Asthma and Weather

A Mathematical Model




Click here to go directly to information about the Asthma Wheel.



Summary


   The researcher, an asthma and hay fever sufferer, has completed a five year investigation to determine whether environmental factors could be used to predict the severity of allergies and estimate the number of asthmatic related clinic visits in various climatic regions. Temperature, pollen, mold, and seasonal changes were found to be the most influential factors, and a simple hand-held device was invented that could be used to regulate an allergy sufferer's medication and excercise.


Abstract


   Having asthma and hay fever, the researcher has noticed the particular climatic conditions of Southeastern Virginia to be troublesome to his symptoms. He has also known other asthmatics who, like himself, needed more medication on some days than they did on others. Therefore, he has completed a five year investigation to determine if weather conditions, environmental irritants such as pollen and mold, or seasonal changes could be used to predict the severity of an asthmatic's asthma symptoms or a hay fever sufferer's symptoms in Newport News, Virginia and other climatic regions of the United States in a simple and practical way -- a study that has not previously appeared in the scientific literature.

   The researcher began by obtaining daily counts of the number of people who received asthma treatment from Southeastern Virginia, New Orleans, and New York City; constant daily weather measurements for the three locations; and pollen and mold spore counts for Southeastern Virginia. Daily surveys were then completed by asthmatics and hay fever sufferers in Southeastern Virginia who rated the severity of their symptoms each day for an entire month. Correlation and regression analyses were performed on the collected data using a statistical computer package. The results were transformed into equations that a computer program, which the researcher designed, could use to predict asthma and hay fever symptoms. From the equations, the researcher designed and built a simple pocket device that asthmatics or hay fever sufferers could use to predict allergy symptoms.

   Temperature, pollen, and mold were found to influence the asthma and hay fever survey ratings more than any other factors. Seasons and several weather factors including temperature and barometric pressure had a strong correlation with the asthma related clinic visits from Southeastern Virginia, New Orleans, and New York City. Although asthmatics accurately predicted the seasons in which their symptoms were the best and the worst, they were, on the average, unable to guess which individual weather factors influenced their symptoms the most. Formulas based upon multiple linear regression were constructed which could be used by an asthmatic and a hay fever sufferer to predict the severity of his or her asthma and hay fever symptoms, and by clinics in various climatic regions of the United States to estimate the number of people who receive asthma treatment. Finally, rather than complex mathematical formulas, a personalized asthma wheel was constructed using a computer program to easily predict asthmatic severity.


Introduction


   Since the age of seven, the researcher has had asthma and another allergy related disease, hay fever. He has known other asthmatics and hay fever sufferers who, like himself, needed more medication some days than they did on other days. The researcher wanted to predict how an asthmatic or a hay fever sufferer would react to a given set of daily weather factors, so that the proper level of medication could have been taken.

   The purpose of the experiment was to determine if weather conditions, environmental irritants such as pollen and mold, or seasonal changes could be used to predict the severity of an asthmatic's asthma symptoms or a hay fever sufferer's symptoms in Newport News, Virginia and other climatic regions of the United States in a simple and practical way. The researcher correlated the effect of the weather factors, pollen, mold, and season against the severity of a hay fever sufferer's symptoms, an asthmatic's asthma symptoms, and the number of people who received asthma treatment from several hospitals in Southeastern Virginia, New Orleans, and New York City. Comparisons between the actual and predicted correlations were conducted to determine if allergy sufferers knew which weather conditions, if any, contributed towards the severity of their symptoms. Afterwards, a computer program was created to design and build a hand-held device that could predict asthma and hay fever symptoms based upon environmental conditions. The researcher hypothesized that weather (especially temperature and barometric pressure), pollen, mold, and seasons could be used to predict the severity of an asthmatic's and a hay fever sufferer's symptoms in the form of a simple mathematical model.


