Severe anemia and red cell product ordering burden among women hospitalized with benign uterine disorders: a retrospective cohort study using MIMIC-IV

This retrospective cohort study was conducted using MIMIC-IV, a large de-identified electronic health record database containing hospital data from patients admitted to Beth Israel Deaconess Medical Center [ 19 ]. The database includes demographic information, diagnoses, procedures, laboratory results, provider order entry records, ICU stay information, and hospital outcomes. Because MIMIC-IV contains de-identified data, individual patient consent was not required for this secondary analysis. The study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement where applicable [ 20 ]. The unit of analysis was hospitalization/admission rather than individual patient. Adult female hospitalizations were screened for benign uterine disorder-related diagnostic codes. The diagnostic code set included uterine leiomyoma codes and uterine endometriosis/adenomyosis-related uterine codes: 2180, 2181, 2182, 2189, 6170, D250, D251, D252, D259, and N800. Codes corresponding to non-uterine diseases or unrelated conditions that appeared during broad code searches were not included. The initial cohort included 3,820 adult female hospitalizations with benign uterine disorder-related diagnoses. Hospitalizations without available hemoglobin measurements were excluded, leaving 2,610 records. To focus on non-obstetric benign uterine disease, 547 hospitalizations with pregnancy- or obstetric-related diagnoses were excluded. We further excluded 203 hospitalizations with gynecologic malignancy-related diagnoses, including malignancies or carcinoma in situ of the cervix, uterus, ovary, fallopian tube, vulva, vagina, and other female genital organs. The final primary analysis cohort consisted of 1,860 non-obstetric, non-malignant hospitalizations with available hemoglobin data (Fig.  1 ). In the final analytic dataset, each hospitalization corresponded to a unique hospital admission identifier and a unique subject identifier. The primary exposure was severe inpatient anemia, defined as a minimum hemoglobin level < 8 g/dL during hospitalization. The minimum hemoglobin value during the hospitalization was used to capture the greatest anemia burden experienced during the inpatient episode. This definition should be interpreted as a measure of inpatient anemia burden and does not necessarily represent hemoglobin status at admission, because the minimum value may have occurred after admission. The threshold was selected to identify clinically important inpatient anemia while preserving sufficient event numbers for analysis and is consistent with the clinical relevance of low hemoglobin thresholds in transfusion and patient blood management literature [ 11 – 13 ]. For sensitivity analyses, moderate-to-severe anemia was defined using an alternative threshold of minimum hemoglobin < 10 g/dL. This threshold was used to assess whether the observed associations were robust when a broader anemia definition was applied. Hospitalizations were classified into disease groups according to diagnostic codes. The fibroid-only group included hospitalizations with uterine leiomyoma-related codes without concurrent uterine endometriosis/adenomyosis-related codes. The uterine endometriosis/adenomyosis-related-only group included hospitalizations with uterine endometriosis or adenomyosis-related codes without fibroid codes. The combined group included hospitalizations with both fibroid-related and uterine endometriosis/adenomyosis-related codes. The N800 code was handled conservatively as a uterine endometriosis/adenomyosis-related diagnostic code rather than as definitive adenomyosis. This approach was used to avoid overinterpreting administrative coding and to maintain consistency with the broader concept of benign uterine disorders. The main outcome was red cell product ordering, identified from Blood Bank provider order entry records. This variable was interpreted as a transfusion-related resource use indicator rather than confirmed red cell administration. The number of red cell product orders was also summarized; this count represented the number of provider order entry records and should not be interpreted as the number of transfused red cell units or confirmed administrations. Additional hospital burden outcomes included hospital length of stay, ICU admission during hospitalization, and in-hospital mortality. ICU admission during hospitalization was defined by the presence of an ICU stay during the same hospital admission. Because this variable reflects whether a hospitalization involved