Abstract
Objective
To identify the determinants of 28-day perioperative mortality and postoperative length of hospital stay among geriatric patients.
Design and setting
A prospective, two-center follow-up study conducted at two tertiary referral hospitals in Ethiopia.
Participants
A total of 1014 consecutive geriatric patients who underwent surgery between January 2019 and January 2022 were included.
Results
This study documented a 28-day perioperative mortality rate of 2.86% (95% CI: 1.9–4.0) and a median postoperative length of hospital stay of six days. Multivariable analyses identified independent determinants for each outcome. For 28-day perioperative mortality, Cox proportional hazards regression revealed significant associations with the presence of comorbidities (aHR = 3.10) and the use of general anesthesia (aHR = 6.12). Determinants of extended postoperative length of hospital stay, evaluated using negative binomial regression, included the following: comorbidities (aIRR= 1.31), emergency surgery (aIRR = 1.28), and higher American Society of Anesthesiologists (ASA) physical status classification (ASA II: aIRR = 1.31; ASA ≥ III: aIRR = 1.38), each additional 60 minutes of anesthesia duration (aIRR = 1.15), and intraoperative blood loss (>500 mL; aIRR = 1.27). Conversely, each 1 g/dL increase in preoperative hemoglobin was found to be protective (aIRR = 0.97).
Conclusion
This study identifies key modifiable determinants of adverse postoperative outcomes in geriatric surgical patients. The 28-day perioperative mortality was strongly associated with general anesthesia and preexisting comorbidities. The postoperative length of hospital stay was influenced by a broader set of variables, including patient factors, surgical urgency, and procedural complexity. These findings highlight critical targets for quality improvement in low-resource settings.
Keywords: Perioperative outcomes, Geriatric surgery, Low-resource settings, Preoperative anemia, Anesthesia
Article summary
Strengths
Prospective two-center design with standardized 28-day follow-up (in-hospital + telephone).
Use of sufficient sample size, enhancing generalizability.
Use of survival analysis and negative binomial regression for rigorous statistical modeling.
Limitations
Potential for residual confounding from unmeasured factors (e.g., socioeconomic status, frailty, preoperative nutritional status).
The sample size, while adequate for the primary analysis, was insufficient for subgroup analyses by specific surgical procedure type.
Background
The global geriatric population is increasing rapidly in both developed and developing nations, with projections indicating it will double by 2050 [1, 2]. This demographic shift is accompanied by a higher prevalence of age-related chronic conditions, including hypertension, diabetes, coronary artery disease, heart failure, and Alzheimer’s disease [3]. Consequently, the demand for surgical care in this age group has risen significantly, with estimates suggesting an 18 to 30% increase [3].
Advanced age is an established risk factor for adverse perioperative outcomes. This elevated risk is largely attributable to the high burden of multimorbidity and the progressive physiological decline associated with aging [4]. Geriatric surgical patients face heightened risks of postoperative complications, prolonged hospital stays, emergency department revisits, readmissions, and increased post-discharge care needs [5, 6]. Hospital length of stay, defined as the duration of an inpatient’s hospitalization, has been a key performance indicator (KPI) for measuring healthcare efficiency and quality of care [7, 8]. Prolonged postoperative length of hospital stay is often a marker of clinical complexity and is associated with excess costs, higher resource utilization, and an increased risk of nosocomial complications [9–12]. Determinants of postoperative length of hospital stay are multifactorial, encompassing patient-related factors (e.g., multimorbidity, polypharmacy, frailty, functional and cognitive impairment, higher ASA physical status), intervention-related factors (e.g., use of general anesthesia, prolonged operative duration, significant blood loss), and system-related factors [13–15]. Strategies such as comprehensive geriatric assessment, structured discharge planning, interdisciplinary care, and clinical pathways have been shown to effectively reduce postoperative length of hospital stay in this population [16, 17].
Despite the growing clinical and economic significance of this issue, there remains a scarcity of published data on 28-day perioperative mortality and postoperative length of hospital stay among geriatric surgical patients in Ethiopia. This study therefore aims to identify the determinants of 28-day perioperative mortality and postoperative length of hospital stay (PLOS) in geriatric patients undergoing surgery at two tertiary hospitals in Ethiopia.
Methods
Study design and setting
This prospective cohort study was conducted between January 1, 2019, and January 1, 2022, at two major tertiary referral hospitals in Ethiopia: Ayder Comprehensive Specialized Hospital (ACSH) in Mekelle, Tigray Region, and Tibebe-Ghion Comprehensive Specialized Hospital (TGCSH) in Bahir Dar, Amhara Region. These institutions are affiliated with Mekelle University and Bahir Dar University, respectively. ACSH serves as one of the largest and principal referral hospitals in northern Ethiopia, with a capacity of 550 beds and 10 operating theaters. Likewise, TGCSH is a leading tertiary teaching and referral hospital in the Amhara Region, comprising 500 beds and 13 operating theaters dedicated to major surgical procedures.
Study population
This study involved all patients aged 65 and older who underwent major surgery at TGCSH and ACSH between January 2019 and January 2022.
Study inclusion and exclusion criteria
The study included all consecutive patients aged ≥ 65 years who underwent major surgery at ACSH or TGCSH during the study period. Patients with incomplete data and those discharged on the same day as surgery (PLOS = 0) were excluded.
Data collection
Data were collected prospectively using the software Research Electronic Data Capture (REDCap) at both hospitals. Each site had one data manager and one information technology (IT) personnel to oversee the local data collection processes. Data were collected by trained anesthesia students from Mekelle and Bahir Dar Universities, alongside affiliated healthcare providers. All personnel, including data collectors, managers, and IT staff, received standardized training on the data collection system and ethical protocols to ensure data quality. All recorded data were securely stored in a protected REDCap database. Data managers were responsible for finalizing patient follow-up data, which included mortality and discharge status following surgery. These finalized data were then uploaded to a centralized server. Patient monitoring involved in-person observation during the hospital stays. For patients discharged from the hospital, subsequent mortality status was assessed through weekly phone calls conducted over a 28-day postoperative period.
