Measles-mumps-rubella vaccination and respiratory syncytial virus-associated hospital contact.

BACKGROUND: The live measles vaccine has been associated with lower non-measles mortality and admissions in low-income countries. The live measles-mumps-rubella vaccine has also been associated with lower rate of admissions with any type of infection in Danish children; the association was strongest for admissions with lower respiratory infections.
OBJECTIVE: To examine whether measles, mumps, and rubella (MMR) vaccination was associated with reduced rate of hospital contact related to respiratory syncytial virus (RSV) in a high-income country.
METHODS: Nationwide cohort study of laboratory-confirmed RSV hospital contacts at age 14-23 months in all children born in Denmark 1997-2002 who had already received the vaccine against diphtheria, tetanus, pertussis (acellular), polio, and Haemophilus influenzae type b (DTaP-IPV-Hib) at the recommended ages of 3, 5, and 12 months.
RESULTS: The study included 888 RSV hospital contacts in 128,588 person years of follow up (rate 6.8/1000 person years). Having MMR as the most recent vaccine was associated with a reduced rate of RSV hospital contacts compared with having DTaP-IPV-Hib as the most recent vaccine (Incidence rate ratio (IRR), 0.75; 95% confidence interval (CI), 0.63-0.89). After adjustment for potential confounders including exact age in days the IRR was 0.78 (95% CI, 0.66-0.93). The adjusted IRR was 0.74 (95% CI, 0.60-0.92) in males and 0.84 (95% CI, 0.66-1.06) in females (P Interaction, 0.42). There was no association in the first month after MMR vaccination (adjusted IRR, 0.97; 95% CI, 0.76-1.24) but the adjusted IRR was 0.70 (95% CI, 0.58-0.85) from one month after MMR vaccination.
CONCLUSIONS: MMR vaccination was associated with reduced rate of hospital contacts related to laboratory-confirmed RSV infection. Further research on the association between MMR vaccination and other unrelated pathogens are warranted.

Introduction

Besides the disease-targeted effects, vaccines may affect morbidity and mortality to unrelated infections by changing the general level of resistance toward infections, the so-called nonspecific effects of vaccines [1], [2], [3]. In low-income countries, live vaccines like bacille Calmette–Guérin (BCG) against tuberculosis and measles vaccine have beneficial effects on all-cause child mortality [4], [5], [6], [7], [8]. In contrast, inactivated vaccines including diphtheria–tetanus–pertussis (DTP) vaccine may increase all-cause child mortality [9], [10], [11]. The nonspecific effects are often most marked in females [2], [8], [10], [11], [12]. Most findings from low-income countries relate to all-cause mortality. However, nonspecific effects of vaccinations on the incidence of infectious diseases and admission rates have been reported from both low-income [13], [14], [15], [16], [17] and high-income countries [18]. Recently, we found that the rate of admissions related to infections and particularly lower respiratory infections was reduced for Danish children following vaccination with the live MMR vaccine against measles, mumps, and rubella [19].

One of the most common causes of acute lower respiratory tract infections in infants is respiratory syncytial virus (RSV) [20], [21], [22]. Worldwide, an estimated 33.8 million new cases occur each year leading to 3.4 million hospital admissions of children under 5 years of age [21]. A study from Guinea-Bissau found that BCG vaccination reduced the risk of severe RSV infection [16]. The aim of the present study was to examine the association between MMR vaccine and the rate of hospital contacts resulting from RSV infection in a high-income setting. In the study period, the Danish recommendations were to administer MMR (Enders Edmonston, Jeryl Lynn, and Wistar RA 27/3) at 15 months of age after three doses of the inactivated vaccine against diphtheria, tetanus, pertussis (acellular), polio, and Haemophilus influenzae type b (DTaP-IPV-Hib) recommended at 3, 5 and 12 months of age (see Supplementary Fig. 1). The prespecified hypothesis was that children most recently vaccinated with MMR have a lower rate of RSV hospital contact compared with children vaccinated most recently with the third dose of DTaP-IPV-Hib (DTaP-IPV-Hib3).

Fig. 1

Flowchart of the cohort. Abbreviations: RSV respiratory syncytial virus; DTaP-IPV-Hib vaccination against diphtheria tetanus pertussis (acellular) polio and Haemophilus influenzae type b; MMR vaccination against measles mumps and rubella. In the parentheses, number of hospital contacts related to RSV from 14 months and until date of censoring for the children included in the study or until 2 years of age for the children excluded from the study. a DTaP-IPV or Hib alone (N = 17,882; 63.0%), not recommended combination of vaccines (N = 5348; 18.8%), fourth dose of DTaP-IPV-Hib (N = 3371; 11.9%), whole cell pertussis vaccine (N = 1011; 3.6%), OPV (N = 435; 1.5%), booster dose against diphtheria and tetanus (N = 252; 0.9%), vaccine against hepatitis B (N = 87; 0.3%), and second dose of MMR (N = 10; 0.0%). b Children receiving MMR before DTaP-IPV-Hib3 were excluded.

