Name: Joey Martinus Sidarta

Student ID: 01071180030

Tutor: dr. Jesslyn Natasha (Group A17)

“Association of Sleep Apnea and Type II Diabetes”

1. Description of details:
PLACE: Wisconsin

PERSON:
1387 Participants from the Wisconsin Sleep Cohort with type II diabetes:
o Those with type 1 diabetes were excluded from the original sample
o Average age is 49 years old
o 779 men, 603 women

TIME:
Unspecified study period (most likely around 2013 – 2015)
o Study was conducted overnight
o Longitudinal analysis was conducted throughout a 4 year period

2. Hypothesis: Sleep apnea is associated with Type II Diabetes

3. Type of Association:

I. CHANCE:
 Both sleep apnea and diabetes share similar risk factors, such as obesity, age, alcohol
consumption, heart disorders, family history, metabolic complications, and high blood
pressure. If they were to both occur coincidentally in a patient due to the same risk
factors, it would be hard to identify the temporality (especially in a cross-sectional
study like this).

 “Other possibilities for our findings include some predisposing risk factor, such as a
genetic predisposition, that leads to an increased likelihood of developing diabetes and
SDB in parallel.” If both SDB and diabetes are genetic in nature, then the odds of
developing either of them (or both) are entirely coincidental.

II. BIAS:
 Measurement Bias
o Loss to follow up: “…half of the participants with diabetes had preexisting diabetes
when they enrolled in the cohort and were excluded from the longitudinal analysis of
incident diabetes.” Some participants were shown to have higher insulin resistance than
others, leading to an earlier development of diabetes. Because of this, they were
excluded from the final longitudinal analysis, which could have adverse effects on the
data.

o “…The possibility of a latent period that could extend beyond the duration of our
study.” Certain participants require longer periods of time to develop diabetes, possibly
longer than the prospective follow-up period (4 years). The investigators even
suspected that 8 years might not have been long enough to demonstrate an exposure
effect.
o The investigators weren’t able to differentiate different types of sleep apnea
(obstructive, central, and mixed), as particular types of SBM might have a stronger
association with diabetes.

o “It’s possible that the most “susceptible” subjects had already developed diabetes,
leaving those more “resistant” to the adverse metabolic effects of sleep apnea in the
prospective analysis.” If the subjects analyzed in the longitudinal follow-up had a
higher resistance towards the development of diabetes, then the investigators should
have anticipated and accounted for this resistance factor.

o The investigators used a venipuncture for fasting plasma glucose to measure diabetes
in participants. Fasting glucose includes the measurement of blood sugar levels after an
overnight fast. However, the investigators did not take into account when each subject
had last eaten before their fast as this could cause their blood sugar levels to vary in the
morning, resulting in inaccurate glucose measurements.

 Selection Bias
o “Because our study population was 96% white, it is uncertain whether our findings can
be generalized to other racial groups…” Prevalence of diabetes does differ in other
ethnic groups (African Americans, American Indians, Alaskans, Asian Americans,
Hawaiian, Hispanics and Latinos were found to be the most common), and because the
sample population had little to none ethnic diversity, this could lead to selection bias
due to the systematic difference in the study sample and the actual population. Sleep
apnea may also have different effects on other ethnic groups.

o “… If sleep apnea only contributes a small additional risk, we may not have had
enough subjects to demonstrate a statistically significant risk.” Sleep apnea does not
solely contribute to the development of diabetes; diabetes has a multitude of risk
factors. By varying the exposure group (varying levels of AHI), the investigators may
be able to identify the significant risks of more severe sleep apnea in comparison to
other risk factors.

o The average age group of the study sample was 49 years, whilst the likelihood of
developing type II diabetes increases drastically after the age of 45. Thus, this study
sample is a lot more susceptible to diabetes or might have already developed diabetes
due to their age.

 Recall Bias:
At the end of each visit, participants were given a questionnaire which included questions
about their personal health, sleep habits and problems, and lifestyle. It is nigh impossible
for a person to remember exactly how many hours he/she slept at an average or details
about their personal health history. As such, the participants might resort to using either
estimates or fabricated memories in order answer this questionnaire, which could lead to
bias in the participant’s bio data.

 Response Bias:
Similarly, participants who lead lives out of the norm could feel pressured to give socially
acceptable answers on their lifestyle and sleep habits. For example, an unemployed person
who sleeps too much and does little to none exercise might feel ashamed to admit his
lifestyle, so he uses fabricated data on the questionnaire in order to avoid social criticism.
This could also lead to bias in the participant’s bio data.

