An investigation of high lithium concentrations

The effects of drugs are studied In the laboratory, In cllnclal trials and In day to day use. The latter has come to be regarded as clinical audit. Many publications attest to the beneficial and toxic effects of lithium, and to the narrow dose range between the two1.Although serum lithium concentration is the best monitor of lithium toxicity, it is only an indirect measure; in particular, in unsteady state conditions it will not be an accurate index of tissue toxicity, and high lithium levels may precede (Sellers et al 1982) or follow (Amdisen, 1980) intoxication. Furthermore, patients as well as their individual organs vary in susceptibility to the toxic effects of lithium. There is therefore no threshold serum concen tration at which lithium can be said to be toxic. Nevertheless, it is convenient to regard concen trations of 1.5 mmol/1 or more as being in the toxic range, since such concentrations are poten tially fatal. This investigation examined the circumstances associated with lithium values in the toxic range in the West Midlands Region during a 22 month period.

The effects of drugs are studied In the laboratory, In cllnclal trials and In day to day use.The latter has come to be regarded as clinical audit.
Many publications attest to the beneficial and toxic effects of lithium, and to the narrow dose range between the two1.Although serum lithium concentration is the best monitor of lithium toxicity, it is only an indirect measure; in particular, in unsteady state conditions it will not be an accurate index of tissue toxicity, and high lithium levels may precede (Sellers et al 1982) or follow (Amdisen, 1980) intoxication.Furthermore, patients as well as their individual organs vary in susceptibility to the toxic effects of lithium.
There is therefore no threshold serum concen tration at which lithium can be said to be toxic.Nevertheless, it is convenient to regard concen trations of 1.5 mmol/1 or more as being in the toxic range, since such concentrations are poten tially fatal.
This investigation examined the circumstances associated with lithium values in the toxic range in the West Midlands Region during a 22 month period.

The study
The 23 participating laboratories sent a short questionnaire to any doctor who had requested a lithium estimate proving to have a value of 1.5 mmol/1 or more.The investigation spanned the period from January 1991 to October 1992.Where more than one estimate in the toxic range arose from the same episode of toxicity, the maximum value was taken as the index value.

Findings
Fifty-two separate instances of toxic values were recorded in 48 patients, one patient experiencing four and another patient two separate episodes (Table 1).The patient with four separate episodes was deliberately maintained at lithium values of about 1.2 mmol/1, because only then, the re spondent reported, was there freedom from disabling and frequent relapse.
The mean age of the patients who attained toxic levels was 58 years (s.d.=16 years, range 18 to 84 years).Thirty-six were female, 12 male.In 47 instances the patient was established on lithium, in four treatment was being initiated, and in one treatment had ended, the overdose coming from the remaining supply.The mean of the toxic values was 2.0 mmol/1 (s.d.=0.80, range 1.5-5).The mean period elapsing between the toxic episodes and a previous lithium estimate was 6.7 weeks with an interquartile range of 2-8.8 weeks.In several patients, more than one reason for an episode of intoxication was given.The value in parentheses gives the number of instances of the quoted reason being associated with another cause.Thus, of the 9 instances of blood being sampled too soon after the last dose, three were associated with some other cause.
Of the 30 respondents who told patients how long they should wait after their last dose before venesection for estimation, 20 quoted Amdisen's (1987) recommendation of 12 hours and seven others periods ranging from seven to 15 hours.Three quoted not a fixed period but a range so wide as to vitiate the estimate.

Some instances of intoxication
One patient who took a deliberate overdose had a maximum lithium value of 9.5 mmol/1 and died, despite a reduction to 2.5 mmol/1 on dialysis, probably because of the slow clearance of lithium.A 67-year-old woman on lithium was admitted to a general hospital with dehydration and a urinary infection.Her blood urea was 36.4 mmol/1 (normal 2.3-8).Lithium was prescribed but she refused it -refusing medication In hospital can sometimes be good for your health.

