Modeling Incidence Rate by Region: Analysis of Asthma and Other Chronic Obstructive Lung Diseases

This study aimed to examine regional differences for asthma and other chronic obstructive lung diseases. The study was based on data collected from annual reports produced by the Ministry of Health over a ten-year period beginning from 2010 to 2019. Incidence rates for eleven regions in the Sultanate of Oman were analyzed using statistical tools including; analysis of variance and binary logistic regression model to determine the effect of region on asthma and other chronic obstructive lung diseases. The incident rates were found to be signicantly different by region (F-value=27.07, p=0.00). There was no signicant variation by year (F-value=1.05, p>0.407). Overall over the ten-year period the incidence rates stagnated between 250 and 300 per 10000 of the population, but showed a reducing trend between 2016 and 2019. The logistic regression model shows that compared to the Muscat region, all the other regions had signicant increased odd ratios. There is a signicant evidence of regional variation in the incidence of asthma and other chronic obstructive lung diseases. This implies probable characteristics in geographical regions that are associated with asthma and other chronic obstructive lung diseases’ exacerbation. Five asthma-region classications were identied from our analysis. Findings of this study may be used to guide decision making towards the management and control of asthma and chronic obstructive lung diseases.


Introduction
The incidence and etiology of asthma in the human populations is exacerbated by factors that may be classi ed as genetic and environmental. Studies have established a strong genetic and environmental linkage that interacts both in the induction and subsequent expression of the disease phenotypes (Thomsen 2015;Yang et al. 2017). In their recent study (Krautenbacher et al. 2021), have established that among farm children, asthma is more determined by genetic polymorphisms, while in non-farm children, it is mainly environmental factor-driven. Multiple genes are involved and sometimes interact, but studies to identify the exact asthma-related gene are still being developed (Altzibar et al. 2015;Lawson et al. 2017; Bayliss et al. 2017; Burnett et al. 1994). On the other hand, the epithelium, which is in contact with the environment, and the underlying cells ( broblasts and dendritic), indicate reactivation of the unit (epithelial mesenchymal trophic) that provides a source for asthma (Holgate 2008; Hellings and Steelant 2020; Mitchell and O'Byrne 2017). Although, studies con rm the effect of environment on exacerbating asthma, not many studies have provided substantial data-centered evidence to support the hypotheses.
In this study, we tested the effect of region to examine its effect on the variation in the incidence rates for asthma and other chronic obstructive lung diseases within Oman.
Asthma and other chronic obstructive lung diseases (ACOLD) within the Sultanate of Oman present a complicated scenario resulting into episodic exacerbations, caused by persistent and variable airway obstruction (Buendía et al. 2018; Tliba and Panettieri Jr 2019). In the medical eld, asthma has been described as di cult due to poorly controlled chronic symptoms, or resulting from patients with refractory asthma who remain di cult to control despite an extensive re-evaluation of diagnosis, management, and following an observational period of at least six months by an asthma specialist (

Study Design
This study was based on data collected from annual reports produced by the Ministry of Health, and archived on their website. Quantitative data on asthma and other chronic obstructive lung diseases' incidence data from eleven regions (governorates) for a ten-year period from 2010 to 2019 were analyzed.

Data Analysis
We evaluated two predictors, region and year to assess the incidence rates among the population. Data analysis was performed primarily to achieve the set speci c objectives mentioned above. The incidence rates were measured as a rate per 10000 of the population for the period of study.
Assessing regional differences using analysis of variance Duddek et al. 1995). We applied this approach to establish if there existed signi cant effects on incidence rates by region and year. Since we have only one observation per (region, year) category, we assume no region-year interaction to be able to assess the effects of year and region. Besides an assumption of interaction is hard to justify, since its existence implies overall regional variation in incidence rates are not meaningful unless they are examined year by year. Since the years in question have passed, a main effects model was tted. Using the Duncan's multiple comparisons test, we identi ed the most signi cant differences. Duncan's test was used rather than least signi cant difference (LSD) to avoid in ating the overall probability of type I error to a level much higher than 0.05. As a follow-up, a one-way analysis of variance and a nonparametric test were also carried out.
Determinants of asthma incidence rates Since incidence rates are proportions resulting from a binary classi cation of the population as diagnosed with Asthma or not, a binary logistics regression analysis is a more natural approach to study the effects of region and year on the odds of asthma. A binary logistic model with Muscat as a reference category was tted. Regions are compared with Muscat because unlike other regions, Muscat is the capital of the country and the most metropolitan region. Year was included in the model to provide an estimate of the annual change in the odds of incidence of asthma and other chronic obstructive lung diseases.

