as CVD, DM, and smoking and alcohol abuse being risk factors for TB, consequently impacting the burden of TB at the population level [7,9].
At the societal level, the TB and NCDs share numerous underlying social determinants, such as high levels of poverty, poor health literacy, social and cultural values, and lack of social protection [10][11][12][13]. Additionally, coexisting communicable and NCDs increases the risk or effect of each other [14]. However, the current healthcare systems of LMICs might not have sufficient resources to respond to this combined burden of diseases derived from communicable and NCDs [15]. At the same time, disease-specific interventions are not the most effective approach for tackling these afflictions [16,17].
A known example is DM, which significantly increases the risk of TB and vice versa. The tight interplay between disease conditions at the biological level complicates the management of TB and contributes to poor TB treatment outcomes [18]. Concomitantly, TB can increase the risk of or aggravate DM status [19]. Other known linkages between TB and NCD typologies include that of CVD and TB [20], chronic lung disease and TB [21], MHI and TB [22,23], and cancer and TB [24]. These mutual predispositions necessitate better integration of health services to place TB and NCDs under control in LMICs after accounting for specific disease burdens within each country context. Therefore, there are multiple ways in which service integration can be manifested. Integration covers disease-specific (vertical) programmes and system-wide (horizontal) pillars and initiatives, overlaying intersecting disease services (e.g., DM and TB treatment services), across community and tertiary level interventions, and between health and nonhealth sectors. This review focuses on integrating service delivery for diseases that are usually delivered separately but often affect the same types of end users. We also acknowledge that service integration goes across and beyond the status quo of a single healthcare entity or stakeholder and have therefore not only consolidated information on the different models of service integration but also provide practical steps (incorporating facilitators and avoiding barriers) when implementing TB and NCD integrated programmes.
Despite the increasing burden of disease derived from TB and NCDs in LMICs and the need for TB and NCD integrated programmes, most systematic reviews on TB service integration investigate the integration of TB-and HIV-related services or the health outcomes of TB and NCD comorbidities, their syndemic interactions, and prevalence [25][26][27][28][29][30][31][32][33][34]. Therefore, there remains a dearth of information regarding the mechanism of integration of TB and NCDs related services in LMICs, which this systematic review aims to address. Hereafter, this systematic review will present integration models that bring together the different services for TB and NCDs; points of service delivery; process modifications; and change management strategies through a quantitative and qualitative lens.

Methods
A protocol for this study has been registered on the PROSPERO international prospective register of systematic reviews (registration number: PROSPERO CRD42020202745). This study is reported as per the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (S1 and S2 Checklists) and illustrated in a flowchart below.
First, the findings from the literature review will be summarised based on its depth of integration, from least to most integrated, with each being delved into detail. A framework that illustrates the models of service integration will be developed from this step. Second, using our organised findings from the first step, WHO's 6 operational "building blocks" of a health system, a well-established framework that identifies health system facilitators and constraints will be used to understand the facilitators and barriers for each integration model [35]. This step involves mapping various integration characteristics derived from the literature review to each dimension of the framework.
Drawing on the definitions proposed by Atun and colleagues [36], Dudley and Garner [16], and Legido-Quigley and colleagues [37], we operationally define programmatic integration as managerial or operational reorganising of health systems components to consolidate inputs, delivery, management, and organisation of particular service functions as a means to improve coverage, access, quality, acceptability, and (cost)-effectiveness [38]. For the purposes of this review, we consider the various tenets of integration that encompass the integration of service provision for both TB-and NCD-related services in terms of physical point-of-care delivery and types of services that include screening, testing or referral, and nonmedically related services like financial and social support.

Study selection
Type of chronic conditions. To be considered for inclusion in the current review, the studies had to integrate TB with NCDs. The list of NCDs included are chronic CVDs (coronary heart disease, cerebrovascular disease, peripheral arterial disease, and hypertension), DM (type 1 and 2), cancers (breast cancer, cervical cancer, colorectal cancer, prostatic cancer, and lung cancer), chronic obstructive pulmonary disease (asthma, chronic bronchitis, and emphysema), and mental health conditions (depressive disorders, vascular dementia, and Alzheimer disease but exclusive of alcohol and substance misuse). We have focused on the integration of care of the established diseases and excluded the risk factors of the diseases for this study.
Inclusion criteria. Publications were included if they described or evaluated management or organisational change strategy made within an existing health system to increase integration between TB and NCD management at the service delivery level in LMICs. Services could be provided in health facilities or the community. Reports had to describe the experience of integration, not a theoretical account of how integration might be implemented. We included articles that studied populations of all age groups at the community, primary, secondary, or tertiary care level or venues that we have detailed in the tables below. We also included articles irrespective of drug sensitivities. The extracted articles included both qualitative and quantitative studies and their subcategories, including but not limited to retrospective and prospective cohort, cross-sectional, case-control, ethnographic, randomised control, and observational studies. We also reviewed studies conducted in languages other than English.
Exclusion criteria. To ensure a comprehensive descriptive review, we did not exclude studies based on their design or absence of outcome measures. Health promotion or disease prevention papers were excluded even if the activity was delivered in a TB or NCD treatment setting. We excluded opinion, editorial, and correspondence pieces on care integration without actual implementation of integrated programmes. Studies not conducted in the LMICs were also excluded.
Search strategy. The search strategy was developed to be consistent with methods used by other authors for systematic reviews of integration of health services. We searched databases such as the Cochrane Library, MEDLINE, Embase, PubMed, Bibliography of Asian Studies, and the Global Index Medicus from their inception to November 4, 2021. The search terms used for these databases are shown in Box 1.
We conducted hand searching and consulted expert reviewers to check the completeness of the electronic searches and included additional papers as appropriate. The World Bank List of Economies 2019 was utilised as reference of the LMICs for the study [39]. The list of the LMICs was not included in the search term, but articles resulting from the database search were reviewed for LMICs during the assessment of the full articles.

Search and retrieval of studies
Two reviewers (PS and LT) independently went through the list of articles from the electronic database search results to identify articles relevant to this systematic review based on title or title and abstract. If either of the 2 reviewers considered a study potentially eligible, the full article was retrieved for further assessment. Further potential studies were also identified by screening the reference lists of included articles. Two reviewers assessed the full texts independently to evaluate whether they met the inclusion criteria for this review. Any disagreements concerning studies to be included were resolved by consensus or by discussion with a third reviewer (QD). There were no language restrictions.

