Thiamin, a water-soluble B-complex vitamin, functions as a coenzyme in macronutrient oxidation and the production of cellular adenosine tri-phosphate (1). Mammals cannot biosynthesize or significantly store thiamin and thus are dependent on continual ingestion. Thiamin deficiency (TD) largely manifests as neurologic (dry beriberi) or cardiovascular (wet beriberi) disease. Thiamin-deficient heart disease is characterized by sodium and water retention, peripheral vasodilatation, and myocardial failure. Therefore, TD would be expected to worsen symptoms in the setting of established congestive heart failure (CHF) (1).

Numerous factors have been reported to increase the risk of developing TD in CHF patients, including diuretic-induced urine thiamin excretion (2- 5), severe heart failure (6), malnutrition (6- 7), advanced age, and frequent hospitalizations (5,8- 12). The prevalence of TD in CHF patients has been reported to range from 3% to 91% (2,5- 7,12- 13). Trials of thiamin supplementation, although clinically important, have been hampered by their small sample sizes, use of indirect measures of thiamin status, and exclusion of patients with less severe disease and those on diuretics other than loop diuretics (2,8,11- 17). Therefore, the objective of this study was to determine the prevalence of TD in a large cross section of hospitalized CHF patients.

In this prospective, cross-sectional, observational study, the prevalence of TD in 100 consecutive patients admitted to the hospital with a primary diagnosis of CHF was determined and compared with that of 50 matched control subjects.

Diagnosis of CHF was made by the attending physician and confirmed based on a left ventricular (LV) ejection fraction (EF) <50%, according to a two-dimensional echocardiogram, or a normal LV function (EF >50%) with classic CHF symptoms according to Framingham criteria (18). Subjects were excluded if they were being treated with high-dose thiamin supplements (200 mg/day) for alcohol abuse or were unable to give informed consent.

Age- (within five years) and gender-matched control subjects were recruited into the study through hospital advertisement in a 2:1 ratio. Control subjects were excluded if they had any known condition that may affect thiamin status or were taking thiamin-containing supplements. Approval for this study was obtained from the St. Michael’s Hospital Research Ethics Board.

Demographic and clinical information (including medication use) was recorded from the medical chart. Oral diuretics were reported as the average daily dose, whereas intravenous diuretics were recorded as twice the actual dose given in order to be equivalent to an oral dose (19).

Within 48 h of admission, fasting blood samples were collected from CHF patients. Blood samples from control subjects were also obtained following an overnight fast. Blood was collected in evacuated tubes with ethylenediamine-tetraacetic acid and centrifuged (within 30 min) for 10 min at 3,000 rpm. Erythrocytes were washed and stored at −70°C for later analysis. Erythrocyte thiamin pyrophosphate (TPP) was determined using high-performance liquid chromatography (HPLC) on an amino column (supelcosil-LC-NH₂, 5 μ particle size, 25.0 cm × 4.6 mm, Supelco product, Sigma Corp., Toronto, Ontario, Canada) according to the methods of Warnock (20). The fluorescence detector was set at an excitation wavelength of 360 nm and an emission wavelength of 440 nm. The coefficient of variation of samples was 5.4%. Thiamin deficiency was defined as an erythrocyte TPP <78 ng/ml packed cells based upon the correlation between erythrocyte TPP concentration and the TPP effect, which indicates TD in alcoholic patients (8,21).

Twenty-four hour urine collections were initiated within 48 h of admission. The urine was immediately transferred to 3-l amber bottles containing 20 ml acetic acid and kept at 4°C, where the volume and creatinine were measured within 24 h; a 50-ml aliquot was stored at −70°C for the determination of urine thiamin concentration.

Urine thiamin was analyzed by HPLC according to the methods developed by Botticher and Botticher (22) using freshly prepared thiamin standards (2 to 30 ng/ml, Sigma Corp.) in 0.1 mol sulphuric acid. Urine thiamin concentration (24 h) was corrected for renal function by calculating urine thiamin as μg/g creatinine. Corrected creatinine clearance (ml/s/SA) was determined using a standard equation (23).

