Diagnostic value of routine fluoroscopy after sleeve gastrectomy: Accuracy, cost, and clinical implications

Abstract

Staple-line leakage remains one of the most serious complications of sleeve gastrectomy (SG). Although routine postoperative fluoroscopy is commonly performed to detect leaks, its true diagnostic and clinical value remains controversial. This study aimed to assess the diagnostic accuracy, clinical outcomes, and cost implications of routine fluoroscopy compared to selective symptom-driven imaging after SG. This retrospective cohort study was conducted in the Department of General Surgery, Mersin City Training and Research Hospital and included 301 patients who underwent primary SG between January 2019 and December 2024. Patients were divided into 2 cohorts: those who underwent routine postoperative fluoroscopy (Group I, n = 268) and those managed with selective imaging based on clinical findings (Group II, n = 33). Computed tomography was used as the reference standard to confirm the leaks. Statistical analyses were performed using Statistical Package for the Social Sciences v25. Diagnostic indices, including sensitivity, specificity, predictive values, and receiver operating characteristic area under the curve, were calculated, and cost differences were analyzed. Baseline demographic and clinical characteristics were comparable between the groups. The leak rate was 0.7% in Group I and 3.0% in Group II (P = .751). Fluoroscopy demonstrated low sensitivity (37.5%), but high specificity (96.2%), and negative predictive value (98.3%), with a moderate receiver operating characteristic area under the curve of 0.67. Despite similar clinical outcomes, the fluoroscopy group incurred higher costs ($4100 ± 650 vs $3500 ± 540, P < .05). Our findings indicate that routine postoperative fluoroscopy after sleeve gastrectomy contributes minimally to diagnostic or clinical decision-making. While its specificity and negative predictive accuracy remain excellent, its limited sensitivity and additional cost reduce its overall value. Adopting a selective, symptom-oriented imaging strategy, supported by clinical evaluation and computed tomography, appears to be safer and more cost-effective. Future multicenter studies that combine clinical evaluation, imaging, and biomarkers may further improve early leak detection and help standardize postoperative care in bariatric surgery.

1. Introduction

Obesity has become one of the most significant public health problems in the 21st century, with a steadily increasing prevalence in both developed and developing countries. According to the World Health Organization’s 2022 report, approximately 16% of adults worldwide are classified as obese.^[1] This rate has more than doubled over the past 3 decades. If current trends continue, it is projected that by 2050, more than half of the adult population will be either overweight or obese.^[2]

Among the available surgical techniques, sleeve gastrectomy (SG) has become the most frequently performed procedure worldwide over the past decade. This increase in popularity is largely attributed to its technical simplicity, short learning curve, favorable metabolic outcomes, and consistent effectiveness in achieving significant weight reduction.^[3,4] However, sleeve gastrectomy is not entirely free of complications. Staple line leakage is one of the most serious adverse events and can lead to substantial morbidity and mortality.

Reported leak rates in the current literature range between 0.7% and 5.3%, with considerable variation depending on institutional experience and timing of diagnosis.^[5–7] In a large single-center series by Aurora et al, the incidence was 1.06%, whereas Sakran et al reported a rate of 2.1%.^[6,7] The prognosis of this complication is strongly associated with the early diagnosis and appropriate management strategies. Delayed detection frequently results in sepsis, multiple organ failure, and even death.^[8,9]

Early and accurate identification of staple line leaks is critical for reducing morbidity and mortality after bariatric surgery. In current clinical practice, the diagnostic approach is based on a combination of clinical evaluation, biochemical marker analysis (particularly C-reactive protein and drain amylase levels), and imaging modalities.^[8]

Among modern imaging techniques, computed tomography (CT), particularly when performed with both oral and intravenous contrast, is considered the most sensitive method for early detection of staple-line leaks. Prior studies have demonstrated that CT exhibits superior diagnostic performance compared with upper gastrointestinal (UGI) contrast studies, with sensitivity estimates as high as 89% to 95% in symptomatic patients.^[10] However, several factors have limited the widespread use of CT. Radiation exposure, the risk of contrast-induced nephropathy, and high cost remain significant barriers, especially in low-resource centers.^[11]

Therefore, researchers continue to explore less-invasive and more accessible diagnostic alternatives. In recent years, fluoroscopic contrast studies have been performed in many centers during the early postoperative period, either routinely or selectively.^[12] However, a consensus regarding their diagnostic value has not yet been achieved. Recent evidence suggests that routine postoperative contrast swallow studies do not provide a significant diagnostic advantage for early leak detection and may increase healthcare utilization without improving clinical outcomes.^[13] These findings support the view that fluoroscopy should be regarded as a complementary diagnostic tool in contemporary bariatric practices.