Study endpoints
The primary outcome measures were 28-day perioperative mortality and PLOS. The determinants of postoperative outcomes were classified into three distinct domains: (1) sociodemographic factors, including patient age, sex, and treating hospital; (2) preoperative clinical characteristics, encompassing comorbidities, American Society of Anesthesiologists (ASA) physical status classification, surgical type, procedural urgency, and baseline vital signs; and (3) intraoperative management variables, consisting of surgical and anesthesia duration, anesthesia type, and estimated blood loss.
Operational definitions
PLOS
Defined as the number of days from the day of surgery to the day of discharge from the hospital [18].
Prolonged PLOS
Defined as any duration equal to or exceeding the 75th percentile of postoperative hospital stays in the study population [8].
Duration of surgery
Defined as the time between skin incision and closure [19].
Anemia
Defined in accordance with World Health Organization (WHO) criteria, defined as a hemoglobin concentration below 12.0 g per deciliter (g/dL) for non-pregnant adult women and below 13.0 g/dL for adult men [20].
Significant intraoperative blood loss
Intraoperative blood loss exceeding 500 mL defined as significant [21, 22].
Major surgery
Defined in accordance with the surgical procedure grading system provided by the National Institute for Health and Care Excellence (NICE) [23]. All major and major-plus surgeries were included, excluding open cardiac surgeries (Table 1).
Table 1.
Classification of surgical procedures by severity, adapted from the National Institute for Health and Care Excellence (NICE) grading system
| Grade & Severity Classification | Representative Procedural Examples |
|---|---|
| Grade 1 (Minor) | Excision of lesion of skin; drainage of breast abscess, cataract, etc. |
| Grade 2 (Intermediate) |
Primary repair of inguinal hernia; excision of varicose vein (s) of leg; knee arthroscopy, patella fracture tension band wiring, hemorrhoidectomy, etc. |
| Grade 3 (Major) |
Total abdominal hysterectomy; Transurethral resection of the prostate (TURP); lumbar discectomy; thyroidectomy, Laparoscopic/open-cholecystectomy/appendectomy, mastoidectomy, Open reduction internal fixation (ORIF) of femur, tibia, or humerus, etc. |
| Grade 4 (Major+) | Total joint replacement, Pulmonary resection (e.g., lobectomy, wedge resection), colonic resection, cholecystectomy with CBD exploration, radical neck dissection; Major neurosurgical procedures (e.g., craniotomy); radical head and neck, gastrointestinal, urological cancer surgeries, etc. |
Statistical analysis and interpretation of results
The analytic dataset, extracted from the secure REDCap platform, underwent validation, coding, and was subsequently imported into STATA version 17.0 (Stata Corp LLC) for statistical analysis. Sociodemographic and baseline clinical characteristics were summarized descriptively in tabular and graphical formats. Categorical data are presented as frequency (percentage), while continuous variables are reported as mean ± standard deviation or median [interquartile range], for non-normally distributed data, based on formal assessment of distribution.
Survival outcomes were compared using Kaplan-Meier estimates and the log-rank test. The determinants of 28-day perioperative mortality were assessed with a multivariable Cox proportional hazards regression model, the appropriateness of which was validated by the non-significance of Schoenfeld residuals. Covariate effects are reported as adjusted hazard ratios (aHR) with corresponding 95% confidence intervals (CI). To analyze time-to-discharge while accounting for the competing risk of postoperative mortality, a Fine-Gray sub distribution hazards regression model was employed. This model specifically treats hospital discharge as the primary event of interest and death as the competing risk, which precludes subsequent discharge. This approach avoids the bias inherent in conventional Cox regression that treats such events as non-informative censoring. The model yields sub-distribution hazard ratios (sHRs) for covariate effects on the cumulative incidence of discharge, and was used to derive the cumulative incidence function (CIF) to illustrate the probability of discharge over time amid competing risks. The CIF was plotted to visually represent discharge dynamics.
For the analysis of PLOS, which is a non-negative integer count variable exhibiting a positively skewed distribution with evidence of overdispersion (variance substantially exceeding the mean). Negative binomial regression was employed as the primary modeling approach. To justify this choice, a preliminary Poisson regression model was fitted, followed by a formal assessment for overdispersion A likelihood ratio test was conducted to compare the Poisson model against the negative binomial model (which includes an additional dispersion parameter, θ or alpha, to account for extra-Poisson variation). The test yielded a highly significant result (P < 0.001), confirming the presence of overdispersion and indicating that the negative binomial model provided a significantly better fit to the data. All models were fitted using maximum likelihood estimation. Potential confounders and clinically relevant covariates were included based on univariate screening (P < 0.20) and a priori knowledge from the literature. Model adequacy was evaluated using goodness-of-fit statistics, including Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC), with lower values favoring the negative binomial specification. Residual diagnostics and predicted versus observed plots were examined to ensure appropriate fit. Results from this model are presented as adjusted incidence rate ratios (aIRR) with 95% CI.
Multicollinearity among predictor variables was assessed using variance inflation factors (VIF). For the negative binomial and Cox proportional hazards models, the mean VIF values were 1.15 and 1.11, respectively, confirming the absence of substantial multicollinearity. Statistical significance was defined as a two-sided p-value < 0.05.
Results
Demographic characteristics of the study subjects
Eligibility assessment screened 1,036 patients, of whom 1,014 were included. Twenty-two (2.2%) patients were excluded due to missing in preoperative and postoperative variables (Fig. 1).
Fig. 1.