A trained registered dietitian determined nutrition status of all CHF patients using subjective global assessment (SGA) (24). Appetite was assessed using a self-reported rating scale (7).

Since biochemical signs of TD occur within two to three weeks of a TD diet, the usual daily thiamin intake of CHF patients from food and supplements was estimated using a semi-quantitative food frequency questionnaire modified from Block 2000 for the month before hospitalization (25- 26). Thiamin intake was determined using the Nutriwatch nutrient analysis program (^© 1981 to 2001, Elizabeth Warwick, Long Creek, PO Box 506, Cornwall, Prince Edward Island, Canada C0A 1H0), which contained the 1997 Canadian Nutrient File. Adequacy of thiamin intake was determined by comparing the estimated intake to the estimated average requirement (EAR) (17).

Continuous data were expressed as medians (range) and categorical data as frequencies and percentages unless otherwise noted. Comparisons between two groups for categorical data were completed using either the Pearson chi-square test or the Fisher exact test, when counts per cell were <5. The Mann-Whitney U test was used for comparisons between continuous variables.

Predictor variables for TD were determined using logistic regression analysis. All variables found significant on univariate logistic regression with a p <0.2 were considered for multiple regression analysis (27). Highly intercorrelated variables were identified by correlation matrix, and only one of these variables was entered into the multiple regression model using the backward selection method. All statistical analyses were performed using the SPSS for Windows, Version 9.0 software (Statistical Package for the Social Sciences, Chicago, Illinois). The p values were two-tailed, and a p value <0.05 was considered statistically significant.

One hundred CHF patients were recruited into the study from April 2001 to June 2002 (Tables 1, 2). Patients represented a community population admitted to hospital, not necessarily under the care of a cardiologist. Despite matching, CHF patients were significantly older than control subjects. The main cause of CHF was coronary artery disease. Twenty-one CHF patients took thiamin-containing supplements, which were predominately multivitamin/mineral supplements with a thiamin content ranging from 1.5 to 2.25 mg/day. Approximately one-third of CHF patients had diabetes mellitus, and the majority had decreased renal function (corrected creatinine clearance <1.24 ml/s/SA). Although 50% of CHF patients were malnourished (using SGA) and reported decreased appetites, the mean intake of thiamin was adequate compared to the EAR.

The median dose of furosemide at home was 60 mg/day (10 to 360 mg/day), which had been taken for 14 months (10 to 300 months) (Table 3). Twenty-four patients (30%) were prescribed multiple diuretics.

One-third of CHF patients (33%) were found to be TD, which was significantly higher than control subjects (p = 0.007) ((Table 4),Figures 1, 2). This difference was strengthened when supplement users were excluded (37.2% [95% confidence interval (CI) 30% to 45%] vs. 12.2% [95% CI 5% to 20%], respectively, p = 0.002) (Figure 3). Although there was no significant difference in erythrocyte TPP between CHF patients and control subjects (87 [range 45 to 222] ng/ml vs. 97 [range 60 to 203] ng/ml, p = 0.109, respectively), when supplement users were excluded, erythrocyte TPP concentrations were significantly lower in the CHF group (85 [range 45 to 203] ng/ml vs. 97 (range 60 to 203) ng/ml, p = 0.014, respectively). Urine thiamin excretion was 252 μg/g creatinine (60 to 8,640) in CHF patients, which was within the previously reported normal range (22) (Figure 4). Forty percent of patients with grade III/IV LV function were TD compared to 25% of those with grade I/II ventricular function (p = 0.1); this difference was accentuated when supplement users were excluded (49% vs. 24% respectively, p = 0.03). The CHF patients with LV dysfunction were younger (p = 0.018) and mostly male (75%); had a lower body mass index (p = 0.001), more malnutrition (p = 0.01), higher creatinine (p = 0.023), and lower TPP (p = 0.07); and stayed in hospital longer (p = 0.06) than those with grade I/II ventricles.