Data extraction
A data extraction format was developed in Microsoft Excel. Data were extracted from each study on the country, study setting, duration of integration programme, model of integration, Box 1. Search strategy used for the Cochrane Library <1945 to November 4, 2021>, Ovid MEDLINE(R) <1946 to November 4, 2021>, Embase <1974 to November 4, 2021>PubMed<1950 to November 4, 2021>, Bibliography of Asian Studies <1946 to November 4, 2021>, and the Global Index Medicus <inception to November 4, 2021> entry into the health system, screening tools used, study methods, outcomes, and facilitators and barriers to programme implementation. Data extraction was performed independently by 2 reviewers (PS and LT) and compared. To further ensure accuracy, a third reviewer (CDF) not involved in searching and retrieving studies checked for disparities. Erroneous and inconsistent data sets reported in the studies were excluded. However, no discrepancies in data were identified.

Quality assessment (risk of bias)
Included studies were independently assessed for quality and risk of bias by 3 reviewers (PS, QD, and CDF). The quality (risk bias) assessment tools used were selected and adapted from a review of tools performed by Ma and colleagues, which provides an overview of the relevant tools to evaluate methodological quality for different types of studies [40]. Therefore, Newcastle-Ottawa Quality Assessment Form for Cohort Studies was employed for retrospective and prospective cohort studies [41]. For case-control studies, Scottish Intercollegiate Guidelines Network (SIGN) critical appraisal checklists for case-control studies was used [42]. For crosssectional studies, Joanna Briggs Institute (JBI) Critical Appraisal Checklist for Analytical Cross-Sectional Studies tool was deployed [43]. For mixed methods studies, the Mixed Methods Appraisal Tool was used [44]. The most relevant and salient features extracted from each study based on the aforementioned tools are reported in the tables below.

Data analysis and synthesis of results
First, we presented the relevant quantitative and qualitative findings in the tables below. As such, we reported the quantitative indicators such as relative risks, odds ratios, prevalence ratios, and prevalence with their associated 95% confidence intervals and number needed to screen and number needed to test where appropriate. We also reported qualitative findings that included the perspectives and accounts of patients and providers regarding the TB and NCD integration programmes, which they were a part of.
Second, 4 reviewers (PS, LT, QD, and CDF) reviewed the data and organised the findings to develop a new framework of integrated services for TB and NCDs through relevant themes extracted from the reviewed studies. To that end, a previously published and validated framework for integrating TB and HIV service delivery by Legido-Quigley and colleagues was reconfigured and updated to address the salient features of integrative TB-and NCD-related services [37]. In the original framework by Legido-Quigley and colleagues, users enter the health system through TB services, where in the least integrated model, the users are referred to other sites for HIV testing and subsequent care. With closer integration, TB clinics can also test for HIV on-site but continue to refer to another service for HIV treatment. Where users enter through HIV services, in the least integrated model, the users are referred to a TB clinic for screening for active TB; in a more integrated model, TB screening is undertaken within the HIV clinic, but users are referred to other sites for TB treatment services. In the most integrated model, treatment for both HIV and TB is provided in one health facility. Hence, the different levels of integration and the types of services provided were concepts drawn from the original framework and incorporated into the newly synthesised TB and NCDs framework below.
Third, the 4 reviewers (PS, LT, QD, and CDF) reanalysed the data thematically to derive themes on the facilitators and barriers for TB and NCD service integration and subsequently mapped the themes into WHO's 6 operational "building blocks" of a health system framework.
However, given the significant heterogeneity in study designs, participants, contexts, and outcomes reported in each study, we were unable to perform meta-analyses. Instead, a descriptive synthesis was performed.

Overview of included studies
The screening process of this review adapted the PRISMA flowchart (Fig 1). Our database search yielded 4,906 articles. After the removal of duplicates, 3,484 articles were screened for titles and abstracts. A total of 184 full articles on integration of care of TB and NCDs were further assessed for eligibility, and 22 articles met the inclusion criteria for this review.

Characteristics of included studies
Of the 22 articles included, 13 were cross-sectional studies, 3 were cohort studies, 1 was a casecontrol study, 1 prospective interventional study, and 4 were mixed methods studies. A brief summary of the key characteristics of each study is provided in Table 1 had a low risk of bias, 11 studies had a moderate risk of bias, 3 studies had a high risk of bias,  and 1 study did not have enough data for risk bias analysis (elaborated in Tables 2-6).
Seven studies were conducted in Asia, 11 in Africa, and 4 in the Americas (Fig 2). Studies conducted in Africa and Asia focused more on the integration of mental health care in TB

Synthesised model of integration based on literature review
A new framework of integrated care for TB and NCDs was developed where the depth of integration is stratified according to distinct levels 1, 2, and 3 in order to distinguish the integration of an additional service from the ones that are preexisting and/or stand-alone. The higher the numerical value, the deeper the level of service integration. Models of integration ranged

PLOS MEDICINE
Integrating TB and NCDs care in LMICs: A systematic review (1) The number of participants lost to follow-up and declined participation was much smaller compared to the total sample size of both case and control groups.
(2) A representative sample of controls were recruited from the same community.
(3) Potential recall bias among positive TB cases as they are more likely to recall a previous NCD diagnosis as compared to a participant without TB.
(4) Possibility of misclassification bias due to the limitations of urinalysis testing for DM and CKD. Restrepo  (2) Inherent biases in data sets might be amplified when the study merged data sets from different population sources.
(3) TB infection was not confirmed by culture screening in a substantial number of participants, which might have led to an overreporting of actual TB cases. Dasa  Low risk of bias overall as: (1) Sample is representative of the population, and the nonresponse rate was 5%.
(2) PHQ-9 tool had been translated into local languages and validated before deployment.
(Continued ) (1) Tools used for screening had a high Cronbach's Alpha.
(2) All analysis models were adjusted for confounders.
(3) Study was limited to 3 selected districts and might not be representative of the entire South African population.
(4) Study used a count of NCDs as a measure of multimorbidity, implying that each chronic condition has an equal impact on an individual, whereas in reality, disease severity, the specific combination of NCDs, and access to healthcare affect multimorbidity.
(Continued ) had uncontrolled blood pressure than men (16%). The authors declared, although without actual evaluation, the feasibility of the integrated programme for the detection of hypertension and DM while acknowledging the need to further strengthen policies and guidelines as advocated in the Bali Declaration Moderate risk of bias overall as: (1) Data were only collected from urban DOT clinics and is not representative of the entire population.
(2) Only TB patients referred to DOT clinic were invited for screening. Thus, TB patients not referred to health services were not represented. The study also reported qualitatively on the positive and negative experiences of patients and facilitators and challenges of healthcare providers.
Low risk for bias overall as: (1) Validated tools were used for data collectio,n and confounders were adjusted for during quantitative analysis.
(2) Qualitative portion of the study made use of maximum variation sampling.
( Moderate risk of bias overall as: (1) 6.6% of patient data on MHIs were missing and not included in the analysis.
(2) Data collectors were trained in information extraction for MHI diagnosis from health cards and patient registers to increase the accuracy of data collected.
(Continued ) Moderate risk of bias: (1) Individuals who were deceased, untraceable and those who attempted but failed to link to care, never linked to care or linked to care but not within the predefined period were categorised as not having linked to care, which led to underreporting of actual numbers that accessed linked care.
(2) Data on the telephonic followups conducted by the mobile unit counsellors 1 week postdiagnosis was of suboptimal quality, and thus the association of successful contact on linkage to care could not be assessed accurately.
(3) Some contact details were lost by the staff of the mobile testing unit, resulting in not all eligible participants being reached.
(4) Trained and experienced counsellors conducted follow-up sessions (telephone calls and/or home visits) to ascertain outcomes in the clients' preferred language to minimise respondent bias.
(Continued ) The authors acknowledge that need for more studies to better understand the magnitude of comorbid TB and depression as well as to elucidate the sensitivity and specificity of the PHQ-9 among individuals affected by TB.
High risk of bias overall as: (1) PHQ-9 tool, which is usually selfadministered, was administered by health providers instead due to poor literacy among study population which might have led to social desirability bias.
(2) A high dropout rate of 24%, whereby patients did not return for sputum culture, making it impossible to diagnose TB accurately. Therefore, the prevalence of TB might have been underreported.

