Author + information
- Received December 22, 2016
- Revision received February 9, 2017
- Accepted March 9, 2017
- Published online July 3, 2017.
- Eric W. Holroyd, MDa,
- Alex Sirker, MB, BChir, PhDb,
- Chun Shing Kwok, MBBS, BSc, MSca,c,
- Evangelos Kontopantelis, PhDd,
- Peter F. Ludman, MDe,
- Mark A. De Belder, MDf,
- Robert Butler, MBChB, MDa,
- James Cotton, MBBS, MDg,
- Azfar Zaman, MBChB, MDh,
- Mamas A. Mamas, BMBCh, DPhila,c,∗ (, )
- British Cardiovascular Intervention Society and National Institute of Cardiovascular Outcomes Research
- aAcademic Department of Cardiology, Royal Stoke Hospital, University Hospital of North Midlands, Stoke-on-Trent, United Kingdom
- bDepartment of Cardiology, University College London Hospitals and St. Bartholomew’s Hospital, London, United Kingdom
- cKeele Cardiovascular Research Group, Institute of Applied Clinical Science, Keele University, Stoke-on-Trent, United Kingdom
- dInstitute of Population Health, University of Manchester, Manchester, United Kingdom
- eQueen Elizabeth Hospital, University Hospital of Birmingham, Birmingham, United Kingdom
- fThe James Cook University Hospital, Middlesbrough, United Kingdom
- gDepartment of Cardiology, The Heart and Lung Centre, The Royal Wolverhampton Hospitals NHS Trust, Wolverhampton, United Kingdom
- hFreeman Hospital and Institute of Cellular Medicine, Newcastle University, Newcastle-upon-Tyne, United Kingdom
- ↵∗Address for correspondence:
Prof. Mamas A. Mamas, Keele Cardiovascular Research Group, Institute of Applied Clinical Science, Keele University, Stoke-on-Trent ST4 7QB, United Kingdom.
Objectives The aims of this study were to examine the relationship between body mass index (BMI) and clinical outcomes following percutaneous coronary intervention (PCI) and to determine the relevance of different clinical presentations requiring PCI to this relationship.
Background Obesity is a growing problem, and studies have reported a protective effect from obesity compared with normal BMI for adverse outcomes after PCI.
Methods Between 2005 and 2013, 345,192 participants were included. Data were obtained from the British Cardiovascular Intervention Society registry, and mortality data were obtained through the U.K. Office of National Statistics. Multiple logistic regression was performed to determine the association between BMI group (<18.5, 18.5 to 24.9, 25 to 30 and >30 kg/m2) and adverse in-hospital outcomes and mortality.
Results At 30 days post-PCI, significantly lower mortality was seen in patients with elevated BMIs (odds ratio [OR]: 0.86 [95% confidence interval (CI): 0.80 to 0.93] 0.90 [95% CI: 0.82 to 0.98] for BMI 25 to 30 and >30 kg/m2, respectively). At 1 year post-PCI, and up to 5 years post-PCI, elevated BMI (either overweight or obese) was an independent predictor of greater survival compared with normal weight (OR: 0.70 [95% CI: 0.67 to 0.73] and 0.73 [95% CI: 0.69 to 0.77], respectively, for 1 year; OR: 0.78 [95% CI: 0.75 to 0.81] and 0.88 [95% CI: 0.84 to 0.92], respectively, for 5 years). Similar reductions in mortality were observed for the analysis according to clinical presentation (stable angina, unstable angina or non–ST-segment elevation myocardial infarction, and ST-segment elevation myocardial infarction).
Conclusions A paradox regarding the independent association of elevated BMI with reduced mortality after PCI is still evident in contemporary U.K. practice. This is seen in both stable and more acute clinical settings.
Obesity is a growing worldwide health concern. In the United States, recent data indicate that more than one-third of adults, and about 1 in 5 children or adolescents, are obese (1,2). The estimated cost of obesity in the United States was $147 billion in 2008, on the basis of Centers for Disease Control and Prevention data (3). Obesity predicts coronary artery disease and premature death (4,5), and it is estimated that obese nonsmokers lose up to 7 years of life expectancy compared with normal-weight nonsmokers (6).
