Effect of bone density evaluation on hormone replacement therapy prescription

JOURNAL OF THE CLIMACTERIC & POSTMENOPAUSE

ELSEVIER

Maturitas 24 (1996) 57-61

Effect of bone density evaluation on hormone replacement therapy prescription S. Rozenberg”,“, “Interdisciplinur~

Group

M. Kroll”,

on Osteoporosis,

‘Biostatistic

Unit,

Institrrt

J. Vandromme”,

M. Paesmansb, H. Ham”

Free

Unicersities of Brussels (VUB-ULB). Sf Peter Hospital, Brussels. Belgium Bordet, Free Unirrersities of Brussels (VUB-ULBJ. Brussek.

rue Haute

322. B- 1000

Belgitmz

Received 12 September 1995; revised 11 January 1996: accepted 17 January 1996

Abstract Ohjectirzs: This study evaluates whether Bone Mineral Density (BMD) results influence HRT prescription. Methods: Successive charts of 29 postmenopausal women were summarised. For each chart, 3 ‘simulated cases’ were created by modifying the BMD result (based on the Z-score) in order to have 4 groups with the same clinical story but a wide range of BMD values (Group I = Z-score > 0, Group II = Z-score between 0 and - 1, Group III = Z-score between - 1 and - 2 and Group IV = Z-score < - 2). The obtained cases were presented to 10 gynaecologists who were asked whether HRT should be prescribed. The gynaecologists were not aware of the above-mentioned manipulation. Results: The overall treatment rate was 74.2(X1, ranging from 65% for women with the highest BMD (Group I), 73% for Group II, 79% for Group III and 80% for Group IV, i.e. women with the lowest BMD (Friedman analysis of variance; chi-square 17.2; P < 0.001). In approximately a third of the patients (11/29). there was agreement for initiation of therapy, regardless of the BMD. Most of these women presented other indications and no contra-indications for therapy. The prescription frequency of the 10 gynaecologists varied between 63% and 87%; Cochran Q Statistic 39.2; P < 0.0001). For some physicians, a trend to increase prescription was observed in relation to the BMD result, but a statistical difference could only be reached for one physician (P < 0.05). Furthermore, for some physicians no modification whatsoever could be observed. Conclusions: BMD appears to be a determinant factor for HRT prescription in only a limited proportion of the patients and a small number of the physicians. From an epidemiological point of view, BMD measurements may be useful in order to help deciding women to start HRT, especially those who are reluctant or to those who present relative contra-indications, provided that their physicians are aware of the usefulness of these investigations. Keywords:

Osteoporosis;

Bone mineral

density; Oestrogen

replacement

therapy

1. Introduction

* Corresponding author. Tel.: + 322 535 34 08; fax: + 32 2 537 59 26.

037%5122/96/$15.00 0 1996 Elsevier PII

SO378-5122(95)01002-X

Science

Women in developed countries have a life expectancy of approximately 30 years beyond their menopause. Hormone replacement therapy

Ireland Ltd. All rights reserved

(HRT) may improve their quality of life by reducing menopausal symptoms. In addition, HRT may reduce mortality and morbidity by decreasing the risk of developing osteoporosis [l-4]. Recent consensus conferences have stated that HRT is the primary prevention for postmenopausal osteoporosis [4]. HRT should therefore be proposed to women at increased risk, i.e. women with low bone mass [3,5]. Few studies, however, have assessed the attitude of physicians towards HRT for the prevention of osteoporosis. The present study was designed to evaluate, in a group of gynaecologists, whether the patient’s Bone Mineral Density (BMD) influences HRT prescription.

