Trends in hospital admission rates for whooping cough in England across five decades: database studies

Abstract

Objectives

Our aim was to report on trends in hospitalisation rates for pertussis in England from the 1960s to 2011; and to provide context for the recent unexpected activity of Bordetella pertussis in the UK.

Design

A retrospective analysis of English national Hospital Episode Statistics (HES, 1968–2011) and the Oxford Record Linkage Study (ORLS, 1963–2011) for people admitted to hospital with whooping cough.

Setting

England and the Oxford Record Linkage Study area.

Main outcome measures

Age- and gender-specific hospital admission rates, and summary age- and sex-standardised rates, for people aged under 25 years per 100,000 population in each age group.

Results

Admission rates declined from the 1960s to the early 1970s. For example, the standardised rates were 12.8 (95% confidence interval 11.2–14.5) per 100,000 in England in 1968 and 4.0 (3.0–4.9) per 100,000 in 1973. They then increased to reach 45.0 (41.4–48.6) per 100,000 in 1978 and 47.4 (43.7–51.1) in 1982. From the late 1980s, admission rates continued to decline, falling to between 1 and 4 per 100,000 in each of the years between 2003 and 2011. While the trend in hospital admissions closely followed that in notifications, the annual ratio between these two measures was not consistent ranging from 1.07 (95% confidence interval 1.00–1.14) to 4.03 (3.79–4.27) notifications per admission over the last 10 years.

Conclusions

Epidemics of whooping cough in the late 1970s and early 1980s were associated with a significant rise in hospital admission rates. Current admission rates are low, by historical comparison. Vaccine programmes must continue to be fully implemented in order to improve control of pertussis activity.

Introduction

Whooping cough, caused by Bordetella pertussis, is an important vaccine-preventable disease. Its incidence has declined in England, and elsewhere, in the era of widespread uptake of immunisation, but it has increased recently in England^1,2 and elsewhere.³ Data on statutory notifications of cases of whooping cough are readily available.² However, the relationship between notifications and rates of disease severe enough to warrant hospital admission have only been examined over relatively brief periods in separate papers.^4–6 We have used an English regional dataset that spans 1963–2011, and an English national dataset that spans 1968–2011, to provide a full five-decade profile of hospital admission rates for whooping cough.

Method

Datasets

Data on trends in hospital admissions were obtained from the Oxford Record Linkage Study (ORLS) which contains linked records of all inpatient admissions between 1963 and 2011 for its study population. The ORLS covered residents of Oxfordshire from 1963, Oxfordshire and West Berkshire from 1968, these plus four other health districts from 1975, and all eight health districts (four counties) that comprised the former Oxford NHS Regional Health Authority from 1991. The counties are Oxfordshire, Buckinghamshire, Berkshire and Northamptonshire, with a population of approximately 2.5 million people. National data on hospital admissions for whooping cough in England (population about 50 million) were analysed in a database containing a combination of data from the Hospital In-patient Enquiry (HIPE – a one-in-ten sample of admissions in England from 1968 to 1985) and Hospital Episode Statistics (HES – a complete national enumeration that covers 1990–2011). The existence of linkable data in the ORLS from its inception in 1963 enabled us to identify any patients with whooping cough who were readmitted to, or transferred between, hospitals in order to count each patient once only. It also provides a continuous run of data across all years, unlike the all-England datasets that have no data for the latter part of the 1980s. The data in the HIPE/HES dataset are only linkable from 1999. The linked data provide the means, as in the ORLS, to identify two or more admissions for the same individual as relating to one person. Prior to 1999, it is not possible to distinguish between numbers of admissions and numbers of individual people admitted. The International Classification of Diseases (ICD) codes used for whooping cough were 056, 033, 033 and A37 in, respectively, the 7th, 8th, 9th and 10th revisions of the ICD, in any diagnostic position on a hospital record.

Analysis

Age- and sex-specific hospital admission rates were calculated for people aged <1 year, 1–4 years, 5–9 years, 10–14 years, 15–19 years and 20–24 years. Summary age- and sex-standardised rates for people aged under 25 years were also calculated, using these age groups, the direct method of standardisation, and the European standard population. Hospitalisation rates were calculated in single calendar years and in five-year grouped time periods.