Flowchart depicting the number of cohorts included in the study (2019–2022; n = 1014)
Exclusions were due to follow-up failure from invalid contact information (n = 4) or incomplete data (n = 18). Incomplete data included missing biochemical profiles (n = 5) and undocumented intraoperative metrics, primarily intraoperative blood loss (n = 13), of whom four also lacked records for surgical/anesthetic duration. Among participants, 67.55% were aged 65–74 years, representing the majority of the cohort. The median age of participants in this study was 70 years. The cohort was predominantly male (75.0%). Participants were recruited in nearly equal proportions from each study site: 52.56% from TGCSH and 47.44% from ACSH (Table 2).
Table 2.
Sociodemographic characteristics of the study cohort (2019–2022; n = 1,014)
| Variable | Category | Frequency (n) | Percentage (%) |
|---|---|---|---|
| Gender | Male | 760 | 74.9 |
| Female | 254 | 25.1 | |
| Age | 65–74 | 685 | 67.6 |
| 75–84 | 280 | 27.6 | |
| ≥ 85 | 49 | 4.8 | |
| Hospital | TGCSH | 533 | 52.6 |
| ACSH | 481 | 47.4 |
Clinical and admission characteristics
Comorbidities were documented in 57.6% (n = 584) of cases, with hypertension (32.8%) and diabetes (14%) being the most prevalent (Fig. 2).
Fig. 2.
Prevalence of coexisting diseases in elderly patients (2019–2022; n = 1014
The majority of the participants (52.3%) were classified as ASA physical status II. The majority of procedures (60.1%) were elective. A quarter of surgeries (25.8%) were for trauma patients. General surgery procedures accounted for the largest proportion of cases (31.8%), followed by Urology surgery (25.7%). Preoperative anemia was observed in 36.9% of the study participants (Table 3).
Table 3.
Preoperative clinical profile of participants (2019–2022; n = 1014)
| Variable | Category | Frequency (n) | Percentage (%) |
|---|---|---|---|
| ASA | I | 319 | 31.4 |
| II | 532 | 52.3 | |
| ≥III | 165 | 16.3 | |
| Trauma | Yes | 262 | 25.8 |
| No | 752 | 74.2 | |
| Type of surgery | General | 322 | 31.8 |
| Orthopedics | 193 | 19.0 | |
| Neuro | 117 | 11.5 | |
| Urology | 261 | 25.7 | |
| Gynecology | 67 | 6.6 | |
| Other | 54 | 5.3 | |
| Urgency | Emergency | 405 | 39.9 |
| Elective | 609 | 60.1 | |
| Anemia | Yes | 374 | 36.9 |
| No | 640 | 63.1 | |
| Transfusion | Yes | 74 | 7.3 |
| No | 940 | 92.7 | |
| Preoperative-oxygen support | Yes | 94 | 9.3 |
| No | 920 | 90.7 | |
| Baseline HR | Mean ± S. D | 81.5 ± 16.4 | |
| Baseline SBP | Mean ± S. D | 132.3 ± 21.3 | |
| Baseline DBP | Mean ± S. D | 77.6 ± 11.3 | |
| Baseline Spo2 | Mean ± S. D | 96.9 ± 3.0 | |
Intraoperative management
General anesthesia was used in 53.8% of cases, compared to 46.2% for regional anesthesia. Most surgeries (69.5%) were performed on weekdays during daytime hours. The median durations of surgery and anesthesia were 90 and 100 min, respectively. The mean estimated blood loss was 328.1 ± 265.7 mls. The WHO surgical safety checklist was utilized in the vast majority of procedures (94.8%). Postoperatively, 12.7% of patients required supplemental oxygen within the first 48 h (Table 4).
Table 4.
Summary of intraoperative procedures and early postoperative interventions (2019–2022; n = 1014)
| Variable | Category | Summary |
|---|---|---|
| Type of anesthesia | General | 546(53.8%) |
| Regional | 468(46.2%) | |
| Timing of surgery | Weekdays | 603(59.5%) |
| Night time | 266(26.2%) | |
| Weekend | 145(14.3%) | |
| WHO surgical safety checklist | Done | 961(94.8%) |
| Not done | 53(5.2%) | |
| Postoperative Oxygen supplement (48 h) | Yes | 129(12.7%) |
| No | 885(87.3%) | |
| Duration of surgery | Median | 90 min |
| Duration of anesthesia | Median | 100 min |
| Total intravenous fluid | (Mean (SD) | 1714.6 ± 1044.5 mL |
| Estimated blood loss | (Mean (SD) | 328.1 ± 265.7 mL |
28-day perioperative mortality and determinants
Patients were followed for 28 days, giving a total of 7442 person-days observation. The overall 28-day perioperative mortality rate among geriatric surgical patients was 2.86% (29/1014; 95% CI: 1.9–4.0). Of all mortalities, 86.2% were in-hospital events, with the remaining 13.8% occurring after discharge. The overall incidence rate of death was 3.9 per 1000 person-days (95% CI:2.7–5.6%). The median time to mortality was six days (IQR = 3–9) (Fig. 3).
Fig. 3.
The Kaplan-Meier failure estimates of mortality in elderly patients (2019–2022; n = 1014)
Multivariable Cox regression analysis identified two factors significantly associated with increased mortality (p < 0.05): the presence of coexisting diseases and the use of general anesthesia. After adjusting for other variables, patients with comorbidities had a 3.1-fold higher mortality compared to those without comorbidities (aHR = 3.1, 95% CI: 1.43–6.64, p < 0.01). The use of general anesthesia, relative to regional anesthesia, was associated with a substantially greater hazard, corresponding to a 6.12-fold increase in mortality (aHR = 6.12, 95% CI: 2.08–17.97, p < 0.01) (Table 5).
Table 5.