Material and methods

The Danish Civil Registration System was established in 1968 and all Danish residents are assigned a unique personal identification number [23]. The personal identification number is used by all Danish national registers and was used to link the registers for the present study.

Vaccination register

In Denmark, all recommended childhood vaccinations are administered free-of-charge by the general practitioner (GP). For the purpose of reimbursement, the GPs report all vaccinations to the counties and from the counties the data are transferred to the Danish National Health Service Register [24]. Based on this information we created a database of childhood vaccinations. Most childhood vaccinations were registered in a child's name, but occasionally childhood vaccinations were registered in a parent's name (5.7%), particularly for young infants who only received their own medical card after they had been named [25]. The recommended childhood vaccinations were only reimbursed by the counties for persons below 18 years of age. Childhood vaccinations registered to an adult can therefore be assumed to have been administered to a child and we assigned such vaccinations to that adult's child, which was closest to the scheduled age of that vaccine. Vaccinations are only registered on a weekly basis. We coded date of vaccination as Wednesday of the registered week of vaccination.

RSV-database

Information on RSV-related hospital contacts was obtained from the Danish nationwide RSV-database, which was established for research purposes by collection of information from the 18 Danish laboratories testing for RSV among patients at the Danish hospitals, described in detail elsewhere [26]. The RSV-database covers the period 1 January 1996 to 1 June 2003 were RSV was examined by ELISA or immunofluorescence. During this period, all admitted children with symptoms consistent with RSV were tested for RSV to facilitate isolation of RSV cases from other admitted children to reduce the risk of transmission. In children born in Denmark and registered in the Danish Civil Registration System, the incidence rates of hospital contacts with RSV were 25.2, 27.5, 16.0, and 6.8 per 1000 person years among the age groups less than 6 weeks, 6 weeks-6 months, 6–14 months and 14–24 months, respectively. We only included information on children born on 1 January 1997 and onwards, because the vaccination schedule changed considerably from 1996 to 1997.

Other register information

The Danish Civil Registration System contains information on vital status and emigration which we used to define inclusion date and follow-up periods [23]. It was also possible to obtain information about the composition of each child's household, and age and country of birth of the parents. The Danish Medical Birth Register contains information about birth weight, mode of delivery, gestational age, and maternal smoking in pregnancy [27]. The Danish National Patient Register contains information about discharge diagnoses [28]; we used this register to obtain information on other types of hospital contacts, including accidents and chronic diseases. We obtained information on household equivalence income [29], maternal education [30], and public childcare from Statistics Denmark.

Design

The study was designed as a cohort study with retrospective identification of children born in Denmark during 1 January 1997 and 31 March 2002 and who were alive and living in Denmark at 14 months of age. In the main analysis we only included children who had followed the recommended vaccination schedule for the first three vaccination visits by receiving DTaP-IPV-Hib1 before 4 months of age, DTaP-IPV-Hib2 before 6 months of age, and DTaP-IPV-Hib3 before 13 months of age. The purpose of this selection was to include children who resemble each other with respect to determinants of vaccination and thereby reduce bias. Further details of the inclusion are given in Fig. 1. Follow-up was stopped at 2 years of age since oral polio vaccine (OPV) was scheduled at 2 years of age until July 1, 2001 (see Supplementary Fig. 1). The Danish Data Protection Agency approved the study.

Statistical methods

To describe determinants of MMR vaccination, we estimated the risk ratios (RRs) of being MMR-vaccinated at 16 months and 2 years of age according to different covariates using Poisson regression with robust variance [31].

To estimate incidence-rate-ratios (IRRs) and 95% confidence intervals (CIs) of RSV hospital contact according to most recent vaccination we used Cox proportional hazard regression analysis. Hence, the children changed vaccination status from DTaP-IPV-Hib3 to MMR on the date of MMR vaccination. We included all RSV hospital contacts, so one child could have several RSV hospital contacts. To minimize the risk that the same episode of RSV infection counted as two hospital contacts we defined the duration of one RSV infection to be 14 days based on the expected maximal period of shedding [32]. These 14 days were excluded from the count of person years.

We used age as the underlying timescale of the Cox regression model and stratified by date of birth such that cases were only compared with children born on the same date and at the same age; hence, we controlled completely for any potential confounding from age, season, and calendar year. Furthermore, the model was adjusted for: sex, birth weight in grams (≤2000, 2001–2500, 2501–3000, 3001–3500, 3501–4000, 4001–4500, or >4500), gestational age (<37 weeks or ≥37 weeks of gestation), caesarean section (no or yes), number of admissions for any cause between 1 month of age and date of DTaP-IPV-Hib3 vaccination (none, one, two, or ≥three), admission for any cause from date of DTaP-IPV-Hib3 vaccination until 14 months of age (no or yes), maternal age at birth of the child in years (≤19, 20–24, 25–29, 30–34, 35–39, or ≥40), parental place of birth (Denmark, Denmark-Foreign, or Foreign), adult composition of the household (two adults, single parent, or no parents), other children in the household (no or yes) and the time-varying variable chronic diseases (no or yes) coded according to Kristensen et al., 2012 [33]. In additional analyses described in the supplementary appendix further adjustment was made for maternal smoking in pregnancy, childcare, household income, and maternal education.