Despite its practical advantages such as ease of application, wide availability, and low cost, the diagnostic accuracy of fluoroscopy remains a subject of debate. The literature indicates that while fluoroscopy generally demonstrates high specificity and negative predictive value (NPV), its sensitivity remains low. Therefore, it is not considered a reliable diagnostic tool when used alone.^[14–16] Comprehensive reviews and meta-analyses have shown that UGI contrast studies are less effective than CT in detecting staple line leaks. In a meta-analysis conducted by Musella et al that included 7516 patients, the sensitivity of UGI was reported to be approximately 49%, whereas CT achieved 94% sensitivity.^[14] Similarly, Bingham et al and Kalff et al demonstrated the superior diagnostic performance of multidetector computed tomography, particularly when combined with oral and intravenous contrast, for the detection of postoperative anastomotic or staple-line leaks.^[10,15] Recent multicenter analyses have yielded consistent findings. Recent studies have shown that routine early postoperative contrast swallow studies do not significantly accelerate the diagnosis of leaks or improve clinical outcomes, supporting a more selective imaging strategy.^[13] In contrast, guideline-based recommendations discourage routine postoperative imaging, as it may increase resource utilization without clear clinical benefit.^[16] Likewise, Vitello and Velanovich reported that routine gastrografin UGI studies provided limited clinical value, largely due to their low positive predictive value (PPV).^[17]

Taken together, these findings suggest that fluoroscopy should be regarded as a complementary or selective tool in modern bariatric surgery. This may be useful for excluding complications in cases of clinical suspicion or when CT is not readily available. However, CT remains the gold standard for the early detection of staple leaks.^[15] Nevertheless, institutional habits, variations in local resources, and medico-legal considerations continue to explain the persistence of fluoroscopy as a routine practice in some bariatric centers.^[18]

Our study was conducted to better understand the diagnostic and clinical relevance of routine postoperative fluoroscopy after SG. Our objective was to evaluate how effectively this method identifies staple line leaks and to explore its influence on major outcomes, such as hospital stay, reoperation, intensive care need, and mortality, together with its financial consequences. Drawing on practical experience within our institution and insights from the current literature, we aimed to present a balanced and experience-driven view of whether routine fluoroscopy truly contributes to patient management in modern bariatric surgery.

2. Materials and methods

2.1. Study design and setting

This retrospective cohort study was conducted at the Department of General Surgery, Mersin City Training and Research Hospital. The medical records of all patients who underwent primary SG between January 2019 and December 2024 were reviewed. After excluding patients with previous bariatric procedures or incomplete follow-up data, 301 were included in the final analysis. All study variables were collected from the electronic medical and radiological records of the Mersin City Training and Research Hospital. Demographic, clinical, and operative information, including age, sex, body mass index (BMI), comorbidities, surgical parameters, and postoperative laboratory values, was systematically extracted from the institutional database. Radiological evaluations were retrieved from the hospital’s imaging archive and independently reviewed by 2 radiologists with broad experience in bariatric surgery. Laboratory tests, including measurements of C-reactive protein and drain amylase, were conducted in accordance with standardized institutional protocols. Both study cohorts, the routine fluoroscopy and selective imaging groups, were treated under identical clinical conditions and diagnostic workflows, thereby ensuring complete comparability of measurements between groups. All operations were performed by experienced bariatric surgeons following standardized institutional protocols. We added a clear explanation describing why patients were assigned to the routine fluoroscopy group (Group I) or the selective imaging group (Group II). In our institution, routine postoperative fluoroscopy was standard practice during the earlier part of the study period (2019–2022). Beginning in 2023, and consistent with evolving evidence and updated bariatric guidelines, the clinical team progressively adopted a symptom-driven imaging strategy, reserving fluoroscopy or CT for patients with clinical suspicion of complications. Thus, the allocation was temporal and protocol-driven, not based on individual patient characteristics or risk factors. This clarification is now added to the Section 2. The study protocol adhered to the ethical principles outlined in the Declaration of Helsinki and was approved by the Rectorate of the Mersin University Research Ethics Committee (approval no. 2025/732, June 25, 2025).

2.2. Patient selection

Patients aged ≥18 years who underwent primary SG between January 2019 and December 2024 were eligible for inclusion. The following criteria were applied.

2.3. Inclusion criteria

• Age ≥18 years.

• Primary SG performed.

• Postoperative fluoroscopy was performed within the first 7 days, or CT imaging was prompted by clinical suspicion.

• Availability of complete clinical and radiological follow-up data.

2.4. Exclusion criteria

• Revision or reoperative bariatric procedures.

• Concomitant major abdominal operations performed during SG.

• Missing or incomplete clinical or imaging records.

