Measuring Health in Patients With Cervical and Lumbosacral Spinal Disorders: Is the 12-Item Short-Form Health Survey a Valid Alternative for the 36-Item Short-Form Health Survey?

Methods 

This cross-sectional study was performed by using patients recruited from an outpatient orthopedic clinic between February 2005 and August 2006. The study was approved by the institutional review board of West Virginia University, and all patients provided written informed consent.

Results 

Table 1 provides the descriptive summary of the sociodemographic and clinical data for both groups. The lumbosacral group tended to be older, have more comorbidities, have a higher proportion of smokers, and have more previous spine surgeries than the cervical group, but this was not statistically significant.

Table 1. Sociodemographic and Clinical Variables

	Variables	Cervical Group (n=24)	Lumbosacral Group (n=74)
	Age (y)	51±9(36–70)	56±17(18–87)
	Sex
	Female	13(54.2)	37(50.0)
	Male	11(45.8)	37(50.0)
	Race (cervical group, n=21; lumbar group, n=52)
	American Indian or Alaskan Native	1(4.8)	0(0.0)
	Black	1(4.8)	0(0.0)
	White	19(90.5)	52(100.0)
	Education (cervical group, n=21; lumbar group, n=52)
	Less than high school	3(14.3)	7(13.5)
	Graduated from high school	7(33.3)	23(44.2)
	Some college	3(14.3)	7(13.5)
	College degree and above	8(38.1)	15(28.8)
	Work
	Homemaker	3(12.5)	2(2.7)
	Working for wages	9(37.5)	20(27.0)
	Not working or sick leave	11(45.8)	23(31.1)
	Student	0(0.0)	2(2.7)
	Retired	1(4.2)	27(36.5)
	CMI (0–100)	9.2±7.4(0.0–23.8)	12.5±8.0(0.0–35.7)
	BMI (kg/m2)	29.9±5.3(19.2–39.4)	28.6±5.4(19.8–43.3)
	Smoking history
	No	13(54.2)	34(45.9)
	Yes	11(45.8)	40(54.1)
	Years smoked	26.5±10.1(14.0–50.0)	21.4±12.7(1.0–55.0)
	ICD-9 diagnosis
	Disk degeneration/displacement	8(33.3)	33(44.6)
	Spinal stenosis	3(12.5)	27(36.5)
	Spondylosis(cervical)/spondylolysis (lumbar)	12(50.0)	1(1.4)
	Spondylolisthesis	0(0.0)	6(8.1)
	Fracture (nonunion/closed)	1(4.2)	3(4.1)
	Other	0(0.0)	4(5.4)
	No. of previous spine surgeries
	None	15(62.5)	37(50.0)
	≥1	9(37.5)	37(50.0)

NOTE. Values are mean ± SD (range) or n (%). Abbreviation: ICD-9, International Statistical Classification of Diseases and Related Health Problems.

Table 2 displays the scale scores and summary scores of the SF-36v2 and SF-12v2 for the cervical and lumbosacral groups. There were strong correlations between the SF-36v2 and SF-12v2 in both groups across the 8 scale scores (cervical group: r range, .81–.99; lumbosacral group: r range, .88–.97), PCS (cervical group, r=.88; lumbosacral group, r=.95), and MCS (cervical and lumbosacral groups, r=.97). The general health score was the only exception, which had moderate but significant correlations between the 2 surveys (cervical group, r=.76; lumbosacral group, r=.69). It was also noted that, despite strong correlations, the correlation values of the corresponding physical function and vitality scores in both the cervical and lumbosacral groups were consistently lower than the other scale scores (r range, .81–.90).