ICU-level care, it is described as ICU admission during hospitalization rather than strictly as transfer from the ward to the ICU. Gynecologic procedure-related outcomes included hysterectomy, uterine or pelvic embolization, and uterine lesion excision or destruction. These variables were identified using procedure codes and were used to describe procedure patterns rather than to infer treatment efficacy. Covariates included age, race/ethnicity, insurance status, marital status, admission type, and disease group. Race/ethnicity was grouped as White, Black, Asian, Hispanic/Latino, and Other/Unknown. Admission type was grouped as Emergency/Urgent, Observation, and Elective/Scheduled. Missing or unclear categorical values were retained as missing or unknown categories when applicable. Continuous variables were summarized as medians with interquartile ranges. Categorical variables were summarized as counts and percentages. Baseline characteristics and hospital burden outcomes were compared between hospitalizations with and without severe anemia. Multivariable logistic regression was used to evaluate the association of severe anemia with red cell product ordering and ICU admission during hospitalization. Adjusted models included age, race/ethnicity, insurance status, marital status, admission type, and disease group. Because in-hospital mortality occurred in only 11 hospitalizations, the mortality analysis was conducted as an exploratory univariable logistic regression model with severe anemia as the only predictor. Sensitivity analyses were performed using hemoglobin < 10 g/dL as an alternative anemia threshold. In these analyses, moderate-to-severe anemia was evaluated in relation to red cell product ordering and ICU admission using the same adjustment covariates as the main multivariable models. To address the limited disease group classification and assess whether findings were driven by the dominant diagnostic subgroup, we conducted a fibroid-only subgroup analysis restricted to hospitalizations with fibroid-only disease. Within this subgroup, multivariable logistic models for red cell product ordering and ICU admission were adjusted for age, race/ethnicity, insurance status, marital status, and admission type. Because the uterine endometriosis/adenomyosis-related-only and combined groups were small, these groups were described but not modeled separately. Results are reported as odds ratios (ORs) with 95% confidence intervals (CIs). A two-sided P value < 0.05 was considered statistically significant. To assess potential selection bias from excluding hospitalizations without hemoglobin measurements, we compared initially identified hospitalizations with and without available hemoglobin measurements before the hemoglobin-based exclusion step. This supplementary comparison used aggregate summaries and standardized mean differences and did not change the primary analysis cohort. To address the timing limitation of the minimum inpatient hemoglobin definition, we performed additional sensitivity analyses using earlier hemoglobin definitions: first inpatient hemoglobin < 8 g/dL, admission-window hemoglobin < 8 g/dL, and early 24-hour hemoglobin < 8 g/dL. Admission-window hemoglobin was defined as the hemoglobin value closest to admission within 6 h before to 24 h after admission. Early 24-hour hemoglobin was defined using the minimum hemoglobin value measured during the first 24 h after admission. These sensitivity models used the same adjustment covariates as the main multivariable models.

A total of 3,820 adult female hospitalizations with benign uterine disorder-related diagnostic codes were initially identified. After excluding hospitalizations without hemoglobin measurements, 2,610 hospitalizations remained. We then excluded 547 pregnancy- or obstetric-related hospitalizations and 203 hospitalizations with gynecologic malignancy-related diagnoses. The final primary analysis cohort included 1,860 non-obstetric, non-malignant hospitalizations with available hemoglobin data (Fig.  1 ). Fig. 1 Flowchart of study population selection. A total of 3,820 adult female hospitalizations with benign uterine disorder-related diagnostic codes were initially identified from MIMIC-IV. After excluding hospitalizations without hemoglobin measurements, pregnancy- or obstetric-related diagnoses, and gynecologic malignancy-related diagnoses, 1,860 hospitalizations were included in the final primary analysis cohort Flowchart of study population selection. A total of 3,820 adult female hospitalizations with benign uterine disorder-related diagnostic codes were initially identified from MIMIC-IV. After excluding hospitalizations