You are viewing the site in preview mode

Skip to main content
Download PDF
Download ePub

Generalizing from qualitative data: a case example using critical realist thematic analysis and mechanism mapping to evaluate a community health worker-led screening program in India

Implementation Science volume 19, Article number: 81 (2024) Cite this article

Abstract

Background

A central goal of implementation science is to generate insights that allow evidence-based practices to be successfully applied across diverse settings. However, challenges often arise in preserving programs’ effectiveness outside the context of their intervention development. We propose that qualitative data can inform generalizability via elucidating mechanisms of an intervention. Critical realist thematic analysis provides a framework for applying qualitative data to identify causal relationships. This approach can be used to develop mechanism maps, a tool rooted in policy that has been used in health systems interventions, to explain how and why interventions work. We illustrate use of these approaches through a case example of a community health worker (CHW)-delivered gestational diabetes (GDM) screening intervention in Pune, India. CHWs successfully improved uptake of oral glucose tolerance tests (OGTT) among pregnant women, however clinical management of GDM was suboptimal.

Methods

Qualitative interviews were conducted with 53 purposively sampled participants (pregnant women, CHWs, maternal health clinicians). Interview transcripts were reviewed using a critical realist thematic analysis approach to develop a coding scheme pertinent to our research questions: “What caused high uptake of GDM screening?” and “Why did most women with GDM referred to clinics did not receive evidence-based management?”. Mechanism maps were retrospectively generated using short- and long-term outcomes as fenceposts to illustrate causal pathways of the CHW–delivered program and subsequent clinical GDM management.

Results

Critical realist thematic analysis generated mechanism maps showed that CHWs facilitated GDM screening uptake through affective, cognitive and logistic pathways of influence. Lack of evidence-based treatment of GDM at clinics was caused by 1) clinicians lacking time or initiative to provide GDM counseling and 2) low perceived pre-test probability of GDM in this population of women without traditional risk factors. Mechanism mapping identified areas for adaptation to improve the intervention for future iterations.

Conclusions

Mechanism maps created by repeated engagement following the critical realist thematic analysis method can provide a retrospective framework to understand causal relationships between factors driving intervention successes or failures. This process, in turn, can inform the generalizability of health programs by identifying constituent factors and their interrelationships that are central to implementation.

Peer Review reports

Introduction

Critical realism is a philosophical approach that has become increasingly applied to the field of implementation science to explain the process and outcomes of implementation. The strength of critical realism in implementation science is that it accounts for the complex nature of evidence-based interventions and focuses on explaining what works under specific conditions and contexts [1]. A critical realist lens interrogates the relationships between individuals and their contexts [2, 3], as well as influences by structures and other agents, to identify causal mechanisms and their effects on health outcomes. The critical realist approach conceives of structures and conditions in specific contexts acting through mechanisms to produce the observed effect or event [4]. While mechanisms are typically not directly observable, they are identified through the process of retroduction [5] – working backwards from empirical events and identifying causal forces that explain the events observed. Through this iterative process, researchers propose multiple explanations (or mechanisms) and investigate their validity through data corroboration [6].

Identifying mechanisms to understand causation within an implementing system can potentially inform generalizability of evidence-based programs, a central aspect of facilitating effective and equitable delivery of evidence-based care [7,8,9]. Generalizability refers to the applicability of findings to an unknown or wider population and is central to the implementation science mission of reducing the evidence-practice gap. To facilitate maximum impact of testing interventions in a specific study context, it is critical to understand the causal mechanisms at play during implementation.

An existing approach in implementation science to identify and evaluate mechanisms empirically is using quantitative data to create directed acyclic graphs (DAGs) [10,11,12], which allows for weighted modeling of mechanisms. A weakness of using hypothetical DAGs, however, is that they are limited in scope to variables known to the researcher. Novel variables are challenging to capture using these methods as the variables are, by definition, generated by the researcher. Qualitative data, in contrast, allows for novel variable capture by including the voices of study participants. Realist evaluations in implementation science have drawn from theories of critical realism to understand how and why interventions work under different circumstances [13]. Using a “context + mechanism = outcome” formula as a guiding principle, realist studies focus on linking contextual drivers with clinical outcomes through theoretical mechanisms of action. One challenge with realist approaches has been a paucity of standardized methods or protocols for this type of analysis, leading to varied analytical methods and confusion among researchers [14]. Pawson & Tilley recommend using mixed methods data to evaluate hypothesized context-mechanism-outcome relationships [15]. However, realist evaluations do not routinely engage with these data to identify mechanisms, with a few notable exceptions [16,17,18].

Fryer (2022) posits that thematic analysis – a method common to qualitative research – should be operationalized under the lens of critical realism to “produce nuanced causal explanations of events, countering the mistaken assumption that qualitative research cannot produce causal knowledge” [19]. Qualitative data provide critical details on the implementation context of an intervention, the possibility for inductive knowledge generation, and describes structural and interpersonal parameters that influence measured (as well as unexpected) study outcomes. These data, therefore, are uniquely positioned to theorize detailed mechanisms, which can be applied to the task of creating generalizable knowledge about implementation. Despite the strength of qualitative data to speak to these issues, very few researchers have engaged with critical realist approaches to qualitative data in this way. Here, we embrace Fryer’s assertion that highly contextual qualitative data can be leveraged to elucidate detailed understandings of causal relationships between implementation structures, context, and outcomes.

