Recording of hospitalizations for acute exacerbations of COPD in UK electronic health care records

Figures & data

Figure 1 Summary of the methods for each of the aims of the study.

Abbreviations: AECOPD, acute exacerbations of chronic obstructive pulmonary disease; CPRD, Clinical Practice Research Datalink; HES, Hospital Episodes Statistics; PPV, positive predictive value.

Table 1 Strategies for identifying admissions to hospital for AECOPD in HES

HES definition of AECOPD hospitalization
1. Specific AECOPD code or COPD code in any position in any FCE during spell
2. Specific AECOPD code in any position or COPD code in first position in any FCE during spell
3. Specific AECOPD code in any position in any FCE during spell
4. Specific AECOPD code in any position in or LRTI code or COPD code in first position in any FCE during spell
5. Specific AECOPD code or LRTI code in any position or COPD code in first position in any FCE during spell
6. Specific AECOPD code in first position in first FCE during spell

Abbreviations: COPD, chronic obstructive pulmonary disease; AECOPD, acute exacerbations of COPD; LRTI, lower respiratory tract infection; HES, Hospital Episodes Statistics; FCE, finished consultant episodes.

Table 2 Possible strategies for identifying hospitalizations for AECOPD using primary care data alone

Definition	Example
Diagnostic code or codes suggesting hospitalization for AECOPD	“Admit COPD emergency”
Diagnostic code(s) suggesting AECOPD (using our previously validated algorithm) and nonspecific code(s) suggesting admission to hospital on the same day	“Acute LRTI” and “admission to hospital” on the same day

Abbreviations: COPD, chronic obstructive pulmonary disease; AECOPD, acute exacerbations of COPD; LRTI, lower respiratory tract infection.

Figure 2 Flow of patients through the study.

Abbreviations: COPD, chronic obstructive pulmonary disease; CPRD, Clinical Practice Research Datalink; HES, Hospital Episodes Statistics.

Table 3 Characteristics of patients included in the study

Characteristic	Overall	Those with hospital discharge information
	N=27,182 (N [%])	N=40 (N [%])
Age group (years)
≤55	5,003 (18.4)	7 (17.5)
55–64	7,746 (28.5)	16 (40.0)
65–74	8,537 (31.4)	12 (30.0)
≥75	5,896 (21.7)	5 (12.5)
Sex
Male	14,556 (53.6)	18 (45.0)
MRC breathlessness scale (kg/m^2; N=21,151)
<3	9,645 (45.6)	21 (53.8)
≥3	11,506 (54.4)	18 (46.2)
BMI (N=26,447)
<19	1,441 (5.5)	1 (2.5)
19–25	9,568 (36.2)	18 (45.0)
≥25	15,438 (58.4)	21 (52.5)
GOLD 2006 grade (N=14,055)
1	2,829 (20.1)	4 (16.7)
2	6,116 (43.5)	6 (25.0)
3	4,075 (29.0)	10 (41.7)
4	1,035 (7.4)	4 (16.7)
Smoking status
Ex-smoker	10,963 (40.3)	19 (47.5)
Current smoker	16,219 (59.7)	21 (52.5)
Index of multiple deprivation quintile (N=25,852)
1 (least deprived)	3,632 (14.1)	8 (20.0)
2	5,259 (20.3)	7 (17.5)
3	4,989 (19.3)	7 (17.5)
4	5,794 (22.4)	6 (15.0)
5 (most deprived)	6,178 (23.9)	12 (30.0)

Abbreviations: MRC, Medical Research Council; BMI, body mass index; GOLD, Global Initiative for Chronic Obstructive Lung Disease.

Figure 3 Diagnostic positions of ICD codes for AECOPD, LRTIs, and COPD in Hospital Episodes Statistics records for hospitalizations for COPD patients.

Abbreviations: ICD, International Classification of Diseases; COPD, chronic obstructive pulmonary disease; AECOPD, acute exacerbations of COPD; LRTI, lower respiratory tract infection.

Table 4 Validity of HES definitions of AECOPD hospitalization

	Discharge summary analysis		Full HES sample analysis
HES definition of AECOPD hospitalization	Number of discharge summary-confirmed AECOPD hospitalizations identified using strategy (N=40 events in 40 patients from discharge letters)	Sensitivity (95% CI) (% of discharge summary-confirmed AECOPD hospitalizations picked up) (N=40 events in 40 patients from discharge letters)	Number of potential AECOPD hospitalization events in total sample identified using strategy (full HES sample for all 27,182 COPD patients included in the study)a
Specific AECOPD code or LRTI code in any position or COPD code in the first position in any FCE during spell	35/40	87.5% (72.4%–94.9%)	40,174
Specific AECOPD code or COPD code in any position in any FCE during spell	34/40	85.0% (69.6%–93.3%)	74,590
Specific AECOPD code in any position or LRTI code or COPD code in the first position in any FCE during spell	34/40	85.0% (69.6%–93.3%)	37,966
Specific AECOPD code in any position or COPD code in the first position in any FCE during spell	31/40	77.5% (61.3%–88.2%)	35,793
Specific AECOPD code in any position in any FCE during spell	31/40	77.5% (61.3%–88.2%)	33,933
Specific AECOPD code in the first position in first FCE during spell	26/40	65.0% (48.5%–78.6%)	21,387

Note:

a These potential events will represent both true and false positives.

