Toward a Useful and Cost-Effective Workup in Chronic Polyneuropathy: The EXPRESS Study Protocol

In the Netherlands, each year more than 13,000 patients are newly diagnosed with polyneuropathy, which has an estimated prevalence of up to 9.4% in the population aged 50 years and older [1‒3]. Some causes are common, such as diabetes, and others uncommon, such as hereditary or immune-mediated polyneuropathy. The frequency of underlying causes and risk factors varies with age, and from the sixth decade onwards, idiopathic (i.e., cryptogenic) polyneuropathy is as common as diabetic polyneuropathy [1, 2, 4‒6]. In population- and hospital-based studies, at least 50% of patients have one common cause or risk factor and approximately 20% have a combination of common causes or risk factors [1, 3, 4, 7].

The evidence-based Dutch Polyneuropathy Guideline (2019) therefore recommends a complete and extensive workup for most patients, consisting of history taking, neurological examination, extensive blood tests, and nerve conduction studies, although the available evidence for this advice was graded as low quality [8]. While a protocolized approach increases the diagnostic yield and reduces tests and expenditures, knowledge gaps remain about the usefulness, extent, and timing of blood tests and nerve conduction studies [7, 9‒15].

We hypothesize that a limited workup could decrease patient burden and costs in the diagnostic pathway without loss of diagnostic reliability or disadvantageous effect on treatment choice in many patients with a clinical diagnosis of chronic polyneuropathy. This would expose patients to fewer hospital visits and investigations, and a shorter time to diagnosis. Consequently, this could likely result in an improvement of patients’ experienced quality of care and reduction in costs.

The primary objective is to ascertain which blood tests are useful to reveal an underlying treatable disease and when nerve conduction studies are necessary in patients with a clinical diagnosis of chronic polyneuropathy (definition in section Study Population). Secondary objectives are (1) to determine the potential cost-effectiveness of a limited workup in patients with chronic polyneuropathy who already have a known cause or risk factor and (2) to evaluate the burden and gains for patients, expressed in a number of investigations, time to diagnosis, hospital visits, sick-leave, loss of (work) productivity, expenses, experienced quality of care.

The EXPRESS study is a prospective, observational, multicenter quality in healthcare evaluation. We compare two diagnostic strategies, both performed on all participants, with each patient being her/his own control. Patients will receive a diagnosis and standard care from their neurologist. Subsequently, study panel members consisting of neurologists with experience in neuromuscular diseases will propose a preferably limited workup and a diagnosis for each patient. The study panel will assess several cases twice, and several cases will be assessed by two study panel members to evaluate the inter-rater and intra-rater variability. To assess the general validity of the study findings, neurologists other than the patients’ neurologists or study panel members and without specific experience in neuromuscular diseases will be asked to review cases and decide upon a preferably limited hypothetical workup and diagnosis. Conducting our study with such a hypothetical minimal workup will avoid potential harm for patients who have, for example, a treatable immune-mediated polyneuropathy. During the study, patients will be asked to fill in questionnaires about productivity loss and other expenses. The duration of the study, including analysis, is 4 years.

Adult patients with suspected symptoms of polyneuropathy who are referred to a neurologist for an outpatient workup in the participating centers (3 neuromuscular expertise centers in university hospitals, and 5 general hospitals of which 4 are also teaching hospitals in the Netherlands) are eligible for this study. The study population will consist of patients in whom a diagnosis of polyneuropathy is confirmed and patients with mimics in whom another diagnosis is made. In this way, real-time clinical practice with differential diagnostic considerations will be reflected. The group of patients with polyneuropathy will encompass those with and without alarm symptoms. We define chronic polyneuropathy as the presence of distal symmetrical symptoms with slow onset and evolution over a period longer than 3–6 months and the absence of alarm symptoms. Alarm symptoms are progression of symptoms in weeks to 6 months resulting in walking problems or limitations of motor function in the arms, asymmetric or non-length distribution of symptoms, pure motor or motor dominant symptoms, ataxia, autonomic symptoms, and severe neuropathic pain [8]. Inclusion criteria are 18 years or older, signed informed consent, sufficient understanding of the Dutch language; in the referral documentation, there should be suspicion of a polyneuropathy based on sensory and/or motor symptoms in the extremities with a duration of progression and/or severity of symptoms of more than 4 weeks. The exclusion criteria are hospitalization for part of the workup, a referral for a second or third opinion, evolution and nadir of symptoms within a period of 4 weeks.