In this paper, we show how critical realist thematic analysis can be utilized to provide a deeper understanding of intervention generalizability through identifying causal mechanisms to generate mechanism maps. Mechanism mapping, which has previously been used to understand outcomes in health systems interventions [20, 21], breaks down an intervention into its constituent parts, relating components of implementation strategies and how these influence intervention outcomes [22]. Mechanism maps expand upon the structure-context-mechanism-outcome framework in critical realism to include non-linear and intertwined relationships of an implementation strategy, the core steps of an intervention, how these pathways interact with each other, and how contextual factors influence one or more mechanisms [21]. Mechanism mapping models intend to explain how a proposed intervention’s theory of change interacts with its context, providing a systematic, data-driven approach to identifying the causal mechanisms driving intervention outcomes [23]. To illustrate this, we provide an example using data from a maternal health intervention in Pune, India where community health workers (CHW) were trained to deliver home-based gestational diabetes (GDM) screening in two slum communities. Through this example, we illustrate how a retrospective critical realist thematic analysis approach was used to create mechanism maps that, in turn, can inform next steps in scaling up this intervention.

Overview of CHW-delivered gestational diabetes study

We conducted an explanatory mixed methods study of gestational diabetes screening in Pune, India, This study was conducted from October 2021 to June 2022 [24]. The work was conducted in collaboration with the Deep Griha Society, a local non-governmental organization with nearly 50 years’ experience providing child welfare, nutrition support, and health promotion programs in Pune’s slum communities (https://deepgriha.org). Details of the intervention are described in Chebrolu et al., 2023 [24]. In brief, we trained five community health workers (CHWs) from Deep Griha Society to conduct oral glucose tolerance tests (OGTTs), the gold standard screening tool for gestational diabetes, in people’s homes. We recruited 248 pregnant women in our study; of these 90% (n = 223) accepted the OGTT delivered by the CHW. Thirty-one women (14%) screened positive for GDM and were referred to antenatal clinics for GDM care by the CHWs. After two weeks, CHWs followed up in person with the women that screened positive for GDM to determine if they had sought care and to ask about clinical management, if any. Nearly all women with GDM had sought clinical care (97%, n = 30); however, only 33% of these (n = 10) received any counseling or treatment from the clinician for GDM. No incentives were provided to pregnant women to encourage acceptance of the OGTT or to attend clinic following screening.

At study completion, a female Ph.D. social scientist not affiliated with the study conducted 53 semi-structured interviews in Marathi (the local language) with a purposive sampled subset of 30 pregnant women (20 of whom had screened positive for GDM), all 5 CHWs, and 18 maternal health clinicians from the Pune area. Participants were identified to represent those with and without GDM; those who screened positive for GDM were purposefully overrepresented so that we could learn about the continuum of care after clinic referral, as well as clinicians from both public and private facilities providing antenatal care in the study region. Characteristics of participants in the qualitative study are shown in Appendix Table 3. Interview sample size was determined based data saturation estimates in the literature and confirmed through data analysis [25]. Qualitative interview participants received a gift (either household staples or snacks) valued at approximately 200 rupees (approximately $2.50 US Dollars).

An interview guide was developed based on the Consolidated Framework for Implementation Research framework and used to ensure consistency of topics across interviews while allowing for novel concepts to arise (S1- S3 Text). Pregnant women were asked about receiving screening from CHWs and experience with clinical GDM care. CHWs were asked about providing GDM screening and counseling in their communities. Clinicians were asked about their perceptions of GDM prevalence, diagnosis, and treatment. Interviews were audio recorded, transcribed, and translated from Marathi into English for analysis by a professional translation service. Both Marathi and English transcripts were produced for each interview. One quarter of all English transcripts were spot-checked against the Marathi transcripts and discussed with the study’s qualitative interviewer – who is fluent in both Marathi and English – to ensure preservation and fidelity of meaning.

We retroductively analyzed these qualitative data using critical realist thematic analysis to empirically create mechanism maps explaining the causal relationships between context, implementation strategies, and observed clinical outcomes. The process of the critical realist thematic analysis and mechanism map creation is presented here as an example to illustrate application of this novel approach.

Methods

Logic model creation

Prior to study initiation, a logic model describing the intervention’s theory of change and proposed facilitators and barriers was created following repeated engagement with the literature on CHW task-shifting interventions, the study team’s prior experience with maternal health and community-based research, and informal conversations with colleagues and community health workers [26]. This logic model summarizes, in broad strokes, the anticipated process, facilitators, barriers, and outcomes of the intervention (Fig. 1). With OGTT administration training and material support, we anticipated that the CHWs would successfully engage in community outreach, home-based testing, and referrals to clinical care. The intended outcomes of this process were uptake of the OGTT among pregnant women and referral to clinical care for those screening positive for GDM (short term), individual GDM management (medium term), and improved population maternal health (long term).

