are sex, age at diagnosis, or mode of presentation prognostic factors for cystic fibrosis?

Pediatric Pulmonology 3288-297 (1987)

Original Articles -

Are Sex, Age at Diagnosis, or Mode of Presentation Prognostic Factors for Cystic Fibrosis?

Irene Hudson, MSC, PhD, and P.D. Phelan, MD, FRACP

Summary. Data on 622 patients with cystic fibrosis born in Victoria, Australia from 1955 to 1980 and on 344 surviving patients in the care of a specialist clinic on June 30, 1983 were analyzed for factors associated with better survival and a less rapid progress of lung disease. Presentation with predominantly gastrointestinal symptoms, other than meconium ileus. was associated with an improved prognosis, whereas presentation with predominandy respiratory symptoms was associated with a worse prognosis. While infants diagnosed before the age of 6 months as a result of routine testing because of a family history of the disease seemed to have less rapid progress of lung disease, their ultimate survival did not seem to be better than that of patients presenting symptomatically after the newborn period. The sex of the patients did not appear to have prognostic significance. Age at diagnosis did not affect rate of progress of lung disease or survival when infants dying within 6 months of birth were excluded. There was a close association between the extent of lung disease at diagnosis and current lung disease. Failure to reverse extensive disease at diagnosis or deterioration of lung disease in the first year after diagnosis was associated with a less favourable course. Pediatr Pulmonol 1987; 3~288-297.

Key words: CF, 622 patients, 344 surviving since birth in 1955-1980; Kaplan-Meler product-limit estimates; logistic analysis; GLlM or BMDP.

INTRODUCTION

Survival of patients with cystic fibrosis (CF) has improved considerably in recent years.' This has followed the introduction of comprehensive care programs in spe- cialized centers. Early and intensive therapy of chest infections, careful attention to pancreatic enzyme re- placement and nutrition, and appreciation of psychoso- cia1 problems have all probably contributed to the better outlook. Patients with minimal acquired complications of cystic fibrosis have been recognized by standard diagnostic criteria, and these have made a further small contri- bution to improved survival figures.2

The various factors associated with improved survival are poorly understood. There is uncertainty whether the age at diagnosis is important in predicting outcome. CF is known to be a variable disorder, but the significance for long-term survival of the different modes of presentation is unclear. One of the problems has been that most studies of survival have come from major referral clinics, and the populations on which their results were based have been difficult to define.

The cystic fibrosis clinic at the Royal Children's Hos- pital, Melbourne, has been in existence since 1953 and @I 1987 Alan R. Liss, Inc.

during that period has been the only specialist clinic serving the State of Victoria, whose population has in- creased from 2.5 to 3.9 million. Around 1965, 60% of patients with CF in the state attended the clinic on a regular basis; by 1975. it was 80%, and currently 90% of diagnosed surviving children and adolescents under the age of 20 years attend the clinic. A similar percentage of older patients attends an associated clinic for adults. During the last 25 years, close collaboration has been maintained with pediatricians at other centers in the state; they have followed similar treatment programs and have willingly provided details on patients under their care. A comprehensive study of the incidence and modes of presentation of cystic fibrosis in all patients born in Victoria

From the Department of Paediatrics. University of Melbourne, and Professorial Department of Thoracic Medicine, Royal Children's Hospital, Melbourne, Australia

Received March 9. 1986; (revision) accepted for publication Septem- ber 10. 1986.

Address correspondence and reprint requests to Dr. P.D. Phelan. Department of Paediatrics. Royal Children's Hospital, Fiemington Road, Parkville, Vic. 3052. Australia.

Prognostic Factors for Cystic Fibrosis 289

the time of last contact. Survival curves for different subgroups defined by age at diagnosis, sex, and mode of presentation were compared using the log rank test.6 These same variables were used in fitting a stepwise proportional hazards model' to the survival times, with the aim of quantifying the effects of these on survival. Analyses of survival times were performed using the SPSS8 and BMDP' statistical packages on the Melbourne University VAX system.

between January 1, 1955 and December 31, 1978 and diagnosed by December 31, 1979 has been previously repo~ted.'.~ These patients provide a valuable source of information about prognostic features of the condition.

