Cohort characteristics. A total of 17,105 people met our study inclusion criteria (Additional Fig. A2); 62.6% were male and 15.8% were current smokers. Median age at IPF diagnosis was 76.7 years (IQR: 69.6–82.7). Nearly a quarter (23.1%) had a GP-recorded diagnosis of GORD prior to their IPF diagnosis, while 1 in 6 or 16% had a prior diagnosis of COPD. A similar proportion (17%) had been previously diagnosed with asthma. Over a quarter (27.2%) also had IHD, pulmonary hypertension, diabetes also ranked as highly prevalent comorbidities in our IPF study population (relative to the general adult population) (Table 1).
Prescribing patterns
Of the 17,105 people with IPF, 5,045 (29.5%) were prescribed a PPI at least once in the year prior to their IPF diagnosis, of whom 1,857 (36.8%) also had a diagnosis of GORD (Table 1). Of the people also diagnosed with hiatus hernia (n = 2,445), 48.5% were regularly or irregularly prescribed PPI in the baseline period. Among those not prescribed a PPI prior to their IPF diagnosis, 1,274 went on to receive a prescription for PPI in the year after their diagnosis, but were classed as “non-users”, and contributed data to the control population (Table 2). Regular and irregular ICS users comprised 16% (n = 2,741) and 8.8% (n = 1,507) of the study cohort, respectively (Table 1). While in absolute terms more men than women were prescribed ICS prior to their IPF diagnosis, the proportion of women was elevated in both the regular and irregular user groups relative to the non-users (42.2% and 40.9% vs 36.0%). Unsurprisingly, patients with a prior diagnosis of COPD or asthma were overrepresented in the ICS user groups, especially the regular users. Only a very low proportion of the cohort (n = 188; 1.1%) were regularly prescribed both ICS and PPIs in the baseline period, and therefore we did not mutually adjust for ICS and PPI use in our subsequent PPI and ICS statistical analyses.
Among the prior non-ICS users (n = 12,857), 1,002 (or around 6% of the whole cohort) were prescribed an ICS-containing inhaler at least once after their IPF diagnosis (Fig. 1). Nearly a half of this subgroup (n = 482; 48.1%) became regular ICS users after their IPF diagnosis. Whereas patients who also had COPD and/or asthma dominated the pre-, post- and pre-and-post IPF ICS user groups (Fig. 1a, 1b and 1c, respectively), among those who started on ICS only after an IPF diagnosis, only a third also had either previous COPD or asthma diagnoses (Fig. 1d). Among the subset of 1,002 new users, 719 were prescribed ICS at least once after their IPF diagnosis without having a concomitant diagnosis of either COPD or asthma, and 342 became regular ICS users (Additional Table A1).
Pneumonia hospitalisation
A total of 3,651 people with a diagnosis of IPF were admitted to hospital on at least one occasion for treatment for pneumonia, of whom 770 (21.1%) were regularly prescribed PPI and 391 (10.7%) were irregularly prescribed PPI in the year prior to IPF diagnosis. Similarly, 770 (21.1%) of the people hospitalised with pneumonia had a history of regular ICS prescription; 30.1% of those hospitalised had been prescribed ICS at least once.
In separate analyses, prescription of PPIs and ICS in the 12-month period prior to IPF diagnosis was associated with an increased risk of hospitalisation for pneumonia. For people regularly prescribed a PPI, the fully-adjusted HR for this association (comparing with people not prescribed PPIs in the baseline period) was 1.14 (95% CI: 1.04–1.24); for people prescribed PPIs less frequently, the corresponding HR was of similar magnitude (HR = 1.12 (95% CI: 1.00–1.26). (Fig. 2; Additional Table A2). However, in the case of ICS, there was some evidence of a dose–response relationship; the adjusted HR for pneumonia hospitalisation for regular users was 1.40 (95%CI, 1.25–1.55) while that for irregular users was 1.19 (95% CI, 1.06–1.33) (Fig. 2; Table A2).