Determinants of 28-day perioperative mortality in geriatric patients: univariable and multivariable cox proportional hazards regression (2019–2022; n = 1014)
| Variable | Category | Survival status | cHR (95%CI) | aHR (95%CI) | |
|---|---|---|---|---|---|
| Died | Censored | ||||
| Comorbidities | Yes | 25 | 559 | 3.21(1.52–6.84) | 3.10(1.43–6.64) * |
| No | 4 | 426 | 1 | 1 | |
| Type of anesthesia | General | 23 | 519 | 5.24(1.82–15.01) | 6.12(2.08–17.97) * |
| Regional | 6 | 466 | 1 | 1 | |
| ASA | 1 | 4 | 314 | 1 | 1 |
| ≥ 2 | 25 | 671 | 0.54(0.22-1.0) | 0.49(0.22–1.10) | |
| Anemia | Yes | 8 | 366 | 0.77(0.36–1.63) | 1.14(0.52–2.48) |
| No | 21 | 619 | 1 | 1 | |
| Intraoperative blood loss | > 500 mL | 6 | 196 | 0.67(0.27–1.67) | 0.74(0.29–1.90) |
| ≤ 500 mL | 23 | 778 | 1 | 1 | |
| Emergency | Yes | 24 | 381 | 1.25(0.59–2.64) | 0.66(0.30–1.44) |
| No | 5 | 604 | 1 | 1 | |
cHR Crude Hazard Ratio, CI Confidence Interval, aHR Adjusted Hazard Ratio, *p-value < 0.05
Postoperative length of hospital stays and its determinants
The median PLOS was six days (IQR: 4–10), with a cohort total of 7,442 days (95% CI: 7159.2–7724.8), corresponding to a mean of 7.34 days per patient. The duration of hospitalization ranged from 1 to 28 days, with 224 patients (22.1%) experiencing a prolonged stay exceeding 10 days (75th percentile).
Evaluation of the cumulative incidence of discharge, accounting for the competing risk of 28-day perioperative mortality, showed a numerical difference between men and women. By postoperative day 5, the cumulative incidence of discharge was 36.2% (95% CI: 30.7–42.5) in women and 44.3% (95% CI: 40.8–47.9) in men. By day 10, these proportions increased to 73.8% (95% CI: 68.2–79.1) for women and 78.1% (95% CI: 75.0–81.0) for men. However, this observed difference was not statistically significant in the Fine-Gray model (Wald test p = 0.136) (Fig. 4).
Fig. 4.
Cumulative incidence of discharge among elderly patients taking death as a competing risk (2019–2022; n = 1014)
Negative binomial regression modeling of PLOS identified several significant independent determinants. In the Univariable analysis, several factors were identified as determinants of PLOS with p-value ≤ 0.2. After adjustment for other variables in the Multivariable model, the following factors remained independently associated with a significantly increased incidence rate ratio (IRR) for longer PLOS. ASA physical status, emergency surgery, the presence of coexisting diseases, significant intraoperative blood loss and duration of anesthesia were identified as determinants of PLOS. The presence of one or more comorbidities was associated with a 31% increase in PLOS (aIRR = 1.31, 95% CI: 1.21–1.41, p < 0.01) compared to patients without comorbidities. Higher ASA physical status predicted progressively longer postoperative stays; relative to ASA I patients, PLOS was 31% longer for ASA II (aIRR = 1.31, 95% CI: 1.20–1.43, p < 0.001) and 38% longer for ASA ≥ III patients (aIRR = 1.38, 95% CI: 1.24–1.54, p < 0.001). Intraoperative blood loss exceeding 500 mL was associated with a 27% increase in PLOS than cases with blood loss below this threshold (aIRR = 1.27, 95% CI: 1.17–1.38, p < 0.001). Each additional 60 min of anesthesia duration was associated with a 15% increase in the expected PLOS (aIRR = 1.15, 95% CI: 1.01–1.30, p = 0.039). Emergency surgeries were also associated with a 28% increase in PLOS (aIRR = 1.28, 95% CI: 1.12–1.38, p < 0.001) compared to elective surgeries. Conversely, higher preoperative hemoglobin was protective, with each 1 g/dL increase associated with a 3% reduction in PLOS (aIRR = 0.97, 95% CI: 0.96–0.99, p = 0.002) (Table 6).
Table 6.
Determinants of postoperative length of hospital stay: univariable and multivariable negative binomial regression analysis in a cohort of geriatric surgical patients (2019–2022; n = 1014)
| Variable | Univariable analysis | Multivariable analysis | |
|---|---|---|---|
| cIRR | aIRR (95% CI) | p-value | |
| Comorbidity (Yes)b | 1.39(1.29–1.51) | 1.31(1.21–1.41) | < 0.001* |
| Emergency (Yes)b | 1.4(1.33–1.54) | 1.28(1.12–1.38) | < 0.001* |
| (ASA class) IIb | 1.33(1.31–1.57) | 1.31(1.20–1.43) | < 0.001* |
| ASA class IIIb | 1.60(1.44–1.79) | 1.38(1.24–1.54) | < 0.001* |
| Hemoglobin (g/dL) a | 0.97(0.96- 0.99) | 0.97 (0.96–0.99) | 0.002* |
| Intraoperative blood loss (mL)>500mLb | 1.48(1.37–1.60) | 1.27(1.17–1.38) | < 0.001* |
| Anesthesia duration (per 60 min) a | 1.03(0.96–1.11) | 1.15(1.01–1.30) | 0.039* |
| Surgery duration (per 60 min) a | 1.00(1.00- 1.01) | 0.95(0.84–1.08) | 0.558 |
| Type of anesthesia(general)b | 1.03(0.96–1.11) | 0.93 (0.86- 1.00) | 0.053 |
aIRR adjusted incidence rate ratio, cIRR crude incidence rate ratio
a = continuous
b =categorical
* statistically significant, aIRR > 1 indicates longer expected PLOS, aIRR < 1 indicates shorter PLOS
Discussion
This prospective study identified key modifiable and non-modifiable determinants of 28-day perioperative mortality and postoperative length of hospital stay among geriatric patients undergoing major surgery. The observed mortality rate of 2.86% (95% CI: 1.9–4.0) aligns with some prior evidence [24] but varies from reports in more selected cohorts, such as those limited exclusively to oncology or emergency surgeries [25–27]. This observed heterogeneity in perioperative mortality rates is most plausibly attributable to differences in cohort composition. Unlike the reference studies, which were limited to narrower subpopulations (e.g., exclusive oncology, very elderly or emergency surgery patients), the present study encompassed a broad, heterogeneous geriatric surgical population.