Interactions

Nonspecific effects of vaccines have previously been found to interact with sex and vitamin A supplementation [2], [3]; other interactions might also be important. For this reason and to assess the generalizability of the results, we introduced interaction terms between MMR and dichotomised forms of the variables adjusted for in the main analysis. The interaction terms were introduced in the model one at a time and statistical significance was examined with Wald test statistics.

Sensitivity

The assumption of proportional hazards between vaccination groups was evaluated by Schoenfeld residuals and no violation was detected. We examined if the effect of MMR was immediate by splitting the time period with MMR at 30 days after MMR vaccination. In addition we examined for trend in the association according to time since MMR vaccination. The stability of the results was examined by continuing follow-up until 3 years of age. In a subgroup analysis, we excluded children who had been admitted to hospital with RSV before inclusion in the study to limit the potential effect of immunity to RSV. We examined the generalizability of the results by using a cohort which also included the children who had not received the first three doses of DTaP-IPV-Hib on time.

We do not believe that accidents are related to vaccination and we therefore performed an analysis in which the outcome was accidents registered at emergency room visits to check for possible registration bias in our data material.

All analyses were performed in Stata 12.

Results

A total of 168,511 children were followed until the first of the following events: age 2 years (N = 129,333; 76.8%), 1 June 2003 (N = 25,475; 15.1%), administration of other vaccines than the first MMR (mainly OPV) (N = 13,370; 7.9%), migration (N = 296; 0.2%), death (N = 31; 0.0%), unknown whereabouts for the Danish authorities (N = 4; 0.0%), and uncertainty about vaccinations for twins (N = 2; 0.0%).

Overall 60.6% were vaccinated with MMR by 16 months of age and 91.8% at 2 years of age. There was little variation in the likelihood of being MMR-vaccinated, but children of single parents, younger mothers, and children who lived together with other children were less likely to be MMR-vaccinated, particularly at 16 months of age (Table 1).

Table 1

Distribution of most recent vaccine and risk ratios of being MMR-vaccinated at 16 months and 2 years of age according to background factors.

