Medical Student Elective
Arjun Chandna
Georgetown & Lethem Public Hospitals, Guyana
January – February 2011
Background to the elective
I recently spent six weeks working in Guyana on my medical elective. I chose to
visit Guyana for many reasons. I was keen to visit South America and after
searching the WHO website, a few countries disease profiles interested me, one of
which was Guyana. Whilst I can converse in rudimentary Spanish, I am not able to
take a medical history and I felt I would gain most from my elective if I undertook
it in a predominantly English speaking country.
Before setting off I had very little idea as to what to expect. Guyana is visited by
few tourists so finding information about the country, from the Internet or the one
available guidebook, was difficult. A significant part of the disease burden is
infectious diseases, particularly HIV/AIDS. I was looking forward to learning
about how these present, particularly their interaction with pregnancy. In
addition, many conditions that are common in the UK are also prevalent in
Guyana, but present much later. I was looking forward to learning about the
advanced presentations of common conditions and the florid clinical signs that
often accompany this.
I spent half my time in Georgetown, the capital of Guyana, working in the public
hospital, which is the major referral centre for the country. I was attached to the
obstetric and gynaecology department and was able to witness first-hand the
challenges facing women’s healthcare. A striking difference between the UK and
Guyana was that whilst in the UK pregnancy is a healthy period of a woman’s life,
in Guyana the strains of pregnancy (both physical and financial) often accentuate
underlying health problems, for example, malnutrition, hypertension, filariasis
and diabetes.
In Lethem, I spent time working with an NGO, Remote Area Medical, who
provide outreach medical services to the people living in the Guyanese savannahs.
This work focuses largely on sexual health education, HIV-testing, basic antenatal
care, eye screening and diabetes and hypertension monitoring.
In Georgetown I was also able to carry out my intended elective research project:
investigating neurological dysfunction in pre-eclampsia using a technique called
saccadometry. I present my results below. My two current career interests are
neurology and obstetrics. This project enabled me to experience the practise of
these disciplines in a new and challenging environment.
Overall my elective was a wonderful experience, which has fuelled my desire to
spend a significant amount of my professional career working abroad.
Word Count: 2731 (excluding figures and references)
A
AB
BSST
TR
RA
AC
CT
T
Introduction
Pre-eclampsia and the consequences thereof remain a leading cause of maternal
death particularly in the developing world. Neurological dysfunction, from
hyperreflexia to eclamptic seizures, is an important yet poorly characterised part of
the disease phenotype. Saccadic reaction time is known to be an excellent
biomarker for general neurological function. In addition, parallels between cortical
control of saccadic reaction time and spinal stretch reflexes (often deranged in preeclampsia) exist. For this reason, we investigated whether neurological
dysfunction in pre-eclampsia is amenable to detection using saccadometry.
Methods
A head-mounted infrared oculometer was used to record 900 saccadic latency
trials from women pre- and post-delivery in the obstetric unit of Georgetown
Public Hospital, Guyana. The pre- and post-delivery reaction times were
compared for each woman.
Results
Reaction times pre- and post-delivery were significantly different in all women
recorded from. In 75% (3/4) of cases the whole distribution was significantly
different (p < 0.05) and in the remaining 25% (1/4) the median latency was
significantly different (p < 0.05).
Insufficient data was collected to compare normal pregnancy and pre-eclamptic
distributions.
Discussion
This study demonstrates altered neurological functioning pre- and post-delivery in
normal pregnancies. This alteration may be due to the delivery process or the
pregnant state itself (with the change post-delivery representing a return to
neurological baseline). Due to problems with data collection it is not yet possible to
answer the primary question of this study as to whether neurological dysfunction
in pre-eclampsia is amenable to detection using saccadometry.
IIN
NT
TR
RO
OD
DU
UC
CT
TIIO
ON
N
Pre-eclampsia occurs in 3-14% of pregnancies (Irminger-Finger, et al., 2008) and is
the fourth leading cause of maternal death, accounting for 12% of maternal
mortality worldwide (WHO, 2005). Improving maternal health was one of eight
Millennium Development Goals committed to by countries at the United Nations
Millennium Summit.
Despite its serious consequences, the pathophysiology of pre-eclampsia is still
incompletely understood and hence prophylactic treatment is difficult to provide.
