The NACC Handbook: A Researcher's Guide

The Uniform Data Set (longitudinal follow-up)

The Uniform Data Set (UDS) is the primary data set used by researchers interested in clinical data. The NIA/NIH Alzheimer's Disease Research Centers (ADRCs) began submitting UDS data to NACC in September 2005, using the UDS Forms to collect standardized clinical data from participants who are evaluated on an approximately annual basis. Since 2005, the UDS forms have undergone three major revisions to reflect advances in the science and incorporate new diagnostic criteria. To combine data across versions of the UDS, we created a Researcher’s Data Dictionary (RDD). The RDD-UDS, should be the first and primary resource for researchers analyzing NACC clinical and demographic data.

As a resource for investigators in their analysis, NACC has provided a CSV file of all RDD-UDS variables and coding.

For more information, please see the section below titled "Advice on research design and best variables to use."

FTLD Module (frontotemporal lobar degeneration)

Beginning in February 2012, a subset of UDS subjects have also been evaluated using the supplemental FTLD Module. At Centers participating in this voluntary effort, subjects with suspected FTLD and/or controls are evaluated with the FTLD Module in addition to the standard UDS Forms.

LBD Module (Lewy body disease)

Beginning in August 2017, a subset of UDS participants have also been evaluated using the supplemental LBD Module. At Centers participating in this voluntary effort, participants with suspected LBD and/or controls are evaluated with the LBD Module in addition to the standard UDS Forms.

Minimum Data Set (MDS) (abstracted records) 

Before the UDS was implemented at Centers in 2005, data on Center participants were collected retrospectively via data abstraction and were included in the MDS. Because of the lack of detailed, longitudinal, and standardized clinical data in the MDS, the utility of the MDS for research is limited, and combining the clinical data in the MDS with the UDS is generally not recommended. For this reason, it is not currently part of the NACC Quick-Access file but is available upon special request. 

Neuropathology Data Set (autopsy data) 

The NP data set comprises subjects who have died and consented to autopsy. The NP data-collection form has undergone numerous revisions to reflect advances in the science and incorporate new diagnostic criteria. To combine data across versions, a Researcher’s Data Dictionary (RDD) was created. The RDD-NP should be the first and primary resource for researchers analyzing NACC neuropathology data. 

As a resource for investigators in their analysis, NACC has provided a CSV file of all RDD-NP variables and coding.

SCAN Imaging Data

The SCAN initiative is a collaboration between NACC, UC Berkeley, Mayo Clinic, University of Michigan, UC Davis, and the Laboratory of Neuro Imaging (LONI), funded by the National Institute on Aging. The goal of SCAN is to enable standardized PET and MRI data collection from across the Alzheimer’s Disease Research Centers (ADRC) Program. ADRCs acquire and upload SCAN-compliant images to a portal hosted by LONI at the University of Southern California where they are de-identified and defaced by the Aging and Dementia Imaging Research (ADIR) laboratory at Mayo Clinic. The PET and MRI laboratories at the University of Michigan and the ADIR Laboratory then process the images for quality assurance and harmonization. Following this, the PET laboratory at UC Berkely and the MRI laboratories at Mayo Clinic and UC Davis analyze the images to produce analysis results such as brain volumes, cortical thickness, surface area, and Standardized Uptake Value Ratio (SUVR) data. summary, QC, analysis (volumes and SUVRs), and access to defaced SCAN images.

The Researchers Data Dictionary – SCAN MRI (RDD - SCAN MRI) includes variables associated with DICOM files collected by SCAN and quality control (QC) variables, as well as volume, cortical thickness, and surface area calculated variables for a subset of MRIs. The Researchers Data Dictionary – SCAN PET (RDD - SCAN PET) includes variables associated with DICOM files collected by SCAN and quality control (QC) variables, as well as Standardized Uptake Value Ratio (SUVR) analysis data for a subset of PETs.  

Mixed Protocol (non-standard) MRI and PET Data 

NACC collects and shares MRI and/or PET data that does not adhere to the SCAN/CLARiTI protocols or was collected prior to January 2021. We have partnered with imaging experts including Dr. Charlie DeCarli at the University of California at Davis, Dr. Beth Mormino at Stanford, and Dr. Tim Hohman at Vanderbilt University to clean, label, and produce harmonized analysis results for mixed protocol MRI and PET data submitted to NACC. 

