Acid–base disturbances, potassium maladjustments, volume dysregulation and osmoregulatory disturbances occur not uncommonly concomitantly and are interrelated. A sound knowledge of electrolyte pathophysiology is important and access to some simple calculations is helpful. A more practicable formula for predicting responses to fluid administration has been proposed that is a genuine assistance in fluid therapy planning [1,2]. We recently encountered a patient who was a model example in this regard.
A 38‐year‐old male was admitted because of weakness and vomiting. He was a binge drinker and claimed that he had vomited for 3 days. During this time, he refrained from alcohol and ingested only small amounts of tea. He had a prior history of a seizure disorder, ostensibly related to alcohol. The patient was oriented but somnolent. His blood pressure was 140/80 mmHg and the heart rate was 80 beats/min. The respiratory rate was 10 breaths/min. A neurologist found no localizing findings. The heart, chest and abdomen were unremarkable except for diminished bowel sounds. There was no oedema and the muscle stretch reflexes were depressed. When moved to an upright position, the patient's blood pressure decreased to 85/60 mmHg and the heart rate increased.
Haemoglobin was 13 g/dl, haematocrit was 39 vol/%, white blood cells numbered 200 000 µ3. Liver enzymes, bilirubin and amylase were slightly elevated. The following values were recorded: creatinine kinase 1742 U, myoglobin 696 µg/l, C‐reactive protein 102 mg/l, creatinine 272 µmol/l, glucose 4 mmol/l and urea 15 mmol/l. The serum electrolytes were (mmol/l): Na 113, Cl 47, K 2.2, Mg 0.86, Ca 1.94 and P 1.63. The arterial pH was 7.58, PaO2 69 mmHg, PaCO2 69 mmHg and HCO3 65 mmol/l. An electrocardiogram was unremarkable, except for a QTc interval of 0.51 s.
Since the patient was conscious, no bladder catheter was placed. We gave him 2 l volume replacement according to the regimen outlined below and 2 h later the patient produced a urine sample with a pH 7, 433 Uosm, 22 mmol/l Na, 0 mmol/l Cl, 66 mmol/l K and 8 mmol/l creatinine. We repeated the serum electrolytes concomitantly and the values were (mmol/l): Na 114, K 2.6, Cl 71 and HCO3 51.
What is the primary disturbance? Is the compensatory response appropriate?
What is responsible for the hypokalaemia?
What is responsible for the hyponatraemia? Can we estimate free water and electrolyte‐free water clearance to gain some further insight into basic mechanisms?
The patient has profound alkalaemia. Since the PaCO2 and HCO3 are elevated, metabolic alkalosis is the only diagnostic option. In metabolic alkalosis, for every mmol/l increase in the serum bicarbonate, an appropriate respiratory compensation would be an increase in the PaCO2 of 0.7 mmHg. The patient's HCO3 is increased by 40 mmol/l. Thus, a 30 mmHg increase in the PaCO2 is perfect compensation. The patient has decreased his alveolar ventilation by 75%, explaining his decreased respiratory rate. The record‐setting decrease in the Cl concentration, coupled with the orthostatic blood pressure decrease make volume contraction with hypochloraemia highly likely [3,4].
The urine specimen was extremely helpful. Our patient could give no urine initially and we were reluctant to catheterize him when the procedure was not clinically absolutely necessary. We elected to treat volume contraction, since we considered volume depletion the most likely life‐threatening problem. The relationship ΔNa=(infusate Na−serum Na)/(total body water+1) led us to conclude that normal saline would be a good choice in terms of volume and hyponatraemia [1,2]. We were also concerned about his hypokalaemia and prolonged QTc interval. Finally, we felt that his chloride deficit had to be replaced. Parenteral omeprazol was given. From the relationship ΔNa=[(infusate Na+infusate K)−serum Na]/(total body water+1), we calculated that 1 l 0.9% saline (Na=154 mmol/l) with 20 mmol KCl in each litre would result in the following: ΔNa=[(154+20)−111]/(40+1)=1.6 mmol/l ΔNa [1,2].