Current management centres upon detection of cardinal signs (pregnancy-induced
hypertension and proteinuria) and control of blood pressure until delivery of fetus
and placenta is appropriate.
Pre-eclampsia is characterised by pregnancy-induced hypertension, proteinuria
and peripheral and central oedema – the first two characteristics being most
discriminative in clinical diagnosis. In addition pre-eclamptic women often have
abnormal neurology, most commonly hyperreflexia.
Work by Sherrington (1924) illustrated the central role of the cerebral cortex in
controlling the gain of spinal stretch reflexes (Liddell and Sherrington, 1924).
Given that eclampsia, a much-feared sequelae to pre-eclampsia, is a manifestation
of severe cerebral dysfunction, it is reasonable to predict that hyperreflexia
exhibited by pre-eclamptic women may be of cerebral origin. In fact, a
retrospective study from 2008 found that 16 of 22 eclamptic seizures were
preceded by hyperreflexia (Boudaya, et al., 2008). Techniques that quantitatively
measure neurological function may therefore be useful in detecting early changes
in pre-eclampsia. Unfortunately it is not easy to get robust, quantitative measures
of hyperreflexia.
In recent years it has become apparent that an excellent quantitative biomarker for
general neurological function can be provided by saccadic reaction time (RT). The
saccadic system is a microcosm of the brain itself (Carpenter, 2004) and hence
subtle changes in neurological function are readily detected by studying RT
distributions. This is due in part to the fact that neural networks spanning wide
areas of the cortex are implicated in the control of saccades. In particular, the
saccadic machinery in the superior colliculus is subject to tonic cortical inhibition
in a manner not dissimilar to cortical inhibition of spinal reflexes (Hikosaka and
Wurtz, 1983).
RT studies have been used to investigate Huntington’s disease and Parkinson’s
disease with great success (Antoniades, et al., 2007; Michell, et al., 2006). In
addition, ongoing work investigating a potential role in traumatic brain injury
(Pearson et al., 2007) is showing promising results.
There are also many practical advantages of using RT. Measurement is noninvasive with a portable, infrared oculometer and as we make three saccades per
second, large amounts of data can be collected without the risk of fatigue that
manual response tasks are exposed to. Our detailed understanding of the saccadic
system has facilitated comprehensive theoretical modelling, which in turn allows
putative mechanistic conclusions to be drawn from observed changes in reaction
time. One such model, LATER (Linear Approach to Threshold with Ergodic Rate),
is discussed in this study (Carpenter and Williams, 1995). Finally, recording RT is
relatively cheap compared with other techniques used to assess neurological
function and thus it is an attractive option for developing countries where the
burden of pre-eclampsia is largest (WHO, 2005).
This is the first study that aims to investigate neurological dysfunction in preeclamptic women in a manner permitting quantitative analysis.
M
MEET
TH
HO
OD
DSS
Participant recruitment
Pre-eclampsia sufferers were recruited from Georgetown Public Hospital (GPH),
Guyana in January 2011. Participants were asked to give informed consent. All
local ethical procedures were adhered to. The number of cases in the study period
determined the sample size.
Diagnosis of pre-eclampsia was according to the following criteria:
1. Pregnancy-induced hypertension
2. Dipstick-positive proteinuria
Participants were excluded if they were:
1. Aged < 16 years or > 40 years
2. Suffering with a medical disorder other than pre-eclampsia
3. Using medication (other than vitamin supplements) during pregnancy
Study design
We performed a longitudinal study comparing RT distributions pre- and postdelivery in pre-eclampsia sufferers. We also carried out identical recordings (preand post-delivery) in age- and gestation-matched normal pregnancy controls.
Data Collection
Figure 1. A saccadometer in situ. The blue nose-piece is adjustable to ensure that infra-red
emitters and detectors are aligned with the participant’s sclera. The three lasers are visible at
the top of the apparatus.
A miniaturised, head-mounted, infrared reflection oculometer (12 bit resolution,
sampling rate 1kHz, low-pass filtered at 250Hz, signal-to-noise ratio 45dB) was
used to measure 100 horizontal saccades. The device is comfortable and requires
no head restraint since the target display moves with the head. The oculometer has
three low-power red lasers that projected high contrast 13 cd.m2 target dots onto a
light-coloured background, subtending 0.1°, in a horizontal line at ± 10° to the
midline; to a first approximation these angles are independent of the distance
between subject and background. Participants sat 1m from a blank, non-reflective
screen.
