Navigating oxygen management challenges amidst COVID-19 pandemic and beyond in India: a modified Total Interpretive Structural Modeling (m-TISM) approach

Abstract

Background

Given the unprecedented surge in COVID-19 infections, the heightened demand for medical oxygen prompted numerous national and global initiatives to bridge the gap between supply and demand. This was crucial for ensuring adequate treatment for patients suffering from acute respiratory distress syndrome and requiring oxygen therapy. This research aims to explore the factors influencing medical oxygen management in India during the pandemic and beyond, examining both facilitators and barriers.

Method

Through a thorough review of literature, secondary research, and interviews with key stakeholders, critical factors affecting oxygen management were identified. These factors were then analyzed using a modified total interpretive structural modeling (m-TISM) approach and MICMAC (Matrice d’ Impacts croises multiplication applique an classment) analysis to comprehend their hierarchical relationships and driving forces.

Results

The study identifies fourteen key factors that act as facilitators and barriers to oxygen management during the COVID-19 pandemic. These factors also influence medical oxygen management during the non-pandemic period. The development of an m-TISM model gives us the hierarchical interrelationships between these factors, including oxygen management as one of the factors itself. The findings identify key strategic levers to strengthen the national oxygen management ecosystem and cross-sectoral collaborations.

Conclusion

Unlike previous studies, this study moves beyond just identifying the challenges to offering a hierarchical map of pandemic-era and post-pandemic facilitators and barriers in the oxygen management ecosystem in India. This study provides insights into strengthening the oxygen ecosystem, enabling policymakers and program implementers to make informed decisions and implement pre-emptive measures to address future threats from the virus or similar crises.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12913-025-13655-z.

Introduction

The COVID-19 pandemic has remained a pressing global crisis, presenting one of the most formidable health challenges, straining public health systems worldwide, resulting in millions of deaths [1–3]. Among the critical vulnerabilities in the global healthcare infrastructure exposed by the pandemic, the acute shortage and mismanagement of medical oxygen, an essential therapeutic for patients with respiratory distress, emerged as a key life threatening issue. To prevent morbidity and mortality, governments and institutions worldwide have tirelessly collaborated to contain transmission, conduct widespread testing and treatment, support overburdened health systems, optimize oxygen supply chains, and ensure vaccine accessibility for their populations [4–7].

This exceptional pandemic underlines the importance of medical oxygen as a critical care component in treating those infected with the virus [2, 8, 9]. Since liquid medical oxygen, the primary source of medical oxygen, is available in limited amounts and not produced at the healthcare facility, meeting the steep increase in demand during the pandemic required a multi-pronged approach involving various stakeholders in the oxygen supply chain [10]. In India, despite multiple national and international efforts to scale up oxygen production and distribution, ranging from the establishment of Pressure Swing Adsorption (PSA) plants to the deployment of digital dashboards like ODTS and OCMIS, several systemic challenges persisted. Managing this heterogeneous medical oxygen crisis required a global, national, and regional strategy to ensure sufficient availability and accessibility of oxygen in different contexts and situations [11–14]. Addressing these challenges requires not just a technical or logistical fix, but a systemic understanding of how different enablers and barriers interact within the medical oxygen ecosystem.

This study aims to identify the factors that facilitate or act as barriers to medical oxygen management in India during the COVID-19 pandemic and non-pandemic period, and understand the hierarchical interrelations between them using the Modified Total Interpretive Structural Modeling (m-TISM) approach [15, 16]. While existing studies have examined discrete aspects of the oxygen crisis, such as supply chain failures, infrastructure gaps, or clinical protocols, none have adopted an integrated perspective [17–19]. This paper addresses that gap by combining qualitative stakeholder insights with structural modeling to understand the hierarchical relationships and influence dynamics among key factors shaping the oxygen ecosystem.

This paper offers actionable insights for prioritizing policy interventions and serves as a replicable analytical framework for other countries. The study’s core contribution lies in constructing a hierarchical model encompassing fourteen critical factors that influence oxygen management during and beyond the COVID-19 pandemic. Thematic analysis of these interviews further contextualizes the m-TISM model, enriching the interpretation of systemic interdependencies and practical implications. By establishing a logical hierarchical framework, interventions in oxygen management can be guided effectively, thereby assisting overwhelmed healthcare facilities in mounting a comprehensive response and evidence-informed decision-making for future pandemic preparedness and health system strengthening.

Research methodology

Study design

The study employs a mixed-method approach, integrating insights from secondary research and expert interviews of key stakeholders. After conducting an extensive literature review and qualitative data collection, this study has pinpointed fourteen crucial factors (described in Fig. 1; detailed in Appendix 1) influencing oxygen management. These factors help in developing a structured interpretive model for analysing interdependencies among various factors.

Fig. 1.

Barriers and facilitators of oxygen management

The modified total interpretive structural modeling (m-TISM) is used, which is an extension of ISM and TISM methodology that has been extensively applied in literature across multiple disciplines [15, 20, 21]. This methodology involves advanced qualitative modelling, widely accepted as a decision-making approach that helps to understand unstructured mental models of complex issues. These models go through multiple iterative steps by identifying and interpreting their hierarchy to understand the system better [22–25]. Furthermore, this approach has been applied in the healthcare industry to explore topics such as manufacturing processes during pandemics and emerging infectious diseases, as well as agile performance within healthcare organizations [26–28]. This study presents a novel application of the m-TISM approach to analyze oxygen management in India, offering both interpretive reasoning and structural visualization [15, 16].

The m-TISM methodology was chosen as it allows modeling of complex, unstructured systems by integrating expert judgment with hierarchical visualization. In the context of medical oxygen management, where interdependencies among policy, logistics, infrastructure, and workforce are high and data availability is limited, m-TISM allows for both systematic structuring and interpretive reasoning [15, 16]. Unlike traditional modeling, m-TISM accommodates diverse stakeholder perspectives and surfaces tacit knowledge that is often not captured in formal datasets. This makes it particularly relevant for pandemic-related health systems analysis. By leveraging m-TISM, using data from exhaustive secondary research and twelve in-depth interviews, we identify foundational drivers as well as dependent variables.