There is scarce literature about the number of patients referred for polyneuropathy and the eventual confirmation of diagnosis in the setting of an outpatient neurological clinic. In one study carried out in five general hospitals and one neuromuscular expertise center, 870 patients were identified with polyneuropathy, but only 310 patients met the stringent inclusion criteria for newly diagnosed chronic polyneuropathy [7]. Two other studies found that of 198 and 231 patients, 74% and 81% of patients were diagnosed with polyneuropathy, respectively [9, 16]. These studies were carried out in a university hospital and a neuromuscular expertise center. For this prospective multicenter quality in healthcare evaluation, a weighted average of the three studies is assumed, and therefore, 50% of the referred patients are expected to be diagnosed with polyneuropathy. A comparison is made between the workup by the patients’ neurologists and the workup from the panel, resulting in paired observations with each patient being their own control, to ascertain that a limited workup result does not result in a significant proportion of discordant or overlooked diagnosis. A significant proportion is defined as more than 5% of patients, i.e., a similar interpretation as a non-inferiority boundary of 5%. No information is known about the test-retest reliability of the workup by the patients’ neurologists according to the recommendations in the Dutch Guideline Polyneuropathy [8], and the correlation between such a workup and a limited workup remains unknown. Assuming a conservative correlation of 0.7, standard sample size calculations for comparing paired proportions show that 500 paired observations are required to detect a difference of 5% or more difference in diagnostic concordance with 90% power and a two-sided alpha of 5%. To account for missing data, the cost-effectiveness evaluation will be done with 750 paired observations.

The primary endpoint is the effectiveness of a limited workup expressed as diagnostic concordance between panel diagnosis and patients’ neurologists’ diagnosis, respectively. This includes differences in type of diagnosis, both in the sense of a diagnosis of polyneuropathy or mimic and underlying etiology.

Differences in diagnostic pathway costs, impact on treatment choice, and other means of patient management will be assessed as secondary endpoints. In addition, an early HTA will be performed using a pathway analysis combining diagnostic pathway costs with the primary endpoint and potential impact with a disconcordance in diagnosis. Because study panel members indicate precisely which diagnostic investigations are needed, the number and related costs of the diagnostic pathway can be ascertained. Also, the number and type of investigations performed in the workup by the patients’ neurologist can be completely ascertained from the patients’ EMR.

Other secondary endpoints are burden and gain for the patient in terms of number of investigations, time to diagnosis, hospital visits, loss of productivity, expenses, and experienced quality of care. Loss of productivity and expenses can be related to hospital visits, which makes it possible to also determine the effect of a limited workup on productivity when the number of hospital visits is reduced.

Healthcare consumption will be collected from the electronic medical records (EMR) in every participating hospital and linked to Dutch unit costs. Dutch unit costs are described in a manual with reference prices of different aspects of the provided health care in the Netherlands [17]. Moreover, costs or expenses such as travel and lunch expenses and productivity loss will be assessed for each individual patient.

An overview of all study procedures is shown in Figure 1. Patients will be identified through referral letters and outpatient consultation registries by the research physician on behalf of the local principal investigator in each participating hospital. Additionally, patients with a diagnosis of a demyelinating polyneuropathy using the administration of diagnosis in the participating academic centers are prospectively identified because an interim descriptive analysis of the study population indicated that the number of inclusions of those patients was less than expected in the sample size calculation. Patients will receive standard clinical care and a diagnosis from their own neurologists. The standard clinical care is guided by the Dutch guideline, which advises to perform at least once a specific set of blood tests and standard nerve conduction studies in every patient. Naturally, in the case of red flags, neurologists can perform additional laboratory tests and more extensive nerve conduction studies. Most treating neurologists were only aware of the fact that a study was conducted on diagnostic accuracy without details on the methods. Patients’ EMR will be consulted to collect data regarding their referral, patient characteristics (age, sex, neurological findings), family history, medical history, medication use, exposure to alcohol or other substances, predisposing conditions to polyneuropathy, results of blood tests, nerve conduction studies, imaging studies, and other ancillary investigations performed in the workup of the polyneuropathy.