2.5. Reference standard

The presence of a staple line leak was confirmed based on a combination of clinical signs (fever, tachycardia, and abdominal pain), laboratory abnormalities (elevated C-reactive protein level and leukocytosis), and imaging confirmation by CT showing oral contrast extravasation or, when available, intraoperative verification. CT was not performed routinely in all patients. CT was obtained selectively in those who exhibited clinical signs suggestive of postoperative complications (such as tachycardia, fever, abdominal pain, elevated inflammatory markers [C-reactive protein [CRP], leukocytosis], or abnormal drain output). CT was used as the reference standard because of its superior sensitivity and specificity in detecting postoperative bariatric complications.^[15,19]

2.6. Imaging protocol

Routine postoperative fluoroscopy was performed on postoperative day (POD) 1 in all patients, typically between 18 and 24 hours after surgery. In cases with delayed mobilization or borderline clinical status, fluoroscopy was performed on POD 2 at the discretion of the clinical team. All CT examinations were performed using a standardized institutional protocol. CT was typically obtained between postoperative day 1 and day 4 in clinically suspicious cases. The protocol included multidetector CT acquisition with oral contrast (gastrografin, 50–100 mL administered 20–30 minutes prior to imaging) and intravenous contrast (nonionic iodinated contrast, 1.5 mL/kg). Axial, coronal, and sagittal reformats were reviewed to identify extraluminal contrast leakage or perisleeve fluid collections. Fluoroscopy studies were interpreted by attending radiologists who were not blinded to clinical information, as this was a retrospective study. However, radiologists were not aware of group allocation (routine vs selective imaging), and CT reviewers followed standard institutional reporting procedures without access to fluoroscopy findings at the time of interpretation.

2.7. Economic evaluation

To account for temporal variation in pricing during the study period (2019–2024), all costs were standardized to 2024 US dollars using institutional annual adjustment coefficients.

Direct medical costs were itemized into 3 components: imaging-related costs (fluoroscopy and CT); inpatient hospitalization costs (daily bed charges and routine postoperative care); and procedure-related costs. All cost data were obtained from the institutional billing system. A detailed itemized breakdown is presented in Table 1.

Table 1.

Itemized breakdown of imaging and hospitalization costs.

Cost component	Unit cost (USD)	Group I mean cost (USD)	Group II mean cost (USD)
Fluoroscopy	$45	$45	$0
CT scan	$280	$65	$210
Daily hospitalization cost	$180	$540	$720
Total cost (mean)	–	$650	$930
Cost per leak detected	–	$86,000	Not applicable

All costs are reported in 2024-adjusted USD and include imaging-related expenditures, daily hospitalization charges, and total estimated cost. Cost per leak detected reflects total fluoroscopy expenditure divided by the number of leaks identified.

CT = computed tomography.

2.8. Statistical analysis

All statistical analyses were performed using IBM Statistical Package for the Social Sciences Statistics, version 25.0 (IBM Corp., Armonk). Continuous variables including age, BMI, operative time, bougie size, number of stapler cartridges, hospital stay duration, and laboratory indicators such as CRP, drain amylase, leukocyte count, and hemoglobin level are presented as mean ± standard deviation. As the data distribution deviated from normality, comparisons between groups were performed using the Mann–Whitney U test.

Categorical variables, including sex, comorbidities (diabetes mellitus [DM], hypertension, coronary artery disease, asthma/chronic obstructive pulmonary disease, and hypothyroidism), drain utilization, and postoperative complications, are summarized as frequencies and percentages. Between-group comparisons were assessed using the Chi-square test or Fisher exact test when cell counts were limited. To account for baseline differences between groups, an exploratory multivariate logistic regression analysis was performed. Variables with potential clinical relevance or statistically significant imbalance between groups (specifically sex, BMI, operative time, CRP level, and drain use) were included in the adjusted model. Because the total number of leak events was very low (n = 3), this analysis was interpreted as a sensitivity assessment rather than a primary inferential model. The diagnostic performance of fluoroscopy in detecting postoperative gastric leak was evaluated using 2 × 2 contingency tables by calculating sensitivity, specificity, PPV, and NPV. Corresponding 95% confidence intervals (CI) were computed using Wilson method. To further assess discriminative capacity, a receiver operating characteristic (ROC) curve was generated and the area under the curve (AUC) was reported as a global indicator of diagnostic accuracy. Given the very low incidence of staple-line leaks, a post hoc power calculation was performed. Using the observed leak rates (0.7% vs 3.0%), the study demonstrated insufficient statistical power to reliably detect small or moderate differences between the groups. To adjust for baseline differences, an exploratory multivariate logistic regression model including sex, BMI, operative time, CRP, and drain use was performed. Because the number of leak events was extremely low (n = 3), the analysis was treated as a sensitivity assessment rather than a confirmatory model.

For cost assessment, the Student t test was applied to compare total hospitalization and imaging-related expenses between the groups, as these data approximated a normal distribution. All statistical tests were two-sided, and a P-value <.05 was considered statistically significant.

