NephJC will do it's fourth article next Tuesday and it is an article that breaks the mold. Clinical practice guidelines, like review articles and editorials are usually not included in the cannon of journal club. However we want NephJC to expand these horizons in order to tackle any document that may influence nephrology. Clinical practice guidelines probably have more influence over medicine than any single article and because of that they deserve the same critical eye that research articles do.
The Subject of this week's NephJC is: Clinical practice guideline on diagnosis and treatment of hyponatraemia. It was published in NDT, The European Journal of Endocrinology and Intensive Care Medicine.
The guideline weighs in at 47 pages but the core guideline is actually only a very meaty 30 pages. It is clearly written by committee with a beautiful contradiction on the third page:
So this CPG is either a definition of the standard of care or its not, who knows.
The scope is hypo-osmolar hyponatremia in adults. The authors start out and focus on the elephant in the room, hyponatremia is defined by an abnormal lab test but the vast majority of patients are not symptomatic and it is unclear how important correcting the sodium is. They created a hierarchy of outcomes which seems well thought out.
The cut-off date for the guidelines was December 2012. They searched the usual subjects for articles. Then they threw in this little nugget, sounds like a sure fire method to generate bias:
We excluded case series that reported on benefit if the number of participants was under 5 but included even individual case reports if they reported an adverse event.
Recommendations were categorized using the GRADE system.
 |
| To interpret the guidelines to come you should major in Grade D, with a minor in Grade C. |
After dispensing with the beurocracy of how the guidelines were written, they start with a review of the physiology of water handling and the etiology of hyponatremia. It is a well written and solid text. A worthy addition to any resident's Evernote.
Hyponatremia occurs in 30% of hospitalized patients. That data comes from Hoorn's exhaustive, but single center, trial at the Erasmus Medical Center in Rotterdam.
According to the American Hospital Association, there are 36 million admissions every year, if we extrapolate Hoorn's data to that number we get 10.8 million U.S. cases of hyponatremia a year and 1 million U.S. cases of profound hyponatremia a year. Keep those numbers in mind as you read the evidence behind the comprehensive guidelines to follow.
The clinical features of hyponatremia.
The authors discuss the commonly expressed symptoms of acute hyponatremia but then spend an entire paragraph focusing on the subtle symptoms of chronic hyponatremia that may easily be missed by history or physical:
Although the more severe signs of acute hyponatraemia are well established, it is now increasingly clear that even patients with chronic hyponatraemia and no apparent symptoms can have subtle clinical abnormalities when analysed in more detail. Such abnormalities include gait disturbances, falls, concentration and cognitive deficits (13). In addition, patients with chronic hyponatraemia more often have osteoporosis and more frequently sustain bone fractures than normonatraemic persons (14, 15, 16). Finally, hyponatraemia is associated with an increased risk of death (17, 18). Whether these are causal associations or merely symptoms of underlying problems such as heart or liver failure remains unclear (19).I wrote an editorial about that last reference for eAJKD.
There is a little blurb about pseudohyponatremia, despite telling us that this guideline will only cover hypo-osmotic hyponatremia. They point out that despite common opinion that this lab abnormality is no longer a problem that it still continues today:
Pseudohyponatraemia was seen more frequently with flame photometric measurement of serum sodium concentration than it is now with ion-selective electrodes, but despite common opinion to the contrary, it still occurs (30), because all venous blood samples are diluted and a constant distribution between water and the solid phase of serum is assumed when the serum sodium concentration is calculated (30)Thought the "it still occurs" loses some punch when one references an 11 year old article.
They have a tidy review of the criteria for SIADH:
Although included in many diagnostic algorithms, hypothyroidism very rarely causes hyponatraemia (49). In 2006, Warner et al. (50) observed that serum sodium concentration decreased by 0.14 mmol/l for every 10 mU/l rise in thyroid-stimulating hormone, indicating that only severe cases of clinically manifest hypothyroidism resulted in clinically important hyponatraemia. Development of hyponatraemia may be related to myxoedema, resulting from a reduction in cardiac output and glomerular filtration rate (51).Then they have a shout out to icodextrin-based PD fluid inducing hyponatremia. This was on my nephrology recertification boards.
I love this figure explaining how the two stimuli for ADH release, volume status and hyperosmolality, interact:
Depending on the kidney’s ability to dilute urine, 50–100 mmol of solutes, such as urea and salts, are required to remove 1 l of fluid. If solute intake is low relative to water intake, the number of available osmoles can be insufficient to remove the amount of water ingested.