without hemoglobin measurements, pregnancy- or obstetric-related diagnoses, and gynecologic malignancy-related diagnoses, 1,860 hospitalizations were included in the final primary analysis cohort Among these hospitalizations, 425 (22.8%) had severe anemia, defined as a minimum inpatient hemoglobin level  = 8 g/dL. In the final analytic dataset, the 1,860 hospitalizations corresponded to 1,860 unique subject identifiers, with no repeated subject identifier observed. Baseline characteristics according to severe anemia status are shown in Table  1 . The median age of the overall cohort was 48 years. Hospitalizations with severe anemia involved slightly younger women than those without severe anemia [47 (40–51) vs. 48 (42–56) years; P  < 0.001]. The number of hemoglobin measurements was higher in the severe anemia group [5 (3–9) vs. 2 (1–4); P  < 0.001], consistent with greater clinical monitoring during hospitalization. Insurance status differed between groups. Medicaid insurance was more frequent among hospitalizations with severe anemia (29.76% vs. 24.88%), whereas private insurance was slightly less frequent (53.10% vs. 56.64%). Race/ethnicity distribution also differed significantly. Black women accounted for 50.12% of the severe anemia group compared with 39.02% of the non-severe anemia group, while White women accounted for 23.29% and 38.68%, respectively. Marital status differed between groups, with a higher proportion of single status in the severe anemia group. Admission type also differed, with emergency or urgent admissions more common among hospitalizations with severe anemia (41.18% vs. 29.97%). Disease group distribution did not differ significantly between the severe anemia and non-severe anemia groups. Table 1 Baseline characteristics of women hospitalized with benign uterine disorders according to severe anemia status Characteristic Overall ( n  = 1,860) Hb > = 8 g/dL ( n  = 1,435) Hb < 8 g/dL ( n  = 425) P value Age, years 48 (42–55) 48 (42–56) 47 (40–51) < 0.001 Hemoglobin tests, n 2 (1–5) 2 (1–4) 5 (3–9) < 0.001 Insurance < 0.001  Medicaid 479 (25.99) 354 (24.88) 125 (29.76)  Medicare 275 (14.92) 227 (15.95) 48 (11.43)  Other 60 (3.26) 36 (2.53) 24 (5.71)  Private 1029 (55.83) 806 (56.64) 223 (53.10) Race/ethnicity < 0.001  Asian 112 (6.02) 78 (5.44) 34 (8.00)  Black 773 (41.56) 560 (39.02) 213 (50.12)  Hispanic/Latino 148 (7.96) 114 (7.94) 34 (8.00)  Other/Unknown 173 (9.30) 128 (8.92) 45 (10.59)  White 654 (35.16) 555 (38.68) 99 (23.29) Marital status 0.009  Divorced 145 (7.91) 119 (8.40) 26 (6.25)  Married 718 (39.17) 562 (39.66) 156 (37.50)  Single 901 (49.15) 674 (47.57) 227 (54.57)  Widowed 69 (3.76) 62 (4.38) 7 (1.68) Admission type < 0.001  Elective/Scheduled 595 (31.99) 471 (32.82) 124 (29.18)  Emergency/Urgent 605 (32.53) 430 (29.97) 175 (41.18)  Observation 660 (35.48) 534 (37.21) 126 (29.65) Disease group 0.586  Combined 95 (5.11) 73 (5.09) 22 (5.18)  Fibroid only 1668 (89.68) 1283 (89.41) 385 (90.59)  Uterine endometriosis/adenomyosis-related only 97 (5.22) 79 (5.51) 18 (4.24) Values are presented as median (interquartile range) or n (%) Hb hemoglobin Baseline characteristics of women hospitalized with benign uterine disorders according to severe anemia status Values are presented as median (interquartile range) or n (%) Hb hemoglobin Hospital burden and procedure-related outcomes are presented in Table  2 . Red cell product ordering was markedly more common in the severe anemia group. Among hospitalizations with severe anemia, 269 of 425 (63.29%) had red cell product orders, compared with 73 of 1,435 (5.09%) hospitalizations without severe anemia ( P  < 0.001). The median number of red cell product order entries was also higher in the severe anemia group [1 (0–2) vs. 0 (0–0); P  < 0.001]. This order count represents provider order entry count and should not be interpreted as transfused red cell units. Hospitalizations with severe anemia had longer hospital stays. The median hospital length of stay was 3.436 (1.911–6.389) days in the severe anemia group and 2.310 (1.335–3.816) days in the non-severe anemia group ( P  < 0.001). ICU admission during hospitalization was more frequent among hospitalizations with severe anemia (15.06% vs. 6.13%; P  < 0.001). In-hospital mortality was uncommon overall but occurred more frequently in the severe anemia group in crude comparisons (1.41% vs. 0.35%; P  = 0.031). Procedure patterns also differed between groups. Hysterectomy was less frequent among hospitalizations with severe anemia (10.82% vs. 18.68%; P  < 0.001), whereas uterine lesion excision or destruction was more frequent (28.24% vs. 19.79%; P  < 0.001). Uterine or