Ir.ia.ac.cn

Int. J. Med. Sci. 2016, Vol. 13 International Publisher International Journal of Medical Sciences 2016; 13(2): 99-107. doi: 10.7150/ijms.13456 Comorbidity Analysis According to Sex and Age in Hypertension Patients in China Jiaqi Liu1†, James Ma3†, Jiaojiao Wang1†, Daniel Dajun Zeng1, Hongbin Song4, Ligui Wang4, Zhidong 1. The State Key Laboratory of Management and Control for Complex Systems, Institute of Automation, Chinese Academy of Sciences, Beijing, China; 2. Cloud Computing Center, Chinese Academy of Sciences, Dongguan, China. 3. College of Business, University of Colorado, Colorado Springs, CO, USA. 4. Institute of Disease Control and Prevention, Academy of Military Medical Sciences, Beijing, China. †These authors contributed equally to this work.  Corresponding author: Zhidong Cao, Institute of Automation, Chinese Academy of Sciences, No 95 Zhongguancun East Road, 100190, Beijing, China. E-mail: [email protected] Ivyspring International Publisher. Reproduction is permitted for personal, noncommercial use, provided that the article is in whole, unmodified, and properly cited. See http://ivyspring.com/terms for terms and conditions. Received: 2015.08.04; Accepted: 2015.11.11; Published: 2016.01.29 Abstract
Background: Hypertension, an important risk factor for the health of human being, is often accom-
panied by various comorbidities. However, the incidence patterns of those comorbidities have not been widely studied.
Aim: Applying big-data techniques on a large collection of electronic medical records, we investigated
sex-specific and age-specific detection rates of some important comorbidities of hypertension, and sketched their relationships to reveal the risk for hypertension patients.
Methods: We collected a total of 6,371,963 hypertension-related medical records from 106 hospitals
in 72 cities throughout China. Those records were reported to a National Center for Disease Control in China between 2011 and 2013. Based on the comprehensive and geographically distributed data set, we identified the top 20 comorbidities of hypertension, and disclosed the sex-specific and age-specific patterns of those comorbidities. A comorbidities network was constructed based on the frequency of co-occurrence relationships among those comorbidities.
Results: The top four comorbidities of hypertension were coronary heart disease, diabetes, hyper-
lipemia, and arteriosclerosis, whose detection rates were 21.71% (21.49% for men vs 21.95% for women), 16.00% (16.24% vs 15.74%), 13.81% (13.86% vs 13.76%), and 12.66% (12.25% vs 13.08%), respectively. The age-specific detection rates of comorbidities showed five unique patterns and also indicated that nephropathy, uremia, and anemia were significant risks for patients under 39 years of age. On the other hand, coronary heart disease, diabetes, arteriosclerosis, hyperlipemia, and cerebral in- farction were more likely to occur in older patients. The comorbidity network that we constructed indicated that the top 20 comorbidities of hypertension had strong co-occurrence correlations.
Conclusions: Hypertension patients can be aware of their risks of comorbidities based on our
sex-specific results, age-specific patterns, and the comorbidity network. Our findings provide useful insights into the comorbidity prevention, risk assessment, and early warning for hypertension patients. Key words: Hypertension, Comorbidity, Electronic Medical Records, Detection Rate, Network Analysis. Background
Hypertension, or high blood pressure, is one of 2]. In China, hypertension is the leading preventable the most important risk factors that can lead to car- risk factor for death among Chinese adults aged 40 diovascular diseases, and is thus regarded as a serious years and older [3, 4]. Moreover, hypertension has a public health problem. The prevalence of hyperten- large number of comorbidities, which greatly affect sion has been increasing in most areas worldwide [1, hypertension patients' quality of life [5-7]. In previous http://www.medsci.org Int. J. Med. Sci. 2016, Vol. 13 years, researchers and medical practitioners have Although the comorbidities of hypertension made a tremendous effort to study the comorbidities have been extensively studied, most existing research of hypertension [8-10]. Specifically, heart disease [2, is based on medical surveys and public census data. 11], diabetes [12, 13], and obesity [14, 15] are the most Census data sets show aggregated facts of the general widely studied comorbidities of hypertension. Some public without detailed information regarding indi- other diseases, such as allergic respiratory disease [9], vidual patients. In contrast, medical surveys while sleep-disordered breathing [16], and chronic kidney include some individual level information usually disease [17], have also been studied as potential involve a limited number of survey participants be- comorbidities of hypertension. Hypertension and cause of limited resources. Those surveys are often set some of its comorbidities have shown high correla- for a confined geographical area (i.e., a city or a tions in terms of their prevalence. An example of this county), and thus cannot claim to be representative of type of correlations is that the prevalence of hyper- a larger and broader area. Due to the nature of medi- tension in patients with diabetes is as high