A Multistate Model Predicting Mortality, Length of Stay, and Readmission for Surgical Patients.

OBJECTIVE: Simultaneously evaluate postoperative mortality, length of stay (LOS), and readmission. DATA SOURCE: National Surgical Quality Improvement Program (NSQIP). DESIGN: Retrospective cohort. METHODS: Data from elective general surgical patients were obtained from the 2012 NSQIP Participant Use File. For each postoperative day, each patient's state was classified as index hospitalization, discharged home, discharged to long-term care (LTC), readmitted, or dead. Transition rates were estimated using exponential regression, assuming constant rates for specified time periods. These estimates were combined into a multistate model, simulated results of which were compared to observed outcomes. FINDINGS: Age, comorbidities, more complex procedures, and longer index LOS were associated with lower rates of discharge home and higher rates of death, discharge to LTC, and readmission. The longer patients had been discharged, the less likely they were to die or be readmitted. The model predicted 30-day mortality 0.38 percent (95 percent CI: 0.36-0.41), index LOS 2.85 days (95 percent CI: 2.83-2.86), LTC discharge 2.76 percent (95 percent CI: 2.69-2.82), and readmissions 5.53 percent (95 percent CI: 5.43-5.62); observed values were 0.39 percent, 2.82 days, 2.87 percent, and 5.70 percent, respectively. CONCLUSIONS: Multistate models can simultaneously predict postoperative mortality, LOS, discharge destination, and readmissions, which allows multidimensional comparison of surgical outcomes.

Rural and urban traffic fatalities, vehicle miles, and population density.

The purpose of this study was to evaluate the effect of population density on the rates of motor vehicle mortality in rural and urban areas, while controlling for vehicle miles traveled (VMT). Rural and urban data for traffic mortality, VMT, and population were obtained for each state from the Federal Highway Administration for 1998-2000. Linear regression was used to estimate the effect of population density, VMT per capita, southern location, and presence of a trauma system on mortality. Variation in rural mortality rate (per 100,000 population) was proportional to rural VMT per capita, but population density and southern location were also independent predictors, together accounting for 91% of this variation. Variation in urban mortality rates was not affected by population density, but urban rates were also higher in the south. The exposure-based rural mortality rate (deaths per 100 million VMT) was inversely proportional to population density, which along with southern location explained 41% of the variation from state to state. The presence of a state trauma system did not measurably affect mortality. After controlling for VMT and southern location, state population density was a moderately strong predictor of rural but not urban traffic mortality rates.

Predicted effect of automatic crash notification on traffic mortality.

OBJECTIVE: To estimate the reduction in traffic mortality in the United States that would result from an automatic crash notification (ACN) system. METHODS: 1997 Fatality Analysis Reporting System (FARS) data from 30,875 cases of incapacitating or fatal injury with complete information on emergency medical services (EMS) notification and arrival times were analyzed considering cases at any time to be in one of four states: (1) alive prior to notification; (2) alive after notification; (3) alive after EMS arrival; and (4) dead. For each minute after the crash, transition probabilities were calculated for each possible change of state. These data were used to construct models with (1) number of incapacitating injuries ranging from FARS cases up to an estimated total for the US in 1997; (2) deaths equal to FARS total; (3) transitions to death from other states proportional to FARS totals and rates and (4) other state transitions equal to FARS rates. The outcomes from these models were compared to outcomes from otherwise identical models in which all notification times were set to 1 min. RESULTS: FARS data estimated 12,823 deaths prior to notification, 1800 after notification, and 14,015 between EMS arrival and 6 h. If notification times were all set to 1 min, a model using FARS data only predicted 10,703 deaths prior to notification, 2,306 after notification, and 15,208 after EMS arrival, while a model using an estimated total number of incapacitating injuries for the US predicted 9,569 deaths prior to notification, 2,261 after notification, and 15,134 after arrival. In the first model, overall mortality was reduced from 28,638 to 28,217 (421 per year. or 1.5%), while in the second model mortality was reduced to 26,964 (1,674 per year, or 6%). CONCLUSIONS: Modest but important reduction in traffic mortality should be expected from a fully functional ACN system. Imperfect systems would be less effective.

Initial presentation of older injured patients to high-volume hospitals is not associated with lower 30-day mortality in Medicare data.

OBJECTIVE: To evaluate whether survival of older patients with severe injuries is positively associated with initial presentation to high-volume trauma hospitals. DESIGN: Historical cohort study. SETTING: We analyzed Medicare fee-for-service records. Cases were classified by maximum Abbreviated Injury Score (AISmax); those with isolated hip fractures or AISmax <3 were excluded. The initial hospital (emergency department or inpatient) for each case was classified by its number of included inpatient cases. PATIENTS: Patients aged >or=65 with principal injury diagnoses (ICD-9 800-959, excluding 905, 930-939, 958) admitted to hospitals or who died in emergency departments during 1999. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Thirty-day mortality was determined using Medicare denominator data and modeled as a function of hospital volume, AISmax, age, gender, and comorbidity. We found that 95,867 patients (74,894 AISmax = 3; 17,932 AISmax = 4; 3,041 AISmax = 5) were managed in 4,391 hospitals. More than 90% of the interhospital transfers were from emergency departments, mostly from low-volume to high-volume hospitals, and were more frequent with greater severity. Regression models showed no difference in 30-day survival between patients taken first to low-volume hospitals (and possibly transferred) vs. patients taken directly to high-volume hospitals. Prior studies showing a positive or negative effect of hospital volume on survival of older patients could be replicated but their findings could not be generalized. CONCLUSIONS: Existing systems of trauma care result in similar survival for older patients with serious injuries seen first at low-volume or high-volume hospitals.