Patients will be asked to complete seven questionnaires regarding symptoms, medical consumption, productivity loss, and experienced quality of care at inclusion (T0) and after 3 (T3) and 6 months (T6). The following questionnaires will be used:

• Douleur Neuropathique 4 (DN4) [18]: T0

• Erasmus Polyneuropathy Symptom Score (EPSS) [19]: T0, T3, T6

• iMTA Medical Consumption Questionnaire (iMCQ) [20]: T3, T6

• iMTA Productivity Cost Questionnaire (iPCQ) [21]: T3, T6

• EuroQol-5D-5L (EQ-5D-5L) [22]: T0, T3, T6

• Inflammatory Rasch-built Overall Disability Scale (I-RODS) [23]: T0, T3, T6

• Questionnaire patients’ experience (constructed questionnaire in collaboration with a patient association, about patients’ experience regarding diagnostic workup): T3, T6

Direct medical costs and other healthcare costs are determined from these data and questionnaires, as well as time to diagnosis. Follow-up time is 6 months for each patient. Six months follow-up should be sufficient to complete data acquisition in all patients. The workup in patients with chronic polyneuropathy usually takes no longer than 6 weeks; however, in patients who have an atypical presentation of chronic polyneuropathy, this can take much longer, as the differential diagnosis is wider.

The panel will base their diagnosis solely on review of the available data obtained from the EMR. For each patient, all these data will be compiled and presented as a case in a standardized and pseudonymized fashion to the panel. Cases are randomly assigned to the panel members, but to avoid potential recognition, panel members will not assess cases from their own hospital. The panel will be blinded for the workup and diagnosis by the patients’ treating neurologist. A panel practice session will be held in the preparatory phase of the project, in which different test patients’ scenarios will be presented to acquaint the panel with the extracted data and rating procedures.

The assessment by the panel is a stepwise process, which is recorded in our electronic data capture tool, Castor.

• Step 1: assess if a patient has a (chronic) polyneuropathy based on clinical features. In this step, the panel member will also conclude on the likelihood of whether the patient has an axonal or demyelinating polyneuropathy, as this will determine the workup in the next step.

• Step 2: determine the (limited) workup. In this step, the panel chooses which tests, for example, laboratory tests (including cerebral spinal fluid analysis), nerve conduction studies, and imaging studies, they consider necessary. Results of the requested tests by the panel member will be shown if they were performed by the patients’ neurologists. If a requested test by the panel is not performed by the patients’ neurologists, no result will be shown.

• Step 3: panel members will receive information about the disease course of a maximum of 6 months. In this step, the panel member can choose additional tests in the same way as in step 2.

• Step 4: conclude the final diagnosis.

If a study panel member has doubts or concerns about the workup or diagnosis, the case will be discussed by the whole study panel in a plenary meeting.

All relevant data will be summarized as means with standard deviations for continuous variables and frequencies and proportions for categorical variables. Medians and IQR will be used for continuous non-normally distributed variables. Classical measures of test-retest reliability and comparisons between paired observations will be used to estimate the correlation and differences between the standard workup and the hypothetical limited workup.

We will use 2 × 2 tables to compare the diagnoses, polyneuropathy or mimic, and calculate diagnostic concordance (i.e., the number of cases where the panel established the same diagnosis as the patients’ neurologists divided by the total number of assessed cases) between the panel and the patients’ neurologist. The same will be done for the underlying etiology of polyneuropathy, and we will evaluate mismatches in diagnosis and underlying etiology between the panel and the patients’ neurologist.

To assess the decrease in healthcare costs, an early HTA will be performed. In addition, patients’ outcomes and diagnosis will be related to average costs per patient depending on the performed investigations and consultations.

