PREDICTORS OF OUTCOME IN HOSPITALIZED PATIENTS: METHODS

Posted by James

Study Setting

This study was done at Grady Memorial Hospital, a large tertiary teaching hospital , that serves a mostly urban population of city of Atlanta and the surrounding counties of Dekalb and Fulton.

Study Design

A retrospective review of patients identified from archived laboratory data using a computerized laboratory information system. All definitions and diagnostic criteria were defined prospectively.
buy prescription drugs without prescription

Population

Over a four-year period (January 1, 1997 to February 2001), we identified 182 patients with a serum sodium concentration of 115mmmol/L or less seen at this hospital using a computerized laboratory information systems (LIS) database. Two patients in the pediatric age group (<14 years) were excluded. In six patients, repeat serum sodium data were in normal range and these cases were considered to be spurious and, therefore, excluded since no specific therapy was given by the treating physicians. In four other patients, data concerning fluid administration and oxygenation status during hyponatremia were incomplete and these were excluded as well. Finally, two more patients were excluded, because hyponatremia was documented in the emergency room, but admission and follow-up data were unavailable.

A total of 168 patients were then selected for inclusion into the study. The medical records of these patients were reviewed by two of us (HBB and CMN) to obtain data on demographics, laboratory findings, fluid therapy, symptoms of hyponatremic encephalopathy before and after therapy, and outcome using a data abstraction form. Over the study period, 5994 patients at this hospital received the International Classification of Diseases (ICD-9) code for hyponatremia -276.1.

Serum sodium, and urine sodium were measured using a Hitachi 747-200 analyzer (Boehringer Manheim, Indianapolis), while Osmolarity was measured by freezing point depression. Determination of the causes of hyponatremia was based on the opinion of the treating physician and/or nephrologist consulted on the case. The rate of correction of hyponatremia in the first 24 hours was derived by assessment of changes in serum sodium (measured at least twice daily in all patients), over time. Absolute levels of serum sodium after 48 hours of therapy also were noted. The comorbid factors in these patients on admission were recorded on a data abstraction sheet.
canadian pharmacy cialis

DEFINITIONS

Hyponatremia was defined as acute if the duration was less than 48 hours, and chronic if the duration was longer than 48 hours, or if previous laboratory data document hyponatremia in the preceeding days or months that was not completely corrected prior to hospital discharge. Correction of hyponatremia was judged to be overly rapid if an increase of > 25mmmol/L in 48 hours occurred, or if correction to hyperna-tremic levels occurred over the same time. A diagnosis of hyponatremia was made by our hospital laboratory for patients with a serum sodium of less than 133mmol/L.

Hypoxia was defined by a partial pressure of oxygen (PaO²) < 60mmHg on arterial blood gas or pulse oximetry showing oxygen saturation of <90% during the first 24 hours of admission with hyponatremia.

Sepsis was defined by the presence of fever (temperature > 100.5° F) or hypothermia in association with tachycardia, tachypnea and leukocytosis with a suspected or confirmed site of infection. The presence of symptoms such as nausea and vomiting, dizziness, altered mentation or seizure activity were attributed to hyponatremic encephalopathy, unless there was a coexisting medical condition (such as hypoglycemia), or medication effect to account for these symptoms, and especially if the symptoms improved with correction of hyponatremia. Chest radiographic and cerebral imaging data (computed tomographic or magnetic resonance scans) also were reviewed.
generic cialis 20mg

ANALYSIS

Data for continuous measurements are expressed as mean + or – (±) SD. The primary outcome of interest was short-term mortality during admission with severe hyponatremia. Differences across demographic, clinical, and treatment characteristics were assessed for association with death using chi-square or student’s t test. In separate logistic regression models that included all the studied risk factors, we evaluated the relationship of these variables to outcome. A P value of <0.05 was considered significant. The associations of identified variables were further reported as odds ratios. Data were analyzed using the SAS package (SAS Cary, NC).