A retrospective analysis of health care utilization for patients with mitochondrial disease in the United States: 2008–2015

The data for this retrospective analysis are based on a commercially insured population, obtained from both the Truven Health Analytics MarketScan database and Milliman’s Consolidated Health Cost Guidelines Sources Database (CHSD) for the calendar years 2008 to 2015. Both sources of data represent patients receiving health insurance coverage through commercial means, including large employers, as well as individual, small- and large-group insurance carriers. Of note, other types of insurance coverage, including Medicaid and Medicare, are not included in this analysis. Given the complexity and chronic and debilitating nature of MD, the authors recognize that many patients ultimately may enroll in state and federally provided healthcare programs.

The Truven MarketScan database includes medical and drug claims data from employers and health plans, encompassing employees, their spouses, and dependents who are covered by employer-sponsored private health insurance. Information accessed included diagnostic codes, procedure codes, diagnosis-related grouping codes, National Drug Classification codes, site of service information, and the total amount paid through patient cost-sharing and by the payer, collectively referred to as the allowed amount. Milliman maintains administrative claims databases that link paid claims and encounter data that give detailed patient information across treatment sites and types of provider over time.

MD patients were identified through commercial claims based on the International Classification of Disease 9th and 10th Revision codes (ICD-9/− 10). Specifically, patients of any age with ICD-9277.87 (disorders of mitochondrial metabolism) or ICD-10 E88.40, E88.41, E88.42, and E88.49 (Table 1) were identified based on having at least 1 occurrence of the ICD-9 or ICD-10 code in the inpatient setting or 2 occurrences of the ICD-9 or ICD-10 code in any of the following settings: outpatient, professional, laboratory. MD patients were required to have at least 6 months of eligibility in the database after the occurrence of the first MD-related claim.

The overall total member population cohort is derived from both the Truven MarketScan database and Milliman’s CHSD for the calendar year 2015 only. The overall baseline population includes 759,024,730 member months for 2015, or an average of 63,252,061 member years. The multiple sclerosis (MS) and amyotrophic lateral sclerosis (ALS) cohorts were used as comparator cohorts for other neuromuscular disorders. Both cohorts were also derived from both the Truven Marketscan database and Milliman’s CHSD for the calendar year 2015.

For resource utilization analysis, utilization values are reported on a claim per thousand basis, which represents the average annual claims utilization for a given service per one thousand patients. Cost figures are presented on an Allowed Per Member Per Month (PMPM) basis. Allowed dollars represents the total health care expenditures for a given service, comprising both patient cost sharing, as established in the patient’s plan of benefits, and the carrier’s paid dollars. Since the MD cohort spans from 2008 to 2015, an adjustment factor was applied to all MD allowed PMPMs to normalize costs to July 1, 2015 as an adjustment for medical inflation. This adjustment factor was based on medical trend factors from the Milliman’s Health Cost Guidelines. Allowed PMPMs for the total population, MS, and ALS cohorts are as of July 1, 2015. Comorbidities were identified based on diagnosis codes used by health care professionals during the member’s eligibility period.

Of the patients who met the selection criteria for MD during the study period, the mean age of the study population (Table 2) was 26.1 years with 46.3% of patients aged 0 to 15 years (pediatric) and 53.7% aged 16 years and greater (adult). Males represented 46.8% (3832) and females represented 53.2% (4348) of the identified patients. A larger proportion of female MD patients (63.1%) were adults. In contrast, a larger proportion of male patients (57%) were of pediatric age. Of the adult patients, 62.5% were female.

Comorbidities in the pediatric and adult MD populations affect multiple bodily systems as shown in Table 2, which is typical of MD. There were 16 comorbid conditions that were present in > 10% of the MD pediatric population. Seven were neurological disorders, with seizure disorders and convulsion presenting in 47.4%, and autistic disorders presenting in 25.4% of MD pediatric patients. Respiratory diseases represented 2 out of the top 16 comorbidities, with asthma being present in 30.2% of MD pediatric patients. Immune disorders included disorders of the immune mechanism (16.4%) and myasthenia gravis/myoneural disorders and Guillain-Barre syndrome/inflammatory and toxic neuropathy (9.8%). Congestive heart failure (CHF) was present in 11.9% of patients, likely due to the high-energy demand of the cardiac muscle.

In the adult MD population there were 11 comorbid conditions that were present in > 10% of adult MD patients. Diabetes was the most prevalent condition, reported in 33% of MD adult patients. Three of the top 11 comorbidities were respiratory disorders, including asthma (22.6%), cardio-respiratory failure and shock (10.4%), and chronic obstructive pulmonary disease (10.4%). Neurological disorders present in > 10% of the adult MD patients included seizures disorders and convulsion (22.5%) and major depressive and bipolar disorders (18.3%). Top cardiac disorders included CHF (17.2%) and cardiac arrhythmias (13.6%). The high number of comorbidities speaks to the cost and burden of care associated with managing these multiple diagnoses.

Table 3 reports the percentages of all claims represented by each drug class. The most prescribed drug class for both pediatric and adult MD patients was anticonvulsants, representing 15.0% of all prescriptions for pediatrics and 10.7% of all prescriptions for adults. For the pediatric population, metabolic modifiers were the second most prescribed drug class, representing 8.0% of claims, while this drug class represented 2.2% of adult prescriptions. Opioid agonists were the second most prescribed class of drug for adults, representing 6.2% of prescriptions. In both pediatric and adult MD patient groups, proton pump inhibitors were the third most prescribed drug class, representing 4.0 and 3.3% in each group, respectively.