Characteristics	16 months of age				2 years of age
	DTaP-IPV-Hib3% (N)a	MMR % (N)a	Adjusted RRb (95%-CI)	P valuec	DTaP-IPV-Hib3% (N)a	MMR % (N)a	Adjusted RRb (95%-CI)	P valuec
Sex
Male	39.7% (33,700)	60.3% (51,136)	1 (Ref)	.03	8.3% (6486)	91.7% (71,693)	1 (Ref)	.10
Female	39.1% (32,328)	60.9% (50,369)	1.01 (1.00–1.02)		8.0% (6147)	92.0% (70,357)	1.00 (1.00–1.01)
Birth weight, gram
≤2000	40.8% (834)	59.2% (1208)	1.02 (0.98–1.06)	.01	8.0% (152)	92.0% (1738)	1.01 (0.99–1.02)	.44
2001–2500	39.5% (2027)	60.5% (3107)	1.02 (1.00–1.05)		8.3% (396)	91.7% (4378)	1.00 (0.99–1.01)
2501–3000	38.3% (7978)	61.7% (12,856)	1.02 (1.01–1.04)		8.2% (1568)	91.8% (17,664)	1.00 (1.00–1.01)
3001–3500	39.5% (21,049)	60.5% (32,298)	1 (Ref)		8.2% (4043)	91.8% (45,178)	1 (Ref)
3501–4000	39.4% (22,254)	60.6% (34,200)	1.01 (1.00–1.02)		8.2% (4264)	91.8% (47,878)	1.00 (1.00–1.01)
4001–4500	40.1% (9634)	59.9% (14,377)	1.01 (0.99–1.02)		8.0% (1781)	92.0% (20,395)	1.01 (1.00–1.01)
>4500	39.4% (2252)	60.6% (3,459)	1.03 (1.00–1.05)		8.2% (429)	91.8% (4819)	1.01 (1.00–1.01)
Gestational age, weeks
<37	40.3% (3465)	59.7% (5139)	0.99 (0.96–1.01)	.18	7.8% (622)	92.2% (7366)	1.01 (1.00–1.02)	.04
≥37	39.4% (62,563)	60.6% (96,366)	1 (Ref)		8.2% (12,011)	91.8% (134,684)	1 (Ref)
Caesarean section
No	39.5% (56,015)	60.5% (85,886)	1 (Ref)	.60	8.2% (10,755)	91.8% (120,192)	1 (Ref)	.53
Yes	39.1% (10,013)	60.9% (15,619)	1.00 (0.99–1.01)		7.9% (1878)	92.1% (21,858)	1.00 (1.00–1.01)
Chronic diseases
No	39.3% (64,154)	60.7% (98,899)	1 (Ref)	.03	8.1% (12,242)	91.9% (138,144)	1 (Ref)	.40
Yes	41.8% (1874)	58.2% (2606)	0.97 (0.95–1.00)		9.1% (391)	90.9% (3906)	1.00 (0.99–1.01)
Number of admissions between 1 month of age and date of DTaP-IPV-Hib3 vaccination
None	39.1% (55,540)	60.9% (86,521)	1 (Ref)	<.001	8.0% (10,439)	92.0% (120,777)	1 (Ref)	<.001
One	40.8% (7858)	59.2% (11,393)	0.98 (0.97–0.99)		9.1% (1617)	90.9% (16,127)	0.99 (0.98–0.99)
Two	42.5% (1747)	57.5% (2367)	0.96 (0.93–0.99)		10.3% (392)	89.7% (3400)	0.98 (0.97–0.99)
Three or more	41.9% (883)	58.1% (1224)	0.98 (0.94–1.02)		9.6% (185)	90.4% (1746)	0.99 (0.98–1.00)
Admission from date of DTaP-IPV-Hib3 vaccination until 14 months of age
No	39.3% (64,073)	60.7% (98,879)	1 (Ref)	<.001	8.1% (12,212)	91.9% (138,262)	1 (Ref)	.001
Yes	42.7% (1955)	57.3% (2626)	0.95 (0.92–0.97)		10.0% (421)	90.0% (3788)	0.98 (0.97–0.99)
Maternal age at birth of the child, years
≤19	41.4% (1027)	58.6% (1452)	0.94 (0.91–0.98)	<.001	11.9% (277)	88.1% (2043)	0.95 (0.94–0.97)	<.001
20–24	37.1% (8509)	62.9% (14,411)	1.01 (1.00–1.02)		8.9% (1858)	91.1% (19,077)	0.98 (0.98–0.99)
25–29	38.1% (24,434)	61.9% (39,708)	1.01 (1.00–1.02)		7.5% (4443)	92.5% (54,593)	1.00 (0.99–1.00)
30–34	40.7% (22,732)	59.3% (33,069)	1 (Ref)		8.0% (4132)	92.0% (47,519)	1 (Ref)
35–39	42.0% (8231)	58.0% (11,357)	0.99 (0.98–1.01)		9.1% (1673)	90.9% (16,635)	0.99 (0.99–1.00)
≥40	42.1% (1095)	57.9% (1508)	0.99 (0.95–1.02)		10.3% (250)	89.7% (2183)	0.98 (0.96–0.99)
Parental place of birth
Denmark	39.2% (55,129)	60.8% (85,372)	1 (Ref)	.004	8.2% (10,583)	91.8% (119,060)	1 (Ref)	<.001
Denmark and foreign	40.5% (5539)	59.5% (8150)	0.98 (0.96–0.99)		8.4% (1063)	91.6% (11,629)	1.00 (0.99–1.00)
Foreign	40.2% (5360)	59.8% (7983)	1.01 (0.99–1.02)		8.0% (987)	92.0% (11,361)	1.01 (1.01–1.02)
Adult composition of the household
Two adults	39.2% (61,640)	60.8% (95,795)	1 (Ref)	<.001	7.9% (11,517)	92.1% (133,767)	1 (Ref)	<.001
Single parent	43.5% (4293)	56.5% (5573)	0.92 (0.91–0.94)		11.9% (1089)	88.1% (8088)	0.96 (0.95–0.97)
No parents	40.9% (95)	59.1% (137)	1.03 (0.92–1.15)		12.2% (27)	87.8% (195)	0.98 (0.93–1.03)
Other children in the household
No	34.9% (27,558)	65.1% (51,327)	1 (Ref)	<.001	6.7% (4892)	93.3% (67,715)	1 (Ref)	<.001
Yes	43.4% (38,470)	56.6% (50,178)	0.87 (0.86–0.88)		9.4% (7741)	90.6% (74,335)	0.97 (0.96–0.97)

Abbreviations: RR—risk ratio; MMR, vaccination against measles, mumps, and rubella.

The number of children do not add to the total number of children in the study, because some children were censored before the specified age.

RR of being MMR vaccinated at the specified age estimated by Poisson regression with robust variance and adjusted for all variables in the table.

P values for the test of the association between the specified variable and vaccination status.

In 128,588 person years of follow-up 888 RSV hospital contacts occurred (rate, 6.8/1000 person years); 886 children had one RSV hospital contact each, while one child had two RSV hospital contacts during follow-up. The rate of RSV hospital contact according to the vaccination status and age are given in Supplementary Table 1. The median duration of RSV hospital contacts was 3 days (inter quartile range, 2–5 days) in both DTaP-IPV-Hib3 and MMR-vaccinated children.

