PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Arch Intern Med. Author manuscript; available in PMC 2013 June 27.
Published in final edited form as:
PMCID: PMC3694340
NIHMSID: NIHMS469228

Tests and Expenditures in the Initial Evaluation of Peripheral Neuropathy

Brian Callaghan, M.D.,(1) Ryan McCammon, AB,(1) Kevin Kerber, M.D.,(1) Xiao Xu, Ph.D.,(1) Kenneth M. Langa, M.D., Ph.D.,(1)(2) and Eva Feldman, M.D., Ph.D.(1)

Abstract

Background

Peripheral neuropathy is a common disorder in which an extensive evaluation is often unrevealing. We sought to define diagnostic practice patterns as an early step in identifying opportunities to improve efficiency of care.

Methods

The 1996–2007 Health and Retirement Study (HRS) - Medicare Claims linked database was used to identify individuals with an incident diagnosis of peripheral neuropathy using ICD-9 codes and required no previous neuropathy diagnosis during the preceding 30 months. Focusing on 15 relevant tests, we examined the number and patterns of tests and specific test utilization performed 6 months before and after the incident neuropathy diagnosis. Medicare expenditures were assessed during the baseline, diagnostic, and follow-up periods.

Results

Of the 12,673 patients, 1,031(8.1%) received a new ICD-9 diagnosis of neuropathy and met our inclusion criteria. Of the 15 tests considered, a median of 4 (inter-quartile range (IQR)=2–5) tests were performed with over 400 patterns of testing. An MRI of the brain and/or spine was ordered in 23.2%, whereas a glucose tolerance test was rarely obtained (1%). Medicare expenditures were significantly higher in the diagnostic period compared to the baseline period (mean $14,363 versus $8,067, p<0.0001).

Conclusions

Patients diagnosed with peripheral neuropathy typically undergo many tests but testing patterns are highly variable. Almost one-quarter of patients receiving neuropathy diagnoses undergo high-cost, low-yield MRIs while very few receive low-cost, high-yield glucose tolerance tests. Expenditures increase substantially in the diagnostic period. More research is needed to define effective and efficient strategies for the diagnostic evaluation of peripheral neuropathy.

Introduction

Peripheral neuropathy is a common and debilitating condition with a prevalence of 2–7% in the general population1,2. The prevalence increases significantly in older adults with a prevalence of approximately 15% in those over age 403. Distal symmetric polyneuropathy (DSP) is by far the most common subtype of neuropathy accounting for the vast majority of cases4. Prior research suggests that a focused and directed evaluation is the optimal diagnostic approach in this patient population5. The best evidence for diagnostic testing in DSP was recently summarized in a systematic review by the American Academy of Neurology (AAN)4. Fasting glucose levels, B12 levels, serum protein electrophoresis (SPEP), and 2 hour oral glucose tolerance tests (GTT) were found to be supported by the literature based on the yield of these tests and the potential for subsequent interventions4. A fasting glucose level is the most frequently used test to diagnose diabetes, which is the most common cause of DSP6. Vitamin B12 deficiency causes a potentially treatable neuropathy with different characteristics than in those with idiopathic neuropathy7. The use of GTT and SPEP testing is supported by evidence that there is a substantially increased prevalence of these abnormalities in patients with neuropathy compared with control groups810. The evidence to support other diagnostic tests in the evaluation of DSP is lacking.

Unfortunately, even after an extensive evaluation, the cause for a substantial number of peripheral neuropathy cases remains unknown11. Further, even when a specific cause is identified, only a few therapies exist. The most common etiology for DSP is diabetes which is treated with glycemic control. Immunosuppressive medications are used for certain rare subtypes of neuropathy such as chronic inflammatory demyelinating polyradiculoneuropathy (CIDP) and mononeuritis multiplex. However, there are few disease-modifying therapies for patients with DSP, and pain management becomes paramount regardless of etiology. Since DSP comprises the vast majority of peripheral neuropathy, many of these cases are idiopathic, and few treatments are available, efficient diagnostic testing is particularly important within this population.