This research will help uncover not only the ‘what’ and ‘how’, but also the ‘why’ behind the interrelationships between barriers and facilitators of oxygen management. By incorporating expert insights in situations where quantitative data is limited, it supports more informed and effective decision-making for policymakers and healthcare administrators. The m-TISM methodology will assist in not just a logical interpretation of successive paired comparisons but also transitive links in the hierarchy [15, 20, 25, 29]. It will be followed by MICMAC analysis to understand the driving and dependence powers of the factors. To enhance the interpretive outputs of m-TISM, MICMAC analysis classifies the factors into driving, dependent, and linkage variables based on their influence and dependence scores. This allows policymakers to visualize not just structural positioning but also strategic leverage points for intervention. The steps of the m-TISM methodology are provided in Fig. 2 [15]. Using thematic analysis, transcripts from the stakeholder interviews were reviewed iteratively by the research team, resulting in the identification of ten themes.

Fig. 2.

Steps in m-TISM methodology [15]

To ensure robustness, the modeling process involved many iterative rounds of discussion among the research team, peer debriefing with subject matter experts, and cross-verification of hierarchical relationships against real-world stakeholder accounts. All pairwise comparisons were justified using interpretive logic tables (Appendix 2), improving traceability and transparency.

Data collection

Participant selection

The selection of participants for this study was guided by a purposive sampling approach designed to ensure both representativeness and relevance to the main objectives of this study [30, 31]. A list of stakeholders is provided in Table 1. Considering the heterogeneous nature of India’s healthcare system and the governance of health service delivery, it was critical to capture insights across diverse geographic, epidemiological, and administrative contexts. This study applied a five-step selection logic that balances conceptual breadth with practical feasibility, thereby maximising the interpretive power of the modified-TISM framework.

Table 1.

List of stakeholders

Sr. N.	Stakeholders	Department/Institutions that will be referred to for various officials (stakeholders) interviewed	No. of interviewees*
1	Government	Ministries/Health Departments/State-District Municipals for regulations, policy decisions, and implementation	4
2	International Organizations	Technical/infrastructure support from organizations like WHO, UNDP, UNICEF, USAID, etc.	5
3	Website/app Developers	Development, operation, and maintenance of apps/websites/dashboards/databases	1
4	State or District technical support teams	State or District officials and technical departments, e.g., NIC	3
5	Academic and technical institutions	Technical and research support	3
6	Doctors/Healthcare providers	Demand assessment by emergency medicine, hospital medicine, critical care, respiratory therapy	2
7	Distributors	Cylinder Suppliers; Transportation departments or companies, etc.	2
8	Private Sector	Private hospitals, NGOs, other countries, start-ups	2
9	Human Resource	Maintenance staff at the health facility and manufacturing/transport facility	2
10	Special Task Forces	For monitoring, communication management, grievances handling, emergency response organizations, workforce training	1

*Note - Some of the interviewees had multiple roles and responsibilities related to the oxygen management during COVID-19; therefore, they were interviewed for more than one role

Health system diversity and capacity

India exhibits vast differences in health system maturity, public health spending, and healthcare infrastructure among different states. In this study, stakeholders from states with relatively robust infrastructure were included alongside those with more constrained capacities. This heterogeneity enabled the research to examine how contextual factors shaped oxygen access, storage, and distribution responses, and what systemic bottlenecks or innovations emerged under varying resource constraints.

Variation in COVID-19 impact and oxygen demand

Stakeholders were purposively chosen to reflect differing COVID-19 trajectories and oxygen demand levels. While some experienced early and intense surges, others witnessed delayed or moderate peaks. This variation allowed the study to capture diverse pressures on local oxygen infrastructure and response systems.

Implementation of distinct oxygen interventions

Some stakeholders pioneered specific policies or technological innovations in response to oxygen shortages. Stakeholders from states that had implemented such distinct interventions were purposively included to enable learning from innovation diffusion and implementation experiences. This included both nationally guided and state-specific programs, helping to assess policy alignment and contextual adaptation.

Accessibility and responsiveness of stakeholders

Stakeholder availability during the data collection period (amid post-pandemic recovery efforts) was a practical consideration. They were selected where key informants across sectors (government, private, technical) were both accessible and willing to participate, ensuring depth and diversity of perspectives.

Expert recommendations and institutional networks

The research team consulted experts and academic collaborators to identify stakeholders from states with operational relevance and feasible stakeholder engagement. Established partnerships also facilitated trusted access to high-quality data.

Interview process

After completing secondary research and expert opinion, semi-structured interviews were conducted with the key stakeholders to gain insights into barriers and facilitators of oxygen management in India and the relations of all the identified variables. The interview discussion was initiated by defining their roles and responsibilities, leading to the factors and challenges that were involved in the smooth functioning of the oxygen management Towards the concluding part, suggestions and recommendations were sought from the respondents that may help strengthen the oxygen ecosystem in India in case any such arise or in general. Each interview lasted for around 45 min to one hour. The data from these interactions were analysed using thematic analysis. Transcripts from the stakeholder interviews were reviewed iteratively by the research team, resulting in the identification of ten themes.

Ethical considerations

This study received ethics approval from the Western Copernicus Group Institutional Review Board. All participants provided informed consent before participating in the study. The study ensured the confidentiality and anonymity of participants, following standard ethical research guidelines.

Modelling

m-TISM

The following steps were followed in the m-TISM methodology:

Step 1

Identification and definition of the factors

This step involves identifying the factors that act as facilitators or barriers to oxygen management in India and defining them. These 14 factors were triangulated across three sources: an extensive review of peer-reviewed literature, national and international health policy documents, and expert validation through stakeholder interviews. This multi-source identification ensured both conceptual relevance and operational resonance. A list of all 14 variables is provided in Table 2.

Table 2.

