Non relationship of climatologic factors and painful sickle cell anemia crisis

J Chron Dis Vol. 39, No. 2, pp. 121-126, 1986 Printed in Great Britain. All rights reserved

Copyright 0

NON RELATIONSHIP OF CLIMATOLOGIC AND PAINFUL SICKLE CELL ANEMIA

0021-9681/86 $3.00 + 0.00 1986 Pergamon Press Ltd

FACTORS CRISIS

COREYM. SLOVIS, J. DAVID TALLEYand ROBERT B. PITTS Department of Medicine (Division of General Medicine), Emory University School of Medicine and Grady Memorial Hospital, Atlanta, GA 30335, U.S.A. (Received in revised form 10 June 1985)

Abstract-Conflicting results have arisen from studies concerning the correlation (if any) between climatological changes and the frequency of painful episodes in the sickle cell population. During a 13 month period records of 71 patients with hemoglobin genotypes SS or SC were reviewed. Data analysis failed to reveal an association between the frequency of painful sickle cell crisis and a number of weather and environmental variables. We were unable to demonstrate relationships between the climatologic factors of temperature, humidity, carbon monoxide level and precipitation in the frequency of 362 pain crises in 71 sickle cell patients during a 13 month period.

INTRODUCTION CONFLICTING results have arisen from studies concerning the existence of a correlation between climatological changes and the incidence of painful episodes in the sickle cell population. Redwood [l] reported a correlation between low temperatures and an increased frequency of painful episodes in Jamaica. Ibrahim [2] reported an increased frequency in Kuwait during periods of extreme heat or cold and during periods of high humidity. An increased incidence of pain episodes during the winter season in Buffalo, New York was reported by Anjad [3]. Diggs [4,5] noted an increased number of complaints when the weather was cold and damp or hot and dry in Memphis, Tennessee. A relationship between pain crises and increasing humidity in Ghana was noted by Konotey-Ahulu [6]. However, Seeler [7] failed to show any seasonality in Chicago. The purpose of this study was to examine the relationship between the incidence of painful episodes and climatological changes in Atlanta, Georgia. METHODS

Selection of patients A list of names was compiled of all the sickle cell patients with painful crises (genotype SS or SC) who were seen at Grady Memorial Hospital’s Medical and Pediatric Emergency Clinics during a 13 month period. The following criteria were used for selection: (1) the patient lived in the Atlanta area from 1 March 1980 to 31 March 1981; (2) the patient received all his medical care at Grady during this span of time; and (3) all the patient’s emergency clinic visits with complaint of pain were considered valid by the attending physicians and nursing staff. A painful crisis was defined as any painful episode typical of that of sickle cell disease and unexplained on any other basis. The records of 87 patients Reprint requests should be addressed to: Corey M. Slovis, M.D., Associate Professor of Medicine, Director, Emergency Medical Services, Grady Memorial Hospital, 80 Butler Street, SE., Atlanta, GA 30335, U.S.A. 121

COREY M. SLOWS et al.

122

were reviewed. Based on the above criteria, 71 patients who ranged in age from 6 to 38 years were identified. Sixty-one of these patients had hemoglobin SS and 10 had SC. Sixteen patients were excluded, 14 of these secondary to a prior history of drug seeking behavior, one secondary to non-resident status and another due to use of more than one hospital. Collection of climatological and carbon monoxide data

Monthly summaries were obtained from the Environmental Data and Information Service of the National Oceanic and Atmospheric Administration. These summaries gave 3 hour interval accounts of the temperature, relative humidity and precipitation from 1 March 1980 to 31 March 1981. The carbon monoxide data was obtained from the Georgia Air Quality Control Bureau. The data gave daily CO (carbon monoxide) levels from four sites in the city and its metropolitan area. Organization of data base

