The incidence of primary glomerulonephritis worldwide: a systematic review of the literature

Abstract

Background. Little is known about the worldwide variation in incidence of primary glomerulonephritis (GN). The objective of this review was to critically appraise studies of incidence published in 1980–2010 so that an overall view of trends of these diseases can be found. This would provide important information for determining changes in rates and understanding variations between countries.

Methods. All relevant papers found through searches of Medline, Embase and ScienceDirect were critically appraised and an assessment was made of the reliability of the reported incidence data.

Results. This review includes 40 studies of incidence of primary GN from Europe, North and South America, Canada, Australasia and the Middle East. Rates for the individual types of disease were found to be in adults, 0.2/100 000/year for membrano-proliferative GN, 0.2/100 000/year for mesangio-proliferative GN, 0.6/100 000/year for minimal change disease, 0.8/100 000/year for focal segmental glomerulosclerosis, 1.2/100 000/year for membranous nephropathy and 2.5/100 000/year for IgA nephropathy. Rates were lower in children at around 0.1/100 000/year with the exception of minimal change disease where incidence was reported to be 2.0/100 000/year in Caucasian children with higher rates in Arabian children (9.2/100 000/year) and Asian children (6.2–15.6/100 000/year).

Conclusions. This study found that incidence rates of primary GN vary between 0.2/100 000/year and 2.5/100 000/year. The incidence of IgA nephropathy is at least 2.5/100 000/year in adults; this disease can exist subclinically and is therefore only detected by chance in some patients. In addition, referral policies for diagnostic biopsy vary between countries. This will affect the incidence rates found.

Introduction

Although much is known about the clinical characteristics and natural history of the primary glomerulopathies, very little information on the epidemiology of these diseases is available from reviews. Insight into the baseline incidence of glomerulonephritis (GN) throughout the world can provide important information on trends of disease occurrence by sex, age and geographical location. New vaccines are being introduced and concerns have been raised about the potential associated risk of autoimmune diseases [1,2]. It is therefore of interest to know what the baseline incidence rates across the world are so that concerns about possibly associated increased incidence rates of autoimmune diseases, such as GN, can be evaluated.

To our knowledge, no other systematic review of incidence of the most common of the primary glomerulopathies has been conducted in the last three decades. In this paper, we perform a systematic review, critically appraising studies of incidence of primary GN throughout the world.

Materials and methods

Searches of the Medline, Embase and ScienceDirect databases (1980–June 2007) were carried out using the search terms ‘glomerulonephritis’, ‘IgA nephropathy’, ‘membranous nephropathy’, ‘membrano-proliferative glomerulonephritis’, ‘mesangial proliferative glomerulonephritis’, ‘minimal change disease’, ‘focal segmental glomerulosclerosis’, ‘post-infectious glomerulonephritis’, ‘idiopathic crescentic proliferative glomerulonephritis’, ‘ANCA-associated necrotising crescentic glomerulonephritis’, ‘anti-glomerular basement membrane disease’, ‘kidney disease’, ‘incidence’, ‘incid*’ and ‘epidemiology’. In Medline, the individual disease names were searched for as well as the term ‘glomerulonephritis’ because this MeSH term does not include all types of GN as daughter terms in its hierarchical structure.

The inclusion criteria were that the studies reported original work, that the study reported incidence of specific forms of GN with reference to a denominator population, that the estimates of population size and person-time contributed were accurate and that efforts had been made to ascertain all incident cases. When assessing the likelihood of missing incident cases, papers were evaluated as follows: (i) for case finding studies, did the authors ensure that all of the subjects contributing to incidence denominator data would have been eligible to have the disease diagnosed and did the authors check all relevant medical records? (ii) For all studies, were cases checked to ensure that they were incident and not prevalent? (iii) For all studies, did the authors ensure that the cause of GN was autoimmune and not secondary to another disease? Where possible, we only included incidence rates for cases of GN caused by autoimmunity, determination of which relied on information given in the paper.