Notwithstanding these statistics, multiple studies have demonstrated an apparently protective effect from obesity, compared with a “normal” body mass index (BMI), when considering in-hospital and even longer term clinical endpoints (7). This so-called obesity paradox has been noted in various settings but was first described in the context of percutaneous coronary intervention (PCI) outcomes by Ellis et al. (8), who noted a decreased risk for in-hospital mortality associated with a BMI of 26 to 34 kg/m2 compared with levels greater or lower than this. Subsequent work by Gruberg et al. (9) looking at more than 9,000 consecutive PCIs between 1994 and 1999 observed that obese patients (BMI >30 kg/m2) were, on average, younger and had a higher incidence of cardiac risk factors, including hypertension, diabetes, high cholesterol, and smoking. There was no difference in PCI acute procedural success among BMI groups in their study. However, obese patients had fewer complications and lower in-hospital and 1-year mortality, and BMI was an independent predictor of favorable clinical outcomes. Several other published studies have supported this association (10–13), including a meta-analysis combining outcomes at 1 to 5 years from 5 separate studies (12). However, there have been some conflicting results from other work (14–16) that has not identified this survival paradox.
The largest single study in this field retrospectively analyzed 50,149 patients with ST-segment elevation myocardial infarction (STEMI) from the U.S. National Cardiovascular Data Registry (17). It demonstrated a “U-shaped” curve relationship between obesity and outcome in unadjusted analyses, with an “obesity paradox” benefit for overweight and obese patients but a detrimental outcome in extreme, class III obese patients (BMI >40 kg/m2). However, when confounding variables were taken into account, the obesity paradox (obese vs. normal BMI) for in-hospital mortality was eliminated, although there remained an increased propensity for major bleeding in both normal-weight patients and class III obese patients, compared with class I obese patients (BMI 30 to 35 kg/m2) (17). Elsewhere, there are very limited data on whether the relationship of PCI outcome to obesity is similar across the different indications for PCI (i.e., STEMI, non–ST-segment elevation myocardial infarction [NSTEMI], and stable angina) (18), and here again, discordant findings have resulted in uncertainty.
We analyzed contemporary U.K. PCI data from an unselected national cohort, derived from the British Cardiovascular Intervention Society (BCIS) database, between 2005 and 2013. Our aim was to explore the “obesity paradox” following PCI. We also analyzed differences in the relationship of obesity to outcomes on the basis of the clinical indication for PCI.
We analyzed PCI data collected on behalf of the BCIS by the National Institute of Cardiovascular Outcomes Research. This dataset records PCI procedures performed in all U.K. hospitals. In 2012, the database contained 99.4% of all PCI procedures performed in the National Health Service hospitals in England and Wales (http://www.bcis.org.uk).
The BCIS–National Institute of Cardiovascular Outcomes Research database records clinical, procedural, and outcome information with a total of 113 variables. Mortality was tracked using each patient’s unique National Health Service number, by linkage with records held by the Office of National Statistics. This process is available only for patients resident in England and Wales, who represent the large majority of the British population, so patients living in Scotland and Northern Ireland were not included in this study (19–21).
We therefore analyzed all patients who underwent PCI in England and Wales between January 1, 2005, and December 31, 2013, with values for BMI and mortality outcomes at 30 days and 1 year. Patients were classified according to BMI groups: <18.5, 18.5 to 24.9, 25.0 to 30.0, and >30 kg/m2. The outcomes were 30-day mortality, 1-year mortality, major adverse cardiovascular events (MACE) and major bleeding. MACE were defined as the composite of in-hospital reinfarction, repeat PCI, emergency coronary artery bypass grafting, and in-hospital mortality.
