mothers' experience from a Finnish case-referent study of birth defects.

mothers' experience NURMINEN KURP K. Office employment, work with video display terminals, and course of preg nancy: Rcferenee mothers' experience from a Finnish case-referent study of birth defects. Scand J Work Environ Health 14 (1988) 293-298. In an examination of the possible harm ful effects of work in an office environment and the use of a video display terminal (VDT) on the course of pregnancy, the experience of I 475 reference mothers from a Finnish case-referent study of birth defects was analyzed. The study was based on the national Register of Congenital Malformations, whose data were supplemented with special interviews on mothers' work conditions. The group which worked in an office environment con sisted of 239 women, of whom 60 had worked with video display terminals; 805 mothers had not worked in an office. Only mothers who had worked during most of their pregnancy and who had a singleton birth were included; hence 431 women were excluded from the analysis. The information on threatened abortion, length of gestation, birthweight, placental weight, and maternal blood pressure was analyzed. Office work involved no elevated risk of threatened abortion when compared with nonoffice work, and among the VDT users the proportion with symptoms related to an impending early termination of preg nancy was similar to that of other office workers. No unfavorable effects on the length of gestation were observed between the compared groups, and there were no differences in the birthweight of the babies when adjustment was made for gestational age or the other aspects under consideration. Thus the results did not suggest that office employment or work with video display terminals would be harm ful for pregnancy.

Concerns about work -relat ed reproductive problems have spread to the office environment largely du e to the publi city given to the fear th at wor k with a video displ ay terminal (VDT) might res ult in birth defects and spo nta neo us abortions. Although it app ear s that the a nxiety concerning the use of such terminals is unfounded when the evidence fro m publ ished studies is con sidered (see referen ces 2, 3,4,6,10,20), much uncertai nty still pre vails among office personnel. Spontaneous abortions among employees in an airline office have been linked to heavy mental work load (14). In view of the large number of women in office work and the widespread anxiety in many coun tries, even descriptive result s from systematic studies are probably of mu ch social value today.
We ha ve previously publi shed result s fro m a caseref erent study of the effect of VDT work during pregnan cy on the risk of con genital defects (8), a nd some preliminar y data on office employment, the use of video displa y term inals , and th e reference mothers' pregnanc y o utcomes (9). In the present study , we have an alyzed the available reference mothers' exper ience more extensively. Specificall y, we ha ve studied th e occur rence of thre atened abo rtion , length o f gestation , 1 Institut e of Occupational Health , Helsinki, Finland .
Reprint requests to: Ms T Nurminen, Department of Epidemiology and Biostat istics, Institute of Occupa tio nal Health , Topeliuksenkatu 41 a A, SF-00250 Helsink i, Finlan d .
birthweight , plac ental weight , and changes in the mother 's blood pr essure during the pregnan cies of a group of women who had done office work with or without video display terminals during th eir pre gnancies and of a group of nonoffice wo rkers.

Subjects and methods
The primary dat a was obtained from 1 475 time-and area-matched case-referent pairs from a Finnish nationwide study on occupational expo sure s and co ngenital mal formations (7). We obtained the ba sic information about the case mothers and their referents from th e Finnish Register of Congenital Mal formations . The pairs of moth ers were interviewed by mean s of standa rd pro cedures of the Register about the pa rticulars o f previou s pregnancies, diseases, consumption of dru gs and alcohol , smoking habi ts, etc (16,17). The Register data were supplemented with interviews on the mothers' work conditions and various exposures related to occupational or leisure-time act ivities, and the interviews were carried out by two trained person s from the Finnish Institute of Occupational Health (5); 96 070 of the mothers participated.
The rules of the Register stated that for each moth er a mat ched reference mother was to be selected whose delivery immediately preceded that of the case mother in th e same maternity welfare dist rict (16). Ho wever, no stillbirths or mal formations were to be included in the refere nce series. Thu s all women whose pregnancy ended in a delivery o f a child not stillborn or malfo rmed in the data collection period of the primar y study between 1976 and 1982 were candidates for the reference series. The actual referents can be considered to represent a random sample from these " noncase" mothers stratified by year of birth and maternity welfare district. The 1 475 reference mothers' pregnancies formed the base of the present study. Table I shows the accumulation of the study pregnancies by year of birth.