Analyses of the rate of hospitalisations for pneumonia revealed similar patterns. An increased rate of hospitalisations due to pneumonia was observed in patients who were regularly prescribed PPI in the baseline period and also in patients irregularly prescribed PPI (relative to non-users) [IRRadj=1.14 (95% CI: 1.03–1.26) and IRRadj=1.18 (95% CI: 1.03–1.35), respectively]. Regular prescriptions of ICS also increased the rate of pneumonia hospitalisation; the rate of hospitalisations was increased by 50% in regular ICS users (IRRadj=1.51; 95% CI: 1.33–1.72) and by 26% in irregular users (IRRadj=1.26; 95% CI: 1.10–1.44), compared with non-users (Additional Table A3).
All-cause mortality
Among our study cohort, all of whom were diagnosed with IPF after 1/1/2010, around a half (n = 8,964; 52.4%) died within the period of study (before the end of 2019). The median survival time for this group of patients was 1.7 years (IQR: 0.68–3.25). Prescription of PPIs (regularly or irregularly) in the baseline period was not significantly associated with all-cause mortality (HRadj= 1.04, 95% CI: 0.90–1.21 and HRadj= 1.05, 95% CI: 0.86–1.28, respectively) (Fig. 2). In a crude Cox regression analysis, we found that people who were prescribed ICS on a regular basis had a small increased risk of dying from any cause relative to non-ICS users (HRcrude=1.06; 95%CI, 1.03–1.25) (Table S2). Furthermore, adjustment for both demographic/lifestyle factors and comorbidities had the effect of increasing the HR for this association (HRadj=1.25; 95%CI, 1.16–1.34), reflecting the tendency in this IPF cohort at least towards a high prevalence of comorbidities such as COPD in the ICS users compared with the non-users (Table 1). Irregular use of ICS was not however associated with a shorter survival time, relative to non-use, in either a crude analysis or adjusted analysis (Table S2).
Sensitivity analyses
Details of our sensitivity analyses are provided in the data supplement. In summary, we found that in the case of our PPI analysis neither GORD or hiatus hernia were effect modifiers of the association between PPI prescribing and our primary outcome, hospitalisation for pneumonia. Likewise, we found only weak evidence to suggest that concomitant diagnoses of COPD and/or asthma had a modifying effect on the relationship between ICS prescribing and risk of pneumonia hospitalization (see Additional Tables S4–S5).