Each trial began with the presentation of a central target, which, after a random
foreperiod (1-2s), disappeared at the same time as a target to either the left or right
(chosen at random to prevent anticipation) appeared. The target remained for
200ms after a resultant correct saccade, or for 1s (with an incorrect or absent
response), whichever was shorter. Participants were instructed to track the
movement of the target with their eyes as quickly as possible without
compromising accuracy. The device is self-calibrating, using five preliminary trials
to each side. The same procedure was used for controls to minimise bias. A further
100 trials were recorded after delivery.
SACCADE
NEXT TRIAL
TIME
RESPONSE
LATENCY
FIXATION
FORE-PERIOD
Figure 2. Schematic illustration of the experimental protocol. Participants were required to
fixate a central target (fixation indicated by a dashed circle, target by red spot). After a random
fore-period (1-2s) the spot would move 10° horizontally, either left or right (in this case to the
right). The participant was required to look at (make a saccade to) the new target location. The
time between target onset and response onset was recorded as the saccadic latency (RT).
The next trial began with the target returning to the central location.
Difficulties with data collection
At the time of data collection a government investigation into the maternal death
rate at GPH made recording from women, particularly sick women (those with
pre-eclampsia) difficult. Whilst it was possible to collect data from normal
pregnancies (800 saccades from 4 women) only 1 woman (100 saccades) with preeclampsia was recorded from pre-delivery.
Analysis
The software application LatencyMeter automatically eliminated trials
contaminated by blinks, head movements and inattention: computerising this
process removed observer bias. The data were then exported to the application
SPIC, that analysed the data, provided estimates of the underlying LATER
Reciprocal Scale
A
Percentage
Percentage
parameters, and generated reciprobit plots.
B
Latency (ms)
ProbitScale
C
Cumulative Percent
Probability
Cumulative Percentage
Latency (ms)
Latency (ms)
D
Latency (ms)
Figure 3. Schematic showing the graphical manipulation from a frequency-histogram (A) to
a reciprobit plot (D). A) A frequency-histogram showing the positively skewed distribution of
saccadic latency. B) Using a reciprocal scale on the abscissa indicates that the reciprocal of
latency may be normally distributed. C) This is further supported by a sigmoidal curve
demonstrated on a cumulative histogram, D) and finally verified by the straight line when a
probit scale is used on the ordinate axis.
Figures adapted from www.cudos.ac.uk/LATER.html
The two fundamental parameters (see Discussion for interpretation) are the
median, µ , and standard deviation, σ , of the main distribution of reciprocal
latencies. For some, but not all, individuals a small sub-population of early
saccades was seen, distinct from the main distribution, which can be characterised
by a third parameter, its standard deviation, σE.
Pre- and post-delivery distributions were compared using a Kolmogorov-Smirnov
test. In addition the median of each distribution was compared using a Student’s
paired t-test.
R
REESSU
ULLT
TSS
Kolmogorov-Smirnov one-sample tests showed that for all participants, observed
distributions conformed (p > 0.05) to the recinormal distribution predicted by
LATER, justifying the description of all distributions by means of the three LATER
parameters, µ, σ and σE.
Patient
Age
Gestation Parity Delivery Anaesthetic Complication Time Post-delivery Medical History
1
22
40+1
0
NVD
Entonox
-
6h
-
2
24
38+2
1
NVD
Entonox
PPH = 550ml
30h
-
3
25
37+6
1
NVD
Entonox
-
17h
-
4
28
39
2
NVD
Entonox
-
22h
-
5
27
37
1
N/A
N/A
N/A
N/A
Previous PIH
Table 1. Tabulated data for the 5 participants. Patients 1 to 4 were normal pregnancies recorded
from pre- and post-delivery. Their age, gestation and parity are recorded along with delivery
method, anaesthetic, complications and the time the post-delivery recording was carried out.
Patient 5 had pre-eclampsia and was only recorded from pre-delivery.
Four normal pregnancy patients were recorded from. In each case the median,
whole distribution or both were significantly different (p < 0.05) pre- and postdelivery. Their distributions are shown below.
Figure 4. Reciprobit plot for patient 1
illustrating significant difference
between RT distributions pre- (red)
and post-delivery (blue), p < 0.001.
The median latency is also
significantly different between the
two groups, p = 0.010.
