Abstract
Introduction
Myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) is a chronic illness characterized by marked functional limitations and fatigue. Electronic health records can be used to estimate incidence of ME/CFS but may have limitations.
Methods
The authors used International Classification of Diseases (ICD) diagnosis codes to identify all presumptive cases of ME/CFS among 9- to 39-year-olds from 2006 to 2017. The authors randomly selected 200 cases for medical record review to classify cases as confirmed, probable, or possible, based on which and how many current clinical criteria they met, and to further characterize their illness. The authors calculated crude annual rates of ME/CFS coding stratified by age and sex using only those ICD codes that had identified confirmed, probable, or possible ME/CFS cases in the medical record review.
Results
The authors identified 522 individuals with presumptive ME/CFS based on having ≥ 1 ICD codes for ME/CFS in their electronic medical record. Of the 200 cases selected, records were available and reviewed for 188. Thirty (15%) were confirmed or probable ME/CFS cases, 39 (19%) were possible cases, 119 (60%) were not cases, and 12 (6%) had no medical record available. Confirmed/probable cases commonly had chronic pain (80%) or anxiety/depression (70%), and only 13 (43%) had completed a sleep study. Overall, 37 per 100,000 had ICD codes that identified confirmed, probable, or possible ME/CFS. Rates increased between 2006 and 2017, with the largest absolute increase among those 30–39 years old.
Conclusions
Using ICD diagnosis codes alone inaccurately estimates ME/CFS incidence.
Keywords: chronic conditions, electronic medical record, epidemiology, ambulatory care, multimorbidity, pain management, teen
Introduction
Myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) is a chronic illness characterized primarily by longstanding functional limitations accompanied by disabling fatigue in the absence of other explanatory disease.1 ME/CFS is typically diagnosed among young and middle-aged adults, with a lesser occurrence among children. Fatiguing comorbid conditions in older adults may confound its diagnosis later in life. A study using population-based registry data in Norway from 2008 to 2012 found that the two peak age ranges for ME/CFS onset were 10–19 and 30–39 years.2 Diagnosis of ME/CFS is symptom-based, as there is currently no clinical test that can confirm the presence of the illness. In 2015, the Institute of Medicine (IOM), now called the National Academy of Medicine, recommended a clinical case definition for ME/CFS focusing on five symptoms; three core criteria are required (impairment in functioning associated with fatigue, post-exertional malaise, and unrefreshing sleep) with at least one of the other two symptoms also present (cognitive impairment or orthostatic intolerance).3 Although there are other case definitions used in research,4–6 the authors used the IOM clinical case definition adopted by the US Health and Human Services. The IOM committee commissioned extensive literature reviews and statistical analyses and considered input from diverse stakeholders before agreeing upon their diagnostic criteria, which are intended for use in a clinical setting.7
A variety of factors have been anecdotally reported as triggers for the onset of ME/CFS symptoms, including viral infection, emotional trauma, hormonal changes, or abrupt changes in health.8,9 Case reports have suggested that vaccines could potentially be linked to the development of ME/CFS, including the human papillomavirus (HPV)10–13 and COVID vaccines.14,15 However, large observational studies of vaccines for HPV,16–18 hepatitis B,19 meningococcus,20 or seasonal influenza have not found any association between vaccination and ME/CFS.21 These studies have utilized diagnosis codes and/or repeated visits for fatigue over extended months as surrogate indicators for ME/CFS. Yet, it is unclear how accurately ME/CFS is captured through diagnosis codes. In the United States, International Classification of Diseases (ICD) codes are also used for billing and reimbursement purposes, practices that have the potential to impact which codes are selected for the clinical record.
The objectives of this study were to determine the accuracy of ME/CFS case ascertainment using electronic health record (EHR) data in an integrated health care system and to describe characteristics of patients who meet most or all IOM diagnostic criteria. The authors also calculated annual rates of ≥ 1 entries of an ME/CFS diagnosis code within the EHR, over a 12-year period, to evaluate use of ME/CFS coding patterns over time.
Methods
Setting and study population
The authors conducted this study within Kaiser Permanente Northwest, an integrated health care delivery system serving > 635,000 members in Oregon and southwest Washington State. The Kaiser Permanente Northwest Institutional Review Board approved the study protocol and granted a waiver of informed consent.
The authors estimated the population at risk for ME/CFS from the number of Kaiser Permanente Northwest members ages 9 through 39 years with at least 30 d of continuous enrollment in the health care system between January 1, 2006, and December 31, 2017. The authors chose 9 years as the lower end of the age range because it is the youngest age at which children can receive the HPV vaccine. The authors chose 39 years as the upper end of the age range to allow adequate follow-up after HPV vaccination and because of the previously identified peak in incidence for those 30–39 years old.2
Case ascertainment
The authors defined all individuals within the study population with ≥ 1 ICD-9 or ICD-10 code for chronic fatigue syndrome, encephalitis, myelitis, encephalomyelitis, or post-viral fatigue syndrome in their Kaiser Permanente Northwest EHR from 2006 through 2017 as presumptive cases. The date of the initial entry of the diagnosis code was considered the “automated index date.” Specific diagnosis codes included ICD-9 codes 780.71 (chronic fatigue syndrome) and 323.9 (unspecified causes of encephalitis, myelitis, and encephalomyelitis) and ICD-10 codes R53.82 (chronic fatigue, unspecified) and G93.3 (post-viral fatigue syndrome).22
Abstraction sample, chart review, and abstraction procedures
The authors selected a random sample of 200 presumptive cases for chart review to confirm ME/CFS diagnoses and to describe the clinical course and symptomatology. Trained chart abstractors manually reviewed medical records and recorded the following data in a REDCap23 abstraction form: demographic information, abstracted index date (ie, initial diagnosis date, if chart review indicated it was different from automated index date), presence of ME/CFS symptoms (based on IOM criteria),3 symptom onset date, medication history, laboratory results, related medical procedures, and family history of ME/CFS or fibromyalgia. Abstractors searched for these data in all available medical records from 7 years before through 7 years after the automated index date.