List of factors and codes

S.No.	Factor	Code
1	Oxygen Purity	F1
2	Medical Equipment and Maintenance	F2
3	Production Capacity	F3
4	Regulatory Framework	F4
5	Oxygen Stewardship	F5
6	Network of Distribution Systems	F6
7	Procurement and Acquisition System	F7
8	Workforce Capacity and Training	F8
9	International Collaboration	F9
10	Public-Private Collaboration	F10
11	Monitoring and Audits	F11
12	Preparedness and Planning	F12
13	Digital Technology	F13
14	Oxygen Management	F14

While Digital Technology, Workforce Capacity and Training and Oxygen Stewardship are interrelated in implementation, they represent distinct conceptual constructs within the model:

• Oxygen Stewardship (F5) focuses on the protocols, guidelines, and monitoring systems that ensure rational, safe, and efficient use of oxygen, including minimizing wastage through audits, SOPs and behavioral practices

• Workforce Capacity and Training (F8) includes human resource availability, skill development and e-learning modules required for the effective handling, maintenance, and administration of oxygen-related systems and equipment

• Digital Technology (F13) refers to the digital infrastructure, tools and platforms (e.g., ODAS, ODTS, IoT-enabled devices, dashboards) used for tracking, forecasting, and optimizing oxygen supply and demand

Given the potential conceptual proximity of certain factors, care was taken to distinguish them based on their unique operational focus. For example, while Digital Technology (F13), Workforce Capacity and Training (F8) and Oxygen Stewardship (F5) may interact during implementation, they are conceptually distinct. Digital Technology encompasses infrastructure and tools used for data collection and system optimization. Workforce Capacity focuses on training and human resource adequacy for oxygen operations. Oxygen Stewardship pertains to the monitoring and behavioral protocols that ensure safe, efficient oxygen use. These distinctions were validated through both thematic analysis and expert feedback and are critical to preserving the interpretive clarity of the m-TISM model.

Step 2

Relationship between the factors identified

This step involves establishing the relationships between the barriers and facilitators that were identified in Step 1. This is done following the existing literature and expert opinions. For instance, Factor F1 influences Factor F2 and vice versa.

Step 3

Interpretation of the relationship

This step involves interpreting the relationship that was defined in Step 2. For instance, it is established how Factor F1 influences or enhances Factor 2. Here, the goal is to interpret how one barrier or facilitator of oxygen management influences the other.

Step 4

Pairwise comparison, reachability, and transitivity matrix

In this step, an interpretive logic-knowledge base (Appendix A) is made to do a pairwise comparison of the factors. This is further used to make the reachability matrix, which is constructed using binary digits 1 and 0. In this matrix, a 1 is placed in case of a direct relationship or impact from one factor to another, and a 0 is placed in case of no relationship. Simultaneously, the factors are checked for transitivity. If Factor A Influences Factor B, then Factor B influences Factor C, then Factor A and Factor C have a transitive relationship. In the reachability matrix, 1* marks the transitive relationships between the factors identified.

Step 5

Level partitioning

Level partitioning is done using the reachability set (factors that are influenced by Factor X), antecedent set (factors that affect or influence Factor X), and intersection set (present in both reachability and antecedent sets). Level partitions are created by identifying factors that have the same reachability and intersection set. This iterative method is repeated till all the m-TISM levels are identified.

Step 6

Diagraph

A Diagraph is made to depict the relationships between the factors with both the direct and indirect linkages between them. The m-TISM digraph is made using the level partitions identified in the above steps.

Step 7

The m-TISM model

The final step or outcome of the above steps is the modified total interpretive structural model. Figure 3 depicts the final m-TISM model along with the interpretation of the relationships between them.

Fig. 3.

Digraph showing the hierarchical relationship among factors

MICMAC

The subsequent analysis after identifying the m-TISM hierarchy is the MICMAC (Matrice d’ Impacts croises multiplication applique an classment) analysis. It is conducted using the dependence and driving powers of the factors using the reachability matrix with transitive links [22]. In this analysis, the driving power is calculated as the row total, that is, the number of factors influenced by a factor. The dependence power is calculated as the column total, that is, the number of factors that influence a particular factor. Further, MICMAC analysis categorizes factors into four quadrants of the two-dimensional Cartesian coordinate system, with driving forces on the Y axis and dependence forces on the X axis. The results of this analysis will assist in making the outcomes of the m-TISM method more meaningful by visually quantifying their driving and dependence powers.

Results

m-TISM analysis

The study identifies fourteen key factors that act as facilitators and barriers to oxygen management during the COVID-19 pandemic. The m-TISM analysis gives us the hierarchical relationship between these factors. Table 3 depicts the reachability matrix. 1* marks the transitive relationships between the factors identified in the reachability matrix. In the level partitioning, the iterative method is repeated till all the m-TISM levels are identified. In Table 4, for the first iteration, it can be noted that for Oxygen Management (F14), the reachability set and the intersection set are the same, placing F14 at Level I. At the next level, it is the same for factors F1, F2, and F5, which are at Level II of the hierarchy. These factors are subsequently removed in the following iteration table, and the process continues until all the levels are identified. The results of the hierarchical level partitioning (Table 4) enabled the identification of different levels in the m-TISM hierarchy.

Table 3.

Reachability matrix

Factor codes	F1	F2	F3	F4	F5	F6	F7	F8	F9	F10	F11	F12	F13	F14
F1	1	1	0	0	1	0	0	0	0	0	0	0	0	1
F2	1	1	0	0	1*	0	0	0	0	0	0	0	0	1*
F3	1*	1	1	0	1*	1	1	1*	1*	1*	0	1	1	1*
F4	1*	1*	1	1	1*	1*	1*	1*	1*	1*	1	1*	1*	1*
F5	1	1*	0	0	1	0	0	0	0	0	0	0	0	1
F6	1*	1	0	0	1*	1	1	1	1	1	0	0	0	1*
F7	1*	1*	0	0	1*	1	1	1*	1*	1*	0	0	0	1*
F8	1*	1*	0	0	1*	1	1*	1	1*	1*	0	0	0	1*
F9	1	1*	0	0	1*	0	0	0	1	1	0	0	0	1*
F10	1*	1*	0	0	1*	0	0	0	1	1	0	0	0	1*
F11	1*	1*	1*	1	1*	1*	1*	1*	1*	1*	1	1	1	1*
F12	1*	1*	1	0	1*	1*	1*	1*	1*	1*	0	1	1	1*
F13	1*	1*	1*	0	1*	1*	1*	1*	1*	1*	0	1	1	1*
F14	0	0	0	0	0	0	0	0	0	0	0	0	0	1

Table 4.

