Measles is a highly contagious viral illness which is transmitted from human to human via the respiratory route; it affects non-immune individuals at any age.1 Measles can lead to serious complications, most prominently encephalitis and subacute sclerosing encephalitis, and causes >100,000 deaths each year.1,2 Peter Aaby and associates carried out measles related studies in West Africa. They noted that measles vaccination was associated with a reduction in mortality from other infectious diseases as well as deaths from measles.3,4 It has been known for over 100 years that measles leads to a transient, but significant immunosuppression, exemplified by negative tuberculin skin reactions for about a month’s duration in individuals who were previously tuberculin positive.5 Mina et al6 looked at national level data from England, Wales, the United States and Denmark for the pre and post measles vaccine eras. They found, following the introduction of measles vaccines in the respective countries, that deaths from other infectious diseases as well as measles deaths fell significantly and that this effect lasted for a 27.3-month period. More recently, the same group of investigators examined blood from 77 Dutch unvaccinated children collected before and ~10 weeks after measles virus infection.7 Mild and severe measles caused elimination of a median of 33% (range: 12%–73%) and 40% (range: 11%–62%), respectively, of their preexisting pathogen specific antibody repertoires. According to these findings, the sustained suppression of the immune system following measles, also referred to as “immune amnesia,” can mainly be attributed to the loss of long lived antibody producing plasma cells. Using the blood of 26 children from the same cohort of Dutch children, Petrova et al8 investigated the qualitative changes in the genetic composition of naive and memory B cells several days to weeks before and ~40 days after beginning of the measles rash. They found incomplete bone marrow reconstitution of naive B cells leading to immunologic immaturity despite resolution of clinical symptoms and a depletion of previously generated immune memory. These immunologic findings provide convincing explanations how measles compromises the immune system,9 consistent with clinical observations of an increased risk of infectious diseases over a prolonged period of time following measles.3,4,10,11
One of us (J.D.C.) has been involved in post measles immune amnesia studies in the Democratic Republic of the Congo (DRC). It was noted that there were increased rates of markers of acute infectious illnesses in children who had measles compared with children who did not have a history of measles.10 Until recently, there have been no studies of post measles immune amnesia in the developed world. Then, in a population-based matched cohort study in England, comprising 2228 children 1–15 years of age who had clinically diagnosed measles, Gadroen et al found an increased risk for infectious diseases up to 5 years after measles.11
Since measles has been both endemic and epidemic in Switzerland during the last 2 decades,12 we took advantage of this fact and have carried out a cohort study to investigate whether children after measles infection had a higher frequency of hospitalization due to other infectious disease than a control group of children who had not had measles.
Patients and Controls
The study was conceptualized in November 2017 with the original idea to conduct it throughout Europe. However, this was not feasible due to lack of resources and the heterogeneity of measles reporting in various countries. We therefore focused on Switzerland, the country of 1 of the principal investigators (U.H.). In Switzerland, measles is a reportable disease. However, due to data protection laws, the names of the reported patients are not disclosed. Therefore we had to limit ourselves to databases of hospitals where patients admitted with measles could be identified individually.
Any patient <18 years of age who was hospitalized with a primary clinical diagnosis of measles between 2000 and 2015 in 11 participating Swiss children’s hospitals (study sites) was eligible for inclusion in the study. Patient records were identified by ICD-10 discharge diagnoses for measles and/or keyword search at each study site by administrative staff. Then the list of cases with their dates of birth and initial date of hospitalization for measles was provided to the study investigator (L.B.) who retrieved the patient records and screened them one by one for the inclusion criterion, that is, that the coded diagnosis “measles” was correct. Patient records were then scrutinized for any exclusion criterion. Cases with chronic underlying disease were excluded because their conditions would lead to an increased risk for rehospitalizations independent of measles. Also, cases with permanent residence outside the catchment area of the hospitals were excluded. This was necessary because rehospitalizations—the study outcome—would not be likely to occur in the same hospital. The remaining cases were categorized as “possible,” “probable,” or “confirmed” cases according to the European Centre for Disease Prevention and Control (ECDC) diagnostic criteria for measles.
We attempted to identify 2 controls for each case. To do so, administrative staff at each study site provided a list of patients who were hospitalized within ± 1 month of each case and had similar age (±6 months). The study investigator retrieved the records of these potential controls, went through them 1 by 1 and checked them for the same inclusion and exclusion criteria as were applied to the cases. Further, the reason for hospitalization of potential control patients had to be a noninfectious illness with duration of up to 10 days. If >2 controls were identified, controls of the same sex and closest age were preferred over cases with a lower fit. Matching criteria were not loosened if <2 controls could be identified per case.
