Interpreting Scores & Use in Clinical Practice

What is Norm-based Interpretation?

• Studies from general populations are used to derive normative data, including estimates of average & median scores along with variability indicators, such as standard deviations. These indicators can be used as "benchmarks" against which to compare observed scores & changes in scores.

• Scores can be examined in relation to estimates of central tendency to determine whether the score falls at, above, or below the average or median score. Scores can be interpreted in relation to the distribution of scores in a reference group, to determine the percentile at which a particular score occurs. A difference in scores or a score change can be also interpreted in terms of standard deviation units.

• Adjusted norms offer even more useful interpretation aides for some purposes, because they account for the socio-demographic characteristics of the respondent population (eg., age, gender, disease, and symptom status).

• Using norms, clinicians & others can readily interpret the burden of disease & the effect of treatment on health. The goal is to keep the patient's scores at or move the patient closer to the norm. The further below the norm a patient's score sits, the greater the burden of disease.

Applications

Comparing Populations

The value of using both a standard measure and norm-based interpretation was demonstrated during early studies of the Sickness Impact Profile (SIP). The SIP user's manual (Damiano, 1996), showed that these studies produced normative SIP data for many disease & condition groups. Selected results across populations for the physical dimension of the SIP (higher scores indicate poorer function).

Scoring in best health is the general population, followed by elderly outpatients & then patients who have specific diagnoses, such as end-stage renal disease (on hemodialysis) & chronic low back pain. In the poorest health are the very frail elderly. The SIP manual includes reference data for populations with many other diseases.

Monitoring Patients Over Time

Norms from health surveys administered to the same individuals over time can be very useful when interpreting health changes. For example, average change scores & other descriptive statistics for 1-year follow-up studies using the SF-36® have been published for patients with different chronic conditions (Ware, Kosinksi, and Keller, 1994). For example, a 70-year-old senior who declines six points in physical functioning (from 52 to 46 as measured by the PCS) has dropped from the 75th percentile of the score distribution for seniors down to the 50th percentile.

Examining the Burden of the Disease

Perhaps most important for clinicians, norms permit meaningful comparison of the burden of different diseases & conditions across patient groups. These comparisons can only be done when the conceptualization of health & the measures used are standardized across the different groups.

Also, before we can meaningfully attribute decline in functional health to a particular disease, we have to account for co-morbid conditions. For example, when Mrs. Smith walks through the door with congestive heart failure (CHF), that diagnosis isn't necessarily the only important thing about her.

Useful norms are adjusted for age, gender & co-morbid conditions. When we recognize that patients with any one chronic disease may have others, & that average age differs across disease groups, we can adjust norms to compare the impact on physical functioning of different chronic diseases.

For example, adjusted scores for adults with CHF, chronic lung disease & arthritis are all below the norm (50) on the SF-36® PCS scale. Patients with congestive failure report significantly poorer physical functioning than do those with allergies Their average score of 35 is more than a full standard deviation - 10 points - below the norm of 47 for those with allergies.

Comparing Physical Health Scores (SF-36® PCS) for Healthy and Chronically Ill Adults

Taken from Davies AR, ed. Interpreting Results: What Do the Numbers Mean? (Video program study guide; Series 1 - Health Status: Concepts, measures, and Applications.) Woodbridge, NJ, HealthStat Productions, Inc., 1998