Methods and Materials


Weather

   From October 1990 through September 1993, constant daily measurements of the following weather factors were obtained from the Federal Aviation Flight Service, located at the Newport News - Williamsburg International Airport, in Newport News, Virginia: high temperature (HTEMP), low temperature (LTEMP), highest wind velocity (WIND), humidity at the time of the highest temperature (HUMID), amount of precipitation (RAIN), high barometric pressure (HBAR), and low barometric pressure (LBAR). Using this data, the following were calculated: average temperature (ATEMP), daily range in temperature (RTEMP), change in temperature from the previous day (CTEMP), average barometric pressure (ABAR), daily range in barometric pressure (RBAR), and change in barometric pressure from the previous day (CBAR). Daily weather measurements (from La Guardia Airport, New York City for 1971, and from New Orleans International Airport for 1975 through 1977) consisting of temperature, barometric pressure, wind, rain, and humidity were also collected from the National Climatic Data Center in Ashville, North Carolina. A lag factor of one day was then created for each of the weather factors which gave each condition for the previous day, and is denoted by a "+1" after the weather condition (for example HTEMP+1).

Pollen and Mold

   From April 1991, through September 1993, daily pollen and mold spore counts were obtained from the Department of Public Health, Bureau of Laboratories, located in Norfolk, Virginia and the Virginia Allergy and Pulmonary Associates, located in Richmond, Virginia.

Clinics

   From October 1990, through September 1993, daily counts of the number of people who received asthma treatment (CLINICS) from Riverside Hospital and Mary Immaculate Hospital, located in Newport News, Virginia were collected and correlated with the weather factors and seasons. Asthma related clinic visits from Charity Hospital emergency room, New Orleans from 1975 through 1977, and Cumberland, Kings County, and Harlem Hospitals in New York City during 1971 were also obtained from the Center of Disease Control (Atlanta, Georgia) and Columbia University School of Public Health (New York City), respectively.

Surveys

   Surveys were constructed for the months of April 1990, 1991, 1992, and 1993, and October 1989, 1990, 1991, 1992, and 1993 and were given to known asthmatics and hay fever sufferers (Appendix A). These surveys included a calendar for the month, along with instructions on how the test subjects should have rated their asthma/hay fever symptoms each day, and spaces were included where they could have indicated which weather factors and seasons most affected his or her asthma/hay fever. Each day the test subjects were to put their asthma/hay fever rating on the correct calendar day. This data was used to find which weather factors most affected his or her asthma/hay fever symptoms and to see if the weather factor they predicted to be most influential was correct. The daily average asthma and hay fever ratings, abbreviated AVASMA and AVHAY, were found by calculating the mean rating for each day.

Statistical Approach

   Statistical computer packages were used to plot graphs and tables and to find the linear correlation coefficients and individual regression lines. A multiple regression analysis was then performed for both the daily average asthma and hay fever ratings, and the daily number of people who received asthma treatment against all of the weather factors simultaneously. The seasonal influence was also considered, and formulas were devised to predict the severity of asthma symptoms, in terms of the number of asthma related clinic visits to Riverside and Mary Immaculate Hospital (for a three year period), to Charity Hospital emergency room (for a three year period), and to Cumberland, Kings County, and Harlem Hospitals (for a one year period). The combined data was then incorporated into a self designed computer program that created an "Asthma Wheel" to be used to predict asthma symptoms. The following were formulas used to perform the correlations:


   The "Asthma Wheel" was designed to be a simple device that asthmatics and hay fever sufferers could use to predict their symptoms. The "Asthma Wheel" was sufficiently accurate to produce reliable results, yet small and light enough to be easily carried. To operate it, an asthmatic or hay fever sufferer would begin by turning the movable disc with the label, RAIN, so that the arrow points to the correct level of rain. Next, without turning the disc for rain, he or she would point the arrow labeled WIND to the correct level and continue for barometric pressure (BAR), high temperature (HTEMP), and low temperature (LTEMP). If the user was unsure of any of the values, pointing the corresponding arrow to the thick black line would represent a three year average measurement and thus compensate for the unknown value. When done, the operator would look into the "window" under the label, LTEMP. The border between the black and white regions would indicate a value ranging from zero, a bad day for asthma, to ten, a good day. By altering options in the computer program that produces the wheels, a new wheel could be created instantly that would allow for more weather conditions, metric and English measurements of weather conditions, and personalization of the device to an individual asthmatic's symptoms.