 Misclassification Bias:
“We used report of physician-diagnosed diabetes, supported by details of the date of
diagnosis and treatment, or fasting glucose rather than glucose tolerance testing, which
may have resulted in misclassification.” Because certain individuals have different glucose
tolerance and intolerances, using fasting glucose solely to identify the presence of diabetes
might be ineffective and cause misclassification of subjects with higher or lower insulin
resistances.

 Surveillance Bias:
After the initial cross–sectional study, the investigators knew who had been exposed to
different severities of sleep apnea and who hasn’t. As such, during the follow-up study, the
groups who were exposed might be followed more closely or given more attention than
those who weren’t exposed, leading to bias in the collection of data.

 Hawthorne Effect:
Because the participants know that they are being studied, they might be influenced to
change certain aspects of their lifestyle, habits, or physical activity. For example, a
participant decides to decrease his/her daily sugar intake and do more exercises before one
of his visits during the follow-up study to decrease his/her probability of developing
diabetes.

III. CONFOUNDING: Sleep apnea is not the only sole risk factor of diabetes.
 “One potential conclusion from our data is that SDB contributes to weight gain and
obesity, and this leads to an increased risk of developing diabetes.” Weight and obesity
are both common risk factors of diabetes, and it is stated that people with SDB gain
more weight that those without sleep apnea. It could be that weight is the confounding
factor that causes diabetes. SDB only contributes to weight gain, but not diabetes
entirely.

 “…It has been shown that people with untreated obstructive sleep apnea had higher
leptin levels, suggesting that obstructive sleep apnea is associated with resistance to the
weight reduction effects of leptin.” Leptin is a hormone secreted by the adipose tissues
that plays a role in the decreasing of appetite by sending signals to the hypothalamus. It
also maintains energy balance by regulating food intake and calorie burn rate; In other
words, decrease obesity and hunger. Sleep apnea causes blood levels of leptin to drop,
making you feel hungry even when your body doesn't need food (which contains
glucose). This increase in glucose and calorie intake leads to diabetes.

 It is suspected that SDM is associated with insulin resistance, which is theorized to be a

predecessor to DM (Diabetes Mellitus). However, in individuals with specific genetics,
insulin secretion might halt entirely, leading to diabetes. If this mechanism of
developing diabetes is based solely on genetics, then it could potentially be a
confounding factor.

IV. REVERSE TIME ORDER:

“It is possible that diabetes, either by autonomic instability or some other undefined
mechanism including inflammatory process, may lead to more breathing abnormalities
 during sleep.” The investigators looked into previous studies by the Sleep Heart Health
Study, which found that diabetic people had more cases of SDB than people without
diabetes.

Thus, the hypothesis that diabetes might be causal in the development of SDB have
been considered to be very much possible. However, because this is a cross-sectional
study, the exposure and outcome are both measured at the same point in time, so we do
not know for sure the exact sequence (temporality).

V. CAUSAL:
A direct causal association between type II diabetes and sleep apnea is yet to be found.
Due to the cross-sectional nature of this study, it is impossible to determine if a cause
and effect relationship even exists. However, due to the 4-year follow-up period, it can
be implied that higher levels of AHI does cause an increase in the incidence risk of
developing diabetes.

1. STRENGTH OF ASSOCIATION (YES)

 “Self reported diabetes was 3 to 4 times more prevalent in subjects with an AHI of 15
or greater than in those with an AHI of less than 5.” Greater prevalence of diabetes was
found in subjects with increasing levels of SDB.

 In other words, more cases of outcomes (diabetes) were found to occur to subjects with
higher exposures (AHI). In subjects with an AHI of 15 or above, 14,7% developed
diabetes, whilst in subjects with an AHI below 5, only 2,8% developed diabetes.
Additionally, the odds ratio for having diabetes (physician-diagnosed) of subjects with
an AHI of 5-15 compared to subjects with an AHI of less than 5 was 1.83 (95% CI,
1.07-3.11; p=0.026), indicating a causal association. Furthermore, the odds ratio when
comparing those with an even more severe sleep apnea (AHI of 15) versus an AHI of
less than 5 was 4.75 (95% CI, 2.62-8.63; p<0.0001), suggesting a stronger association.

 In the longitudinal analysis, the results showed a similar association. When comparing
those with an initial AHI of 5-15 compared with those with an AHI of less than 5, the
OR for developing diabetes was 2.81 (95% CI, 1.51-5.23; p=0.001), and the OR
between those with an AHI 15 or greater and those with an AHI of less than 5 was 4.06
(95% CI, 1.86-8.85; p=0.004), both similarly strong associations.

2. CONSISTENCY (NO)
 No other study has ever conducted a longitudinal analysis of a large cohort to identify
if an association exists between SBM and diabetes using a full nocturnal PSG-
evaluation. Previous prospective studies used snoring to measure levels of SDB.
 “There are no published prospective studies demonstrating an increased propensity for
persons with SDB to develop insulin resistance or glucose tolerance.” Although insulin
resistance and glucose tolerance could affect the developent of diabetes, they do not
indicate an association between SDB and diabetes. Thus, this causal association is very
inconsistent.