pelvic embolization was performed in 3.29% of hospitalizations with severe anemia and 4.60% of hospitalizations without severe anemia, with no statistically significant difference between groups ( P  = 0.304). Table 2 Hospital burden and procedure-related outcomes according to severe anemia status Outcome Overall ( n  = 1,860) Hb > = 8 g/dL ( n  = 1,435) Hb < 8 g/dL ( n  = 425) P value Red cell product order count 0 (0–0) 0 (0–0) 1 (0–2) < 0.001 Hospital length of stay, days 2.503 (1.446–4.294) 2.310 (1.335–3.816) 3.436 (1.911–6.389) < 0.001 Red cell product order < 0.001  No 1518 (81.61) 1362 (94.91) 156 (36.71)  Yes 342 (18.39) 73 (5.09) 269 (63.29) ICU admission during hospitalization < 0.001  No 1708 (91.83) 1347 (93.87) 361 (84.94)  Yes 152 (8.17) 88 (6.13) 64 (15.06) In-hospital mortality 0.031  No 1849 (99.41) 1430 (99.65) 419 (98.59)  Yes 11 (0.59) 5 (0.35) 6 (1.41) Hysterectomy < 0.001  No 1546 (83.12) 1167 (81.32) 379 (89.18)  Yes 314 (16.88) 268 (18.68) 46 (10.82) Uterine or pelvic embolization 0.304  No 1780 (95.70) 1369 (95.40) 411 (96.71)  Yes 80 (4.30) 66 (4.60) 14 (3.29) Uterine lesion excision or destruction < 0.001  No 1456 (78.28) 1151 (80.21) 305 (71.76)  Yes 404 (21.72) 284 (19.79) 120 (28.24) Values are presented as median (interquartile range) or n (%). Red cell product order count represents provider order entry count and should not be interpreted as transfused units ICU intensive care unit, Hb hemoglobin Hospital burden and procedure-related outcomes according to severe anemia status Values are presented as median (interquartile range) or n (%). Red cell product order count represents provider order entry count and should not be interpreted as transfused units ICU intensive care unit, Hb hemoglobin Logistic regression results are summarized in Table  3 . After adjustment for age, race/ethnicity, insurance status, marital status, admission type, and disease group, severe anemia was associated with red cell product ordering. Compared with hospitalizations without severe anemia, hospitalizations with severe anemia had substantially higher odds of red cell product ordering (adjusted OR 33.61, 95% CI 24.23–47.29; P  < 0.001). Severe anemia was also associated with ICU admission during hospitalization. After adjustment for the same covariates, hospitalizations with severe anemia had higher odds of ICU admission during hospitalization than those without severe anemia (adjusted OR 2.69, 95% CI 1.84–3.91; P  < 0.001). This association should not be interpreted as evidence that anemia caused ICU admission. In an exploratory univariable logistic regression model, severe anemia was associated with higher in-hospital mortality (OR 4.10, 95% CI 1.23–14.27; P  = 0.020). This result should be interpreted cautiously because the total number of death events was small. Sensitivity analyses using hemoglobin < 10 g/dL as an alternative threshold showed directionally consistent results. Moderate-to-severe anemia was associated with red cell product ordering after adjustment for age, race/ethnicity, insurance status, marital status, admission type, and disease group (adjusted OR 74.64, 95% CI 36.04-190.32; P  < 0.001). Moderate-to-severe anemia was also associated with ICU admission during hospitalization (adjusted OR 2.01, 95% CI 1.39–2.95; P  < 0.001). In the fibroid-only subgroup ( n  = 1,668), severe anemia remained associated with red cell product ordering after adjustment for age, race/ethnicity, insurance status, marital status, and admission type (adjusted OR 35.55, 95% CI 24.96–50.64; P  < 0.001). Severe anemia also remained associated with ICU admission during hospitalization in this subgroup (adjusted OR 2.68, 95% CI 1.83–3.92; P  < 0.001). Descriptive analyses by disease group are provided in Additional file 1; the uterine endometriosis/adenomyosis-related-only and combined groups were small and were therefore not modeled separately. To evaluate potential selection bias from excluding hospitalizations without hemoglobin measurements, we compared the 2,610 hospitalizations with available hemoglobin data with the 1,210 initially identified hospitalizations without available hemoglobin data. Hospitalizations with available hemoglobin data had longer median length of stay [2.82 (1.70–4.72) vs. 1.56 (1.06–2.47) days], more frequent red cell product ordering (15.8% vs. 1.9%), and more frequent ICU admission during hospitalization (7.0% vs. 0.2%). This comparison is summarized in Additional file 1: Supplementary Table S4 and supports the possibility of selection related to hemoglobin testing. In additional sensitivity analyses using earlier hemoglobin definitions, first inpatient