Fig. 1
figure 1

Logic model summarizing the proposed theory of change of CHW-facilitated GDM screening

Full size image

Our quantitative data demonstrated that our CHW-delivered screening program resulted in achievement of anticipated short term clinical outcomes but did not align with the hypothesized medium–term outcome of women receiving evidence-based GDM management, and therefore would preclude achievement of the long-term goal of improved population maternal health. This prompted us to critically re-evaluate our qualitative data with a focus on explaining why our program failed to achieve distal outcomes, a “pre-mortem” approach that has been suggested to leverage hindsight to generate an explanation and prevent a poor outcome in the future [27].

Qualitative data analysis

All steps in our retrospective analysis process are summarized in Table 1 and discussed in detail below.

Table 1 Summary of mechanism map creation
Full size table

Critical realist thematic analysis

Interview transcripts were reviewed, using a critical realist thematic analysis approach described by Fryer [19], to develop a coding scheme pertinent to our research questions: “What caused high uptake of the CHW-delivered GDM screening?” and “Why did most women GDM referred from this study not receive evidence-based treatment at the clinics?”.

We followed Fryer’s approach to critical realist thematic analysis as follows:

  • Step 1: We began by establishing our events (the experience of CHW-delivered GDM screening) and defining our two research questions, as above.

  • Step 2: Authors KB, MP, and AC familiarized themselves with the qualitative data, skimming a large portion of the interviews and taking general notes. Initial descriptive codes were then generated using a data-led approach. Identified codes and themes were organized using Microsoft Word and Excel. Pertinent quotes were highlighted in Word then copied and pasted into Excel. Each participant ID was placed in an Excel row and initial descriptive codes placed in separate columns. Illustrative quotes were placed to provide “evidence” for the descriptive codes.

  • Step 3: Authors KB, MP, and AC reviewed codes to ensure standardization (using the same word for similar codes) and facilitate consolidation (bringing similar thematic codes together into themes). Disagreements or discrepancies in codes were resolved through discussion between the three authors and senior author RS. An updated Excel sheet was created with consolidated codes explaning the two research questions: 1) high uptake of the CHW-delivered screening and 2) paucity of evidence-based treatment for women with GDM referred to local clinics.

  • Step 4: Causal explanations were drafted from these revised codes and iteratively tested for validity by re-reading the interviews in full through the lens of the themes to test for validity. While Fryer describes causal explanations as “themes”, we present them here as “mechanisms” insofar as these are factors we identified as influencing outcomes within the study context.

  • Step 5: Graphs of causal mechanisms and their relationships to intervention context/outcomes were created and then reviewed across the authorship team to interrogate the validity of the conclusions made. For cases where the authorship team did not agree on causal mechanisms or the map components, we reverted to Steps 3 and 4 again to review the primary data. After mechanisms and maps were finalized through consensus, we initiated discussions on how best to disseminate these findings and agreed on creating this methodology paper to report our process and results.

Mechanism map creation

Drafts of the mechanism maps were created by authors KB and AV based on the ‘themes’ identified during analysis to illustrate the linkage of component steps to mechanisms within the study context [28]. The intervention’s short- and medium-term outcomes – identified by the initial logic model – were placed as fenceposts for the maps (Fig. 2). Mechanism maps were then generated by positioning the key steps in relation to one another and connecting the identified causal relationships as identified by critical realist thematic analysis. Repeated engagement with the data throughout the process ensured that maps reflected participant experiences [29]. Representative quotes supporting the development of the mechanistic pathways can be found in Appendix Tables 4 and 5.

Fig. 2
figure 2

Anticipated short- and medium-term outcomes serve as fenceposts for the intervention’s mechanism map

Full size image

Discordance of observed outcomes with logic model

In cases where the outcomes predicted by our implementation logic model were not observed, mechanism mapping interrogated the gap between initial and empirically derived mechanisms. First, authors KB and AV revisited the logic model and utilized the thematic analysis to identify which stage(s) the intervention failed to align (Appendix Figure 5). Authors then reengaged with qualitative analysis to characterize the contextual factors that precipitated the observed breakdown and to generate ideas regarding context modifications that would bridge the implementation gap. This included consideration of outcomes that aligned with the logic model but worked through unanticipated mechanisms. This learning intends to target improvement of the intervention to more effectively achieve long-term goals of improving population health.

Case example

Research question 1: What caused high uptake of the CHW-delivered GDM screening?

CHWs trained in the delivery of home-based OGTT successfully improved GDM screening through three distinct mechanisms: affective, cognitive, and logistic influences (Fig. 3). CHWs operated affectively in the context of low social distance, as peers and members of the same community, which facilitated the formation of social bonds and allowed them to serve in the role of informal advisors. Pregnant participants described looking to CHWs for guidance on health-related topics and stated that CHWs explained the OGTT and GDM in a manner they understood, given low baseline knowledge on GDM in the communities [30, 31]. Lastly, participants cited the need for transportation and time away from the household as barriers to clinic based GDM screening care.