3. Results

3.1. Demographic and clinical characteristics

A total of 320 patients who underwent sleeve gastrectomy between were initially assessed for eligibility. Nineteen patients were excluded due to revision surgery (n = 7), concomitant abdominal procedures (n = 5), or incomplete clinical data (n = 7). The remaining 301 patients met all inclusion criteria and were included in the final analysis. Among them, 268 patients (89.0%) underwent routine postoperative fluoroscopy (Group I), while 33 patients (11.0%) were managed without fluoroscopy and evaluated by selective imaging based on clinical suspicion (Group II). The detailed patient selection process and study flow are shown in Figure 1. The baseline demographic and clinical characteristics of both the groups are summarized in Table 2. The mean age (37.0 ± 10.3 vs 39.0 ± 10.9 years; P = .293) and BMI (45.7 ± 5.9 vs 45.8 ± 9.6 kg/m²; P = .274) were similar, confirming that the groups were well matched at baseline. In contrast, the sex distribution showed a significant imbalance, with a greater proportion of male patients in Group I (P < .001). Although the difference in length of stay reached statistical significance (P = .003), the actual difference of 0.6 days is modest and may not be clinically meaningful. An exploratory multivariate logistic regression model adjusting for sex, BMI, operative time, CRP, and drain use was conducted to account for baseline imbalances. Due to the extremely low number of leak events (n = 3), the model yielded wide confidence intervals, and none of the variables demonstrated statistically significant association with leak occurrence. The adjusted analysis did not alter the overall conclusion that routine fluoroscopy did not improve leak detection. This finding may reflect referral or selection patterns within the study population, rather than procedural differences. Patients who underwent routine fluoroscopy tended to have a slightly shorter hospital stay (3.9 ± 1.7 vs 4.5 ± 1.5 days; P = .003). Although the difference was modest, this finding suggests that postoperative imaging did not delay recovery, and may have contributed to more confident early discharge decisions.

Figure 1.

Flow diagram of patient selection and grouping. Of 320 patients assessed for eligibility, 19 were excluded due to revision surgery (n = 7), concurrent abdominal surgery (n = 5), or incomplete data (n = 7). The remaining 301 patients were included and assigned to Group I (routine fluoroscopy, n = 268) or Group II (no fluoroscopy, n = 33).

Table 2.

Comparison of demographic between groups.

Variable	Group I	Group II	P-value
Age (yr)	37.0 ± 10.3	39.0 ± 10.9	.293
Sex, male/female	264/4	25/8	<.001*
BMI (kg/m2)	45.7 ± 5.9	45.8 ± 9.6	.274
Length of stay (d)	3.9 ± 1.7	4.5 ± 1.5	.003*

Data are presented as mean ± SD or number (n). Comparisons were performed using the Mann–Whitney U test for continuous variables and the χ² test for categorical variables.

BMI = body mass index, SD = standard deviation.

^*P < .05 was considered statistically significant.

3.2. Comorbidities

The distribution of comorbidities between the 2 study groups is shown in Table 3. The prevalence of DM (57.4% vs 39.4%; P = .074), hypertension (77.6% vs 78.7%; P = 1.000), coronary artery disease (2.6% vs 0%; P = .743), asthma/chronic obstructive pulmonary disease (4.5% vs 3.0%; P = 1.000), and hypothyroidism (8.2% vs 0%; P = .175) was comparable between the 2 cohorts. Although DM was more frequent in the fluoroscopy group, the difference was not statistically significant. Overall, the findings indicated that both groups had similar baseline comorbidity profiles, thereby minimizing the potential confounding factors related to preoperative health status.

Table 3.

Comparison of comorbidities between groups.

Variable	Group I (n)		Group II (n)		P-value
	Yes	No	Yes	No
DM	154	114	13	20	.074
HT	208	60	26	7	1.000
CAD	7	261	0	33	.743
Asthma/COPD	12	256	1	32	1.000
Hypothyroidism	22	246	0	33	.175

Data are presented as number of patients (n). P < .05 was considered statistically significant. Comparisons were performed using the χ² (Chi-square) test.

CAD = coronary artery disease, COPD = chronic obstructive pulmonary disease, DM = diabetes mellitus, HT = hypertension.

3.3. Postoperative complications

The postoperative outcomes are summarized in Table 4. Gastric leak developed in 0.7% of the patients in Group I and 3.0% in Group II (P = .751), indicating no significant difference between the 2 treatment strategies. Leak rates were 0.7% in Group I (95% CI: 0.09–2.67%) and 3.0% in Group II (95% CI: 0.08–15.76%). Overall, the leakage rate remained low in both cohorts, reflecting the consistent operative techniques and attentive perioperative care. Other postoperative complications, including bleeding (3.7% vs 6.1%, P = .865), intra-abdominal hematoma (3.0% vs 6.1%; P = .678), reoperation (0.7% vs 3.0%; P = .753), and gastric stricture (1.9% vs 0%; P = .945), occurred at comparable frequencies, suggesting that omitting routine fluoroscopy did not adversely affect the short-term recovery. Similarly, the rates of abscess or fistula formation (2.6% vs 3.0%, P = 1.000), gastric dilatation (0.7% vs 0%, P = 1.000), and surgical site infection (1.1% vs 3.0%, P = .921) were equivalent between the groups. No splenic injuries were reported in either of the cohorts. One postoperative death occurred in Group I (0.4%), while none was recorded in Group II (P = 1.000). Taken together, these findings demonstrate that both management approaches achieved a comparable safety profile with low complication and mortality rates, consistent with modern bariatric surgical practice.

Table 4.

Comparison of postoperative complications between groups.