hemoglobin, admission-window hemoglobin, and early 24-hour minimum hemoglobin were available for 1,635, 1,431, and 1,402 of the final 1,860 hospitalizations, respectively. Early 24-hour hemoglobin < 8 g/dL remained associated with red cell product ordering after multivariable adjustment (adjusted OR 22.57, 95% CI 15.54–32.80; P  < 0.001), and similar associations were observed using admission-window hemoglobin < 8 g/dL and first inpatient hemoglobin < 8 g/dL. In contrast, associations with ICU admission were attenuated and not statistically significant under these earlier hemoglobin definitions (Additional file 1: Supplementary Table S5). Table 3 Logistic regression analyses for anemia-related hospital burden outcomes Analysis Exposure Outcome Model OR 95% CI P value Main analysis Severe anemia, Hb < 8 g/dL Red cell product order Adjusted 33.61 24.23–47.29 < 0.001 Main analysis Severe anemia, Hb < 8 g/dL ICU admission during hospitalization Adjusted 2.69 1.84–3.91 < 0.001 Exploratory analysis Severe anemia, Hb < 8 g/dL In-hospital mortality Univariable 4.10 1.23–14.27 0.020 Sensitivity analysis Moderate-to-severe anemia, Hb < 10 g/dL Red cell product order Adjusted 74.64 36.04-190.32 < 0.001 Sensitivity analysis Moderate-to-severe anemia, Hb < 10 g/dL ICU admission during hospitalization Adjusted 2.01 1.39–2.95 < 0.001 Fibroid-only subgroup Severe anemia, Hb < 8 g/dL Red cell product order Adjusted 35.55 24.96–50.64 < 0.001 Fibroid-only subgroup Severe anemia, Hb < 8 g/dL ICU admission during hospitalization Adjusted 2.68 1.83–3.92 < 0.001 Adjusted models included age, race/ethnicity, insurance status, marital status, admission type, and disease group, except fibroid-only subgroup models, which omitted disease group. The in-hospital mortality model was univariable because of the small number of death events. Red cell product order was identified from Blood Bank provider order entry records and should be interpreted as a transfusion-related resource use indicator rather than confirmed transfusion administration CI confidence interval, Hb hemoglobin, ICU intensive care unit, OR odds ratio Logistic regression analyses for anemia-related hospital burden outcomes Adjusted models included age, race/ethnicity, insurance status, marital status, admission type, and disease group, except fibroid-only subgroup models, which omitted disease group. The in-hospital mortality model was univariable because of the small number of death events. Red cell product order was identified from Blood Bank provider order entry records and should be interpreted as a transfusion-related resource use indicator rather than confirmed transfusion administration CI confidence interval, Hb hemoglobin, ICU intensive care unit, OR odds ratio

In this retrospective cohort study of women hospitalized with benign uterine disorders, severe inpatient anemia was common and was associated with greater hospitalization-level resource use. More than one in five non-obstetric, non-malignant hospitalizations with available hemoglobin data met the definition of severe anemia. Severe anemia was associated with a much higher frequency of red cell product ordering, longer hospital stay, and greater likelihood of ICU admission during hospitalization. These findings frame severe inpatient anemia as a practical marker of hospital burden in this women’s health population, while the observational design precludes causal interpretation. The strongest association observed in this study was between severe anemia and red cell product ordering. This finding is expected clinically, but the magnitude of association underscores how strongly inpatient hemoglobin burden aligns with transfusion-related resource use among women hospitalized with benign uterine disorders. Importantly, the outcome in this study was red cell product ordering rather than confirmed transfusion administration. Provider order entry data reflect transfusion-related preparation and resource use, but they do not necessarily prove that red cells were administered. Even with this conservative interpretation, the finding remains clinically meaningful because red cell product ordering itself represents a measurable hospital process and a practical indicator of anemia-related care escalation. Additional analyses using first inpatient, admission-window, and early 24-hour hemoglobin definitions further supported the robustness of the red cell product ordering outcome. The association between early 24-hour hemoglobin < 8 g/dL and red cell product ordering remained strong after multivariable adjustment, suggesting that the main red cell product-ordering result was not solely driven by the use of the lowest hemoglobin