Th ere were no specific questions about VDT work in the questionnaire of the exposure interv iew. Therefore we used occupational titles to distinguish mothers in office work with potential VDT exposure. Among the 1 475 mothers there were 255 women with occupational titles indicating potential VDT use (table 2). An indu strial hygienist and two experts in occupational medicine assessed the VDT exposures of these mothers using descriptions of the mother' s ordinary workday, which had been recorded in detail during the interview.

294
In all, 64 mothers were assessed as exposed. VDT use was explicitly mentioned in the work description of 43 moth ers. VDT work, though not actuall y ment ioned in the description, was deemed obvious for 18 mothe rs.
For three mothers such work was considered pr obable. Exposure time was categorized as follows: an average of 4 h or more per workda y (33 mother s), less than 4 h but at least 1 h per workday (10 mothers), or less than 1 h per work day (21 mothers). The exposure of four moth ers had taken place only aft er the first trimester of pregnancy.
Of the 255 mothers with pot ential VDT exposure, 242 had worked during most of their pregnancy. The thir teen mothers who had worked only tempora rily were excluded from the analysis. Of these, seven had stopped working because their employment had terminated , one had left for maternity leave, one for sick leave (becau se of hypertension), and four had other reasons or the reason was missing. Two of the excluded mothers had worked with video display terminal s. Thr ee of the 242 working mothers had twin birth s and were also excluded. Afte r the exclusions, the final group with office work consisted of 239 mothers. Of these, 60 had used video display termin als, and 179 were not VDT users.
Of the remaining 1 220 interview forms belonging to moth ers not potentially VDT-exposed, we selected a random sample of 50 for inspection . Non e of the mothers in this sample appeared to have used video display terminals. In all, 270 mothers had not worked durin g their pregnancy, and 135 had worked only temporarily. Of the latter gro up 67 had sto pped working because their employment had terminat ed, II had left for maternity leave, 21 had been given sick leave (seven because of threatened abortion and one due to hypertension), an d the remainin g 36 had ot her reaso ns or the reason was missing. These moth ers were not included in the final analysis. A total of 815 moth ers had been in nonoffice work during mo st o f their pregnancy. Ho wever nine of these mothers had given birth to twins , and in the case of one woman inform ation on the number of fetuses was missing. These LO women were also excluded from the stud y. Thu s the final group in non office wor k consisted of 805 mot hers.
We obtained information pertaining to the out come of the pregnancies under study through the Register, from the standard questionn aire, and from the antenatal record of the maternity health care center. The question of whether the mother had had a threatened abortion during her pregnancy was asked in the Register interview by a midwife . The mother was also asked to specify if she had had vaginal bleeding or lower abdomi nal pain, or both , and when these symptoms had occurred. For 75 lIlo of the mothers who had had bleeding with or without pain durin g pregnancy, the symptom s had occur red in the first trim ester. For 71 alo of the moth ers who had had onl y lower abdominal pain during pregnancy, the symptoms had occurred in the th ird trimester. The duration of gestat ion was calcu-lated from the firs t da y of th e last normal menstru al period and was expre ssed in com pleted days. An increase of at least 20 mm Hg (3 kP a) in the mean arterial blood pressur e between the mother's first visit to the maternity health care cent er and her last visit before delivery was con sidered to indicate pr egnan cyinduced hypertension (1) . The mean arterial blood pressure was calculated as on e-third of the pulse pressure plus the diastolic pressure (ie, one-third of the systolic pressure plus two-thirds of the diastolic pressure).
Estimates and confidence limits were ca lculated fo r the rate ra tios accor di ng to th e Miettinen-Nurminen chi-square funct io n procedu re for un str atif ied and st ratified data ( 13). The mean s were compared with the t-t est , and the med ians with the Wilcoxon test.
Po wer calc ulatio ns for single 2 x 2 table s were carri ed out with method s deri ved by Miettinen (11, (5 ) with a significance level of a = 0.025 and a one-sided test.
