Quality of laboratory biomarker monitoring during treatment with lithium in patients with bipolar disorder

Clinical guidelines recommend monitoring of creatinine and lithium throughout treatment with lithium. We here assessed the extent to which this occurs in healthcare in Sweden.


| BACKG ROU N D
3][4][5] Lithium has unique anti-suicidal properties 6 not shared by alternative mood stabilizers such as valproate. 7wever, a downside of lithium treatment is the need of close monitoring.Monitoring of serum concentration of lithium is important because of a narrow therapeutic window.While subtherapeutic lithium concentrations may result in relapse or recurrence, supratherapeutic concentrations may be associated with a higher risk of adverse effects.Potential adverse effects include kidney function decline, hypothyroidism, and hyperparathyroidism (the latter often presenting as hypercalcemia). 8,9[12][13][14] In an international survey completed by 117 healthcare from 24 countries, 97% reported to monitor lithium levels and creatinine regularly and according to guideline recommendations. 15ports from routine clinical practice in the United Kingdom, 11,12,16 Japan, 17 and the Netherlands 18 suggest otherwise: rates of laboratory testing of creatinine, lithium, and TSH were highest during the first year of therapy, but decreased thereafter.Monitoring was not consistent with guideline recommendations, and fewer than 15% of patients were monitored for two or all three parameters annually throughout the therapy course.In Sweden, a singlecenter study reported that in 2010, fewer than 70% of patients on lithium therapy had a creatinine or lithium test taken, but consistency over time or in more contemporary settings for Sweden is unknown.
The aim of this study was to evaluate the quality of lithium monitoring among contemporary patients with bipolar disorder in the region of Stockholm, Sweden.

| Data sources
We used data from the Stockholm CREAtinine Measurements (SCREAM) project, a healthcare utilization cohort including all adult residents in Stockholm between 2006 and 2018. 19There is a sole and unified health system in the region of Stockholm that provides universal healthcare.The core repository of SCREAM is an extraction of laboratory measurements performed at the three main central laboratories, which account for about 95% of all measurements performed in Stockholm healthcare, including all tests of lithium, creatinine, calcium and TSH.Using the personal identification number of each citizen, these data were linked to regional and national administrative databases with complete information on demographic data, healthcare use, diagnoses, vital status, and dispensed prescriptions at Swedish pharmacies.The study used only de-identified data and thus was deemed not to require informed consent, being approved by the regional ethical review boards, the Swedish National Board of Welfare and the healthcare provider (Region Stockholm).

| Study design and patient selection
This is an observational study of persons with bipolar disorder initiating lithium therapy.We created a cohort of all adults (≥18 years old) in the Stockholm region who had a diagnosis of bipolar disorder and who initiated lithium therapy between January 1, 2007 and December 31, 2018.The study exposure was initiation of lithium therapy, which was defined as the first lithium dispensation, after a minimum gap of 18 months of observation (dispensed prescriptions are recorded since June 2005) to ensure it was not a temporary therapy interruption.The date of the first lithium dispensation was considered the index date.Annual periods were defined by the anniversary dates of initiation.Patients were followed from index until lithium therapy discontinuation, death, migration outside of Stockholm region or end of follow up (December 2018).

| Study covariates
Baseline covariates included age, sex, eGFR, prescriber characteristics, comorbidities hyperthyroidism, hypothyroidism, diabetes mellitus, cardiovascular diseases, mental disorders due to psychoactive substance abuse and manic episodes (see Table S1 for detailed definitions).We then extracted information on laboratory measurements performed in connection with any healthcare consultation.
In addition, we extracted information of the centers prescribing the first lithium fill, classifying them as psychiatric specialist care or other types of care.Estimated glomerular filtration rate (eGFR) was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) 2009 creatinine equation 20 without correction for race.
Serum or plasma creatinine was measured with either enzymatic or corrected Jaffe methods, both traceable to IDMS standards.

| Outcomes
The study outcome was the concordance with guideline recommendations regarding monitoring within recommended intervals of serum lithium, creatinine, calcium, and TSH during lithium therapy.
Like international guidelines, [10][11][12] Swedish guidelines recommend monitoring creatinine before treatment initiation (to adjust lithium dose).During lithium treatment, the recommendation is to monitor levels of: lithium, creatinine, calcium, and TSH at least once per year. 14We only searched for the presence of laboratory testing in patients that had lithium dispensations estimated to provide treatment for more than 6 months during each year of therapy.
To evaluate concordance between routine clinical practice and guideline recommendations, we ascertained the presence of performed laboratory tests through distinct laboratory monitoring periods.First, pre-initiation testing: a measurement of creatinine within 120 days prior to lithium therapy initiation.Second, monitoring: a measurement of lithium, creatinine, TSH or calcium at least once per every year of therapy until therapy cessation, switching to another agent, or censoring (death or end of follow-up).We evaluated each biomarker separately, and we also created a combined indicator of appropriate monitoring if all biomarkers were monitored annually during therapy.
The end of lithium therapy was defined at 6 months after the end of estimated pill supply, when there was no intervening dispensation.Therapy switch was defined as initiation of any other psychoactive medication with an indication for bipolar disorder, that is, valproate and atypical antipsychotics, after a last recorded dispensation of lithium.