as 92.7% cal surveys, usually only certain types of participants are willing to reveal their private and sensitive medi- Moreover, the sex-specific and age-specific cal related situations. People with a stronger sense of analyses of comorbidities of hypertension have re- privacy are normally reluctant to reveal their medical sulted in various important findings [1, 19-21]. Spe- history or health-related conditions. Therefore, med- cifically, the incident rates of comorbidities in hyper- ical surveys on a voluntary basis may have a biased tension patients with a different sex and age can sig- participant population. The data points that are col- nificantly differ. An example of this difference is that lected in a medical survey also largely depend on the the incidence of hypertension and hypercholesterol- participant's availability during the time of the sur- emia combined is 20% for women versus 16% for vey, the participant's mood at the time, and the sur- men, and ranges from 1.9% for those aged 20–29 to vey collector's attitude and human interaction skills. 56% for those aged 80 years and older [22]. Addition- Too many human-related factors can affect the quality ally, patient's age and sex need to be considered for of medical surveys. Furthermore, to reduce the survey treatment of these comorbidities [23, 24]. An example participant's reluctance, a medical survey is usually of the situation is that treatment for hypertension pa- composed of a limited number of survey questions so tients who are 80 years or older with indapamide has that an interview or a questionnaire can be completed been proved to be effective and can also reduce the within a short time period. This greatly reduces the patient's risk of stroke [23]. Research has shown that versatility of the survey when analyzing the survey untreated male hypertension patients are more likely results. In summary, because of the time-consuming to suffer from cognitive impairment than untreated and labor-intensive nature of medical surveys, the female hypertension patients do [25]. Thus, hyper- limited number of, and possibly biased, survey par- tension should be treated and controlled as early as ticipants and survey questions can lead to biased possible for male patients before they encounter de- analysis results, and possibly overlook important patterns and relationships in the occurrence of dis- There has been increasing interest in analyzing eases. disease relationships using network theory [26, 27]. In the current study, we leveraged a large, relia- The disease network is particularly useful when ana- ble, and extensive data set and analyzed the occur- lyzing the co-occurrence of different diseases. Specif- rence patterns of hypertension comorbidities. We also ically, the disease network denotes an individual investigated the common comorbidities of hyperten- disease with a vertex, and the co-occurrence of two sion with respect to the patient's sex and age. The diseases with an edge connecting those two diseases. co-occurrence relationships among comorbidities of The disease network summarizes the connections hypertension are also discussed using the disease among diseases and shows progress of disease pref- network approach. erentially along the edges or links [28]. The frequency of co-occurrence relationships among important Methods
comorbidities could provide useful insight into de- scribing the disease development process, and thus Study population
result in doctor's and patient's awareness of diseases Our data set was obtained from a Chinese Na- at the early stage of development. Studying the tional Surveillance System, which was initially im- comorbidity co-occurrence of hypertension using the plemented by the Chinese government in 2010. This disease network may be an effective tool for deter- surveillance system collects electronic medical records mining meaningful comorbidity relationships that from hospitals and aims to oversee the overall health other approaches have not reported. conditions of the Chinese population. Since 2010, this http://www.medsci.org Int. J. Med. Sci. 2016, Vol. 13 system has been adopted by 192 hospitals located Statistical analysis
throughout China. Although we had access to all 192 The occurrences of comorbidities were counted hospitals' data in the surveillance system, we inten- in hypertension-related electronic medical records. tionally excluded some hospitals that did not appear The comorbidity's occurrence was then utilized to to present a sufficient and continuous data stream. derive the detection rate of the comorbidity which Some obvious errors and incomplete data points were better reflected the comorbidity's prevalence in hy- also removed to maintain the data integrity. pertensive patients. The detection rate of a comorbid- Eventually, we decided to use 6,371,963 hyper- ity was defined as the ratio of the number of the tension-related high-quality data records from the comorbidity's records to the number of hyperten- 110,528,991 electronic medical records that we had sion-related records: access to. Those medical records were dated between 2011 and 2013, and were from 106 hospitals located in 72 cities in China (Figure S1). Those cities are geo- graphically distributed in 29 of 31 provinces in China The sex-specific detection rate was determined (excluding two underpopulated