Our analysis identified two primary determinants of 28-day perioperative mortality: general anesthesia and preexisting comorbidities. General anesthesia was associated with a six-fold increased hazard of mortality compared to regional techniques (aHR = 6.12, 95% CI: 2.08–17.97). This finding aligns with previous evidence [28]. Regional anesthesia likely reduces risk by attenuating the surgical stress response, thereby lowering the incidence of perioperative complications, especially in high-risk patients with chronic illness [29]. In contrast, general anesthesia is associated with higher rates of complications, including postoperative pulmonary complications, prolonged ventilatory dependence, and unplanned intensive care unit admission [30]. The strong association between general anesthesia and mortality is susceptible to confounding by indication, as this technique is typically required for more complex and invasive procedures. Our statistical adjustments cannot fully account for this inherent risk, meaning the hazard ratio represents a combination of anesthetic and procedural risk. Preexisting comorbidities were also a significant risk factor, reinforcing the well-established connection between a patient’s physiological burden and postoperative survival [31]. Comorbidities act as a central mediating mechanism, whereby the cumulative burden of chronic diseases progressively erodes multisystem physiological reserves [32]. These conditions can increase mortality risk during follow-up through pathways independent of the surgery itself. Nevertheless, it remains plausible that surgical intervention may influence the complex interplay and pathophysiological balance among coexisting multi-organ diseases [33, 34].
PLOS is a critical indicator of clinical efficiency and resource utilization [5]. For geriatric surgical patients, prolonged hospitalization strongly elevates the risk of hospital-acquired complications [35]. In our cohort, 22.1% of patients experienced a prolonged PLOS. The determinants of prolonged PLOS identified here can be categorized as modifiable and non-modifiable. Among modifiable factors, preoperative hemoglobin level was a key predictor. Each 1 g/dL increase was associated with a 3% reduction in PLOS, aligning with a substantial body of evidence [36–40]. Preoperative anemia was highly prevalent in this cohort (36.9%), representing a major and modifiable target for prehabilitation. This finding is consistent with prior Ethiopian studies [41]. In geriatric patients, anemia is frequently rooted in nutritional deficits or chronic disease [42]. It is a well-documented risk factor for frailty, functional decline, elevated fall risk, and cognitive impairment [43], justifying targeted intervention. Furthermore, significant intraoperative blood loss is associated with extended PLOS in geriatric patients. Significant intraoperative blood loss is often managed with transfusion [44], which itself is correlated with a higher incidence of postoperative complications and prolonged recovery periods [45].
Key non-modifiable determinants included clinical acuity and baseline physiology. Emergency admission was strongly associated with increased PLOS, a consistent finding attributed to the lack of opportunity for preoperative optimization [46]. In modern medicine, geriatric interventions like Comprehensive Geriatric Assessment and Enhanced Recovery After Surgery protocols are standard practice to speed up postoperative recovery and shorten hospital stays [15]. However, the urgent nature of these procedures leaves no time to identify or manage pre-existing health conditions and functional limitations, directly contributing to prolonged recovery and PLOS [47, 48]. Similarly, we observed a graded association between higher ASA physical status and increased PLOS. After adjustment, patients classified as ASA II had a 31% longer stay, and those classified as ASA ≥ III had a 38% longer stay compared to the reference group. This observation is consistent with prior research on how ASA status influences postoperative recovery and hospital duration [49–52].The presence of comorbidities significantly prolonged PLOS. Patients with at least one comorbidity had a 31% longer PLOS than those without comorbidities (aIRR = 1.31, 95% CI: 1.21–1.41). These data underscore the preeminent role of preoperative. Physiological reserve in determining postoperative recovery trajectory, thereby validating and extending prior epidemiological observations [53–55]. Finally, anesthesia duration was a technical determinant, For every additional 60 min of anesthesia duration, PLOS increased by 15%, consistent with prior research on operative complexity and tissue trauma [56].
The determinants of 28-day mortality and PLOS identified in this study should be viewed in the context of Ethiopia’s surgical ecosystem, which faces persistent structural challenges. The health system is characterized by a critically low procedure rate (43 per 100,000 population), inequitable access to specialist care, and pervasive infrastructure deficits, this system is acutely sensitive to inefficiencies [57–61]. In this context, prolonged PLOS functions not merely as a patient outcome but as a primary system stressor. It directly constrains bed availability, disrupts surgical scheduling, and necessitates last-minute cancellations, imposing substantial burdens on patients and the health system [8, 59]. Ultimately, by blocking essential care pathways, it likely exacerbates the risk of preventable mortality, further straining systemic resilience [62]. A crucial consideration is that these insights derive from two tertiary referral centers managing a high volume of complex cases. Therefore, while the findings elucidate important modifiable and non-modifiable risk factors, their direct generalizability to primary and general surgical settings in Ethiopia and similar low-resource contexts may be limited, given probable differences in patient acuity, resource availability, and surgical complexity.
Strength and limitation of study
This study possesses several strengths, particularly its two-center design and a sufficiently representative sample size, which enhance the generalizability of the findings. Nevertheless, important limitations must be acknowledged. Despite statistical adjustment for known confounders, residual unmeasured confounding remains a plausible concern. Furthermore, the comparatively low number of mortality events, while sufficient for our final model, limits the statistical precision of our estimates and precludes more complex multivariable modeling or subgroup analyses.
Conclusion
This prospective study of Ethiopian geriatric surgical patients confirms that perioperative mortality and prolonged hospital stay are multifactorial. Mortality is strongly linked to the presence of comorbidities and the use of general anesthesia, often a marker of complex surgery. A prolonged postoperative stay is independently driven by patient physiology (ASA status, anemia, comorbidities), surgical urgency, and procedural factors (anesthesia duration, blood loss). These findings create a risk-stratification framework. Preoperative optimization of anemia and chronic conditions emerges as a critical, an actionable focus for improving outcomes in resource-constrained surgical systems.
