Letters to the editor

tends to have a predictable effect on smokingcessation. However, from a social perspective,it is fairly certain (to within relatively narrowconfidence limits) how many cases of lungcancer or cardiac events will be avoided if10000 persons stop smoking. What this il-lustrates is that social benefit is more certainthan individual benefit. In other words, if weare concerned with a social perspective, it ismore consistent (when we are making healthdecisions on a social level) to use a discountrate which is lower than the simple aggregateof individual discount rates.

TOM MARSHALL,Northamptonshire Health Authonty,

Northampton NNJ SDN.

1 West R. Discounting the future: influence of theeconomic model. J Epidemiol CommunityHealth 1996; 50: 239-44.

Reply

I welcome the opportunity to explain someof the "misapprehensions" and "mis-understandings" detailed by Dr Marshall.

1. Economics for some may be the "sci-ence of choice" but the Oxford English Dic-tionary is a little more specific and describesit as the, "practical and theoretical science ofproduction and distribution of wealth".

2. I had not intended in the introductoryparagraph to claim that economic theory ap-plied only to paid employment. I simply citedan observation in a widely used economictext' that the history of economics is prettymuch the history of paid employment.

3. In Dr Marshall's third point (rewrittenand expanded since I wrote my original reply)there seems to be a misunderstanding aboutthe role of theory in science. In the cycle of"systematic enquiry, conjecture, and re-futation", do we not accept the theory thatbest fits (explains) the observations, questionthe theory that fails "to fit all observations",and reject the theory when a new one seemsto fit the observations better?-and thus didexit the "flat earth society".

4. Economics has certainly contributed toour understanding of community behaviourwithin the environment of paid employment.The suggestion was that other social sciencesmight contribute more (before or) outwiththe environment of paid employment.

5. I understand that the theoretical basis ofneoclassical economics is the individual-andin my naive view that may be the originof the limitations of the neoclassical theory,because it overdoes the extrapolation fromthe individual to the community (the OxfordEnglish Dictionary describes "economy" as theadministration of resources of a community).That is the substance of my thesis.

6. Dr Marshall's discussion seems to dem-onstrate the dominance of "money" and the"perfect market" in economic theory.

7. The investment illustration appears toshow the importance for economics of "fin-ance". Since this non-health care examplehas been introduced, may I suggest that itshows simply that the purchase of a large"good" like a house is spread over time: thereis no need to "ignore" or "assume".

8. It is most certainly not my perceptionthat people who do not produce are notvalued. The inclusion of the "human capital"model and the decision rules of "rationaleconomic man" was to demonstrate that, asa moderately "compassionate caring society",we do not restrict health care planning choicesto middle aged men in gainful employment.2

9. I 'recognised the possibilities of es-timating implicit social value in health caredecision making some 20 years ago,3 and theidea was very rapidly adopted in the earlydays of health economics. Before that cost-benefit analyses (mostly in the field of trans-port economics) tended to enter an empiricalor arbitrary social value and then comparecost-benefit ratios, but it seemed to me to bemore logical to turn it round and estimate an"unknown" from an established equilibriumor status quo.

In conclusion, the closing sentence of DrMarshall's letter that "social benefit is morecertain than individual benefit" seems to en-dorse my thesis that "society's future is morecertain than an individual's future". Thismight suggest that Dr Marshall may be be-ginning to drop some of the trappings ofneoclassical economics-sovereignty of theindividual and an assumption that society'schoice may be predicted by the individual's.Let us hope that other economists will followand continue to question the wisdom of dis-counting the future.

ROBERT WESTUniversity of Wales College of Medicine,

Cardiff CF4 4XN

1 Samuelson P, Nordhaus W. Economics. McGrawHill, New York 1985.

2 West RR. Market forces. BMJ 1988;296:711-12.3 Buxton MJ, West RR. Cost benefit analysis of

long-term haemodialysis for chronic renal fail-ure. BMJ 1975; ii:376-79.

Ionizing radiation andoffspring sex ratio

SIR-Dickinson et al' report a high sex ratio(proportion male) of offspring born to menexposed to ionizing radiation at Sellafield.These authors were undecided in their in-terpretation of this. I should like to suggestthat their data represent strong further evi-dence that ionizing radiation directly (butadmittedly weakly) affects offspring sex ratio.

