# Crisis-time efficiency in Eastern Poland's regional hospitals (2015–2024): a data envelopment analysis

## TL;DR

This study analyzed how efficiently three regional hospitals in Eastern Poland operated and managed finances from 2015 to 2024, especially during the pandemic.

## Contribution

The paper introduces a decade-long, crisis-focused efficiency analysis of hospitals in a strategically sensitive border region using DEA with bootstrap confidence intervals.

## Key findings

- Operational efficiency dropped during the pandemic but partially recovered afterward.
- Financial efficiency remained relatively stable, showing a divergence from operational performance.
- Scale inefficiencies were a consistent issue across all hospitals and time periods.

## Abstract

Hospitals in Poland's border regions face persistent staffing pressures and rising costs, and the COVID-19 pandemic further disrupted activity. We assessed year-by-year changes in operational and financial efficiency in three regional hospitals (BIA, LOM, SUW) in Podlaskie Voivodeship, on NATO's eastern flank, over 2015–2024. Input-oriented Data Envelopment Analysis with CCR (CRS) and BCC (VRS) models was applied in two domains (operational and financial), and 95% bootstrap confidence intervals were calculated for efficiency scores. Operational efficiency declined during the pandemic and partially recovered thereafter. Before COVID-19, mean TE_CCR ranged from 0.607 (95% CI: 0.571–0.643) in LOM to 0.909 (0.833–0.989) in SUW. In 2020–2021, TE_CCR fell to 0.746 (0.704–0.783) in BIA and 0.399 (0.371–0.410) in LOM, with SUW decreasing to 0.810 (0.731–0.870). Post-pandemic values showed partial rebound: 0.858 (0.781–0.946) in BIA, 0.602 (0.565–0.634) in LOM, and 0.830 (0.758–0.913) in SUW. For LOM, operational TE_CCR dropped as low as 0.399 while financial TE_CCR remained at or above 0.94, illustrating a marked divergence between service delivery and financial performance. Across periods, PTE_BCC remained high, indicating scale efficiency as an important source of inefficiency. Financial efficiency showed a similar trough and recovery. Pandemic-period TE_CCR declined to 0.785 (0.766–0.798) in BIA and 0.951 (0.925–0.967) in SUW, while LOM remained relatively stable at 0.960 (0.947–0.970). Post-pandemic values increased to 0.928 (0.863–0.959) in BIA and 0.949 (0.901–0.979) in SUW, with LOM at 0.940 (0.919–0.958). Several confidence intervals did not include 1.00, indicating persistent inefficiency components. Overall, the decade shows a distinct pandemic-related dip followed by partial normalisation, with between-hospital heterogeneity and recurrent scale-related shortfalls in both domains. These results support routine, domain-specific efficiency monitoring as a tool for transparent performance tracking in strategically sensitive border regions. However, findings are constrained by the very small three-hospital sample, low discriminatory power of annual frontiers and the use of aggregated administrative data. Key methodological constraints are summarised in the Limitations section and should be considered when interpreting the findings.