Statistical analysis was performed using Stata version 13.1 (StataCorp, College Station, Texas). To account for missing data among included patients, we used the mi impute procedure to perform multiple imputation using chained equations and generate 10 datasets. Across the 2 groups, we used 1-way analysis of variance and chi-square tests to compare continuous and categorical variables, respectively. Multiple imputation logistic regressions (mi estimate: logistic) were used to calculate crude and adjusted odds of 30-day mortality according to BMI group. Data were adjusted for age, sex, year, race, smoking, family history of coronary artery disease, hypertension, hypercholesterolemia, diabetes, peripheral vascular disease, previous myocardial infarction, previous stroke, valvular heart disease, renal disease, previous PCI, previous coronary artery bypass surgery, left ventricular ejection fraction, receipt of ventilation, receipt of circulatory support, cardiogenic shock, left main coronary artery intervention, use of drug-eluting stents, radial access, glycoprotein IIb/IIIa inhibitor use, and diagnosis. We also performed the analysis stratifying by the diagnosis (stable angina, NSTEMI, or STEMI). In view of the number of possible analytic approaches to control for potential measured confounding, in addition to multiple regression, we performed propensity score matching followed by simple regression and inverse of the propensity score as weighting in simple regression, as these all have been shown to be methodologically valid (22). In the inverse probability treatment weighting approach, we first calculated propensity scores for all binary BMI combinations we investigated: 25 to 30 kg/m2 versus 18.5 to 24.9 kg/m2, >30 kg/m2 versus 18.5 to 24.9 kg/m2, and <18.5 kg/m2 versus 18.5 to 24.9 kg/m2. The inverses of the propensity scores were used as probability weights in multiple imputation logistic regressions where only the respective treatment was included as a predictor. We conducted a sensitivity analysis considering risk for adverse outcomes for different grades of patients with BMIs >30 kg/m2 (30.0 to 34.9, 35.0 to 39.9, and ≥40 kg/m2). In addition, we performed sensitivity analysis for the subgroup of participants who were admitted before or in 2009, participants who were admitted after 2009, participants without diabetes, and participants with diabetes.
Between 2005 and 2013, there were 345,192 records in the BCIS database with data for BMI for patients who underwent PCI procedures. The extent of available and missing data is shown in Online Table 1.
Proportion of patients undergoing PCI who were obese from 2005 to 2013
The percentage of participants undergoing PCI with BMIs >30 kg/m2 showed a small but significant increase (p = 0.002) from 30% in 2005 to 32% in 2013 (Figure 1).
Baseline clinical characteristics, periprocedural factors, and unadjusted outcomes
Table 1 presents baseline characteristics, divided by BMI into 4 groups as defined by the World Health Organization: lean (<18.5 kg/m2), normal (18.5 to 24.9 kg/m2), overweight (25 to 30 kg/m2), and obese (>30 kg/m2). We compared characteristics in each group. Obese patients were significantly younger (p < 0.001) compared with other groups. Obese patients, compared with those with normal BMIs, were more often smokers (66% vs. 63%; p < 0.001) and had features of the metabolic syndrome associated with obesity, namely, hypertension (62% vs. 49%; p < 0.001), hypercholesterolemia (61% vs. 53%; p < 0.001), and diabetes (29% vs. 13%; p < 0.001). Left ventricular ejection fraction was somewhat more likely to be good in obese patients (76% had good left ventricular function vs. 72% of patients with normal BMIs). Radial access was slightly more commonly used in obese patients (48% vs. 44% in patients with normal BMI; p < 0.001), but there was no significant difference in the proportion receiving drug-eluting stents. A much larger proportion of PCIs for STEMI were performed in lean compared with obese patients (24% vs. 12%; p < 0.001). Conversely, a greater proportion of PCIs in obese patients were performed for stable angina compared with patients with lean BMIs (52% vs. 30%; p < 0.001).
Lean patients (BMI <18.5 kg/m2) were significantly older, were less likely to be male (46% vs. 71%), tended to have poorer left ventricular function (67% had good left ventricular function vs. 72%), had less radial access use (41% vs. 44%), and had less frequent treatment with drug-eluting stents (66% vs. 69%) compared with patients with normal BMIs.
Unadjusted crude mortality suggested the presence of an obesity paradox with better survival in obese patients, and this relationship was more evident at longer follow-up time frames. Crude 30-day mortality was 1% in obese patients compared with 2% in those with normal BMIs and 4% in lean patients (p < 0.001). At 1 year, mortality in obese patients was 3%, compared with 6% in those with normal BMIs and 14% in lean patients (p < 0.001). At 5 years, crude mortality was 19% in obese patients, 28% in patients with normal BMIs, and 53% in lean patients (p < 0.001).