25-item Johns Hopkins Symptom Checklist to screen for depression and anxiety MSPSS to screen for low social support
In the study population, 9% received HAP counselling, 85% received information materials, 59% received an education session, and 36% received at least 1 group session. Eight group sessions were conducted in total. The qualitative component of the study reported that the intervention packages were acceptable to the patients, including the screening, information provided, support group and counselling services. The programme was also able to successfully train individuals with no prior experience of psychological counselling to deliver the HAP services.
Low risk of bias overall as: (1) 25-item Johns Hopkins Symptom Checklist was selected as it had previously been translated and culturally adapted for use Nepal for the identification of depression and anxiety among internally displaced people, has been shown to have moderate validity.
(2) MSPSS tool was translated from English to Nepali by the research team and verified by back-translation.
(3) Untested interrater reliability between the 2 counsellors that performed the screening at each centre might lead to discrepancies in screening.
(Continued ) High risk of bias overall as: (1) Not every clinics had an MHP available for participation in the study. Thus, findings might not be generalisable.
(2) Qualitative coding process reported a high level of agreement between coders.
(3) During the data collection phase, a very high-profile event involving the death of at least 94 mental health patients occurred. This could have caused an abnormal increase in mental health awareness, and participant responses might not have been representative of past procedures in patient management.
(4) The purposive selection of clinics may have introduced bias. The study team and participating districts identified rural and urban clinics subjectively rather than using a specific metric to define these regions. General physical and mental health were assessed with the SF-12 K-10 to measure anxiety or depressive disorders PC-PTSD Screen to assess PTSD symptoms 10-item AUDIT Participants were asked if they had other NCDs and if they were on any related treatments. No particular tool was used for the screening of nonmental health-related conditions. At baseline, 34.1% of the study population had severe psychological distress with a higher K-10 cutoff score (�28), 81.1% had moderate psychological distress with a lower K-10 cutoff score (�16), and 29.4% had PTSD symptoms (2 or more). At the 6-month follow-up, severe psychological distress with a higher K-10 cutoff score (�28) was significantly reduced by 12.3%, moderate psychological distress with a lower K-10 cutoff score (�16) was reduced by 24.9%, and PTSD was reduced by 20.0%. The authors commented that mental health services and social welfare services should be integrated into TB treatment programmes over a period of time to observe significant results.
Moderate risk of bias overall as: (1) Cronbach's alpha for the K-10 screening tool was 0.92 at baseline and 0.87 at followup.
(2) Cronbach for the PC-PTSD screening tool was 0.89 at baseline and 0.92 at followup.
(3) More than twice the number of male than female participants; thus, the difference in screening and treatment outcomes between genders might not be pronounced.
(4) Most study variables were assessed by interview, and socially desirable responses might have been provided. (5) Findings were limited to TB patients with alcohol problems on anti-TB treatment in public primary healthcare facilities.
(Continued ) Low risk of bias overall as: (1) AKUADS screening tool used, which was indigenously developed, was deployed in the local Urdu language and was validated in the local population.
(2) High level of data consolidation allowed for 96.6% of patient TB treatment outcome measurements to be analysed. from the type of services provided and point of entry into the healthcare system via TB or NCDs or both. We identified 7 models of integrated care and illustrated them in the framework below (Fig 3). At level 1, the least integrated model, (i) patients entering the system via TB service are offered the additional service for screening for NCDs, and likewise (ii) the patients entering via NCDs service are screened for TB.

PLOS MEDICINE
Integrating TB and NCDs care in LMICs: A systematic review At level 2, moderate level of integration, (iii) patients entering from TB services are screened for NCDs and patients with potential or positive diagnosis are referred for care at appropriate healthcare facilities, and likewise (iv) patients entering the system from NCD services are screened and referred for TB treatment. Another model is also identified at this level, where (v) patients enter the system via both TB and NCDs services and are referred for the necessary additional TB-or NCD-related care after bidirectional screening is performed.
At level 3, the highest level of integration, (vi) patients entering the system via TB services are screened for NCDs, and (vii) patients entering the system via NCDs services are screened for TB and both models provide treatment or care management for both TB and NCDs at the same location (colocation).
The number of articles found at each level of integration is expressed in brackets, and entry points differentiated based on underlying known condition, be it TB or NCDs.