Compared with children having DTaP-IPV-Hib3 as the most recent vaccination, having MMR as the most recent vaccination was associated with a reduced rate of RSV hospital contacts (IRR, 0.75; 95% CI, 0.63–0.89; Table 2); the adjusted IRR was 0.78 (95% CI, 0.66–0.93; Table 2). Further adjustment for maternal smoking in pregnancy, childcare, household income, and maternal education did not alter the results (Supplementary Table 3). In males, the adjusted IRR of RSV hospital contacts for MMR compared with DTaP-IPV-Hib3 as the most recent vaccine was 0.74 (95% CI, 0.60–0.92), while the adjusted IRR for females was 0.84 (95% CI, 0.66–1.06; P Interaction, 0.42; Table 3). Also, none of the other examined interactions were statistically significant (Table 3).

Table 2

Incidence and incidence-rate-ratios of RSV hospital contact according to vaccination status.

Characteristics	RSV hospital contacts per 1000 person years (RSV hospital contacts/person yearsa)		Unadjusted IRRb (95%-CI)	P Valuec	Adjusted IRRd (95%-CI)	P Valuec
Vaccination status
DTaP-IPV-Hib3	8.9	(320/35,995)	1 (Ref)	.001	1 (Ref)	.006
MMR	6.1	(568/92,593)	0.75 (0.63–0.89)		0.78 (0.66–0.93)

Abbreviations: RSV—respiratory syncytial virus; IRR—incidence-rate-ratio; DTaP-IPV-Hib3—vaccination with the third dose against diphtheria, tetanus, pertussis (acellular), polio, and Haemophilus influenzae type b; MMR, vaccination against measles, mumps, and rubella.

The distribution of person years for each category indicates the demographic features of the included children.

Cox proportional hazards model with age as underlying time and stratified by date of birth thereby controlling for age and season.

P values for the test of the association between the most recent vaccine and RSV hospital contacts.

Cox proportional hazards model with age as underlying time, stratified by date of birth and adjusted for sex, birth weight, gestational age, caesarean section, chronic diseases, number of admissions between 1 month of age and date of DTaP-IPV-Hib3 vaccination, admission from date of DTaP-IPV-Hib3 vaccination until 14 months of age, maternal age at birth of the child, parental place of birth, adult composition of the household, and other children in the household (the IRR estimates for these variables are given in Supplementary Table 2).

Table 3

Results for two-way-interactions between vaccination status and dichotomised forms of the variables included in the main analysis.