From the medical records of 71 patients, every emergency clinic admission form was reviewed to log the dates each patient was in the clinic with a diagnosis of sickle cell pain crisis. We arbitrarily set the length of a crisis at 4 days so that a patient that made multiple visits to the clinic in a 4-day period with the same complaints was logged as a single crisis on the first day of that 4 day period. Thus, the 362 sickle cell related visits made by the 71 patients during the 13 months of the study were logged as only 308 crises. We then tabulated, with computer assistance, the number of crises, the maximum and minimum temperature, the maximum and average relative humidity (average for the eight 3-hour readings), the amount of precipitation and the CO (carbon monoxide) levels for each of the 396 days of the study. Computation of incidence rates

All rates were computed with the following formula: Number of Crises on Days in Question ’ = (No. of Pts) (No. of Eligible Days)

OR

Number of Crises (71) (No. Eligible Days)

[Eligible days are defined as those days for which a specific number of days with rainfall]. We also computed the incidence rates on the days following example, we compared the rate on the days following rainfall to These one day lag computations were an attempt to reveal a different variables may have on the instigation of a sickle cell

event occurred, i.e. total each of the criteria. For the rate on all other days. delayed effect any of the pain crisis.

Statistical analysis

For each of the computed rates the approximate standard error was computed with the following formula: SE=

fl J N (where N = the number of patients and P is as defined above). It should be noted that this formula assumes a constant episodic rate for each patient. Thus, statistical analysis was performed assuming a Poisson distribution of events. This assumption is not the case for some of the patients (for example; a patient having a crisis during a high humidity day may be more likely to have other crises in the future on days of high humidity). Nominal significance testing was performed by evaluating the frequency of patient visits against each weather variable. The presence or absence of significance was determined by checking for an overlap in frequency of visits ( f 2 SD).

Climatologic Factors and Painful Sickle Cell Crisis

123

n I* MAR

8

11

APR

MAY

a JUN

J”,

s

*

*

I

a

AUG

SEP

DC1

NW

OEC

*I

0

*

JAN

FEE

MAR

MONTH

FIG. 1. Relationship of number of painful crises to average temperature on a monthly basis.

RESULTS

A 13 month account of the number of episodes and the average temperature for each month is given in Fig. 1. No correlation between the mean temperature for the month and the frequency of painful episodes is evident. Since the humidity fluctuates widely in a day’s time, monthly averages are meaningless and thus not tabulated. Table 1 compares days having measurable rainfall with all other days and shows no significant difference in the rates. Likewise, no significant rate change is seen on “cool rainy days” (maximum temperature of 65 and measurable rainfall) as portrayed in Table 2. Tables 3 and 4 show the rates for different intervals of average and maximum humidity. Again, none of the rate differences are significant. Similarly, as shown in Tables 5 and 6, TAI~~_E I.RELATIONSHIP0~ srcKL~CELL cmls TO *Aysw~H RAINFALL

RAINFALLcoMpARm

TO mys WITHOUT

One day lag Number

rain

on

eligible

day

days

Number

eligible days

Rate*

+ 2 SD

II0

0.0 I04

+_ 0.0024

81

110

0.0104

286

0.0112

+ 0.0014

226

285

0.01 I2 + 0.0014

81

*No

Number

episodes Rain No

Number

227

statistically

significant

crises

difference

(p

next

Rate*

g 0.05).

TABLE 2. RELATIONSHIPOF SICKLECELL CRISIS TO COOL RAINY DAYS* MMPARED

TO ALL OTHER DAYS

One Number

Cool All

rainv other

tNo

Number

eligible days

Ratet

+ 2 SD

32

0.0145

+ 0.0050

28

364

0.0106;

0.0012

279

33

hays

*Maximum

215

temperature

statistically

Number

episodes davs

$65

significant

with

+ SD It 0.0024

lag

Number on

eligible

day

days

Ratet

32

0.0123

+ 0.0046

363

0.0108

;0.0012

crises next

day

f

2 SD

rainfall.

difference

(p

g 0.05).