Statistical analysis of adverse outcome according to BMI group
Statistical multivariate analysis of outcome data at 30 days and 5 years post-PCI is presented in Table 2. At 30 days, the unadjusted odds ratio (OR) in the obese group was 0.49, compared with 1 in patients with normal BMI and 1.68 in lean patients (p < 0.001). After adjusting for the factors listed earlier, the odds of 30-day mortality remained significantly decreased in both the BMI 25 to 30 kg/m2 (OR: 0.86; 95% confidence interval [CI]: 0.80 to 0.93; p = 0.001) and BMI >30 kg/m2 (OR: 0.90; 95% CI: 0.82 to 0.98; p = 0.016) groups but did not reach statistical significance in the BMI <18.5 kg/m2 group (OR: 1.23; 95% CI: 0.98 to 1.54; p = 0.077) (Figure 2). Similar observations were recorded at 1 year, with independent decreases in the odds of mortality in the BMI 25 to 30 kg/m2 (OR: 0.70; 95% CI: 0.67 to 0.73; p < 0.001) and BMI >30 kg/m2 (OR: 0.73; 95% CI: 0.69 to 0.77; p < 0.001) groups and independent increases in odds of mortality in the BMI <18.5 kg/m2 group (OR: 1.85; 95% CI: 1.63 to 2.10; p < 0.001). Similar trends were recorded at 3 and 5 years.
Although in-hospital MACE were more likely in the unadjusted data with rising BMI (OR: 0.68 vs. 1 in patients with normal BMIs and 1.24 in lean patients), after adjustments for differences in baseline covariates, the effect of BMI on MACE was no longer significant. Nevertheless, the odds for in-hospital bleeding complications were significantly less in obese patients following multivariate analysis compared with patients with normal BMIs and lean patients (0.87 vs. 1.00 vs. 1.24; p < 0.001).
Adjusted odds of adverse outcome in obese patients depending on clinical syndrome
We then divided the recorded PCI data on the basis of clinical presentation (i.e., stable angina, unstable angina or NSTEMI, or STEMI), to examine the differential effect of BMI according to PCI indication (Table 3). Multivariate regression analysis yielded similar results to the overall PCI data. Unadjusted 30-day mortality was lower with higher BMI in patients with stable angina, unstable angina or NSTEMI, and STEMI but this effect was no longer significant after statistical adjustment in the STEMI group. However, at 1, 3, and 5 years, the adjusted odds of mortality in patients with obesity were significantly less than in patients with normal BMIs in PCIs for stable angina, unstable angina or NSTEMI, and STEMI. Similarly, in lean patients (BMI <18.5 kg/m2) taking into account comorbidity, the adjusted odds for mortality were significantly increased at 1, 3, and 5 years. There were significantly fewer in-hospital bleeding events in obese patients compared with normal and lean patients for all 3 clinical syndromes, an effect that remained significant even after statistical adjustment.
Inverse probability weighting by propensity scores analysis of adverse outcomes and BMI
Inverse probability weighting (by propensity scores) analysis of adverse outcomes directly comparing different BMI groups is shown in Table 4. Using this method of analysis, both overweight and obese groups are seen to have a significantly lower odds of mortality than the normal BMI group at all studied time points (30 days to 5 years), whereas the lean group had an increased odds of mortality at 1, 3, and 5 years.
Sensitivity analysis considering patients with BMIs ≥30 kg/m2 compared with those with normal BMIs
Online Table 2 shows the risk for adverse outcomes among participants with BMIs ≥30 kg/m2 by BMI group. Unadjusted estimates suggest that participants in all elevated BMI groups have lower odds of mortality, MACE, and bleeding compared with subjects with normal BMIs. However, after adjustment, it appears that participants with BMIs ≥40 kg/m2 have no significant difference in the odds of in-hospital MACE, in-hospital major bleeding, or 5-year mortality.
Additional analysis considering the subgroup of participants admitted before or in 2009, after 2009, and those with and without diabetes showed similar trends as overall results (Online Tables 3 to 6).
Our data show significant differences in short-, medium-, and long-term mortality independently associated with baseline BMI group, with greater survival being seen in patients classified as overweight (BMI 25 to 30 kg/m2) or obese (BMI >30 kg/m2) as opposed to those with normal BMIs (18.5 to 24.9 kg/m2) at the time of PCI. In patients with BMIs <18.5 kg/m2, worse clinical outcomes were observed both in the short and longer term. This significant effect persisted (albeit with reduced magnitude) even after adjustment for multiple potential confounding factors, as described. Furthermore, there was overall consistency between the findings of our main analysis, using multivariate logistic regression, and the alternative methodology using inverse probability weighting by propensity scores. The very large patient numbers involved also allowed us to undertake a meaningful subgroup analysis based on clinical presentation; here too a consistent pattern of findings was seen, with better outcomes observed in overweight patients and worse outcomes recorded in those with BMIs <18.5 kg/m2, even after adjustment for differences in baseline covariates.