MD-related health care cost in comparison to the cost for the total member population is summarized in Table 4. The mean total MD allowed PMPM cost was $3958 ($455 per claim), which was significantly greater (p < 0.001) than the allowed PMPM cost for the total member population, with a mean of $427 ($206 per claim). These findings were consistent in both the pediatric and adult member population. Allowed PMPM cost for MD patients was higher for pediatric patients ($4829) than for adult patients ($3100), whereas in the total member population, the allowed PMPM cost was greater for the adults ($486) than pediatrics ($202). The average allowed per claim was approximately twice as much in the MD population compared to the total member population.

Health care resource utilization and cost were greater for MD patients than the total member population for all settings (Table 5). The most frequently used health resources in patients diagnosed with MD were prescription drugs (31%), physician visits (20%), and home health (10%). The top 3 greatest resources used in the total member population came from prescription drugs (41%), physician visits (20%), and laboratory testing (11%). The highest MD-related PMPM per claim were in the categories of inpatient hospitalizations ($769; 19%), surgery ($746; 19%), and prescription drugs ($582; 15%). Health care cost was higher for MD patients than for the total member population in all categories, with the exception of colonoscopy.

Pediatric MD patients utilized more physician specialties compared to the total member pediatric population as shown in Table 6. The top 3 physician specialties visited by MD pediatric patients were pediatric (24%), general practice (9%) and neurology (6%). The greatest increase in resource utilization in the physician setting for pediatric MD patients compared to the total member pediatric population was physical therapy (108.3-fold). Other specialties that showed a greater than 50-fold increase in utilization for MD patients were gastroenterology (61.0 fold), intensivist (56.5 fold), and neurology (52.0 fold).

Across all physician specialties, resource utilization was greater for the MD adult population compared to the total member adult population, as shown in Table 6. The top 3 physician specialties seen by MD patients were internal medicine (15%), general practice (10%) and family practice (9%). The greatest increase in resource utilization in the physician setting for MD compared to total member population was neurology (19.2-fold), which is consistent with the large proportion of neurologic comorbidities in the MD population (Table 2).

Health care resource utilization and cost for adult MD patients were compared to those of 125,434 multiple sclerosis members and 4579 amyotrophic lateral sclerosis members for calendar year 2015 (Fig. 1). Resource utilization per 1000 for MD was 95,404, which was lower than ALS (107,133) but greater than MS (71,859). Overall MD allowed PMPM costs were comparable to MS and ALS, although the allowed PMPM per claim for MS was higher. Given the relatively small number of pediatric members with MS (n = 443) and ALS (n = 46), comparison of a pediatric cohort was not conducted.

As shown in Fig. 1, the top 3 drivers of health care cost for MD and ALS patients were inpatient, surgery and prescription drugs. In contrast, the most significant expense for MS patients was prescriptions drugs (69%). As compared to the total population, MD, MS and ALS patients utilized the emergency room and inpatient resources at a higher percentage than the total member population (Fig. 2). A greater percentage of MD patients (54.9%) experienced an ER visit compared to MS (30.7%) or ALS (34.1%) patients. Inpatient visits were comparable in MD (27.8%) and ALS patients (20.9%) but both were greater than the 9.1% of MS utilizers in the inpatient setting.

This study has several limitations. There are inherent faults in choosing the restrictive ICD codes. Given the relative rarity of diagnosis, there are likely patients with certain mitochondrial disorders that are never included in the database either because they have never been diagnosed with a mitochondrial disease, or the provider may not be familiar with the specific ICD terminology, or may have chosen a more general ICD code such as a code for myopathy or encepholopathy. This factor will result in patients with real mitochondrial diseases not being included in this data analysis. Conversely, because the choice of the ICD terminology does not require a high level of clinical certainty that a patient has MD, there could be patients included in the database that were suspected of having a MD but never had the clinical evidence to justify that choice of the specific ICD code. This issue has been outlined in recent publications [7]. We also recognize that the ICD codes may not capture the entire spectrum of MDs. Inclusion into this analysis required either 1 inpatient or 2 outpatient mitochondrial disease-specific claims and may have missed MDs with a legitimate single outpatient claim. Therefore, the MD cohort may have been underrepresented or have been preselected to identify a more severe patient population. In addition, mitochondrial disease is unique when compared to other well-studied neuromuscular disorders in three key ways: a) characterized by multisystem disease, b) MD has both acute and chronic aspects, with patients potentially living many years after diagnosis with minimal or progressive symptoms while simultaneously experiencing multiple acute exacerbations, c) there is not a uniform approach to disease management, leading to a variety in the types of prescription drugs and therapies prescribed, due to the overall heterogeneity of the disease. This analysis is limited to patients covered by private insurance plans. There may be inherent differences between these patients and patients covered by public insurance plans such as Medicare and Medicaid. The National Health Interview Survey 2017 reports that the percent of insured adults under 65 years is much higher under private plans (65.4%) as compared to public plans (25.3%). This may affect the generalizability of this analysis [21]. We did not analyze the differences in claims and utilization based on geographical areas/regions. This is because there are limited MD specialists and providers in the US, and most patients need to travel to the nearest specialist provider to receive treatment. Thus, conclusions from a geographical analysis would be driven by location of specialists rather than provisions of care based on state or region [22]. These variables contribute to the complexity of the disease and to the difficulties associated with a large population-based claims analysis.