Investigators examined whether insulin delivery method correlates with early hospital readmission risk among adults with type 1 diabetes. Using the 2022 Nationwide Readmissions Database, they compared patients using insulin pump therapy (with or without injectable insulin) to those treated with injectable insulin only. Index admission was defined as the first eligible hospitalization within the calendar year; December and elective admissions were excluded to preserve a full 30-day follow-up window. The primary outcome was time to first all-cause readmission within 30 days.
The analysis included a weighted cohort of 230,958 adult index admissions. Absolute readmission rates differed: 14.64% among pump users versus 20.63% among injectable-only users. After multivariable adjustment with survey-weighted Cox regression, pump therapy was associated with a lower hazard of 30-day readmission (adjusted hazard ratio [aHR], 0.77; 95% CI, 0.74–0.81; P < .001). The study team reported these findings as independent of the covariates included in the models.
The authors measured in-hospital mortality during the index admission and any subsequent readmission. Reported point estimates show lower in-hospital mortality for pump users during the index hospitalization: 0.90% (95% CI, 0.78–1.05) compared with 1.31% (95% CI, 1.21–1.42) in injectable-only patients. In adjusted logistic regression models, insulin pump therapy remained associated with reduced in-hospital mortality (adjusted odds ratio [aOR], 0.78; 95% CI, 0.66–0.93; P = .006).
The authors also noted that mortality differences were assessed both during the index admission and during readmission episodes, though the abstract presents aggregate in-hospital mortality estimates for index admissions specifically. The source did not report granular causes of death or adjudicated mortality mechanisms within the abstract.
This was a retrospective cohort study using a nationally representative, all-payer inpatient database. Adults were identified by ICD-10-CM coding for type 1 diabetes. Index admissions were the first eligible hospitalization per patient within 2022; the analytic design excluded December and elective admissions to allow complete 30-day follow-up.
For time-to-event analysis of readmission, investigators applied survey-weighted Cox regression to estimate adjusted hazard ratios. For in-hospital mortality, they used survey-weighted logistic regression to estimate adjusted odds ratios. The study abstract reports adjusted effect estimates and accompanying 95% confidence intervals and P values for primary and key secondary outcomes.
The source did not report certain methodological details in the abstract: specific covariates included in the final multivariable models are not listed here, nor are details about imputation for missing data, if any. Additionally, the abstract does not specify whether facility-level clustering beyond survey weights was modeled.
The observed absolute difference in 30-day readmission rates—roughly 6 percentage points lower among pump users—translated to an adjusted hazard ratio of 0.77 (95% CI, 0.74–0.81). That association reached conventional statistical significance (P < .001).
For in-hospital mortality during the index admission, pump users had a reported mortality of 0.90% (95% CI, 0.78–1.05), and injectable-only users had 1.31% (95% CI, 1.21–1.42). Adjusted models yielded an aOR of 0.78 (95% CI, 0.66–0.93; P = .006) favoring pump therapy.
These effect sizes persisted in secondary analyses reported by the authors, as described below.
The authors performed secondary analyses that extended the readmission window. Data indicated the inverse association between insulin pump therapy and readmission risk persisted for 90-day readmission outcomes, according to the abstract. Specific numerical estimates for the 90-day analyses are not provided in the abstract.
Details about additional sensitivity tests, subgroup analyses (for example by age strata, comorbidity burden, or payer type), or methods to address residual confounding were not enumerated in the abstract text available here.
The study uses administrative claims data. That yields large, nationally representative sample sizes but introduces potential for misclassification because case ascertainment depends on ICD-10-CM codes. The design is observational and retrospective; causal inference is limited. Investigators adjusted for covariates using survey-weighted regression, yet unmeasured confounding may persist.
Important clinical details were not reported in the abstract. The source did not provide information on pump model or manufacturer, duration of pump use before admission, outpatient glycemic control (for example, HbA1c levels), insulin dosing, or specific reasons for readmission beyond the all-cause classification. Cause-specific readmission rates and adjudicated mortality causes were not presented in the abstract.
These gaps constrain interpretation: the association between insulin pump therapy and lower 30-day readmission and in-hospital mortality is robust in adjusted models in this dataset, but the abstract cannot resolve whether the effect reflects the device itself, patient selection, access to outpatient specialty care, socioeconomic factors, or other unmeasured variables. Further detail would require the full text, supplementary materials, or prospective studies designed to capture clinical and device-level covariates.