Participation in “The Airways Program” (TAP) and
associated mortality reduction
Pradeep Paul
1Health
1*
George ,
1
Heng ,
2
Ling ,
Bee Hoon
Mavis Yeow Bee
2,3
2,3
4
Yeow , Fong Seng Lim , Lim Tow Keang
Loo See
Services & Outcomes Research (HSOR), National Healthcare Group, Singapore, 2Disease management, National Healthcare Group,
3National Healthcare Group Polyclinics, Singapore, Singapore, 4National University Hospital, Singapore, Singapore
BACKGROUND
Table:3 Risk of death, hazard ratio and 95% CI
Chronic Obstructive Pulmonary Disease (COPD) is a major cause of chronic morbidity
and mortality worldwide and in Singapore. In Singapore, there are about 60,000 cases
of COPD. They constitute about 1/5 of all deaths and 7th principal cause of death and
7th most common condition for hospitalization in Singapore (2010). The Airway
Programme (TAP) was started in 2008 in three hospitals – Tan Tock Seng Hospital,
National University Hospital and Alexandra Hospital. The programme aimed to
optimise resources to enable better management of patients afflicted with COPD and
Community Acquired Pneumonia (CAP). The objectives of TAP were
• To reduce hospitalisations for acute exacerbations of COPD and prevent premature
hospitalisations for CAP
• To reduce the average length of stay (ALOS) for patients with COPD; and improve
quality of life of patients with COPD and/or afflicted with CAP and
• Improve quality of life of patients with COPD and/or afflicted with CAP
RESEARCH OBJECTIVE
To assess the impact of “The air way program” (COPD) on patients length of stay,
readmission and mortality.
METHODS
TAP patients were compared with controls, who were COPD patients (DRG 177) and
fulfilled the programme’s inclusion criteria but were not enrolled in the programme.
Control patients were identified from Operations Data Store and Central Clinical
Research Database. Patients who had hospitalisations after enrolment or refusal till
December 2009 were analysed. Outcomes of interest were hospital length of stay, readmission and mortality. T-test and chi-square test were used to compare continuous
and categorical variables respectively.
Risk of death and risk of readmission was estimated using cox and competing risk
regression respectively. Propensity score was estimated to identify the predictors of
program enrolment. Variables used for propensity score computation were age,
gender, race, hospital, ward class, comorbidities: asthma, diabetes, hypertension,
stroke, congestive heart failure, dyslipidemia and obesity. TAP patients and controls
were matched on their propensity score using nearest neighbor without replacement
methodology.
RESULTS
There were 170 matched TAP patients and controls patients, they had 287 and 207
hospitalizations respectively. TAP patients and controls were similar with regards to
age and gender and average hospital length of stay (P<0.05). No of days admitted to
hospital per 100 person days were higher for TAP patients (Table 1).
Variables
Age
Females
NUH
TTSH
Ward class B
Ward class C
Readmission
Asthma
Diabetes Mellitus
Hypertension
Stroke
Coronary Heart Disease
Heart Failure
Dyslipidemia
Obesity
TAP patients
Hazard 95% Confidence Interval (CI)
P value
Ratio
Lower
Upper
1.04
1.00
1.08
0.05
0.50
0.17
1.46
0.21
0.83
0.31
2.22
0.71
0.83
0.35
1.99
0.68
0.62
0.13
2.89
0.54
1.15
0.26
5.05
0.86
1.99
1.09
3.61
0.03
0.37
0.18
0.78
0.01
2.27
1.20
4.30
0.01
1.00
0.50
2.01
1.00
0.98
0.38
2.49
0.96
1.06
0.49
2.29
0.88
1.00
0.44
2.27
1.00
1.18
0.57
2.43
0.66
0.22
0.08
0.65
0.01
0.38
0.21
0.69
0.00
Table:4 Risk of readmission, sub hazard ratio and 95% CI
Variables
Age
Females
Malay
Indian
NUH
TTSH
Ward class B
Ward class C
Asthma
Diabetes Mellitus
Hypertension
Stroke
Coronary Heart Disease
Heart Failure
Dyslipidemia
Obesity
TAP patients
SHR*
1.01
1.19
0.57
1.49
0.51
1.16
0.86
1.12
1.55
0.44
0.98
0.91
0.83
0.87
1.09
1.37
3.40
95% Confidence Interval (CI)
P value
Lower
Upper
0.97
1.04
0.65
0.57
2.51
0.65
0.17
1.84
0.35
0.60
3.73
0.39
0.17
1.58
0.25
0.51
2.63
0.73
0.28
2.68
0.80
0.38
3.30
0.84
0.81
2.96
0.19
0.19
1.02
0.05
0.53
1.79
0.94
0.30
2.73
0.87
0.36
1.92
0.67
0.32
2.37
0.79
0.51
2.31
0.83
0.69
2.72
0.36
1.81
6.39
0.00
* SHR: Sub Hazard ratio
Table 1 Comparison of characteristics, TAP patients vs. controls
Variables
Episodes
ALOS ± SD (days)
Total hospital days
Avg. hospital days / patient
Total person-years of follow up
Total hospital days per 100 person-days¶
Total no of deaths, n (%)
Deaths due to respiratory system diseases
All cause mortality rate per person-year.