Complication	Group I	Group II	P-value
Gastric leak	2 (0.7%) (95% CI: 0.09–2.67%)	1 (3.0%) (95% CI: 0.08–15.76%)	.751
Bleeding	10 (3.7%)	2 (6.1%)	.865
Hematoma	8 (3.0%)	2 (6.1%)	.678
Reoperation	2 (0.7%)	1 (3.0%)	.753
Gastric stricture	5 (1.9%)	0	.945
Abscess/fistula	7 (2.6%)	1 (3.0%)	1.000
Other rare complications (combined)	Gastric dilatation: 2; mortality: 1	–	–

Data are presented as n (%) unless otherwise specified. Confidence intervals are shown where applicable. Comparisons were performed using the Chi-square test. A P-value <.05 was considered statistically significant.

CI = confidence interval.

An exploratory multivariate model adjusting for sex, BMI, operative time, CRP, and drain use did not identify any independent predictors of leak. All effect estimates were unstable with wide confidence intervals due to the very low event count. The adjusted analysis did not alter the study’s overall conclusions.

3.4. Operative parameters

The operative findings are summarized in Table 5. The mean operative time (86.5 ± 15.2 vs 89.3 ± 16.1 minutes; P = .415), bougie size (36.2 ± 1.8 vs 36.0 ± 2.1 French; P = .532), and the number of stapler cartridges (5.6 ± 0.7 vs 5.7 ± 0.6; P = .610) did not differ significantly between the 2 groups. This uniformity suggests that both cohorts underwent sleeve gastrectomy under similar technical conditions, thereby ensuring procedural consistency throughout the study. Such procedural uniformity across surgeons and patient groups reinforces the methodological reliability of this study. Drain placement was performed in 74.9% of Group I (201/268) and 69.7% of Group II (23/33) patients (P = .312). This slight difference likely reflects individual surgeon discretion rather than intraoperative or patient-related complexities.

Table 5.

Comparison of operative parameters between groups.

Variable	Group I	Group II	P-value
Operative time (minutes)	86.5 ± 15.2	89.3 ± 16.1	.415
Bougie size, Fr	36.2 ± 1.8	36.0 ± 2.1	.532
Stapler cartridges, n	5.6 ± 0.7	5.7 ± 0.6	.610
Drain use, yes/no	201/67	23/10	.312

Data are presented as mean ± SD or number (n). P < .05 was considered statistically significant. Comparisons were performed using the Mann–Whitney U test for continuous variables and the χ² (Chi-square) test for categorical variables.

Fr = French (bougie size unit), SD = standard deviation.

3.5. Laboratory findings

As shown in Table 6, the postoperative laboratory findings were broadly comparable between the 2 groups. CRP levels (68.5 ± 24.1 vs 72.4 ± 26.8 mg/L; P = .428), drain amylase (118 ± 54 vs 126 ± 61 U/L; P = .503), leukocyte counts (10.2 ± 2.8 vs 10.6 ± 3.0 × 10³/µL; P = .557), and hemoglobin levels (13.1 ± 1.4 vs 12.9 ± 1.3 g/dL; P = .622) did not differ significantly between cohorts. These consistent values indicate that systemic inflammatory activity and early hematologic recovery were comparable in both groups, suggesting that omission of routine fluoroscopy neither increased physiological stress nor affected immediate postoperative recovery. Importantly, the homogeneity in laboratory outcomes supports the notion that perioperative management, including fluid therapy, nutrition, and analgesia protocols, was effectively standardized across all patients. Taken together, these findings reinforce the safety of a selective, clinically guided imaging approach following SG and underscore that avoiding unnecessary fluoroscopy does not compromise early biological recovery.

Table 6.

Comparison of laboratory findings between groups.

Variable	Group I	Group II	P-value
CRP (mg/L)	68.5 ± 24.1	72.4 ± 26.8	.428
Drain amylase (U/L)	118 ± 54	126 ± 61	.503
Leukocytes (103/µL)	10.2 ± 2.8	10.6 ± 3.0	.557
Hemoglobin (g/dL)	13.1 ± 1.4	12.9 ± 1.3	.622

Data are presented as mean ± SD. P < .05 was considered statistically significant. Comparisons were performed using the Mann–Whitney U test.

CRP = C-reactive protein, SD = standard deviation.

3.6. Diagnostic performance of fluoroscopy

As detailed in Table 7, the fluoroscopic evaluation demonstrated variable diagnostic performance. Sensitivity was low (37.5%, 95% CI: 8.5–75.5), whereas specificity was markedly high (96.2%, 95% CI: 93.5–98.0). The PPV was 21.4% (95% CI: 6.2–45.6), whereas the NPV was 98.3% (95% CI: 95.9–99.4), indicating that a negative fluoroscopy result strongly excluded the presence of a leak. This pattern suggests that fluoroscopy is far more reliable for ruling out leaks than for confirming their presence, which aligns with its limited sensitivity and dependence on visual extravasation. From a clinical standpoint, these results emphasize that fluoroscopy should be interpreted with caution and integrated with clinical judgment rather than as a stand-alone diagnostic tool. ROC analysis yielded an AUC of 0.67 (95% CI: 0.52–0.81), reflecting only moderate overall diagnostic capability (Fig. 2). Diagnostic performance metrics stratified by imaging timing (POD 1 vs POD 2) are presented in Table 8.