Comment
How might lithium intoxication be avoided?Educat ing both patient and doctor is crucially important.Peet & Harvey (1981) advocate the use of videotapes to educate patients.As with electroconvulsive therapy, there is a strong case for devoting a special session to lithium in postgraduate psychiatric training.Letters to family doctors when treatment is initiated might contain a concise statement of the monitoring recommendations and a reminder of adverse Interactions and of the dangers of lithium during pregnancy.
The most difficult decision falls to the family doctor -whether, during Intercurrent illness, to continue lithium, possibly at a reduced dose, or to stop It.The risk of affective relapse has to be weighed against that of lithium intoxication.Mander & Loudon (1988) reported that 50% of a group of patients with a history of mania relapsed within two weeks of lithium being deliberately stopped.Margo & McMahon (1982) reported rapid relapse of four patients taken off lithium In a drugs trial and Cordess (1982) gives a brief but useful review of other Instances.Even reducing the dose of lithium may precipitate relapse (Tyrer et al 1983).Suppes et al (1991) published a risk of relapse curve of patients taken off lithium but without confidence Intervals for the very early relapse risk, possibly because there were too few Instances to provide acceptably narrow limits.It Is not known how many patients have suffered morbidity or died from lithium toxicity during intercurrent illness.
In practice the problems likely to be encoun tered are:  1991), it Is safer to assume that it does and to stop lithium temporarily.
On these considerations, the general practi tioner might reasonably resolve to stop lithium during any but the most trivial intercurrent illness, until a lithium estimate has been done, and to expect a psychiatrist to contribute to further management.

D. H. Myers and M. J. Hallworth
The quality of lithium monitoring In a health district serving a population of 450 000 was studied over a period of a year.The following Instances of poor monitoring were found: too frequent monitoring In stabilised patients, and failure to take action when lithium values fen below 0.3 mmol/l or rose above 1.0 mmol/l.Ways of Improving the standard of monitoring are considered.
Precautionary measures can be overdone.The British National Formulary (BNFi British Medical Association & Royal Pharmaceutical Society of Great Britain, 1995) used to advocate routine monthly monitoring of serum lithium.Schou (1988) suggested that routine monitoring is not worth the cost or effort, a refreshing statement on a topic that needed simplification.The issue turns on whether patients, once stabilised, will remain so.Kehoe & Mander (1992) found that in 18 of 458 patients there was a gradual Increase in serum lithium concentration during the course of a year, sufficient to require dose reduction.Such patients were found to have an unexplained 18% lower creatinlne clearance rate than controls {Kehoe, 1994).This Is important because 95% of ingested lithium is excreted renally (Dyson et al, 1987).Until more is known, It is reasonable to follow the current BNF(1995) recommendation of monitoring at 3-monthly Intervals or even 6monthly in young and middle-aged patients who are reliable (Ferrier et al 1995).
However, there is no disagreement that it is essential to estimate serum lithium at the start of treatment, during intercurrent illness, and If there is the slightest suspicion of Intoxication.

The study
The Biochemistry Laboratory of the Shropshire District Health Authority, one of the 23 district laboratories in the West Midlands Region, keeps a record of all lithium estimates, together with information culled from the request forms.The quality of monitoring was assessed from these data which covered the interval from 1st January 1991 to 31st October 1992.All estimates were Included In the study.
The time elapsing between two consecutive estimates is referred to as the 'Interval' of the second estimate.
The estimates were considered as four classes, the boundaries being derived from the BNF( 1995): (a) 0-0.3 mmol/l: probably too low to be effective; (b) 0.4-1.0mmol/l: the recommended thera peutic range; (c) 1.1-1.4mmol/l: effective but high, and usually incurring unnecessarily severe side-effects; and (d) 1.5 and above, 'toxic'.
If any lithium value lay outside the BNF range (0.4-1.0 mmol/l), or if any Interval was six months or more, or less than one month, further information was sought from case notes and from family doctor records.
Psychiatric Bulletin (1996), 20. 333-334 (a) Diarrhoea and vomiting.Either may be manifestations of toxicity.Both can result In marked loss of fluid and electrolytes, vomiting Interferes with lithium absorp tion, and mild diarrhoea may itself be caused by lithium even at therapeutic doses.These complex circumstances col lectively constitute a good case for stopping lithium temporarily.(b) Relapse of affective disorder with signifi cantly reduced food and fluid intake.Relapse of this severity requires in-patient care.(c) Fever.Unless there are domiciliary facil ities for frequent monitoring and for en suring adequate fluid intake, lithium Is better stopped or the patient admitted.(d) Urinary infection.Since It is not always possible at onset for a family doctor to know whether the infection extends be yond the bladder (Davlson & Lambie,