SF-36v2 Health Survey: eight-domain health profile

What this assessment is

The SF-36v2 Health Survey is a 36-item adult self-report measure of general health status and health-related quality of life.

It produces eight health-domain scores. Those domains can also be combined into a Physical Component Summary and a Mental Component Summary.

There is no standard 36-item total.

No total exists.

The survey also includes a separate rating of perceived change in health. That rating does not enter the eight domain scores.

This page covers version 2. SF-36v2 differs from the original SF-36, RAND-36, and shorter forms in the same family.

Keep versions separate.

Use v2.

The profile describes reported health. It cannot diagnose a condition.

Who it was designed for

The SF-36 family grew from the Medical Outcomes Study. John Ware and colleagues developed the survey to compare health across diseases, treatments, and populations.

Version 2 was introduced in 1996 with revised response choices and presentation. The current IQVIA instrument record identifies the standard version in the present survey family.

SF-36v2 is used in clinical trials, health services research, population surveys, and outcome monitoring.

It is intended for adults. The person reports their own health, although a matching interview mode can be used.

The measure is generic. It can compare broad health across conditions, but it may not capture the symptoms most important in a specific disease.

General is not complete.

How it is administered

The respondent answers 36 questions about health, function, and the effect of health on daily life.

The standard form uses a four-week recall period. The acute form uses the previous week.

Use one recall form throughout a study or episode of care.

Recall matters.

Completion usually takes several minutes. Reading ability, fatigue, illness, and interview administration can change the time required.

Use the complete validated version and preserve its response coding. Do not combine items from SF-36v1, SF-36v2, RAND-36, or SF-12v2.

Names matter.

When comparing scores over time, keep the version, recall period, language, mode, scoring algorithm, and norm set constant.

Consistency matters.

What it measures

SF-36v2 reports eight health domains:

• physical functioning
• role limitations due to physical health
• bodily pain
• general health
• vitality
• social functioning
• role limitations due to emotional problems
• mental health

The profile shows how different parts of health relate to daily life.

Domains can diverge.

PCS combines information from all eight domains, with greater emphasis on the physical domains.

MCS also combines all eight, with greater emphasis on mental and emotional domains.

The separate health-change rating asks for a comparison with an earlier point in time. It is not part of the eight-domain profile or either component summary.

The longer profile gives more detail than SF-12v2, but it remains a general health measure rather than a detailed symptom inventory.

Use the full profile.

How scoring works

Responses are recoded so that higher values represent better reported health.

Items assigned to each domain are then combined under the SF-36v2 scoring rules.

The eight domains can be reported on the original 0-to-100 metric. On that scale, higher values indicate better health in the domain.

They can also be reported with norm-based scoring. Under QualityMetric's norm guidance, these T-scores use 50 as the named reference mean and 10 points as one reference standard deviation.

QualityMetric's guidance notes that norm-based scores do not run from 0 to 100. Their achievable ranges differ by domain and recall form.

Label the metric.

PCS and MCS use weighted combinations of the eight domain scores. They are not averages of the domains.

Weights differ.

Do not add all 36 responses into a total. Do not mix 0-to-100 domain scores with norm-based values without clear labels.

Use licensed scoring.

Follow the matching scoring system for missing responses. Do not create a local replacement rule.

How to interpret SF-36v2 scores

Begin with the eight-domain profile.

A higher domain score indicates better reported health in that area. A lower score indicates more limitation, symptoms, or poorer perceived health in that area.

For norm-based scores, 50 is a reference mean. It is not a pass mark or clinical cutoff.

PCS and MCS provide broad summaries. They should not replace the domain profile when the pattern matters.

There are no universal mild, moderate, or severe bands for SF-36v2. A score's meaning depends on the norm set, population, condition, and purpose.

Scores need context.

The United States norms study illustrates how population estimates can differ from an earlier reference norm.

When tracking change, consider measurement uncertainty and whether the person's circumstances or administration method changed.

Read the pattern, not one number.

Context still matters.

What the scores cannot tell you

SF-36v2 cannot diagnose a physical or mental health condition.

The mental health domain and MCS are not depression or anxiety diagnoses. The physical domains and PCS do not establish physical capacity, disability status, or fitness for work.

The scores cannot identify the cause of a problem. Pain, illness, mood, support, environment, expectations, and treatment can all affect responses.

A high broad score does not rule out a serious symptom outside the survey's focus. A low score does not identify which treatment is needed.

The profile also cannot represent every personal priority. Two people with similar scores may care about different activities and outcomes.

Ask what matters.

Use the results with clinical information and a direct conversation about what matters.

Evidence and limitations

SF-36v2 is widely used and supports comparisons across many health conditions.

Its eight-domain profile is more informative than one general health total. The common scoring system can also support comparisons when the same version and norms are used.

The measure has limits. Thirty-six questions still provide only a broad view. Condition-specific symptoms, cognition, sleep, relationships, and participation may need separate assessment.

Norm-based scoring improves reference comparisons but adds complexity. Results can appear different when analysts change the norm set or score metric.

Self-report captures the person's experience, which may differ from clinical tests or another observer's view.

Use the survey to identify the health domains that need attention. Then ask what is driving the score and what change would be meaningful.

Profiles guide questions.