For threatened abortion, we used adjusted odds ratio estimates fro m logistic regression analy ses as the estimates of the rat e ratios becau se the outcome was rare. In the com parisons, we controlled for the pot ential confo unding effects of age, o utcome of previou s pr egnancies, history of menstru al irregularity , alcohol consumption, smo king, and int a ke of dru gs.
We der ived the differences between the gro ups with respect to the mean length o f gesta tion for term pregnancies from regression procedures after excluding preterm births (length of gestation less than 259 d) and prolonged pregnancies (gestation 294 d or longer). To adj ust for confounding, we used the same vari ables as for threatened abortion.

Results
O f th e 239 mothers who had don e o ffice wor k, 8.4 0/ 0 had had sympto ms of threat ened abor tion (bleeding wit h or without lower abdominal pain ) dur ing pregnancy as against 9.8 % of the 805 mothers in nonoffice wor k. Th e cr ude ra te rati o was 0.8 with a 95 % confidence interval (95 % CI) of 0.5-1.3, and the adjusted rate ratio derived from the logistic regression analysis was 0.8 (95 OJo CI 0.5-1.4). Thirteen percen t of the mothers in office work and 10.9 % of the mothers in non office work had had only lo wer abdominal pa in du ring their pregnancy. Th e cru de rate ratio was 1.2 (95 OJo C I 0. 8-1.7), and the adj usted rat e rat io was 1.3 (95 % CI 0. 8-2.0 ).
Twe nty percent of the 60 mothers who had worke d with video d isplay term inals had had eit her bleedi ng or pain or bot h du ring pregnancy as again st 21.8 OJo of the 179 mothers in office work who had not used such terminals. The unadjusted rate ratio was 0.9 with a 95 % con fidence interval of 0.5-1 .6. The power to detect a two fold risk was 89 OJo . Controlling for t he effects of the mothers' previou s pregnancies d id not essentia lly cha nge the estimates (ta ble 3), nor did th ey vary upon contro l for other poss ible confounders. T he moth ers in office and nonoffice work had similar pro portions of preterm , term , and prolonged pr egnancie s (ta ble 4) . Th e mean du ration of the term pregna ncies was 280 (SD 7) d for both groups , and the adju sted difference in th e means betw een these groups was 0.4 [standard err or (SE) 0.6] d. Th e mean length of the term pregn an cies of the mothers in VDT work was 282 (SO 6) d . Th e adjusted difference in the mean s bet ween the VOT and no n-VOT users in office wo rk was 2.9 (SE 1.2) d .
The occur rence o f preterm birth was 3.0 OJo among th e moth ers in office work and 2.5 % among those in nono ffice work (rate rati o 1.2, 95 % CI 0.5-2.7) (ta ble 4). Th e power to detect a twofold risk was 50 % .
The re were no statistically significant differen ces in the birthweight of the babies born to mothers in the study gro ups when adjustment was made for gestational age (tables 4 and 5). In a ll, 21.5 % of th e mothers in office work had babies with a birthweight sma ller tha n the lower quartile birthweight of the babies bo rn to mothers in nonoffice wor k in the sa me gestatio na l age gro up (ta ble 4). For mothers in VOT wor k the cor respond ing proportion of small babie s was 25.4 % when their ba bies' birth weights were compared with those of the babi es born to mo thers who worked in an o ffice environment but did not use video display terminals (table 5). The power to detect a twofold risk o f the VOT user s giving birth to a small baby was 92 % .
On th e average, t he placental weight s did not differ between th e groups, the mean pla cent al weight bein g 613 (SO 122) g for th e mothers in VOT work, 619 (SO 122) g for th e non-VOT users in office wo rk , and 614 (SO 128) g for th e mothers in nonoffice wor k. a Smaller tha n t he lower quartile birthweight of t he babies born to moth ers in nonoffice wo rk in th e same gest ational age gro up. b There were two mo t hers in office work and 15 moth ers in nonoff ice work with missing data on the length of gesta tion . • Smaller t han the lowe r quartile bi rth weigh t of the babies born to mothe rs wit h no VOT work in the same gest ati ona l age group . Infants with a birthweigh t below the 10th , the 5th , or the 2.5th percentiles are considered small for the ir gestational age. The group in office work was too small for thes e c rit eria. b The re was one mo ther in VOT wor k and one mother in non-VOT work wi th mi ssing data on the length of gest ation .
