We identified a total of 23 cases of Nocardia in heart and lung transplant recipients at St Vincent’s Hospital during June 2015–March 2021; of those, 16 were HTR and 7 were LTR (Figure 1, panel A). Compared with the years before and after the outbreak period (2018–2019), we saw no significant change in the annual total of heart and lung transplants performed (40–50/year). Over the entire study period, the total number of transplants performed was similar across the 2 services: 289 lung transplants and 275 heart transplants. There was no increase in the number of cases of Nocardia across other hospital departments during the study period: 2 cases during the preoutbreak period (June 2015–December 2017), 3 during the outbreak period (January 2018–December 2019), and 2 during the postoutbreak period (January 2020–March 2021).

Overall, the median age at Nocardia diagnosis was 59 (range 38–71) years; 15 (65.2%) patients were male (Table 1). Sites of Nocardia infection were lung (11 HTR, 4 LTR); bloodstream (1 HTR); skin (1 HTR); lung and bloodstream (1 LTR); lung and brain (2 HTR); skin and brain (1 LTR); lung, skin, and brain (1 LTR); and lung, brain, and kidney (1 HTR).

When comparing heart and lung transplant recipients, we saw no substantial differences in age at Nocardia diagnosis, sex, CMV mismatch status, use of intravenous immunoglobulin, CMV viremia, significant CMV viremia, diagnosis of other respiratory infections within 6 months preceding the Nocardia diagnosis, or use of sulfamethoxazole/trimethoprim for Pneumocystis jirovecii prophylaxis. In addition, all heart and lung transplant recipients received a similar combination of maintenance immunosuppression therapy with tacrolimus, mycophenolate mofetil, and prednisone.

We identified significant differences between LTR and HTR that suggested greater immunosuppression in HTR before and at the time of Nocardia diagnosis (Table 1). These differences included a shorter median time from transplant to Nocardia diagnosis (4.8 months, vs. 22.8 months in LTR; p<0.05), higher prevalence of diabetes at the time of Nocardia diagnosis (81.3%, vs. 28.6% in LTR; p<0.05), greater use of basiliximab for induction immunosuppression (100%, vs. 40.0% in LTR; p<0.05), increased rates of any grade of cellular rejection at any time before Nocardia diagnosis (100%, vs. 42.9% in LTR; p<0.05), and increased rates of moderate to severe rejection at any time before Nocardia diagnosis (62.5%, vs. 0 in LTR; p<0.05). In addition, azithromycin prophylaxis rates were lower in HTR (6.25%) than in LTR (100%) (p<0.05).

We found no significant difference in the types of Nocardia species between HTR and LTR (Table 2) and no apparent clustering of Nocardia species over time to suggest a single source of the outbreak (Figure 1, panel B). We also found no differences between HTR and LTR in Nocardia susceptibilities to amikacin, amoxicillin/clavulanic acid, ceftriaxone, ciprofloxacin, clarithromycin, doxycycline, imipenem, linezolid, minocycline, moxifloxacin, and tobramycin (Table 3). Nocardia in LTR were more likely to be susceptible to cefepime than in HTR (50% vs. 6.7%; p<0.05), however, this difference may have been related to differences in Nocardia species across type of transplant.

Six Nocardia cases were located outside Greater Sydney, and 17 were within the Greater Sydney region (Figure 2). During June 2015–March 2021, no Nocardia patients changed their residential address within 3 months before Nocardia diagnosis. The 6 Nocardia cases outside Greater Sydney were situated in northern coastal NSW (n = 4), central NSW (n = 1), and Tasmania (n = 1). In western Sydney, 2 clusters of Nocardia were in geographic proximity (Figure 2, panel B). However, those clusters did not represent an outbreak because different species were identified within each cluster and cases diagnosed in the clusters were separated in time by >6 months.