Level 1 integration
Entry via TB service, testing for NCDs. This model was described in 8 studies whereby patients who entered the system by TB services were screened for 1 or more NCDs to establish the presence of comorbidity between TB and NCDs. In all 8 studies, existing and/or new TB patients were screened for NCDs at the location where patients received care for TB. A total of 5 studies screened for single NCDs, of which 4 screened for MHIs and 1 for DM. Three other studies screened TB patients for multiple NCDs such as DM, hypertension, and chronic kidney disease (CKD).
This integration model was reported in a case-control study by Byrne and colleagues in Peru, which explored the feasibility of screening for NCDs in patients receiving treatment for TB. In this study, the programme offered point-of-care screening for NCDs such as DM, hypertension, and CKD to 177 TB patients and 161 controls through the provision of detailed medical history taking, obtaining anthropometric measurements, clinical assessment with standardised measurements of resting blood pressure, and urinalysis. The authors reported the possibility of missing the diagnosis of CKD as the urinalysis procedures used offered only a moderate specificity of 53% as a shortfall in the study [45].
Similar issues of screening accuracy were raised in a cross-sectional study by Restrepo and colleagues in Mexico. This study aimed to estimate the risk of TB attributable to DM and identify opportunities for TB prevention among patients with TB and DM. This study employed EDTA-treated blood and batch tested it for glycosylated hemoglobin (HbA1c) to test for DM for 333 TB patients. However, the study reported overestimation of prevalence of DM from transient hyperglycemia triggered by TB [46].
Dasa and colleagues in Ethiopia recognised that the cost of screening and treatment was a barrier to patients seeking health services. The cross-sectional study assessed the prevalence and associated factors for depression among TB patients and highlighted the difficulty in covering costs indirectly related to treatment, despite anti-TB drugs being provided for free, which hindered the effectiveness of TB programmes. This was due to the expenses for additional nutritional needs and transportation costs that might result in other knock-on effects such as missed workdays due to fatigue, chest pain, and symptoms of TB, which meant lower earnings and, in turn, heightened psychological distress [47]. Closely related to the concept of financial support, the study conducted in Manila, Philippines by Masumoto and colleagues stated the use of social support scale (Duke-UNC Functional Social Support Questionnaire) being deployed to understand the level of social support received when measuring the depressive states of 561 TB patients [48].
The relatively low number of TB patients needed to be screened for a positive diagnosis such as 24 for hypertension and 5 for proteinuria had been illustrated by Byrne and colleagues as a potentially feasible screening protocol for an integrated TB and NCD model [45]. We acknowledge that the absence of process indicators for evaluating the programme's feasibility for other NCDs, such as the number needed to screen for MHIs, prevents us from concluding the programme's feasibility for all NCDs. However, level 1 integration models assist in the early detection of cases otherwise missed. The low cost of implementation, by virtue of using simple and easily deployable screening tools and existing human resources, empowered this integration model [45,47].
The validity of the screening tools used and the probability of socially desirable responses provided by the participants for questionnaire type screening tools were concerns identified by some authors [45,[48][49][50]. Some studies also reiterated other weaknesses of this model such as the lack of confirmatory tools for definitive diagnosis of NCDs after an initial round of positive screening and having heavier workloads on healthcare teams due to additional integrated services could compromise existing TB programmes [45]. However, the authors of 3 studies reported that despite the lack of accuracy of screening tools used, the ease and practicability of screening for multiple NCDs simultaneously proved beneficial for patient management [45,48,51].
Other challenges to the level 1 integration model include the lack of human resources and poor data collection systems for providers, whereas the cost of the added services and logistical barriers (repeated visits) deterred patients from enrolling into the integrated programmes [45,51].
Entry via NCDs, testing for TB. There were no studies on an integrated care model with entry via NCDs leading to screening for TB.

Level 2 integration
Entry via TB service, testing and referral for NCD treatment services. Two studies described this integration model where patients entered the system via TB programmes. One study screened for DM and hypertension and the other for DM and MHIs. Both studies included the assessment of risk factors for NCDs and referred the patients for required treatments at appropriate care facilities.
This model was observed in a cross-sectional study by Contreras and colleagues in Peru. Here, TB Cero, a collaborative programme between the Health Ministry and a nongovernmental organisation (NGO) "Socios En Salud" (SES), screened for NCDs such as DM and MHI, and HIV for 192 TB patients. The programme mobilised a field team that functioned using case managers and patient advocates that connected TB patients with appropriate public health facilities for DM and MHI treatment. The costs of treatment were covered by the NGO and TB patients also received social assistance for their unmet basic needs [49].
In the mixed methods study by Anand and colleagues in India, a 2-staged integrated screening for NCDs such as DM and hypertension and their risk factors among 410 TB patients was conducted to assess the yield, feasibility, and acceptability of the programme using process indicators. The patients diagnosed with the DM and/or hypertension were referred to the NCD clinic within the hospital [50].
Manpower and training regarding the screening for NCDs as well as making appropriate referrals were essential for this model to succeed. The collaboration and partnership between the involved healthcare facilities (NCD and TB clinics) are the strengths of this model [49]. Both the patients and the providers appreciated this model of care integration as the added services led to the diagnosis of NCDs otherwise missed [49,50]. Incorporating such services was deemed relevant and important in controlling the TB epidemic as studies have shown comorbidities with NCDs and their risk factors such as low socioeconomic status affect TB treatments and patient outcomes [49,50].
The model by Contreras and colleagues in Peru mentioned above had incorporated an additional yet salient module of managing TB and NCDs, which addressed certain social aspects of patients. This encompassed the provision of food vouchers and home-based treatment support. The lack of social support and basic necessities such as food are known risk factors for TB and NCDs, making this aspect a crucial area that will be delved further in our discussion section [49]. However, the providers indicated the absence of standardised reporting for NCDs services in TB clinics to be a weakness in this integration model. Limitations within the health facilities also affected service delivery as reported by Contreras and colleagues and included the lack of a separate room for NCD screening, which raised privacy concerns among patients and which impeded the uptake of NCD screening services [49].
Other barriers to the level 2 integration model were much like the barriers found in the level 1 integration model, which included long waiting time for screening and referrals, and out-of-pocket (OOP) cost for patients, whereas the challenges for the providers were the additional workload, lack of training to conduct integrated screening services, and shortage of staff and medical supplies for NCD screening.
Entry via NCDs, testing and referral for TB treatment services. Three studies elaborated on this model of care integration whereby patients entered the system via NCDs and were subsequently referred to TB clinics for necessary treatment. Patients with DM were screened in 2 studies and 1 study explored the presence of TB in patients with MHIs.
In a study by Zhang and colleagues in China, a specific timeline was allocated for elderly patients with DM from different counties to participate in a free TB screening programme embedded within a health examination exercise conducted at the village level primary care facilities. A total of 93,094 elderly residents from 3 counties participated in the health examinations. The large number of participants enrolling into the programme at the same time overwhelmed the hospital and providers, lowering overall screening quality for TB. This led to a higher possibility of missing out on potential participants requiring the TB treatment service. Many patients also dropped out of the programme and were lost to follow-up due to a weak referral mechanism [52].
A cross-sectional study by Qader and colleagues in Afghanistan noted a 20% higher prevalence of TB among patients with MHIs as compared to the general population. The trained nurses screened 8,073 patients with MHIs for TB during their regular follow-up visits. The patients with presumptive TB were then provided with a sputum test, and all patients with a positive diagnosis of TB were enrolled in nearby direct observation therapy (DOT) centre for further examination and treatment [53].
As this level 2 integration model aims to establish confirmatory diagnosis of TB suspects through further assessments at TB clinics, a more definitive diagnostic tool was not required at the point of entry at NCD clinics. The providers at the NCD clinics could effectively run this integration model after receiving basic training to conduct the TB screening services. This integrated screening and referral model reiterates the impetus for integrated services in settings facing a high prevalence of TB among patients with DM and MHIs [51][52][53]. The ability to provide opportunistic screening when patients present themselves for NCDs is deemed a key gateway into TB-related interventions, which can help mitigate the spread of TB in LMICs [51,52].
Cross-testing and referral for both TB and NCDs treatment services. Three studies describe this model of care integration whereby screening was performed for both TB and NCDs within the same health facility and patients with positive screening results were referred to other care facilities for treatment. Two studies included screening services for DM and 1 study included screening services for multiple NCDs such as DM and hypertension in addition to TB screening services. Two of the studies also screened for HIV, which is of relatively high prevalence in LMICs where the studies were conducted. HIV, however, is not included in our review.
The study by Jerene and colleagues in Ethiopia focused on the lack of the health system's ability to detect prospective DM cases. This cross-sectional study conducted a tridirectional screening to detect previously missed or undiagnosed TB and DM among 3,439 participants. The study stated that 32.4% of TB patients had abnormal blood sugar suggestive of DM, but the existing health system had detected only 3.5% of these cases, reemphasising the urgency for such integrated screening processes into existing TB or NCD programmes. The same study also highlighted the weakness in its model for TB diagnosis as it was mainly done through symptom screening followed by sputum microscopy. Since most DM patients with TB are asymptomatic, Jerene and colleagues acknowledged the potential underestimation of TB diagnosis among patients with DM [54].
The study by Govindasamy and colleagues in South Africa screened 9,806 participants for TB and HIV and NCDs such as DM and hypertension using a mobile testing unit. The nurses in the mobile testing unit provided the participants that screened positive for TB and/or NCDs with a referral letter to link to healthcare facilities for treatment and follow-up management [55]. Similar to the challenges facing other models, this study also observed a significant loss of continuity of care due to patients losing referral letters or failing to attend follow-up sessions at referred clinics on weekdays due to work commitments. More flexible clinic hours and patient advocacy to improve understanding of the importance of the referral letter and follow-up care were some recommendations provided by the authors [55].
This model of screening and referral for both communicable and NCDs was considered an effective platform for diagnosing TB and NCDs concomitantly, especially in resource-limited settings. The need for such a platform is further emphasised in the cross-sectional study by Araia and colleagues in Eritrea, which points out the shared risk factors such as BMI for the development of TB and DM whereby only 54.5% of the patients in the study were aware that they have both conditions [56]. Furthermore, free screening using mobile units was deemed effective to reach participants in rural areas where screening for diabetes was performed using handheld glucometers and screening for hypertension performed with electronic sphygmomanometers [57]. Provision of trained staff for the integrated screening and referral processes and coordination between the involved healthcare facilities was recognised as a facilitator for the success of this model [57,58].

