A comparison of statistical methods for modeling count data with an application to hospital length of stay

Relationship between preoperative hemoglobin levels and length of stay in elderly patients with hip fractures: A retrospective cohort study

Globally, hip fractures in elderly individuals are a prevalent and serious issue. Patients typically have a longer length of stay (LOS), which increases the risk of complications and increases hospitalization costs. Hemoglobin (Hb) is a routine blood test that is associated with disease prognosis. This study aimed to investigate the relationship between preoperative Hb and LOS in elderly hip fracture patients and to determine a reliable transfusion threshold. The clinical data of hip fracture patients (aged ≥ 60 years) admitted to the Department of Orthopaedics, Shenzhen Second People's Hospital, between January 2012 and December 2021 were retrospectively analyzed. Multiple linear regression analysis was used to assess the linear relationship between preoperative Hb and LOS. Smooth curve fitting was performed to investigate potential nonlinear relationships. In the case of discovering nonlinear relationships, a weighted two-piecewise linear regression model was built, and the inflection points were determined using a recursive algorithm. Subgroup analyses were conducted based on age and gender. A total of 1444 patients with an average age of (77.54 ± 8.73) years were enrolled. After adjusting for covariates, a nonlinear relationship was found between preoperative Hb and LOS. The two-piecewise linear regression model revealed an inflection point of 10 g/dL. On the left of the inflection point (Hb < 10 g/dL), the LOS was reduced by 0.735 days for every 1 g/dL increase in Hb (β = -0.735, 95% confidence interval: -1.346 to -0.124, P = .019). On the right side of the inflection point (Hb > 10 g/dL), the relationship was not statistically significant (β = 0.001, 95% confidence interval: -0.293 to 0.296, P = .992). In elderly hip fracture patients, there is a nonlinear association between preoperative Hb and LOS. However, when Hb levels were <10 g/dL, there was a negative correlation with the LOS. No correlation was observed when Hb levels were >10 g/dL. These findings underscore the importance of timely intervention to manage Hb levels in elderly patients with hip fractures, potentially reducing hospitalization durations and associated complications.

Socio-environmental predictors of diabetes incidence disparities in Tanzania mainland: a comparison of regression models for count data

BACKGROUND

Diabetes is one of the top four non-communicable diseases that cause death and illness to many people around the world. This study aims to use an efficient count data model to estimate socio-environmental factors associated with diabetes incidences in Tanzania mainland, addressing lack of evidence on the efficient count data model for estimating factors associated with disease incidences disparities.

METHODS

This study analyzed diabetes counts in 184 Tanzania mainland councils collected in 2020. The study applied generalized Poisson, negative binomial, and Poisson count data models and evaluated their adequacy using information criteria and Pearson chi-square values.

RESULTS

The data were over-dispersed, as evidenced by the mean and variance values and the positively skewed histograms. The results revealed uneven distribution of diabetes incidence across geographical locations, with northern and urban councils having more cases. Factors like population, GDP, and hospital numbers were associated with diabetes counts. The GP model performed better than NB and Poisson models.

CONCLUSION

The occurrence of diabetes can be attributed to geographical locations. To address this public health issue, environmental interventions can be implemented. Additionally, the generalized Poisson model is an effective tool for analyzing health information system count data across different population subgroups.

Low back pain-driven inpatient stays in the United States: a nationwide repeated cross-sectional analysis

BACKGROUND

Low back pain (LBP)-driven inpatient stays are resource-intensive and costly, yet data on contemporary national trends are limited.

MATERIALS AND METHODS

This study used repeated cross-sectional analyses through a nationally representative sample (US National Inpatient Sample, 2016-2019). Outcomes included the rate of LBP-driven inpatient stays; the resource utilization (the proportion of receiving surgical treatments and hospital costs) and prognosis (hospital length of stay and the proportion of nonroutine discharge) among LBP-driven inpatient stays. LBP was classified as overall, nonspecific, and specific (i.e. cancer, cauda equina syndrome, vertebral infection, vertebral compression fracture, axial spondyloarthritis, radicular pain, and spinal canal stenosis). Analyses were further stratified by age, sex, and race/ethnicity.

RESULTS

292 987 LBP-driven inpatient stays (weighted number: 1 464 690) were included, with 269 080 (91.8%) of these for specific LBP and 23 907 (8.2%) for nonspecific LBP. The rate of LBP-driven inpatient stays varied a lot across demographic groups and LBP subtypes (e.g. for overall LBP, highest for non-Hispanic White 180.4 vs. lowest for non-Hispanic Asian/Pacific Islander 42.0 per 100 000 population). Between 2016 and 2019, the rate of nonspecific LBP-driven inpatient stays significantly decreased (relative change: 46.9%); however, substantial variations were found within subcategories of specific LBP-significant increases were found for vertebral infection (relative change: 17.2%), vertebral compression fracture (relative change: 13.4%), and spinal canal stenosis (relative change: 19.9%), while a significant decrease was found for radicular pain (relative change: 12.6%). The proportion of receiving surgical treatments also varied a lot (e.g. for overall LBP, highest for non-Hispanic White 74.4% vs. lowest for non-Hispanic Asian/Pacific Islander 62.8%), and significantly decreased between 2016 and 2019 (e.g. for nonspecific LBP, relative change: 28.6%). Variations were also observed for other outcomes.