Table 1 | Baseline characteristics of study cohort
Characteristic
|
Whole cohort
|
|
PPI
|
|
ICS
|
Regularly prescribed
|
Irregularly prescribed
|
Not prescribed
|
Regularly prescribed
|
Irregularly prescribed
|
Not prescribed
|
n (% of total)a
|
n (% of total)a
|
n (% of total)a
|
n (% of total)a
|
|
n (% of total)a
|
n (% of total)a
|
n (% of total)a
|
Total
|
17,105 (100.0%)
|
|
3,396 (19.9%)
|
1,649 (9.6%)
|
12,060 (50.5%)
|
|
2,741 16.0%)
|
1,507 (8.8%)
|
12,857 (75.2%)
|
Demographic and lifestyle
|
|
|
|
|
|
|
|
|
Male
|
10,706 (62.6)
|
|
2,058 (60.6%)
|
1,000 (60.6%)
|
7,648 (63.4%)
|
|
1,584 (57.8%)
|
894 (59.3%)
|
8,228 (64.0%)
|
Female
|
6,399 (37.4)
|
|
1,338 (39.4%)
|
649 (39.4%)
|
4,412 (36.58%)
|
|
1,157 (42.2%)
|
613 (40.9%)
|
4,629 (36.0%)
|
Age (years)
|
|
|
|
|
|
|
|
|
|
Median (IQR)
|
76 (69–83)
|
|
77 (70– 83)
|
76 (69 - 83)
|
77 (70– 83)
|
|
76 (68 –81)
|
75 (67–81)
|
77 (70–83)
|
40–59
|
1,263 (7.4%)
|
|
188 (5.5%)
|
129 (7.8%)
|
859 (7.12%)
|
|
241 (8.8%)
|
148 (9.8%)
|
874 (6.8%)
|
60–69
|
3,214 (18.8%)
|
|
598 (17.6%)
|
322 (19.5%)
|
2,147 (17.8%)
|
|
570 (20.8%)
|
358 (23.8%)
|
2,286 (17.8%)
|
70–79
|
6,516 (38.1%)
|
|
1,304 (38.4%)
|
585 (35.5%)
|
4,489 (37.22%)
|
|
1,104 (40.3)
|
546 (36.2%)
|
4,866 (37.9%)
|
>= 80
|
6,112 (35.7%)
|
|
1,306 (38.5%)
|
613 (37.2%)
|
4,565 (37.9%)
|
|
826 (30.1%)
|
455 (30.2%)
|
4,831 (37.6%)
|
Index of Multiple Deprivation
|
|
|
|
|
|
|
|
1 (least deprived)
|
3,530 (20.7%)
|
|
658 (19.4%)
|
301 (18.3%)
|
2,571 (21.3%)
|
|
458 (16.7%)
|
319 (21.2%)
|
2,753 (21.4%)
|
2
|
3,676 (21.5%)
|
|
675 (19.9%)
|
345 (21.0%)
|
2,656 (22.0%)
|
|
509 (18.6%)
|
307 (20.4%)
|
2,860 (22.2%)
|
3
|
3,204 (18.8%)
|
|
611 (18.0%)
|
312 (19.0%)
|
2,281 (18.9%)
|
|
492 (18.0%)
|
304 (20.2%)
|
2,408 (18.7%)
|
4
|
3,211 (18.8%)
|
|
659 (19.5%)
|
313 (19.0%)
|
2,239 (18.6%)
|
|
561 (20.5%)
|
265 (17.6%)
|
2,385 (18.6%)
|
5 (most deprived)
|
3,464 (20.3%)
|
|
785 (23.2%)
|
375 (22.7%)
|
2,304 (19.1%)
|
|
713 (26.1%)
|
311 (20.6%)
|
2,440 (19.0%)
|
Missing
|
20 (0.1%)
|
|
8 (0.2%)
|
3 (0.2%)
|
9 (0.1%)
|
|
8 (0.29%)
|
1 (0.07%)
|
11 (0.09%)
|
Body mass index (kg/m2)
|
|
|
|
|
|
|
|
|
Underweight
|
466 (2.7)
|
|
72 (2.1%)
|
41 (2.5%)
|
353 (2.9%)
|
|
79 (2.9%)
|
47 (3.1%)
|
340 (2.6%)
|
Normal (baseline)
|
3,775 (22.1)
|
|
710 (20.9%)
|
395 (24.0%)
|
2,671 (22.1%)
|
|
641 (23.4%)
|
313 (20.8%)
|
2,821 (21.9%)
|
Overweight
|
4,502 (26.3)
|
|
969 (28.5%)
|
438 (26.6%)
|
3,095 (25.7%)
|
|