Acknowledgements
We extend our sincere gratitude to Impact Africa for providing access to the data and for their permission to publish these research findings. We are also profoundly grateful to the dedicated data collectors and information management team, whose meticulous work made this study possible.
Abbreviations
- AHR
Adjusted hazard ratio
- aIRR
Adjusted incidence rate ratio
- CHR
Crude hazard ratio
- cIRR
Crude incidence rate ratio
- IQR
Interquartile range
- IRR
Incidence rate ratio
- PLOS
Postoperative length of hospital stay
- SD
Standard deviation
Authors’ contributions
**Sitotaw Tesfa Zegeye** : Conceptualization, Methodology, Investigation, Data Curation, Formal Analysis, Writing – Original Draft. **Masresha Gebru Teklehaimanot** : Conceptualization, Methodology, Investigation, Project Administration, Writing – Original Draft. **Fekrey Berhe Gebru** : Methodology, Validation, Writing – Review & Editing, Supervision. **Bantayehu Sileshi** : Funding Acquisition, Methodology, Formal Analysis, Writing – Review & Editing, Supervision. All authors: reviewed and approved the final manuscript.
Funding
We received no financial aid from any funder.
Data availability
The de-identified participant data that support the findings of this study, including data on demographics, clinical variables, and outcomes, are available from the corresponding author up on reasonable request.
Declarations
Ethics approval and consent to participate
All procedures were conducted in strict compliance with the ethical principles set forth in the Declaration of Helsinki. Ethical approval for the study was obtained from the Institutional Review Board (IRB) of the College of Medicine and Health Sciences, Bahir Dar University (Reference No.: 0163/2018). The requirement for written informed consent was waived by the IRB of the College of Medicine and Health Sciences, Bahir Dar University and by TGCSH.
Consent for publication
Not applicable
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Lim B-G, Lee I-O. Anesthetic management of geriatric patients. Korean J anesthesiology. 2020;73(1):8–29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Kotekar N, Shenkar A, Hegde AA. Anesthesia issues in geriatrics. Anesthesiology: A Practical Approach. Springer; 2018. pp. 795–825.
- 3.Wolfe JD, Wolfe NK, Rich MW. Perioperative care of the geriatric patient for noncardiac surgery. Clin Cardiol. 2020;43(2):127–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Barbosa TA, Souza AMFd, Leme FCO, Grassi LDV, Cintra FB, Gumieiro DN. Perioperative complications and mortality in elderly patients following surgery for femoral fracture: prospective observational study. Revista brasileira de anestesiologia. 2020;69:569–79. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Saripella A, Wasef S, Nagappa M, Riazi S, Englesakis M, Wong J, et al. Effects of comprehensive geriatric care models on postoperative outcomes in geriatric surgical patients: a systematic review and meta-analysis. BMC Anesthesiol. 2021;21(1):127. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Payton P, Shook JE. Perioperative understanding of geriatric patients. Clin Podiatr Med Surg. 2019;36(1):131–40. [DOI] [PubMed] [Google Scholar]
- 7.Cannoodt L, Knickman J. The effect of hospital characteristics and organizational factors on pre-and postoperative lengths of hospital stay. Health Serv Res. 1984;19(5):561. [PMC free article] [PubMed] [Google Scholar]
- 8.Hurisa Dadi H, Habte N, Mulu Y. Length of hospital stay and associated factors among adult surgical patients admitted to surgical wards in Amhara Regional State Comprehensive Specialized Hospitals, Ethiopia. PLoS ONE. 2024;19(8):e0296143. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Lisk R, Yeong K, Fluck D, Robin J, Fry CH, Han TS. An orthogeriatric service can reduce prolonged hospital length of stay in hospital for older adults admitted with hip fractures: a monocentric study. Aging Clin Exp Res. 2023;35(12):3137–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Awad A, Bader–El–Den M, McNicholas J. Patient length of stay and mortality prediction: a survey. Health Serv Manage Res. 2017;30(2):105–20. [DOI] [PubMed] [Google Scholar]
- 11.Gruskay JA, Fu M, Bohl DD, Webb ML, Grauer JN. Factors affecting length of stay after elective posterior lumbar spine surgery: a multivariate analysis. Spine J. 2015;15(6):1188–95. [DOI] [PubMed] [Google Scholar]
- 12.Devalapalli AP, Kashiwagi DT. Perioperative care of geriatric patients. Hosp Pract. 2020;48(sup1):26–36. [DOI] [PubMed] [Google Scholar]
- 13.Miller RL, Barnes JD, Mouton R, Braude P, Hinchliffe R. Comprehensive geriatric assessment (CGA) in perioperative care: a systematic review of a complex intervention. BMJ open. 2022;12(10):e062729. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Sibia US, MacDonald JH, King PJ. Predictors of hospital length of stay in an enhanced recovery after surgery program for primary total hip arthroplasty. J Arthroplast. 2016;31(10):2119–23. [DOI] [PubMed] [Google Scholar]
- 15.Hirani R, Podder D, Stala O, Mohebpour R, Tiwari RK, Etienne M. Strategies to Reduce Hospital Length of Stay: Evidence and Challenges. Medicina. 2025;61(5):922. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Siddique SM, Tipton K, Leas B, Greysen SR, Mull NK, Lane-Fall M, et al. Interventions to reduce hospital length of stay in high-risk populations: a systematic review. JAMA Netw open. 2021;4(9):e2125846–e. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Abuzied Y, Maymani H, AlMatouq B, AlDosary O. Reducing the length of stay by enhancing the patient discharge process: using quality improvement tools to optimize hospital efficiency. Global J Qual Saf Healthc. 2021;4(1):44–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Healy GL, Stuart CM, Dyas AR, Bronsert MR, Meguid RA, Anioke T, et al. Association between postoperative complications and hospital length of stay: a large-scale observational study of 4,495,582 patients in the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) registry. Patient Saf Surg. 2024;18(1):29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Reames BN, Bacal D, Krell RW, Birkmeyer JD, Birkmeyer NJ, Finks JF. Influence of median surgeon operative duration on adverse outcomes in bariatric surgery. Surg Obes Relat Dis. 2015;11(1):207–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Beutler E, Waalen J. The definition of anemia: what is the lower limit of normal of the blood hemoglobin concentration? Blood. 2006;107(5):1747–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.McGillicuddy EA, Schuster KM, Davis KA, Longo WE. Factors predicting morbidity and mortality in emergency colorectal procedures in elderly patients. Arch Surg. 2009;144(12):1157–62. [DOI] [PubMed] [Google Scholar]
- 22.Chiappa A, Zbar AP, Bertani E, Biella F, Audisio RA, Staudacher C. Surgical outcomes for colorectal cancer patients including the elderly. Hepatogastroenterology. 2001;48(38):440–4. [PubMed] [Google Scholar]
- 23.Karim HMR, Sahoo SK, Prakash A, Rajaram N, Kumar S, Narayan A, editors. Abnormal routine preoperative test results and their perioperative anesthetic impact in patients aged 60 years and more: An observational study. The Indian Anaesthetists Forum; 2018.
