Author

Data Source

Diagnosis/Treatment Codes and Algorithms

Study Cohort

Validation Methodology

Comments

Katz et al. (1997)

Country: USA

Source: Medicare physicians claims

Years: March 1, 1993 to October 31, 1993

Codes: ICD-9-CM

RA: 714.0, 714.1, 714.2, 714.30, 714.31, 714.32, 714.33

OA: 715.0 or 715.09, 715.1, 715.15, 715.16, 715.2, 715.25, 715.26, 715.3, 715.35, 715.36, 715.8, 715.85, 715.86, 715.89, 715.9

Fibromyalgia: 729.0, 729.1, 729.2, 729.4, 729.5

Current Procedural Terminology (CPT) codes:

Joint and soft tissue injection and aspiration: 20550, 20600, 20605, 20610

Age of cohort was not specified.

Patient office records (using a standardized data collection form).

Medicare Part B physician claims:

Max Sens = 90% (RA); 90% (OA)

Max PPV = 95%(RA); 88% (OA in the knee)

“For OA, when specific 5-digit ICD-9-CM codes indicting the anatomic location of OA (e.g., hip or knee) were required, the sensitivities of the physician claims were ≤0.50. However, when either the 5-digit site-specific codes or the less specific 3- and 4-digit codes (indicating OA without specification of anatomic site) were accepted, the proportion of medical record diagnoses identified in claims improved to at least 0.90 for the hip and knee.” p.1597

Harrold et al. (2000)

Country: USA

Source: Administrative data from one health maintenance organization (HMO)

Years: 1994-1996

Codes: ICD-9

OA 715.00-715.99

Algorithm: continuously enrolled in the health care plan and at least one health care encounter with an OA diagnosis.

18 years and older

OA related clinical, laboratory, and radiology data from medical records reviewed by two trained nurse reviewers and four clinicians.

Max PPV = 83% (for persons with a consultation with a rheumatologist and/or orthopedic surgeon).

“The positive predictive value of an administrative OA diagnosis was 62%...When our sample was limited to those members who had a consultation with a rheumatologist and/or orthopedic surgeon, the positive predictive value increased.” p.1884

Powell et el. (2003)

Country: USA

Source: Kaiser Permanente Georgia Region (KPGR) clinical information (including primary care encounters, hospital and emergency department use, pharmacy data and lab tests). Georgia Medicaid (GAMC) data, including outpatient and hospital claims

Years: 1995-1999

Codes: ICD-9-CM (the complete list included 616 diagnostic codes; these are not provided in the article)

CPT codes (not specified)

Algorithm: Arthritis had to be listed as one of the first 3 diagnoses

KPGR- all ages

GAMC patients less than 64 year were excluded

The algorithms were not validated.

“Arthritis was defined as any diagnosis on the National Arthritis Data Workgroup (NADW) list of diagnoses and ICD-9-CM codes.” p.292

“A single year’s worth of data estimates the prevalence of persons with arthritis seeking medical care, but underestimates the prevalence of arthritis within the population by twofold to threefold.” p.298

Losina et al. (2003)

Country: USA

Source: Medicare claims, including inpatient and surgical data.

Years: 1995

Codes: ICD-9-CM

RA: 714

Age of cohort not specified.

Medical records and patient survey data

Max Sens = 65% for medical records

Max PPV = 98% for medical records.

Max κ = .73

Ruof et al. (2004)

Country: Germany

Source: Administrative claims data, including prescription drug data

Years: July 2000 to June 2001

Codes: ICD-10

RA: M05 and M06

Algorithms: Group A - patients enrolled in a randomized clinical trial about the introduction of a disease management plan for. Group B - patients for whom administrative claims data reported the diagnosis of RA, and who had at least one prescription of a disease-modifying anti-rheumatic drug (DMARD) in the year under observation. Group C - patients for whom administrative data had a RA diagnosis, but who had no DMARD prescription in the one-year observation period

All ages

For group A, clinical data were available for validation, while for groups B and C no clinical data were available for validation.

Specificity, sensitivity and predictive values were not reported.

One of the major outcomes was, “administrative claims data in conjunction with treatment based algorithms (i.e., DMARD therapy yes/no) provide an estimate of disease related medical care costs in RA which is closely related to that of a clinically defined cohort of RA patients.” p.67

Singh et al. (2004)

Country: USA

Source: Minneapolis VA administrative database, including pharmacy and laboratory data.

Years: Jan. 2001 to July 2002

Codes: ICD-9 RA: 714

Algorithms: For patients seen at the VA rheumatology clinic. Evaluated five algorithms: 1) ICD code 714; 2) ICD code 714 plus ≥ 3 month prescription of a DMARD; 3) ICD code 714 plus a positive RF titer; 4) ≥3 month prescription of a DMARD plus positive RF; and 5) ICD code 714, a DMARD prescription, and a positive RF titer.

Age range of cohort was not specified, but mean age was 64.4 years.

Chart review. A chart diagnosis of RA by a rheumatologist on 2 separate occasions > 6 weeks apart was the gold standard.

Max Sens = 88.2 %

Max Spec = 97.1%

Max PPV = 97.0%

Rector et al. (2004)

Country: USA

Source: Medicare and HMO data: Physician, facility (i.e., hospital), and pharmacy claims

Years: 1999 to 2000

Codes: ICD-9-CM codes 714.xx, 715.xx, 716.xx, 720.xx, 721.1-721.4x, 721.9x in physician claims in one of up to four diagnosis fields and hospital claims in one of up to nine diagnosis fields. CPT codes and National Drug Codes (not listed)

Algorithms: 38 validated

Age of cohort was not specified.

Survey data was the validation source.

Max Sens = 70%

Max Spec = 99%

“The algorithm that required at least two face-to-face claims with a first-listed diagnosis and at least one prescription for a medication commonly used to treat the condition produced the highest specificity” p.1849