Imaging data collection and acquisition protocols vary by ADRC; thus, packages of mixed protocol files at NACC include several different scan types and naming conventions. 

A subset of mixed-protocol MRI files stored at NACC for UDS participants have standardized calculated volume values (e.g., hippocampal volume) and cortical thicknesses. These data are provided to NACC by the IDeA lab at the University of California, Davis. Investigators requesting these data should review the description of the calculation methods and protocols. 

The Researcher’s Data Dictionary — Imaging Data (RDD-ID) includes MRI calculated variables, as well as variables associated with the DICOM and NIfTI files stored at NACC. For specific methods used to perform these calculations, please see this guidance provided by the IDeA Lab. 

The Alzheimer’s Disease Sequencing Project Phenotype Harmonization Consortium (ADSP-PHC) Data 

The Alzheimer's Disease Sequencing Project Phenotype Harmonization Consortium (ADSP-PHC, U24-AG074855) was established to harmonize the rich endophenotypic data across cohort studies to enable modern genomic analyses of ADRD. Drs. Timothy Hohman, Mike Cuccaro, and Arthur Toga lead this consortium. Only a subset of NACC participants have harmonized cognitive, fluid biomarker, neuropathology, cardiovascular risk factor, and neuroimaging scores available in this dataset. (Note that NACC was not involved in this harmonization project.) 

CSF Biomarker Data (CSF Aβ, total Tau, p-Tau) 

For a small sample of UDS participants, NACC stores CSF biomarker values from a single lumbar puncture or longitudinal lumbar punctures. The Data Element Dictionary — CSF (DED-CSF) describes the variables related to CSF biomarker data. Please note that these data come from a small number of Centers. 

Standard Data File Download

NACC data are provided via secure links. You will receive a username, password, and link to download your data files. Your download link remains active for a limited period of time. If you find that your link has expired and you want us to reactivate it, please submit a support request to the research support team.

Image File Download

NACC will provide your images (both SCAN-compliant and mixed-protocol) on Amazon Web Service, S3. NACC will send you an email containing an access key ID and secret access key. 

Here are some options for downloading your MR image files:

• Software applications - some popular options used by researchers are Cyberduck and S3 Browser. Both offer free versions and can connect to the S3 Bucket, allowing you to download all the image files to your local computer.
• Command line - download your images using the Amazon Command Line Interface (CLI). If you haven't used this before, first you will need to install the CLI. Then, you can configure the CLI with the credentials NACC provided and download the bucket using the "aws s3 sync" command.

Before getting started with analysis, here are some things to keep in mind:

• If you need to merge two or more NACC data sets, be sure to merge them by NACCID.
• If you are focusing your analysis on neuropathology data, you will need to eliminate any longitudinal visits that you are not interested in.
  • For example, if you are interested in matching a participant's neuropathology data to the clinical data from the most recent visit before death, you will need to delete from your file any previous visits for that participant; for instance, if a participant has had five visits to date, you will need to restrict your file to the fifth visit for this participant, deleting visits 1 through 4.

Overview

If you would like to narrow down the participants in your file by certain criteria, please see the appropriate section below. For example, if you are interested in focusing on participants with CSF biomarker data, be sure to request from NACC the pertinent CSF variable data set when submitting your data request.

Please keep the following important notes in mind when using NACC variables to restrict your sample.

Restricting Based on Cognitive Status and Etiologic Diagnosis

On the UDS Clinician Diagnosis dorm, participants receive a diagnosis corresponding to cognitive status: normal cognition, impaired-not-MCI (mild cognitive impairment), MCI, or demented). Participants also receive an etiologic diagnosis — what the clinician suspected to be the cause (whether primary, contributing, or non-contributing) of any cognitive impairment. Both the variable on cognitive status (NACCUDSD) and one or more variables concerning etiologic diagnosis (for example, NACCALZD) will often be required to focus on a specific diagnostic group of interest.

For example, participants with an etiologic diagnosis of Alzheimer’s disease (NACCALZD=1) can have a cognitive status of impaired-not-MCI, MCI, or dementia. The only way to focus on those with AD dementia is to use both the cognitive status variable (NACCUDSD=4) and the etiologic diagnosis variable (NACCALZD=1).