MATERIALS AND METHODS Survival Cohort and Statistical Analysis

Six hundred twenty-two patients with CF were born in Victoria. Australia between January 1, 1955 and Decem- ber 3 1 , 1980 and were diagnosed by December 3 1. 198 1. The diagnosis in all cases was based on compatible clinical features and the finding of a sweat sodium and chloride in excess of 60 mmol/L using the Gibson Cooke method of pilocarpine iontophoresis or by finding typical pathological features at autopsy. Details of date of birth, date of diagnosis, the major presenting symptom complex at dignosis, and, where applicable, the date of death were known for all but 18 patients. Ages at diagnosis were classified into those less than 6 months, 6-12 months, 12-24 months, and 24 months and over. Patients were further grouped according to the symptoms re- corded by the medical social worker at the time of diagnosis or obtained from the referring practitioner's letter or hospital medical record. The groups were: predominantly respiratory symptoms; predominantly gastrointestinal symptoms; meconium ileus; routine sweat testing of siblings; and other or unknown. Full clinical details at diagnosis and subsequently were unavailable for some patients who were not managed by the cystic fibrosis clinic.

Survival curves from birth to December 3 I , 1983 were constructed using the Kaplan-Meier product-limit estimates' of the survival distributions. The five patients dying from causes unrelated to cystic fibrosis (one drowning, one acute gastroenteritis, one subendocardial fibroelastosis, and two neontal deaths with complex con- genital anomalies) and the 15 patients about whom no current information could be obtained were censored at

Clinic Population and Statistical Methods

Three hundred forty-four patients were under the care of the cystic fibrosis clinic, Royal Children's Hospital, Melbourne and its associated clinic for adults at the Alfred Hospital, Melbourne, on June 30, 1983. They represented 90% of known living patients with CF in the state. They were graded at diagnosis and on June 30, 1983, according to the extent of lung disease using the classification shown in Table 1. 'O There were five grades of lung disease: 0 = no established disease; 1 = minimal disease; 2 = mild disease; 3 = moderate disease; and 4 = severe disease. In patients less than 7 years of age. pulmonary function tests were not used as part of the classification. Current pulmonary status of the 344 patients was dichotomized into those with good pulmonary status grades (grades 0, 1, and 2) and those with poor pulmonary status (grades 3 and 4).

We performed a cross tabulation of the numbers of patients exhibiting good pulmonary status by sex, mode of presentation, age at diagnosis grouping, and current age groupings - 0 to (but not including) 5 years; 5 years up to (but not including) 10 years; 10 years up to (but not including) 15 years; 15 years up to (but not including) 20 years; 20 years up to (but not including) 25 years; and 25 years and over. Only those siblings diagnosed before the age of 6 months were included in the analysis of the effect of the mode of presentation. Age-adjusted predic- tions of good pulmonary status in relation to sex, mode

TABLE 1-Criteria for Grades of Pulmonary Status

Grade"

~- 0 I 2 3 4 ~~ ~ ~ ~ _ _ _ _ _ _ _ _ _

Cough No persistent Persistent cough Persistent cough Persistent cough Chronic cough cough

Sputum (ml/day) None None < 10 10 -100 > I f x ) Chekt radiograph Normal Minimal strcaking Bronchial wall Patchy infiltrates, Widespread infiltrates.

thickening and cysts. moderate pronounced minimal hyperinflation hyperinflation hyperinflation

hl monary function FEV I ( '% predicted) > 80 65-80 65-80 50-65 < 50

'Grade 0 = presence of all4 criteria; grade I = any I ; grades 2-4 at least 2.

290 Hudson and Phelan

m

m

ao

0 3 4 0 0 u) 0 Y l o 1. 20 12 2. AGE (monthr)

Fig. 1. Survival curve from 6 months of age of the cohort (A), survival curve from birth of cohort, excluding all patients with meconium ileus (B), and survival curve from birth of the whole cohort (C).

TABLE 2-Deaths in the Victorian Group Under 6 Months According to Years and Modes of Presentation

Years of presentation Modes of 1955- 1961- 1966- 1971- 1976- presentat ion 60 65 70 75 80

Respiratory 15 3 2 I Gastrointestinal 2 I Meconium ileus I8 20 2 I 2 Routine sweat

testing of siblings

~~ ~.~ -~ .______.