The titles and abstracts of all of the studies produced by the searches were reviewed, and those papers accepted for inclusion in the study were appraised. Studies published in English, French, German, Spanish or Dutch were included. Review papers identified were searched for secondary references reporting on original research; secondary references found from any of the other papers reviewed were also included.

A standard data abstraction form was used to record all details of the papers reviewed; a copy of this is given in the  Appendix (Figure A1). Each study was scored for accuracy of the incidence rates it presented and was classified as being at low, medium or high risk for under- or overestimation of reported incidence rates by considering the reliability of numerator and denominator data. For instance, inclusion of prevalent cases or those thought not to be caused by autoimmunity will have led to overestimated rates as well as underestimated denominator data. Conversely, missing cases or an overestimated denominator (e.g. a catchment area from which not all inhabitants had access to hospital services) would be considered to have resulted in underestimated incidence rates. Explanations provided by the papers’ authors as to why incidence rates were as expected or whether they were considered to be an over- or underestimate of the true incidence rate were taken into account in this process. If the extent of likely error was considered to be very great, the study was excluded. To minimize subjectivity, this assessment was agreed upon between two of the authors, and random checks were performed to ensure consistency. Rates are presented as the number of cases/100 000/year, and where sufficient data were given in the paper, rates were checked for accuracy. Guidelines were followed in the reporting of this study to ensure that key information was presented [3].

Results

The results of the database searches with the number of included and excluded papers are given in Figure 1; the excluded references are available on request from the authors.

Most of the papers rejected at abstract review stage did not report on primary and autoimmune GN and had been found from the search using ‘kidney disease’ as the search term. Of the remaining papers, reasons for rejecting included those reporting on an ill sub-group of the population (e.g. those with systemic lupus erythematosus), those that reported on prevalence and not incidence, review papers and those that gave incidence rates as a percentage of people who had a renal biopsy.

For some types of GN, the terms used by the authors varied, some using the term to describe clinical presentation or syndrome, for example crescentic proliferative [4] or crescentic GN [5–7], nephrosis [4,8], rapidly progressive GN [9,10], acute GN [11] and acute nephritis [12], whereas others used non-specific terms such as vasculitis [13–17] and total GN [11]. These incidence rates have been excluded from the review because it was impossible to determine the extent to which these papers reported on primary and autoimmune GN.

Descriptions of the studies are given in Table 1, and incidence rates are displayed in forest plots showing rates for children in Figure 2 and adults or all ages in Figure 3.  Appendix 1 (Table A1) includes a table of all rates presented in the forest plots. Most of the studies included in this review investigated populations in Australasia [5,13,18], Europe [4,6–8,11,16,19–34] and North America [35–40] with five studies from the Middle East [15,41–44], two studies from South America [17,45], one study from Japan [46] and one study from Tunisia [47]. Findings for the different subtypes of GN are summarized below:

IgA nephropathy Four studies reported rates in children and teenagers: 0.03/100 000/year (CI₉₅ 0–0.1) in Venezuela [45], 0.08 (CI₉₅ 0.01–0.46)/100 000/year (1975–84) to 0.57 (CI₉₅ 0.23–1.18)/100 000/year (1985–94) (0–17 years) in Tennessee, USA [37], 4.5/100 000/year (0–15 years) in Japan [46] and 0.31/100 000/year (0–15 years) in Italy [27] (CI₉₅ indicates 95% confidence interval). The difference in these rates may be explained by the study type: the Japanese study was a screening study and therefore detected subclinical cases of IgA nephropathy as well as including three possibly prevalent cases (total 37) that did not have previous screening results recorded. In contrast, the studies from Tennessee and Italy were based on renal biopsy, and the study from Venezuela used chart reviews of symptoms and some biopsy results.