Fig. 3
figure 3

Mechanism map explaining high OGTT uptake and high rates of presentation to clinic for GDM care

Full size image

CHW-delivered counseling resulted in nearly all participants who screened positive for GDM presenting to clinic within two weeks. The context of CHW-provided brokered information and counseling served as an important backdrop. Affectively, the social bond between the CHW and the participant was important because the recommendation to seek a higher level of care came from a trusted peer. Cognitively, participants’ new understanding of their risk of GDM led to concern for her fetus. The confluence of both affective and cognitive factors led to the prioritization of attending clinic despite the aforementioned barriers in this low-resource setting.

Research question 2: Why did most women with GDM referred from this study not receive evidence-based treatment at the clinics?

Once pregnant women who screened positive for GDM presented to clinic, interviews indicated numerous drivers of inconsistent management by maternal health clinicians. The standard of care involves a fasting plasma glucose OGTT to confirm the diagnosis of GDM [32]. Clinician participants, however, expressed skepticism regarding the validity of our CHW-delivered GDM screening test and were uncertain or unaware of a protocol to repeat an OGTT with a venous blood draw. Clinicians also had low concern for GDM in this low-income population, given their lack of traditional risk factors such as obesity, thereby reducing their pre-test probability that our CHW-delivered tests were accurate.

Very few women who screened positive for GDM by the CHW-delivered test were provided with any recommendations regarding the need for medication or dietary changes. Some clinicians expressed concern that individual counseling would be ineffective in the cultural and socioeconomic context of slum communities, citing beliefs that poor pregnant women were active and did not have access to unhealthy foods – and therefore were not at risk for GDM. Moreover, clinicians operated within the context of temporal scarcity, operating above capacity and without adequate staffing, so some stated they did not have the time to provide lifestyle or nutritional counseling (Fig. 4).

Fig. 4
figure 4

Mechanism map of clinical management for participants screening positive for GDM

Full size image

Discussion

Mechanism mapping using qualitative data through a critical realist lens provided insight into the causal relationships driving the observed outcomes of our CHW-delivered GDM screening program. To our knowledge, this is the first study to retrospectively create mechanism maps to understand drivers of implementation using critical realist thematic analysis of qualitative data. We posit that qualitative data can be used to identify mechanistic relationships explaining intervention outcomes, brings to the forefront important considerations for generalizability of findings. Our work also addresses the call to interrogate challenges in implementation science research [27, 33]; the mismatch between the assumptions in our original logic model and observed clinical outcomes was the nidus from which this reflexive engagement grew.

Generalizability is a central component of scalability [34] – continuously increasing reach or adoption of an intervention across populations. We believe that the methodological approach illustrated here can facilitate progress towards study generalizability by illustrating determinants of implementation. In their consideration of ‘scaling out’ of evidence-based programs, Aarons et al. (2017) state implementation scientists must determine if “there is sufficient empirical evidence or justification that this evidence-based program would impact health as expected” in a new context [35]. Our use of critical realist thematic analysis and mechanism mapping can provide an evidence base for generalization, by explaining how the program worked (and did not work) and illustrating the core mechanisms contributing to desired clinical outcomes of screening uptake and evidence based GDM management. Further, the task of generalizability is strongly linked to a thorough understanding of context [36]. Our approach accounted for the influence of highly detailed, local experiences on study outcomes while extrapolating on causality in overarching way that can be used to frame evaluations of other study settings. Our approach to mechanism mapping informs generalizability of implementation research by methodically evaluating the intervention and teasing apart what aspects of intervention success or failure could be modifiable and which were inextricably tied to contextual factors. We interrogated our original theory of change, organizing data to elucidate the steps and contextual elements that contributed to the actual implementation outcomes. The retroductive observation of mechanisms from the data by building a mechanism map allowed for reflection on the assumptions underlying our theory of change. In turn, the comparison between theoretical and indirectly observed mechanisms facilitated changes to the intervention to better match the causal mechanisms found through this critical realist analysis. This approach aligns with the “bottom up” strategy of generalization described by Borgstede & Scholz (2021) [37].

Unfortunately, our initial logic model was misaligned with the implementation context, a circumstance that has been described in other study circumstances [20, 21, 27], highlighting the need for rigorous and thorough pre-implementation work to identify these types of barriers. Our findings also demonstrate the importance of bringing empirical data to challenge and potentially deconstruct errors in researchers’ assumptions. Reflexivity of the researcher is essential to advancing an equitable approach to implementation science [38], and a central component of reflexivity is laying bare one’s assumptions. Reflexivity is even more critical when conducting global health research where health systems are already vulnerable and under-resourced, and pragmatic research – such as ours – places additional strain on these systems [39]. Our critical realist thematic analysis and mechanism maps demonstrated the need to support clinicians caring for women with GDM, which we have integrated into a central pillar of the ongoing cluster-randomized clinical trial which expands upon our pilot work (NCT06209411).

