Ionized Calcium, Parathormone, and in Critically Ill Surgical Patients
KENNETH W. BURCHARD, M.D.,* DONALD S. GANN, M.D.,t JULIE COLLITON, M.D.,4 and JAMESON FORSTER, M.D.§ From the Departments of Surgery, Dartmouth-Hitchcock Medical Center,* Hanover, New Hampshire; University of Maryland,t Baltimore, Maryland; Rhode Island Hospital/ Brown University,f Providence, Rhode Island; and University of Kansas, § Kansas City, Kansas
A prospective study measured ionized calcium and parathormone sequentially at 48- to 72-hour intervals in 25 surgical intensive care unit patients. Twelve patients (48%) died at mean day 40 and median day 26. Levels of ionized calcium, parathormone, blood urea nitrogen, creatinine, albumin, magnesium, and phosphate for patients who lived were compared with levels for patients who died. The incidence of hypotension, renal failure (creatinine 2 3.0), and bacteremia, as well as the amount of red cell, crystalloid, and colloid administration for the two groups was compared. Hypotension, bacteremia, red cells, crystalloid, and colloid were no different. On days 1 and 2 ionized calcium levels were significantly lower and parathormone levels significantly higher in nonsurviving patients; this difference persisted through days 3 and 4. Blood urea nitrogen and creatinine levels increased early in nonsurviving patients but renal failure, which occurred in nine nonsurviving patients, did not develop until mean day 14, median day 18. The phosphate level was slightly higher but still within normal range in nonsurviving patients. By days 5 and 6 ionized calcium and parathormone levels were no different in nonsurviving patients, despite there being no improvement in renal function. Magnesium and albumin levels were no different between groups. Ionized calcium levels are lower and parathormone levels higher early in nonsurviving patients. This difference is not readily explained by associated clinical conditions, including renal dysfunction. Although etiology remains unclear, low ionized calcium and elevated parathormone are early predictors of mortality in critically ill surgical patients.
T n HE ASSOCIATION OF decreased serum total cal-
cium (TC) and ionized calcium (IC) levels with critical illness is well documented.'" Because calcium is important for many cellular activities (e.g., electrical conduction, contractility, enzyme and hormone activity), reduced calcium concentration has been sugPresented at the 1 10th Annual Meeting of the American Surgical Association, Washington, D.C., April 5-7, 1990. Address reprint requests to Kenneth W. Burchard, M.D., Section of General Surgery, Department of Surgery, Dartmouth-Hitchcock Medical Center, Hanover, NH 03756. Accepted for publication April 12, 1990.
gested to cause decreased cellular and organ performance, particularly cardiovascular dysfunction.23,61,12 Hypovolemic hypoperfusion, sepsis, hypomagnesemia, rhabdomyolysis, and renal failure, as well as albumin and red cell transfusions, are all associated with diminished IC but are difficult to consolidate into one pathogenic The frequent coexistence of exmechanism tracellular conditions fostering hypocalcemia (elevated phosphate levels, citrate infusion, 'relative' hypoparathyroidism, albumin infusion) with diseases that permit calcium migration into cells (hemorrhagic hypoperfusion, sepsis, pancreatitis, muscle injury) accentuates this difficulty.'1,3,101215-18 The clinical significance of hypocalcemia also is difficult to assess. Patients with the conditions listed above have a significant mortality risk and the relationship of hypocalcemia and parathormone (PTH) response to this risk is controversial.4"0 Similarly controversial is the benefit of calcium administration, especially when intracellular calcium levels may be increasing.'0"12"19-22 Our interest in calcium and PTH in critical illness began with a study of hypercalcemia in patients documented to have had multiple previous episodes of hypocalcemia.7 23 This study and several others have used one-time or retrospective measures of calcium and PTH and therefore do not describe the 'natural history' of these variables as they relate to clinical factors and prognosis. The purpose of the present investigation was to study IC and PTH sequentially in critically ill surgical patients. Prospective, sequential data would be expected to clarify relationships with clinical factors and illness severity.
BURCHARD AND OTHERS
Materials and Methods Patient Selection All patients admitted to the Surgical Intensive Care Unit (SICU) in the Rhode Island Hospital during a 3-month period were considered for study. Twenty-five consecutive patients expected to require SICU management for more than 72 hours were selected regardless of admitting diagnosis or surgical interventions.
Ionized Calcium and Parathormone Measurements Serum IC level was determined using the NOVA ion selective electrode (NOVA Biomedical, Newton, MA), whose normal range is 4.8 to 5.2 mg/dL and coefficient of variance is 0.7%. Parathormone level was determined using the PTH-MM mid molecule radioimmunoassay (INCSTAR, Stillwater, MN), whose normal range is 29 to 85 pmol/L, minimum detectable level is 10 pmol/L, and coefficient of variance is 9.5%. Samples were collected within 48 hours of admission and thereafter at 48- to 72hour intervals until discharge from the SICU. Most of the data were collected during the first 10 SICU days and provide the measurements for the analysis of variance (ANOVA) matrix described below. Data collected after 10 days were too few to be used for this analysis but were combined with the first 10 days' data for analysis by t test.
Clinical Data Clinical data recorded prospectively were age; sex; pridiagnosis by disease category (e.g., gastrointestinal, trauma); surgery by category (e.g., gastrointestinal, cardiovascular); episodes of hypotension (systolic blood pressure less than 90 mmHg) and bacteremia; red cell transfusion; intake of total parenteral nutrition (TPN); incidence of renal failure (creatinine 2 3.0 mg/dL); and serum magnesium (normal, 1.2 to 2.3 mg/dL), serum phosphate (normal, 2.7 to 4.5 mg/dL), albumin (normal, 3.4 to 4.8 g/dL), and TC (normal, 8.5 to 10.5 mg/dL) levels. Data collected retrospectively were daily crystalloid and colloid administration, urine output, antibiotics administered, serum blood urea nitrogen (BUN), and serum creatinine. mary
Data Analysis For ANOVA data were organized using 48-hour groupings (i.e., days 1 and 2, 3 and 4, 5 and 6, 9 and 10), which allowed potentially 8 data points per patient and 200 data points for each measured variable. Because not all of the data were collected each day, no variable was measured 200 times. For the data listed under results, the total number of data points available for ANOVA is given
Ann. Surg. * October 1990
in the tables. For instance, of 200 potential creatinine determinations, 142 were available for analysis. Data for surviving patients compared with data for those who did not survive during this hospital admission were analyzed using ANOVA corrected for repeated measures with individual means compared using the Newman-Keuls method.24 Linear regression analysis was used to correlate individual PTH and IC values in both patient groups. Data collected after 10 days were too few for ANOVA. To analyze all IC and PTH data gathered in the study, including that beyond the first 10 ICU days, a t test was used. Data are presented as mean plus or minus standard error. Data were considered significantly different when p