Cognitive impairment predicts medication discrepancies in Huntington’s Disease: patient self-report compared to pharmacy records

Abstract

Background

Proper medication reconciliation (= comparing the accuracy of patient-reported medication use with pharmacy records) could prevent potentially dangerous situations such as drug–drug interactions and hospitalization. This is particularly important when patients rely on multiple medications, such as in neurodegenerative disorders like Huntington’s Disease (HD). Currently, it is unknown how often medication discrepancies occur in HD patients and which factors contribute to the discrepancies.

Objective

Identify prognostic factors of medication discrepancies in HD, using patient-reported medication use and local pharmacy records.

Methods

With 134 pre- and manifest HD patients, we performed a multivariable logistic regression analysis with medication discrepancy as dependent variable (pharmacy records as reference value) and sex, CAP score, disease status (pre- or manifest HD), number of concomitant medications taken, presence of an informal caregiver, Unified Huntington’s Disease Rating Scale-Total Functioning Capacity, unified cognitive Z-score and Problem Behaviors Assessment-short characteristic scores for depression, anxiety, and apathy as independent variables.

Results

Medication discrepancies were reported frequently, both in premanifest (43.2%) and manifest HD subjects (36.7%). Impaired cognition significantly predicted medication discrepancies (beta = −0.688, SE 0.27, p = 0.011). All other variables were non-significant.

Conclusions

Regardless of HD disease status and stage, patient self-reported medication use is not a reliable source, especially in those with impaired cognitive function. The presence of an informal caregiver and the absence of polypharmacy, depression, anxiety and apathy do not influence self-reported medication accuracy. Objective verification of medication use with HD patients’ local pharmacy is recommended.

Trial registration number

ICTRP-NL-OMON55123, HD-med, registered 26-Nov-2019.

Introduction

Huntington’s Disease (HD) is a rare, autosomal dominantly inherited neurodegenerative disorder, caused by an expansion of the cytosine-adenine-guanine (CAG) repeat, which causes profound neuropsychiatric, cognitive, and motor symptomatology [1]. Although chorea (unvoluntary, dance-like movement) is the hallmark symptom of HD, cognitive impairments can start up to 15 years prior to motor diagnosis [2]. Cognitive dysfunctions include decreased psychomotor performance, executive dysfunction (impairment in decision making, problem solving, planning and processing speed) and attention problems [2]. Neuropsychiatric symptoms vary widely between patients and include among others depression, anxiety and apathy [1]. As there is no disease modifying treatment available for HD, symptoms are being treated with a variety of pharmacological agents as well as varying psychological therapies [3, 4].

Both in the context of clinical practice and research, healthcare professionals request patients to provide information about their actual medication use. Medication reconciliation, the process of comparing the accuracy of patient-reported medication use with pharmacy records, is vital in this process. Data collected through self-report are prone to recall inaccuracy, potentially leading to unreliable healthcare data that could result in dangerous situations, such as increased risk of adverse events and emergency room admittance. In research settings, unreliable data could lead to misclassification and bias undermining research validity [5]. Inaccuracies in self-reported medication use are likely even more prevalent in daily clinical practice than in standardized research environments, since patients may not consistently anticipate the need to recall their medication use. Patient-related factors such as age, sex, polypharmacy and the presence of an informal caregiver may be relevant when reviewing self-reported medication use in HD. Previous research has shown higher error variance and lower reliability in patient-reported outcomes in patients with more advanced HD. [6] Moreover, informal caregivers could function as mnemonics, and when taking a higher number of medications, increased memory capacity is needed [7]. Furthermore, polypharmacy is common in HD [8] and could lead to increasing side-effects and dangerous drug interactions [9].

Currently, there is no overview of the accuracy of patient-reported medication use in HD, nor is it known what factors contribute to potential discrepancies. Therefore, the aim of this study is to analyze medication reconciliation in HD, by comparing patient-reported medication use to the current medication overview given by local pharmacy records (used as the golden standard), and investigate prognostic factors of medication discrepancies in HD.