Mechanism mapping done in this manner can also unveil the relative importance of correctly predicted mechanisms. For example, based on our engagement with prior data, we believed that reducing logistical barriers was the primary benefit of this community-based screening program; however, our data showed that CHWs’ peer advisory roles was a major driver of OGTT uptake and presentation to clinical care for women screening positive for GDM. Finding this affective pathway can inform generalizability and transferability. Table 2 summarizes the mechanisms of action in identified in our analysis, and contextual considerations regarding generalization to other settings. These considerations of how mechanisms affect implementation outcomes are not limited to CHW-delivered or community-based programming. For example, our study notes that activating social networks through near-peer relationships between CHWs and women in this study was an important mechanism facilitating uptake of evidence-based GDM screening. Peer-delivered counseling may facilitate adoption by activating the affective mechanism more successfully than nurse- or physician-delivered programs. This is borne out in the literature, where peer support has been described as a highly effective approach in both community- [40,41,42] and facility-based [43, 44] interventions.

Table 2 Summary of observed mechanisms and contextual considerations for generalizing to other contexts
Full size table

Using a critical realist thematic analysis approach to identify causal mechanism to explain intervention outcomes can also provide targets for adaptation. While qualitative data used to inform adaptation is highly contextual, the process of adaptation potentially increases intervention generalizability by creating interventions that are more responsive and aligned with structural and social constraints [45]. While our intervention was highly successful in its originally stated goal (GDM screening uptake), our medium-term goal of improved GDM management was not achieved due to observed clinical management that felt short of gold standard, evidence-based management. By using mapping to further understanding what those factors are and how they influenced study outcomes through qualitative data, adaptations can be designed overcome barriers or replicate facilitators to foster intervention success in more generalized settings. By disaggregating the intervention into component, related parts, our analysis pinpointed areas for targeted adaptations to improve medium- and long- term participant outcomes in various contexts. Our illustration of targeted adaptation intends to contribute to literature on implementation generalizability by highlighting how adaptations can improve engagement as well as clinical effectiveness [46].

Conclusion

Mechanism maps generated through critical realist thematic analysis of qualitative data can provide a detailed understanding of causal relationships driving implementation of evidence-based practices. Qualitatively derived mechanism maps could be used to generalize implementation of evidence-based programs across global contexts.

Data availability

The datasets generated and/or analysed during the current study are not publicly available but are available from the corresponding author on reasonable request. The TIDieR and COREQ reporting checklists are available as supplemental materials.

References

  1. Williams L, Rycroft-Malone J, Burton CR. Bringing critical realism to nursing practice: Roy Bhaskar'scontribution. Nurs Philos. 2017;18(2). https://doiorg.publicaciones.saludcastillayleon.es/10.1111/nup.12130.

  2. Bhaskar, R. A Realist Theory of Science (1st ed.). London: Routledge; 2008.

  3. Bunge M. How Does It Work?: The Search for Explanatory Mechanisms. Philos Soc Sci. 2004;34(2):182–210.

    Article  Google Scholar 

  4. Sayer A. Realism and Social Science. Sage Publications; 1999

  5. Mingers J. Real-izing information systems: critical realism as an underpinning philosophy for information systems. Inf Organ. 2004;14(2):87–103.

    Article  Google Scholar 

  6. Wynn Jr D, Williams CK. Principles for conducting critical realist case study research in information systems. MIS quarterly. 2012:787-810.

  7. Lewis CC, Klasnja P, Powell BJ, Lyon AR, Tuzzio L, Jones S, Walsh-Bailey C, Weiner B. From Classification to Causality: Advancing Understanding of Mechanisms of Change in Implementation Science. Front Public Health. 2018;7(6):136. https://doiorg.publicaciones.saludcastillayleon.es/10.3389/fpubh.2018.00136.PMID:29868544;PMCID:PMC5949843.

    Article  Google Scholar 

  8. Lewis CC, Powell BJ, Brewer SK, Nguyen AM, Schriger SH, Vejnoska SF, Walsh-Bailey C, Aarons GA, Beidas RS, Lyon AR, Weiner B, Williams N, Mittman B. Advancing mechanisms of implementation to accelerate sustainable evidence-based practice integration: protocol for generating a research agenda. BMJ Open. 2021;11(10):e053474. https://doiorg.publicaciones.saludcastillayleon.es/10.1136/bmjopen-2021-053474.PMID:34663668;PMCID:PMC8524292.

    Article  PubMed  PubMed Central  Google Scholar 

  9. Lewis CC, Frank HE, Cruden G, Kim B, Stahmer AC, Lyon AR, Albers B, Aarons GA, Beidas RS, Mittman BS, Weiner BJ, Williams NJ, Powell BJ, MNoE Group. A research agenda to advance the study of implementation mechanisms. Implement Sci Commun. 2024;5(1):98. https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s43058-024-00633-5. PMID: 39285504; PMCID: PMC11403843.

    Article  PubMed  PubMed Central  Google Scholar 

  10. Rodrigues D, Kreif N, Lawrence-Jones A, Barahona M, Mayer E. Reflection on modern methods: constructing directed acyclic graphs (DAGs) with domain experts for health services research. Int J Epidemiol. 2022;51(4):1339–48.