Latency (ms)
Figure 5. Reciprobit plot for patient 2
illustrating significant difference
between RT distributions pre- (red)
and post-delivery (blue), p = 0.007.
The median latency is not
significantly different between the
two groups, p = 0.808.
Latency (ms)
Figure 6. Reciprobit plot for patient 3
illustrating significant difference
between RT distributions pre- (red)
and post-delivery (blue), p = 0.046.
The median latency is not
significantly different between the
two groups, p = 0.275.
Latency (ms)
Figure 7. Reciprobit plot for patient 4
illustrating no significant difference
between RT distributions pre- (red)
and post-delivery (blue), p = 0.088.
The median latency is significantly
different between the two groups, p =
0.018.
Latency (ms)
D
DIISSC
CU
USSSSIIO
ON
N
Due to the difficulties collecting data it has not been possible to answer the
primary question of this study: to determine whether neurological dysfunction
associated with pre-eclampsia is amenable to detection using saccadometry.
However, the results suggest, somewhat unexpectedly, that the delivery process or
the pregnant state itself may affect neurological function. Given the preliminary
nature of the results, discussion at this stage is tentative, however a number of
observations can be made.
Before further discussion it is useful to have a framework within which these
results can be interpreted. As previously noted, RT is an excellent biomarker for
general neurological function. In fact, RT is equivalent to decision time (Carpenter,
2004) and hence decision-making models have been used to characterise RT
distributions with much success. The simplest of these models (Nakahara, et al.,
2006), LATER, will be used in this discussion.
The LATER Model
µ
ST
σ
θ
S0
Response
Stimulus
Figure 8. The LATER Model. The initial decision signal (S0) represents a prior probability that
a particular hypothesis is true. Accumulation of sensory information updates this probability, at
rate r, and the decision signal rises or falls depending on whether the information is supportive
of the initial hypothesis. A decision (for our purposes, a saccade) is made once the signal
reaches a threshold level (ST).
LATER proposes a decision signal rising at rate r. Trial-by-trial this rate varies
about a median (µ ) and standard deviation (σ). Different conditions can alter the
shape of the decision profile. Reducing θ, either by increasing S0 (altering the prior
probability of a certain hypothesis (Carpenter and Williams, 1995)), or lowering ST
(placing increased importance on the urgency of response (Reddi and Carpenter,
2000)), causes clockwise swivel about the infinite time axis on the reciprobit graph.
Providing more supportive information increases r and, whilst maintaining the
slope, causes a parallel leftward shift (Reddi, et al., 2003).
LATER also offers an explanation for the subpopulation of early responses seen in
saccades, characterised by their distinct σE. Their latencies are consistent with a
putative subcortical pathway through the superior colliculus. It could be that a
second collicular LATER unit is usually suppressed by cortical inhibition, but
occasionally it reaches threshold first, triggering an early saccade. This hypothesis
is supported by the fact that a distracting task increases the proportion of early
responses (Halliday and Carpenter, in submission).
Alterations in reaction time distributions
All four patients have significantly different RT pre- and post-delivery. The
median, whole distribution or both are significantly different pre- and postdelivery in the same individual. In 75% (3/4) of cases the whole distribution is
significantly different. The remaining case, patient 4 (figure 7), has a significantly
different median RT pre- and post-delivery. In addition, they demonstrate a clear
shift in their whole distribution. It is therefore likely that with more trials (for
example, 200 saccades pre- and post-delivery) the difference in the whole
distributions may also achieve significance.
This finding is important because whilst RT distributions are idiosyncratic, they
are remarkably consistent within an individual. Hence, the finding suggests an
alteration in neurological functioning pre- and post-delivery.
At this stage it is difficult to comment on how or why the distributions are
changing. Whilst 200 trials per patient allow effective analyses of changes within
an individual, it is difficult to comment on population trends with a sample of 4
patients. Within our cohort, different patients demonstrate swivel and shift and
median RT sometimes increases and sometimes decreases post-delivery. With a
larger sample, trends in swivel vs. shift or the different parameters may become
apparent, allowing putative mechanistic interpretation in the context of LATER or
other decision-making models.