Restricting by Diagnosis at a Certain Visit (e.g., MCI at the initial UDS visit)

To restrict to participants with a diagnosis at the initial visit, select those who meet your criteria and have NACCVNUM=1.

To include participants who have ever received the diagnosis of interest, you would look across all visits and determine whether the participant ever received that diagnosis at any UDS visit.

To restrict to subjects who have the diagnosis of interest at the most recent UDS visit, you would use the variables for visit number and/or visit date (NACCVNUM=NACCAVST, VISITMO, VISITYR).

Defining Cognitive Status Based on the MMSE, MoCA, or Clinical Dementia Rating (CDR®) Score

Although many researchers choose to define cognitive status according to the clinician diagnosis provided on UDS Form D1, others choose to use the global CDR® score, the MMSE, or the MoCA. The following are examples of the commonly used CDR® cutpoints: Normal cognition: CDR®=0; Mild cognitive impairment: CDR®=0.5; Demented: CDR®=1 (mild), 2 (moderate), or 3 (severe). For more information, see the section on Form B4 in the UDS Coding Guidebook. The MMSE score is sensitive to demographic and educational differences. Please consult the research literature to determine the best cut points for establishing cognitive status for your sample.

Clinicopathological Studies

If you are conducting a clinicopathological study, it may be advisable to restrict your sample to those who have clinical measures within one or two years of autopsy. The NACCINT variable, which indicates the months between the last UDS visit and death, can be used for this purpose.

Using the UDS Data Cross-Sectionally

To restrict to data from the initial visit, focus on visit data corresponding to NACCVNUM=1.

To restrict to data from the most recent visit, use the variables for visit number and/or visit date to determine the most recent visit (NACCVNUM=NACCAVST, VISITMO, VISITYR).

Restricting to Those with Non-Missing Data

The Researcher Data Dictionaries (RDD) includes a number of missing codes to indicate why data are missing. To focus your analyses on participants with non-missing data for a particular variable, be sure to exclude participants with the missing codes, such as –4, 9, 99, and 995–998. Also, be sure to exclude those missing value codes from your analysis; otherwise, they may skew your findings. Please refer to the section below titled "Missing data and data collection changes between UDS versions" for additional details about missing codes.

Restricting by Number of Visits Made

Use the variables NACCAVST (total number of UDS visits) and NACCNVST (total number of in-person visits, excluding telephone visits) to restrict to participants with a minimum number of UDS visits completed.

Restricting to Data from a Particular UDS Version

Use the FORMVER variable to determine the form version (e.g., FORMVER=1, which corresponds to version 1). For example, to focus on participants assessed with UDS version 3 forms, use FORMVER>=3.

NACC Data Structure

NACC provides Quick Access Files that include both longitudinal clinical data from the RDD-UDS (PDF) and neuropathology data from the RDD-NP (PDF).

UDS data is provided in long format, with one row of data per participant visit (so a participant with more than one visit completed will have more than one row of data). The illustration below shows how the data might look in your file, followed by a brief explanation.

• Visit number: The chronological order of the visits is indicated by the NACCVNUM variable. For example, NACCVNUM=1 is the Initial Visit.
• Number of visits completed: As you can see above, the first participant has completed 2 UDS visits, the second participant has completed 1 UDS visit, and the third participant has completed 9 UDS visits.
• Form version: You can see that the first participant has only UDS version 2 data (FORMVER=2), the second participant has only UDS version 1 data (FORMVER=1), and the third participant has both UDS version 1 and version 2 data collected over time.
• Neuropathology data is duplicated: The neuropathology data values are repeated/duplicated for each of a participant's UDS visits (see NACCNEUR and NACCBRAA variables). For example, the third participant has the same value for NACNEUR for all nine of the participant's visits. Make sure you analyze the data at the participant level and do not double count those participants who have more than one UDS visit.

Data Freeze

NACC data are frozen and archived approximately every three months. Your data set includes UDS data up through the data freeze indicated in the email sent by NACC. For example, if you are using data from the June 2025 data freeze, your file will include visit data collected and submitted to NACC from the beginning of the Uniform Data Set (September 2005) through the end of May 2025. If at any time you would like an updated version of your file, please submit an online request for an update.