One abstractor attempted to review medical records of all 200 presumptive cases, with a second abstractor reviewing 100 presumptive cases, plus a random 10% sample of the remaining presumptive cases, for quality assurance. The study team discussed discrepancies to achieve consensus. The primary adjudicator (MLH) reviewed abstractions to determine whether the four symptom criteria of ME/CFS were reported (ie, the three core criteria, then one of either cognitive impairment or orthostatic intolerance). The authors categorized findings as “confirmed ME/CFS” when medical documentation described all four criteria, “probable ME/CFS” with description of three core criteria, “possible ME/CFS” with description of two core criteria, and “not ME/CFS” for description of one or no core criteria. A clinical adjudicator with experience providing health care for patients with ME/CFS (EGL) reviewed all possible and probable cases and 10% of all remaining cases for quality assurance. Abstraction and adjudication forms are available on request.
Statistical Analyses
Descriptive analyses of abstraction sample
The authors described the proportions of presumptive ME/CFS cases classified as confirmed, probable, possible, and not ME/CFS, stratified by specific ICD-9 or ICD-10 code entered. The authors similarly described demographic characteristics, diagnostic tests and settings, symptomatology, clinical workup, comorbid conditions, and medications—all stratified by chart-confirmed ME/CFS case status.
Starting in 2016, ICD-10 code R53.82 was one of the approved indications for outside referral to a naturopath. In a post hoc analysis, the authors compared the proportions of those with confirmed/probable ME/CFS and possible/not ME/CFS referred to a naturopath. The authors used the Fisher’s exact test to evaluate an association between case likelihood status and naturopathic referral.
Annual rates of ME/CFS Index ICD Coding
The authors conducted a population-based analysis of the annual rate of first (index) ME/CFS code entry. The authors included as cases all 9- to 39-year-old males and females who had received ≥ 1 ICD-9 or ICD-10-Clinical Modification (CM) codes used for ME/CFS (ICD-9 code 780.71, chronic fatigue syndrome, and ICD-10 code R53.82, chronic fatigue, unspecified); in the chart review, these had been the only two diagnosis codes associated with confirmed, probable, or possible cases of ME/CFS (Figure 1). Cases were enrolled in the Kaiser Permanente Northwest health plan at the time of initial (index) code entry. The authors calculated crude annual rates of index ME/CFS code entry by dividing the number of index codes used each calendar year by the number of eligible person-years. The authors calculated age- and sex-stratified annual rates of index ME/CFS code entry per 100,000 person-years for each calendar year over the 12-year period from 2006 to 2017. The authors used the age strata of 9–19, 20–29, and 30–39 years, calculating age as of the index date for cases. Among non-cases, the authors used birth date to determine the year of age for person-time accrual. Censoring of cases occurred on the date of first entry of an ME/CFS diagnosis code (automated index date). Person-time calculation began at the start of the study period, the ninth birthday, or the first health care system enrollment date, whichever occurred last. Person-time calculation ended at date of index code entry among cases or at date of health care system disenrollment, death, the 40th birthday, or the end of the study period, whichever occurred first, for non-cases. The authors included individuals who did not have their sex available in the medical record in overall incidence rates but excluded them from sex-stratified analyses.
Figure 1:
Identification and confirmation of ME/CFS cases from electronic medical records of Kaiser Permanente Northwest health care system members 9–39 years of age between 2006 and 2017. Analysis of annual coding rate was based on a subset of presumptive cases. ICD = International Classification of Diseases; MF/CFS = myalgic encephalomyelitis/chronic fatigue syndrome.
Statistical software
The authors exported abstraction data from REDCap and performed both descriptive and incidence analyses using SAS statistical software (version 9.4; SAS Institute, Inc).
Results
Characterization of the abstraction sample
Chart confirmation of ME/CFS case status
Of the present study’s 200-case abstraction sample, the authors completed a full chart review for 188 presumptive cases (Figure 1). The authors could not fully classify the remaining 12 (6%) presumptive cases due to there being no record available for review. For these presumptive cases, the authors categorized ME/CFS status as “unable to determine.” For 63 individuals, there was indication from medical documentation of a coding error at the index encounter, with no information to support the single ME/CFS diagnosis code entry. These cases, and others with no reported core diagnostic criteria of ME/CFS in full abstraction, were categorized by the adjudicators as “not ME/CFS” (n = 119; 60%). The authors classified 6 (3%) of the 200-case abstraction sample as “confirmed ME/CFS” cases, 24 (12%) as “probable” cases, and 39 (19%) as “possible” cases. Given the small numbers of confirmed and probable cases, the authors combined these groups for descriptive analyses.
Demographic characteristics
Table 1 summarizes the demographic characteristics of the abstraction sample, stratified by chart-confirmed ME/CFS classification, compared to the general population of Kaiser Permanente Northwest members aged 9–39 years. The abstracted population was older, more likely to be female, and more likely to be identified as White than the general Kaiser Permanente Northwest population in the same age range. Similarly, most cases with confirmed or probable ME/CFS (76.6%) were female and nearly all were White (96.6%). Patients in this study’s abstraction sample had been enrolled in the health care system for a mean of 3.4 years [standard deviation (SD) = 5.2; range = 0–24], compared to 1.5 years (SD = 2.5; range = 0–15) for the general patient population.