- 24.Turrentine FE, Wang H, Simpson VB, Jones RS. Surgical risk factors, morbidity, and mortality in elderly patients. J Am Coll Surg. 2006;203(6):865–77. [DOI] [PubMed] [Google Scholar]
- 25.Novello M, Mandarino FV, Di Saverio S, Gori D, Lugaresi M, Duchi A et al. Post-operative outcomes and predictors of mortality after colorectal cancer surgery in the very elderly patients. Heliyon. 2019;5(8):e02363. https://www.sciencedirect.com/science/article/pii/S2405844019360232. [DOI] [PMC free article] [PubMed]
- 26.Lees MC, Merani S, Tauh K, Khadaroo RG. Perioperative factors predicting poor outcome in elderly patients following emergency general surgery: a multivariate regression analysis. Can J Surg. 2015;58(5):312. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Hamel MB, Henderson WG, Khuri SF, Daley J. Surgical outcomes for patients aged 80 and older: morbidity and mortality from major noncardiac surgery. J Am Geriatr Soc. 2005;53(3):424–9. [DOI] [PubMed] [Google Scholar]
- 28.Desai V, Chan PH, Prentice HA, Zohman GL, Diekmann GR, Maletis GB, et al. Is anesthesia technique associated with a higher risk of mortality or complications within 90 days of surgery for geriatric patients with hip fractures? Clin Orthop Relat Research®. 2018;476(6):1178–88. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Roy RC. Choosing general versus regional anesthesia for the elderly. Anesthesiol Clin North America. 2000;18(1):91–104. [DOI] [PubMed] [Google Scholar]
- 30.Hausman MS Jr, Jewell ES, Engoren M. Regional versus general anesthesia in surgical patients with chronic obstructive pulmonary disease: does avoiding general anesthesia reduce the risk of postoperative complications? Anesth Analgesia. 2015;120(6):1405–12. [DOI] [PubMed] [Google Scholar]
- 31.Endeshaw AS, Kumie FT, Debas SA, Sileshi B. Perioperative outcomes among older surgical patients with multimorbidity: a longitudinal study from Ethiopia. BMC Public Health. 2025;25(1):2641. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Chou W-C, Chang P-H, Lu C-H, Liu K-H, Hung Y-S, Hung C-Y, et al. Effect of comorbidity on postoperative survival outcomes in patients with solid cancers: a 6-year multicenter study in Taiwan. J Cancer. 2016;7(7):854. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Fabbian F, De Giorgi A, Ferro S, Lacavalla D, Andreotti D, Ascanelli S, et al. editors. Post-operative all-cause mortality in elderly patients undergoing abdominal emergency surgery: role of Charlson Comorbidity Index. Healthcare: MDPI; 2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Thanakiattiwibun C, Siriussawakul A, Virotjarumart T, Maneeon S, Tantai N, Srinonprasert V, et al. Multimorbidity, healthcare utilization, and quality of life for older patients undergoing surgery: a prospective study. Medicine. 2023;102(13):e33389. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Liu H, Xing F, Jiang J, Chen Z, Xiang Z, Duan X. Random forest predictive modeling of prolonged hospital length of stay in elderly hip fracture patients. Front Med. 2024;11:1362153. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Fontanals M, O’Leary JD, Zaarour C, Skelton T, Faraoni D. Preoperative anemia increases the risk of red blood cell transfusion and prolonged hospital length of stay in children undergoing spine arthrodesis surgery. Transfusion. 2019;59(2):492–9. [DOI] [PubMed] [Google Scholar]
- 37.Kunz JV, Spies CD, Bichmann A, Sieg M, Mueller A. Postoperative anaemia might be a risk factor for postoperative delirium and prolonged hospital stay: A secondary analysis of a prospective cohort study. PLoS ONE. 2020;15(2):e0229325. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Endeshaw AS, Diress FT, Endeshaw MS, Kefale WM, Addisu BT, Demilie AE, et al. The relationship between preoperative anemia and length of hospital stay among patients undergoing orthopedic surgery at a teaching hospital in Ethiopia: a retrospective cohort study. Perioperative Med. 2024;13(1):120. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Willems JM, de Craen AJ, Nelissen RG, van Luijt PA, Westendorp RG, Blauw GJ. Haemoglobin predicts length of hospital stay after hip fracture surgery in older patients. Maturitas. 2012;72(3):225–8. [DOI] [PubMed] [Google Scholar]
- 40.Elsamadicy AA, Adogwa O, Ongele M, Sergesketter AR, Tarnasky A, Lubkin DE, et al. Preoperative hemoglobin level is associated with increased health care use after elective spinal fusion (≥ 3 levels) in elderly male patients with spine deformity. World Neurosurg. 2018;112:e348–54. [DOI] [PubMed] [Google Scholar]
- 41.Beyable AA, Berhe YW, Nigatu YA, Tawuye HY. Prevalence and factors associated with preoperative anemia among adult patients scheduled for major elective surgery at University hospital in Northwest Ethiopia; a cross-sectional study. Heliyon. 2022;8(2):e08910. 10.1016/j.heliyon.2022.e08910. [DOI] [PMC free article] [PubMed]
- 42.Sim YE, Abdullah HR. Implications of anemia in the elderly undergoing surgery. Clin Geriatr Med. 2019;35(3):391–405. [DOI] [PubMed] [Google Scholar]
