The Taiwanese Census Data
Taiwan has been excluded from the United Nations and its related organizations since 1971. Taiwan has not been privy to the health related work of the World Health Organization (WHO). There is no Taiwanese census data available from WHO file. The Taiwanese National Health and Nutritional Survey for the period from 1999 to 2000 were not published until 2005. One of the Taiwanese sleep medicine studies covering the period from 2002 to 2003 was published by Tang (2007). The control data used for Tang’s study (2007) was from a national survey, which was not published in Taiwan until 1999. Another report entitled ‘Elderly Nutrition and Health Survey in Taiwan (1999-2000): research design, methodology and content’ was not published until 2005.
The Filipinos Census Data
The WHO report for 2006, showing the 2004 estimates, for the Philippines is presented in http://www.who.int/tb/publications/global_report/2006/pdf/full_report_correctedversion (accessed 1 September 2009).
The updated estimates for 2005 in the Philippines 2007 WHO report can be located at the following site (accessed 1 September 2009):
http://www.who.int/tb/publications/global_report/2007/download_centre/en/index.html
Clinical Research Materials used in this Article for Comparison
A related literature review follows. It is nested within case series studies of sleep disturbance of two different societies. Tang’s case-serious study (2007) was used as the framework for discussion. Tang’s (2007) study mainly is on total sleep time and PSG findings. Tang’s (2007) study is compared with data of other studies in Taiwan and the Philippines. The relationship among snoring, total sleep time, apnea-hypopnea index (AHI), body mass index (BMI), sleep efficiency, etc are evaluated and compared in this article (Tables 1 -6). To understand the relative impact of each set of independent variables (such as demographic, socioeconomic, and health characteristics, along with sleep variables) on the risk of being overweight or underweight (versus normal weight), examples were derived from related studies of both societies. The comparison was then made.
The Classification of Obstructive Sleep Apnea Syndrome
The OSAHS was classified according to the criteria of severity of sleep apnea. The degrees in severity of sleep apnea were defined, bases on the protocol of American Academy of Sleep Medicine Task Force (1999). These include: 1] zero (0) degree, apnea/hypopnea index (AHI) < 5, 2] first degree, 5 <= AHI < 15, second degree, and 3] 15 <= AHI < 30, and third degree, AHI > 30.
It is worthwhile to evaluate the difference between 1999 and 2007 two different American Academy of Sleep Medicine (AASM) criteria. The difference between AASM’s 1999 and 2007 criteria in AHI follows. For example, the recommended hypopnea definition of 2007 has to meet the following criteria:
1) The nasal pressure signal excursions drop by not less then 30% of pre-event baseline.
2) The duration of this drop occurs for a period lasting 10 seconds.
3) There is a not less than 4% desaturation for pre-event baseline.
4) At least 90% of the event’s duration must meet the amplitude reduction of criteria for hypopnea.
The aforementioned definition of AASM’s 2007 criteria is somehow different from that of 1999′s criteria. It would cause some hypopnea cases to be deleted in the classification, if one is using the criteria based on 2007′s instead on 1999 ‘s criteria. Such a consideration in the aforementioned difference is applicable to all the clinical classification (Tang, 2009a).
Criteria Commonly used in Applying Apnea-Hypopnea-Index (AHI)
For the indication for clinical treatment or not, the criteria include:
1] AHI < 5, SDB can probably be ruled out , 2] 10< AHI < 15 gray zone, 3] AHI = 15 warranted treatment, and 4] 30<AHI<50 definitely needed to be treated clinically.
Sleep in the PSG findings were staged according to Rechtschaffen et al’s criteria (1963)
Results
Sleeping Apnea in a Taiwanese Sleep Study
Sleeping apnea was found 81.3 % with age of 65 years and over suffering from sleep disturbance found in a Taiwanese study (Tang, 2007) according to the result of PSG. Tang (2007) reported that 15.32% of his subjects had an AHI<5; 14 out of 19 were obese and each had an AHI>30. Obesity is defined by the U S National Institute of Health. There were 45 overweight patients (39.52%), only 7 had an AHI<5. Sixteen out of 38 subjects subject w2ere overweight and each had an AHI not less than 30. (Tang, 2007)
Obstructive Sleep Apnea (OSA)
Notwithstanding the fact that obstructive sleep apnea (OSA) may be common in Asian and Southern Pacific patients as compared with what has been reported for Western patients, there is hardly any information of large study. No prior publication relating height measurement with AHI exists. Apnea (elevated AHI) and snoring are associated with obesity and increased neck size has been widely reported in the literature, as well as has the relationship of AHI to hypertension (Paw et al, 1996; Fletcher et al, 1995, Hader et al, 1998; Hoffstein et al, 19 91; Wilcox et al, 1003, Shahar et al 2003; Johnson et al, 1984).
AHI: Decile and the Quartile Distributions
Decile Distributions
In this study, using Tang’s 2007 clinical data, study, a decile, instead of quartile, distribution of height was found with a significant negative correlation with AHI. There was no such a relationship found between the latter (AHI) and weight. In the tenth decile (D10), the range of height was from 169 to 174 cm, and that of AHI was from 0.3 to 106 per hour. In D10, the Spearman rank correlation test revealed that between height and AHI, there was a significant negative correlation coefficient rs – 0.74, with two tailed p value = 0.0058, which was significant.