Our study findings are consistent with the results of 3 recent systematic reviews and meta-analyses of the published research for outcomes based on BMI after coronary revascularization (23,24). Those studies involved 91,582 patients (who had detailed medication use data available) and 242,377 patients, respectively, and hence each was significantly smaller than our cohort, in which 30-day post-PCI mortality data were available in more than 350,000 patients. The findings also are consistent with those of a recent meta-analysis of more than 1.3 million patients that re-examined the link between mortality and BMI in patients with coronary artery disease (not restricted solely to a PCI or revascularization setting) (25). This too found short- and long-term mortality advantages for overweight or obese groups compared with patients with normal BMIs. Sharma et al. (26) conducted a systematic review and meta-analysis of the relationship between BMI and mortality, cardiovascular mortality, and myocardial infarction after revascularization. There review of 36 coronary artery bypass graft and PCI studies found that patients with low BMIs had the highest risk for adverse events, whereas those with high BMI had the fewest events. Our present study provides further evidence that supports their findings.
The confirmation of a BMI paradox (for overweight and obese patients) in this large contemporary PCI population raises questions about potential unrecognized confounders, for which adjustment has not been made in our analysis. This is a feature common to all registry-based studies. For our BCIS cohort, 4 specific aspects are recognized as limitations. First, there is only limited recording of other (noncardiac) comorbidities, which are pertinent to mortality at all time points post-PCI (27,28).
Second, we do not have access to accurate recording of guideline-recommended medical therapy use for these patients. Differences in their use would potentially affect clinical outcomes, and recent published work confirms that this may explain some, although seemingly not all, of the observed obesity paradox (23).
Third, measures of frailty or comorbidity (27,28) were not recorded in this dataset and may represent unmeasured confounders and therefore contribute to the poorer outcomes reported in the low BMI group. This is of particular importance in this analysis because weight loss may be a manifestation of underlying ill health for a wide variety of reasons, including heart failure or malignancy, which in its most marked form may present as cachexia. Inclusion of such patients in the “low BMI” group will contribute to a higher rate of adverse clinical outcomes compared with those with greater BMIs. By extension, some of those in the “normal weight” group may likewise have experienced prior weight loss because of comorbidity. However, the very large patient numbers involved in our study should ameliorate the impact from such an influence, because “hitherto healthy” normal-weight patients are likely to account for the majority of patients in this BMI grouping.
Finally, the BCIS dataset does not capture data on post-discharge secondary prevention therapies prescribed, and differences in the provision of secondary care prevention among patients of different BMI may contribute to the outcomes reported.
A separate, more relevant issue, however, is the acknowledged limitation of BMI as a measure of obesity. Important additive prognostic information comes from knowledge of fat distribution, with a recognized detrimental impact from “central obesity” (29). Relevant data, such as waist circumference, are not available in the BCIS dataset. Hence, it is not possible to identify those who would fall into the category of “normal-weight central obesity” to refine our group classification system beyond BMI alone. Whether this would change our key findings is currently unknown. Indeed, in a previous analysis of more than 15,000 patients derived from the European Prospective Investigation Into Cancer Norfolk cohort who were prospectively followed up, waist-to-hip ratio was the strongest predictor of incident cardiovascular disease and mortality compared with either BMI or body fat percentage (30).
In considering other explanations for our key study findings on mortality, we note that in-hospital major bleeding complications were lower in overweight and obese patients, and with bleeding independently associated with worse short-term and longer term mortality outcomes (31,32). A reduction in bleeding is likely driven to some extent by higher rates of radial access in patients with greater BMIs. However, other potential mechanisms for bleeding differences between BMI groups include appropriate dosing of peri-PCI anticoagulant therapy and differences in sheath-to-artery ratios (18). Indeed, dosing of anticoagulation may be particularly relevant in acute settings such as STEMI, in which opportunities to gain an accurate measure of a patient’s weight may be limited, resulting in overdosing of patients with low BMIs, whose weight might be overestimated.
