What Does “Success” Mean in the Cost-Effectiveness of Fetal Fibronectin Testing? [Response to Letter] Authors: Dina Abushanab, Nader Al-Dewik, Reem AlNasr, Reem Alenany, Lana Kattan, Ghaith Alali, Alaa Al-Naama, Moza Al-Hail, Thomas Farrell, Haseebur Rahman Khan Mohammed, Wessam Elkassem, Abdulrouf Pallivalapila, Binny Thomas, Daoud Al-Badriyeh Published: 2026 Dear editor We thank Dr. Sarıoğlu for his careful and constructive appraisal of our study and for recognizing the importance of generating real-world, Qatar-specific evidence on the clinical and economic implications of fetal fibronectin (fFN) testing. We appreciate the opportunity to clarify the three issues raised regarding the definition and interpretation of the effectiveness endpoint, the relationship between the decision-analytic model and observed clinical practice, and the costing calculations presented in Table 2. These points are addressed individually below.1 Definition and Interpretation of the Effectiveness Endpoint We thank Dr. Sarıoğlu for emphasizing the importance of clearly defining and specifying the direction of the effectiveness endpoint. In our economic model, the effectiveness event was defined as delivery occurring within 48 hours of admission. This definition was applied consistently throughout the underlying analysis and served as the basis for the decision-analytic model. The model included six patient pathways across two comparator arms: fFN testing and no-fFN testing. Within the fFN arm, pathways were further categorized by whether the fFN result was positive or negative, and whether delivery took place within 48 hours. Similarly, Table 3 reported delivery within 48 hours as the primary clinical endpoint. This occurred in 94% of women in the fFN group and 80% in the no-fFN group, whereas delivery beyond 48 hours was reported as a secondary outcome. The probabilities and costs incorporated in the economic model were derived using this endpoint definition. We acknowledge, however, that the wording used to describe this outcome was not consistent across the manuscript. Specifically, the terms “successful delivery within 48 hours” and “successful prolongation of delivery within 48 hours” were used in the Methods, while the Discussion referred to “successful prolongation of pregnancy beyond 48 hours.” We agree that these phrases are not equivalent and could create confusion about the intended direction of the endpoint. To clarify, the endpoint analyzed in the study was delivery within 48 hours, rather than prolongation of pregnancy beyond 48 hours. The references to “prolongation” in the narrative did not reflect a different definition or coding of the endpoint used in the analysis. We appreciate the opportunity to clarify this distinction and the intended interpretation of the reported outcome. We also agree that delivery within 48 hours should not be interpreted as a therapeutic effect attributable to fFN testing itself. fFN is a diagnostic and risk-stratification tool, not an intervention designed to accelerate or delay delivery. Its main clinical value is to inform the short-term risk of preterm delivery and to guide subsequent management decisions. In particular, the high negative predictive value of fFN may help identify women at low risk of imminent delivery, potentially reducing unnecessary interventions such as hospitalization, transfer, corticosteroid use, or tocolytic treatment. Therefore, the higher observed rate of delivery within 48 hours in the fFN group should be understood as an observed short-term outcome associated with the fFN-tested clinical pathway, rather than as a therapeutic effect attributable to the test. This is especially relevant given our retrospective observational design. As noted in the original article, the fFN and no-fFN groups differed in several baseline characteristics, and the decision to perform fFN testing was based on clinical judgment rather than random assignment. As a result, selection and indication bias cannot be ruled out, and the between-group differences should not be interpreted causally. The 48-hour timeframe was chosen because it represents a clinically meaningful short-term period in the management of threatened preterm labor and captures the key decisions and resource use relevant to our hospital-perspective economic evaluation. Accordingly, the study aimed to assess short-term clinical and economic outcomes associated with fFN testing in routine clinical practice, rather than any longer-term therapeutic effect on pregnancy duration. We thank Dr. Sarıoğlu for highlighting this distinction, which offers an important opportunity to clarify both the endpoint definition and its appropriate clinical interpretation. Relationship Between the Modeled Pathways and Observed Clinical Practice We also appreciate Dr. Sarıoğlu’s comments about the link between the decision-tree framework and the treatment frequencies observed in our retrospective cohort. This economic analysis was designed as a pragmatic assessment using real-world clinical and resource-utilization data from the Women’s Wellness and Research Center at Hamad Medical Corporation, Qatar. Clinical inputs were collected retrospectively from the Cerner electronic medical record system for the study period, while healthcare resource use and direct medical costs were obtained from the local hospital environment using a micro-costing approach. The decision tree was therefore structured using real-world clinical and resource-utilization data, while necessarily simplifying the observed management pathways for the purposes of the economic evaluation. It