When the patient made urine 2 h later, we calculated the FENa at 0.6%. The FENa showed that the patient's azotaemia was pre‐renal and strongly corroborated our clinical impressive of severe volume contraction. The urine chloride concentration of zero was extremely helpful. This value told us that the patient's kidneys were resorbing all the sodium they could with the resorbable anion, chloride. Thus, the vomiting was solely responsible for the chloride (and hydrogen ion) loss. Frequently, the source of the chloride loss is not as obvious as it was in our patient. Some persons ingest loop and thiazide diuretics surreptitiously. In such cases, the urine chloride may be elevated depending upon the last dose of drug. Patients can be extremely resourceful in hiding surreptitious drugs. Drug metabolites are helpful here. Patients that induce vomiting chronically as part of an eating disturbance may have decreased or absent enamel on their teeth, particularly the inner aspects. Occasionally, adults lose hydrogen ions in the form of chloridorrhoea from a villous adenoma; however, chloridorrhoea is generally seen in infants and is related to a genetic defect in the Cl/HCO3 exchanger . Lastly, some patients will have Bartter's syndrome. Three genes have been found responsible. Each genetic discovery has contributed substantially to our understanding of ascending limb function.
Metabolic alkalosis can come about through a gain of bicarbonate or a loss of hydrogen ions [3,4]. A gain of bicarbonate outside the hospital is uncommon, although rarely, bicarbonate gluttons ingesting NaHCO3 or CaCO3 (milk‐alkali syndrome) are identified. Decreased renal function is the rule in such patients, since with a normal glomerular filtration rate, renal bicarbonate excretion is excellent. Loss of hydrogen ions is common and occurs via the gastrointestinal tract or kidneys. Hydrogen ions are lost and bicarbonate generation is increased when mineralocorticoid effects are increased. In primary aldosteronism, Cushing's syndrome and monogenic forms of hypertension, metabolic alkalosis and hypokalaemia occur . However, the extracellular fluid volume and blood pressure are elevated and the urinary chloride is not zero, but rather reflects dietary intake. Our patient had signs consistent with volume contraction.
Patients with severe vomiting will develop metabolic alkalosis only when their renal function is decreased to some degree. Otherwise, they would be able to excrete the generated bicarbonate. With a decreasing serum chloride concentration, the option of resorbing sodium with a resorbable anion becomes progressively problematic. For that reason, the metabolic alkalosis is sustained until the chloride deficit is replaced (Figure 1) [7–9].
Potassium disturbance is best elucidated by the transtubular potassium gradient (TTKG) that is calculated from the relationship: TTKG=UK/(Uosm/Posm)÷serum K. Thus, if urine osmolarity is equal to plasma or serum osmolarity, the calculation is simple. If the urine is concentrated or dilute, this fact must be taken into account to adjust the gradient accordingly. Our patient's TTKG was 1:19. Clearly, the potassium losses occurred through the kidneys and not through vomitus. The loss results from the effort to retain sodium and preserve volume at all costs. Our patient had alkaline urine; however, acid urine (paradoxical aciduria) can also occur. The replacement formula we employed accommodates potassium deficits. Potassium is an ‘effective osmolyte’ since it does not traverse cell membranes unless commanded to do so. Since the intracellular as well as the extracellular space must be repleted in effective osmolytes, the sodium concentration and potassium concentration of the infusate can be summed.
Our patient had hyponatraemia and therefore had a defect in free‐water excretion by definition. Does he have syndrome of inappropriate ADH or ‘sodium depletion’ or both, or does it matter? Our patient had a urine osmolarity of 433 mosm/l. This value is not exactly ‘concentrated urine’; however, it is prima facie evidence that some ADH is present. The formula for free‐water clearance is: Cl(water)=V(1−Uosm/Posm). The plasma osmolarity was 2×(113+2)+4+15=249 mosm/l. We do not know the timed urine volume since we only measured concentrations. However, one thing is very clear, the Cl(water) is a negative number. Thus, our patient must be resorbing free water, namely his Cl(water) is negative. Is this inappropriate ADH? Well, if osmolarity is driving ADH, an appropriate ADH value for our patient would be zero and he should make dilute urine to excrete the excess water. However, our patient is volume contracted and has a decreased glomerular filtration rate, both not features of classical SIADH. Possibly, our patient's ADH response is volume driven, although a 15% decrease in effective circulating fluid volume is necessary for volume‐mediated ADH release. Alternatively, vomiting stimulated ADH release, a not uncommon stimulus, particularly in women.