Quartile Distribution
It has, as well, shown that a quartile distribution by the height with the Spearman correlation coefficient being – 0.34, albeit the two tailed p (0.01) was marginal insignificant. (p = 0.06). Another novel finding reported here is that AHI relates with weight only but not with height.
A Novel Finding Of AHI: A Quantile-Quantile Plots (Q-Q Plots)
The quantile-quantie plots (q-q plots) of AHI suggest deviations from normality, supported with kurtosis and skewness that are greater/less than a -2 to +2 range when their standard deviations are considered, imply that the assumption of normality is not met; notwithstanding, none of the diagnostics of kurtosis and skewness on multiple linear regressions of AHI being significant or worrisome. (Figs. 1-4)
Among the four figures, the Fig.1 shows regressional residual versus filted diagnostic values. The Fig. 2 shows a relationship between theoretical quantiles and standardized residuals in a Q Q curve. The Fig.3 shows a relationship between the standardized residuals and regressional filted values. The Fig.4 shows relationship between the standardized residuals and leverage.
Figure 1. (upper left) There is a slight increase in variability with increasing fitted values in the Fig. 1 as compared with Fig. 3, but it is not problematic
Figure 2. (upper right) Harrell-Davis plot not only gives empirical probability intervals, it does not depend on normality of the distribution of interest.
Figure 3. (lower left) There is a slight decrease in variability with decreasing fitted values in the fig. 3 as compared with Fig. 1, but it is not problematic either.
Figure 4. (lower right) Judging from the Residual Diagnostics (right bottom) we can fairly assume that outliers do not distort the parameters (even though the residuals have extreme values).
AHI and BMI
BMI was calculated as weight (kilograms) divided by height squared (meters squared). It has relevance to OSA (upper limit of normal BMI in Far-East Asian is 23.5 kg/m2 from current WHO data.) For the correlation between AHI and BMI, after checking with Spearman’s rank correlation coefficient (rs), using the same data as that of one of Taiwanese studies (Tang, 2007, 2008, 2008a, 2009, and 2009a), the rs revealed 0.295. This rs is close to 0.330, that is, the rs between AHI and neck circumstance. As another novel application, the relationship between AHI and Neck Circumference can be used to approximate with that between AHI and BMI. It is noted that in Taiwan, distributions of body composition are usually generated for children, adolescent, and middle-aged groups, but not for the older people.
The Taiwanese study (Tang, 2007) revealed that the AHI mode was 9.4 per hour; its count for 40 subjects. The highest number of AHI was 106 per hour in that study. There was an insignificant effect on frequency of snoring by grouping factor of AHI. Spearman’s rank correlation coefficient (rs) test revealed a mildly positive correlation between AHI and snoring (rs was 0.26).
Characteristics of BMI in subjects with sleep disturbance follow. Females had mean height, weight and cervical circumference less than the counterparts of the males. The female subjects had mean BMI of 26.550 kgs/m2, and male 25.19. The mean BMI of total 124 subjects was 25.574 +_4.521 kgs/m2, while that of control was 23.768 +_3.662 Kgs/ m2, the difference was significant (the p value < 0.0009). The Pearson’s correlation coefficient between height and BMI was significantly different between males (- 0.227) and females (0.0854). In these data, the female subjects were about one year older than the male in the average. However, the female subjects had average (mean) height, weight and cervical circumference, which were respectively less than the counter parts of the males. The 95% confidential interval for mean BMI was 24.746 – 26.402. The minimum BMI was 15.20, and maximum was 39.15.
AHI was correlated with BMI, snoring, body height and weight, and cervical circumference respectively. There was a highly significant correlation between BMI and snoring. The subjects whose BMIs were more than 25 had more frequent snoring than those whose BMIs were less than 25 in the studied population. BMI of patients was higher than control subjects. There were significant positive correlation between AHI and BMI; Spearman’s rank correlation test revealed that the relationship between snoring and BMI was highly significant.
The Effect of BMI Grouping on the Frequency of Snoring in a Taiwanese Study
BMIs were classified as following 3 groups: (BMI_group 1, BMI >30), (BMI_group 2, BMI between 25 and 30), and (BMI_group 3, BMI <25).
Mann-Whitney test reveals as follows. The two groups of BMI were selected for comparison. Group differences were insignificant between group 1 and group 2 (BMI_group = 1, BMI >30, and BMI_group = 2, BMI between 25 and 30), (Grouping factor BMI: Mann-Whitney test, degree of freedom (d.f.) = 1, P = 0.432). Group difference was significant between group 1 and group 3, also significant between group 2 and group 3. Grouping factor BMI: Mann-Whitney test, d.f. = 1, p < 0.001, p = 0.001, respectively significant. (Tang, 2007) The groups of subjects with their BMI >30, and a BMI between 25 and 30 snored more frequent than those in the group with a BMI <25.