Hierarchical partitioning

Factor codes	Reachability set	Antecedent set	Intersection set	Level
Iteration 1
F1	1,2,5,14	1,2,3,4,5,6,7,8,9,10,11,12,13	1,2,5
F2	1,2,5,14	1,2,3,4,5,6,7,8,9,10,11,12,13	1,2,5
F3	1,2,3,5,6,7,8,9,10,12,13,14	3,4,11,12,13	3,12,13
F4	1,2,3,4,5,6,7,8,9,10,11,12,13,14	4,11	4,11
F5	1,2,5,14	1,2,3,4,5,6,7,8,9,10,11,12,13	1,2,5
F6	1,2,5,6,7,8,9,10,14	3,4,6,7,8,11,12,13	6,7,8
F7	1,2,5,6,7,8,9,10,14	3,4,6,7,8,11,12,13	6,7,8
F8	1,2,5,6,7,8,9,10,14	3,4,6,7,8,11,12,13	6,7,8
F9	1,2,5,9,10,14	3,4,6,7,8,9,10,11,12,13	9,10
F10	1,2,5,9,10,14	3,4,6,7,8,9,10,11,12,13	9,10
F11	1,2,3,4,5,6,7,8,9,10,11,12,13,14	4,11	4,11
F12	1,2,3,5,6,7,8,9,10,12,13,14	3,4,11,12,13	3,12,13
F13	1,2,3,5,6,7,8,9,10,12,13,14	3,4,11,12,13	3,12,13
F14	14	1,2,3,4,5,6,7,8,9,10,11,12,13,14	14	I
Iteration 2
F1	1,2,5	1,2,3,4,5,6,7,8,9,10,11,12,13	1,2,5	II
F2	1,2,5	1,2,3,4,5,6,7,8,9,10,11,12,13	1,2,5	II
F3	1,2,3,5,6,7,8,9,10,12,13	3,4,11,12,13	3,12,13
F4	1,2,3,4,5,6,7,8,9,10,11,12,13	4,11	4,11
F5	1,2,5	1,2,3,4,5,6,7,8,9,10,11,12,13	1,2,5	II
F6	1,2,5,6,7,8,9,10	3,4,6,7,8,11,12,13	6,7,8
F7	1,2,5,6,7,8,9,10	3,4,6,7,8,11,12,13	6,7,8
F8	1,2,5,6,7,8,9,10	3,4,6,7,8,11,12,13	6,7,8
F9	1,2,5,9,10	3,4,6,7,8,9,10,11,12,13	9,10
F10	1,2,5,9,10	3,4,6,7,8,9,10,11,12,13	9,10
F11	1,2,3,4,5,6,7,8,9,10,11,12,13	4,11	4,11
F12	1,2,3,5,6,7,8,9,10,12,13	3,4,11,12,13	3,12,13
F13	1,2,3,5,6,7,8,9,10,12,13	3,4,11,12,13	3,12,13
Iteration 3
F3	3,6,7,8,9,10,12,13	3,4,11,12,13	3,12,13
F4	3,4,6,7,8,9,10,11,12,13	4,11	4,11
F6	6,7,8,9,10	3,4,6,7,8,11,12,13	6,7,8
F7	6,7,8,9,10	3,4,6,7,8,11,12,13	6,7,8
F8	6,7,8,9,10	3,4,6,7,8,11,12,13	6,7,8
F9	9,10	3,4,6,7,8,9,10,11,12,13	9,10	III
F10	9,10	3,4,6,7,8,9,10,11,12,13	9,10	III
F11	3,4,6,7,8,9,10,11,12,13	4,11	4,11
F12	3,6,7,8,9,10,12,13	3,4,11,12,13	3,12,13
F13	3,6,7,8,9,10,12,13	3,4,11,12,13	3,12,13
Iteration 4
F3	3,6,7,8,12,13	3,4,11,12,13	3,12,13
F4	3,4,6,7,8,11,12,13	4,11	4,11
F6	6,7,8	3,4,6,7,8,11,12,13	6,7,8	IV
F7	6,7,8	3,4,6,7,8,11,12,13	6,7,8	IV
F8	6,7,8	3,4,6,7,8,11,12,13	6,7,8	IV
F11	3,4,6,7,8,11,12,13	4,11	4,11
F12	3,6,7,8,12,13	3,4,11,12,13	3,12,13
F13	3,6,7,8,12,13	3,4,11,12,13	3,12,13
Iteration 5
F3	3,12,13	3,4,11,12,13	3,12,13	V
F4	3,4,11,12,13	4,11	4,11
F11	3,4,11,12,13	4,11	4,11
F12	3,12,13	3,4,11,12,13	3,12,13	V
F13	3,12,13	3,4,11,12,13	3,12,13	V
Iteration 6
F4	4,11	4,11	4,11	VI
F11	4,11	4,11	4,11	VI

Figure 4 demonstrates the m-TISM digraph of barriers and facilitators of oxygen management during COVID-19 in India. Relevant transitive links were retained in the final digraph, and the rest were dropped. Figure 4 depicts the final m-TISM model along with the interpretation of the relationships between them.

Fig. 4.

Digraph showing direct and transitive linkages between the factors

MICMAC analysis

MICMAC analysis categorizes factors into four quadrants of the two-dimensional Cartesian coordinate system, with driving forces on the Y axis and dependence forces on the X axis. Table 5 details the calculation of the degree of driving and dependence powers. Figure 5 depicts the MICMAC analysis quadrants and the positioning of the factors affecting oxygen management.

Table 5.

Degree of driving and dependence power of factors

Factor codes	F1	F2	F3	F4	F5	F6	F7	F8	F9	F10	F11	F12	F13	F14	Driving power (Y)
F1	1	1	0	0	1	0	0	0	0	0	0	0	0	1	4
F2	1	1	0	0	1*	0	0	0	0	0	0	0	0	1*	4
F3	1*	1	1	0	1*	1	1	1*	1*	1*	0	1	1	1*	12
F4	1*	1*	1	1	1*	1*	1*	1*	1*	1*	1	1*	1*	1*	14
F5	1	1*	0	0	1	0	0	0	0	0	0	0	0	1	4
F6	1*	1	0	0	1*	1	1	1	1	1	0	0	0	1*	9
F7	1*	1*	0	0	1*	1	1	1*	1*	1*	0	0	0	1*	9
F8	1*	1*	0	0	1*	1	1*	1	1*	1*	0	0	0	1*	9
F9	1	1*	0	0	1*	0	0	0	1	1	0	0	0	1*	6
F10	1*	1*	0	0	1*	0	0	0	1	1	0	0	0	1*	6
F11	1*	1*	1*	1	1*	1*	1*	1*	1*	1*	1	1	1	1*	14
F12	1*	1*	1	0	1*	1*	1*	1*	1*	1*	0	1	1	1*	12
F13	1*	1*	1*	0	1*	1*	1*	1*	1*	1*	0	1	1	1*	12
F14	0	0	0	0	0	0	0	0	0	0	0	0	0	1	1
Dependence Power (X)	13	13	5	2	13	8	8	8	10	10	2	5	5	14

Fig. 5.