No prior studies have described the evaluation of peripheral neuropathy in routine clinical care. This information is important because it can provide insights into opportunities for optimizing care and setting future research priorities. In this study, we used a large, nationally representative health survey that is linked with claims data to identify a cohort with incident peripheral neuropathy and to determine evaluation practices by all physicians.

Methods

Population

Data for our analysis came from respondents to one or more waves of the Health and Retirement Study (HRS) biennial interview between 1998 and 2006, with linked Medicare Standard Analytic File data. This database combines the rich demographic detail from the HRS with the extensive health care utilization data available in Medicare claims. We identified individuals with incident peripheral neuropathy defined as persons who had an ICD-9 diagnosis of peripheral neuropathy and no previous diagnosis during the preceding 30 months (incident diagnoses range from 1998–2007). All ICD 9 codes for peripheral neuropathy were included (354.5, 356.0–9, and 357.0–9). Individuals were included if they were at least 65 years-old at the start of the baseline period, were continuously enrolled in Medicare parts A & B fee-for-service from 30 months preceding the index diagnosis through 6 months following the index diagnosis, and completed a HRS interview within 3 years prior to the diagnosis date. We also identified a matched comparison group using a propensity score method (eMethods).

Demographics/Health measures

Key demographic variables that were identified from the HRS survey included age, gender, race/ethnicity, education, BMI, alcohol intake, and limitations in activities of daily living (ADLs). The Medicare claims database provided the diabetes status of the patient based on the Chronic Condition Warehouse (CCW) definition (at least 1 inpatient, skilled nursing or home health claim or 2 outpatient or Carrier claims with diagnostic codes 249.x, 250.x, 357.2, 362.01, 362.02, or 366.31 during the 2-year matching period)12. Moreover, ICD-9 diagnosis codes identified patients with diabetic complications other than neuropathy. Medicare claims also provided information on chronic kidney disease, rheumatoid arthritis/osteoarthritis, and/or cancer.

Diagnostic tests

Tests were identified by CPT codes and included fasting glucose level, hemoglobin A1C (HA1C), glucose tolerance test (GTT), serum protein electrophoresis (SPEP), B12, ANA, ESR, TSH, CBC, and comprehensive metabolic panel. Electrodiagnostic tests and MRI studies (brain, cervical, thoracic, or lumbosacral spine) were also identified. These tests were selected based on their relevance to the diagnostic evaluation of DSP.

Medicare expenditures

Medicare payment information was obtained from the Medicare Standard Analytic Files, and included all payments found in the MedPAR, outpatient, carrier, home health, hospice, and durable medical equipment files. We evaluated expenditures during the baseline (6–18 months prior to diagnosis), diagnostic (6 months before and after diagnosis), and follow-up periods (6–18 months after diagnosis).

Statistical analysis

The number and patterns of testing were assessed during the diagnostic period (6 months before and after the index diagnosis). This time frame was chosen because tests are frequently ordered for this condition prior to the firm establishment of diagnosis. Medicare expenditures were calculated during the baseline, diagnostic, and follow-up periods. T-tests were employed when comparing continuous variables. Sensitivity analyses were conducted after exclusion of patients with a diagnosis of mononeuritis multiplex, demyelinating neuropathy or hereditary neuropathy. All analyses were performed with SAS 9.1 (Cary, NC).

Results

Population

Of the 12,673 patients within the HRS-Medicare claims database, 1,031 (8.1%) received a new ICD-9 diagnosis of peripheral neuropathy during the 10-year study period and met our inclusion criteria. Demographic and other characteristics of the population are presented in Table 1. Mean age of this population was 77.6 years and 54.0% were female. Twelve percent were non-Hispanic black and 8.0% were Hispanic. 41.5% met the CCW definition of diabetes, and 16.3% had other diabetic complications. Demographics from a matched comparison group are presented in eTable1.

Table 1
Demographic and clinical variables in the population of patients with neuropathy.