Level 3 integration
Entry via TB service, colocation of testing and treatment for NCDs. Six studies on this model of integration were found, out of which 5 studies explored the integration of care between TB and various MHIs and 1 study focused on pharmacological care of TB and DM. In this model, the participants entered the system via the TB programme and received care for both TB and NCDs at the same location.
A mixed methods study by Gnanasan and colleagues in Malaysia identified 53 patients with TB and DM comorbidity from their system data and 35 of them that agreed to take part in the study were then provided with a pharmacist-led pharmaceutical care service within the same tertiary hospital to manage their medical conditions. The integrated programme led by pharmacists offered DOT and medication therapy adherence clinical services which were reported to be useful for patients with multiple diseases, namely TB and DM, in managing their medication regiment. This study illustrated that pharmacists could also conduct the integrative services as a form of task sharing. Like other levels of integration, a lack of information flow and coordination was reported. In this study, the disjointed data availability between DM clinics and TB clinics due to the practice of single disease management had made coordination of services across diseases an impediment to integration. In addition, the dilemma of whether to recommend certain monitoring tests or to wait and see if a particular test was ordered by a physician displayed the challenges due to the power dynamics between pharmacists and physicians [59].
Completion of TB treatment can be promoted in programmes with this level of integration as reported in the study by Pasha and colleagues in Pakistan, whereby the integration of an MHI intervention by assimilating integrated practice units (IPUs) into existing TB treatment facilities showed not only improved symptoms of depression and anxiety, but also a significant increase in proportion of TB patients completing their TB treatment with each subsequent MHI intervention session [60]. Additionally, services that include how to manage social stigma and carry positive dialogue regarding mental health conditions with family and friends proved effective in alleviating the effects of depression in patients with TB. The IPU used a task-shifting approach to service delivery, utilising lay health counsellors for the provision of mental health services. The IPUs were designed to embed mental health screening and counselling services into the existing treatment flow at each facility, so as to not affect normal TB related operations in the facilities [60].
Resources such as trained health professionals and medical supplies are essential for providing the added care in this integration model. For example, a nurse and 2 medical students were trained for the integrated programme as reported in a study by De Castro-Silva and colleagues in Brazil [61], and a designated team of pharmacists and physicians worked together to design an individualised pharmaceutical care plan for TB and DM in the study conducted in Malaysia by Gnanasan and colleagues [59]. De Castro-Silva and colleagues had further reported that certain tools such as the Patient Health Questionnaire-9 (PHQ-9), which was originally a selfadministered questionnaire had required providers to administer due to the population's low literacy rate, thus showing the need to train providers to offer tailored services based on requirements of the target population.
A mixed methods study by Walker and colleagues in Nepal had involved the services of an NGO. This study screened 135 patients receiving MDR-TB treatment for MHIs. This integration model enabled Nepal's National Tuberculosis Programme (NTP) to identify and develop contextually relevant psychosocial interventions that could be delivered within routine NTP services. The services administered were in a stepped-care format. The first step was to provide information and educational materials pertaining to MDR-TB and its treatment options to patients and their family followed by a screening process for NCDs during follow-up monthly visits. Subsequently, brief counselling sessions for overall health based on behavioural activation psychological therapy, group support for selected patients to reduce their negative emotions, and provision of social support for eligible patients were also provide in this integration model. An NGO that delivered mental health services also trained providers to provide group counselling services to patients. However, the authors reported that group counselling was not suitable for all patients as some faced difficulties expressing themselves in a group setting, while others became more worried after hearing the problems faced by fellow patients. Language barriers during counselling sessions also surfaced. The infrastructural limitation due to the lack of an appropriate, private space to conduct the counselling at the centre (which was also the case with screening) often created distractions for the patients and counsellors during the counselling sessions. The counsellors had to improvise and used rooms that were shared with other services or an outside space that provided more privacy to conduct the counselling sessions. Despite the difficulties mentioned, patients had expressed the benefits of hearing experiences of a cured patient during the group counselling sessions [58].
The mixed methods study by Lovero and colleagues in South Africa evaluated the integration of mental healthcare into TB and maternal-child healthcare services in 4 different districts involving 40 clinics (10 per district) [57]. A stepped-care approach was employed where patients were screened for MHIs and treated at the same clinic or referred to specialised care based on severity of the condition. Additionally, after the patient's condition was stabilised, a down-referral back to the original primary care clinic was performed. In-depth interviews were conducted with 9 district-level programme managers (DPMs), 17 mental health practitioners (MPHs), and 59 nurses to understand the feasibility and barriers of the programme. This study homed in on more meso and macro-level barriers such as the absence of coordination across health programmes in district-level administration, lack of material and human resources, and low mental health awareness in both the district administration and general population. Low mental health literacy in both the district administration and general population made mental health a low priority for integration into other existing programmes. Providers also highlighted that poor coordination within and across governing bodies and limited understanding of the need for integrated care at the administrative and governance level were weaknesses of this integration model. Therefore, despite TB and NCD services being colocated or within close physical proximity, providers highlighted that poor interprogramme coordination and communication plus conflicting directives from different programmes were areas that required improvement [57].
Entry via NCDs, colocation testing and treatment for TB. There were no studies on an integrated care model with entry into the system via NCDs and colocation for treatment.