2. Material

and methods

2.1. The uses

Twenty nine successive charts of out-patients, which attended a menopause clinic and had been seen by the same gynaecologist (SR) were reviewed and the relevant clinical, biological, mammography and BMD data were summarised in a short case report. These summaries included thorough descriptions of patients menopausal symptoms, risk or history of breast cancer (personal and familial), endometrial cancer, cardiovascular disease, osteoporotic fractures and lifestyle, age at menopause and previous use of HRT. The history of these patients had been taken using a standardised method and were presented in a standardised way. The BMD had been measured at the Lumbar spine L2-L4 using a Hologic 1000. The presentation of the results included the BMD result expressed in g/cm’ of hydroxyapatite equivalent, the Z-score and the T-score. In addition, the same graphical illustration which is provided by the manufacturer was reproduced and also provided to the 10 gynaecologists. We further will refer to the summaries as the ‘original charts’. For each original chart, three ‘simulated cases’ were created. The clinical history, laboratory, mammography and echography results remained unchanged, even if the wording was slightly modified: e.g. ‘a 48-year-old woman who had her

last menstruation 2 years ago’ would become ‘the women is 48 and had her menopause at the age of 46’. The BMD results were modified in these ‘simulated cases’ in order to have 4 groups of women with exactly the same clinical story and a wide range of BMD values. The BMD modifications were based on Z score: one result would be higher than 0 Z-score (Group I), one between 0 and - 1 Z-score (Group II), one between - 1 and - 2 Z-score (Group III) and one lower than - 2 Z score (Group IV). For example, if a women had a ‘real Z score of - 0.26’, the three simulated cases would have Z-scores of + 0.26, - 1.26 and - 2.26, respectively. All the cases were presented to 10 physicians in maps of 20 cases at a time. The maps were constituted in such a way as to be certain that the cases issued from the same original chart would not be in one map. A time interval of approximately 1 week elapsed between the evaluation of two maps. 2.2. The physicians

Ten gynaecologists (6 women and 4 men, age range 35-55 years) who have been trained in the same University Hospital were asked to participate in the study. Half of them were full-time staff members. Half were private physicians, used to referring their patients to the same University Hospital for specialised examinations and who performed operations in the same institution. In addition, these gynaecologists regularly attended staff meetings at the hospital, many of which discussed the importance of osteoporosis prevention and the use of bone mineral measurements. They were told that the aim of the study was to assess the attitude of physicians towards HRT prescription. They were not aware of the manipulation of the cases (change in wording and in Z-score). The question asked was whether a HRT should be prescribed to the patient. 2.3. Statistical

analyses

In order to evaluate the effect of BMD (in the four different Z-classes) on the frequency of HRT

S. Rozenherg

et ul. / Maturitas

prescription for the 29 clinical stories, a paired one-way Friedman analysis of variance was used. Since significant results where obtained, interclass comparison between two Z-classes was performed using the Matched-Pairs Wilcoxon test. In order to assess whether differences between prescription habits existed between the 10 gynaecologists, the Cochran test was used. Since differences existed we also assessed the effect of BMD (in the four different Z-classes) on HRT prescription by each individual physician. 3. Results Overall, the 10 physicians prescribed a treatment in 74.2% of the 116 cases. When the number of treated cases are analysed in relation to the bone density, it appears that physicians treat, respectively, 65% of women who had the highest bone mass (Group I = Z score > 0), 73% of women of Group II (who had a Z score between 0 and - l), 79% of those of Group III (who had a Z score between - 1 and - 2) and 80% of those of Group IV (who had a Z score lower than - 2). These are significant differences using a Friedman analysis of variance (chi-square 17.2; P < 0.001). Applying the Matched-Paired test of Wilcoxon, the distribution of the number of physicians prescribing HRT to the patients was different between Group I and Group III, between Group I and Group IV (P < 0.01) as well as between Group I and Group II and between Group II and Group IV (P < 0.05). In five original charts there was an absolute agreement by all physicians for treatment whatever the bone mass was (for each of these chart, 40 answers in favour of therapy versus zero against it were obtained). In 6 others, there was strong agreement (at least 36 answers out of the 40 in favour of therapy). Thus, in approximately a third of the original charts (1 l/29) there was agreement for the initiation of therapy, regardless of the bone mass result. These women presented other indications for therapy such as climacteric symptoms and neither contra-indication to nor hesitation about this therapy. We therefore repeated the analysis after exclusion of these 11 original charts.