- - - -

- - - - -

- Others and 6 I I - unknown

of presentation, and age at diagnosis groupings were then obtained by a log-linear analysis. ' I

Pulmonary status at diagnosis and its change in tht: succeeding 12 months were also related to current pulmonary status by a separate logistic analysis of the data, using either the GLIM'* or BMDP9 statistical packages on the Melbourne University VAX system. The patients were divided into two groups for analysis of the effect of change in status. The first group of 255 comprised those patients with a pulmonary status of 0, 1, or 2 at diagnosis, who remained unchanged or improved, and those with a pulmonary status of 3 or 4 who improved. The second group of 66 comprised those patients with a pulmonary status of 0, 1, or 2, whose condition deteriorated, and

those with a pulmonary status of 3 or 4 who remained unchanged or deteriorated. The possible effects of current age, mode of presentation, and sex were also incor- porated in the logistic model.

In all analyses, the significance level was taken at P < 0.05.

RESULTS Factors Associated With Better Survival

A survival curve for all patients is given in Figure 1. There was a high mortality in the first year of life. This was due to deaths within 6 months of birth of infants with meconium ileus, almost all deaths being due to surgical complications, and deaths of infants presenting within 6 months of birth with respiratory symptoms and dying soon after presentation. Infants dying in these two ways were seen mainly in the first 1 1 years of the study (Table 2). As infants dying in this manner were rare in the last 10 years of the study, a survival curve was recalculated from the age of 6 months, thereby excluding the 75 deaths at less than 6 months of age (Fig. I ) . A curve was also calculated excluding all deaths from meconium ileus.

Survival for males and females excluding deaths before the age of 6 months in infants with meconium ileus was not significantly different (P > 0.20) (Fig. 2). The exclu- sion of all patients with meconium ileus also failed to demonstrate any significant difference. Calculating sur-

Prognostic Factors for Cystlc Flbrosls 291

0

I 2 2 8

Q3 1

0.6

0.5

F 269

2 4 6 8 1 0 1 2 14 18 18 2 0 2 2 2 4 AGE (months)

o.2j . . , , . . . . , , . .

Fig. 2. Survival curves from birth of males (M) and females (F) with cystic fibrosis: patients presenting with meconium iieus and dying before the age of 6 months are excluded.

vival curves for males and females who survived beyond nosed on routine sweat testing, but it excludes patients 10 years again did not demonstrate any significant dif- presenting and dying within 6 months of birth. Appropri- ference . ate analyses showed no significant difference in the sur-

Figure 3 gives survival curves from birth according to vival of these four groups. Those diagnosed before 24 age at diagnosis. The group under 6 months of age months appeared to have a significantly worse survival includes patients with meconium ileus and siblings diag- from birth than those diagnosed after 24 months (P <


i '.

6-12 months 12-24 months

> >24 months

O f 3 months

"'i

Fig. 4. Survival curves from age at diagnosis by age at diagnosis. All patients with meconium ileus were excluded, but other patients dying within the first 6 months of life were included. Numbers are th9 age ranges and the numbers of patients in each group.

0.05). However, when the more appropriate analysis of survival from 24 months for both groups was done, there was no significant difference. If patients diagnosed and dying within 6 months of birth were included, survival of patients diagnosed before the age of 6 months was significantly worse than for all other groups (P < 0.001).

Survival from diagnosis was also calculated for the groups by age at diagnosis including deaths in the group diagnosed before 6 months but excluding all meconium ileus patients. The results are given in Figure 4. The only significant difference was that those diagnosed after 24 months of age had an improved survival in comparison to those diagnosed before 6 months (P < 0.02)-but this was due to differences in the first 12 years from diagnosis. After that time the curves crossed.

Survival curves for patients presenting with predominantly chest symptoms, with predominantly gastrointestinal symptoms, and with meconium ileus, and siblings diagnosed prior to the age of 6 months on routine sweat testing are shown in Figure 5. Patients dying under 6 months of age were excluded. Patients presenting with gastrointestinal symptoms had a significantly improved

survival compared with those presenting with predominantly respiratory symptoms (P < 0.001) or with meconium ileus (P < 0.005). The survival of patients with meconium ileus who did not succumb within 6 months of birth was not significantly different from that of patients presenting with respiratory symptoms nor from the whole group excluding those dying under 6 months of age. The survival of siblings diagnosed prior to the age of 6 months on routine sweat testing was not significantly different from that of patients presenting with predominantly gastrointestinal symptoms but was better than that for patients presenting with predominantly respiratory symptoms or with meconium ileus (P < 0.05). However, the survival of siblings diagnosed before 6 months of age was not different from those presenting with respiratory or gastrointestinal symptoms grouped together.