was not intended to represent an idealized treatment protocol or to imply deterministic treatment allocation at the individual-patient level. A key point concerns the level at which probabilities and treatment frequencies were reported. As outlined in the Methods, within the fFN arm the modeled probabilities were conditional on the fFN result (positive versus negative). By contrast, the reported treatment frequencies for atosiban use (12.1%) and corticosteroid administration (45.7%) were aggregate measures across the entire cohort in which fFN testing was performed, and were not presented separately by positive and negative fFN results. Accordingly, these overall treatment frequencies cannot be directly compared with specific branches of the decision tree. As Dr. Sarıoğlu correctly notes, the aggregate treatment frequencies by themselves do not demonstrate that the decision model was misspecified. That said, we recognize that the Methods wording indicating that women with positive fFN “received one cycle of atosiban and antenatal corticosteroids,” while those with negative fFN received neither, could be interpreted as implying deterministic treatment allocation based solely on the fFN result. This was not the intended interpretation of the modeled pathways. In routine practice, clinical management is not determined exclusively by the fFN result. Decisions regarding hospitalization, corticosteroids, tocolysis, and other interventions are guided by the overall clinical context, including gestational age, cervical findings, membrane status, symptoms, and clinician judgment. The decision tree was intended to summarize the principal test-associated clinical and management pathways relevant to the economic evaluation, without suggesting that every woman with a positive fFN result necessarily received identical treatment or that every woman with a negative result received no intervention. This clarification is particularly important because, as Dr. Sarıoğlu emphasizes, the economic impact of a diagnostic test depends on how it influences subsequent clinical management and resource utilization. We therefore agree that a more explicit description of how the modeled pathways correspond to the observed management patterns would enhance the transparency of the analysis. Accordingly, our results should be interpreted as reflecting the clinical and economic outcomes associated with fFN testing in the real-world study setting, based on a decision model that necessarily simplifies the complexity of individual clinical management, and within the limitations typical of retrospective observational designs. They should not be taken as evidence of a rigid protocol in which the fFN result alone determines subsequent treatment. Clarification of the Costing Calculations in Table 2 We thank Dr. Sarıoğlu for his careful review of the costing information in Table 2 and for highlighting the “negative fFN, no delivery within 48 hours” pathway. After re-examining the final source economic model, we identified a discrepancy between the values reported for this pathway in Table 2 and the values used in the underlying analysis. In the final model, the branch cost for “negative fFN, no delivery within 48 hours” was QAR 88,384, with a pathway probability of approximately 0.0233, which produced a probability-weighted cost of approximately QAR 2,056. Any small difference that may occur when multiplying the displayed probability by the branch cost reflects the use of the full probability values in the underlying model. As a result, the QAR 75,980 figure shown in Table 2 as the branch cost and the QAR 172.68 shown as the proportional cost do not match the values applied in the final source economic model. We appreciate Dr. Sarıoğlu’s comment, which led us to revisit the underlying model and detect this reporting issue. Importantly, this discrepancy is restricted to the presentation of this pathway in Table 2 and does not alter the underlying economic analysis. Using the correct branch values, the final source model continues to show a net saving of QAR 6,470.12 per person, consistent with the QAR 6,470 per-person saving reported in the published article. Therefore, the study’s overall economic findings and conclusions remain unchanged. We agree that transparent and reproducible reporting of model inputs and probability-weighted costs is essential, and we welcome the opportunity to clarify the values used in the underlying analysis. Conclusion We thank Dr. Sarıoğlu again for his thoughtful and constructive comments. His remarks offer a valuable opportunity to clarify key elements of both the presentation and interpretation of our analysis. In brief, the effectiveness endpoint in our economic model was defined as delivery within 48 hours, although the wording used to describe this endpoint was inconsistently presented in some parts of the manuscript. The decision tree was built using real-world clinical and resource-utilization data, while necessarily simplifying the complexity of individual clinical decision-making; it should not be interpreted as a deterministic treatment protocol based solely on the fFN result. Finally, our review of the final source model identified a reporting discrepancy in one pathway in Table 2, but the correct values were used in the underlying economic analysis. The reported net saving (approximately QAR 6,470 per person) and the study’s overall conclusions therefore remain unchanged. We appreciate the opportunity for this scientific discussion and thank Dr. Sarıoğlu for helping to promote a clearer and more transparent interpretation of our work.