3. SPECIFICITY (NO)
Diabetes is a multifactorial disease; sleep apnea is not a sole risk factor. Other risk factors
include being overweight or obese, aged 45 or older, having a family member with type II
diabetes, race and ethnicity, heart disease, depression, and lifestyle. Hence, sleep apnea
(exposure) is not a specific cause to type II diabetes (outcome).

4. TEMPORALITY (UNCERTAIN)
Because this is a cross-sectional study set in a single point in time, it is difficult to
determine if the exposure or disease occurred first. In this case, both possibilities are
equally plausible and have been considered and explained in further detail by the
investigators. Thus, we are uncertain on the temporality in this case.

5. DOSE RESPONSE/ BIOLOGICAL GRADIENT (YES)

The data has shown that subjects with a more severe case of sleep apnea (AHI of 5-15 or
higher) had a greater prevalence of developing diabetes than subjects with a less severe
case of sleep apnea (AHI of 5 or lower). This indicated that the more intense the exposure
(Severity of sleep apnea or AHI levels), the greater the risk of disease (type II diabetes).
Thus, a dose response association is present here.

6. PLAUSIBILITY (YES)
A plausible cause and effect mechanism between sleep apnea and type II diabetes is
present. Insulin resistance and impaired glucose tolerance, both direct causes of SDB, are
often thought to precede diabetes, suggesting a causal association albeit an indirect one.
SDB also contributes to weight gain, obesity, and leptin blood levels, all which are
common risk factors to diabetes. This shows that a plausible theory does exist, although
severely limited by insufficient research and scientific evidence.

7. COHERENCE (NO)
“…The preponderance of evidence supports worsened insulin sensitivity and glucose
utilization in those with SDB that this would lead to more diabetes. However our data does
not clearly support this and it underscores the complexity of the pathway from insulin
resistance to the diabetic state and the difficulty in predicting who will go on to develop
diabetes.”

The data obtained here clearly shows that the prevalence of cases of type II diabetes is
higher in those with more severe SDB (AHI>15) than in those with less severe SDB
(AHI<5). This indicates that this data is incoherent with previously established theories
that state sleep apnea associated with insulin resistance would go on to further develop
diabetes.

8. INTERFERENCE/ EXPERIMENTAL EVIDENCE (NO)

There was no interference performed in this study, as this is strictly an observational study
and not an experimental one.

9. ANALOGY (NO)
There were no analogical studies showing another causal relationship between sleep apnea
and diabetes to compare with. Previous studies have only shown cross-sectional
associations between sleep apnea and insulin resistance or glucose intolerance rather than
prospective ones.

4. Best association:
The best association here would be bias. Even though an indirect causal relationship
between sleep apnea and type II diabetes is very much possible, this study contains too
many inaccurate data measurements, leading to numerous potential biases. Thus, any direct
or indirect causal association would be dismissed as biased, due to a lack of valid and
accurate information present.

5. Prevention
 Prevention of Diabetes:
o Exercise - Do more physical activities and stay active
o Control your weight (BMI) - Avoid being overweight or obese
o Plan a balanced diet accordingly – Reduce your sugar intake from sweet foods or
sugary drinks, especially carbonated drinks like Coca-Cola. Avoid processed and red
meat.
o Lifestyle changes – Avoid consuming alcohol and quit smoking

 Prevention of BIAS/CONFOUNDING:
1. BIAS:
o In order to prevent potential biases, I suggest future researches to select a more valid
study design, such as a cohort or case-control study. Although they are expensive and
require more resources, they are also less susceptible to biases than cross-sectional
studies.
o To prevent selection bias, it is best if the researchers used a larger and more diverse
study population. Because the odds of developing diabetes greatly affect ethnicity, it
is not wise to use a cohort consisting of primarily white ethnicity. I suggest that future
studies also use different age groups in contrast to this study, which only used one.
o The investigators should’ve blinded both the participants and the data analysts in
order to avoid surveillance bias and the Hawthorne effect altogether.
o The investigators should’ve taken into account the fact that insulin resistance and
glucose tolerance vary between individuals.

2. CONFOUNDING:
o The best method to reduce the presence of confounding factors is to restrict the study
to a narrower range to differentiate people into different groups based on the severity
of their exposure to other risk factors of type II diabetes, such as BMI index or leptin
blood levels, and perform stratification (layering) of these different exposure groups
to find their individual relative risks and odd ratios. That way, the effects of potential
confounding factors could easily be identified and accounted for.