| Statistical analysis
Continuous variables are presented as means with standard deviation or median with interquartile range (IQR) depending on the distribution, and categorical variables as numbers and percentages.
We calculated the cumulative incidence of therapy cessations or switches to another mood stabilizer agents.During each year of lithium therapy, we estimated the proportion of total patients who underwent testing for each of the guideline recommended tests.In a second step, we evaluated the proportion of patients who had their testing done consistently once annually.The consistency of testing for lithium levels and creatinine was graphically depicted through an alluvial plot.Lastly, we fitted a multivariable logistic regression model to compute odds ratios with 95% CIs to identify baseline characteristics (at time of lithium initiation) associated with the presence of a monitoring consistent with guideline recommendations.
Analyses were performed using R version 4.0.5.

| Baseline characteristics of new users of lithium
Among 8684 individuals who initiated lithium therapy in our region, we identified 4428 adults who had a diagnosis of bipolar disorder and were included in our study.Their characteristics are depicted in Table 1.Median age was 39 years (IQR: 29-50) and 64% were women.As many as 52% of participants had a concurrent diagnosis of anxiety and 22% a history of mental disorders due to psychoactive substance abuse.Their median (IQR) eGFR at time of therapy initiation was 103 (90-114) ml/min/1.73m 2 ; 1.0% of the cohort had eGFR < 60 ml/min/1.73m 2 , which is the GFR threshold for chronic kidney disease.The majority of patients (98%) initiated treatment as prescribed by psychiatric care specialist units.

| Length of treatment
The median time on lithium therapy was 4.3 (IQR: 1.9-7.45)years.   of patients who did not undergo testing: in the 1st year, 12% for lithium, and 13% for creatinine; in the 11th year, 33% for lithium and 33% for creatinine did not undergo testing for these biomarkers.

| Testing for recommended laboratory biomarkers during lithium therapy
Figure 3 depicts the proportion of patients tested for TSH and the proportion tested for calcium each year, among those who remained on therapy.The frequency of testing for these biomarkers was lower than that for creatinine or lithium.At therapy initiation, 24% and 33% of patients were not tested for TSH or calcium, respectively.During treatment, an increasing proportion of patients (from 16% in year 1 to 35% in year 11 for TSH and between 33% and 46% for calcium) did not undergo testing for these biomarkers.

| Consistent monitoring of recommended laboratory biomarkers throughout lithium therapy
As a next step, we evaluated the proportion of patients for whom monitoring for both lithium and creatinine occurred once annually throughout their treatment, in keeping with guidelines.In the 1st F I G U R E 1 Cumulative incidence of therapy discontinuation events, defined as the composite of lithium cessation or switching to another mood stabilizer.
F I G U R E 2 Proportion of patients tested for blood creatinine or lithium each year, among those that remained on therapy. | year, 21% did not, and in the 11th year, 37% did not.The overall trajectory of laboratory monitoring of lithium and creatinine is illustrated through an alluvial plot (Figure 4).The plot shows that the unmonitored patients differ from year to year.The proportion of patients who were adequately monitored with at least one testing for both lithium and creatinine in every year of follow up  2).

