Table 2 Subgroup analysis of all included studies.

From: Accuracy of portable spirometers in the diagnosis of chronic obstructive pulmonary disease A meta-analysis

Factor

Studies

Sensitivity (95% CI)

I2

Model used

Specificity (95% CI)

I2

Model used

SROC (95% CI)

All studies

31

0.85 (0.81–0.88)

77.87%

Random

0.85 (0.81–0.88)

95.48%

Random

0.91 (0.89–0.94)

Type of the device

 COPD-6

14

0.84 (0.80–0.88)

79.22%

Random

0.85 (0.79–0.90)

95.85%

Random

0.91 (0.88–0.93)

 Piko-6

6

0.89 (0.76–0.96)

84.17%

Random

0.88 (0.75–0.94)

96.11%

Random

0.95 (0.92–0.96)

 PEF

4

0.77 (0.73–0.81)

1.13%

Fixed

0.83 (0.71–0.91)

94.04%

Random

0.82 (0.78–0.85)

 Hi-checker

2

0.75 (0.63–0.84)

91.20%

Random

0.86 (0.82–0.90)

92.80%

Random

–

 IQ-spiro

1

0.94

–

–

0.92

–

–

–

 SP10BT

1

0.83

–

–

0.75

–

–

–

 Medikro SpiroStar

1

0.92

–

–

0.84

–

–

–

 MS01 Micro spirometer

1

0.89

–

–

0.58

–

–

–

 Spirobank Smart

1

0.83

–

–

0.92

–

–

–

Detection indicators

 PEF

4

0.77 (0.73–0.81)

1.13%

Fixed

0.83 (0.71–0.91)

94.04%

Random

0.82 (0.78–0.85)

 FEV1/FVC

4

0.85 (0.81–0.88)

6.47%

Fixed

0.80 (0.65–0.90)

96.14%

Random

0.87 (0.84–0.90)

 FEV1/FEV6

23

0.85 (0.81–0.89)

82.09%

Random

0.86 (0.81–0.90)

95.54%

Random

0.92 (0.90–0.94)

Threshold selection method

 Fixed

4

0.83 (0.65–0.92)

90.39%

Random

0.89 (0.79–0.94)

94.40%

Random

0.93 (0.90–0.95)

 Cutoff

19

0.86 (0.81–0.89)

80.62%

Random

0.85 (0.80–0.90)

95.43%

Random

0.92 (0.89–0.94)

 P

 

0.673

  

0.385

  

0.581

Type of indicators

 Multi-index

27

0.85 (0.82–0.88)

79.21%

Random

0.85 (0.81–0.89)

95.60%

Random

0.92 (0.89–0.94)

 Single index

4

0.77 (0.73–0.81)

1.13%

Fixed

0.83 (0.71–0.91)

94.04%

Random

0.82 (0.78–0.85)

 P

 

<0.001

  

0.716

  

<0.001

Country

 Developed

23

0.85 (0.81–0.88)

73.60%

Random

0.83 (0.78–0.87)

95.52%

Random

0.90 (0.88–0.93)

 Developing

8

0.84 (0.77–0.89)

88.47%

Random

0.90 (0.85–0.93)

96.57%

Random

0.94 (0.91–0.95)

 P

 

0.778

  

0.023

  

0.015

Executive place

 Tertiary hospital

10

0.89 (0.83–0.93)

81.27%

Random

0.92 (0.89–0.95)

82.86%

Random

0.96 (0.94–0.97)

 Primary care/community

21

0.83 (0.79–0.87)

76.98%

Random

0.80 (0.75–0.85)

95.97%

Random

0.89 (0.86–0.91)

 P

 

0.066

  

<0.001

  

<0.001

Study setting

 Hospital-based

27

0.85 (0.81–0.88)

77.64%

Random

0.85 (0.80–0.89)

95.81%

Random

0.91 (0.89–0.93)

 Population-based

4

0.85 (0.78–0.90)

86.53%

Random

0.87 (0.84–0.90)

94.79%

Random

0.93 (0.90–0.95)

 P

 

0.999

  

0.469

  

0.222

Population

 No COPD

13

0.86 (0.78–0.91)

80.71%

Random

0.87 (0.81–0.92)

94.67%

Random

0.93 (0.90–0.95)

 The whole crowd

18

0.84 (0.81–0.87)

76.22%

Random

0.83 (0.78–0.87)

95.81%

Random

0.90 (0.87–0.93)

 P

 

0.584

  

0.271

  

0.133

Publication year

 2000–2016

15

0.83 (0.79–0.86)

66.10%

Random

0.83 (0.77–0.87)

95.94%

Random

0.89 (0.86–0.92)

 2017–2021

16

0.86 (0.80–0.90)

83.90%

Random

0.87 (0.81–0.91)

95.02%

Random

0.92 (0.89–0.95)

 P

 

0.336

  

0.267

  

0.178

  1. The study divides countries into developed and developing countries according to HDI. When HDI > 0.80, it is classified as a developed country and if otherwise, developing country59.
  2. SROC summary receiver operating characteristic curve.
  3. aOwing to the complexity of the bivariate model and the limited number of studies, the groups with n ≥ 4 were pooled using a bivariate model. The remaining data were pooled by univariate random-effects logistic regression model. We also tested for the difference in the sensitivity or specificity between the two groups using the bivariate model. The AUC difference in the area under the SROC between the groups was obtained by Z test.
Back to article page