Among those patients with diabetes, the most common ICD-9 neuropathy diagnoses were polyneuropathy in diabetes (44.4%) followed by different idiopathic classifications (47.8%). Additionally, 6.6% were classified as neuropathy due to other diseases (including toxins, drugs, or inflammatory conditions), 1.4% hereditary neuropathies, and 0.2% acute inflammatory demyelinating polyradiculoneuropathy (AIDP). In those patients without diabetes, 80.0% had an ICD-9 diagnosis of idiopathic neuropathy, followed by 11.7% with neuropathy due to other diseases, 4.7% hereditary neuropathies, 3.5% due to diabetes (new diagnosis of diabetes), 1.0% mononeuritis multiplex, 1.0% AIDP, and 0.3% CIDP.

Test Utilization

Of the 15 relevant tests assessed, the median number of tests performed per patient was 4 (inter-quartile range (IQR)=2–5). There were over 400 patterns of testing within this population with no single pattern occurring in more than 4.8% of patients. Furthermore, no particular test was common to all of the top patterns.

A fasting glucose level was ordered in 23.4% of neuropathy patients, and the HA1C was ordered in 43.2% (Figure 1A). B12 levels were ordered in 32.6% of neuropathy patients and SPEP was performed in 13.3%. In the non-diabetic population, HA1C was ordered in only 17.1% of cases, B12 levels in 40.6%, and SPEP in 18.7% (Figure 1B). Only 10 (1.0%) cases received a glucose tolerance test. In contrast, 23.2% of cases received at least one MRI of the brain and/or spinal cord (Figure 2). The most common types of MRI performed were brain (13.7%), lumbar spine (9.6%), cervical spine (5.0%), and thoracic spine (1.9%). An electromyogram (EMG) was performed in 19.8% of neuropathy cases. Of those receiving an electrodiagnostic test, the mean number of nerves evaluated on nerve conduction studies (NCS) was 8.79 (standard deviation 6.89), median 7.0 (IQR=5–10), with a range 1–48. Patients with 1–14 nerves evaluated on NCS were 2.8 (95% CI 2.1–3.9) times more likely to have a MRI then those who received no test. Those with 15 or more nerves evaluated (greater than one standard deviation greater than the mean) were 5.2 (95% CI 2.6–10.3) times more likely to have a MRI then those who received no test. The complete blood count was ordered in 73.1% of cases, TSH in 55.2%, comprehensive metabolic panel in 53.2%, ESR in 28.7%, and the ANA in 11.2% of cases, (Figure 3). Test utilization from a matched comparison group is presented in eFigures 1–3.

Figure 1
Utilization of diabetic and AAN recommended tests in neuropathy patients
Figure 2
EMG and MRI utilization in neuropathy patients
Figure 3
Utilization of common diagnostic tests in neuropathy patients

Medicare Expenditures

In the baseline period, prior to ICD-9 diagnosis of neuropathy, the Medicare expenditures were $8,067 (Table 2). During the diagnostic period the expenditures increased significantly to $14,362 (p<0.0001). This increase was also observed after excluding patients with diabetes (mean $12,190 versus $6,633, p<0.0001). In the follow-up period, the Medicare expenditures remained higher ($11,748) than the baseline period but were lower than during the diagnostic period. Expenditures from a matched comparison group are presented in eTable2.

Table 2
Medicare expenditures in the entire sample and in the sub-sample of patients without diabetes

Discussion

Using a nationally representative sample of older US adults, we found that over 8.1% of the individuals had a new diagnosis of neuropathy and met our inclusion criteria during the 10 years of our study. Many tests were ordered during the diagnostic period for peripheral neuropathy, but the evaluation was highly variable. MRIs of the brain and/or spine were frequently ordered, whereas the glucose tolerance test was rarely ordered. Significant increases in cost occurred during the diagnostic period compared to the baseline period. These findings suggest substantial opportunity to improve efficiency in the evaluation of peripheral neuropathy.