Visit Number and Visit Date

If you are using the longitudinal UDS data in your analysis, you can identify the order of the visits using the NACCVNUM variable (NACCVNUM=1 is initial visit; NACCVNUM=2 is first follow-up completed, etc.). However, the variables for the visit date (VISITMO, VISITDAY, and VISITYR) and days since Initial Visit (NACCFDYS) are generally better to use than the NACCVNUM variable when doing the following:

• Selecting pertinent MRIs or other biomarker data
• Time to event analyses
• Longitudinal analyses such as linear mixed modeling
• Calculating variables such as time since cognitive symptom onset to most recent visit

Relating Imaging Data to UDS Visits

NACC does not associate MRI or amyloid PET scans with a particular UDS visit because often the scans and the UDS visit occur at different times, sometimes even years apart. We leave it up to investigators to match the scans to the UDS visit based on their study criteria. Investigators can determine which UDS visit is closest in time to a scan by comparing the UDS visit date variables (VISITMO, VISITDAY, VISITYR) and the MRI scan date variables (MRIMO, MRIYR, MRIDY in mixed-protocol MRIs; STUDYDATE or SCANDT in SCAN MRIs) or PET scan date variables (APETMO, APETDY, APETYR in mixed-protocol PETs, SCAN_DATE or SCANDATE in SCAN PETs).

Telephone Visits

Telephone visits are completed by the participant's co-participant when the participant is unable to attend an in-person visit. can Use the PACKET variable (PACKET=T) to identify visits completed over the telephone. If desired, use this variable to exclude telephone visits.

Clinical Diagnosis Groups

The manner of collecting the clinical diagnosis changed substantially from v1-v2 to v3 of the Clinician Diagnosis form; therefore, it is helpful to compare the Clinician Diagnosis forms for v2, v3, and v4.

Please also see the important note about "Cognitive status and etiologic diagnosis" under the section on "Narrow your data set based on your eligibility criteria."

Missing Data and Changes in Data Collection among UDS Versions

Please review the Researchers Data Dictionary for UDS data (RDD-UDS) (PDF) and Neuropathology data (RDD-NP) (PDF) to determine the missing/unknown codes for each variable. In most cases, the missing code is 8, 88, 888, 9, 99, or 999. Variables are coded as -4 if those particular variables were not collected for a given version of the UDS, or if a skip pattern on the form resulted in a missing value that could not be replaced by an implicit value based on the preceding question. To date, the UDS has been implemented in four versions. Use the FORMVER variable to determine the version of the forms used for a participant's visit.

Medication Data

Variables DRUG1 through DRUG40 correspond to each of the medications (up to 40) the participant reported, per visit. For example, if a participant reports taking atenolol and losartan, then DRUG1=atenolol and DRUG2=losartan.

Data Values that Can Vary Over Time (e.g., age of onset of cognitive decline)

Some variables are collected at each UDS visit, and because different clinicians perform the assessments over time, or because the participant's symptoms have changed over time, the values for these longitudinally collected variables change over time. In particular, if you are using the Clinician Judgment of Symptoms form variable for age of onset of cognitive decline (DECAGE) or first predominant symptoms, be sure to ascertain whether values for these variables have changed over time. In some cases, it may be best to use the most recent non-missing value, because it may represent the best data available. In other cases, it may be better to use the value at the initial UDS visit, depending on your study design.

UDS Participant Health History Form Data

The data collected using the Participant Health History form is usually focused on health conditions reported by the participant or co-participant; therefore, be careful when using the associated variables in your analysis. If the health condition reported on the Health History form is also collected on other UDS forms, we advise you to use data from the other form, especially if the other form is based on clinician judgment (e.g., Clinician Diagnosis form). For example, the Participant Health History form collects data on Parkinson’s disease; however, in most cases, it is advisable to use the data on Parkinson’s disease from the Clinician Diagnosis form instead, because it is diagnosed by the clinician, not reported by the participant. Alternatively, you may want to consider examining all relevant data collected on the health condition to determine the best variable(s) to use.

Mini Mental State Exam (MMSE, UDSv1-2) or MoCA (UDSv3-4) versus Clinician Dementia Rating (CDR®)

The MMSE/MoCA and CDR® capture different information and have different measurement properties. The choice of the instrument depends on your analysis goals. The MMSE (replaced by the MoCA in UDS version 3) is good for screening and staging moderate and severe dementia, whereas the CDR® can measure (non-subtle) progression of cognitive and functional decline.