Table 1:
Demographic characteristics of 200 presumptive ME/CFS cases, compared to the general Kaiser Permanente Northwest population1
| Characteristic | Confirmed or probable cases (n = 30) [n (%)] |
Possible cases (n = 39) [n (%)] |
Not ME/CFS (n = 119) [n (%)] |
Unable to determine (n = 12) [n (%)] |
Total (N = 200) [n (%)] |
General Kaiser Permanente Northwest population (N = 223,611) (n) a |
|---|---|---|---|---|---|---|
| Age, y | ||||||
| 9–19 | 2 (6) | 2 (5) | 16 (13) | 2 (17) | 22 (11) | 71,516 (32) |
| 20–29 | 8 (27) | 18 (46) | 43 (36) | 3 (25) | 72 (36) | 72,916 (33) |
| 30–39 | 20 (67) | 19 (49) | 60 (50) | 7 (58) | 106 (53) | 79,179 (35) |
| Sex b | ||||||
| Female | 23 (77) | 34 (87) | 76 (64) | 12 (100) | 145 (73) | 114,158 (51) |
| Male | 7 (23) | 5 (13) | 43 (36) | 0 (0) | 55 (28) | 109,338 (49) |
| Race c | ||||||
| White | 29 (97) | 34 (87) | 94 (79) | 4 | 161 (81) | 145,467 (65) |
| Black | 0 (0) | 1 (3) | 6 (5) | 0 (0) | 7 (4) | 9037 (4) |
| American Indian or Alaskan Native | 0 (0) | 1 (100) | 0 (0) | 0 (0) | 1 ( < 1) | 1692 ( < 1) |
| Native Hawaiian/ Pacific Islander | 1 (3) | 0 (0) | 1 (1) | 0 (0) | 2 (1) | 2649 (1) |
| Asian | 0 (0) | 0 (0) | 6 (5) | 0 (0) | 6 (3) | 15,284 (7) |
| Not listed above | 0 (0) | 0 (0) | 1 (1) | 1 (50) | 2 (1) | 1765 ( < 1) |
| Unknown | 0 (0) | 4 (10) | 14 (12) | 7 (29) | 25 (13) | 47,717 (21) |
| Ethnicity | ||||||
| Hispanic | 0 (0) | 3 (8) | 10 (8) | 1 (8) | 14 (7) | 25,920 (12) |
| Non-Hispanic | 5 (17) | 4 (10) | 9 (8) | 0 (0) | 18 (9) | 28,026 (13) |
| Unknown | 25 (83) | 32 (82) | 100 (84) | 11 (92) | 168 (84) | 169,665 (76) |
| Mean enrollment, y d | 3.6 | 2.6 | 3.9 | 1.6 | 3.4 | 1.5 |
General population data based on snapshot of Kaiser Permanente Northwest Virtual Data Warehouse population for ages 9–39 years on December 31, 2017.
For n = 115 individuals in the Kaiser Permanente Northwest population, sex was unknown.
Race categories are not mutually exclusive.
Mean length of Kaiser Permanente Northwest health plan enrollment through index diagnosis date, in years.
MF/CFS, myalgic encephalomyelitis/chronic fatigue syndrome.
Diagnostic characteristics of ME/CFS
Table 2 shows specific diagnosis codes, clinical settings, and practitioner types involved for the initial diagnosis code entry among those in the abstraction sample, organized by chart-confirmed ME/CFS case status. Most (93%) confirmed or probable cases received their index code during outpatient or telephone visits, and the majority occurred in one of three departments: internal medicine (33%), family practice (20%), or neurology (20%).
Table 2:
Diagnostic codes, settings, and index diagnosis practitioner types among 200, presumptive ME/CFS cases, by chart-confirmed ME/CFS case status
| Characteristic | Confirmed or probable cases (n = 30) [n (%)] |
Possible cases (n = 39) [n (%)] |
Not ME/CFS (n = 119) [n (%)] |
Unable to determine (n = 12) [n (%)] |
Total (N = 200) [n (%)] |
|---|---|---|---|---|---|
| Index diagnosis code (Col%) (Col%) | |||||
| ICD-9 code 780.71 (chronic fatigue syndrome) | 24 (80) | 23 (59) | 58 (49) | 7 (58) | 112 (56) |
| CD-10 code R53.82 (chronic fatigue, unspecified) | 6 (20) | 16 (41) | 35 (29) | 5 (42) | 62 (31) |
| CD-9 code 323.9 (encephalitis NOS) | 0 (0) | 0 (0) | 22 (18) | 0 (0) | 22 (11) |
| CD-10 code G93.3 (post-viral fatigue syndrome) | 0 (0) | 0 (0) | 4 (3) | 0 (0) | 4 (2) |
| Index diagnosis setting | |||||
| Outpatient (fatigue secondary diagnosis) | 15 (50) | 22 (56) | 51 (43) | 10 (83) | 98 (49) |
| Outpatient (fatigue primary diagnosis) | 9 (30) | 12 (31) | 36 (30) | 2 (17) | 59 (30) |
| Inpatient | 2 (7) | 1 (3) | 15 (13) | 0 (0) | 18 (9) |
| Telephone | 4 (13) | 2 (5) | 9 (7) | 0 (0) | 15 (8) |
| Emergency | 0 (0) | 1 (3) | 8 (7) | 0 (0) | 9 (4) |
| Unknown | 0 (0) | 1 (3) | 0 (0) | 0 (0) | 1 ( < 1) |
| Index diagnosis practitioner type | |||||
| Family practice | 6 (20) | 14 (36) | 28 (24) | 5 (42) | 53 (26) |
| Internal medicine | 10 (33) | 8 (21) | 25 (21) | 0 (0) | 43 (22) |
| Hospital | 2 (7) | 2 (5) | 23 (19) | 1 (8) | 28 (14) |
| Neurology | 6 (20) | 3 (8) | 3 (3) | 0 (0) | 12 (6) |
| Mental health | 1 (3) | 3 (8) | 5 (4) | 0 (0) | 9 (5) |
| Urgent care | 1 (3) | 1 (2) | 5 (4) | 1 (8) | 8 (4) |
| Rheumatology | 1 (3) | 4 (10) | 1 ( < 1) | 0 (0) | 6 (3) |
| Gastroenterology | 0 (0) | 0 (0) | 3 (3) | 1 (8) | 4 (2) |
| Other | 3 (10) | 4 (10)\ | 25 (22) | 4 (33) | 37 (18) |
ICD, International Classification of Diseases; MF/CFS, myalgic encephalomyelitis/chronic fatigue syndrome; NOS, not otherwise specified.