- 43.Girelli D, Marchi G, Camaschella C. Anemia in the elderly. Hemasphere. 2018;2(3):e40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Desai N, Schofield N, Richards T. Perioperative patient blood management to improve outcomes. Anesth Analgesia. 2018;127(5):1211–20. [DOI] [PubMed] [Google Scholar]
- 45.Mikhail C, Pennington Z, Arnold PM, Brodke DS, Chapman JR, Chutkan N, et al. Minimizing blood loss in spine surgery. Global spine J. 2020;10(1suppl):S71–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Issa ME, Al-Rashedy M, Ballester P, Ammori BJ. Predictors of duration of postoperative hospital stay in patients undergoing advanced laparoscopic surgery. Surg Laparoscopy Endoscopy Percutaneous Techniques. 2005;15(2):90–3. [DOI] [PubMed] [Google Scholar]
- 47.Merani S, Payne J, Padwal RS, Hudson D, Widder SL, Khadaroo RG. Predictors of in-hospital mortality and complications in very elderly patients undergoing emergency surgery. World J Emerg Surg. 2014;9(1):43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Ibitoye S, Braude P. Geriatric perioperative care. Medicine. 2021;49(1):51–5. [Google Scholar]
- 49.Papalia R, Zampogna B, Torre G, Papalia GF, Vorini F, Bravi M, et al. Preoperative and perioperative predictors of length of hospital stay after primary total hip arthroplasty—our experience on 743 cases. J Clin Med. 2021;10(21):5053. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Ahmad M, Shaikh A, Ahmad AS, Awan N. Enhanced recovery program–impact of ASA grade on length of hospital stay in patients undergoing hip and knee arthroplasties. Surgeon. 2023;21(1):16–20. [DOI] [PubMed] [Google Scholar]
- 51.Kastanis G, Topalidou A, Alpantaki K, Rosiadis M, Balalis K. Is the ASA score in geriatric hip fractures a predictive factor for complications and readmission? Scientifica. 2016;2016(1):7096245. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Schneider AM, Denyer S, Brown NM. Risk factors associated with extended length of hospital stay after geriatric hip fracture. JAAOS Global Res Reviews. 2021;5(5):e21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Inneh IA, Iorio R, Slover JD, Bosco JA III. Role of sociodemographic, co-morbid and intraoperative factors in length of stay following primary total hip arthroplasty. J Arthroplast. 2015;30(12):2092–7. [DOI] [PubMed] [Google Scholar]
- 54.Farley KX, Anastasio AT, Premkumar A, Boden SD, Gottschalk MB, Bradbury TL. The influence of modifiable, postoperative patient variables on the length of stay after total hip arthroplasty. J Arthroplast. 2019;34(5):901–6. [DOI] [PubMed] [Google Scholar]
- 55.Rodrigues LP, de Oliveira Rezende AT, Delpino FM, Mendonça CR, Noll M, Nunes BP, et al. Association between multimorbidity and hospitalization in older adults: systematic review and meta-analysis. Age Ageing. 2022;51(7):afac155. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56.Phan K, Kim JS, Kim JH, Somani S, Di’Capua J, Dowdell JE, et al. Anesthesia duration as an independent risk factor for early postoperative complications in adults undergoing elective ACDF. Global spine J. 2017;7(8):727–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 57.Burssa D, Teshome A, Iverson K, Ahearn O, Ashengo T, Barash D, et al. Safe surgery for all: early lessons from implementing a national government-driven surgical plan in Ethiopia. World J Surg. 2017;41(12):3038–45. [DOI] [PubMed] [Google Scholar]
- 58.Alidina S, Sydlowski MM, Ahearn O, Andualem BG, Barash D, Bari S, et al. Implementing surgical mentorship in a resource-constrained context: a mixed methods assessment of the experiences of mentees, mentors, and leaders, and lessons learned. BMC Med Educ. 2022;22(1):653. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 59.Corren YB, Merrill J, Wilkinson R, Cannon C, Bickel J, Reis BY. Predicting surgical department occupancy and patient length of stay in a paediatric hospital setting using machine learning: a pilot study. BMJ Health Care Inf. 2022;29(1):e100498. [Google Scholar]
- 60.Meshesha BR, Sibhatu MK, Beshir HM, Zewude WC, Taye DB, Getachew EM, et al. Access to surgical care in Ethiopia: a cross-sectional retrospective data review. BMC Health Serv Res. 2022;22(1):973. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 61.Alemayehu TY, Boled AT, Taye MG, Kibret S, Habtie MF. Assessment of emergency and essential surgical service in primary hospitals of South Gondar zone, Amhara region, Ethiopia, 2019. Int J Surg Open. 2020;23:35–8. [Google Scholar]
- 62.Gebregzi AH, Teko EB, Tantu AD, Jale NT, Getahun GK, Asemu YM. Assessment of surgical capacity and productivity in high-volume Ethiopian hospitals: mixed method study. BMC Health Serv Res. 2025;25(1):760. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The de-identified participant data that support the findings of this study, including data on demographics, clinical variables, and outcomes, are available from the corresponding author up on reasonable request.