Abbreviations: CI, confidence interval; COPD, chronic obstructive pulmonary disease; AECOPD, acute exacerbations of COPD; LRTI, lower respiratory tract infection; HES, Hospital Episodes Statistics; FCE, finished consultant episodes.

Table 5 PPV and sensitivity of CPRD strategies to identify hospitalizations for AECOPD using different HES definitions as reference standard allowing 30 days after HES record of hospitalization for AECOPD

HES AECOPD definition	CPRD strategy	PPV (95% CI)	Sensitivity (95% CI)
AECOPD hospitalization or LRTI code in any position or COPD in the first position in any FCE	AECOPD hospitalization code	50.2% (48.5%–51.8%)	4.1% (3.9%–4.3%)
	AECOPD identified using validated algorithm and hospitalization code	43.3% (42.3%–44.2%)	5.4% (5.1%–5.7%)
Either specific AECOPD code in any position or COPD code in the first position	AECOPD hospitalization code	49.0% (47.3%–50.6%)	4.6% (4.5%–4.9%)
	AECOPD identified using validated algorithm and hospitalization code	38.5% (37.6%–39.4%)	5.5% (5.2%–5.9%)
Either specific AECOPD code in the first position in any FCE	AECOPD hospitalization code	45.9% (44.2%–47.6%)	4.7% (4.4%–4.9%)
	AECOPD identified using validated algorithm and hospitalization code	37.2% (36.3%–38.1%)	5.7% (5.4%–6.0%)

Abbreviations: CI, confidence interval; COPD, chronic obstructive pulmonary disease; AECOPD, acute exacerbations of COPD; LRTI, lower respiratory tract infection; HES, Hospital Episodes Statistics; FCE, finished consultant episodes; CPRD, Clinical Practice Research Datalink; PPV, positive predictive value.

Table S1 PPV and sensitivity of CPRD strategies to identify hospitalizations for AECOPD using different HES definitions as reference standard using day of admission in HES only

HES AECOPD definition	CPRD strategy	PPV (95% CI)	Sensitivity (95% CI)
AECOPD hospitalization or LRTI code in any position or COPD code in the first position in any FCE	AECOPD identified using algorithm	0.7% (0.7–0.7)	7.2% (6.95–7.6)
	Nonspecific hospitalization code	10.3% (10.1–10.6)	27.1% (26.5–27.7)
Either specific AECOPD code in any position or COPD code in the first position	AECOPD identified using algorithm	0.6% (0.6–0.6)	7.2% (6.9–7.6)
	Nonspecific hospitalization code	9.1% (8.9–9.3)	27.6% (27.0–28.2)
Either specific AECOPD code in the first position in any FCE	AECOPD identified using algorithm	0.6% (0.6–0.6)	7.4% (7.1–7.9)
	Nonspecific hospitalization code	8.7% (8.5–8.9)	28.1% (27.4–28.7)

Abbreviations: CI, confidence interval; COPD, chronic obstructive pulmonary disease; AECOPD, acute exacerbations of COPD; LRTI, lower respiratory tract infection; HES, Hospital Episodes Statistics; FCE, finished consultant episodes; CPRD, Clinical Practice Research Datalink; PPV, positive predictive value.

Table S2 PPV and sensitivity of CPRD strategies to identify hospitalizations for AECOPD using different HES definitions as reference standard using day of admission in HES only

HES AECOPD definition	CPRD strategy	PPV (95% CI)	Sensitivity (95% CI)
AECOPD hospitalization code or LRTI code in any position or COPD in the first position in any FCE	AECOPD hospitalization code	47.6% (44.3–50.8)	1.9% (1.7–2.0)
	AECOPD identified using validated algorithm and hospitalization code	41.9% (39.8–44.0)	3.7% (3.5–4.0)
Either specific AECOPD code in any position or COPD code in the first position	AECOPD hospitalization code	43.6% (40.5–46.9)	2.1% (1.9–2.3)
	AECOPD identified using validated algorithm and hospitalization code	36.8% (34.7–38.9)	3.7% (3.5–4.0)
Either specific AECOPD code in the first position in any FCE	AECOPD hospitalization code	46.5% (44.9–48.2)	2.1% (1.9–2.3)
	AECOPD identified using validated algorithm and hospitalization code	35.3% (33.2–37.4)	3.8% (3.5–4.1)

Abbreviations: CI, confidence interval; COPD, chronic obstructive pulmonary disease; AECOPD, acute exacerbations of COPD; LRTI, lower respiratory tract infection; HES, Hospital Episodes Statistics; FCE, finished consultant episodes; CPRD, Clinical Practice Research Datalink; PPV, positive predictive value.