The number of hospital visits is inferred by assuming that a standard workup would require at least 3 hospital visits within an average time span of 6 weeks. First visit is for neurological examination and blood tests, second visit is for nerve conduction studies, and third visit is for discussing results and further management or treatment. In case the panel indicated no further workup, the number of visits would reduce to just one as the patients receive the diagnosis immediately. In case the panel indicated a limited workup, the number of visits would reduce to one (blood tests only) or two (nerve conduction studies with or without blood tests) plus a telephone consultation after 1 week for discussing the results and further management or treatment.

Outcomes will be descriptive and can be used to further inform the development of a cost-effective pathway. The average costs for diagnostics per patient in both groups will be calculated, i.e., standard care by the patients’ neurologists and the hypothetical workup proposed by study panel members. The outcomes of the panel will be compared with the performed workup and established diagnosis by the patients’ neurologists. In case of mismatches, we will select a number of representative cases (or personas) because of the heterogeneity in the underlying causes of polyneuropathy and thereby treatment impact and prognosis. In these representative cases, we will estimate the potential impact of mismatch at combining literature data and expert opinion of panel members. This potential impact will be related to the cost-savings in the diagnostic pathway.

The aim of our study is to use data generated in the real-world/daily practice in this patient population that will consist of polyneuropathy patients and mimics. Therefore, as neurologists inevitably have different approaches, this can lead to variation in history taking, examination, reporting, and which tests are performed. Since the clinical decision-making of our panel will be dependent on the information gathered by the patients’ neurologist, it is possible that our panel might miss or interpret certain information in a different manner. Furthermore, we are aware of the distinction between seeing a patient at the outpatient clinic and reading about a patient. However, we believe that neurologists report the required relevant and essential information in the patients’ EMRs to be able to decide if a patient has polyneuropathy. To prevent loss of potential essential data, the exact reports made by the patients’ neurologist in the patients’ EMRs are presented in a standardized manner to the panel.

This prospective observational multicenter quality in healthcare study will provide level III evidence about what a cost-effective workup in patients with chronic polyneuropathy entails. Improving the cost-effectiveness of the workup of patients with polyneuropathy is especially important as the number of these patients is expected to grow, because of (1) the aging population and polyneuropathy mostly affecting the middle-aged and elderly patients and (2) a concomitant expected increase in risk factors for polyneuropathy [1, 3, 4]. As a consequence, this will inevitably result in increased healthcare costs, while healthcare resources and costs are limited.

We are very grateful for the participation of the representative of patient association Spierziekten Nederland, Ger van de Wijdeven. We are also very grateful for the participation and time of our study panel members (A.M. Dekker, P.T.C. van Doormaal, M. Eurelings, M.F.G. van der Meulen, S. Piepers, E. Verstraete, and L.L. Teunissen), without them this study would not be possible. Likewise, we are really thankful for the participation of M.P. van der Meulen.

The Medical Ethics Committee Utrecht appraised this study as not obligatory with the Medical Research Involving Human Subjects Act (WMO), and therefore, our study protocol was not reviewed by the Medical Ethics Committee. Written informed consent will be obtained from all participating patients.

The authors have no conflicts of interest to declare. F. Eftimov, P.A. van Doorn, NC. Notermans, and A.F.J.E. Vrancken participated in the working group for developing a guideline for diagnosing polyneuropathy in the Netherlands.

This study is supported by “Evaluatieonderzoek Zorg Evaluatie & Gepast Gebruik” (project number 10330022010005). The funder had no role in the design, data collection, data analysis, and reporting of this study.

M. Wiersma: project administration and writing – original draft. G.M. van der Star: project administration and writing – review and editing. R.PA. van Eijk and G.W.J. Frederix: methodology. F. Eftimov and P.A. van Doorn: conceptualization, methodology, and writing – review and editing. NC. Notermans: conceptualization and writing – review and editing. A.F.J.E. Vrancken: conceptualization, methodology, funding acquisition, and writing – review and editing.

All data generated or analyzed during this study will be included in future articles. Further inquiries can be directed to the corresponding author.