    Article  PubMed  PubMed Central  Google Scholar 

  11. Tsai CL, Camargo CA. Methodological considerations, such as directed acyclic graphs, for studying “acute on chronic” disease epidemiology: chronic obstructive pulmonary disease example. J Clin Epidemiol. 2009;62(9):982–90.

    Article  PubMed  Google Scholar 

  12. Rosenberg NE, Westreich D, Bärnighausen T, Miller WC, Behets F, Maman S, et al. Assessing the effect of HIV counselling and testing on HIV acquisition among South African youth. AIDS Lond Engl. 2013;27(17):2765–73.

    Article  Google Scholar 

  13. Dalkin SM, Greenhalgh J, Jones D, Cunningham B, Lhussier M. What’s in a mechanism? Development of a key concept in realist evaluation. Implement Sci. 2015;10(1):49.

    Article  PubMed  PubMed Central  Google Scholar 

  14. Salter KL, Kothari A. Using realist evaluation to open the black box of knowledge translation: a state-of-the-art review. Implement Sci. 2014;9(1):115.

    Article  PubMed  PubMed Central  Google Scholar 

  15. Pawson R, Tilley N. Realistic Evaluation. SAGE publications; 1997.

  16. Haynes A, Gilchrist H, Oliveira JS, Tiedemann A. Using Realist Evaluation to Understand Process Outcomes in a COVID-19-Impacted Yoga Intervention Trial: A Worked Example. Int J Environ Res Public Health. 2021;18(17):9065.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  17. Tsang JY, Blakeman T, Hegarty J, Humphreys J, Harvey G. Understanding the implementation of interventions to improve the management of chronic kidney disease in primary care: a rapid realist review. Implement Sci. 2016;11(1):47.

    Article  PubMed  PubMed Central  Google Scholar 

  18. Bick DE, Rycroft-Malone J, Fontenla M. A case study evaluation of implementation of a care pathway to support normal birth in one English birth centre: anticipated benefits and unintended consequences. BMC Pregnancy Childbirth. 2009;9(1):47.

    Article  PubMed  PubMed Central  Google Scholar 

  19. Fryer T. A critical realist approach to thematic analysis: producing causal explanations. J Crit Realism. 2022;21(4):365–84.

    Article  Google Scholar 

  20. Geng EH, Baumann AA, Powell BJ. Mechanism mapping to advance research on implementation strategies. PLoS Med. 2022;19(2):e1003918. https://doiorg.publicaciones.saludcastillayleon.es/10.1371/journal.pmed.1003918.

  21. Kilbourne AM, Geng E, Eshun-Wilson I, Sweeney S, Shelley D, Cohen DJ, et al. How does facilitation in healthcare work? Using mechanism mapping to illuminate the black box of a meta-implementation strategy. Implement Sci Commun. 2023;4(1):53.

    Article  PubMed  PubMed Central  Google Scholar 

  22. Williams MJ. External Validity and Policy Adaptation: From Impact Evaluation to Policy Design. Health Nutr Popul. 2020;35(2):158–91.

    Google Scholar 

  23. Williams MJ. External Validity and Policy Adaptation: A Five-step Guide to Mechanism Mapping. Policy Memo, Blavatnik School of Government, Oxford University, Oxford, UK, https://www.bsg.ox.ac.uk/research/publications/external-validity-and-policy-adaptation-0 . 2017.

  24. Chebrolu P, Chalem A, Ponticiello M, Broderick K, Vaidyanathan A, Lorenc R, et al. A community health worker-led program to improve access to gestational diabetes screening in urban slums of Pune, India: Results from a mixed methods study. PLOS Glob Public Health. 2023;3(10): e0001622.

    Article  PubMed  PubMed Central  Google Scholar 

  25. Hennink MM, Kaiser BN, Marconi VC. Code Saturation Versus Meaning Saturation: How Many Interviews Are Enough? Qual Health Res. 2017;27(4):591–608.

    Article  PubMed  Google Scholar 

  26. Thompson HM, Clement AM, Ortiz R, Preston TM, Quantrell ALW, Enfield M, et al. Community engagement to improve access to healthcare: a comparative case study to advance implementation science for transgender health equity. Int J Equity Health. 2022;21(1):104.

    Article  PubMed  PubMed Central  Google Scholar 

  27. Beidas RS, Dorsey S, Lewis CC, Lyon AR, Powell BJ, Purtle J, Saldana L, Shelton RC, Stirman SW, Lane-Fall MB. Promises and pitfalls in implementation science from the perspective of US-based researchers: learning from a pre-mortem. Implement Sci. 2022;17(1):55. https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13012-022-01226-3.PMID:35964095;PMCID:PMC9375077.

    Article  PubMed  PubMed Central  Google Scholar 

  28. Lewis CC, Klasnja P, Lyon AR, Powell BJ, Lengnick-Hall R, Buchanan G, Meza RD, Chan MC, BoyntonMH, Weiner BJ. The mechanics of implementation strategies and measures: advancing the study of implementation mechanisms. Implement Sci Commun. 2022;3(1):114. https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s43058-022-00358-3.