771 (28.1%)
|
400 (26.5%)
|
3,331 (25.9%)
|
Obese
|
3,399 (19.9)
|
|
723 (21.3%)
|
294 (17.8%)
|
2,122 (17.6%)
|
|
705 (25.7%)
|
373 (24.8%)
|
2,321 (18.1%)
|
Missing
|
4,963 (29.0)
|
|
922 (27.2%)
|
481 (29.2%)
|
3,820 (31.7%)
|
|
545 (19.9%)
|
374 (24.8%)
|
4,044 (31.5%)
|
Smoking status
|
|
|
|
|
|
|
|
|
|
Never
|
1,950 (11.4%)
|
|
335 (9.9%)
|
185 (11.2%)
|
1,430 (11.9%)
|
|
218 (8.0%)
|
173 (11.5%)
|
1,559 (12.1%)
|
Ex-smoker
|
12,431 (72.7%)
|
|
2,558 (75.3%)
|
1,221 (74.0%)
|
8,667 (71.8%)
|
|
2,024 (73.8%)
|
1,091 (72.4%)
|
9,316 (72.5%)
|
Current smoker
|
2,712 (15.8%)
|
|
503 (14.8%)
|
243 (14.8%)
|
1,951 (16.2%)
|
|
499 (18.2%)
|
243 (16.1%)
|
1,970 (15.3%)
|
Missing
|
12 (0.1%)
|
|
0 (0.0%)
|
0 (0.0%)
|
12 (0.1%)
|
|
0 (0.0%)
|
0 (0.0%)
|
12 (0.1%)
|
Comorbidities
|
|
|
|
|
|
|
|
|
|
GORD
|
3,957 (23.1%)
|
|
1,330 (39.2%)
|
527 (68.0%)
|
2,100 (17.4%)
|
|
763 (27.8%)
|
372 (24.7%)
|
2,822 (22.0%)
|
COPD
|
2,777 (16.2%)
|
|
654 (19.26%)
|
276 (16.7%)
|
1,847 (15.3%)
|
|
1,396 (50.9%)
|
384 (25.5%)
|
997 (7.8%)
|
Asthma
|
2,911 (17.0%)
|
|
695 (20.5%)
|
291 (17.7%)
|
1,925 (16.0%)
|
|
1,662 (60.6%)
|
481 (31.9%)
|
768 (6.0%)
|
PH
|
313 (1.8%)
|
|
66 (1.6%)
|
29 (1.8%)
|
218 (1.8%)
|
|
56 (2.0%)
|
25 (1.7%)
|
232 (1.8%)
|
Lung cancer
|
137 (0.8%)
|
|
28 (0.85%)
|
15 (0.9%)
|
94 (0.8%)
|
|
32 (1.2%)
|
11 (0.7%)
|
94 (0.7%)
|
IHD
|
4,657 (27.2%)
|
|
1,242 (36.6%)
|
493 (29.9%)
|
2,922 (24.2%)
|
|
735 (26.8%)
|
387 (25.7%)
|
3,535 (27.5%)
|
Hiatus hernia
|
2,445 (14.3%)
|
|
894 (26.3%)
|
293 (17.8%)
|
1,258 (10.4%)
|
|
456 (16.6%)
|
212 (14.1%)
|
1,777 (13.8%)
|
Heart failure
|
1,844 (10.8%)
|
|
423 (12.5%)
|
182 (11.0%)
|
1,239 (10.3%)
|
|
284 (10.4%)
|
117 (7.8%)
|
1,443 (11.2%)
|
Diabetes
|
4,052 (23.7%)
|
|
908 (26.7%)
|
416 (25.2%)
|
2,728 (22.6%)
|
|
633 (23.1%)
|
357 (23.7%)
|
3,062 (23.8%)
|
COPD, chronic obstructive pulmonary disease; GORD, gastro-oesophageal reflux disease; ICS, inhaled corticosteroid; IHD, ischaemic heart disease; PH, pulmonary hypertension; PPI, proton pump inhibitor
a Unless otherwise specified.
Table 2 | PPI prescribing patterns pre- and post- IPF diagnosis
Prior prescription
|
|
|
Prescription in year post IPF diagnosis
|
|
|
No PPI
|
Irregular PPI
|
Regular PPI
|
|
Total
|
No PPI
|
|
10,680
|
900
|
480
|
|
12,060
|
Irregular PPI
|
|
639
|
528
|
482
|
|
1,649
|
Regular PPI
|
|
228
|
609
|
2,559
|
|
3,396
|
|
|
|
|
|
|
|
Total
|
|
11,547
|
2,037
|
3,521
|
|
17,105
|