Finally, when trying to interpret our findings, consideration should be given to evidence of potentially protective effects from adipose tissue itself in various post-operative and post-procedural settings. Adipose tissue is important in the production of various hormones and cytokines, including tissue necrosis factor, adiponectin, and leptin (33). Whether these factors or others may be involved in the protective mechanisms against PCI-related complications is unclear (34). Some experimental models indicate a protective effect of obesity against ischemia-reperfusion injury; for example, a hyperphagia-induced obese rat model has been shown to have smaller infarcts and improved functional recovery following reperfusion, with increased signaling shown in the reperfusion injury salvage kinase pathway (35). Obesity-inducing diets in rats (sucrose-supplemented or high-fat diet) have also been shown to be cardioprotective (36). Harvested hearts were less susceptible to ischemia-reperfusion injury and had smaller infarct sizes, an effect not due to reperfusion injury salvage kinase signaling. Although a role for such pathways in influencing clinical outcomes is plausible in acute presentations (particularly STEMI), their relevance to PCI in stable settings is questionable. Nevertheless, our confirmation of earlier studies demonstrating a “BMI paradox” should provide support for mechanistic studies to explore this observation.
In this largest study to date examining the relationship between BMI and PCI outcomes, an obesity paradox is still evident in contemporary PCI. This paradox is encountered with PCI in both stable coronary disease and in more acute clinical situations. Factors underlying this phenomenon remain uncertain and controversial, and this study provides support for further exploration.
WHAT IS KNOWN? The clinical relevance of this study for PCI operators is that despite the added difficulty of a procedure on a patient with high BMI, in terms of radiation exposure, contrast use, technical difficulty, and even airway management, the obesity paradox remains apparent. Obese patients have better outcomes 1 year after PCI, for both acute coronary syndromes and angina treatment.
WHAT IS NEW? This study should encourage radial access further and should also highlight the potentially negative implication of low BMI, leading physicians to carefully dose medications and investigate weight loss or very low BMI.
WHAT IS NEXT? Future research is necessary to understand the pathological mechanism of this effect or identify which confounding variables are missing from the analysis.
The authors acknowledge the University Hospitals of North Midlands Charity for supporting this study.
For supplemental tables, please see the online version of this article.
This work is funded by the University Hospitals of North Midlands Charity. The authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Holroyd, Sirker, and Kwok contributed equally to this work.
- Abbreviations and Acronyms
- British Cardiovascular Intervention Society
- body mass index
- confidence interval
- major adverse cardiovascular event(s)
- non–ST-segment elevation myocardial infarction
- odds ratio
- percutaneous coronary intervention
- ST-segment elevation myocardial infarction
- Received December 22, 2016.
- Revision received February 9, 2017.
- Accepted March 9, 2017.
- 2017 American College of Cardiology Foundation
- Finkelstein E.A.,
- Trogdon J.G.,
- Cohen J.W.,
- Dietz W.
- Gruberg L.,
- Weissman N.J.,
- Waksman R.,
- et al.
- Buschur M.E.,
- Smith D.,
- Share D.,
- et al.
- Akin I.,
- Tölg R.,
- Hochadel M.,
- et al.
- Das S.R.,
- Alesander K.P.,
- Chen A.Y.,
- et al.
- Sirker A.,
- Mamas M.,
- Robinson D.,
- et al.
- Mamas M.A.,
- Anderson S.G.,
- Carr M.,
- et al.
- Mamas M.A.,
- Nolan J.,
- de Belder M.A.,
- et al.
- Bundhun P.K.,
- Li N.,
- Chen M.H.
- Tan X.F.,
- Shi J.X.,
- Chen A.M.
- Wang Z.J.,
- Zhou Y.J.,
- Galper B.Z.,
- Gao F.,
- Yeh R.W.,
- Mauri L.
- Sharma A.,
- Vallakati A.,
- Einstein A.J.,
- et al.
- Rashid M.R.,
- Kwok C.S.,
- Gale C.P.,
- et al.
- Myint P.K.,
- Kwok C.S.,
- Luben R.N.,
- Wareham N.J.,
- Khaw K.T.
- Kwok C.S.,
- Rao S.V.,
- Myint P.K.,
- et al.
- Kwok C.S.,
- Khan M.A.,
- Rao S.V.,
- et al.
- Hajer G.R.,
- van Haeften T.W.,
- Visseren F.L.
- Donner D.,
- Headrick J.P.,
- Peart J.N.,
- du Toit E.F.
- Salie R.,
- Huisamen B.,
- Lochner A.