At each site, the study investigator screened the hospital’s admission database for any further hospitalization of each case and control within the 3 individual years following the original hospitalization and retrieved the records of re-hospitalizations for further analyses. Each rehospitalization was classified as an infectious or noninfectious episode.
Ethical approval was granted by the leading ethics committee “Ethikkommission Nordwest- und Zentralschweiz” (EKNZ 2018-00514).
We performed a sample size calculation based on the following assumptions: the hospitalization rate per year in non-exposed children (ie, controls) is 10%. A 20% increase in the hospitalization rate in exposed children (ie, cases) would correspond to a rate ratio of 1.2, that is, 12% per year. For a 2:1 ratio of non-exposed to exposed children, 5912 non-exposed and 2956 exposed would be needed to have 80% power. However, for reasons stated above, the study could not be carried out throughout Europe and therefore these numbers were not realistically achievable. We then decided to conduct the study with limited case numbers as a pilot project in Switzerland.
For comparing binary outcomes between groups, proportions (risks) and risk ratios (RRs) are reported along with the corresponding RR 95% confidence bounds. The P values for comparing proportions were computed using Fisher exact test. For events such as the number of rehospitalizations, where the same subject can have >1 event, rates per person and rate ratios are reported along with the corresponding 95% confidence bounds for the rate ratio. These confidence bounds and P values for comparing rates were computed assuming the number of events followed the Poisson distribution (Poisson test).
Due to different data storage policies, measles cases could not be identified for the complete study periods (2000–2015) in all study sites. Six sites provided cases back to year 2000 and 1 site each back to years 2003, 2005, 2007, 2009 and 2010. The search resulted in 156 records of patients hospitalized with measles, of whom 113 (72.4%) fulfilled the inclusion criteria (Fig. 1).
Cases and Controls
The mean age of the 113 cases (63 = 56% males) was 9.0 years (median 8.3 years, age range: 2 weeks to 17.8 years) at the time of admission for measles. Of these, 10 (9%) were categorized as “possible,” 25 (22%) as “probable” and 68 (60%) as “confirmed” measles according to ECDC definitions. The remaining 10 (9%) cases had a physician diagnosis of measles based on fever and a maculo-papular rash but records did not specify whether at least one of the 3 further signs qualifying for an ECDC diagnosis of possible measles, that is, cough, coryza or conjunctivitis, was present.
For 92 (81%) cases 2 controls were identified, for 12 (11%) cases (including 1 pair of twins) 1 control was identified and for the remaining 9 cases no control could be identified. Mean age of the 196 controls (117 = 60% males) was 9.0 years (median: 8.6 years, range: 0–18.0 years).
Rehospitalizations due to Infectious Diseases
Eleven (9.7%, 6 = 55% males) of 113 cases and 6 (3.1%, 5 = 83% males) of 196 controls had rehospitalizations due to an infectious disease. The RR was 3.18 (95% CI: 1.16–10.22, P = 0.02). These infections affected various organ systems without any specific notable pattern (Table 1). One case had 2 rehospitalizations and the remaining 10 cases and all 6 controls had 1 rehospitalization each (episode rates 0.106 per person versus 0.031 per person, ratio 3.47; 95% CI: 1.20–11.3; P = 0.012).
Figure 2 shows the time sequence of infectious disease rehospitalizations of cases and controls over the 3 year study period. As can be seen, rehospitalizations due to infectious diseases in cases peaked during the first year after measles with 9 of all 12 episodes compared with 3 of 6 episodes among controls [rates 0.080 per person vs 0.015 per person, ratio 5.20 (95% CI: 1.30–29.9); P = 0.012]. In years 2 and 3 after measles, the risk was not increased any longer [year 2: ratio 0.87 (95% CI: 0.015–16.7, P = 0.999); year 3: ratio 3.47 (95% CI: 0.181–206.4, P = 0.303); years 2 and 3 combined: ratio 1.73 (95% CI: 0.232–13.0, P = 0.675)].
Rehospitalizations due to Noninfectious Diseases
There were a total of 8 episodes of noninfectious rehospitalizations among 113 measles cases and 34 episodes among 196 control patients [rates: 0.071 per person vs 0.173 per person; ratio 0.41 (95% CI: 0.163–0.90); P = 0.023]. However, 24 (67.6%) of the 34 episodes among control patients were directly related to the diagnosis of the initial hospitalization when matched to the measles cases or a previous rehospitalization and 2 rehospitalizations in a measles case were also related to each other (Table 1). When those rehospitalizations were excluded, the episode rate in measles cases and controls were 0.062 and 0.051 per person, respectively, resulting in a case-control ratio of 1.214 (95% CI: 0.392–3.534; P = 0.802).