The large variation in testing indicates little consensus on an appropriate testing strategy in this population. With over 400 total patterns of tests and no pattern accounting for more than 4.8% of the total number, no standard approach to the evaluation of peripheral neuropathy currently exists. Similarly, the number of nerves tested on nerve conduction studies exhibited substantial variation and was close to the recommended number of nerves for patients entering a clinical trial as suggested by a 2009 AAN practice parameter. Substantial utilization of diagnostic tests was observed, exhibited by a median of 4 tests ordered out of the 15 tests evaluated. Interestingly, those with more nerves evaluated on NCS also had a higher chance of receiving a MRI, another expensive test. More research is needed to determine the optimal approach to this prevalent condition and to disseminate this information to the physicians that care for these patients.

When examining test utilization, two significant deviations from expected clinical practice and the tests supported by the best available evidence were discovered. The first was that a large proportion of these patients received MRIs of the brain and/or spine. In fact, each segment of the neuroaxis (brain, cervical, thoracic, and lumbar spine) was performed at a higher than expected frequency. When combining all MRI tests together, the utilization was even more dramatic with nearly one in four receiving at least one MRI. For a condition that affects the peripheral nervous system, this degree of utilization is substantial and suggests that many physicians have significant uncertainty when localizing neuropathy symptoms to the peripheral nervous system. The use of MRI may also result from the large proportion of patients with idiopathic neuropathy, the fact that electrodiagnostic studies can be non-diagnostic or normal, or from patient preferences. Another possibility is that neuropathy patients are at higher risk for other conditions or symptoms that warrant MRI.

The second deviation from expected practice is that GTTs are rarely ordered. In fact, only 1% of this neuropathy population received the GTT. The prevalence of impaired glucose tolerance in otherwise idiopathic neuropathy patients is higher compared to historical controls and the type of neuropathy in these patients is different (more sensory and painful neuropathies)9,10. Therefore, emerging data supports impaired glucose tolerance as potentially one of the most common etiologies of neuropathy although controversy still exists1315. This condition is also one of the few potentially treatable causes of neuropathy with diet and exercise preventing a large percentage of patients from going on to develop diabetes and its inherent risk of neuropathy progression16. One potential reason for the extremely low utilization of this test includes the fact that many physicians use hemoglobin A1C to identify those with pre-diabetes16. Unfortunately, the cut point used to define pre-diabetes with this test has a low sensitivity and many of the patients in the Diabetes Prevention Program trial would not have been included using this criterion17. These results indicate that two of the first steps in increasing the effectiveness and efficiency of the evaluation of peripheral neuropathy may be investigating why so many MRIs are ordered and determining the barriers to utilization of the glucose tolerance test.

The other three tests supported by the AAN systematic review (fasting glucose, B12, SPEP) were ordered less frequently than expected. In fact, only 49.8% of neuropathy patients received one or more of these three tests and only 17.3% received two or more. Even though some patients with peripheral neuropathy may not need these tests if they have a well-established cause, these numbers are still significantly lower than if the 25–40% of patients that end up with an idiopathic diagnosis were evaluated11. Interestingly, B12 levels are ordered much more frequently than SPEPs, emphasizing the fact that many physicians do not recognize the evidence in support of ordering this test. While these data were collected from a time period prior to the release of the AAN review, they highlight that physicians were not ordering the tests with the highest levels of evidence to support their use. Understanding the obstacles to the utilization of these tests will be paramount to improving the efficiency of diagnostic testing in this population.

Medicare expenditures in this population rose substantially during the diagnostic period. The expenditures decreased during the 12-month follow up period, but did not return to baseline. This pattern is not surprising given our findings that patients with a new diagnosis of neuropathy undergo an extensive evaluation. These expenditures, however, may also reflect other broad expenditures related to their disabling condition including orthotics, walking-assist devices, office visits, and hospitalizations to name a few. These other expenditures likely explain the persistent increase in expenditures in this population, but the transient increase in the diagnostic period is at least partially explained by costs associated with diagnostic tests. Therefore, understanding the relative impact of these tests is important in allowing physicians to practice efficient care, especially in a patient population where the etiology frequently remains unclear and there are few disease modifying therapies. Future studies examining which diagnostic tests are driving the costs and whether they are effective and useful within this population will be essential.