Characteristics	RSV hospital contacts per 1000 person years (RSV hospital contacts/person years)		Unadjusted IRRa (95%-CI)	Adjusted IRRb (95%-CI)
Female
DTaP-IPV-Hib3	7.7	(136/17,637)	1 (Ref)	1 (Ref)
MMR	5.8	(265/45,774)	0.82 (0.65–1.04)	0.84 (0.66–1.06)
Male
DTaP-IPV-Hib3	10.0	(184/18,358)	1 (Ref)	1 (Ref)
MMR	6.5	(303/46,819)	0.70 (0.57–0.87)	0.74 (0.60–0.92)
P interactionc			.28	.42
No other children in the household
DTaP-IPV-Hib3	8.5	(132/15,580)	1 (Ref)	1 (Ref)
MMR	6.0	(270/44,896)	0.78 (0.61–0.99)	0.82 (0.65–1.04)
Other children in the household
DTaP-IPV-Hib3	9.2	(188/20,415)	1 (Ref)	1 (Ref)
MMR	6.2	(298/47,696)	0.74 (0.60–0.91)	0.76 (0.61–0.94)
P interactionc			.70	.57
No single parenthood
DTaP-IPV-Hib3	8.7	(294/33,604)	1 (Ref)	1 (Ref)
MMR	5.9	(519/87,445)	0.74 (0.62–0.89)	0.77 (0.64–0.92)
Single parenthood
DTaP-IPV-Hib3	10.9	(26/2391)	1 (Ref)	1 (Ref)
MMR	9.5	(49/5148)	0.96 (0.59–1.57)	1.00 (0.61–1.63)
P interactionc			.30	.30
Only Danish born parents
DTaP-IPV-Hib3	9.7	(294/30,157)	1 (Ref)	1 (Ref)
MMR	6.4	(502/77,870)	0.72 (0.60–0.86)	0.75 (0.63–0.90)
At least one parent born outside Denmark
DTaP-IPV-Hib3	4.5	(26/5838)	1 (Ref)	1 (Ref)
MMR	4.5	(66/14,723)	1.11 (0.69–1.77)	1.12 (0.70–1.80)
P interactionc			.08	.10
Maternal age <30 years
DTaP-IPV-Hib3	8.9	(168/18,871)	1 (Ref)	1 (Ref)
MMR	6.3	(313/50,078)	0.76 (0.61–0.94)	0.79 (0.64–0.98)
Maternal age ≥30 years
DTaP-IPV-Hib3	8.9	(152/17,124)	1 (Ref)	1 (Ref)
MMR	6.0	(255/42,514)	0.74 (0.59–0.93)	0.77 (0.62–0.97)
P Interaction c			.86	.88
Normal birth weight (>2500 g)
DTaP-IPV-Hib3	8.5	(294/34,484)	1 (Ref)	1 (Ref)
MMR	5.9	(522/88,670)	0.75 (0.63–0.90)	0.79 (0.66–0.94)
Low birth weight (≤2500 g)
DTaP-IPV-Hib3	17.2	(26/1511)	1 (Ref)	1 (Ref)
MMR	11.7	(46/3923)	0.69 (0.42–1.13)	0.73 (0.44–1.22)
P interactionc			.71	.79
Term (gestational age≥37 weeks)
DTaP-IPV-Hib3	8.6	(293/34,173)	1 (Ref)	1 (Ref)
MMR	5.8	(512/87,914)	0.74 (0.62–0.88)	0.77 (0.64–0.92)
Premature (gestational age<37 weeks)
DTaP-IPV-Hib3	14.8	(27/1822)	1 (Ref)	1 (Ref)
MMR	12.0	(56/4679)	0.87 (0.54–1.40)	0.94 (0.58–1.52)
P interactionc			.53	.42
No caesarean section
DTaP-IPV-Hib3	8.5	(260/30,582)	1 (Ref)	1 (Ref)
MMR	5.6	(441/78,714)	0.72 (0.60–0.87)	0.75 (0.62–0.90)
Caesarean section
DTaP-IPV-Hib3	11.1	(60/5413)	1 (Ref)	1 (Ref)
MMR	9.2	(127/13,879)	0.88 (0.63–1.22)	0.94 (0.67–1.31)
P interactionc			.25	.21
No chronic diseases
DTaP-IPV-Hib3	8.6	(301/34,985)	1 (Ref)	1 (Ref)
MMR	5.9	(528/90,110)	0.74 (0.62–0.89)	0.77 (0.65–0.93)
Chronic diseases
DTaP-IPV-Hib3	18.8	(19/1010)	1 (Ref)	1 (Ref)
MMR	16.1	(40/2483)	0.90 (0.51–1.57)	0.94 (0.53–1.67)
P interactionc			.52	.50
No admissions between 1 month of age and date of DTaP-IPV-Hib3 vaccination
DTaP-IPV-Hib3	7.4	(225/30,285)	1 (Ref)	1 (Ref)
MMR	4.9	(385/78,784)	0.72 (0.59–0.88)	0.74 (0.60–0.90)
Admissions during 1 month of age and date of DTaP-IPV-Hib3 vaccination
DTaP-IPV-Hib3	16.6	(95/5710)	1 (Ref)	1 (Ref)
MMR	13.3	(183/13,809)	0.86 (0.65–1.12)	0.89 (0.68–1.17)
P interactionc			.28	.23
No admission from date of DTaP-IPV-Hib3 vaccination until 14 months of age
DTaP-IPV-Hib3	8.5	(298/34,936)	1 (Ref)	1 (Ref)
MMR	5.8	(523/90,193)	0.74 (0.62–0.88)	0.76 (0.64–0.91)
Admission from date of DTaP-IPV-Hib3 vaccination until 14 months of age
DTaP-IPV-Hib3	20.8	(22/1059)	1 (Ref)	1 (Ref)
MMR	18.7	(45/2400)	1.06 (0.62–1.80)	1.12 (0.66–1.92)
P interactionc			.19	.16

Abbreviations: RSV—respiratory syncytial virus; IRR—incidence-rate-ratio; DTaP-IPV-Hib3—vaccination with the third dose against diphtheria, tetanus, pertussis (acellular), polio, and Haemophilus influenzae type b; MMR, vaccination against measles, mumps, and rubella.

Cox proportional hazards model with age as underlying time and stratified by date of birth thereby controlling for age and season.

Cox proportional hazards model with age as underlying time, stratified by date of birth and adjusted for sex, birth weight, gestational age, caesarean section, chronic diseases, number of admissions between 1 month of age and date of DTaP-IPV-Hib3 vaccination, admission from date of DTaP-IPV-Hib3 vaccination until 14 months of age, maternal age at birth of the child, parental place of birth, adult composition of the household, and other children in the household.

P value for the interaction term between vaccination status and the specified variables. This corresponds to testing if the IRR for MMR varies between categories of the other variables; for example, if the IRR for MMR differs between females and males.