MICMAC analysis (driving and dependence relationship of factors)

Quadrant 1 (Autonomous variables)

The factors in this quadrant are weaker and have a lower driving and dependence power. They are generally disconnected from the other variables and are difficult to influence even with concentrated efforts. In this analysis, there are no autonomous variables.

Quadrant 2 (Dependent variables)

The factors in this quadrant have high dependence power and low driving power. These factors are driven by the factors that have a high driving power. In this analysis, the top three levels are dependent factors. Oxygen Management (F14) is at Level I, Oxygen Purity (F1), Medical Equipment and Maintenance (F2), Oxygen Stewardship (F5) at Level II, and International Collaboration (F9) and Public Private Collaboration (F10) at Level III are dependent variables. These are driven by factors that have a high driving power (Quadrant 4).

Quadrant 3 (Linkage variables)

The factors that lie in this quadrant have both high driving and dependence powers and stay connected with factors in Quadrant 2 and Quadrant 4. These are critical in the hierarchy since any changes in them impact all the other variables. In this analysis, the Network of Distribution Systems (F6), Procurement and Acquisition System (F7), and Workforce Capacity and Training (F8) are linkage variables.

Quadrant 4 (Independent/driving variables)

The factors in this quadrant have high driving power and play a very important role as key influencers in the model. They ultimately drive the other factors (dependent factors). The critical driving factors in this analysis are Production Capacity (F3), Regulatory Framework (F4), Monitoring and Audits (F11), Preparedness and Planning (F12), and Digital Technology (F13). These factors are placed at the lower levels of the m-TISM hierarchy.

Thematic analysis

The qualitative data from the interviews conducted were analysed using thematic analysis, following the six-phase framework, providing a structured yet flexible approach for identifying, analyzing, and interpreting patterns of meaning across qualitative interviews. This analysis was chosen for its theoretical flexibility and its suitability for unpacking the complexity of cross-sectoral stakeholder perspectives in medical oxygen and health systems research, especially since the goal is to explore both manifest content (what was said) and underlying meanings (why it matters) [32].

The thematic analysis process involved six recursive phases [32]:

1. Familiarisation with the data: Each transcript was read multiple times by members of the research team to ensure immersion and note initial observations.

2. Generating initial codes: Codes were inductively developed and applied manually across the dataset, capturing meaningful units of text relevant to the study’s aim and context.

3. Searching for themes: Codes were grouped into potential themes using a collaborative process that examined relationships between codes and how they clustered conceptually with each other.

4. Reviewing themes: Themes were refined by assessing their internal coherence and distinctiveness from each other, ensuring that each theme accurately reflected the data.

5. Defining and naming themes: Themes were clearly defined with concise labels and detailed descriptions, representing both challenges and enabling conditions for oxygen management.

6. Producing the report: The final themes were organised into a narrative framework with representative quotations from interviewees to support each theme’s interpretation.

The analysis was primarily inductive, although familiar categories from oxygen management policy literature informed interpretation in later stages. Coding and theme development were conducted iteratively and collaboratively by multiple researchers to enhance analytical rigour, ensure intersubjective consistency, and incorporate insights from diverse disciplinary standpoints.

The final set of ten themes (detailed in Table 6) included both barriers and enablers, and these informed the development and interpretation of the m-TISM model. The identified challenges include infrastructure gaps, lack of coordination, logistical delays, insufficient workforce training, and safety concerns. Conversely, the enablers comprise inter-sectoral support, emergency preparedness, monitoring tools, online training, digital technology, and effective resource allocation. By triangulating these themes with factors identified through secondary research and MICMAC analysis, the thematic analysis helped contextualise the hierarchical structure of oxygen ecosystem variables, strengthening both the construct validity and interpretive richness of the study.

Table 6.

Themes identified from the stakeholders’ interview

Themes	Stakeholder types	Quotes
Infrastructure Gaps	Hospital Admin	“There were insufficient cylinders to meet the demand and limited refining facilities.”
Lack of Coordination	International Organization	“The government, hospital, and private supplier were all working in silos.”
Safety Concerns	Doctors	“Sometimes, oxygen purity is in question; it did not feel safe to administer.”
Lack of Trained Workforce	Healthcare Providers	“Even if we received cylinders, no one really knew how to handle or monitor them.”
Logistics delays	Cylinder Suppliers	“There were delays in oxygen supply due to interstate permissions, which led to trucks stuck for hours.”
	Transportation departments	“There were mismatches in container specifications with liquid medical oxygen leading to delays in oxygen transportation.”
Inter-sectoral support	Government	“Development partners helped us in resource allocation, maintenance, and training of human resources.”
Emergency Preparedness	State or District technical support teams	“Our preparedness for emergency measures ensured prompt oxygen delivery to needy patients. Availability of trained workforce and funding were the main challenges for us”
Online Training	NGO Representatives	“Online training of human resources on handling concentrators and leak prevention helped oxygen wastage using e-learning modules.”
Digital Technology	Website/app developers	“Devices with IoTs, RFID, and GPS enabled us to track oxygen movement in real time and helped redirect it based on the requirement.”
	State Technical Unit	“Oxygen dashboards by states were assisting in procurement and quick mobilization of oxygen.”
Data Analytics	Academic institutional support	“Prediction and forecasting of demand helped in taking production decisions. Also, data-backed decisions strengthened the preparedness for non-pandemic aspects as well.”

The findings of the thematic analysis from stakeholder interviews were systematically mapped onto the factors identified through the literature review and secondary research, ensuring both empirical grounding and conceptual clarity. Each theme either directly informed the selection of factors included in the m-TISM model or provided interpretive context to understand the interrelationships among these factors within the structural hierarchy. Thus, the thematic analysis served as both an input for identifying relevant variables and a qualitative lens for interpreting the model.

Discussion

m-TISM levels

The m-TISM analysis gives us the hierarchical relationship between these factors. The results of the hierarchical level partitioning (Table 4) enabled the identification of different levels in the m-TISM hierarchy.