Guidelines for the diagnosis of the etiology of hyponatremia
Finally on page 17 of the guideline we get the first guideline: The Classification of hyponatremia, they have three different systems for classifying hyponatremia, none of them based on volume status: |
| Table 5. What? No hiccups? |
- Biochemical severity
- Mild: Na 130-135
- Moderate: 125-130
- Profound: less than 125
- Time-based
- Acute: duration less than 48 hours
- Chronic: duration at least 48 hours
- Symptom based
- moderately symptomatic (based on Table 5)
- Severely symptomatic (based on Table 5)
An explanation for why they went with Mild, Moderate and Profound (rather than severe):
Authors mostly use the terms ‘mild’, ‘moderate’ and ‘severe’ (61, 62, 63). We have chosen to replace ‘severe’ by ‘profound’ to avoid confusion with the classification based on symptoms, for which we have reserved the term ‘severe’.
Note to authors: changing the classic hierarchy of mild moderate and severe with mild moderate and profound does not avoid confusion.
In regards to symptoms, the authors explain that they very specifically excluded a category of asymptomatic:
The authors defend not using volume status to categorize hyponatremia at least partly due to the pathetic job physicians do at assessing volume status:
Another interesting statement was in regards to using the urine osmolality to determine if ADH is present or not, It is clear that ADH is present when the urine osmolality is greater than serum osmolality and it is clear that ADH is absent when urine osmolality is less than 100 mOsm/Kg, but they identify a grey area from a urine osmolality of 100 to a urine osmolality that isotonic where they say one is unable to definitely say there is ADH or not. (refs: Nephron Physiology and Journal of Emergency Medicine) I found myself in that no man's land in this case.
Urine sodium was examined as a test to separate volume depleted hypovolemia from SIADH. Urine sodium less than 30 was highly sensitive for hypovolemia (0.87-1.0) even in the one series that looked at patients on diuretics separately. However urine sodium has disappointing specificity (i.e. low urine sodium in a patient with SIADH) (0.24 to 0.83).
The authors gave cautious enthusiasm for the fractional excretion of urea less than 12% indicating SIADH. One case series had the FEUrea with a sensitivity of 0.86 and specificity of 1.0.
It is interesting though, that when they made their flow chart, the lack of precision of urine sodium, urine osmolality, physician ability to discriminate volume status are all thrown out the window. Flow charts are loved by students but never allow for the subtle Bayesian realities of clinical medicine.
They use a flow chart to map out when to use the various treatment recommendations. This seems like an intelligent to build what is essentially a table of contents for using the guidelines. Nice job.In regards to symptoms, the authors explain that they very specifically excluded a category of asymptomatic:
We have purposefully omitted the category ‘asymptomatic’ as we believed this might create confusion. Patients are probably never truly ‘asymptomatic’ in the strictest sense of the word. Very limited and subclinical signs such as mild concentration deficits are seen even with mild hyponatraemia (13).They use the Hillier correction factor for hyperglycemia (add 2.4 mmol/L per 100 mg/dl of glucose). I think that is wrong.
The authors defend not using volume status to categorize hyponatremia at least partly due to the pathetic job physicians do at assessing volume status:
We found two studies indicating that in patients with hyponatraemia, clinical assessment of volume status has both low sensitivity (0.5–0.8) and specificity (0.3–0.5) (89, 103). Similarly, it seems that clinicians often misclassify hyponatraemia when using algorithms that start with a clinical assessment of volume status (88). Using an algorithm in which urine osmolality and urine sodium concentration are prioritized over assessment of volume status, physicians in training had a better diagnostic performance than senior physicians who did not use the algorithm (104).Hard to believe that nephrologists are no better at determining hypovolemia than a coin toss. Incredible.
Another interesting statement was in regards to using the urine osmolality to determine if ADH is present or not, It is clear that ADH is present when the urine osmolality is greater than serum osmolality and it is clear that ADH is absent when urine osmolality is less than 100 mOsm/Kg, but they identify a grey area from a urine osmolality of 100 to a urine osmolality that isotonic where they say one is unable to definitely say there is ADH or not. (refs: Nephron Physiology and Journal of Emergency Medicine) I found myself in that no man's land in this case.
Urine sodium was examined as a test to separate volume depleted hypovolemia from SIADH. Urine sodium less than 30 was highly sensitive for hypovolemia (0.87-1.0) even in the one series that looked at patients on diuretics separately. However urine sodium has disappointing specificity (i.e. low urine sodium in a patient with SIADH) (0.24 to 0.83).