provinces, Qinghai as the ratio of the number of each comorbidity in and Ningxia). Our data set covers 33.90% of the city males or females to the number of hypertension cases population in China. The city population data is based in the corresponding sex group. The odds ratios and on the sixth Chinese population census published by their 95% confidence interval (CI) of each sex-specific the National Bureau of Statistics of the People's Re- detection rate were also calculated. For the public of China (http://www.stats.gov.cn). age-specific analysis, every 10-year age range between This study was approved by the institutional re- 0 and 99 years was considered an age group (e.g., 0–9 view board of the Institute of Automation, Chinese years, 10–19 years). Ages greater than or equal to 100 Academy of Sciences. The data set was collected by years were considered as one age group. Because the the Chinese government for disease control. All pa- numbers of each comorbidity in the 0–9 years group tients gave their informed consent. The patient's pri- and above 99 years group were small, the age-specific vacy was strictly preserved in our study. We only detection rates were calculated and analyzed only used the patient's sex, age, and clinical diagnostic from 10 years to 99 years. Similarly the age-specific information to perform our analysis. Patients' identi- detection rate was determined as the ratio of the ty-related information was masked before we started number of each comorbidity in each age group to the number of hypertension cases in the corresponding Data normalization
age group. Their 95% CIs were calculated. To analyze The clinical diagnosis in the original electronic the age-specific prevalence trends of the top 20 medical records was not coded using uniformed and comorbidities, the expectation maximization class in standardized text terms. An example was that some Weka [32] version 3.7.7 was used to cluster those 20 doctors had used "upper infection" as an abbreviation trends. The expectation maximization [33] algorithm for "upper respiratory tract infection" and others had assigns a probability distribution to each trend, which chosen a different abbreviation for the same diagno- indicates the probability of it belonging to each clus- sis. To standardize the diagnosis, we applied a natural language processing technique [29, 30] and developed Network analysis
several in-house Python scripts for Chinese text pro- When two comorbidities of hypertension ap- cessing and mining. Python [31] has been proved an peared in one electronic medical record, we consid- effective tool for handling similar tasks. Specifically ered that there was a co-occurrence relationship be- for our study, each electronic medical record was au- tween this comorbidity pair. The number of tomatically segmented into a series of Chinese words, co-occurrences between a couple comorbidities can be and these words were then combined to form Chinese an important factor to reveal the relationship of those phrases according to the probability distribution of two comorbidities. Thus, we constructed a weighted those words. In addition to automatic normalization comorbidity network [34, 35] to study the comorbidi- of data, many text ambiguities and synonyms were ties of hypertension and the co-occurrence relation- handled manually. Finally, all medical diagnostic ships among those comorbidities. records were converted to standardized and coded The nodes of the network represented comor- diagnostic terms that could be easily manipulated and bidities and the diameter of each node was propor- tional to the detection rate of each comorbidity. An edge in the network indicated the co-occurrence of two comorbidities whom that edge was connecting. http://www.medsci.org Int. J. Med. Sci. 2016, Vol. 13 The weight of an edge was the number of sufficiency and uremia, and respiratory-related dis- co-occurrences of those two comorbidities. When an eases, such as respiratory tract infection, upper res- electronic medical record contained more than two piratory tract infection, and tracheitis had a high de- comorbidities of hypertension, the count of every re- tection rate, which indicated that those comorbidities lationship between each possible pair of comorbidities were of a higher risk in hypertension patients than in that record would have an increment of one (e.g., other comorbidities were. Moreover, the detection when the record was "hypertension, A, B, C", the rates of comorbidities reduced with rank. The detec- count of relationships A-B, A-C, and B-C would all tion rate of the last comorbidity, arthritis, was only encounter an increment of one). After investigating all hypertension-related electronic medical records, we retained the high-frequency relationships among the top 20 comorbidities. The high-frequency relation- Table 1. Detection rates of the top 20 comorbidities of hyper-