Notification data

Annual national data on statutory notifications for whooping cough in England were obtained for England and Wales⁷ and those for England alone were calculated by subtracting the data for Wales.⁸ Notifications were compared with the hospital admission data from 1999 (when wholly comparable data became available). The ratio of notifications to hospital admissions was calculated in each year.

Results

Figure 1 shows the age-standardised rates of hospital admissions for whooping cough from 1963 to 2011 in the ORLS area and from 1968 to 2011 in England. They show the expected periodicity with peaks every three or four years. Age- and sex-standardised admission rates declined from the 1960s to the early 1970s. For example, they were 12.8 (95% confidence interval 11.2–14.5) per 100,000 in England in 1968 and 4.0 (3.0–4.9) per 100,000 in 1973. They then increased to reach 45.0 (41.4–48.6) per 100,000 in 1978 and 47.4 (43.7–51.1) in 1982.

Figure 1.

Whooping cough: hospital admission rates in England and in ORLS for people aged under 25 years: directly age-standardised admission rates per 100,000 population under 25.

The annual rates declined in the 1980s and 1990s, still showing year-on-year fluctuations, with low rates of (for example) 8.1 (5.9–10.3) per 100,000 in 1984 and 6.8 (5.8–6.9) in 1991, interspersed with higher rates of (for example) 21 (18.0–25.1) in 1985 and 14.8 (13.9–15.6) in 1990. They have fallen further to between 1 and 4 per 100,000 in each of the years between 2003 and 2011.

Table 1 shows the all-England data grouped into five-year periods by age group and sex (See Appendix 1 for primary data). The majority of cases were in the <1 year age group. Whooping cough was generally a little more common in female patients than male patients (unlike most infectious diseases). Table 1 also shows that the trends seen in the Figure 1 were found in every age group shown.

Table 1.

Age-specific hospital admission rates (events per 100,000 population) for whooping cough in England in people aged under 25 years, 1968–2011.

Year of admission	Age (years)					Overall rate (crude)	Overall rate (standardised)	95% CI (l)	95% CI (u)
	0–4	5–9	10–14	15–19	20–24
Female patients
1968–70	45.2	3.6	0.8	0.0	0.0	10.9	10.9	9.7	12.1
1971–75	51.0	3.3	0.5	0.0	0.0	11.0	12.1	11.0	13.1
1976–80	108.9	5.8	0.4	0.1	0.0	19.4	25.4	23.7	27.2
1981–85	101.8	10.6	0.6	0.0	0.2	19.6	24.8	23.1	26.6
1986–90	58.7	8.5	0.9	0.0	0.0	13.4	14.9	14.0	15.8
1991–96	28.6	4.3	0.8	0.0	0.0	6.9	7.4	7.1	7.6
1997–01	27.4	2.5	0.5	0.1	0.0	5.9	6.7	6.3	7.0
2002–06	15.2	0.9	0.4	0.0	0.0	3.1	3.6	3.4	3.8
2007–11	11.8	0.1	0.3	0.0	0.0	2.4	2.7	2.5	2.9
Male patients
1968–70	43.4	3.6	0.8	0.0	0.0	10.4	10.5	9.3	11.7
1971–75	47.3	1.7	0.0	0.0	0.0	9.9	10.8	9.9	11.8
1976–80	88.2	3.2	0.3	0.0	0.0	15.5	20.3	18.8	21.8
1981–85	94.9	7.5	0.6	0.0	0.0	17.9	22.7	21.1	24.3
1986–90	49.7	5.4	1.0	0.0	0.0	11.1	12.3	11.5	13.1
1991–96	24.4	2.6	0.6	0.0	0.0	5.7	6.1	5.8	6.3
1997–01	24.1	1.6	0.5	0.1	0.0	5.2	5.8	5.4	6.1
2002–06	12.9	0.6	0.3	0.1	0.0	2.6	3.1	2.9	3.3
2007–11	9.7	0.1	0.2	0.0	0.0	2.0	2.2	2.1	2.4