Methods

This study used a subset of data from the HD-MED study [10] for which cognition data, Unified Huntington’s Disease Rating Scale—Total Functional Capacity [11] (UHDRS-TFC) and the short version of the Problem Behaviors Assessment [12] (PBA-s) from the Enroll-HD study [13] were also available. All measurements were taken on the same day. Both studies were approved by a Medical Ethical Research Committee. All subjects have given written informed consent. Between January 2020 and September 2023, adult HD expanded gene carriers, who were at baseline using medication, were included from the departments of Neurology of Leiden University Medical Center, University Medical Center Groningen and Maastricht University Medical Center + .

At baseline, participants were asked about their medication use. Medication use was assessed for all participants using a standard questionnaire. Informal caregivers were allowed to provide additional information or clarify responses of participants. All medication details recalled, including clarifications and additions from informants, were compared to the pharmacy record for discrepancy findings. Prescription drugs, including as-needed prescriptions were included. Probe questions about as-needed medication or seasonal medication were used to ensure that medication such as antihistamines, which are often used seasonally, were included. To analyze the accuracy of patient-reported current medication use, we compared patient-reported data with local pharmacy medication records of current medication use (reference value). Deviations from the pharmacy record regarding the total number, name (e.g. pantoprazole instead of omeprazole) or dose of medication were considered a discrepancy. Dose discrepancies included the total daily dose and number of daily intake moments. Over-the-counter medications, although registered in HD-MED, and spelling errors were not considered discrepancies. Polypharmacy was defined as the concurrent use of five or more medications [14].

Subjects were divided in premanifest and manifest HD based on UHDRS diagnostic confidence levels. [11] Scores range from 0 to 4, indicating the certainty of clinical HD diagnosis based on motor symptoms. A score of ≤ 3 is considered premanifest and a score of 4 manifest. To illustrate group characteristics, manifest subjects were classified in disease stages based on the UHDRS-TFC. UHDRS-TFC scores range from 0 to 13. Lower scores indicate a decline in functionality: UHDRS-TFC score 11–13 (stage 1), 7–10 (stage 2), 3–6 (stage 3), 1–2 (stage 4) and a 0 (stage 5). To quantify HD disease progression, we derived the CAG-Age Product score [15] (CAP score = age * ((CAG—30)/6.27)). Cognitive functioning was assessed by the Symbol Digit Modality Test [16] (SDMT), Categorical Fluency Test [17], Stroop Test [18], Trail Making Test [19] (TMT) and the Letter Fluency Test [17]. Herewith, executive function (processing speeds, psychomotor speed, cognitive flexibility, planning and inhibitory ability), attention and language (semantic and phonemic) were determined. A total composite score for each participant was calculated by converting the raw SDMT, Categorical Fluency Test, Stroop Test, TMT and the Letter Fluency Test scores, for each cognitive test into a unified Z-score using the calculator from Mühlbäck et al. [20] Herewith, a cognitive z-score was generated for each participant. For education level, the International Standard Classification of Education 1997 [21] (ISCED) score was determined for each participant ranging from 0 to 6, with higher numbers representing a higher level of education. To assess neuropsychiatric characteristics, depression, anxiety and apathy subscores from the PBA-s were used. For each characteristic, a total score was calculated (total score = frequency*severity), with a total score ≥ 4 indicating the presence of that neuropsychiatric characteristic. The presence of an informal caregiver (ranging from spouse, parent, sibling, child, other family member, friend, neighbor to professional caregiver) was recorded. All informal caregivers were invited to complement medication histories.

Statistical analysis

Data were analyzed using multivariable logistic regression with medication discrepancy as dependent variable and with sex, CAP score, number of concurrent medications, presence of an informal caregiver, UHDRS-TFC (as continuous variable), (pre-)manifest status based on DCL classification (1–3 = premanifest, 4 = manifest), unified cognitive Z-score and PBA-s characteristic scores (continuous variables) for depression, anxiety, and apathy as independent variables. Statistical significance is determined by p < 0.05. Variables were chosen based on expected clinical relevance by three experts in the field (SF, MTDV, LCMK). Model assumptions were checked. Only participants with complete data were used. All analyses were carried out using R version 4.3.2 [22] and packages easystats v0.7.0 [23] and tidyverse v2.0.0 [24].