Most of the other studies were prospective, reporting rates from 0.2/100 000/year to 2.8/100 000/year [4–8,15,20,21,23,29,47]; the retrospective studies reported a similar range of rates, from 0.4/100 000/year to 2.9/100 000/year [11,13,16,22,31,32,39,40,42,43]. Two studies reported incidence rates of 5.0/100 000/year [30] and 5.7/100 000/year in males [13] but both had high biopsy rates which is likely to have contributed to the greater incidence rates found. Many of the studies included in this review were conducted between 1970 and 1990; no trend in incidence rates with time was discernable. Sixteen studies [5–8,11,15,16,22,23,27,29,31,36,39,42] reporting rates used immunofluorescence in diagnosing IgA nephropathy.

Membranous nephropathy Two studies [27,35] gave incidence rates in children and adolescents: 0.05/100 000/year (1985–93) to 0.09/100 000/year (1993–2002) (6 months–19 years) in Ottawa-Hull Canada [35] and 0.02/100 000/year (0–15 years) in Italy [27]. For the other studies, most were prospective and reported incidence rates between 0.3/100 000/year and 1.4/100 000/year [4–8,15,20,22,23,29,30]; the retrospective studies reported incidence to be between 0.2/100 000/year and 1.3/100 000/year [11,13,16,17,22,28,31–33,39,42,43]. Not many studies reported on differences in rates between males and females. In those that did, the numbers were low and no confidence intervals or indications of statistical significance for differences in incidence rates were available. Three studies [8,11,13] suggested that the rates were higher in males than females whereas El Reshaid et al. [15] reported the opposite. Taking into account the methods used in the studies presented, it is not thought that there has been a change in incidence between 1970 and 2000. The studies that used retrospectively collected data are likely to have missed cases and therefore underestimate the incidence rates. There was insufficient information to conclude reliably whether there is a difference in risk between males and females. Our best estimate of the incidence is 1.2/100 000/year.

Membrano-proliferative GN Thirteen studies were retrospective and six were prospective; the range of incidence rates in all studies was between 0.14/100 000/year and 0.93/100 000/year. Over time, incidence appears to have decreased from around 0.7/100 000/year in the 1970s [5,8,20] to 0.2 in the 1990s [4,7,11,16] with the exception of Covic et al. [28] who report a rate of 0.93/100 000/year in 2004. Covic et al. [28] link this to the higher rates of streptococcal infection and hepatitis B and C in their population; they also note a decrease in the prevalence of membrano-proliferative GN between 1995 and 2004, which they believe is associated with improvements in income, sanitation, social and medical infrastructure. Simon et al. [4,8] noted that there was an association between streptococcal infection and the onset of membrano-proliferative GN and that the decreases in incidence rates of membrano-proliferative and post-streptococcal GN were closely linked. Other differences in rates are noticeable: Hachicha et al. [47] found the highest rate of membrano-proliferative GN although this is probably an over-estimate of the true rate; in New Zealand [5] an incidence rate five times higher in Polynesians than in non-Polynesians was reported. This may indicate an infection-related glomerular-specific injury. Two studies [4,5] reported on the incidence of type I membrano-proliferative GN, two studies [11,16] reported that 68% of their cases were type I and in a third study [32] 90% were type I; the remaining studies [5,7,8,13,20,21,27–29,34,35,42,43] did not specify the type of membrano-proliferative GN diagnosed.

Mesangial proliferative GN Four studies included were prospective and three were retrospective; all gave incidence rates between 0.2/100 000/year and 1.1/100 000/year [5,7,8,28,29,33]. In order to differentiate mesangial proliferative GN from IgA nephropathy, immunofluorescence should be used. All seven studies used immunofluorescence in their studies but four reported that this was not performed on all biopsies [28,29,32,33], giving a potential overestimation of the incidence rate. The only study to make the distinction between IgA nephropathy and mesangial proliferative GN without IgA deposits was that by Schena et al. [7], who found an incidence rate of 0.16/100 000/year; therefore, it is likely that the ‘true’ incidence rate of mesangial proliferative GN is at the lower end of the range given.

