Is pertussis in infants a potential cause of retinal haemorrhages?
- 1Emergency Department, Sant Joan de Deu Hospital, Barcelona, Spain
- 2Ophthalmologic Department, Sant Joan de Deu Hospital, Barcelona, Spain
- 3Pediatrc Department, Sant Joan de Deu Hospital, Barcelona, Spain
- Correspondence to Dr Ana Isabel Curcoy, Emergency Department, Sant Joan de Deu Hospital, passeig Sant Joan de deu 2 08950, Barcelona, Spain;
- Received 4 July 2011
- Accepted 18 November 2011
- Published Online First 18 December 2011
Aim To determine the prevalence of retinal haemorrhages in infants with pertussis infection with the purpose of clarifying the differential diagnosis of the cases of abusive head trauma.
Methods Prospective study of children aged 15 days to 2 years admitted to our hospital with a diagnosis of pertussis over a period of 4 years (May 2004–May 2008). All children underwent one detailed ophthalmological examination within 72 h of admission. If retinal haemorrhages were detected, further investigation was undertaken to rule out systemic disorder or maltreatment.
Results 35 children with pertussis infection were examined. None was found to have retinal haemorrhages. Therefore, applying Wilson's method, the data suggest with 95% confidence that the true effect estimate for retinal haemorrhage occurring due to symptomatic pertussis infection requiring admission to hospital is no higher than 9.9%.
Conclusions Pertussis infections are unlikely to cause retinal haemorrhages in children under 2 years of age.
Retinal haemorrhages are among the most common findings in abusive head trauma (AHT) but these may be associated with a variety of systemic illnesses as well as accidental head trauma.1
In 2006, Geddes and Talbert2 suggested that retinal haemorrhages commonly attributed to ‘shaking’ can result from paroxysmal cough injury. They based their hypothesis on a computer modelling approach. This model seems to demonstrate that coughing causes intracranial pressure to build up exponentially, approaching a maximum that is proportional to the amount of pressure the musculature of the thorax can produce, as venous return is impeded. This circumstance can be dangerous during paroxysmal coughing because it does not allow for the re-establishment of normal pressure between coughs, with the result that very high-luminal pressure may be generated, sufficient to damage veins. However, recently, Goldman et al3 examined 100 infants with severe cough and none demonstrated retinal haemorrhages. They estimate that the chance of retinal haemorrhages occurring as a result of severe cough is at most 0.03. A limitation of that study, which may account for the low proportion found, is that the coughing episodes included were not powerful enough to produce retinal haemorrhages. That study included many cough aetiologies. In our opinion, the condition that is more appropriate to either prove or reject the Geddes and Talbert model2 is pertussis alone. Pertussis typically has a paroxysmal phase characterised by severe and repetitive coughing spells, followed by an inspiratory whoop and post-tussive vomiting.
These cough characteristics make easier the assumption that pertussis is associated more strongly than other forms of cough with venous and arterial hypertension, hypoxia and apnoea,4 and these factors would be expected to increase the risk of retinal haemorrhage. But in the Goldman3 study only one patient had pertussis.
No reports of retinal haemorrhages induced by pertussis in infants have been published. Therefore, the objective of this prospective study was to determine the prevalence of retinal haemorrhages in pertussis with the purpose of clarifying the differential diagnosis of AHT cases.
This was a prospective study of children aged 15 days to 2 years admitted to our hospital with a diagnosis of pertussis over a period of 4 years (May 2004–May 2008). All children were examined by an experienced ophthalmologist using indirect ophthalmoscopy within 72 h of admission. The pupils of the patients were dilated with short-acting mydriatic agents.
A detailed history was taken for each child to investigate the mode of delivery (vaginal delivery; caesarean section, forceps delivery and spatula delivery) and pertussis.
Pertussis was defined as a paroxysmal whooping cough and at least one of the following: a positive nasopharyngeal culture, PCR or a positive immunoglobulin M antibody for Bordetella pertussis.5 All patients underwent a nasopharyngeal swab on admission.
Our a priori protocol required that if retinal haemorrhages were detected, then further investigation would be undertaken to rule out systemic disorder or maltreatment, following our hospital's protocol. The protocol includes at least one head CT and/or MRI to rule out subdural and/or subarachnoid haemorrhage, a skeletal survey and/or bone scintigraphy to rule out fractures, and a full blood count and coagulation study to rule out blood dyscrasias. Patients suspected of having been abused are normally evaluated by a multidisciplinary team composed of paediatricians, psychologists, neurosurgeons, ophthalmologists and social workers. Their conclusions are forwarded to the local government's management agency for children and adolescents.
Data analysis was conducted using SPSS 12.0 software. The study population was described using frequencies and percentages for categorical variables, and means and SD or medians and ranges for continuous variables. As we assumed that pertussis retinal haemorrhage is rare, we used the binomial test to estimate the statistical probability of such an event. If none of ‘n’ patients showed retinal haemorrhages, statistical analysis was undertaken using Wilson's method.6
From May 1, 2004 to April 30, 2008, 35 patients were admitted to our hospital with pertussis infection; 18 (51%) were boys and 17 (48%) were girls. Their ages ranged from 21 to 130 days with a median age of 54 days. Their mode of delivery was as follows: normal vaginal delivery, 14; spatula delivery, 9; caesarean section, 6; forceps delivery, 6. The duration of coughing before admission ranged from 2 to 21 days, with a median of 7 days.
No patient was found to have retinal haemorrhages. Therefore, using Wilson method, we can be confident to an upper limit of 95% that the chance of retinal haemorrhages occurring as a result of pertussis whooping cough is at the most 0 .099.
The aim of this study was to ascertain whether pertussis alone could produce retinal haemorrhages and we did not find retinal haemorrhages in any of the 35 enrolled cases. Therefore these results show that only on scant occasions (prevalence lower than 0.1) is a symptomatic pertussis infection requiring admission to hospital associated with retinal haemorrhages. These findings support the conclusion of Goldman et al3 that if one finds retinal haemorrhage in infants and young children with cough, it is unlikely that coughing is a sufficient cause for any identified retinal haemorrhage.
By contrast, the finding of retinal haemorrhages in children with non-accidental head trauma is quite common, occurring in as many as 75% of cases.7 Given the low likelihood of cough producing retinal haemorrhages and the much greater probability that such a haemorrhage is the result of maltreatment, we disagree with Talbert et al: we feel what needs to be demonstrated in such cases is that there was not maltreatment. Talbert et al4 explain that the irritation caused by the powerful coughing paroxysms of whooping cough can produce retinal haemorrhages and concludes in their article that additional objective evidence of inflicted trauma is necessary to distinguish between abusive head trauma and paroxysmal cough injury. We have tried to show that additional objective evidence of inflicted trauma is not necessary.
The main limitation of our study is the small size of the sample, given the low prevalence of whooping cough due to the systematic vaccination of the population. On the other hand, we also have to exclude the children under 15 days of age due to the difficulty in the differential diagnosis of retinal haemorrhages associated with delivery.8
In conclusion, in our study we did not detect retinal haemorrhages in patients with pertussis infection. Therefore, if retinal haemorrhages are detected, investigation into the possibility of non-accidental injury is essential to safeguard the patient and his or her siblings.
Competing interests None.
Ethics approval This study was approved by the Sant Joan de Déu Hospital Committee.
Provenance and peer review Not commissioned; externally peer reviewed.