Facilitators and barriers to integration of TB and NCD services
Salient and overarching themes regarding facilitators and barriers facing integrated service delivery models were mapped to the WHO health systems framework which comprises six dimensions are summarised in Table 7 below.

Discussion
This review identified the potential applicability of integrating TB and NCD services in LMICs. Using the data extracted from the included studies, we derived key elements of the different levels of integration for TB and NCD services, followed by categorising them into 3 distinct levels, differentiated according to entry points into the health system. Our findings suggest that higher levels of integration conferred more benefits to patients in terms of managing TB and NCDs. However, we are also cognizant that numerous aspects spanning the 3 levels of integration are overlapping yet differ in certain operational areas depending on the study setting. Due to the pluralistic approaches taken by integration models that vary from how they exploit opportunistic screenings at either point of disease entry to onwards care treatment, the application of a complex systems approach to integrated services planning, delivery, and evaluation is, therefore, central to deciphering intervention implementation and impacts in realworld settings. We accomplished this by mapping the operational features of the various models of integration to WHO health systems framework, which provided a validated structure to organise our findings. As such, we elucidated the overarching facilitators and barriers that fall within the dimensions of service delivery, human resources, sustainable financing and social protection, information, and leadership and governance.
Our findings suggest that care integration is more common with entry via TB to the system, with 19 studies supporting this. Eight studies were conducted at level 1 integration, 8 studies at level 2 integration, and 6 at level 3 integration. A higher level of integration that requires longer chains of care as compared to a lower level of integration, whereby less services or services catering to one disease only is provided, comes with increasing challenges that need to be iteratively surmounted at all levels of the health system. More integrated models need to be mindful that additional integrated services do not compromise the functionality of existing TB or NCD programmes or clinical operations.
Integration aims to address fragmentation in services, enabling better coordinated and more continuous care. In short, integration is achieved by careful planning and financing, with a shared vision centred on a target patient population [62]. All studies pointed to integration as the genesis of structural links between previously separate services, encompassing Table 7. Summary of facilitators and barriers mapped into WHO health system framework.

Barriers Facilitators
Service delivery • Cultural beliefs and preconceptions about disease conditions posed huge hurdles at the personal level • Poor public awareness of TB-and NCD-related comorbidities led to poor uptake • Absence of monitoring parameters • Lack of a conducive space to conduct screening and/or counselling services due to privacy concerns of patients • Prolonged waiting times at clinics and also extended intervals before receiving referral care at onwards care facilities • Gender of the provider might be a concern for particular groups of patients especially when taking anthropometric measurements • Missed screenings due to undefined roles resulting in providers being unclear of job scope and responsibility in the cascade of integrated care