Level VI of the hierarchy consists of ‘Regulatory Framework’ and ‘Monitoring and Audit’, followed by ‘Production Capacity’, ‘Preparedness and Planning’, and ‘Digital Technology’ on Level V. The factors placed at the lower levels of the m-TISM hierarchy are critical elements in the model. As highlighted in previous studies, multi-sectoral coordination is essential for mitigating oxygen shortage in the country [33, 34]. This TISM analysis further emphasized that a strong regulatory framework, along with coordination, is essential.

Research supports the findings that a country’s regulatory framework plays an indispensable role in managing the production as well as the logistic supply chain while also ensuring policy refinements wherever necessary [35, 36]. It directly impacts the monitoring and audit of oxygen supply by ensuring appropriate guidelines and standards are in place for production, allocation, and distribution to ensure that an adequate amount of safe medical oxygen reaches the patients [14, 37].

Robust audit and monitoring mechanisms would help to design the interoperability of the digital systems, universal accessibility of these systems, and streamlining of information and communication of guidelines or SOPs. Moreover, the regulations and monitoring would help in capacity building to generate medical oxygen at different levels, such as hospitals, districts, or states, and low third-party dependency that may assist in-house production of medical oxygen, such as PSA plants, due to its cost-effectiveness [38].

The regulatory framework also dictates how different digital tools and interventions will be implemented, managed, and monitored to comply with the needs of COVID-19 patients’ oxygen needs in different states [2, 17, 39, 40]. The data analytics and data-backed decisions would help prepare for non-pandemic loads, contingent situations, and crises. The medical oxygen demand aggregation and forecasting led to the decision to determine the production capacity, as it happened during the pandemic when industrial oxygen production was diverted towards medical oxygen production. Transportation of medical oxygen or LMO tankers through railways and airways plays a crucial role in oxygen management.

At level V, the limited production capacity of oxygen in a certain country or region impacts their procurement and acquisition systems to minimize shortages and ensure oxygen is available at affordable prices [37, 41]. This requires a prompt response by both the procurement and the network of distribution systems by coordinating with all the stakeholders in the medical oxygen management supply chain to plan and stay prepared with a well-functioning, flexible, transparent, and efficient system to provide oxygen to patients in need [17, 41, 42].

A well-planned and thoroughly prepared response will assist the authorities in employing digital tools like oxygen dashboards to procure or acquire adequate oxygen supplies and ensure quick mobilization of those oxygen resources by transporting them to health facilities on time to reduce delays in treatment and mortality rates. This emphasis on preparedness was echoed by stakeholders, with one state technical support team official explaining, “Even though there were shortages, our emergency preparedness measures helped us mobilize oxygen quickly where it was most needed. Training and prior planning made a big difference in responding fast.” Being planned and organized will also help in making sure that healthcare workers, trainees, and volunteers know the proper use of oxygen equipment (local and internationally procured) as well as the protocols to be followed while administering and adjusting medical oxygen supply to COVID-19 patients [43–45].

Digital tools like Oxygen Digital Tracking System (ODTS), Oxygen Demand Aggregation System (ODAS), OxyCare- Management Information System (OCMIS), platforms to manage oxygen concentrators, IoT devices, GPS, QR codes, and RFID based tracking of oxygen cylinders, and virtual training will assist in surveillance and monitoring, providing telemedicine facilities, remote learning and training of the workforce, predictive analysis to forecast demand, communication and information sharing as well as supply chain management [37]. Automated ODTS (oxygen demand tracking system), GPS, and SIM cards to track the vehicle through the generation of an e-way bill, which provided details of the driver and the vehicle that helped to communicate and coordinate. Digital interventions helped in making data-backed decisions, leading to the optimal usage and allocation of medical oxygen. E-Learning modules can reach different parts of the country in local languages, not limited to the times of a crisis, but for the long run too. As one academic institutional stakeholder explained, “Digital dashboards and forecasting tools were crucial. They helped us predict demand surges and make production decisions quickly. Even beyond the pandemic, these data-backed systems have strengthened preparedness for routine and emergency oxygen needs.” This underscores how digital interventions have become embedded in health system preparedness, extending their value beyond pandemic-specific contexts.

Next, on Level IV are ‘Procurement and Acquisition System’, ‘Network of Distribution Systems’, and ‘Workforce Capacity and Training’, followed by ‘International Collaboration’ and ‘Public-Private Collaboration’ at Level III. Some of the challenges faced during the COVID-19 oxygen crisis were the transportation of medical oxygen from the manufacturing plant to the health facility, a mismatch in the container specifications containing liquid medical oxygen from different sources, and a lack of availability of tankers for distribution of medical oxygen. Another key concern was the inadequate infrastructure in the medical oxygen ecosystem, which was not sufficient to cater to the medical oxygen demand during the peak surge. For instance, cylinders were not enough to support the demand, and a limited number of cylinder refilling facilities and testing facilities were not available. As one healthcare administrator described, “During the peak of the crisis, we simply didn’t have enough cylinders to meet the skyrocketing demand, and there were also very few refilling and testing facilities available. This left many hospitals struggling to maintain a reliable oxygen supply for patients.” Such accounts emphasize on how infrastructural gaps directly impacted timely treatment and patient safety, emphasizing the urgent need for investment in decentralized production and robust logistical systems.

Efficient procurement and acquisition systems play a crucial role in ensuring sufficient quantities of high-quality medical oxygen are available to be distributed by coordinating with oxygen producers and suppliers, setting up quick contracts, and monitoring the requirements and deliveries [8, 37]. Human Resources was another major area that required attention in terms of training, awareness, and motivation. Coordination among human resources helps minimize time and optimize resource sharing.

Lack of awareness about the technology, PSA plants, oxygen rationing, medical oxygen wastage, or leakage prevention, and the requirement of an appropriate purity level of medical oxygen were areas that are directly related to human resource capacity building. Logistic management requires a trained workforce to manage storing, handling, and distributing oxygen-delivery equipment. This will facilitate timely and cost-effective distribution domestically and also internationally to other countries with a higher caseload and medical oxygen deficiency [12, 46–48].