value over the entire hospitalization. However, the ICU admission association was attenuated and not statistically significant under earlier hemoglobin definitions, reinforcing that ICU-related findings should be interpreted as markers of overall inpatient complexity rather than as temporally established effects of anemia. The clinical relevance of anemia-related resource use in gynecologic care is supported by studies outside MIMIC-IV. In a cohort of nonpregnant women presenting to the emergency department with iron deficiency anemia caused by uterine blood loss, Boone et al. described frequent transfusion support and opportunities for improved iron-directed management [ 21 ]. In surgical fibroid populations, myomectomy has been associated with nontrivial transfusion risk and postoperative morbidity [ 22 , 23 ]. Cao et al. identified perioperative transfusion risk factors among women undergoing total laparoscopic hysterectomy [ 24 ]. More specifically, Tercan et al. reported that preoperative hematological and inflammatory markers and myoma characteristics were associated with postoperative erythrocyte suspension requirements following myomectomy [ 25 ]. These studies evaluated surgical or emergency-care populations and generally measured confirmed transfusion requirements, whereas our primary outcome was provider order entry. Nevertheless, they collectively support the clinical importance of hematological assessment and blood-product planning in women with benign uterine disease. Severe anemia was also associated with ICU admission during hospitalization. This finding should not be interpreted as evidence that anemia caused ICU admission. Rather, severe anemia may mark greater inpatient complexity, closer monitoring needs, hemodynamic assessment, procedural planning, or management of acute bleeding or related complications. The association between severe anemia and ICU admission remained significant after adjustment for age, race/ethnicity, insurance, marital status, admission type, and disease group, and it was also observed in the fibroid-only subgroup. These findings suggest that the association was not driven solely by the broad disease group distribution, although residual confounding remains possible. The disease group distribution did not differ significantly between hospitalizations with and without severe anemia. Most hospitalizations were in the fibroid-only group, while the uterine endometriosis/adenomyosis-related-only and combined groups were smaller. The fibroid-only subgroup analysis showed directionally consistent associations for both red cell product ordering and ICU admission. Because the other disease strata were small, they were summarized descriptively rather than modeled separately. Administrative diagnostic codes do not capture lesion size, fibroid location, bleeding duration, symptom severity, imaging findings, medication exposure, or prior treatment history, all of which may influence anemia risk. Prior studies have shown that fibroid characteristics, particularly submucosal fibroids, may be relevant to bleeding and anemia risk [ 9 ]. The procedure-related findings are descriptive and should be interpreted cautiously. Hysterectomy was less frequent in the severe anemia group, whereas uterine lesion excision or destruction was more frequent. Uterine or pelvic embolization was uncommon and did not differ significantly between groups. These findings should not be interpreted as comparative effectiveness evidence for any procedure. Procedure selection in hospitalized patients is influenced by clinical stability, patient preference, fertility considerations, procedural availability, surgical risk, bleeding severity, and timing of admission. Instead, these results should be viewed as descriptive evidence of how procedure patterns differ across anemia severity groups in real-world inpatient care. This study has several practical implications for women’s health and gynecologic inpatient care. First, severe inpatient anemia among women hospitalized with benign uterine disorders is not rare and may identify hospitalizations with greater resource needs. Second, severe anemia may help clinicians anticipate red cell product ordering, longer hospitalization, and possible ICU-level care, although it should not be used as a causal explanation for these outcomes. Third, women with benign uterine disorders and substantial anemia burden may require coordinated inpatient management involving gynecology, interventional radiology, transfusion medicine, anesthesia, and critical care when clinically appropriate. Fourth, these findings may be relevant to pre-procedural planning and anemia