Pre-admission spironolactone use, preserved renal function, and non-use of thiamin-containing supplements were significantly associated with TD (p < 0.05). Left ventricular function, poor and fair appetite ratings, mild to moderate malnutrition (SGA score B or C), and inadequate thiamin intake tended to be associated with TD (Table 5).

As many variables were intercorrelated, representative variables were selected for multivariate analysis. Thiamin-containing supplement use was chosen over other variables related to thiamin intake (SGA, intake < EAR, appetite rating) as it had the strongest relationship with TD (Table 5). Similarly, urine thiamin excretion (μg/g creatinine) reflected the most accurate estimate of thiamin excretion. Corrected creatinine clearance was unrelated to any other variable and therefore was entered into multivariate analysis (Table 6). Of these three variables, urine thiamin excretion was found to be the strongest negative predictor of TD (p = 0.025 [95% CI 0.94 to 0.99]) (Table 6).

One-third of hospitalized CHF patients at our institution were found to have erythrocyte TPP concentrations indicative of TD. This study represents a large, typical cross section of hospitalized patients in various stages of CHF and on varying types and doses of diuretics using a direct measure of thiamin status.

Previous studies have focused on diuretic-induced urine thiamin excretion as the main cause of TD in CHF patients (2- 5). Increased urinary losses of thiamin in response to diuretic therapy have been observed in both normal and TD rats (3,28), in healthy human volunteers (4), and in CHF patients on high doses of furosemide (2,5). Surprisingly, we failed to find a significant relationship between TD and furosemide dose, urine volume, or urine thiamin excretion. In fact, urine thiamin excretion was the only significant positive predictor of thiamin status, suggesting that increased urine thiamin excretion independently predicts better thiamin status. Since studies in healthy adults suggest that urine thiamin excretion is positively correlated with dietary thiamin intake and blood thiamin concentrations, our urinary thiamin excretion may simply reflect the amount of thiamin consumed in the diet and supplements (Figure 5) (1,25,29- 30). Supporting this relationship is the observed trend between TD and inadequate thiamin intakes, a finding previously observed in CHF patients as well as in healthy volunteers following a thiamin-restricted diet (6,25). This finding is not surprising, as 50% of our CHF patients were malnourished and had decreased appetites. Multivitamin vitamin supplement use was found to be protective against TD, suggesting that even small doses of thiamin (1.5 mg/day) may be effective in reducing rates of TD. Finally, the finding that decreased renal function was significantly associated with better thiamin status in CHF patients implies that decreased renal function may prevent excessive thiamin loss, thereby protecting against TD. Further research is required to confirm our findings regarding renal function and TD. The significant relationship between TD and spironolactone use before hospitalization supports previous findings linking TD and increased disease severity (5,12,31). Furthermore, the increased prevalence of TD in patients with LV dysfunction, despite similar intakes of dietary thiamin, suggests that these patients may represent a higher-risk population for TD.

Our sample size may have been inadequate to determine all the significant relationships between factors and TD, as the sample size was based on an expected prevalence of TD. In addition, urinary thiamin data from control subjects was not collected. Finally, given the nature of our population, we were unable to use the most accurate estimate of usual food intake and were limited to using a semi-quantitative food frequency questionnaire to estimate usual thiamin intake.

This is the largest study of the prevalence of TD in a cross section of typical hospitalized CHF patients with varying degrees of disease severity and taking a variety of diuretics. We found that the prevalence of TD in CHF patients in an acute care hospital is high (>30%). Factors significantly related to TD included pre-admission spironolactone use, preserved renal function, and non-use of thiamin-containing supplements. These factors could be used to identify hospitalized CHF patients at risk for TD. In contrast to previous studies, we found that increased urinary thiamin excretion predicted improved thiamin status rather than TD in hospitalized CHF patients. The protective effect of supplements with relatively low thiamin content suggests that routine thiamin supplementation may preserve thiamin status. This should be considered as routine treatment for hospitalized CHF patients. Future studies are warranted to determine the possible benefits of thiamin supplementation in CHF patients.