Table 2. Correlations Between SF-36 and SF-12 in Cervical and Lumbosacral Groups

		Cervical Group (n=24)			Lumbosacral Group (n=74)
	Scores	SF-36	SF-12	r	SF-36	SF-12	r
	Scale
	Bodily pain	34.2±8.8(24.1–55.4)	31.1±12.4(16.7–57.4)	0.92⁎(0.82–0.97)	34.3±10.1(19.9–62.1)	31.0±12.2(16.7–57.4)	0.94⁎(0.91–0.96)
	General health	45.0±12.0(28.2–62.5)	44.7±12.0(29.7–62.0)	0.76⁎(0.52–0.89)	47.5±8.6(23.4–62.5)	47.5±9.6(18.9–62.0)	0.69⁎(0.55–0.79)
	Mental health	37.6±12.0(16.2–58.5)	39.6±11.9(15.8–58.5)	0.93⁎(0.84–0.97)	43.2±13.7(16.2–64.1)	44.4±12.8(21.9–64.5)	0.94⁎(0.91–0.96)
	Physical functioning	32.1±11.4(14.9–57.0)	31.4±9.4(22.1–56.5)	0.82⁎(0.62–0.92)	29.9±11.1(14.9–54.9)	30.8±10.0(22.1–56.5)	0.90⁎(0.84–0.94)
	Role−emotional	39.4±13.3(9.2–55.9)	41.2±12.5(11.4–56.1)	0.99⁎(0.98–1.00)	40.0±14.1(9.2–55.9)	42.1±13.5(11.4–56.1)	0.97⁎(0.95–0.98)
	Role−physical	27.7±9.5(17.7–54.4)	29.5±8.8(20.3–52.6)	0.96⁎(0.91–0.98)	29.9±9.8(17.7–56.9)	30.5±9.1(20.3–57.2)	0.96⁎(0.94–0.97)
	Social functioning	33.4±11.9(13.2–56.9)	33.8±10.8(16.2–56.6)	0.88⁎(0.74–0.95)	37.3±12.6(13.2–56.9)	38.0±12.7(16.2–56.6)	0.91⁎(0.86–0.94)
	Vitality	35.4±8.9(20.9–58.3)	36.4±9.1(27.6–57.8)	0.81⁎(0.60–0.91)	40.3±11.5(24.0–67.7)	41.6±11.2(27.6–67.9)	0.88⁎(0.82–0.92)
	Summary
	PCS	32.8±8.9(17.3–55.6)	31.3±8.8(20.1–53.8)	0.88⁎(0.74–0.95)	32.0±8.5(16.4–55.6)	31.0±8.5(15.4–54.8)	0.95⁎(0.92–0.97)
	MCS	39.9±13.3(11.0–62.0)	42.3±12.6(11.7–60.1)	0.97⁎(0.93–0.99)	45.3±13.7(16.7–71.7)	47.4±12.9(22.2–72.2)	0.97⁎(0.95–0.98)

NOTE. Values are mean ± SD (range).

⁎ P<.05.

SF-36v2 general health scores had only moderate correlations (r<.80) with the corresponding SF-12v2 scores in both groups. Therefore, stepwise linear regression analyses were performed to determine the best predictors for the SF-36v2 general health scores. The independent variables examined were SF-12v2 general health score, age, CMI, BMI, and smoking history. Table 3 presents the stepwise regression analysis results. Both groups had additional predictors besides the SF-12v2 general health scores that contributed significantly to the variance of the SF-36v2 general health scores. The additional predictors included CMI (negative direction effect) in both groups and age (positive direction effect) and smoking history (negative direction effect) in the lumbosacral group. The multivariate models explained a larger proportion of variance in the SF-36v2 general health scores (cervical group, r2=.66; lumbosacral group, r2=.61) than the SF-12v2 general health scores alone (cervical group, r2=.58; lumbosacral group, r2=.48).

Table 3. Stepwise Regression Analysis for SF-36v2 General Health Score

	Groups	Model Adjusted R2	Independent Variables	Standard Coefficient β (Cl)	Partial R2	P
	Cervical	.66	SF-12v2 general health	0.61(0.32to0.89)	.49	<.001
			CMI	−0.37(−1.06to−0.13)	.27	.01
	Lumbosacral	.61	SF-12v2 general health	0.68(0.47to0.76)	.51	<.001
			Age	0.32(0.08to0.25)	.19	<.001
			Smoking history(y)	−0.22(−0.23to−0.04)	.11	.01
			CMI	−0.18(−0.04to−0.01)	.06	.04

Abbreviation: CI, confidence interval.