Results

We included 134 adult HD expanded gene carriers, of whom 44 were premanifest and 90 manifest. Manifest subjects were mostly in early disease stages. On average, demographic characteristics were evenly distributed between the premanifest and manifest group (Table 1). Notably, the unified cognitive Z-score differed between premanifest (1.9; SD 0.7) and manifest (0.6; SD 0.8) participants, although education level was comparable. In premanifest participants, a TFC lower than 11 was observed in a minority of participants. This was partly due to pre-existing comorbidities explaining loss of function at work and at home, for example fibromyalgia symptoms inhibiting a participant’s ability to perform household chores. For others, TFC reduction was due to fatigue and cognitive problems, which may be a non-motor sign of HD. Manifest classification however, was done based on DCL which is derived from the total motor score (TMS).

Table 1.

Demographic and clinical characteristics, and number of medication discrepancies

	Premanifest HD (n = 44)	Manifest HD (n = 90)
Age (years, mean, SD)	49.3 (11.2)	54.2 (10.3)
Sex (n,% male)	21 (47.72%)	48 (53.3%)
Discrepancies (n, %)	Total: 19 (43.2%) peoplea	Total: 33 (36.7%) peoplea
In number of medications	13 (68.4%)	26 (78.8%)
In name of medications	1 (5.3%)	1 (3.0%)
In dosage	8 (42,1%)	12 (36.4%)
Number of medications per subject (mean, SD)	3 (1.9)	3.3 (2.5)
Polypharmacy (n, %)	8 (18.2%)	15 (16.7%)
Informal caregiver present (n, %)	17 (38.6%)	41 (45.6%)
UHDRS-TFC score
11—13	37 (84.14%)	33 (36.7%)
7—10	7 (15.9%)	49 (54.4%)
3—6	0	8 (8.9%)
1—2	0	0
0	0	0
CAG repeat length (mean, SD)	41.5 (2.9)	42.9 (2.2)
Unified cognitive Z-score (mean, SD)	1.9 (0.7)	0.6 (0.8)
ISCED score (first quartile, median, third quartile)	3, 4, 5	3, 4, 5
PBA-s
Depression (n, %)	14 (31.8%)	33 (36.7%)
Anxiety (n, %)	19 (43.2%)	35 (38.9%)
Apathy (n, %)	7 (15.9%)	24 (26.7%)

CAG cytosine-adenine-guanine; HD Huntington’s Disease; ISCED International Standard Classification of Education; n number; PBA-s Problem Behaviors Assessment short version; SD standard deviation; UHDRS-TFC Unified Huntington’s Disease Rating Scale—Total Functioning Capacity. ^a Participants can have multiple types of discrepancies. Consequently, the cumulative totals of discrepancy types is higher than the amount of participants with discrepancies

Discrepancies between self-reported medication use and the pharmacy record were found in 43.2% of premanifest and 36.7% of manifest HD subjects. These included a discrepancy in the number of medications, name of medications and/or dosage. Discrepancies in the number of medications taken occurred most frequently. The medications most commonly overlooked, contributing to number discrepancies, were primarily as-needed medications for allergies (e.g. desloratadine and fluticasone), along with temporary medications for fungal infections (e.g. miconazole and itraconazole). Citalopram was most frequently used (50 of 134 participants, data not shown), and also the most occurring medication in the dose discrepancies (4 out of 20). Name discrepancies were rare.

Predictors of failure to self-report medication use in HD

Multivariable logistic regression analysis revealed that lower scores on cognitive tests were significantly related to the presence of medication discrepancies (beta = −0.688, SE = 0.27, p = 0.011). All other variables were not significantly related to the presence of medication discrepancies (Table 2). The model has adequate fit (Hosmer–Lemeshow GOF chi^2 = 4.30 (df = 8), p = 0.829), with VIF values all < 2.1. It was not possible to analyze each discrepancy type (number/name/dose of medication) separately in the model due to limited sample size.