Mortality due to respiratory system
diseases per person-year
¶
TAP patients
(n = 170)
Controls
(n = 170)
287
4.2 ± 4.7
1207
7.1
152.02
2.2
18 (11)
15 (9)
0.12
207
4.3 ± 4.9
891
5.2
129.49
1.9
35 (21)
26 (15)
0.27
0.0012
0.0105
0.0664
0.0036
0.09
0.20
0.0282
P value
0.822
Person time (days) – Program: 55527 and Controls: 47295
Risk of death was lower for program patients (Table 3). After controlling for
competing risk death, the risk of readmission was 3.4 times higher for TAP patients
when compared to controls (Table 4). No of hospital days per 100 person- days was
significantly higher for TAP patients when compared to controls (Table 5).
Table:5 No of days admitted to hospital per 100 person-days
Hospital PersondaysΦ
AH
NUH
TTSH
Total
7167
15389
32556
55112
TAP patients
Controls
Hospital
Hospital
Total
Total
days per 100 Persondays per 100 P value
hospital
Φ hospital
persondays
personΦ
Φ
days
days
┼
days
days┼
0.00
282
3.93
6078
81
1.33
0.76
257
1.67
12881
209
1.62
0.54
667
2.05
28336
601
2.12
1206
2.19
47295
891
1.88
0.0007
┼
# of days admitted to hospital per100 person days, Φ One patient from Singhealth not included in the total, hence
the total person time and total hospital days would not add up to 55527 person-days and 1207 (total hospital days)
CONCLUSIONS
Table : 2 Readmission rate among TAP patients and controls
Status
TAP patients
Control patients
n
170
170
Patient with
Person-time readmissions Readmission
(years)
rate
(n)
140.99
126.12
51
22
0.36
0.17
95% CI
0.27 0.48
0.11 0.26
All cause mortality and mortality due to respiratory system diseases were lower
(Table 1) and readmission was higher for program patients (Table 2). 30-day
readmission was significantly higher for program patients (0.14/person-year, 95%
CI:0.09 – 0.22) when compared to controls (0.03/person-year, 95% CI: 0.01 – 0.08).
Participation in “TAP” was associated with lower all-cause mortality when compared
to the controls. The survival gain in the TAP patients was associated with an increase
in total hospital days. Risk of readmission was higher for TAP patients after adjusting
for mortality (competing risk). Evaluation of such program by means of administrative
databases may yield meaningful results. Further studies are required to ensure
corrections for in case mix and time bias.
Limitations
The evaluation focuses only on patients with hospitalization and hence not
representative of all COPD patients. The analysis has not been adjusted for disease
severity, but only adjusted for demographics and comorbidities. Readmission that
were analyzed were only from AH, NUH and TTSH, admissions/ readmissions that
happened outside of these are not captured. Lastly, the evaluation assessed only
association, not causality.
* For further information please contact Pradeep_Paul_G_Gunapal@nhg.com.sg