Cox's proportional hazards regression model7 was then used to investigate the interrelationships of the factors of mode of presentation, age at diagnosis, and sex on survival for the whole cohort excluding all deaths within 6 months of birth. Only mode of presentation was a significant prognostic factor related to survival (P < 0.001).


L. ! . -

2 4 6 8 Ib 1; I 4 16 (s 20 22 24

AGE

Fig. 5. Survival curves from age of 6 months of patients with cystic fibrosis according to mode of presentation. G-I, predominantly gastrointestinal symptoms; S, diagnosis by age 6 months because of affected older sibling; RESP, predominantly respiratory symptoms; MI, meconium ileus. Numbers are number of patients with each presentation.

TABLE 3-Number of Patients in the Clinic Group by Current Pulmonary Status According to Current Age

Current age (years)

0-5 5- 1 0 10-15 15-20 20-25 > 25 Total

~~

Grades of pulmonary status 0 I 2 3 4

~ _____

46 27 5 I 1 32 23 13 4 4 19 33 19 17 10 6 h 16 9 I 1 I 2 I 1 4 7 0 3 5 3 6

I04 04 69 38 39

TABLE 4-Odds Ratio Calculations of the Likelihood of Good Pulmonary Status According to Mode of Presentation

Presentation ratio interval P value Odds 95% confidence

.-

Respiratory versus 0.415 (0.254-0.732) <0.001 gastrointestinal, meconium ileus, and sibling < 6/ I2

meconium ileus and sibling < 6/ 12

sibling <6/12

Gastrointestinal versus 2.540 (1.200-5.400) <0.01

Meconium ileus versus 1.420 (0.220-5.991) >0.43

TABLE 5-Probability of Good Current Pulmonary Status According to Change in Pulmonary Status in the 12 Months From Diagnosis

Patients who

improved status or poor status or Current age maintained gcmd status deteriorated (years) (probability) (probability)

0-5 0.987 0.959 5- 10 0.929 0.802 10-15 0.777 0.518 15-20 0.648 0.362 20-25 0.653 0.368 > 25 0.570 0.291

Patients who failed to improve

In agreement with earlier log rank analysis, those presenting with predominantly respiratory symptoms or with meconium ileus had the worst survival. Survival was not significantly related to a patient's sex or age at diagnosis (P > 0.05).

Factors Associated With Less Rapid Progress of Lung Disease

Log-linear analysis of the clinic group showed a highly significant negative relationship between the likelihood of good pulmonary status and current age grouping (,y2


c d d m 0.6 .

0.5

0.2 i- 1 -

0-5 5 -10 l0-15 15-20 20-25 25*

AGE GROUPS

Fig. 6. Probability (ordinate) of good pulmonary status in clinic patients according to current age (abscissa) and modes of presentation. Resp, predominantly respiratory symptomr; GI, predominantly gastrointestinal symptoms; MI, meconlum ileus; SIB, diagnosis by age 6 months because of affected older sibling.

= 48.96, df = 5, P < 0.001) (Table 3). After adjusting for possible differences in current age between the modes of presentation, logit analysis showed a significant association between current pulmonary status and presenting mode (x2 = 14.3, df = 3, P < 0.005) (Fig. 6). Patients presenting with respiratory symptoms had the least likelihood of exhibiting good pulmonary status for all age groupings (Fig. 6). The same factors of current age and mode of presentation were found to be significantly related to good pulmonary status by a stepwise logistic regression of the data in its generic form, i.e., without stratification of patients into current age and age at diagnosis groupings.I3 Odds ratio and the corresponding 95% confidence intervalI2 were derived for various groupings according to mode of presentation (Table 4). In this type of analysis, if the confidence limits include 1 .O the value of P is greater than 0.05. We demonstrated that those presenting with predominantly respiratory symptoms were least likely to have good pulmonary status (P < 0.001) and those presenting with predominantly gastrointestinal symptoms were less likely to have good pulmonary status than those presenting with meconium ileus or diagnosed as siblings aged under 6 months of age (P < 0.01). The latter two classifications were not significantly different from each other. Age at diagnosis and sex did not significantly affect the likelihood of good pulmonary status in the clinic group after adjustment for the effects of current age and mode.