If ionizing radiation induced sex linkedlethal mutations, one would expect irradiatedfathers to produce an excess of sons, andirradiated mothers to produce an excess ofdaughters.2 In accordance with this theo-rising, Cox3 reviewed the literature and re-ported that in 12 data sets describing theoffspring of irradiated fathers, there were 10in which sex ratios were raised in contrast withthose ofcontrols: and in 10 sets describing theoffspring of irradiated mothers, 9 reportedlowered offspring sex ratios (X2 = 9, p<0.OO5).So there is good prior evidence that ionizingradiation has the effects described. The dataof Dickinson et al' conform with this gen-eralisation. The sex ratio of offspring born totheir most heavily irradiated men (201 sons,144 daughters) is highly significantly differentfrom that of the offspring of the women whohad received preconceptual radiation (454sons, 461 daughters) (X2=7.l, p<0.01). Sothe data of Dickinson et all add to the alreadyexisting strong evidence that ionizing ra-diation induces sex linked lethal mutations inman which are reflected in the offspring sexratios of irradiated parents.The effects of ionizing radiation are quite

different from those ofnon-ionizing radiationin this regard. For instance, men working inhigh voltage power stations or substationsreportedly produce an excess of daughters,"possibly as a consequence of hormonally me-diated effects of such radiation. In general,

illness and adverse industrial/occupational ex-posures to men are associated with low tes-tosterone and/or high gonadotrophin levels,and with low offspring sex ratios (presumablyas a consequence of the hormonal dis-turbance).7 Thus-as far as I know-ionizingradiation is the only reproductive hazardwhich causes men to sire an excess of sons.

WILLIAM H JAMESThe Galton Laboratory, University Colege London,

Wolfson House, London NWI 2HE

1 Dickinson HO, Parker L, Binks K, Wakeford R,Smith J. The sex ratio of children in relationto patemal preconceptual radiation dose. JEpidemiol Community Health 1996;50:645-52.

2 Schull WJ, Neel JV. Radiation and the sex ratioin man. Science 1958;128:343-48.

3 Cox DW. An investigation of possible geneticdamage in the offspring of women receivingmultiple diagnostic pelvic X-rays. Am J HumGenet 1964;16:214-30.

4 Nordstrom S, Birke E, Gustavsson L. Re-productive hazards among workers at highvoltage substations. Bioelectromagnetics 1983;4:91-101.

5 Knave B, Gamberale F, Bersgstrom EE, Birke E,Iregren A, Kolmodin-Hedman B et al. Longterm exposure to electric fields: a cross-sec-tional epidemiologic investigation of oc-cupationally exposed workers in high voltagesubstations. ScandJ7 Work Environ Health 1979;5:115-25.

6 Mubarak AAS, Mubarak AAS. Does high voltageelectricity have an effect on the sex distributionof offspring? Hum Reprod 1996;11:230-31.

7 James WH. Occupations associated with low off-spring sex ratios. Am J Ind Med 1994;25:607-8.

Reply

We thank Dr James for his letter and, inparticular, for reminding us of the theorywhich Schull and Neel put forward in 1958that irradiated fathers and mothers would beexpected to produce excesses of sons anddaughters respectively.' As knowledge ofgen-etics progressed, Schull and Neel re-considered this theory,2" their most recentpostulate being that while maternal exposurewould, in principle, reduce the sex ratio, it isdifficult to predict what effect irradiation offathers might have since, "With the re-cognition that one X chromosome is inactivein the somatic cells of women (the Lyon-ization phenomenon), it became clear thatsex linked mutations induced in males wereunlikely to have a dominant lethal effect infemales."4James also brings to our attention the study

by Cox, who claims that 10 out of 12 data setsdescribing the offspring of irradiated fathersshowed an increased sex ratio in exposedgroups.5 All the studies of paternal irradiationto which Cox referred were summarised intable 4 of our paper,6 where we gave theminimum detail necessary for critical as-sessment. Such a review must inevitably bedominated by Schull and Neel's study of53 691 children born during 1948-55 to thesurvivors of the atomic bombs dropped onHiroshima and Nagasaki.' Subsequent toCox's review, Schull et al updated this workby further considering the 47 624 childrenborn during 1956-62 to the Japanese atomicbomb survivors,2 the relevant results againbeing summarised in table 4 of our study.6Many of the smaller studies appeared to clas-sify children as "exposed" if they were born,rather than conceived, after paternal ir-radiation.7'0 Several were questionnairebased studies9'-2 whose response rate rangedfrom 37%9 to 64%."The studies of the atomic bomb survivors'2

considered children conceived at least 18

340

Book reviews

months after the bombings and thus excludedthe effects of radiation doses received by fath-ers during the period of spermatogenesis-ieimmediately before conception-whereas we

reported an association with the dose es-timated to have been received by fathers in the90 days before conception. Referring again totable 4 of our paper,6 it will be seen that onlyin the studies of chronic exposure is it likelythat the father was irradiated in the immediatepreconceptional period and that these studiesyield equivocal results.