This study has important limitations. ICD-9 diagnosis codes were used to identify patients with peripheral neuropathy, which may lead to misclassification bias. Peripheral neuropathy is a heterogeneous condition, and certain rare subtypes of neuropathy may require a different evaluation than those with DSP. However, very few patients in this study had an ICD-9 diagnosis indicating a rarer subtype of neuropathy, and a sensitivity analysis excluding patients with a diagnosis of mononeuritis multiplex, demyelinating neuropathy or hereditary neuropathy, did not significantly change our results. Many patients were included who have a known cause of neuropathy, and these patients may not require any work up for the cause of their neuropathy. On the other hand, 80% of the patients without diabetes were coded as idiopathic neuropathy, and the patterns observed in this group were similar to the entire cohort. Another limitation is that there are likely patients with neuropathy that did not receive an ICD-9 diagnosis and our population may be biased towards those with more severe neuropathy. Yet the high incidence of neuropathy in our cohort gives support to the likelihood that we are capturing a large proportion of the population with neuropathy. An additional limitation is that we were unable to investigate detailed information on why patients are receiving specific tests. For instance, some of the patients who had MRIs performed may have had another indication for this test, such as spinal arthritis. It is also possible that patients with a neuropathy diagnosis were more likely to see a specialist in neurology, which subsequently led to an increase in neurology-specific tests. However, the magnitude of the MRI utilization makes these factors unlikely to account for all of these tests. We also do not know whether the increase in Medicare expenditures is specifically related to the evaluation of neuropathy. We investigated total expenditures, which includes other non-diagnostic test related expenditures. On the other hand, the expenditures increased substantially around the time of diagnosis, and then decreased towards but not entirely back to that of the baseline period in the subsequent year. We also studied a Medicare population made up largely of patients over the age of 67. How these results apply to a younger population or one with private or no insurance is unclear.

Conclusions

In routine practice from 1998–2007, the evaluation of peripheral neuropathy involved substantial use of diagnostic tests with wide variation in testing patterns. MRI, a costly and low yield test, is frequently performed during the diagnostic period for neuropathy. On the other hand, the GTT, the optimal test for identifying one of the most common and treatable causes of DSP (IGT and diabetes), is rarely performed. The evaluation and management of peripheral neuropathy is associated with substantial increases in health expenditures. These findings indicate an important opportunity to improve the effectiveness and efficiency of the diagnostic evaluation of this prevalent disease.

Supplementary Material

Supplementary

Footnotes

Presented at the American Academy of Neurology meeting in Oahu, Hawaii on 4/12/2011

Dr. Kerber received speaker honoraria from the American Academy of Neurology 2010 and 2011 annual meetings, and performed consulting work for the American Academy of Neurology.

The other authors have no financial disclosures to report.

Conflicts of Interest:

Funding/support: Drs. Callaghan and Feldman are supported by a NIH T32 grant, the Katherine Rayner Program, and by the Taubman Medical Institute. The Health and Retirement Study is supported by the National Institute on Aging (U01 AG09740), and performed at the Institute for Social Research, University of Michigan. Dr. Langa is supported by National Institute on Aging grant R01 AG030155. Dr. Kerber is supported by NIH/NCRR #K23 RR024009 and AHRQ #R18 HS017690. Dr. Kerber also received speaker honoraria from the American Academy of Neurology 2010 and 2011 annual meeting, and performed consulting work for the American Academy of Neurology.

Authorship contributions: Brian Callaghan was involved in the design, interpretation of the statistical analysis, and wrote the manuscript. Ryan McCammon contributed to the study design and manuscript, performed and aided in the interpretation of the statistical analysis. Kevin Kerber helped with statistical interpretation and contributed to the manuscript. Xiao Xu contributed to the study design and to writing the manuscript. Ken Langa and Eva Feldman were integrally involved in study design, interpretation of the data, and were significantly involved in writing the manuscript.

Non-author contributions: James Burke reviewed and edited the manuscript.