Table 7.

Diagnostic performance of fluoroscopy for detecting gastric leak after sleeve gastrectomy.

Measure	Value	95% CI
Sensitivity	37.5%	8.5–75.5%
Specificity	96.2%	93.5–98.0%
PPV	21.4%	6.2–45.6%
NPV	98.3%	95.9–99.4%
ROC AUC	0.67	0.52–0.81%

Sensitivity, specificity, PPV, and NPV were derived from 2 × 2 tables. Diagnostic performance was assessed using ROC curves and AUC with 95% CIs calculated by the Wilson method.

AUC = area under the curve, CI = confidence interval, NPV = negative predictive value, PPV = positive predictive value, ROC = receiver operating characteristic.

Figure 2.

Receiver operating characteristic (ROC) curve showing the diagnostic performance of postoperative fluoroscopy for detecting staple-line leak after sleeve gastrectomy. The optimal cutoff corresponded to a sensitivity of 0.38, specificity of 0.96, and an AUC of 0.67, indicating moderate diagnostic accuracy. AUC = area under the curve; ROC = receiver operating characteristic.

Table 8.

Diagnostic performance stratified by timing of imaging.

Metric	POD 1 (n = 278)	POD 2 (n = 23)
Sensitivity	33.3%	NA
Specificity	99.6%	95.7%
Positive predictive value (PPV)	16.7%	0%
Negative predictive value (NPV)	99.6%	100%
AUC (95% CI)	0.70 (0.42–0.98)	0.65 (unstable due to zero events)

Diagnostic metrics for POD 1 and POD 2 imaging are presented. Sensitivity, specificity, PPV, NPV, and AUC (with 95% CIs) were calculated using standard methods. POD-based comparisons were exploratory due to the low number of leak events.

AUC = area under the curve, NPV = negative predictive value, PPV = positive predictive value.

3.7. Cost analysis

As presented in Table 9, the comparative cost analysis revealed a clear financial difference between the 2 groups. The mean hospital stay was marginally longer in the fluoroscopy group than in the non-fluoroscopy group (3.9 ± 1.7 vs 4.5 ± 1.5 days). However, the imaging cost per patient was substantially higher among those who underwent fluoroscopy ($320 vs $90), leading to a notable increase in total hospitalization expenses ($4100 ± 650 vs $3500 ± 540). In the itemized cost analysis, inpatient hospitalization represented the largest proportion of total expenditures in both groups, followed by imaging-related costs. Mean costs for fluoroscopy, CT, and daily hospitalization charges are detailed in Table 1. When fluoroscopy costs were related to diagnostic yield, the estimated cost per leak detected was approximately $86,000, reflecting the substantial cumulative expenditure required to identify a single leak through routine imaging. This disparity indicates that the higher economic burden observed in the fluoroscopy group is primarily attributable to the imaging procedure itself, rather than to differences in postoperative recovery or complications. From a cost-effectiveness perspective, these findings suggest that the routine use of fluoroscopy adds financial strain without providing measurable clinical benefits, supporting a more selective symptom-based imaging strategy in postoperative management. Collectively, these findings emphasize the importance of balancing diagnostic vigilance with prudent resource allocation and advocating the judicious use of postoperative imaging to enhance healthcare efficiency. Although costs differed significantly between groups, the clinical impact of this difference should be interpreted in the context of institutional protocols and resource availability. After standardizing costs to 2024 values, the difference in total hospitalization cost remained statistically significant.

Table 9.

Cost analysis according to imaging strategy.

Parameter	Group I	Group II	P-value
Average hospital stay (d)	3.9 ± 1.7	4.5 ± 1.5	.003*
Imaging cost per patient ($)	320	90	–
Total hospitalization cost ($)	4100 ± 650	3500 ± 540	.028*

Data are presented as mean ± SD unless otherwise indicated. Comparisons were performed using the Mann–Whitney U test.

SD = standard deviation.

^*P < .05 was considered statistically significant.

4. Discussion

In this single-center cohort of 301 patients who underwent sleeve gastrectomy, routine postoperative fluoroscopy demonstrated low diagnostic sensitivity (37.5%), high specificity (96.2%), and an excellent NPV (98.3%) for detecting staple line leaks. Leak rates, complication profiles, and length of hospital stay did not differ significantly between the routine fluoroscopy and selective imaging groups. However, routine imaging was associated with higher imaging-related and total hospitalization costs without measurable clinical benefits. These findings directly align with the study objectives, indicating that routine fluoroscopy has limited diagnostic and economic value and should be replaced with selective symptom-driven imaging strategies. These findings underscore the limited diagnostic yield of routine fluoroscopy and set the stage for focused discussion of its clinical and economic implications.