Features of Chart-Confirmed or Probable ME/CFS Cases
Symptom onset date
Symptom onset dates could be determined (3%) or estimated (97%) for all 30 confirmed/probable cases. For 43% of the confirmed/probable cases, symptom onset was ≥ 5 years before the automated index date. The median length of time between symptom onset and index code entry was 5 years; mean was 7.3 years (range = 0–28 years).
Comorbid medical conditions and medications
Table 3 lists comorbid medical conditions among those with confirmed or probable ME/CFS. The most common comorbid medical conditions for confirmed/probable cases included chronic pain (80%), major depression (70%), and anxiety disorders (70%). Two confirmed/probable cases (7%) had only fibromyalgia as a cause of chronic pain, 30% had another cause of chronic pain (without fibromyalgia), and 43% had both fibromyalgia and another underlying cause of chronic pain.
Table 3:
Frequency of comorbid medical conditions among 125 abstracted, presumptive ME/CFS cases, by chart-confirmed ME/CFS case status
| Medical condition | Confirmed or probable cases (n = 30) [n (%)] | Possible cases (n = 39) [n (%)] | Not ME/CFS (n = 56) [n (%)] |
Total (N = 125) [n (%)] |
|---|---|---|---|---|
| Other chronic pain | 22 (73) | 26 (67) | 35 (62) | 83 (66) |
| Major depression | 21 (70) | 22 (56) | 27 (48) | 70 (56) |
| Anxiety disorder | 21 (70) | 25 (64) | 22 (39) | 68 (54) |
| Fibromyalgia | 15 (50) | 14 (36) | 8 (14) | 37 (30) |
| Bowel disorder | 8 (27) | 16 (41) | 6 (11) | 30 (24) |
| Hypothyroidism | 7 (23) | 7 (18) | 8 (14) | 22 (18) |
| Obstructive or central sleep apnea | 8 (27) | 2 (5) | 7 (12) | 17 (14) |
| Anemia | 3 (10) | 3 (8) | 8 (14) | 14 (11) |
| Other autoimmune condition (besides lupus, rheumatoid arthritis) | 1 (3) | 3 (8) | 2 (4) | 6 (5) |
| Post-concussion syndrome | 0 (0) | 3 (8) | 3 (5) | 6 (5) |
| Lyme disease or other tick-borne infections | 1 (3) | 0 (0) | 3 (5) | 4 (3) |
| Chiari malformation or cervical spine stenosis | 0 (0) | 1 (3) | 1 (2) | 2 (2) |
| Narcolepsy | 0 (0) | 0 (0) | 2 (4) | 2 (2) |
| Hyperthyroidism | 1 (3) | 0 (0) | 0 (0) | 1 (1) |
| Adrenal insufficiency | 1 (3) | 0 (0) | 0 (0) | 1 (1) |
| Athletic overtraining syndrome | 0 (0) | 0 (0) | 0 (0) | 0 (0) |
| Systemic lupus erythematosus | 0 (0) | 0 (0) | 0 (0) | 0 (0) |
Most (90%) individuals categorized as confirmed/probable cases received ≥ 1 prescriptions for medications that could contribute to fatigue within the abstraction period (Table 4). Some patients received repeated dispensations of multiple fatigue-causing medications (data not shown). These included 73% of cases receiving opioids and 66% receiving either a benzodiazepine, muscle relaxant, first-generation antihistamine, or non-benzodiazepine hypnotic. Common types of fatigue-treating medications included stimulants (30% of confirmed/probable cases), bupropion (20%), and modafinil or armodafinil (17%).
Table 4:
Medication history for 125 abstracted, presumptive ME/CFS cases, by chart-confirmed ME/CFS case status
| Medication type | Confirmed or probable cases (n = 30) [n (%)] |
Possible cases (n = 39) [n (%)] |
Not ME/CFS (n = 56) [n (%)] |
Total (N = 125) [n (%)] |
|---|---|---|---|---|
| Potentially fatigue-causing medications | ||||
| Antidepressants | 11 (37) | 15 (38) | 19 (34) | 45 (36) |
| Benzodiazepines | 8 (27) | 11 (28) | 12 (21) | 31 (25) |
| Non-benzodiazepine hypnotics | 9 (30) | 4 (10) | 2 (4) | 15 (12) |
| First-generation antihistamines | 8 (27) | 9 (23) | 6 (11) | 23 (18) |
| Antipsychotics | 1 (3) | 2 (5) | 2 (4) | 5 (4) |
| Muscle relaxants | 8 (27) | 9 (23) | 9 (16) | 26 (21) |
| Antispasmodics | 1 (3) | 1 (3) | 2 (4) | 4 (3) |
| Beta-blockers | 1 (3) | 3 (7) | 6 (11) | 10 (8) |
| Montelukast | 1 (3) | 0 (0) | 2 (4) | 3 (2) |
| Melatonin | 2 (7) | 2 (5) | 2 (4) | 6 (5) |
| Opioids | 22 (73) | 19 (49) | 27 (48) | 68 (54) |
| Cancer therapiesa | 0 (0) | 3 (8) | 1 (2) | 4 (3) |
| Medications to treat fatigue | ||||
| Bupropion | 6 (20) | 3 (8) | 6 (11) | 15 (12) |
| Sertraline | 7 (23) | 6 (15) | 8 (14) | 21 (17) |
| Stimulantsb | 9 (30) | 9 (23) | 4 (7) | 22 (18) |
| Modafinil or armodafinil | 5 (17) | 5 (13) | 5 (9) | 15 (12) |
| Amantadine | 1 (3) | 1 (3) | 0 (0) | 2 (2) |
Examples include tamoxifen, aromatase inhibitors, hydroxyurea, and others.
Examples include amphetamine–dextroamphetamine, methylphenidate, and others.
ME/CFS, myalgic encephalomyelitis/chronic fatigue syndrome.