optimization, especially for patients being considered for uterine-sparing interventions or other definitive treatments, including uterine artery embolization and myomectomy in appropriately selected patients [ 16 – 18 ]. The present analysis intentionally avoided framing the study as a model to predict the presence of fibroids or adenomyosis, or as a causal evaluation of treatment effects. MIMIC-IV is a hospital-based database and is not designed to represent a general gynecologic screening population. The study therefore focused on hospitalizations with existing benign uterine disorder-related diagnoses and evaluated anemia-related hospital burden. This framing aligns the research question with the strengths of inpatient electronic health record data. Several limitations should be acknowledged. First, this was a retrospective observational study, and causal inference cannot be established. Associations between severe anemia and red cell product ordering, ICU admission, length of stay, or mortality should not be interpreted as evidence that anemia caused these outcomes. Second, the unit of analysis was hospitalization/admission. Although the final analytic dataset contained unique subject identifiers without repeated hospitalizations, the cohort was constructed and analyzed at the hospitalization level, and the results should be interpreted accordingly. Third, diagnostic and procedure definitions relied on administrative codes, which may be subject to misclassification. In particular, N800 was treated as a uterine endometriosis/adenomyosis-related code but should not be interpreted as definitive adenomyosis in every case. The disease grouping was broad and did not capture important gynecologic details such as uterine size, fibroid number or location, menstrual bleeding severity, imaging findings, medication use, outpatient management, patient preference, or detailed procedural indications. Fourth, red cell product ordering was identified from provider order entry records and should not be equated with confirmed transfusion administration. Similarly, red cell product order count represents provider order entry count and not the number of transfused units. Fifth, severe anemia was defined using the minimum inpatient hemoglobin value. This approach was chosen to capture the greatest anemia burden during hospitalization, but the minimum value may have occurred after admission. Therefore, the exposure should not be interpreted as admission hemoglobin status or as proof that severe anemia preceded all downstream outcomes. We performed additional analyses using first inpatient, admission-window, and early 24-hour hemoglobin definitions. These analyses supported the robustness of the red cell product ordering association but not the ICU admission association, so temporal interpretation remains cautious. Sixth, hospitalizations without hemoglobin measurements were excluded from the primary analysis. This exclusion may introduce selection bias because hospitalizations with hemoglobin testing may differ from those without testing in clinical severity, monitoring intensity, or procedural context. We performed a supplementary comparison of initially identified hospitalizations with and without available hemoglobin measurements, which showed differences in length of stay, red cell product ordering, and ICU admission. However, this comparison cannot eliminate residual selection bias or fully establish the direction of bias in the primary estimates. Seventh, the study was based on a single de-identified hospital database and did not include independent external validation; therefore, the findings may not generalize to outpatient populations, other health systems, or settings with different transfusion-ordering practices. Finally, although crude in-hospital mortality was higher among hospitalizations with severe anemia, death events were rare, and the mortality analysis was exploratory and should not guide clinical inference. Despite these limitations, the study provides a focused description of anemia-related inpatient burden among women hospitalized with benign uterine disorders. By excluding pregnancy-related admissions and gynecologic malignancies, the analysis attempted to isolate a clinically coherent non-obstetric, non-malignant population. The consistency of the main findings using both hemoglobin < 8 g/dL and hemoglobin < 10 g/dL thresholds, the fibroid-only subgroup results, and the earlier hemoglobin sensitivity analyses for red cell product ordering support the robustness of the observed relationship between inpatient anemia burden and red cell product ordering/resource use within the available data.