TABLE 3. RELATIONSHIPOF PAINFULCRISIS TO MAXIMUM DAILY HUMIDITY One Number

Number

Maximum

Number

eligible

humidity

crises

days

Rate’

I

0.0423

MHc40

3

40&MH<50

crises

5

6

< 60

26

28

0.0131

33

38

0.0122

70 6 MH

< 80

48

60

0.01

80 6 MH

< 90

66

89

72

102

50 g MH 60gMH<70

90$MH
statistically

55 significant

0.01

72 difference

(p

f

2 SD

+ 0.0488 17 f

0.0104

next

day

lag

Number on

eligible

day

days

Rate*

I

0.0141

5 0.0242

0.0141

+O.OI

I 6

6

+0.0052

22

28

f

f

2 SD

0.01 I I + 0.0048

0.0042

40

37

0.0152

rt 0.0048

I3 k 0.0032

46

60

0.0108

_+ 0.0032

0.0104

+ 0.0026

71

a9

0.01 I2 f

0.0099

+ 0.0024

16

102

0.0105

5 0.0024

0.0108

+ 0.0030

45

72

0.0088

+ 0.0026

~$0.05).

0.0026

lb

COREY M. SLOWS et al.

124

TABLE 4. RELATIONSHIPOF PAINFUL CRISIS -ro AVERAGEDALLYHUM~JITY one day lag Average humidity = AH

Number crises

eligible days

3 32 35 124 56 33 25

31 46 166 76 49 27

*No statistically

significant

difference

crises on next day

Rate* f 2 SD

I

AH<30 30$AH<40 40 z$ AH < 50 50 g AH < 70 70 6 AH < 80 80gAH<90 90~AH
Number

Number

Number

0.0423 0.0145 0.0107 0.0105 0.0107 0.0095 0.0130

+_0.0488 * 0.0052 k 0.0036 k 0.0018 f 0.0028 + 0.0034 f 0.0052

eligible days

Rate* + 2 SD

I

I 33 41 128 57 37 I2

0.0141 0.0150 0.0128 0.0107 0.0106 0.0106 0.0063

31 45 I66 76 49 27

f 0.0282 + 0.0052 f 0.0040 *0.0019 rt 0.0028 f 0.0034 + 0.0036

(p Q 0.05).

TABLE 5. RELATIONSHIP OF PAINFUL CRISIS TO MAXWJM DAILY TEMPERATURE One day lag

Max. temp. = MX MX<30 30sMX<40 4OsMX<50 5O&MX<70 70QMXt85 85&MX
Number crises

Number days

Rate* + 2 SD

Number crises on next day

0 6 23 I05 80 89 5

I IO 27 126 102 I21 9

0.0085 + 0.0070 0.0120 f 0.0050 0.0117 + 0.0022 0.01 I I + 0.0024 0.0104 f 0.0022 0.0078 It 0.0070

I I 26 104 79 85 5

significant

Number eligible days I IO 27 126 102 120 9

Rate’ +- 2 SD 0.0141 0.0099 0.0126 0.01 I6 0.0109 0.0100 0.0078

f + f + f + f

0.0282 0.0074 0.0054 0.0022 0.0024 0.0022 0.0070

(p 6 0.05).

difference

TABLE 6. RELATIONSHIP OF PAINFUL CRISIS TO MINIMUMDAILY TEMPERATURE One day lag

Min. temp. = MT MT< =I0 lO80 ‘No statistically

Number crises 0 4 47 II3 50 35 59 0

significant

Number eliaible days

-

2 5 52 I35 59 60 82 difference

Rate* + 2 SD

Number crises on next day

0.0113 0.0127 0.01 I8 0.01 I9 0.0082 0.0101

I (p s 0.05).

