Inpatient Data

Emergency Dept

Observation Stays

Physician Demographics

Resource Ctr & Library

Data Locator

Srvcs by Members

Request Data Now

Data Request Form

Request Form Tips

Data Use Agreement (PDF)

A Primer of Helpful Hints on the Types of Questions That Users Should Ask When Requesting Data

Accessing large data sets can be expensive and time consuming! By structuring your data request, with assistance from your health data organization, you will avoid potential problems and maximize the usefulness of each data retrieval.

I. How many years of data will be included?

A. By Calendar Year (CY): January 1 through December 31

B. By Fiscal Year (FY)

1. Municipal: July 1 through June 30
   

2. Non-municipal: October 1 through September 30

C. Otherwise, requestor should define study period, if it is different

Hint: A request for two calendar years requires data pulls from three fiscal year data bases

Hint: For multi-year studies; Verify changes in hospital universe, coding methods used for each data variable requested, and other additions/deletions which may have occurred between data years

II. Which hospitals will be included in this request?

A. Are data available for each hospital you are requesting?

(e.g. Chronic, Long Term care, Rehabilitative and other non-acute hospitals may not be required to report data to any public data sources.)

B. Has each hospital reported data for the study period requested?

(i.e. Are there missing hospital data for particular months or quarters)

C. Have there been any changes in services &/or operations for each requested hospital during the study period? (An addition of a new Intensive Care Unit (ICU) or hospital merger, for example)

D. Will hospitals be identified by name will they be aggregated into comparison groups? (e.g. Teaching status, number of beds, or location)

E. Do all hospitals use the same coding methodology to report the types of illnesses and treatments seen in their facility? (i.e. ICD-9 codes are the standard for diagnoses and inpatient procedures)

III. Are there specific socio-demographic characteristics needed for this request?

A. Sex

B. Age/Birth Date

1. In years

2. In weeks, if patient is less than 1 year old

   Hint: The coding of many hospital procedures may not be based on age or gender considerations important to the study (for example, breast biopsies are also done on males; Tonsil and Adenoid procedures for older adults may be handled differently than in children)
   

   Hint: Birth date may be considered confidential and therefore unavailable for specific research, Calculated age or month/year (without day) may be available as substitute information.
   

C. Zip Code

1. Zip code ranges may be aggregated into town, county, and other specified geographic areas.

   Hint: Many requestors may omit zip codes for a particular study area and may exclude residents by unintentionally including invalid zip codes and/or excluding new valid zip codes

D. Race

1. Coding at each hospital is usually based on individual response, appearance, or other subjective criteria and may be considered "soft data"

   Hint Race categories may need to be aggregated into larger groups (e.g. black, white, other)

E. Expected Type of Payment Data (Insurance Carrier)

1. The Expected payment source recorded on the face sheet of the medical record may not have been the actual source of payment

2. Because of the method in which these data were collected there may be no way to determine if the charge for the hospital stay and/or medical services provided were covered by the principal expected source of payment; the actual payment amount may vary

F. Expected Source of Payment Data (Health Plan)

1. Inpatient data reporting guidelines for Massachusetts acute care hospitals, amended by the state in January 1994, now require that in addition to reporting the expected type of payment data, hospitals must also report the "expected source of payment" data

2. Expected source of payment data uniquely identify 159 different health plans (e.g. Aetna, HMO Blue, Tufts Health Plan, Kaiser, Medicare Plus, etc.). Each expected source of payment code has also been assigned to a corresponding expected type of payment code, which identify each of the insurance carriers, including Blue Cross and Blue Cross Managed Care, Commercial Insurance, Commercial Insurance Managed Care, HMO plus ten additional managed and non-managed care payor type categories

   Hint: For each "expected source of payment" code reported, consider analyzing whether the corresponding data by reported "expected type of payment" code matches the same payor type assignments defined by the state, to verify the accuracy and reliability of the various payor data reported

   Hint: Consider aggregating carriers into broader categories (e.g. aggregating all Blue Cross non-managed carriers and all Blue Cross managed care carriers into Blue Cross)

IV. Other Clinical Data Set Elements

A. Length of Hospital Stay (L.O.S.)

1. Based on the sum of gross patient days during one hospitalization. However, some hospitals do include "Leave of Absence Days" as part of its length of stay calculations

   a. Leave of Absence Days are the number of days of a patient's absence with physician approval during a hospital stay without formal discharge and readmission to the facility

   Hint: Since not all hospitals code Leave of Absence Days, calculate length of stay using Net Days, as gross Days minus Leave of Absence Days

   Hint: Researchers should decide whether they want to exclude records with lengths of stay of 0 days or lengths of stay greater than 999 days, &/or assign these cases to a length of stay based on predetermined criteria