  29. Lewis CC, Boyd MR, Walsh-Bailey C, Lyon AR, Beidas R, Mittman B, Aarons GA, Weiner BJ, Chambers DA. A systematic review of empirical studies examining mechanisms of implementation in health. Implement Sci. 2020;15(1):21. https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13012-020-00983-3.

  30. Prabh JK, Kondamuri SD, Samal S, Sen M. Knowledge of gestational diabetes mellitus among pregnant women in a semiurban hospital - A cross -sectional study. Clin Epidemiol Glob Health.2021(12). https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.cegh.2021.100854.

  31. Shriraam V, Rani MA, Sathiyasekaran BWC, Mahadevan S. Awareness of gestational diabetes mellitus among antenatal women in a primary health center in South India. Indian J Endocrinol Metab. 2013;17(1):146–8.

    Article  PubMed  PubMed Central  Google Scholar 

  32. International Association of Diabetes and Pregnancy Study Groups Consensus Panel; Metzger BE, Gabbe SG, Persson B, Buchanan TA, Catalano PA, Damm P, Dyer AR, Leiva Ad, Hod M, Kitzmiler JL, Lowe LP, McIntyre HD, Oats JJ, Omori Y, Schmidt MI. International association of diabetes and pregnancy study groups recommendations on the diagnosis and classification of hyperglycemia in pregnancy. Diabetes Care. 2010;33(3):676-82. https://doiorg.publicaciones.saludcastillayleon.es/10.2337/dc09-1848.

  33. Chambers DA, Emmons KM. Navigating the field of implementation science towards maturity: challenges and opportunities. Implementation Sci. 2024;19:26. https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13012-024-01352-0.

    Article  Google Scholar 

  34. Milat AJ, King L, Bauman AE, Redman S. The concept of scalability: increasing the scale and potential adoption of health promotion interventions into policy and practice. Health Promot Int. 2013;28(3):285–98.

    Article  PubMed  Google Scholar 

  35. Aarons GA, Sklar M, Mustanski B, Benbow N, Brown CH. “Scaling-out” evidence-based interventions to new populations or new health care delivery systems. Implement Sci. 2017;12(1):111.

    Article  PubMed  PubMed Central  Google Scholar 

  36. Nilsen P, Bernhardsson S. Context matters in implementation science: a scoping review of determinant frameworks that describe contextual determinants for implementation outcomes. BMC Health Serv Res. 2019;19(1):189.

    Article  PubMed  PubMed Central  Google Scholar 

  37. Borgstede M, Scholz M. Quantitative and Qualitative Approaches to Generalization and Replication–A Representationalist View. Front Psychol. 2021Feb;5(12):605191.

    Article  Google Scholar 

  38. Snell-Rood C, Jaramillo ET, Hamilton AB, Raskin SE, Nicosia FM, Willging C. Advancing health equity through a theoretically critical implementation science. Translational behavioral medicine. 2021;11(8):1617–25.

    Article  PubMed  PubMed Central  Google Scholar 

  39. Adsul P, Shelton RC, Oh A, Moise N, Iwelunmor J, Griffith DM. Challenges and Opportunities for Paving the Road to Global Health Equity Through Implementation Science. Annu Rev Public Health. 2024;45(1):27–45. https://doiorg.publicaciones.saludcastillayleon.es/10.1146/annurev-publhealth-060922-034822. (Epub 2024 Apr 3 PMID: 38166498).

    Article  PubMed  Google Scholar 

  40. Wright JL, Achieng F, Tindi L, Patil M, Boga M, Kimani M, Barsosio HC, Juma D, Kiige L, Manu A, Kariuki S, Mathai M, Nabwera HM. Design and implementation of a community-based mother-to-mother peer support programme for the follow-up of low birthweight infants in rural western Kenya. Front Pediatr. 2023;3(11):1173238. https://doiorg.publicaciones.saludcastillayleon.es/10.3389/fped.2023.1173238.PMID:37465422;PMCID:PMC10352086.

    Article  Google Scholar 

  41. Lu S, Leduc N, Moullec G. Type 2 diabetes peer support interventions as a complement to primary care settings in high-income nations: A scoping review. Patient Educ Couns. 2022;105(11):3267–78. https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.pec.2022.08.010. (Epub 2022 Aug 20 PMID: 36038395).

    Article  PubMed  Google Scholar 

  42. Chang LW, Nakigozi G, Billioux VG, Gray RH, Serwadda D, Quinn TC, Wawer MJ, Bollinger RC, Reynolds SJ. Effectiveness of peer support on care engagement and preventive care intervention utilization among pre-antiretroviral therapy, HIV-infected adults in Rakai, Uganda: a randomized trial. AIDS Behav. 2015;19(10):1742–51. https://doiorg.publicaciones.saludcastillayleon.es/10.1007/s10461-015-1159-y.

  43. Liebling EJ, Perez JJS, Litterer MM, Greene C. Implementing hospital-based peer recovery support services for substance use disorder. Am J Drug Alcohol Abuse. 2021;47(2):229–37. https://doiorg.publicaciones.saludcastillayleon.es/10.1080/00952990.2020.1841218. (Epub 2020 Nov 20).