CONCLUSIONS

In the US, the burden of LBP-driven inpatient stays (i.e. rates of LBP-driven inpatient stays, resource utilization, and prognosis among LBP-driven inpatient stays) is enormous. More research is needed to understand why the burden varies considerably according to the LBP subtype (i.e. nonspecific and specific LBP as well as subcategories of specific LBP) and the subpopulation concerned (i.e. stratified by age, sex, and race/ethnicity).

Association between fat-free mass and survival in critically ill patients with COVID-19: A prospective cohort study

BACKGROUND

Most critically ill patients with COVID-19 experience malnutrition and weight loss associated with negative clinical outcomes. Our primary aim was to assess body composition during acute and late phase of illness in these patients in relation to clinical outcome and secondary to tailored nutrition support.

METHODS

This prospective cohort study included adult critically ill patients with COVID-19. Body composition (fat-free mass [FFM] [exposure of interest], fat mass [FM], skeletal muscle mass [SMM], and phase angle [PA]) was determined with multifrequency bioelectrical impedance analyses in the acute and late phase. Nutrition support data were collected simultaneously. Clinical outcome was defined as intensive care unit (ICU) survival (primary outcome) and 30-90 days thereafter, duration of mechanical ventilation, and length of ICU stay and length of hospital stay (LOS). Nonparametric tests and regression analyses were performed.

RESULTS

We included 70 patients (73% male, median age 60 years). Upon admission, median BMI was 30 kg/m2 , 54% had obesity (BMI > 30 kg/m2 ). Median weight change during ICU stay was -3 kg: +3 kg FM and -6 kg FFM (-4 kg SMM). Body composition changed significantly (P < 0.001). Regarding clinical outcome, only low PA was associated with prolonged LOS (odds ratio = 0.83, 95% CI = 0.72-0.96; P = 0.015). Patients with optimal protein intake (>80%) during acute phase maintained significantly more FFM (2.7 kg, P = 0.047) in the late phase compared with patients who received <80%.

CONCLUSION

FFM decreased significantly during acute and late phase of illness, but we observed no association with ICU survival. Only low PA was associated with prolonged LOS. FFM wasting likely occurred because of disease severity and immobility.

In-hospital complications of work-related musculoskeletal injuries

BACKGROUND, OBJECTIVES

Work-related musculoskeletal (MSK) injuries are a major contributor to morbidity worldwide and frequently result in hospitalisation. Hospital complications are common, costly, and largely preventable, but relevant data is required to address this. This study aimed to identify the incidence and factors associated with in-hospital complications of work-related MSK injuries.

METHODS

This study is based on work-related MSK hospital admission data from Victorian Admitted Episodes Database, 2016-2022. Complications were identified based on ICD-10-AM coding using CHADx (Classification of Hospital Acquired Diagnoses). Negative binomial and logistic regression analyses were performed to identify factors related to in-hospital complications.

RESULTS

In-hospital complications occurred in 6.3 % of work-related MSK injury admissions. In the adjusted models, ages ≥45 years, female sex, and area-level disadvantage were associated with in-hospital complications. Stay at public (vs private) hospitals, comorbidity, emergency admissions, and general anaesthesia were also associated. Complication rates were higher in hospitalised workers with direct head, neck, and trunk injuries and cumulative MSK disorders than those with direct extremities injuries and acute MSK conditions. The most common complications were cardiovascular, gastrointestinal complications and adverse drug events.

CONCLUSION

This study identified patient, injury and hospital-related characteristics associated with in-hospital complications of work-related MSK injuries for informing prevention strategies and risk estimation by hospital staff and workers' compensation schemes. The results demonstrate a sizable rate of complications given the relatively young and healthy study population.

Long-term exposure to air pollution and severe COVID-19 in Catalonia: a population-based cohort study

The association between long-term exposure to ambient air pollutants and severe COVID-19 is uncertain. We followed 4,660,502 adults from the general population in 2020 in Catalonia, Spain. Cox proportional models were fit to evaluate the association between annual averages of PM_2.5, NO₂, BC, and O₃ at each participant's residential address and severe COVID-19. Higher exposure to PM_2.5, NO_2, and BC was associated with an increased risk of COVID-19 hospitalization, ICU admission, death, and hospital length of stay. An increase of 3.2 µg/m³ of PM_2.5 was associated with a 19% (95% CI, 16-21) increase in hospitalizations. An increase of 16.1 µg/m³ of NO₂ was associated with a 42% (95% CI, 30-55) increase in ICU admissions. An increase of 0.7 µg/m³ of BC was associated with a 6% (95% CI, 0-13) increase in deaths. O₃ was positively associated with severe outcomes when adjusted by NO₂. Our study contributes robust evidence that long-term exposure to air pollutants is associated with severe COVID-19.