Osmoles are fascinating. However, our clinical focus should be on ‘effective’ osmoles. When we looked at the ‘effective’ free‐water clearance (electrolyte‐free water clearance) relationship: Cl(water e)=V(1−UNa+UK/SNa), we observed that (1 – 88/114) gives a positive number. In other words, the sum of urine Na and urine K was less than his serum Na. Therefore, his effective clearance of free water was positive and we could expect his serum Na to rise faster than our infusion equation would otherwise predict. If the sum of UNa and UK is greater than SNa, the SNa must fall. If the sum is less than SNa, the SNa must rise. Had we made timed collections of urine (say hourly), we could have estimated exactly how fast SNa will increase. However, our patient had no bladder catheter and could not cooperate to allow timed voidings. If the amount of urine is considerable (say 200 ml/h), the increase in SNa might be such that we might have to consider giving a hyponatraemic patient hypoosmolar solutions. Thus, the free water and ‘electrolyte‐free’ water clearances provide us with completely different important pieces of information. The free‐water clearance tells us whether or not ADH is active. The ‘electrolyte‐free’ water clearance tells us if the serum sodium concentration is going up or down. Both pieces of information are highly clinically important.
In conclusion, our patient had severe contraction metabolic alkalosis on the basis of vomiting. His volume contraction resulted in pre‐renal azotaemia that contributed to his problems. The alkalosis was perfectly compensated since the increase in PaCO2 was ∼0.7 mmHg for every 1 mmol/l increase in HCO3. The chloride deficit was responsible for ‘maintenance’ of the metabolic alkalosis, which will remain until the chloride deficit is repleted. The urine chloride concentration of zero was the key to place the blame on the gastrointestinal tract. The hypokalaemia was engendered by the kidney's determination to maintain circulating fluid volume at all costs. The calculation of the TTKG is the clue to finding the site of the potassium loss. The infusion formula that works for hypo‐ and hypernatraemia, with or without potassium deficits, is a valuable tool to adequately control the therapy [1,2]. The urine Na and urine K concentrations are the key to determining whether or not the serum Na will rise or fall, while the urine osmolality tells us whether or not ADH is present. Our patient's serum phosphate plummeted with volume expansion and caloric substitution. Hypophosphataemia is common in poorly nourished alcoholics and may contribute to muscle weakness, including the respiratory musculature, in such patients . Total body potassium and magnesium deficit was also substantial in our patient.
Finally, there are some flies in our ointment. Calculation of the anion gap in our patient on admission discloses a gap of almost zero. The repeat HCO3 2 h later was 51 mmol/l, which gives a gap in the expected range. Perhaps the machinery is not perfect at such extremes from normal. Calculating pulmonary status from the values on admission is also interesting. If we apply the modified alveolar gas equation (Berlin resides at ∼70 M), we observed the following: PaO2=150−PaCO2×1.25. The equation gives us a PaO2 of 64 mmHg, whereas his measured PaO2 was 69 mmHg, suggesting that he may be producing O2 and secreting it over his alveolae! Numbers do not always fit the expected physiology, but in our patient they were pretty close. He recovered uneventfully after ∼8 l of fluid substitution guided by the formula provided.
Correspondence and offprint requests to: Ralph Kettritz, Franz Volhard Clinic, Wiltberg Strasse 50, D‐13125 Berlin, Germany. Email: kettritz@fvk‐berlin.de
We appreciate Friedrich C. Luft's infectious enthusiasm for ‘salt and water’ problems. He claims to have been in turn infected by Jerome Kassirer, Robert Narins and Mitchell Halperin. Finally, we have the sneaking suspicion that Luft actively recruits these patients!