    Article  PubMed  Google Scholar 

  44. White S, Foster R, Marks J, Morshead R, Goldsmith L, Barlow S, et al. The effectiveness of one-to-one peer support in mental health services: a systematic review and meta-analysis. BMC Psychiatr. 2020;20:534. https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12888-020-02923-3.

    Article  Google Scholar 

  45. Chambers DA. Advancing adaptation of evidence-based interventions through implementation science: progress and opportunities. Front Health Serv. 2023;3:1204138.

    Article  PubMed  PubMed Central  Google Scholar 

  46. Lau AS, Huey SJ, Baumann AA. Advances in the adaptation and implementation of evidence-based interventions for historically marginalized groups. Behav Res Ther. 2023;168: 104377.

    Article  PubMed  Google Scholar 

Download references

Acknowledgements

We appreciate the collaboration of Deep Griha, our NGO partner, in piloting this work. We are grateful to the CHWs who are the core of this intervention for bringing their skills, knowledge, and curiosity in addition to telling their stories through interviews. We also thank all of our study participants for sharing their time and insights.

Funding

Funding for this project was through the Weill Cornell Primary Care Innovation Grant and Weill Cornell Medicine Area of Concentration (AOC) program. Funders had no role in the conceptualization, design, data collection, analysis, decision to publish, or preparation in the manuscript.

Author information

Author notes

  1. Kathryn Broderick and Arthi Vaidyanathan contributed equally to this work and should be considered co-first authors.

Authors and Affiliations

  1. Department of Family Medicine, Sydney Kimmel Medical College, Philadelphia, PA, USA

    Kathryn Broderick

  2. School of Medicine, Duke University, Durham, NC, USA

    Arthi Vaidyanathan

  3. School of Medicine, Yale University, New Haven, CT, USA

    Matthew Ponticiello

  4. Center for Global Health, Weill Cornell Medicine, New York, NY, USA

    Misha Hooda, Puja Chebrolu, Jyoti Mathad & Radhika Sundararajan

  5. Deep Griha Society, Pune, Maharashtra, India

    Vaishali Kulkarni & Ashlesha Onawale

  6. Department of Epidemiology, Gillings School of Global Public Health, University of North Carolina, Chapel Hill, NC, USA

    Andrea Chalem

  7. Division of Public Health Sciences, Department of Surgery, Washington University in St Louis, St Louis, MO, USA

    Puja Chebrolu & Jyoti Mathad

  8. Department of Medicine, Weill Cornell Medicine, New York, NY, USA

    Ana Baumann

  9. Department of Emergency Medicine, Weill Cornell Medicine, New York, NY, USA

    Radhika Sundararajan

Authors
  1. Kathryn Broderick
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Arthi Vaidyanathan
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Matthew Ponticiello
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Misha Hooda
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Vaishali Kulkarni
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Andrea Chalem
    View author publications

    You can also search for this author inPubMed Google Scholar

  7. Puja Chebrolu
    View author publications

    You can also search for this author inPubMed Google Scholar

  8. Ashlesha Onawale
    View author publications

    You can also search for this author inPubMed Google Scholar

  9. Ana Baumann
    View author publications

    You can also search for this author inPubMed Google Scholar

  10. Jyoti Mathad
    View author publications

    You can also search for this author inPubMed Google Scholar

  11. Radhika Sundararajan
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

RS and JM conceived of, designed, and secured funding for the parent study. RS, JM, PC, and AO oversaw conduct of the parent study with assistance from AC and MP. Data collection was overseen by VK. Qualitative data analysis was primarily performed by KB, MP, and AC under the supervision of RS. Mechanism maps were drafted by KB and AV and iteratively refined through discussions with AB and RS. KB and AV wrote the first draft of this manuscript. MH assisted with creating figures and manuscript revisions. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Radhika Sundararajan.

Ethics declarations

Ethics approval and consent to participate

This study was approved by the Weill Cornell Medicine Institutional Review Board (Protocol 21–06023615), and Sahara Aalhad, an Institutional Review Board in Pune, India. All participants were provided with written, informed consent. Qualitative interviews were conducted in secure, private locations to maintain participant confidentiality.

Consent for publication

This is not applicable as all data presented here is deidentified.

Competing interests

The authors declare they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Appendix

Appendix

Table 3 Parent study qualitative interview participant characteristics
Full size table
Table 4 Representative quotes illustrating mechanisms of OGTT uptake and clinic presentation
Full size table
Table 5 Representative quotes illustrating mechanisms of inconsistent clinical management of GDM
Full size table
Fig. 5
figure 5

Revised logic model summarizing proposed intervention changes to address inconsistent clinical management of GDM

Full size image

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Broderick, K., Vaidyanathan, A., Ponticiello, M. et al. Generalizing from qualitative data: a case example using critical realist thematic analysis and mechanism mapping to evaluate a community health worker-led screening program in India. Implementation Sci 19, 81 (2024). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13012-024-01407-2

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13012-024-01407-2

Keywords

Implementation Science

ISSN: 1748-5908