James' post-hoc comparison ofthe childrenborn to mothers and fathers who had receiveda preconceptional radiation dose is invalid:our analysis considered an a priori hypothesiscomparing both these groups independentlywith the remainder of the population, afterallowing for the effect of year and paternalemployment at Sellafield. Since the sex ratioof children ofmothers with a preconceptionalradiation dose was 0.985 (95% CI: 0.865,1.121), not significantly different from thatof children of mothers without such a dose,1.046 (95% CI: 0.961, 1.140), our data pro-vide no evidence to support the hypothesisthat irradiated mothers are more likely toproduce girls.One of the caveats concerning our results

which we would like to reiterate is our concernabout the estimation from annual dose sum-

maries of radiation doses received during the90 days before conception-a process whichinevitably leads to dose misclassification.While random misclassification generallybiases the results towards the null hypothesis,in this instance it is quite possible that thereis differential dose misclassification and henceuncertainty about the direction of bias.Measurement error is clearly an importantissue in this regard and is an area we are

actively exploring.To summarise, studies of the possible as-

sociation between parental preconceptionalirradiation and an altered sex ratio do not yetsatisfy the Bradford Hill criteria for inferringa causal relationship."3We would like to take this opportunity to

thank Mr Les Scott, Dosimetry Data Man-ager at Westlakes Research Institute, for es-

timating the preconceptional radiation doseson which the study was based.

H 0 DICKINSONL PARKER

University of Newcastle,Department of Child Health,

Royal Wctona Infirmary,Newcastle upon Tyne NE) 4LP

K BINKSDepartment of Occupational

Health and Medical Statistics,Westlakes Research Institute,

Whitehaven,Cumbria CA24 3LN

R WAKEFORDBritish Nuclear Fuels,

Risley, Wamngton,Cheshire WA3 6AS

1 Schull WJ, Neel JV. Radiation and the sex ratioin man. Science 1958;128:343-48.

2 Schull WJ, Neel JV, Hashizume A. Some furtherobservations on the sex ratio among infantsborn to survivors of the atomic bombings ofHiroshima and Nagasaki. Am Hum Genet1966;8:328-38

3 Schull WJ, Otake M, Neel JV. Genetic effectsof the atomic bombs: a reappraisal. Science1981;213:1220-27.

4 Neel JV, Schull WJ, Awa AA, Satoh C, Kato H,

OtakeM et al. The children ofparents exposedto atomic bombs: estimates of the geneticdoubling dose of radiation for humans. AmHum Genet 1990;46:1053-72.

5 Cox DW. An investigation of possible geneticdamage in the offspring of women receivingmultiple diagnostic pelvic X-rays. Am HumGenet 1964;16:214-30.

6 Dickinson HO, Parker L, Binks K, WakefordR, Smith J. The sex ratio of children in relationto paternal preconceptional radiation dose: astudy in Cumbria, northern England. Jf Epi-demiol Community Health 1996;50:645-52.

7 Scholte PJL, Sobels FH. Sex ratio shifts amongprogeny from patients having received thera-peutic X-radiation. Hum Genet 1964;16:26-37.

8 Muller C, Rericha V, Kubit M. Zur Frage dergenetischen Auswirkung der ionisierendenStrahlung bei Joachimstaler Bergleuten. Zen-tralbl Gynakol 1962;84:558-60

9 Turpin R, Lejeune J, Rethorpe M-O. Etude dela descendance de sujets traites par radio-therapies pelvienne. Acta Genet (Basel) 1956;6:204-16

10 Kitabatake T. Sterility, stillbirth, infant death,and sex ratio of offsprings of X-ray workers.Nagoya J Med Sci 1960;2:226-37

11 Macht SH, Lawrence PS. National survey ofcongenital malformations resulting from ex-posure to Roentgen radiation. Am Jf Roentgen1955;73:442-66.