Minimalchange disease In children, minimal change disease has been found to cause over 75% of cases of nephrotic syndrome [48]. Seven studies reported on incidence of nephrotic syndrome, which would produce a slight overestimation of incidence of minimal change disease [24,26,38,41,44,45]; three studies reported on incidence of minimal change nephrotic syndrome [25,27,35]. Incidence rates in children were between 0.23/100 000/year and 15.6/100 000/year [24–27,35,38,41,44,45]. The rates were reported with respect to ethnic origin and differences were noted: 0.23–2.8/100 000/year in Caucasian children [24,26,27,35], 2.4/100 000/year in Hispanic children [45], 3.4/100 000/year in Afro-Caribbean children [24], 7.2–11.6/100 000/year in Arabian children [41,44] and 6.2–15.6/100 000/year in Asian children who resided in the UK [24,25].

In the remaining studies, retrospective and prospective studies reported similar rates between 0.2/100 000/year and 0.8/100 000/year in adults [5,7,11,13,17,20,23,28,29,31–33,39,42,43]; no trend of changes over time was found. Taking into account the accuracy of these rates, our best estimate of incidence of minimal change disease in adults is 0.6/100 000/year.

Focal segmental glomerulosclerosis Most rates, whether from prospective or retrospective studies, were between 0.2/100 000/year and 1.1/100 000/year; the Australian study reported the highest rates of 2.5/100 000/year in males and 1.8/100 000/year in females. The latter may be due to the fact that in Australia, people are referred for biopsy more often than in other countries included in this review [13]. In the three studies that investigated differences in rates between males and females, incidence appeared to be higher in males [11,13,15]. However, no indication was given of the statistical significance of these differences and given the fact that the numbers of cases were low, they may have arisen by chance.

Other types of GN Incidence rates were presented for other types of GN; however, these were limited to just one or two studies per disease type. Details are given in the forest plots; generally, these rates were low at < 1.0/100 000/year. However, Becquet et al. [18], in their study of post-infectious GN in French Polynesia, reported an incidence rate of 18/100 000/year; this is likely to be an underestimate of the true rate. This country had a higher rate of bacterial infections due to the climatic conditions, greater numbers of people sharing residences, low socio-economic level and a lower use of medical care due to cultural beliefs; these factors are all thought to contribute to this higher incidence rate.

Accuracy of incidence rates

There are a number of components to assess when considering the incidence rates presented. Indication of likely accuracy of rates has been given in Table 1. Of key importance is whether the cases included in the numerator were new cases: six studies reported in their methods section that only new cases were included [4,6,8,11,24,26,40,41]. As GN is diagnosed by biopsy, the more liberal the biopsy policy, the greater the possibility of detecting all cases of the disease. Some studies gave their biopsy rate per head of population; these are given in Table 1 and vary between 1.08 and 24.7/100 000/year [8,13,16,20,21,23,28–33,39,42,43].

Discussion

This literature review found incidence rates for different types of primary autoimmune GN to be between 0.2/100 000/year and 2.5/100 000/year in adults. Most studies were from the USA and France; therefore, it is difficult to draw conclusions regarding variability of rates with geographical location or ethnicity. Given that GN can exist subclinically, and given differences in access to renal biopsy between different healthcare systems, it is likely that geographical variations in incidence rates found can be explained by differences in diagnosis rather than by genuine difference in disease frequency.

It is useful to note the type of studies undertaken to determine incidence of GN; in reviews of incidence of other autoimmune diseases [1,2], prospective studies are thought to have given more accurate rates than retrospective studies. However, in the studies presented here, the rates reported were consistent irrespective of the study type.

Most studies used biopsy to diagnose the disease for the majority of cases; however, only five studies [5,23,29,40,44] reported the guidelines used for diagnosis. Classification of the disease and detection threshold of signs and symptoms may cause variation in incidence rates between studies [13,16,23], and the lack of a central histopathology review may have caused within-study variations [16]. Wyatt et al. [36] reported that over time there was an increased recognition of IgA nephropathy; therefore, the incidence rates from the end of the study may be more accurate. Similarly, Mazzuchi et al. [17] reported rates using a national registry and found an increase in incidence with time; they are reported to be due to a greater awareness of the disease and earlier diagnosis.