A logit analysis showed that, after adjusting for the significant effect of current age, (x2 = 54.43, df = 5, P < 0.0005), current pulmonary status was significantly related to pulmonary status at diagnosis (x = 17.64, df = 4, P < 0.005). A similar analysis showed that patients who either improved pulmonary status or maintained good status in the 12 months after diagnosis had a significantly greater likelihood of good current pulmonary status than those who lost status or failed to improve poor status (x2 = 11.849, df = 1, P < 0.0005) (Table 5 ) .

DISCUSSION

The study identified a number of factors that were associated with improved survival and less rapid progress of lung disease from cystic fibrosis. As it is a progressive disease, it was not surprising that there was a close association between current age and current pulmonary status. After allowing for this association, presentation with predominantly gastrointestinal symptoms indicated a better prognosis than presentation with predominantly respiratory symptoms. The extent of current lung disease was significantly related to lung disease at diagnosis. Success in reversal of lung disease present at diagnosis was associated with an improved prognosis, whereas failure to improve extensive lung disease present at diagnosis or deterioration in lung disease in the first year after diagnosis was associated with a less favourable


course. The sex of the patients did not appear to have essentially similar therapy. In the present report the pos- prognostic significance. Age at diagnosis did not affect sible effect of changes in therapy have not been consid- either rate of progress of lung disease or survival, if ered, as they would have added considerably to its com- infants presenting and dying within 6 months of birth plexity and not affected its major conclusions. were excluded. Such infants were seen mainly in the The classification at the time of presentation into either early years of the study. While routine diagnosis before predominantly respiratory or predominantly gastrointes- the age of 6 months because of a family history seemed tinal symptoms had been used in the clinic for many to be associated with less rapid progress of lung disease, years. Many children have both gastrointestinal and res- it did not appear to improve ultimate survival in compar- piratory symptoms; there have been a few in whom it has ison with patients whose diagnosis was the result of been difficult to determine which symptom complex is presentation with symptoms. the predominant one. While there is some evidence to

These conclusions have particular validity because they suggest that those patients with respiratory symptoms and were based on a comprehensive ascertained population. some residual pancreatic function have milder disease, It is believed that almost all patients diagnosed in Victoria so that their inclusion in patients presenting with respi- between between 1955 and 1980 were It is ratory symptoms could obscure some important factors, unlikely that substantial numbers of patients were dying only about 5 % of patients in this study had residual unrecognized, at least since the early 1960s. All deaths pancreatic function. There is no ready explanation of within the first 28 days of life in Victoria were specifi- why this percentage is lower than the 15% reported from cally investigated, and over 90% had an autopsy. Over the Hospital for Sick Children, Toronto, except that the 70% of infants dying unexpectedly after the age of 28 latter clinic is a referral center, and the age distribution days had an autopsy performed by one experienced pe- of its patients (with the peak group in the teenage years) diatric pathologist, who specifically looked for evidence suggests bias in referral.” The small number with resid- of cystic fibrosis. Most infants and children with compli- ual pancreatic function is unlikely to have a major effect cated disease were looked after in two pediatric centers on the conclusions of the study. in Melbourne, and, excluding children dying unexpect- The classification of the extent of lung disease is a edly, most deaths in the pediatric age group occurred in simple one, developed in this clinic.” It has been found these two institutions. The possibility that large numbers to be highly reproducible between three independent ob- of children died in the state with unrecognized cystic servers. Its simplicity gives it considerable advantages fibrosis, as could be inferred from the hypothesis of over more complex grading systems, and it readily al- Warwick,I4 had been syifically investigated and found lows patients to be dichotomized into those with none or to be highly unlikely.’ The incidence of cystic fibrosis minor lung disease and those with extensive disease. in the state based on case finding had not changed signif- Survival analysis was recalculated from 6 months of icantly during the period of the study,4 and it was vir- age to allow for the apparent change in the early course tually identical to that of 1 in 2,400 live births found of cystic fibrosis. As very few deaths now occur under elsewhere in Australia by newborn screening using serum the age of 6 months, excluding these seemed to provide immunoreactive trypsin, a test with high sensitivity and a cohort more appropriate for comparison with patients high specificity.I6 While in the early years of the study a presenting in recent time. In the analysis of the effect of significant number of patients were looked after by other the patient’s sex on survival, those presenting with me- pediatricians, they followed similar treatment programs conium ileus and dying under 6 months were excluded, to those of the clinic. More recently, almost all patients in case sex had some effect on surgical outcome. How- in the state had been looked after by the one group of ever, if these 43 patients were included, the conclusion pediatricians and internists. Therefore the conclusions of was unaltered. the study are probably more valid for the whole cystic The failure to demonstrate a significant difference in fibrosis population than those based on referral clinics, survival or in current pulmonary status between males which may see selected populations. and females was unexpected, as it is widely believed that