Table 2.

Multivariable logistic regression analysis for predictors of failure to self-report medication use in HD

	beta (SE)	p-value
Sex (0 = woman, 1 = man)	−0.345 (0.38)	0.364
CAP score	−0.009 (0.012)	0.429
Disease status (0 = manifest, 1 = premanifest)	−0.895 (0.56)	0.110
Number of concurrent medications	0.097 (0.08)	0.251
Informal caregiver present (0 = no, 1 = yes)	−0.584 (0.39)	0.138
UHDRS-TFC score	0.016 (0.11)	0.879
Unified cognitive Z-score	−0.688 (0.27)	0.011*
Total PBA-s depression score	−0.014 (0.07)	0.847
Total PBA-s anxiety score	0.097 (0.08)	0.220
Total PBA-s apathy score	−0.005 (0.07)	0.948

CAP CAG-Age Product; PBA-s Problem Behaviors Assessment short version; SE Standard Error; UHDRS-TFC Unified Huntington’s Disease Rating Scale—Total Functioning Capacity. Bold and * = p < 0.05

Discussion

We are the first to analyze medication reconciliation in HD. Our results showed that medication discrepancies are highly prevalent in both premanifest and manifest HD (up to 43.2%). Impaired cognitive function, as assessed by tests measuring executive functioning, attention, and language, was the only significant predictor for medication discrepancies (p = 0.011). Contrary to our expectations, the absence of an informal caregiver, disease status, polypharmacy, and characteristics of depression, anxiety, and apathy did not contribute to predicting more medication discrepancies. This study shows that medication use reported by HD patients is insufficiently reliable, both in clinical practice and research settings.

In general, our findings are in line with previous studies, both in the general population and in neurodegenerative disorders. Medication discrepancies occurred in almost half (45.4%) of general patients admitted to the hospital, with the level of education and number of concomitant medications significantly predicting the occurrence of medication discrepancies. [25] Of note, the number of concomitant medications in our sample (n = 3) was lower than in comparable research (n = 4–5). This could be a reason why the number of concomitant medications did not contribute significantly to our model. Previous but sparse research in comparable neurodegenerative disorders, showed that in Parkinson’s Disease more medication discrepancies occurred (up to 60%), mostly due to errors in time-sensitive intake-moments of antiparkinsonian medication [26]. However, HD-related medication is less intake-moment dependent, explaining for this difference. In contrast to our HD sample, agreement between self-reported and reimbursement-approval medication data in multiple sclerosis was high (κ = 0.87) without clinical predictors of discrepancies [27].

In our study, cognitive data analysis focused solely on executive function, attention and language domains, neglecting memory. Although we believe that tests of executive functioning are highly appropriate for addressing our research question due to their sensitivity to early cognitive changes in HD on critical abilities such as planning and processing speed, memory tests could have been informative as well. Memory dysfunction is known to be an important feature in the clinical presentation of HD [28]. Consequently, future research should incorporate a comprehensive cognitive dataset that includes memory assessments to provide a more in-depth understanding of multi-domain cognitive functioning in the context of HD. Furthermore, our premanifest cohort has remarkably high unified cognition Z-scores. This may be explained by a higher average level of education in our sample than in the general Dutch population [29] and test–retest effect due to the widely used UHDRS standardized HD battery in research [13, 30–32] and Dutch HD outpatient clinic settings. Although manifest HD subjects had a similar level of education as premanifest subjects, their cognitive abilities are overall more affected by HD, as expected.

As cognitive symptoms may have an earlier onset than motor symptoms, using the DCL as classification tool for disease manifestation is flawed. This can be seen in the TFC scores of the premanifest group, where there are surprisingly low TFC scores. These scores can be attributed to comorbidities, side effects of medication for other diagnoses (both also responsible for some outliers in TMS > 5), but also to an early decline in cognitive functions. Most of the lower TFC scores are attributable to cognitive and/or psychiatric problems, without HD-specific motor symptoms. The use of the DCL = 4 cutoff for manifest patients resulted in people with clear, non-motor symptoms being included in the premanifest category. This may, therefore, also (partly) explain why we have shown a relatively comparable percentage of medication discrepancies in the premanifest group compared to the manifest group, due to early cognitive decline. However, this did not influence our multivariable logistic regression analysis, as the TFC was used as a continuous variable.