We analyzed the risk of rehospitalizations due to infectious diseases following measles in children during a 3 year period in Switzerland, an industrialized country in Europe where measles has remained endemic and epidemics also occur regularly.12 Most rehospitalizations occurred in the first year of the observation period. The prolonged time intervals between measles and rehospitalizations suggest that these were not immediate secondary complications due to measles but unrelated, nonspecific events. Individual diagnoses leading to rehospitalization were mainly respiratory tract infections (10 of 12), as is typically the case in children in this part of the world, and none of them were particularly severe or fatal.
Measles in the developing world is considerably different from that in the developed world.1,3,4,10,13–16 In the developed world, deaths and complications are less common than in the developing world. Therefore, one might expect that the beneficial effects of measles immunization on preventing post measles immune amnesia might be different in the developing and developed worlds. We, in this cohort study and also the results in a retrospective population based study in England,11 suggest that post measles immune amnesia is also important today in the developed world where known risk factors (malnutrition and vitamin A deficiency) are not common.
In the DRC, where the majority of children under 5 years of age are malnourished, Ashbaugh et al10 found that children with past measles infection had a higher likelihood (with odds of 1.80) for fever in the 2 weeks preceding interviews with mothers than children without a history of measles. For years 1 and 2 following measles, the odds were 2.08 and 2.14. These findings also indicate an increased risk of infectious diseases following measles for at least 2 years duration. In an analysis of a general practice database in England, Gadroen et al11 tested the hypothesis that measles infection would increase the incidence of infectious diseases other than measles over a prolonged period of time. In their population-based cohort study, 2228 children, 1–15 years of age, had measles (mainly based on a clinical diagnosis) between 1990 and 2014. These cases were matched on age, sex, general practitioner practice and calendar year with 19,930 control children without a history of measles. The investigators found incidence rate ratios (IRR) for non-measles infectious diseases which were significantly increased up to 5 years after measles with IRR values decreasing from 1.43 (95% CI: 1.22–1.68) in the first month to 1.15 (95% CI: 1.06–1.25) 2.5–5 years after measles. Interestingly, IRR of hospitalization was increased only in the month following measles but not thereafter (IRR: 2.83; 95% CI: 1.72–4.67).
Mina et al6 analyzed non-measles infectious disease mortality time sequence data collected during periods when measles was common in England, Wales and Denmark in children 1–9 years of age and in the United States in children 1–14 years of age in comparison to periods after introduction of universal measles immunization programs in the respective countries. Indeed, measles incidence was significantly associated with mortality in all 4 countries. By use of a gamma-distributed transformation program applied to the data from England and Wales, the best-fit for duration of measles virus induced immunomodulation leading to mortality was 28.3 months.
The fact that the increased risk in our study (which was restricted to hospitalizations due to infectious diseases) was only apparent in year 1 whereas the findings from the United Kingdom and the DRC demonstrated a clinical impact of impaired immune functions beyond 1 year following measles can be explained by a gradual recovery of the immune system. The immune system’s recovery may prevent infectious diseases severe enough to require hospitalization already after 1 year whereas recovery may take up to 5 years to prevent those that are less severe and can be managed in an ambulatory setting.
Our study has several strengths. First, in contrast to several other investigations, most cases were laboratory confirmed measles and the time point of the disease could be determined exactly. Second, the design of the study with preset criteria for inclusion of cases and controls makes selection bias unlikely although unknown confounders cannot be ruled out completely. Third, including hospitalizations due to noninfectious diseases allowed us to control for different hospitalization habits which could be different between cases and controls. Apparently, this was not the case as the RRs were similar in cases and controls. Fourth, restricting our analysis to infectious diseases that led to hospitalizations make our findings of an increased risk clinically more significant than including any presumed infectious disease or nonspecific fever episodes.
Our study also has some limitations. As some of our cases were not laboratory confirmed measles, misclassifications are possible. However, if this was the case, the true increased risk after measles might have been even higher than what we calculated. Further, since we did not contact cases and controls, some of them might have moved outside the hospital’s catchment area during the 3 year observational period. In this case, rehospitalizations could have been missed. However, this uncertainty applies to cases and controls alike and there is no plausible reason to assume that it would be of different magnitude in the 1 group or the other. Finally, due to the fact that the study was carried out in Switzerland and restricted to hospitalized cases (for reasons mentioned in the Methods), the overall rate of rehospitalizations was low so that confidence intervals of some of our findings were wide. Yet, the main finding of an increased risk for rehospitalization for infectious diseases was significant in year 1 after measles. Further studies in high-income countries with measles epidemics should be carried out to confirm our findings.
Statistical advice by Dr. Jeffrey Gornbein, Department of Medicine Statistical Unit, University of California, Los Angeles, is gratefully acknowledged. The following staff from study sites assisted in identification of clinical records of cases and controls: Sara Bernhard-Stirnemann, Aarau; Jenny Breitschmid and Silke Schwarzelühr, Basel; Patrizio Baldi, Bellinzona; Olga Endrich, Bern; Fadri Bisatz, Chur; Mario Petrini, Lucerne; Christine Gasser, Doris Dürr and Siegmund Grosse-Honebrinck, St. Gallen.