The main results of the sensitivity analyses are given here; further data is available on request. In the first month after MMR vaccination, the adjusted IRR for MMR compared with DTaP-IPV-Hib3 was 0.97 (95% CI, 0.76–1.24) but this changed significantly after one month to an adjusted IRR of 0.70 (95% CI, 0.58–0.85). There was a significant trend for lower IRRs with longer time since MMR vaccination (Table 4). In the sensitivity analysis in which the children were followed to 3 years of age, an additional 129 cases of RSV hospital contacts were included and the estimate was unchanged (adjusted IRR, 0.78; 95% CI, 0.66–0.93). The adjusted IRR of RSV hospital contact was 0.75 (95% CI, 0.63–0.90) when the 3243 children who had been admitted with RSV before 14 months of age were excluded. The results were also stable when all children were included in the analysis irrespective of whether they had DTaP-IPV-Hib at the recommended ages (adjusted IRR, 0.79; 95% CI, 0.71–0.89).

Table 4

Results according to time since MMR vaccination.

Characteristics	RSV hospital contacts per 1000 person years (RSV hospital contacts/person years)		Unadjusted IRRa (95%-CI)	Adjusted IRRb (95%-CI)
Vaccination status
DTaP-IPV-Hib3	8.9	(320/35,989)	1 (Ref)	1 (Ref)
Days since MMR vaccination
0–30	8.7	(105/12,107)	0.98 (0.77–1.25)	1.01 (0.79–1.28)
31–90	7.0	(164/23,469)	0.78 (0.62–0.97)	0.82 (0.65–1.03)
91–210	5.6	(239/42,461)	0.62 (0.49–0.79)	0.64 (0.50–0.82)
>210	4.1	(60/14,546)	0.40 (0.27–0.60)	0.43 (0.29–0.64)
P trend			<0.001	<0.001

Abbreviations: RSV—respiratory syncytial virus; IRR—incidence-rate-ratio; DTaP-IPV-Hib3—vaccination with the third dose against diphtheria, tetanus, pertussis (acellular), polio, and Haemophilus influenzae type b; MMR, vaccination against measles, mumps, and rubella.

Cox proportional hazards model with age as underlying time and stratified by date of birth thereby controlling for age and season.

Cox proportional hazards model with age as underlying time, stratified by date of birth and adjusted for sex, birth weight, gestational age, caesarean section, chronic diseases, number of admissions between 1 month of age and date of DTaP-IPV-Hib3 vaccination, admission from date of DTaP-IPV-Hib3 vaccination until 14 months of age, maternal age at birth of the child, parental place of birth, adult composition of the household, and other children in the household.

There was no statistically significant association between MMR vaccination and emergency room visits resulting from accidents (adjusted IRR, 1.02; 95% CI, 0.98–1.06).

Discussion

As hypothesised children who had received MMR as their most recent vaccine had a lower rate of RSV hospital contacts compared with children who had DTaP-IPV-Hib3 as their most recent vaccination. The association was only apparent from 1 month after MMR vaccination. The association was similar for females and males and did not vary significantly according to any other background factors.

The main analyses of the present study included only children who had been vaccinated on time with DTaP-IPV-Hib to limit the possibility for confounding by factors related both to delayed vaccination and RSV hospital contact. There were too few RSV hospital contacts among children reversing the sequence of DTaP-IPV-Hib3 and MMR to perform an analysis in this group. It is important to emphasize that the statistical model had age in days as the underlying time scale securing elimination of any potential confounding by age by comparing the incidence of RSV hospital contacts between children of the exact same age but with different vaccination status. Inclusion of a wide range of additional potential confounders did not change the main estimate appreciably, indicating that confounding was limited in the present study.

Since there was no association between MMR vaccination and accidents, it seems unlikely that the association between MMR vaccination and RSV hospital contacts can be explained by a bias in health-seeking behavior.

Vaccinations were registered by GPs to obtain reimbursement; this provides an economic incentive to report all vaccines [24]. However, there might have been some underreporting of vaccines [34] and travel vaccines are not reimbursed and therefore not reported [24], but we believe travel vaccines are limited before two years of age. Any misclassification or underreporting of MMR vaccinations would bias the estimates toward no association. The outcome was RSV detected by ELISA or immunofluorescence and included in the RSV database, which cover 96% of RSV hospital contacts in Denmark [26]. RSV tests performed by ELISA or immunofluorescence are not as specific and sensitive as tests performed by PCR [35] indicating that some misclassification is present in the current study. However, there is no reason to believe that the sensitivity and specificity is affected by vaccination status; any misclassification would bias the results toward no association. There was a minimum of loss to follow-up because of the high quality of the Danish population-based registries [23].

Our previous observation of reduction in the rate of all-cause lower respiratory infections following MMR vaccination [19] can now partly be explained by a reduced rate of RSV hospital contacts. However, reduction in hospital contacts related to bacterial pathogens might also be important; a British self-controlled case-series study reported the risk of lobar pneumonia to be reduced 0–90 days following MMR vaccination [36]. In a similarly designed study of premature children from the US, the risk of wheezing lower respiratory disease in the first 30 days following MMR vaccination was reduced compared with 45–90 days after MMR vaccination [37]. These studies indicate a short-lived association. However, in the present study the association between MMR vaccination and RSV hospital contacts was only present from 30 days after MMR vaccination and continued throughout the third year of life.