Relying on pharmacy records as a golden standard overlooks several complexities. For instance, a patient may not collect medication from the pharmacy or deviates from the prescription (non-compliance). We explicitly removed over-the-counter medication from the analysis since over-the-counter medications are per definition not prescribed, thus not included in pharmacy records. However, to have a complete overview of all medications taken by the HD patients, over-the-counter medications should be included.

Strengths of our study include the thorough and detailed information gathered. Self-report was queried via a simple, short questionnaire (including consistent probe questions about as-needed medication or seasonal medication), minimizing investigator bias. We included the additions and clarification of the caregivers into the patient self-reported medication use to best reflect clinical care and research settings. Missing data were limited by repeated attempts (phone calls) to contact the participant and gather data. Furthermore, we have a relatively large cohort, considering the rare nature of HD. Study limitations include limited cognitive domains, overlooking the cognitive domain of memory and a lack of patients in later disease stages, limiting the generalizability of our results.

In conclusion, clinicians need to be aware of medication discrepancies being related to impaired cognitive function in premanifest and manifest HD. Remarkably, the presence of an informal caregiver does not compensate for the patient’s cognitive abilities. The absence of polypharmacy, depression, anxiety and apathy does not influence self-reported medication accuracy. Since cognitive decline may not be immediately observable, we recommend a closer collaboration between neurologists, neuropsychologists and pharmacies to improve patient safety and quality of care. Objective verification of medication use in HD with patients’ local pharmacy is recommended, both in clinical practice and research settings.

Authors’ contributions

Conceptualization SF and MTDV; methodology SF, MTDV and LCMK; formal analysis SF and LCMK; investigation SF, LCMK; resources STdB; data curation SF and LCMK; writing—original draft SF, MTDV and LCMK; writing—review and editing SF, MTDV, LCMK, RACR, STdB; visualisation SF and LCMK; supervision RACR and STdB; project administration SF; funding acquisition SF and STdB. All authors gave their final approval on the version to be submitted.

Funding

Leiden University Medical Center receives grants from the European Huntington’s Disease Network (EHDN) and Cure HD Initiative (CHDI), participates in an EU Horizon 2020 project: Innovative Medicines Initiative (IMI) 2 (IDEA_FAST), and participates in clinical trials sponsored by PRILENIA, PTC Therapeutics, WAVE and VICO Therapeutics. The aforementioned sponsors had no role in the design, execution, interpretation, or writing of this manuscript. RACR is a member of the Data Safety Monitoring Board of Enroll-HD and the uniQure CT-AMT-130 trial. Author MTDV has nothing to declare. Private Huntington’s Disease community donations.

Data availability

HD-MED data will not be publicly released. Parties that wish to collaborate are invited to contact author S.T. de Bot. Researchers may request access to Enroll-HD data via the process outlined on https://enroll-hd.org/for-researchers/access-data-biosamples/.

Declarations

Conflict of interest

This research was supported by private HD community donations. We received no funding from commercial or not-for-profit parties. HD-MED is an investigator-initiated study set up by the Department of Neurology and Department of Clinical Pharmacy and Toxicology of the Leiden University Medical Center. The authors declare that there are no conflicts of interest relevant to this work.

Ethics approval, and consent to participate and publish

All participants in the HD-MED and ENROLL-HD study gave written informed consent to participate in the study and have their data published in international journals. The HD-MED study is approved by the medical research ethics committee Leiden Den Haag Delft under number NL70391.058.19 and is registered in the International Clinical Trials Registry Platform under number NL-OMON55123. The HD-med study is in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. All sites participating in the ENROLL-HD study have obtained local ethical approval. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this work is consistent with those guidelines.