Skip to main content

Table 2 Impact of propensity score matching for prospective cohort

From: The feasibility of matching on a propensity score for acupuncture in a prospective cohort study of patients with chronic pain

 

Referred for Acupuncture

(n = 173)

Not Referred for Acupuncture

(n = 350)

Standardized differencea

Propensity score characteristicsb

 Opioid therapy plan

20.2%

30.6%

−0.239

 Physical therapy past 30 days

1.2%

3.1%

−0.141

 Physical therapy past 31–180 days

11.6%

8.6%

0.010

 Physical therapy past 181–365 days

16.2%

8.3%

0.244

 Nonspecific chronic pain

30.1%

30.0%

0.013

 Substance abuse

5.2%

3.7%

0.072

 Sleep problem

26.0%

20.0%

0.143

 History of tobacco use

24.9%

27.1%

−0.052

 Anxiety

21.7%

18.5%

−0.080

 Pain treatment procedure

18.5%

22.0%

−0.087

 Pain diagnosis procedure

46.2%

51.4%

−0.104

 Pain medication

57.8%

56.0%

0.036

 Age (years)

49.6 (11.8)

52.8 (11.5)

−0.283

 Number of outpatient visits

10.3 (8.0)

10.8 (8.2)

−0.062

 Months since cohort entry

42.4 (27.7)

45.5 (26.9)

−0.114

 Ambulatory Charlson score

1.1 (1.4)

1.5 (2.0)

−0.264

Characteristics that did not contribute to the propensity score

Demographic Characteristics

 Female

71.1%

73.7%

−0.059

 White

88.5%

93.5%

−0.176

 Hispanic

3.8%

4.1%

−0.016

Medical and Psychiatric Comorbidities

 Depression

12.7%

16.6%

−0.109

Types of Nonmalignant Chronic Pain (NCP)

 Back and/or neck pain

71.1%

60.6%

0.223

 Joint pain (including osteoarthritis)

66.5%

70.9%

−0.095

 Fibromyalgia/other myofascial pain

30.6%

16.9%

0.327

 Headaches

20.2%

11.1%

0.252

 Neuropathy

3.5%

9.1%

−0.240

 Temporomandibular disorders

2.9%

2.6%

0.020

 Carpal tunnel syndrome

2.9%

5.1%

−0.116

 Abdominal pain

11.0%

9.4%

0.051

 Other NCP

5.2%

5.1%

0.003

 Two of above NCP types

74.0%

68.0%

0.132

Pharmacotherapy

 Any use of an opioid

20.2%

21.7%

−0.036

 Opioid morphine equivalent dose (MED)

0.3 (1.1)

0.3 (1.1)

−0.026

  ≥ 120 MED

7.5%

8.6%

−0.039

Mental health related

 Any antidepressant use

54.9%

49.4%

0.110

 Any anxiolytic use

28.3%

28.3%

0.001

 Any benzodiazepine use

28.9%

27.1%

0.039

  1. aStandardized difference expressed as (difference in means)/(pooled standard deviation) for continuous measures and as 2*(arcsin(√P1)-arcsin(√P2)) for binary data. For propensity score decile adjusted data, standardized differences calculated using same standard deviation as for unadjusted data in order to make comparison of standardized differences with and without adjustment more meaningful
  2. bContinuous data expressed as mean (standard deviation)