These data are not directly comparable with those pro- females, particularly after the age of 10 years, have a vided by the patient registry of the USA Cystic Fibrosis poorer survival. Certainly in the present survey the sur- Foundation. Its survival curves are based on patients seen vival rate for males was better than that for females, but over 5- or 10-year periods,” whereas the present study this did not reach statistical significance. This may have covers all patients diagnosed in a defined population over been because numbers were too small, or, alternatively, a 26-year period. because some undetermined factor is improving the out-

During the time covered by the study, there were major look for females. The difference in survival between changes in therapy that will have affected survival. How- males and females in the United States was less in 1978 ever, during a given period all patients would have had than in 1969.” A study from the United Kingdom also


failed to show any difference in survival between males and females.20

It had been hoped that early diagnosis, particularly before symptoms developed, would improve prognosis. Shwachman et aI2' reported optimistically on a group of 130 patients diagnosed under 3 months of age. Robinson and Norman2' found no difference in survival of patients diagnosed before and after 12 months. In the present study, those diagnosed before 6 months of age had the worst outcome, but this observation was due to patients seen in the early years who presented with and died from respiratory disease before 6 months of age. Whether survival was calculated from birth, which may be more appropriate, as the risk of death from CF is present from birth, or from date of diagnosis did not affect this conclusion.

Orenstein et al", studying 16 sibling pairs, found that when comparison was undertaken at age 7 years, the younger diagnosed sibling had less lung disease. How- ever, almost all younger siblings had some lung disease, and the workers were unable to comment on long-term survival. There was conflicting evidence in our study about the value of routine diagnosis of siblings before 6 months of age. While children diagnosed in this way seemed to have less current lung disease than those presenting with respiratory symptoms, the ultimate outcome appeared to be no different from that of other symptomatically presenting patients. These data should give caution to those who suggest that routine newborn screening is likely to produce a major improvement in outlook for the disease. Much more data are required before the value of newborn screening is proven.

Patients presenting with predominantly respiratory symptoms seem to have a more serious form of the disease than those presenting with predominantly gastrointestinal symptoms. This infers that they had an early onset of respiratory disease, and, once established, it was more difficult to control. These findings cannot be directly compared with those reported from the Hospital for Sick Children, Toronto, where the prognosis of patients presenting with gastrointestinal symptoms only was no different from that of patients with respiratory symptoms only.I8 The data on patients with meconium ileus also produced some conflicting results. Their survival was no different from other patients' but, again, like routinely diagnosed siblings, they seemed to have less rapid progress of lung disease than those presenting with respiratory disease. It is impossible to determine whether it was the earlier diagnosis or some intrinsic difference in the cystic fibrosis that contributed to this variation. Better management during the postoperative period in patients presenting in the last 10-15 years may have delayed the establishment of lower respiratory infection.

fants with meconium ileus who survived surgery, but the USA Cystic Fibrosis Foundation Registry figures show no difference in survival."

The extent of irreversible lung disease at the time of diagnosis has previously been reported as being of prognostic significance.2' Data from the present study support this conclusion.

ACKNOWLEDGMENTS

The assistance of Jean Allan and Helen Saunders in the preparation of the data on which this study is based is gratefully acknowledged.