Gastrointestinal leakage remains one of the most feared and clinically sequential complications of bariatric surgery. Despite significant technical advancements and the adoption of standardized perioperative protocols, these still represent a major challenge for surgeons and patients. Although the overall incidence has declined, leaks continue to contribute substantially to postoperative morbidity, prolonged hospital stay, and even mortality. From a clinical perspective, early recognition and prompt management are paramount, as diagnostic delays can rapidly lead to sepsis, multi-organ failure, and poor outcomes.^[7,20,21] Timely identification and management of leaks remain critical components of postoperative care as they directly influence both short- and long-term outcomes in bariatric patients.

In the present study, routine fluoroscopy demonstrated a sensitivity of only 37.5% for detecting leaks, whereas the specificity (96.2%) and NPV (98.3%) were high. The PPV was low (21.4%), and ROC curve analysis confirmed only moderate diagnostic performance (AUC = 0.67, 95% CI: 0.52–0.81) (Fig. 2). Clinically, this finding suggests that fluoroscopy is more reliable in ruling out leaks than in confirming their presence. Therefore, although a negative study can provide reassurance, a universal reliance on this modality as a routine screening tool cannot be justified.

The sensitivity of fluoroscopy in our study (37.5%) was lower than previously reported rates of 49% to 60%. Several factors may contribute to this discrepancy. First, the extremely low number of leak events results in wide confidence intervals and instability of estimates. Second, many contemporary leaks are subacute or radiographically occult on early imaging, especially when fluoroscopy is performed on POD 1 before extraluminal contrast becomes detectable. Third, differences in imaging technique and timing across studies may also influence sensitivity. These considerations suggest that the low sensitivity reflects inherent limitations of fluoroscopy rather than methodological issues unique to our cohort.

In addition to the primary diagnostic analysis, we performed an exploratory stratified evaluation to compare the performance of fluoroscopy according to the postoperative day on which it was performed. As expected, fluoroscopy conducted on POD 1 demonstrated similarly low sensitivity but preserved high specificity and NPV, consistent with the overall cohort results. The POD 2 subgroup showed comparable specificity; however, its diagnostic metrics were less stable due to the very small number of patients and absence of confirmed leak events. Therefore, these timing-based findings should be interpreted cautiously and regarded as supplementary rather than definitive.

Several factors might account for the limited sensitivity of fluoroscopy. Many leaks after sleeve gastrectomy are not immediate technical failures but evolve later as a result of ischemia, tissue tension, or increased intraluminal pressure. Such processes often occur beyond the typical POD 1 to 2 window when fluoroscopy is performed, resulting in false negative results. Additionally, rapid contrast passage, subtle inflammatory changes, and small or contained leaks may prevent detection by fluoroscopy, highlighting the limitations of this modality in early postoperative assessment.^[18,22,23]

The clinical implications of this study are clear. Although fluoroscopy plays a role in carefully selecting high-risk patients owing to its high specificity and NPV, CT with oral contrast continues to demonstrate superior diagnostic accuracy when clinical suspicion arises. Meta-analyses have consistently shown a CT sensitivity exceeding 90% compared with 50% to 60% for UGI series.^[24] In line with this evidence, our findings reinforce the shift toward a selective, symptom-driven, CT-based strategy, rather than the routine use of fluoroscopy.

From an economic perspective, our cost analysis (Table 9) demonstrated that routine fluoroscopy was associated with greater imaging expenditures, higher total hospitalization costs, and a modest prolongation of hospital stay. These findings align with international evidence suggesting that omitting routine contrast studies can meaningfully reduce healthcare spending without compromising patient safety.^[25,26] Within the framework of enhanced recovery after surgery, minimizing nonessential investigations not only enhances cost-effectiveness but also strengthens the principles of value-based clinical practice. The itemized cost assessment further demonstrated that the majority of total expenditures were driven by inpatient hospitalization rather than imaging costs, reinforcing that the financial impact of routine fluoroscopy stems from high-volume use rather than per-test cost. The extremely high cost per leak detected (≈$86,000) underscores the limited value of routine fluoroscopy in this setting. Despite the low unit price of each study, the near-universal application to low-risk patients results in a disproportionately high financial burden relative to the minimal diagnostic gain. These findings support the adoption of a selective, symptom-triggered imaging strategy. Inflation adjustment minimized the risk that observed cost differences were driven by temporal pricing changes rather than genuine differences in resource utilization.

From a cost-effectiveness perspective, the limited diagnostic yield of routine fluoroscopy raises important questions regarding its value in modern bariatric practice. Although each individual fluoroscopy study is relatively inexpensive, the cumulative cost of routinely imaging a large number of low-risk patients is substantial, whereas the number of additional clinically actionable diagnoses is extremely small. In contrast, a selective, symptom-based imaging strategy concentrates resource utilization in patients with a higher pretest probability of complications and may therefore provide a more favorable cost–utility profile.