Numbers of admissions in England were compared with statutory notifications from 1999 onwards (when person-based data, derived from record linkage, became available nationally). There was a tendency for the ratio of notifications per admission to increase in years with high numbers of cases (Table 2). For example, in 2006, 270 people were admitted to hospital with whooping cough; there were 539 notifications; and the ratio of notifications to people admitted was 2.00 (95% confidence interval 1.83–2.17) (Table 2). The corresponding figures in 2008 were 419 people, 1472 notifications and a ratio of 3.53 (3.35–3.71) (Table 2). The confidence intervals in the two years do not overlap demonstrating that the difference between the ratios in 2006 and 2008 are statistically significant.

Appendix 1.

Numbers on which Table 1 is based.

Year of admission	Age (years)					Total
	0–4	5–9	10–14	15–19	20–24
Female patients
1963–70	272	21	4	0	0	297
1971–75	460	32	5	0	0	497
1976–80	800	52	4	1	0	857
1981–85	732	74	5	0	2	813
1986–90	912	125	12	0	0	1049
1991–96	2684	398	65	2	2	3151
1997–01	1192	114	23	4	0	1333
2002–06	1068	63	32	2	0	1165
2007–11	915	7	22	3	0	947
Male patients
1963–70	271	22	4	0	0	297
1971–75	450	18	0	0	0	468
1976–80	684	30	3	0	0	717
1981–85	719	55	5	0	0	779
1986–90	807	82	14	0	0	903
1991–96	2397	252	53	3	1	2706
1997–01	1099	79	23	3	0	1204
2002–06	954	45	26	5	1	1031
2007–11	795	10	19	0	0	824

Table 2.

Number of notifications, number of people admitted to hospital (and number of episodes of hospital care), and ratio of notifications per person admitted, for whooping cough, England, 1999–2011.

Year of admission	Notifications	People admitted (episodes)	Ratio (95% CI)
1999	1077	846 (976)	1.27 (1.20–1.35)
2000	658	545 (630)	1.21 (1.12–1.30)
2001	851	797 (934)	1.07 (1.00–1.14)
2002	858	552 638)	1.55 (1.45–1.66)
2003	386	292 (343)	1.32 (1.19–1.46)
2004	475	329 (380)	1.44 (1.32–1.58)
2005	567	446 (529)	1.27 (1.17–1.38)
2006	539	270 (307)	2.00 (1.83–2.17)
2007	1071	281 (338)	3.81 (3.59–4.05)
2008	1479	419 (506)	3.53 (3.35–3.71)
2009	1123	279 (333)	4.03 (3.79–4.27)
2010	377	156 (180)	2.42 (2.18–2.67)
2011	810	349 (414)	2.32 (2.16–2.49)
Total	10,271	5561 (6508)	1.85 (1.81–1.88)

The contribution of record linkage can be judged from comparisons of numbers of hospital admissions per year and numbers of individual people admitted per year (Table 2). In England from 1999 to 2011, there were 6508 episodes of care for 5561 individuals.

Discussion

Admission rates showed the periodicity associated with whooping cough that has been recognised for decades.^9,10 Prior to immunisation, introduced routinely in the early 1960s, in epidemic years there were about 160,000 notifications for whooping cough in England.² Our data add the dimension of trends in whooping cough cases that were deemed sufficiently severe to warrant hospital admission. Immunisation rates fell substantially from the 1960s to the mid-1970s: in 1972, when coverage of infants and children was around 80%, there were only about 2000 notifications,² and, in our data, there were admission rates of 3 per 100,000 people aged under 25 years. Widespread doubts about the efficacy and safety of immunisation against whooping cough were raised in the early 1970s following publications in The Lancet and elsewhere^11,12 and vaccine coverage fell to about 30%.¹² There were epidemics of whooping cough in 1977–1979 and 1981–1983 in which annual numbers of notifications reached 60,000²; and, in our data, annual admission rates reached 45 and 47 per 100,000, respectively, in 1978 and 1982. When fears about the safety of whooping cough vaccine were allayed, immunisation rates rose again and the incidence of whooping cough declined from the mid-1980s. Immunisation rates are currently a little over 90% in England, nationally, in the 2000s and hospitalisation rates are around 2–4 per 100,000. Notably, admission rates in those under 10 years have fallen from the 1990s to 2011 and are likely the result of the accelerated vaccination schedule introduced in 1990 and the pre-school booster introduced in 2001.⁶