Diagnostic laboratory tests and procedures
Confirmed/probable cases had received blood testing or other diagnostic tests. The duration of time over which these tests occurred ranged from 1 to 17 years (mean = 3.57 years, SD = 4.75). Tests included complete blood count (100% of cases), thyroid function (98%), vitamin B12 (96%), comprehensive metabolic profile (87%), testosterone (78% of males), C-reactive protein (69%), basic metabolic profile (56%), hepatitis C surface antigen (56%), tissue transglutaminase (51%), and hepatitis B surface antigen (49%). Some of these blood tests had been ordered repeatedly; among confirmed/probable cases, complete blood count was ordered a mean of 13.4 times (median = 11; interquartile range = 5–16), and thyroid function a mean of 7.8 times (median = 5; interquartile range = 4–9). Common diagnostic procedures or screenings included depression screening [18 (60%) of confirmed/probable cases], electrocardiogram [16 (53%)], polysomnography or home sleep tests [13 (43%)], and echocardiogram [9 (30%)]. Formal screening for anxiety was comparably uncommon [6 (20%)], as was tilt table testing [3 (10%)].
Family history of ME/CFS or fibromyalgia
Family history of ME/CFS or fibromyalgia was not well documented in the EHR. Among those queried about family history of these conditions, 25% (3 of 12) reported a family history of ME/CFS and 36% (4 of 11) reported a family history of fibromyalgia.
Post hoc comparison of naturopathic referral proportions
A higher proportion of individuals with possible/not ME/CFS (29%) had a referral to a naturopath as compared to individuals with confirmed/probable ME/CFS (17%); however, this finding was not statistically significant (risk difference = 12.7%, 95% CI: –37% to 34%, p = 0.66).
Annual Rates of ME/CFS Index ICD Coding
From January 2006 through December 2017, the authors identified 455 individuals with index ICD code entries for ME/CFS in the study population with either ICD-9 code 780.71, or ICD-10 code R53.82. The overall crude annual rate of index ME/CFS code entry was 21.5 index code entries per 100,000 person-years. This varied by age group, increasing with older age from 5.2 per 100,000 person-years among those aged 9–19 years to 37.0 per 100,000 person-years among those aged 30–39 years (Table 5). Crude annual rates of index ME/CFS code use also varied by sex (9.7 per 100,000 person-years for males, 32 per 100,000 person-years for females), with higher rates among females than males in every age group.
Table 5:
Crude annual index coding rates for ME/CFS among individuals 9–39 years of age, using ICD diagnosis codes, overall, and by sex and age group—2006–2017.1
| Age group, ya | No. individuals b | Person-years | ME/CFS cases [group n (%)] | ME/CFS annual coding rates c | |
|---|---|---|---|---|---|
| Overall | 9–19 | 235,402 | 772,452 | 40 (8.8) | 5.2 |
| 20–29 | 300,414 | 644,632 | 143 (31.4) | 22.2 | |
| 30–39 | 280,746 | 736,048 | 272 (59.8) | 37.0 | |
| Total population | 653,215 | 2,153,132 | 455 (100) | 21.5 | |
| Males | 9–19 | 119,496 | 393,890 | 18 (17.6) | 4.6 |
| 20–29 | 143,966 | 305,688 | 31 (30.4) | 10.2 | |
| 30–39 | 138,325 | 349,357 | 53 (52.0) | 15.2 | |
| All males | 322,786 | 1,048,936 | 102 (100) | 9.7 | |
| Females | 9–19 | 115,891 | 378,561 | 22 (6.2) | 5.8 |
| 20–29 | 156,420 | 338,944 | 112 (31.7) | 33.1 | |
| 30–39 | 142,400 | 386,691 | 219 (62.0) | 56.7 | |
| All females | 330,376 | 1,104,196 | 353 (100) | 32.0 |
Age at first identified automated ME/CFS diagnosis code.
Due to age progression over the study period, the age groups were not restricted to mutually exclusive individuals and do not sum to the overall/sex totals
Annual index coding rates per 100,000 person-years; automated diagnosis codes included ICD-9 code 780.71 and ICD-10 code R53.82.
ICD, International Classification of Diseases; ME/CFS, myalgic encephalomyelitis/chronic fatigue syndrome.
Between 2006 and 2017 the use of index ME/CFS codes increased in each of the three age groups; these increases appeared to occur during the second half of the study period (Figure 2). Those in the 30- to 39-year-old age group had the largest absolute increase in ME/CFS coding, from 28.4 index codes entries per 100,000 in 2006 to 67.6 per 100,000 in 2017. Rates in the 9- to 19-year-old age group increased from 4.6 per 100,000 in 2006 to 15.1 per 100,000 in 2017. Annual index code entry rate in the 20- to 29-year-old age group also increased > 2-fold, from 18 code entries per 100,000 in 2006 to 41.7 per 100,000 in 2017. Figure 3 shows annual index code rates stratified by age group and sex. Changes in coding rates over time varied in the different subgroups; the greatest increase occurred among 30- to 39-year-old females, from an annual rate of 29.1 per 100,000 in 2006 to 93.3 per 100,000 in 2017.
Figure 2:
Annual coding rates for prespecified ME/CFS ICD-9 diagnosis code 780.71 and ICD-10 diagnosis code R53.82, per 100,000 person-years; age at first entry of ME/CFS code. ICD = International Classification of Diseases; ME/CFS = myalgic encephalomyelitis/chronic fatigue syndrome.
Figure 3:
Annual coding rates for prespecified ME/CFS ICD-9 diagnosis code 780.71 and ICD-10 diagnosis code R53.82, per 100,000 person-years; age at first entry of ME/CFS code. ICD = International Classification of Diseases; ME/CFS = myalgic encephalomyelitis/chronic fatigue syndrome.