ships were defined as relationships with a weight of tension in China. more than 1% of the total number of hyperten- sion-related records. Coronary Heart Disease Several network measures have been adopted to identify the importance of nodes [36]. Three primary methods, namely degree centrality, average degree, 4 Arteriosclerosis and average path length [37], were used to analyze the Cerebral Infarction Move With Difficulty comorbidity network. Degree centrality is the most 7 readily calculated and understood concept of node 8 Respiratory Tract Infection centrality. The degree centrality of a comorbidity is 9 Cerebral Circulation Insufficiency 3.87 the total number of relationships that are directly as- Upper Respiratory Tract Infection 3.43 Renal Insufficiency sociated with that comorbidity. A comorbidity with a high degree centrality has more co-occurrence rela- tionships with other comorbidities in the network 14 [38]. The average degree of a network is an overall 16 Anemia evaluation about the connections among comorbidi- ties [39]. In addition, path length focuses on the least number of relationships in order to connect two 19 Osteoarthropathy comorbidities. A comorbidity pair with a low path length and high edge weights along the path has a higher risk of co-occurrence in hypertension patients. The path length of any two directly connected Sex-specific detection rates
comorbidities is one and the number of comorbidities The sex-specific detection rates of the top 20 on the shortest path is path length minus one. Similar comorbidities of hypertension and their odds ratios to the average degree of a network, the average path were shown in Table 2 and Figure S2. Osteoporosis length of a network is also used to describe the aver- showed the largest difference between males and fe- age distance between each comorbidity pair in the males, which suggested that female hypertension pa- network [40]. A frequently used force-directed layout tients have a 73.12% higher risk than male hyperten- algorithm, the Fruchterman–Reingold algorithm, was sion patients in developing osteoporosis. Other used to layout the network. bone-related diseases, such as arthritis and osteoar- thropathy, also had a higher incidence in female hy- pertension patients than in male hypertension pa- Detection rates of the top 20 comorbidities
tients (40.64% vs 36.29%). In addition, insomnia and The top 20 comorbidities of hypertension with difficulty with movement threated the health of fe- the highest detection rates were identified (Table 1). males more than males (39.78% vs 29.92%). Surpris- Coronary heart disease (CHD), which is one of the ingly, two cerebral diseases showed different risks in most important cardiovascular diseases, had the males and females. Cerebral circulation insufficiency highest detection rate. Diabetes, hyperlipemia, and was 40.15% more likely to occur in females, while arteriosclerosis had a detection rate that was higher cerebral infarction was 19.05% more likely to occur in than 10%. Cerebral diseases, such as cerebral infarc- males. Moreover, several diseases related to the kid- tion and cerebral circulation insufficiency, kid- ney had a higher morbidity in male hypertension pa- ney-related diseases, such as nephropathy, renal in- tients than in female hypertension patients. More at- tention should be paid to renal insufficiency, uremia, http://www.medsci.org Int. J. Med. Sci. 2016, Vol. 13 and nephropathy in male hypertension patients pertension patients was cerebral infarction being (35.39%, 25.56%, and 17.99%) than in female hyper- ranked in the top five comorbidities between 50 and tension patients. The sex-specific detection rates of 89 years of age and the fourth at 90–99 years of age. other top comorbidities, including CHD, diabetes, The age-specific detection rates of the top 20 hyperlipemia, and arteriosclerosis, were relatively comorbidities of hypertension (Figure 2) were clus- uniform, with no significant differences between male tered into five classes. First, the age-specific detection and female patients. rates of CHD, arteriosclerosis, cerebral infarction, in- somnia, arrhythmia, gastritis, osteoarthropathy, and Age-specific detection rates
arthritis gradually increased as patients got older. The The age-specific occurrence distribution of hy- detection rates of these comorbidities at 90–99 years pertension patients was shown in Figure 1. Based on were several times (relative ratio: CHD, 25.91; arteri- 6,371,963 electronic medical records, the proportion of osclerosis, 22.65; cerebral infarction, 18.55; insomnia, hypertension patients who were aged between 50 and 12.58; arrhythmia, 8.17; gastritis, 3.03; osteoarthropa- 79 years was 71.27% (95% CI: 71.23–71.31%). Only thy, 62.16; and arthritis, 8.02) higher than those at the 5.99% of hypertension patients were younger than 40 age of 10–20 years. years. In addition, because there was only a small number of patients who were aged 9 years or older than 100 years, these two age groups were removed from the analysis. The top five detection rates of comorbidities in each age group were different (Table 3). Nephropa- thy, uremia, and anemia were the three biggest risks for hypertension patients who were younger than 39 years, while renal insufficiency was a potential risk to hypertension patients who were younger than 29 years. Hyperlipemia was always in the top five comorbidities through all age groups and was the top comorbidity in the 40–49-year age group. Addition- ally, CHD, diabetes, and arteriosclerosis became a major risk when hypertension patients were older than 40 years. Another significant risk for older hy- Figure 1. Age-specific distribution of hypertension patients in China.
Table 2. Sex-specific distribution of the top 20 comorbidities of hypertension in China.