Discussion 

The SF-12v2 is an efficient method of collecting health outcome data for patients with spinal disorders. The SF-12v2 provides clinicians with the assessment of the broad health summaries and the detailed individual health profiles that are not available in SF-12v1. However, there is a paucity of evidence supporting the validity of the SF-12v2 in patients with spinal disorders. This study validated the use of the SF-12v2 in patients undergoing elective spinal surgery and determined additional sociodemographic and clinical factors that would improve the prediction of SF-36v2 scores in scales with less than strong correlations with the SF-12v2.

The results of the study supported our first hypothesis that there would be strong correlations (r≥.80) between the SF-36v2 and SF-12v2 for the 8 scale and 2 summary scores, with the exception of the general health scores in both spinal groups. The finding that the 2 summary scores (PCS, MCS) of the SF-12v2 adequately reflect those of the SF-36v2 is consistent with previous validity studies on the SF-12v1.13, 14, 15, 16 The 8 scale scores, on the other hand, were not available in the previous version of the SF-12v1 because of the limited range of item responses. Thus, this study is the first to date to show that the 8 scale scores of the SF-12v2 are valid alternatives for the corresponding scores of the SF-36v2 in patients with spinal disorders.

The availability of the 8 scale scores in the SF-12v2 allows clinicians to have a detailed yet efficient and valid assessment of individual health profiles.11 Nonetheless, clinicians should take into account the trend we observed between the SF-12v2 and SF-36v2, which suggested that lower correlations between the corresponding scale scores were likely caused by a lower proportion of items from the SF-36v2 being included in the SF-12v2. Examples include the general health, physical function, and vitality scale scores, which consistently had the lowest correlations among the 8 health profiles in both spinal groups. These domains also happen to have the lowest proportion of representation of the SF-36v2 in the SF-12v2. The general health and physical function scales each had 20%, and the vitality scale had 25% of their respective domains in SF-36v2 represented in the SF-12v2, as compared with 40% to 67% of representation in other domains. Thus, clinicians should interpret the general health, physical function, and vitality scale scores within the context of their respective summary scores because they encompass broader health profiles.

Clinicians should interpret the SF-12v2 general health score differently than the other scale scores because the SF-12v2 general health score had only a moderate correlation and the lowest correlation with SF-36v2 across both spinal groups. The stepwise regression revealed that other significant demographic (age) and clinical factors (comorbidities, smoking history) could provide clinicians with a more accurate reflection of the SF-36v2 general health score above and beyond the SF-12v2 general health score alone.

The level of comorbidities is a common factor that should be considered in both spinal groups, with a stronger inverse relationship with the SF-36v2 general health score in the cervical group than the lumbosacral group. Nonetheless, clinicians should also consider age and smoking history in patients with lumbosacral disorders, which had positive and negative relationships, respectively, on the SF-36v2 general health score. The predictive significance of age and smoking history for the SF-36v2 general health score in only the lumbosacral group may have been attributed to the group being older and having a higher proportion of smokers than the cervical group.

Conclusions 

This study shows for the first time that the SF-12v2 is a valid alternative for the SF-36v2 in assessing the health status of patients with spinal disorders. The SF-12v2 adequately reflects both the health summary scores and the individual health profiles of SF-36v2. However, important factors including age, level of comorbidities, and smoking history need to be taken into consideration when assessing the SF-12v2 general health profile. These important factors are often part of the standard clinical intake data and thus do not impose additional time for collection. Taken together, the SF-12v2 is a comprehensive yet time-efficient health survey, with minimal loss of information and precision for the health status assessment of patients with spinal disorders.