Discussion
In a sample of 200 presumptive cases with an ME/CFS diagnosis code in an integrated health care system from 2006 to 2017, only 3% met enough diagnostic criteria to be considered a confirmed case of ME/CFS following medical record review; 12% met enough criteria to be considered probable cases. The reasons for the low case confirmation rate are not clear. It is possible that ME/CFS is not as common as previously reported.24,25 It is also possible that clinicians do not routinely query patient s about or fully document symptoms of ME/CFS. A previous study from the UK using patient-reported questionnaires on symptom and functional capacity rather than medical record review found 66.8% of cases coded as ME/CFS were confirmed as meeting at least one definition of ME/CFS criteria.24 The UK has had national guidelines in place that could have helped clinicians recognize the illness. In addition, patient-reported questionnaires allow a more in-depth view of symptoms that may not be documented in medical records. A consistent diagnostic approach, with standardization of documentation, is challenging for any symptom-based diagnosis, especially when the criteria for making the diagnosis change. In the last 20 years, there have been several different ME/CFS case definitions,26 though the 2015 IOM criteria are now commonly recommended.7,27 Awareness of these criteria may increase with the recent creation of a new diagnosis code specific for ME/CFS28 to replace previously ambiguous ICD-9- or ICD-10-CM codes (ie, ICD-9-CM “malaise and fatigue” and ICD-10-CM “chronic fatigue, unspecified”).
This study found that diagnosis of confirmed/probable ME/CFS occurred almost entirely in outpatient settings, suggesting that EHR data could be used to identify cases. Yet, the date of the first code entry of ME/CFS was not a reliable indicator of ME/CFS symptom onset and was delayed by a median of 5 years from the date of first reported symptoms. Further, it seems that a long follow-up period may be necessary to accurately identify ME/CFS cases using EHR data without case confirmation. In an informal post hoc analysis, the authors found that the diagnosis code specificity for confirmed or probable ME/CFS tended to increase with the number of entries of the code in the medical record. Lengthening the time over which codes occurred, from 2 to 5 years, also improved case-finding accuracy. For example, the specificity of 6 ME/CFS code entries over 5 years of time was 84% and sensitivity was 70% (data not shown). Taken together, the present study’s results indicate that studies of ME/CFS relying on retrospective EHR data would require decades of data to evaluate potential causative factors and subsequent ME/CFS diagnoses in a large, closed population (and probably still require some chart review for case confirmation).
The authors found that cases they categorized as confirmed/probable ME/CFS had received some diagnostic workup, although only 43% of probable or confirmed cases had completed polysomnography or a home sleep study. Comorbid conditions were common, and their relationship to fatigue symptoms was often complex. Eighty percent of confirmed or probable cases in the current study had chronic pain, and 70% had comorbid depression or anxiety (aligning with the results of prior studies).24,29,30 Consistent with this, 73% of confirmed/probable cases received ≥ 1 dispensations of an opioid, 30% a muscle relaxant, and 37% an antidepressant—all medications that are also potentially sedating.31 Some patients received repeated dispensations of multiple fatigue-causing medications. Fatigue and cognitive impairment are known side effects of medications that act on the central nervous system,32–34 but they are also among the core symptoms of ME/CFS.26 The treatment of common ME/CFS comorbid conditions (eg, chronic pain, depression, insomnia) may confound establishing a diagnosis of ME/CFS in many patients. Also, the interplay between ME/CFS symptoms, comorbid conditions and their treatments can make EHR-based use of exclusionary diagnosis codes (to improve ME/CFS case finding) problematic, though prior studies have used them.30,35,36
The present study and two comparably designed studies showed an increased use of the diagnosis code for ME/CFS over the last two decades, even if overall incidence of those with significant fatigue symptoms may have been stable. The authors found an annual rate of 21.5 ME/CFS initial diagnoses per 100,000 person-years among 9- to 39-year-olds in the Kaiser Permanente Northwest population during the study period, with an increase in ME/CFS initial diagnosis coding between 2006 and 2017. Given this study’s low case confirmation rates, the actual rates of IOM-defined ME/CFS are likely to be significantly lower than this. A study of ME/CFS coding within National Health System primary care practices in the UK found that ME/CFS diagnosis for adults of any age increased as a percentage of all fatigue diagnoses (excluding fibromyalgia) from 31.1% in 2001 to 59.9% in 2013.37 A study in South Korea found an ME/CFS diagnosis rate of 44.7 per 100,000 among all adults, with a 1.53-fold increase between 2010 and 2020.38 Consistent with existing literature on ME/CFS incidence, rates of ME/CFS code entry in this study’s study population were higher among females than males.22,25,30,37,38
This study has several limitations. First, limiting the age range of the present study’s population to 9- to 39-year-olds limits generalizability to individuals outside of this age range. A second limitation is that this was a single-site study in the United States, employing the IOM diagnostic criteria3 and so the findings may not be generalizable to other health care systems in the United States or to other countries with different standards for diagnosing ME/CFS patients. Third, the transition from ICD-9 to ICD-10 coding systems during this study period could have impacted the authors’ diagnosis coding rate estimates, given the change in the name of the diagnosis code from ICD-9, “chronic fatigue syndrome,” in 2006 to 2015 to ICD-10, “chronic fatigue, unspecified,” in 2016 and 2017. The latter, less-specific name may have resulted in increased diagnosis coding at the end of the study period, for example, as an indication to refer for naturopathic services. Fourth, this study employed chart review to validate the ICD coding for ME/CFS. As the medical record may not record all symptoms, the authors recognize that patient-reported questionnaires on symptom and functional capacity, as used in the UK study, may have identified more cases among those receiving ME/CFS ICD codes.
Conclusion
Only 15% of abstracted cases with an ME/CFS diagnosis code met all or most diagnostic criteria. Therefore, historical ICD coding patterns have significant limitations for estimating incidence of ME/CFS. The authors found an increase in ME/CFS code use, without confirming each coded diagnosis as a case, between 2006 and 2017, consistent with previous studies. Patients with ME/CFS with chart-reviewed confirmed diagnoses were predominantly female and had the central feature of disabling fatigue that had endured for years. Comorbid chronic pain, depression, and anxiety were common.
Acknowledgments
The authors acknowledge Stacy Harsh, RN, and Kim Olson for assistance with chart abstraction and Mara Kalter, MA, for project oversight and chart abstraction quality assurance. The authors also acknowledge Steven Olson, MD, at Kaiser Permanente Northern California, for serving as a subject matter expert during the early phases of the study, including developing the chart abstraction and adjudication forms. The authors also acknowledge Jeanne Bertolli, PhD, from the Centers for Disease Control and Prevention for review of an earlier version of the manuscript. No copyrighted materials were adapted or used for this article.