Among women hospitalized with benign uterine disorders in MIMIC-IV, severe inpatient anemia was common and was associated with greater red cell product ordering and longer hospital length of stay, and with more frequent ICU admission under the primary minimum-hemoglobin definition. Earlier hemoglobin sensitivity analyses supported the robustness of the red cell product ordering association, whereas ICU-related findings were less consistent and should be interpreted cautiously. These findings suggest that severe inpatient anemia may serve as a practical marker of greater anemia-related hospital burden and red cell product ordering in this defined hospitalized population. The results should be interpreted as observational associations rather than causal effects, and further studies with confirmed transfusion administration, richer gynecologic details, and external validation are warranted.

Benign uterine disorders, particularly uterine leiomyomas and uterine endometriosis or adenomyosis-related conditions, represent an important source of gynecologic morbidity and a major women’s health issue among reproductive-aged and peri-menopausal women [ 1 – 6 ]. Although these disorders are non-malignant, their clinical consequences are not always mild. Heavy menstrual bleeding, chronic pelvic symptoms, acute exacerbations, and procedure-related admissions may lead to clinically meaningful anemia, increased use of hospital resources, and, in selected cases, escalation of care [ 7 – 10 ]. Anemia is one of the most direct and measurable consequences of bleeding-related benign uterine disease. In routine practice, hemoglobin level often influences the urgency of inpatient evaluation, the need for blood product preparation, timing of operative or interventional management, and peri-procedural risk assessment [ 11 – 13 ]. Severe anemia may also reflect delayed diagnosis, prolonged disease course, acute bleeding, socioeconomic barriers to care, or greater physiologic vulnerability. Despite its clinical relevance, the inpatient burden of severe anemia among hospitalized women with benign uterine disorders has not been sufficiently described using large-scale real-world hospital data. Most existing discussions of benign uterine disease focus on symptom control, fertility preservation, surgical management, or disease-specific treatment strategies [ 14 – 18 ]. Less attention has been paid to the hospitalization-level burden created by anemia itself, including red cell product ordering, length of stay, ICU admission, and short-term mortality. Such information is clinically useful because hospitalizations differ from outpatient encounters: they may represent more urgent presentations, more complex care needs, and greater resource use. The MIMIC-IV database provides detailed de-identified hospital data and allows investigation of inpatient clinical burden in a real-world setting [ 19 ]. In this study, we used MIMIC-IV to evaluate severe inpatient anemia and red cell product ordering among women hospitalized with benign uterine disorders. Rather than developing a disease prediction model or evaluating treatment efficacy, we focused on a clinically practical women’s health question: whether severe inpatient anemia is associated with greater anemia-related hospital burden and resource use.

Additional file 1: Supplementary Tables S1-S5. Fibroid-only subgroup adjusted logistic regression, descriptive outcomes by disease group, repeated hospitalization check, comparison of hospitalizations with and without available hemoglobin measurements, and early hemoglobin sensitivity analyses. Additional file 1: Supplementary Tables S1-S5. Fibroid-only subgroup adjusted logistic regression, descriptive outcomes by disease group, repeated hospitalization check, comparison of hospitalizations with and without available hemoglobin measurements, and early hemoglobin sensitivity analyses.