The authors gave cautious enthusiasm for the fractional excretion of urea less than 12% indicating SIADH. One case series had the FEUrea with a sensitivity of 0.86 and specificity of 1.0.
It is interesting though, that when they made their flow chart, the lack of precision of urine sodium, urine osmolality, physician ability to discriminate volume status are all thrown out the window. Flow charts are loved by students but never allow for the subtle Bayesian realities of clinical medicine.
Treatment Guidelines
Treatment of patients with severe symptoms from hyponatremia:
If they don't get better:
If you can't tell if they got better, assume they did.
This is how I put these recommendations together:
It was interesting to see them review all of the cases of osmotic demyelination, to paraphrase Churchill:
Never has so much been done by so many to avoid a complication that occurs in so few.
Some characteristics of the people who developed osmotic demyelination:
The speed limit for correction of hyponatremia:
Treatment of patients with moderately severe symptoms from hyponatremia:
Remember moderate symptoms include headache, nausea (without vomiting), and confusion. I would be reluctant to use 3% in a lot of patients that look like they fit this recommendation. They even recommend 3% saline in some cases of for acute yet asymptomatic patients. The speed limit, 10 mmol/L the first day and 8 mmol/L per day after that remains the same.
Treatment of patients without severe of moderately severe symptoms from hyponatremia:
What the rest of the world calls asymptomatic hyponatremia.
They sure like their 3%. Here again, even in asymptomatic patients they recommend possibly giving 3%. The most surreal moment in this clinical practice guideline is when they reference a randomized trial of 8 ultra-marathoners in the Western States Endurance Run. Like this has any relevance to my patients in the hospital, where endurance is measured in steps, not miles.
Treatment of patients with chronic hyponatremia without severe or moderately severe symptoms:
They divide their recommendations into general recommendations, and then recommendations for patients with volume expansion, SIADH, and reduced circulating volume.
7.4.1. general management. They recommend against treating the serum sodium if it is above 130. This implies that sodiums below 130 should be corrected for the sake of the number alone.
In 7.4.2 they drill down into patients with expanded extracellular volume. Here they advise against treating sodiums above 125. They also recommend against using vasopressin antagonists.
This seems crazy to me. Vaptans are the best studied drugs in the whole field of hyponatremia. No other therapy for low sodiums has any type of reasonable rigorous clinical trials to support its use. Look at the description of the studies they sight to support the use of hypertonic saline in acute symptomatic hyponatremia:
It's all case series and retrospective crap. Compare that to the thousands of patients who have used vaptans safely to raise their sodiums. The authors reference two reviews, the first with 15 RCTs of vaptans with 1,619 patients and the other with 11 RCTs and 1,094 patients. In neither of these reviews were there any episodes of osmotic demyelination despite increased risk for overly rapid correction of sodium with vaptans. The authors found 5 more trials outside of those reviews bringing the total number of patients examined in RCTs to 2,900. The authors own analysis of the data found a signal indicating a possible increased risk of death in hypervolemic patients. See appendix 6:
 |
| appendix 1-5 then 7 and then 6. Cause that's how European's count. |
So what do they mean by signal indicating a possible increase risk of death? Here is the forrest plot from the appendix:
1.19.4 is the analysis they called out. Three studies, two using oral lixivaptan and one using satavaptan. I guess I was reassured to see that this signal (code for non-signifigant data) does not come from any of the licensed vaptans in the US. One of the reason it feels a bit like a hit job is they could have just as easily pointed out this chart where it looks like there is a survival signal for conivaptan.
Of note: the manufacturer of tolvaptan, has warned that some patients with rapid increases of sodium due to tolvaptan have suffered neurologic sequelae, though the authors could not find any reports in the literature.
The next guidelines are for SIAD where again they recommend against using vaptans, but now add lithium or demeclocycline to the list. They recommend using fluid restriction and if that doesn't work, use urea, or the combination of a loop diuretic and salt tablets. No randomized controlled trials of any of these therapies were found. In addition they reported two cases of osmotic demyelination in patients who had fluid restriction, though it seems unlikely that fluid restriction was the primary culprit.
In their defense of the guidelines they had these wonderful paragraphs:
The last section of the guideline addresses what to do if the sodium goes up too fast.
And that is the last guideline. I look forward to discussing this on Tuesday night at NephJC at 9pm eastern.