Cases of IgA nephropathy can exist subclinically and therefore will only be diagnosed through routine urinary tests or if a patient presents with severe symptoms [29]. Simon et al. [8] reported that 60% of cases of idiopathic IgA nephropathy were discovered by chance through routine testing as part of a medical examination in employment. Screening populations for conditions that can exist subclinically will produce higher and more accurate incidence rates especially when done routinely so that prevalent cases are not included in the incidence rate. In this review, one study [46] used a regular screening programme to find cases, and the rate produced was greater by nearly an order of magnitude than other comparable incidence rates. Sehic et al. [37] reported that many cases of IgA are never diagnosed and that limited access to medical care for those of lower socio-economic status may explain some failure to diagnose in this study.

There is the possibility of cases of secondary GN being included as primary cases particularly in retrospective studies; this would lead to an overestimation of incidence rates. There are links, for example between membrano-proliferative GN and hepatitis B or hepatitis C. Variations between countries in terms of rates of infections will contribute to differences in incidence rates.

One of the key factors in explaining differences in incidence rates of GN is the difference in referral and biopsy policies between different countries and even between regions of countries [16,23,30,31]. Briganti et al. [13] found higher incidence rates of minimal change disease and focal segmental glomerulosclerosis than other studies. They claim that this may be as a result of a more liberal biopsy policy, leading to the detection of less well-defined or asymptomatic cases. In contrast, The New Zealand GN Study [5] noted that children and teenagers with post-infectious GN or minimal change nephropathy are rarely biopsied and older patients may also be less likely to have biopsies taken than younger adults. There is evidence that Polynesians have poorer access to healthcare resources in New Zealand than others, and in this study, a high proportion of Polynesian patients presented for the first time with end-stage renal failure, making it impossible to reach a specific diagnosis. Lack of access to healthcare may also have resulted in prevalent cases being included in the incidence rates for GN in this population. Naumovic et al. [32] reported that their low biopsy rate (1.08/100 000/year) may be due to economic sanctions and that elderly people and those with diabetes do not routinely undergo biopsy. Incidence was found to be lower in African-American children compared to Caucasian children [37]. The study authors suggested that this may be because the decision to perform a biopsy for those suspected to have mild disease was referred to their parents who may have been less likely than Caucasian parents to give their consent; records were not available for the number of biopsies refused [19].

Renal biopsy policy also affects diagnosis rates in the elderly: a number of studies reported increases in the number of elderly patients biopsied as a result of a change in policy and increasing age of the population [4,6,11,36] with Stratta et al. [11] reporting that rates of elderly patients diagnosed increased from 1.6% in 1970 to 20.4% in 1994. Wyatt et al. reported an increase in incidence at 45 years and over which was thought to be due to a more proactive attitude towards conducting diagnostic procedures in the elderly. It is likely that incidence in older people is underestimated as not all cases were referred to specialists or underwent biopsies [11].

Conclusion

Reported incidence rates of GN in adults varied between 0.2 and 2.5/100 000/year depending on the type of GN. Incidence in children was generally lower with most rates around 0.1/100 000/year; two exceptions to this were that incidence of minimal change disease in children was around 2.0/100 000/year and a screening study that reported a rate of IgA nephropathy of 4.5/100 000/year in children in Japan.

The reported incidence rates are likely to underestimate true rates of IgA nephropathy as this disease can exist subclinically and may never be detected; however, other types of GN may be overestimated due to relapses and prevalent cases being counted as incident. There is variation in biopsy policy between countries, which affects the incidence rates found. Incidence in older people appears to have increased over time; this is considered to be due to greater inclusion of this age group in referrals for biopsy rather than due to a genuine increase in disease occurrence.

This work was supported by a grant from GSK Biologicals.

Conflict of interest statement. None declared.

Appendix

References