PLOS MEDICINE
Integrating TB and NCDs care in LMICs: A systematic review aspects from organisational modifications, resource planning to physical colocation. At the lowest level of integration, only screening for either TB or NCDs is provided. In most studies, low-cost and straightforward screening techniques for TB and NCDs are used by trained staff. Level 1 integration studies had illuminated that providing point-of-care screening by either disease entry point, but without receiving the needed treatment through onwards referral or in-house treatment is insufficient to mitigate the increase in prevalence of either TB or NCDs effectively. The lack of providing referral to onwards care postscreening could be attributable to reasons such as the lack of comprehensive operational planning since such integrated programmes are relatively new in LMIC settings and cost concerns whereby the funds allocated for such programmes might have limited the prospects of linked care which may be considered unethical and a wasteful expenditure of scarce resources for diagnostic tests without any impact on patient management or outcomes. Therefore, level 2 integration is more optimal in that regard but has its own challenges. Despite having referrals made to appropriate providers for both TB and NCD treatment postscreening, level 2 integration models face new difficulties that the level 1 integration models did not operationally face. This includes lost to follow-up due to loss of referral documents and patients facing the inconvenience of attending another visit, which might be constrained by work commitments and physical distance to the referred clinic [55]. Hence, full integration at level 3, whereby screening and treatment occur at the same location, has been shown in our reviewed studies to provide the best point of care for both disease groups and promotes, to an extent, care integration and continuity of care. Several studies have shown that colocation of services derives better health outcomes, follow-up care adherence, and overall higher patient satisfaction [63,64]. High levels of integration have also been shown to reduce fragmentation of care, thus minimising resource wastage and inefficiency by avoiding duplication of services and contradictory decisions by providers [65]. However, as direct outcome measures were not reported in the reviewed studies, we can only surmise that having the deepest levels of service integration confer the most benefits to both patients and the health system. TB and NCDs control programmes must touch every aspect of a nation's health system to manage the diseases comprehensively. Herein, we will discuss the facilitators and barriers with the health systems components in mind to further unpack the operational features of a high functioning integration model embedded in the health system of LMICs. Notably, 2 studies highlighted the need to factor in the social and financial elements for integrated models of care. Both factors fall beyond the remit of traditional TB control programmes. As sustainable financing and social protection is one of the dimensions of WHO health systems framework, we will use this as the entry point to delve deeper into other health systems dimensions using the facilitators and barriers summarised in Table 7 above.
First, sustainable financing and social protection is central to the management of TB and NCD integrated programmes as TB and NCDs have close links to poverty and social activation [66]. Hence, integration models will require more from health systems than the traditional medically oriented interventions that do not focus on the financial and social aspects of patients. OOP payment is shown to reduce the uptake of essential health services, where patients in LMICs might prioritise nonmedical yet essential aspects of their lives such as the purchasing of food and shelter. This will, in turn, become a missed opportunity if costs are the limiting factor to the uptake of TB and NCDs services. Therefore, health systems must explore robust financing and social protection mechanisms to promote the uptake of services at all levels of integration to ensure maximum reach and impact of integrated programmes [67]. From the providers and policymakers point of departure, the terms universal health coverage (UHC) and limited fiscal space lie in a luminal space, whereby finite resources need to be allocated to various aspects of the health system [68]. The study by Contreras and colleagues conducted in Peru showed that meeting the high demands for screening services necessitated the engagement of private laboratories where the costs were covered by the programme, while socioeconomic assistance such as transportation reimbursement and food vouchers were rendered to the patients by an NGO [49]. Therefore, having TB and NCD services provided at one facility can potentially reduce transport costs due to multiple visits. However, we are unclear of how the medical costs incurred by the patients will be divided in an integrated programme since certain medical conditions such as TB might already be covered by an existing programme but not others.
More importantly, risk factors for TB and NCDs are often multifaceted, and the most costeffective way to mitigate both epidemics might be through early detection. Thus, financial attributes of screening services must consider the socioeconomic status of the targeted population and the ability of the country to finance it sustainably. This was further emphasised in another study by Byrne and colleagues in Peru, whereby glycosuria was used as a biomarker for undiagnosed DM and inadequate glycemic control for DM and CKD patients. Despite having moderate sensitivity, the study failed to employ more specific assessment tools for renal function impairment such as serum creatinine, due to cost and implementation feasibility concerns [45].
Second, at the service delivery level, contextual nuances such as cultural beliefs and preconceptions about a disease condition pose huge hurdles at the personal level and also impede the full functionality of an integrated programme. Poor or lack of public education and awareness of TB and NCDs can create misconceptions about the diseases such as "leaving it to fate." This holds true not only for patients but also for policymakers, as seen in the disconnect from TB and NCDs agendas mentioned in the study by Lovero and colleagues [57]. A facilitator that can be employed to overcome this is to increase health literacy through community health workers [69]. The deployment of validated guidelines ensures that evidence-based process and outcome indicators are being monitored and fulfilled as cornerstones for effective patient-centred care, especially for TB and NCDs, which require prolonged management [70]. Furthermore, accessibility to care can also pose a barrier due to transportation costs or heightened opportunity costs for informal labourers when attending follow-up services. These challenges can be surmounted by deploying mobile testing units as used in the study by Govindasamy and colleagues in South Africa to meet patients where they are [55]. The lack of conducive space to screen for NCDs and provide counselling services might cause privacy concerns. As a result, patients might be less willing to take up these additional services, be it for mental health treatment or screening for other NCDs where anthropometric measurements are taken. Although not reported in the studies reviewed, colocation with TB services necessitates extra infectious disease precautions to reduce the nosocomial spread of TB when non-TB patients present themselves for NCDs [71].
Third, at the human resources level, it was apparent in nearly all reviewed studies that the increased workload posed a considerable roadblock to the operations of integrated programmes. This is especially salient in LMIC settings whereby training of health workers typically involve only a particular disease group such as maternal and child health or specific infectious diseases, while cross-training (i.e., TB and NCD services) for more than 1 disease condition is rare [72]. Additionally, the lack of manpower and training or expanded job scope places more stress on existing providers, who are already stretched by their normal clinical operations [73]. The overextending of these providers might lead to screening misses and treatment errors. As such, providers must not only be well trained but also motivated to provide the services. Attitude, professionalism, and commitment create stronger patient-provider relationships, which confer better patient outcomes and, hence, essential elements to any integration model [74].
Fourth, the availability of medical products and vaccines is essential to the implementation of integration programmes. However, aside from Bacillus Calmette-Guérin (BCG) vaccine, whose effect wanes after childhood, there are effectively no viable vaccines for TB. NCDs are also primarily developed due to genetic or lifestyle predisposition, which can only be prevented through behavioural modifications and medications [75]. Thus, the availability of efficacious medicines for TB, administered mainly through DOT, is an integral element to any TB programme. A break in the medication regime or shortage of medicines at all levels of integration can lead to the generation of MDR-TB [76]. Likewise, the provision of high-quality medication for NCDs is required to ensure optimal management of NCDs, which are often prolonged and require extended medication regimes and monitoring. Additionally, there is also a need for medical resources that aid in screening of patients for TB and NCDs. For example, the availability of handheld glucometers used for screening DM and electronic sphygmomanometers for hypertension as explicated by Govindasamy and colleagues provided a practical way to screen patients for NCDs [55]. Without the provision of medical equipment for screening, linkage to onwards care and treatment will be rendered ineffective in an integrated service delivery model. Although not discussed in the articles reviewed, having integration models that are equipped with adequate medical resources for both TB and NCDs is essential to managing both disease conditions concurrently. This is because the inability to control one might lead to the exacerbation of the other. For example, uncontrolled DM is associated with worsening of TB control as these 2 conditions are closely associated biologically [19].