Estimation And Results
We present the study ndings beginning with the general distribution of asthma incidence by region and year, followed by analysis of variance (ANOVA) and the logistic regression model. The analyses were aimed at determining the effect of region on the incidence of asthma and its possible distribution over the regions.
Incidence rates by region and year Figure 1 displays the incidence rates by year and region. It shows that the incidence rates were relatively stable over the ten-year study period for most of the regions except Dhofar and Al Wusta. Whereas incidence rates for Dhofar decreased, those for Al Wusta increased between the year 2010 and 2019. The rates for Musandam were consistently higher than other regions followed by those in South Ash Sharqiyah region. The gure provides no clear evidence of region-year interaction. Table 1, gives the mean (standard deviation) for regions and years. The standards deviations show the erratic variability of incidence rates in Dhofar and Musandam.  Analysis of variance of incidence rates of asthma and other chronic obstructive lung disease To test the hypothesis for existence of differences in asthma incidence among the eleven regions, and the ten years, we used analysis of variance (ANOVA) and the null hypotheses of no differences in mean incidence rates for regions or years were tested (see Table 2). There were highly signi cant differences in asthma incidence rates among regions (F = 27.072, p = 0.001); however, these differences were not signi cant among the years (F = 1.051, p = 0.407). Further our model shows that region and year explain about 75% of the variation in incidence rates (R2 = 0.757). Since there is no evidence of variation among years, a one-way ANOVA was tted using regions only, resulting to the same conclusion of highly signi cant differences among regions (F = 26.948, p = 0.001). Here, regions explain about 73% of the total variation in incidence rates (R2 = 0.73) indicating that yearly variations account for about 3% only of the total variability in incidence rates. This region hosts the capital city, it is urban, metropolitan, with plenty of government and private clinics and health care facilities.
2. The second group is made up of six geographical regions, that is North As Sharqiya, Ad Dakhliya, Al Buraimi, Ad Dhahira, South Al Batina and North Al Batina with average rates ranging from 158 to 220 per 10000. They are basically located in the central and interior area of Oman.
3. The third group is made up of only the South As Sharqia region with an average rate of 312 per 10000. This region is located on the East coast.
4. The forth group is made up of two regions, Dhofar and Al Wusta with average rates of 395 and 435 per 10000. These two are located in the Central and Southern parts of the country.

5.
In the fth group is the Musandam region with average rate of about 500 per 10000. The Musandam region is located in the far north, separated from the mainland by the United Arab Emirates, it is a mountainous peninsula.
Overall, the incidence rate in Muscat region was by far the lowest in the country, followed by those regions in the Central, then the South and the far North of Oman. There was evidence of lack of homogeneity of variance (Levene's test statistic = 4.867, p = 0.000). Evidence of lack of homogeneity suggested performing analysis based on a nonparametric method and the Friedman's nonparametric test was used to test differences between regions. This led to the same conclusion (Friedman's Test Statistics = 37.843, p = 0.000).
Determinants of asthma and other chronic obstructive lung disease incidence rates Binary logistic regression is more appropriate technique for examining the effects of a number of qualitative and quantitative explanatory variables on a binary response and hence for rates and proportions. It models the logarithm of the odds ratio as a linear function of the explanatory variables.
Muscat was taken as the reference category for other regions and year was entered as a quantitative variable. Table 3 summarizes the results. The negative coe cient for year shows that overall incidence rate is decreasing over the years. The odds on incidence of asthma and other chronic obstructive lung disease decrease by about one percent per year. All the coe cients for regions are positive, con rming that they all have higher incidence rates than Muscat. The estimated odds of incidence for regions range from twice the odds for Muscat in North Sharqiya, to six fold as in Al Wusta and Musandam. We noted that the regions ordered from lowest to highest estimated odds of incidence is the same as the order of the observed raw incidence rates.  Sharma et al. 2018). In this study, we have tested for region and tted a related prediction model for ACOLD incidence rates.
The study established ve ACOLD geographical regions, which are based on the incidence rates. Generally, the incidence rates remained fairly the same over the period, but started to decrease between 2016 and 2019. The lowest incidence rate was in the year 2019, possibly in uenced by demographic and other socio-economic characteristics that were outside the scope of this study.

Asthma incidence rates by geographical regions
Using the analysis of variance (ANOVA) (Andrusaityte et al. 2016) we identi ed ve groups for asthma incidence rates by regions. The grouping of eleven geographical regions into ve regions was premised on incidence rates, their standard deviations, correlations and trends of the incidence rates over the period. There was a general consensus for the regional groups for the incidence rates.

Asthma incidence using the logistic regression model
The results from the logistic model corroborate with those obtained from the Analysis of variance. The logistic regression provided an estimate of the annual decrease by about 1% of the odds of asthma and other chronic obstructive lung disease. It also readily provides a comparison of two regions using the ratio of the two odds of asthma and other chronic obstructive lung disease for the two regions. This is obtained directly by dividing the odds of the two regions relative to Muscat by each other.
The general climate in Oman is tropical desert, almost everywhere, characterized with some summer rains in the northern and southern parts, with clouds along the eastern coast brought by the Monsoon winds. These winds, though have limited effects in terms of rain, but tend to in uence the climate of the country. There are also childhood risk factors that include among others, phenotypes of asthma, for example early childhood wheezing and age at onset, breastfeeding, poor lung function, family structure, low socio-economic status, exposure to environmental tobacco, exposure to animals, sex and gene-environment interactions. However, our ndings show that though, region was signi cant, its interaction with year was not signi cant (Akinbami et al. 2016).
Availability of time series data for a relatively longer period would possibly provide trends and seasonality of the incidence rates among all categories of, for example; age, region and gender. Nonetheless, this study highlights the signi cance of distributive evidence of asthma by region.

Conclusion And Policy Implications
The fact that asthma is chronic and categorized as one of the most hazardous health threats to the economic contribution by the workforce points to the urgent need for it to be prioritized among common health risks. Its association with COVID-19 severity and poor outcome is also compelling (Eger et al. 2021). Combatting this health threat primarily requires information on incidence rates; distribution and other related risk factors so as to inform decisions on human and nancial resource mobilization and allocation. This study draws the conclusion that geographical region of residence has signi cant association with incidence rates of asthma and chronic obstructive lung diseases. As to whether, this effect may vary over time and space is yet a gap to be ascertained. The implication of our ndings is that both health practitioners and policy makers should focus resources to the identi ed regions to abate the effect of incidence rates, since different regions have different de ning characteristics, which in turn affect asthma incidences. Secondly, grouping incidence rates by region helps to focus resources and maximize reduction in the disease incidence rates because of the targeted interventions. The Ministry of Health should consider the ve-asthma regions when making resources allocation decisions.