Table S3 PPV and sensitivity of record of AECOPD or nonspecific hospitalization code to identify hospitalizations for AECOPD using different HES definitions as reference standard allowing 30 days after HES record of hospitalization for AECOPD

HES AECOPD definition	CPRD strategy	PPV (95% CI)	Sensitivity (95% CI)
AECOPD hospitalization code or LRTI code in any position or COPD code in first position in any FCE	AECOPD identified using algorithm	1.8% (1.7–1.8)	34.2% (33.7–34.6)
	Nonspecific hospitalization code	14.5% (14.3–14.6)	53.5% (53.0–54.0)
Either specific AECOPD code in any position or COPD code in the first position	AECOPD identified using algorithm	1.5% (1.5–1.6)	34.6% (34.1–35.1)
	Nonspecific hospitalization code	12.6% (12.5–12.8)	54.1% (53.6–54.6)
Either specific AECOPD code in the first position in any FCE	AECOPD identified using algorithm	1.5% (1.4–1.5)	35.1% (34.6–35.6)
	Nonspecific hospitalization code	12.0% (11.9–12.2)	54.8% (54.3–55.3)

Abbreviations: CI, confidence interval; COPD, chronic obstructive pulmonary disease; AECOPD, acute exacerbations of COPD; LRTI, lower respiratory tract infection; HES, Hospital Episodes Statistics; FCE, finished consultant episodes; CPRD, Clinical Practice Research Datalink; PPV, positive predictive value.

Table S4 PPV and sensitivity of CPRD strategies to identify hospitalizations for AECOPD using different HES definitions as reference standard allowing 60 days after HES record of hospitalization for AECOPD

HES AECOPD definition	CPRD strategy	PPV (95% CI)	Sensitivity (95% CI)
AECOPD hospitalization code or LRTI code in any position or COPD in the first position in any FCE	AECOPD hospitalization code	50.7% (49.1–52.3)	4.4% (4.3–4.6)
	AECOPD identified using validated algorithm and hospitalization code	46.1% (44.3–47.8)	6.1% (5.8–6.4)
Either specific AECOPD code in any position or COPD code in the first position	AECOPD hospitalization code	49.5% (48.0–51.1)	5.0% (4.8–5.3)
	AECOPD identified using validated algorithm and hospitalization code	41.2% (39.5–42.9)	6.2% (5.9–6.6)
Either specific AECOPD code in the first position in any FCE	AECOPD hospitalization code	46.6% (45.0–48.2)	5.0% (4.8–5.3)
	AECOPD identified using validated algorithm and hospitalization code	39.5% (37.8–41.2)	6.4% (6.1–6.8)

Abbreviations: CI, confidence interval; COPD, chronic obstructive pulmonary disease; AECOPD, acute exacerbations of COPD; LRTI, lower respiratory tract infection; HES, Hospital Episodes Statistics; FCE, finished consultant episodes; CPRD, Clinical Practice Research Datalink; PPV, positive predictive value.

Table S5 PPV and sensitivity of record of AECOPD or nonspecific hospitalization code to identify hospitalizations for AECOPD using different HES definitions as reference standard allowing 60 days after HES record of hospitalization for AECOPD

HES AECOPD definition	CPRD strategy	PPV (95% CI)	Sensitivity (95% CI)
AECOPD hospitalization code or LRTI code in any position or COPD in the first position in any FCE	AECOPD identified using algorithm	2.3% (2.3–2.3)	46.3% (45.8–46.7)
	Nonspecific hospitalization code	14.5% (14.4–14.7)	55.9% (55.4–56.4)
Either specific AECOPD code in any position or COPD code in the first position	AECOPD identified using algorithm	2.0% (2.0–2.1)	46.8% (46.3–47.3)
	Nonspecific hospitalization code	12.7% (12.5–12.9)	56.5% (56.0–57.0)
Either specific AECOPD code in the first position in any FCE	AECOPD identified using algorithm	1.9% (1.9–1.9)	47.3% (46.8–47.8)
	Nonspecific hospitalization code	12.1% (11.9–12.2)	57.2% (56.6–57.7)

Abbreviations: CI, confidence interval; COPD, chronic obstructive pulmonary disease; AECOPD, acute exacerbations of COPD; LRTI, lower respiratory tract infection; HES, Hospital Episodes Statistics; FCE, finished consultant episodes; CPRD, Clinical Practice Research Datalink; PPV, positive predictive value.