Fifth, the information provided within and across healthcare entities is an essential tenet for the smooth running of any health programme, let alone one that integrates services catering to multiple disease conditions across different providers. Many of the studies reviewed highlighted the lack of coordination across providers, especially across different healthcare institutions, which hindered integration on many fronts. There is often a lack of common electronic medical record system or standardised medical information storage platform that tracks process and clinical outcome parameters of the patients and data access to providers within and across echelons of care was also reported to be limited. For example, Gnanasan and colleagues had expressed that certain clinical parameters such as blood pressure were not checked at the TB clinic but could have been extracted from DM or cardiovascular clinic's records instead in order to give providers a more comprehensive medical history of the patient. However, the lack of coordination across provider settings stymied the exchange of patient data [59]. Furthermore, the information tracked must be evidence informed and relevant for clinical management of the patients' condition. To this end, validated guidance principles and reporting measures/frameworks need to be deployed by providers. However, many reviewed studies had reported the lack of a former reporting system. Many had to turn to operationalised indicators to track the patient pathway, which might not be optimal. An example is using a modified WHO STEPwise Approach to NCD Risk Factor Surveillance (STEPS) framework that Anand and colleagues reported [50]. Although only pilot tested once in the study's context, the operationalised questionnaire that screens for NCD predisposition was deemed quick and easy by both patients and providers. Therefore, despite the absence of thorough validation, rapid screening procedures can be deployed in resource constraint settings to streamline protocols to refer patients for onwards care.
Last, leadership and governance are central to implementing any integrated programme vis-à-vis other core functions that ensure the programme's sustainability and credibility within a health system. Governance also cuts across all other aforementioned health system functions, making transparent, accountable and committed stewardship paramount in TB and NCD control. Good governance will also require taking the lead in upholding newly minted partnerships that include formal and informal productive relationships that span within and beyond the health sector and between the health system, its intermediaries and end users (patient and providers) through the accountable provision of evidence-driven guidelines and adequate and responsible incentivisation. An example of active collaboration is displayed in the study by Contreras and colleagues whereby the TB Cero programme was implemented as an integrative partnership model between public hospitals and an NGO in Peru [49]. At the national level, having a platform to bring together intersectoral actions driven by strong political commitment coupled with a definition of integration that captures an explicit agenda and a clear roadmap for implementation undergirds organisational change towards more assimilated TB and NCD service models. Specifically, governments can also promulgate national TB and NCD plans/guidelines/frameworks and earmark the needed funds to run plans into fruition. This was alluded to in the study by Ekeke and colleagues in Nigeria, whereby patients with DM were screened for TB, in line with the national TB guidelines [51]. Thus, there is a need to update these guidelines to include NCDs to prevent any missed opportunities in identifying NCD cases. Lönnroth and colleagues had also proposed a TB elimination framework to draw key takeaways from prospective TB and NCD integration models. This encompasses universal access to high-quality TB and NCD service with a lens that places groups at the highest risk first and reducing underlying vulnerabilities by incorporating health lenses in all policies, bolstering long-term political will for locally spearheaded integrated programmes, and garnering cross-sectorial and multilevel buy-in for the implementation of global TB and NCD action plans [77]. Finally, governance bodies must be mindful of the interconnectedness of interventions along the "cascade of care" from detection of TB or NCDs to successful completion of treatment or long-term management. This implies that governance bodies must be able to evaluate programmes using both process and clinical outcome measurements in a joined-up rather than a standalone manner to truly ascertain the effectiveness of the integration of TB and NCD services within one programme and at a common locale. As the effectiveness of the integrated programmes go beyond a single entity and across different sectors outside the health system, the need to incorporate systems thinking concepts as well as adopting relevant disciplines such as social sciences and health policy analysis will serve as cornerstones for a joined-up systems approach to evaluation [78].
This systematic review foregrounds the gaps in both policy and research in the areas of TB and NCD service integration. Policymakers and researchers have focused mainly on integrating TB and HIV services, as seen from the various existing and well-evaluated programmes and systematic reviews performed [33,37,[79][80][81][82].
The research gaps can be plugged with more monitoring and evaluation works that track the process and clinical outcomes through validated indicators using a prospective or retrospective cohort study, case-control, or randomised control study approaches. Furthermore, the experiences and expectations of patients and providers can be explored through qualitative studies to confer appropriate models of care integration that meet the needs of all stakeholders. The urgency is now more than ever in the face of a swelling TB and NCD burden, and prospective research findings must be able to empower policymakers to drive integration models using implementation science and sustained political will.
Policymakers and programme implementers must recognise that the inability to attain sustainable and effective integration might be less attributed to ineffective programmes per se but rather due to the impediments underlying the health systems building blocks. Hence, integration necessitates stakeholders to envision the channelising of resources and agendas beyond individual programmes and to an extent sectors, while concomitantly investing in all health systems building blocks.
Although there is no magic bullet to achieve a fully integrated configuration for all settings, some actionable steps surfaced. We have summarised the key facilitators to be accounted for and barriers to be avoided during the implementation of integration models in Table 7 above. The high start-up costs to initiate such programmes require policymakers to place long-term returns at the forefront when deciding to allot resources into these models as return on investment in terms of better population health outcomes can only be seen in the longer term. Spillover effects such as better infrastructure and trained manpower can also enhance the health sector and improve access to care for the general population. Uptake must also be sustainable through carefully tailored integration models that account for the needs and preferences of the local population. A salient point that did not appear from our review but is crucial for the uptake of integrated services is the reduction of stigma when seeking care for certain services associated with TB and MHIs, which might affect the extent to which such models of integration are palatable.
This systematic review was limited by the dearth of research exploring the experiences of patients and providers, and specific process indicators that evaluate the feasibility of integrated interventions, as most studies focused on reporting outcomes at a particular time period. Furthermore, the main bibliographic databases used in this review may have had a limited coverage of material from LMICs regarding TB and NCD integrated programmes, and we acknowledge that the databases might have lacked coverage of material not published English [83][84][85]. In addition, the heterogeneity of the study designs, contexts of study, participants, sample sizes, outcomes, interventions, and diseases studied can potentially limit the generalisability of the study findings. Importantly, this review also revealed that most integration models were contextually bound, with each LMIC having their very own set of healthcare and political agendas writ large. Therefore, our findings have extracted the overarching themes that are cross-cutting and salient to LMICs, particularly in Asia, Africa, and the Americas.
One strength of this review is the usage of a widely deployed health system framework to guide the data analysis and reorganisation of themes for the facilitators and barriers, reducing the variability across research team members. However, we must be cognizant of the oversimplification of intricate process pathways and organisational culture. In practice, many of the key issues span across multiple building blocks, as to be anticipated when employing a health systems lens. However, we managed to utilise this framework for mapping integrated service options to each health system building block to derive high-quality and pragmatic design options for integration elaborated above.
Another strength is the development of a new framework that categorises models of integration according to their depth of integration. This will enable both policymakers and researchers to understand where a country's TB and NCD programmes lie on the spectrum of integration and, with that knowledge, strive towards the deepest levels of integration by avoiding the barriers and incorporating the facilitators into their programmes.

Conclusions
The burgeoning number of TB and NCD cases has made a strong case for tackling this combined burden of TB and NCD prevalence with a comprehensive and integrated approach commensurate to its disease burden. Limitations notwithstanding, this systematic review provides valuable insights into integrating TB-and NCD-related services into one common programme in LMICs. This has the potential to improve health service delivery across disease conditions and levels of care that can lead to population health benefits.