Causes
The blood vessels of the retina begin to develop about 3 months into pregnancy. They complete development at the time of normal birth. The eyes may not develop properly if a baby is born very early. The vessels may stop growing or grow abnormally from the retina into the back of the eye. The vessels are fragile. They can leak and cause bleeding in the eye.
Scar tissue may develop and pull the retina loose from the inner surface of the eye. In severe cases, this can result in vision loss.
In the past, the use of too much oxygen in treating premature babies caused vessels to grow abnormally. Better methods are now available for monitoring oxygen, so this problem is rare.
Today, the risk of developing ROP depends on the degree of prematurity. Smaller babies with more medical problems are at higher risk.
Almost all babies who are born before 30 weeks or weigh fewer than 3 pounds at birth are screened for the condition. Some high-risk babies who weigh 3 - 4.5 pounds or who are born after 30 weeks should also be screened.

See link for full article:
https://www.nlm.nih.gov/medlineplus/ency/article/001618.htm

Abstract

PURPOSE:To present the occasional observation of retinal haemorrhages in premature babies, as a diagnostic alternative to those observed as part of shaken baby syndrome.

METHODS:

We carried out an observational study on 11 infants in whom retinal and/or vitreous haemorrhages had been observed within their first months of life. Ten infants were under surveillance for retinopathy of prematurity (ROP), with gestational ages and birth weights in the ranges of 27--34 weeks and 790--1665 g, respectively. One infant was diagnosed with Zellweger's syndrome and one received substitution therapy for coagulation factor II deficiency. The last child had been delivered at 38 weeks, weighing 2070 g; he died on day 5 from severe brain oedema with incarceration and extensive bilateral fundus bleeding.
RESULTS:Four of the 11 infants had some evidence of ROP, and two later received retinal ablation therapy. Contrary to the quick absorption (<1-2 weeks only) usually seen in most newborn term infants, the ocular bleeding in preterms was generally longstanding. A quick increase in intracranial pressure probably played a role in the lethal case with delivery near term, and one infant received lung physiotherapy for pneumonia at the age of 6 months. Some bleeding appeared to be truly postnatal (i.e. it was observed as a new occurrence during the course of surveillance).
CONCLUSIONS:In the series under study there was no suspicion of child abuse. In term infants, retinal haemorrhages are extremely rare except when due to shaking, but other diseases should be ruled out, coagulopathies in particular. We suggest that prematurity as such is added to the list of possibly underlying causes when retinal bleedings are evaluated in very small infants and shaken baby mechanisms are suspected.



For article link see:
http://www.ncbi.nlm.nih.gov/pubmed/16029264

Examination induced retinal haemorrhage in preterm infants is uncommon. It may occur independently of retinopathy of prematurity (ROP) associated neovascularisation. Retinal haemorrhages due to ROP tend to occur on the surface of the neovascular ridge, which represents an arteriovenous shunt formed by the anastomosis of primitive retinal vessels.1 However, ROP related retinal haemorrhages occurring posteriorly,2 as well as vitreous haemorrhage,3 have also been described in advanced disease. Other common causes of retinal haemorrhage in infancy such as birth4 and shaken baby syndrome should be excluded. Retinal haemorrhage in a premature infant after RetCam photography has been reported.5 We report retinal haemorrhages in a premature infant immediately following examination for ROP.

Please see link for full article:
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1860199/

Abstract 

Spontaneous subarachnoid haemorrhage (SAH) in children is uncommon, but is sometimes seen after rupture of aneurysms, and in different disorders. Traumatic SAH is common after serious accidental head injury, but is also reported after child abuse with vigorous shaking. To avoid unnecessary accusations of innocent care givers, it is important not to misinterpret the findings as abusive head trauma in small children with SAH.

In the presented case, a nearly two-year-old girl was brought to the hospital after a fall witnessed by her father. The girl was unconscious, with elevated intracranial pressure, SAH and bilateral retinal haemorrhage (RH). She was pronounced dead after 9 h. Premortem angiography revealed a dissection of the right vertebral artery, and postmortem examination revealed a traumatic lesion deep in the neck, at the base of the skull. Cerebral edema, in combination with SAH and RH, is highly suggestive of abusive head trauma. However, no external lesions, no skeletal lesions, especially no long bone metaphyseal lesions, or subdural haematomas occurring at the same time as SAH, were found. There was no report of previous child abuse in the family. Based on the radiological and postmortem findings, we believe that an accidental fall caused a blunt force trauma with a subsequent dissection of the right vertebral artery.

To our knowledge, accidental tear of one of the vertebral arteries, leading to SAH in a toddler, has previously not been described. Child abuse is an important exclusion diagnosis with serious legal implications.



http://www.sciencedirect.com/science/article/pii/S0379073812001946

The English-language medical case literature was searched for cases of apparent or alleged child abuse between the years 1969 and 2001. Three-hundred and twenty-four cases that contained detailed individual case information were analyzed yielding 54 cases in which someone was recorded as having admitted, in some fashion, to have shaken the injured baby. Individual case findings were tabulated and analyzed with respect to shaking as being the cause for the injuries reported. For all 54 admittedly-shaken-infant cases, the provided details regarding the shaking incidents and other events are reported. Data in the case reports varied widely with respect to important details. Only 11 cases of admittedly shaken babies showed no sign of cranial impact (apparently free-shaken). This small number of cases does not permit valid statistical analysis or support for many of the commonly stated aspects of the so-called shaken baby syndrome.

Leestma, Jan E. MD, MM

http://journals.lww.com/amjforensicmedicine/Abstract/2005/09000/Case_Analysis_of_Brain_Injured_Admittedly_Shaken.1.aspx

A 4-month-old male infant presented to the emergency room with a history of choking while bottle feeding at home, and was found by emergency medical services (EMS) to be apneic and pulseless. He subsequently developed disseminated intravascular coagulopathy and died. Computed tomography (CT) and magnetic resonance imaging (MRI) showed subdural hemorrhages (SDHs), subarachnoid hemorrhage (SAH), and retinal hemorrhages (RHs), along with findings of hypoxic-ischemic encephalopathy (HIE). The caretaker account appeared to be inconsistent with the clinical and imaging features, and a diagnosis of nonaccidental injury with “shaken baby syndrome” was made. The autopsy revealed diffuse anoxic central nervous system (CNS) changes with marked edema, SAH, and SDH, but no evidence of “CNS trauma.” Although NAI could not be ruled out, the autopsy findings provided further evidence that the child’s injury could result from a dysphagic choking type of acute life threatening event (ALTE) as consistently described by the caretaker.

http://www.sciencedirect.com/science/article/pii/S1071909110000069