Results


Part 1: Surveys

   For the first part of the experiment (Appendix B), the researcher, using the statistical computer program MINITAB, correlated each of the weather factors with the average asthmatic survey rating (AVASMA) and the average hay fever sufferer's survey rating (AVHAY). A correlation of greater magnitude than the critical value (r) indicated a strong correlation (rejecting the claim of no significant linear correlation at an alpha=.05 level of significance).

   Many weather factors were highly correlated with the AVASMA and AVHAY, but, temperature, followed by barometric pressure, was found to influence the AVASMA and AVHAY more than any other weather factor. However, mold greatly influenced the AVASMA and AVHAY, while pollen, with one lag day, in April 1991, was found to have the highest correlation with hay fever with an extremely high correlation of -.782. It was discovered that record high levels of pollen were experienced in Southeastern Virginia for that month, and could have been responsible for the strong correlation.

   Graphs were constructed to better illustrate the findings of the correlation analysis. The first graph of each section showed both the high temperature (HTEMP) and the AVASMA rising at nearly the same rate throughout April 1990. These pointed out that the higher the temperature was in April 1990, the higher the AVASMA. This illustrated that the warmer the temperature was, the better an asthmatic felt. The second graph showed the relationship between the AVHAY and pollen with one lag day (POLLEN+1) in April 1991. The pollen count reached a record high on the seventh day of the month, and the AVHAY reached an all time low on the following day. These graphs indicated that as the pollen count rose, the AVHAY dropped, meaning that the more pollen in the air, the worse a hay fever sufferer felt.

   Next, the researcher wanted to determine if asthmatics and hay fever sufferers knew which individual weather factors most influenced their asthma. The large number of returned surveys indicated a strong interest in the topic of the relationship weather had with asthma and hay fever, considering that it took an entire month of daily recordings by over two hundred volunteers. Of the asthmatics surveyed, 64% believed that weather affected his or her asthma symptoms, while only 23% were correct in guessing which weather factor was most influential. Of the hay fever sufferers surveyed, 63% believed that weather affected his or her hay fever symptoms, while only 14% were correct in guessing which weather factor was most influential.

Part 2: Clinics

   The researcher wanted to determine if the number of asthma related clinic visits to Riverside Hospital in Newport News, Virginia, from October 1, 1990, to September 30, 1993, were being affected by similar weather factors as the AVASMA. He also wanted to find out how the results in Southeastern Virginia compared to New Orleans and New York City. In the second part of the experiment (Appendix C), correlation analysis was performed on 1096 days of clinic data from Newport News and New Orleans and on 365 days of clinic data from New York City. The critical value was much less than with the surveys due to the large sample sizes. It was found that the same weather factors (temperature and barometric pressure) were affecting the AVASMA as well as the asthma related clinic visits, but with a relatively stronger correlation in the latter.

   Of greater importance, however, was that correlations with weather in Southeastern Virginia, New Orleans, and New York City were almost identical! This proved that the findings in this experiment were not localized in one particular area or climate, but were valid in other climatic regions of the United States. The bars on the national climatic comparison graph represented the correlations that each of the tested clinics had with the weather factors. Whenever the correlations were of a strong significant value in Newport News, the correlations were the same in each of the other cities.

   The three line graphs illustrated the results of the correlations with the clinics. The first graph showed the number of asthma related clinic visits in Newport News. The next graph plotted the low temperature (LTEMP), and the last graph illustrated the barometric pressure (BAR) for the same period of time. Although the general trends of the clinic and low temperature graphs looked exactly opposite, the fact that when one was rising the other always fell accounted for the strong negative correlations. The barometric pressure graph almost exactly matched the graph for the clinics. These showed that the higher the temperature was, the fewer asthmatics needed to go to the clinics, indicating once again that the warmer the weather, the better an asthmatic felt, and that the lower the barometric pressure was, the fewer asthmatics came to the clinics, which again was most likely due to the fact that drops in barometric pressure usually brought rain which would carry off large amounts of pollen.