12 Tanaka K, Ohkura K. Evidence for geneticeffects of radiation in offspring of radiologicaltechnicians. JpnJ Hum Genet 1958;3:135-45

13 Hill AB. The environment and disease: as-sociation or causation? Proc R Soc Med 1965;58:295-300.

NOTICES

11th Conference of the European HealthPsychology Society, 3-5 September, 1997,Bordeaux, France. For further informationcontact: 11th EHPS, Congress Rive Droite,23 rue Baudrimont, 33100 Bordeaux, France.Tel: +33 05 56 32 82 29. Fax: +33 05 5632 79 53.

International Congress on BiomedicalPeer Review and Global Com-munications, 17-21 September 1997,Prague, Czech Republic. For further in-formation, contact: Annette Flanagin, Work-shop on Peer Review, 515 N State St,Chicago, IL 60610, USA; fax: +312 4645824; e-mail: jama-peer@ama&enr;assn.org.

BOOK REVIEWS

Lecture notes on epidemiology andpublic health medicine. 4th ed. RichardFarmer, David Miller, and Ross Lawrenson(Pp 288; £13.95) Oxford: Blackwell Science,1996 ISBN 0-86542-611-2.

The fourth edition of this well establishedtext bears little relation to the thin tome thatI used at university in the early 80s. The bookhas been expanded and is well laid out withgood illustrative diagrams and extensive use

of bullet points. The latest edition has beenexcessively updated and includes a new chap-ter on health targets with a particular em-

phasis on the priority conditions identified inThe Health of the Nation. Many of the latestchanges in the Health Service are described,although the pace of change is such thatthe most recent mergers, of district healthauthorities and family health service au-

thorities, have not been included.In general the book is an excellent in-

troduction to the subject and any criticismsare fairly minor. In the chapter on surveymethods, it was sometimes not clear whetherthe term "bias" was being used in a generalor technical sense with a definition only ap-

pearing towards the end of the chapter. The

discussion of this important concept could

have been expanded within one section ratherthan spread through subsequent chapters. Indiscussing Health Service structure, there isa heavy emphasis on describing the HealthService in England, the other countries in theUnited Kingdom meriting only one sentence.One diagram is erroneously labelled as il-lustrating the Health Service in the UnitedKingdom when, in fact, it only shows thestructure in England. A further chapter givingsome general international perspectives inhealth service organisation could also havebeen usefully added.

Notwithstanding these comments, thebook should be a recommended initial textfor medical students, and serve as useful sum-mary of the subject for clinicians wishing toupdate their knowledge. Postgraduate stu-dents commencing study in public healthmight also find the book useful.

STEPHEN KISELYLecturer in Public Health Medicine,

Manchester University

Adult mortality in Latin America. Ian MTimaeus, Juan Chackiel, and Lado Ruzicka.(Pp 367; £45.00). Oxford: Clarendon Press,1996. ISBN 0-19-828994-4

Latin America, as other world regions, ischanging its demographic and epidemio-logical pictures. The adult population is grow-ing and it is essential to analyse health con-ditions of this group. This volume is focusedon the mortality of that segment of the popu-lation. The book originated from the FirstScientific Seminar of the Adult MortalityCommittee held in Chile in 1991. In spite ofthe old information on which the papers arebased, the diversity of subjects, and diverseauthor approaches, it is a good summary ofthe burdens and contrasting situations char-acterising the changes in this region.The book has four parts. The first presents

a comprehensive overview ofthe book contentand an expert's paper with a complete de-scription of trends and demographic char-acteristics of Latin America. The four paperscomprising the second part concentrate onthe different methods of studying adult mor-tality and on the limitations of data sourcesin underdeveloped environments. Part threeconcerns itself with current changes in theepidemiological patterns in selected coun-tries. The corresponding analysis is a re-minder of the complexity of the transitionphenomena in countries with vulnerable eco-nomies and increasing social conflicts. Thenegative effects of global crisis on publichealth programmes and the big contrastswithin the countries, are important con-siderations made by the authors. The lastsection is the longest and is concerned withspecific causes of death and their prevention.The quality of the papers is variable-it is

worth mentioning, however, the remarkablepaper on maternal mortality by Rajs. The re-duction in cardiovascular mortality seen insome countries is difficult to explain because ofthe scarcity ofeffective preventive interventionand health promotion programmes. The ex-cessive deaths due to all kinds of violence arestriking, even more when it is considered thatviolence ranks first in morbidity and mortalityin several countries. Some of the conclusionsarrived at by the authors in this section arefascinating and challenging, particularly thoseby Frejka and Atkin on abortion.

This book is a valuable contribution to thesubject ofadult health in Latin America. How-ever, several questions are still unanswered. 1.

341