At the same time, efficient procurement and distribution networks combined with a competent healthcare workforce will ensure coordination, sufficient funding, innovation, and adequate resource allocation by incentivizing and providing subsidies to private players to reduce some of the burdens on the public healthcare system through public-private partnerships and collaborations with international organizations. Moreover, these international collaborations and public-private partnerships will assist in the sharing of oxygen resources and expertise, effective coordination, joint funding efforts, necessary technology, and oxygen equipment transfers to enable a cohesive global approach to managing oxygen supplies during the pandemic [49–51]. These were helpful in many ways, such as getting technical assistance in maintenance, establishing global alliances, additional funds and grants, and medical equipment donations by international agencies or other countries. Whereas, public-private collaboration and including inter-sectoral support among different government departments helped in diverting industrial oxygen to medical oxygen to optimize oxygen rationing and allocation. Corporate funding in the form of CAPEX is supported in building the infrastructure, maintenance, and medical equipment availability.

Finally, at Level II are ‘Medical Equipment and Maintenance’, ‘Quality and Purity’, and ‘Oxygen Stewardship’, with ‘Oxygen Management’ at Level I. The global collaborations and public-private partnerships (Level III) enabled the sharing of best practices and knowledge of oxygen management, along with easing the strain on healthcare systems by providing the required medical equipment like oxygen concentrators, oxygen tanks, and ventilators that meet high technical standards [13, 39, 52, 53]. The private sector, third-party organizations such as technology companies or NGOs, played a supportive role in developing training modules prepared for the quality and purity of the medical oxygen required for the treatment.

Technology platforms equipped the human resource and emergency response teams to minimize the oxygen wastage and improve oxygen stewardship, which was central to the generation of quality data. This ensures patients get consistent and reliable oxygen therapy treatment with minimal or no wastage of this critical care component during the pandemic. Availability and proper maintenance of the right medical equipment further ensure that they are functioning correctly and providing safe and effective oxygen support to patients while also ensuring minimum wastage and following the protocols and guidelines set by international organizations like the WHO and different government bodies [14, 37, 54, 55]. Figure 6 depicts the m-TISM digraph with hierarchical relationships.

Fig. 6.

m-TISM digraph of barriers and facilitators of oxygen management

Comparison with existing literature

The findings of this study contribute to and extend the existing literature on medical oxygen systems, especially in the context of pandemic response and health systems resilience in low- and middle-income countries like India. By using an m-TISM approach, this study provides a hierarchical understanding of how various systemic, technical, and institutional factors interact to shape the oxygen management ecosystem in India.

Our model positions Regulatory Framework, Production Capacity, and Monitoring and Audit systems at the bottom of the hierarchy as the strongest driving factors, enabling all other components of oxygen management. This is consistent with existing literature that highlights the critical role of regulatory reforms and data systems in enabling rapid scale-up in the Southeast Asia region, as well as real-time dashboards and policy waivers during the second wave of COVID-19 in India [14, 37]. These studies attribute India’s rapid capacity expansion to emergency regulatory flexibility and relaxation, as well as real-time audit loops. Unlike some studies that treat production and regulation as critical, but largely operational issues, our model shows they exert the highest driving power, cascading through every other node in the m-TISM hierarchy. Unlike their narrative emphasis on infrastructure gaps, the results of our m-TISM model reveal that regulatory and production factors drive the overall ecosystem, affecting not only supply but also workforce preparedness, distribution efficiency, and even stewardship practices [18].

The procurement and acquisition system, distribution network, and workforce capacity and training occupy intermediate levels in our model, influenced by upstream systems, yet critical to enabling downstream outcomes, as suggested in the diagraph. This aligns with recent research work that describes these factors as both enablers and pressure points in LMIC oxygen supply chains [11, 34]. However, although researchers argue that workforce shortages were the primary barrier in rural Nepal, our findings suggest that India’s use of surge staffing, task-sharing optimization, and digital training partially mitigated these constraints [46]. These differences underscore the importance of context-specific analysis in pandemic systems modeling.

The upper levels of the m-TISM model consist of Oxygen Purity, Medical Equipment and Maintenance, Oxygen Stewardship, and collaborative mechanisms as dependent outcomes. While existing studies have often considered these as immediate operational bottlenecks, this paper reframes them as functions that are heavily dependent on upstream policy, infrastructure, and monitoring systems [19, 56]. This perspective diverges from prior hospital-level analyses by highlighting that local oxygen crises often reflect national-level governance or supply-chain challenges, not just technical or facility-level problems.

This study builds on and adds to existing applications of interpretive structural modelling in health systems research by extending the methodology to the complex medical oxygen infrastructure [26, 27]. While earlier studies have considered vaccination strategies or hospital agility within single institutions, this model addresses interdependencies across multiple government tiers, public-private boundaries, and stages of the oxygen supply chain. The combination of qualitative insights with a formal structural model offers a replicable tool for other countries planning for future health contingencies.

Implications

The outcomes derived from the m-TISM modelling will serve as a roadmap for international organizations and government authorities, assisting them in formulating and executing policies and regulations to ensure the safe and reliable supply of medical oxygen. By prioritizing the driving factors identified at the lower levels of the m-TISM model, efforts can be directed towards influencing the dependent factors at the higher levels. For instance, streamlining regulatory frameworks can expedite oxygen production, procurement, and distribution, while emergency measures can be implemented to ensure prompt delivery of medical oxygen to patients in need. Additionally, readiness to scale up production capacity, ongoing monitoring, and integration of digital solutions will facilitate the efficient organization and distribution of medical oxygen, thereby minimizing delays and barriers to access for both the general public and frontline healthcare workers.

Furthermore, policymakers must address linkage factors and broader healthcare considerations such as the availability of medical equipment, workforce training, capacity building, and international collaborations, all of which have overarching implications for oxygen management across various levels of healthcare delivery. While existing WHO reports and national guidelines offer foundational recommendations, this study provides new, context-specific insights by highlighting how these elements interact in practice and identifying overlooked implementation challenges and enablers. Key areas for attention include infrastructure and human resource capacity building, targeted workforce training initiatives, optimization and consistent operation of infrastructure like Pressure Swing Adsorption (PSA) plants, enhancement of data accuracy and quality, establishment of actionable and locally adaptable audit guidelines, and fostering partnerships with the private sector and international organizations for resource sharing and technical assistance. Additionally, this study highlights the need to actively dispel persistent misconceptions regarding the costs, operational functionality, and installation requirements of PSA plants—issues often inadequately addressed in broader policy frameworks.