Further work
Meaningful further work will require the collection of more data. As previously
stated, this may reveal how RT distributions are changing and in turn, decisionmaking models would allow speculation as to why. At birth there are a number of
changes occurring (hormonal, psychological, physical) and any or all of these
could contribute to the altered neurological function that has been suggested. It
will also be important to examine how the other parameters, σ and σE, are affected
in order to fully characterise how the distributions are changing.
Further work should be carried out with a refined methodology. Firstly, more
trials should be recorded at each session to increase the statistical power of the
analyses. Secondly, post-delivery recording time should be standardised. It is
important to consider lingering effects of anaesthesia on the brain and ensure that
recording is undertaken after these have worn off. One explanation for the slightly
odd shape to the RT distribution of patient 1 is that the post-delivery recording
was at 6 hours. Finally, consideration should be given to carrying out a control
study pre- and post-delivery by caesarean section. Most pre-eclamptic deliveries
are likely to be by caesarean and hence it will be important to have a baseline to
compare these results to.
This study has revealed that delivery or pregnancy itself may alter neurological
function. Hence, recordings in pre-eclamptic patients will not be straightforward
to interpret: potential changes in the pre-eclamptic population must be compared
to changes in normal deliveries. It may be particularly interesting to investigate
any change in early saccades (σE). The cortex normally suppresses these but given
the impaired suppression of spinal reflexes in pre-eclampsia (evidenced by
hyperreflexia) there may be salient changes in this parameter.
Finally, given that pregnancy may affect neurological function, it may be useful to
characterise this more thoroughly. One method for doing this could be to perform
saccadometry regularly throughout pregnancy, ideally starting prior to
conception.
Summary
This study has been unable to answer the primary question: is neurological
dysfunction in pre-eclampsia amenable to detection using saccadometry.
However, it has demonstrated, albeit preliminarily, that the delivery process or the
pregnant state itself may be associated with altered neurological function. There is
much scope for further work in this field.
R
REEFFEER
REEN
NC
CEESS
1.
Irminger-Finger, I., 2008. Pre-eclampsia: a danger growing in disguise. Int J
Biochem Cell Biol., 40(10),1979-83
2.
World Health Organisation, 2005. The world health report 2005 – Make every
mother and child count, Geneva: World Health Organization
3.
Liddell, E., 1924. Reflexes in Response to Stretch (Myotatic Reflexes.
Proceedings of the Royal Society of London. Series B, Containing Papers of a
Biological Character, Vol. 96, No. 675, pp. 212-242
4.
Boudaya, F., 2008. Eclampsia: epidemiological aspects and management of
28 patients. Tunis Med., Jul;86(7):685-8
5.
Carpenter, RHS., 2004. The saccadic system: a neurological microcosm.
Advances in Clinical Neuroscience and Rehabilitation, 4:6-8
6.
Hikosaka, O., 1983. Visual and Oculomotor Functions of Monkey Substantia
Nigra Pars Reticulata. I. Relation of Visual and Auditory Responses to
Saccades. Journal of Neurophysiology, Vol. 49, No. 5
7.
Antoniades, CA., 2007. Saccadometry: a new tool for evaluating presymptomatic Huntington patients. Neuroreport, 18:1133-6
8.
Michell, AW., 2006. Saccadic latency distributions in Parkinson's disease
and the effects of L-dopa. Experimental Brain Research, 169:237-45
9.
Pearson, BC., 2007. Saccadometry: the possible application of latency
distribution measurement for monitoring concussion. British Journal of
Sports Medicine. 41:610-2
10.
Carpenter, RHS., 1995. Neural computation of log likelihood in the control
of saccadic eye movements. Nature, 377:59-62
11.
Nakahara, H., 2006. Extended LATER model can account for trial-by-trial
variability of both pre- and post-processes. Neural Networks, 19:1027-1046
12.
Reddi, B.A.J., 2000. The influence of urgency on decision time. Nature
Neuroscience, 3, 827-831
13.
Reddi, B.A.J., 2003. Accuracy, Information, and Response Time in a Saccadic
Decision Task. Journal of Neurophysiology, 90, 3538-3546
14.
Halliday, J., In submission. Audio-verbal distraction leading to over-fast,
inaccurate saccadic responses: implications for driving and a possible
neural mechanism.
This project would not have been possible without the support of the Royal College of
Obstetricians & Gynaecologists, UCL Medical School, Amulree Bursary, the Royal Free Association,
the Wellcome Trust, the Wellbeing of Women and the British Dental & Medical Students’ Trust.