Part 3: Seasonal Effects

   For the third part of the experiment (Appendix D), the researcher wanted to determine if seasons also affected asthma. There was a total of 26,106 asthma related visits to the clinics in Newport News, New Orleans, and New York City during the years that data was collected. They were distributed with respect to the seasons in the following way:

 Newport News  New Orleans  New York City 
Winter
Spring
Summer
Fall
:
:
:
:
861
817
473
906
3096
2516
2392
2130
3290
2577
2130
3962

In most cases, summer was the least difficult season for asthmatics, followed by spring; fall was the worst, followed by winter. Graphs showed the number of asthma related clinic visits to each of the cities and polynomial curves of their general trends.

   The results of summer being the best season and fall being the worst were also indicated by examining the surveys of the asthmatics who responded to the question concerning seasons. On a scale of 1 - 5 where 1 = very bad, 3 = average, 5 = very good, a general overall rating of the severity of asthma symptoms for each of the four seasons was requested. The average rating for each season resulted in winter having a rating of 3.35, spring with 2.64, summer with 3.71, and fall with 2.56. This was surprising because although asthmatics were unable to predict which individual weather factor influenced them the most, it appeared that since asthmatics were correct in predicting summer as their best season and fall as their worst, they understood the seasonal affects on their asthma symptoms.

Part 4: Regression Formulas

   Combining the clinic and survey results, the researcher created several multiple regression formulas so that a clinic, an asthmatic, and a hay fever sufferer could use all of the weather factors simultaneously to most accurately predict the number of asthma related clinic visits or the severity of an asthmatic's or hay fever sufferer's symptoms (Appendix E). An asthmatic or hay fever sufferer could have entered the weather factors into their corresponding formula, evaluated the formula, and end up with a predicted survey rating (10 = good and 0 = bad) which could have been used to regulate the level of medication or limit the stress and exercise an asthmatic or hay fever sufferer could withstand on a given day.

Part 5: The Asthma Wheel

   After it was found that weather conditions could be used to accurately predict the severity of asthma and allergy symptoms, the multiple regression formulas were given to asthmatics and hay fever sufferers. However, to use the formulas, each factor needed to be known, the equations could not be customized, and a lot of mathematical calculations had to be performed by hand. A solution to these problems was to create a simple, hand-held devise that used an incorporated mathematical model to quickly predict the severity of an allergy sufferer's symptoms by using one or many weather factors. The difficulty in designing this type of device arose from the fact that the regression formulas represent a multi-dimensional structure. The researcher eventually discovered that by piling several small disks with "windows" that list varying degrees of certain weather conditions and contain an arc proportional in size to the correlation of the corresponding weather factor, a simple two-dimensional model of the regression formulas could be built that would require no arithmetic calculations by the user. Although similar to a circular slide rule in that the relationship in position of scales would yield a simple result to a complex formula, this "Asthma Wheel" used multiple scales simultaneously to represent the multi-dimensional matrix of the multiple regression formula.

   The process of building an accurate Asthma Wheel began with an original Macintosh computer program called TheAsthmaWheel. The computer program allowed allergy sufferers to enter, each day for up to a month, the severity of their allergy symptoms and the corresponding weather data in the form of a spreadsheet. Using the program, the user could choose which weather factors he or she wanted in the final Wheel. After a minimum of five days of data were entered, the program created from the data a personalized Asthma or Hay Fever Wheel that was optimized for the particular asthmatic or hay fever sufferer. The computer program could then be used to either print out a physical Wheel, try a virtual Wheel, or determine which weather factor was most influential in predicting symptom severity (Appendix F). For allergy sufferers that did not have access to a computer, the average survey ratings for a month were entered into the program, and generic Asthma Wheel was created that would give accurate predictions to the typical allergy sufferer.


The Asthma Wheel (Click for a more detailed image.)

   By using a personalized Asthma or Hay Fever Wheel, an allergy sufferer could quickly, easily, and accurately predict the severity of their symptoms, and clinics around the country could estimate the number of asthmatics requiring treatment. Graphs demonstrate the accuracy of the generic Asthma Wheel in predicting asthmatic symptoms (Appendix G).