REFERENCES

I . Phelan PD, Allan JL, Landau LI. Barnes GL. Improved survival o f patients with cystic fibrosis. Med J Aust. 1979; I :261-263.

2. Stern RC, Boat TF. Dtxrshuk CF. Tucker AS, Miller RB, Mat- thews LW. Cystic fibrosis diagnosed after age 13. Twenty five teenage and adult patients including three asymptomatic men. Ann Intern Med. 1977; 87:188-191.

3. Danks DM, Allan J. Anderson CM. A genetic study of fibrocys- tic disease of the pancreas. Ann Hum Genet. 1965; 28:323-356.

4 . Allan JL, Robbie M. Phclan PD, Danks DM. The incidence and presentation of cystic fibrosis in Victoria, 1955- 1978. Aust Pac- diatr J. 1980; 16270-273.

5. Kaplan E. Meier P. Nonparamctric estimation from incomplete observations. J Am Stat Assoc. 1958; 53:457-481.

6. Breslow N . A generalized Kruskal-Wallis test for comparing k samples subject to unequal patterns o f censorship. Biometrika.

7. Cox DR. Regression models and life-tables. J R Stat S w (Series B). 1972; 34: 187-220.

8. Hull CH. Nic NH. SPSS Update 7-9. New Procedures and Facilities for Release 7-9. New York: McGraw-Hill. 1981:205- 219.

9. Dixon WJ ct al. BMDP Statistical Software. revised edition. Berkeley: University of California Press. 1983:575-594.

10. Holzer F, Olinsky A. Phelan PD. Variability o f airway hypcr- reactivity and allergy in cystic fibrosis. Arch Dis Child. 1981; 56:455-459.

1 1 . Cox DR. The Analysis of Binary Data. London: Methuen. 1970:30-42.

12. Baker RJ, Nelder JA. The GLIM System Release 3. Oxford: Numerical Algorithms Group. 1978:2-4.

13. Lane PW. Nelder JA. Analysis o f covariance and standardization as instances of prediction. Biomctrics. 1982; 38:613-621.

14. Warwick WJ. The natural history of cystic fibrosis. In: Warwich WJ, ed. 100 Years of Cystic Fibrosis: Collected Papers. Minne- apolis: University o f Minnesota. 1982: 13-26.

15. Danks DM, Allan JL. Phelan PD, Chapman C. Mutations at more than one locus may be involved in cystic fibrosis-evidence based on first cousin data and direct counting of cases. Am J Hum Genet. 1983; 35:838-844.

16. Wilcken B. Towns SJ, Mellis CM. Diagnostic delay in cystic fibrosis: Lessons from newborn screening. Arch Dis Child. 1983. 58:863-866.

1970; 57~579-594.

17. Cystic Fibrosis Foundation. Report of the Patient Registry, Vol. 1978. Rockville. Maryland. 1980:8-9.

18. Gaskin K. Gurwitz D. Durie P. Corev M. Levison H. Fnrstner 1 - Robinson and Norman22 found impaired survival in in- 2 ~-


Report of 130 patients diagnosed under 3 months of age over a 20 year period. Pediatrics. 1970; 46:335-343.

22. Robinson MJ, Norman AP. Life tables for cystic fibrosis. Arch

23. Orenstein DM, Boat TF, Stern RC. Tucker AS. Charnock EL. Matthews LW, Dwrshuk CF. The effect of early diagnosis and treatment in cystic fibrosis. Am J Dis Child. 1977; 131:973-975.

G. Improved respiratory prognosis in patients with cystic fibrosis with normal fat absorption. J Pediatr. 1982; 100:857-862.

19. Gurwitz D, Corey M. Francis PWJ. Crozier D, Levison H. Perspcctivcs in cystic fibrosis. Pediatr Clin North Am. 1979; Dis Child. 1975; 50:962-965. 26:603-615.

20. Wilmott RW. Tyson SL, Dinwiddie R, Matthew DJ. Survival rates in cystic fibrosis. Arch Dis Child. 1983; 58:835-836.

? I . Shwachman H. Redmond A. Khaw KT. Studies in cystic fibrosis.

are sex, age at diagnosis, or mode of presentation prognostic factors for cystic fibrosis?

Documents