Male Detection Rate(%) 95% CI Female Detection Rate(%) 95% CI Odds ratios 95% CI Coronary Heart Disease 21.44-21.53 21.95 21.90-22.00 0.973 0.970-0.977 <.00001 16.20-16.28 15.74 15.70-15.78 1.038 1.034-1.042 <.00001 13.82-13.89 13.76 13.72-13.80 1.008 1.003-1.012 0.00057 Arteriosclerosis 12.22-12.29 13.08 13.05-13.12 0.928 0.923-0.932 <.00001 Cerebral Infarction 1.200-1.215 <.00001 Move With Difficulty 0.761-0.773 <.00001 1.179-1.198 <.00001 Respiratory Tract Infection 0.892-0.907 <.00001 Cerebral Circulation Insufficiency 3.24 0.698-0.710 <.00001 Upper Respiratory Tract Infection 3.36 0.945-0.962 <.00001 Renal Insufficiency 1.355-1.380 <.00001 0.946-0.963 <.00001 0.563-0.573 <.00001 0.702-0.715 <.00001 1.252-1.276 <.00001 0.969-0.988 2.5E-05 0.908-0.927 <.00001 0.868-0.886 <.00001 Osteoarthropathy 0.721-0.737 <.00001 0.698-0.714 <.00001 http://www.medsci.org Int. J. Med. Sci. 2016, Vol. 13 Table 3. Top five comorbidities of hypertension in each age group.
Renal Insufficiency Renal Insufficiency Arteriosclerosis Arteriosclerosis Cerebral Infarction Arteriosclerosis Cerebral Infarction Arteriosclerosis Cerebral Infarction Arteriosclerosis Cerebral Infarction Arteriosclerosis Cerebral Infarction Figure 2. Age-specific patterns of the top 20 comorbidities of hypertension in China.
Second, the age-specific detection rate of diabe- Third, moving with difficulty, respiratory tract tes, hyperlipemia, and cerebral circulation insuffi- infection, upper respiratory tract infection, tracheitis, ciency increased with an increase in age but decreased and osteoporosis had different rates of detection rate in older patients. For diabetes and hyperlipemia, the increase, and occasionally showed a slight decline detection rate reached a peak at 70–79 years, with depending on age. The detection rate of moving with detection rates of 19.55% and 15.33%, respectively. difficulty greatly increased at 20–29 years (228.86%), The detection rate of diabetes continued to increase 50–59 years (172.18%), and 70–79 years (153.28%) over time, with the highest detection at 40–49 years. compared with the previous age group. Respiratory However, the detection rate of hyperlipemia flattened tract infection and upper respiratory tract infection off from 50–79 years (14.56–15.33%). Moreover, the had the same patterns in detection rate of greatly in- detection rate of cerebral circulation insufficiency creasing at 20–29 years (171.34% and 153.13%, respec- flattened off from 50–69 years (3.92%) and then tively) and 50–59 years (128.87% and 127.20%, respec- peaked at 70–89 years (4.62–4.70% for the two age tively). The detection rate of tracheitis greatly in- groups). The detection rates of these three comorbidi- creased at almost all age ranges (131.13–181.66%) and ties greatly decreased in older people after the peak declined at 60–69 years (103.18%) compared with (diabetes: 29.72%; hyperlipemia: 26.01%; and cerebral previous age groups. The detection rate of osteoporo- circulation insufficiency: 8.05%). http://www.medsci.org

Int. J. Med. Sci. 2016, Vol. 13 sis was also unique in that it decreased below 40 tion ensured individual patient's medical records to years and quickly increased by 50–59 years (212.36%). be reliable, extensive, and timely. Compared with Fourth, the trend for detection rate did not al- other similar research [1, 19, 20, 22], our study was ways show an upward trend. The detection rate of based on a much larger patient base. Our data records kidney-related diseases continued to fall at most age were collected while patients were hospitalized, thus ranges. The age-specific detection rates of nephropa- the records contained detailed and extensive coverage thy, uremia, and anemia fell from 16.81% (10–19 on patient's medical-related information. Because the years) to 1.94% (90–99 years), 10.47% (20–29 years) to medical-related information was for medical diag- 0.56% (90–99 years), and 8.57% (10–19 years) to 1.64% nostic purposes, the information was highly reliable (80–89 years), respectively. The decline in detection and objective. rate of these diseases in each age group compared with the previous age group was similar. The last class only contained renal insufficiency whose detection rate reached a peak at 20–29 years (6.15%) and showed a U-shaped curve with increasing age. At 50–69 years old, hypertension patients had the lowest risk in developing renal insufficiency with a detection rate of only 2.66%.