Discussion and Conclusion


   The results confirmed that weather factors, pollen, and seasons had an effect on the severity of an asthmatic's symptoms and a hay fever sufferer's symptoms. Statistical analysis showed that high temperature and barometric pressure were the most highly correlated weather factors. Wind, which blew dust and pollen into the air; and rain and humidity, which sped up the growth of mold spores were also strongly correlated with asthma and hay fever. Pollen, mold spore counts and seasons were extremely correlated with asthma and hay fever. Similar findings were concluded in several other studies. Not shown in the literature, however, was that weather and seasons were also affecting asthma in Southeastern Virginia in ways similar to what the researcher had discovered for other climatic regions of the United States.

   The conclusions of this research were as follows: temperature was found to influence the asthma and hay fever survey ratings more than any other weather factor, while airborne irritants such as pollen and mold had an extremely high correlation with the asthma and hay fever survey ratings; although allergy sufferers could not predict which weather conditions were most influential to their symptoms, they could accurately determine which seasons were the worst for asthma and hay fever; seasons and several weather factors including temperature and barometric pressure had a strong correlation with the three years of localized asthma related clinic visits; asthma related clinic visits in Southeastern Virginia, New Orleans, and New York City all had very similar correlations with weather factors and seasonal effects; the constructed formulas based upon multiple linear regression could be used by an asthmatic and a hay fever sufferer to predict the severity of his or her asthma and hay fever symptoms, and by clinics in various climatic regions of the United States to estimate the number of people who receive asthma treatment; rather than complex mathematical formulas, a personalized "Asthma Wheel" was constructed using an original computer program to easily predict asthmatic severity; and finally, weather conditions, environmental irritants such as pollen and mold, or seasonal changes can be used, to some extent, to predict the severity of an asthmatic's asthma symptoms or a hay fever sufferer's symptoms in Newport News, Virginia and other climatic regions of the United States in a simple and practical way.



Bibliography


    Journals

  1. Derrick, E.H., 1967, "The Relation of the Seasonal Variation of Asthma to the Weather", Austr. Paediat. J., 3:82-89.

  2. Greenburg, L., 1964, "Asthma and Temperature Change", Health, 8:642-647.

  3. Hobday, J. 1973, "The Relationship Between Daily Asthma Attendance, Weather Parameters, Spore Count and Pollen Count", Aust. N.Z.J. Med., 3:552-556.

  4. Khan, A., 1977, "The Role of Air Pollution and Weather Changes in Childhood Asthma", Allergy, 39:397-400.

  5. Khot, A., 1988, "Biometeorological Triggers in Childhood Asthma", Clinical Allergy, 18:351-358.

  6. Salvaggo, J., 1970, "Relationships of Climatologic and Seasonal Factors to Outbreaks", Allergy, 45:257-265.

  7. Seabury, J., 1971, "New Orleans Asthma and Weather", Allergy, 48:96-114.

  8. Tromp, T.W., 1968, "Influence of Weather and Climate on Asthma and Bronchitis", Rev. Allergy, 22:1027-1044.


    Lungs

  9. Crocco, John, 1977, Gray's Anatomy, New York City.

  10. Lambert, Mark, 1988, The Lungs and Breathing, Englewood Cliffs, New Jersey.

  11. Marr, John, 1971, A Breath of Air and a Breath of Smoke, Philadelphia, Pennsylvania.


    Allergies

  12. Riedman, Sarah, 1978, Allergies, USA.

  13. Silverstein, Alvin, 1977, Allergies, New York City.


    Asthma

  14. Haas, Sheila, 1987, The Essential Asthma Book, New York City.

  15. Plaut, Thomas, 1988, Asthma, Amherst, Massachusetts.

  16. Sorvino, Paul, 1985, How to Become a Former Asthmatic, New York City.




Appendixes


Sample Survey
Survey Results
Clinics Results
Seasonal Effects
Regression Formulas
Screen Shots of TheAsthmaWheel Program
Accuracy of the Asthma Wheel
Appendix A
Appendix B
Appendix C
Appendix D
Appendix E
Appendix F
Appendix G

Download a limited version of TheAsthmaWheel for Macintosh computers.