Because 74% of cases were from Greater Sydney, our climate data analysis focused on that region. We chose the suburb of Greystanes as the reference location for all climate data comparisons because of its central geographic location. Five Nocardia cases were reported from Greater Sydney during June 2015–December 2017 (0.16/month), of which 1 (20%) was an HTR. During January 2018–December 2019, a total of 11 Nocardia cases (0.46/month) were diagnosed, of which 10 (91%) were HTR; during January 2020–March 2021, there was 1 Nocardia case (0.07/month) in an HTR (Figure 3). The increased incidence of Nocardia from January 2018 to December 2019 occurred during times of lower rainfall and a drier surface (Figure 3).

When comparing averaged values of climate parameters in Greater Sydney between the time of the outbreak (January 2018–December 2019), and the periods before (June 2015–December 2017) and after (January 2020–March 2021), we found the outbreak coincided with average monthly precipitation of 40.0 mm/month, significantly lower than that for the time periods before and after (p<0.01) (Table 4). During the same period, average monthly dryness reached its lowest (driest) value of 0.3 compared with the time periods before and after (p<0.01). We observed no significant differences in the average monthly temperature, windspeed, or erodibility during the outbreak period (January 2018–December 2019) (Table 4), compared with the time periods before and after. However, when we compared months with a Nocardia diagnosis with months without a Nocardia diagnosis, windspeed was higher (0.4 vs. 0.2 m/s; p<0.01) and surface levels were drier (0.3 vs. 0.4; p<0.01); because erodibility is derived from dryness and windspeed, erodibility was also higher (0.4 vs. 0.1 m/s for months without a Nocardia diagnosis; p<0.01). The average temperature was also higher in months with a Nocardia diagnosis (18.9°C vs. 17.4°C in months without a Nocardia diagnosis), although this difference was not statistically significant (p<0.24) (Table 5).

In this study, we observed no differences in Nocardia species isolated from HTR compared with LTR and no clustering of Nocardia species in space or time to suggest a single source for the outbreak. We found that HTR had evidence of greater immunosuppression before their Nocardia diagnosis than LTR, including higher use of basiliximab for induction immunosuppression, higher rates of cellular rejection, and a shorter median time from transplant to Nocardia diagnosis. Our analysis of climate data revealed that low local precipitation and drier surface levels correlated with increased incidence of Nocardia diagnosis.

Previous studies have shown that high-dose steroids, calcineurin inhibitor usage, CMV disease in the 6 months before diagnosis, and increased age are risk factors for nocardiosis (3,14). In our cohort, HTR and LTR did not have demonstrable differences in age, CMV disease status, or use of tacrolimus, mycophenolate mofetil, or prednisone. However, our data indicate that HTR had more characteristics indicating immunosuppression than LTR, including higher rates of basiliximab induction, diabetes, and cellular rejection (including treated moderate-to-severe rejection). Because basiliximab is an immunosuppressant with interleukin-2R α antagonist properties, increased use of basiliximab among HTR compared with LTR may have increased the risk for Nocardia infection through a diminished T-cell response in HTR. Similarly, additional immunosuppression in the setting of treating moderate to severe acute organ rejection may have further increased the risk for Nocardia infection among HTR.

We found no difference in the proportion of macrolide susceptible Nocardia isolates in HTR versus LTR despite increased use of azithromycin for prophylaxis in LTR compared with HTR. The possibility that azithromycin use lowered the risk for Nocardia infections in LTR compared with HTR warrants further investigation.

Seasonality and changes in climate conditions are known to affect the incidence and distribution of various infectious pathogens, particularly those acquired by inhalation or exposure to the respiratory tract (15). Respiratory viruses and environmental fungi, such as Coccidioides immitis, Aspergillus spp., and Cryptococcus gattii, are highly influenced by climate patterns, having enhanced infection risk in the setting of increased aerosolised dust particulate matter (9,10,16,17). In a study of Aspergillus infections among stem cell transplant recipients in the United States, Panackal et al. identified an increased incidence of invasive Aspergillosis in drier and warmer conditions by comparing incidence rates across different seasons (16). Tong et al. identified a correlation between increased frequencies of dust storms, precipitated by low moisture level soils, and higher rates of inhaled soil-dwelling fungi such as Coccidioides immitis and C. posadasii (17). Similarly, Majeed et al. (9) and Saubolle et al. (10) observed that the greatest number of Nocardia infections occur in dry, warm climates, such as in the Southwest United States.