Limitations and future research

This study offers valuable insights into the systemic challenges and enablers of oxygen management during COVID-19 in India, yet several constraints must be acknowledged to contextualise the findings and guide subsequent inquiry. Firstly, the m-TISM methodology employed in this study, while effective for mapping structural relationships, does not quantify the strength or directionality of influence between factors. Future research could combine interpretive structural modeling with quantitative techniques such as fuzzy logic or structural equation modeling to validate and strengthen the model’s predictive utility. Secondly, while this purposive sampling approach is appropriate for this research, it limits the generalisability of findings to other states or countries with different health system capacities or arrangements. Furthermore, although the study included multiple institutional stakeholder perspectives, such as government officials, logistics teams, and development partners, it did not incorporate the voices of patients, who are most directly affected by oxygen availability and use. Including these perspectives in future research could offer a more holistic, user-centric understanding of the oxygen ecosystem, especially with regard to the operational, ethical, and equity-related challenges faced during crises. In the future, multi-site comparative studies that integrate diverse stakeholder groups would strengthen the external validity of the model and help refine or adapt the driver–linkage–dependent hierarchy across different epidemiological and institutional contexts.

Since this study primarily captured experiences largely from the pandemic wave, a longitudinal design that tracks how factor interdependencies shift over successive waves or post-pandemic recovery phases would add important temporal nuance and assess system resilience beyond acute phases. Further, this study highlights high-level system relationships but does not empirically assess the implementation fidelity of interventions in the oxygen ecosystem (whether PSA plants were functional, dashboards were used effectively, and so on). Future studies could also incorporate frameworks to evaluate the on-the-ground effectiveness and sustainability of oxygen-related interventions. Additionally, in terms of future research, this m-TISM framework could be applied to other critical aspects of pandemic preparedness, such as vaccination logistics, ICU and bed capacity planning, or the integration of digital health systems. Cross-country comparative studies, particularly among low- and middle-income countries, would help assess the transferability of this model. This would help set up a holistic pandemic-preparedness architecture. Modeling healthcare management for other essential services, such as oxygen for neonatal care, cancer treatment, or surgical procedures, could further extend the methodology’s applicability.

The m-TISM approach was particularly suitable for this study due to the complex and unstructured context of medical oxygen management in India during a crisis. Unlike other modeling methods that require large datasets or predefined quantitative relationships, m-TISM accommodates expert judgment and tacit knowledge, allowing for the modeling of interdependencies based on stakeholder interpretation. This makes it well suited for the current COVID-19 scenario where data availability is constrained but expert experience is rich. However, we recognize that our findings are derived from a purposive sample of twelve stakeholder interviews. While the factors were triangulated with secondary literature and cross-validated with experts, the small sample size constrains the breadth of perspectives. Therefore, the generalizability of the model should be interpreted with caution. Nonetheless, the core structural insights align with challenges seen across other low- and middle-income countries. This suggests that the model may offer transferable lessons for LMICs with similar infrastructural and governance constraints. By addressing these limitations and exploring the proposed research avenues, future scholarship can build on the present study to deliver evidence-based, context-sensitive strategies for strengthening medical oxygen ecosystems and broader health-service supply chains.

Conclusion

This study employed a modified Total Interpretive Structural Modelling approach to examine the facilitators and constraints of medical oxygen management in India during the COVID-19 pandemic. By analysing 14 interdependent factors across six hierarchical levels, the model illustrates how foundational elements, such as regulatory frameworks, audit systems, production capacity, preparedness, and digital infrastructure, collectively influence the effectiveness of oxygen delivery. The findings highlight that oxygen management is not a product of isolated inputs but the outcome of dynamic, multi-level coordination across policy, logistics, and workforce systems.

By offering a structured understanding of these interconnections, the study provides a practical framework for policymakers, health administrators, and implementation partners seeking to strengthen oxygen ecosystems across the world. The m-TISM model can inform prioritisation of high-leverage interventions, promote alignment between technological tools and institutional governance, and support the design of integrated, responsive oxygen supply chains. As health systems worldwide invest in long-term resilience, this research contributes actionable insights for building more equitable and durable access to medical oxygen in crisis and non-crisis contexts.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ADB

Asian Development Bank

AIIGMA

All India Industrial Gas Manufacturers Association

GPS

Global Positioning System

ICU

Intensive Care Unit

IoT

Internet of Things

LMIC

Low and middle-income countries

LMO

Liquid medical oxygen

m-TISM

Modified Total Interpretive Structural Modeling

MICMAC

Matrice d’ Impacts croises multiplication applique an classment

NGO

Non-governmental Organization

NIC

National Informatics Centre

OCMIS

OxyCare- Management Information System

ODAS

Oxygen Demand Aggregation System

ODTS

Oxygen Digital Tracking System

PSA

Pressure swing adsorption

SOP

Standard operating procedure

UNICEF

United Nations International Children’s Emergency Fund

USAID

United States Agency for International Development

WHO

World Health Organisation

Author contributions

MS: Made substantial contributions to the conception and design of the work; the acquisition, analysis, and interpretation of data; has drafted the work and has approved the submitted version. SD: Made substantial contributions to the conception and design of the work; the acquisition, analysis, and interpretation of data; has drafted the work and has approved the submitted version. JK: Made substantial contributions to the conception and design of the work; the acquisition, analysis, and interpretation of data; has drafted the work and has approved the submitted version. LS: Made substantial contributions to the conception and design of the work; the acquisition, analysis, and interpretation of data; has drafted the work and has approved the submitted version. RC: Made substantial contributions to the conception and design of the work; the acquisition, analysis, and interpretation of data; has drafted the work and has approved the submitted version. NT: Made substantial contributions to the conception and design of the work; the acquisition, analysis, and interpretation of data; has drafted the work and has approved the submitted version.

Funding

The study is conducted with the collaboration of PATH and IIT Delhi, PATH being the funding agency.

Data availability

All data generated or analysed during this study are included in this published article [and its supplementary information files].

Declarations

Ethics approval and consent to participate

The study adhered to the Declaration of Helsinki. This study received ethics approval from the Western Copernicus Group Institutional Review Board (WCG IRB Tracking Number: 20235125) on 28th November 2023. The IRB was in full compliance with Good Clinical Practices, FDA regulations, HHS regulations, and ICH guidelines. All participants provided informed consent before participating in the study. A detailed consent form, titled " Informed Consent Document”, was approved by the WCG IRB on 28th November 2023 and used during data collection. This includes a clear explanation of the study’s purpose, risks, and confidentiality measures. This document has been uploaded with supplementary documents.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.