In contrast to studies from West Africa of measles vaccination [38], [39], we did not find any statistically significant interaction between MMR vaccination and sex. MMR was associated with significant reduction in RSV hospital contacts in boys but not in girls. Boys have a higher risk of severe RSV infection [40], [41] as also seen in the present study; therefore there is greater statistical power to detect differences in boys. Furthermore, MMR vaccination may benefit boys more because of their higher risk of severe RSV infection. Also, there are numerous differences between high and low-income settings that might interact with the nonspecific and sex-differential effects of vaccinations, including family size, nutritional status, micronutrient supplementation, sex-differential access to care, the presence of atopic disease, exposure to infections in public child care institutions, and genetic differences.

The present study suggests that the MMR vaccine may have beneficial nonspecific effects by reducing the rate of hospital contacts related to RSV in a high-income setting. Only the most severe cases of RSV are cared for at the hospital, so the reduction in hospital contacts could be related to a reduction in the severity of RSV infection, enhanced resistance to RSV infection or both. The finding contradicts the current perception of vaccines as an intervention with a disease-specific effect only. Our observation might be interpreted as a chance finding or a finding resulting from uncontrolled confounding of factors related to delay in vaccination and the risk of RSV hospital contact. However, there were hardly any difference in characteristics between children who were and were not MMR-vaccinated at 16 and 24 months of age (Table 1). Furthermore, if the beneficial effect of MMR resulted from confounding, the same benefit should have been observed in the first month after vaccination. Several randomized trials from low-income countries have documented that nonspecific effects of vaccines are indeed possible [1]. In one of these trials from Guinea-Bissau [5] measles vaccine was particularly effective in reducing hospital admissions related to respiratory infections [17].

Nonspecific effects of vaccines have often been dismissed as biologically implausible. However, immunological studies have shown that a pathogen encounter may alter the subsequent resistance toward other infections [2], [3]. These changes can occur at the level of the innate immune system; for example, BCG may enhance the nonspecific innate resistance through epigenetic reprogramming of monocytes [42]. Numerous animal studies have shown that pathogens can also induce nonspecific alterations of the adaptive immune system, through cross-reacting T-cell epitopes [43]. Notably, measles virus, mumps virus, and RSV all belong to the family paramyxoviridae and one study in mice found T-cell cross reactivity between these viruses [44] and a study in children found that prior RSV infection was related to higher antibody titres against measles, mumps, and rubella following vaccination with MMR [45]. Further studies on the biological mechanisms are clearly needed, in particular whether antigen-specific responses or non-specific inflammatory/innate mechanisms are more important for the protection we observed in the present study.

The current WHO recommendations do not consider nonspecific effects of vaccines. However, WHO's Strategic Advisory Group of Experts has recently recommended further research into nonspecific effects of vaccines [46]. Ideally, the age schedule for vaccinations should be evaluated in trials testing both the specific effect and the nonspecific effects on overall morbidity and mortality. In high-income countries it might reduce the risk of hospital contacts and reduce health care costs to lower the age of MMR vaccination or introduce an additional dose of MMR vaccination at an earlier age, particularly bearing in mind that the greatest burden of RSV occurs in children below 12 months of age [21]. If confirmed in further studies in high-income settings, the nonspecific beneficial effect of MMR vaccination could be used as an additional incentive for parents to have their children vaccinated on time.

In conclusion, the present nationwide cohort study showed that the rate of RSV hospital contact was reduced in Danish children who had received MMR rather than DTaP-IPV-Hib3 as their most recent vaccination. The finding has major public health implications. Further studies are needed to test this observation and to examine whether earlier administration of MMR might be beneficial.

Funding

This work was supported by the [2009B132 to S.S.]; Rosalie Petersens Foundation [to S.S.]; [DNRF108 to S.S., C.S.B., L.G.S. and H.R.]; [ERC-2009-StG, grant agreement no. 243149 to C.S.B.]; and the Novo Nordisk Foundation [research professorship grant to P.A.]. The funders had no role in study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the article for publication

Author's contributions

S.S. conceptualized and designed the study, acquired the vaccination and confounder data, analyzed and interpreted the data, drafted the initial manuscript. C.S.B. and P.A. conceptualized and designed the study, interpreted the data and reviewed and revised the manuscript. L.G.S. conceptualized and designed the study, acquired the RSV hospital contact data, interpreted the data, reviewed and revised the manuscript. H.R. conceptualized and designed the study, acquired the vaccination and confounder data, interpreted the data, reviewed and revised the manuscript, All authors read and approved the final manuscript.