Incidence and Prevalence Rates

NACC data are not suitable for an analysis of dementia incidence or prevalence at a city, state, or national level. This is because the sample is not population-based. Recruitment protocols differ by Center. Depending on the ADRC, participants may or may not have been randomly selected. Therefore, the NACC data are best viewed as a case series. Please exercise caution when developing research aims surrounding the NACC data and when interpreting the results.

Death, Dropout, and Discontinuation

ADRCs periodically notify NACC, via the Milestones Form, about UDS participants who have died, have dropped out of the study, or have been discontinued by the Center for other reasons. The frequency with which Centers provide these data varies by Center; therefore, researchers should be cautious in the interpretation of any analysis in which data on death, dropout, or discontinuation are used.

Differences in the Neuropathology Database among UDS and MDS Participants

All UDS participants who died and consented to autopsy have data collected using NACC’s Neuropathology (NP) Form (version 8). In the ADRCs’ earlier data set, the Minimum Data Set (MDS), some autopsied participants have data based on earlier versions of the NP Form, if their autopsies were conducted after the NP Form was implemented in 2002. The remaining autopsied MDS participants have limited NP data that were collected within the MDS Form. If you will be obtaining autopsy data for MDS participants, a NACC consultant will help guide you through the data availability.

Difference Between Primary Neuropathologic Diagnosis and Neuropathologic Features

As of version 10, the NP form includes the most current AD and FTLD neuropathological criteria. The NP form now focuses on assessing neuropathological features rather than the primary neuropathological diagnosis, as had been the case in previous versions of the NP Form. Upon request, researchers can obtain the data on primary neuropathological diagnosis as was collected in NP Form versions 1 through 9.

Inconsistencies among the CDR®, Neuropsychological Testing, and the Clinical Diagnosis

A small number of UDS participants have seeming inconsistencies among their scores on the CDR®, their neuropsychological test results, and/or their clinical diagnoses. For example, a participant may have a clinical diagnosis of dementia on the Clinician Diagnosis form but a CDR® of 0 (no impairment). These inconsistencies are generally verified as correct by Centers and may occur because different clinicians complete different parts of the UDS assessment.

ZIP Codes

To protect the confidentiality of the data, NACC generally does not provide ZIP code data to researchers.

Generalizability

Because UDS recruitment methods vary by ADRC and over time, UDS participants are best described as a clinical case series of patients from each ADRC.

Non-Imaging Biospecimens and Genetic Data

See NACC Partnerships for information on requesting biospecimens from NCRAD or genetic data from ADGC or NIAGADS.

Biospecimens Stored by ADRCs

Brain tissue is stored by ADRCs on a subset of their UDS participants. Some Centers may be willing to share specimens with outside investigators. Researchers can use the NACC biospecimen locator or submit a tissue location request when submitting a Data Request.

• If you mention any previous contact with NACC in your communication with the Center, please include a statement like the following: "Please note that this is neither a NACC request nor a NACC initiative, and your participation is entirely voluntary."

Other ADRC Data

Centers may store additional data of interest, such as scores on neuropsychological tests that were performed as part of the UDS visit but not submitted to NACC because they are not part of the standardized UDS protocol. NACC is not usually aware of additional data stored at the Centers. Generally, the best way to identify participants with the desired data desired is to determine which Centers may have data you require, based on published research found through a PubMed search, or a search of the individual Centers’ websites, and then to contact the Center(s) directly. 

• If you mention any previous contact with NACC in your communication with the Center, please include a statement like the following: "Please note that this is neither a NACC request nor a NACC initiative, and your participation is entirely voluntary."

Genetic Data (APOE genotype, availability of genetic data) 

APOE genotype data are available for a large subset of UDS participants. The Researcher’s Data Dictionary — Genetic Data (RDD-Gen) describes variables relating to APOE genotype data as well as variables indicating the availability of genetic data at the Alzheimer’s Disease Genetics Consortium (ADGC) and the NIA Genetics of Alzheimer’s Disease Data Storage Site (NIAGADS). 

As a resource for investigators in their analysis, NACC has provided a CSV file of all RDD-Gen variables and coding.