Comorbidity network of hypertension
The comorbidity network comprising high co-occurrence frequency relationships among comor- bidities of hypertension was presented in Figure 3 and Table S1. The core of the network included CHD, hy- perlipemia, arteriosclerosis, and diabetes whose de- gree centrality was 13, 7, 7, and 6, respectively. Those comorbidities were directly connected to 75% of all comorbidities. Therefore, hypertension patients who had one of those four comorbidities had a greater health risk. Uremia and anemia were connected to the Figure 3. Comorbidity network of hypertension.
core network through nephropathy, which indicated that nephropathy was an important indicative varia- ble between those two comorbidities and the core To the best of our knowledge, this study was the network. Hypertension patients with CHD, hyper- first to investigate the prevalence of comorbidities of lipemia, arteriosclerosis, and diabetes had a relatively hypertension through a large amount of electronic low risk of developing uremia and anemia. Gastritis medical record data rather than using just medical and the comorbidity pair of arthritis and osteoar- survey or census data. Our findings on the detection thropathy were isolated in the core network. The rates of comorbidities were sufficiently representative morbidity risk of these three comorbidities was rela- for Chinese population and were insightful for doc- tively independent to comorbidities in the core net- tors and hypertension patients. The top 20 comorbid- work. Moreover, because the average degree was 3.3 ities in terms of the detection rate and their and the average path length was 2.09 in the comor- co-occurrence relationships implied important health bidities network, the comorbidity network showed risks to hypertension patients. From our study, more that the top 20 comorbidities had a strong correlation targeted measures can be taken into consideration in with each other. Each top 20 comorbidity was directly order to prevent the deterioration of health of hyper- connected to an average of 3.3 other comorbidities tension patients. and the number of comorbidities between any two The sex-specific and age-specific detection rates comorbidities was only approximately one. of comorbidities described the different risks of comorbidities in hypertension patients with a differ- Discussion
ent sex and age range. We found that female hyper- In our study, we obtained a large collection of tension patients were more likely suffer from osteo- electronic medical records from 106 prestigious hos- porosis, while male hypertension patients were more pitals located in 72 cities in China. The data collection likely to develop renal insufficiency. Nephropathy, process was mostly automatic and involved little uremia, and anemia were important risk factors in human intervention. The automation of data collec- hypertension patients younger than 39 years, while http://www.medsci.org Int. J. Med. Sci. 2016, Vol. 13 CHD, diabetes, hyperlipemia, arteriosclerosis, and medical data collection process need to be executed to cerebral infarction were high risk factors in older hy- ensure a more comprehensive data collection. pertension patients. Those findings can provide guidelines for the prevention of comorbidities of hy- Conclusions
pertension. An example of a preventative measure is In summary, our analysis of comorbidities of with diabetes, one of the most frequently observed hypertension in China between 2011 and 2013 pro- comorbidities of hypertension, where reducing sugar vided an overview of the detection rate of comorbidi- intake is a common proposal for hypertension pa- ties among hypertension patients. Variate detection tients. In addition, kidney disease is the most im- rates of comorbidities regarding age and sex were portant risk factor in young hypertension patients. presented, and the co-occurrence relationships among Therefore, patient's sex and age should be accounted comorbidities were analyzed. Our findings can sup- in when proposing prevention measures for hyper- port doctors and patients to make more specific di- tension patients. agnoses and treatment plans by considering patient's Currently, medically-aided diagnostic technolo- age, sex and comorbidity conditions. Our results can gies primarily focus on preliminary statistics and a also increase people's awareness of the comorbidities probabilistic computing system. In the era of large of hypertension. Further study on hypertension and amounts of data, network-based recommendation its comorbidities will likely improve the life quality of technology continues to be developed. The character- hypertension patients, and be helpful for the preven- istics of comorbidities that are calculated from a tion of hypertension. comorbidity network could provide valuable infor- mation about the relationships between each comor- bidity pair. Our findings could rapidly promote the CHD: Coronary Heart Disease; CI: Confidence development of new diagnostic technologies, not only for hypertension, but also for other diseases. In the current study, the high-frequency Supplementary Material