Campbell et al.⁶ reported on notifications and hospital admissions for whooping cough from 1998 to 2009 in England and Wales. They showed that pertussis admission rates continued to be highest among infants and to peak every three to four years as reported in previous papers. From 1998 to 2009, they found a downward trend in hospitalisation rates among infants, with an overall 9% annual decrease (p < 0.001) and a corresponding decline in annual notification rates of 7% (p = 0.001). Nevertheless, there were 39 pertussis-related deaths during 2002–2009, of which 36 (92%) were among infants. More recently, in 2012 there were more deaths than in any single year since 1982, with 10 deaths, all in infants under 12 months of age.¹³ Thus, despite a long-term decline in incidence (as measured by both notifications and admission rates) in infants, deaths from whooping cough still occur. The persistence of whooping cough is in part explained by an increase in whooping cough cases in adolescents due to the waning of immunity over time.¹⁴ As a result, infants are still at risk of infection; and unvaccinated children are more likely to be admitted to hospital than vaccinated children and are more likely to have complications following admission.² Because of this, a temporary vaccination programme for pregnant women was introduced by the Department of Health in October 2012 following a record 9368 cases of whooping cough the previous year. The programme aims to provide passive immunity to neonates prior to their first vaccination. As of March 2013, around 57% of pregnant women had taken up the vaccine. As yet its effects are unclear but notifications for whooping cough have declined (e.g. 682 in May 2012 to 380 in May 2013).¹⁵

The scale of hospitalisation in the 2000s was about one admission for every 1.8 notifications. Generally, in recent years, hospitalisation rates for whooping cough rose and fell as notification rates rose and fell, as noted in an English regional dataset covering the 1970s.⁴ There was a tendency that, when the figures for whooping cough increased, the number of notifications increased proportionately more than the number of admissions. This, too, has been previously noted in the regional data for the 1970s.⁴ Two explanations are possible (which we cannot distinguish). First, as numbers of cases increase, doctors may become more aware of the disease and they may notify a higher percentage of cases. Second, as incidence increases, it is possible that a higher percentage of cases are relatively mild and do not warrant hospital admission. However, in those aged under six months Van Buynder et al.⁵ found that hospital admissions for whooping cough from 1995 to 1997 were greater than notifications (369 vs. 238) leading them to suggest that laboratory reports and notification data significantly underestimate the burden of pertussis-related morbidity in those under six months. Campbell et al.⁶ have also reported this phenomenon in those aged under three months consistently in every year from 1998 to 2009. Perhaps this reflects increased awareness and experience of the disease among paediatric hospital specialists. On the other hand, it may represent the difficulty in diagnosing pertussis in the very young in primary care. As Polymerase chain reaction laboratory diagnosis becomes increasingly widespread, it is expected that the discrepancy between hospital admissions and notifications in this age group will diminish.¹⁶

In summary, whooping cough has never been close to eradication in England for any prolonged period of time, and our study demonstrates that a high proportion of people diagnosed with whooping cough are admitted to hospital.

Declarations

Competing interests

None declared

Funding

The Unit of Health-Care Epidemiology has been funded by the English National. Institute for Health Research to build the datasets and undertake trend studies (RNC/035/002). This paper had no specific funding. The views expressed in this paper do not necessarily reflect those of the funding body.

Ethical approval

Ethical approval for analysis of the record-linkage study data was obtained from the Central and South Bristol Multi–Centre Research Ethics Committee (04/Q2006/176).

Guarantors

NH and MJG

Contributorship

MJG proposed the study, specified the analyses, and directed the building of the linked datasets. UH contributed to analysis. NH reviewed the literature and wrote the first draft. All contributed to interpretation of the findings and to the final manuscript.

Acknowledgements

None

Provenance

Not commissioned; peer-reviewed by Daniel Gibbons