24 (1996)

57-61

59

Physicians were willing to treat, respectively, 49% of the remaining women of Group I, 58% of women of Group II, 67% of Group III and 69% of those of Group IV. As expected, these differences appear significant using a Friedman analysis of variance (chi-square 9.2; P < 0.05). Applying the Matched-Paired test of Wilcoxon, the distribution of the number of physicians prescribing HRT to the patients was different between Group I and Group III, between Group I and Group IV as well as between Group II and Group IV (P < 0.05). When we analysed whether differences existed between the prescription frequency of the 10 gynaecologists, an overall significant variation of the 116 cases was observed between physicians (range in favour of treatment 63% to 87%; Cochran Q Statistic 39.2; P < 0.0001). As expected, when we exclude the 11 stories where a strong agreement to treat is met, the variation increases and ranges between 44% to 81% for the 72 cases which concern the remaining 18 clinical stories. We were not able to find any differences in prescription habits between private physicians or staff members. In order to assess the BMD effect on individual physicians, we calculated the individual percentage of treated cases per physician and per Z-class for the 72 remaining cases (shown in Fig. I ). The figure suggests that there are differences in attitudes between physicians. For some physicians, a trend to increase prescription is observed in relation to the BMD result (physicians numbers 2, 3, 4, 5 and lo), but a statistical difference could only be reached for physician number 2 (P < 0.05). Moreover, for the other physicians no moditication whatsoever could be observed.

4. Discussion

Despite extensive evidence that HRT is effective in preventing osteoporosis, very few studies have assessed whether physicians are inclined to prescribe HRT for this reason. Surveys of physicians have shown variable results about their views concerning HRT initiation and continuation. In a study evaluating the views

of 25 family physicians and 25 gynaecologists about HRT, using written case descriptions and a questionnaire, Holzman et al [6] found that osteoporosis was not a significant factor in a multivariate analysis of prescription probability. More recently, Grisso et al [7] asked physicians to what proportion of white postmenopausal women they would recommend some prevention for osteoporosis (having the choice between calcium, HRT, vitamin D, fluoride and exercise). While about 37% of the interviewed gynaecologists would recommend HRT in a majority of postmenopausal women, less than 7% of the general internists and 5% of the cardiopulmonary specialists would do so [7]. But in Grisso’s study, one might question whether the gynaecologists were reporting their use of HRT in general (including for relief of climacteric symptoms) rather than its specific use for osteoporosis prevention. Norman and Studd [8] have assessed the attitude of general practitioners and consultant gynaecologists towards HRT. While about 97% of the physicians answered that they knew that HRT decreases the risk of osteoporosis, only 48% considered treating women over 60 who had experienced fractures with HRT, assuming that there were no medical contraindications for HRT.

100

% Treated r- --

cases Z Class n>o O-1 o<-1

&r-2

i-m<-2

Physicians Fig. I. After exclusion of the 11 stories where a strong agreement to treat is met. 18 clinical stories are left. For these stories. the individual percentages of treated cases are shown per physician and for each Z-class. A wide variation existed between physicians ranging between 44% and 81%. The figure suggests that, for some physicians, a trend to increase prescription is observed in relation to the BMD result (physicians numbers 2, 3, 4, 5 and 10). but a statistical difference could only be reached for physician number 2 (P < 0.05). Moreover, for the other physicians no modification whatsoever could be observed.

Currently. we are not aware of data specifically concerning the attitude of gynaecologists towards BMD results, even though it has been shown that BMD measurements predict fracture risk [9 1 I] and it has been suggested that women with the lowest BMD may have the highest benefit from HRT in terms of osteoporosis prevention [ 12- 141. The present study was therefore designed to evaluate the effect of BMD on HRT prescription. To minimise the bias of self-reported practices and theoretical attitudes, we used simulated cases [6]. Bone density appeared, in average, to be a factor influencing the prescription of HRT since. for the same cases, the percentage of treated cases increased from 65% when the BMD result was above average for age (Z score > 0) to 73% when it was lower than average and even 80% when exceptionally low (Z score lower than - 3). In approximately one third of the patients, a very high proportion of treatment was advised independently of bone mass values. One may hypothesise that, in these cases, an agreement existed between the physicians about the potential benefits of HRT (for example in reducing climacteric symptoms). When these cases are excluded, the percentages of treated cases increased from 49% when the BMD result was above average for age (Z score > 0) to 58% when it was lower than average and even 69% when exceptionally low (Z score lower than - 2). Another important observation of this study is the large variation in therapeutic attitude between physicians. In a rather homogenous group of physicians, the likelihood of having treatment proposed varied from 63% to 87%. For some physicians, no variation at all in treatment frequency according to the BMD results could be observed. This heterogeneity suggests that some physicians do not consider BMD to be a predictor of osteoporosis and do not use BMD to decide whether to initiate HRT. The fact that this observation is made in an optimal clinical setting should be stressed. Indeed, in Belgium, the BMD measurement is completely funded by social security. This concerns almost all postmenopausal women since, in Belgium, more than 95% of the population is covered by social security. Belgium is one of the countries with the highest proportion