Disclaimer.
The findings and conclusions in this article are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.
Footnotes
Author Contributions: Author contributions were as follows: study design—Michelle L Henninger, PhD, Stephanie A Irving, MHS, Allison L Naleway, PhD, Bradley Crane, MS, Padma Koppolu, MPH, Julianne Gee, MPH, and Elizabeth R Unger, PhD, MD; data collection—Michelle L Henninger, PhD, Padma Koppolu, MPH, Elizabeth G Liles, MD, MCR, Bradley Crane, MS, and Stephanie A Irving, MHS; data analysis—Padma Koppolu, MPH, Stephanie A Irving, MHS, and Bradley Crane, MS; manuscript preparation—Elizabeth G Liles, MD, MCR, Michelle L Henninger, PhD, Stephanie A Irving, MHS, Neon B Brooks, PhD, Bradley Crane, MS, Allison L Naleway, PhD, Elizabeth R Unger, PhD, MD, Padma Koppolu, MPH, and Julianne Gee, MPH.
Conflict of Interest: None declared
Funding: This work was supported by a contract from the Centers for Disease Control and Prevention (200-2012-53584/0011; PI: Michelle L Henninger, PhD), “Describing Myalgic Encephalomyelitis/Chronic Fatigue Syndrome in the Vaccine Safety Datalink.” The funding source provided review and feedback on the study protocol and manuscript but was not involved in the conduct of the study.
Data-Sharing Statement: The data that support the findings of this study are not openly available due to reasons of sensitivity and are available from the corresponding author upon reasonable request.
References
- 1. Lim WT, Torpy DJ. Chronic fatigue syndrome. In: Feingold KR, Anawalt B, et al., eds. Endotext. South Dartmouth (MA). MDText.Com, Inc. ; 2000. Accessed https://www.ncbi.nlm.nih.gov/books/NBK279099/ [Google Scholar]
- 2. Bakken IJ, Tveito K, Gunnes N, et al. Two age peaks in the incidence of chronic fatigue syndrome/myalgic encephalomyelitis: A population-based registry study from Norway 2008-2012. BMC Med. 2014;12:167. 10.1186/s12916-014-0167-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Committee on the Diagnostic Criteria for Myalgic Encephalomyelitis/Chronic Fatigue Syndrome, Board on the Health of Select Populations, Institute of Medicine . Beyond Myalgic Encephalomyelitis/Chronic Fatigue Syndrome: Redefining an Illness. National Academies Press (US); 2015. [PubMed] [Google Scholar]
- 4. Fukuda K, Straus SE, Hickie I, Sharpe MC, Dobbins JG, Komaroff A. The chronic fatigue syndrome: A comprehensive approach to its definition and study. Ann Intern Med. 1994;121(12):953. 10.7326/0003-4819-121-12-199412150-00009 [DOI] [PubMed] [Google Scholar]
- 5. Carruthers BM, Jain AK, Meirleir KL, et al. Myalgic encephalomyelitis/chronic fatigue syndrome: Clinical working case definition, diagnostic and treatment protocols. J Chronic Fatigue Syndr. 2003;11(1):7–115. 10.1300/J092v11n01_02 [DOI] [Google Scholar]
- 6. Carruthers BM, van de Sande MI, De Meirleir KL, et al. Myalgic encephalomyelitis: International consensus criteria. J Intern Med. 2011;270(4):327–338. 10.1111/j.1365-2796.2011.02428.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Grach SL, Seltzer J, Chon TY, Ganesh R. Diagnosis and management of myalgic encephalomyelitis/chronic fatigue syndrome. Mayo Clin Proc. 2023;98(10):1544–1551. 10.1016/j.mayocp.2023.07.032 [DOI] [PubMed] [Google Scholar]
- 8. Chu L, Valencia IJ, Garvert DW, Montoya JG. Onset patterns and course of myalgic encephalomyelitis/chronic fatigue syndrome. Front Pediatr. 2019;7:12. 10.3389/fped.2019.00012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Johnston SC, Staines DR, Marshall-Gradisnik SM. Epidemiological characteristics of chronic fatigue syndrome/myalgic encephalomyelitis in Australian patients. Clin Epidemiol. 2016;8:97–107. 10.2147/CLEP.S96797 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Chandler RE, Juhlin K, Fransson J, Caster O, Edwards IR, Norén GN. Current safety concerns with human papillomavirus vaccine: A cluster analysis of reports in VigiBase® . Drug Saf. 2017;40(1):81–90. 10.1007/s40264-016-0456-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Tomljenovic L, Colafrancesco S, Perricone C, Shoenfeld Y. Postural orthostatic tachycardia with chronic fatigue after HPV vaccination as part of the “autoimmune/auto-inflammatory syndrome induced by adjuvants”: Case report and literature review. J Investig Med High Impact Case Rep. 2014;2(1):2324709614527812. 10.1177/2324709614527812 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Brinth L, Theibel AC, Pors K, Mehlsen J. Suspected side effects to the quadrivalent human papilloma vaccine. Dan Med J. 2015;62(4):A5064. [PubMed] [Google Scholar]
- 13. HPV Vaccine Safety and Effectiveness Data. 2021. Accessed 27 September 2020. https://www.cdc.gov/hpv/hcp/vaccine-safety-data.html
- 14. Manysheva K, Sherman M, Zhukova N, Kopishinskaia S. Myalgic encephalomyelitis/chronic fatigue syndrome: First described complication after Gam-COVID-Vac vaccine. Psychiatr Danub. 2022;34(Suppl 8):189–190. [PubMed] [Google Scholar]
- 15. In Rare Cases, Coronavirus Vaccines May Cause Long Covid–Like Symptoms. 2022. Accessed 3 February 2023. https://www.science.org/content/article/rare-cases-coronavirus-vaccines-may-cause-long-covid-symptoms
- 16. Donegan K, Beau-Lejdstrom R, King B, Seabroke S, Thomson A, Bryan P. Bivalent human papillomavirus vaccine and the risk of fatigue syndromes in girls in the UK. Vaccine. 2013;31(43):4961–4967. 10.1016/j.vaccine.2013.08.024 [DOI] [PubMed] [Google Scholar]