Another clinically relevant consideration was the role of intra-abdominal pain. Historically employed as an early warning mechanism for leaks or bleeding, drains are increasingly viewed as unnecessary in bariatric procedures. Numerous randomized controlled trials and large cohort studies have repeatedly demonstrated no diagnostic benefit and, in certain cases, a potential association with increased morbidity.^[27,28] In our cohort, drain output did not aid the early detection of leaks or enhanced postoperative monitoring, and in some patients, it created a misleading sense of reassurance. Taken together, these results further reinforce the latest recommendations of the American Society for Metabolic and Bariatric Surgery that favor a selective, symptom-driven strategy for postoperative imaging rather than routine drain placement or fluoroscopic screening.^[29]

The integration of novel diagnostic markers and clinical algorithms offers an exciting opportunity to refine the early leak detection after bariatric surgery. Biomarkers such as C-reactive protein and procalcitonin have repeatedly demonstrated prognostic value in identifying postoperative complications.^[30] Embedding these indicators within standardized postoperative pathways, along with vigilant clinical assessment and selective imaging, could enhance diagnostic precision, while minimizing unnecessary interventions and associated costs. Future multicenter investigations exploring dynamic biomarker trajectories and optimal cutoff thresholds may further expand their clinical applicability, advancing a more personalized and economically sustainable model of postoperative care.

A notable sex imbalance was observed between the study groups, with a higher proportion of male patients in Group I. This disparity is likely related to temporal referral patterns rather than differences in clinical characteristics or operative risk. In our cohort, sex was not associated with the occurrence of staple-line leaks, and prior studies have similarly reported no sex-based differences in leak risk. Nevertheless, this imbalance may modestly limit the generalizability of the findings, and this limitation has been explicitly acknowledged.

When interpreting the statistically significant findings, it is important to emphasize their clinical magnitude. The observed differences in length of stay and cost, while statistically significant, were relatively small and may have limited practical implications.

This single-center retrospective analysis inevitably had certain constraints. The limited number of leak events reduced the statistical power and broadened the confidence intervals for the sensitivity estimates, especially in subgroup analyses. Although an adjusted multivariate model was performed to account for baseline differences, the extremely low number of leak events limited the stability of the effect estimates. Therefore, the results of this model should be interpreted cautiously and regarded as supportive sensitivity analysis rather than confirmatory evidence. Because patient allocation for routine or selective imaging was nonrandom, some selection bias may have existed, and fluoroscopic interpretation remains operator-dependent. Nevertheless, the close agreement of our findings with those of multicenter cohorts and systematic reviews supports their robustness and external validity. Because CT was performed selectively in symptomatic patients rather than routinely in all cases, a potential verification bias may exist. This selective confirmation by CT may have resulted in either overestimation or underestimation of diagnostic performance measures, particularly sensitivity. This methodological limitation has been clearly acknowledged. The study was underpowered for leak detection due to the extremely low number of events, which limits the precision of sensitivity estimates and contributes to wide confidence intervals. Although a multivariate model was performed to adjust for baseline differences, the extremely low event count limited the stability of the estimates. Cost analysis was adjusted for temporal variation, reducing the likelihood of inflation-related bias.

Our findings indicate that routine postoperative fluoroscopy after sleeve gastrectomy contributes little to improving the patient outcomes. Although the test maintains high specificity and excellent NPV, its limited sensitivity substantially reduces its reliability for early detection of staple line leaks. Given the retrospective design of this study and relatively small number of leak events, the results should be regarded as informative rather than conclusive. The close agreement between our observations and those of multicenter investigations and recent meta-analyses lends additional credibility, although some bias may persist because of the nonrandom allocation of imaging practices. Overall, these findings support the adoption of a more selective symptom-based imaging approach guided by clinical assessment and pertinent laboratory data as a rational and cost-effective strategy for postoperative management.

Although this study was conducted in a single tertiary bariatric center, the surgical technique, postoperative protocol, and leak management strategy were consistent with the widely accepted international standards. Therefore, the results are likely applicable to comparable high-volume centers performing sleeve gastrectomy under similar conditions. However, caution should be exercised when extrapolating these findings to low-volume units or settings with different perioperative care pathways, where variations in surgeon experience or imaging access may influence diagnostic performance and clinical outcomes.

5. Conclusion

The findings of this study suggest that routine postoperative fluoroscopy after sleeve gastrectomy provides limited additional diagnostic value and contributes minimally to clinical decision-making. Although the test maintains high specificity and an excellent NPV, its low sensitivity restricts its usefulness for early leak detection. However, given the retrospective design, nonrandom allocation, and particularly the extremely low number of leak events, these results should be interpreted as hypothesis-generating rather than practice-changing. The generalizability of our findings is further limited by the small number of events and the single-center design. Overall, a selective, symptom-driven imaging strategy appears reasonable and potentially more cost-efficient, but larger multicenter prospective studies are needed to confirm these preliminary observations.

Author contributions

Methodology: Haci Vural Soyer.

Project administration: Sinan Aslan.

Software: Haci Vural Soyer.

Visualization: Haci Vural Soyer.

Writing – original draft: Sinan Aslan.

Writing – review & editing: Sinan Aslan.