S. Rozenberg

et al. 1 Maturitas

of physicians and densitometers per person. Therefore, BMD measurements are available to almost all postmenopausal women. In summary, while most studies have shown that BMD is the best predictor of osteoporotic fractures and that HRT may prevent these fractures, few studies have assessed the attitude of physicians towards HRT for prevention of osteoporosis. The present study suggests that, in clinical practice, HRT is often initiated for many other reasons and independently of the BMD. The BMD result appears to be a determinant factor for HRT prescription in only a limited proportion of the patients and a small proportion of the physicians. From an epidemiological point of view, BMD measurements may be proposed to women who are reluctant to start HRT or to those who present relative contra-indications, provided that their physicians are aware of the use of these investigations.

24 (1996)

1993: 240-245. [4]

[5]

[6]

[7] [8] [9] [lo]

[ll]

References [12] [l] Ettinger B, Genant HK, Cann CE. Long term estrogen replacement therapy prevents bone loss and fractures. Ann Intern Med 1985; 102: 319-324. [2] Christiansen C. Hormone replacement therapy for established osteoporosis in elderly women. In: Christiansen C (Ed). Hormone replacement and its impact on osteoporosis Balliere’s Clin Obstet Gynaecol 1991; 5: 853-856.

61

[3] Lindsay R. Prevention of osteoporosis. In Favus MJ, Christakos S, Gage1 RF, Kleerekoper M, Langman CB. Shane E, Stewart AF, Whyte MP (Eds). Primer on the Metabolic Bone Diseases and Disorders of Mineral Metabolism. Raven Press, New York, Second Edition

Acknowledgements This study was supported by a grant of Novo Nordisk, Belgium.

57-61

[13] [14]

Consensus Development Conference: Diagnosis, Prophylaxis and Treatment of Osteoporosis. Am J Med 1993; 94: 6477650. Johnston CC, Melton JL. Bone density measurements and the management of osteoporosis. In Favus MJ. Christakos S, Gage1 RF, Kleerekoper M, Langman CB, Shane E, Stewart AF, Whyte MP. (Eds). Primer on the Metabolic Bone Diseases and Disorders of Mineral Metabolism. Raven Press, New York, Second Edition 1993: 1377 146. Holzman GB, Ravitch MM, Metheny W, Rothert ML, Holmes M, Hoppe RB. Physicians’ judgments about estrogen replacement therapy for menopausal women. Obstet Gynecol 1984; 63: 3033311. Grisso JA, Baum CR, Turner BJ. What do physicians in practice do to prevent osteoporosis? J Bone Miner Res 1990; 3: 213-220. Norman SG, Studd JWW. A Survey of views on hormone replacement therapy. Br J Obstet Gynaecol 1994: 101, 879-887. Hui SL, Slemenda CW, Johnston CC Jr. Baseline measurement of bone mass predicts fracture in white women. Ann Intern Med 1989; 111: 355-361. Wasnich RD, Ross PD, Davis JW. Vogel JM. A comparison of single and multi-site BMC measurements for assessment of spine fracture probability. J Nucl Med 1989; 30: 1166-1171. Cummings SR, Black DM, Nevitt MC HK et al. Bone density at various sites for prediction of hip fractures. Lancet 1993; 341: 72-75. Cummings SR. Bone mass and bone loss in the elderly: a special case? Int J Fertil 1993: 38: 92297. Rozenberg S, Gevers R. Peretz A et al. Decrease of bone mineral density during estrogen substitution therapy. Maturitas 1993; 17: 2055210. Armamento-Villareal R, Civitelli R. Estrogen action on the bone mass of postmenopausal women is dependent on body mass and initial bone density. J Clin Endocrinol Metab 1995; 80: 776-782.