- 17. Feiring B, Laake I, Bakken IJ, et al. HPV vaccination and risk of chronic fatigue syndrome/myalgic encephalomyelitis: A nationwide register-based study from Norway. Vaccine. 2017;35(33):4203–4212. 10.1016/j.vaccine.2017.06.031 [DOI] [PubMed] [Google Scholar]
- 18. Schurink-van’t Klooster TM, Kemmeren JM, van der Maas NAT, et al. No evidence found for an increased risk of long-term fatigue following human papillomavirus vaccination of adolescent girls. Vaccine. 2018;36(45):6796–6802. 10.1016/j.vaccine.2018.09.019 [DOI] [PubMed] [Google Scholar]
- 19. Duclos P. Safety of immunisation and adverse events following vaccination against hepatitis B. Expert Opin Drug Saf. 2003;2(3):225–231. 10.1517/14740338.2.3.225 [DOI] [PubMed] [Google Scholar]
- 20. Magnus P, Brubakk O, Nyland H, et al. Vaccination as teenagers against meningococcal disease and the risk of the chronic fatigue syndrome. Vaccine. 2009;27(1):23–27. 10.1016/j.vaccine.2008.10.043 [DOI] [PubMed] [Google Scholar]
- 21. Magnus P, Gunnes N, Tveito K, et al. Chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME) is associated with pandemic influenza infection, but not with an adjuvanted pandemic influenza vaccine. Vaccine. 2015;33(46):6173–6177. 10.1016/j.vaccine.2015.10.018 [DOI] [PubMed] [Google Scholar]
- 22. Valdez AR, Hancock EE, Adebayo S, et al. Estimating prevalence, demographics, and costs of ME/CFS using large scale medical claims data and machine learning. Front Pediatr. 2018;6:412. 10.3389/fped.2018.00412 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42(2):377–381. 10.1016/j.jbi.2008.08.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Nacul LC, Lacerda EM, Pheby D, et al. Prevalence of myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) in three regions of England: A repeated cross-sectional study in primary care. BMC Med. 2011;9:91. 10.1186/1741-7015-9-91 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Reeves WC, Jones JF, Maloney E, et al. Prevalence of chronic fatigue syndrome in metropolitan, urban, and rural Georgia. Popul Health Metr. 2007;5:5. 10.1186/1478-7954-5-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Lim EJ, Son CG. Review of case definitions for myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS). J Transl Med. 2020;18(1):289. 10.1186/s12967-020-02455-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Gluckman S. Clinical Features and Diagnosis of Myalgic Encephalomyelitis/Chronic Fatigue Syndrome. 2022. Accessed 3 February 2023. https://www.uptodate.com/contents/clinical-features-and-diagnosis-of-myalgic-encephalomyelitis-chronic-fatigue-syndrome
- 28. Centers for Disease Control and Prevention . Diagnosis of ME/CFS. 2021. Accessed 3 February 2023. https://www.cdc.gov/me-cfs/symptoms-diagnosis/diagnosis.html
- 29. Rimes KA, Goodman R, Hotopf M, Wessely S, Meltzer H, Chalder T. Incidence, prognosis, and risk factors for fatigue and chronic fatigue syndrome in adolescents: A prospective community study. Pediatrics. 2007;119(3):e603–e609. 10.1542/peds.2006-2231 [DOI] [PubMed] [Google Scholar]
- 30. Vincent A, Brimmer DJ, Whipple MO, et al. Prevalence, incidence, and classification of chronic fatigue syndrome in Olmsted County, Minnesota, as estimated using the Rochester epidemiology project. Mayo Clin Proc. 2012;87(12):1145–1152. 10.1016/j.mayocp.2012.08.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Jones JF, Lin J-M, Maloney EM, et al. An evaluation of exclusionary medical/psychiatric conditions in the definition of chronic fatigue syndrome. BMC Med. 2009;7:57. 10.1186/1741-7015-7-57 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Shaiova L. The management of opioid-related sedation. Current Science Inc. 2005;9(4):239–242. 10.1007/s11916-005-0030-7 [DOI] [PubMed] [Google Scholar]
- 33. Stranks EK, Crowe SF. The acute cognitive effects of zopiclone, zolpidem, zaleplon, and eszopiclone: A systematic review and meta-analysis. J Clin Exp Neuropsychol. 2014;36(7):691–700. 10.1080/13803395.2014.928268 [DOI] [PubMed] [Google Scholar]
- 34. Lader M. Benzodiazepine harm: How can it be reduced? Br J Clin Pharmacol. 2014;77(2):295–301. 10.1111/j.1365-2125.2012.04418.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Boneva RS, Lin J-M, Maloney EM, Jones JF, Reeves WC. Use of medications by people with chronic fatigue syndrome and healthy persons: A population-based study of fatiguing illness in Georgia. Health Qual Life Outcomes. 2009;7:67. 10.1186/1477-7525-7-67 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Reyes M, Nisenbaum R, Hoaglin DC, et al. Prevalence and incidence of chronic fatigue syndrome in Wichita, Kansas. Arch Intern Med. 2003;163(13):1530–1536. 10.1001/archinte.163.13.1530 [DOI] [PubMed] [Google Scholar]
- 37. Collin SM, Bakken IJ, Nazareth I, Crawley E, White PD. Trends in the incidence of chronic fatigue syndrome and fibromyalgia in the UK, 2001–2013: A clinical practice research datalink study. J R Soc Med. 2017;110(6):231–244. 10.1177/0141076817702530 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Lim E-J, Lee J-S, Lee E-J, et al. Nationwide epidemiological characteristics of chronic fatigue syndrome in South Korea. J Transl Med. 2021;19(1):502. 10.1186/s12967-021-03170-0 [DOI] [PMC free article] [PubMed] [Google Scholar]