References

1. Bharucha NE, Bharucha AE, Bharucha EP. Prevalence of peripheral neuropathy in the Parsi community of Bombay. Neurology. 1991 Aug;41(8):1315–1317. [PubMed]
2. Savettieri G, Rocca WA, Salemi G, et al. Prevalence of diabetic neuropathy with somatic symptoms: a door-to-door survey in two Sicilian municipalities. Sicilian Neuro-Epidemiologic Study (SNES) Group. Neurology. 1993 Jun;43(6):1115–1120. [PubMed]
3. Gregg EW, Sorlie P, Paulose-Ram R, et al. Prevalence of lower-extremity disease in the US adult population >=40 years of age with and without diabetes: 1999–2000 national health and nutrition examination survey. Diabetes Care. 2004 Jul;27(7):1591–1597. [PubMed]
4. England JD, Gronseth GS, Franklin G, et al. Practice Parameter: evaluation of distal symmetric polyneuropathy: role of laboratory and genetic testing (an evidence-based review). Report of the American Academy of Neurology, American Association of Neuromuscular and Electrodiagnostic Medicine, and American Academy of Physical Medicine and Rehabilitation. Neurology. 2009 Jan 13;72(2):185–192. [PubMed]
5. Smith AG, Singleton JR. The diagnostic yield of a standardized approach to idiopathic sensory-predominant neuropathy. Arch Intern Med. 2004 May 10;164(9):1021–1025. [PubMed]
6. Palumbo PJ, Elveback LR, Whisnant JP. Neurologic complications of diabetes mellitus: transient ischemic attack, stroke, and peripheral neuropathy. Adv Neurol. 1978;19:593–601. [PubMed]
7. Saperstein DS, Wolfe GI, Gronseth GS, et al. Challenges in the identification of cobalamin-deficiency polyneuropathy. Arch Neurol. 2003 Sep;60(9):1296–1301. [PubMed]
8. Kelly JJ, Jr, Kyle RA, O’Brien PC, Dyck PJ. Prevalence of monoclonal protein in peripheral neuropathy. Neurology. 1981 Nov;31(11):1480–1483. [PubMed]
9. Novella SP, Inzucchi SE, Goldstein JM. The frequency of undiagnosed diabetes and impaired glucose tolerance in patients with idiopathic sensory neuropathy. Muscle Nerve. 2001 Sep;24(9):1229–1231. [PubMed]
10. Singleton JR, Smith AG, Bromberg MB. Increased prevalence of impaired glucose tolerance in patients with painful sensory neuropathy. Diabetes Care. 2001 Aug;24(8):1448–1453. [PubMed]
11. McLeod JG, Tuck RR, Pollard JD, Cameron J, Walsh JC. Chronic polyneuropathy of undetermined cause. J Neurol Neurosurg Psychiatry. 1984 May;47(5):530–535. [PMC free article] [PubMed]
12. Weiss KH, Johanssen C, Tielsch A, et al. Malformation of the radial glial scaffold in the dentate gyrus of reeler mice, scrambler mice, and ApoER2/VLDLR-deficient mice. J Comp Neurol. 2003 May 19;460(1):56–65. [PubMed]
13. Dyck PJ, Klein CJ, Weigand SD. Does impaired glucose metabolism cause polyneuropathy? Review of previous studies and design of a prospective controlled population-based study. Muscle Nerve. 2007 Oct;36(4):536–541. [PubMed]
14. Hughes RA, Umapathi T, Gray IA, et al. A controlled investigation of the cause of chronic idiopathic axonal polyneuropathy. Brain. 2004 Aug;127(Pt 8):1723–1730. [PubMed]
15. Kissel JT. Peripheral neuropathy with impaired glucose tolerance: a sweet smell of success? Arch Neurol. 2006 Aug;63(8):1055–1056. [PubMed]
16. Knowler WC, Barrett-Connor E, Fowler SE, et al. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med. 2002 Feb 7;346(6):393–403. [PMC free article] [PubMed]
17. Santos-Rey K, Fernandez-Riejos P, Mateo J, Sanchez-Margalet V, Goberna R. Glycated hemoglobin vs. the oral glucose tolerance test for the exclusion of impaired glucose tolerance in high-risk individuals. Clin Chem Lab Med. 2010 Dec;48(12):1719–1722. [PubMed]