In Greater Sydney, the increase in Nocardia cases occurred during a time of decreased rainfall and a very dry surface (evaporation/potential evaporation), supporting our hypothesis that the increase in Nocardia infections may have been driven by extreme climate conditions. In NSW, 2019 was the warmest and driest year, with the annual mean temperature 1.95 degrees above average and average rainfall 55% below average (6); 2018 was NSW’s second warmest and seventh driest year, with an annual mean temperature 1.68 degrees above average and average rainfall 40% below average (5).

The Australian Therapeutic Guidelines for treating moderate nocardiosis recommend dual therapy with sulfamethoxazole/trimethoprim and either ceftriaxone or linezolid; for severe nocardiosis, a third agent (amikacin, imipenem or meropenem), is added to this regimen (18). Most Nocardia isolates identified in this cohort were reassuringly susceptible to sulfamethoxazole/trimethoprim (95.7%); however, only 30.4% of isolates were susceptible to ceftriaxone and 43.5% to imipenem, corresponding well with previous data (19). All isolates were susceptible to linezolid and amikacin, supporting the usage of empiric linezolid or amikacin in addition to sulfamethoxazole/trimethoprim over ceftriaxone or imipenem for moderate and severe disease in our cohort.

Our study’s first limitation is that our small study population size limited our analyses to primarily descriptive statistics and correlations with climate data. Most of our patients were from the Greater Sydney region and underwent heart or lung transplantation at a single medical center, further limiting the generalizability of our findings. The retrospective study design and reliance on data from the medical record limited our ability to comment on patient practices, including use of personal protective equipment such as masks when gardening and participating in other activities that may increase the risk for infection. However, postoperative patient education on use of personal protective equipment did not change during the study period; we would therefore not expect to see changes in patient practices that would affect the incidence of Nocardia infections. The retrospective study design meant that data on intraoperative care, such as changes in procedures, personnel, and infection control practices, were not available. However, because the surgical teams and theater conditions remained consistent throughout the study period, differences in intraoperative care that may have affected risk for Nocardia infection would be unlikely. Data on environmental sources within and around the hospital, such as construction projects and water sources, were also not available. However, given the broad range in times from transplant to diagnosis and the multispecies nature of the outbreak, it is unlikely that environmental contamination was a potential source of the outbreak. Although more surveillance biopsies are routinely performed among HTR than LTR at our institution, which may have contributed to the higher rates of any grade of rejection detected in HTR, treated rejection episodes (i.e., moderate to severe rejection) are likely to be symptomatic and therefore unlikely to be biased by differences in routine surveillance biopsy schedules. Nocardia infections can be indolent and subclinical for some time, which limited our analyses comparing climate conditions in months with and without Nocardia diagnoses. Larger studies and studies in other regions are needed to confirm the correlations we identified between climate conditions and Nocardia infections.

In conclusion, tThis study examined the association between climate conditions and Nocardia infection in HTR and LTR. Periods of low precipitation and dryer surface levels were associated with an increased incidence of Nocardia diagnoses, suggesting that environmental conditions affected the risk for Nocardia infection. In addition, HTR had a shorter time to Nocardia diagnosis and increased rates of markers of immunosuppression, suggesting that HTR had greater susceptibility to infection at the time of diagnosis than LTR. We hypothesize that the increased environmental risk from climate conditions during 2018–2019, coupled with increased host susceptibility related to immunosuppression in HTR, may explain the increased incidence of Nocardia infections during 2018–2019. Further studies should evaluate the influence of climate characteristics on Nocardia infections in immunocompromised hosts, as well as potential screening or other preventive measures that might decrease disease burden in these vulnerable patients. Because Nocardia is an environmental organism, these results highlight the importance of wearing personal protective equipment around soil exposures, particularly for immunocompromised patients during periods of increased soil dryness.