TABLE 7. RELATIONSHIP OF PMNFUL CRY

~0.0102 f 0.0038 It 0.0022 + 0.0034 f 0.0028 iO.0026 -

Number

eligible days

0 3 56 103 52 29 64 0

TO AVERAGE DALY

Rate’ + SD

2

-

5 52 I35 58 60 82 I

0.0085 + 0.0152 + 0.0107 * 0.0126 + 0.0068 + 0.01 IO k -

0.0098 0.0040 0.0021 0.0036 0.0026 0.0028

CARBON MONOXIDE LEVEL

One dav lae Number euisodes Elevated CO levels days All other days *No statistically

significant

Number eligible davs

I2 I96 difference

9 387

Rate* + 2 SD

Number crises on next dav

0.0188 f 0.0108 0.0108 f 0.0012

8 299

Number eligible Davs 9 386

Rate* + 2 SD 0.0125 + 0.0088 0.0109 + 0.0012

(p g 0.05).

we failed to demonstrate any correlation between daily maximum or minimum temperature and the frequency of painful episodes. We also could not show any correlation between elevated carbon monoxide levels and the incidence of painful episodes (Table 7). Moreover, it should be noted that there is no statistically significant difference between any of the same day rates and their counterpart one day lag rates. DISCUSSION

The frequency of sickle cell pain crises were examined in relation to a number of weather variables. We were unable to demonstrate relationships between the climatologic factors of temperature, humidity, carbon monoxide level and precipitation in the frequency of 308 pain crises experienced by 71 sickle cell patients during the 13 month period of 1 March 1980 and 31 March 1981.

Climatologic Factors and Painful Sickle Cell Crisis

125

There are a number of biological considerations that would seem to make such relationships plausible. Cooler weather causes a redistribution of circulating blood volume away from more superficial organs to those more internal. Rubenstein [8] demonstrated that increased blood viscosity occurred when blood from five patients with SS disease was cooled from 37 to 30°C in vitro. He reasoned that the increased viscosity causes local tissue hypoxia, resulting in an increased amount of sickling in susceptible areas. Hot weather could also be postulated to cause adverse blood viscosity changes. One could reason that elevations in temperature might result in dehydration with subsequent decreased blood volume. This could cause capillary stasis and tissue hypoxia thus increasing the likelihood of sickling. Several factors might account for the discrepancy between the above explanations and the results obtained from our study. These can be divided into two separate categories: environmental and methodical. It is possible that only much colder weather (such as occurs in Buffalo, New York during the winter months) significantly affects the likelihood of a sickle cell patient developing a pain crises [3]. This is an unlikely explanation however, because both Redwood [l] and Ibrahim [2] detected an association in climates as mild or milder than Atlanta’s. It is also possible that variables measured in terms of regional area do not reflect the local environment of the patient. Additionally, other environmental factors not measured could be more influential in initiating a pain crisis than those we selected. We doubt these latter two explanations, though they are possible. It is also possible that the relative contribution of the variables we chose and their relationship to the onset of the pain would be significant if analyzed for “lag times” of more than three days. The methodical design of our study could also potentially explain our contradictory results as compared to those cited in the literature. It is possible that some of our 71 patients received their care either at home for minor painful crises or that they utilized another hospital during major painful episodes. Follow-up interviews have shown that the majority of the patients sought hospital based care for crises they considered moderate to severe; however, they attempted to treat minor pain at home with fluids and analgesics. The number of mild crises treated at home varied significantly among patients. Numbers of mild crises and their reported relationship to temperature and rainfall were too diverse to draw any conclusions. It is possible that environmntal variables could possibly play a role in the frequency of mild crises, but our study was not designed to evaluate this. We believe that utilization of another facility is unlikely because the majority of our patients are indigent and Grady Memorial Hospital is the sole provider of their care within metropolitan Atlanta. Follow-up interviews supported the use of Grady as the sole provider of health care. Another compelling question is whether we successfully excluded patients who visited the emergency room in order to receive narcotics but were not experiencing pain crises. If we failed to do so, this could have resulted in our minimizing or obscuring any true climatological effect. This question can unfortunately not be answered definitively as there is no available objective test to “prove” that a sickle cell pain crisis is or is not occurring. Of interest is that one patient who was initially excluded for drug seeking behavior subsequently died of a narcotic overdose. Additionally, only one patient originally included in the data analysis for the crises group is now being treated in our emergency department as a “drug seeker”. Thus, 70 of the original 71 patients included in this report are still considered reliable historians by independent physicians and nurses. In trying to explain why our results differ with the previously published studies we carefully reviewed the previously published studies and found significant methodological errors and omissions. Redwood et al. in Jamaica retrospectively reviewed casualty clinic charts and hospital records to assess the frequency of moderate and severe crises [ 11.Sickle cell clinic notes were reviewed to assess the number of mild crises treated at home. The only variables analyzed were temperature and rainfall expressed as mean monthly values. These investigators utilized predefined criteria to classify patients as in crisis. Ibrahim in Kuwait retrospectively reviewed hospital admissions from a 5 year period

126

COREYM.