| DISCUSS ION
Among approximately 4428 patients with bipolar disorder newly initiating lithium therapy in Stockholm and followed for up to 11 years, we report that a considerable proportion were not monitored according to the Swedish guidelines.In any given year, between 21% and 33% of patients lack either a creatinine or a lithium level, or both.This proportion increased with longer treatment duration.Monitoring is not consistent, and as few as 16% of patients had lithium and creatinine tested at least once annually throughout their follow-up time.However, this also implies that the people with missing values vary from year to year, and that over several years a larger proportion of people have had some values in each category.
Laboratory monitoring of lithium treatment in our study was suboptimal.A previous study from Sweden evaluated lithium monitoring in a southern region during 1981-2010, finding that compliance with monitoring of both lithium and creatinine by calendar year rose from 36% in 1981 to 68% in 2010. 12However, patient inclusion in this study was based on the presence of a lithium measurement rather than a dispensation of lithium, and treatment discontinuation was defined as an absence of lithium measurements for more than 365 days.Both these factors could increase the proportion that was estimated as adherent.These low rates of monitoring are, however, not exclusive of Sweden.Similar observations have been made in the Netherlands, 18 while studies from the United Kingdom 11 and Japan, 17 report even lower rates, with annual lithium and creatinine monitoring being performed in as few as 30% and 15% of patients, respectively.We speculate that this low monitoring likely puts patients at risk of supratherapeutic lithium levels and adverse effects.
Lithium is fully excreted by the kidneys. 21Monitoring of kidney function is a prerequisite for safe dosing of lithium.Besides, once kidney function starts declining, there is likely a limited time frame during which lithium cessation may increase GFR. 22With time, there is an increased risk of kidney function decline, due to ageing, 23   direct nephrotoxicity attributed to lithium.In our study, people of older age (>=75 years) were less likely to undergo yearly monitoring for creatinine and lithium, despite guidelines being equal for all age groups. 13,24In a Canadian study of patients with bipolar disorder aged >65 years, the proportion of patients without a creatinine or a lithium level in the first year was 33% and 25%, respectively. 24The corresponding proportion in our overall population was 14% and 14%.Lower adherence to monitoring in older people is a matter of concern, because older adults may be at highest risk of both dose-related adverse events and potential lithium-related nephrotoxicity. 25,26 are unable to determine whether the levels are ordered and not done, or not ordered.There is likely an element of patient-related non-adherence.We observed that patients with previous diagnoses of anxiety or mental disorders due to psychoactive substance use were less likely to be adequately monitored.Alternatively, the expected benefits from lithium therapy in these patients may be seen as larger than the potential harms, altering the risk-benefit ratio of continued prescription without monitoring.
Lithium discontinuation was common in our study, reaching 50% within 4 years, a proportion that aligns with previous reports. 27The most common reason for therapy cessation was switching to another mood stabilizer.Previous studies cite adverse drug effects as the most common reason for lithium discontinuation in patients with bipolar disorder. 22,27If the decision to switch to another agent was driven by intolerance or adverse effects requiring therapy cessation, there are therapeutic opportunities to maintain patients on therapy.
For instance, thiazides or amiloride can be used to alleviate polyuria, 28,29 the most frequently reported adverse effect of lithium. 21,30 is also possible that non-adherence with monitoring is a reason for switching in some patients.
Strengths of our study include the evaluation of all eligible cases in our region, its contemporary period of data collection and a large sample size with long follow-up (up to 11 years), no loss to follow-up, and complete collection of performed laboratory tests.Another strength is the ascertainment of the exposure through pharmacy dispensations, which is a better surrogate of drug intake than prescription claims or the presence of a lithium level.A limitation is the lack of information on the reasons-whether provider or patient related-for the absence of laboratory monitoring or cessation of therapy, which makes our study descriptive.We also recognize that some patients may have picked up their dispensation first, and had a creatinine level shortly after, deferring the actual start of therapy until the results returned and were reviewed by their psychiatrist.Nevertheless, our study emphasizes the need for strategies

Figure 1
Figure 1 depicts the cumulative incidence curve for cessation of lithium therapy (combining both permanent therapy cessation or switch to other mood stabilizers).About 35% of patients interrupted therapy within 3 years, and 50% at 4 years.Approximately 60% of

Figure 2
Figure 2 depicts the proportion of patients tested for creatinine and the proportion tested for lithium each year, among those who remained on therapy.At initiation, 21% had no prior testing for creatinine.During treatment, there was an increasing proportion

F I G U R E 3 F I G U R E 4
was low: 16% (n = 687).If we evaluate the consistency of monitoring for three biomarkers (lithium, creatinine, and TSH) suggested by clinical guidelines, 12% (n = 546) met this criterion every year (not shown).Proportion of patients tested for TSH or calcium each year, among those that remained on therapy.Alluvial plot showing the consistency of annual monitoring for blood lithium and creatinine.Using multivariable logistic regression model, we evaluated selected sociodemographic or clinical factors associated with adherence to the recommended monitoring frequency for lithium and creatinine.Higher age and a previous diagnosis of anxiety or mental disorders due to psychoactive substance use were the only baseline factors associated with lower odds of being monitored (Table to improve adherence and persistence to this potentially life-saving treatment.Finally, our results are representative of the region of Stockholm during 2007-2018, and extrapolation to other regions, countries or periods should be done with caution.To conclude, in a contemporary evaluation of laboratory monitoring among patients on lithium therapy, we observe considerable discordance with guideline recommendations.As a practical application, this study thus evidences a clinical gap leading to increased risk of poor-quality care and missed opportunities to detect adverse drug reactions.Further research is needed to understand the reasons for non-adherence with monitoring recommendations, and to implement strategies with the goal of optimizing safety and effectiveness for patients treated with lithium.FU N D I N G I N FO R M ATI O NResearch reported in this publication was supported by the Swedish Research Council (2019-01059) and the Martin Rind's Foundation.