The changes in maternal blood pressure during pregnancy were similar in the study groups. For systolic blood press ure the median of all three gro ups was 120 mm Hg (16 kPa) for both the first visit to the maternit y hea lth care center and the last visit before delivery . The median diastolic blood pre ssure was 70 mm Hg (9 kPa) for all three groups on the first visit and 75 mm Hg (10 kPa) for the VDT users and 80 mm Hg 296 (11 kPa) for the two other groups on th e last visit. The mean blood pressure of 8.3 % of the mothers in VDT work , 5.0 % of the non -VDT users in office work , and 8.2 % of the mothers in nonoffice work increased during pregn an cy to a level suggesting hypert ension . Th e po wer to detect a twofold greater risk o f hyperten sion amo ng the mothers in office work th an among the mothers in nonoffice work was 93 010, whereas the cor-responding power of detection between mothers with and those without VOT work in an office environment was 31 0,70 .

Discussion
It can be con sidered unconventional to form the base population of a study from subjects that have primarily been chosen as referents for cases in another study. We became interested in making use of the re ference mothers' experience because both relevant information on the course of their pregnancy and the associated exposure data were readily available. Our re ference subjects repr esented "noncases" of malformations and stillbirt hs , and there was no indication that th e midwives who selected the referents would have used any other criteria for exclusion. The reference information corresponded well with data provided by other Finnish studies (18,19). In fact, if the midwives would have chosen child ren healthier in other respects, then the reference series would have shown diluted ab solute rates of occurrence of, eg, preterm birth or babies with low birthweights. However, becau se the midwives hardly even knew the mothers' expo sures during pregnancy, an y differential selectivity according to exposure was improbable. Th erefore, the relative risk estimates can be considered to be unbiased. (See refe rence 12, p 68.) Nevertheless, inefficiency ca n be considerable when rare outcomes are studied and a small-sized reference series is used, as was the case in our original study with one-to-one matching . For example, our data did not allow us to study the occurrence of babies with very low birthweight s (tables 4 and 5). In addition th e VOT gro up was so small that it wo uld have been difficult to detect minor differences in the rates of po ssible complications.
In our data, the symptoms inquired about in connection with the question co ncern ing threatened abortion in the interview seemed to relate to two different phenomena. Bleeding with or without lower abdominal pain had mostly occurred during early pregnancy and thus wa s probably a manifestation of threatened abortion . Lower abdominal pain only had mainl y occur red late in pre gnancy, but its relation to threatened preterm labor is apparently less clear. We found no difference in the risk of threatened abortion between the mothers in office and nonoffice work. For VOT wo r k the proportion of mo th er s with symptoms was sim ilar to th at of mothers in other office work, but the VOT group was too small for us to study the se symptoms separately.
The duration of pregnancy did not differ between the mothers in office and nonoffice work, but the term pregnancies of mothers in VOT work were slightly longer than those of the mothers with non-VOT office wor k. However the data were too limited to permit a proper statistical evaluation of the occurrence of pre-term deli very. The birthweight s of the babies in the three groups were similar when adjustment was made for gestational age, but it was not po ssible to study the occurrence of premature babies (birthweight less than 2500 g) because their number was too small.
The systolic and diastolic blood pre ssures during pregnancy were equal in the compared groups, and the proportions of women with changes in blood pressure suggesting pregnancy-induced hypertension were similar.
A working group appointed by the World Health Or ganization to evaluate the effect of video display terminals on workers' health concluded that the studies it examined provided no evidence of adverse effects of video display terminals on pregnancy (20). Neither have the studies published after the appearance of thi s evaluation supported the suggestion that VOT use might increase the risk of congenital defects or spontaneous abortion (3,4,10). The analysis of our data also produced no indication of work-related reproductive problems in office environments in general or in VOT work in particular.