co-occurrence relationships among comorbidities of Figures S1-S2, Table S1. hypertension were analyzed and presented by the http://www.medsci.org/v13p0099s1.pdf comorbidity network. Relationships with a high edge weight indicated those two comorbidities had a high co-existing correlation. The core network comprising This study was funded by National Natural the top four comorbidities verified a strong Science Foundation of China (Nos. 91024030, co-occurring relation and high risk of those four 71025001, 91224008, 91324007) and Important Na- comorbidities to hypertension patients. Anemia and tional Science & Technology Specific Projects (Nos. uremia had a relatively lower relevance with comor- 2012ZX10004801, 2013ZX10004218). bidities in the core network than nephropathy did. Moreover, arthritis, osteoarthropathy, and gastritis Competing interests
had a relatively independent morbidity risk with the core network. In overall, the high-frequency The authors declared that they had no compet- co-occurrence relationships among comorbidities ing interests. could be important for prevention and treatment of References
hypertension and its comorbidities. 1. Gu DF, Reynolds K, Wu XG, Chen J, Duan XF, Muntner P, et al. Prevalence, There are limitations of this study. First, the de- awareness, treatment, and control of hypertension in China. Hypertension. tection rates achieved from different cities had shown 2002; 40: 920-7. 2. Wolf-Maier K, Cooper RS, Banegas JR, Giampaoli S, Hense HW, Joffres M, et different patterns. Further study, such as spatial al. Hypertension, prevalence and blood pressure levels in 6 European analysis, could reveal the reasons for those differ- countries, Canada, and the United States. JAMA. 2003; 289: 2363-9. 3. He J, Gu DF, Wu XG, Reynolds K, Duan XF, Yao CH, et al. Major causes of ences. Second, some ambiguous or casually typed death among men and women in China. N Engl J Med. 2005; 353: 1124-34. records were ignored due to the insufficiency of the 4. Sheng CS, Liu M, Kang YY, Wei FF, Zhang L, Li GL, et al. Prevalence, awareness, treatment and control of hypertension in elderly Chinese. natural language processing techniques that we had Hypertens Res. 2013; 36: 824-8. utilized. Adopting more effective text mining tools 5. Al-Tuwijri AA, Al-Rukban MO. Hypertension control and co-morbidities in primary health care centers in Riyadh. Ann Saudi Med. 2006; 26: 266-71. might increase the validity of rules that we used and 6. Hirani V, Zaninotto P, Primatesta P. Generalised and abdominal obesity and the likelihood of finding new rules. Third, some risk of diabetes, hypertension and hypertension-diabetes co-morbidity in England. Public Health Nutr. 2008; 11: 521-7. seemingly unrelated or undetected patient symptoms 7. Wang R, Zhao Y, He X, Ma X, Yan X, Sun Y, et al. Impact of hypertension on health-related quality of life in a population-based study in Shanghai, China. might not have been completely and thoroughly rec- Public Health. 2009; 123: 534-9. orded in the system. More detailed inspections on the 8. in't Veld AJM. Symptomatic BPH and hypertension: Does comorbidity affect quality of life? Eur Urol. 1998; 34: 29-36. http://www.medsci.org Int. J. Med. Sci. 2016, Vol. 13 9. Aung T, Bisognano JD, Morgan MA. Allergic respiratory disease as a potential 38. Opsahl T, Agneessens F, Skvoretz J. Node centrality in weighted networks: co-morbidity for hypertension. Cardiol J. 2010; 17: 443-7. Generalizing degree and shortest paths. Social Networks. 2010; 32: 245-51. 10. Prudenzano MP, Monetti C, Merico L, Cardinali V, Genco S, Lamberti P, et al. 39. Tang CL, Wang WX, Wu X, Wang BH. Effects of average degree on The comorbidity of migraine and hypertension. A study in a tertiary care cooperation in networked evolutionary game. EPJB. 2006; 53: 411-5. headache centre. J Headache Pain. 2005; 6. 40. Fronczak A, Fronczak P, Holyst JA. Average path length in random networks. 11. Dzudie A, Kengne AP, Mbahe S, Menanga A, Kenfack M, Kingue S. Chronic PhRvE. 