SLOWS

et al.

[2]. The only statistical analysis performed was a graphic representation of the mean number of patients admitted per month plotted against monthly means of maximum temperature, minimum temperature and mean maximum humidity. All environmental data was obtained from weather charts. Amjad and co-investigators in Buffalo retrospectively reviewed hospital admission data from a 5 year period [3]. Their inclusion criteria was acute pain as the “sole or main reason” for admission. Monthly averages for admissions were analyzed for a relationship to mean averages of humidity, temperature and relative humidity. Diggs and Flowers in Memphis longitudinally studied 30 families by contacting each family in person or by phone at least once per week for 2 years [5]. Their results are reported as generalized observations without statistical analysis in terms of “cold and wet” and “hot and dry” weather parameters. There was no stated inclusion or exclusion criteria, or description of how the study patients were identified or what weather parameters were followed. The brief report by Konotey-Ahulu which relates “atmospheric water vapour” content to musculoskeletal pain crisis is based on data from an unknown number of patients who had genotypes SS, SA, SThal or SC [6]. No notation is made concerning the percentage of patients in each category. Highs and lows in vapour pressure are used to explain why some patients seem to have only two crises per year. No statistical methods, experimental design, numbers of visits or percentage of patients with genotype SA are provided. In a letter by Seeler, the monthly number of children less than 15 years of age who registered in a sickle cell clinic was used to conclude that cold weather was not a factor in Chicago, Illinois. No inclusion criteria, weather variables, or statistical analysis is provided [7]. Although many difficulties are inherent in retrospective heterodemic studies such as ours, we feel that the analysis of the additional environmental factors chosen as well as the attempt to exclude patients not meeting the study design make our results more tenable than those of others. The statistical analysis using standard deviations and “lag times” are additional features not seen previously. It is also possible that the relative contribution of the variables we chose and their relationship to the onset of the pain would be significant if analyzed for “lag times” of more than 3 days. In summary, we were unable to detect a relationship between visits to the emergency room for painful sickle cell crises and several climatological factors, including temperature extremes during a thirteen month observation period. The relevance of our results show that hospitals which provide care to significant numbers of sickle cell patients do not need to anticipate major changes in utilization of emergency departments or an increase in hospital admissions during periods of changing climate. Additional studies in other cities would be helpful in assessing the conflicting information available regarding the risks of a pain crisis and other environmental factors. Acknowledgements-The

authors wish to express their gratitude to Dr David Simon for his help and opinions. REFERENCES

1. Redwood AM, Williams EN, Des01 P, Sergeant GR: Climate and painful crisis in sickle-cell disease in Jamaica. Br Med J 1: 66-68, 1976 2. Ibrahim AS: Relationship between meteorological changes and occurrence of painful sickle cell crises in Kuwait. Trans R Sot Trop Med Hyg 74: 159-161, 1980 3. Amjad H, Bannerman RM, Judisch JM: Sickling pain and season. Br Med J 2: 54, 1974 4. Diggs LW: Sickle cell crises. Am J Clin Path 44: 1-19, 1965 5. Diggs LW, Flowers E: Sickle cell anemia in the home environment. Clin Pediatr 10: 697-700, 1971 6. Konotey-Ahulu FID: Sicklemic human hygrometers. Lancet 1: 1003-1004, 1965 7. Seeler RA: Non-seasonality of sickle cell crisis. Lancet 2: 743, 1973 8. Rubenstein E: Studies on the relationship of temperature to sickle cell anemia. Am J Med 30: 95--98, 1961