APPENDIX: THE SWISS MEASLES IMMUNE AMNESIA STUDY GROUP
Christoph Aebi, Department of Pediatrics, Inselspital, Bern University Hospital, University of Bern, Switzerland; Christoph Berger, Division of Infectious Diseases and Hospital Epidemiology and Children’s Research Center, University Children’s Hospital Zurich, Switzerland; Sara Bernhard-Stirnemann, Cantonal Hospital Aarau, Children’s Hospital, Aarau, Switzerland; Pierre Alex Crisinel, Woman-Mother-Child Department, Lausanne University Hospital and University of Lausanne, Lausanne, Switzerland; Alex Donas, Department of Pediatrics, Children’s Hospital Lucerne, Lucerne, Switzerland; Lisa Kottanattu, Department of Pediatrics, Regional Hospital Bellinzona, Bellinzona, Switzerland; Anita Niederer-Loher, Children’s Hospital of Eastern Switzerland, St. Gallen, Switzerland; Christian Mann, Department of Pediatrics, Cantonal Hospital Graubünden, Chur, Switzerland; Noémie Wagner, Department of Pediatrics, Geneva University Hospital, Geneva, Switzerland; Franziska Zucol, Department of Pediatrics, Cantonal Hospital Winterthur, Winterthur, Switzerland.
1. Cherry JD, Lugo D. Cherry JD, Demmler-Harrison G, Kaplan S, et al. Measles
virus. In: Feigin and Cherry’s Textbook of Pediatric Infectious Diseases
. 2018:8th ed. Philadelphia: Saunders/Elsevier; 1754–1770.
2. Wendorf KA, Winter K, Zipprich J, et al. Subacute sclerosing panencephalitis: the devastating measles
complication that might be more common than previously estimated. Clin Infect Dis. 2017;65:226–232.
3. Aaby P, Garly ML, Balé C, et al. Survival of previously measles
-vaccinated and measles
-unvaccinated children in an emergency situation: an unplanned study. Pediatr Infect Dis J. 2003;22:798–805.
4. Aaby P, Samb B, Simondon F, et al. Non-specific beneficial effect of measles
immunisation: analysis of mortality studies from developing countries. BMJ. 1995;311:481–485.
5. Starr S, Berkovich S. Effects of measles
, gamma-globulin-modified measles
and vaccine measles
on the tuberculin test. N Engl J Med. 1964;270:386–391.
6. Mina MJ, Metcalf CJ, de Swart RL, et al. Long-term measles
-induced immunomodulation increases overall childhood infectious disease mortality. Science. 2015;348:694–699.
7. Mina MJ, Kula T, Leng Y, et al. Measles
virus infection diminishes preexisting antibodies that offer protection from other pathogens. Science. 2019;366:599–606.
8. Petrova VN, Sawatsky B, Han AX, et al. Incomplete genetic reconstitution of B cell pools contributes to prolonged immunosuppression after measles
. Sci Immunol. 2019;;4:eaay6125.
9. Wesemann DR. Game of clones: how measles
remodels the B cell landscape. Sci Immunol. 2019;4:eaaz4195.
10. Ashbaugh HR, Cherry JD, Hoff NA, et al. Association of previous measles
infection with markers of acute infectious disease among 9- to 59-month-old children in the Democratic Republic of the Congo. J Pediatric Infect Dis Soc. 2018;8:531–538.
11. Gadroen K, Dodd CN, Masclee GMC, et al. Impact and longevity of measles
-associated immune suppression: a matched cohort study using data from the THIN general practice database in the UK. BMJ Open. 2018;8:e021465.
12. Richard JL, Masserey Spicher V. Large measles
epidemic in Switzerland from 2006 to 2009: consequences for the elimination of measles
in Europe. Euro Surveill. 2009;14:pii=19443Available from: https://www.eurosurveillance.org/content/10.2807/ese.14.50.19443-e
13. Hussey GD, Klein M. A randomized, controlled trial of vitamin A in children with severe measles
. N Engl J Med. 1990;323:160–164.
14. Barclay AJ, Foster A, Sommer A. Vitamin A supplements and mortality related to measles
: a randomised clinical trial. Br Med J (Clin Res Ed). 1987;294:294–296.
15. Sommer A, Tarwotjo I, Djunaedi E, et al. Impact of vitamin A supplementation on childhood mortality. A randomised controlled community trial. Lancet. 1986;1:1169–1173.
16. Paules CI, Marston HD, Fauci AS. Measles
in 2019 - going backward. N Engl J Med. 2019;380:2185–2187.