2004; 70. heart failure, selected risk factors and co-morbidities among adults treated for hypertension in a cardiac referral hospital in Cameroon. Eur J Heart Fail. 2008; 12. Channanath AM, Farran B, Behbehani K, Thanaraj TA. State of Diabetes, Hypertension, and Comorbidity in Kuwait: Showcasing the Trends as Seen in Native Versus Expatriate Populations. Diabetes Care. 2013; 36: E75-E. 13. Weiderpass E, Persson I, Adami HO, Magnusson C, Lindgren A, Baron JA. Body size in different periods of life, diabetes mellitus, hypertension, and risk of postmenopausal endometrial cancer (Sweden). Cancer Causes Control. 2000; 11: 185-92. 14. Lukas A, Kumbein F, Temml C, Mayer B, Oberbauer R. Body mass index is the main risk factor for arterial hypertension in young subjects without major comorbidity. Eur J Clin Invest. 2003; 33: 223-30. 15. Uretsky S, Messerli FH, Bangalore S, Champion A, Cooper-DeHoff RM, Zhou Q, et al. Obesity paradox in patients with hypertension and coronary artery disease. Am J Med. 2007; 120: 863-70. 16. Bixler EO, Vgontzas AN, Lin HM, Ten Have T, Leiby BE, Vela-Bueno A, et al. Association of hypertension and sleep-disordered breathing. Arch Intern Med. 2000; 160: 2289-95. 17. Sarafidis PA, Li S, Chen SC, Collins AJ, Brown WW, Klag MJ, et al. Hypertension awareness, treatment, and control in chronic kidney disease. Am J Med. 2008; 121: 332-40. 18. Abougalambou SSI, Abougalambou AS. A study evaluating prevalence of hypertension and risk factors affecting on blood pressure control among type 2 diabetes patients attending teaching hospital in Malaysia. Diabetes Metab 19. Wu YF, Huxley R, Li LM, Anna V, Xie GQ, Yao CH, et al. Prevalence, Awareness, Treatment, and Control of Hypertension in China Data from the China National Nutrition and Health Survey 2002. Circulation. 2008; 118: 20. Wang J, Ning X, Yang L, Lu H, Tu J, Jin W, et al. Trends of hypertension prevalence, awareness, treatment and control in rural areas of northern China during 1991-2011. J Hum Hypertens. 2014; 28: 25-31. 21. Schillaci G, Pirro M, Vaudo G, Gemelli F, Marchesi S, Porcellati C, et al. Prognostic value of the metabolic syndrome in essential hypertension. J Am Coll Cardiol. 2004; 43: 1817-22. 22. Wong ND, Lopez V, Tang S, Williams GR. Prevalence, treatment, and control of combined hypertension and hypercholesterolemia in the United States. Am J Cardiol. 2006; 98: 204-8. 23. Beckett NS, Peters R, Fletcher AE, Staessen JA, Liu LS, Dumitrascu D, et al. Treatment of hypertension in patients 80 years of age or older. N Engl J Med. 2008; 358: 1887-98. 24. Tanushi H, Dalianis H, Nilsson GH. Calculating prevalence of comorbidity and comorbidity combinations with diabetes in hospital care in sweden using a health care record database. 3rd International Workshop on Health Document Text Mining and Information Analysis 2011, LOUHI 2011, July 6, 2011 - July 6, 2011. Bled, Slovenia: Sun SITE Central Europe CEUR-WS; 2010: 25. Kilander L, Nyman H, Boberg M, Hansson L, Lithell H. Hypertension is related to cognitive impairment - A 20-year follow-up of 999 men. Hypertension. 1998; 31: 780-6. 26. Barabasi AL. Network medicine - From obesity to the "Diseasome' . N Engl J Med. 2007; 357: 404-7. 27. Goh KI, Cusick ME, Valle D, Childs B, Vidal M, Barabasi AL. The human disease network. Proc Natl Acad Sci U S A. 2007; 104: 8685-90. 28. Hidalgo CA, Blumm N, Barabási A-L, Christakis NA. A Dynamic Network Approach for the Study of Human Phenotypes. PLoS Comput Biol. 2009; 5: 29. Shi J, Hu M, Shi X, Dai G-Z. Text segmentation based on model LDA. Chin J Comp. 2008; 31: 1865-73. 30. Fu GH, Kit C, Webster JJ. Chinese word segmentation as morpheme-based lexical chunking. Inf Sci. 2008; 178: 2282-96. 31. Sanner MF. Python: A programming language for software integration and development. J Mol Graph Model. 1999; 17: 57-61. 32. Hall M, Frank E, Holmes G, Pfahringer B, Reutemann P, Witten IH. The WEKA data mining software: an update. ACM SIGKDD explorations newsletter. 2009; 11: 10-8. 33. Moon TK. The expectation-maximization algorithm. ISPM. 1996; 13: 47-60. 34. Roque FS, Jensen PB, Schmock H, Dalgaard M, Andreatta M, Hansen T, et al. Using Electronic Patient Records to Discover Disease Correlations and Stratify Patient Cohorts. PLoS Comput Biol. 2011; 7. 35. Jensen AB, Moseley PL, Oprea TI, Ellesoe SG, Eriksson R, Schmock H, et al. Temporal disease trajectories condensed from population-wide registry data covering 6.2 million patients. Nat Commun. 2014; 5: 4022. 36. Bavelas A. Communication patterens in task-oriented groups. J Acoust Soc Am. 1950; 22: 723-30. 37. Newman MEJ. The structure of scientific collaboration networks. Proc Natl Acad Sci U S A. 2001; 98: 404-9. http://www.medsci.org

Source: http://ir.ia.ac.cn/bitstream/173211/11433/1/2016IJMS_Comorbidity%20analysis%20according%20to%20sex%20and%20age%20in%20hypertension%20patients%20in%20China.pdf