   Hint: For length of stay "outliers," i.e. patients with extreme length of stay, consider using "trimmed data;" Patients with lengths of stay longer than other patients with similar clinical characteristics and treatments may be excluded from average length of stay calculations because the inclusion of these cases will tend to inappropriately inflate these averages. These "outlier cases" can be the result of such factors as coding errors, patients awaiting nursing home placement, physician practices or an unusual set of circumstances. "Trimming" is a technique which reduces the variability in the data (Consult with technical staff for "trimming" methodologies)

B. Discharge Disposition Status

1. Patient status at the time of discharge from the hospital (i.e. patient sent home, patient sent to another hospital, long term or rehabilitation facility for additional (post-acute) services, patient expired during hospitalization)

C. Type of Admission: Indicates the priority status of the admission

D. Hospital Inpatient Admission Source: The source referring or transferring the patient to inpatient status in the hospital (For Emergency, Elective or Other Type of Admission)

Caveat: Definitions and assignment of these variables may be unique to each facility reporting information

Caveat: Information regarding these variables may be unavailable at the time the patient is discharged and therefore may be understated

Hint: Broad category definitions and data validation using other data sources should be considered as these data variables may be considered to be "soft data" by some hospitals

E. Admission Date/Discharge Date

1. These data may be considered confidential and therefore unavailable for specific data research

2. Day of the week, and month/year may be available as substitute information

F. Diagnoses

1. Principal Diagnosis Code: The ICD-9-CM (5 digit) diagnosis code corresponding to the condition established to be chiefly responsible for the admission of the patient for hospital care.

2. Associated Diagnosis Codes (up to 14): Conditions that co-exist with the principal diagnosis at the time of admission, or develop subsequently, which affect the treatment received or the length of the patient's hospital stay

   Hint: Be sure to define your diagnostic criteria fully; For example, newborns may be classified using ICD-9 diagnosis codes V37-V37.2 & V39-V39.2, however, other newborn diagnoses include weight (coded in grams) and/or by using a fifth digit, in addition to the four digit diagnosis code. Newborns can also be defined using DRG

G. Procedures

1. Principal Procedure Code: The ICD-9-CM (4 digit) procedure code corresponding to the procedure most related to the principal diagnosis and performed for definitive treatment of the principal diagnosis rather than for diagnostic or exploratory purposes, or was necessary to treat a complication of the principal diagnosis

2. Associated Procedure Code (up to 14): Corresponds to the additional procedures performed during the length of the patient's hospital stay.

H. Diagnosis Related Groups (DRGs)

1. A software program established as the basis for the Medicare hospital reimbursement system which aggregates all diagnoses and procedures coded during a patient's hospital stay. (HCFA usually revises the DRG categories annually, for each data year there are at least 495 inpatient DRGs.)

2. The DRGs are a patient classification scheme which provides a means of relating the type of patients a hospital treats to the costs incurred by the hospital

   Hint: Since HCFA could change DRGs annually, multi-year studies should include a matrix of DRG revisions, including both old and new DRGs

   Hint: Consider other, more meaningful, aggregations based on your study criteria. (e.g. Aggregating by hospital department; Dividing Cardiology into Cardiac Medicine and Cardiac Surgery)

V. Charges

A. Total Charges: Thew full undiscounted charges summarized by prescribed revenue center codes. The total charges should not include charges for telephone services, television or private duty nurses. Any charges for a leave of absence period are to be included in the routine accommodation charges for the appropriate services (medical/surgical, psychiatry) from which the patient took the leave of absence. Any routine admission charges or daily charges under which expenses are allocated to the routine, special care & ancillary revenue centers are included in the total charges

Caveat: Charges are not costs nor payments

Hint: Calculate median as well as average charges when using charge data

Hint: Researchers should consider whether they want to exclude records with charges of $0 dollars, or records with missing or erroneous financial data (e.g. Per Diem charges less than $100) &/or assign those cases to a charge, based on predetermined criteria

Hint: Consider using criteria to "Trim" &/or exclude cases with extremely high charges; Patients high charges (outlier cases) may be the result of such factors as coding errors, physician practices or an unusual set of circumstances. Trimming charges is a technique based on predetermined criteria which may reduce the variability when using these data. (Consult with technical staff for "trimming" methodologies)

VI. Unique Physician Number (UPN)

A. Attending Physician Number

1. The unique number assigned to the physician primarily responsible for the care of the patient from the beginning of the hospital episode. The patient's personal physician who arranged for the patient's admission and directed the patient's care is normally designated the attending physician.

2. If the patient does not have a person physician, the hospital staff physician to whom the patient is assigned and who is responsible for the patient's care is the attending physician.

B. Operating Physician Number

1. The unique number assigned to the physician responsible for performing the principal procedure on the patient.

   Caveat: UPN data are based on the actual license number of each attending &/or operating physician reported in the data base. However, UPN data are encrypted by the state prior to public release, so that the identity of an individual patient may not be directly or indirectly identified.

   Caveat: Since UPN data are encrypted, these data cannot be used to determine the name of the actual attending &/or operating physician.

For more information on MHDC Data, please contact us via e-mail or at (781) 419-7800. We welcome your further questions & look forward to your participation in our work and our events!